paperID
stringlengths 36
36
| pwc_id
stringlengths 8
47
| arxiv_id
stringlengths 6
16
⌀ | nips_id
float64 | url_abs
stringlengths 18
329
| url_pdf
stringlengths 18
742
| title
stringlengths 8
325
| abstract
stringlengths 1
7.27k
⌀ | authors
stringlengths 2
7.06k
| published
stringlengths 10
10
⌀ | conference
stringlengths 12
47
⌀ | conference_url_abs
stringlengths 16
198
⌀ | conference_url_pdf
stringlengths 27
199
⌀ | proceeding
stringlengths 6
47
⌀ | taskID
stringlengths 7
1.44k
| areaID
stringclasses 688
values | embedding
stringlengths 9.26k
12.5k
| umap_embedding
stringlengths 29
44
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
c3fe9d83-260a-4119-91ad-9f56f23582e0 | inter-beat-interval-estimation-with-tiramisu | 2107.00693 | null | https://arxiv.org/abs/2107.00693v1 | https://arxiv.org/pdf/2107.00693v1.pdf | Inter-Beat Interval Estimation with Tiramisu Model: A Novel Approach with Reduced Error | Inter-beat interval (IBI) measurement enables estimation of heart-rate variability (HRV) which, in turns, can provide early indication of potential cardiovascular diseases. However, extracting IBIs from noisy signals is challenging since the morphology of the signal is distorted in the presence of the noise. Electrocardiogram (ECG) of a person in heavy motion is highly corrupted with noise, known as motion-artifact, and IBI extracted from it is inaccurate. As a part of remote health monitoring and wearable system development, denoising ECG signals and estimating IBIs correctly from them have become an emerging topic among signal-processing researchers. Apart from conventional methods, deep-learning techniques have been successfully used in signal denoising recently, and diagnosis process has become easier, leading to accuracy levels that were previously unachievable. We propose a deep-learning approach leveraging tiramisu autoencoder model to suppress motion-artifact noise and make the R-peaks of the ECG signal prominent even in the presence of high-intensity motion. After denoising, IBIs are estimated more accurately expediting diagnosis tasks. Results illustrate that our method enables IBI estimation from noisy ECG signals with SNR up to -30dB with average root mean square error (RMSE) of 13 milliseconds for estimated IBIs. At this noise level, our error percentage remains below 8% and outperforms other state of the art techniques. | ['Hassan Ghasemzadeh', 'Behrooz A. Shirazi', 'Roozbeh Jafari', 'Seyed Iman Mirzadeh', 'Ali Akbari', 'Asiful Arefeen'] | 2021-07-01 | null | null | null | null | ['heart-rate-variability'] | ['medical'] | [ 3.14562649e-01 -3.96713823e-01 2.17536598e-01 -1.13719717e-01
-6.00889683e-01 -3.40324283e-01 -1.88589498e-01 1.61971465e-01
-3.31647247e-01 9.02452767e-01 2.14611441e-01 1.16319679e-01
-3.06278616e-01 -5.47297716e-01 -1.93713397e-01 -1.00592291e+00
-3.22873056e-01 -1.70983911e-01 -4.54349667e-01 -2.68965424e-03
-7.42163658e-02 3.57876509e-01 -9.49929774e-01 1.09520080e-02
8.79991412e-01 1.24037611e+00 -1.49631262e-01 7.62813747e-01
3.75514239e-01 3.46721292e-01 -1.27862597e+00 -6.08104132e-02
2.74458025e-02 -7.05961049e-01 -2.96918362e-01 -2.31139526e-01
-1.11970976e-01 -5.32092094e-01 -4.29490298e-01 9.41065252e-01
1.10040522e+00 -8.14756006e-02 4.54587519e-01 -6.36989415e-01
-1.32873043e-01 2.05195218e-01 -5.57102621e-01 6.12336397e-01
1.84941694e-01 1.70576096e-01 2.55609453e-01 -5.22586942e-01
2.32765749e-01 5.33586740e-01 1.40097010e+00 2.63830066e-01
-1.35367393e+00 -5.15731990e-01 -5.35101593e-01 3.70103329e-01
-1.44719362e+00 -2.37205088e-01 1.13939643e+00 -2.28974491e-01
5.04233837e-01 5.07655323e-01 9.88678336e-01 1.01163614e+00
6.15034878e-01 2.70787209e-01 1.08648133e+00 -1.42919764e-01
6.86970502e-02 -3.70730400e-01 2.65076328e-02 3.64966765e-02
4.04816329e-01 1.40666023e-01 -2.50946283e-01 -8.37505795e-03
8.57414961e-01 3.43496561e-01 -6.84554696e-01 5.11891305e-01
-1.39356136e+00 3.41111124e-01 4.15905505e-01 5.34679651e-01
-9.02953088e-01 8.03226829e-02 5.71382940e-01 4.73980397e-01
3.75497103e-01 2.81194091e-01 -3.41693163e-01 -3.25837433e-01
-1.02450109e+00 6.23390526e-02 5.87048173e-01 2.42764577e-01
1.44944608e-01 5.50474286e-01 -2.54670143e-01 6.57716990e-01
2.99625564e-02 6.67789578e-01 6.30388081e-01 -1.07248986e+00
7.75338113e-02 2.72178024e-01 2.80186176e-01 -1.42881918e+00
-5.80385387e-01 -1.00991499e+00 -1.69726872e+00 -2.33526424e-01
4.52886969e-01 -3.69926631e-01 -4.65071797e-01 1.35969365e+00
5.58788121e-01 3.99715722e-01 -5.34085408e-02 1.19344628e+00
9.03670549e-01 6.67190552e-01 3.04107182e-02 -8.15478444e-01
1.30358303e+00 -3.66689148e-03 -1.22017193e+00 -9.45400000e-02
1.31506816e-01 -4.84072298e-01 5.26607752e-01 6.98299348e-01
-7.89639235e-01 -8.78520310e-01 -1.09731436e+00 2.31401548e-01
4.58009660e-01 1.26020417e-01 2.31181279e-01 5.95402122e-01
-6.12732470e-01 1.01099491e+00 -8.08658838e-01 9.00372267e-02
3.63655120e-01 1.09804079e-01 -3.16211402e-01 1.45527884e-01
-1.43215442e+00 6.12762809e-01 3.91802117e-02 8.24851573e-01
-4.70196277e-01 -8.46226752e-01 -5.41164994e-01 -4.03327122e-03
7.06311688e-02 -6.70203686e-01 7.84924209e-01 -9.65680838e-01
-1.27272737e+00 3.57085168e-01 -1.92192778e-01 -4.65174824e-01
4.82422322e-01 -4.72966492e-01 -8.41107965e-01 3.16286176e-01
-2.16356382e-01 -2.31119126e-01 1.41201723e+00 -8.78081620e-01
1.36238605e-01 -5.63953638e-01 -6.01731062e-01 -5.25633544e-02
-1.56969994e-01 -2.52026707e-01 5.01963198e-02 -9.23059702e-01
5.08216619e-01 -7.01278627e-01 -7.34640881e-02 -1.70103267e-01
-2.08929420e-01 3.26732755e-01 4.17664081e-01 -1.33736622e+00
1.65711701e+00 -2.39644432e+00 -1.23557188e-02 2.46100470e-01
4.62675780e-01 5.81243813e-01 2.71759152e-01 2.34700277e-01
-6.38936460e-02 1.45843402e-01 -3.46557438e-01 3.43001604e-01
-5.58606863e-01 5.09453006e-02 6.75546080e-02 7.86388218e-01
1.16708346e-01 6.51143730e-01 -6.46835566e-01 -2.55840153e-01
2.43130520e-01 8.60316634e-01 -4.31802869e-02 1.86339915e-01
5.77592969e-01 1.07135451e+00 -3.15412730e-01 5.44166028e-01
6.14519238e-01 -2.02696715e-02 2.44275287e-01 -9.24498975e-01
1.39125809e-01 -2.93336585e-02 -1.31168997e+00 1.39703798e+00
-2.48832256e-01 7.15782762e-01 1.42648518e-01 -1.31662035e+00
1.01742458e+00 6.74935520e-01 6.03096664e-01 -8.21758926e-01
2.69112855e-01 7.28843361e-02 2.64533967e-01 -8.84697974e-01
-3.28809381e-01 -4.26967919e-01 2.03814074e-01 1.83060139e-01
-3.63589078e-01 2.19154730e-01 -2.02535659e-01 -4.29666191e-01
1.02975273e+00 -5.87292872e-02 5.76135814e-01 -1.69614524e-01
4.58433568e-01 -5.28429508e-01 9.76623952e-01 7.68256903e-01
-5.58688164e-01 6.94197416e-01 1.35253444e-01 -9.23665345e-01
-7.26574183e-01 -9.03588355e-01 -3.79947513e-01 1.80635661e-01
-2.21675150e-02 -4.56082486e-02 -6.51046157e-01 -3.91713344e-02
-7.83984456e-03 1.54606581e-01 -3.42675269e-01 -2.45593905e-01
-8.69377613e-01 -1.00451446e+00 6.87389195e-01 7.74008811e-01
5.22957504e-01 -9.88525033e-01 -8.40924621e-01 6.99291229e-01
-6.95804775e-01 -9.66180325e-01 -2.30337501e-01 -1.37761859e-02
-1.33895457e+00 -9.09217596e-01 -1.06571448e+00 -3.33548367e-01
3.81820172e-01 -1.08304240e-01 1.14831221e+00 2.85036415e-01
-7.89048553e-01 1.17394432e-01 -5.08424900e-02 -6.22020423e-01
-2.19522834e-01 -3.83223981e-01 9.08262730e-02 1.98419690e-01
1.33013353e-01 -8.68350148e-01 -1.36399460e+00 1.45643964e-01
-6.57615244e-01 -3.47631246e-01 5.00309825e-01 9.10259128e-01
5.35464287e-01 1.99729934e-01 1.16043723e+00 -3.70305568e-01
7.34989524e-01 -2.84740269e-01 -2.61343181e-01 -3.52580279e-01
-4.39428568e-01 -4.62144226e-01 6.40796423e-01 -5.26813924e-01
-8.09387207e-01 -3.59192818e-01 -2.86693573e-01 -1.27185851e-01
-2.66544074e-01 5.34772456e-01 -4.01017554e-02 2.97550559e-01
7.26605237e-01 3.75309587e-01 3.81226152e-01 -2.17117205e-01
-3.05918515e-01 7.71420717e-01 7.63467610e-01 -1.79774895e-01
3.83278370e-01 4.03836757e-01 1.56767815e-01 -1.34613395e+00
-5.06566465e-01 -5.11871397e-01 -3.33675563e-01 -3.49553823e-01
8.22877467e-01 -1.13832951e+00 -9.86939132e-01 7.07571507e-01
-1.05911708e+00 1.09064328e-02 -9.35155377e-02 7.17719018e-01
1.45676574e-02 6.70995653e-01 -8.65591288e-01 -1.01123750e+00
-1.02622581e+00 -6.96525514e-01 6.00871086e-01 2.23258317e-01
-5.99829853e-01 -7.51839459e-01 -5.81593141e-02 5.13316512e-01
6.53146148e-01 9.51427937e-01 7.70286798e-01 -2.39445210e-01
-1.18102185e-01 -5.91916323e-01 1.07396662e-01 6.47361755e-01
2.17959344e-01 -3.51840675e-01 -9.31671917e-01 -2.50580281e-01
6.66408479e-01 2.99505711e-01 4.20649558e-01 7.96669126e-01
1.08409560e+00 -3.23625833e-01 -9.73286107e-02 5.55700839e-01
1.40247607e+00 3.51117432e-01 9.69743967e-01 3.55362818e-02
5.39174736e-01 3.31681132e-01 2.39085913e-01 7.33145714e-01
-2.94018593e-02 2.72452056e-01 1.87636226e-01 -2.48311996e-01
4.99849916e-02 6.10813498e-01 7.74302334e-02 1.07193708e+00
-6.12708449e-01 3.75493057e-02 -7.05113888e-01 5.47775626e-01
-1.39923310e+00 -1.01303148e+00 -6.16448224e-01 2.30394626e+00
1.01167309e+00 -6.51991665e-02 1.41460776e-01 8.62942636e-01
6.93857968e-01 -3.03416587e-02 -7.43654549e-01 -1.68999031e-01
4.04220633e-03 4.37264949e-01 7.92161524e-02 5.91909364e-02
-1.00962389e+00 -2.03406364e-01 5.60531807e+00 -1.08766090e-02
-1.46424532e+00 2.85088294e-03 6.56965435e-01 8.91598240e-02
2.29148805e-01 -5.10198712e-01 -1.88814774e-01 7.22014844e-01
1.17890704e+00 1.46018431e-01 2.84200639e-01 4.97748315e-01
5.63233495e-01 -3.51141244e-02 -7.05705106e-01 1.47148752e+00
-1.61433175e-01 -9.40975726e-01 -6.02214277e-01 -2.62727022e-01
3.45737249e-01 -4.43465054e-01 -1.40064329e-01 1.33861363e-01
-1.01054978e+00 -1.06903076e+00 2.18653992e-01 9.73291636e-01
8.61638069e-01 -9.25473213e-01 1.21718001e+00 4.24078435e-01
-9.74423170e-01 -2.12094992e-01 -1.84306920e-01 -2.08437532e-01
7.28514269e-02 1.54126489e+00 -3.82404596e-01 4.94540453e-01
7.35185564e-01 6.73367381e-01 1.34191466e-02 1.07992899e+00
-1.39161259e-01 1.03552842e+00 -2.76410818e-01 2.96852350e-01
-5.86046994e-01 -2.57029891e-01 6.70730114e-01 8.97444308e-01
5.39124310e-01 5.42542458e-01 -5.01035363e-04 6.46436453e-01
-3.35573219e-02 -1.76169351e-01 -3.62161964e-01 1.30428046e-01
4.51561719e-01 1.08030748e+00 -4.29664075e-01 -3.85759592e-01
-1.98614046e-01 8.80065262e-01 -5.84506452e-01 4.78266090e-01
-7.66589105e-01 -6.90698326e-01 4.46542948e-01 3.38428259e-01
-1.17963962e-01 -7.96897262e-02 -5.13308287e-01 -9.60401416e-01
3.50036681e-01 -1.26316798e+00 3.07330233e-03 -3.72854501e-01
-1.08629584e+00 5.34690917e-01 -3.96535397e-01 -1.33551049e+00
-2.39079773e-01 -1.57900695e-02 -7.25744903e-01 9.54504550e-01
-1.26218379e+00 -3.65689039e-01 -5.82529724e-01 3.97826046e-01
2.23391995e-01 3.45092505e-01 1.02042603e+00 6.85799837e-01
-5.13793349e-01 2.69530594e-01 5.20492829e-02 2.48405814e-01
5.87960243e-01 -1.13833058e+00 2.75755152e-02 7.37850964e-01
-1.68844253e-01 7.64332771e-01 8.55456173e-01 -5.87118685e-01
-1.44004023e+00 -1.04770410e+00 7.33713567e-01 1.25990948e-02
1.67655721e-01 2.19950095e-01 -1.16225076e+00 -5.19462721e-03
1.07144088e-01 4.07828212e-01 8.09457600e-01 -1.72890320e-01
3.45938742e-01 -8.19443464e-01 -1.04254580e+00 3.44377220e-01
5.51077843e-01 -4.95334893e-01 -6.09758198e-01 -9.71749946e-02
-4.01656069e-02 -2.97024488e-01 -1.43149412e+00 7.12562561e-01
8.97136569e-01 -8.69515657e-01 1.15266418e+00 -1.16772555e-01
4.64639254e-02 -3.92019004e-01 3.61769557e-01 -1.37936652e+00
-3.43365997e-01 -7.92449355e-01 -3.75856370e-01 1.00809658e+00
-1.91801041e-01 -4.86201704e-01 4.43503737e-01 3.40660661e-01
2.59035118e-02 -5.75891912e-01 -8.76292646e-01 -4.82517391e-01
-5.20929456e-01 -3.39292347e-01 1.80410650e-02 8.69333625e-01
-3.05873156e-01 2.05387697e-01 -6.26458824e-01 2.02082008e-01
9.47426677e-01 -5.09245731e-02 4.32608455e-01 -1.47082877e+00
-2.66446084e-01 9.71853361e-02 -4.53423500e-01 -5.30557930e-01
-2.70207763e-01 -3.25208873e-01 5.11812232e-02 -1.53358126e+00
-1.75926119e-01 9.03707668e-02 -6.26844406e-01 2.07957879e-01
-5.69360435e-01 5.18277109e-01 -4.19311188e-02 1.17845252e-01
-7.54490634e-03 3.16855431e-01 1.17406905e+00 -2.56022811e-01
-5.35065711e-01 2.21176833e-01 -4.40884650e-01 9.34971333e-01
6.81050599e-01 -5.36024094e-01 -3.01837057e-01 -1.28397539e-01
1.38584882e-01 6.44709766e-01 4.17359561e-01 -1.36316848e+00
-1.44659519e-01 3.91153395e-01 1.03369868e+00 -5.27104199e-01
1.19618140e-01 -8.03933680e-01 7.24510908e-01 6.00985467e-01
-1.18567571e-02 1.15197478e-02 1.74453810e-01 6.08842671e-01
-4.55431342e-01 1.80805445e-01 6.71518266e-01 -1.53652087e-01
-2.36451820e-01 6.53289333e-02 -6.50597870e-01 1.42237440e-01
5.76375961e-01 -3.97429705e-01 1.59398869e-01 -4.02284026e-01
-1.02277410e+00 -2.45140046e-01 -3.30027908e-01 -2.33579621e-01
9.07087445e-01 -1.17981112e+00 -8.24085295e-01 1.22240625e-01
-2.60481864e-01 2.28842371e-03 8.07100058e-01 1.50869215e+00
-6.85887098e-01 -1.56094983e-01 -9.89734083e-02 -7.63014615e-01
-1.22307861e+00 1.77599296e-01 2.80344456e-01 -1.31549701e-01
-1.01181364e+00 4.71214861e-01 -3.61419678e-01 5.40074229e-01
2.30842620e-01 -4.72958386e-01 -2.33062580e-01 2.66321361e-01
8.17227364e-01 6.79328501e-01 2.58056283e-01 -3.21028173e-01
-4.23403144e-01 6.33798122e-01 2.99003154e-01 4.33426499e-01
1.33444691e+00 -3.54669809e-01 -4.10889648e-02 5.00906169e-01
1.00810349e+00 -1.63127750e-01 -1.04906356e+00 8.30033794e-02
-1.04260802e-01 -1.78840846e-01 3.77579272e-01 -7.43027091e-01
-1.16211152e+00 1.06026137e+00 1.15619600e+00 2.61949897e-01
1.59076548e+00 -8.80884409e-01 1.30207384e+00 2.10511521e-01
1.80362403e-01 -9.33639824e-01 1.41064212e-01 -9.97425914e-02
8.65414202e-01 -1.04547942e+00 1.70514047e-01 9.46106091e-02
-5.30707955e-01 1.24368060e+00 -1.50779411e-01 -2.25684017e-01
7.47489214e-01 2.16767684e-01 4.22354162e-01 1.84497431e-01
-1.02544412e-01 2.44750753e-01 2.29191378e-01 7.30267227e-01
7.04576969e-01 -3.97403538e-02 -5.44043779e-01 8.14998507e-01
1.96445897e-01 3.72424096e-01 3.45066309e-01 8.10975552e-01
-2.23808512e-01 -4.96590674e-01 -6.25775695e-01 5.10948896e-01
-1.22355235e+00 -2.27602031e-02 3.34370583e-01 4.59350884e-01
4.98198569e-02 1.12157142e+00 -2.55721241e-01 -4.86590266e-02
3.32569331e-01 1.50269330e-01 3.41150403e-01 -5.89862242e-02
-6.88689530e-01 5.13533294e-01 1.01305428e-03 -5.33528566e-01
-4.98143226e-01 -2.94539511e-01 -1.11960053e+00 -7.51428828e-02
-3.00488114e-01 -4.97898273e-02 6.69595182e-01 8.79811287e-01
4.13991898e-01 9.72421825e-01 5.71923435e-01 -6.63201809e-01
-5.84332168e-01 -1.02495527e+00 -7.29518950e-01 6.41386986e-01
8.36692035e-01 -1.63043380e-01 -2.98481911e-01 3.55444789e-01] | [14.259604454040527, 3.177182912826538] |
29266f30-49c8-40f0-b5ac-97617e885311 | image-to-image-translation-for-autonomous | 2209.11673 | null | https://arxiv.org/abs/2209.11673v1 | https://arxiv.org/pdf/2209.11673v1.pdf | Image-to-Image Translation for Autonomous Driving from Coarsely-Aligned Image Pairs | A self-driving car must be able to reliably handle adverse weather conditions (e.g., snowy) to operate safely. In this paper, we investigate the idea of turning sensor inputs (i.e., images) captured in an adverse condition into a benign one (i.e., sunny), upon which the downstream tasks (e.g., semantic segmentation) can attain high accuracy. Prior work primarily formulates this as an unpaired image-to-image translation problem due to the lack of paired images captured under the exact same camera poses and semantic layouts. While perfectly-aligned images are not available, one can easily obtain coarsely-paired images. For instance, many people drive the same routes daily in both good and adverse weather; thus, images captured at close-by GPS locations can form a pair. Though data from repeated traversals are unlikely to capture the same foreground objects, we posit that they provide rich contextual information to supervise the image translation model. To this end, we propose a novel training objective leveraging coarsely-aligned image pairs. We show that our coarsely-aligned training scheme leads to a better image translation quality and improved downstream tasks, such as semantic segmentation, monocular depth estimation, and visual localization. | ['Mark Campbell', 'Kilian Q Weinberger', 'Bharath Hariharan', 'Wei-Lun Chao', 'Josephine Monica', 'Youya Xia'] | 2022-09-23 | null | null | null | null | ['visual-localization'] | ['computer-vision'] | [ 5.91059566e-01 -4.09465730e-02 -1.30357891e-01 -6.23569906e-01
-6.77222788e-01 -7.89489388e-01 5.22156179e-01 -3.66951317e-01
-1.74560770e-01 5.96149385e-01 -2.77233154e-01 -5.05318940e-01
2.92769194e-01 -8.57134044e-01 -1.29542482e+00 -5.87364733e-01
4.87373620e-01 3.98985177e-01 1.40444905e-01 -1.47649139e-01
3.94107141e-02 3.78202289e-01 -1.67128778e+00 -1.38236642e-01
1.05915344e+00 8.61811876e-01 5.92136562e-01 7.83863783e-01
2.73533821e-01 4.71661538e-01 -4.55822617e-01 -3.60119402e-01
5.36894202e-01 -2.82469690e-01 -3.98596406e-01 6.27987504e-01
7.11818397e-01 -6.29065216e-01 -4.39375997e-01 1.06874895e+00
-6.66474625e-02 2.08263189e-01 3.43014866e-01 -1.60396671e+00
-3.86925280e-01 -1.46244556e-01 -6.09606683e-01 4.18552384e-02
3.64165038e-01 4.60061789e-01 5.06342351e-01 -4.16320920e-01
4.71804023e-01 9.76153970e-01 3.28722090e-01 2.91376412e-01
-9.20599699e-01 -5.98960161e-01 3.91573817e-01 2.20750377e-01
-1.24733543e+00 -5.21825969e-01 6.93897724e-01 -1.26572758e-01
4.69460130e-01 3.76972824e-01 5.77129722e-01 1.12764001e+00
8.90289471e-02 6.98458314e-01 1.25688851e+00 -5.96675277e-02
2.68233120e-01 2.28652537e-01 -2.62673289e-01 4.15730417e-01
3.64430398e-01 3.42445701e-01 -6.47179425e-01 4.47201997e-01
7.30904281e-01 3.13659579e-01 -2.93285280e-01 -2.84392983e-01
-1.29108667e+00 4.08299387e-01 5.81840575e-01 -5.83100878e-02
-3.13618690e-01 -9.68876854e-03 -1.83329687e-01 2.30227262e-01
2.86733121e-01 1.63824424e-01 -2.74583101e-01 -1.72304899e-01
-8.72547030e-01 3.27994704e-01 4.53546017e-01 1.37657356e+00
1.05385911e+00 -9.52088758e-02 3.42974573e-01 2.42074788e-01
3.87942940e-02 9.99590874e-01 -5.20095304e-02 -1.12587571e+00
8.94825220e-01 3.78476411e-01 4.91951972e-01 -1.03785264e+00
-1.73867121e-01 -2.70683646e-01 -5.74640632e-01 7.83994347e-02
6.54142916e-01 -1.45894438e-01 -1.26414394e+00 1.68048406e+00
4.69729036e-01 5.14620304e-01 1.79028332e-01 1.34198213e+00
3.09974134e-01 6.93457425e-01 -2.25956887e-01 -5.07590035e-03
1.28095961e+00 -1.06050849e+00 -5.65374970e-01 -9.69705582e-01
3.90206367e-01 -7.02850521e-01 1.10432136e+00 2.06864432e-01
-9.38642263e-01 -5.18457234e-01 -1.14625788e+00 -5.53399213e-02
-2.88789719e-01 -9.91235524e-02 3.81309599e-01 5.01373827e-01
-8.42520833e-01 1.94285497e-01 -8.64804208e-01 -4.36745763e-01
1.42971531e-01 -6.07935898e-02 -4.68888015e-01 -6.75003290e-01
-1.08173239e+00 8.94076407e-01 4.28002812e-02 2.05758169e-01
-9.95707750e-01 -5.39223552e-01 -8.77414882e-01 -3.73883694e-01
6.87098384e-01 -6.90636814e-01 1.30521667e+00 -1.06988668e+00
-1.09603775e+00 9.15979564e-01 -6.07050598e-01 -3.96496892e-01
7.10213006e-01 -2.82202333e-01 -2.64998257e-01 2.20718116e-01
4.97131407e-01 7.49905407e-01 9.19541419e-01 -1.66356790e+00
-9.76914585e-01 -6.13343298e-01 2.81258792e-01 7.19967246e-01
-1.61905047e-02 -3.64196062e-01 -7.90316224e-01 -2.84994811e-01
2.77692884e-01 -1.17694116e+00 -2.05395296e-01 5.23824804e-02
-5.88292360e-01 5.46511352e-01 9.10027087e-01 -5.50774813e-01
5.62863529e-01 -2.11530089e+00 -5.95549271e-02 1.85758591e-01
5.41144945e-02 -4.33301628e-02 6.08224906e-02 1.54340668e-02
2.74058521e-01 -9.62468907e-02 -3.94758165e-01 -6.40331328e-01
-1.99136376e-01 8.03161502e-01 -3.47671092e-01 5.93242049e-01
8.80127549e-02 1.03432035e+00 -1.09106612e+00 -4.02388155e-01
5.14902115e-01 2.88588077e-01 -8.15620422e-02 4.23564494e-01
-1.30794719e-01 9.27531362e-01 -4.39027220e-01 7.37287581e-01
8.82096529e-01 5.11801988e-02 -8.81067067e-02 -9.39566419e-02
-1.24648839e-01 5.69166467e-02 -1.01882124e+00 1.48555243e+00
-7.63293922e-01 8.03133130e-01 2.36899316e-01 -8.44387352e-01
6.64501548e-01 -4.58558314e-02 1.82187706e-01 -1.16835785e+00
7.15107098e-02 1.97257817e-01 -4.41154808e-01 -5.99586368e-01
6.28547251e-01 -1.17669679e-01 -2.47916639e-01 2.83186615e-01
-6.53961778e-01 -4.67714101e-01 -5.28571233e-02 4.63739857e-02
8.31770241e-01 7.15543106e-02 -1.40874937e-01 1.85952753e-01
-2.90619247e-02 2.68129140e-01 7.28454113e-01 7.40356922e-01
-3.58385265e-01 1.00623739e+00 1.53561503e-01 -1.84116378e-01
-1.26880574e+00 -1.32101643e+00 2.71122921e-02 7.37417758e-01
1.02155244e+00 3.65537368e-02 -8.92889917e-01 -4.92155820e-01
-2.04914913e-01 7.45658040e-01 -4.48491216e-01 -1.92310005e-01
-5.27592182e-01 -4.45728958e-01 3.71195853e-01 6.53097749e-01
8.85024130e-01 -4.69056726e-01 -9.73593056e-01 -3.82574648e-02
-6.50420070e-01 -1.62520361e+00 -6.11277640e-01 4.02945019e-02
-5.73504627e-01 -9.80649054e-01 -6.36323988e-01 -5.57049870e-01
8.36370647e-01 1.01580954e+00 1.12045956e+00 2.42146011e-03
1.81078929e-02 2.31618062e-01 -3.89085203e-01 -2.92422980e-01
-1.46434233e-01 -1.81224734e-01 -3.36292386e-02 2.61052877e-01
2.66529799e-01 -4.53215241e-01 -8.63459826e-01 6.83505356e-01
-9.83888566e-01 4.91569340e-01 5.91161788e-01 5.25652647e-01
7.26218462e-01 3.18520695e-01 -2.52225380e-02 -5.31021893e-01
-1.99174862e-02 -5.18346310e-01 -4.99129355e-01 1.05108902e-01
-3.21300715e-01 -3.41864318e-01 6.70760512e-01 -1.30243093e-01
-1.14879215e+00 2.48406664e-01 2.22665057e-01 -5.72237074e-01
-6.07695222e-01 4.69648130e-02 -4.22677428e-01 -3.84728611e-02
4.90376472e-01 3.77749354e-01 -1.45742267e-01 9.66892112e-03
4.35812533e-01 7.30344653e-01 9.41580176e-01 -4.38292623e-01
1.12073636e+00 9.94450748e-01 -5.03369346e-02 -8.28048825e-01
-8.82596076e-01 -5.80046952e-01 -5.41367769e-01 -4.37992066e-01
1.07589412e+00 -1.10417354e+00 -1.81352034e-01 6.64848089e-01
-8.85667264e-01 -6.17082715e-01 1.50063679e-01 3.66440475e-01
-5.55317998e-01 2.30119348e-01 -1.19934075e-01 -8.37242126e-01
3.41910809e-01 -1.26261353e+00 1.54896581e+00 5.02907872e-01
6.46572700e-03 -8.03107202e-01 -4.40075487e-01 8.43621671e-01
1.53856456e-01 2.94500500e-01 3.27172369e-01 5.95976487e-02
-1.01162589e+00 -9.01172459e-02 -3.28919023e-01 1.80817470e-01
2.16986001e-01 -1.79278016e-01 -1.03686905e+00 -1.90233648e-01
1.12120710e-01 -1.16181567e-01 6.58657730e-01 2.90511221e-01
8.16192627e-01 -1.19115919e-01 -5.08590519e-01 7.82512486e-01
1.24502504e+00 3.22839439e-01 7.06453085e-01 4.03833508e-01
1.03952491e+00 7.77332842e-01 1.17549777e+00 9.61912721e-02
7.82940030e-01 9.50767398e-01 7.01436996e-01 -3.93983275e-01
3.20958346e-02 -4.70404208e-01 2.75407016e-01 1.51021481e-01
3.11845988e-01 -5.86962044e-01 -8.54900122e-01 7.19293535e-01
-1.89813840e+00 -6.21730387e-01 -3.09543073e-01 2.28067875e+00
3.52765024e-01 2.98427284e-01 -1.61780715e-01 -9.11890045e-02
7.13083684e-01 1.56078652e-01 -8.78502190e-01 1.47323972e-02
-3.00511688e-01 -3.67877543e-01 1.02509713e+00 6.29380584e-01
-1.04661524e+00 7.96320200e-01 5.13186312e+00 4.94525701e-01
-1.04069829e+00 1.52312785e-01 9.26342070e-01 -1.89759925e-01
-5.34343719e-01 1.32711232e-01 -6.93782568e-01 8.04107666e-01
7.96278059e-01 2.22978011e-01 4.55066741e-01 6.56914771e-01
5.14331400e-01 -5.68295062e-01 -9.52567577e-01 1.13423848e+00
2.49645069e-01 -1.00292218e+00 -2.26072848e-01 2.15638988e-02
1.03910089e+00 1.96961798e-02 3.08735907e-01 -1.26439586e-01
1.85786828e-01 -1.02236974e+00 1.13372803e+00 3.75838339e-01
8.02627563e-01 -6.44547582e-01 6.09576941e-01 5.14302194e-01
-1.20838094e+00 2.20504496e-02 -9.55052599e-02 -2.50801831e-01
5.85167348e-01 6.99125648e-01 -5.62235832e-01 7.38247097e-01
7.20402777e-01 6.94455802e-01 -4.26313013e-01 7.82445610e-01
-4.80619580e-01 2.71751732e-01 -5.32492459e-01 3.31030369e-01
3.49700570e-01 -4.57982600e-01 5.33325434e-01 6.03607774e-01
3.74254167e-01 1.76872090e-01 2.73394674e-01 8.26613963e-01
2.21790254e-01 -6.01757407e-01 -9.03731525e-01 3.59945416e-01
5.82359493e-01 1.02111840e+00 -7.75075912e-01 -2.86644787e-01
-5.18044710e-01 1.35522103e+00 -1.07847460e-01 6.49241209e-01
-1.15755296e+00 4.65733893e-02 9.65192199e-01 3.76098484e-01
1.35326564e-01 -3.59174401e-01 -7.73155212e-01 -1.21533060e+00
5.23906350e-01 -6.12466812e-01 -1.60883516e-01 -1.13602126e+00
-9.74015653e-01 5.55725694e-01 5.48972897e-02 -1.28765190e+00
-1.97199881e-01 -3.94629061e-01 -5.98551929e-01 7.95958579e-01
-1.79676008e+00 -1.26817465e+00 -8.33693326e-01 6.16568446e-01
5.70022523e-01 4.38153297e-01 1.20715022e-01 4.94402677e-01
-6.50026917e-01 5.04036486e-01 1.06342666e-01 -1.02313221e-01
6.33995473e-01 -1.16009200e+00 5.87788284e-01 1.44572985e+00
4.95467223e-02 2.08470911e-01 8.39498878e-01 -6.69847190e-01
-1.38872683e+00 -1.55802810e+00 6.63180411e-01 -5.82461059e-01
2.30310500e-01 -3.99116427e-01 -6.75873280e-01 6.83136642e-01
-2.07753301e-01 -7.92924389e-02 -4.01190221e-02 -5.67067623e-01
-2.72353124e-02 -1.94346189e-01 -1.04222393e+00 9.32758927e-01
1.25678575e+00 -6.56193912e-01 -3.10789943e-01 4.68928099e-01
6.66726172e-01 -7.90728569e-01 -2.91601241e-01 2.98864543e-01
1.61355570e-01 -1.07622480e+00 1.01271367e+00 -7.80029446e-02
3.87094855e-01 -5.84580004e-01 -2.83884108e-01 -1.19797778e+00
4.05099452e-01 -5.55400610e-01 2.38395825e-01 9.91671622e-01
3.50878030e-01 -8.04012299e-01 9.42645907e-01 8.61790717e-01
-3.77368301e-01 -6.08650148e-01 -1.00389731e+00 -9.09743607e-01
-2.09022075e-01 -6.75160587e-01 6.96467578e-01 6.69378400e-01
-4.23710883e-01 8.33308846e-02 -4.89429444e-01 6.28859460e-01
7.15865493e-01 4.22745913e-01 1.01993215e+00 -7.19724536e-01
-1.06128819e-01 -2.99834292e-02 -4.23448592e-01 -1.56238520e+00
1.11163266e-01 -3.61704111e-01 4.90894109e-01 -1.46668184e+00
3.40798050e-02 -7.30352879e-01 3.82197618e-01 2.33053908e-01
-3.99458230e-01 5.77883422e-01 3.90358754e-02 2.15538755e-01
-4.84745383e-01 3.31044793e-01 1.28672731e+00 1.03936368e-03
-8.13117549e-02 1.47175670e-01 -6.65382326e-01 5.64678788e-01
8.95338595e-01 -1.65857300e-01 -6.74229681e-01 -7.32225358e-01
2.90682167e-01 1.15679458e-01 6.83459878e-01 -1.05075932e+00
1.65520802e-01 -4.10168082e-01 2.47331128e-01 -4.83755201e-01
5.61044455e-01 -8.75375211e-01 2.82671750e-01 1.77097604e-01
9.29909572e-02 8.67704898e-02 -1.79041978e-02 8.10709119e-01
-2.54085302e-01 1.01890303e-02 6.26168370e-01 -2.04035819e-01
-1.01795745e+00 3.99762750e-01 -1.85242817e-01 6.48219213e-02
1.37780261e+00 -8.57199430e-01 -4.60015357e-01 -6.37569964e-01
-3.93299937e-01 5.42464495e-01 1.05591166e+00 5.61119854e-01
6.59924448e-01 -1.15988898e+00 -4.92130220e-01 5.06529331e-01
2.62386143e-01 5.29760540e-01 3.96642923e-01 1.01908779e+00
-4.83987838e-01 3.48286510e-01 -2.96410732e-02 -8.93393934e-01
-1.20319235e+00 3.15114141e-01 3.51349205e-01 1.43875107e-01
-6.04207873e-01 7.22662866e-01 6.55184686e-01 -3.50094438e-01
9.10016075e-02 -5.15238523e-01 3.55913311e-01 -2.67561704e-01
4.92411256e-01 1.78468704e-01 2.53569752e-01 -8.67688298e-01
-2.80948758e-01 7.35706687e-01 1.18476465e-01 -2.08724111e-01
8.17653477e-01 -8.32766712e-01 3.61294925e-01 3.24088186e-01
1.17395997e+00 -3.29554856e-01 -1.91928208e+00 -1.37409836e-01
-7.22523212e-01 -1.02685630e+00 6.10447042e-02 -5.84600389e-01
-1.19447100e+00 8.45707059e-01 5.05709469e-01 -7.17289746e-02
1.16206622e+00 1.08109318e-01 1.17253327e+00 1.31348908e-01
6.81510806e-01 -1.10236394e+00 -1.45028725e-01 2.97250658e-01
3.06088150e-01 -1.47162068e+00 -4.23166633e-01 -5.31874478e-01
-8.79643679e-01 6.83991671e-01 7.31842935e-01 2.58738577e-01
1.43182248e-01 2.20738217e-01 2.59595841e-01 -8.28076228e-02
-3.69403273e-01 -4.52790082e-01 -3.50383483e-02 9.12126243e-01
-3.80702972e-01 3.55896503e-01 5.33959091e-01 1.22250393e-01
-3.95049334e-01 -2.90774405e-01 6.43764138e-01 9.94484663e-01
-3.05838287e-01 -5.65406144e-01 -7.77566433e-01 3.85563731e-01
1.08013205e-01 1.68782771e-01 -3.66871178e-01 5.89734495e-01
3.05708319e-01 1.30274916e+00 3.32126409e-01 -4.74339843e-01
4.86066848e-01 -4.47806001e-01 4.13554192e-01 -4.90787476e-01
1.40249744e-01 -3.83316100e-01 8.36855695e-02 -7.97828674e-01
-3.92334878e-01 -7.24541187e-01 -1.08316660e+00 -5.83398879e-01
-2.89342310e-02 -2.65702128e-01 6.58760130e-01 1.22948694e+00
1.95260435e-01 4.87980783e-01 7.98494220e-01 -1.12615764e+00
-4.41568494e-02 -4.20360327e-01 -5.22355974e-01 5.86445391e-01
6.07521892e-01 -6.62686288e-01 -4.63432074e-01 2.77322114e-01] | [8.468131065368652, -2.138725996017456] |
1b548b8b-dc5a-42b2-ba5e-f7ae619146df | boosting-rgb-d-saliency-detection-by | 2201.001 | null | https://arxiv.org/abs/2201.00100v1 | https://arxiv.org/pdf/2201.00100v1.pdf | Boosting RGB-D Saliency Detection by Leveraging Unlabeled RGB Images | Training deep models for RGB-D salient object detection (SOD) often requires a large number of labeled RGB-D images. However, RGB-D data is not easily acquired, which limits the development of RGB-D SOD techniques. To alleviate this issue, we present a Dual-Semi RGB-D Salient Object Detection Network (DS-Net) to leverage unlabeled RGB images for boosting RGB-D saliency detection. We first devise a depth decoupling convolutional neural network (DDCNN), which contains a depth estimation branch and a saliency detection branch. The depth estimation branch is trained with RGB-D images and then used to estimate the pseudo depth maps for all unlabeled RGB images to form the paired data. The saliency detection branch is used to fuse the RGB feature and depth feature to predict the RGB-D saliency. Then, the whole DDCNN is assigned as the backbone in a teacher-student framework for semi-supervised learning. Moreover, we also introduce a consistency loss on the intermediate attention and saliency maps for the unlabeled data, as well as a supervised depth and saliency loss for labeled data. Experimental results on seven widely-used benchmark datasets demonstrate that our DDCNN outperforms state-of-the-art methods both quantitatively and qualitatively. We also demonstrate that our semi-supervised DS-Net can further improve the performance, even when using an RGB image with the pseudo depth map. | ['Yueting Zhuang', 'Yi Yang', 'Fei Wu', 'Ping Li', 'Huazhu Fu', 'Siliang Tang', 'Lei Zhu', 'Xiaoqiang Wang'] | 2022-01-01 | null | null | null | null | ['rgb-d-salient-object-detection'] | ['computer-vision'] | [ 1.69812620e-01 3.36705863e-01 -3.73539597e-01 -4.86153007e-01
-7.54141510e-01 -1.38018116e-01 1.99548259e-01 -2.44836528e-02
-2.69491673e-01 2.94922739e-01 7.39625171e-02 -1.40871331e-01
4.01395857e-01 -6.45928741e-01 -8.09204400e-01 -8.87029707e-01
6.07878625e-01 6.20639622e-02 9.25225079e-01 -1.11936450e-01
1.58506811e-01 4.42625791e-01 -1.68279910e+00 1.00015718e-02
8.65776062e-01 1.64207757e+00 7.62673497e-01 6.14484996e-02
-2.77925670e-01 9.72437978e-01 -1.59400612e-01 3.46301198e-02
5.02596080e-01 -3.42071772e-01 -5.59924185e-01 3.91306907e-01
2.44178370e-01 -7.18970358e-01 -4.11770642e-01 9.97207820e-01
5.86705148e-01 3.67863216e-02 2.90556908e-01 -1.48102152e+00
-6.61203921e-01 2.34158307e-01 -9.42038774e-01 3.20917636e-01
9.73480269e-02 2.35785142e-01 7.47296691e-01 -1.26808274e+00
3.71224821e-01 1.11740661e+00 2.55911499e-01 6.20925546e-01
-7.77395487e-01 -6.56343281e-01 2.92621702e-01 1.13209963e-01
-1.10640359e+00 5.99989295e-02 1.48145294e+00 1.25727365e-02
5.18107951e-01 -6.11457005e-02 9.41788435e-01 6.20300412e-01
-3.78614038e-01 1.62775636e+00 1.09951031e+00 -3.27886432e-01
1.49557069e-01 1.69550389e-01 -2.81444751e-02 9.06923234e-01
7.77618960e-02 1.59838229e-01 -6.46122098e-01 4.40338671e-01
1.10996985e+00 3.14111471e-01 -1.06917344e-01 -8.77929389e-01
-1.06323588e+00 6.89438879e-01 1.39652467e+00 -9.86709297e-02
-3.49251211e-01 -4.40693721e-02 5.00824116e-02 -5.72299697e-02
5.92818081e-01 3.32677574e-03 -4.76128399e-01 3.93018574e-01
-7.79217422e-01 1.03512693e-05 1.64028972e-01 9.21965361e-01
1.21563768e+00 -6.85335770e-02 2.53956579e-02 5.47052264e-01
6.05460286e-01 7.15211749e-01 4.10467029e-01 -8.22090387e-01
6.09176874e-01 1.22008944e+00 1.05081387e-01 -7.67478228e-01
-5.03258228e-01 -3.47378075e-01 -6.68160200e-01 2.71712542e-01
3.78516316e-01 1.74447328e-01 -1.13278615e+00 1.46185219e+00
7.58542955e-01 5.27851023e-02 -4.29910682e-02 1.60993946e+00
1.15407324e+00 3.33498269e-01 -7.78820962e-02 4.53463309e-02
9.98310149e-01 -1.34558761e+00 -2.93422908e-01 -5.88677347e-01
3.85573834e-01 -4.65170741e-01 1.33479381e+00 -6.93485979e-03
-1.14361763e+00 -5.97654164e-01 -1.04094148e+00 -7.18270421e-01
-2.15980992e-01 3.56905580e-01 7.60360658e-01 3.65880519e-01
-9.62751806e-01 1.30078599e-01 -1.05147219e+00 3.55060287e-02
8.16751361e-01 2.85039365e-01 -3.15974690e-02 -1.15364037e-01
-1.13208282e+00 7.27870762e-01 3.67523283e-01 1.99362695e-01
-1.26253223e+00 -6.85127854e-01 -1.12272191e+00 -1.27441555e-01
3.58652085e-01 -3.60262334e-01 1.22883952e+00 -1.09773135e+00
-1.29032278e+00 1.19415104e+00 -7.36781359e-02 -7.86449835e-02
3.84895623e-01 -5.33906594e-02 3.15693207e-02 4.61825997e-01
3.44378263e-01 1.12216580e+00 8.77190948e-01 -1.53499091e+00
-9.98794496e-01 -5.94629228e-01 9.41182598e-02 4.84334588e-01
-2.62358308e-01 -3.78197044e-01 -6.51181936e-01 -4.63894337e-01
8.02221000e-01 -6.77381992e-01 -2.32924983e-01 5.09784222e-01
-7.37771451e-01 -1.73154339e-01 9.19919252e-01 -4.20818448e-01
6.83240652e-01 -2.13747263e+00 2.60656625e-01 -2.09839158e-02
4.14344281e-01 1.92318648e-01 3.91806103e-02 -3.76011521e-01
-4.50676903e-02 -4.34448004e-01 -3.00064534e-01 -6.86126351e-01
-2.38964096e-01 7.77563453e-02 -5.00435472e-01 4.61989373e-01
5.42865694e-01 1.16088331e+00 -1.26231027e+00 -7.52852559e-01
5.40618479e-01 3.43451381e-01 -2.89617896e-01 3.86634618e-01
-3.37870181e-01 4.01930839e-01 -7.90564716e-01 1.26492953e+00
7.40503371e-01 -5.19256055e-01 -3.91728520e-01 -2.69068986e-01
-2.21870497e-01 4.07546103e-01 -8.75231087e-01 2.03181863e+00
-2.21147045e-01 4.57464695e-01 -1.51981965e-01 -9.02002692e-01
1.07546568e+00 -2.66555309e-01 3.81285250e-01 -9.15610194e-01
2.09017009e-01 4.65524375e-01 -4.77992862e-01 -1.85101554e-01
3.06606472e-01 -2.42302381e-02 -1.61537062e-02 6.39142692e-01
-5.21535287e-03 -5.61613679e-01 -2.91707009e-01 3.23958546e-01
5.54487348e-01 2.45059088e-01 -1.01346396e-01 3.84017229e-02
5.02005875e-01 9.09191817e-02 6.67873919e-01 3.17090929e-01
-5.23556948e-01 8.40571463e-01 4.67573285e-01 -5.80236018e-01
-9.07428026e-01 -1.42627990e+00 4.93850447e-02 9.45679367e-01
1.07872045e+00 2.60978550e-01 -6.13294184e-01 -1.04662061e+00
1.50377348e-01 3.61729532e-01 -7.98532605e-01 -4.21310455e-01
-2.76357561e-01 -4.07299638e-01 -7.70743489e-02 8.02531123e-01
9.45569158e-01 -1.22140312e+00 -8.69397521e-01 -8.40145200e-02
-1.57101259e-01 -9.46309805e-01 -4.56190020e-01 6.61651194e-01
-1.09709823e+00 -9.08133388e-01 -9.59275842e-01 -1.15276575e+00
9.40104544e-01 1.04713333e+00 8.43401730e-01 8.70300531e-02
-1.28239155e-01 1.11958787e-01 -2.49027759e-01 -5.69495618e-01
1.18370555e-01 1.60166591e-01 -9.74433348e-02 -1.35137975e-01
4.90069747e-01 -5.18264592e-01 -1.10253549e+00 4.61523533e-01
-8.87517750e-01 4.83209968e-01 9.52388465e-01 6.10043585e-01
1.01429653e+00 -4.08798456e-01 5.35684526e-01 -2.80669391e-01
-2.66680777e-01 -3.04538757e-01 -5.88823974e-01 1.47098362e-01
-4.59373176e-01 -2.43768692e-02 9.90144387e-02 -3.09258342e-01
-8.97754967e-01 5.96875489e-01 1.93855278e-02 -7.89108932e-01
2.20845696e-02 7.03688264e-02 -3.53512287e-01 -2.46929884e-01
2.72778779e-01 5.69001198e-01 1.06038235e-01 -3.83104742e-01
2.92411417e-01 7.30794370e-01 5.05863845e-01 -7.20770657e-02
9.79312122e-01 7.18676269e-01 -1.84541315e-01 -5.17105579e-01
-1.63875401e+00 -5.37687659e-01 -6.57158136e-01 -2.29843199e-01
8.64906311e-01 -1.34188783e+00 -3.42549771e-01 8.23798656e-01
-9.00436461e-01 -6.84265137e-01 -3.57972175e-01 1.83745295e-01
-4.79732096e-01 1.39505282e-01 -4.61428374e-01 -6.44453585e-01
-2.93120682e-01 -1.29695737e+00 1.70858717e+00 7.26049244e-01
5.28513849e-01 -6.85031950e-01 -2.73429573e-01 5.05829632e-01
8.22080970e-02 1.01594292e-01 6.55711234e-01 -1.53699666e-01
-1.01128626e+00 -1.18397148e-02 -8.54393482e-01 4.09946561e-01
3.14632714e-01 -5.65676689e-01 -1.14527118e+00 8.64465255e-03
-7.12996535e-03 -8.18900824e-01 1.04082835e+00 3.87240618e-01
1.39547062e+00 1.84002295e-01 -3.61552358e-01 8.24678183e-01
1.32152402e+00 -3.63853812e-01 3.22490275e-01 4.08746302e-01
1.15744555e+00 4.90707666e-01 9.66710627e-01 3.20041001e-01
8.08255494e-01 4.23394203e-01 9.40172732e-01 -6.41232073e-01
-3.96408468e-01 -5.66450536e-01 1.50984734e-01 5.73156595e-01
2.44622171e-01 2.14831740e-01 -8.57869029e-01 7.11463690e-01
-1.66904938e+00 -3.95395696e-01 1.66133828e-02 1.89935863e+00
1.16661084e+00 3.20766836e-01 3.24741453e-01 2.56998628e-01
5.58998168e-01 8.24351758e-02 -1.15225434e+00 3.84607762e-01
-3.00008804e-01 -8.71343613e-02 4.99554068e-01 1.63370043e-01
-1.15567875e+00 9.61177528e-01 4.73462677e+00 6.18405461e-01
-1.37843728e+00 1.20919608e-01 7.82977462e-01 -2.53638148e-01
-5.19440055e-01 -2.06754982e-01 -8.61252367e-01 4.80767578e-01
3.58007438e-02 1.05632424e-01 1.08892769e-01 1.28018773e+00
8.61000195e-02 -3.83978367e-01 -7.58825898e-01 1.10764158e+00
1.61601424e-01 -1.14601755e+00 1.57196280e-02 -1.19286425e-01
1.06385994e+00 3.30246359e-01 2.11212188e-01 8.26602578e-02
2.51107633e-01 -5.45445561e-01 1.01637805e+00 2.08002135e-01
6.18731201e-01 -8.06436121e-01 6.09890461e-01 3.52602810e-01
-1.20530117e+00 -1.91754103e-01 -5.98956048e-01 9.62561592e-02
2.10256502e-02 7.17112601e-01 -4.64864880e-01 3.14228654e-01
8.81270051e-01 1.14585388e+00 -8.10037196e-01 1.01857364e+00
-7.92443037e-01 1.08933210e-01 -3.29413325e-01 -2.06296027e-01
4.49189633e-01 -2.62535866e-02 2.02816322e-01 5.25275409e-01
1.17981303e-02 8.54341909e-02 2.61672109e-01 9.92708802e-01
-1.14031658e-01 -2.84415632e-01 -1.24130234e-01 3.86774480e-01
6.09753132e-01 1.32904267e+00 -9.76485610e-01 -2.37787709e-01
-4.31408226e-01 1.16156638e+00 5.05054295e-01 2.92670786e-01
-7.53650248e-01 -3.48801106e-01 4.55839872e-01 5.97280301e-02
4.77016836e-01 7.98643604e-02 -5.26410282e-01 -1.10458553e+00
9.94195603e-03 -3.12237769e-01 1.80762678e-01 -1.31343138e+00
-1.08196115e+00 4.16218281e-01 -2.24349216e-01 -1.52565217e+00
1.83466956e-01 -5.32459617e-01 -5.35822213e-01 8.36322486e-01
-2.25109768e+00 -1.40832949e+00 -9.36702430e-01 6.73098087e-01
3.43999237e-01 1.41633332e-01 1.71402648e-01 2.92397533e-02
-6.29266679e-01 3.96110773e-01 -2.52393961e-01 1.29986137e-01
4.56246257e-01 -1.42291844e+00 2.52720535e-01 8.33436787e-01
-1.10999584e-01 2.04871729e-01 2.51425683e-01 -4.85610574e-01
-1.37749970e+00 -1.39942670e+00 4.68013555e-01 -3.91999960e-01
4.03438747e-01 -4.33315605e-01 -7.60961175e-01 4.70496446e-01
-4.42295432e-01 5.83040059e-01 2.43660241e-01 -5.78781486e-01
-2.01790467e-01 -2.81081080e-01 -1.11323071e+00 4.01572883e-01
1.08593667e+00 -7.43658483e-01 -7.14858353e-01 3.99448663e-01
1.31827128e+00 -7.62846589e-01 -4.32829022e-01 2.02182651e-01
3.12655509e-01 -1.12786412e+00 1.23575079e+00 -4.68464792e-02
9.07663405e-01 -5.58050275e-01 -9.28814709e-02 -1.03022754e+00
2.52036750e-01 -1.06621673e-02 -2.13848338e-01 1.03945148e+00
2.33265609e-01 -2.59951949e-01 1.15500915e+00 5.26711226e-01
-4.61276680e-01 -1.06714582e+00 -7.98097551e-01 -4.37928021e-01
-2.52083153e-01 -4.09424335e-01 5.16866446e-01 6.80301845e-01
-3.43584478e-01 2.02399686e-01 2.38589756e-02 2.48892248e-01
9.90661502e-01 7.25460410e-01 6.61363482e-01 -1.04012787e+00
1.12305760e-01 -4.14999098e-01 -4.10290867e-01 -1.55064726e+00
8.29300098e-03 -8.55685413e-01 3.02969962e-01 -1.61261666e+00
2.32435510e-01 -7.32231855e-01 -4.68806684e-01 6.77695751e-01
-5.07820964e-01 5.41620791e-01 -1.23379426e-02 2.66862482e-01
-9.14428711e-01 1.08288968e+00 1.67041588e+00 -1.70285285e-01
-4.59345162e-01 -1.30935144e-02 -7.28763938e-01 7.24506199e-01
6.38746679e-01 -4.36819047e-01 -5.18163264e-01 -3.94756556e-01
-7.26611391e-02 -6.33074790e-02 7.63062239e-01 -8.88898075e-01
2.85082668e-01 -1.23634599e-01 7.53327310e-01 -1.15737367e+00
2.02607095e-01 -6.55004084e-01 -8.95910442e-01 4.34345186e-01
-2.07123488e-01 -5.26758134e-01 -5.93943289e-03 5.33542514e-01
-1.68751746e-01 1.57726198e-01 8.52624893e-01 4.58307331e-03
-1.07067895e+00 6.17652118e-01 4.43464488e-01 9.69795734e-02
1.11264265e+00 -4.04504955e-01 -2.13154584e-01 -3.54181938e-02
-3.03879172e-01 5.70059776e-01 7.46365488e-01 4.51890498e-01
1.02740610e+00 -1.45650649e+00 -1.71295956e-01 4.74576473e-01
3.86087835e-01 8.52469683e-01 1.94155216e-01 9.20711100e-01
-4.20269281e-01 2.51025349e-01 -3.62935275e-01 -9.46352720e-01
-7.10646093e-01 7.31129289e-01 2.34888256e-01 1.61965042e-01
-5.10068536e-01 1.22122550e+00 3.88535917e-01 -4.95259255e-01
5.42880833e-01 -6.60231411e-01 6.85544014e-02 -4.94156703e-02
5.11904478e-01 2.52880324e-02 1.14911934e-02 -4.83555704e-01
-5.73897839e-01 6.12389505e-01 1.00663811e-01 1.31839573e-01
1.41935515e+00 -5.15071273e-01 -6.39897957e-02 5.34811556e-01
1.19797504e+00 -5.21147490e-01 -1.96808040e+00 -5.64014018e-01
-3.24008942e-01 -4.87047404e-01 4.89620715e-01 -5.56459665e-01
-1.50153625e+00 1.07702112e+00 7.38970101e-01 -8.31781924e-02
1.44831896e+00 3.07068944e-01 1.02305484e+00 2.61028886e-01
3.88065636e-01 -8.28452468e-01 6.96767271e-01 1.72303066e-01
6.71439826e-01 -1.79098678e+00 3.22328471e-02 -4.57615107e-01
-8.47696841e-01 7.87466705e-01 1.00501430e+00 -2.42060140e-01
5.94469428e-01 -2.77338326e-02 2.74471581e-01 -1.50586322e-01
-2.56806791e-01 -5.15132964e-01 2.83190846e-01 7.54330277e-01
-1.49686769e-01 -2.73088634e-01 4.50667977e-01 7.22019970e-01
7.33898953e-02 3.94767597e-02 2.69203871e-01 9.49856281e-01
-6.78903818e-01 -6.73092127e-01 -8.77652541e-02 3.08054805e-01
1.19552083e-01 -9.63247195e-02 -4.54725653e-01 7.88943768e-01
1.52784273e-01 5.97877502e-01 1.36542276e-01 -4.70985591e-01
1.94597676e-01 -4.23213005e-01 3.42080086e-01 -6.70873046e-01
-3.06766570e-01 -1.10123128e-01 -6.18496597e-01 -7.09703505e-01
-6.11928701e-01 -4.79940832e-01 -1.65149665e+00 5.71581870e-02
-4.64844197e-01 -1.99916333e-01 6.53640926e-01 9.13462579e-01
1.20408952e-01 4.53765541e-01 1.07448971e+00 -1.26495671e+00
-3.25535029e-01 -6.79517865e-01 -6.88095093e-01 2.80642718e-01
7.87254810e-01 -9.30514216e-01 -5.24880946e-01 -1.53406546e-01] | [9.67101764678955, -0.7761887907981873] |
c46ae9d2-246b-4564-9060-7515b59586d6 | scalable-algorithms-for-string-kernels-with | null | null | http://papers.nips.cc/paper/3441-scalable-algorithms-for-string-kernels-with-inexact-matching | http://papers.nips.cc/paper/3441-scalable-algorithms-for-string-kernels-with-inexact-matching.pdf | Scalable Algorithms for String Kernels with Inexact Matching | We present a new family of linear time algorithms based on sufficient statistics for string comparison with mismatches under the string kernels framework. Our algorithms improve theoretical complexity bounds of existing approaches while scaling well with respect to the sequence alphabet size, the number of allowed mismatches and the size of the dataset. In particular, on large alphabets with loose mismatch constraints our algorithms are several orders of magnitude faster than the existing algorithms for string comparison under the mismatch similarity measure. We evaluate our algorithms on synthetic data and real applications in music genre classification, protein remote homology detection and protein fold prediction. The scalability of the algorithms allows us to consider complex sequence transformations, modeled using longer string features and larger numbers of mismatches, leading to a state-of-the-art performance with significantly reduced running times. | ['Pai-Hsi Huang', 'Vladimir Pavlovic', 'Pavel P. Kuksa'] | 2008-12-01 | null | null | null | neurips-2008-12 | ['genre-classification'] | ['computer-vision'] | [ 6.42693341e-01 -4.97496188e-01 -1.26468703e-01 -1.74740762e-01
-7.92704582e-01 -1.00635529e+00 2.39152580e-01 7.83838391e-01
-5.79944849e-01 6.61605716e-01 -2.59721279e-01 -3.60659838e-01
-1.11085892e-01 -6.19333565e-01 -9.30006444e-01 -6.60552204e-01
-2.67359436e-01 7.21546173e-01 7.53317237e-01 -3.45158279e-01
6.57614946e-01 4.75952148e-01 -1.90685952e+00 2.67703205e-01
7.27607489e-01 8.11745226e-01 1.83496438e-02 1.02241206e+00
-2.23687682e-02 -2.11740304e-02 -4.26332444e-01 -4.54930753e-01
4.19758052e-01 -5.95005810e-01 -8.37885141e-01 -3.49586427e-01
6.35564446e-01 2.27102831e-01 -1.08581983e-01 1.11983371e+00
7.18198895e-01 1.43005416e-01 3.27371627e-01 -9.35784698e-01
-2.12276042e-01 6.33968711e-01 -2.90768296e-01 1.50346100e-01
8.66064668e-01 1.04136720e-01 1.21794999e+00 -7.81368136e-01
7.13138402e-01 1.04358697e+00 9.72293198e-01 5.38028218e-02
-1.71678138e+00 -5.33273160e-01 -2.76800245e-01 6.22643471e-01
-1.43311667e+00 -6.49891123e-02 6.26808628e-02 -3.86344016e-01
1.35886550e+00 7.40275621e-01 4.40732747e-01 7.69599319e-01
-1.49119735e-01 -5.36791161e-02 7.89444864e-01 -7.63111413e-01
3.15806776e-01 -2.79558331e-01 4.58243340e-01 5.14985502e-01
1.50600150e-01 -1.04936017e-02 -4.82312649e-01 -9.76918697e-01
4.68528509e-01 -1.64676964e-01 -1.23872191e-01 -3.45674068e-01
-1.56872022e+00 7.27936566e-01 -3.09546262e-01 1.86146453e-01
4.52434607e-02 -1.52517691e-01 6.36425197e-01 6.42130196e-01
1.47080004e-01 4.30726469e-01 -4.69347596e-01 -4.65121955e-01
-7.81986177e-01 5.39816201e-01 1.25970602e+00 1.11740541e+00
5.10513723e-01 -5.48641741e-01 1.13695957e-01 9.05171633e-01
-5.29451191e-01 4.25857097e-01 4.42706794e-01 -8.26279104e-01
3.11311483e-01 4.10079032e-01 2.58723557e-01 -8.67106974e-01
-4.72558975e-01 -2.44804874e-01 -6.94320023e-01 4.56436612e-02
8.07033956e-01 6.12508774e-01 -4.15953666e-01 1.81313431e+00
3.85030478e-01 2.66819805e-01 -7.96525367e-03 5.92025340e-01
2.37479821e-01 5.54336667e-01 -3.63677233e-01 -6.03168190e-01
1.46139622e+00 -6.92746580e-01 -2.84983665e-01 2.63258696e-01
8.60539019e-01 -1.09966719e+00 1.30911374e+00 6.52751207e-01
-1.04266846e+00 -4.76452947e-01 -1.05008888e+00 8.31903964e-02
-2.26165384e-01 -1.54751658e-01 4.38585401e-01 7.49850273e-01
-5.86129189e-01 1.20900750e+00 -6.89012468e-01 -6.22077882e-01
-1.16479270e-01 5.28612971e-01 -4.74458843e-01 1.30880266e-01
-9.05511677e-01 5.41825831e-01 6.92110837e-01 -4.94131655e-01
1.60122603e-01 -7.82523692e-01 -4.36994702e-01 3.82497124e-02
2.27656960e-02 -3.79402161e-01 1.14137220e+00 -8.02081823e-01
-1.34073925e+00 9.93122816e-01 -1.22054525e-01 -5.72748780e-01
6.12519085e-01 -2.70477384e-02 -2.16536537e-01 8.34050477e-02
-3.88469577e-01 9.93382558e-02 1.30050480e-01 -2.14201003e-01
-3.56883645e-01 -2.21529886e-01 -1.20839544e-01 -1.90486349e-02
-1.82891503e-01 3.36244285e-01 -3.23060155e-01 -9.28505659e-01
1.01496384e-01 -1.20678425e+00 -3.80788952e-01 1.54481735e-02
-1.90080553e-01 -1.37374833e-01 3.17371607e-01 -4.52656120e-01
1.22785580e+00 -2.15814042e+00 5.01302719e-01 3.69024158e-01
-2.24676639e-01 2.95275837e-01 -3.50673407e-01 1.00445640e+00
-3.74333620e-01 -2.59493321e-01 -3.47755581e-01 3.42705309e-01
1.80868402e-01 1.49048716e-01 -3.67550552e-01 6.01555407e-01
-1.78536698e-01 3.57685864e-01 -6.72330201e-01 -1.77529901e-01
-1.43707842e-01 -5.53332232e-02 -7.60809660e-01 2.81763077e-01
-3.80673736e-01 1.98488578e-01 -1.43609896e-01 2.97814906e-01
8.54256809e-01 -4.58318353e-01 5.51451027e-01 -1.52038187e-01
2.60107219e-03 3.94055605e-01 -1.54299784e+00 1.69094634e+00
-1.49242664e-02 2.11838111e-01 -4.40652847e-01 -1.15546525e+00
9.21468914e-01 -1.16202440e-02 4.43053603e-01 -3.39811802e-01
-1.67216837e-01 5.90903521e-01 1.76296234e-01 -2.54747659e-01
2.02324495e-01 -3.60372150e-03 4.64797951e-02 4.48960245e-01
-1.22777358e-01 2.21390754e-01 5.50020695e-01 3.73077877e-02
1.26010799e+00 1.16572101e-02 6.91778064e-01 -4.47556049e-01
5.51918149e-01 -6.79841340e-02 5.37966132e-01 8.44009280e-01
3.86891723e-01 4.21643078e-01 7.24552333e-01 -5.31308472e-01
-1.58793497e+00 -8.27337861e-01 -3.37849736e-01 1.46507788e+00
1.90896437e-01 -6.22389734e-01 -7.88155735e-01 -1.47888452e-01
1.73722029e-01 2.54080087e-01 -2.93777525e-01 -6.49368018e-02
-8.09569836e-01 -9.98434544e-01 7.97531843e-01 5.21398604e-01
-5.52739911e-02 -8.03120852e-01 -5.77248871e-01 3.67560655e-01
-5.23195527e-02 -1.22281218e+00 -6.60441458e-01 1.66718140e-01
-8.49083483e-01 -1.34315288e+00 -5.65898597e-01 -8.60341549e-01
2.37163395e-01 -9.19694975e-02 1.03581142e+00 -5.89509234e-02
-7.66129315e-01 -1.58043906e-01 -3.57931137e-01 -5.59415147e-02
-6.88187063e-01 7.56430998e-02 3.06798220e-01 -2.83211440e-01
3.04118484e-01 -1.09211349e+00 -4.55752403e-01 6.04730546e-01
-1.14624715e+00 1.85835897e-03 2.95013934e-01 1.16047013e+00
7.84843206e-01 -3.84598553e-01 3.19778055e-01 -8.67579877e-01
2.15006039e-01 -1.94241315e-01 -1.01635361e+00 3.85606080e-01
-4.97009367e-01 2.53455609e-01 9.04633164e-01 -9.96134818e-01
-3.48330796e-01 3.55979323e-01 -3.47063571e-01 5.15565425e-02
3.74167114e-02 1.54646471e-01 -1.26671493e-01 -4.25644994e-01
8.58186603e-01 5.66507161e-01 -3.21769714e-02 -7.46110678e-01
2.47509122e-01 5.47423780e-01 8.05852473e-01 -7.53887057e-01
4.19540197e-01 1.80614322e-01 3.78241211e-01 -6.53726220e-01
-2.57542521e-01 -6.45813763e-01 -8.04746747e-01 3.75796467e-01
3.12851191e-01 -4.90350306e-01 -1.23557401e+00 5.38621545e-01
-9.13646877e-01 -2.42403187e-02 4.32343632e-02 5.29575050e-01
-1.11121845e+00 1.12575102e+00 -6.69088900e-01 -4.58272219e-01
-2.94000834e-01 -1.07133806e+00 8.27558935e-01 -2.44523570e-01
-2.78294533e-01 -5.34195125e-01 4.93228048e-01 1.70213953e-02
5.18357009e-02 4.38146681e-01 1.50573123e+00 -1.10360253e+00
-3.97974849e-01 -2.25754425e-01 -1.35157295e-02 5.92306070e-02
-1.92418262e-01 -1.65608764e-01 -4.33531880e-01 -4.23179537e-01
-3.96098077e-01 -2.55641609e-01 7.60171413e-01 -1.83139026e-01
1.22711635e+00 -3.93449873e-01 -2.68692225e-01 8.38519216e-01
1.44481671e+00 -3.73747759e-02 4.79337692e-01 3.92115802e-01
2.55607814e-01 3.55626643e-01 7.62629271e-01 8.66551638e-01
-4.39362556e-01 1.20717049e+00 1.05516627e-01 3.11489165e-01
3.43973666e-01 2.38406174e-02 1.61797091e-01 8.76742840e-01
-1.85552351e-02 -2.84102082e-01 -7.23258674e-01 2.30172575e-01
-2.14851642e+00 -9.82238650e-01 -3.31223279e-01 2.81480098e+00
1.02510321e+00 3.61265987e-02 6.08172178e-01 4.43065226e-01
7.84413815e-01 -3.18638027e-01 -7.25130260e-01 -6.42123282e-01
-4.40662026e-01 5.84750056e-01 7.76363850e-01 2.60854363e-01
-9.36810672e-01 6.30552828e-01 7.28299427e+00 1.07672179e+00
-6.45667195e-01 -2.83056259e-01 1.14199162e-01 -1.76924005e-01
2.89937705e-02 1.42779443e-02 -6.54954076e-01 4.92704928e-01
1.01671267e+00 -3.48185450e-01 5.44844329e-01 6.98269427e-01
-2.80980468e-01 1.02285430e-01 -1.30331397e+00 1.17098510e+00
-2.65180990e-02 -1.47315478e+00 8.19467846e-03 1.80470794e-02
4.25344169e-01 5.01793288e-02 -1.73346803e-01 -3.64256710e-01
1.56525865e-01 -8.14054132e-01 5.32656372e-01 2.81381011e-01
8.70760739e-01 -9.18239474e-01 4.58019227e-01 3.38712126e-01
-1.00230849e+00 -5.16818278e-02 -6.16479814e-01 -1.67663381e-01
-1.06957614e-01 5.91797054e-01 -4.22172248e-01 5.60367167e-01
5.22512555e-01 3.22912425e-01 -3.72313976e-01 1.27861702e+00
4.60980654e-01 5.48007429e-01 -8.21725667e-01 -3.04770499e-01
-8.72551203e-02 -2.90360510e-01 6.18066728e-01 1.46577442e+00
4.47504640e-01 3.59313577e-01 3.48611295e-01 2.10540131e-01
8.69363323e-02 7.48770297e-01 -1.76750213e-01 -7.93009698e-02
5.36083281e-01 7.02513039e-01 -6.65993989e-01 -4.00701553e-01
-3.03431869e-01 1.09153163e+00 3.74862254e-01 4.77534160e-02
-7.44754136e-01 -9.36717749e-01 8.08137476e-01 -4.78603244e-02
6.25748575e-01 -2.52109528e-01 7.01569468e-02 -1.07913148e+00
3.91856879e-01 -1.37417972e+00 6.45769060e-01 -1.42358243e-01
-1.31562269e+00 4.36678767e-01 -1.18974634e-01 -1.32472456e+00
-2.31360018e-01 -8.81965101e-01 -2.16671363e-01 6.82705283e-01
-8.55133355e-01 -5.87117374e-01 1.26702398e-01 1.43314287e-01
1.81624502e-01 -1.29566938e-01 1.34784269e+00 2.97916681e-01
-2.07809001e-01 1.09462178e+00 9.44802165e-01 -3.26190025e-01
8.23452532e-01 -1.01992261e+00 8.68703127e-01 4.49351072e-01
2.60315597e-01 5.50817251e-01 1.16943657e+00 -3.64662677e-01
-1.52576709e+00 -6.52050614e-01 8.84089291e-01 -1.75739542e-01
7.37933218e-01 -6.97156370e-01 -1.23011577e+00 3.51348758e-01
-3.73063803e-01 -1.87661424e-01 1.15720475e+00 1.04833513e-01
-1.07641017e+00 1.09897271e-01 -1.06559861e+00 3.65597248e-01
1.52745831e+00 -4.17203993e-01 -3.98177177e-01 8.31375003e-01
6.39821529e-01 -4.61262107e-01 -1.18204606e+00 5.36202133e-01
1.07362378e+00 -1.03465831e+00 1.00232494e+00 -1.04329813e+00
-8.60036016e-02 -2.50569314e-01 -3.78032744e-01 -7.02808678e-01
-4.33800280e-01 -9.51449871e-01 2.70299405e-01 7.36898720e-01
3.29368234e-01 -6.66840851e-01 5.29893577e-01 -1.68709084e-02
2.35795766e-01 -6.52935982e-01 -1.07034731e+00 -1.29269075e+00
-5.45437820e-03 -3.15073431e-01 7.52707005e-01 8.80994856e-01
6.45307958e-01 -4.20035869e-02 -4.19296861e-01 1.11090608e-01
4.89766926e-01 6.98149145e-01 8.17101419e-01 -1.27924573e+00
-1.08819938e+00 -5.85787058e-01 -9.97731149e-01 -1.01765490e+00
5.39072081e-02 -8.77547741e-01 -1.76121101e-01 -5.89392483e-01
6.51434600e-01 -2.25278258e-01 -2.51455933e-01 3.55220288e-01
-2.46989682e-01 5.01606405e-01 1.69787034e-02 1.49551317e-01
-4.77948099e-01 2.62776278e-02 4.17403340e-01 1.11268558e-01
-1.00604407e-02 1.34523377e-01 -1.48035437e-01 5.52948415e-01
6.19562566e-01 -4.89865899e-01 6.26195669e-02 5.99178448e-02
4.73240405e-01 1.12269614e-02 1.28310338e-01 -9.15409207e-01
6.58003166e-02 -2.89594352e-01 -2.03892644e-02 -5.75154543e-01
2.96882540e-01 -3.40687513e-01 4.97617662e-01 7.53931344e-01
-6.53895497e-01 4.52577353e-01 3.33958209e-01 6.82549536e-01
9.55048501e-02 -2.20512196e-01 1.00011265e+00 2.61482149e-01
-1.52412742e-01 7.63676316e-02 -2.73382723e-01 -1.47796227e-02
9.29594576e-01 -2.44289130e-01 -1.19965978e-01 -1.46568939e-01
-7.80735910e-01 -2.44497135e-01 8.68797243e-01 3.25172633e-01
2.11616635e-01 -1.27546000e+00 -7.75560498e-01 3.09295893e-01
5.54980874e-01 -8.08753908e-01 1.13892639e-02 6.82344377e-01
-9.07983780e-01 4.60794896e-01 -3.59673202e-01 -6.60329401e-01
-2.09682775e+00 8.17697167e-01 3.34540755e-02 -2.27211311e-01
-6.57212853e-01 6.89448833e-01 1.90373454e-02 -3.78277898e-01
4.59733933e-01 -2.74565995e-01 3.61921489e-01 -3.41677278e-01
7.26276696e-01 5.27790308e-01 3.43924761e-01 -4.59526211e-01
-3.68287474e-01 8.91153634e-01 2.18421929e-02 1.33138508e-01
1.13212097e+00 4.06475872e-01 -2.58583844e-01 4.14037943e-01
1.28738654e+00 3.05003976e-03 -6.03561103e-01 -4.09159511e-01
3.15128982e-01 -5.69340169e-01 -9.25269961e-01 -5.06846845e-01
-1.80764198e-01 5.90344727e-01 7.57171452e-01 -2.66058855e-02
1.12698209e+00 -1.58321057e-02 8.61501276e-01 9.83449042e-01
4.41559374e-01 -8.46808434e-01 -2.88601547e-01 5.25470614e-01
4.88514900e-01 -7.62116194e-01 1.48126844e-03 -6.67159379e-01
-1.25400349e-01 1.30654848e+00 1.96226668e-02 -1.03964850e-01
2.67686903e-01 5.03924370e-01 -2.68222868e-01 4.48734403e-01
-9.36363161e-01 -9.03543532e-02 2.24672630e-01 3.52832347e-01
4.46877807e-01 1.91672891e-02 -9.46375012e-01 3.54786694e-01
-3.70584935e-01 -2.72473305e-01 2.49700382e-01 7.52691567e-01
-7.52720714e-01 -1.78962910e+00 -2.99972981e-01 3.07675868e-01
-6.62625968e-01 -3.06110322e-01 -5.38579464e-01 3.36211145e-01
-2.20456108e-01 5.81440985e-01 -6.32974431e-02 -2.38049999e-01
4.65109438e-01 3.57139260e-01 9.17396963e-01 -3.43410283e-01
-6.55881524e-01 4.18518297e-02 1.48662195e-01 -6.80922031e-01
-2.14106798e-01 -7.47543156e-01 -1.21282446e+00 -4.78855759e-01
-4.64560956e-01 3.45434576e-01 4.56591487e-01 6.93877280e-01
6.14218175e-01 -2.22887844e-01 7.00857937e-01 -3.35052729e-01
-1.16070604e+00 -8.63986433e-01 -5.63069582e-01 7.10601687e-01
-9.49325934e-02 -2.64117956e-01 -1.39296576e-01 1.66956767e-01] | [4.860790729522705, 5.202752590179443] |
588caeb7-e1d5-4b8a-afb6-339b7959de3d | sparse-gaussian-process-temporal-difference | 1810.01217 | null | http://arxiv.org/abs/1810.01217v1 | http://arxiv.org/pdf/1810.01217v1.pdf | Sparse Gaussian Process Temporal Difference Learning for Marine Robot Navigation | We present a method for Temporal Difference (TD) learning that addresses
several challenges faced by robots learning to navigate in a marine
environment. For improved data efficiency, our method reduces TD updates to
Gaussian Process regression. To make predictions amenable to online settings,
we introduce a sparse approximation with improved quality over current
rejection-based sparse methods. We derive the predictive value function
posterior and use the moments to obtain a new algorithm for model-free policy
evaluation, SPGP-SARSA. With simple changes, we show SPGP-SARSA can be reduced
to a model-based equivalent, SPGP-TD. We perform comprehensive simulation
studies and also conduct physical learning trials with an underwater robot. Our
results show SPGP-SARSA can outperform the state-of-the-art sparse method,
replicate the prediction quality of its exact counterpart, and be applied to
solve underwater navigation tasks. | ['John Martin', 'Jinkun Wang', 'Brendan Englot'] | 2018-10-02 | null | null | null | null | ['marine-robot-navigation'] | ['robots'] | [ 1.04133487e-01 2.20677346e-01 7.52495751e-02 -8.56427327e-02
-1.18342721e+00 -3.08899432e-01 2.99909920e-01 -5.65068088e-02
-6.32521272e-01 1.07822633e+00 5.67869507e-02 -3.75427663e-01
-3.65404814e-01 -4.94271070e-01 -1.12994981e+00 -1.01957214e+00
-7.66222179e-01 6.41938567e-01 1.51471317e-01 -1.59037858e-01
3.14022273e-01 1.39284357e-01 -1.23331761e+00 -3.82653266e-01
9.76238072e-01 1.02293229e+00 5.93214750e-01 5.48816860e-01
2.77480334e-01 7.19979525e-01 -9.94614735e-02 2.62423363e-02
6.20151579e-01 -8.64722282e-02 -3.39227140e-01 -3.37706566e-01
2.22582132e-01 -4.13604379e-01 -3.40202928e-01 9.78345037e-01
8.96801531e-01 7.16573656e-01 7.09842861e-01 -8.56402397e-01
-4.86557670e-02 4.12271887e-01 -6.12174034e-01 8.43334384e-03
9.25955698e-02 3.98296677e-02 7.20217109e-01 -8.82720530e-01
2.51657724e-01 1.67757201e+00 1.13614190e+00 6.95077837e-01
-1.29164791e+00 -7.29084253e-01 4.41197276e-01 -1.67429492e-01
-8.95212531e-01 -4.95355219e-01 3.03521665e-04 -2.97713369e-01
8.07952881e-01 -4.82929558e-01 8.32839549e-01 1.00251174e+00
6.10743284e-01 9.56536174e-01 1.19598627e+00 1.06029503e-01
9.61600065e-01 -5.36318660e-01 -3.55529428e-01 9.29942310e-01
5.63800395e-01 4.42705274e-01 -7.89579034e-01 -5.88381290e-01
7.27291882e-01 -2.78953761e-01 -5.24569035e-01 -7.59519577e-01
-7.43598521e-01 1.05085850e+00 1.56404637e-02 -7.38253117e-01
-5.83754241e-01 5.81987560e-01 1.72014460e-01 3.77513230e-01
3.88797045e-01 4.21962410e-01 -8.06599081e-01 -5.92291355e-01
-6.44400835e-01 6.84227467e-01 1.63124883e+00 9.62985814e-01
6.69332922e-01 5.59272230e-01 1.20743379e-01 6.97858155e-01
7.50725448e-01 1.09992409e+00 2.97916204e-01 -1.75125730e+00
3.07986408e-01 -6.22724056e-01 6.67758703e-01 -5.16309261e-01
-5.15194476e-01 -5.34899890e-01 -5.83737493e-01 4.52652007e-01
1.64998755e-01 -9.20866907e-01 -5.61392784e-01 1.68115270e+00
2.56057173e-01 6.92440271e-01 7.03603625e-01 7.61786520e-01
1.04218759e-01 1.00286555e+00 -7.11961985e-02 -4.11320060e-01
5.81192672e-01 -8.47721517e-01 -3.85379255e-01 -6.02162957e-01
5.77861369e-01 -2.56017387e-01 5.08881450e-01 7.60998249e-01
-8.79867196e-01 5.35515174e-02 -8.65233004e-01 5.98973215e-01
5.12517869e-01 -3.39016199e-01 6.93845928e-01 3.14459711e-01
-1.23158312e+00 9.12329614e-01 -1.31674433e+00 -2.63031512e-01
3.15416634e-01 4.22707677e-01 -1.19874813e-01 -9.81763303e-02
-8.48764122e-01 1.14117420e+00 4.36545163e-02 5.24441525e-03
-1.55910957e+00 -8.62243831e-01 -1.13586879e+00 -1.72882274e-01
3.53999645e-01 -7.62145281e-01 2.03068614e+00 -3.48324269e-01
-2.23107195e+00 -6.41336292e-02 -3.77271742e-01 -1.11583972e+00
3.06631953e-01 -7.46316671e-01 4.14832264e-01 1.73403069e-01
2.78715879e-01 4.77924585e-01 9.38542604e-01 -1.51975608e+00
-9.61419404e-01 1.25542164e-01 -2.01449275e-01 4.80696172e-01
2.30482128e-02 -6.64594054e-01 -1.55464346e-02 -3.28865111e-01
2.16816440e-01 -1.23175251e+00 -1.02799571e+00 2.37724483e-01
2.92577505e-01 1.56381622e-01 6.00196123e-01 -3.99423212e-01
3.56341213e-01 -1.76554155e+00 3.13951641e-01 2.02807710e-01
-2.01378345e-01 -1.38428090e-02 -2.08106220e-01 5.62968135e-01
7.40199029e-01 -2.54989088e-01 -4.64623302e-01 -8.73215139e-01
3.74790691e-02 9.62027311e-01 -6.93588853e-01 8.27138782e-01
3.66800651e-02 2.07995415e-01 -1.44490242e+00 -2.63709068e-01
-8.18508193e-02 3.10906738e-01 -9.82481360e-01 1.15517914e-01
-2.17836782e-01 6.02537215e-01 -8.64794195e-01 5.32077849e-01
4.32730854e-01 1.64394919e-02 1.46858767e-01 1.80752382e-01
-3.22117120e-01 8.59950036e-02 -1.10625529e+00 1.69608212e+00
-7.82668114e-01 3.79310101e-01 8.08078229e-01 -1.08840883e+00
9.41705048e-01 -7.84760248e-03 6.58319116e-01 -2.79812068e-01
-1.15157567e-01 4.03686017e-01 -2.39253670e-01 -3.48465025e-01
6.58544600e-01 -3.46373469e-01 7.71100074e-02 1.37421846e-01
1.64420068e-01 -5.55347443e-01 -3.80525179e-02 1.15478560e-01
1.04110599e+00 5.33624709e-01 1.62009358e-01 -6.79741442e-01
-3.10982391e-02 -3.02982070e-02 1.01316833e+00 1.37478340e+00
-1.46951467e-01 2.10519299e-01 3.12910855e-01 -1.29309520e-01
-7.73863554e-01 -1.14249206e+00 -3.13037857e-02 1.02834582e+00
4.21965271e-01 -1.42163038e-01 -9.26252902e-02 -1.63450837e-01
5.56880116e-01 5.37126005e-01 -4.91003275e-01 -2.50892108e-03
-4.95443583e-01 -7.06646025e-01 2.85989910e-01 4.41608965e-01
2.10553616e-01 -5.23581564e-01 -7.33046710e-01 6.30456507e-01
1.99832901e-01 -7.47837901e-01 -1.93955064e-01 7.20206320e-01
-1.08232629e+00 -7.51992941e-01 -6.38199747e-01 -7.25796521e-01
2.46890098e-01 1.85021698e-01 7.15955496e-01 -6.44491315e-01
4.53518629e-01 8.44933748e-01 -3.75438064e-01 -7.22891092e-01
-1.61927283e-01 -2.91375309e-01 7.51782894e-01 -3.08763653e-01
-3.98648292e-01 -7.83933938e-01 -5.60278893e-01 1.56999096e-01
-3.67235094e-01 -3.27332914e-01 7.20258117e-01 1.10101688e+00
8.31877530e-01 -2.55462468e-01 5.55090547e-01 -3.94324511e-01
6.71740234e-01 -8.16576004e-01 -1.05714476e+00 -3.15616488e-01
-5.74380159e-01 5.53607225e-01 5.25750220e-01 -6.06603444e-01
-1.12991655e+00 2.93495148e-01 -1.31321386e-01 -4.96337593e-01
4.75244731e-01 8.44276965e-01 4.86250192e-01 -6.12392783e-01
3.96054983e-01 4.31014270e-01 5.25270462e-01 -4.84310567e-01
1.32729426e-01 2.91274607e-01 5.79416454e-01 -1.23944318e+00
7.22737312e-01 6.29111648e-01 3.04116040e-01 -1.05714202e+00
-1.05802190e+00 -4.64587182e-01 1.26043499e-01 -3.93565595e-02
3.27140063e-01 -1.46102643e+00 -8.96734178e-01 2.32639566e-01
-8.47413599e-01 -1.06066096e+00 -4.64948922e-01 7.23287284e-01
-1.17753899e+00 4.66998994e-01 -6.91756964e-01 -1.51002872e+00
-3.76455665e-01 -8.58667731e-01 1.09879982e+00 3.95975798e-01
3.39885384e-01 -1.08357739e+00 6.45251215e-01 -4.06608015e-01
5.63062012e-01 -1.54413218e-02 1.41219035e-01 -5.04978538e-01
-3.02102268e-01 3.22268814e-01 2.90598243e-01 1.54862955e-01
-2.89856464e-01 -4.31776404e-01 -4.88247901e-01 -8.04095924e-01
1.41605586e-01 -5.66399693e-01 9.67767477e-01 7.31513560e-01
5.51614106e-01 -6.28518641e-01 -4.52840298e-01 9.00216639e-01
1.31284654e+00 1.67138472e-01 8.32465664e-02 5.68914175e-01
2.13443696e-01 3.42492580e-01 9.78647351e-01 1.03948057e+00
4.92168218e-01 1.92958072e-01 7.18044043e-01 4.78356361e-01
7.51324296e-01 -4.31392789e-01 8.71337056e-01 7.67477632e-01
1.64225362e-02 -1.11067154e-01 -6.94007516e-01 6.18034244e-01
-2.24006033e+00 -8.49938154e-01 2.82713801e-01 2.10838675e+00
7.99128890e-01 -3.93157639e-02 -2.02405244e-01 -6.46964133e-01
2.20336825e-01 4.20910418e-02 -8.32234859e-01 -2.86132932e-01
8.83589871e-03 2.44852632e-01 1.12831771e+00 6.54068112e-01
-1.03958440e+00 8.03591788e-01 7.25811815e+00 6.00604355e-01
-7.98019648e-01 9.74828936e-03 1.11597441e-01 3.25789559e-03
-1.07290581e-01 5.32669239e-02 -9.97522533e-01 2.19252288e-01
1.12599325e+00 -3.73139888e-01 4.00461316e-01 1.31082261e+00
6.52927816e-01 -2.20730394e-01 -9.26392734e-01 9.18735325e-01
-2.02557191e-01 -1.09745634e+00 -1.98341131e-01 1.16573744e-01
1.07939065e+00 7.39345074e-01 4.42499444e-02 6.04922652e-01
1.21946168e+00 -7.98571408e-01 7.60230303e-01 7.50855505e-01
3.15986186e-01 -6.44742310e-01 9.02756453e-01 4.72410768e-01
-1.25938153e+00 -5.71956336e-01 -8.42223883e-01 -3.60059142e-01
6.12870038e-01 2.53928721e-01 -6.31739736e-01 4.30262864e-01
9.31762516e-01 1.01801097e+00 4.54284877e-01 1.66196942e+00
-3.22686471e-02 8.78776073e-01 -8.78580749e-01 -3.06241959e-01
6.62401974e-01 -3.05925965e-01 9.51492131e-01 9.95751321e-01
8.74854982e-01 1.32084981e-01 6.21664643e-01 2.83233553e-01
3.06813717e-01 -1.63358375e-01 -5.92977226e-01 3.27909201e-01
7.30558932e-01 7.53300667e-01 7.02125654e-02 -1.84910461e-01
-3.43533486e-01 4.26456809e-01 2.67564029e-01 5.30864716e-01
-3.93696934e-01 -7.10938424e-02 1.11567438e+00 -4.66714472e-01
7.86590397e-01 -6.11815393e-01 2.04999238e-01 -1.01604748e+00
-4.37541366e-01 -8.34476173e-01 3.17024649e-03 -5.65855324e-01
-1.61052155e+00 2.67866790e-01 8.65459740e-02 -1.55476737e+00
-4.38444108e-01 -6.89915299e-01 -4.46716338e-01 6.31107926e-01
-1.74642861e+00 -6.02687061e-01 1.61291301e-01 2.23082349e-01
7.22621620e-01 -2.05008298e-01 8.68881643e-01 -2.53391713e-01
-1.27600908e-01 1.13865405e-01 8.59371185e-01 -4.02554750e-01
5.63426316e-01 -1.35308528e+00 1.61282927e-01 6.96821988e-01
-3.96099448e-01 4.62301314e-01 1.33037376e+00 -8.39613914e-01
-2.00370908e+00 -1.17701650e+00 -2.95322210e-01 -1.95322707e-02
1.22152054e+00 2.03918412e-01 -6.34470880e-01 6.48693919e-01
-1.64680809e-01 -5.18360361e-02 4.98326451e-01 -3.97620201e-02
1.17703132e-01 -3.09608597e-02 -9.57803428e-01 5.33567011e-01
8.65479708e-01 3.72525528e-02 -6.91631138e-01 9.09294859e-02
7.06930757e-01 -7.84356952e-01 -9.86685514e-01 5.19035101e-01
8.76638114e-01 -2.99849212e-01 7.69377112e-01 -4.68402922e-01
-4.31112275e-02 -2.46730000e-01 -4.19583857e-01 -1.81629455e+00
-7.23740757e-02 -1.26695168e+00 -5.57465911e-01 4.04211015e-01
3.58604938e-01 -1.06824338e+00 8.75964761e-01 3.14825982e-01
-6.35215044e-01 -8.20680618e-01 -1.28893065e+00 -1.10633790e+00
5.12937546e-01 -3.26878369e-01 -1.44947723e-01 3.20255220e-01
2.71025971e-02 -1.21396221e-02 -8.90867770e-01 8.84638309e-01
1.24709094e+00 1.70272496e-02 6.99161053e-01 -1.25474179e+00
-7.58917689e-01 -1.22855529e-02 -3.38861085e-02 -1.83924425e+00
3.29423189e-01 -2.08295688e-01 9.11962450e-01 -1.72942924e+00
-2.29348734e-01 -7.10755885e-01 4.92974557e-02 2.62077123e-01
8.62956345e-02 -1.80136874e-01 3.16659398e-02 1.67281002e-01
-7.28427887e-01 1.23106956e+00 1.07319498e+00 -7.25597590e-02
-3.20357352e-01 2.16041744e-01 -4.62823570e-01 9.10992444e-01
7.79247403e-01 -6.52057946e-01 -3.91361743e-01 -5.81398785e-01
1.62008837e-01 4.92573380e-01 -6.75988197e-02 -1.37608969e+00
6.03099883e-01 -3.84523213e-01 3.82651854e-03 -3.84260356e-01
5.74650109e-01 -4.90851462e-01 -1.98597714e-01 9.52189982e-01
-3.00066590e-01 -1.76433221e-01 3.57295305e-01 1.38525903e+00
-3.64642218e-03 -3.89106840e-01 7.04830229e-01 -1.08497880e-01
-8.29157591e-01 5.68560839e-01 -1.10116971e+00 2.12423772e-01
5.35348177e-01 2.05248132e-01 -2.75965959e-01 -8.42289627e-01
-6.80395007e-01 9.32892382e-01 3.05801123e-01 -2.12278053e-01
8.07057679e-01 -7.36988127e-01 -6.76908970e-01 -2.49337494e-01
-3.90848875e-01 2.72858202e-01 9.05517638e-02 8.07367682e-01
-4.92696464e-01 1.47337139e-01 1.00546025e-01 -5.18375993e-01
-5.53575456e-01 2.04925425e-02 5.93336403e-01 -5.12183867e-02
-7.36009657e-01 9.51201558e-01 1.29622445e-01 -5.87840736e-01
3.62642616e-01 -3.88875961e-01 -1.81014799e-02 -4.92137402e-01
4.28353518e-01 2.59026080e-01 -5.21254301e-01 -1.25494912e-01
-3.88080478e-02 5.90597093e-01 3.54090959e-01 -4.61002588e-01
1.74783051e+00 -3.81271958e-01 2.53388733e-01 5.30977249e-01
6.30694091e-01 1.96924154e-02 -2.43139005e+00 -2.87050992e-01
-9.63220075e-02 -2.00176343e-01 -1.30799338e-02 -2.58735448e-01
-6.98301852e-01 4.91418272e-01 5.44732153e-01 -2.20268041e-01
5.45560598e-01 -2.46745393e-01 6.92977846e-01 1.41741264e+00
1.14785004e+00 -9.05012667e-01 1.87508073e-02 1.25675917e+00
6.67161286e-01 -1.24734175e+00 7.21162856e-02 8.56132731e-02
-6.56331539e-01 1.01340961e+00 5.03922164e-01 -6.06467545e-01
8.59091699e-01 6.53697491e-01 -1.48583606e-01 2.38866359e-01
-1.17656457e+00 -1.61714241e-01 -2.86404133e-01 8.78200889e-01
-2.53564924e-01 -2.43234038e-01 -1.36774937e-02 5.59211254e-01
-2.10628718e-01 -1.90302104e-01 6.87998116e-01 1.39393556e+00
-1.08909714e+00 -7.50255883e-01 -2.00409502e-01 6.71484232e-01
-3.94921809e-01 -1.62004799e-01 5.72685778e-01 5.16810000e-01
-6.02474988e-01 9.00267184e-01 -1.04544368e-02 -1.05113789e-01
9.83606353e-02 -4.02390659e-01 3.36120039e-01 -5.30560136e-01
1.34674400e-01 3.01728118e-02 3.71823490e-01 -8.99715781e-01
-5.69069743e-01 -9.87296641e-01 -1.32837081e+00 -5.30689955e-02
-1.35876298e-01 7.77478158e-01 6.18749619e-01 1.01354408e+00
3.58294904e-01 1.52332306e-01 4.97002065e-01 -1.60364771e+00
-1.28495502e+00 -9.47560012e-01 -5.94998956e-01 -5.53697705e-01
6.31748617e-01 -1.07129109e+00 -6.80800796e-01 -2.29028568e-01] | [4.163841247558594, 2.332655191421509] |
e11bce48-5f55-4b32-a3c1-6f9dcabdf881 | continuous-mdp-homomorphisms-and-homomorphic | 2209.07364 | null | https://arxiv.org/abs/2209.07364v1 | https://arxiv.org/pdf/2209.07364v1.pdf | Continuous MDP Homomorphisms and Homomorphic Policy Gradient | Abstraction has been widely studied as a way to improve the efficiency and generalization of reinforcement learning algorithms. In this paper, we study abstraction in the continuous-control setting. We extend the definition of MDP homomorphisms to encompass continuous actions in continuous state spaces. We derive a policy gradient theorem on the abstract MDP, which allows us to leverage approximate symmetries of the environment for policy optimization. Based on this theorem, we propose an actor-critic algorithm that is able to learn the policy and the MDP homomorphism map simultaneously, using the lax bisimulation metric. We demonstrate the effectiveness of our method on benchmark tasks in the DeepMind Control Suite. Our method's ability to utilize MDP homomorphisms for representation learning leads to improved performance when learning from pixel observations. | ['Doina Precup', 'David Meger', 'Prakash Panangaden', 'Rosie Zhao', 'Sahand Rezaei-Shoshtari'] | 2022-09-15 | null | null | null | null | ['policy-gradient-methods'] | ['methodology'] | [-1.88507468e-01 1.57964766e-01 -7.23378897e-01 1.02657929e-01
-5.55015564e-01 -6.94994450e-01 9.49579477e-01 4.43653800e-02
-4.69218194e-01 9.06403959e-01 3.95194024e-01 -5.08581698e-01
-1.08053647e-01 -7.72835255e-01 -9.63823199e-01 -7.69086480e-01
-3.48164767e-01 2.14484587e-01 3.93666625e-02 -3.95694971e-01
2.39855394e-01 6.44237518e-01 -1.22869134e+00 2.83935070e-02
6.58880293e-01 7.46959209e-01 -3.66845839e-02 5.99245906e-01
1.79308385e-01 8.28876615e-01 -3.32200795e-01 3.36444497e-01
7.79393196e-01 -3.72144371e-01 -8.34778786e-01 -1.28865056e-02
2.91420192e-01 -5.54759443e-01 -5.52539825e-01 9.98349905e-01
3.08602273e-01 4.83318418e-01 4.46925133e-01 -1.40808070e+00
-2.73128301e-01 7.05798924e-01 -4.40183878e-01 -4.99807186e-02
-8.96159001e-03 4.82654661e-01 1.23224294e+00 -3.40981156e-01
6.86797678e-01 1.53828776e+00 2.25807682e-01 7.02694654e-01
-1.69216335e+00 -5.06984949e-01 5.48136294e-01 -4.98121716e-02
-1.04711032e+00 -2.46197730e-01 3.80874336e-01 -4.04363334e-01
1.23791838e+00 -1.58836350e-01 7.74401963e-01 1.15629113e+00
4.02980387e-01 8.30564618e-01 1.19716144e+00 -3.84293377e-01
7.46667445e-01 -5.83399273e-02 -1.70075163e-01 9.36867177e-01
1.98677212e-01 8.16493452e-01 -3.16595823e-01 -3.23548257e-01
9.71306145e-01 -1.82866603e-01 -8.93707126e-02 -9.98577774e-01
-1.01057112e+00 1.02146077e+00 4.25789893e-01 -2.16224134e-01
-3.91219884e-01 8.10190976e-01 6.10620022e-01 7.05208719e-01
1.04372784e-01 8.56928229e-01 -2.64860034e-01 -3.63933861e-01
-3.06894839e-01 7.13939786e-01 1.03108549e+00 7.71047533e-01
6.22015417e-01 4.18661207e-01 -4.40343618e-01 3.20255131e-01
4.24034223e-02 5.25182843e-01 2.19073787e-01 -1.59287071e+00
4.06545013e-01 2.31518403e-01 4.79835749e-01 -5.90664566e-01
-2.78216928e-01 -4.11974937e-01 -3.86795610e-01 7.94423103e-01
2.52072662e-01 -2.11814091e-01 -7.17487216e-01 2.18591285e+00
3.49948794e-01 1.99636608e-01 2.72592396e-01 5.90285301e-01
-6.23779118e-01 7.09824681e-01 -5.04715741e-02 -2.52692074e-01
7.88416445e-01 -8.54247630e-01 -5.12645721e-01 -9.87700000e-02
7.57736981e-01 1.11137144e-01 1.30558133e+00 3.21792811e-01
-1.11685681e+00 -2.67686546e-01 -1.27017117e+00 1.88582495e-01
-1.26979634e-01 -4.21552509e-01 5.53189695e-01 3.15236479e-01
-1.00871074e+00 9.85496223e-01 -1.33215058e+00 -1.62495255e-01
5.05511284e-01 4.89373267e-01 -1.45737097e-01 3.46051127e-01
-1.02606332e+00 1.18866003e+00 6.93634033e-01 -5.26469588e-01
-1.74062169e+00 -8.16090822e-01 -7.23814249e-01 2.83362955e-01
7.65364230e-01 -7.81310737e-01 1.73865509e+00 -6.12016499e-01
-2.03381968e+00 2.77016789e-01 2.35117003e-01 -1.06600523e+00
4.73177642e-01 -6.55443817e-02 1.15058385e-01 1.80825070e-01
-9.87176746e-02 4.05714214e-01 9.96195078e-01 -9.91681039e-01
-8.75248849e-01 -2.91295528e-01 5.76301992e-01 3.19659323e-01
-2.89577842e-01 -4.69236612e-01 2.36532241e-01 -2.55948633e-01
-3.47315282e-01 -1.15052152e+00 -5.06822705e-01 4.63439114e-02
-1.81329712e-01 -8.41434970e-02 7.65096545e-01 -2.19581753e-01
9.18867648e-01 -2.13889909e+00 6.20074034e-01 3.24680001e-01
1.22732393e-01 1.70309335e-01 -1.21406019e-01 6.49482012e-01
7.94148445e-02 1.03694357e-01 -2.66702443e-01 -3.03672612e-01
5.44578731e-01 6.08649790e-01 -9.41147268e-01 7.01737702e-01
1.44569546e-01 7.71229565e-01 -1.07961226e+00 -2.41951898e-01
1.67506859e-01 -3.03453542e-02 -9.03766453e-01 2.31844455e-01
-8.22025180e-01 4.78341132e-01 -6.44470274e-01 3.26463580e-02
3.00261021e-01 1.71736881e-01 4.10259634e-01 4.91247416e-01
-3.09744954e-01 3.67471367e-01 -1.03034616e+00 1.84180629e+00
-7.81690896e-01 1.76101074e-01 1.88301593e-01 -1.18733394e+00
6.27710223e-01 2.40967527e-01 5.33082783e-01 -6.75526321e-01
-1.31030202e-01 2.63622440e-02 -9.04880166e-02 -1.16412349e-01
3.12419683e-01 -1.98301896e-01 -2.39294693e-01 6.18836880e-01
3.77316736e-02 -2.94938654e-01 1.92077637e-01 1.78954765e-01
1.14189994e+00 5.40360868e-01 4.45242852e-01 -6.26140416e-01
3.87694001e-01 -1.97010767e-02 5.05407870e-01 1.10539281e+00
-1.04178697e-01 -3.20967227e-01 1.27975273e+00 -2.52315998e-01
-1.25954831e+00 -1.39246964e+00 7.79678077e-02 1.15946138e+00
-2.46179193e-01 -5.71859419e-01 -5.74778140e-01 -7.87356079e-01
4.22888786e-01 8.35682511e-01 -7.89735258e-01 -5.62607586e-01
-7.51626849e-01 -1.35638282e-01 5.10986865e-01 6.36307716e-01
5.67491531e-01 -7.13549018e-01 -7.95475185e-01 2.77018219e-01
4.66083050e-01 -8.72505963e-01 -4.27912772e-01 1.99560270e-01
-9.19401705e-01 -8.31397891e-01 -3.16943347e-01 -4.40872103e-01
2.43883237e-01 -1.04407176e-01 6.56673312e-01 -2.97286987e-01
3.93813923e-02 7.92072177e-01 1.07657775e-01 -4.04162914e-01
-7.48999476e-01 2.63463825e-01 3.15377206e-01 -2.64110774e-01
-1.72365293e-01 -7.43697703e-01 -2.88509697e-01 6.12720214e-02
-7.92635322e-01 5.51654138e-02 1.95624799e-01 1.06928194e+00
6.23260081e-01 -2.58598089e-01 4.59586352e-01 -6.44867659e-01
7.70578623e-01 -4.42866504e-01 -1.40541506e+00 3.97373028e-02
-7.62974858e-01 1.00438511e+00 9.58409011e-01 -3.03384095e-01
-1.00761449e+00 9.35957506e-02 2.32892841e-01 -6.19392753e-01
2.54601419e-01 3.11784297e-01 2.21210415e-03 -5.28863035e-02
6.58137500e-01 2.71890700e-01 5.00684202e-01 -3.62691015e-01
5.65981865e-01 1.42303348e-01 3.65051657e-01 -1.58192956e+00
7.08687007e-01 5.92284441e-01 5.74932992e-01 -6.40133202e-01
-7.16266811e-01 2.06625909e-02 -1.58102244e-01 2.32303917e-01
5.98914146e-01 -7.67738581e-01 -1.07128882e+00 -2.33834032e-02
-7.35337794e-01 -8.94072711e-01 -7.11894512e-01 4.54979151e-01
-1.29929078e+00 1.54663518e-01 -5.18026173e-01 -8.28063011e-01
1.99031364e-02 -1.32147241e+00 7.29090035e-01 -1.54688805e-01
2.02307731e-01 -9.39554572e-01 5.71489573e-01 -6.01810336e-01
4.00949389e-01 3.65437865e-01 1.06048536e+00 -4.73123401e-01
-6.98803365e-01 3.22670609e-01 1.86916366e-01 5.66162825e-01
-6.68190345e-02 -2.32026324e-01 -7.08956182e-01 -6.71164572e-01
2.48829797e-02 -3.51611614e-01 8.40524137e-01 1.76286593e-01
1.42195427e+00 -6.32686496e-01 -1.51824847e-01 7.99018621e-01
1.52012181e+00 1.08144417e-01 4.53958511e-01 5.58123946e-01
4.53730315e-01 2.59934306e-01 6.13642573e-01 7.50923336e-01
-8.41465220e-02 7.51404107e-01 6.64012849e-01 3.96217287e-01
2.59293973e-01 -6.52666271e-01 8.30650032e-01 1.92588568e-01
-5.75641617e-02 3.78518313e-01 -7.35011637e-01 4.47997451e-01
-2.16550827e+00 -9.59937871e-01 7.38007843e-01 2.30694675e+00
9.56742287e-01 1.29021332e-01 4.22901362e-01 -3.92175525e-01
4.27977622e-01 2.04955637e-01 -8.10886085e-01 -8.08612466e-01
3.67861271e-01 5.95142782e-01 7.75961578e-01 7.72456765e-01
-7.97503829e-01 9.19773757e-01 6.62612534e+00 5.23705006e-01
-1.02628422e+00 7.42790550e-02 1.26556093e-02 -2.27651447e-01
-7.45849684e-02 1.90466046e-01 -8.52240324e-01 3.62010479e-01
1.05707419e+00 -5.66453516e-01 9.79080796e-01 1.09246480e+00
3.51388544e-01 2.03122526e-01 -1.51398492e+00 4.79720294e-01
-6.29061997e-01 -1.45051694e+00 -8.63084644e-02 4.81277078e-01
9.12413776e-01 -5.82593046e-02 4.35313046e-01 6.86605990e-01
9.22526836e-01 -9.08576369e-01 5.61506450e-01 5.06244116e-02
5.39477825e-01 -1.02283978e+00 3.80609967e-02 2.87610114e-01
-9.58699465e-01 -5.65147877e-01 -3.48561347e-01 -3.33498448e-01
-1.27509996e-01 -2.23297805e-01 -1.05449891e+00 3.60486567e-01
4.73420620e-02 7.37146020e-01 6.00510929e-03 6.34321451e-01
-5.08923531e-01 4.67859030e-01 -3.92653197e-01 6.84896410e-02
7.48305500e-01 -4.05467302e-01 6.37310266e-01 9.43046570e-01
2.36049034e-02 -2.90262222e-01 5.56467175e-01 1.05282414e+00
-7.22525716e-02 -2.56601870e-01 -9.42091525e-01 -1.23770639e-01
3.60637039e-01 7.31776714e-01 -1.27192870e-01 -2.72230387e-01
-2.02686995e-01 5.81696630e-01 6.00474834e-01 5.23575604e-01
-9.71744835e-01 -1.36010885e-01 1.39892733e+00 -1.00770853e-01
5.38684487e-01 -6.05602860e-01 8.86203423e-02 -1.15958321e+00
-1.58030391e-01 -1.16468513e+00 4.01407242e-01 -8.84571671e-02
-8.77259851e-01 -2.28023037e-01 5.13215065e-01 -9.89455223e-01
-6.90056205e-01 -6.61088526e-01 -4.82958376e-01 7.42766857e-01
-1.34545875e+00 -6.11740291e-01 3.77331376e-01 6.93051040e-01
2.05269784e-01 -2.24503353e-01 8.78149390e-01 -2.76529849e-01
-5.75919151e-01 4.47475046e-01 5.02420783e-01 -8.97593647e-02
3.42389435e-01 -1.65167367e+00 3.45796108e-01 8.14652741e-01
5.87021466e-03 7.25054681e-01 8.03600430e-01 -3.11010063e-01
-1.83412480e+00 -1.01247227e+00 -2.07136601e-01 -2.00340420e-01
1.08005142e+00 -3.07449162e-01 -6.70411825e-01 1.16640508e+00
2.89691925e-01 3.90337668e-02 -2.51484960e-02 1.89810857e-01
-5.49395442e-01 -3.89900118e-01 -1.03663480e+00 8.77756178e-01
1.04360831e+00 -6.68963730e-01 -6.33374512e-01 1.77650183e-01
1.02141416e+00 -6.11856699e-01 -1.02632999e+00 -1.01500884e-01
3.47524315e-01 -4.71449822e-01 1.01308692e+00 -1.34426451e+00
3.56029928e-01 -1.75212055e-01 -2.70917833e-01 -1.83753717e+00
-2.31794015e-01 -9.88327444e-01 -7.25392342e-01 5.59461534e-01
1.79076806e-01 -1.10777950e+00 4.43365157e-01 2.66730934e-01
-2.48944059e-01 -8.88831139e-01 -1.07091951e+00 -1.14083219e+00
5.95867872e-01 -4.66928929e-02 6.84477091e-01 5.47964215e-01
3.59857172e-01 1.18547000e-01 -3.58754963e-01 2.62125075e-01
7.20705390e-01 1.83961481e-01 7.40767419e-01 -6.23976171e-01
-9.52923477e-01 -5.43521285e-01 -1.93403691e-01 -1.06208897e+00
6.80430412e-01 -1.04058969e+00 -1.46889389e-01 -1.08109748e+00
-1.00173593e-01 -3.24190080e-01 -4.96087253e-01 5.58521450e-01
4.24609154e-01 -6.41878188e-01 4.73514885e-01 -7.16200694e-02
-4.58178103e-01 8.62801790e-01 1.35587668e+00 -9.20866728e-02
-3.05042863e-01 -3.94488648e-02 -4.93641913e-01 4.14203048e-01
1.21813154e+00 -3.94123137e-01 -7.52953351e-01 -3.27601433e-01
1.62204325e-01 4.38491963e-02 4.35372710e-01 -8.96118343e-01
-4.95368093e-02 -6.83107853e-01 -1.11546054e-01 1.12549856e-01
3.42670143e-01 -6.48312569e-01 -2.76075333e-01 9.50050473e-01
-8.29182148e-01 1.34160012e-01 2.60846019e-01 6.77734077e-01
1.69251785e-01 -6.60496652e-02 9.42358434e-01 -2.26396561e-01
-6.54797852e-01 4.62467790e-01 -3.37112278e-01 4.52453762e-01
1.05675805e+00 4.65606391e-01 -3.88297170e-01 -2.33285218e-01
-6.10537291e-01 3.62231582e-01 6.80791557e-01 1.46923229e-01
2.98774660e-01 -1.28323150e+00 -4.82343733e-01 1.13827057e-01
-7.09890723e-02 -2.08039448e-01 -3.64011973e-01 7.69077778e-01
-3.66565615e-01 4.02050167e-01 -5.36769450e-01 -3.26882839e-01
-7.49027133e-01 8.25899363e-01 7.36167490e-01 -4.64343697e-01
-8.89203668e-01 6.63563460e-02 1.67302817e-01 -1.14658356e-01
3.56482983e-01 -7.29632556e-01 4.35309917e-01 -3.83010149e-01
5.92993200e-01 3.36286157e-01 -2.44587094e-01 2.76162058e-01
-1.22953986e-03 5.61922565e-02 -2.30810165e-01 -6.89481080e-01
1.23515499e+00 2.05120504e-01 7.61144012e-02 4.06844467e-01
1.24829614e+00 -3.45049113e-01 -1.81258607e+00 -2.28811845e-01
2.58255191e-02 -5.97894490e-01 8.21276233e-02 -3.88163358e-01
-7.99662173e-01 7.72153199e-01 4.89702612e-01 3.18738103e-01
7.31711388e-01 -2.20822558e-01 3.36650997e-01 9.37544584e-01
5.99522591e-01 -1.35937893e+00 1.00736506e-01 7.08608866e-01
9.06613886e-01 -7.50782132e-01 -4.09134384e-03 3.09694588e-01
-6.30708575e-01 9.59833860e-01 5.64911306e-01 -5.43547630e-01
3.78293544e-01 3.86620343e-01 -6.20213926e-01 1.86090857e-01
-1.10317099e+00 -2.00744495e-01 -3.99660438e-01 7.05774486e-01
-2.06785545e-01 1.47000909e-01 -1.25659794e-01 -9.71930921e-02
7.37346560e-02 -5.22454344e-02 7.24708796e-01 1.27144873e+00
-6.62204266e-01 -1.42367685e+00 -2.19273325e-02 1.52644202e-01
-2.22245410e-01 2.82366961e-01 4.21998799e-02 1.01955771e+00
-4.58903670e-01 2.80037493e-01 2.30799057e-02 -2.64545046e-02
3.48252654e-01 1.93197712e-01 8.85746121e-01 -6.67989492e-01
-3.48324299e-01 -3.87024671e-01 5.32862470e-02 -1.11969459e+00
-3.93256247e-02 -5.84799826e-01 -1.41557920e+00 -4.77770835e-01
3.31557840e-01 2.01152295e-01 7.52273738e-01 7.72158623e-01
4.75446224e-01 4.86765742e-01 8.95246446e-01 -4.83647525e-01
-1.85913479e+00 -5.10252416e-01 -7.11772025e-01 1.83102012e-01
7.43024409e-01 -8.50216806e-01 -2.58345813e-01 -3.21055353e-01] | [4.173600673675537, 2.044827938079834] |
d074922d-177e-42be-b3ff-5703ac9126f8 | iterative-patch-selection-for-high-resolution | 2210.13007 | null | https://arxiv.org/abs/2210.13007v2 | https://arxiv.org/pdf/2210.13007v2.pdf | Iterative Patch Selection for High-Resolution Image Recognition | High-resolution images are prevalent in various applications, such as autonomous driving and computer-aided diagnosis. However, training neural networks on such images is computationally challenging and easily leads to out-of-memory errors even on modern GPUs. We propose a simple method, Iterative Patch Selection (IPS), which decouples the memory usage from the input size and thus enables the processing of arbitrarily large images under tight hardware constraints. IPS achieves this by selecting only the most salient patches, which are then aggregated into a global representation for image recognition. For both patch selection and aggregation, a cross-attention based transformer is introduced, which exhibits a close connection to Multiple Instance Learning. Our method demonstrates strong performance and has wide applicability across different domains, training regimes and image sizes while using minimal accelerator memory. For example, we are able to finetune our model on whole-slide images consisting of up to 250k patches (>16 gigapixels) with only 5 GB of GPU VRAM at a batch size of 16. | ['Aravindh Mahendran', 'Christoph Lippert', 'Benjamin Bergner'] | 2022-10-24 | null | null | null | null | ['multiple-instance-learning'] | ['methodology'] | [ 4.58998203e-01 5.74493501e-03 -9.21363160e-02 -1.68245718e-01
-8.59071910e-01 -2.97140509e-01 3.36182684e-01 5.10954022e-01
-7.17901945e-01 4.10921723e-01 -4.14611101e-01 -3.99477094e-01
-4.72798124e-02 -9.97455537e-01 -8.70021880e-01 -7.60786235e-01
1.85045600e-01 3.70814830e-01 4.84544337e-01 -4.73993504e-03
3.76660258e-01 5.94476581e-01 -2.00161123e+00 3.96497011e-01
7.05496013e-01 1.16419494e+00 4.69992012e-01 6.28652036e-01
1.21370897e-01 4.68781620e-01 -5.22507548e-01 -2.27808077e-02
2.09957048e-01 1.04402930e-01 -6.58720970e-01 1.11721084e-01
6.50237739e-01 -1.82985619e-01 -4.92847450e-02 8.38040113e-01
6.04960740e-01 1.00859255e-01 2.46861726e-01 -5.99242508e-01
-2.02583313e-01 4.81055640e-02 -8.21631789e-01 4.08964336e-01
-1.56412557e-01 1.48723632e-01 7.44890392e-01 -7.81217873e-01
4.72212940e-01 8.49285245e-01 5.62009215e-01 3.46623182e-01
-1.27756000e+00 -5.91708124e-01 6.21371418e-02 1.81570992e-01
-1.27004564e+00 -3.72023672e-01 4.43792969e-01 -2.51879781e-01
1.40114915e+00 3.49946320e-01 6.87614799e-01 8.32992613e-01
3.69129956e-01 4.93942767e-01 1.03502274e+00 -3.65865827e-01
3.68946373e-01 -5.04388176e-02 2.44975120e-01 6.33901656e-01
2.72537440e-01 -1.72208562e-01 -5.84666193e-01 -1.80533513e-01
1.04510415e+00 8.48793164e-02 -9.25270021e-02 5.93800936e-03
-1.28062844e+00 8.75896811e-01 4.48797852e-01 1.02536805e-01
-3.19258302e-01 4.60805604e-03 5.69766343e-01 1.63383126e-01
2.57630527e-01 4.52572882e-01 -3.75483602e-01 4.11991104e-02
-7.81869113e-01 2.75276192e-02 3.74884456e-01 5.74062467e-01
9.95942175e-01 -1.84509441e-01 -5.92409559e-02 1.07742560e+00
-2.48761490e-01 4.13982034e-01 6.19428515e-01 -7.02039242e-01
2.48814777e-01 6.95453465e-01 -1.65348232e-01 -9.83781397e-01
-5.92006207e-01 -5.05226552e-01 -1.19366145e+00 3.85920316e-01
2.74155378e-01 2.06235549e-04 -9.92855787e-01 1.36662126e+00
6.86294854e-01 1.18985131e-01 -1.42197937e-01 9.89653707e-01
7.91335046e-01 5.96098125e-01 2.59617511e-02 2.82505830e-03
1.87803996e+00 -1.06294429e+00 -7.70268217e-02 -5.53297102e-01
6.20955229e-01 -5.15246391e-01 1.17531657e+00 5.97972453e-01
-9.92790461e-01 -7.39368618e-01 -1.20680559e+00 -4.18075621e-01
-2.50175297e-01 2.22217023e-01 6.03286862e-01 3.87116492e-01
-1.04793799e+00 7.00630724e-01 -1.05921865e+00 -2.52623349e-01
4.44375783e-01 7.54400373e-01 -4.51413274e-01 8.36346485e-03
-6.89974308e-01 4.17588770e-01 2.61950344e-01 9.02264267e-02
-4.10928249e-01 -8.74836028e-01 -6.74025178e-01 1.04386635e-01
1.94195792e-01 -6.79890752e-01 7.29933202e-01 -1.02483070e+00
-1.55357933e+00 9.23992693e-01 -1.59325540e-01 -5.68454385e-01
1.29199162e-01 -7.53083378e-02 -1.65922299e-01 1.34684965e-01
-8.49223044e-03 7.57456660e-01 1.22232461e+00 -6.55799150e-01
-7.58172035e-01 -5.38062096e-01 -1.24872245e-01 2.61226833e-01
-5.46101332e-01 -8.83814320e-02 -7.34371126e-01 -3.89546871e-01
8.31736252e-02 -9.46851850e-01 -4.46981966e-01 -1.73484206e-01
-1.85065761e-01 -4.33353409e-02 7.01416314e-01 -2.24853665e-01
7.74071097e-01 -2.33569074e+00 1.37160063e-01 2.93794751e-01
3.10492665e-01 1.85520619e-01 -1.06970267e-02 -1.25919521e-01
2.93119494e-02 -3.73193830e-01 -7.39795417e-02 -1.64777935e-01
-3.79786521e-01 1.53176129e-01 -2.66472131e-01 4.29403067e-01
3.40722054e-01 7.12820470e-01 -5.78975797e-01 -3.69694024e-01
2.48014897e-01 6.20517194e-01 -6.96238518e-01 1.21768616e-01
-5.01073971e-02 4.45753187e-01 -4.38212425e-01 3.69720638e-01
6.98007762e-01 -8.02847564e-01 3.41973752e-01 -3.15478563e-01
-1.73652142e-01 1.91715866e-01 -1.07425296e+00 1.53045797e+00
-6.27688348e-01 5.48552215e-01 5.26649877e-02 -1.03188062e+00
8.85303259e-01 -1.69578105e-01 2.97541946e-01 -1.14853811e+00
2.24310398e-01 2.26572543e-01 -1.33087859e-01 -1.67800307e-01
7.87130117e-01 1.15260482e-01 -1.19672880e-01 4.01693702e-01
-1.54594049e-01 2.65699565e-01 1.65895030e-01 -2.29140908e-01
1.25517023e+00 -3.49117756e-01 3.50865334e-01 -4.70231831e-01
3.49022955e-01 1.56433240e-01 4.39433783e-01 7.33167052e-01
1.65689871e-01 5.60132802e-01 4.87337232e-01 -7.95936525e-01
-1.11157203e+00 -5.77224731e-01 -4.38452899e-01 1.20627189e+00
2.69432753e-01 -3.58140737e-01 -7.14215636e-01 -4.13915724e-01
-2.54268199e-02 -2.77652545e-03 -6.18387043e-01 5.72474599e-02
-8.59930336e-01 -1.02892506e+00 2.32657790e-01 5.53018510e-01
3.74317437e-01 -1.16323102e+00 -1.16898119e+00 2.54978150e-01
4.22815144e-01 -1.12664592e+00 -1.19281761e-01 3.17265987e-01
-9.72202182e-01 -8.19011807e-01 -6.50593281e-01 -7.36568630e-01
9.22024608e-01 3.62937659e-01 1.15652180e+00 2.68597752e-01
-6.77267373e-01 -1.06149372e-02 -8.69301427e-03 -6.96099550e-02
-1.02471188e-01 3.97020876e-01 -3.20102870e-02 7.71967322e-02
1.63753301e-01 -6.01899505e-01 -9.22282517e-01 3.87957454e-01
-7.75135756e-01 2.79538125e-01 9.18003917e-01 1.22937953e+00
1.17049110e+00 -1.57971814e-01 3.98112923e-01 -1.08104229e+00
2.02553272e-01 -3.41977358e-01 -9.55661833e-01 3.15159559e-02
-3.51346850e-01 2.02156734e-02 9.02104080e-01 -6.15438879e-01
-7.11958647e-01 1.36924252e-01 -2.58845538e-01 -4.07659590e-01
-3.25784326e-01 2.37444207e-01 1.77092999e-01 -4.29342180e-01
7.16736734e-01 2.13880509e-01 1.79550841e-01 -3.44741166e-01
6.23646053e-03 4.94146168e-01 5.01116097e-01 -5.91695905e-01
2.92384654e-01 4.75654930e-01 1.29006561e-02 -1.24358582e+00
-4.76203352e-01 -2.52758324e-01 -3.46834391e-01 4.44782972e-02
6.12466097e-01 -1.01422477e+00 -9.81367528e-01 4.03454095e-01
-7.75822639e-01 -6.14115298e-01 -1.46270350e-01 2.50952333e-01
-3.43196869e-01 1.87552005e-01 -7.59448886e-01 -3.44268590e-01
-6.70711100e-01 -1.46199286e+00 1.43140614e+00 3.71542692e-01
-1.39826715e-01 -5.46691895e-01 -1.77679300e-01 3.18233311e-01
3.65490586e-01 5.06650954e-02 1.07713866e+00 -2.36800179e-01
-8.82568359e-01 -1.02846466e-01 -5.18567920e-01 -9.78723913e-02
-6.28575385e-02 -2.41569847e-01 -9.18501079e-01 -6.02284431e-01
-2.01212138e-01 -4.13687021e-01 8.30621243e-01 3.61773729e-01
1.50015581e+00 -3.46622691e-02 -5.72134018e-01 7.23186612e-01
1.54548013e+00 -1.49295600e-02 7.44085968e-01 3.76834840e-01
8.55121255e-01 4.45123762e-01 5.97873032e-01 4.13510740e-01
4.90383580e-02 9.30263579e-01 2.87848383e-01 -3.41912031e-01
3.85850221e-02 3.10864121e-01 -1.07192509e-02 7.40745842e-01
-3.97575162e-02 -5.47806621e-02 -8.72336447e-01 4.57850814e-01
-1.59583819e+00 -5.97884834e-01 6.31550103e-02 2.46591353e+00
6.31276667e-01 1.62731349e-01 6.55916112e-04 4.01662588e-02
5.57646334e-01 1.76188871e-01 -7.99341142e-01 -3.38663608e-01
1.65368572e-01 5.87137341e-01 6.46989167e-01 2.49930516e-01
-1.08757865e+00 7.91630447e-01 6.27058125e+00 1.08394182e+00
-1.61234844e+00 3.66393663e-02 1.09767616e+00 -3.61461669e-01
1.33719370e-01 -3.43586266e-01 -1.16617727e+00 4.40225035e-01
9.44820821e-01 2.81374753e-01 2.35782877e-01 1.00929177e+00
-1.86229765e-01 -3.18908751e-01 -8.76571059e-01 1.12828195e+00
-8.75086337e-02 -1.63009775e+00 -8.03106353e-02 2.72290707e-01
5.79493761e-01 3.56487304e-01 1.70622006e-01 8.15718621e-03
-1.10410266e-01 -9.62148726e-01 4.11676198e-01 -5.17738014e-02
1.10910583e+00 -8.68825376e-01 5.03075659e-01 2.94368774e-01
-1.01003611e+00 -1.21692970e-01 -6.25741780e-01 1.16537392e-01
-1.59965798e-01 7.27524817e-01 -5.18728912e-01 1.54442042e-01
1.01761734e+00 3.85210425e-01 -5.03159642e-01 6.80281878e-01
4.42601800e-01 2.30303168e-01 -6.53206825e-01 -2.12760810e-02
7.01181144e-02 -1.92796588e-02 1.62075654e-01 1.11125016e+00
5.26375175e-01 1.78948611e-01 1.68702733e-02 3.53665173e-01
-8.44720453e-02 2.28075787e-01 -3.32425326e-01 4.51525867e-01
3.17027718e-01 1.40996540e+00 -1.08791065e+00 -3.73379022e-01
-4.64659244e-01 9.55409944e-01 8.71670604e-01 -3.08790319e-02
-7.22258985e-01 -3.80147576e-01 8.20913494e-01 2.18936145e-01
7.21370220e-01 -1.56380028e-01 -4.52520937e-01 -9.97877181e-01
2.42604703e-01 -9.65814531e-01 3.59648556e-01 -2.31475249e-01
-8.98969114e-01 8.75724435e-01 -4.71131146e-01 -1.20604360e+00
2.90279668e-02 -7.98378050e-01 -4.47614729e-01 5.27963817e-01
-1.52698910e+00 -7.58999586e-01 -5.88576078e-01 4.57788885e-01
4.86454666e-01 -4.45582382e-02 8.25015187e-01 4.41630721e-01
-8.54812205e-01 7.71605194e-01 1.40528917e-01 -1.99012905e-01
5.50437272e-01 -8.80347848e-01 6.25426531e-01 4.69433039e-01
5.72552495e-02 4.99505877e-01 5.01662314e-01 -4.03400391e-01
-1.70078111e+00 -1.23806953e+00 3.49317908e-01 -1.06214747e-01
4.17820752e-01 -4.00174618e-01 -1.25340676e+00 2.80811369e-01
1.46632586e-02 4.97652113e-01 5.59888601e-01 1.98509321e-01
-3.26465875e-01 -3.94878656e-01 -9.84896779e-01 5.58011234e-01
7.76153445e-01 -2.38836795e-01 1.35563821e-01 5.29765666e-01
4.22648370e-01 -8.23118508e-01 -9.55064595e-01 3.44524801e-01
4.74941820e-01 -1.05570245e+00 9.97426808e-01 -1.57946914e-01
3.06914866e-01 -3.22210789e-01 1.27687618e-01 -8.08783531e-01
-4.94751126e-01 -3.75402123e-01 1.20938495e-02 5.87317765e-01
5.15047252e-01 -8.52485299e-01 9.56620455e-01 3.50664377e-01
-2.35258080e-02 -1.08421338e+00 -1.04726410e+00 -4.38551307e-01
-1.73094004e-01 -2.66928643e-01 5.22896349e-01 6.70246959e-01
-9.38740894e-02 3.72533262e-01 -1.93676829e-01 4.14362043e-01
6.12240613e-01 4.66538250e-01 7.77964294e-01 -1.07178271e+00
-6.63532615e-01 -4.72572029e-01 -6.06839538e-01 -1.18892193e+00
-4.40427139e-02 -4.70463425e-01 7.47345686e-02 -1.03270817e+00
1.07849136e-01 -8.74996006e-01 -2.71992922e-01 5.55586815e-01
-7.95276389e-02 7.89228797e-01 -1.55101836e-01 2.90598959e-01
-5.51001310e-01 1.55621111e-01 1.02995300e+00 -1.05697364e-01
-2.49378651e-01 -2.17702493e-01 -5.33870637e-01 4.47665244e-01
6.71430707e-01 -3.39556307e-01 -2.17180535e-01 -6.40157998e-01
1.27969310e-01 1.23591930e-01 3.94355208e-01 -1.29940295e+00
3.10353100e-01 4.36829813e-02 4.67052281e-01 -3.19159359e-01
4.64187920e-01 -4.98434186e-01 2.76785642e-01 3.75858426e-01
-8.83413181e-02 4.70095798e-02 5.91405690e-01 5.26392817e-01
-1.91975936e-01 -2.84480322e-02 1.02546382e+00 2.59595457e-02
-7.63174236e-01 3.43585223e-01 -2.17586130e-01 -3.03476274e-01
1.02959192e+00 -2.12277889e-01 -4.78986531e-01 2.84144402e-01
-5.73269963e-01 -1.82290059e-02 5.58239579e-01 2.83515513e-01
4.32310075e-01 -1.12812877e+00 -5.08324265e-01 6.20205760e-01
1.70366824e-01 3.88422489e-01 7.98890471e-01 8.93383145e-01
-7.11220980e-01 3.66227359e-01 -2.63199031e-01 -1.05489028e+00
-1.51907825e+00 5.98063350e-01 1.34753168e-01 -3.41864824e-01
-1.18225241e+00 7.26518571e-01 5.27763486e-01 7.62435570e-02
-9.16596577e-02 -2.22751856e-01 -3.37425023e-01 -1.84317619e-01
9.07872319e-01 7.72455558e-02 5.83310485e-01 -5.09630978e-01
-2.94810504e-01 9.10608709e-01 -5.22684336e-01 3.47183734e-01
1.26282763e+00 1.93932086e-01 -9.60437804e-02 -3.92979383e-02
1.16623640e+00 -1.94196224e-01 -1.40168834e+00 -1.65905401e-01
-2.10526124e-01 -3.92459154e-01 2.02885106e-01 -1.11663669e-01
-1.22599506e+00 7.76636064e-01 7.20466614e-01 1.73018783e-01
1.33392251e+00 -6.49027526e-02 9.47618902e-01 5.42302847e-01
6.92224085e-01 -1.00546145e+00 -7.45724589e-02 3.22981775e-01
5.19694865e-01 -1.25385058e+00 1.79606110e-01 -3.85111183e-01
-4.17069048e-01 1.05244482e+00 7.20242500e-01 -4.12212342e-01
4.17379200e-01 5.20377994e-01 -7.25020543e-02 -1.70339480e-01
-9.13352549e-01 8.10105503e-02 1.95375368e-01 2.76219040e-01
1.63153231e-01 1.90539155e-02 -2.17826571e-02 2.24128962e-01
4.31891298e-03 -1.47344500e-01 2.20806718e-01 8.87407899e-01
-4.18460906e-01 -9.93303895e-01 -2.27900788e-01 7.32706904e-01
-4.16517019e-01 -1.75387424e-03 3.52233768e-01 5.15137315e-01
-2.53389068e-02 3.63753229e-01 5.80123782e-01 -3.24817866e-01
2.37824306e-01 -2.82335788e-01 4.68091428e-01 -5.41538537e-01
-6.56542778e-01 2.49871183e-02 -1.88050389e-01 -7.03884661e-01
-1.55731261e-01 -4.73058522e-01 -1.01192701e+00 -3.03507835e-01
-1.68609142e-01 -6.23559766e-02 6.34031296e-01 8.39567006e-01
1.04986262e+00 6.19373441e-01 4.16578382e-01 -1.17156851e+00
-7.36345053e-02 -6.15019202e-01 -3.49988729e-01 1.12235002e-01
3.24142814e-01 -5.33954382e-01 -1.66687310e-01 -1.40971512e-01] | [9.467338562011719, 0.296062707901001] |
45632730-e93e-4a51-a261-0cfd674be36f | continuous-episodic-control | 2211.15183 | null | https://arxiv.org/abs/2211.15183v3 | https://arxiv.org/pdf/2211.15183v3.pdf | Continuous Episodic Control | Non-parametric episodic memory can be used to quickly latch onto high-rewarded experience in reinforcement learning tasks. In contrast to parametric deep reinforcement learning approaches in which reward signals need to be back-propagated slowly, these methods only need to discover the solution once, and may then repeatedly solve the task. However, episodic control solutions are stored in discrete tables, and this approach has so far only been applied to discrete action space problems. Therefore, this paper introduces Continuous Episodic Control (CEC), a novel non-parametric episodic memory algorithm for sequential decision making in problems with a continuous action space. Results on several sparse-reward continuous control environments show that our proposed method learns faster than state-of-the-art model-free RL and memory-augmented RL algorithms, while maintaining good long-run performance as well. In short, CEC can be a fast approach for learning in continuous control tasks. | ['Aske Plaat', 'Mike Preuss', 'Thomas M. Moerland', 'Zhao Yang'] | 2022-11-28 | null | null | null | null | ['continuous-control'] | ['playing-games'] | [-6.26154989e-02 -8.65770318e-03 -4.66286659e-01 7.58479908e-02
-7.82113314e-01 -2.64223546e-01 6.66117132e-01 2.17098683e-01
-8.06525409e-01 1.45309997e+00 9.70934778e-02 9.70574915e-02
-4.19739276e-01 -9.47378635e-01 -8.00960183e-01 -7.98976600e-01
-3.01771402e-01 9.59770441e-01 2.08653778e-01 -1.50049224e-01
4.41320658e-01 1.99338570e-01 -1.82163870e+00 2.58107763e-02
7.74736822e-01 7.77885377e-01 5.61773360e-01 3.83240253e-01
-1.85222059e-01 1.23628032e+00 -4.73199785e-01 2.07017884e-01
2.42142635e-03 -7.86732793e-01 -7.80827463e-01 -1.84991524e-01
-2.55863219e-01 -4.99843627e-01 -4.38245922e-01 5.37997425e-01
6.61585927e-01 7.18655586e-01 2.47461945e-01 -9.05357778e-01
-4.73551810e-01 7.67534971e-01 -5.64374961e-02 1.59515679e-01
4.40736115e-01 3.33488107e-01 6.93385959e-01 -7.56065607e-01
8.02516878e-01 1.11994863e+00 3.57287198e-01 6.99658215e-01
-1.37274969e+00 -6.29950702e-01 4.49341327e-01 4.44128364e-01
-8.65438581e-01 -2.22537205e-01 4.54580337e-01 2.96597797e-02
1.46041703e+00 -1.70877367e-01 1.40037549e+00 1.32560456e+00
5.30116677e-01 1.21416247e+00 1.51534665e+00 -1.81598753e-01
9.51852500e-01 -2.63005793e-01 -5.07909477e-01 5.63816011e-01
-1.33028343e-01 1.03471041e+00 -9.28963959e-01 -4.13873494e-02
9.79334712e-01 -4.38577263e-03 1.44165143e-01 -7.13975012e-01
-1.21227109e+00 1.14747715e+00 3.77377272e-01 1.37367547e-01
-8.69239807e-01 7.55924702e-01 4.05313671e-01 8.11347485e-01
1.21484779e-01 9.63383019e-01 -4.25556421e-01 -7.63432205e-01
-8.96527946e-01 8.80191267e-01 6.60773039e-01 6.39225066e-01
6.02520704e-01 5.29884398e-01 -5.75680017e-01 7.55451322e-01
-3.12955081e-02 4.98681128e-01 7.99333334e-01 -1.15050304e+00
1.16890319e-01 1.29511759e-01 5.20215511e-01 -4.75937814e-01
-4.19957519e-01 -6.73263133e-01 -4.83425438e-01 4.73267972e-01
2.59105653e-01 -2.60786384e-01 -7.59740114e-01 1.69715798e+00
3.31870794e-01 3.39195520e-01 3.46365035e-01 8.86968732e-01
1.60312280e-01 7.54966676e-01 1.66864619e-01 -5.98262489e-01
5.78618169e-01 -1.02440357e+00 -1.01053023e+00 -4.91840959e-01
4.24206793e-01 -2.22538203e-01 1.04552078e+00 6.29891038e-01
-1.43840206e+00 -3.99014324e-01 -9.34288025e-01 4.59261060e-01
-2.48508140e-01 -3.09145659e-01 8.57807398e-01 2.45935753e-01
-9.26249743e-01 9.67389286e-01 -1.02620411e+00 3.59532386e-02
5.13688326e-01 4.15459931e-01 2.40043432e-01 -2.80051082e-01
-1.39594316e+00 1.38270593e+00 6.44277394e-01 -4.52413559e-02
-1.63683474e+00 -6.02719367e-01 -6.20526552e-01 1.49399051e-02
9.90642011e-01 -4.45320815e-01 1.69755089e+00 -7.88458347e-01
-2.13986635e+00 1.43412620e-01 2.01276585e-01 -9.34996784e-01
3.69078428e-01 -4.52657968e-01 -2.07231879e-01 1.97055377e-03
-8.06907117e-02 8.36380422e-01 9.95928168e-01 -9.83377039e-01
-5.82724869e-01 6.76793605e-02 2.95981374e-02 4.98102218e-01
8.08936656e-02 -3.77490193e-01 1.31911874e-01 -5.71301997e-01
-1.08234167e-01 -8.35074902e-01 -6.17441535e-01 -2.58773565e-01
3.34531456e-01 -1.94906801e-01 5.27560234e-01 -1.91355839e-01
1.23280466e+00 -1.86050057e+00 5.85070491e-01 1.33286178e-01
-3.17524344e-01 1.27388746e-01 -3.37088972e-01 6.86467528e-01
2.68219948e-01 -4.02181894e-01 -2.98548102e-01 -6.67005926e-02
3.37802052e-01 6.85693145e-01 -5.61537504e-01 2.01619536e-01
9.32263806e-02 1.24230206e+00 -1.36897767e+00 -3.22325081e-01
1.47760615e-01 1.65299401e-01 -5.78793645e-01 3.22531939e-01
-9.16121662e-01 4.48448062e-01 -4.87193018e-01 4.49570507e-01
5.38583584e-02 7.68275931e-02 4.58276898e-01 8.48608673e-01
-2.32501522e-01 3.18525970e-01 -1.36313152e+00 1.90631068e+00
-6.63129628e-01 5.57731576e-02 -2.00539351e-01 -1.02082086e+00
1.01818919e+00 2.85331458e-01 5.00106990e-01 -1.54750979e+00
-1.62379146e-01 3.12377185e-01 -1.28212348e-01 -1.30055234e-01
6.37050688e-01 -2.14315146e-01 -1.45225987e-01 5.83529413e-01
8.37022364e-02 -4.00060177e-01 3.97245854e-01 -1.39854163e-01
1.20793390e+00 6.82377100e-01 3.10802788e-01 2.70589218e-02
3.39316390e-02 2.25206599e-01 6.76240921e-01 1.08185625e+00
4.75882441e-02 6.98773414e-02 5.50263464e-01 -4.42655385e-01
-7.94993162e-01 -1.15464818e+00 7.78826699e-02 1.22413719e+00
-1.57001372e-02 -3.18840325e-01 -3.62553179e-01 -4.61065859e-01
1.36699021e-01 1.02759790e+00 -6.52056515e-01 -4.86975580e-01
-7.76849627e-01 -6.02588952e-01 3.78940068e-02 6.52963102e-01
5.95860302e-01 -1.92456961e+00 -1.15551627e+00 9.79919016e-01
2.11156160e-01 -2.65342206e-01 -1.82633460e-01 8.05148780e-01
-1.22620344e+00 -8.49678814e-01 -6.58186972e-01 -6.36187971e-01
2.37202972e-01 -3.26629728e-01 1.19998753e+00 2.10894085e-02
-1.19018808e-01 5.60386300e-01 -3.15928876e-01 -1.84146002e-01
-2.67904967e-01 -3.20000239e-02 1.14289239e-01 -3.93104225e-01
8.20873380e-02 -5.16910195e-01 -5.18533289e-01 2.04625562e-01
-7.05473065e-01 -2.07955554e-01 7.88572729e-01 1.42605627e+00
1.09709859e+00 4.81023416e-02 1.22242928e+00 -7.21006453e-01
8.96753073e-01 -4.98519480e-01 -8.57348740e-01 1.21027850e-01
-9.94858682e-01 4.15126592e-01 5.24086356e-01 -7.37524450e-01
-1.13030541e+00 4.13272142e-01 1.45444676e-01 -4.17052388e-01
4.55076620e-02 8.87304425e-01 3.65702868e-01 2.01615557e-01
6.58608139e-01 6.77045703e-01 3.44747156e-01 -2.15998486e-01
4.55581665e-01 -1.70083448e-01 2.57054895e-01 -8.10427248e-01
3.58062118e-01 -1.63055331e-01 6.13829903e-02 -3.26101631e-01
-6.86758041e-01 -3.04329656e-02 -3.88230622e-01 -3.20084125e-01
6.37036145e-01 -8.61935318e-01 -5.38253963e-01 4.12183732e-01
-5.59845269e-01 -1.16428959e+00 -9.60550189e-01 5.48561096e-01
-1.43279934e+00 -3.31362158e-01 -6.53230369e-01 -1.02761936e+00
2.47510254e-01 -8.26565564e-01 5.60339510e-01 1.40898973e-01
-2.80592948e-01 -8.56973231e-01 5.42998910e-01 -3.18355560e-01
7.13842928e-01 2.08638459e-01 7.42742002e-01 -2.25367352e-01
-7.56812155e-01 1.42540872e-01 5.27495623e-01 -2.15577856e-01
-2.88963228e-01 -6.37890339e-01 -5.16346872e-01 -5.43666184e-01
-5.29008433e-02 -1.04462731e+00 9.22836781e-01 4.21923548e-01
1.00511265e+00 -4.36598718e-01 -8.63545835e-02 9.72499251e-02
1.36435318e+00 4.94599968e-01 8.66602480e-01 6.23544693e-01
-5.96139394e-02 2.50100076e-01 1.37569475e+00 8.93113852e-01
7.74856582e-02 6.61047518e-01 5.38940370e-01 2.09857687e-01
1.21933311e-01 -6.07569039e-01 6.10554695e-01 4.39921200e-01
7.69032314e-02 1.32398501e-01 -6.27008855e-01 5.95702350e-01
-2.31323123e+00 -1.45324945e+00 4.58920568e-01 2.18920970e+00
1.21738243e+00 2.16855019e-01 3.28643799e-01 1.23423390e-01
3.03128660e-01 2.09909841e-01 -1.05225861e+00 -6.02959931e-01
-6.22117296e-02 6.85233414e-01 2.72072434e-01 3.97099257e-01
-8.70798826e-01 1.21786487e+00 6.89815760e+00 8.52778673e-01
-1.09886920e+00 2.07157090e-01 2.38876119e-01 -5.90088069e-01
-6.12934642e-02 -1.40002683e-01 -5.78085244e-01 2.85912871e-01
1.34216392e+00 -2.02678531e-01 9.74967122e-01 1.10743272e+00
6.16694614e-02 -5.51069677e-01 -1.03311336e+00 8.63032758e-01
-4.28988397e-01 -1.44050360e+00 -3.80546391e-01 -3.64621170e-02
1.03627145e+00 -1.17663592e-01 4.46426690e-01 1.09940159e+00
7.50614941e-01 -1.26715708e+00 6.89121306e-01 7.78680325e-01
5.59352815e-01 -1.20504940e+00 3.99213523e-01 4.45139468e-01
-8.91968548e-01 -7.44456112e-01 -5.84130287e-01 -2.96076417e-01
8.29277337e-02 8.11811537e-02 -5.40413976e-01 1.31164849e-01
5.94431639e-01 7.73172557e-01 -2.64487296e-01 1.18039536e+00
-4.65738744e-01 5.46028793e-01 -9.48959216e-02 -4.84980583e-01
6.10758781e-01 6.08320348e-02 2.28788063e-01 5.97890735e-01
3.39721411e-01 1.68102577e-01 3.85833234e-01 8.97130966e-01
2.54351676e-01 -8.86013508e-02 -8.53062749e-01 -2.42550194e-01
5.32931149e-01 6.27260387e-01 -6.69320226e-01 -2.14870572e-01
-1.39884632e-02 9.27968264e-01 6.21607661e-01 2.63820022e-01
-8.23751271e-01 -2.70359796e-02 3.02366585e-01 3.87513787e-02
7.78617442e-01 -5.11875927e-01 2.95238681e-02 -7.46061802e-01
-3.97354215e-01 -1.05049527e+00 2.81506687e-01 -5.27539432e-01
-9.51911449e-01 2.10327238e-01 8.07928480e-03 -1.27229464e+00
-9.09158051e-01 -5.59554994e-02 -3.86924416e-01 4.23021853e-01
-1.58735752e+00 -3.89988482e-01 2.61317313e-01 7.37990499e-01
7.05276132e-01 -3.31270337e-01 1.30892050e+00 -1.65532887e-01
-2.60217994e-01 2.98400313e-01 6.26111269e-01 -5.13531804e-01
4.59329069e-01 -1.42187488e+00 -2.86809318e-02 2.76111931e-01
1.76622182e-01 3.43919098e-01 5.88428438e-01 -8.57522905e-01
-1.75232506e+00 -9.85636711e-01 4.57104951e-01 -9.16085094e-02
4.94433343e-01 -2.69851446e-01 -9.04974282e-01 4.22682703e-01
2.78172433e-01 -6.93875402e-02 2.70116985e-01 1.28211617e-01
6.57624602e-02 1.83067657e-02 -1.07917011e+00 4.96245325e-01
9.37626839e-01 -3.48630279e-01 -7.42971420e-01 3.21254551e-01
4.99581397e-01 -6.91073418e-01 -8.03149223e-01 -2.09807884e-02
3.64155769e-01 -8.39467645e-01 9.57027912e-01 -7.39049911e-01
4.03223932e-01 -1.25936717e-01 1.30667794e-03 -1.81594861e+00
-4.21906680e-01 -6.63153410e-01 -7.42578089e-01 6.39080524e-01
3.25899005e-01 -5.44484735e-01 6.91326261e-01 3.56795669e-01
-2.48481199e-01 -1.10697722e+00 -1.19648898e+00 -1.05320561e+00
1.95367664e-01 -1.23537965e-01 5.18651128e-01 7.05802500e-01
7.54335430e-03 4.79511730e-02 -7.15213299e-01 -3.85344356e-01
4.23202008e-01 3.81040841e-01 4.30860549e-01 -1.03875828e+00
-5.40944755e-01 -2.42902189e-01 1.74107566e-01 -8.53804529e-01
3.74500155e-01 -6.91557646e-01 3.45580339e-01 -1.54300165e+00
-1.34143829e-01 -6.47891283e-01 -5.09220600e-01 6.89680934e-01
1.41285002e-01 -1.78913221e-01 1.87848628e-01 7.80558661e-02
-1.02876937e+00 1.25922382e+00 1.53518748e+00 -5.93879707e-02
-5.89965761e-01 -1.00299917e-01 -2.83560455e-01 2.42791146e-01
1.06823719e+00 -7.22039700e-01 -7.62639225e-01 -1.98069979e-02
4.06391591e-01 6.50502205e-01 2.58868486e-01 -1.20958495e+00
3.08010399e-01 -6.09771073e-01 5.91684461e-01 -5.50371766e-01
4.52191532e-01 -6.20291293e-01 1.43229127e-01 8.77954841e-01
-6.15087509e-01 3.49244952e-01 3.30541849e-01 9.93034720e-01
-2.16598317e-01 -2.41651654e-01 6.90793633e-01 -4.30883050e-01
-1.00950181e+00 1.82763696e-01 -9.54245448e-01 2.46956334e-01
1.15990472e+00 -1.24031238e-01 -1.76838964e-01 -3.77540201e-01
-1.17671967e+00 5.04869044e-01 1.42066777e-01 4.02297795e-01
1.08523118e+00 -1.47082186e+00 -4.50716376e-01 -2.62847776e-03
-3.26013833e-01 -1.44031480e-01 3.16809177e-01 6.81064665e-01
-4.99670021e-02 3.60930771e-01 -7.17826188e-01 -3.20360184e-01
-5.31927645e-01 8.81817341e-01 3.95206898e-01 -5.37020981e-01
-8.26891243e-01 5.37297964e-01 -5.51796019e-01 -3.55038226e-01
5.53303897e-01 -1.68675900e-01 -2.38622397e-01 3.23470026e-01
4.80096340e-01 2.86110312e-01 -1.69279844e-01 2.19794393e-01
6.85899984e-03 2.08062902e-01 -1.67347342e-02 -5.14621198e-01
1.62726331e+00 1.33690596e-01 1.71605378e-01 8.08918536e-01
4.90969360e-01 -6.34347260e-01 -1.62756813e+00 -1.96218580e-01
1.34859100e-01 -5.94868124e-01 2.19606176e-01 -1.25810421e+00
-8.92147124e-01 5.95364571e-01 7.11003304e-01 -1.63092494e-01
1.01051092e+00 -4.15903986e-01 5.98507166e-01 9.38685656e-01
8.09515715e-01 -1.86104691e+00 7.52075553e-01 8.51763189e-01
1.21129489e+00 -9.24793780e-01 -8.20291191e-02 4.60346311e-01
-9.02485967e-01 8.63391399e-01 7.72285104e-01 -3.76696497e-01
3.80248010e-01 1.87102512e-01 -2.75284320e-01 1.66829918e-02
-1.36345696e+00 -3.62246305e-01 -3.56376439e-01 7.39794731e-01
9.98511016e-02 -9.87305939e-02 -2.85194159e-01 2.73656398e-01
-4.65849377e-02 3.12412173e-01 4.78253543e-01 1.42111325e+00
-6.59765601e-01 -1.38969350e+00 -1.96240023e-01 5.39086282e-01
-1.35523841e-01 7.52071887e-02 -1.35747090e-01 8.17630827e-01
-3.12198102e-01 7.16957033e-01 1.01592224e-02 -4.30494100e-02
2.81592578e-01 2.45435566e-01 8.31109345e-01 -5.39098203e-01
-9.64766979e-01 5.60729988e-02 1.04974799e-01 -1.06950736e+00
-1.89672112e-01 -9.61713612e-01 -1.68035901e+00 -1.51016504e-01
1.24698512e-01 2.94236600e-01 3.98834258e-01 6.53000355e-01
3.16235721e-01 7.57284999e-01 5.24157166e-01 -7.70912588e-01
-1.05357242e+00 -6.81272447e-01 -7.19311833e-01 -5.12570590e-02
2.05934539e-01 -1.21260333e+00 -1.33712422e-02 -5.58209062e-01] | [4.087830066680908, 1.7945263385772705] |
119816f4-3efa-4894-8a87-a5f59c5a7a3d | do-deep-learning-models-really-outperform | 2302.07134 | null | https://arxiv.org/abs/2302.07134v3 | https://arxiv.org/pdf/2302.07134v3.pdf | Do Deep Learning Models Really Outperform Traditional Approaches in Molecular Docking? | Molecular docking, given a ligand molecule and a ligand binding site (called ``pocket'') on a protein, predicting the binding mode of the protein-ligand complex, is a widely used technique in drug design. Many deep learning models have been developed for molecular docking, while most existing deep learning models perform docking on the whole protein, rather than on a given pocket as the traditional molecular docking approaches, which does not match common needs. What's more, they claim to perform better than traditional molecular docking, but the approach of comparison is not fair, since traditional methods are not designed for docking on the whole protein without a given pocket. In this paper, we design a series of experiments to examine the actual performance of these deep learning models and traditional methods. For a fair comparison, we decompose the docking on the whole protein into two steps, pocket searching and docking on a given pocket, and build pipelines to evaluate traditional methods and deep learning methods respectively. We find that deep learning models are actually good at pocket searching, but traditional methods are better than deep learning models at docking on given pockets. Overall, our work explicitly reveals some potential problems in current deep learning models for molecular docking and provides several suggestions for future works. | ['Guolin Ke', 'Hang Zheng', 'Zhifeng Gao', 'Shuqi Lu', 'Yuejiang Yu'] | 2023-02-14 | null | null | null | null | ['molecular-docking'] | ['medical'] | [-3.65817696e-01 -2.97885388e-01 -4.13684636e-01 -1.83456138e-01
-7.54580975e-01 -7.60551035e-01 -2.74683554e-02 1.83852434e-01
-3.39643538e-01 1.08245707e+00 -7.25424737e-02 -8.81182313e-01
2.58978784e-01 -6.91041291e-01 -1.06469691e+00 -1.00512660e+00
-1.91583201e-01 6.02103829e-01 1.31158784e-01 -3.22302341e-01
1.87938139e-01 7.70080328e-01 -6.70258939e-01 3.14397961e-01
6.01318538e-01 3.50916356e-01 9.55520272e-02 2.31668323e-01
3.51533890e-02 3.40709507e-01 -5.01521170e-01 -5.28614581e-01
-6.20539933e-02 -3.06080431e-01 -9.17673528e-01 -5.27620137e-01
3.23063463e-01 -2.76729703e-01 -4.38211150e-02 8.54084849e-01
1.04637718e+00 -7.34444633e-02 6.67782605e-01 -5.40926695e-01
-5.12384057e-01 -1.06929243e-01 -3.85248303e-01 1.19952589e-01
5.73252439e-01 2.13018090e-01 1.01173437e+00 -1.15410864e+00
5.81684649e-01 1.01778209e+00 1.17024601e+00 7.47243226e-01
-1.22219336e+00 -8.61750245e-01 -4.47588600e-02 1.21849164e-01
-1.37202585e+00 -1.22258238e-01 2.70741045e-01 -7.19004691e-01
1.27813399e+00 7.67974406e-02 5.58995843e-01 1.18552136e+00
8.99337709e-01 4.03627008e-01 5.16506374e-01 -1.24315955e-01
5.31621277e-01 -3.89741182e-01 -9.94946733e-02 6.70848906e-01
1.40580339e-02 2.41091341e-01 -4.03946847e-01 -8.46112788e-01
8.79757583e-01 3.89455646e-01 -4.17911381e-01 -5.91914415e-01
-9.87277985e-01 1.12281966e+00 7.79176474e-01 1.48991168e-01
-3.83142173e-01 4.36118469e-02 4.21604067e-01 1.65734347e-02
8.36749375e-02 4.85874146e-01 -8.35378826e-01 2.05011636e-01
-5.50717890e-01 7.14228809e-01 8.35266829e-01 3.87742549e-01
7.37736940e-01 -4.24504787e-01 -6.78721517e-02 4.34451103e-01
5.32007694e-01 -2.24426776e-01 3.07342529e-01 -3.17665011e-01
1.41331449e-01 5.27618766e-01 4.23563749e-01 -8.12238574e-01
-7.54381418e-01 -1.68508247e-01 -8.31434250e-01 4.84104514e-01
5.72768867e-01 -3.61579388e-01 -7.96742618e-01 1.58115840e+00
3.12775463e-01 2.07238868e-01 2.15983108e-01 1.00202203e+00
1.07500410e+00 6.60156965e-01 4.46534634e-01 -9.10340250e-02
1.36041355e+00 -1.01070273e+00 -5.05902052e-01 1.30177559e-02
9.26567614e-01 -9.09339130e-01 9.64639544e-01 4.38911051e-01
-7.13374138e-01 -4.12127137e-01 -1.14104295e+00 -1.95554465e-01
-5.33332348e-01 -6.10116031e-03 1.20939147e+00 5.98005295e-01
-8.18696022e-01 8.87161314e-01 -8.20309162e-01 -2.78861612e-01
4.02699649e-01 1.10266781e+00 -5.97975492e-01 6.23370335e-02
-1.35494220e+00 9.77993011e-01 3.55121911e-01 5.41705713e-02
-1.05850482e+00 -5.92821121e-01 -4.67181832e-01 -6.32561669e-02
-8.13161135e-02 -7.13003278e-01 1.23746622e+00 -6.20303094e-01
-1.48902738e+00 7.87340879e-01 -1.68261111e-01 -2.68713057e-01
5.61651066e-02 1.61219221e-02 4.27175239e-02 -4.95326310e-01
-2.13605061e-01 5.84980428e-01 -9.92686898e-02 -8.88849378e-01
-2.68056840e-01 -5.46763301e-01 4.95796382e-01 2.55702078e-01
3.63810718e-01 1.68640777e-01 -3.27981502e-01 -4.50682670e-01
-8.95612687e-02 -9.11844671e-01 -7.67703891e-01 -1.05614811e-01
-4.21371639e-01 -4.30963218e-01 2.67940253e-01 -2.54846007e-01
1.12479436e+00 -1.74616265e+00 2.43307635e-01 2.35958457e-01
5.33625960e-01 4.70889956e-01 -4.43876795e-02 8.45913231e-01
-5.61840713e-01 -3.67343277e-02 1.89962909e-01 1.53007075e-01
-2.10519880e-01 6.32624403e-02 -1.92355081e-01 7.52870083e-01
-2.84337431e-01 9.48380232e-01 -7.19172418e-01 5.38646206e-02
-1.30042851e-01 7.19116688e-01 -8.87807846e-01 1.99029326e-01
-5.62296271e-01 6.22180641e-01 -6.52211726e-01 9.63418841e-01
7.71523058e-01 -3.91008973e-01 6.63615525e-01 -2.59089768e-01
1.00433491e-02 3.82862628e-01 -7.69762278e-01 1.77273083e+00
2.50606507e-01 2.44462371e-01 -2.98058152e-01 -7.35571206e-01
7.53616571e-01 5.46697676e-01 5.15586495e-01 -5.18202960e-01
7.00310245e-02 3.58366460e-01 4.65140224e-01 -1.95403501e-01
-3.07388276e-01 -5.48325360e-01 3.92808229e-01 1.56849653e-01
-1.69437975e-02 6.41224921e-01 -2.37017199e-01 -2.83588618e-01
1.03262126e+00 3.18653524e-01 4.99405205e-01 -2.29055136e-01
4.05154049e-01 2.03677773e-01 4.95719582e-01 2.31474414e-01
-7.08240494e-02 3.17876488e-01 5.72671950e-01 -1.22109306e+00
-7.50569105e-01 -6.11780941e-01 -3.50017428e-01 1.44011521e+00
1.56528801e-01 -7.24742770e-01 -1.03563905e+00 -6.71933591e-01
1.89400658e-01 -2.32271224e-01 -8.35089982e-01 -4.36445475e-02
-3.83281231e-01 -1.44733369e+00 6.18960202e-01 3.52328062e-01
-1.57956064e-01 -1.09226370e+00 -2.54078954e-01 5.37137806e-01
1.14205770e-01 -2.69644767e-01 -3.53103518e-01 8.20094287e-01
-7.64212549e-01 -1.56940556e+00 -8.04520130e-01 -1.00702226e+00
3.39333147e-01 8.28739330e-02 1.09165549e+00 1.49998695e-01
-1.17584608e-01 -5.86284339e-01 -2.89375596e-02 -4.96639132e-01
6.43792152e-02 -2.84258686e-02 2.85275549e-01 -3.94585967e-01
9.98419821e-01 -6.02966785e-01 -8.77481222e-01 4.78322148e-01
-5.85669696e-01 -2.81445324e-01 6.46351874e-01 1.03552616e+00
9.80717242e-01 -3.33036095e-01 2.66766578e-01 -8.36532116e-01
7.81808376e-01 -5.63187957e-01 -5.14437497e-01 2.02369075e-02
-2.69607365e-01 -1.90985929e-02 6.51903152e-01 -4.54541773e-01
-2.95577139e-01 3.96666408e-01 -9.43375647e-01 -1.64771467e-01
-1.94498450e-01 8.90779853e-01 -3.64719123e-01 -6.90845072e-01
9.32909548e-01 2.29629260e-02 -1.08043164e-01 -9.82559502e-01
-1.80943012e-01 3.08075845e-01 5.09577803e-02 -6.35895729e-01
3.53987187e-01 2.27805942e-01 3.85088138e-02 -9.57502574e-02
-6.66944563e-01 -4.98812765e-01 -8.07189107e-01 5.11042714e-01
9.68427062e-01 -9.64733183e-01 -1.47928357e+00 8.42124894e-02
-1.38231063e+00 -4.87537742e-01 5.80706000e-01 5.47229588e-01
-4.84083414e-01 5.02773762e-01 -6.34990871e-01 1.67175494e-02
-3.12779039e-01 -1.73054576e+00 1.18245864e+00 -5.91118373e-02
-1.96999788e-01 -1.38144326e+00 8.54081392e-01 1.74454629e-01
9.45216492e-02 5.03157973e-01 1.20342374e+00 -1.00626636e+00
-4.77413118e-01 -3.61452520e-01 -5.66227622e-02 -1.58478156e-01
1.99117050e-01 -2.57415354e-01 -8.65562618e-01 -5.48607469e-01
-2.45148957e-01 -4.87583399e-01 5.77688813e-01 4.40114886e-01
1.06101704e+00 -2.56175041e-01 -8.35105836e-01 9.93293881e-01
1.40387416e+00 6.88924432e-01 9.95524824e-01 6.58464909e-01
7.11874902e-01 7.86268041e-02 4.57759559e-01 1.88269481e-01
9.95279625e-02 9.44229603e-01 7.13617384e-01 -7.06781089e-01
4.40217346e-01 2.19167788e-02 4.25449342e-01 3.18601131e-02
-4.92561907e-01 -2.59775370e-01 -1.14430428e+00 -2.31569692e-01
-1.78497040e+00 -8.56701910e-01 -7.39257932e-02 2.18291450e+00
1.32829213e+00 -2.76165932e-01 3.07888448e-01 -2.73546070e-01
1.44036800e-01 -4.24256384e-01 -8.36239874e-01 -2.91551918e-01
3.20473015e-02 5.13484299e-01 3.35680991e-01 7.00769484e-01
-1.20687711e+00 1.00930417e+00 7.71239614e+00 8.82580876e-01
-1.46489370e+00 5.95919564e-02 6.63609266e-01 2.10020155e-01
2.62983069e-02 4.89499560e-03 -9.93124723e-01 3.62437487e-01
9.25347090e-01 2.17969567e-01 1.85075670e-01 1.05115438e+00
1.82538047e-01 2.79688507e-01 -1.71201730e+00 1.05792427e+00
-4.29918468e-01 -1.95922458e+00 2.91982263e-01 6.19061947e-01
2.08526596e-01 1.94360152e-01 3.48928198e-02 2.80520469e-01
2.20543474e-01 -1.78297508e+00 4.15596813e-02 2.41970450e-01
4.79865193e-01 -9.37173784e-01 9.90780115e-01 3.70359868e-01
-8.75085533e-01 3.85704994e-01 -7.53526330e-01 -2.19950870e-01
-4.71045375e-01 7.94373304e-02 -8.21659207e-01 3.45985085e-01
6.44998193e-01 6.04395568e-01 -3.77711892e-01 1.13774061e+00
2.33690068e-01 2.77235478e-01 -1.12365987e-02 8.44395310e-02
5.80932379e-01 -2.81854093e-01 -1.43906519e-01 1.34281361e+00
-2.06951484e-01 1.81092173e-01 5.70475280e-01 7.08987415e-01
-2.39046484e-01 4.89406377e-01 -4.08202916e-01 1.63036153e-01
1.90840125e-01 8.93231332e-01 -4.38726723e-01 -4.03245762e-02
-6.46965921e-01 9.26954031e-01 3.71952176e-01 4.39034432e-01
-1.06614077e+00 -3.78156751e-01 1.29237413e+00 1.45197913e-01
-7.12481737e-02 -7.61540234e-02 2.31960475e-01 -1.07724130e+00
-2.62842059e-01 -1.14375615e+00 2.27080271e-01 -4.71815169e-01
-1.27902281e+00 2.93884933e-01 -6.94247186e-01 -9.84274149e-01
3.65772337e-01 -1.23908532e+00 -7.43066847e-01 1.26585495e+00
-1.47282422e+00 -9.35793459e-01 9.18547884e-02 8.10782135e-01
2.14600503e-01 -2.12161675e-01 1.31089520e+00 6.81007206e-01
-7.29506910e-01 7.73973048e-01 6.18026018e-01 1.18323803e-01
9.38178539e-01 -1.16788089e+00 4.38134700e-01 1.12064546e-02
-2.46007934e-01 1.38603628e+00 5.77895403e-01 -6.21357799e-01
-1.66100347e+00 -9.27258432e-01 6.74053013e-01 -8.36169422e-01
4.23211157e-01 -3.11979443e-01 -1.10373926e+00 7.00077236e-01
3.17866765e-02 -1.48267135e-01 1.51050115e+00 2.79110163e-01
-2.56649822e-01 3.65167528e-01 -1.00315166e+00 6.09010279e-01
6.70195282e-01 -3.67960006e-01 -3.51178825e-01 1.04147446e+00
4.34403509e-01 -8.64021778e-01 -1.04922640e+00 3.82834733e-01
7.24935353e-01 -1.12466502e+00 1.40678060e+00 -1.25231814e+00
-1.87606271e-02 -5.81160128e-01 -2.52490461e-01 -9.09646511e-01
-7.23168910e-01 -5.59987307e-01 9.89115164e-02 2.77755529e-01
5.62789142e-01 -3.83425832e-01 1.35627460e+00 3.02861959e-01
-1.52085945e-01 -1.30344558e+00 -9.78446543e-01 -5.47869980e-01
6.27720535e-01 5.55586442e-02 6.66179240e-01 1.12298477e+00
1.24576323e-01 7.49166667e-01 -2.52542883e-01 2.97650605e-01
2.06958517e-01 -2.58706789e-02 9.51063097e-01 -1.43788111e+00
-4.01790440e-01 -3.59003037e-01 -3.58016968e-01 -1.10411620e+00
9.44566280e-02 -8.46215308e-01 -3.81394058e-01 -1.49213111e+00
5.72930038e-01 -2.34229952e-01 -3.67079705e-01 9.10389602e-01
-3.55607904e-02 2.77712971e-01 -4.94664907e-01 5.30650795e-01
-4.64890391e-01 3.25868666e-01 1.35092592e+00 -2.30595946e-01
-3.38673413e-01 9.50676650e-02 -7.14455605e-01 8.16336215e-01
5.22084296e-01 -3.89911205e-01 2.11767401e-04 -3.34690139e-02
4.83973980e-01 -1.89461827e-01 2.01901659e-01 -7.08410025e-01
8.00908506e-02 -4.05155540e-01 7.29974449e-01 -3.33504587e-01
2.46809468e-01 -6.95204496e-01 2.92712659e-01 6.36439085e-01
2.74852246e-01 -3.81372273e-02 4.12110150e-01 4.85967278e-01
-1.37056500e-01 1.52947977e-01 8.74130189e-01 -2.76632190e-01
-4.81103748e-01 6.54956639e-01 -4.95438844e-01 -5.50813854e-01
8.56656134e-01 -6.16227210e-01 1.50565699e-01 2.16659948e-01
-1.24302161e+00 -7.29592051e-03 7.78625846e-01 -1.10686503e-01
5.05369365e-01 -1.26979017e+00 -1.54794037e-01 1.89697482e-02
9.19770002e-02 -4.35463078e-02 -2.04812974e-01 6.80250585e-01
-1.09788275e+00 8.70857179e-01 -2.49363109e-01 -3.61975402e-01
-1.36681092e+00 7.62728691e-01 1.08098388e+00 -1.91463023e-01
-1.91401049e-01 1.01248729e+00 8.22184026e-01 -6.55788660e-01
6.56765640e-01 -4.21875745e-01 -2.41509989e-01 -1.57825828e-01
5.85250914e-01 -1.89936429e-01 4.18248504e-01 -4.67518806e-01
-6.60626531e-01 7.32372284e-01 -3.40534627e-01 8.86673093e-01
1.28907645e+00 5.51342487e-01 -3.60480934e-01 -2.19312340e-01
1.04688442e+00 -3.39715704e-02 -1.03732598e+00 -1.50519637e-02
3.77570689e-02 -1.10818908e-01 -1.68730184e-01 -1.04873788e+00
-7.28269339e-01 1.04821551e+00 8.72525096e-01 -4.73082572e-01
6.33445024e-01 -8.23457912e-02 8.01052451e-01 1.00071895e+00
4.63449389e-01 -3.11617583e-01 1.17156170e-01 6.65402114e-01
7.30070055e-01 -1.30312169e+00 3.35892886e-02 -1.85857192e-01
-2.49719843e-01 1.46634436e+00 6.70677662e-01 3.22082005e-02
5.80584049e-01 5.62627167e-02 1.13757104e-01 -6.33345246e-01
-5.37847459e-01 3.57205197e-02 2.68888056e-01 6.62497222e-01
1.18838704e+00 -2.39263713e-01 -1.43644243e-01 9.86546636e-01
1.23539954e-01 1.98413432e-01 -3.42116877e-03 8.72899652e-01
-6.44375741e-01 -1.70985794e+00 -5.75829685e-01 -2.79628336e-01
-7.98896313e-01 -3.49470019e-01 -9.55074787e-01 6.86577260e-01
5.02923906e-01 4.21630144e-01 -3.64481509e-01 -3.82291555e-01
4.61366594e-01 6.32889867e-02 5.75270176e-01 -8.45111012e-01
-7.22390115e-01 3.04592252e-01 -3.87048006e-01 -5.82191169e-01
-1.94110811e-01 -7.72010861e-03 -1.46609724e+00 -6.66178286e-01
-4.88419473e-01 6.52524352e-01 3.97584409e-01 9.24071789e-01
6.43082082e-01 1.46203205e-01 2.20937744e-01 -1.39522624e+00
-2.27702767e-01 -6.05663896e-01 -5.63504398e-01 -1.74482003e-01
5.87849975e-01 -8.40736866e-01 1.34330556e-01 6.76881243e-03] | [4.916896343231201, 5.6805620193481445] |
c5024753-d946-4ebe-b729-3dbcff4b6e32 | fr-net-a-light-weight-fft-residual-net-for | 2305.11875 | null | https://arxiv.org/abs/2305.11875v1 | https://arxiv.org/pdf/2305.11875v1.pdf | FR-Net:A Light-weight FFT Residual Net For Gaze Estimation | Gaze estimation is a crucial task in computer vision, however, existing methods suffer from high computational costs, which limit their practical deployment in resource-limited environments. In this paper, we propose a novel lightweight model, FR-Net, for accurate gaze angle estimation while significantly reducing computational complexity. FR-Net utilizes the Fast Fourier Transform (FFT) to extract gaze-relevant features in frequency domains while reducing the number of parameters. Additionally, we introduce a shortcut component that focuses on the spatial domain to further improve the accuracy of our model. Our experimental results demonstrate that our approach achieves substantially lower gaze error angles (3.86 on MPII and 4.51 on EYEDIAP) compared to state-of-the-art gaze estimation methods, while utilizing 17 times fewer parameters (0.67M) and only 12\% of FLOPs (0.22B). Furthermore, our method outperforms existing lightweight methods in terms of accuracy and efficiency for the gaze estimation task. These results suggest that our proposed approach has significant potential applications in areas such as human-computer interaction and driver assistance systems. | ['Di Huang', 'Yun Zhou', 'Ruilong Fan', 'Bo Wu', 'Tao Xu'] | 2023-05-04 | null | null | null | null | ['gaze-estimation'] | ['computer-vision'] | [ 1.12278700e-01 -2.31358469e-01 1.91327259e-02 -4.39402461e-01
-2.43160620e-01 -2.06103697e-01 3.83422971e-02 -1.86520040e-01
-7.25301027e-01 4.26440954e-01 -3.69387537e-01 -4.90206689e-01
-7.61883482e-02 -1.98019594e-01 -4.65753406e-01 -6.56827152e-01
2.96376854e-01 -5.05998552e-01 4.78289187e-01 -8.50124508e-02
7.61005282e-01 3.32400143e-01 -2.23958468e+00 -4.72678035e-01
1.19885242e+00 1.29162025e+00 3.32783937e-01 5.58708787e-01
2.74096638e-01 4.41165686e-01 -6.81426883e-01 -1.27468228e-01
-3.54327494e-03 -1.26697868e-01 -3.45261425e-01 -2.51013070e-01
7.77738392e-01 -5.28233171e-01 2.17512064e-02 1.01241100e+00
5.96483827e-01 2.17939109e-01 3.22842687e-01 -1.46107650e+00
-3.57273757e-01 -3.73680741e-01 -1.23728955e+00 3.01360250e-01
4.47174162e-01 1.99222818e-01 4.89280909e-01 -9.01268721e-01
8.75351503e-02 9.21466470e-01 7.37008333e-01 2.65976459e-01
-7.82786250e-01 -1.10272884e+00 1.26393706e-01 5.96796930e-01
-1.47708213e+00 -7.06422806e-01 4.76598710e-01 -2.82968789e-01
8.55739236e-01 2.70601749e-01 4.81442809e-01 3.88542086e-01
2.64428884e-01 5.53440690e-01 1.26170111e+00 -5.11262238e-01
-9.12638605e-02 6.12011142e-02 4.12421495e-01 7.44595349e-01
3.75224203e-01 -3.43758687e-02 -9.92893636e-01 1.17960975e-01
5.88661373e-01 3.45213003e-02 -5.60181201e-01 -5.40668741e-02
-1.07251525e+00 6.78604960e-01 4.59187984e-01 -2.28290200e-01
-2.92269737e-01 1.22863743e-02 -1.47572509e-05 1.67314392e-02
5.02485752e-01 1.70488760e-01 -1.32254168e-01 -7.04873145e-01
-7.93453991e-01 3.10375839e-01 4.70590204e-01 1.22908616e+00
8.03888679e-01 -2.96495169e-01 3.30971479e-02 7.99811482e-01
6.10809326e-01 9.69872594e-01 2.97729254e-01 -9.95486140e-01
3.24409723e-01 5.13143241e-01 3.58570397e-01 -1.06146896e+00
-6.51167393e-01 -8.81564990e-02 -3.42510015e-01 3.04665864e-01
4.92235839e-01 -2.80247092e-01 -6.17004097e-01 1.49919808e+00
5.43804526e-01 1.45569369e-01 -4.34795409e-01 1.14250827e+00
6.84076130e-01 3.56934547e-01 -8.23155493e-02 -3.47008705e-01
1.60665429e+00 -1.00258183e+00 -9.57670331e-01 -1.87823147e-01
5.00350118e-01 -1.17771411e+00 1.24623084e+00 4.64867264e-01
-1.05727339e+00 -5.24450839e-01 -1.08192241e+00 -2.13795573e-01
6.09040149e-02 3.89748931e-01 5.97942293e-01 9.74821031e-01
-1.18884182e+00 2.78487802e-02 -7.16493189e-01 -4.53987211e-01
2.73069739e-01 6.90437496e-01 1.16395913e-01 1.61727980e-01
-6.37585282e-01 7.61373818e-01 -1.75637290e-01 1.68846697e-01
9.39155370e-02 -6.54866755e-01 -9.19370472e-01 2.01509833e-01
4.65627879e-01 -3.60373646e-01 1.44701862e+00 -6.01405144e-01
-1.83765912e+00 4.13292289e-01 -1.04875565e+00 -2.07455426e-01
1.04416065e-01 -5.49897015e-01 -4.45847958e-01 1.87028363e-01
-1.50079057e-01 6.67760074e-01 1.00401127e+00 -7.88331568e-01
-1.18159056e+00 -4.33442414e-01 3.43014836e-01 3.89821500e-01
-5.70352495e-01 2.70177752e-01 -5.85528135e-01 -1.48863435e-01
-1.76188105e-03 -1.18445015e+00 2.59713054e-01 2.04460800e-01
-1.09152749e-01 -4.57988977e-01 1.04118621e+00 -2.98784465e-01
1.67083216e+00 -2.20019746e+00 -3.92229915e-01 9.49332267e-02
6.14603043e-01 4.87143993e-01 2.33325064e-01 1.38811348e-02
3.32438916e-01 -2.44403809e-01 2.17665419e-01 -4.25588310e-01
-1.29918590e-01 -3.51793319e-01 9.51341838e-02 6.09501183e-01
-4.83995769e-03 6.29704475e-01 -6.58055365e-01 -4.24722284e-01
3.72051328e-01 6.64947927e-01 -4.78590965e-01 1.39183521e-01
4.49145764e-01 2.24945366e-01 -2.60119408e-01 4.70493138e-01
1.01699722e+00 -3.19609076e-01 -1.51544556e-01 -2.32639685e-01
-6.40919745e-01 2.55425394e-01 -9.58963752e-01 1.37470388e+00
-4.75649655e-01 1.19045484e+00 -1.52522013e-01 -3.45447063e-01
7.13459849e-01 -1.39396071e-01 1.24347538e-01 -9.07974601e-01
5.45120716e-01 1.24467462e-01 1.56143218e-01 -6.27018809e-01
7.94964373e-01 4.71525669e-01 3.60374600e-01 5.93423545e-01
-3.19977641e-01 3.99521112e-01 3.06118280e-01 -1.74184233e-01
6.82271481e-01 9.03219283e-02 3.42432588e-01 -2.87011266e-01
6.64648235e-01 -4.49097514e-01 2.81381398e-01 2.73060352e-01
-6.42899096e-01 3.01699996e-01 3.43203902e-01 -1.74084291e-01
-4.78985786e-01 -4.49843466e-01 -1.84909508e-01 1.21651125e+00
6.71710372e-01 -5.38466871e-01 -1.21336257e+00 -3.59820873e-01
-1.54663011e-01 5.61246157e-01 -4.76113528e-01 7.76698962e-02
-5.37527263e-01 -3.93400162e-01 2.48130977e-01 4.45785820e-01
5.44896841e-01 -7.15813398e-01 -1.33555532e+00 -3.67538244e-01
-2.39853755e-01 -1.08465004e+00 -8.56089413e-01 -6.55689895e-01
-6.44693255e-01 -1.21196139e+00 -6.59720123e-01 -6.91995859e-01
8.51521552e-01 1.12631834e+00 5.78421474e-01 1.90308899e-01
-1.19713224e-01 1.70285061e-01 -1.50473759e-01 -8.71007562e-01
4.59179342e-01 1.99903503e-01 2.49221489e-01 9.28364173e-02
1.05180645e+00 -1.42954528e-01 -1.11279917e+00 7.19458520e-01
-1.92358613e-01 1.46969678e-02 5.80758631e-01 6.56149328e-01
2.68893033e-01 -2.48561963e-01 3.12231153e-01 -6.64859712e-01
6.58828914e-01 -1.16201535e-01 -1.00547230e+00 1.13392202e-02
-1.03240585e+00 -1.17642231e-01 2.34004110e-01 -4.79454100e-01
-1.26879966e+00 -2.99050361e-01 1.53021127e-01 -2.18398198e-01
-7.38012195e-02 2.03423023e-01 2.87342519e-01 -6.12032890e-01
6.91999376e-01 -9.40309018e-02 3.48525107e-01 -2.93991625e-01
-5.46153896e-02 1.02383733e+00 3.20621312e-01 2.83973888e-02
5.60874343e-01 3.45710903e-01 9.51986536e-02 -1.03803277e+00
-7.44561553e-01 -7.41424739e-01 -3.08355600e-01 -3.29663664e-01
5.92621684e-01 -8.70075107e-01 -1.69264376e+00 6.78788126e-01
-9.11623776e-01 1.08292557e-01 4.48983520e-01 8.14391375e-01
-3.06845039e-01 3.53292048e-01 -1.95267960e-01 -9.97619092e-01
-5.95761895e-01 -1.27619410e+00 1.08747220e+00 8.04024518e-01
-2.05010772e-01 -5.42008817e-01 -3.48845154e-01 4.88622457e-01
5.81324399e-01 -2.13485539e-01 2.98886776e-01 1.55231103e-01
-6.03472352e-01 8.47142539e-04 -7.15604067e-01 -6.14389731e-03
1.80737451e-01 2.56385505e-02 -1.21631658e+00 -4.66372162e-01
1.71816081e-03 -8.13949853e-02 3.40215564e-01 6.45229042e-01
1.20173967e+00 1.33832060e-02 -4.77122396e-01 8.68685186e-01
1.13008213e+00 2.78631955e-01 4.74668473e-01 4.45592672e-01
6.92011178e-01 6.16248131e-01 1.14941823e+00 6.04178309e-01
7.40667164e-01 6.58019602e-01 3.53059679e-01 -9.45875272e-02
-1.50016388e-02 1.56035498e-01 1.64860457e-01 5.69783032e-01
-4.74354714e-01 -2.72245616e-01 -9.31649864e-01 4.00869489e-01
-1.76454282e+00 -6.75023139e-01 -4.09029573e-01 2.37340117e+00
6.38352513e-01 1.21265002e-01 3.03070694e-01 1.43957332e-01
6.48251772e-01 -1.01082727e-01 -7.14169085e-01 -2.68279403e-01
6.05997443e-01 2.31995836e-01 6.43217504e-01 3.56926858e-01
-8.73681962e-01 6.37897551e-01 6.47389078e+00 5.10977387e-01
-1.39828634e+00 -1.17546823e-02 9.83523130e-02 -4.89345968e-01
2.99189985e-01 -3.57081771e-01 -1.18160021e+00 7.39459336e-01
1.12578237e+00 -1.84963331e-01 4.89080817e-01 7.04532504e-01
3.21246237e-01 -5.85754693e-01 -8.40610623e-01 1.45456982e+00
3.83495957e-01 -8.10192049e-01 -6.62282884e-01 2.81566560e-01
2.50175416e-01 -6.89874664e-02 3.77336532e-01 -4.87487130e-02
-4.37199950e-01 -8.91396344e-01 5.03555536e-01 4.02576298e-01
1.08850241e+00 -1.00268888e+00 7.72138834e-01 2.40812629e-01
-1.13728929e+00 1.63223520e-02 -2.82994568e-01 -3.92341763e-01
-9.54195857e-03 2.18328208e-01 -8.18893135e-01 1.49325669e-01
1.01173198e+00 5.72077334e-01 -6.02635026e-01 1.04563498e+00
-7.87044764e-02 5.10026276e-01 -4.71095890e-01 -3.79380375e-01
-4.90916781e-02 -2.58311573e-02 2.91600406e-01 7.38712907e-01
4.81527895e-01 2.08724648e-01 -3.27988178e-01 4.75865662e-01
2.21706759e-02 -6.53974190e-02 -1.83630481e-01 3.69695485e-01
7.76589632e-01 1.26401579e+00 -4.13662940e-01 3.44189443e-02
-6.67420447e-01 6.20547593e-01 2.80361563e-01 4.80282426e-01
-1.07730210e+00 -9.86570477e-01 9.58290935e-01 2.22102776e-01
4.39491630e-01 -2.48161048e-01 -3.73720139e-01 -8.82105112e-01
3.16145957e-01 -6.24824524e-01 -1.82931304e-01 -8.50514889e-01
-6.10903203e-01 6.52930796e-01 -9.47690308e-02 -1.51657164e+00
-2.95798987e-01 -7.36386418e-01 -3.54503214e-01 1.11791062e+00
-1.98928201e+00 -1.03337252e+00 -1.01726580e+00 6.66803062e-01
4.17836547e-01 7.06978841e-03 5.36676884e-01 2.57040054e-01
-7.98074007e-01 1.15960419e+00 -8.15331656e-03 -3.64816129e-01
1.11656535e+00 -8.96962881e-01 4.44282144e-01 8.87666702e-01
-4.02991414e-01 1.07096612e+00 5.98000944e-01 -2.81930957e-02
-1.35740387e+00 -5.30254424e-01 1.04245234e+00 -2.77061075e-01
3.78757179e-01 -1.98351592e-01 -7.40141988e-01 3.16684902e-01
2.64916629e-01 -4.10674661e-02 8.20591986e-01 3.70403528e-01
-9.52173844e-02 -2.76566505e-01 -9.60998118e-01 7.13365853e-01
9.88797545e-01 -3.77957046e-01 -2.28635207e-01 -6.29372299e-02
4.22855794e-01 -7.89832711e-01 -7.17511833e-01 1.03645749e-01
1.01169872e+00 -1.29516840e+00 6.61909819e-01 1.86618045e-01
-5.16527519e-02 -4.70063120e-01 3.92142147e-01 -9.29384947e-01
-2.85368443e-01 -8.83492291e-01 -4.75125223e-01 9.11285341e-01
9.47642103e-02 -1.16930497e+00 5.25435448e-01 7.26524830e-01
1.32714793e-01 -9.67258751e-01 -5.30291498e-01 -4.47195709e-01
-6.90060079e-01 -3.59315485e-01 7.05880642e-01 4.40139383e-01
7.78823942e-02 3.60223800e-01 -2.22539485e-01 1.69447258e-01
6.63486362e-01 6.40897686e-03 1.09514177e+00 -1.31066895e+00
2.10897133e-01 -3.56369317e-01 -5.59898674e-01 -1.47439325e+00
-7.64139444e-02 1.39208645e-01 1.88851617e-02 -1.01922631e+00
3.14667001e-02 -4.48497593e-01 -3.78664106e-01 3.33388925e-01
-5.60550570e-01 4.76848602e-01 2.03026742e-01 4.07815099e-01
-4.78744805e-01 3.18023235e-01 1.08002198e+00 3.81426781e-01
-2.19686702e-01 1.74584359e-01 -8.25933397e-01 8.96787763e-01
7.33248591e-01 -2.52790034e-01 -7.32487857e-01 -5.73244333e-01
1.19715936e-01 -3.50105762e-01 2.00457156e-01 -1.03341568e+00
5.24185300e-01 -1.71506688e-01 1.76176399e-01 -7.51961768e-01
4.47935671e-01 -6.76810622e-01 -3.91807407e-01 1.43132910e-01
1.80050686e-01 2.09360659e-01 5.05312085e-01 4.44147021e-01
-8.28388110e-02 -1.53953373e-01 7.29130507e-01 7.48307765e-01
-7.33355522e-01 1.61829099e-01 -1.43943265e-01 -1.50174975e-01
1.29249549e+00 -5.79456747e-01 -7.29656577e-01 -2.78287679e-01
1.13173183e-02 1.99416146e-01 7.46024311e-01 3.32002938e-01
6.06632113e-01 -9.58926916e-01 -1.75690591e-01 5.47086298e-01
2.33676836e-01 -2.43289564e-02 2.61229604e-01 1.29425442e+00
-7.08415449e-01 7.56198347e-01 -2.95249581e-01 -9.78825688e-01
-1.75770450e+00 2.44219363e-01 -7.58098811e-02 3.41328532e-01
-3.34746271e-01 9.21516359e-01 3.75032783e-01 2.64025807e-01
2.56065458e-01 -4.30621386e-01 -4.85883772e-01 -2.80357525e-02
9.51007307e-01 7.50755370e-01 9.47494581e-02 -7.64563084e-01
-5.06066859e-01 9.83413279e-01 -2.45391950e-01 1.72459409e-01
9.82307613e-01 -6.32040501e-01 3.18982303e-02 1.90620691e-01
9.55449700e-01 2.27561638e-01 -1.44750893e+00 -1.58880770e-01
-3.19028497e-01 -8.49031746e-01 3.41948837e-01 -4.75212991e-01
-7.48931110e-01 9.05450165e-01 9.19443369e-01 -1.59590654e-02
1.59527612e+00 -4.38416421e-01 9.57198799e-01 3.85615706e-01
3.72772276e-01 -9.69075799e-01 -3.04931849e-01 3.13539177e-01
4.25389409e-01 -1.51881313e+00 2.44802594e-01 -7.30653286e-01
-4.99827653e-01 9.69298601e-01 8.73048067e-01 1.85554773e-01
6.58173442e-01 -3.41571565e-03 3.26241910e-01 -2.35472709e-01
-5.13387620e-01 -3.28552544e-01 6.56259775e-01 5.92144668e-01
5.78447223e-01 -2.56175935e-01 -4.55254048e-01 1.46668583e-01
-4.14618284e-01 1.37709484e-01 4.05673057e-01 1.08597040e+00
-4.22726631e-01 -7.05278158e-01 -4.60202187e-01 4.33041751e-01
-4.45702314e-01 -9.56428051e-02 2.05078110e-01 7.60547400e-01
-6.60308152e-02 1.33274913e+00 2.59152889e-01 -4.31059092e-01
3.89375091e-01 -1.65071443e-01 4.64808077e-01 -3.00715566e-01
-2.56815106e-01 6.93470612e-02 -1.43073618e-01 -9.15793777e-01
-8.99022639e-01 -5.52224815e-01 -9.42365348e-01 -8.23007822e-01
-7.65081167e-01 -2.58061588e-01 8.34874272e-01 7.21225321e-01
9.54117537e-01 5.14059663e-01 5.69569886e-01 -9.01532829e-01
-1.67678133e-01 -1.11246085e+00 -3.01316470e-01 7.09696487e-02
5.90994418e-01 -1.28592658e+00 -2.51827657e-01 1.99839547e-01] | [14.110392570495605, 0.10402733832597733] |
3d9127c9-8c82-42d0-b293-5491946fa71a | milestones-in-autonomous-driving-and-2 | 2306.0198 | null | https://arxiv.org/abs/2306.01980v1 | https://arxiv.org/pdf/2306.01980v1.pdf | Milestones in Autonomous Driving and Intelligent Vehicles Part II: Perception and Planning | Growing interest in autonomous driving (AD) and intelligent vehicles (IVs) is fueled by their promise for enhanced safety, efficiency, and economic benefits. While previous surveys have captured progress in this field, a comprehensive and forward-looking summary is needed. Our work fills this gap through three distinct articles. The first part, a "Survey of Surveys" (SoS), outlines the history, surveys, ethics, and future directions of AD and IV technologies. The second part, "Milestones in Autonomous Driving and Intelligent Vehicles Part I: Control, Computing System Design, Communication, HD Map, Testing, and Human Behaviors" delves into the development of control, computing system, communication, HD map, testing, and human behaviors in IVs. This part, the third part, reviews perception and planning in the context of IVs. Aiming to provide a comprehensive overview of the latest advancements in AD and IVs, this work caters to both newcomers and seasoned researchers. By integrating the SoS and Part I, we offer unique insights and strive to serve as a bridge between past achievements and future possibilities in this dynamic field. | ['Fei-Yue Wang', 'Nanning Zheng', 'Dongpu Cao', 'Jinjun Wang', 'Zixuan Li', 'Yuchen Li', 'Xiaoxiang Na', 'Bai Li', 'Siyu Teng', 'Long Chen'] | 2023-06-03 | null | null | null | null | ['ethics'] | ['miscellaneous'] | [-1.18053183e-01 1.82953939e-01 -4.06474710e-01 -5.36268413e-01
-2.45540235e-02 -4.99406427e-01 6.43296599e-01 -1.00734644e-02
-2.75981098e-01 4.03334081e-01 -1.48315921e-01 -8.38853598e-01
1.54087916e-01 -8.09864163e-01 -6.29117310e-01 -3.21641505e-01
8.86290297e-02 6.16054647e-02 4.56428587e-01 -6.81317985e-01
3.12545151e-01 6.77994192e-01 -2.23053765e+00 -2.80503273e-01
8.48470449e-01 9.88385558e-01 3.39047283e-01 4.72256392e-01
2.88353026e-01 7.98379481e-01 -3.41976970e-01 -4.71181050e-02
-7.56838964e-03 -1.75551468e-04 -5.50630629e-01 -1.03432097e-01
9.78227630e-02 -5.41130841e-01 -5.91462255e-01 5.60031056e-01
3.03748369e-01 -6.45326748e-02 3.19903374e-01 -2.21580720e+00
-3.84627759e-01 -9.47079882e-02 -2.01969873e-02 -1.02750435e-01
1.69077456e-01 5.34815133e-01 4.02850956e-01 -5.45332253e-01
4.92519349e-01 9.65688109e-01 6.27117455e-01 7.65528738e-01
-7.50787199e-01 -8.22355211e-01 2.24951953e-01 6.09513938e-01
-1.23741555e+00 -7.29781687e-01 3.80234540e-01 -7.30208874e-01
1.03120983e+00 9.79507864e-02 9.12741005e-01 9.04456913e-01
5.14457822e-01 1.02959406e+00 7.82826960e-01 -1.92444205e-01
5.17482638e-01 7.94759035e-01 5.12822986e-01 3.36193979e-01
4.47846919e-01 5.42345941e-01 -3.43297631e-01 2.52302408e-01
-1.14640228e-01 -3.50862890e-01 2.92885393e-01 -4.75985050e-01
-8.44832242e-01 7.55657434e-01 -6.97675049e-02 -3.83792892e-02
-2.97092915e-01 1.39422879e-01 5.71255147e-01 1.13483272e-01
-9.33707133e-02 -6.66763335e-02 -1.20968595e-01 -5.34988821e-01
-2.34135821e-01 5.23107052e-01 7.43616939e-01 1.18815601e+00
8.24246585e-01 2.37327501e-01 1.16656549e-01 6.08473957e-01
6.25943542e-01 9.86772180e-01 -7.54258931e-02 -1.33038247e+00
8.29985589e-02 4.19455349e-01 3.87996197e-01 -9.42767262e-01
-6.27190769e-01 7.47350045e-03 -3.68128419e-01 7.32304871e-01
-3.60723972e-01 -5.77277064e-01 -5.85332215e-01 1.41169298e+00
1.80559695e-01 -4.93224740e-01 1.73212424e-01 6.30301774e-01
8.60920846e-01 5.91878533e-01 1.52827114e-01 -5.21087162e-02
1.25631738e+00 -8.82622600e-01 -9.00415599e-01 -4.85838145e-01
3.88305932e-01 -3.81288797e-01 7.88369596e-01 1.24489002e-01
-8.99183571e-01 -8.19867194e-01 -1.57069385e+00 3.16450261e-02
-6.60193861e-01 -2.25768492e-01 4.62118536e-01 9.28639233e-01
-1.34900546e+00 -3.99704091e-02 -8.15242708e-01 -7.68839836e-01
1.78003192e-01 2.38153800e-01 -4.39523943e-02 -1.28850982e-01
-1.31409609e+00 1.47511184e+00 -1.78864494e-01 -2.66652584e-01
-6.76503479e-01 -5.20452917e-01 -1.00424409e+00 -5.88261127e-01
3.24538112e-01 -5.67351222e-01 1.53373265e+00 -3.14683579e-02
-1.52626014e+00 7.08264351e-01 -3.69931102e-01 -4.82310891e-01
2.24041343e-01 2.38918653e-03 -8.85079563e-01 -4.47076485e-02
5.10922313e-01 9.46095943e-01 4.94111851e-02 -1.30791652e+00
-1.37537205e+00 -4.21518743e-01 -1.16885386e-01 1.36378169e-01
2.20161855e-01 1.29387006e-01 -2.88151085e-01 4.40200239e-01
-3.43690276e-01 -9.84573781e-01 -2.28992075e-01 1.82656962e-02
6.07835054e-02 -4.09657478e-01 1.31527317e+00 5.34888916e-02
1.11385036e+00 -2.47916770e+00 -6.53549135e-01 3.03247850e-02
1.88539419e-02 5.77721238e-01 2.79122710e-01 9.05432820e-01
5.86083174e-01 -2.20746174e-01 2.35728219e-01 8.54955614e-03
5.25955319e-01 5.89650929e-01 -3.09874713e-01 3.51614416e-01
9.57866013e-03 7.77143538e-01 -8.24761271e-01 -9.06931907e-02
1.06933546e+00 3.03008020e-01 -4.30441834e-02 -1.04816288e-01
1.12621814e-01 3.17380935e-01 -3.51193368e-01 5.34553587e-01
7.14628816e-01 3.42220545e-01 -1.48413152e-01 8.22665170e-02
-1.04856551e+00 9.91740450e-02 -8.15011561e-01 8.82897794e-01
-1.78675041e-01 1.10577106e+00 3.26975971e-01 -8.65514755e-01
9.80787992e-01 2.58999169e-01 4.37204599e-01 -1.37054157e+00
1.56493396e-01 4.89070833e-01 -3.06986839e-01 -9.35451806e-01
7.37319231e-01 2.37695247e-01 -1.57379523e-01 -5.99554516e-02
-5.61548591e-01 -3.54490370e-01 3.36256295e-01 5.04514500e-02
9.30606067e-01 2.40259040e-02 7.52409101e-02 -1.90820411e-01
5.80420375e-01 3.69048685e-01 4.08742309e-01 6.63482010e-01
-9.51573491e-01 -2.39212513e-01 3.48685384e-01 -4.98894066e-01
-8.28690112e-01 -1.02232683e+00 -3.74150068e-01 8.80205274e-01
8.42907012e-01 -3.13820362e-01 -7.16582060e-01 -1.77333966e-01
4.32281762e-01 1.14105093e+00 -4.11336005e-01 -1.03400253e-01
-1.97476372e-01 -2.75142670e-01 5.56697547e-01 7.83822656e-01
8.65919352e-01 -7.56782830e-01 -8.99953544e-01 8.32333043e-02
-1.62059516e-01 -1.41282570e+00 2.65769601e-01 -1.19626209e-01
-2.55928338e-01 -1.00723135e+00 8.09893385e-03 -8.45539212e-01
1.17292926e-01 7.62163639e-01 5.41564703e-01 -1.25515729e-01
-7.18337893e-02 7.43633330e-01 -1.41906634e-01 -9.78655040e-01
-4.89827216e-01 -2.73368627e-01 2.99667478e-01 -3.66941512e-01
7.66817749e-01 -3.28051686e-01 -6.35167003e-01 7.06510067e-01
-2.97401041e-01 -5.97596318e-02 5.14309287e-01 2.44501710e-01
2.60195792e-01 -7.36571699e-02 6.42063737e-01 -3.61336738e-01
2.93641299e-01 -7.02145934e-01 -7.05581725e-01 -2.27576464e-01
-1.20096231e+00 -5.01383781e-01 2.55844533e-01 4.99707013e-01
-7.11837769e-01 -9.39473435e-02 -3.74478728e-01 3.50254983e-01
-7.09153593e-01 -1.30230337e-02 -1.75768524e-01 -2.64716089e-01
5.85360944e-01 7.48166293e-02 6.92246556e-01 -1.36946393e-02
2.40196332e-01 1.25884175e+00 6.28688753e-01 -2.06355259e-01
5.59106350e-01 5.42479396e-01 2.31473148e-02 -1.07035697e+00
-1.93054199e-01 -6.01910830e-01 -1.98734611e-01 -7.49604821e-01
8.30256164e-01 -7.87582934e-01 -1.17934763e+00 5.85158527e-01
-8.76431048e-01 -4.56722647e-01 -2.08895355e-01 6.77621722e-01
-6.14754975e-01 9.81277749e-02 -2.75718808e-01 -1.09770679e+00
1.13261454e-01 -1.58210182e+00 1.01030338e+00 3.67569685e-01
-2.47755393e-01 -7.47785807e-01 6.25126157e-03 6.56007469e-01
6.52673483e-01 3.01951110e-01 5.06016195e-01 -6.58416823e-02
-5.19509614e-01 -4.50977027e-01 -3.05746347e-01 3.26941043e-01
-2.92401344e-01 3.05445522e-01 -1.06577086e+00 4.79289740e-02
-3.27376157e-01 -3.52739356e-02 2.70015001e-01 3.68703783e-01
5.21788359e-01 7.90104046e-02 -7.39277422e-01 1.14506662e-01
1.14604485e+00 1.10362256e+00 6.69867992e-01 7.52263546e-01
1.15886219e-01 8.34129155e-01 1.17424965e+00 3.11421514e-01
1.15476573e+00 6.17297173e-01 6.47273242e-01 -4.14307788e-02
-1.60449475e-01 -2.17980221e-01 3.84423792e-01 5.44197142e-01
3.91581170e-02 -1.35739595e-01 -9.78925586e-01 7.15292931e-01
-1.71399927e+00 -8.25629592e-01 -4.74491358e-01 1.93963480e+00
-6.35411143e-02 1.44729048e-01 3.40713948e-01 2.22171843e-01
7.08998203e-01 -8.80545750e-02 -6.92853808e-01 -7.32884884e-01
2.84682572e-01 -3.75693440e-01 7.15440273e-01 5.19956827e-01
-8.02305043e-01 6.94501758e-01 7.76255465e+00 3.88419360e-01
-1.11532605e+00 -1.16474576e-01 3.48731399e-01 3.46677363e-01
-2.82024801e-01 5.40856346e-02 -1.08982635e+00 4.71227735e-01
1.25761747e+00 -2.46737584e-01 1.34680480e-01 1.05142355e+00
7.67597139e-01 -4.88690108e-01 -8.68904889e-01 6.43413126e-01
-9.85776484e-02 -1.28378689e+00 -4.55602139e-01 1.81329280e-01
5.03564596e-01 4.83652085e-01 1.71466768e-01 5.86930275e-01
2.00838760e-01 -7.21302927e-01 1.13885498e+00 1.96248457e-01
7.56822824e-01 -8.50558579e-01 8.89790416e-01 1.31880194e-01
-1.06008101e+00 -3.53090674e-01 -6.98935017e-02 -5.52293718e-01
3.63217860e-01 1.94411844e-01 -5.63497841e-01 3.76953065e-01
8.37366045e-01 4.28758204e-01 -8.73064250e-02 8.74079525e-01
6.38495535e-02 2.75664747e-01 -5.83970584e-02 -5.43379903e-01
3.39527220e-01 -7.13820755e-02 4.97505248e-01 1.20383108e+00
-2.14285463e-01 1.10958628e-01 1.96966350e-01 5.16048253e-01
6.99872136e-01 -2.70755380e-01 -9.46762323e-01 1.01570385e-02
7.31345534e-01 1.09307206e+00 -1.77236065e-01 -3.65268409e-01
-7.51811445e-01 7.67598897e-02 -3.32088411e-01 3.00523549e-01
-8.90967429e-01 -7.78653860e-01 1.11353970e+00 5.19168079e-01
-1.22936472e-01 -6.03909135e-01 -8.10203969e-01 -1.56039402e-01
-6.96145520e-02 -4.33623284e-01 -1.74131989e-01 -7.68150449e-01
-3.60946536e-01 3.22801411e-01 4.13190067e-01 -1.21682405e+00
-3.08632493e-01 -5.91507614e-01 -3.90603244e-01 3.45803648e-01
-1.68439972e+00 -1.00272882e+00 -8.12493920e-01 1.08330362e-01
4.40815270e-01 -2.66434193e-01 5.05156100e-01 5.28693914e-01
-6.48665607e-01 4.40470040e-01 1.84913412e-01 -5.52088022e-01
4.44811583e-01 -5.83575130e-01 6.74649954e-01 4.71423686e-01
-1.03347683e+00 2.36720696e-01 1.00676548e+00 -4.26282316e-01
-1.84478223e+00 -1.03203487e+00 8.42695117e-01 -4.17985111e-01
5.53254843e-01 -2.25410357e-01 -8.99881050e-02 6.62196636e-01
3.12765002e-01 -4.51160878e-01 3.84776324e-01 -1.27007902e-01
1.74124300e-01 -6.78683281e-01 -1.24183786e+00 8.51302803e-01
7.87744761e-01 -2.35838354e-01 -5.82719855e-02 7.59314448e-02
6.47525609e-01 -3.03656459e-01 -5.14530420e-01 5.59199691e-01
8.99152994e-01 -1.13603115e+00 3.96507323e-01 3.99480201e-02
-9.87182856e-02 -5.15669346e-01 -3.30865353e-01 -8.31277907e-01
-2.76969522e-01 -7.37178981e-01 4.96699810e-01 8.61239433e-01
3.92219007e-01 -1.17407274e+00 5.22742152e-01 9.27856147e-01
-1.05399191e+00 -6.50482953e-01 -6.49821877e-01 -7.48556435e-01
-1.40078306e-01 -1.03596652e+00 5.52016973e-01 4.19374198e-01
4.63641763e-01 2.16089770e-01 -1.27189569e-02 9.07722190e-02
6.34836316e-01 -4.43932503e-01 1.25409210e+00 -9.85798001e-01
4.68004256e-01 -3.60201269e-01 -1.08567607e+00 -1.04144728e+00
-1.81983843e-01 -2.77555555e-01 3.01222324e-01 -1.93452179e+00
-1.94310278e-01 -3.82481009e-01 3.27328801e-01 3.82848769e-01
4.23360586e-01 2.84867376e-01 -3.23830545e-02 -9.23640877e-02
-5.75794697e-01 3.47283751e-01 9.25202310e-01 -3.73042864e-03
-1.44599497e-01 2.19954595e-01 -8.96008492e-01 6.96482360e-01
1.14763045e+00 3.36807311e-01 -7.10406959e-01 -1.94741830e-01
-4.08588424e-02 -1.03807487e-02 5.50596178e-01 -1.28193712e+00
6.65921867e-01 -5.16332567e-01 -1.86095819e-01 -1.03606033e+00
3.63107383e-01 -8.07632327e-01 1.92876652e-01 7.26865649e-01
2.74022520e-01 -1.11867301e-02 5.05239844e-01 3.50352854e-01
-1.97999477e-01 2.10418552e-01 8.21699619e-01 1.76608279e-01
-1.40425265e+00 1.51823303e-02 -1.31000972e+00 -4.03837979e-01
1.60289454e+00 -7.69715667e-01 -6.68481171e-01 -3.88074249e-01
-4.55661923e-01 7.81435966e-01 4.11016256e-01 8.37658763e-01
4.88776743e-01 -1.15812206e+00 -3.98717016e-01 4.30542529e-01
4.55627203e-01 -3.03744346e-01 3.76309842e-01 9.54630971e-01
-5.44952214e-01 9.56198752e-01 -4.52005416e-01 -6.92064464e-01
-9.69957948e-01 5.20476639e-01 1.38039455e-01 7.19216824e-01
-4.35995638e-01 1.59471467e-01 1.64221749e-01 -3.36666793e-01
4.13908422e-01 4.83650565e-02 -3.20284277e-01 -3.76257479e-01
5.27556777e-01 9.68063235e-01 2.76939943e-02 -8.86224091e-01
-5.32934666e-01 5.36146879e-01 6.81776181e-02 -3.19837004e-01
7.02875137e-01 -9.22563434e-01 1.07781149e-01 6.73448980e-01
9.03257310e-01 -2.17882946e-01 -1.08452022e+00 4.48767841e-01
-2.86752969e-01 -2.21177429e-01 -7.17224851e-02 -6.45390451e-01
-6.76224530e-01 6.75041676e-01 7.92941988e-01 2.35638902e-01
8.52141201e-01 -2.98062321e-02 9.64599490e-01 5.95061630e-02
6.36953294e-01 -1.49327469e+00 -2.93476373e-01 7.48395562e-01
6.23508573e-01 -1.17559934e+00 -4.15399611e-01 -4.08893794e-01
-8.69450748e-01 7.97468960e-01 8.36402595e-01 2.78395921e-01
9.78967309e-01 4.94271100e-01 3.17925423e-01 -2.31905058e-01
-7.44235694e-01 -3.16093326e-01 -2.59772301e-01 1.31503832e+00
1.48077816e-01 2.68368363e-01 -4.83421981e-01 3.20925921e-01
-4.06620055e-01 2.14027464e-01 4.18613166e-01 1.04601741e+00
-1.05705583e+00 -9.41535473e-01 -3.25119793e-01 1.69149026e-01
1.19390197e-01 6.62733614e-01 -2.55028516e-01 9.93956208e-01
5.36071122e-01 1.73388624e+00 1.85541436e-01 -1.00382769e+00
8.76723051e-01 -2.04378828e-01 -6.25614077e-02 6.77705631e-02
-1.16228268e-01 -6.29733860e-01 5.60060084e-01 -7.54067421e-01
-2.60557115e-01 -8.13619494e-01 -1.38425350e+00 -1.02081692e+00
5.87691441e-02 1.10110827e-01 1.35225105e+00 8.93616021e-01
6.58121347e-01 5.55133522e-01 8.74787927e-01 -6.05003893e-01
-1.23924352e-01 -2.97380358e-01 -5.87456405e-01 -3.65053624e-01
3.98790091e-01 -7.54696131e-01 -9.36668441e-02 -3.17570418e-01] | [5.690428733825684, 1.0457582473754883] |
2197e1c1-6164-4c2a-bdf3-b92e624eca08 | amuse-multilingual-semantic-parsing-for | 1802.09296 | null | http://arxiv.org/abs/1802.09296v1 | http://arxiv.org/pdf/1802.09296v1.pdf | AMUSE: Multilingual Semantic Parsing for Question Answering over Linked Data | The task of answering natural language questions over RDF data has received
wide interest in recent years, in particular in the context of the series of
QALD benchmarks. The task consists of mapping a natural language question to an
executable form, e.g. SPARQL, so that answers from a given KB can be extracted.
So far, most systems proposed are i) monolingual and ii) rely on a set of
hard-coded rules to interpret questions and map them into a SPARQL query. We
present the first multilingual QALD pipeline that induces a model from training
data for mapping a natural language question into logical form as probabilistic
inference. In particular, our approach learns to map universal syntactic
dependency representations to a language-independent logical form based on
DUDES (Dependency-based Underspecified Discourse Representation Structures)
that are then mapped to a SPARQL query as a deterministic second step. Our
model builds on factor graphs that rely on features extracted from the
dependency graph and corresponding semantic representations. We rely on
approximate inference techniques, Markov Chain Monte Carlo methods in
particular, as well as Sample Rank to update parameters using a ranking
objective. Our focus lies on developing methods that overcome the lexical gap
and present a novel combination of machine translation and word embedding
approaches for this purpose. As a proof of concept for our approach, we
evaluate our approach on the QALD-6 datasets for English, German & Spanish. | ['Soufian Jebbara', 'Sherzod Hakimov', 'Philipp Cimiano'] | 2018-02-26 | null | null | null | null | ['knowledge-base-question-answering'] | ['natural-language-processing'] | [-1.02431573e-01 6.86746895e-01 -2.77220964e-01 -6.84356451e-01
-1.14731336e+00 -7.64868498e-01 1.05596375e+00 6.50437236e-01
-5.18376470e-01 7.68575191e-01 6.82374477e-01 -4.49323386e-01
-4.05074507e-01 -1.32332253e+00 -1.12019479e+00 6.83844686e-02
2.79496193e-01 1.13911295e+00 4.83421534e-01 -6.25736058e-01
-9.86530259e-02 2.05619782e-01 -1.45402241e+00 6.55040979e-01
9.92005169e-01 7.79152572e-01 3.31284329e-02 4.43582207e-01
-8.43975723e-01 1.15621006e+00 -2.99255908e-01 -7.86643326e-01
-1.58488795e-01 -3.42274010e-01 -1.50643730e+00 -7.56092608e-01
4.55833703e-01 2.02484708e-02 -5.77781014e-02 9.11918759e-01
1.63999826e-01 1.45303041e-01 7.44228780e-01 -9.14032876e-01
-6.82026148e-01 8.60939562e-01 4.01833296e-01 -1.43971378e-02
9.44055915e-01 -2.52026141e-01 1.62445700e+00 -8.85905087e-01
9.90107536e-01 1.62041104e+00 4.43144649e-01 5.48882067e-01
-1.34705937e+00 1.08950421e-01 -1.59387618e-01 7.71227896e-01
-1.03854144e+00 -1.69567168e-01 3.66874337e-01 -3.16548169e-01
1.47510970e+00 2.88242787e-01 1.67835698e-01 8.49935830e-01
1.36869460e-01 5.34217417e-01 1.17969239e+00 -9.35231686e-01
5.43379784e-01 4.13628876e-01 4.70801860e-01 8.09375644e-01
2.96598911e-01 -2.46177346e-01 -4.88316834e-01 -2.75422603e-01
-8.78874660e-02 -5.64295650e-01 -7.70130306e-02 -4.73797649e-01
-8.61799836e-01 1.18955755e+00 4.45747226e-01 3.88788164e-01
-3.04247260e-01 1.61279738e-01 3.60965461e-01 4.11265612e-01
2.48227686e-01 4.39670891e-01 -7.76514232e-01 2.79666305e-01
-5.43897569e-01 6.16364241e-01 1.47564471e+00 7.48117983e-01
1.02383232e+00 -8.20488751e-01 -2.97929764e-01 5.31761408e-01
6.72822654e-01 5.00311196e-01 2.99135059e-01 -8.75880480e-01
6.47748411e-01 8.83957148e-01 1.34803683e-01 -6.68228865e-01
-2.77387142e-01 2.77341425e-01 1.51733622e-01 -1.30011402e-02
4.69485044e-01 2.68314838e-01 -3.78751755e-01 1.80524445e+00
7.28268862e-01 -2.11330742e-01 5.22815287e-01 6.63387001e-01
8.65513265e-01 7.09419489e-01 3.04825544e-01 2.65196949e-01
1.89916420e+00 -5.71778536e-01 -6.33496940e-01 -1.33344680e-01
8.73195708e-01 -4.89361256e-01 1.35812879e+00 8.38265717e-02
-8.05745900e-01 -2.52784520e-01 -9.84301329e-01 -5.99682271e-01
-6.88607991e-01 -5.85587770e-02 2.96426654e-01 7.41152763e-01
-9.08483684e-01 4.08575207e-01 -7.44447410e-01 -7.77246714e-01
3.51073556e-02 2.30956823e-02 -4.42390889e-01 -4.93414462e-01
-1.63339543e+00 1.37021804e+00 7.39746809e-01 -3.46132100e-01
-5.94229817e-01 -8.01870525e-01 -1.13500297e+00 7.68312886e-02
5.19549549e-01 -1.03896976e+00 1.15282226e+00 -5.83345592e-01
-1.49025249e+00 8.73109698e-01 -2.67708898e-01 -7.58643210e-01
7.70254508e-02 -2.47228980e-01 -4.14373308e-01 1.97303876e-01
2.69551367e-01 3.98499966e-01 4.82755721e-01 -9.15916145e-01
-6.12640262e-01 -5.72848320e-01 7.99029827e-01 -3.23757939e-02
-1.20872058e-01 2.98067212e-01 -2.92745709e-01 -1.07512116e-01
-3.41333538e-01 -7.40707517e-01 1.08563647e-01 -2.84568578e-01
-1.51998401e-01 -8.89760971e-01 3.32413644e-01 -9.19538140e-01
1.13571203e+00 -1.53184450e+00 4.49183255e-01 1.74996972e-01
-5.52682057e-02 -2.14815531e-02 3.99513775e-03 7.18763649e-01
1.06547765e-01 1.82212040e-01 -3.63705546e-01 9.59301442e-02
6.55206084e-01 5.77988327e-01 -5.72517514e-01 9.86683741e-02
5.41054308e-01 1.02495766e+00 -9.09707725e-01 -4.94866788e-01
5.42203784e-02 3.21803004e-01 -9.35193539e-01 3.54673892e-01
-1.04609489e+00 -4.80966680e-02 -4.57871675e-01 1.94881082e-01
5.13596117e-01 -1.70959875e-01 6.03786469e-01 -6.86023355e-01
1.23821586e-01 9.69460428e-01 -1.02466452e+00 1.96515334e+00
-9.00451362e-01 1.25929415e-01 -5.32057405e-01 -9.06185925e-01
7.27743387e-01 2.44315341e-01 -2.77995765e-02 -6.88075900e-01
-1.30847588e-01 3.74282897e-01 -5.83481789e-01 -8.63044739e-01
4.40483391e-01 -2.48314232e-01 -3.45504552e-01 5.27154565e-01
5.72745144e-01 -1.33259490e-01 5.92716813e-01 4.27378416e-01
1.13452470e+00 6.35608375e-01 4.22537804e-01 -4.34861153e-01
8.65981936e-01 3.20936859e-01 1.28682539e-01 4.81651902e-01
4.72141325e-01 1.02300525e-01 7.21601605e-01 -3.67841661e-01
-9.36395347e-01 -1.31038666e+00 -4.87530380e-02 1.03135669e+00
-2.58508086e-01 -6.97065771e-01 -8.74790728e-01 -1.15295875e+00
1.49443537e-01 1.43623281e+00 -6.61957860e-01 -4.39151414e-02
-6.68280780e-01 -3.47305447e-01 5.86488366e-01 2.72218913e-01
-4.15082946e-02 -1.01227605e+00 -5.35446465e-01 2.69011080e-01
-2.11340874e-01 -1.39615595e+00 7.89490938e-02 9.61143523e-02
-6.21239126e-01 -1.37915003e+00 -2.29235203e-03 -4.70550805e-01
4.21767473e-01 -6.89243078e-01 1.56442523e+00 -3.11952651e-01
-8.59266818e-02 6.93585277e-01 -4.55670327e-01 -2.15425491e-01
-8.18168521e-01 1.89815342e-01 -1.77643597e-02 -3.26681733e-02
8.04205656e-01 -2.27256611e-01 -2.59738445e-01 -1.55016035e-01
-1.37769961e+00 -3.34215045e-01 2.72169501e-01 5.13629854e-01
6.10750020e-01 -3.82692873e-01 4.08633381e-01 -1.16097724e+00
7.24766076e-01 -7.80004978e-01 -8.07490945e-01 8.08062196e-01
-7.05695152e-01 9.66396868e-01 7.46038735e-01 1.44695997e-01
-1.26994407e+00 -1.51091456e-01 -3.42328876e-01 5.47904484e-02
-5.31687699e-02 1.05979514e+00 -4.90751743e-01 2.53838301e-01
1.03269684e+00 -2.12905943e-01 -1.79552138e-01 -6.27064705e-01
1.02921164e+00 4.59762096e-01 3.59500587e-01 -1.16916490e+00
8.01035106e-01 4.40008521e-01 -2.58993842e-02 -4.75635052e-01
-1.05251217e+00 -2.66926289e-01 -4.72319573e-01 1.54337615e-01
1.09984124e+00 -7.25964904e-01 -6.59837663e-01 -4.76219833e-01
-1.31643367e+00 -6.46035895e-02 -6.43685520e-01 5.42817712e-01
-5.42595565e-01 4.23299789e-01 -3.70871633e-01 -4.22585577e-01
-2.93697208e-01 -8.04422200e-01 1.01803350e+00 3.67787480e-02
-2.93325245e-01 -1.25632429e+00 7.01276302e-01 4.72532272e-01
3.24812710e-01 -7.10027069e-02 1.67580092e+00 -9.53130424e-01
-6.24762952e-01 -5.17834611e-02 -2.71942794e-01 4.72092152e-01
-6.73746243e-02 -2.84811080e-01 -8.21895301e-01 2.72971727e-02
-1.18818089e-01 -5.60555577e-01 5.60703933e-01 -1.25886232e-01
2.65376240e-01 -4.86621082e-01 -3.03979591e-03 2.05272749e-01
1.83334064e+00 -5.25334299e-01 5.47248602e-01 4.36837882e-01
3.64997596e-01 8.38202477e-01 4.49422091e-01 -2.85646934e-02
1.05928862e+00 7.51537085e-01 3.16280514e-01 6.16301358e-01
-2.54423082e-01 -6.90789461e-01 5.27513504e-01 8.04513872e-01
2.42798701e-01 -9.59092379e-02 -1.09669864e+00 5.67484081e-01
-1.68627143e+00 -6.61561012e-01 -1.89165905e-01 2.23921466e+00
1.06808889e+00 -2.88300931e-01 -2.21642107e-01 -3.06757569e-01
2.55616993e-01 -3.78091075e-02 -1.58767819e-01 -4.79063630e-01
-2.58644614e-02 8.16605270e-01 2.57139087e-01 8.54729354e-01
-6.80005848e-01 9.58887577e-01 5.05569887e+00 5.36707580e-01
-6.38321102e-01 3.90062869e-01 -6.74428269e-02 2.35888869e-01
-1.01121545e+00 6.17732882e-01 -1.01968646e+00 1.70278981e-01
1.39787161e+00 -2.93428689e-01 5.05407035e-01 5.95138967e-01
-3.20426285e-01 -5.85331172e-02 -1.36683440e+00 4.53040808e-01
2.13192537e-01 -1.52085793e+00 5.25293946e-01 -3.76948386e-01
2.76843697e-01 1.04997240e-01 -4.92220193e-01 6.55754566e-01
7.41662383e-01 -1.02178609e+00 7.82690644e-01 8.45242321e-01
7.50211298e-01 -5.63395739e-01 6.90518737e-01 2.41959646e-01
-1.00853026e+00 6.92397356e-02 -5.34700990e-01 2.11396381e-01
2.29436412e-01 5.49848616e-01 -8.67353261e-01 1.22640908e+00
6.63814366e-01 2.28473529e-01 -5.56539178e-01 3.14439625e-01
-7.43730605e-01 5.37051380e-01 -4.57442373e-01 -2.23761320e-01
1.54405218e-02 -1.23252109e-01 3.25916588e-01 1.12891185e+00
3.87684524e-01 -1.15060493e-01 -1.63111985e-01 1.12085724e+00
-2.17600793e-01 5.84752679e-01 -4.32180732e-01 -3.96292619e-02
2.20284626e-01 1.09240866e+00 -5.59982844e-02 -3.26995343e-01
-6.67543650e-01 6.49811983e-01 7.78001130e-01 2.50852793e-01
-6.37404084e-01 -2.98963398e-01 3.47652316e-01 7.63634443e-02
3.28188926e-01 -5.06599993e-03 3.46719354e-01 -1.49157214e+00
2.98782408e-01 -1.00641632e+00 8.05795193e-01 -7.96912968e-01
-1.33544481e+00 6.41163945e-01 2.21310019e-01 -5.21220267e-01
-7.58367896e-01 -6.72603607e-01 -1.75497219e-01 1.05701745e+00
-1.88795257e+00 -1.19936073e+00 -5.30636907e-02 5.86673319e-01
6.10133186e-02 -6.90243319e-02 1.20196581e+00 3.11719865e-01
-2.00607747e-01 2.94604212e-01 -2.09860533e-01 -6.30953982e-02
7.37644196e-01 -1.48085678e+00 3.21348637e-01 6.37462616e-01
5.39906263e-01 7.19942153e-01 8.61541986e-01 -5.17044485e-01
-1.66308177e+00 -1.03247666e+00 1.68543303e+00 -1.00880015e+00
9.96755004e-01 -3.46256495e-01 -1.07622981e+00 8.00610363e-01
4.10523772e-01 1.19181275e-01 5.74641705e-01 2.31381819e-01
-7.32516825e-01 -2.04069689e-01 -1.11730409e+00 2.80673593e-01
6.46869779e-01 -9.56342340e-01 -1.28361678e+00 3.82952511e-01
8.21744084e-01 -2.59798646e-01 -1.16935694e+00 3.17845374e-01
2.67679423e-01 -7.89686024e-01 8.22774053e-01 -1.21218157e+00
3.81678820e-01 -4.70046401e-01 -6.11938238e-01 -1.09867489e+00
1.64943993e-01 -3.06267262e-01 -2.64012367e-01 1.18425977e+00
7.05060959e-01 -6.51172280e-01 5.53622782e-01 6.46360755e-01
2.83948421e-01 -5.99830270e-01 -1.22416186e+00 -6.49320960e-01
2.44246811e-01 -6.28408253e-01 8.51459146e-01 6.41335607e-01
3.73024121e-02 7.41340339e-01 2.81992048e-01 3.71803075e-01
4.86472696e-01 2.33235389e-01 7.48683512e-01 -1.24080682e+00
-4.31554377e-01 4.71568964e-02 -3.19904476e-01 -8.02423835e-01
5.78651845e-01 -1.45277786e+00 -2.44155392e-01 -1.92726314e+00
-4.74648401e-02 -3.75956476e-01 -1.16455354e-01 3.10209394e-01
-3.66493501e-02 -3.56127590e-01 1.65819954e-02 -1.53050303e-01
-5.21083295e-01 5.25516570e-01 4.06155735e-01 -2.15383098e-01
3.55268598e-01 -4.32420254e-01 -5.72168589e-01 3.93252015e-01
5.07768810e-01 -7.75895357e-01 -4.49766487e-01 -4.85845685e-01
1.20034766e+00 1.07246168e-01 5.49679399e-01 -5.85905254e-01
2.87344962e-01 5.98825254e-02 -4.11440760e-01 8.22255109e-03
5.19544398e-03 -8.40454400e-01 -6.61980361e-02 3.10291141e-01
-4.62396145e-01 -3.26248296e-02 1.37298070e-02 5.78041494e-01
-4.09208626e-01 -7.38664985e-01 3.48731846e-01 4.03946079e-02
-7.01417208e-01 -2.10418291e-02 6.28162697e-02 6.62957013e-01
6.97705507e-01 4.79181886e-01 -3.10576677e-01 -1.34471849e-01
-6.84097588e-01 8.05327669e-02 3.34074348e-01 4.30438399e-01
2.71485686e-01 -1.29828429e+00 -9.20607626e-01 -4.10763063e-02
4.83382314e-01 -2.81941563e-01 8.27182829e-02 5.89014471e-01
-5.92975199e-01 8.00750196e-01 9.25521329e-02 -2.50619352e-01
-7.82749295e-01 5.70719600e-01 3.35561872e-01 -8.11379671e-01
-1.78510338e-01 5.46089411e-01 -2.25332066e-01 -1.03003109e+00
-2.71525234e-01 -6.28875077e-01 -4.85047817e-01 1.15086243e-01
3.81353050e-01 1.36879027e-01 4.12007481e-01 -4.20757353e-01
-5.40582478e-01 3.45629513e-01 1.56875849e-01 -3.40278178e-01
1.26343298e+00 8.81481692e-02 -5.59516668e-01 3.04065257e-01
1.21409249e+00 2.88899153e-01 -5.84632218e-01 -6.67937279e-01
6.79233134e-01 -1.28121689e-01 -2.57600635e-01 -7.79031277e-01
-3.67518574e-01 6.58053935e-01 4.03815746e-01 1.95468202e-01
6.67877436e-01 6.49630189e-01 6.09662950e-01 8.00139844e-01
4.67576176e-01 -9.85398889e-01 -2.66149342e-01 7.55696535e-01
9.03855026e-01 -1.01300418e+00 -1.48949936e-01 -2.11580142e-01
-2.70682871e-01 1.26555479e+00 2.04527229e-02 -3.79693694e-02
4.62788731e-01 3.83897834e-02 -2.98034633e-03 -4.40533787e-01
-9.08690810e-01 -3.68723571e-01 5.04976451e-01 5.18964350e-01
4.61948067e-01 -2.43002828e-02 -4.31419551e-01 8.19469333e-01
-1.82412222e-01 2.85839409e-01 2.61489987e-01 7.62572944e-01
-3.51550192e-01 -1.69473457e+00 -1.18968025e-01 3.07794929e-01
-3.77754927e-01 -3.03435564e-01 -1.51279092e-01 7.18601227e-01
-1.04906857e-01 8.00391734e-01 -2.97898173e-01 -5.60023785e-02
6.52202725e-01 5.26374459e-01 7.61885643e-01 -8.72579277e-01
-5.32046378e-01 -9.53713536e-01 5.60978353e-01 -7.39468217e-01
-4.41645801e-01 -6.37809336e-01 -1.38771534e+00 -7.31562600e-02
-2.33670920e-02 4.44721371e-01 8.57426345e-01 1.25083113e+00
4.47295606e-01 2.49756247e-01 1.04430951e-01 8.19798708e-02
-8.76691997e-01 -7.15678751e-01 -1.26291275e-01 5.31422317e-01
-1.91722378e-01 -3.19165975e-01 -1.76080868e-01 9.10912305e-02] | [10.2982816696167, 7.896028995513916] |
269939d9-615f-4174-b449-d39b8cf8e9f0 | a-discourse-aware-graph-neural-network-for | null | null | https://aclanthology.org/2021.findings-emnlp.252 | https://aclanthology.org/2021.findings-emnlp.252.pdf | A Discourse-Aware Graph Neural Network for Emotion Recognition in Multi-Party Conversation | Emotion recognition in multi-party conversation (ERMC) is becoming increasingly popular as an emerging research topic in natural language processing. Prior research focuses on exploring sequential information but ignores the discourse structures of conversations. In this paper, we investigate the importance of discourse structures in handling informative contextual cues and speaker-specific features for ERMC. To this end, we propose a discourse-aware graph neural network (ERMC-DisGCN) for ERMC. In particular, we design a relational convolution to lever the self-speaker dependency of interlocutors to propagate contextual information. Furthermore, we exploit a gated convolution to select more informative cues for ERMC from dependent utterances. The experimental results show our method outperforms multiple baselines, illustrating that discourse structures are of great value to ERMC. | ['Guohong Fu', 'Nan Yu', 'Yang Sun'] | null | null | null | null | findings-emnlp-2021-11 | ['emotion-recognition-in-conversation'] | ['natural-language-processing'] | [ 1.43472642e-01 2.78324157e-01 -1.40620759e-02 -6.94632947e-01
-4.46294338e-01 -5.47589302e-01 7.12923169e-01 2.27083847e-01
-2.84271091e-01 4.30247962e-01 1.03297246e+00 -2.55317301e-01
2.08662316e-01 -3.90558779e-01 -2.23377302e-01 -3.24868888e-01
-3.25835228e-01 -2.51205694e-02 -2.77531713e-01 -6.43921018e-01
2.74764150e-01 2.20308170e-01 -9.88739371e-01 6.92705691e-01
8.90929103e-01 7.85370290e-01 2.67663691e-03 7.83850491e-01
-3.94891322e-01 1.21508098e+00 -8.67748916e-01 -4.02301490e-01
-4.12852019e-01 -7.42092609e-01 -1.10270452e+00 2.99478531e-01
-2.07041010e-01 -3.09881475e-02 -2.48115540e-01 7.79020548e-01
5.24589419e-01 5.64564049e-01 5.41849852e-01 -1.02999902e+00
-5.69945574e-01 1.19926274e+00 -4.69437122e-01 3.34979177e-01
4.11308438e-01 -3.15687582e-02 1.25045502e+00 -7.11268783e-01
6.78586125e-01 1.60020840e+00 4.01498377e-01 6.30936027e-01
-8.02205563e-01 -4.73551422e-01 8.11095834e-01 4.08797115e-01
-8.20861995e-01 -5.99765539e-01 1.48407149e+00 -3.04143041e-01
1.03168404e+00 4.33821440e-01 4.05532449e-01 1.58675516e+00
-1.89374641e-01 1.18612421e+00 8.30029249e-01 -4.25971448e-01
8.81428942e-02 -7.46107567e-03 5.29834330e-01 2.92028457e-01
-4.90063697e-01 -3.39820921e-01 -7.12633073e-01 7.35494681e-03
3.64055455e-01 -2.96008646e-01 -6.15728021e-01 3.31044525e-01
-8.41431022e-01 1.17490160e+00 3.89023006e-01 6.06119573e-01
-3.11494619e-01 6.38754889e-02 9.46068168e-01 5.12066543e-01
9.51746166e-01 5.87810278e-01 -2.24011973e-01 -4.31586802e-01
-2.39500940e-01 -1.64311379e-01 9.95347142e-01 9.95269954e-01
3.53936523e-01 7.72696454e-03 -5.36519468e-01 1.20259273e+00
3.91664088e-01 -7.37525746e-02 3.63479763e-01 -9.86655951e-01
8.18283200e-01 8.08058441e-01 -3.35444570e-01 -1.33711541e+00
-6.51788175e-01 -1.11728981e-01 -8.48133683e-01 -6.04080796e-01
-7.15718567e-02 -5.61494291e-01 -1.75505280e-01 1.75044513e+00
1.99532464e-01 -2.91689318e-02 2.94646442e-01 1.00285804e+00
1.31289744e+00 8.73265803e-01 1.20625384e-01 -3.97170484e-01
1.32382202e+00 -1.23968506e+00 -1.19315398e+00 -3.11981350e-01
5.67950189e-01 -6.46059513e-01 8.43367398e-01 7.17919916e-02
-8.35579336e-01 -2.59347558e-01 -6.81590617e-01 -2.11325303e-01
-1.45968691e-01 -1.56636670e-01 7.58871019e-01 3.30879301e-01
-8.96155596e-01 1.99759007e-01 -5.74116230e-01 -1.86445698e-01
2.32608438e-01 -5.87030035e-03 -4.90716882e-02 5.26411310e-02
-1.53584909e+00 8.50876331e-01 3.04271996e-01 5.25254905e-01
-4.66356039e-01 -2.99969286e-01 -1.19389546e+00 1.43932179e-01
5.40017843e-01 -1.45872265e-01 1.39292765e+00 -1.16372299e+00
-2.04869771e+00 5.41894138e-01 -4.02857810e-01 -3.53143960e-01
2.86371112e-01 -2.28983507e-01 -6.43676400e-01 3.73601824e-01
-2.73170888e-01 3.21818501e-01 6.81011200e-01 -1.25708199e+00
-2.38490745e-01 -1.26320869e-01 2.70732939e-01 5.79465449e-01
-4.45533752e-01 5.19708812e-01 -5.44497609e-01 -6.43819928e-01
-1.10875547e-01 -6.14923656e-01 -1.62186772e-01 -6.69179440e-01
-7.51163244e-01 -7.75519073e-01 8.50512445e-01 -6.04842603e-01
1.47370553e+00 -2.20423841e+00 3.29531819e-01 8.46447200e-02
3.27587932e-01 1.63494915e-01 -3.66522789e-01 8.69683981e-01
1.43363297e-01 2.57080138e-01 -2.63542142e-02 -7.48670399e-01
1.07429072e-01 1.37677059e-01 -1.69089481e-01 3.11576933e-01
4.86336470e-01 1.10458219e+00 -7.78549373e-01 -4.94271040e-01
6.05172738e-02 5.65403998e-01 -4.39178467e-01 4.53279018e-01
-3.59819412e-01 6.35808706e-01 -6.46651089e-01 2.94712484e-01
4.62928951e-01 -4.22543079e-01 6.16886437e-01 1.66159272e-02
-4.42906134e-02 5.75925231e-01 -5.76263011e-01 1.47534454e+00
-5.25851965e-01 9.57798481e-01 4.20443416e-01 -1.04265785e+00
9.96672451e-01 5.42381823e-01 9.82166901e-02 -5.91578424e-01
5.04023015e-01 -2.77899772e-01 8.16735178e-02 -7.54121900e-01
8.16365361e-01 4.27468233e-02 -3.74806523e-01 5.28336942e-01
2.24853065e-02 1.45418987e-01 7.99884796e-02 4.73877758e-01
8.67131770e-01 -3.51352990e-01 4.31174904e-01 -5.95453605e-02
6.32717609e-01 -4.48735595e-01 7.19968081e-01 4.79947895e-01
-5.19308388e-01 5.14723778e-01 8.62425387e-01 -1.16126463e-01
-1.86342254e-01 -2.21476048e-01 3.13657582e-01 1.40012014e+00
1.26161054e-01 -4.88689631e-01 -6.42951310e-01 -7.80419469e-01
-2.72787839e-01 8.15909743e-01 -7.24155188e-01 -1.20459296e-01
-9.95805085e-01 -3.92288893e-01 4.82866675e-01 5.24702430e-01
4.87508804e-01 -1.30140805e+00 -2.14658394e-01 3.62130076e-01
-7.65330255e-01 -1.32739449e+00 -8.33116055e-01 7.97735974e-02
-4.33375567e-01 -1.07345390e+00 -2.58157402e-01 -7.27476001e-01
3.12122524e-01 3.97405982e-01 1.22891068e+00 4.37716544e-02
2.68676043e-01 6.59464240e-01 -8.18914473e-01 -3.05017978e-01
-4.91128594e-01 3.80368710e-01 -4.97960299e-01 2.72311896e-01
4.15801525e-01 -5.62702417e-01 -5.36333740e-01 1.42708853e-01
-5.54562747e-01 9.61275473e-02 9.74965692e-02 6.55930042e-01
-8.97467360e-02 -2.83886284e-01 8.99929047e-01 -1.22852552e+00
1.35710824e+00 -6.31581366e-01 -1.17007889e-01 2.47322127e-01
-1.16544060e-01 -2.51746207e-01 6.89897239e-01 -4.01405096e-01
-1.52918077e+00 -4.13222104e-01 -1.03971004e-01 -1.53429329e-01
-1.97226211e-01 7.97276139e-01 -2.99014121e-01 4.67475504e-01
1.91715091e-01 -2.49410912e-01 -1.68492019e-01 -2.63990909e-01
6.00492358e-01 8.36286843e-01 1.61524177e-01 -7.90843844e-01
8.65252514e-04 1.26758084e-01 -6.29059911e-01 -1.09908891e+00
-1.00758398e+00 -7.44521081e-01 -2.54896909e-01 -6.15973651e-01
1.14608979e+00 -7.87431002e-01 -1.09496260e+00 2.36286432e-01
-1.61784852e+00 -3.04189682e-01 8.73051137e-02 3.98014009e-01
-1.40163332e-01 2.95055121e-01 -9.83876169e-01 -1.14398324e+00
-3.77932936e-01 -1.04189742e+00 9.54278529e-01 3.30008358e-01
-4.39971745e-01 -1.43913591e+00 -6.82102069e-02 5.66465378e-01
4.90453124e-01 4.55968559e-01 7.10069954e-01 -1.05399132e+00
-3.58028203e-01 1.51016116e-01 -2.30146259e-01 2.44208947e-01
7.23451748e-02 -1.19707517e-01 -1.09983110e+00 6.88168481e-02
8.49183649e-02 -4.40596849e-01 1.03079283e+00 3.39578152e-01
1.04743600e+00 -4.48228568e-01 -1.79947540e-01 2.49119699e-01
8.30420136e-01 2.48464122e-01 3.72197092e-01 5.69877140e-02
7.63769329e-01 1.08297622e+00 2.68563747e-01 5.39596915e-01
8.58629346e-01 3.25887144e-01 3.01540345e-01 5.37288487e-02
1.17451340e-01 1.21923350e-02 4.46670830e-01 1.58207440e+00
-7.16583878e-02 -7.39865959e-01 -7.94949055e-01 4.98022109e-01
-2.03808832e+00 -9.56129074e-01 -3.11601311e-01 1.32070363e+00
9.10458565e-01 -9.73334610e-02 -2.30834261e-01 -3.36436331e-01
1.03238869e+00 7.43245661e-01 -3.76773894e-01 -7.02166378e-01
-3.03698719e-01 -1.20937109e-01 -2.32337132e-01 5.97154796e-01
-1.17876065e+00 9.38186169e-01 5.64979362e+00 4.11263227e-01
-1.00288951e+00 -4.90027480e-03 7.38778412e-01 8.47574398e-02
-4.66005683e-01 -2.06833750e-01 -6.56273007e-01 2.99549550e-01
9.04392481e-01 -6.70081973e-02 3.32561076e-01 6.40264273e-01
4.09790814e-01 8.00712779e-02 -1.05543053e+00 7.42476285e-01
4.04141694e-01 -1.26436424e+00 -3.09944451e-01 -1.94995612e-01
6.05928719e-01 -1.44802541e-01 -2.19108060e-01 6.92500353e-01
5.82884133e-01 -9.41051543e-01 2.46213779e-01 3.27841014e-01
1.46795690e-01 -8.01501155e-01 8.08378220e-01 1.86173916e-01
-1.07911038e+00 1.14928745e-01 2.30341535e-02 -2.27625474e-01
4.10600662e-01 4.18984264e-01 -9.03846383e-01 5.89831591e-01
4.42557365e-01 8.47938895e-01 -4.11652952e-01 3.45785201e-01
-5.30897319e-01 1.07793915e+00 1.18811904e-02 -5.02096474e-01
4.34069842e-01 -1.08941562e-01 5.44023335e-01 1.75414348e+00
-2.52714247e-01 3.88105631e-01 3.62089783e-01 7.09640622e-01
-5.39885879e-01 3.23841661e-01 -5.00978172e-01 -1.95034698e-01
6.76630676e-01 1.32833171e+00 -6.25908136e-01 -2.52113827e-02
-4.70378578e-01 1.03186691e+00 8.08831930e-01 6.23331726e-01
-5.86010456e-01 -2.76782185e-01 5.80628216e-01 -6.81913555e-01
3.75795782e-01 -1.33032933e-01 -3.01675014e-02 -1.17068124e+00
3.44245136e-02 -8.43251824e-01 4.41305965e-01 -3.02215368e-01
-1.64999676e+00 6.52142286e-01 -2.94404477e-01 -7.26345181e-01
-3.68944317e-01 -3.32775354e-01 -1.22416806e+00 7.07071841e-01
-1.77657712e+00 -1.17608202e+00 -1.54003114e-01 5.34026265e-01
8.52082074e-01 -3.33409831e-02 8.38804781e-01 -6.56287149e-02
-8.44684899e-01 5.01323581e-01 -2.22776905e-01 5.88985443e-01
7.14347959e-01 -1.23805416e+00 2.68999636e-01 7.62082398e-01
-1.92297980e-01 8.65570247e-01 5.30730367e-01 -4.51961488e-01
-1.27451360e+00 -1.07836354e+00 1.02013969e+00 -3.45044434e-02
7.47820973e-01 -6.01437569e-01 -1.12971604e+00 7.13856757e-01
1.01745677e+00 -2.75622278e-01 1.16945243e+00 7.36372292e-01
-4.00056750e-01 2.57288814e-01 -7.08323240e-01 6.30444288e-01
8.98463786e-01 -7.48319209e-01 -6.31841958e-01 1.24318309e-01
1.32609105e+00 -3.83643955e-01 -8.07462275e-01 1.63158745e-01
1.08644761e-01 -8.18494022e-01 5.78444898e-01 -6.21856809e-01
6.50243819e-01 1.08304389e-01 -8.49049389e-02 -1.52650380e+00
-2.29382068e-02 -9.24960554e-01 -2.10481793e-01 1.77731597e+00
4.13569093e-01 -6.07288957e-01 2.95855701e-01 7.60970831e-01
-5.47257423e-01 -5.65750957e-01 -7.14124084e-01 -3.01311880e-01
-9.23492666e-03 -6.19056344e-01 4.69072104e-01 1.45054770e+00
6.29733503e-01 9.81497824e-01 -4.50033605e-01 -6.17906228e-02
3.66866356e-03 4.10403073e-01 6.12312376e-01 -1.15614688e+00
-2.68252105e-01 -6.41566277e-01 1.77330241e-01 -1.16942346e+00
7.91722596e-01 -8.85587096e-01 2.50300080e-01 -1.55362856e+00
-3.88154760e-02 -6.69948086e-02 -1.51508272e-01 8.82074609e-02
-5.18722892e-01 -4.31620955e-01 3.25539589e-01 -1.75709546e-01
-1.08284068e+00 9.69320893e-01 1.38294399e+00 -2.17072040e-01
-4.66454715e-01 -5.98451272e-02 -1.02836406e+00 7.15230227e-01
1.03660595e+00 -8.06563720e-02 -4.34213221e-01 -3.57786417e-01
1.08629622e-01 3.34546566e-01 3.28570046e-02 -1.71783626e-01
3.75119597e-01 -1.63817108e-01 -1.19994424e-01 -6.52662218e-01
3.83583635e-01 -5.04142821e-01 -5.76226413e-01 -3.45103770e-01
-8.98403645e-01 -1.38445497e-01 1.32685542e-01 7.06163824e-01
-5.40556252e-01 -2.37498105e-01 2.30629161e-01 -1.27475038e-01
-6.94110811e-01 4.03199485e-03 -8.07632744e-01 4.80112314e-01
6.44095361e-01 2.70702392e-01 -5.32974243e-01 -9.00350511e-01
-6.14457786e-01 7.03099430e-01 -1.91791892e-01 7.26017118e-01
6.72031641e-01 -1.27292800e+00 -7.98756123e-01 -1.55531809e-01
2.29552746e-01 9.44730490e-02 4.10498083e-01 7.68791258e-01
6.15111068e-02 3.29015166e-01 4.32215482e-01 -3.80420119e-01
-1.50457406e+00 3.14783931e-01 3.44665535e-02 -3.72827947e-01
-6.13690913e-01 9.92620289e-01 9.05430987e-02 -6.42625272e-01
5.08309543e-01 -4.16319460e-01 -8.57364953e-01 4.02850121e-01
4.39731896e-01 1.31153569e-01 -2.01060742e-01 -6.81697488e-01
-2.16168955e-01 -8.33960325e-02 -3.27318221e-01 -1.07268192e-01
1.46516669e+00 -4.98013437e-01 -3.34175795e-01 7.50890553e-01
1.44158494e+00 1.03194676e-01 -1.11086500e+00 -4.63551730e-01
4.08026248e-01 -4.35559005e-02 -7.28900358e-02 -6.36430979e-01
-9.38836932e-01 8.61822724e-01 -1.71108082e-01 6.29890919e-01
8.89061689e-01 1.37141630e-01 6.77878499e-01 5.52967966e-01
-1.56712040e-01 -1.35371566e+00 2.99453795e-01 1.05666292e+00
1.21426678e+00 -1.43717289e+00 -2.79587984e-01 -6.10015690e-01
-1.20693278e+00 1.24745476e+00 7.82612383e-01 1.32019386e-01
5.71350753e-01 6.93171797e-03 3.49361688e-01 -4.77246255e-01
-1.05843341e+00 -2.43954256e-01 2.74701446e-01 5.03683627e-01
1.00056827e+00 1.08483151e-01 -2.85391599e-01 7.65812755e-01
-1.60547838e-01 -6.03154361e-01 5.12706280e-01 7.68412530e-01
-1.85171142e-01 -9.85730827e-01 -1.25275716e-01 1.41731814e-01
-3.80351156e-01 -2.59597331e-01 -8.71793628e-01 7.50463068e-01
-3.75914603e-01 1.52418494e+00 1.51802786e-02 -2.31125951e-01
2.91717172e-01 1.71957031e-01 -4.66527417e-02 -6.25964701e-01
-1.03937876e+00 2.73424536e-01 7.99514949e-01 -5.25965512e-01
-8.96449029e-01 -6.12200856e-01 -1.32626414e+00 -1.29564837e-01
-3.84479582e-01 3.42900068e-01 5.32973766e-01 9.72055674e-01
5.51705539e-01 9.78640378e-01 8.11626792e-01 -6.50566697e-01
-1.38832226e-01 -1.29187477e+00 -1.54449925e-01 4.55667466e-01
4.32538688e-01 -4.41463262e-01 -4.23068076e-01 -1.75906852e-01] | [12.93421745300293, 6.374436378479004] |
0c150f54-ac0e-4e56-973d-8395b48776ef | low-resource-neural-machine-translation-a | 2003.14402 | null | https://arxiv.org/abs/2003.14402v1 | https://arxiv.org/pdf/2003.14402v1.pdf | Low Resource Neural Machine Translation: A Benchmark for Five African Languages | Recent advents in Neural Machine Translation (NMT) have shown improvements in low-resource language (LRL) translation tasks. In this work, we benchmark NMT between English and five African LRL pairs (Swahili, Amharic, Tigrigna, Oromo, Somali [SATOS]). We collected the available resources on the SATOS languages to evaluate the current state of NMT for LRLs. Our evaluation, comparing a baseline single language pair NMT model against semi-supervised learning, transfer learning, and multilingual modeling, shows significant performance improvements both in the En-LRL and LRL-En directions. In terms of averaged BLEU score, the multilingual approach shows the largest gains, up to +5 points, in six out of ten translation directions. To demonstrate the generalization capability of each model, we also report results on multi-domain test sets. We release the standardized experimental data and the test sets for future works addressing the challenges of NMT in under-resourced settings, in particular for the SATOS languages. | ['Matteo Negri', 'Marco Turchi', 'Surafel M. Lakew'] | 2020-03-31 | null | null | null | null | ['low-resource-neural-machine-translation'] | ['natural-language-processing'] | [ 9.15639028e-02 -3.00680131e-01 -6.05943978e-01 -3.79055530e-01
-1.73313475e+00 -8.44869077e-01 8.80432069e-01 -3.35945964e-01
-5.17046332e-01 1.29754210e+00 2.67063588e-01 -9.76612151e-01
2.49362439e-01 -1.45056173e-01 -9.15077567e-01 -2.82864630e-01
3.50160003e-01 1.14234698e+00 -4.42444414e-01 -6.39557421e-01
5.85304992e-03 1.70015305e-01 -3.51175338e-01 5.81635118e-01
1.24999464e+00 2.03502446e-01 2.27462769e-01 4.79223967e-01
2.52346937e-02 2.89332747e-01 -4.75401551e-01 -7.74835765e-01
5.32789826e-01 -6.33007407e-01 -9.16190147e-01 -5.96077740e-01
6.96519315e-01 -4.80732322e-02 -1.30188195e-02 6.92001045e-01
7.21391976e-01 -2.74181217e-01 6.95396125e-01 -7.91792870e-01
-1.12272847e+00 9.73054588e-01 -6.72575295e-01 1.57762945e-01
3.96495193e-01 1.96928270e-02 9.75618124e-01 -1.42123020e+00
9.83228445e-01 1.40347397e+00 7.59247243e-01 5.65548539e-01
-1.20355487e+00 -7.63334513e-01 -2.73678094e-01 -4.34607603e-02
-1.18795967e+00 -7.16004968e-01 1.06542990e-01 -1.28238484e-01
1.49288249e+00 1.62863925e-01 1.56980380e-02 1.32773912e+00
4.95610684e-01 7.70231366e-01 1.76334500e+00 -9.73939598e-01
-3.25674653e-01 3.78848106e-01 -4.12632048e-01 2.88979292e-01
1.90239295e-01 -1.56124565e-03 -7.97860324e-01 -1.70590263e-02
4.15627211e-01 -6.57803297e-01 -1.01290792e-02 2.27396950e-01
-1.74988520e+00 7.32909024e-01 7.32207522e-02 6.04770243e-01
-1.83414847e-01 -3.13314408e-01 4.48592335e-01 1.02015793e+00
1.07894039e+00 5.22186160e-01 -9.72123861e-01 -3.42237473e-01
-9.82869387e-01 -9.40540507e-02 8.04280281e-01 1.23022425e+00
4.79429305e-01 4.15919488e-03 -8.20455551e-02 1.39029598e+00
6.98082373e-02 1.10774541e+00 5.86698353e-01 -4.25966948e-01
1.36897254e+00 2.09108159e-01 5.71638979e-02 -2.12667644e-01
-1.01651609e-01 -3.71929944e-01 -7.42646933e-01 -4.60452765e-01
4.68288839e-01 -3.95728409e-01 -8.28738630e-01 1.79960346e+00
-9.79837254e-02 -5.25116086e-01 5.90352714e-01 6.23418331e-01
5.25916398e-01 1.02525854e+00 -1.67304754e-01 -3.86266649e-01
1.05129325e+00 -1.38466585e+00 -6.32363141e-01 -4.01383072e-01
1.10181749e+00 -1.51914358e+00 1.33517003e+00 6.22074306e-02
-1.24673629e+00 -5.52229285e-01 -9.55590725e-01 -1.07208826e-01
-3.88851643e-01 6.92799032e-01 3.21525544e-01 4.42452669e-01
-1.38324738e+00 3.98274153e-01 -7.51993001e-01 -9.63140130e-01
-6.89779818e-02 4.58642781e-01 -6.32492185e-01 -4.07247692e-01
-1.57988560e+00 1.46811223e+00 2.04063386e-01 -1.86534086e-03
-6.78695798e-01 -5.55792451e-01 -5.07387519e-01 -6.14838481e-01
-2.00626999e-01 -4.06673908e-01 1.31665730e+00 -1.03502965e+00
-1.55317020e+00 1.20322871e+00 -2.13474303e-01 -3.06835443e-01
7.61433601e-01 -3.65828842e-01 -6.80784464e-01 -4.28410500e-01
4.04303819e-01 7.67199457e-01 2.14437738e-01 -8.07465971e-01
-5.86678684e-01 -1.79595396e-01 -3.42844784e-01 5.09472251e-01
-1.85737535e-01 6.96994007e-01 -2.51843154e-01 -7.52240539e-01
-6.38892725e-02 -1.35760522e+00 8.23200960e-03 -9.25578058e-01
-2.54972100e-01 -1.12604924e-01 3.10269862e-01 -1.31873977e+00
9.72453535e-01 -1.70393288e+00 3.88671786e-01 -2.45033711e-01
-6.28972411e-01 3.06391656e-01 -5.43844879e-01 8.21373701e-01
1.41077619e-02 2.08598047e-01 -1.66245446e-01 -4.17417049e-01
-1.78201348e-01 2.48880565e-01 -1.76704541e-01 3.10183644e-01
3.58888805e-01 1.31726742e+00 -7.51160443e-01 -3.81292105e-01
-2.18382195e-01 1.90988109e-01 2.75332611e-02 2.87825260e-02
5.19790268e-03 9.15894389e-01 1.12506449e-01 8.02197754e-01
4.50994045e-01 2.27154821e-01 3.70632619e-01 1.35361329e-01
-3.05811793e-01 9.06503737e-01 -2.97641903e-01 2.07929611e+00
-9.59554076e-01 7.53467083e-01 -3.03190053e-01 -4.44299042e-01
9.53827322e-01 5.18582106e-01 3.34357731e-02 -1.00127125e+00
-2.82515585e-01 1.22179568e+00 3.80922794e-01 -2.47911233e-02
5.86879671e-01 -7.73729831e-02 -1.87253296e-01 7.27403402e-01
3.19795400e-01 -5.39654195e-02 2.90511250e-01 -4.81473133e-02
5.96583545e-01 5.32581985e-01 3.02631199e-01 -6.24813080e-01
3.72604847e-01 2.88260370e-01 4.91158545e-01 4.75289583e-01
3.96740437e-02 5.04713833e-01 9.47870407e-03 -2.06225529e-01
-1.45379567e+00 -1.07795751e+00 -5.79322949e-02 1.29174113e+00
-4.76298898e-01 -1.64654508e-01 -8.72330606e-01 -8.26118469e-01
-3.66244763e-01 8.71068656e-01 -2.31019586e-01 9.71745253e-02
-1.22284937e+00 -1.06229293e+00 8.52408946e-01 2.35759124e-01
4.23865587e-01 -1.04994857e+00 2.71121800e-01 1.50895000e-01
-7.68304229e-01 -1.31331754e+00 -7.02581942e-01 7.17106238e-02
-1.04234278e+00 -2.91546732e-01 -1.10793400e+00 -1.18854153e+00
4.02047396e-01 1.09456465e-01 1.26563275e+00 -5.99656999e-01
3.08105737e-01 -2.73172885e-01 -3.12680542e-01 -3.59411001e-01
-9.79203820e-01 7.80217409e-01 4.93699580e-01 -4.60979670e-01
5.67320764e-01 -1.93631902e-01 3.25025357e-02 3.64210308e-01
-1.85519964e-01 2.83498645e-01 9.68688965e-01 9.19581890e-01
5.42616904e-01 -8.81271839e-01 7.56238580e-01 -8.50237548e-01
7.28993237e-01 -5.31196654e-01 -3.03113729e-01 7.51549304e-01
-7.37100899e-01 -3.91516760e-02 6.04146838e-01 -4.21076119e-01
-1.17518365e+00 -4.29099262e-01 9.96851400e-02 1.23700641e-01
1.47206157e-01 6.22563601e-01 -6.84081540e-02 2.30111312e-02
7.44050205e-01 3.17347616e-01 -3.27202737e-01 -7.89956868e-01
4.61520374e-01 1.14441919e+00 4.21595067e-01 -9.09056365e-01
8.15499604e-01 -2.39245266e-01 -3.30626428e-01 -4.44663167e-01
-4.48351771e-01 -1.22353487e-01 -9.74424541e-01 3.01718805e-02
5.67769587e-01 -1.24131978e+00 2.64147878e-01 4.63386446e-01
-1.38280308e+00 -6.00682616e-01 1.16112538e-01 8.80587399e-01
-5.85725009e-01 -9.01800171e-02 -1.16830850e+00 -3.49343210e-01
-7.85771012e-01 -1.38789868e+00 1.13710713e+00 -1.34780213e-01
-4.58931148e-01 -1.15024257e+00 4.72340882e-01 5.43953717e-01
6.38396084e-01 -1.91061229e-01 1.21922553e+00 -9.46992755e-01
-9.32730734e-02 2.24286299e-02 -2.08047956e-01 4.49700743e-01
3.32789987e-01 -3.11905921e-01 -4.92084056e-01 -6.82928979e-01
-2.07138985e-01 -6.23824656e-01 3.75998825e-01 7.28007555e-02
-1.42964497e-01 -2.03337058e-01 -1.01707526e-01 4.47609931e-01
1.29755461e+00 2.11974189e-01 3.94081384e-01 5.50683379e-01
6.63536251e-01 5.17024696e-01 8.82574737e-01 -3.47485483e-01
4.43054914e-01 9.22604501e-01 -4.53597933e-01 -4.63049620e-01
-4.49106902e-01 -4.23208684e-01 1.06807005e+00 1.81421697e+00
-2.41143331e-01 -2.85662085e-01 -1.21116912e+00 5.54192960e-01
-1.67907894e+00 -2.71661520e-01 -1.60054758e-01 2.32566595e+00
1.22913885e+00 -1.38319924e-01 -4.33063433e-02 -5.83225548e-01
6.47368133e-01 -2.93376774e-01 -1.99160531e-01 -1.04360485e+00
-4.83870775e-01 4.47820961e-01 6.34854317e-01 6.77739620e-01
-6.78857028e-01 1.68662667e+00 6.34318161e+00 1.02345884e+00
-1.12252569e+00 6.52989209e-01 7.40600109e-01 -5.12659224e-03
-1.89311698e-01 1.32444739e-01 -9.07546043e-01 2.32431516e-01
1.66062653e+00 -8.13062638e-02 7.33212113e-01 3.23299825e-01
3.26402217e-01 2.38022029e-01 -1.31827426e+00 6.49641693e-01
2.22689673e-01 -9.27839100e-01 1.49109930e-01 1.02486573e-01
1.34762001e+00 9.00250256e-01 1.61783010e-01 5.62283278e-01
4.02646214e-01 -1.01786351e+00 6.14817381e-01 4.74598669e-02
1.38014138e+00 -7.67445207e-01 8.11022520e-01 3.56939554e-01
-7.07112134e-01 4.84660596e-01 -5.05775392e-01 1.37006894e-01
1.07974723e-01 6.01560250e-02 -1.14642262e+00 1.03395987e+00
4.74223495e-01 7.61044562e-01 -5.24063051e-01 4.11748827e-01
-3.96399677e-01 1.01424980e+00 -2.11089462e-01 1.25445396e-01
5.16272366e-01 -4.97162670e-01 5.11473954e-01 1.60125446e+00
5.01930416e-01 -5.34903109e-01 4.97519337e-02 4.46221650e-01
-3.70245814e-01 8.25043440e-01 -7.66121924e-01 -2.05169678e-01
3.29838067e-01 9.98891532e-01 -2.80387253e-01 -2.52390087e-01
-5.18331170e-01 1.33410311e+00 5.11191428e-01 4.18838501e-01
-6.68964565e-01 -2.36181542e-01 4.98488247e-01 -1.94956109e-01
-2.76745319e-01 -5.29635012e-01 -3.05631340e-01 -1.32201493e+00
1.22606628e-01 -1.51269007e+00 1.06740063e-02 -5.64598560e-01
-1.23738515e+00 9.78479981e-01 4.14529815e-02 -1.33833587e+00
-6.20314717e-01 -6.25269234e-01 -2.06577092e-01 1.49830091e+00
-1.31585777e+00 -1.82200885e+00 7.08604038e-01 1.23592064e-01
9.96888757e-01 -7.76288509e-01 1.06611955e+00 5.96612275e-01
-4.06875193e-01 1.11522532e+00 6.84402764e-01 1.13682643e-01
1.36590242e+00 -1.06243694e+00 1.17025483e+00 9.40538764e-01
4.74248886e-01 8.18172991e-01 4.05587137e-01 -7.60922253e-01
-1.25548589e+00 -1.14378369e+00 1.82266009e+00 -7.53559709e-01
7.30483472e-01 -6.13424659e-01 -4.98198569e-01 9.88106728e-01
7.11632729e-01 -6.65601194e-01 6.44986629e-01 2.96885610e-01
-4.49875057e-01 1.21130623e-01 -1.00316000e+00 8.21692705e-01
9.68991697e-01 -6.82478070e-01 -4.90129501e-01 7.70205915e-01
8.46865177e-01 -3.39665771e-01 -1.15113437e+00 4.83273864e-01
6.65931880e-01 -2.68893927e-01 6.38613522e-01 -1.03300035e+00
5.75085580e-01 -7.76220858e-02 -5.71848094e-01 -1.81306016e+00
-1.19046174e-01 -7.54126132e-01 4.85345036e-01 1.40019035e+00
1.20905650e+00 -7.42452979e-01 1.47293955e-01 -1.80749565e-01
-1.76689684e-01 -6.42263710e-01 -1.16036081e+00 -1.07833850e+00
8.47934544e-01 -2.08152041e-01 3.53948623e-01 1.16835773e+00
-7.00537711e-02 8.18039894e-01 -7.57600486e-01 -2.55564660e-01
2.61819869e-01 3.16043035e-03 7.81642973e-01 -5.86026967e-01
-4.96802181e-01 -2.21533209e-01 1.74438775e-01 -7.64245450e-01
3.35866958e-01 -1.49339068e+00 -1.16290376e-02 -1.55792630e+00
3.16771448e-01 -4.94858265e-01 -1.91935614e-01 6.42287076e-01
-1.02654450e-01 6.90677166e-01 3.37888479e-01 6.40236318e-01
-2.97072738e-01 1.15968667e-01 1.18792045e+00 -8.62079337e-02
-1.31841689e-01 -1.83356583e-01 -3.61400902e-01 2.47584432e-01
1.02345741e+00 -5.22191107e-01 -1.40013441e-01 -1.17653811e+00
9.77493078e-02 -5.43182641e-02 -5.96321225e-01 -6.68806016e-01
-1.16799235e-01 -1.78518012e-01 1.92474440e-01 -3.53489697e-01
1.61663041e-01 -2.86904275e-01 1.65893450e-01 5.48776567e-01
-5.31075299e-01 7.81242013e-01 3.53680283e-01 -2.34109119e-01
-1.22010134e-01 1.36062369e-01 6.65476024e-01 -9.37701985e-02
-2.26219699e-01 1.65725052e-02 -2.62888640e-01 2.11683929e-01
5.82668662e-01 1.25938967e-01 -3.70595753e-01 -3.25011641e-01
-2.28226691e-01 1.29773661e-01 5.30152559e-01 8.46476614e-01
1.12363778e-01 -1.43800664e+00 -1.58834159e+00 1.53111100e-01
3.04290414e-01 -6.67863131e-01 -4.43713754e-01 1.11799932e+00
-5.09736836e-01 8.78746569e-01 -4.48625058e-01 -4.71545547e-01
-1.04795432e+00 1.34494724e-02 2.00309262e-01 -6.87640548e-01
8.86249531e-04 5.95804811e-01 -1.83705181e-01 -1.22576165e+00
-2.14480653e-01 -1.64148156e-02 3.04651648e-01 -2.78939545e-01
1.13944121e-01 4.22180563e-01 3.53019923e-01 -1.02799833e+00
-3.44771564e-01 5.03416419e-01 -4.43975240e-01 -6.75274611e-01
9.52646792e-01 -2.15437472e-01 -4.48401541e-01 8.36083710e-01
1.24027264e+00 2.65493661e-01 -2.68525153e-01 -4.00566101e-01
3.39671493e-01 -9.79299471e-02 -4.52829659e-01 -1.48108506e+00
-4.54221815e-01 9.25205827e-01 4.91445333e-01 -7.50388265e-01
9.37839031e-01 -1.71945974e-01 9.15419638e-01 3.60367209e-01
8.11501801e-01 -1.11188126e+00 -6.31727040e-01 9.42876458e-01
8.54909301e-01 -1.35021412e+00 -4.14057910e-01 -1.64433233e-02
-7.44248688e-01 1.03854847e+00 4.30660248e-01 2.11541131e-01
6.69109747e-02 1.26082510e-01 5.84980607e-01 4.88717645e-01
-9.01051164e-01 2.71215141e-01 3.53506535e-01 3.66835713e-01
1.16568065e+00 4.19436216e-01 -8.87909055e-01 3.71370539e-02
-3.52351457e-01 -1.96410030e-01 2.24251673e-01 6.43969297e-01
8.04920420e-02 -1.75657248e+00 -3.46605122e-01 2.70620942e-01
-7.73982584e-01 -6.81388974e-01 -7.90773094e-01 9.75031972e-01
-9.20951068e-02 9.88757193e-01 -1.61226794e-01 -4.49202716e-01
2.12458730e-01 3.40329260e-01 6.39228046e-01 -6.08899117e-01
-8.77511322e-01 2.96013862e-01 4.77445394e-01 -1.17961571e-01
-1.77446485e-01 -7.72079766e-01 -8.48672092e-01 -4.06159341e-01
-2.85430968e-01 3.98244590e-01 1.00649905e+00 9.80783761e-01
1.96859956e-01 -2.42556054e-02 5.69816530e-01 -6.28111541e-01
-7.18782902e-01 -1.57233918e+00 1.52299419e-01 1.32346883e-01
-2.08352581e-01 -9.40363482e-02 -2.02843659e-02 -1.97169274e-01] | [11.542854309082031, 10.390960693359375] |
598c2b33-a2e0-41cf-b340-7908e7e0538c | inducing-semantic-grouping-of-latent-concepts | 2108.11761 | null | https://arxiv.org/abs/2108.11761v2 | https://arxiv.org/pdf/2108.11761v2.pdf | A Framework for Learning Ante-hoc Explainable Models via Concepts | Self-explaining deep models are designed to learn the latent concept-based explanations implicitly during training, which eliminates the requirement of any post-hoc explanation generation technique. In this work, we propose one such model that appends an explanation generation module on top of any basic network and jointly trains the whole module that shows high predictive performance and generates meaningful explanations in terms of concepts. Our training strategy is suitable for unsupervised concept learning with much lesser parameter space requirements compared to baseline methods. Our proposed model also has provision for leveraging self-supervision on concepts to extract better explanations. However, with full concept supervision, we achieve the best predictive performance compared to recently proposed concept-based explainable models. We report both qualitative and quantitative results with our method, which shows better performance than recently proposed concept-based explainability methods. We reported exhaustive results with two datasets without ground truth concepts, i.e., CIFAR10, ImageNet, and two datasets with ground truth concepts, i.e., AwA2, CUB-200, to show the effectiveness of our method for both cases. To the best of our knowledge, we are the first ante-hoc explanation generation method to show results with a large-scale dataset such as ImageNet. | ['Vineeth N Balasubramanian', 'Anindya Sarkar', 'Deepak Vijaykeerthy', 'Anirban Sarkar'] | 2021-08-25 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Sarkar_A_Framework_for_Learning_Ante-Hoc_Explainable_Models_via_Concepts_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Sarkar_A_Framework_for_Learning_Ante-Hoc_Explainable_Models_via_Concepts_CVPR_2022_paper.pdf | cvpr-2022-1 | ['explainable-models'] | ['computer-vision'] | [ 1.21638231e-01 9.16474283e-01 -2.48883530e-01 -5.99613607e-01
-2.96173275e-01 -9.52221230e-02 7.02698946e-01 1.14969738e-01
-5.12634590e-02 8.47614884e-01 1.49538785e-01 -5.64190328e-01
-2.46099874e-01 -7.03880847e-01 -9.25229371e-01 -2.65432924e-01
-1.01243801e-01 7.01929688e-01 -3.83189991e-02 -1.74519420e-01
4.16571736e-01 -5.10882996e-02 -1.80179155e+00 3.70761514e-01
1.41735721e+00 7.05194414e-01 1.82882994e-01 3.48331392e-01
-1.58803195e-01 6.81816220e-01 -3.63353938e-01 -5.68335712e-01
-2.88325585e-02 -5.99239230e-01 -8.68019760e-01 3.54445666e-01
5.22671402e-01 -2.57388204e-01 -1.12163108e-02 5.57998002e-01
-1.11770868e-01 1.35356054e-01 7.14688480e-01 -1.58141410e+00
-1.12730265e+00 9.34659183e-01 -2.31325969e-01 -1.21465363e-01
1.83406696e-01 -1.17007703e-01 1.28691483e+00 -9.83208239e-01
4.03432488e-01 1.12553489e+00 4.95019436e-01 9.69171882e-01
-1.22670174e+00 -8.94802988e-01 4.42553580e-01 3.74075085e-01
-8.98060799e-01 -5.16780056e-02 7.36946464e-01 -1.64623290e-01
1.11382568e+00 -2.12293733e-02 7.12363660e-01 1.12316239e+00
-1.44091425e-02 8.27104926e-01 8.64234030e-01 -4.94914353e-01
4.67681944e-01 3.22331637e-01 4.21705097e-01 7.18115866e-01
5.29910862e-01 2.71722883e-01 -6.54070079e-01 7.58613124e-02
6.36151373e-01 3.69250685e-01 -2.74007827e-01 -5.82848489e-01
-1.16930103e+00 1.12819040e+00 8.47103179e-01 3.46890718e-01
-3.40856045e-01 4.00193363e-01 -9.07958820e-02 1.84149683e-01
5.17906547e-01 7.10049748e-01 -7.01714873e-01 3.18014592e-01
-1.00712454e+00 1.67826205e-01 6.21792674e-01 1.23446381e+00
9.60640490e-01 2.70797193e-01 1.17917791e-01 4.50952739e-01
3.55247021e-01 2.11078465e-01 7.71698117e-01 -8.32249761e-01
4.43763435e-01 7.45584369e-01 -1.31805882e-01 -8.19972038e-01
-3.90622079e-01 -7.87172675e-01 -8.15923870e-01 -5.87625280e-02
-6.32128417e-02 -3.57021838e-02 -1.14424336e+00 1.85225403e+00
-1.53061926e-01 4.86492634e-01 5.44799864e-01 9.10092831e-01
1.00743258e+00 3.25454563e-01 1.72734290e-01 -7.22289756e-02
1.11777997e+00 -1.40888929e+00 -6.75727904e-01 -3.79387081e-01
5.51806927e-01 -2.01178432e-01 1.21481597e+00 3.96990240e-01
-8.73184443e-01 -6.66967213e-01 -1.27196610e+00 1.43409789e-01
-3.79819542e-01 1.59543917e-01 1.46967709e+00 4.40845937e-01
-1.18466306e+00 6.44840419e-01 -7.78527200e-01 -4.84333932e-01
5.58502853e-01 5.71230114e-01 -5.02642572e-01 -4.29415591e-02
-1.08469677e+00 6.17226362e-01 6.43767357e-01 -3.76284599e-01
-9.05534267e-01 -7.67608166e-01 -9.70488012e-01 5.56560755e-01
3.79677981e-01 -9.86653090e-01 1.14995992e+00 -1.13599873e+00
-1.08976805e+00 3.80373597e-01 -2.42146581e-01 -9.71807599e-01
1.02751724e-01 -4.22560871e-01 -3.27136397e-01 1.24668613e-01
2.60434866e-01 1.37890565e+00 6.47432268e-01 -1.60302222e+00
-3.65012318e-01 -3.41518484e-02 4.78669077e-01 -2.52594259e-02
-4.80949938e-01 -6.53339565e-01 -1.67918548e-01 -5.53154171e-01
5.74641705e-01 -9.25060868e-01 -3.61406475e-01 -3.10837984e-01
-6.66266739e-01 -2.29958460e-01 5.68551123e-01 -6.89420477e-02
8.29738021e-01 -1.86334753e+00 -1.73231632e-01 2.96589248e-02
4.53426212e-01 3.71790379e-02 -3.19235265e-01 2.77988464e-01
-6.24811769e-01 5.23319542e-01 -5.15498757e-01 -7.39357054e-01
-3.62938419e-02 4.67043459e-01 -4.70857531e-01 -3.00241895e-02
4.42898959e-01 9.80767310e-01 -1.00875449e+00 -3.02146435e-01
1.50092646e-01 4.37059224e-01 -1.20250249e+00 1.38559103e-01
-5.06389976e-01 3.93668890e-01 -4.16280150e-01 3.82908404e-01
4.22599316e-01 -5.86044312e-01 1.48703799e-01 1.51000276e-01
2.67537534e-01 3.87109607e-01 -8.28296781e-01 1.93389392e+00
-5.44206440e-01 6.38362169e-01 -7.71139920e-01 -1.04986441e+00
1.03543019e+00 5.35635412e-01 2.01664343e-01 -5.21068394e-01
-1.20848611e-01 2.50758857e-01 1.30154416e-01 -1.21895023e-01
5.24407148e-01 -3.46262366e-01 3.13150376e-01 4.92381752e-01
3.43149096e-01 3.11043169e-02 1.29976004e-01 6.77580655e-01
9.82259989e-01 1.25603735e-01 3.59504223e-01 -3.25548798e-01
4.22385395e-01 1.90256953e-01 3.09984684e-01 8.15618336e-01
1.84885994e-01 8.16183567e-01 3.16190004e-01 -5.54809690e-01
-7.45324969e-01 -8.86685014e-01 5.22944592e-02 5.62318623e-01
2.54078537e-01 -7.23042369e-01 -6.50080204e-01 -9.32739019e-01
-6.60223141e-02 1.09859574e+00 -7.43515551e-01 -2.36617640e-01
-8.28870833e-02 -4.48483437e-01 1.57275483e-01 7.31360316e-01
6.16831541e-01 -1.17069507e+00 -4.26399797e-01 3.24267447e-02
-2.06102908e-01 -1.03025413e+00 -7.78774824e-03 3.15237790e-01
-1.38659978e+00 -1.20263386e+00 -1.80131584e-01 -5.85957050e-01
1.04307604e+00 4.46808517e-01 1.32984281e+00 7.46610880e-01
1.40852258e-01 2.87183851e-01 -5.64976037e-01 -4.82845277e-01
-2.62578577e-02 7.14578852e-02 -1.42976686e-01 -3.06925774e-01
2.83997536e-01 -8.05753589e-01 -6.99963272e-01 2.64355361e-01
-9.47478652e-01 5.35423815e-01 7.46296525e-01 1.00002706e+00
5.60148656e-01 1.67779639e-01 7.28945732e-01 -1.37292290e+00
3.75201762e-01 -7.69193590e-01 -1.00102924e-01 7.80765899e-03
-1.17582607e+00 4.01827693e-01 7.22935438e-01 -2.59822577e-01
-1.03138936e+00 7.54373968e-02 1.00546353e-01 -3.27101260e-01
-4.65101928e-01 5.95405400e-01 -1.54633507e-01 4.44474488e-01
7.35491216e-01 9.78923813e-02 -2.65986621e-01 -4.78193432e-01
6.25397325e-01 2.23842427e-01 6.15352929e-01 -4.07051504e-01
1.00850832e+00 5.98126709e-01 -1.38490453e-01 -1.67997077e-01
-1.12983894e+00 -3.24927062e-01 -6.00984097e-01 3.33450735e-01
7.61062205e-01 -9.90261793e-01 -3.87562007e-01 -2.67414659e-01
-1.32433438e+00 -8.66143629e-02 -2.32677117e-01 6.81608498e-01
-7.58074701e-01 6.61601871e-02 -3.24524611e-01 -5.83495855e-01
-2.73638010e-01 -8.01516235e-01 8.01586211e-01 1.67824343e-01
-3.65653247e-01 -1.22418773e+00 -2.06257731e-01 4.70648915e-01
4.03222293e-01 2.64808059e-01 8.89925778e-01 -1.07178485e+00
-9.02467489e-01 7.61468783e-02 -2.70601183e-01 5.57333715e-02
4.84308489e-02 -3.76432776e-01 -1.16455054e+00 -8.64064544e-02
-1.87383428e-01 -3.50761950e-01 1.34063280e+00 2.82251656e-01
1.45385504e+00 -5.72086811e-01 -3.86782229e-01 4.36999500e-01
1.46394217e+00 2.88193696e-03 7.45384872e-01 4.46730644e-01
6.67646110e-01 5.97962439e-01 8.05185676e-01 2.82268614e-01
5.67661762e-01 4.23191398e-01 9.13857877e-01 -4.08336788e-01
-1.45397037e-01 -4.71688569e-01 7.80299678e-02 5.37981570e-01
-1.49365142e-01 -5.27829289e-01 -7.57043362e-01 7.32818067e-01
-2.19641995e+00 -9.84123290e-01 -3.73616129e-01 1.84089446e+00
5.09130478e-01 8.18847120e-02 -2.99763560e-01 2.99495041e-01
3.97622436e-01 -1.57593369e-01 -5.20829320e-01 -3.75134647e-01
1.02297768e-01 3.56746227e-01 -8.05748347e-03 5.67279220e-01
-8.39537203e-01 1.17259371e+00 6.10297871e+00 3.94346118e-01
-7.06337214e-01 2.23174617e-01 5.14274418e-01 -3.15969996e-02
-8.74734700e-01 5.08706093e-01 -4.29013759e-01 8.81225318e-02
8.72172117e-01 1.09305836e-01 1.74343929e-01 1.09580839e+00
2.03670096e-02 2.51890421e-01 -1.35930228e+00 7.03350306e-01
2.20132440e-01 -1.40948474e+00 3.59337717e-01 1.05994612e-01
1.00185215e+00 -3.10030878e-01 -5.29749691e-02 4.06447023e-01
3.39551449e-01 -1.23831189e+00 6.06908739e-01 3.62850398e-01
4.87045825e-01 -8.44000220e-01 8.99338007e-01 3.40173453e-01
-7.93769240e-01 -6.12789169e-02 -5.44405460e-01 -2.91405499e-01
3.67477424e-02 7.77268648e-01 -8.62784028e-01 6.74741507e-01
4.74120617e-01 1.05672896e+00 -7.12128520e-01 8.88802707e-01
-8.19773316e-01 8.36746573e-01 1.74081728e-01 2.00841039e-01
2.36985236e-01 1.60584196e-01 1.90836787e-01 8.79484117e-01
5.37918150e-01 2.38720834e-01 -1.76966842e-02 1.26238263e+00
-5.97649887e-02 -9.28429887e-02 -5.22543311e-01 -3.12864706e-02
3.43618661e-01 1.20319486e+00 -7.53287375e-01 -6.59258544e-01
-3.81046623e-01 9.59843278e-01 4.54566538e-01 3.35047841e-01
-9.86207902e-01 -8.26635212e-02 5.37314475e-01 1.38573855e-01
2.68806815e-01 -1.03412732e-01 -5.79893470e-01 -1.41981077e+00
-2.89320886e-01 -6.71948612e-01 3.79594892e-01 -1.03908336e+00
-1.23262799e+00 9.00974870e-01 1.21838883e-01 -1.03500056e+00
-4.96369243e-01 -5.95280409e-01 -9.01534438e-01 6.18991554e-01
-1.72693539e+00 -1.19497335e+00 -5.46112418e-01 5.89573264e-01
6.59707963e-01 -4.00185674e-01 8.55332553e-01 -4.69159409e-02
-2.39136547e-01 5.25035441e-01 -3.97798091e-01 -2.67816871e-01
4.88529652e-01 -1.55513310e+00 4.83838081e-01 7.20548570e-01
6.09058976e-01 1.08565092e+00 8.68992329e-01 -6.92416787e-01
-8.13968956e-01 -1.01432621e+00 1.12140965e+00 -4.78036612e-01
3.78392696e-01 -3.09109151e-01 -9.97555256e-01 8.15729380e-01
2.95153081e-01 -3.25932771e-01 9.14926291e-01 5.74675560e-01
-3.54340136e-01 1.76039487e-01 -9.79047835e-01 5.29993057e-01
1.30572367e+00 -6.22082688e-02 -8.22178781e-01 3.32422018e-01
1.17966890e+00 -1.83010131e-01 -3.72178197e-01 4.01596963e-01
3.26960146e-01 -1.35278213e+00 9.52999949e-01 -7.26853907e-01
1.06723130e+00 -3.73940885e-01 -5.50590865e-02 -1.39732695e+00
-3.82042468e-01 -1.32098719e-01 -2.84370035e-01 1.20783567e+00
7.98496366e-01 -8.17636669e-01 9.42167759e-01 6.86654806e-01
-5.23953736e-01 -7.92337537e-01 -4.08828795e-01 -7.47090816e-01
-9.48705599e-02 -5.83603740e-01 1.05434155e+00 9.21750128e-01
5.41686043e-02 5.44006050e-01 -3.03032279e-01 2.21938759e-01
6.14100933e-01 4.59798515e-01 8.16543460e-01 -1.30286431e+00
-4.56231475e-01 -1.24326803e-01 -3.94849539e-01 -8.49936783e-01
5.15371382e-01 -1.03611374e+00 -1.66730806e-02 -1.86256289e+00
5.97390354e-01 -3.77133369e-01 -3.47156078e-01 1.01958966e+00
-3.87362927e-01 2.35426888e-01 2.32087344e-01 3.64819884e-01
-6.45724773e-01 8.22359502e-01 1.23749757e+00 -6.38791844e-02
-9.72996280e-02 -1.89501330e-01 -1.36878836e+00 6.71678245e-01
1.22591579e+00 -6.74704611e-01 -1.09660542e+00 -5.04595459e-01
6.34474866e-03 -2.53893733e-01 6.73166752e-01 -1.15341151e+00
7.42171258e-02 -1.24196291e-01 3.76352966e-01 -4.97657061e-01
6.82118684e-02 -8.20999980e-01 1.45268500e-01 4.92753267e-01
-5.22999406e-01 -2.31437422e-02 2.29751319e-01 7.27368593e-01
-4.33958739e-01 -1.57717571e-01 3.22560132e-01 -9.11649764e-02
-7.75068879e-01 3.70140642e-01 -3.51282135e-02 -3.05300713e-01
7.62513518e-01 -3.77822876e-01 -5.19524157e-01 -6.88623905e-01
-6.40005410e-01 3.40665460e-01 3.83207381e-01 5.67339003e-01
9.00341928e-01 -1.45943689e+00 -5.75030386e-01 1.75672144e-01
3.78573656e-01 -2.08889917e-02 1.10008195e-01 4.74063426e-01
-1.26893476e-01 8.62528265e-01 -1.87737182e-01 -4.81316537e-01
-8.81796777e-01 8.16458464e-01 1.16833203e-01 -2.63108134e-01
-7.19085693e-01 6.32413387e-01 6.49125099e-01 -5.94678581e-01
-3.71456705e-02 -4.17379737e-01 -3.75544488e-01 -4.75518107e-01
3.65900248e-01 -4.67623509e-02 -1.41373232e-01 -2.35114321e-01
-3.13803583e-01 2.01290920e-01 -1.71344019e-02 -1.15094163e-01
1.38883233e+00 -5.58852293e-02 1.44975185e-01 1.42875999e-01
7.99114883e-01 -3.04207921e-01 -1.23945379e+00 6.89375866e-03
5.55165708e-02 -5.03515303e-01 -1.28719732e-01 -1.03428698e+00
-1.27585137e+00 1.13895047e+00 2.73547560e-01 -1.86962318e-02
1.18018067e+00 2.18604624e-01 5.29065788e-01 4.85341251e-01
3.40470105e-01 -4.97282475e-01 5.20006359e-01 3.38470459e-01
9.92026091e-01 -1.45175517e+00 -8.01429152e-02 -5.85813344e-01
-7.04266846e-01 9.37217593e-01 1.09026229e+00 -1.91941202e-01
3.91369820e-01 -3.50378960e-01 -6.37680665e-02 -4.79210198e-01
-1.09442818e+00 -4.08778071e-01 6.30897939e-01 7.17042387e-01
6.44148588e-01 1.23154424e-01 -3.60981017e-01 1.08652544e+00
-5.31613648e-01 -1.09423354e-01 4.74228382e-01 3.76271725e-01
-4.42448914e-01 -1.21188748e+00 -3.63343656e-02 3.12453240e-01
-6.91708773e-02 -4.58985060e-01 -5.70834160e-01 1.07207131e+00
3.14878225e-01 1.22044170e+00 1.01637654e-02 -3.24931175e-01
1.04509570e-01 7.69923255e-02 1.18378088e-01 -9.64385271e-01
-3.64345491e-01 -3.69102597e-01 -8.21000189e-02 -6.33872330e-01
-5.96831203e-01 -3.13197732e-01 -1.85872543e+00 -3.32411617e-01
-5.59400141e-01 4.08425540e-01 6.30231857e-01 1.28497648e+00
5.66525638e-01 6.97182536e-01 2.21607953e-01 -5.96439958e-01
-2.04714350e-02 -8.98584902e-01 -4.17976707e-01 6.22288465e-01
2.16017693e-01 -8.58583033e-01 -4.15948480e-01 6.73652962e-02] | [9.02340316772461, 5.703380107879639] |
0db958ae-4354-42f5-890f-b9fcf6c2baa7 | s2gan-share-aging-factors-across-ages-and | null | null | http://openaccess.thecvf.com/content_ICCV_2019/html/He_S2GAN_Share_Aging_Factors_Across_Ages_and_Share_Aging_Trends_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/He_S2GAN_Share_Aging_Factors_Across_Ages_and_Share_Aging_Trends_ICCV_2019_paper.pdf | S2GAN: Share Aging Factors Across Ages and Share Aging Trends Among Individuals | Generally, we human follow the roughly common aging trends, e.g., the wrinkles only tend to be more, longer or deeper. However, the aging process of each individual is more dominated by his/her personalized factors, including the invariant factors such as identity and mole, as well as the personalized aging patterns, e.g., one may age by graying hair while another may age by receding hairline. Following this biological principle, in this work, we propose an effective and efficient method to simulate natural aging. Specifically, a personalized aging basis is established for each individual to depict his/her own aging factors. Then different ages share this basis, being derived through age-specific transforms. The age-specific transforms represent the aging trends which are shared among all individuals. The proposed method can achieve continuous face aging with favorable aging accuracy, identity preservation, and fidelity. Furthermore, befitted from the effective design, a unique model is capable of all ages and the prediction time is significantly saved.
| [' Xilin Chen', ' Shiguang Shan', ' Meina Kan', 'Zhenliang He'] | 2019-10-01 | null | null | null | iccv-2019-10 | ['face-age-editing'] | ['computer-vision'] | [-1.54118448e-01 -1.80709392e-01 -2.17081860e-01 -6.47541583e-02
4.24334586e-01 -2.58334398e-01 1.23644508e-01 -3.16840746e-02
-1.90405976e-02 8.74216735e-01 3.20700556e-01 2.87677169e-01
8.97737667e-02 -8.80845487e-01 -4.79980439e-01 -8.23579550e-01
-3.03340815e-02 -8.53199586e-02 1.60334513e-01 -3.61132413e-01
2.00302135e-02 1.32525459e-01 -1.78483987e+00 -3.11727732e-01
1.26085508e+00 1.00654936e+00 -2.57136412e-02 2.97945321e-01
4.88116592e-02 1.23821609e-01 -3.41338009e-01 -4.27015156e-01
-9.07591879e-02 -5.36912918e-01 -7.39553571e-02 -1.47939771e-01
3.05415750e-01 -6.05057061e-01 -5.29550612e-01 8.84286880e-01
6.07545316e-01 -2.16505080e-01 7.81808436e-01 -1.27063012e+00
-9.08052444e-01 5.28708100e-01 -8.44438314e-01 -4.91962075e-01
3.38558763e-01 5.14431186e-02 6.41261578e-01 -5.65616429e-01
4.48277861e-01 1.43938875e+00 7.95331717e-01 9.65114295e-01
-9.86781776e-01 -1.06593263e+00 4.83496189e-01 2.40955144e-01
-1.42165685e+00 -2.58307278e-01 7.31324613e-01 -3.53154212e-01
-3.19320172e-01 2.20332682e-01 1.27063811e+00 1.07200801e+00
7.23636031e-01 7.56298661e-01 1.04052556e+00 -2.26884589e-01
2.23012082e-02 -1.98598281e-01 1.04517698e-01 7.82383263e-01
6.98992014e-01 1.76180393e-01 -7.15930521e-01 3.29259560e-02
6.19719148e-01 1.31211862e-01 -3.66860092e-01 -4.33380008e-02
-1.06098223e+00 8.16808864e-02 1.63326249e-01 1.78432480e-01
-4.09274139e-02 1.66836157e-02 2.48612672e-01 -8.07483941e-02
3.04220229e-01 -1.67273477e-01 -4.87247407e-01 2.39115015e-01
-8.46189797e-01 5.25646210e-01 4.53774780e-01 9.59172726e-01
1.11314905e+00 7.08201602e-02 -1.86837047e-01 7.81504333e-01
5.73665917e-01 7.03165293e-01 6.99877441e-01 -8.76337230e-01
-3.24160218e-01 5.78910589e-01 1.68687794e-02 -1.12545753e+00
-4.49324667e-01 -5.44676423e-01 -1.18858016e+00 1.76128179e-01
3.56336445e-01 -1.84492648e-01 -8.38681579e-01 2.27866316e+00
5.59759676e-01 2.77328193e-01 -1.44473821e-01 3.71653795e-01
5.60794652e-01 5.03649056e-01 4.72856492e-01 -7.25186706e-01
1.67114449e+00 -3.81977022e-01 -9.90193546e-01 -3.75057086e-02
-3.98941524e-03 -6.87123597e-01 8.78766358e-01 3.90818685e-01
-8.66270185e-01 -6.91089988e-01 -1.45059228e+00 1.48160517e-01
-8.27094447e-03 1.34282976e-01 5.17910600e-01 9.23865736e-01
-7.52244711e-01 8.18773210e-01 -5.08798599e-01 -6.23991430e-01
2.61140615e-01 7.66655132e-02 -2.67274845e-02 6.76297247e-02
-1.37859547e+00 4.52123046e-01 2.64290452e-01 1.29389510e-01
-7.51400888e-01 -1.18832719e+00 -5.56733727e-01 -3.45878124e-01
-2.19707400e-01 -1.04640937e+00 9.98346746e-01 -7.04880714e-01
-1.40734684e+00 6.20855927e-01 -3.66221994e-01 7.05099339e-03
6.64897084e-01 -4.91266370e-01 -7.76431322e-01 -2.10604668e-01
-1.40534062e-02 5.24510562e-01 8.83014083e-01 -1.48772049e+00
-5.05402446e-01 -7.09868491e-01 -2.51520395e-01 5.44521138e-02
-9.58324313e-01 -1.86031029e-01 -2.28125721e-01 -7.68539011e-01
2.98463285e-01 -7.96894729e-01 9.78961214e-03 6.26843512e-01
-3.97580802e-01 2.34848976e-01 1.13294387e+00 -1.14375889e+00
1.73701131e+00 -2.07824874e+00 1.81695268e-01 1.10888921e-01
4.76959229e-01 -1.45866826e-01 1.98279947e-01 3.98897618e-01
9.30136144e-02 1.74866647e-01 -1.76945567e-01 -1.00314170e-01
-1.31460845e-01 8.53817612e-02 -5.45091694e-03 3.09422553e-01
-2.91485876e-01 5.49858153e-01 -7.18162417e-01 -8.48681033e-01
-2.99231589e-01 4.76705909e-01 -3.68522763e-01 4.28960919e-02
-1.91465154e-01 5.41614890e-01 -5.41912913e-01 7.48307526e-01
1.00779045e+00 2.74378359e-01 1.04968950e-01 -6.61165535e-01
-3.36407036e-01 -7.49119878e-01 -9.83399808e-01 1.51220059e+00
-2.19416633e-01 -1.18675284e-01 4.82557416e-02 -2.58588970e-01
1.36605072e+00 2.14018449e-01 7.10816145e-01 -4.50685799e-01
1.19296342e-01 3.58356684e-01 -1.95408985e-01 -2.12213308e-01
3.54543597e-01 -3.18745106e-01 2.62160331e-01 4.10970926e-01
-3.63926977e-01 2.58703679e-01 9.97580364e-02 -2.36171223e-02
4.73819673e-01 2.49214873e-01 1.94308847e-01 -6.09353662e-01
7.69509256e-01 -6.37140691e-01 1.19674194e+00 -5.93844615e-02
-2.54210293e-01 2.96909124e-01 2.36162931e-01 -4.17720407e-01
-1.18221998e+00 -1.28805494e+00 -2.47808158e-01 5.97756147e-01
6.59986854e-01 -4.39635485e-01 -9.10675049e-01 -7.76768178e-02
1.24393299e-01 2.89907008e-01 -6.92297995e-01 -7.63619602e-01
-4.03915435e-01 -7.60085106e-01 5.94016910e-01 3.12225699e-01
1.04280484e+00 -6.32253885e-01 -1.07476451e-01 2.16017678e-01
3.11552975e-02 -5.93635499e-01 -8.53120804e-01 -7.98580825e-01
-1.04137540e+00 -9.23929274e-01 -1.07907152e+00 -7.12391198e-01
7.22896636e-01 -1.45303518e-01 7.29460776e-01 3.67486805e-01
-4.25941020e-01 1.78677753e-01 -2.86876351e-01 -4.74944562e-01
-2.19239384e-01 -2.83051678e-03 6.36054277e-01 3.43700439e-01
-1.40572086e-01 -1.07505143e+00 -1.23011136e+00 3.66745412e-01
-6.48617506e-01 2.02072769e-01 3.59148979e-01 5.95167339e-01
4.14490104e-01 3.10670823e-01 6.97804749e-01 -4.20304775e-01
6.20910048e-01 -1.86725408e-01 1.35443863e-02 4.55173731e-01
-1.01845860e+00 -7.03240260e-02 7.15300679e-01 -6.09310150e-01
-1.32274461e+00 -5.65122701e-02 -3.65693797e-03 -9.24743488e-02
2.98676223e-01 1.40478713e-02 -7.25342691e-01 -6.71552792e-02
4.60893422e-01 4.06462401e-01 2.87013859e-01 -4.53453541e-01
3.62104893e-01 4.22349364e-01 5.84763408e-01 -1.08813310e+00
1.09606743e+00 3.63164842e-01 2.25650877e-01 -8.79112184e-01
-4.27393287e-01 4.69650477e-01 -3.67547601e-01 -8.31185162e-01
8.73243868e-01 -6.98965490e-01 -9.42485690e-01 1.36026335e+00
-9.06688869e-01 1.02322297e-02 -4.21238318e-02 3.05425465e-01
-3.07667345e-01 6.33436263e-01 -9.80507970e-01 -1.01831436e+00
-5.32621324e-01 -5.81280470e-01 7.73909628e-01 9.35399771e-01
-3.19142967e-01 -8.38663459e-01 8.88734162e-02 -1.88351408e-01
2.44016007e-01 4.33975369e-01 1.19553471e+00 3.01167607e-01
-3.36036891e-01 1.67509824e-01 -2.41905842e-02 3.05363297e-01
4.89253014e-01 7.20809102e-01 -8.84760678e-01 -2.77095377e-01
-3.22742105e-01 2.16682762e-01 5.45754969e-01 2.64019191e-01
1.04525459e+00 -4.04504597e-01 -5.02662599e-01 6.28489494e-01
1.24775910e+00 5.23583829e-01 8.33491981e-01 3.38246338e-02
6.00084424e-01 8.55319858e-01 5.35353899e-01 6.22339785e-01
3.88925612e-01 3.09480846e-01 1.32730499e-01 -1.13479145e-01
-1.26228511e-01 -8.15583825e-01 3.36583555e-01 1.04677272e+00
-3.38785350e-01 1.59916461e-01 -5.49574971e-01 4.10469808e-02
-1.57530177e+00 -8.88138771e-01 4.10110597e-03 2.25061703e+00
9.94931698e-01 -1.39013737e-01 1.34523764e-01 1.59148023e-01
1.10239100e+00 1.80007704e-02 -1.07832706e+00 1.27481520e-01
-3.09363544e-01 -3.29234526e-02 3.32100332e-01 1.69217139e-01
-5.69431245e-01 5.85801661e-01 7.00626135e+00 7.18028903e-01
-9.35720444e-01 -2.68432498e-01 8.74714017e-01 1.24827653e-01
-7.25461364e-01 3.35004702e-02 -8.55536282e-01 8.06572676e-01
2.66011715e-01 -9.06826735e-01 3.86362642e-01 5.81487715e-01
3.42969209e-01 1.58459961e-01 -6.78818882e-01 8.08641672e-01
-1.61487907e-01 -7.94465125e-01 2.82635331e-01 -7.16966251e-03
5.29833257e-01 -1.15551341e+00 1.64613694e-01 4.27151173e-02
2.26258882e-03 -6.20310664e-01 9.18281436e-01 1.09446561e+00
1.09589052e+00 -6.92168176e-01 1.92546830e-01 1.40812665e-01
-1.75398219e+00 -1.48885816e-01 -8.26120451e-02 2.64718343e-04
1.61941588e-01 8.08765709e-01 2.02338323e-01 7.96500146e-01
6.84898317e-01 9.38318312e-01 -5.46933472e-01 9.50338125e-01
-4.78294007e-02 3.42371166e-01 -1.02713279e-01 2.12515388e-02
-7.38583922e-01 -7.23211288e-01 1.88982174e-01 6.00906730e-01
1.00146544e+00 1.57661751e-01 -1.88961979e-02 6.21345401e-01
1.39594972e-01 3.19822520e-01 -2.14599356e-01 9.53197330e-02
9.66792583e-01 1.08020091e+00 -3.32448244e-01 -1.17543794e-01
-2.91186217e-02 8.06126237e-01 -2.23723561e-01 4.88807350e-01
-8.25462759e-01 -3.37896198e-01 1.07138455e+00 5.15689373e-01
-3.97291839e-01 -1.52034611e-01 -2.43598089e-01 -9.52298105e-01
6.41462067e-03 -6.79401875e-01 2.65259445e-01 -7.80800939e-01
-1.30769563e+00 2.93376595e-01 -2.53034562e-01 -1.22202396e+00
5.18837690e-01 -4.47124809e-01 -9.60930586e-01 7.79759228e-01
-9.70832586e-01 -1.35073721e+00 -7.05340743e-01 4.39705491e-01
1.69957448e-02 -1.28336355e-01 5.57783544e-01 5.79060197e-01
-8.87152791e-01 7.61015594e-01 1.59211867e-02 -3.82210851e-01
9.09580529e-01 -8.06880355e-01 4.81425487e-02 5.99849343e-01
-7.40118921e-01 9.82158065e-01 8.73720050e-01 -9.64444101e-01
-1.30033863e+00 -8.67804825e-01 4.76694733e-01 2.64404267e-01
5.39677441e-01 -1.10501021e-01 -8.51431310e-01 4.24865991e-01
-9.32527930e-02 -6.22833014e-01 4.75014240e-01 2.49299496e-01
-4.86969799e-01 -9.48565185e-01 -1.12888610e+00 1.10100091e+00
1.55456483e+00 -1.97022825e-01 -1.21545687e-01 -5.13582267e-02
9.03395653e-01 -1.50710523e-01 -9.96543169e-01 6.11784518e-01
1.35170829e+00 -8.98502231e-01 8.43936622e-01 -6.00214750e-02
3.66652638e-01 -6.87934399e-01 1.57482237e-01 -1.07587934e+00
-5.95074177e-01 -6.82038903e-01 -2.29305193e-01 1.79748058e+00
-5.72063308e-03 -8.82602632e-01 7.85125434e-01 4.68051553e-01
-6.72577024e-02 -8.94347250e-01 -7.51338840e-01 -8.08307827e-01
5.47726043e-02 2.87530601e-01 1.19052017e+00 6.57591581e-01
-2.07511485e-01 1.13101810e-01 -6.98485911e-01 4.26171310e-02
9.00134563e-01 -2.38351166e-01 5.98289073e-01 -1.44462204e+00
-5.34872748e-02 -3.69239509e-01 -4.00222510e-01 -9.45541322e-01
-7.43247345e-02 -4.26345170e-01 -1.70354411e-01 -1.24593067e+00
2.97619671e-01 -6.23902380e-01 -3.30000728e-01 3.11757047e-02
-4.17842776e-01 -1.71245739e-01 -8.14097300e-02 2.00512975e-01
7.10618049e-02 8.73583376e-01 1.88188756e+00 -1.00841098e-01
3.07956133e-02 -2.10611612e-01 -8.95993054e-01 5.84204555e-01
8.71689260e-01 1.80907980e-01 -5.45961916e-01 9.60530248e-03
1.81019843e-01 -7.93697983e-02 5.17500713e-02 -1.29674184e+00
5.76684773e-02 -2.21956685e-01 5.41513026e-01 -3.93460274e-01
2.75536269e-01 -5.71758628e-01 7.19387412e-01 1.01054776e+00
2.20902890e-01 -8.32147002e-02 2.45424472e-02 6.53749049e-01
1.60527080e-01 6.05505146e-02 8.57645214e-01 8.23535100e-02
-5.24604261e-01 1.01233816e+00 -1.37241930e-01 -3.14728618e-01
1.12270081e+00 -4.56684858e-01 -4.40956533e-01 -2.86909461e-01
-6.44801021e-01 3.70457411e-01 9.46485281e-01 2.76283264e-01
3.78800094e-01 -1.70820761e+00 -7.24089861e-01 2.34092787e-01
9.89299193e-02 -3.59358937e-01 8.14551771e-01 4.10817981e-01
-4.85631675e-01 -3.67577791e-01 -7.08025217e-01 -2.65196711e-01
-1.33773601e+00 6.00756824e-01 3.41251165e-01 1.17282085e-01
-3.25264454e-01 5.49806356e-01 4.41760004e-01 -1.14456557e-01
-1.28792331e-01 8.05921331e-02 -3.38884443e-01 2.77851254e-01
5.83636940e-01 6.65163815e-01 -6.10837698e-01 -3.98773938e-01
-1.85389310e-01 1.33123016e+00 -4.15564515e-02 1.19817629e-01
8.49247158e-01 -3.14698964e-01 -4.14505810e-01 6.12298429e-01
6.12692654e-01 1.90154180e-01 -1.26276982e+00 1.61505803e-01
-4.59576696e-01 -4.45261300e-01 -6.10030591e-01 -4.39686149e-01
-1.15792441e+00 4.90859926e-01 7.27426112e-01 2.46926807e-02
1.46267653e+00 -2.98738927e-01 1.15795600e+00 -3.07205021e-01
7.12594271e-01 -1.07348621e+00 -1.44593953e-03 5.77871762e-02
7.08062410e-01 -3.84506375e-01 3.32557887e-01 -6.73041940e-01
-1.64348513e-01 9.15205777e-01 8.95370722e-01 -5.98807633e-03
7.35562265e-01 3.98354381e-02 -8.39603916e-02 3.05776477e-01
-6.12470448e-01 3.32970321e-01 -7.91056603e-02 8.39971304e-01
5.05599618e-01 2.54776835e-01 -1.01023781e+00 8.95009577e-01
-4.11130816e-01 6.59921914e-02 1.11726284e-01 4.44212973e-01
-5.92535675e-01 -1.40421784e+00 -4.19221371e-01 5.00496924e-01
-5.98359201e-03 1.71068385e-01 -8.28531757e-02 5.62359750e-01
6.92818820e-01 6.16297543e-01 -9.10615325e-02 -7.78494895e-01
3.01602483e-01 7.59818479e-02 5.64326942e-01 -4.14091870e-02
-6.08540103e-02 -2.35554442e-01 -1.13640897e-01 -4.13639516e-01
-7.83268958e-02 -8.35061908e-01 -1.11013186e+00 -8.11667740e-01
-2.29576249e-02 -7.81507045e-02 2.85677642e-01 4.86153007e-01
2.04975918e-01 5.77482343e-01 7.49657393e-01 -2.65123665e-01
-1.97113767e-01 -6.29978955e-01 -1.15145421e+00 3.62541050e-01
-3.67855020e-02 -8.35334659e-01 -3.41105521e-01 3.42442304e-01] | [13.146327018737793, 0.43954065442085266] |
1c704b56-414d-47bf-b3d5-d41d8b78664b | neural-inventory-control-in-networks-via | 2306.11246 | null | https://arxiv.org/abs/2306.11246v1 | https://arxiv.org/pdf/2306.11246v1.pdf | Neural Inventory Control in Networks via Hindsight Differentiable Policy Optimization | Inventory management offers unique opportunities for reliably evaluating and applying deep reinforcement learning (DRL). Rather than evaluate DRL algorithms by comparing against one another or against human experts, we can compare to the optimum itself in several problem classes with hidden structure. Our DRL methods consistently recover near-optimal policies in such settings, despite being applied with up to 600-dimensional raw state vectors. In others, they can vastly outperform problem-specific heuristics. To reliably apply DRL, we leverage two insights. First, one can directly optimize the hindsight performance of any policy using stochastic gradient descent. This uses (i) an ability to backtest any policy's performance on a subsample of historical demand observations, and (ii) the differentiability of the total cost incurred on any subsample with respect to policy parameters. Second, we propose a natural neural network architecture to address problems with weak (or aggregate) coupling constraints between locations in an inventory network. This architecture employs weight duplication for ``sibling'' locations in the network, and state summarization. We justify this architecture through an asymptotic guarantee, and empirically affirm its value in handling large-scale problems. | ['Yash Kanoria', 'Daniel Russo', 'Matias Alvo'] | 2023-06-20 | null | null | null | null | ['management'] | ['miscellaneous'] | [-1.20173067e-01 -2.05191299e-02 -6.80825830e-01 -2.17945844e-01
-7.41884053e-01 -8.45669985e-01 3.04233819e-01 1.31000236e-01
-6.02051318e-01 1.04246294e+00 1.89818531e-01 -7.67196476e-01
-4.79844064e-01 -6.00877464e-01 -1.07673371e+00 -7.05758333e-01
-5.20269096e-01 9.53515351e-01 -1.01569660e-01 -2.41534784e-01
4.42975760e-01 5.98689437e-01 -1.19812536e+00 -8.14647079e-02
7.63665617e-01 1.25267017e+00 2.61918247e-01 5.55337131e-01
1.15979433e-01 1.00613856e+00 -8.31082761e-01 -1.10208280e-01
5.23081362e-01 -1.21126093e-01 -6.81956708e-01 1.00944877e-01
3.18241417e-01 -7.68329382e-01 -4.03988570e-01 9.99154627e-01
5.38450360e-01 3.17648053e-01 4.46272582e-01 -1.28880084e+00
-6.13877237e-01 9.48584378e-01 -6.19564235e-01 3.64530206e-01
-1.21264532e-02 6.83063447e-01 1.38908815e+00 -2.31800489e-02
3.99481922e-01 1.26957798e+00 4.80988264e-01 2.79872566e-01
-1.44006109e+00 -3.65920335e-01 7.82848060e-01 3.73972207e-02
-7.66867340e-01 -4.18237388e-01 5.13776302e-01 -1.98408470e-01
1.21558499e+00 -8.16119164e-02 4.62989569e-01 1.20086396e+00
3.01274955e-01 1.01532757e+00 1.03956974e+00 -6.02621175e-02
6.46384299e-01 -4.97198403e-02 -1.20230787e-01 5.87238193e-01
3.54541004e-01 3.82922500e-01 -2.56077379e-01 -1.51464492e-01
7.21368730e-01 -6.94277212e-02 -1.67460006e-03 -3.86258841e-01
-1.00674534e+00 8.25459063e-01 3.13169271e-01 -3.22004050e-01
-6.29909456e-01 5.94975114e-01 4.41312224e-01 6.06479883e-01
1.65649951e-01 1.26027620e+00 -8.04616153e-01 -1.96917951e-01
-8.23241293e-01 4.33007270e-01 1.03363168e+00 9.67849791e-01
4.19224709e-01 4.81741607e-01 -2.47822925e-01 4.13476199e-01
-1.42459288e-01 6.76031113e-01 3.69808823e-01 -1.50513196e+00
7.23291934e-01 3.90435234e-02 6.56307936e-01 -8.37730348e-01
-6.08224690e-01 -7.32062399e-01 -5.21482468e-01 2.38122180e-01
4.42359149e-01 -5.99439800e-01 -7.08440900e-01 2.07739305e+00
-3.84536083e-03 -1.65462986e-01 -6.18870836e-03 8.05394053e-01
-3.81254315e-01 7.05654860e-01 -2.74908602e-01 -4.26454842e-01
9.18970704e-01 -9.34523821e-01 -3.40337545e-01 -5.41876674e-01
5.19508302e-01 -2.53761351e-01 1.01928949e+00 6.06966317e-01
-1.45595503e+00 -1.74538538e-01 -9.57829297e-01 5.07599354e-01
-1.68092251e-01 -1.90366089e-01 6.82017326e-01 1.67051509e-01
-1.11512935e+00 9.80060875e-01 -9.30553496e-01 4.06450545e-03
2.55257279e-01 6.25177443e-01 2.62230754e-01 1.88390493e-01
-1.16207683e+00 1.07826781e+00 3.97190303e-01 2.11711019e-01
-1.29243207e+00 -5.89585245e-01 -5.57734072e-01 4.46765989e-01
1.12641788e+00 -5.17127454e-01 1.79858851e+00 -9.71914589e-01
-1.56011689e+00 3.80915701e-02 2.23857731e-01 -8.95181835e-01
5.65307021e-01 -2.98587829e-01 -2.55625367e-01 6.38807938e-02
6.28743023e-02 3.58803749e-01 6.25998616e-01 -1.19595683e+00
-9.76778090e-01 -2.35696718e-01 4.03650671e-01 2.17865542e-01
-1.29295319e-01 -4.03440356e-01 -1.67163223e-01 -4.81419414e-01
-1.62069216e-01 -9.14639354e-01 -5.98574221e-01 -4.62372273e-01
-6.01082087e-01 -1.18880168e-01 2.44721308e-01 -4.99313354e-01
1.12647235e+00 -1.73236871e+00 1.03777938e-01 5.15067518e-01
-2.98923589e-02 4.37963419e-02 -3.17527473e-01 5.08570194e-01
2.46901974e-01 3.83043438e-02 9.43403989e-02 -9.40339118e-02
4.46299136e-01 4.65807080e-01 -5.71696818e-01 4.60853547e-01
3.03899884e-01 9.15625572e-01 -1.13529408e+00 -1.02987468e-01
-1.92167182e-02 -4.42729801e-01 -8.29661191e-01 4.83893640e-02
-7.19514787e-01 1.18828110e-01 -4.40662414e-01 6.36607111e-01
1.14518248e-01 -3.23157787e-01 5.90759993e-01 6.36056587e-02
-1.08360043e-02 3.74164283e-01 -1.17233169e+00 1.21312737e+00
-5.66186786e-01 2.58933395e-01 2.83714205e-01 -1.33125472e+00
5.69577634e-01 1.23146707e-02 6.51082039e-01 -9.83182371e-01
8.17477256e-02 2.32655734e-01 3.83196846e-02 -3.78529578e-01
5.53610086e-01 -1.41855508e-01 -3.89953285e-01 7.51876056e-01
3.69867757e-02 9.04450491e-02 2.92438388e-01 1.04428660e-02
1.19636345e+00 -1.31965047e-02 9.52912588e-03 -3.63590658e-01
-1.81104764e-01 1.31370738e-01 7.40161002e-01 1.30758226e+00
-2.78488219e-01 -1.43757433e-01 1.08620834e+00 -4.70138252e-01
-1.20517004e+00 -1.22249258e+00 3.52198511e-01 1.33007812e+00
-2.87699886e-03 3.22553754e-01 -3.75071287e-01 -6.98851526e-01
7.33854711e-01 9.70273137e-01 -6.10439956e-01 -2.52134055e-01
-6.34397864e-01 -5.50036907e-01 3.10751617e-01 8.94036531e-01
2.10786834e-01 -9.61133540e-01 -6.58288181e-01 5.62741160e-01
3.13672125e-01 -8.92953277e-01 -6.41324699e-01 6.89912617e-01
-8.76134217e-01 -9.22044098e-01 -4.82541353e-01 -5.51996946e-01
4.45007324e-01 4.03164364e-02 1.34811676e+00 -2.61494964e-01
1.46654561e-01 2.77379572e-01 1.34618908e-01 -1.95219323e-01
-3.12646210e-01 5.45465827e-01 5.69837570e-01 -4.51381594e-01
9.28141549e-02 -5.94762683e-01 -6.38657987e-01 3.71398538e-01
-6.24160707e-01 -4.30699527e-01 8.78791451e-01 8.73134494e-01
4.27617878e-01 1.43096298e-01 8.35450649e-01 -7.85817623e-01
1.01233566e+00 -5.20435393e-01 -1.15629327e+00 4.16918695e-01
-8.73345196e-01 5.83642066e-01 9.64345753e-01 -4.83323067e-01
-6.20065749e-01 -2.45195881e-01 1.61110640e-01 -6.31797373e-01
1.97850883e-01 4.75689769e-01 1.08581334e-01 5.92400670e-01
3.82933736e-01 2.40677431e-01 2.46527106e-01 -4.48088229e-01
5.23877680e-01 2.82396376e-01 7.99199998e-01 -1.31254315e+00
4.66268271e-01 1.14703745e-01 -6.16259268e-03 -2.18721256e-01
-1.02146411e+00 -4.09755558e-02 -1.91690713e-01 3.61598469e-02
4.46345210e-01 -6.37460411e-01 -1.38090169e+00 9.99197587e-02
-6.49603963e-01 -7.95149386e-01 -5.58829606e-01 3.47938955e-01
-9.63271379e-01 -3.39713469e-02 -7.47816563e-01 -9.95231271e-01
-4.91533801e-03 -1.17391634e+00 7.18069792e-01 6.86364770e-02
4.90761399e-02 -1.02282941e+00 -4.01400924e-02 -8.33802000e-02
4.87932354e-01 1.46300584e-01 1.09814000e+00 -8.38303506e-01
-3.75252753e-01 -2.02078726e-02 1.18479598e-02 5.41864097e-01
7.85685107e-02 -4.98122647e-02 -5.21286607e-01 -7.79255867e-01
-1.71784699e-01 -5.08721292e-01 6.75650716e-01 5.75029492e-01
1.43258274e+00 -9.10718799e-01 -1.68617666e-01 4.73921150e-01
1.36075878e+00 4.13439959e-01 1.19059272e-02 6.43406212e-01
3.33517730e-01 5.59633970e-01 5.13247132e-01 6.92077398e-01
2.88482249e-01 4.40631419e-01 5.98231792e-01 1.20553918e-01
5.78241050e-01 -4.47877020e-01 4.49936628e-01 5.89730263e-01
1.38357654e-01 -2.84379065e-01 -7.44060934e-01 4.90983307e-01
-1.96482551e+00 -1.12607598e+00 7.47184396e-01 2.25652766e+00
8.05971265e-01 7.97642529e-01 5.24088144e-01 -3.91153663e-01
6.00758910e-01 1.91111296e-01 -1.43091893e+00 -8.94149899e-01
-1.09932929e-01 -2.05931962e-02 1.27101874e+00 4.70081151e-01
-7.55312085e-01 8.05397213e-01 7.48153210e+00 6.43346250e-01
-1.09070158e+00 -2.31606200e-01 8.71813059e-01 -7.20428407e-01
-2.71764696e-01 -2.76231587e-01 -8.95169914e-01 6.57632172e-01
1.18220985e+00 -1.57057390e-01 1.06423712e+00 1.00821316e+00
4.05916899e-01 -1.18540945e-02 -1.34944785e+00 3.23254138e-01
-3.77655298e-01 -1.46626472e+00 -2.30283644e-02 2.76383936e-01
9.60800350e-01 1.31368786e-01 3.53791803e-01 6.70115769e-01
1.18846381e+00 -9.75374520e-01 9.13968086e-01 2.83266664e-01
5.21148801e-01 -1.09768999e+00 5.62551498e-01 3.46853822e-01
-6.80559516e-01 -7.91027844e-01 -3.76066953e-01 -1.77036434e-01
1.50291741e-01 1.73131973e-01 -7.84264207e-01 2.53078878e-01
4.30982918e-01 2.09952533e-01 -1.20201327e-01 7.31961250e-01
-1.58841163e-01 5.03851414e-01 -5.49623609e-01 -1.83083594e-01
9.52001035e-01 -1.12665690e-01 2.03127339e-01 9.12697077e-01
3.22409486e-03 -2.53959000e-01 5.45313001e-01 6.64742351e-01
-1.58575680e-02 -4.49852943e-01 -5.04055738e-01 -2.85672307e-01
6.47587121e-01 7.20812440e-01 -4.43346918e-01 -3.23677450e-01
-2.01955944e-01 4.94341314e-01 3.85465503e-01 5.71729720e-01
-8.31534147e-01 -3.39499950e-01 1.03600574e+00 -1.61524639e-01
5.40693283e-01 -3.41274410e-01 -1.03802107e-01 -7.06130385e-01
3.67367417e-02 -1.11394596e+00 4.03906941e-01 -5.45904040e-01
-1.44756532e+00 1.39322162e-01 -2.06621736e-02 -8.70612323e-01
-5.77849448e-01 -8.57719302e-01 -4.03923184e-01 5.52384138e-01
-1.52686667e+00 -3.71048808e-01 6.16251707e-01 3.25295657e-01
5.51554382e-01 -1.98963836e-01 3.62070680e-01 4.66258302e-02
-9.52302873e-01 8.04777145e-01 7.36719787e-01 -2.59456336e-02
4.00376588e-01 -1.60631955e+00 5.28597295e-01 6.03565335e-01
-3.10051203e-01 5.43898761e-01 9.35854256e-01 -4.35448378e-01
-1.83080757e+00 -9.62219238e-01 1.17566660e-01 -3.55979800e-01
1.09074748e+00 -1.53273106e-01 -5.50847888e-01 9.23637509e-01
3.63938808e-02 -3.28978688e-01 1.61230460e-01 3.61769259e-01
-2.76178479e-01 -3.97271544e-01 -1.01739764e+00 6.69885278e-01
1.06030285e+00 -2.54529655e-01 -4.06928658e-01 4.62916791e-01
9.50550497e-01 -3.32377136e-01 -9.43499625e-01 1.18848294e-01
5.42372942e-01 -7.14370668e-01 7.79426336e-01 -1.40614736e+00
4.84891146e-01 4.72369827e-02 -3.43542099e-01 -1.60449493e+00
-4.35875267e-01 -9.59758341e-01 -4.85403180e-01 8.09398532e-01
6.65012360e-01 -7.97269583e-01 8.21576953e-01 6.06576681e-01
-1.68363363e-01 -1.19979274e+00 -7.58857787e-01 -1.25905252e+00
3.67363065e-01 -7.59189054e-02 8.69974613e-01 7.26099789e-01
-1.90957412e-01 1.99301079e-01 -5.13217211e-01 3.30584973e-01
5.06391406e-01 3.16497743e-01 5.18600583e-01 -8.24821055e-01
-8.38228285e-01 -8.82655501e-01 8.11502710e-02 -1.35903561e+00
2.83601552e-01 -4.86331880e-01 2.74528593e-01 -1.40418649e+00
-1.20790236e-01 -6.81243062e-01 -7.41600573e-01 4.99110043e-01
1.75108835e-01 -5.58440864e-01 2.22726390e-01 1.06992900e-01
-9.02614594e-01 3.52529138e-01 1.18841660e+00 -1.66620642e-01
-2.52523422e-01 2.24608153e-01 -9.89245474e-01 4.55062926e-01
1.08557999e+00 -2.70244896e-01 -4.78462130e-01 -7.08474696e-01
4.62647080e-01 3.16129714e-01 -7.89410342e-03 -6.00248337e-01
2.04896897e-01 -7.28325963e-01 3.25772732e-01 -4.72511381e-01
7.61960745e-02 -5.40822625e-01 -2.35897899e-01 6.96567595e-01
-6.71553969e-01 7.35799372e-01 1.36460543e-01 5.96036494e-01
3.17178994e-01 -1.98775634e-01 6.02127254e-01 -4.06284422e-01
-6.67834997e-01 1.71835333e-01 -4.29449856e-01 4.17575389e-01
8.72620881e-01 9.72077623e-02 -5.71396053e-01 -5.29907823e-01
-4.41000968e-01 1.01124728e+00 4.85405177e-01 1.97787225e-01
1.60072029e-01 -1.03437495e+00 -3.52309257e-01 -1.70575287e-02
-3.29646587e-01 -9.44078267e-02 -1.04799636e-01 7.19412923e-01
-3.26540887e-01 5.87928951e-01 -3.60901535e-01 -3.82578313e-01
-2.03323826e-01 1.00248814e+00 5.16364634e-01 -5.28060257e-01
-3.57313722e-01 6.27969146e-01 -1.60151452e-01 -2.28896365e-01
6.28240764e-01 -4.60709989e-01 4.04479831e-01 1.42187446e-01
3.37470978e-01 4.01342064e-01 -2.59345230e-02 2.28810057e-01
-1.45860031e-01 -3.02900188e-02 -3.02323222e-01 -3.32835585e-01
1.47591567e+00 3.13652456e-02 3.33414078e-01 4.58885789e-01
1.00764060e+00 -3.61957312e-01 -1.85423148e+00 -3.52588177e-01
2.66966790e-01 -2.54076689e-01 1.03112467e-01 -9.98247623e-01
-1.31911039e+00 5.36983550e-01 3.02112520e-01 6.88873231e-01
9.31248724e-01 -2.75362611e-01 7.00034082e-01 9.94685709e-01
4.08017635e-01 -1.65312433e+00 4.23546843e-02 5.32559216e-01
4.48083937e-01 -1.08720720e+00 7.85977915e-02 7.18016565e-01
-8.47117782e-01 8.71450484e-01 5.84958196e-01 -3.50514621e-01
2.44658589e-01 3.19189787e-01 -1.29854769e-01 -7.13574365e-02
-1.27540469e+00 -1.25091970e-01 -3.22586358e-01 4.66256529e-01
-1.61170542e-01 2.61276871e-01 1.85334712e-01 2.44114533e-01
-2.33133763e-01 -3.42883974e-01 3.82446676e-01 1.08541048e+00
-7.90873408e-01 -7.06958592e-01 -1.12986207e-01 8.45741868e-01
-3.74338388e-01 1.65314332e-01 -1.09840184e-01 8.40314090e-01
-4.79291499e-01 7.91910827e-01 4.13976312e-01 -1.99700519e-01
4.86493707e-01 -2.07829192e-01 4.20013636e-01 -3.41914594e-01
-4.25990850e-01 -3.58685143e-02 1.88532144e-01 -8.93824697e-01
1.82360113e-01 -6.87742472e-01 -8.29685628e-01 -6.65601313e-01
-2.64380760e-02 1.85300633e-01 6.22772455e-01 9.14267361e-01
4.05353278e-01 6.09388173e-01 1.05049098e+00 -8.12649965e-01
-1.79327118e+00 -5.84738851e-01 -7.23853528e-01 8.96447673e-02
5.52249134e-01 -7.63320208e-01 -3.39769721e-01 -6.61605656e-01] | [4.215375900268555, 2.3922066688537598] |
a2be6792-a350-4652-8e00-3d7832c6e067 | audio-transformers-transformer-architectures | 2105.00335 | null | https://arxiv.org/abs/2105.00335v1 | https://arxiv.org/pdf/2105.00335v1.pdf | Audio Transformers:Transformer Architectures For Large Scale Audio Understanding. Adieu Convolutions | Over the past two decades, CNN architectures have produced compelling models of sound perception and cognition, learning hierarchical organizations of features. Analogous to successes in computer vision, audio feature classification can be optimized for a particular task of interest, over a wide variety of datasets and labels. In fact similar architectures designed for image understanding have proven effective for acoustic scene analysis. Here we propose applying Transformer based architectures without convolutional layers to raw audio signals. On a standard dataset of Free Sound 50K,comprising of 200 categories, our model outperforms convolutional models to produce state of the art results. This is significant as unlike in natural language processing and computer vision, we do not perform unsupervised pre-training for outperforming convolutional architectures. On the same training set, with respect mean aver-age precision benchmarks, we show a significant improvement. We further improve the performance of Transformer architectures by using techniques such as pooling inspired from convolutional net-work designed in the past few years. In addition, we also show how multi-rate signal processing ideas inspired from wavelets, can be applied to the Transformer embeddings to improve the results. We also show how our models learns a non-linear non constant band-width filter-bank, which shows an adaptable time frequency front end representation for the task of audio understanding, different from other tasks e.g. pitch estimation. | ['Jonathan Berger', 'Prateek Verma'] | 2021-05-01 | null | null | null | null | ['unsupervised-pre-training'] | ['methodology'] | [ 1.33098543e-01 5.08214300e-03 4.10441071e-01 -3.90825123e-01
-7.48498380e-01 -5.61523974e-01 4.62393016e-01 1.33532807e-01
-5.43342173e-01 1.29398674e-01 5.45513332e-01 -5.46855479e-02
-1.70204416e-01 -8.00821126e-01 -6.22650266e-01 -4.87902194e-01
-5.74994028e-01 -1.09702908e-01 3.40751112e-01 -3.63196224e-01
2.27080837e-01 2.85646081e-01 -1.82354164e+00 7.20619917e-01
-2.65436530e-01 1.41247797e+00 5.16114458e-02 1.06680882e+00
8.32379051e-03 6.47421479e-01 -5.95233440e-01 -2.21116856e-01
1.68889761e-02 -4.63268347e-02 -9.65503454e-01 -3.61499131e-01
6.70037210e-01 2.93625779e-02 -4.09769565e-01 7.35616624e-01
7.96674311e-01 2.45111495e-01 4.36575621e-01 -1.04071426e+00
-7.40683138e-01 9.83880997e-01 8.61342698e-02 4.93333668e-01
2.83346981e-01 -7.75981843e-02 1.55101752e+00 -1.13151181e+00
-4.49988097e-02 1.51535416e+00 8.64049137e-01 5.46913624e-01
-1.17285848e+00 -8.10333371e-01 1.02266692e-01 5.06737113e-01
-1.13385820e+00 -5.36274672e-01 8.53221178e-01 -3.55146080e-01
1.22856224e+00 2.01495171e-01 7.53860414e-01 1.00476277e+00
1.86537147e-01 6.40027702e-01 7.93477476e-01 -5.26180625e-01
1.42890871e-01 -1.23384736e-01 1.12735622e-01 4.32851076e-01
-5.05084634e-01 2.07458124e-01 -1.07171535e+00 2.25195549e-02
4.79428917e-01 -1.92056283e-01 -3.17666411e-01 -2.20146868e-02
-1.38067794e+00 8.06501269e-01 6.87944710e-01 5.91407120e-01
9.43790078e-02 7.31260300e-01 8.31124187e-01 6.14747345e-01
4.67101872e-01 7.51804948e-01 -8.06992948e-01 -2.83831805e-01
-8.30519676e-01 2.00569913e-01 7.56117761e-01 4.61273223e-01
4.81057376e-01 5.17601132e-01 -2.45370436e-02 9.48086977e-01
1.69755995e-01 5.18875569e-02 8.66930366e-01 -1.03707826e+00
1.10051602e-01 -6.94674775e-02 -3.96408945e-01 -8.20655048e-01
-6.92351043e-01 -7.30507731e-01 -7.28049755e-01 1.07474066e-01
4.26922619e-01 2.07426831e-01 -9.03748333e-01 1.77167416e+00
-2.82698691e-01 4.29524004e-01 1.10091932e-01 8.02239180e-01
9.05703306e-01 8.13437462e-01 -2.01306120e-02 2.43067190e-01
1.68176377e+00 -7.96365798e-01 -6.03983760e-01 -9.48214978e-02
3.05383682e-01 -9.51182187e-01 1.30217814e+00 7.98412561e-01
-9.06150579e-01 -9.25547183e-01 -1.07706046e+00 -3.28090310e-01
-6.55230522e-01 -1.28984004e-01 7.75487840e-01 6.72359943e-01
-1.24635828e+00 7.82940447e-01 -7.06564009e-01 -3.51534516e-01
5.70946157e-01 4.85463858e-01 -1.84069857e-01 4.44126248e-01
-1.29544485e+00 6.70568705e-01 2.99882650e-01 -1.53381750e-01
-1.18972147e+00 -1.20239782e+00 -8.02139223e-01 3.98579448e-01
-8.56109038e-02 -4.90583718e-01 1.69056571e+00 -8.80139172e-01
-1.76732874e+00 5.47373772e-01 3.29582468e-02 -1.07032776e+00
-1.61722422e-01 -5.01607239e-01 -5.07739246e-01 1.35722846e-01
-2.36131877e-01 9.71988499e-01 9.92158234e-01 -6.06609881e-01
-6.25616729e-01 5.74503541e-02 2.64623076e-01 -1.73184246e-01
-8.24049532e-01 1.05130129e-01 1.67148516e-01 -8.34927380e-01
3.45309339e-02 -6.90533221e-01 -1.36867642e-01 -3.48186493e-02
4.82721590e-02 -4.57313895e-01 8.64979744e-01 -2.56363958e-01
1.02547383e+00 -2.48528886e+00 -2.25045718e-02 -8.97895321e-02
1.15822837e-01 -4.26812023e-02 -3.66201371e-01 3.89686197e-01
-3.05518776e-01 3.01643074e-01 -1.64205566e-01 -2.12532356e-01
2.56173074e-01 2.61015128e-02 -9.04181302e-01 1.33784398e-01
4.80689853e-01 6.03175282e-01 -8.29509735e-01 1.12933712e-02
1.93072349e-01 6.44328713e-01 -9.43125725e-01 6.07728623e-02
-1.61005318e-01 1.56678349e-01 4.68252314e-04 3.81628245e-01
1.96531370e-01 8.12641084e-02 -2.88053453e-01 -3.11519086e-01
-7.14913160e-02 7.54390240e-01 -9.04759467e-01 2.06559968e+00
-1.04366875e+00 1.26657665e+00 -3.82648073e-02 -1.23807883e+00
9.30756211e-01 7.79537916e-01 3.98163646e-01 -6.00771308e-01
-4.84218914e-03 1.43341556e-01 3.92630249e-01 -2.30326474e-01
5.43464661e-01 -3.92543614e-01 -1.18563846e-01 2.68131733e-01
6.95821881e-01 -6.74743354e-01 -1.16070308e-01 -8.81064683e-03
1.09179139e+00 -2.76105165e-01 9.56589133e-02 -3.32593590e-01
3.83497894e-01 -4.62538004e-01 1.13994487e-01 6.42707109e-01
-6.14849031e-02 7.94524968e-01 2.78735250e-01 -7.56424665e-01
-8.93754959e-01 -1.23320234e+00 -5.73665023e-01 1.76702452e+00
-4.39764202e-01 -7.82755435e-01 -5.92353046e-01 -7.34116808e-02
-1.11424617e-01 2.81521529e-01 -5.25943935e-01 -2.27554277e-01
-5.11313021e-01 -5.45874834e-01 1.03953302e+00 7.16654003e-01
2.76227444e-01 -1.36558366e+00 -7.43871689e-01 6.02565765e-01
1.19795784e-01 -1.16937053e+00 -1.25913158e-01 7.77853549e-01
-5.91472983e-01 -6.06017947e-01 -4.48908210e-01 -9.09551740e-01
-3.58166575e-01 1.17574381e-02 1.35149598e+00 -1.65198207e-01
-3.94551814e-01 7.09129989e-01 -3.90703470e-01 -8.77498329e-01
-2.31042221e-01 2.87661880e-01 3.76444370e-01 -6.44817129e-02
4.19103503e-01 -1.03086758e+00 -5.49172163e-01 4.45660576e-02
-8.85229230e-01 -4.41188544e-01 3.78610820e-01 8.18644047e-01
3.71592283e-01 6.91780169e-03 1.01739669e+00 -3.76545280e-01
6.65839791e-01 -3.04865688e-01 -1.39530987e-01 -2.67465532e-01
-1.62132144e-01 2.55242046e-02 8.02134752e-01 -5.94109356e-01
-3.77619952e-01 -7.15462351e-03 -5.76301575e-01 -4.10952955e-01
-2.97033370e-01 2.47311801e-01 2.28229329e-01 4.17119218e-03
7.65193939e-01 9.13831592e-02 -2.76441574e-01 -6.45276129e-01
6.08206868e-01 7.54017770e-01 6.10264480e-01 -5.04865885e-01
6.26699984e-01 4.17546898e-01 -1.27281532e-01 -1.08926153e+00
-1.08727336e+00 -3.89958650e-01 -4.96822178e-01 6.52841106e-02
1.01690078e+00 -9.84560609e-01 -9.75326717e-01 1.96713537e-01
-1.12836802e+00 -1.85015604e-01 -6.41472578e-01 7.27325022e-01
-8.02360415e-01 -1.05149217e-01 -7.63267696e-01 -6.90689802e-01
-1.85583696e-01 -9.51159835e-01 1.09543645e+00 4.92780358e-02
-3.66600066e-01 -9.89007711e-01 1.73999578e-01 -1.01778857e-01
8.08844864e-01 -1.70990273e-01 9.45583701e-01 -7.78739750e-01
-4.28173095e-01 1.29096493e-01 3.71218473e-02 7.66433179e-01
3.64154801e-02 -7.87269473e-02 -1.84764683e+00 -1.35498881e-01
1.06214494e-01 -6.26293600e-01 1.37304068e+00 4.08427209e-01
1.74792230e+00 -1.68602029e-03 2.17351139e-01 5.76416254e-01
1.00488651e+00 1.44146919e-01 4.09648627e-01 1.87158242e-01
5.45665503e-01 4.90750074e-01 1.33796066e-01 3.18182945e-01
2.23372310e-01 6.50350153e-01 4.63173270e-01 -1.34299293e-01
-3.70303601e-01 -2.02397585e-01 4.78564888e-01 1.29152262e+00
7.80841783e-02 4.63554710e-02 -7.95931041e-01 7.72044003e-01
-1.32275295e+00 -1.05018771e+00 3.08609575e-01 1.84655643e+00
7.53152192e-01 3.77844751e-01 6.13858216e-02 8.88824344e-01
1.32739618e-01 3.06853563e-01 -2.05139726e-01 -9.45520520e-01
2.63253041e-02 1.16934884e+00 1.05681598e-01 4.20147926e-01
-1.31001532e+00 9.66353655e-01 6.88721085e+00 8.66316378e-01
-1.40651035e+00 2.65934289e-01 2.41101101e-01 -2.20065981e-01
-1.72813937e-01 -1.67243704e-01 -4.22541261e-01 5.42926863e-02
1.44073784e+00 -5.74877262e-02 4.79068875e-01 7.59816468e-01
-1.11467347e-01 5.04133344e-01 -1.48401356e+00 1.01837099e+00
-4.24732491e-02 -1.46825635e+00 5.08407690e-02 -1.55768618e-01
2.54680306e-01 1.43226713e-01 4.53916728e-01 5.17303526e-01
1.86053127e-01 -1.42238498e+00 1.03689766e+00 3.62162977e-01
5.68196833e-01 -8.39732409e-01 4.38827097e-01 -1.04774244e-01
-1.60710609e+00 -3.93541098e-01 -6.60714924e-01 -4.87117171e-01
-8.85294303e-02 6.45494938e-01 -9.39887822e-01 2.99189538e-01
1.22249746e+00 8.95518839e-01 -4.40304846e-01 9.26313162e-01
-2.93348935e-02 9.75464761e-01 -3.38180572e-01 -2.86786985e-02
2.71289945e-01 5.05456805e-01 4.29401755e-01 1.58346093e+00
2.99361736e-01 -3.61490011e-01 -9.12144259e-02 5.07414341e-01
-2.01716408e-01 6.11376092e-02 -6.99731290e-01 -5.38311377e-02
4.20515239e-01 1.24002481e+00 -4.51101065e-01 -2.10546881e-01
-4.01630372e-01 4.01763171e-01 2.11288497e-01 2.51071721e-01
-5.39412677e-01 -6.36928439e-01 9.62416410e-01 -1.55660491e-02
5.64514935e-01 -3.90065402e-01 -1.13683805e-01 -7.71476269e-01
-2.70593882e-01 -8.47291946e-01 2.52703637e-01 -6.31013095e-01
-1.32122755e+00 8.27003479e-01 -2.69210726e-01 -1.31102872e+00
-3.35469723e-01 -8.80724847e-01 -6.69184566e-01 5.98423898e-01
-1.56882238e+00 -8.84314120e-01 8.79167169e-02 5.90306342e-01
8.17557693e-01 -3.56287837e-01 1.19659543e+00 3.80011082e-01
8.87581930e-02 5.14033735e-01 -2.49194086e-01 2.50696212e-01
7.25029290e-01 -1.49525321e+00 5.91504276e-01 5.35727203e-01
1.04402423e+00 3.89439613e-01 6.23744786e-01 4.54318374e-01
-1.12193727e+00 -9.42571163e-01 6.14243150e-01 -3.58720183e-01
1.09090054e+00 -8.16314638e-01 -8.99428666e-01 5.06412148e-01
3.65920216e-01 1.76950112e-01 9.19848859e-01 5.02084613e-01
-7.83225060e-01 -4.23538953e-01 -5.75975358e-01 2.44310305e-01
1.01248038e+00 -1.12570345e+00 -7.19591916e-01 2.07110658e-01
1.30149567e+00 -1.65979937e-01 -9.45241749e-01 3.34607184e-01
7.62737155e-01 -1.00702929e+00 1.29495525e+00 -7.48917937e-01
4.37750608e-01 -1.68783337e-01 -5.32769680e-01 -1.47781038e+00
-4.43484485e-01 -5.22126615e-01 1.02748483e-01 1.07909310e+00
3.11362118e-01 -4.24559951e-01 4.03006166e-01 -2.92248309e-01
-5.78301132e-01 -7.32569337e-01 -1.27006376e+00 -7.66738474e-01
3.85902882e-01 -1.13254821e+00 5.77168047e-01 6.00630164e-01
1.02831041e-02 4.22817111e-01 -2.37467587e-01 1.66624010e-01
2.69370645e-01 4.03073058e-03 5.70044458e-01 -1.40653300e+00
-4.39621270e-01 -6.43016040e-01 -9.49155450e-01 -1.04385662e+00
3.11594486e-01 -1.03887093e+00 1.48573712e-01 -1.08747232e+00
-4.14230198e-01 -2.39557907e-01 -8.53465438e-01 5.93635321e-01
3.49708706e-01 6.95453286e-01 3.19939315e-01 -1.47164479e-01
-2.32494399e-01 5.06991029e-01 1.04507601e+00 -3.05335641e-01
5.41354809e-03 -4.05914783e-02 -6.68179512e-01 9.33930397e-01
9.28323269e-01 -5.24880826e-01 -4.33291316e-01 -5.65379083e-01
2.78547078e-01 -2.09040597e-01 5.56826234e-01 -1.57523715e+00
1.39056012e-01 3.35135758e-01 3.10619056e-01 -1.91644713e-01
7.55180299e-01 -6.51430726e-01 -3.40997249e-01 4.84987080e-01
-7.60223508e-01 -7.18510523e-02 3.84537995e-01 4.18927342e-01
-7.08793461e-01 -6.92009777e-02 6.77769721e-01 -2.53145844e-02
-7.34250546e-01 1.29866332e-01 -5.23168445e-01 5.93240783e-02
3.23150933e-01 2.90115271e-02 -2.37666652e-01 -4.62511599e-01
-1.02550173e+00 -2.10565656e-01 -2.89540857e-01 7.08010972e-01
6.75714731e-01 -1.36492896e+00 -7.65794098e-01 2.46134102e-01
7.45997652e-02 -2.23527297e-01 1.06857769e-01 5.07327974e-01
-1.84423119e-01 4.34200495e-01 -2.65671909e-01 -8.63700867e-01
-1.01583290e+00 2.66824782e-01 4.42030281e-01 5.28475009e-02
-5.24949908e-01 1.13263690e+00 4.95976150e-01 -2.23823279e-01
5.10614693e-01 -1.04956126e+00 -2.90185362e-01 2.49798834e-01
6.92790747e-01 1.03674401e-02 3.12159061e-01 -2.90430576e-01
-4.64382470e-01 7.55847335e-01 1.07533067e-01 -3.31983119e-01
1.58002460e+00 2.24943042e-01 9.53636616e-02 1.01374876e+00
1.44281733e+00 3.58304307e-02 -1.00125527e+00 -1.56518996e-01
1.10320393e-02 -7.82954022e-02 1.56082332e-01 -5.57174385e-01
-1.11012280e+00 1.61837959e+00 8.54301453e-01 7.42298067e-01
1.21006048e+00 1.92932963e-01 6.12305641e-01 6.17953897e-01
2.21728474e-01 -8.55793118e-01 5.18378794e-01 9.03537869e-01
1.02229738e+00 -9.32187438e-01 -4.09293532e-01 3.95043641e-02
-2.60946542e-01 1.51116800e+00 1.24157108e-01 -4.60378230e-01
1.00436747e+00 4.53145236e-01 1.84740394e-01 -1.75298572e-01
-1.17472446e+00 -2.71169424e-01 4.22380507e-01 6.76468074e-01
8.24624181e-01 1.30922854e-01 2.76029468e-01 7.27886021e-01
-1.05865312e+00 -2.90011346e-01 5.60456336e-01 4.48190957e-01
-7.43939817e-01 -1.03105998e+00 -2.62197316e-01 3.80121350e-01
-7.97837138e-01 -3.08476210e-01 -1.30196944e-01 5.59015989e-01
1.61232218e-01 9.93274271e-01 4.33736503e-01 -6.90226793e-01
2.80792356e-01 3.67271781e-01 4.56901461e-01 -8.31425607e-01
-9.63007867e-01 -1.12264128e-02 -9.42629352e-02 -4.80953842e-01
-5.40263891e-01 -1.29520506e-01 -1.25426054e+00 8.47311541e-02
-7.65625387e-02 2.42259279e-01 6.69279277e-01 7.50823140e-01
7.30268052e-03 9.16079521e-01 6.59217358e-01 -1.14546084e+00
-4.57865626e-01 -1.18550730e+00 -5.11058807e-01 9.17463005e-02
7.55826354e-01 -5.44206262e-01 -5.37934065e-01 2.68692881e-01] | [15.423295021057129, 5.328564167022705] |
56915359-8b11-43bf-8ae9-98f1ac989305 | label-relation-graphs-enhanced-hierarchical | 2201.03194 | null | https://arxiv.org/abs/2201.03194v2 | https://arxiv.org/pdf/2201.03194v2.pdf | Label Relation Graphs Enhanced Hierarchical Residual Network for Hierarchical Multi-Granularity Classification | Hierarchical multi-granularity classification (HMC) assigns hierarchical multi-granularity labels to each object and focuses on encoding the label hierarchy, e.g., ["Albatross", "Laysan Albatross"] from coarse-to-fine levels. However, the definition of what is fine-grained is subjective, and the image quality may affect the identification. Thus, samples could be observed at any level of the hierarchy, e.g., ["Albatross"] or ["Albatross", "Laysan Albatross"], and examples discerned at coarse categories are often neglected in the conventional setting of HMC. In this paper, we study the HMC problem in which objects are labeled at any level of the hierarchy. The essential designs of the proposed method are derived from two motivations: (1) learning with objects labeled at various levels should transfer hierarchical knowledge between levels; (2) lower-level classes should inherit attributes related to upper-level superclasses. The proposed combinatorial loss maximizes the marginal probability of the observed ground truth label by aggregating information from related labels defined in the tree hierarchy. If the observed label is at the leaf level, the combinatorial loss further imposes the multi-class cross-entropy loss to increase the weight of fine-grained classification loss. Considering the hierarchical feature interaction, we propose a hierarchical residual network (HRN), in which granularity-specific features from parent levels acting as residual connections are added to features of children levels. Experiments on three commonly used datasets demonstrate the effectiveness of our approach compared to the state-of-the-art HMC approaches and fine-grained visual classification (FGVC) methods exploiting the label hierarchy. | ['Yuntao Qian', 'Jian Liu', 'Peng Wang', 'Jingzhou Chen'] | 2022-01-10 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Chen_Label_Relation_Graphs_Enhanced_Hierarchical_Residual_Network_for_Hierarchical_Multi-Granularity_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Chen_Label_Relation_Graphs_Enhanced_Hierarchical_Residual_Network_for_Hierarchical_Multi-Granularity_CVPR_2022_paper.pdf | cvpr-2022-1 | ['fine-grained-image-classification'] | ['computer-vision'] | [ 3.47916245e-01 2.56280303e-01 -2.96794981e-01 -4.06038195e-01
-5.68489850e-01 -4.01994437e-01 3.76895458e-01 3.62259597e-01
-1.07466191e-01 8.94409716e-01 -1.09741770e-01 1.69908166e-01
-5.38038850e-01 -1.16896498e+00 -7.80614257e-01 -9.28764045e-01
-2.50474542e-01 3.40220124e-01 3.55363071e-01 3.79903555e-01
1.39518738e-01 6.17481112e-01 -2.06951284e+00 7.80636549e-01
8.35939288e-01 1.53389215e+00 9.90290120e-02 2.60797203e-01
-3.07912305e-02 9.29953754e-01 -6.77536070e-01 -4.44853187e-01
2.90011823e-01 -2.23626480e-01 -9.54164088e-01 3.55069339e-01
8.39966536e-01 2.49841325e-02 1.91281050e-01 1.34518504e+00
-8.54021758e-02 -1.44590074e-02 1.01853359e+00 -1.40876698e+00
-6.04158819e-01 5.71272910e-01 -7.92394340e-01 -6.14010654e-02
-3.12385350e-01 -9.43359509e-02 1.43949914e+00 -6.00928485e-01
3.82216573e-01 1.42769110e+00 6.12813473e-01 3.20326686e-01
-1.38427305e+00 -7.68682301e-01 6.22735620e-01 2.83701271e-01
-1.66304028e+00 1.05299003e-01 6.50767386e-01 -7.73853958e-01
1.56303704e-01 1.99400082e-01 5.50416350e-01 7.93542802e-01
1.87139109e-01 6.89600408e-01 1.63349879e+00 -2.19585091e-01
3.07263494e-01 2.86588728e-01 4.00475442e-01 9.09465492e-01
3.07464898e-01 3.15491021e-01 -5.56071401e-01 -8.51412937e-02
5.17495334e-01 1.26933932e-01 -1.67893514e-01 -5.54177105e-01
-9.32138145e-01 9.54484463e-01 9.14312541e-01 1.51431590e-01
-3.45313609e-01 6.29489869e-02 2.29355201e-01 1.71266004e-01
4.76066083e-01 1.44206971e-01 -3.92908365e-01 5.30519187e-01
-9.84785974e-01 -1.10630542e-01 5.50808847e-01 1.09599185e+00
1.20233548e+00 -2.41240636e-01 -4.03351337e-01 8.77441227e-01
3.58427435e-01 -6.12874143e-02 1.78088278e-01 -9.77214515e-01
4.04495627e-01 8.92008662e-01 -1.43875182e-01 -8.02649677e-01
-3.33482057e-01 -8.82598937e-01 -1.06579292e+00 5.40271699e-01
2.11819544e-01 1.80419266e-01 -1.06590772e+00 1.79292500e+00
5.46225250e-01 4.44025509e-02 1.54455351e-02 7.31030703e-01
6.86871946e-01 5.59306979e-01 3.69824708e-01 -1.07420154e-01
1.57704377e+00 -9.38213766e-01 -2.55988091e-01 1.71012431e-01
2.14637384e-01 -3.07313144e-01 1.08481705e+00 3.10543537e-01
-6.62074924e-01 -8.84057045e-01 -1.03910720e+00 1.72639519e-01
-6.69871390e-01 2.26460665e-01 4.36871916e-01 5.64422965e-01
-8.41991365e-01 6.57733977e-01 -3.89021397e-01 -1.02141492e-01
4.59264606e-01 2.23007694e-01 -2.70124525e-01 2.27751471e-02
-1.10216761e+00 5.01180828e-01 6.65147483e-01 -9.05884802e-02
-1.14967597e+00 -7.44269073e-01 -7.42493033e-01 3.55052114e-01
4.41185921e-01 -4.60055560e-01 7.00974703e-01 -1.00989890e+00
-8.41462970e-01 9.57637250e-01 1.95603371e-01 -4.07108814e-01
3.98395211e-01 1.66272312e-01 -2.79714286e-01 3.11097980e-01
5.05849600e-01 1.07803917e+00 8.79016221e-01 -1.77653909e+00
-1.53618705e+00 -5.19695163e-01 5.82503200e-01 2.16623113e-01
-2.04763010e-01 -4.90809709e-01 -1.24791063e-01 -7.85432279e-01
1.78263202e-01 -7.79149532e-01 2.82190591e-01 1.23503171e-01
-4.20088559e-01 -4.80168730e-01 7.78380036e-01 -2.93721825e-01
1.16473842e+00 -2.17884374e+00 1.27639798e-02 2.03854054e-01
3.30298692e-01 -2.34851182e-01 3.14269066e-02 1.23564661e-01
1.16999075e-01 2.48271257e-01 -2.84066826e-01 6.49557710e-02
2.23191082e-01 2.78273165e-01 -1.86681688e-01 3.16808313e-01
1.88988104e-01 4.92291063e-01 -7.23682642e-01 -8.81633639e-01
1.37040675e-01 3.22402596e-01 -3.31479669e-01 1.00908317e-01
-1.31056443e-01 3.01811963e-01 -4.23991054e-01 9.44399416e-01
5.48439622e-01 -4.49039578e-01 -5.86887710e-02 -6.83030128e-01
-1.30212903e-01 -1.50904059e-01 -1.13943624e+00 9.94914114e-01
-5.34435749e-01 6.89513311e-02 7.70977139e-02 -9.39489186e-01
8.55442286e-01 7.73032531e-02 3.48772138e-01 -3.12533498e-01
-1.30505025e-01 -4.97672483e-02 -6.87960535e-02 -5.88455833e-02
3.00246030e-01 -3.38435739e-01 -3.13058555e-01 -6.81692809e-02
2.73267757e-02 3.01555712e-02 2.59864926e-01 -1.03920415e-01
7.19609916e-01 2.25457847e-02 6.05866671e-01 -3.98851573e-01
6.26184285e-01 -9.62731689e-02 9.84010458e-01 8.85757327e-01
-2.43052751e-01 4.22876060e-01 6.24142349e-01 -3.68127584e-01
-7.64428079e-01 -1.19721532e+00 -6.11590922e-01 1.27360225e+00
3.41839284e-01 -1.94638580e-01 -8.30589592e-01 -1.09665084e+00
2.63569385e-01 4.59811330e-01 -1.04271543e+00 -8.57988223e-02
-5.90250231e-02 -5.59275448e-01 4.96354699e-01 5.33439219e-01
9.65690196e-01 -1.05221093e+00 -6.89720333e-01 2.78563574e-02
-1.51782095e-01 -1.04584658e+00 -3.10557514e-01 3.68662983e-01
-5.80894768e-01 -1.04667068e+00 -4.91468906e-01 -6.74937844e-01
6.90131128e-01 -2.56620366e-02 1.26308179e+00 2.89936829e-02
-3.97504598e-01 1.87912956e-01 -5.93163490e-01 2.88879182e-02
-1.13299124e-01 9.50538516e-02 -1.81279629e-01 6.67228699e-01
1.20683342e-01 -4.06363904e-01 -4.49804813e-01 5.20246267e-01
-8.92612696e-01 -6.78187376e-03 7.70446777e-01 1.03611302e+00
9.52624381e-01 7.13406980e-01 5.14905095e-01 -1.23849916e+00
6.80462569e-02 -5.02833128e-01 -7.31053233e-01 5.86753845e-01
-6.73984110e-01 1.09508755e-02 8.28997493e-01 -4.06030148e-01
-9.28826332e-01 2.29879282e-02 3.01352799e-01 -4.32896942e-01
-5.31789899e-01 2.33593985e-01 -5.74663699e-01 -9.35289860e-02
3.16587538e-01 -2.68075876e-02 -7.90157735e-01 -3.55707616e-01
2.48857304e-01 6.89276457e-01 3.88230801e-01 -1.10190725e+00
5.76525271e-01 4.17912632e-01 3.93406332e-01 -6.01154804e-01
-1.30352974e+00 -2.05611095e-01 -7.39490867e-01 -2.65965372e-01
1.15618074e+00 -8.72847319e-01 -8.50339890e-01 3.07855666e-01
-7.80121803e-01 -1.12659864e-01 -5.48009574e-01 2.21722439e-01
-4.43511099e-01 1.34963483e-01 -4.86540109e-01 -8.01205873e-01
6.42768294e-02 -1.15949774e+00 1.38490295e+00 4.01857883e-01
1.02227263e-01 -7.90425658e-01 -6.50796175e-01 4.62572575e-01
-9.87134725e-02 4.69761789e-01 1.39121974e+00 -2.66061097e-01
-8.09102774e-01 -1.66628242e-03 -5.93555212e-01 6.30607724e-01
1.42574146e-01 -1.34209335e-01 -9.78216350e-01 -3.75779718e-01
-1.50190875e-01 -6.84317708e-01 9.08906519e-01 2.16026157e-01
1.60306728e+00 -5.80906093e-01 -2.77652144e-01 5.86655676e-01
1.61052573e+00 2.01359496e-01 2.19824895e-01 1.43127307e-01
6.74086213e-01 9.82224941e-01 6.54486716e-01 3.89054239e-01
5.24125814e-01 5.96746087e-01 6.82032406e-01 -2.31070742e-02
-2.98927337e-01 -3.41607809e-01 4.16022688e-02 2.10361212e-01
2.24250257e-02 -1.93266392e-01 -5.57785451e-01 4.41306800e-01
-1.58842230e+00 -7.47243583e-01 7.96763226e-02 2.36807036e+00
7.49478698e-01 1.05822079e-01 -6.10788129e-02 4.08161953e-02
1.03561509e+00 1.41635612e-01 -6.73046112e-01 -6.68533966e-02
-7.76614398e-02 -1.27055030e-02 4.19475794e-01 4.71707165e-01
-1.39559853e+00 7.32662559e-01 4.73993349e+00 1.26340246e+00
-6.96320891e-01 1.22303322e-01 7.98167229e-01 1.97147831e-01
3.03180311e-02 6.00471236e-02 -1.29792321e+00 7.13106632e-01
3.47498983e-01 2.85074919e-01 2.58553654e-01 8.00748944e-01
-4.13435370e-01 -7.41536766e-02 -1.28047252e+00 7.73514807e-01
-1.32146910e-01 -1.01525772e+00 3.63541633e-01 3.52440268e-01
8.60668659e-01 -4.59837735e-01 1.41557992e-01 5.91181338e-01
4.99133408e-01 -8.59602571e-01 9.20503378e-01 4.76235509e-01
9.57992256e-01 -8.29723179e-01 5.44673681e-01 4.65184510e-01
-1.73815024e+00 -3.42062920e-01 -6.04509413e-01 3.60354841e-01
-2.61245370e-01 6.25208676e-01 -2.18082502e-01 7.90742278e-01
8.18117976e-01 4.24221218e-01 -7.39168227e-01 7.14979410e-01
-3.17613631e-01 6.08978868e-01 -1.08219124e-01 2.99824774e-01
5.11685491e-01 -1.44005269e-01 1.40542209e-01 1.09135616e+00
1.44030944e-01 5.55871725e-02 7.69179881e-01 8.66091788e-01
-2.90507406e-01 -8.15203264e-02 -2.90756047e-01 3.21971774e-01
5.32307982e-01 1.37066662e+00 -8.20714593e-01 -4.60886478e-01
-3.92289549e-01 7.35661745e-01 5.91833532e-01 3.69240910e-01
-8.25300813e-01 -4.03916657e-01 3.63086373e-01 1.20293915e-01
3.37847024e-01 4.63153690e-01 -3.39703649e-01 -9.58937943e-01
3.85662983e-03 -7.38244057e-01 8.46186996e-01 -4.73856032e-01
-1.65217996e+00 5.51368356e-01 1.80049360e-01 -1.43004441e+00
2.08300889e-01 -6.92443013e-01 -1.20338656e-01 8.75965059e-01
-1.40765917e+00 -1.52953136e+00 -4.64357793e-01 5.82326114e-01
5.52813888e-01 -8.19771364e-02 5.69001436e-01 2.81035483e-01
-2.50917464e-01 7.15333343e-01 -1.56278789e-01 3.75475809e-02
2.27556184e-01 -1.44637430e+00 -4.98537153e-01 5.74438930e-01
1.16773114e-01 3.70838284e-01 8.37295204e-02 -6.01503134e-01
-3.32932472e-01 -1.56696451e+00 6.94491684e-01 -7.80633232e-03
5.90237081e-01 -4.40540433e-01 -9.15392160e-01 4.68782932e-01
-1.74382776e-01 2.67939925e-01 6.13597691e-01 2.37014681e-01
-7.08557069e-01 -5.99521637e-01 -1.49294901e+00 2.45065227e-01
9.38389838e-01 -6.12173200e-01 -4.39372331e-01 2.13500291e-01
6.79795623e-01 1.44944772e-01 -1.35934901e+00 8.95711243e-01
6.41124070e-01 -9.69177663e-01 8.70431900e-01 -5.23507357e-01
4.57173198e-01 -6.14734709e-01 -5.49698651e-01 -1.10383046e+00
-6.27868891e-01 1.80538192e-01 -4.52817865e-02 1.53697610e+00
2.07104817e-01 -5.44245541e-01 7.89456010e-01 1.54049918e-01
-1.58620790e-01 -8.03722858e-01 -1.09606862e+00 -1.01820910e+00
2.06554770e-01 7.12876841e-02 7.22200155e-01 8.71162415e-01
-5.88428378e-01 4.37070668e-01 -2.16711596e-01 6.54002249e-01
1.13926184e+00 4.79077041e-01 2.47996524e-01 -1.71776450e+00
-4.37878907e-01 -4.69101846e-01 -6.21076524e-01 -8.09359372e-01
3.47728521e-01 -1.00313020e+00 7.32544288e-02 -1.36886895e+00
3.77312124e-01 -7.04862773e-01 -5.98875344e-01 5.80107749e-01
-8.97088870e-02 3.47056657e-01 3.21401030e-01 3.55576783e-01
-5.16798139e-01 4.21358913e-01 1.30081081e+00 -4.84119534e-01
3.88309121e-01 -6.35271892e-03 -6.21006072e-01 8.43592823e-01
5.14407337e-01 -4.82556731e-01 -4.04231250e-01 -2.52604987e-02
-1.63251206e-01 6.07345514e-02 6.68317974e-01 -1.22743881e+00
1.04669988e-01 -1.42518565e-01 5.05628228e-01 -7.59276450e-01
4.03773725e-01 -1.06136155e+00 1.63833916e-01 3.52254540e-01
-5.96820354e-01 -4.60772365e-01 -2.54318684e-01 5.82036018e-01
-4.48058128e-01 -4.74642992e-01 1.33627486e+00 -1.90951899e-01
-6.81843102e-01 4.20095563e-01 -1.05095744e-01 1.14207193e-01
1.10755050e+00 -2.48102397e-01 -4.11456138e-01 2.28881836e-01
-8.28916728e-01 2.53378898e-01 4.19808418e-01 1.63022697e-01
2.77547956e-01 -1.38503277e+00 -6.69323802e-01 1.88336030e-01
5.40959656e-01 8.57463256e-02 2.01737136e-01 4.99878794e-01
-7.22515881e-02 1.97152525e-01 -3.21091205e-01 -7.44901419e-01
-1.17410600e+00 7.65922546e-01 3.18469822e-01 -6.32168889e-01
-4.41680402e-01 8.53689909e-01 1.14933240e+00 -4.98638302e-01
4.63961214e-01 -3.08712572e-01 -3.53122324e-01 3.74383003e-01
8.02663863e-02 4.49823171e-01 -7.39741698e-02 -6.90628171e-01
-1.69758886e-01 8.74499679e-01 -1.74515888e-01 2.51853198e-01
7.21193075e-01 -2.07154125e-01 -2.07849696e-01 6.85790002e-01
1.19491065e+00 -3.42948645e-01 -1.38326001e+00 -2.08185598e-01
4.60153483e-02 -3.80309194e-01 -7.75028244e-02 -1.06700385e+00
-1.04644370e+00 9.86027300e-01 8.55441272e-01 3.75211388e-01
1.32325900e+00 2.72828758e-01 2.32800320e-01 1.68563277e-01
7.77340472e-01 -1.10024202e+00 -9.00809392e-02 3.13027024e-01
7.04323888e-01 -1.00550842e+00 -1.48993477e-01 -5.85809052e-01
-5.04634500e-01 7.80957937e-01 7.93290854e-01 -3.33171384e-03
5.08853614e-01 1.80544220e-02 -2.70329326e-01 -2.17783689e-01
-6.71446323e-01 -4.50282186e-01 4.89035726e-01 4.63328511e-01
2.30396688e-01 4.87338185e-01 -2.91599602e-01 5.27804375e-01
-5.22455433e-03 -2.92601556e-01 1.65293857e-01 5.29602110e-01
-6.48166537e-01 -8.24648619e-01 -5.37686944e-01 6.89990103e-01
-3.83422494e-01 -6.99581951e-02 -2.85894275e-01 8.43520105e-01
1.04274201e+00 9.10851300e-01 9.37069878e-02 -2.54969746e-01
3.57285500e-01 -7.42245466e-02 4.28271979e-01 -6.74413383e-01
-6.00347161e-01 -1.01317372e-02 -7.39523992e-02 -3.03103387e-01
-4.27904427e-01 -6.61635160e-01 -1.10339403e+00 -1.18570169e-02
-1.93007708e-01 3.61538649e-01 1.99374661e-01 7.06820726e-01
7.73738697e-02 6.00801289e-01 8.37586224e-01 -5.12024164e-01
-4.47252572e-01 -8.53734672e-01 -1.12791169e+00 5.81052065e-01
3.97032738e-01 -1.07548332e+00 -5.39753139e-01 1.55441150e-01] | [9.601076126098633, 2.3980371952056885] |
ec9160a0-2ec3-475a-b9b7-3337ce13e687 | generative-entity-typing-with-curriculum | 2210.02914 | null | https://arxiv.org/abs/2210.02914v2 | https://arxiv.org/pdf/2210.02914v2.pdf | Generative Entity Typing with Curriculum Learning | Entity typing aims to assign types to the entity mentions in given texts. The traditional classification-based entity typing paradigm has two unignorable drawbacks: 1) it fails to assign an entity to the types beyond the predefined type set, and 2) it can hardly handle few-shot and zero-shot situations where many long-tail types only have few or even no training instances. To overcome these drawbacks, we propose a novel generative entity typing (GET) paradigm: given a text with an entity mention, the multiple types for the role that the entity plays in the text are generated with a pre-trained language model (PLM). However, PLMs tend to generate coarse-grained types after fine-tuning upon the entity typing dataset. Besides, we only have heterogeneous training data consisting of a small portion of human-annotated data and a large portion of auto-generated but low-quality data. To tackle these problems, we employ curriculum learning (CL) to train our GET model upon the heterogeneous data, where the curriculum could be self-adjusted with the self-paced learning according to its comprehension of the type granularity and data heterogeneity. Our extensive experiments upon the datasets of different languages and downstream tasks justify the superiority of our GET model over the state-of-the-art entity typing models. The code has been released on https://github.com/siyuyuan/GET. | ['Yanghua Xiao', 'Jingyue Huang', 'Jinxi Liu', 'Zhixu Li', 'Jiaqing Liang', 'Deqing Yang', 'Siyu Yuan'] | 2022-10-06 | null | null | null | null | ['entity-typing'] | ['natural-language-processing'] | [-1.12056389e-01 1.89094007e-01 -3.53751779e-01 -3.98484558e-01
-6.03944242e-01 -6.27447605e-01 6.49360418e-01 2.47329399e-01
-6.98625863e-01 9.85196471e-01 -3.96923721e-02 -3.49573106e-01
2.17693835e-01 -1.15457606e+00 -7.37271249e-01 -4.37950999e-01
3.07893097e-01 8.69488299e-01 2.72328287e-01 -3.47642422e-01
-6.80364743e-02 -4.43505853e-01 -1.71974719e+00 3.23736429e-01
1.36965561e+00 7.30371594e-01 4.60465640e-01 5.05719364e-01
-6.66050315e-01 4.57751662e-01 -6.26697719e-01 -8.37094784e-01
-1.97307289e-01 -1.05832897e-01 -7.91707873e-01 -5.42994320e-01
2.21751511e-01 1.17949266e-02 -6.49793819e-02 1.08020258e+00
6.05384350e-01 5.82219362e-02 8.33990455e-01 -1.40588224e+00
-1.03256309e+00 8.63637209e-01 -4.04981881e-01 4.91649704e-03
4.30086553e-01 1.09198302e-01 1.26631773e+00 -1.10485196e+00
8.72051120e-01 1.01268375e+00 8.29423547e-01 1.06486917e+00
-1.01142073e+00 -7.06575394e-01 1.80426776e-01 3.47425938e-01
-1.30626822e+00 -3.04524481e-01 3.51922035e-01 -5.34814477e-01
8.69464815e-01 2.90609807e-01 3.08338583e-01 1.36755729e+00
-1.53128430e-01 8.29700470e-01 9.03003871e-01 -3.63195211e-01
3.19737256e-01 3.60230267e-01 5.53026915e-01 5.47985196e-01
5.15251398e-01 -4.21631522e-02 -2.27310658e-01 -3.60162199e-01
2.54679233e-01 -2.16357455e-01 -2.78026946e-02 2.30531227e-02
-1.15556157e+00 6.45049214e-01 1.31107435e-01 3.22721899e-01
-1.21391848e-01 -7.94065893e-02 6.69029236e-01 1.06623799e-01
3.39107811e-01 5.02016962e-01 -9.62952495e-01 -2.45838672e-01
-5.70386350e-01 4.23730522e-01 9.38714921e-01 1.63919401e+00
9.78407562e-01 -1.23847924e-01 -3.04198325e-01 1.04117751e+00
1.55116245e-01 5.12679100e-01 6.60114765e-01 -1.78924039e-01
7.84228683e-01 9.34093416e-01 2.54075706e-01 -3.39450330e-01
-2.92903572e-01 -1.33389533e-01 -9.60503757e-01 -2.20011219e-01
4.76325721e-01 -5.85951805e-01 -8.72203648e-01 1.75494432e+00
6.22985005e-01 3.62174213e-01 1.80397257e-01 5.19160628e-01
1.26138413e+00 6.56888247e-01 3.98140788e-01 1.11390173e-01
1.71481621e+00 -6.64125025e-01 -8.05786669e-01 -3.76908362e-01
7.90105581e-01 -4.23328191e-01 1.32574522e+00 -1.87495984e-02
-6.94854617e-01 -5.36385357e-01 -8.51888537e-01 -1.78420797e-01
-7.23086655e-01 2.20131055e-01 4.82220262e-01 6.80835545e-01
-3.25912148e-01 2.51766026e-01 -4.86884803e-01 -1.32065058e-01
2.10340664e-01 2.93419547e-02 -3.00487150e-02 1.02595754e-01
-1.78548419e+00 5.60896158e-01 5.65220177e-01 7.90574029e-02
-5.40519357e-01 -1.14819717e+00 -8.70799780e-01 1.20168172e-01
6.49956882e-01 -1.01447904e+00 1.47658777e+00 -7.18242645e-01
-1.10035741e+00 7.23343015e-01 -3.97111803e-01 -6.97586313e-02
4.40617323e-01 -1.59655467e-01 -4.56846565e-01 -5.53537071e-01
2.75836110e-01 8.25681314e-02 3.76083612e-01 -1.21663165e+00
-1.21331024e+00 -4.03268859e-02 2.16642320e-01 1.55201957e-01
-2.45057046e-01 -2.76858397e-02 -4.32189703e-01 -8.14662457e-01
-3.79525006e-01 -8.28650653e-01 -1.10803738e-01 -5.85736930e-01
-4.36809033e-01 -8.03389013e-01 4.63057578e-01 -3.52518916e-01
1.70273638e+00 -1.88475513e+00 9.19838026e-02 -1.27661362e-01
2.12922439e-01 3.26273054e-01 4.25773375e-02 4.13506418e-01
2.03300029e-01 4.23605591e-01 -2.49672934e-01 -3.53489041e-01
3.86040330e-01 2.95584559e-01 -2.67758727e-01 -5.07845655e-02
-3.59652564e-02 8.56033266e-01 -1.33549750e+00 -5.82597673e-01
-3.66216153e-01 1.52781889e-01 -6.22975826e-01 6.88655019e-01
-5.70734143e-01 1.19736955e-01 -6.54028714e-01 7.06393242e-01
5.60654223e-01 -1.59406066e-01 3.01565230e-01 -1.15872137e-01
-9.20207500e-02 5.41072369e-01 -1.44348669e+00 1.31298244e+00
-5.15088499e-01 4.62289974e-02 -2.26974756e-01 -5.51845670e-01
6.63442552e-01 4.56575841e-01 1.57554984e-01 -4.48707312e-01
-5.49523942e-02 3.97484243e-01 -6.61828071e-02 -8.50220680e-01
8.82256627e-01 -2.62591302e-01 -5.74552417e-01 4.49773967e-01
2.63509840e-01 4.26350355e-01 4.43079203e-01 1.90075472e-01
8.70945811e-01 1.56196460e-01 2.27359876e-01 9.50676873e-02
2.99879074e-01 -3.33766751e-02 1.16474271e+00 9.20317531e-01
1.02767467e-01 3.84919107e-01 5.88177621e-01 -2.66419798e-01
-1.14710605e+00 -7.28864253e-01 -2.79236704e-01 1.50111043e+00
2.80261219e-01 -5.73036373e-01 -6.05135739e-01 -9.63400900e-01
4.33326997e-02 7.23936617e-01 -7.51575112e-01 1.34444013e-01
-5.48167944e-01 -1.08729100e+00 7.79513657e-01 5.81141293e-01
2.89708734e-01 -1.41714585e+00 -1.45040482e-01 3.87887061e-01
-5.94579160e-01 -9.34771955e-01 -5.04488289e-01 7.87384510e-02
-2.12825209e-01 -1.10572112e+00 -5.21727979e-01 -9.28545117e-01
6.70873404e-01 -1.51052788e-01 1.24568582e+00 2.63229698e-01
1.16765462e-01 -1.77823633e-01 -6.86164200e-01 -4.43265408e-01
-5.34145117e-01 6.64511383e-01 7.63154924e-02 -6.60787970e-02
6.37717068e-01 -1.98918968e-01 -2.31023937e-01 1.93308949e-01
-6.43089473e-01 4.91940677e-02 5.24451733e-01 1.21401215e+00
4.00934994e-01 1.03305921e-01 7.49577999e-01 -1.91061068e+00
5.80256164e-01 -9.51547325e-01 -3.89241755e-01 4.69647914e-01
-8.97033811e-01 7.97227919e-02 8.21731031e-01 -7.45065331e-01
-1.47287452e+00 -3.38387102e-01 -2.66827345e-01 5.64335147e-03
-1.24550737e-01 8.39903295e-01 -6.01746678e-01 6.15048826e-01
6.41777933e-01 8.76070485e-02 -6.13673687e-01 -5.84861517e-01
4.54431772e-01 9.14887428e-01 5.91935158e-01 -1.16786921e+00
8.34661245e-01 -2.51219481e-01 -6.27240717e-01 -3.51069331e-01
-9.00270224e-01 -5.28359652e-01 -8.11382353e-01 9.69304740e-02
6.35501385e-01 -9.08170938e-01 -4.88533497e-01 8.28797996e-01
-1.04821014e+00 -6.55110002e-01 -4.89886142e-02 9.65707004e-02
-3.15651566e-01 3.09647501e-01 -6.97342038e-01 -7.33497322e-01
-5.41493833e-01 -9.07744288e-01 9.05759275e-01 4.42515463e-01
-1.31699681e-01 -1.02503657e+00 1.67876810e-01 6.34895638e-02
2.58940279e-01 6.67746663e-02 1.29030132e+00 -7.74474740e-01
-4.43000853e-01 -1.38316870e-01 -4.79293838e-02 -1.83666885e-01
1.36983767e-01 1.32474169e-01 -8.36888492e-01 -1.21157773e-01
-5.31552792e-01 -3.58114094e-01 4.20150399e-01 -1.64395615e-01
9.21990871e-01 -7.53499925e-01 -4.57027137e-01 7.19141006e-01
1.42700744e+00 6.06015176e-02 5.28946579e-01 5.51670492e-01
1.06275141e+00 5.08081615e-01 9.90306437e-01 4.67389762e-01
8.81430149e-01 6.76981747e-01 -7.92495534e-02 -3.30008343e-02
1.69212315e-02 -6.54211640e-01 4.18061391e-02 8.40539336e-01
-9.06640366e-02 -4.22173232e-01 -1.02920532e+00 6.49774492e-01
-2.03140068e+00 -1.22988260e+00 -3.91502947e-01 2.11980438e+00
1.69112706e+00 1.95702866e-01 1.98356599e-01 -3.01230013e-01
8.39488506e-01 -1.42321676e-01 -6.21903241e-01 -9.31280330e-02
5.43457456e-02 -5.94132394e-02 4.59933251e-01 4.91836071e-01
-1.07350385e+00 1.06238997e+00 5.39569139e+00 9.16501999e-01
-8.18343520e-01 1.94344863e-01 2.33749449e-01 3.79801631e-01
-6.22433603e-01 9.77208763e-02 -1.37457252e+00 1.10523093e+00
7.91674256e-01 -6.34986997e-01 2.37518728e-01 9.46306646e-01
-2.81422377e-01 1.57829553e-01 -1.08973598e+00 5.73110044e-01
-2.88076848e-01 -1.13932252e+00 1.63647963e-03 -7.50463828e-02
5.16212106e-01 -6.85615018e-02 -2.06157506e-01 1.02171063e+00
6.82580709e-01 -7.48359144e-01 9.35311377e-01 4.46808487e-01
1.19393110e+00 -5.76552331e-01 9.76379395e-01 7.65250921e-01
-1.39204502e+00 -3.02549172e-02 -5.66145539e-01 2.44757548e-01
-7.48791248e-02 5.03162682e-01 -9.08242881e-01 7.24761128e-01
5.85194528e-01 4.76057440e-01 -5.33542097e-01 9.12723958e-01
-4.31291997e-01 6.04270279e-01 -4.84946296e-02 -3.57700706e-01
-9.94817391e-02 -1.40026556e-02 4.17618334e-01 1.52780497e+00
2.23546043e-01 3.05345982e-01 9.26845521e-02 8.41525674e-01
-1.27613336e-01 -3.00389528e-02 -2.20171019e-01 2.46777296e-01
1.10706961e+00 1.53055060e+00 -9.16427933e-03 -8.83583307e-01
-5.97565114e-01 5.65357625e-01 6.45947874e-01 3.18974465e-01
-8.22806537e-01 -6.64848268e-01 6.01001263e-01 5.89728765e-02
3.27450037e-01 8.57172981e-02 -4.00705129e-01 -1.59082615e+00
4.63699363e-02 -1.05170751e+00 7.82727063e-01 -5.60555637e-01
-1.76807356e+00 6.47539556e-01 1.23911828e-01 -1.18618178e+00
-3.77413362e-01 -3.34888697e-01 -5.43070138e-01 1.03425920e+00
-1.50782096e+00 -1.18585896e+00 -4.15081412e-01 2.95355707e-01
5.69036126e-01 -1.89189255e-01 8.90008628e-01 6.62760019e-01
-9.02141929e-01 1.26640093e+00 -2.80773565e-02 6.11896634e-01
9.24200773e-01 -1.80825782e+00 6.93850160e-01 8.26821685e-01
-3.90945017e-01 8.98659110e-01 8.03838611e-01 -7.82430947e-01
-1.36654353e+00 -1.41077518e+00 1.34777594e+00 -7.29453027e-01
7.09314823e-01 -7.01223731e-01 -1.38338780e+00 8.61448050e-01
1.13348022e-01 -1.05650380e-01 8.58288288e-01 6.95031762e-01
-6.91038132e-01 9.26975086e-02 -1.00938904e+00 4.18184906e-01
1.07781577e+00 -3.19399446e-01 -1.05862236e+00 4.14065234e-02
5.38665473e-01 -7.85508573e-01 -9.62805808e-01 2.88098723e-01
5.49497187e-01 -4.03361142e-01 5.20268977e-01 -7.69122541e-01
4.98345524e-01 -3.10811996e-01 1.37437314e-01 -1.66041148e+00
-4.80344474e-01 -4.79073614e-01 -2.76306301e-01 1.95859802e+00
7.48763382e-01 -7.35972822e-01 5.40593088e-01 9.83310580e-01
-2.99113095e-01 -6.92602336e-01 -6.29202902e-01 -7.60208786e-01
1.75659418e-01 -3.93996201e-02 1.04684937e+00 1.21571171e+00
2.01120928e-01 5.92210889e-01 -1.77879974e-01 2.31609285e-01
4.70183551e-01 2.29448989e-01 1.01219189e+00 -1.12527764e+00
-4.11468387e-01 -3.16372424e-01 6.76834956e-02 -8.90140057e-01
2.37341389e-01 -1.11041367e+00 3.17184448e-01 -1.33657265e+00
3.71255487e-01 -1.45317197e+00 -2.47553751e-01 9.15106118e-01
-9.89422917e-01 -3.36361527e-01 -1.51755050e-01 2.81054735e-01
-4.79898602e-01 3.58153939e-01 8.49264205e-01 -2.03464881e-01
-6.18668534e-02 7.84913152e-02 -8.90946090e-01 4.17417377e-01
6.08754635e-01 -7.34074533e-01 -2.62662202e-01 -4.40553218e-01
5.88246644e-01 2.59319507e-02 1.44718727e-02 -3.94633085e-01
3.64095688e-01 -4.21844155e-01 2.30588168e-02 -5.45461774e-01
-2.64041156e-01 -2.86470234e-01 3.11730832e-01 9.13080275e-02
-4.07157898e-01 1.31849244e-01 -1.37062520e-01 3.79257202e-01
1.62187440e-03 -6.33862615e-01 4.90624219e-01 -2.15028837e-01
-1.07633817e+00 6.20826960e-01 -1.32772937e-01 6.91634119e-01
7.15755403e-01 9.21244249e-02 -6.94259524e-01 3.22602659e-01
-2.94547409e-01 4.45603043e-01 6.65132105e-01 6.46842718e-01
9.44688767e-02 -1.32449722e+00 -8.34895730e-01 1.96762741e-01
5.55000901e-01 4.75666851e-01 2.92345315e-01 3.34257662e-01
-1.24338120e-02 -4.82470822e-03 -5.92510886e-02 -3.28365624e-01
-1.12020814e+00 7.10435629e-01 3.38145554e-01 -3.57654303e-01
-6.55010998e-01 7.72416830e-01 2.61494666e-01 -8.74211192e-01
2.10919827e-01 -1.29147753e-01 -2.97393680e-01 3.07500809e-01
5.51055670e-01 6.57935590e-02 1.54626727e-01 -3.79995316e-01
-3.12326729e-01 1.18599072e-01 -2.85866737e-01 1.32474065e-01
1.29814827e+00 9.10130739e-02 -3.23416404e-02 7.01901853e-01
9.11679387e-01 2.40307167e-01 -1.06705666e+00 -4.12709266e-01
1.37142777e-01 -3.56801838e-01 -4.62973297e-01 -8.72533679e-01
-7.83275485e-01 4.30982262e-01 1.51098808e-02 3.91785353e-01
6.13270402e-01 4.57779020e-02 9.63886559e-01 2.82275110e-01
5.42971909e-01 -1.16117573e+00 -3.93668085e-01 8.34847450e-01
4.80129331e-01 -1.11174762e+00 -2.22617313e-01 -4.72756952e-01
-7.30197370e-01 7.76735067e-01 1.00305402e+00 2.47284383e-01
4.29842591e-01 5.28758645e-01 1.70413911e-01 1.01762004e-01
-1.20232296e+00 -3.36646467e-01 2.18635038e-01 6.32777810e-01
6.57052100e-01 3.67629290e-01 -4.73957658e-01 1.16959190e+00
-4.90816951e-01 -3.89546528e-03 5.96828461e-01 7.87282407e-01
-2.22417638e-01 -1.40035272e+00 7.34789744e-02 5.65618277e-01
-3.72503251e-01 -3.07233453e-01 -1.09218694e-01 7.30929255e-01
4.51548964e-01 6.44317329e-01 -1.21325500e-01 -3.30753058e-01
5.76807559e-01 1.17674544e-01 2.45246544e-01 -9.87722874e-01
-8.28096151e-01 -2.77027935e-01 3.67980212e-01 4.32521440e-02
1.07865073e-02 -8.27251673e-01 -1.36687219e+00 -3.41916382e-01
-3.96707892e-01 2.30940357e-01 2.09180772e-01 8.32349658e-01
4.31474090e-01 4.22176450e-01 6.63428783e-01 -1.53413132e-01
-5.37107825e-01 -1.13468587e+00 -4.38333154e-01 7.82826483e-01
1.66185334e-01 -8.93914402e-01 -3.44657987e-01 1.27486885e-01] | [9.644502639770508, 8.758882522583008] |
b1a9f1b7-f0a1-4db4-ba08-03cc1da24902 | reinforcement-learning-finetuned-vision-code | 2305.14637 | null | https://arxiv.org/abs/2305.14637v1 | https://arxiv.org/pdf/2305.14637v1.pdf | Reinforcement Learning finetuned Vision-Code Transformer for UI-to-Code Generation | Automated HTML/CSS code generation from screenshots is an important yet challenging problem with broad applications in website development and design. In this paper, we present a novel vision-code transformer approach that leverages an Encoder-Decoder architecture as well as explore actor-critic fine-tuning as a method for improving upon the baseline. For this purpose, two image encoders are compared: Vision Transformer (ViT) and Document Image Transformer (DiT). We propose an end-to-end pipeline that can generate high-quality code snippets directly from screenshots, streamlining the website creation process for developers. To train and evaluate our models, we created a synthetic dataset of 30,000 unique pairs of code and corresponding screenshots. We evaluate the performance of our approach using a combination of automated metrics such as MSE, BLEU, IoU, and a novel htmlBLEU score, where our models demonstrated strong performance. We establish a strong baseline with the DiT-GPT2 model and show that actor-critic can be used to improve IoU score from the baseline of 0.64 to 0.79 and lower MSE from 12.25 to 9.02. We achieved similar performance as when using larger models, with much lower computational cost. | ['Tianyi Zhou', 'Khalid Saifullah', 'Davit Soselia'] | 2023-05-24 | null | null | null | null | ['code-generation'] | ['computer-code'] | [ 3.85940403e-01 1.21314332e-01 2.52668291e-01 -2.60064781e-01
-1.32293105e+00 -8.97094250e-01 5.60110807e-01 -2.51108199e-01
1.81377809e-02 3.76625001e-01 2.15432689e-01 -4.74861622e-01
3.00044656e-01 -5.37580729e-01 -1.07618451e+00 6.31019920e-02
4.40559864e-01 1.61894038e-03 1.87799662e-01 -5.21899275e-02
3.80985737e-01 -3.75501484e-01 -1.32035542e+00 7.35826075e-01
1.01899457e+00 6.54362023e-01 3.35372239e-01 1.18890858e+00
-9.61580351e-02 9.69865322e-01 -7.98822045e-01 -1.11730731e+00
1.93549171e-01 -7.61080980e-01 -5.93734741e-01 -2.37259921e-03
7.43359685e-01 -3.84080619e-01 -1.48478955e-01 9.13303971e-01
5.49626350e-01 -5.33230245e-01 4.27768499e-01 -9.43794131e-01
-1.26279652e+00 7.63221025e-01 -7.27041781e-01 -1.49606913e-01
6.06221378e-01 2.76684880e-01 1.15470362e+00 -1.00936449e+00
7.28694677e-01 1.08966112e+00 9.62658703e-01 5.93196511e-01
-1.31852949e+00 -2.90199339e-01 -2.61797428e-01 -9.82041806e-02
-1.02376938e+00 -5.04039526e-01 4.32700992e-01 -7.01829433e-01
1.08235633e+00 9.36990082e-02 4.81607258e-01 1.38139176e+00
1.03100851e-01 8.47641110e-01 9.07989383e-01 -8.62423241e-01
-2.99584214e-03 3.38379562e-01 -1.20855741e-01 1.02488673e+00
2.69444138e-01 -2.19095111e-01 -4.25610900e-01 -3.53441276e-02
6.06743932e-01 -2.76481748e-01 -2.05906123e-01 -2.02836230e-01
-1.16494346e+00 7.39747703e-01 8.80949795e-02 7.31450086e-03
-1.30463034e-01 5.95181227e-01 3.12792391e-01 2.82219768e-01
2.07864076e-01 7.35326648e-01 -3.63295853e-01 -6.78036690e-01
-1.08791089e+00 2.35091627e-01 8.67713213e-01 1.29681921e+00
5.08884192e-01 2.48993542e-02 -7.36659050e-01 1.06801915e+00
2.66103506e-01 5.56156158e-01 4.34512943e-01 -1.08729291e+00
8.68253052e-01 4.24458236e-01 1.15835398e-01 -6.89476252e-01
2.54540175e-01 -2.68752664e-01 -1.85485393e-01 1.67788029e-01
2.87203252e-01 -3.35416675e-01 -9.82277811e-01 1.44035983e+00
-2.76802540e-01 -3.27120751e-01 -9.23432410e-02 3.97848785e-01
6.77674830e-01 6.65740728e-01 -2.87531972e-01 2.29094297e-01
1.20269060e+00 -1.51219678e+00 -4.65770900e-01 -3.19423616e-01
5.20060778e-01 -1.07822394e+00 1.68558490e+00 3.53842735e-01
-1.46952319e+00 -5.31542242e-01 -9.89965200e-01 -6.38133958e-02
7.68502206e-02 7.29357719e-01 2.90836513e-01 8.51391494e-01
-1.35924482e+00 5.05011559e-01 -7.76337326e-01 -4.20680195e-01
5.50057650e-01 -1.40387118e-01 1.03871906e-02 -2.49952190e-02
-4.16728854e-01 5.58627784e-01 -4.88524958e-02 -5.41169643e-01
-1.03800678e+00 -8.08993340e-01 -6.60858035e-01 3.47913802e-01
2.89119601e-01 -7.44178712e-01 1.83036661e+00 -1.09786987e+00
-1.57568574e+00 7.50339389e-01 -4.08999622e-04 -3.58058184e-01
6.09720826e-01 -2.97886372e-01 -1.57644629e-01 -4.90345880e-02
2.31683806e-01 5.36971807e-01 8.05065393e-01 -1.11885834e+00
-6.55773938e-01 1.63066596e-01 1.80549055e-01 -1.55470401e-01
-6.44466877e-01 1.92486107e-01 -9.92555022e-01 -7.02588975e-01
-6.24730229e-01 -1.02529097e+00 9.75821018e-02 8.95830840e-02
-5.76972961e-01 2.57167518e-01 3.47900867e-01 -1.10763633e+00
1.47029555e+00 -2.01616263e+00 -1.00377619e-01 -7.04272538e-02
1.21418387e-01 3.52284998e-01 -5.25119066e-01 5.98997474e-01
1.33201167e-01 3.79192322e-01 -2.87339509e-01 -3.05561900e-01
8.49349722e-02 -3.89860958e-01 -7.83350766e-02 -2.96644151e-01
4.20186162e-01 1.13459921e+00 -7.18078732e-01 -3.07926327e-01
-2.74192095e-01 4.85193342e-01 -9.62485313e-01 4.33371305e-01
-4.38266128e-01 -1.36030644e-01 -1.04483306e-01 7.15916753e-01
2.44559050e-01 -6.21604323e-01 2.16960251e-01 8.43472779e-02
-6.14941046e-02 2.29464084e-01 -6.71452641e-01 1.80461466e+00
-8.69382679e-01 1.25382864e+00 -3.47302884e-01 -3.18925977e-01
9.70848501e-01 1.69918269e-01 8.88204575e-03 -9.70301032e-01
-2.48649061e-01 1.83787271e-02 -2.09604979e-01 -7.98197389e-01
5.61227202e-01 4.80343461e-01 -1.47100436e-02 5.39731503e-01
2.41968445e-02 -1.06999531e-01 5.49306154e-01 2.85766870e-01
1.68029618e+00 5.37530482e-01 4.06689756e-02 1.91881895e-01
1.54698431e-01 7.44119212e-02 -1.45751797e-03 8.94073308e-01
1.80240184e-01 1.05741167e+00 9.72656012e-01 -2.19038770e-01
-1.63811445e+00 -1.00428975e+00 3.55131418e-01 1.06011605e+00
-4.35676783e-01 -8.28615308e-01 -1.24364245e+00 -9.48954701e-01
-2.31491983e-01 9.63899076e-01 -6.01479530e-01 -9.74131450e-02
-4.24239516e-01 -5.96588016e-01 7.00401425e-01 5.76933265e-01
3.79366517e-01 -8.68873596e-01 -6.23331249e-01 1.56822830e-01
-2.38827944e-01 -1.02890265e+00 -8.25992942e-01 -1.96888074e-01
-4.11365151e-01 -1.01207292e+00 -1.08529055e+00 -8.83826971e-01
6.34692609e-01 2.41448715e-01 1.57658148e+00 4.79912013e-02
-2.52730131e-01 4.90003943e-01 -5.69793463e-01 -4.86293793e-01
-9.87170577e-01 8.59485269e-02 -7.03578413e-01 -3.62355173e-01
2.28652284e-02 -2.76189029e-01 -6.69518113e-01 1.46720499e-01
-7.73619294e-01 3.55275631e-01 7.95404673e-01 8.82263422e-01
2.71760017e-01 -4.65831935e-01 2.84599900e-01 -1.03786767e+00
9.28354323e-01 -2.88226306e-01 -8.17998171e-01 6.63197339e-01
-8.66557956e-01 2.88823873e-01 7.66400874e-01 -2.43448719e-01
-1.21529329e+00 3.60886753e-02 -1.86734170e-01 -2.14131683e-01
3.05357397e-01 5.89272082e-01 1.32868320e-01 -2.58917473e-02
1.06050014e+00 1.67205140e-01 -1.75547346e-01 -4.89184052e-01
4.82571393e-01 9.47515130e-01 4.78496075e-01 -3.90882552e-01
8.59889209e-01 -5.01020700e-02 -6.37382746e-01 -2.66264975e-01
-6.71898603e-01 2.60402486e-02 -1.72607407e-01 -3.87429744e-02
6.62059247e-01 -1.00541306e+00 -2.49788523e-01 2.38754958e-01
-1.35590613e+00 -5.57395458e-01 -8.60412121e-02 3.02104615e-02
-5.02234340e-01 3.57660770e-01 -6.45164490e-01 -3.97150487e-01
-5.34787297e-01 -1.16318846e+00 1.13925910e+00 1.73218787e-01
-1.42770663e-01 -7.88451195e-01 3.66333008e-01 4.08805192e-01
5.88229895e-01 3.89248550e-01 8.19611132e-01 -3.29356611e-01
-6.39733016e-01 -2.62165248e-01 -4.98920530e-01 4.70353425e-01
1.38864042e-02 6.44794106e-01 -8.75666261e-01 -2.20412999e-01
-4.86513972e-01 -3.18476021e-01 6.73055947e-01 9.16339457e-02
1.20715046e+00 -4.56707567e-01 -4.39140201e-02 6.87866211e-01
1.68743587e+00 3.82767260e-01 8.05533051e-01 5.60762882e-01
8.92540693e-01 2.35276461e-01 3.78359407e-01 6.07913017e-01
6.04541421e-01 7.92994380e-01 2.71375209e-01 -1.64107755e-02
-4.85169023e-01 -5.48225522e-01 7.07896650e-01 8.62124085e-01
1.08178690e-01 -4.72939730e-01 -1.13420510e+00 6.37350321e-01
-1.68451023e+00 -8.95296216e-01 -1.48663655e-01 2.05461025e+00
9.95116651e-01 2.80510426e-01 1.02120474e-01 -2.50195801e-01
6.90504313e-01 -1.25375196e-01 -3.16761643e-01 -6.29222572e-01
1.58644915e-01 2.47089595e-01 3.83128405e-01 1.29744112e-01
-7.60639250e-01 7.82297492e-01 6.78628969e+00 5.40791988e-01
-8.49553466e-01 1.07846387e-01 5.24674714e-01 -1.33890823e-01
-5.99422812e-01 6.55609146e-02 -6.29809499e-01 7.21276820e-01
1.36161208e+00 -3.05117548e-01 7.76224554e-01 1.17081904e+00
1.45563066e-01 2.58917570e-01 -1.14803195e+00 9.69749689e-01
3.02444220e-01 -1.65607798e+00 -2.82256342e-02 -7.81277418e-02
1.19701660e+00 -4.81130183e-02 3.34331505e-02 4.14510876e-01
5.84243178e-01 -7.55028009e-01 9.95628834e-01 4.74654108e-01
1.20256698e+00 -3.58062148e-01 4.31638956e-01 -1.07989669e-01
-8.77219200e-01 -1.47210151e-01 -2.59857506e-01 1.82044804e-01
-4.00528796e-02 5.70084929e-01 -1.26821733e+00 1.75255924e-01
7.04287827e-01 7.82792687e-01 -1.14540184e+00 1.27595520e+00
-3.60406458e-01 7.64004111e-01 1.86452225e-01 -2.53848016e-01
4.59026918e-02 1.98589116e-01 1.26450926e-01 1.58950186e+00
7.84346342e-01 -6.25727534e-01 -2.88129926e-01 1.27140617e+00
-4.92525041e-01 -6.62940554e-03 -6.03822470e-01 -3.48396063e-01
5.19183159e-01 1.32272863e+00 -4.50002909e-01 -4.27078247e-01
-7.20942080e-01 1.21601152e+00 3.42502773e-01 3.26824069e-01
-1.31205392e+00 -8.86002064e-01 3.09192061e-01 -5.39496448e-03
7.76724577e-01 -6.67643920e-02 -2.87719667e-01 -1.20568538e+00
3.40924740e-01 -1.20827007e+00 -9.31890756e-02 -1.13799024e+00
-7.86949158e-01 6.74358249e-01 -3.98943365e-01 -1.32788563e+00
-2.91889131e-01 -4.75888461e-01 -7.92253435e-01 6.49089217e-01
-1.41790569e+00 -1.07947302e+00 -6.88495696e-01 4.70921807e-02
8.32495034e-01 -3.26523840e-01 4.89339113e-01 3.56983095e-01
-6.63529158e-01 1.12478256e+00 5.16785324e-01 2.11494371e-01
7.69321382e-01 -1.56181777e+00 1.19876242e+00 1.13152885e+00
3.26239318e-01 6.39960825e-01 4.65187818e-01 -5.13238251e-01
-1.34480047e+00 -1.31145084e+00 5.96936762e-01 -7.30009437e-01
5.76092899e-01 -5.15723169e-01 -4.49344873e-01 6.25101328e-01
6.06704652e-01 -4.16351795e-01 4.43154693e-01 -1.25843883e-01
-5.37027121e-01 8.09875503e-02 -8.19913983e-01 7.64531493e-01
9.41561341e-01 -5.11414230e-01 -2.46470481e-01 4.17395771e-01
1.06115174e+00 -4.53606546e-01 -9.74675059e-01 -2.47031674e-01
7.72460580e-01 -1.25685728e+00 7.55334198e-01 -2.51559287e-01
1.29891431e+00 -1.27825767e-01 8.04268289e-03 -1.31470370e+00
-4.55174446e-01 -8.39145899e-01 -1.21956080e-01 1.50245440e+00
8.94453287e-01 -4.15794477e-02 6.58764064e-01 5.16989172e-01
-3.03719699e-01 -6.54284477e-01 -2.60823905e-01 -7.29513407e-01
-1.38970613e-01 -1.96971908e-01 5.68140805e-01 5.13800442e-01
-6.13424517e-02 4.52844173e-01 -4.37095881e-01 -3.37320268e-01
4.61751401e-01 -4.15543318e-02 9.65388536e-01 -6.47070527e-01
-7.06406415e-01 -4.54857469e-01 6.18197117e-03 -8.98051143e-01
-3.36166680e-01 -8.66027296e-01 1.84242249e-01 -1.78928196e+00
6.31585002e-01 3.48737910e-02 4.67896387e-02 5.88213205e-01
-2.55750179e-01 2.86918253e-01 3.64829481e-01 1.85298577e-01
-7.56290793e-01 3.37654322e-01 1.02525759e+00 -2.86175251e-01
-1.41653001e-01 -1.49007320e-01 -1.00704932e+00 4.51952487e-01
7.03984618e-01 -4.32429522e-01 -3.74136925e-01 -8.74048352e-01
5.80649018e-01 3.54179926e-02 1.62956312e-01 -1.16370916e+00
-1.26246586e-01 2.17052788e-01 2.89209664e-01 -1.20192327e-01
-1.57222480e-01 -3.53977263e-01 -4.45416495e-02 3.37068588e-01
-6.76126122e-01 4.02496129e-01 1.92204982e-01 3.97829413e-01
3.88351008e-02 -5.27903318e-01 5.74492097e-01 -2.65003115e-01
-5.07709384e-01 -2.40731344e-01 -2.81057775e-01 3.19663972e-01
8.48112941e-01 -2.74912328e-01 -6.35269165e-01 -4.22468305e-01
-1.89767897e-01 -7.99210891e-02 8.02397072e-01 5.93300223e-01
7.76531219e-01 -1.20702672e+00 -7.84339905e-01 3.13773304e-02
4.06098962e-01 -5.33377647e-01 -2.17213094e-01 5.34918725e-01
-9.47198033e-01 4.11953449e-01 -2.64213562e-01 -5.28333068e-01
-1.34006023e+00 3.85331392e-01 1.29308879e-01 -2.94037253e-01
-3.00667048e-01 7.93003201e-01 9.02364999e-02 -2.75343031e-01
2.04560801e-01 -4.88077790e-01 8.53995308e-02 -3.06029260e-01
5.16302228e-01 4.19000238e-01 1.37884676e-01 8.41032714e-02
-1.30413741e-01 5.83972871e-01 -1.86887592e-01 -2.12536067e-01
1.43656790e+00 -4.71293256e-02 1.97235286e-01 1.52909150e-02
1.31051362e+00 1.26404315e-01 -1.59389853e+00 1.16118252e-01
1.17431439e-01 -6.45252645e-01 -2.13184208e-01 -1.26646495e+00
-1.06192875e+00 7.05219090e-01 5.92346668e-01 3.58824193e-01
9.86997783e-01 -1.23663604e-01 8.70924950e-01 3.76916498e-01
1.91339791e-01 -1.06423068e+00 6.42406225e-01 3.78599882e-01
9.21060681e-01 -1.28186488e+00 -2.05732271e-01 -8.68230537e-02
-9.79118049e-01 1.19530606e+00 6.98866963e-01 7.81961605e-02
-6.33376539e-02 4.00091738e-01 4.90562990e-02 -6.78201169e-02
-1.10772777e+00 1.52683333e-02 2.86638737e-01 5.81589937e-01
8.50646794e-01 -3.12567890e-01 -1.21869788e-01 4.44006860e-01
-2.14234322e-01 1.65726051e-01 1.08982968e+00 9.38801885e-01
-2.54980981e-01 -1.14902067e+00 -2.39308387e-01 5.81513762e-01
-6.23157263e-01 -4.84277308e-01 -4.32191342e-01 4.70424324e-01
-2.17944771e-01 9.14916873e-01 -2.04572320e-01 -5.99798679e-01
5.00971794e-01 -1.11253023e-01 3.67740542e-01 -6.77344084e-01
-7.66016185e-01 -6.05359627e-03 4.11482394e-01 -6.36044860e-01
-1.38380095e-01 -6.93114638e-01 -7.29985118e-01 -7.47209266e-02
-2.97379583e-01 -1.04887731e-01 9.40686822e-01 3.97086173e-01
7.41770923e-01 9.04018998e-01 6.81212008e-01 -4.76415753e-01
-5.93675852e-01 -7.88486838e-01 9.25526172e-02 3.73308331e-01
1.98222339e-01 4.11173794e-03 -4.61939722e-04 7.08831489e-01] | [7.783567905426025, 7.795492172241211] |
c009cbf6-3779-483b-98bf-487ced354785 | correcting-discount-factor-mismatch-in-on | 2306.13284 | null | https://arxiv.org/abs/2306.13284v1 | https://arxiv.org/pdf/2306.13284v1.pdf | Correcting discount-factor mismatch in on-policy policy gradient methods | The policy gradient theorem gives a convenient form of the policy gradient in terms of three factors: an action value, a gradient of the action likelihood, and a state distribution involving discounting called the \emph{discounted stationary distribution}. But commonly used on-policy methods based on the policy gradient theorem ignores the discount factor in the state distribution, which is technically incorrect and may even cause degenerate learning behavior in some environments. An existing solution corrects this discrepancy by using $\gamma^t$ as a factor in the gradient estimate. However, this solution is not widely adopted and does not work well in tasks where the later states are similar to earlier states. We introduce a novel distribution correction to account for the discounted stationary distribution that can be plugged into many existing gradient estimators. Our correction circumvents the performance degradation associated with the $\gamma^t$ correction with a lower variance. Importantly, compared to the uncorrected estimators, our algorithm provides improved state emphasis to evade suboptimal policies in certain environments and consistently matches or exceeds the original performance on several OpenAI gym and DeepMind suite benchmarks. | ['A. Rupam Mahmood', 'Gautham Vasan', 'Fengdi Che'] | 2023-06-23 | null | null | null | null | ['policy-gradient-methods', 'openai-gym'] | ['methodology', 'playing-games'] | [-9.10883695e-02 -1.56497121e-01 -5.79787314e-01 -4.31941450e-01
-4.52479005e-01 -4.98283565e-01 5.36944449e-01 1.76244915e-01
-1.03474605e+00 1.18583429e+00 1.82008758e-01 -6.97194219e-01
-7.33587295e-02 -3.83435398e-01 -7.67381608e-01 -7.56949663e-01
5.35894781e-02 8.01946521e-02 5.59219956e-01 -1.94922820e-01
4.79488671e-01 1.08183153e-01 -1.61358654e+00 -2.22745627e-01
1.01407194e+00 9.43786800e-01 1.59804538e-01 5.54271400e-01
-1.64270490e-01 7.98842907e-01 -6.94152296e-01 -1.78278357e-01
5.06332934e-01 -5.68557918e-01 -5.35369992e-01 -2.64047951e-01
5.19559205e-01 -8.46450627e-01 -2.04754889e-01 1.25808048e+00
4.34948146e-01 5.04616141e-01 4.98248935e-01 -1.15405345e+00
-3.51249456e-01 4.62145001e-01 -6.58881366e-01 5.62604368e-01
1.22860588e-01 1.20250121e-01 7.63612211e-01 -2.11493149e-01
3.12652647e-01 1.35750973e+00 6.80346906e-01 6.57584071e-01
-1.15206790e+00 -5.31892121e-01 7.37169743e-01 5.08515835e-02
-8.74747634e-01 -3.25439930e-01 3.17161947e-01 -2.21305639e-01
1.08243501e+00 -6.38959408e-02 7.28289545e-01 1.07487619e+00
6.71981394e-01 9.08887148e-01 1.46502292e+00 -2.27054253e-01
6.39379859e-01 1.16374590e-01 1.22655155e-02 7.12052941e-01
3.71114016e-01 5.22704124e-01 -3.54655504e-01 -3.13015103e-01
8.53567481e-01 -1.19972505e-01 -1.87013298e-01 -4.88032162e-01
-8.41639638e-01 7.61845469e-01 1.04486234e-01 -1.71143532e-01
-3.02673668e-01 5.62139452e-01 4.44316745e-01 3.29600900e-01
6.18646145e-01 2.59963572e-01 -5.11402786e-01 -6.39157116e-01
-7.72322178e-01 6.87820792e-01 7.73017347e-01 6.39431953e-01
5.60527742e-01 4.16400582e-01 -5.58617830e-01 6.01089656e-01
3.63517731e-01 4.49129105e-01 5.74232578e-01 -1.45711851e+00
2.76236743e-01 2.47719899e-01 6.87291801e-01 -4.42885727e-01
-3.32278222e-01 -6.41309619e-01 -1.24169677e-01 7.65388131e-01
1.04530072e+00 -4.20427412e-01 -9.96898592e-01 2.21081591e+00
3.87039363e-01 1.09423086e-01 -1.81605071e-01 9.09749746e-01
-1.63758233e-01 2.85661459e-01 2.23561421e-01 -3.27208132e-01
1.03511155e+00 -8.64623368e-01 -8.94998193e-01 -5.42387366e-01
5.04506648e-01 -5.03694355e-01 1.27061677e+00 3.25562209e-01
-1.00562942e+00 -1.43197151e-02 -1.16808951e+00 2.61099517e-01
-1.60748646e-01 -2.76539147e-01 8.71583045e-01 7.30410576e-01
-1.11711109e+00 1.13323700e+00 -9.51313913e-01 -3.00980002e-01
1.58510372e-01 1.65776640e-01 2.66657472e-01 2.70956099e-01
-1.11640799e+00 1.45768023e+00 3.37246299e-01 -1.63761988e-01
-7.86382437e-01 -7.03789532e-01 -7.15497375e-01 -7.73992911e-02
5.80710053e-01 -4.28549260e-01 1.85112548e+00 -9.17216361e-01
-2.09040308e+00 2.65691847e-01 6.91035241e-02 -6.46704495e-01
9.10011351e-01 -4.69145179e-01 -2.25674175e-02 -1.64098427e-01
-2.78601535e-02 3.95695508e-01 9.47627485e-01 -8.21534574e-01
-7.84266889e-01 -3.03396016e-01 1.91081271e-01 6.66503966e-01
-8.81208405e-02 -2.79381096e-01 -1.13705270e-01 -4.68575507e-01
-7.64503479e-02 -1.03950655e+00 -4.20377314e-01 1.15638733e-01
1.66867778e-01 -2.25814000e-01 5.11464119e-01 -5.36825478e-01
1.20786989e+00 -1.96068072e+00 -2.72056103e-01 -8.14211182e-03
-1.23657264e-01 3.85638744e-01 1.74041986e-01 2.15119779e-01
2.44255528e-01 -8.36664811e-02 -1.54691175e-01 -2.31238946e-01
2.24209845e-01 4.98049676e-01 -3.88012320e-01 7.31063426e-01
-1.80423990e-01 3.36578578e-01 -1.35419548e+00 -1.66242763e-01
1.46911204e-01 2.00203404e-01 -7.41985798e-01 -1.43631116e-01
-3.48073572e-01 3.13337177e-01 -4.79951292e-01 1.10365301e-01
4.47768301e-01 9.94545966e-02 2.05763862e-01 3.46975863e-01
-4.84683037e-01 6.58967733e-01 -1.39258361e+00 1.67253160e+00
-4.52996343e-02 3.02479416e-01 1.81957945e-01 -1.02189386e+00
5.68115294e-01 1.95067123e-01 5.24920404e-01 -6.30568862e-01
1.47488832e-01 3.34798247e-01 1.56642556e-01 -1.17391363e-01
6.26734972e-01 -3.53845328e-01 2.20587552e-01 5.17641425e-01
-2.50440519e-02 -1.94755137e-01 3.60900134e-01 1.27382264e-01
1.05214083e+00 7.26276815e-01 1.90590501e-01 -7.84548163e-01
-1.34266131e-02 -1.46381512e-01 9.27240074e-01 1.08973980e+00
-8.83126557e-01 1.34145826e-01 7.98336446e-01 -2.39173412e-01
-9.04586673e-01 -1.08397830e+00 -6.60185367e-02 1.24668562e+00
2.21055567e-01 -2.90383399e-01 -6.75471127e-01 -7.21587300e-01
3.86742711e-01 8.89228880e-01 -5.14058828e-01 -6.22677863e-01
-2.25454524e-01 -8.27468276e-01 5.56664944e-01 6.81286514e-01
4.83264178e-01 -7.03638077e-01 -9.27568793e-01 3.94813985e-01
4.92843380e-03 -7.10678577e-01 -6.62639678e-01 4.06905651e-01
-1.07724035e+00 -7.90912390e-01 -7.80443072e-01 -1.49486467e-01
3.54975611e-01 -3.30436304e-02 9.59929883e-01 -3.07786822e-01
1.34812668e-01 4.84582335e-01 -6.37695119e-02 -7.68816888e-01
-3.95877719e-01 -2.46816665e-01 5.77449381e-01 -3.83037269e-01
2.53744125e-01 -3.62459540e-01 -8.11868548e-01 4.39917222e-02
-5.70796013e-01 -2.98663348e-01 3.57163310e-01 8.28798115e-01
3.02302003e-01 -1.92432180e-01 6.58147216e-01 -6.76351309e-01
8.34417641e-01 -3.70493591e-01 -8.73748422e-01 -7.26240873e-02
-1.31596410e+00 6.57144606e-01 4.96231735e-01 -6.73017979e-01
-1.27229130e+00 -1.72429278e-01 -1.32012501e-01 -3.24290514e-01
2.69849688e-01 1.34576157e-01 3.88945431e-01 9.62349400e-02
7.19312429e-01 6.62066936e-02 3.97708535e-01 -5.38497329e-01
2.19016075e-01 2.75926977e-01 3.66402924e-01 -9.85047221e-01
2.47624099e-01 3.99989516e-01 -8.67340341e-02 -2.69957066e-01
-9.57466960e-01 -2.13847235e-01 -1.60852186e-02 -9.66847390e-02
4.86004263e-01 -8.28274965e-01 -8.19261014e-01 5.43417692e-01
-7.69708037e-01 -6.76302314e-01 -7.05218256e-01 8.71729553e-01
-7.92559147e-01 4.11920726e-01 -8.14904869e-01 -1.12750900e+00
-1.68708429e-01 -1.11982346e+00 5.94727755e-01 6.40282214e-01
6.36851788e-02 -9.45986390e-01 1.81010053e-01 -3.17993224e-01
6.89619482e-01 -1.75863281e-02 7.74785995e-01 -3.32677454e-01
1.25407875e-01 7.15816626e-03 3.18757184e-02 6.98240340e-01
8.31217915e-02 -3.78631838e-02 -6.94432914e-01 -5.80416262e-01
4.60662656e-02 -2.05794498e-01 9.15244460e-01 6.64120674e-01
6.81927204e-01 -2.28251562e-01 1.00299791e-02 3.13230366e-01
1.28586483e+00 4.39374030e-01 3.95881921e-01 5.38182080e-01
1.79835498e-01 1.94994569e-01 5.64701617e-01 8.32914352e-01
3.58866930e-01 3.55416596e-01 4.44844365e-01 2.33158037e-01
1.84682295e-01 -3.34438473e-01 8.16388071e-01 3.64894569e-01
-1.79130346e-01 1.02416255e-01 -6.65565789e-01 3.80738735e-01
-2.03730536e+00 -1.00337625e+00 1.19838186e-01 2.54632950e+00
1.11491442e+00 5.25395632e-01 2.26237774e-01 -1.17598578e-01
5.49268186e-01 2.39337832e-01 -1.11177647e+00 -6.89051747e-01
1.97657928e-01 1.91099048e-02 1.02235150e+00 7.01575339e-01
-7.92398214e-01 8.67879391e-01 7.86521912e+00 6.45764291e-01
-1.09008169e+00 1.72394246e-01 2.29703248e-01 -2.52990901e-01
-2.23604307e-01 2.94671863e-01 -9.70762789e-01 6.32999301e-01
9.87693489e-01 -4.29881692e-01 5.62760890e-01 1.12949717e+00
2.27220491e-01 -5.84285378e-01 -9.17052269e-01 5.34186900e-01
-3.38555455e-01 -8.77556980e-01 -3.42559069e-01 1.05429642e-01
6.29937530e-01 2.04878628e-01 1.67136222e-01 7.44427264e-01
8.30880046e-01 -3.80661905e-01 1.05651557e+00 3.05084288e-01
4.89308238e-01 -7.75923908e-01 4.53773260e-01 4.71690029e-01
-7.98592389e-01 -3.39662284e-01 -4.48204815e-01 -5.35433412e-01
-2.07000058e-02 4.33508158e-01 -6.17374778e-01 3.67984585e-02
6.87106550e-01 4.05603588e-01 -2.23651543e-01 9.94233906e-01
-2.87682474e-01 5.62740564e-01 -5.33111751e-01 -1.20507278e-01
5.75958729e-01 -3.22634012e-01 7.35839307e-01 8.68251979e-01
1.56711221e-01 -2.51818240e-01 2.59169310e-01 5.79282880e-01
3.09642851e-01 -2.28061602e-01 -3.61871421e-01 3.59533690e-02
3.47035319e-01 7.72939980e-01 -7.12888598e-01 -5.27153373e-01
-2.18141913e-01 8.48905981e-01 4.27165002e-01 6.03964746e-01
-9.53536034e-01 -2.31293872e-01 1.03423119e+00 3.96421999e-02
2.51427948e-01 -3.80009264e-01 3.76818329e-03 -1.21579218e+00
9.37438011e-02 -8.59002352e-01 3.87682378e-01 -3.37733805e-01
-9.36296344e-01 1.61988452e-01 4.25471991e-01 -9.63726223e-01
-5.82296610e-01 -5.55738986e-01 -3.81796360e-01 7.54643440e-01
-1.58344543e+00 -1.32676333e-01 2.62091607e-01 5.94230354e-01
4.04594243e-01 2.02072114e-01 5.74445784e-01 3.33744615e-01
-6.39959693e-01 5.07232547e-01 4.79514688e-01 -4.40506250e-01
9.38471973e-01 -1.72974503e+00 7.25881830e-02 9.33692515e-01
-4.93091702e-01 6.31541848e-01 1.17861032e+00 -6.42613828e-01
-1.08661425e+00 -6.34442806e-01 3.53291661e-01 -2.42633507e-01
6.44425631e-01 6.55662864e-02 -8.77078474e-01 8.30799639e-01
4.62325923e-02 -1.43710803e-03 2.80740976e-01 7.78733343e-02
-1.92590401e-01 -3.55874375e-02 -1.17202950e+00 7.60852754e-01
9.74423647e-01 -3.11105013e-01 -5.36774576e-01 1.45697013e-01
5.22251248e-01 -6.84568405e-01 -6.42904758e-01 2.54010051e-01
6.67977393e-01 -1.01420212e+00 5.31454742e-01 -8.25174332e-01
-1.17565446e-01 -1.33125797e-01 -8.43638554e-02 -1.62352860e+00
-2.98594385e-01 -7.67837703e-01 -5.27867019e-01 8.51669967e-01
1.95318148e-01 -8.79119992e-01 6.95281446e-01 9.51463521e-01
-2.79544413e-01 -8.44955146e-01 -1.11419487e+00 -1.17089593e+00
3.24309289e-01 -3.48286688e-01 3.28714460e-01 6.06477797e-01
1.93726495e-02 -2.86698993e-03 -4.43906903e-01 -2.41447419e-01
7.42556572e-01 -2.20975250e-01 2.28159934e-01 -9.18359339e-01
-4.29698199e-01 -6.99872017e-01 2.28067078e-02 -1.35992956e+00
1.57336518e-01 -4.35573071e-01 3.37562561e-01 -1.43935478e+00
-1.40938789e-01 -4.30350721e-01 -7.06488311e-01 5.26136458e-01
-3.87907624e-01 -3.01805884e-01 2.00670078e-01 -9.94977206e-02
-4.84889686e-01 7.55910695e-01 1.30454731e+00 2.19179437e-01
-3.74794632e-01 6.32204860e-02 -7.34086633e-01 9.63779032e-01
8.90756667e-01 -6.38566852e-01 -5.90101838e-01 -4.60155129e-01
1.25621483e-01 -1.89575121e-01 1.50427833e-01 -9.42800343e-01
1.82548851e-01 -5.42122424e-01 2.41071075e-01 -1.78978205e-01
2.09799722e-01 -4.62070733e-01 -2.79390305e-01 7.30818033e-01
-3.60121131e-01 2.85041571e-01 2.81991839e-01 8.62919807e-01
3.97339255e-01 -1.83485255e-01 9.70166743e-01 -1.78802252e-01
-7.19380260e-01 1.19888872e-01 -7.26772428e-01 2.16574088e-01
7.68380940e-01 -9.01370570e-02 -3.94491524e-01 -4.68329519e-01
-3.14716041e-01 3.84132415e-01 4.45162505e-01 4.54498976e-01
1.73947111e-01 -1.23571074e+00 -3.68929863e-01 -5.49407825e-02
-3.52076411e-01 -3.77960473e-01 -5.89226335e-02 9.64870453e-01
-2.97290385e-01 1.04687624e-01 -2.24910617e-01 -2.82411247e-01
-5.58774710e-01 2.86326349e-01 6.20422482e-01 -4.01633769e-01
-6.65119052e-01 6.60637736e-01 -3.63114923e-01 -1.90696344e-01
4.70100790e-01 -6.10292435e-01 1.41436845e-01 -1.05479218e-01
6.52400136e-01 5.46642303e-01 -6.90207854e-02 -1.86416537e-01
-3.63549531e-01 -2.08654199e-02 -1.89212024e-01 -5.96589804e-01
1.02107358e+00 -9.96252671e-02 4.53703910e-01 7.41584063e-01
7.51801431e-01 -2.64762819e-01 -2.11121345e+00 -2.42050841e-01
-2.39064340e-02 -5.23788095e-01 3.36332440e-01 -8.20735693e-01
-7.68392682e-01 4.43276048e-01 7.89582253e-01 5.91829792e-02
7.45724797e-01 -6.43133581e-01 6.28260016e-01 3.61237586e-01
4.74461317e-01 -1.72916853e+00 -1.97160646e-01 8.68994117e-01
4.47095603e-01 -1.20088863e+00 3.39421213e-01 3.39478910e-01
-7.56070137e-01 7.01153278e-01 7.41363406e-01 -2.88917929e-01
6.95184112e-01 2.91911006e-01 1.01462893e-01 3.67513627e-01
-8.33079815e-01 -2.85106599e-01 -1.01566397e-01 4.32724983e-01
3.53660196e-01 -5.76624982e-02 -9.37915504e-01 3.93736154e-01
3.05398293e-02 7.88759291e-02 4.15317148e-01 1.53088403e+00
-5.88268578e-01 -1.04559124e+00 -1.63876399e-01 5.96802056e-01
-7.15429127e-01 2.67139226e-02 1.81848228e-01 6.48178697e-01
-1.42745525e-01 8.49701464e-01 2.04115078e-01 -1.41936749e-01
4.13468033e-01 2.82117754e-01 6.13113523e-01 -2.91577935e-01
-5.71199656e-01 1.50877365e-03 -4.97443341e-02 -8.91386151e-01
-3.04194152e-01 -5.62698007e-01 -1.37387943e+00 -4.49409544e-01
-2.81570345e-01 2.10690916e-01 9.12093043e-01 1.07034004e+00
3.84038925e-01 4.63358313e-01 3.16864640e-01 -9.10821199e-01
-1.55209327e+00 -8.27617824e-01 -8.33061755e-01 3.33396494e-01
4.55317557e-01 -1.12114179e+00 -6.29691303e-01 -3.85817260e-01] | [4.110219955444336, 2.373169183731079] |
349e37e4-ff18-4855-a0f5-ccf1f88dc9a0 | neural-rankers-for-effective-screening | 2212.09017 | null | https://arxiv.org/abs/2212.09017v1 | https://arxiv.org/pdf/2212.09017v1.pdf | Neural Rankers for Effective Screening Prioritisation in Medical Systematic Review Literature Search | Medical systematic reviews typically require assessing all the documents retrieved by a search. The reason is two-fold: the task aims for ``total recall''; and documents retrieved using Boolean search are an unordered set, and thus it is unclear how an assessor could examine only a subset. Screening prioritisation is the process of ranking the (unordered) set of retrieved documents, allowing assessors to begin the downstream processes of the systematic review creation earlier, leading to earlier completion of the review, or even avoiding screening documents ranked least relevant. Screening prioritisation requires highly effective ranking methods. Pre-trained language models are state-of-the-art on many IR tasks but have yet to be applied to systematic review screening prioritisation. In this paper, we apply several pre-trained language models to the systematic review document ranking task, both directly and fine-tuned. An empirical analysis compares how effective neural methods compare to traditional methods for this task. We also investigate different types of document representations for neural methods and their impact on ranking performance. Our results show that BERT-based rankers outperform the current state-of-the-art screening prioritisation methods. However, BERT rankers and existing methods can actually be complementary, and thus, further improvements may be achieved if used in conjunction. | ['Guido Zuccon', 'Bevan Koopman', 'Harrisen Scells', 'Shuai Wang'] | 2022-12-18 | null | null | null | null | ['document-ranking'] | ['natural-language-processing'] | [ 6.47104621e-01 3.81299049e-01 -7.41374731e-01 -3.33215445e-01
-1.41989064e+00 -6.19506836e-01 6.40096188e-01 7.54570842e-01
-6.90374553e-01 6.82348788e-01 4.94422525e-01 -6.01242006e-01
-6.57574713e-01 -7.71739900e-01 -3.63704115e-01 -2.78027385e-01
-8.80555660e-02 9.50635135e-01 3.03677022e-01 -9.97900069e-02
5.25477946e-01 2.62827247e-01 -1.54215491e+00 8.58050764e-01
8.08395326e-01 7.12833762e-01 2.34952688e-01 6.71475708e-01
-1.78880304e-01 8.31906140e-01 -6.96976900e-01 -2.70512193e-01
-5.47441766e-02 -4.06074375e-01 -1.03864932e+00 -4.74996060e-01
4.73717004e-01 -5.37329912e-01 7.55348951e-02 9.50164080e-01
6.34178519e-01 2.91884504e-02 7.87979424e-01 -3.18683058e-01
-9.22234416e-01 8.40947926e-01 -4.39557582e-01 3.25483561e-01
6.03563190e-01 -4.98240471e-01 1.29920447e+00 -8.98909926e-01
6.20617986e-01 1.29755688e+00 5.17774940e-01 6.47993624e-01
-7.50913560e-01 -5.45334280e-01 2.20254317e-01 -4.81928065e-02
-9.34141576e-01 -4.82255012e-01 2.96841323e-01 -4.58254009e-01
9.53967631e-01 4.72347587e-01 4.29998100e-01 6.26942635e-01
3.32809508e-01 7.84962237e-01 9.87018406e-01 -6.84045613e-01
1.77001953e-01 1.41100615e-01 6.03527367e-01 6.20340586e-01
7.04936087e-01 5.55094220e-02 -5.46186805e-01 -5.83090484e-01
2.18030691e-01 5.81945963e-02 -2.34383583e-01 5.12570031e-02
-1.03679931e+00 9.62613642e-01 2.54968882e-01 2.46421039e-01
-4.05341893e-01 6.53832927e-02 6.08274996e-01 3.18305552e-01
6.44359767e-01 1.05750251e+00 -4.80460972e-01 2.52789874e-02
-1.39791310e+00 3.17902356e-01 6.28952444e-01 5.30053914e-01
3.82113755e-01 -7.77747273e-01 -7.79907882e-01 1.10201383e+00
3.42509747e-01 1.46186993e-01 5.82923889e-01 -7.11326599e-01
6.23368800e-01 9.66976523e-01 1.61400303e-01 -7.33667672e-01
-6.21082246e-01 -2.13166565e-01 -6.83760464e-01 -1.24534450e-01
5.74702658e-02 -2.13263556e-01 -1.16285062e+00 1.28820288e+00
-1.56970814e-01 -6.44256353e-01 -3.14213894e-02 7.62626827e-01
1.18334186e+00 5.54865539e-01 1.71725154e-01 -4.03446883e-01
1.61016548e+00 -8.75724316e-01 -8.82753432e-01 -3.21638286e-01
7.23970175e-01 -7.59183228e-01 8.57554972e-01 6.12195194e-01
-1.41363823e+00 -2.95125544e-01 -9.68712687e-01 -1.07088424e-01
-4.97351766e-01 4.10668939e-01 7.66933501e-01 6.56416535e-01
-1.34851670e+00 6.92308366e-01 -4.83353108e-01 -1.81019202e-01
3.34908605e-01 5.39968729e-01 -1.24118045e-01 -2.06719190e-01
-1.57595360e+00 1.11108327e+00 1.33066580e-01 3.42508972e-01
-7.20840752e-01 -5.44882357e-01 -6.83468580e-01 9.07965153e-02
4.55103129e-01 -6.37111425e-01 1.58857286e+00 -4.90982920e-01
-1.04142475e+00 1.05603969e+00 -2.35417485e-01 -1.66842714e-01
2.33839825e-01 -3.63702446e-01 -1.16898462e-01 2.66456008e-01
2.31408849e-01 4.78048474e-01 5.26932515e-02 -9.04287994e-01
-8.05745840e-01 -2.69573927e-01 2.87825614e-01 4.94331688e-01
-4.54301149e-01 6.14045918e-01 -7.34953701e-01 -4.27486390e-01
-1.33261770e-01 -7.92579889e-01 -5.79584837e-01 -2.55186200e-01
-3.41450363e-01 -4.95938450e-01 -1.34539921e-02 -5.97375453e-01
1.69136715e+00 -1.59628749e+00 -1.26934782e-01 2.74264485e-01
2.47442022e-01 4.67785954e-01 -3.06664795e-01 5.44309318e-01
-9.44004282e-02 7.45883703e-01 7.47006340e-03 -1.35969728e-01
-1.99395090e-01 -3.75703692e-01 -6.40473962e-02 1.89482406e-01
3.22742313e-01 8.78401935e-01 -1.25201511e+00 -5.69151998e-01
-2.91694134e-01 2.01387778e-01 -3.61233711e-01 -1.88176092e-02
8.20106342e-02 -3.07478577e-01 -6.68678820e-01 7.03698158e-01
3.23562115e-01 -4.39555854e-01 2.99174875e-01 1.00715481e-01
-1.90187860e-02 1.11471748e+00 -8.45401525e-01 1.15520728e+00
-3.89086455e-01 5.66051424e-01 -3.10504675e-01 -8.88281047e-01
7.88459122e-01 3.69561374e-01 3.53107393e-01 -8.19522738e-01
-2.15048656e-01 4.53610897e-01 2.45280951e-01 -2.97513455e-01
7.03106403e-01 -2.47513615e-02 -2.54701212e-04 7.29569554e-01
-4.45093393e-01 -6.74145371e-02 5.84166348e-01 2.70623714e-01
1.44279814e+00 -3.18933457e-01 5.34719825e-01 -3.10016721e-01
4.33749050e-01 2.00023636e-01 1.39940962e-01 1.34528625e+00
1.91362262e-01 7.92892039e-01 7.37001538e-01 -3.13126355e-01
-5.22956669e-01 -7.33445823e-01 -4.30769831e-01 1.17791152e+00
-2.19118029e-01 -4.98379797e-01 -4.16976184e-01 -9.65440512e-01
-1.08193256e-01 5.17244995e-01 -8.85647714e-01 -2.70890743e-01
-5.26243091e-01 -9.28267956e-01 4.33385104e-01 5.26891530e-01
-1.11619681e-01 -1.35816610e+00 -8.18731248e-01 3.28294814e-01
9.25463662e-02 -3.24282795e-01 -5.10571599e-01 6.26462042e-01
-1.04685307e+00 -1.32690156e+00 -1.16894352e+00 -7.94198990e-01
8.37392330e-01 2.66043782e-01 1.53067970e+00 5.16234338e-01
2.36941446e-02 2.08141342e-01 -4.20670062e-01 -6.19996250e-01
-5.63456297e-01 3.14559519e-01 -1.67723075e-01 -7.38292277e-01
5.47879100e-01 3.07885796e-01 -8.51746202e-01 1.38364658e-01
-9.42582846e-01 -1.87948287e-01 8.86884451e-01 9.25608575e-01
6.73105240e-01 -2.25677863e-02 7.80133486e-01 -1.44455373e+00
1.42332304e+00 -3.18234652e-01 -4.51963425e-01 6.28359735e-01
-1.00485194e+00 2.13771775e-01 2.76879966e-01 -4.94862616e-01
-8.55112135e-01 -4.91501153e-01 -3.61393653e-02 1.81707904e-01
3.80714506e-01 1.21874309e+00 3.98186296e-01 3.19234222e-01
7.86405563e-01 -2.23705992e-01 -2.30445653e-01 -3.66160244e-01
9.36345663e-03 7.31715858e-01 -1.21388473e-01 -3.58739555e-01
2.46410012e-01 1.22954056e-01 -2.08019778e-01 -2.36534044e-01
-9.26405549e-01 -7.81274557e-01 -2.15255708e-01 -1.15068089e-02
5.51588833e-01 -7.73843765e-01 -3.29368055e-01 -1.66149959e-01
-1.32774389e+00 -2.03044400e-01 -1.84752122e-01 4.11238134e-01
-4.86496044e-03 2.54770964e-01 -7.33026028e-01 -8.04313719e-01
-7.19169974e-01 -1.39061177e+00 1.25291121e+00 1.55150127e-02
-6.32681668e-01 -7.19276428e-01 2.89987773e-01 1.78694963e-01
2.97890335e-01 -3.24908435e-01 1.14239681e+00 -9.58664119e-01
-1.88772559e-01 -6.18582010e-01 -3.39039803e-01 1.97484255e-01
3.06735337e-01 -5.38001209e-02 -7.26062953e-01 -4.43787724e-02
-2.37557933e-01 -4.31190342e-01 1.28932023e+00 7.39974976e-01
1.21896505e+00 -3.31716001e-01 -7.05948293e-01 -3.75835039e-02
9.12683904e-01 5.61560810e-01 6.49954855e-01 5.20159900e-01
3.87975186e-01 1.01298070e+00 9.34186637e-01 6.61276877e-02
3.20288122e-01 5.06394506e-01 3.37087549e-02 -3.00900728e-01
3.53994630e-02 4.54460159e-02 1.37686446e-01 7.36728549e-01
6.34791702e-02 -3.13494027e-01 -1.14514017e+00 7.45551467e-01
-1.71451211e+00 -7.17716932e-01 2.22212728e-02 2.44637895e+00
1.26327133e+00 2.84898818e-01 -9.75889340e-02 9.93330479e-02
6.72788560e-01 -2.20674518e-02 -2.78919160e-01 -8.25005114e-01
1.01562314e-01 2.41887301e-01 4.14096057e-01 4.43365663e-01
-1.02475488e+00 4.93622750e-01 7.21263504e+00 6.01806521e-01
-8.46516073e-01 -1.29460692e-01 1.05886281e+00 -2.70610929e-01
-6.05939209e-01 -8.06486383e-02 -1.10207212e+00 1.51954010e-01
1.00825644e+00 -1.05424106e-01 6.31760508e-02 6.36455834e-01
1.87466532e-01 -2.13231862e-01 -1.46342635e+00 4.94903922e-01
1.25569105e-01 -1.49353600e+00 3.53967458e-01 4.47852910e-02
8.44461203e-01 -5.46066351e-02 -6.71148524e-02 4.94926363e-01
3.84842366e-01 -1.07552969e+00 4.30310011e-01 4.27861571e-01
9.16506708e-01 -5.57595909e-01 1.08828330e+00 4.79710549e-02
-8.35711420e-01 1.32624786e-02 -4.09396470e-01 1.56291798e-01
-2.58337241e-02 8.61632466e-01 -8.25328887e-01 3.77379149e-01
7.66871274e-01 4.43693697e-01 -8.20889592e-01 1.15955830e+00
-1.87849656e-01 5.28195918e-01 3.50186159e-03 -6.08024299e-01
3.58630985e-01 2.47903496e-01 2.12530196e-01 1.40340304e+00
2.69377142e-01 8.01756978e-03 -7.04795569e-02 4.55107301e-01
-1.93649322e-01 2.28172779e-01 -5.27442515e-01 -2.96197832e-01
3.63599896e-01 9.07141328e-01 -8.18624735e-01 -5.24435401e-01
-2.90606827e-01 2.32017532e-01 3.41266215e-01 3.77385706e-01
-2.84139097e-01 -4.92198557e-01 5.67573169e-03 1.92150176e-01
5.93761355e-02 4.71753299e-01 -3.14011127e-01 -6.41118824e-01
6.85998127e-02 -1.14866221e+00 7.38839805e-01 -6.51427627e-01
-1.15850341e+00 6.53317869e-01 2.59713173e-01 -1.02856171e+00
-5.13240039e-01 -6.86313152e-01 -3.53538722e-01 9.19710994e-01
-1.76753557e+00 -5.05054295e-01 1.71946943e-01 -1.59810215e-01
5.53281546e-01 -1.60607249e-01 8.52115214e-01 3.37022543e-01
-1.96433157e-01 5.42756855e-01 -9.77623463e-03 -9.99598652e-02
8.57464790e-01 -1.44940531e+00 1.48177475e-01 5.11408091e-01
-2.03784153e-01 1.26743817e+00 2.66882569e-01 -9.64612782e-01
-1.13838148e+00 -8.71816516e-01 1.49379146e+00 -6.12801909e-01
4.01460111e-01 -4.86650579e-02 -9.20783937e-01 1.21651471e-01
1.27278939e-01 -5.10483205e-01 7.53172100e-01 6.15043879e-01
-2.79052015e-02 -1.16533153e-02 -6.83470547e-01 7.95761943e-01
8.21488917e-01 -6.10614717e-01 -7.87333488e-01 4.88153428e-01
6.68521404e-01 -3.77152592e-01 -4.74380016e-01 6.61608160e-01
7.29818761e-01 -5.77202439e-01 1.03266251e+00 -5.32582879e-01
9.74453211e-01 -6.58579841e-02 3.09716165e-01 -1.10909724e+00
-4.32827175e-01 -4.04687852e-01 -4.15607169e-02 9.38337445e-01
1.02932084e+00 -3.95326376e-01 6.50489569e-01 6.47928953e-01
-8.17350298e-02 -1.33144343e+00 -6.31723523e-01 -3.38580072e-01
2.25730613e-01 -1.92981303e-01 5.30930758e-01 5.76178432e-01
1.63300052e-01 3.98979038e-01 -3.81874852e-02 -1.83906987e-01
1.17971487e-01 -5.76332696e-02 1.18394151e-01 -1.39914298e+00
-2.12084904e-01 -8.85408819e-01 3.63952935e-01 -7.54316807e-01
-1.19933628e-01 -8.62954974e-01 4.39178646e-01 -2.36960769e+00
5.85235357e-01 -5.47308028e-01 -8.03561687e-01 6.41499162e-01
-7.34691203e-01 -4.69006971e-02 -1.62710220e-01 2.79253721e-01
-8.23233306e-01 4.43784297e-02 1.26875854e+00 -4.48051542e-01
-5.26613653e-01 1.19141519e-01 -1.34381962e+00 4.77402717e-01
5.31890333e-01 -7.80185938e-01 -5.17322123e-01 -4.56302971e-01
8.98359239e-01 1.35400280e-01 -1.33433953e-01 -3.52174878e-01
3.65695715e-01 -8.20022970e-02 3.16899270e-01 -7.94468522e-01
-2.12815311e-02 -3.27163279e-01 -3.51618350e-01 5.32046199e-01
-1.01427269e+00 4.37590390e-01 2.99887210e-01 2.75442094e-01
-3.09811503e-01 -8.26719940e-01 2.89472818e-01 -3.73791099e-01
-1.41811520e-01 -2.13309773e-03 -6.98889375e-01 4.79691476e-02
2.31333539e-01 -1.77353308e-01 -3.82989496e-01 -2.64768481e-01
-1.12247750e-01 3.63336861e-01 3.00466828e-02 4.23918247e-01
7.70408154e-01 -9.94070709e-01 -7.19540715e-01 -3.43691200e-01
2.66088039e-01 1.58016875e-01 -1.18472651e-01 6.97587132e-01
-3.72135550e-01 9.06918526e-01 4.21399593e-01 -2.81130433e-01
-1.23312116e+00 3.94284010e-01 7.26925954e-02 -1.13603711e+00
-2.94290960e-01 8.61688375e-01 4.59041595e-01 -5.27662039e-01
5.10179877e-01 -5.21124601e-01 -8.91367555e-01 4.97507453e-01
8.35859239e-01 2.30002075e-01 4.85025197e-01 1.64300799e-01
-6.89733565e-01 4.64079529e-01 -5.68406224e-01 -2.24994972e-01
1.28703701e+00 2.05376789e-01 -3.30721796e-01 2.49164969e-01
9.19696867e-01 3.45117822e-02 -2.30343193e-01 -3.08483709e-02
3.27171743e-01 -2.64549460e-02 3.41992348e-01 -1.28319860e+00
-7.21297860e-01 6.94172502e-01 3.95012796e-01 2.29954824e-01
1.19471705e+00 5.25435321e-02 1.89700723e-01 7.47606814e-01
3.30162756e-02 -1.13428330e+00 6.17855787e-02 4.03372437e-01
1.07759988e+00 -1.32743680e+00 5.24327457e-01 -1.13383338e-01
-6.03696167e-01 9.64937687e-01 2.76538849e-01 9.18905810e-02
5.37627816e-01 3.32929334e-03 1.65553123e-01 -7.20761716e-01
-9.08558905e-01 -5.05817570e-02 1.04335594e+00 2.08945081e-01
1.09292769e+00 -1.31405577e-01 -1.00235271e+00 7.75516927e-01
1.58579379e-01 1.29621714e-01 4.18114543e-01 9.50531840e-01
-1.22820728e-01 -1.21783400e+00 -4.54255760e-01 1.21506834e+00
-9.76515412e-01 -5.97436845e-01 -7.54869401e-01 6.13837123e-01
-2.36668393e-01 1.17946911e+00 -2.10552722e-01 -4.54435796e-02
4.17686135e-01 -2.65319854e-01 6.99968860e-02 -1.22018564e+00
-9.95966613e-01 2.40279749e-01 6.16806209e-01 -2.93344170e-01
-4.16430384e-01 -7.07416594e-01 -9.68638659e-01 3.46024096e-01
-6.99044108e-01 5.21111429e-01 4.84620184e-01 8.54925275e-01
2.25688487e-01 7.73249090e-01 2.42305025e-01 -5.58043838e-01
-6.42764568e-01 -8.90285730e-01 -1.67215571e-01 4.30921055e-02
5.23615897e-01 -6.78740323e-01 -2.85625935e-01 -4.51332480e-01] | [8.796343803405762, 8.569063186645508] |
66706eb2-fa30-4295-8df9-c64dfc1267c0 | imagenet-pretrained-cnns-for-jpeg | null | null | http://www.ws.binghamton.edu/Fridrich/Research/Alaska-2-Revised.pdf | http://www.ws.binghamton.edu/Fridrich/Research/Alaska-2-Revised.pdf | ImageNet Pretrained CNNs for JPEG Steganalysis | In this paper, we investigate pre-trained computervision deep architectures, such as the EfficientNet, MixNet, and
ResNet for steganalysis. These models pre-trained on ImageNet
can be rather quickly refined for JPEG steganalysis while offering
significantly better performance than CNNs designed purposely
for steganalysis, such as the SRNet, trained from scratch. We
show how different architectures compare on the ALASKA
II dataset. We demonstrate that avoiding pooling/stride in the
first layers enables better performance, as noticed by other
top competitors, which aligns with the design choices of many
CNNs designed for steganalysis. We also show how pre-trained
computer-vision deep architectures perform on the ALASKA I
dataset | ['Jessica Fridrich', 'Eugene Khvedchenya', 'Jan Butora', 'Yassine Yousfi'] | 2020-11-24 | null | null | null | null | ['steganalysis', 'image-steganography'] | ['computer-vision', 'computer-vision'] | [ 4.63553280e-01 4.91108119e-01 2.66155154e-01 1.87624231e-01
-4.21628088e-01 -3.51860136e-01 8.45156133e-01 -6.81053400e-01
-6.71263933e-01 3.24454993e-01 2.85919398e-01 -7.14267313e-01
5.51909924e-01 -7.76778162e-01 -8.88619065e-01 -5.92087626e-01
-1.80603206e-01 -7.73719996e-02 3.72288644e-01 -6.15194201e-01
3.07198763e-01 1.97541788e-01 -1.06896102e+00 2.33580664e-01
3.62185478e-01 7.02708483e-01 -1.69663951e-01 1.06321406e+00
6.43898427e-01 1.11737692e+00 -5.79005778e-01 -7.10422277e-01
5.24892569e-01 -5.69860280e-01 -5.54551780e-01 -1.14787430e-01
7.62354434e-01 -8.75463784e-01 -1.15958214e+00 1.17085946e+00
3.94718915e-01 -3.97925705e-01 5.00584424e-01 -7.91780531e-01
-7.55970120e-01 1.06065404e+00 -2.42587656e-01 3.94459695e-01
-1.96655422e-01 8.94710422e-01 8.52542460e-01 -5.43934226e-01
6.12733364e-01 1.36181140e+00 1.04656136e+00 6.04283690e-01
-8.14510107e-01 -8.71809959e-01 -3.27946961e-01 2.49726221e-01
-1.12859738e+00 -8.34868431e-01 5.56446493e-01 6.13668328e-03
1.26297414e+00 4.08139229e-02 8.04030001e-01 1.52551603e+00
5.73098421e-01 7.12990940e-01 7.41116107e-01 -1.59015417e-01
-1.93267643e-01 -3.73638868e-01 -2.69675642e-01 9.82306004e-01
6.15759194e-01 6.78046823e-01 2.70094350e-02 2.46862248e-01
1.00704205e+00 -2.00114191e-01 -4.46011871e-01 -1.30402386e-01
-1.49556720e+00 1.04385769e+00 7.45088458e-01 2.89425433e-01
-7.44595677e-02 1.07120121e+00 6.72744274e-01 7.64523447e-01
1.34027854e-01 8.78692806e-01 -3.35410595e-01 1.27216309e-01
-1.17635334e+00 -3.62234861e-02 8.37873161e-01 8.15388978e-01
6.83985114e-01 8.53088677e-01 1.53371409e-01 2.25237966e-01
5.38485706e-01 3.65994602e-01 3.91586721e-01 -9.89244819e-01
3.99963588e-01 -1.16777368e-01 -4.15244132e-01 -1.15676856e+00
-2.17134684e-01 -5.94388485e-01 -1.29650044e+00 4.17901665e-01
3.33810240e-01 -7.63310567e-02 -1.51163054e+00 1.38540328e+00
-5.11403382e-01 4.88333881e-01 4.02445793e-01 6.01121008e-01
7.55228102e-01 4.81311291e-01 8.25923979e-02 5.08666396e-01
1.31888485e+00 -1.16237235e+00 -2.99855381e-01 -6.10378265e-01
8.17028761e-01 -5.63769639e-01 5.06647408e-01 3.22013885e-01
-1.01650107e+00 -6.35008931e-01 -1.51310146e+00 -1.31398693e-01
-3.42160344e-01 -3.34938973e-01 3.93278539e-01 1.14130640e+00
-1.62420928e+00 8.50711763e-01 -8.63258481e-01 -3.24346304e-01
6.75756931e-01 2.78654099e-01 -3.08651298e-01 -1.38143841e-02
-1.57811141e+00 1.04993153e+00 9.50022161e-01 6.02796562e-02
-1.77381122e+00 -4.52095538e-01 -1.16676486e+00 5.10440469e-01
1.32055938e-01 -6.62227809e-01 9.76556659e-01 -1.00386262e+00
-1.48314667e+00 1.04819059e+00 7.98369527e-01 -1.29102910e+00
5.63586593e-01 1.34893015e-01 -4.78951752e-01 2.99279958e-01
-4.73968208e-01 1.08985972e+00 1.36497104e+00 -1.03404939e+00
-4.49145049e-01 3.43375593e-01 1.55062795e-01 -2.80488163e-01
-1.31616339e-01 -1.00355797e-01 -3.96595508e-01 -8.12940419e-01
-2.09862426e-01 -9.71010804e-01 -6.16466641e-01 -1.78167358e-01
-5.60984910e-01 3.37903649e-01 1.10512900e+00 -9.61292088e-01
9.35713708e-01 -2.20229340e+00 -2.45887890e-01 3.73140723e-01
6.48856461e-01 8.42642248e-01 -5.89826643e-01 3.22221518e-01
-3.86146337e-01 6.12705886e-01 -1.72158331e-01 -1.25731423e-01
-9.04483721e-03 -6.75872639e-02 -2.91473597e-01 8.65566134e-01
2.91716754e-01 1.41672432e+00 -7.73138523e-01 -3.07539970e-01
4.33221072e-01 3.98223191e-01 -7.00242460e-01 -1.51649535e-01
-2.17627347e-01 2.27874056e-01 -3.29556242e-02 4.11423326e-01
8.09028149e-01 -5.90238571e-01 3.61132860e-01 -1.25756532e-01
2.80874699e-01 3.57011914e-01 -2.81474888e-01 1.43521798e+00
-3.65401089e-01 1.20289433e+00 5.23765646e-02 -1.02687919e+00
7.66886592e-01 4.28150058e-01 3.84625942e-02 -7.47523427e-01
4.59985137e-01 3.23012948e-01 2.72055954e-01 -1.28862575e-01
6.54873312e-01 2.21349761e-01 -1.33944139e-01 3.29988748e-01
3.11814994e-01 -2.84249544e-01 -1.24655031e-01 3.11747253e-01
1.50727630e+00 -3.34066898e-01 4.11153197e-01 -3.50666493e-01
5.05794108e-01 -8.55500549e-02 6.19940870e-02 1.15629649e+00
-1.35523319e-01 6.95479333e-01 5.66496551e-01 -8.07352483e-01
-1.80046344e+00 -6.19323313e-01 2.59209782e-01 5.24671197e-01
1.78693116e-01 -5.95108747e-01 -6.26764119e-01 -8.56514573e-01
-4.68509912e-01 3.08521688e-01 -5.34855545e-01 -5.33162415e-01
-9.53142881e-01 -2.26745650e-01 1.37849140e+00 1.65870979e-01
1.36736917e+00 -1.16753066e+00 -7.26387680e-01 2.16460928e-01
1.01399541e-01 -1.28298974e+00 -3.12205821e-01 1.09548204e-01
-7.73456872e-01 -1.15707064e+00 -8.66584957e-01 -8.92318308e-01
2.44344547e-01 1.77252874e-01 1.39532614e+00 7.02207267e-01
3.15060467e-01 -4.12050076e-02 -3.63403261e-01 -6.35091215e-02
-1.02206194e+00 4.17112052e-01 -4.30930078e-01 -5.81516922e-01
8.79806876e-02 -4.94116217e-01 -7.65742064e-01 2.83281147e-01
-1.17848146e+00 1.60722751e-02 9.09700155e-01 9.13141131e-01
-1.48038149e-01 2.86263317e-01 -1.59783989e-01 -9.40838635e-01
2.67248869e-01 -4.48527783e-01 -5.61507285e-01 5.31745516e-02
-5.79271257e-01 2.15266660e-01 5.68032324e-01 -2.89514363e-01
-3.36183250e-01 -2.81564355e-01 -5.64795196e-01 -7.59133160e-01
-3.62256318e-02 3.92739862e-01 7.33347014e-02 -7.18227863e-01
6.72036827e-01 4.31061059e-01 2.87756592e-01 -1.51034683e-01
1.49359494e-01 2.34276161e-01 7.42141545e-01 1.63047135e-01
1.32961559e+00 5.21317482e-01 -1.44766299e-02 -6.73382819e-01
-3.08176309e-01 2.33676344e-01 6.50948891e-03 2.03570828e-01
1.07987261e+00 -1.36523771e+00 -5.62322736e-01 9.93925750e-01
-1.14371955e+00 -5.61328471e-01 -2.21711561e-01 2.28018165e-01
-4.96786565e-01 8.49362969e-01 -8.70434225e-01 -1.22696750e-01
-4.77285773e-01 -1.23747861e+00 7.83613801e-01 -1.16550572e-01
2.13911403e-02 -1.24427307e+00 -6.59040436e-02 -3.69284838e-03
9.45252001e-01 3.50397915e-01 6.18753314e-01 -7.82736659e-01
-9.15046930e-01 -2.11447611e-01 -4.78032678e-01 6.95038795e-01
-3.33112091e-01 -1.22677527e-01 -8.40276301e-01 -7.69311905e-01
2.56583020e-02 -4.01386172e-01 1.88219988e+00 3.85386705e-01
1.20282519e+00 -5.47353625e-01 -8.73298347e-02 1.32797682e+00
1.44105852e+00 1.75779928e-02 1.67616427e+00 7.00254798e-01
9.41073477e-01 4.83729169e-02 -1.98287606e-01 8.22749268e-03
3.38228583e-01 1.26194820e-01 1.01714480e+00 -3.01203012e-01
-3.96478057e-01 -3.22717011e-01 5.84208727e-01 5.32187581e-01
-8.72713514e-03 -7.59864986e-01 -8.59079242e-01 3.89841706e-01
-1.48868787e+00 -9.96980309e-01 -2.72576734e-02 1.47922361e+00
3.38180035e-01 4.24811006e-01 -3.38371903e-01 7.64560997e-02
6.67415798e-01 9.16292369e-01 -1.99283957e-01 -3.73570979e-01
-2.81240046e-01 5.82226217e-01 1.31303358e+00 3.68038297e-01
-1.49583113e+00 1.37319422e+00 7.93771935e+00 1.05875778e+00
-1.12590182e+00 1.80018514e-01 7.91289926e-01 5.13778329e-01
-4.00275677e-01 9.82330516e-02 -3.88895005e-01 3.71451229e-01
1.28343594e+00 4.16647106e-01 3.71999413e-01 6.06757879e-01
-2.39680573e-01 4.72156554e-01 -8.59940171e-01 9.05164421e-01
-3.57345231e-02 -1.86694539e+00 5.22653423e-02 4.59328592e-01
8.66766810e-01 4.71160978e-01 4.82270807e-01 3.22018862e-01
9.03570175e-01 -1.31133866e+00 6.70124054e-01 -4.50823922e-03
1.15315819e+00 -4.94133919e-01 8.60500097e-01 -1.82398900e-01
-7.97025681e-01 -7.51176178e-02 -6.80773735e-01 2.32783675e-01
6.89086542e-02 3.38295698e-01 -8.08480382e-01 3.46686393e-01
5.92315853e-01 1.06687295e+00 -6.75350904e-01 8.25560153e-01
-4.84546483e-01 9.43402529e-01 -2.16634706e-01 4.29148883e-01
8.72154415e-01 3.53369772e-01 6.67335868e-01 1.20492983e+00
5.02386332e-01 -4.27342236e-01 -3.87893945e-01 8.16585124e-01
-3.74756843e-01 -7.54673481e-01 -9.11132693e-01 -3.32156748e-01
3.73995118e-02 8.86745453e-01 -6.40642881e-01 -5.44227362e-01
-2.50274122e-01 8.80798876e-01 -2.39493757e-01 4.13791984e-01
-1.09622049e+00 -3.01315010e-01 7.88030088e-01 -5.42558283e-02
1.01624084e+00 -4.09788281e-01 -1.02765813e-01 -1.46838069e+00
-8.40324998e-01 -1.42616701e+00 3.45475435e-01 -6.04086816e-01
-8.50510180e-01 6.03280008e-01 -3.90812457e-01 -1.25063086e+00
-3.53980869e-01 -7.98165083e-01 -8.25242996e-01 6.06013834e-01
-1.71081650e+00 -1.20228171e+00 -1.31610602e-01 6.28061652e-01
4.95515436e-01 -5.84379852e-01 5.49884617e-01 -2.10909005e-02
-3.38755488e-01 7.36915767e-01 5.03112115e-02 6.45365715e-01
2.57306784e-01 -8.19496989e-01 1.66499579e+00 1.40175891e+00
1.13605939e-01 2.47681707e-01 7.67703950e-01 -6.95236504e-01
-1.19741058e+00 -1.10119593e+00 4.35568154e-01 -2.22978994e-01
6.29681170e-01 -2.01817662e-01 -7.15316594e-01 9.82820451e-01
7.11874902e-01 -8.30781981e-02 -3.45216170e-02 -5.68397045e-01
-7.91356206e-01 3.73177469e-01 -1.15545607e+00 6.36161208e-01
1.11343169e+00 -5.82181752e-01 -1.82571515e-01 -8.52371380e-03
9.51564014e-01 -4.17438090e-01 -5.68852007e-01 2.72843689e-01
4.65689898e-01 -1.19452572e+00 1.27233648e+00 -5.22238374e-01
8.51120710e-01 4.61537652e-02 1.47371199e-02 -1.34615982e+00
-6.93084300e-01 -1.03860188e+00 -1.93839371e-01 3.69898736e-01
2.72912353e-01 -7.71298885e-01 1.13347948e+00 -3.80606540e-02
-2.57740200e-01 -6.38565049e-02 -7.78314292e-01 -7.99767554e-01
2.34879345e-01 -2.18041092e-01 6.58175051e-01 9.21769321e-01
-7.32195139e-01 -1.47112504e-01 -9.99109805e-01 1.85478508e-01
8.92170787e-01 -7.23992527e-01 7.84387290e-01 -7.67626166e-01
-2.43953422e-01 -6.10131383e-01 -8.73211682e-01 -1.23849797e+00
1.44434631e-01 -6.42090559e-01 -3.52209114e-04 -9.63458002e-01
-2.93307394e-01 -1.42806217e-01 -3.02737385e-01 4.76346701e-01
-3.21778208e-02 9.15941179e-01 3.30535531e-01 3.34336579e-01
-4.42408353e-01 3.21109593e-01 1.34725308e+00 -4.80226427e-01
4.08632189e-01 -4.64645177e-01 -8.96513104e-01 6.95090055e-01
9.19643879e-01 -5.38112462e-01 -4.57547717e-02 -7.22301900e-01
3.24841082e-01 -1.96146443e-01 6.39398813e-01 -1.46116114e+00
1.13878354e-01 2.30900481e-01 4.58643794e-01 -3.36679876e-01
1.09012552e-01 -8.31893206e-01 4.11496490e-01 1.09570885e+00
-2.71436155e-01 -1.02246955e-01 -1.72798391e-02 5.27794063e-01
-5.54863401e-02 -5.11879288e-02 1.07909870e+00 -6.48929179e-01
-1.11956978e+00 3.69255513e-01 -6.76145315e-01 -1.24386609e-01
6.05842113e-01 -5.70144117e-01 -7.33505070e-01 -7.82053292e-01
-4.10163760e-01 3.85728292e-02 7.25052536e-01 2.33335540e-01
9.49538231e-01 -1.11253679e+00 -9.53444004e-01 3.62248927e-01
-2.00678125e-01 -3.42836082e-01 1.24320358e-01 7.02316821e-01
-1.41022623e+00 5.77087402e-01 -6.20224178e-01 -3.31614792e-01
-9.46408927e-01 6.18464112e-01 6.67310953e-01 -7.56345868e-01
-7.23865211e-01 9.14593935e-01 2.55074114e-01 -9.19950679e-02
-8.14359337e-02 -3.04101437e-01 -2.59748921e-02 -5.28944552e-01
4.26895231e-01 2.57930309e-01 -1.36961460e-01 -4.52570379e-01
-2.19825223e-01 2.46070191e-01 -3.19583178e-01 1.54147997e-01
1.31202602e+00 -1.71276197e-01 -8.05442035e-02 -6.20206535e-01
1.35624814e+00 -5.00738323e-01 -1.30933154e+00 -1.30566344e-01
-1.36607602e-01 -3.27009112e-01 2.16300085e-01 -3.34237397e-01
-1.70203590e+00 8.54218483e-01 4.39664662e-01 2.64396578e-01
9.37255442e-01 -2.43185863e-01 1.08539438e+00 6.38033509e-01
2.26994321e-01 -5.91005921e-01 2.77377009e-01 8.44030499e-01
4.99109387e-01 -1.12571001e+00 -1.31024078e-01 -6.55882135e-02
-4.05880243e-01 1.20076323e+00 4.18053806e-01 -8.41703773e-01
6.75214887e-01 2.88263381e-01 -1.07893445e-01 -4.85962838e-01
-6.01692855e-01 -1.66501880e-01 -2.95654535e-02 7.00070739e-01
-1.05653360e-01 -3.58745664e-01 8.31700787e-02 -3.33109647e-01
-3.91759068e-01 -2.25550488e-01 8.30812931e-01 7.50832856e-01
-5.04011631e-01 -6.62419617e-01 -2.72907406e-01 3.55750054e-01
-7.94037759e-01 -6.03838623e-01 -2.91592032e-01 1.02751124e+00
-5.18075861e-02 7.07494974e-01 1.43384159e-01 -8.81376922e-01
-2.46008277e-01 -3.30882132e-01 3.21772099e-01 -3.57449323e-01
-9.68952000e-01 -1.71961278e-01 4.09607410e-01 -7.61128247e-01
-2.38071099e-01 5.90920150e-02 -4.74559665e-01 -1.00652087e+00
-2.34107301e-01 -2.01012716e-01 4.53352004e-01 8.08413029e-01
2.58307643e-02 5.46868980e-01 5.33876598e-01 -1.09442604e+00
-6.45940781e-01 -9.50448811e-01 -4.96576577e-01 1.00223377e-01
7.27582216e-01 2.34826073e-01 -7.02747822e-01 -4.60118130e-02] | [4.334000587463379, 8.04135513305664] |
bc5d3d30-8948-42ab-aa0f-13c9f97495fc | consistent-and-elastic-registration-of | null | null | https://link.springer.com/chapter/10.1007/11889762_8 | https://repositorio.uam.es/bitstream/handle/10486/666430/consistent_arganda-carreras_LNCS_2006_ps.pdf | Consistent and elastic registration of histological sections using vector-spline regularization | Here we present a new image registration algorithm for the alignment of histological sections that combines the ideas of B-spline based elastic registration and consistent image registration, to allow simultaneous registration of images in two directions (direct and inverse). In principle, deformations based on B-splines are not invertible. The consistency term overcomes this limitation and allows registration of two images in a completely symmetric way. This extension of the elastic registration method simplifies the search for the optimum deformation and allows registering with no information about landmarks or deformation regularization. This approach can also be used as the first step to solve the problem of group-wise registration. | ['Carlos Ortiz-de-Solorzano', 'José María Carazo', 'Ignacio Arganda-Carreras', 'Roberto Marabini', 'Jan Kybic', 'Carlos O. S. Sorzano'] | 2006-05-12 | null | null | null | computer-vision-approaches-to-medical-image | ['birl-cima'] | ['medical'] | [ 8.79950672e-02 2.76348572e-02 -1.53181306e-03 -4.20913219e-01
-6.27552927e-01 -4.35017884e-01 3.34822029e-01 2.67757148e-01
-8.00951600e-01 5.97115576e-01 -3.61334607e-02 -4.10542898e-02
-3.97371233e-01 -6.94349527e-01 -3.11374158e-01 -1.05298042e+00
-1.94098845e-01 5.24214268e-01 6.19329393e-01 -3.65653157e-01
4.09910321e-01 7.83312678e-01 -8.23808193e-01 -1.00771941e-01
6.14996552e-01 4.11254615e-01 2.15216041e-01 4.69914526e-01
-9.13679600e-02 -1.04597531e-01 -2.05748174e-02 -1.17013697e-02
3.01203877e-01 -3.19082975e-01 -1.21364498e+00 6.63598925e-02
4.86980289e-01 -3.00328106e-01 7.87851959e-03 9.53361511e-01
5.63834429e-01 2.75908291e-01 6.36078835e-01 -6.53439164e-01
-3.79991859e-01 1.06502101e-01 -6.27786756e-01 2.20053971e-01
4.00009751e-01 -3.79664898e-01 3.40636104e-01 -6.61527753e-01
6.73126161e-01 9.30009246e-01 1.14401543e+00 5.96499920e-01
-1.77143908e+00 -1.75145671e-01 -4.04968560e-01 -2.28429273e-01
-1.43650270e+00 -3.99147034e-01 8.02924812e-01 -6.57777905e-01
5.54432094e-01 5.99902153e-01 6.87935054e-01 2.76732296e-01
4.06086057e-01 -3.16552371e-02 1.22594953e+00 -6.62374973e-01
-4.50243093e-02 -1.59625441e-01 3.50092560e-01 6.12589657e-01
2.30812639e-01 1.22566685e-01 1.78401899e-02 -3.87073129e-01
1.24131322e+00 -1.65215835e-01 -3.58344227e-01 -5.24051130e-01
-1.31897080e+00 6.31334066e-01 3.60992491e-01 1.05480218e+00
-3.15979838e-01 1.37641385e-01 2.80873686e-01 2.08816588e-01
5.09556115e-01 2.49930590e-01 -8.12205970e-02 4.76927727e-01
-9.46558297e-01 8.49549845e-02 4.68386769e-01 3.05860311e-01
6.04299486e-01 -1.06011078e-01 1.25620201e-01 6.39204144e-01
7.23793805e-01 2.41801247e-01 7.16073811e-01 -1.09954417e+00
-4.98641953e-02 2.45535612e-01 -1.55772984e-01 -1.13195324e+00
-5.26156545e-01 2.68096209e-01 -8.95085275e-01 5.52242458e-01
8.94662559e-01 2.04218015e-01 -6.76417172e-01 1.49817383e+00
5.34690559e-01 3.02576512e-01 -2.11326227e-01 7.10691273e-01
7.08806753e-01 1.49373263e-01 1.54276546e-02 -4.68199074e-01
1.27387333e+00 -3.34106743e-01 -1.01920998e+00 2.89479762e-01
4.65921283e-01 -1.02775586e+00 5.46652615e-01 -1.26604885e-01
-1.34096301e+00 -2.64902800e-01 -7.60281801e-01 -9.47597697e-02
-1.65530071e-01 -4.00332510e-01 4.16979402e-01 4.59521949e-01
-1.43102789e+00 9.32524264e-01 -1.24844980e+00 -3.41333210e-01
-5.95985632e-03 9.15707707e-01 -8.94334316e-01 4.89650398e-01
-8.18612278e-01 1.30260611e+00 2.38454401e-01 4.11160409e-01
1.16454840e-01 -6.47018373e-01 -7.01873302e-01 -4.70101267e-01
-3.42973560e-01 -4.95826066e-01 4.53038067e-01 -5.94541132e-01
-1.50182009e+00 1.52006817e+00 -3.90776724e-01 -1.03707306e-01
7.16163993e-01 3.28072250e-01 -3.88256349e-02 7.67728388e-02
-1.54153615e-01 4.20404166e-01 4.14613932e-01 -1.24016345e+00
2.31554449e-01 -6.11157179e-01 -5.10231555e-01 -2.34418884e-02
4.60151993e-02 3.13316405e-01 -1.61087722e-01 -6.70158029e-01
7.40325212e-01 -1.11107457e+00 -5.73857844e-01 3.08102936e-01
-8.62181652e-03 8.21238607e-02 4.84962463e-01 -9.57591891e-01
7.84918487e-01 -2.20179534e+00 2.90568799e-01 6.72766209e-01
1.44972637e-01 1.00024402e-01 -1.30451620e-01 1.22480415e-01
-5.29522419e-01 1.32545575e-01 -4.41238910e-01 -2.35627368e-01
-5.98315537e-01 2.85861611e-01 1.64969400e-01 1.03676760e+00
-3.55922163e-01 8.60693514e-01 -7.69398630e-01 -7.89491773e-01
1.06002130e-01 7.76247978e-01 -2.00150609e-01 1.84990495e-01
7.50187993e-01 1.15683150e+00 -4.03326213e-01 6.34927303e-02
7.92508364e-01 2.48854607e-01 2.59135246e-01 -3.91778678e-01
-4.03243452e-01 -1.37565974e-02 -1.18740392e+00 1.70433414e+00
-2.32168332e-01 5.28227210e-01 3.22624922e-01 -1.25254190e+00
1.20613766e+00 8.57032955e-01 1.09559441e+00 -1.61693737e-01
2.03447923e-01 5.12019336e-01 -8.70736912e-02 -4.69074547e-01
-1.03690818e-01 -4.93745089e-01 3.85221660e-01 3.97489160e-01
-1.10462308e-01 -3.83900315e-01 4.60501388e-02 -5.10856688e-01
5.73901892e-01 9.84166786e-02 6.08449161e-01 -8.96850109e-01
9.36135888e-01 -2.07512021e-01 4.68892366e-01 1.59665093e-01
-1.21442072e-01 7.29353964e-01 6.44851327e-02 -8.25565934e-01
-8.43168795e-01 -1.08005226e+00 -7.56158113e-01 1.57482713e-01
3.06913584e-01 1.91997439e-01 -7.60617256e-01 -4.04764205e-01
9.15269554e-02 -7.42064044e-02 -7.54814923e-01 1.02396704e-01
-9.68098521e-01 -1.05081785e+00 1.79598629e-01 1.43741906e-01
1.41710117e-01 -8.80884469e-01 -3.06575865e-01 2.10348487e-01
-3.35254103e-01 -6.59526467e-01 -5.65235317e-01 2.36085448e-02
-1.68081880e+00 -1.14954484e+00 -8.95515084e-01 -1.11273789e+00
1.43203187e+00 1.50529787e-01 7.93537557e-01 6.54026806e-01
-3.30270469e-01 2.70653367e-01 2.22460583e-01 1.50379390e-01
-7.57765949e-01 -3.74807537e-01 1.19027123e-01 -4.00059707e-02
-1.84766040e-03 -9.36461568e-01 -5.21473110e-01 6.64691448e-01
-9.06752944e-01 -4.00185287e-01 1.46637470e-01 7.99827635e-01
1.02861118e+00 -1.76753998e-01 1.99434906e-01 -8.05742204e-01
6.56263471e-01 1.23194596e-02 -6.39730752e-01 3.23717296e-01
-5.61239004e-01 4.18570526e-02 6.81829751e-02 -4.87898052e-01
-7.21588671e-01 3.80868077e-01 -4.47429240e-01 -4.15989645e-02
-2.72624165e-01 3.58938634e-01 2.55736351e-01 -1.21084714e+00
7.59342730e-01 1.05111293e-01 5.97442448e-01 -5.81192195e-01
1.11807980e-01 2.83121496e-01 6.89299583e-01 -3.86199117e-01
9.09865141e-01 7.84873128e-01 5.85575700e-01 -7.73863673e-01
-1.43769518e-01 -7.03020692e-01 -1.52974498e+00 -2.58116037e-01
1.00211453e+00 -3.50134999e-01 -7.60404289e-01 3.90482455e-01
-1.30553341e+00 -2.36739293e-01 -6.05528355e-01 7.93904006e-01
-8.01021039e-01 9.41170812e-01 -4.59158599e-01 -1.89393774e-01
-3.64566863e-01 -1.39489675e+00 7.17666149e-01 -1.47125378e-01
-3.58689815e-01 -1.50459611e+00 6.42129779e-01 2.84530744e-02
4.71702009e-01 6.41435266e-01 6.55470133e-01 -5.43951452e-01
-2.16955274e-01 -6.42646432e-01 2.06357464e-01 1.72060415e-01
4.47607696e-01 8.80154148e-02 -4.53133762e-01 -3.42394024e-01
4.93840963e-01 3.53588343e-01 3.53621215e-01 6.97783589e-01
7.11026430e-01 -2.06566930e-01 -3.57988000e-01 7.24494219e-01
1.63931656e+00 2.00329691e-01 9.26181257e-01 3.53255600e-01
5.00675321e-01 7.81947076e-01 2.22191304e-01 -2.05673561e-01
1.00153588e-01 1.27079165e+00 1.10636733e-01 -5.10626972e-01
-1.98088810e-01 4.49699938e-01 -1.53335482e-01 1.07954144e+00
-8.79202962e-01 6.78971887e-01 -8.59856784e-01 4.73651677e-01
-1.68234062e+00 -1.10185719e+00 -8.26206505e-01 2.48795152e+00
1.10420597e+00 -4.46127087e-01 1.25981225e-02 1.68840289e-01
9.28642094e-01 -3.35293263e-01 2.96655029e-01 -3.66647243e-01
2.93163687e-01 1.25031412e-01 6.37543440e-01 1.02025068e+00
-1.01222908e+00 2.75333554e-01 8.37476730e+00 4.77945864e-01
-1.09693670e+00 2.29810596e-01 2.24693596e-01 7.42400646e-01
-3.03803593e-01 2.23291293e-01 -5.04640877e-01 4.00858700e-01
4.86391544e-01 -2.12888062e-01 3.49540293e-01 2.38941729e-01
3.54463935e-01 -1.98941961e-01 -8.90611470e-01 5.12360454e-01
-1.62045553e-01 -1.31717527e+00 -3.53853643e-01 3.67234647e-01
5.60966551e-01 -2.30402127e-01 -2.66149908e-01 -7.50873446e-01
-1.40798524e-01 -6.97514355e-01 2.50835687e-01 8.22972715e-01
7.50608921e-01 -2.05685198e-01 7.72729337e-01 9.68420953e-02
-1.13326764e+00 6.61657810e-01 -4.28493261e-01 2.82523334e-01
5.20632625e-01 4.96745974e-01 -6.64322853e-01 5.05562961e-01
4.22675252e-01 3.78121048e-01 -1.89098030e-01 1.21146297e+00
-4.11196053e-02 -9.87919196e-02 -4.62447256e-01 5.95285416e-01
-2.95878619e-01 -8.27326357e-01 7.63035774e-01 9.32249248e-01
2.84726769e-01 2.45468512e-01 1.14358798e-01 7.41572320e-01
5.77194452e-01 4.19909149e-01 -5.05279541e-01 7.27882922e-01
1.61050279e-02 1.27501261e+00 -1.09935582e+00 1.63149387e-01
-3.84295970e-01 7.07232416e-01 2.63766032e-02 6.52175248e-02
-2.89150923e-01 -3.08874901e-02 2.31479764e-01 5.08577168e-01
-1.71742842e-01 -6.10508025e-01 -3.05739701e-01 -9.13270712e-01
-1.41343072e-01 -2.43805647e-01 2.70960897e-01 -3.29953492e-01
-1.11766839e+00 6.15506172e-01 3.64019990e-01 -1.12048018e+00
-1.83930159e-01 -3.20846558e-01 -8.90158772e-01 1.20816362e+00
-1.30585611e+00 -1.11281085e+00 -2.35526860e-01 7.73148060e-01
-1.92377821e-01 2.77385741e-01 1.16069090e+00 2.91072994e-01
1.68882273e-02 3.05770367e-01 1.84088141e-01 1.59349903e-01
8.74526978e-01 -1.31898272e+00 8.86393338e-02 6.09026015e-01
-1.61280960e-01 9.17537749e-01 7.00149655e-01 -7.06698358e-01
-8.10107112e-01 -5.45774221e-01 1.20746446e+00 -3.81801993e-01
5.67482769e-01 2.70080864e-01 -1.21027350e+00 6.99393928e-01
1.15228370e-02 5.08918583e-01 8.65328431e-01 -1.65587395e-01
2.64150292e-01 -9.10091549e-02 -1.46078610e+00 3.54844928e-01
4.55446810e-01 -2.40138412e-01 -7.40992486e-01 5.21621227e-01
7.98454806e-02 -5.57904661e-01 -1.59555149e+00 3.35141063e-01
6.93221092e-01 -7.24140048e-01 1.45374858e+00 -2.51587808e-01
-3.31205189e-01 -3.91808420e-01 3.95975977e-01 -9.63138700e-01
-3.57178152e-01 -8.22529972e-01 7.12936997e-01 1.04203343e+00
4.23676036e-02 -8.55324805e-01 5.59943557e-01 7.41868377e-01
-4.99699004e-02 -3.80184233e-01 -1.39562821e+00 -9.86620128e-01
1.56068847e-01 2.79883981e-01 2.39793226e-01 1.07457650e+00
3.33332032e-01 -5.06551087e-01 -1.17777333e-01 -9.62247625e-02
8.12507212e-01 -1.02951877e-01 3.53389353e-01 -1.59675753e+00
7.01673105e-02 -3.99093509e-01 -8.05202186e-01 -5.52179575e-01
3.44277024e-01 -1.09181511e+00 2.42284521e-01 -1.35855198e+00
-4.97053638e-02 -8.55160236e-01 -2.35682782e-02 6.03510976e-01
3.03708985e-02 5.93257427e-01 -2.11684555e-01 7.04542160e-01
2.24442273e-01 -9.40966904e-02 1.50930500e+00 2.46816918e-01
-4.05802876e-01 2.04948023e-01 -2.08218858e-01 7.95374990e-01
6.55666530e-01 -6.73133671e-01 1.30601540e-01 -1.71469018e-01
-1.55305825e-02 1.55854627e-01 3.40152293e-01 -5.84168434e-01
3.65463912e-01 -2.07866117e-01 3.01794056e-02 -7.36422790e-03
-1.27178177e-01 -1.21755588e+00 8.32617640e-01 6.94009185e-01
-2.75025666e-01 1.44111395e-01 -1.46950604e-02 2.13432938e-01
-3.39239299e-01 -7.81832874e-01 1.16559291e+00 -2.28540733e-01
-2.25813478e-01 3.04407537e-01 -3.29801232e-01 -4.92123634e-01
1.04941440e+00 -4.86212879e-01 1.83027938e-01 6.70384988e-02
-1.30927408e+00 -2.69588619e-01 6.62508190e-01 -2.25660518e-01
4.49414998e-01 -1.60467386e+00 -7.50795484e-01 2.42464378e-01
-4.36184406e-01 5.20753749e-02 1.92908973e-01 1.69303715e+00
-1.07057619e+00 1.70974687e-01 -6.19506299e-01 -6.53804064e-01
-1.75888801e+00 5.28182268e-01 7.05509365e-01 -4.53557938e-01
-5.67682326e-01 5.14088273e-01 -7.84465596e-02 -3.07332397e-01
-3.94042909e-01 1.22221066e-02 -5.41546702e-01 -1.37489527e-01
3.37664098e-01 4.21018779e-01 2.99339652e-01 -1.23794913e+00
-5.56933880e-01 1.31335115e+00 3.50757092e-01 -1.66096017e-02
1.37610209e+00 -3.36829811e-01 -9.37079608e-01 1.88264787e-01
1.21257949e+00 3.85052562e-01 -8.24407220e-01 2.05251798e-02
-8.82819295e-02 -4.83307183e-01 2.57648677e-01 -1.56550035e-01
-1.23796499e+00 5.74940264e-01 6.36425138e-01 3.02324742e-01
9.73047614e-01 -1.82875946e-01 3.65798473e-01 -2.22058427e-02
1.96150750e-01 -7.10038483e-01 -4.26544517e-01 1.07659698e-01
1.10946941e+00 -9.93873537e-01 4.11299467e-01 -8.77316833e-01
9.64419097e-02 1.54990184e+00 1.39842406e-02 -4.81131643e-01
1.00498617e+00 4.73716974e-01 2.16927618e-01 -1.37109295e-01
1.41106948e-01 3.34814042e-02 8.37087452e-01 8.50597918e-01
8.69611144e-01 -2.51284271e-01 -1.29211283e+00 -3.85101102e-02
2.84615099e-01 -1.73361208e-02 3.62949520e-01 8.49971592e-01
-2.43056864e-01 -1.64299262e+00 -7.43778825e-01 -1.21124059e-01
-6.27986252e-01 2.10704982e-01 -6.13813987e-03 8.98815036e-01
-2.95800548e-02 3.91194969e-01 1.39215693e-01 2.29173318e-01
3.78051966e-01 -8.57658759e-02 8.08253229e-01 -2.74289995e-01
-6.53545737e-01 5.07501662e-01 -3.30565393e-01 -5.64262927e-01
-1.07751107e+00 -8.86656642e-01 -1.35608947e+00 -2.25033551e-01
-6.89225793e-01 1.88450187e-01 9.20970738e-01 9.25068617e-01
-1.04604483e-01 1.12046525e-01 6.62646234e-01 -1.10134435e+00
-3.39179397e-01 -4.78682160e-01 -7.44800746e-01 6.06037199e-01
2.51309842e-01 -4.86248463e-01 -3.41747314e-01 5.56135535e-01] | [13.97729778289795, -2.595978021621704] |
afe6140e-896b-412c-af4f-2aaf3f7dcea9 | predictive-process-model-monitoring-using | 2011.02819 | null | https://arxiv.org/abs/2011.02819v3 | https://arxiv.org/pdf/2011.02819v3.pdf | Predictive Process Model Monitoring using Recurrent Neural Networks | The field of predictive process monitoring focuses on case-level models to predict a single specific outcome such as a particular objective, (remaining) time, or next activity/remaining sequence. Recently, a longer-horizon, model-wide approach has been proposed in the form of process model forecasting, which predicts the future state of a whole process model through the forecasting of all activity-to-activity relations at once using time series forecasting. This paper introduces the concept of \emph{predictive process model monitoring} which sits in the middle of both predictive process monitoring and process model forecasting. Concretely, by modelling a process model as a set of constraints being present between activities over time, we can capture more detailed information between activities compared to process model forecasting, while being compatible with typical predictive process monitoring objectives which are often expressed in the same language as these constraints. To achieve this, Processes-As-Movies (PAM) is introduced, i.e., a novel technique capable of jointly mining and predicting declarative process constraints between activities in various windows of a process' execution. PAM predicts what declarative rules hold for a trace (objective-based), which also supports the prediction of all constraints together as a process model (model-based). Various recurrent neural network topologies inspired by video analysis tailored to temporal high-dimensional input are used to model the process model evolution with windows as time steps, including encoder-decoder long short-term memory networks, and convolutional long short-term memory networks. Results obtained over real-life event logs show that these topologies are effective in terms of predictive accuracy and precision. | ['Jochen De Weerdt', 'Johannes De Smedt'] | 2020-11-05 | null | null | null | null | ['predictive-process-monitoring'] | ['time-series'] | [ 6.97811902e-01 5.27110603e-03 -1.33674592e-01 -3.97535890e-01
-3.57963890e-02 -1.55875355e-01 1.21383786e+00 5.79387605e-01
8.26018378e-02 1.82277188e-01 3.87598336e-01 -2.19086006e-01
-6.69533134e-01 -1.03010786e+00 -2.50427336e-01 -3.56295466e-01
-4.29897845e-01 5.79925358e-01 1.36379704e-01 2.78044432e-01
3.39113802e-01 6.40389502e-01 -1.72316802e+00 6.76754415e-01
1.80251062e-01 1.25223374e+00 2.70774215e-01 8.42622101e-01
-5.35010338e-01 1.46898139e+00 -4.75656271e-01 -9.63383541e-02
-7.84945190e-02 -3.60510021e-01 -5.85431993e-01 4.14726406e-01
-4.90330577e-01 1.02830708e-01 -3.91758382e-01 3.55428427e-01
-2.76163429e-01 1.81752115e-01 4.87798512e-01 -1.20239902e+00
-1.83249995e-01 6.13905609e-01 -5.46193384e-02 2.89185613e-01
6.04358256e-01 3.25057119e-01 7.32799351e-01 -5.42914271e-01
4.96749729e-01 1.25536966e+00 5.12077749e-01 4.01807785e-01
-1.23939860e+00 -9.74476710e-02 4.66344237e-01 2.26319000e-01
-1.08315182e+00 -9.50712264e-02 6.80195451e-01 -7.00874686e-01
1.46310389e+00 4.49806511e-01 7.90819824e-01 1.20133388e+00
1.02985382e+00 7.19637990e-01 7.36075163e-01 -2.95810699e-01
4.34742898e-01 -3.64568919e-01 8.34145546e-02 4.08787191e-01
-2.59530902e-01 1.38788447e-01 -7.90241599e-01 -1.56774044e-01
8.37650836e-01 7.69702137e-01 1.38919279e-02 9.42949429e-02
-1.36858273e+00 2.43161052e-01 -4.68586236e-01 5.88833272e-01
-8.17143619e-01 3.62574697e-01 5.49252570e-01 5.49461842e-01
3.56307089e-01 3.08006346e-01 -4.78042901e-01 -5.80592036e-01
-1.08048224e+00 1.75111398e-01 1.41074228e+00 9.28727448e-01
5.08420110e-01 3.07069629e-01 -6.81971967e-01 3.25877458e-01
3.50196272e-01 8.84887949e-02 6.76516354e-01 -8.27424586e-01
3.46318036e-01 8.71534824e-01 2.19761088e-01 -6.59111738e-01
-3.09860855e-01 -2.67941862e-01 -9.15466309e-01 1.50178690e-02
8.65249187e-02 8.78686458e-02 -7.08255649e-01 1.38956392e+00
-2.73313314e-01 5.59502244e-01 -3.56996506e-02 6.84970245e-02
-1.38394862e-01 1.23525190e+00 2.64735430e-01 -9.21275616e-01
1.21917105e+00 -8.39555085e-01 -1.07480216e+00 -1.38884142e-01
3.28661293e-01 -2.56968290e-01 4.96668547e-01 7.07459092e-01
-1.32565820e+00 -8.76531959e-01 -6.64244115e-01 6.72037423e-01
-3.84117693e-01 -1.83119804e-01 5.64778745e-01 2.97715217e-01
-1.06598687e+00 1.11598217e+00 -1.37652707e+00 -5.25868714e-01
3.92431300e-03 4.32045221e-01 -2.15015374e-02 2.06708476e-01
-8.29335093e-01 8.19689095e-01 5.89574277e-01 3.04818451e-01
-1.34703004e+00 -5.12641191e-01 -6.47023559e-01 5.96124887e-01
5.83419979e-01 -4.83222902e-01 1.41661787e+00 -7.17602909e-01
-1.61034346e+00 2.37447038e-01 -5.80956399e-01 -7.63080537e-01
5.57464123e-01 -2.72136629e-01 -1.08343959e+00 -2.03299478e-01
-4.20647144e-01 -2.52367333e-02 9.64653432e-01 -9.03258681e-01
-1.03098571e+00 -1.78627729e-01 -3.59160632e-01 -4.67645526e-02
-2.43140221e-01 3.33881080e-01 -5.62158227e-01 -1.91153929e-01
1.17652595e-01 -6.46727502e-01 -5.75920224e-01 -4.57109809e-01
-3.56686473e-01 -3.22742522e-01 7.90783346e-01 -5.79193950e-01
1.74316752e+00 -1.90182245e+00 1.42492548e-01 4.44717318e-01
4.86975498e-02 2.46209037e-02 1.29726157e-01 1.18377829e+00
-1.57343686e-01 6.54693693e-02 -1.56540260e-01 -6.58954859e-01
1.71805367e-01 5.63007832e-01 -5.89843512e-01 9.40901861e-02
3.56476784e-01 7.28394091e-01 -5.98677754e-01 -2.46693090e-01
5.84858775e-01 3.09077144e-01 2.30511025e-01 5.83222032e-01
-5.94303012e-01 5.41669309e-01 -5.28393745e-01 6.79963827e-01
-3.43723521e-02 -1.47074416e-01 4.78998214e-01 5.43629527e-01
-4.58777666e-01 1.21694570e-02 -1.27638865e+00 1.28449762e+00
-8.12026858e-01 3.14135522e-01 -2.74524152e-01 -7.29819953e-01
1.32142317e+00 8.13732088e-01 9.52363849e-01 -5.74262738e-01
-2.90488720e-01 -4.42261323e-02 -3.26490164e-01 -5.39588451e-01
4.89540905e-01 -1.29599690e-01 8.23105425e-02 6.10506356e-01
-1.56215519e-01 2.06215858e-01 5.60227811e-01 -4.91253972e-01
1.49102950e+00 4.59782720e-01 4.30293292e-01 1.09315313e-01
9.46488738e-01 -3.91480953e-01 6.34835780e-01 6.42957866e-01
1.47423476e-01 1.96023911e-01 8.14121604e-01 -8.27666759e-01
-9.13640916e-01 -9.68966663e-01 3.58026385e-01 1.11712027e+00
-1.65597379e-01 -6.85212910e-01 -1.98961675e-01 -2.67036647e-01
-2.27230102e-01 9.55476761e-01 -7.23627806e-01 1.37991151e-02
-7.93124855e-01 -4.92391348e-01 2.44509354e-01 6.11424983e-01
8.84315670e-02 -1.45959437e+00 -9.97826815e-01 9.20689762e-01
4.86061901e-01 -9.38684583e-01 -4.06520367e-02 5.91464221e-01
-1.21664965e+00 -9.67346966e-01 -2.29805964e-03 -2.73799866e-01
9.65536982e-02 -4.14149135e-01 1.12295449e+00 -2.94865102e-01
8.34154859e-02 5.07184923e-01 -1.46089301e-01 -5.97350061e-01
-7.47895539e-01 -2.82470465e-01 -1.01293199e-01 4.42254663e-01
4.52406138e-01 -8.59084129e-01 -8.03482533e-02 1.89538613e-01
-1.13353026e+00 9.94733274e-02 5.98680556e-01 3.49005759e-01
7.60398328e-01 4.16303158e-01 3.83915365e-01 -6.35606825e-01
9.39260900e-01 -5.56918383e-01 -4.35589820e-01 5.99659920e-01
-1.00004756e+00 6.68166503e-02 8.94900858e-01 -6.07737839e-01
-1.37424898e+00 3.48998010e-02 2.01262295e-01 -7.13993907e-01
-5.23236871e-01 8.26694548e-01 -1.67565122e-01 1.01697683e+00
7.25260302e-02 8.72588336e-01 -1.49083599e-01 -3.99175644e-01
-1.52051792e-01 1.82106689e-01 3.73740017e-01 -5.10857642e-01
4.99728948e-01 6.69454277e-01 3.09210867e-01 -4.30119008e-01
-4.34239447e-01 -6.21517777e-01 -6.60902917e-01 -6.64929330e-01
7.89716363e-01 -5.92427790e-01 -1.06976414e+00 3.09314132e-01
-1.09563267e+00 -3.61995369e-01 -7.24679291e-01 4.95634645e-01
-9.56417322e-01 -1.40071893e-02 -1.09945440e+00 -1.34179854e+00
-3.46397012e-01 -7.47430861e-01 8.65444601e-01 -5.88747673e-03
-6.79650843e-01 -1.47525275e+00 1.05950303e-01 -1.32287264e-01
4.52840477e-01 3.79212111e-01 8.64010096e-01 -1.07769966e+00
-6.12328351e-01 -6.75806999e-01 3.72774541e-01 3.08580905e-01
1.55913845e-01 1.37793019e-01 -6.48608267e-01 -6.38256818e-02
6.18685424e-01 5.65904200e-01 3.57804954e-01 2.95637131e-01
1.37180912e+00 -3.13082010e-01 -5.48748016e-01 3.34441274e-01
1.47698486e+00 8.48057270e-01 7.37638175e-01 2.25725785e-01
5.44937551e-01 7.63274193e-01 5.60025871e-01 9.19196248e-01
3.07675172e-03 3.56315523e-01 4.41090375e-01 4.33574736e-01
4.98907208e-01 -3.90762389e-01 8.48087072e-01 8.15970659e-01
-5.86675227e-01 -3.09265286e-01 -1.12644982e+00 3.45864773e-01
-2.12199616e+00 -1.53015721e+00 -4.32369649e-01 2.19905424e+00
1.08094320e-01 5.01759291e-01 -1.38848424e-01 2.99245983e-01
7.37272680e-01 4.59329605e-01 -4.64731693e-01 -9.32559967e-01
2.06772327e-01 -1.49368560e-02 1.88392594e-01 3.66495699e-01
-1.00584269e+00 4.23168242e-01 6.08429193e+00 5.97969651e-01
-8.63359153e-01 8.82103741e-02 5.07169366e-01 -2.67500848e-01
-1.06330469e-01 4.66153622e-02 -9.72309649e-01 5.98017037e-01
1.83935010e+00 -3.48655850e-01 2.79882431e-01 7.53263116e-01
9.72127855e-01 4.91227433e-02 -1.83886623e+00 7.24830747e-01
-1.16769850e-01 -1.46420109e+00 1.57735601e-01 3.88626963e-01
4.78754461e-01 -2.34893441e-01 -3.63118321e-01 3.77374828e-01
2.65358388e-01 -1.26751912e+00 8.81786168e-01 1.29960549e+00
3.13362390e-01 -7.16967165e-01 7.28202224e-01 7.00402856e-01
-1.55732620e+00 -7.21356511e-01 8.03258419e-02 -4.90441084e-01
6.25253320e-01 6.86416030e-01 -7.44915247e-01 8.39707851e-01
3.62611711e-01 1.03529394e+00 -6.77034408e-02 5.61785102e-01
-5.13868518e-02 6.30710781e-01 -1.03253551e-01 1.18634954e-01
1.93915695e-01 -3.52402091e-01 5.36369264e-01 1.46423769e+00
7.18618691e-01 -4.69566524e-01 3.20136219e-01 1.01185036e+00
6.24093235e-01 -2.53450722e-01 -6.26104772e-01 -1.91501722e-01
1.63142681e-01 9.63706255e-01 -6.82380557e-01 -4.48041886e-01
-4.45140392e-01 7.69428790e-01 -2.52779990e-01 3.97251219e-01
-7.60334313e-01 4.89592075e-01 4.33388710e-01 2.47346371e-01
2.01245070e-01 -4.54794854e-01 -2.40882277e-01 -8.13943028e-01
8.90117586e-02 -4.58246261e-01 5.12492597e-01 -7.44891644e-01
-1.28301370e+00 6.15707397e-01 -9.05102119e-03 -1.52955282e+00
-7.85811305e-01 -3.09675574e-01 -1.02104533e+00 9.92233157e-01
-1.24325633e+00 -1.18394589e+00 -1.44954875e-01 6.44223332e-01
1.00738490e+00 -1.34845003e-01 9.55706596e-01 -2.86499381e-01
-6.52819157e-01 -5.69240987e-01 -2.42050365e-02 -4.54776824e-01
6.72005415e-02 -1.15788376e+00 2.74031699e-01 8.38318825e-01
1.05147958e-01 4.50001299e-01 8.76312912e-01 -9.87137675e-01
-1.67415130e+00 -1.25038016e+00 1.30991673e+00 -5.23963690e-01
1.08499169e+00 -1.21017762e-01 -1.16101241e+00 1.04617858e+00
4.14611995e-02 -4.03485775e-01 6.63902044e-01 -2.67067552e-02
1.32652193e-01 -2.01467574e-01 -5.96123993e-01 4.52979952e-01
9.21289861e-01 -6.26969635e-01 -7.78801978e-01 1.44971907e-01
5.01942158e-01 -8.57271552e-02 -1.30195630e+00 2.31874555e-01
2.22209334e-01 -9.95829999e-01 6.01359308e-01 -5.79805195e-01
5.55013001e-01 -2.52419412e-01 -2.18257099e-01 -8.55728447e-01
-4.02061075e-01 -9.36586857e-01 -1.08764875e+00 1.17679334e+00
1.38349518e-01 -2.44240522e-01 7.89446831e-01 7.58135378e-01
-2.78434992e-01 -9.86672521e-01 -9.27937090e-01 -8.89973998e-01
-5.33853531e-01 -9.90748644e-01 8.30054641e-01 7.49106765e-01
1.12038255e-01 1.79947019e-01 -4.72894430e-01 2.32652843e-01
5.61290644e-02 2.40543082e-01 3.96173477e-01 -1.50928974e+00
-4.70557451e-01 -4.86080289e-01 -3.37513745e-01 -7.09961951e-01
1.50883496e-01 -5.09058297e-01 -1.62994385e-01 -1.85864604e+00
-1.72995403e-02 6.88248947e-02 -4.97908354e-01 2.88420022e-01
3.74436319e-01 -7.88974643e-01 2.84948409e-01 6.71521842e-01
-6.03898108e-01 5.58642387e-01 9.61037099e-01 -2.82717049e-02
-5.29074371e-01 5.84128022e-01 2.28606667e-02 6.95452929e-01
7.58725762e-01 -2.60677963e-01 -5.48430681e-01 1.45228561e-02
2.97287852e-01 8.44290614e-01 2.32906580e-01 -1.53917253e+00
6.08549893e-01 -5.87125480e-01 3.76600236e-01 -6.39182866e-01
6.43678546e-01 -1.21428871e+00 9.74408925e-01 7.16667175e-01
-6.96884573e-01 4.14054006e-01 -1.13417611e-01 1.10090172e+00
-6.15995049e-01 -1.35236040e-01 7.07290769e-02 -2.11875394e-01
-9.34243500e-01 4.01316255e-01 -9.76415396e-01 -8.40018690e-01
1.29517472e+00 -6.15770698e-01 8.30713809e-02 -4.79969054e-01
-1.21576893e+00 1.58933237e-01 -6.23423532e-02 3.77174437e-01
5.28830469e-01 -7.38911152e-01 -5.55623949e-01 1.43695921e-01
5.87422140e-02 -1.73827186e-02 3.66604358e-01 8.31083119e-01
-1.23974733e-01 5.69430470e-01 -1.30819768e-01 -6.05487704e-01
-9.60999846e-01 8.27666461e-01 3.93384904e-01 -1.16614306e+00
-7.67822385e-01 1.17747433e-01 -3.08919430e-01 -4.95594786e-03
1.30121797e-01 -5.35703182e-01 -4.88604635e-01 2.18049511e-01
7.32037067e-01 3.81282926e-01 -6.90864623e-02 -2.33342513e-01
1.06034346e-01 1.02444664e-01 3.06266636e-01 -1.33257672e-01
1.60486412e+00 -1.69882774e-01 -3.84433389e-01 1.30879772e+00
4.56624746e-01 -3.28829080e-01 -1.70403993e+00 -1.71610177e-01
8.60192358e-01 -4.51571196e-02 -3.90351951e-01 -6.90803587e-01
-7.00444281e-01 8.88205588e-01 3.86616409e-01 7.24142313e-01
1.11456907e+00 -3.20610441e-02 3.97417158e-01 3.75220031e-01
3.38645786e-01 -1.25983000e+00 -6.74541816e-02 7.51800835e-01
8.46650898e-01 -4.61408824e-01 -2.94786245e-01 -3.28004152e-01
-6.01947665e-01 1.45702267e+00 3.96353155e-01 1.91773444e-01
6.60121322e-01 4.67983186e-01 -5.52357078e-01 -2.95025796e-01
-1.52943587e+00 1.11982785e-01 -1.28619671e-01 4.95952427e-01
3.16136271e-01 2.76917964e-01 1.30516514e-01 6.49889350e-01
1.76988229e-01 5.64066887e-01 6.17152631e-01 1.28201783e+00
-3.06345820e-01 -1.11017191e+00 -4.46344018e-01 7.59933472e-01
-4.76140231e-01 2.20111355e-01 1.11964352e-01 5.93575418e-01
9.80910212e-02 1.05372083e+00 5.59883893e-01 -4.40787554e-01
6.01695061e-01 6.87649012e-01 6.57485202e-02 -7.41931617e-01
-1.03250599e+00 1.31588027e-01 2.85179436e-01 -8.05135429e-01
-3.54848623e-01 -8.72977972e-01 -1.17088997e+00 -3.41110229e-01
4.23921674e-01 -2.04058409e-01 5.67822933e-01 1.19308674e+00
1.61227852e-01 1.01768517e+00 5.57109833e-01 -6.84095383e-01
-3.55865657e-01 -1.10563910e+00 -7.92308688e-01 4.33088511e-01
-1.06577218e-01 -1.27066553e-01 -1.05733536e-01 5.23376405e-01] | [8.58228588104248, 5.946831703186035] |
267c9704-1a9a-47ca-9efa-2d640b36297f | segment-everything-everywhere-all-at-once | 2304.06718 | null | https://arxiv.org/abs/2304.06718v3 | https://arxiv.org/pdf/2304.06718v3.pdf | Segment Everything Everywhere All at Once | Despite the growing demand for interactive AI systems, there have been few comprehensive studies on human-AI interaction in visual understanding e.g. segmentation. Inspired by the development of prompt-based universal interfaces for LLMs, this paper presents SEEM, a promptable, interactive model for Segmenting Everything Everywhere all at once in an image. SEEM has four desiderata: i) Versatility: by introducing a versatile prompting engine for different types of prompts, including points, boxes, scribbles, masks, texts, and referred regions of another image; ii) Compositionality: by learning a joint visual-semantic space for visual and textual prompts to compose queries on the fly for inference as shown in Fig 1; iii)Interactivity: by incorporating learnable memory prompts to retain dialog history information via mask-guided cross-attention; and iv) Semantic-awareness: by using a text encoder to encode text queries and mask labels for open-vocabulary segmentation. | ['Yong Jae Lee', 'Jianfeng Gao', 'Linjie Li', 'Feng Li', 'Hao Zhang', 'Jianwei Yang', 'Xueyan Zou'] | 2023-04-13 | null | null | null | null | ['personalized-segmentation'] | ['computer-vision'] | [ 4.65957642e-01 6.51061833e-01 -5.84531836e-02 -5.60550272e-01
-3.72739673e-01 -9.19807076e-01 9.69369173e-01 2.46329159e-01
-3.48921418e-01 2.15814933e-01 5.49524307e-01 -3.76148015e-01
1.32178932e-01 -2.97415167e-01 -6.52124524e-01 -1.73342019e-01
2.28532508e-01 8.93396854e-01 5.75076818e-01 -2.16935620e-01
2.28539377e-01 3.95061672e-01 -1.64081585e+00 9.16445374e-01
6.88261569e-01 8.91415954e-01 6.66805387e-01 9.30689752e-01
-6.75083518e-01 1.12920320e+00 -7.11865246e-01 -3.57477754e-01
9.09318402e-03 -6.54592216e-01 -1.37875044e+00 4.97651935e-01
5.49278736e-01 -5.71198225e-01 -2.04728097e-01 7.58628011e-01
7.50216246e-02 3.31939697e-01 5.87440848e-01 -1.22549784e+00
-1.02865589e+00 9.01386678e-01 -1.96405947e-01 1.43675715e-01
8.71761084e-01 5.77135026e-01 1.21149838e+00 -7.04861999e-01
1.04717386e+00 1.49683273e+00 1.72160193e-02 8.68111312e-01
-1.38235462e+00 -7.08896071e-02 5.76162338e-01 3.04067463e-01
-1.02794111e+00 -3.37129086e-01 5.80238044e-01 -6.76818550e-01
1.09114361e+00 7.03154266e-01 9.60430801e-01 1.05020118e+00
-1.13212921e-01 1.37359202e+00 9.08968091e-01 -5.58573127e-01
9.55782682e-02 4.24317539e-01 1.30516827e-01 9.02580321e-01
-7.53692031e-01 1.10447109e-02 -7.57146537e-01 2.34861389e-01
1.29078209e+00 -1.25241317e-02 -2.60115489e-02 -3.69848192e-01
-1.31245399e+00 5.59820712e-01 4.96796340e-01 2.66742259e-01
-1.89212948e-01 1.97169974e-01 4.13088143e-01 4.38901186e-01
3.13778788e-01 6.93944335e-01 -2.68524915e-01 1.41173407e-01
-8.02484810e-01 3.59694391e-01 7.15802670e-01 1.34048998e+00
7.86383986e-01 -7.66722932e-02 -8.45725417e-01 6.93331778e-01
2.54065931e-01 4.24854189e-01 1.62382752e-01 -1.13646197e+00
3.28177661e-01 7.37492144e-01 2.31337875e-01 -6.25796318e-01
-4.68175113e-01 1.22791365e-01 -3.42549056e-01 1.11699991e-01
4.52625185e-01 -3.93913016e-02 -1.26974869e+00 1.69681168e+00
2.87680835e-01 -8.15826431e-02 -2.67604232e-01 1.08674598e+00
1.23147488e+00 6.94316566e-01 5.96970916e-01 2.15320051e-01
1.63906074e+00 -1.19272852e+00 -8.41944039e-01 -5.67476451e-01
4.10012811e-01 -4.99994665e-01 1.62909055e+00 1.56458199e-01
-1.11786020e+00 -7.04171121e-01 -6.63796842e-01 -7.98188269e-01
-6.98942542e-01 -1.40875742e-01 6.37379885e-01 1.22271515e-01
-1.33309102e+00 1.72281533e-01 -6.19713068e-01 -5.85104167e-01
3.58543158e-01 1.33663774e-01 -7.29520693e-02 2.49858335e-01
-8.40705216e-01 8.50580812e-01 2.57101476e-01 -2.11350605e-01
-8.80563974e-01 -6.86030686e-01 -9.71743643e-01 1.64780244e-01
7.42538393e-01 -8.59469891e-01 1.66381180e+00 -1.57169294e+00
-1.65643096e+00 1.20597351e+00 -3.78323048e-01 -2.94345826e-01
3.31795931e-01 -5.18098712e-01 -7.67596588e-02 4.55707490e-01
1.28052328e-02 1.51751590e+00 8.66625488e-01 -1.50245321e+00
-5.80194712e-01 -4.17119265e-01 3.60630900e-01 6.58157408e-01
1.89425245e-01 3.01374882e-01 -7.77914584e-01 -6.88081980e-01
-1.06013417e-01 -5.48294067e-01 -3.63017738e-01 3.21243852e-01
-7.00731874e-01 -3.98124725e-01 1.00780737e+00 -8.23402822e-01
1.16580081e+00 -2.17474794e+00 5.09912014e-01 -2.51610149e-02
3.44893247e-01 1.30045310e-01 -2.74312705e-01 5.02992213e-01
9.16922539e-02 6.61325231e-02 -1.36453599e-01 -5.18067658e-01
2.41990685e-01 3.23978782e-01 -5.19888341e-01 -1.96275100e-01
2.60216773e-01 1.55687952e+00 -9.18231726e-01 -7.18015432e-01
7.66958296e-01 3.11873704e-01 -4.72317547e-01 5.66851020e-01
-1.09864569e+00 6.41550362e-01 -3.08453023e-01 4.34133530e-01
7.84535185e-02 -4.68825221e-01 -1.23607501e-01 -1.38828859e-01
-2.21120924e-01 1.29722640e-01 -9.03999150e-01 2.05770946e+00
-2.03815803e-01 9.09832180e-01 3.73341203e-01 -5.55702984e-01
4.97246712e-01 3.74853373e-01 8.31132680e-02 -9.63160932e-01
1.15963720e-01 -3.39907885e-01 -5.45328736e-01 -7.85088062e-01
7.24328697e-01 2.28514075e-01 -2.43376285e-01 4.99985367e-01
1.39448956e-01 -4.52528447e-01 1.69283092e-01 6.19223416e-01
5.79949260e-01 4.13135976e-01 3.64135988e-02 -1.91631958e-01
2.18470499e-01 1.67281672e-01 -2.21863896e-01 8.92187774e-01
1.62755698e-02 5.76278627e-01 4.43413466e-01 -5.57301879e-01
-8.59508097e-01 -1.13983035e+00 2.37772644e-01 1.81308103e+00
4.87741917e-01 -2.94553429e-01 -9.59226608e-01 -4.06986147e-01
-1.16959237e-01 1.00539792e+00 -6.07755303e-01 3.15093160e-01
-4.31829154e-01 2.77783900e-01 1.33473083e-01 7.04484940e-01
2.95239538e-01 -1.68180096e+00 -1.29983354e+00 -1.30126700e-01
-1.80781603e-01 -1.06785643e+00 -8.09086084e-01 3.08647513e-01
-6.71266556e-01 -8.92676175e-01 -6.93458080e-01 -1.06406307e+00
7.83757150e-01 5.58104254e-02 1.42148399e+00 8.54830593e-02
-4.59768921e-01 9.66380775e-01 -3.07260275e-01 -3.08091521e-01
-3.53956431e-01 -1.43257022e-01 -8.44005346e-01 -6.90356344e-02
5.20500839e-02 -1.52596578e-01 -6.79104745e-01 2.21284106e-01
-9.96782660e-01 7.88350582e-01 3.25450838e-01 4.44624335e-01
5.30309677e-01 -8.11764300e-01 1.60575002e-01 -1.19407129e+00
6.56768024e-01 -1.17411815e-01 -5.06917596e-01 4.43870813e-01
-6.76615462e-02 4.64498252e-02 2.61945963e-01 -4.61561859e-01
-1.24064755e+00 2.16057092e-01 -7.52572492e-02 -2.77453095e-01
-6.74613655e-01 6.67956173e-02 -9.69986320e-02 4.00463343e-01
5.96937180e-01 1.34374648e-01 -1.00289867e-03 -3.07492226e-01
1.48104751e+00 5.55896819e-01 9.61439431e-01 -5.74188888e-01
2.87947297e-01 4.19744670e-01 -5.75075924e-01 -1.01713312e+00
-6.95475876e-01 -5.72862566e-01 -8.16796958e-01 -3.95167768e-01
1.41762578e+00 -6.75444305e-01 -1.03223193e+00 2.08236560e-01
-1.35411513e+00 -1.00622153e+00 -7.28688061e-01 -2.73059130e-01
-8.17426324e-01 6.42822012e-02 -7.88846254e-01 -7.22412348e-01
-2.44609267e-01 -1.26320469e+00 1.34264982e+00 4.37255502e-01
-8.34357142e-01 -1.03760540e+00 -4.88809913e-01 2.78960317e-01
2.67948061e-01 1.27710283e-01 1.02429044e+00 -7.49104142e-01
-9.46856439e-01 2.57310212e-01 -5.68048775e-01 -1.06450118e-01
-7.43861720e-02 -1.32382467e-01 -9.09526229e-01 1.86361909e-01
-3.75812292e-01 -4.68362987e-01 5.47103822e-01 3.12679052e-01
1.25145853e+00 -6.76534057e-01 -6.14735067e-01 5.37651241e-01
1.02773416e+00 6.02490842e-01 4.78210270e-01 -4.86468896e-02
8.52689147e-01 7.33261466e-01 2.81178266e-01 3.27924132e-01
6.44492745e-01 6.18090510e-01 3.81876200e-01 -5.55782259e-01
-3.97154242e-01 -4.39869612e-01 -1.54556558e-01 1.39105991e-01
3.75909299e-01 -4.88476813e-01 -8.42839718e-01 6.80166662e-01
-1.87157333e+00 -8.47357571e-01 7.13476166e-02 1.79242253e+00
8.36705625e-01 -1.46130130e-01 1.45938843e-01 -4.52631950e-01
3.79184783e-01 4.40436661e-01 -7.18103111e-01 -5.72751582e-01
-1.00634210e-01 1.51777714e-01 1.30846545e-01 8.06658268e-01
-1.06958020e+00 1.47551250e+00 6.94975328e+00 4.67045844e-01
-1.07562137e+00 9.46108624e-02 7.38936007e-01 4.37943377e-02
-6.39383435e-01 1.50232345e-01 -4.31157380e-01 1.17124453e-01
3.78404260e-01 1.69918910e-01 7.04804242e-01 4.91424978e-01
-5.09814546e-02 -3.94254506e-01 -1.41814053e+00 8.98331046e-01
2.77668595e-01 -1.50503469e+00 2.91573822e-01 -5.37536561e-01
4.69700992e-01 -1.89538285e-01 3.89378071e-02 1.17255315e-01
6.07066572e-01 -1.06478274e+00 1.14919770e+00 6.06312931e-01
1.00782645e+00 -2.87706703e-01 -9.66946930e-02 2.05673531e-01
-9.04576004e-01 -1.65916651e-01 2.20709160e-01 -2.50234455e-02
3.75076920e-01 -2.23386347e-01 -9.11807001e-01 1.48917452e-01
4.89446491e-01 4.44271028e-01 -6.10061467e-01 6.52771592e-01
-4.18296576e-01 1.75821513e-01 -1.05617858e-01 -1.43494278e-01
3.46750975e-01 -5.84486164e-02 4.00380880e-01 1.36936772e+00
-4.09237325e-01 1.80160210e-01 4.23042208e-01 1.13943958e+00
1.71367005e-01 3.27462144e-02 -4.54784185e-01 -2.40134686e-01
4.49627221e-01 1.00655079e+00 -1.01776040e+00 -5.38643479e-01
-4.42113101e-01 1.56379008e+00 6.21541440e-02 7.94474483e-01
-5.06713629e-01 -3.52481604e-01 2.86957800e-01 3.19142193e-01
2.76346862e-01 -2.45894343e-01 -6.35619402e-01 -6.70957208e-01
-4.41619635e-01 -8.89794290e-01 5.07201076e-01 -1.34900248e+00
-7.97431767e-01 6.78525567e-01 3.15200984e-01 -3.25987816e-01
-2.77597874e-01 -5.82241476e-01 -4.93246138e-01 8.66462827e-01
-7.44830608e-01 -1.37107980e+00 -3.45993340e-01 6.87472403e-01
1.13403201e+00 2.64414430e-01 7.80547798e-01 -1.08652890e-01
-9.02276337e-02 9.29199383e-02 -5.06346405e-01 1.14366598e-01
5.19349754e-01 -1.51493132e+00 8.13837945e-01 5.46343327e-01
5.21415353e-01 5.02864182e-01 7.87543416e-01 -6.86411142e-01
-1.21107805e+00 -5.95510364e-01 7.43576884e-01 -9.43522871e-01
4.02752012e-01 -7.67291546e-01 -8.30540538e-01 1.19086099e+00
8.98152947e-01 -5.43401420e-01 3.78058970e-01 3.54274362e-02
-4.49835598e-01 2.47734457e-01 -7.78691947e-01 1.06418228e+00
1.09485090e+00 -8.55361879e-01 -7.38100827e-01 5.17063916e-01
1.22005773e+00 -8.26411843e-01 -2.22150698e-01 3.46407034e-02
4.54430968e-01 -8.84851515e-01 1.01490235e+00 -5.80311418e-01
2.78427720e-01 -1.62379906e-01 1.02818213e-01 -9.46497083e-01
-2.68004507e-01 -9.22553062e-01 1.16993682e-02 1.10290861e+00
3.93826514e-01 -1.57077476e-01 6.55397415e-01 9.13498938e-01
-3.14583302e-01 -5.52976727e-01 -6.06221676e-01 -7.37799853e-02
-3.70585680e-01 -4.12608504e-01 5.12756526e-01 5.66603005e-01
2.93196052e-01 7.17736900e-01 -2.41090849e-01 -1.05636925e-01
7.77111501e-02 1.75707906e-01 8.20148885e-01 -1.04797018e+00
-2.58568048e-01 -7.80841470e-01 1.18515849e-01 -1.82425928e+00
2.31537595e-03 -8.51522982e-01 2.59250700e-01 -1.95692432e+00
1.85153872e-01 -1.44764736e-01 3.25033695e-01 7.03872740e-01
-2.44415015e-01 -3.19584981e-02 5.56498706e-01 1.20000571e-01
-1.04244733e+00 2.18164429e-01 1.57657647e+00 -1.60220772e-01
-5.14455974e-01 -4.04401898e-01 -7.26712286e-01 6.35006666e-01
2.38268957e-01 1.25772431e-01 -8.20272326e-01 -7.67764628e-01
3.49771641e-02 4.65273082e-01 6.42141342e-01 -6.97237432e-01
4.24740016e-01 -3.89225036e-01 3.79997611e-01 -7.33045876e-01
5.28844893e-01 -7.48700917e-01 8.78926665e-02 -1.20133918e-04
-9.23821211e-01 6.91839159e-02 4.45963979e-01 5.45551300e-01
2.03002524e-02 -1.42403552e-02 4.08044815e-01 -5.10432422e-01
-1.20769286e+00 5.44608049e-02 -6.73683286e-01 1.90226018e-01
8.46581578e-01 -4.49464351e-01 -3.30071628e-01 -5.66318154e-01
-1.08981776e+00 5.48448265e-01 5.12331903e-01 6.66331947e-01
5.64819694e-01 -8.03879797e-01 -1.54397085e-01 3.26833725e-01
1.96393341e-01 2.74722606e-01 3.52475852e-01 2.70737976e-01
-3.86018544e-01 4.82382208e-01 -9.97058526e-02 -7.73909807e-01
-1.31107843e+00 8.12008798e-01 5.61964996e-02 1.11292124e-01
-8.02866042e-01 1.26127136e+00 7.47913957e-01 -1.46955445e-01
6.91825271e-01 -4.97071713e-01 -8.98700301e-03 -5.30161289e-03
5.66342533e-01 -6.79971129e-02 -5.02898455e-01 -5.13866186e-01
-6.92040920e-02 3.48053277e-01 7.45153278e-02 -6.25971735e-01
7.59658873e-01 -5.01141727e-01 -6.24349080e-02 9.25106823e-01
6.22305334e-01 -3.28927994e-01 -1.62824059e+00 -3.33206713e-01
2.31796086e-01 -1.63681373e-01 -4.20586675e-01 -1.34352660e+00
-5.13384879e-01 9.85399425e-01 2.57600069e-01 4.05759096e-01
1.05448782e+00 6.13259017e-01 5.56127548e-01 1.95423409e-01
-3.11871711e-03 -1.05275989e+00 5.33692181e-01 5.67456484e-01
1.31206441e+00 -8.64414215e-01 -4.12786484e-01 -4.71683353e-01
-1.20838714e+00 9.69596505e-01 7.68338740e-01 3.71468544e-01
2.38513812e-01 2.06807137e-01 4.60634559e-01 -5.45839310e-01
-8.55876923e-01 -4.29311872e-01 4.98865306e-01 7.53474057e-01
4.41580147e-01 9.06300098e-02 1.81584463e-01 4.93936837e-01
-3.26990560e-02 -2.82952487e-01 -2.57547982e-02 6.82969928e-01
-6.46788538e-01 -8.54159832e-01 -1.49957612e-01 2.95608014e-01
-2.67770197e-02 -1.81901529e-01 -8.49110544e-01 6.48282290e-01
6.47593960e-02 9.70369577e-01 5.24140358e-01 -8.27473104e-02
1.15015127e-01 4.94603842e-01 5.14323473e-01 -9.50307906e-01
-9.08205926e-01 1.91841405e-02 1.91056300e-02 -7.89885879e-01
-2.54930437e-01 -4.92439836e-01 -1.45072949e+00 3.85794342e-01
6.45804256e-02 4.17553745e-02 5.58374107e-01 9.83660102e-01
5.10650814e-01 5.77864885e-01 -2.17803530e-02 -8.66967857e-01
-9.59616341e-03 -8.71089935e-01 -1.61144540e-01 6.75159574e-01
3.85265529e-01 -2.92970717e-01 4.18260805e-02 5.73790133e-01] | [10.915609359741211, 1.6670310497283936] |
b774c4c0-56bd-49bf-880f-9ef7489dbd9e | deep-vfx-deep-action-recognition-driven-vfx | 2007.11257 | null | https://arxiv.org/abs/2007.11257v1 | https://arxiv.org/pdf/2007.11257v1.pdf | Deep-VFX: Deep Action Recognition Driven VFX for Short Video | Human motion is a key function to communicate information. In the application, short-form mobile video is so popular all over the world such as Tik Tok. The users would like to add more VFX so as to pursue creativity and personlity. Many special effects are added on the short video platform. These gives the users more possibility to show off these personality. The common and traditional way is to create the template of VFX. However, in order to synthesis the perfect, the users have to tedious attempt to grasp the timing and rhythm of new templates. It is not easy-to-use especially for the mobile app. This paper aims to change the VFX synthesis by motion driven instead of the traditional template matching. We propose the AI method to improve this VFX synthesis. In detail, in order to add the special effect on the human body. The skeleton extraction is essential in this system. We also propose a novel form of LSTM to find out the user's intention by action recognition. The experiment shows that our system enables to generate VFX for short video more easier and efficient. | ['Feng Jiang', 'Ning Xie', 'Ao Luo', 'Zhijia Tao'] | 2020-07-22 | null | null | null | null | ['template-matching'] | ['computer-vision'] | [ 1.81950003e-01 -1.01176724e-01 -7.58600011e-02 -1.21772595e-01
1.81691200e-02 -1.60629645e-01 3.45217168e-01 -6.87623024e-01
-3.08422834e-01 4.38848913e-01 1.36996880e-01 -9.42475814e-03
5.57581000e-02 -7.90809333e-01 -4.75633532e-01 -4.48084503e-01
2.71213055e-01 -1.44808784e-01 3.14091146e-01 -4.25471038e-01
4.14459020e-01 2.03054160e-01 -1.53763545e+00 4.61345464e-01
6.36976242e-01 7.63009727e-01 8.85381699e-01 3.83108139e-01
-3.81393880e-01 5.73594391e-01 -5.07249594e-01 -2.61981666e-01
2.55457371e-01 -6.46671295e-01 -7.29759395e-01 1.42514333e-01
-6.32708147e-03 -5.32958567e-01 -3.86566848e-01 1.15083826e+00
6.88601077e-01 7.30643123e-02 3.62724781e-01 -1.21930885e+00
-2.33785793e-01 7.80060768e-01 -6.20172918e-01 1.05380919e-02
6.54062390e-01 2.52107203e-01 1.42896786e-01 -3.31384569e-01
6.85156703e-01 1.23765481e+00 7.17164814e-01 7.92433441e-01
-3.22375238e-01 -7.08632052e-01 -8.04062113e-02 5.79633832e-01
-1.42496943e+00 -6.15090907e-01 9.85028684e-01 -1.84070274e-01
6.32710636e-01 3.54870170e-01 1.10093284e+00 1.33350313e+00
4.97902185e-01 9.11131740e-01 8.18597317e-01 -5.46277702e-01
-2.62998521e-01 4.45626117e-02 -8.67539719e-02 7.43027031e-01
-1.83675259e-01 -2.41278529e-01 -3.40738595e-01 4.33050841e-01
1.17169631e+00 3.47835034e-01 -4.48722363e-01 6.22684285e-02
-1.44105184e+00 3.02557707e-01 1.30814746e-01 8.19831729e-01
-3.31620842e-01 2.53062159e-01 4.29787248e-01 2.11460665e-01
-2.71787077e-01 2.74261445e-01 -3.38897407e-01 -8.75589490e-01
-9.29878116e-01 1.21724829e-01 4.63378996e-01 8.94402802e-01
3.55328351e-01 1.91062942e-01 -7.89271593e-02 7.95127451e-01
4.88665074e-01 4.85578239e-01 1.15956604e+00 -1.21572459e+00
3.09567600e-01 6.72325790e-01 -5.71764037e-02 -1.32172906e+00
-2.34514117e-01 -2.47160032e-01 -8.88924062e-01 2.29492977e-01
2.99107641e-01 -1.62106425e-01 -8.39877784e-01 1.42401421e+00
7.53748268e-02 2.12292194e-01 -3.51485044e-01 1.02037859e+00
1.09828627e+00 8.96291852e-01 -2.41990939e-01 -4.18385714e-01
1.54573739e+00 -8.18515241e-01 -1.26749575e+00 -6.55153394e-02
6.48969948e-01 -9.92413163e-01 1.43612480e+00 6.08767807e-01
-1.06021380e+00 -9.66506898e-01 -1.20900595e+00 2.04852179e-01
-1.50696963e-01 3.54826957e-01 5.34030616e-01 7.04355299e-01
-5.74879646e-01 9.64036107e-01 -6.65908515e-01 -5.94924927e-01
6.29496351e-02 5.57602406e-01 -3.57863098e-01 4.38623011e-01
-1.50665843e+00 8.61310303e-01 4.60767120e-01 3.82178754e-01
-1.66943282e-01 -1.06996283e-01 -7.20401764e-01 -1.19777158e-01
3.84927064e-01 -8.55023801e-01 1.43811631e+00 -1.33797181e+00
-1.96969688e+00 6.63893402e-01 -1.35242313e-01 -8.60287473e-02
6.96199954e-01 -1.63649827e-01 -4.70302552e-01 4.99469303e-02
1.34611884e-02 6.55259311e-01 9.67688918e-01 -6.59336507e-01
-6.56057358e-01 -2.31258824e-01 9.09009799e-02 4.15403754e-01
-3.75496864e-01 -8.97450969e-02 -7.27711260e-01 -7.94961870e-01
6.84470385e-02 -9.43057179e-01 7.40053803e-02 -1.42351598e-01
-1.75228491e-01 -1.01810150e-01 1.15726662e+00 -9.57463205e-01
1.64496589e+00 -2.09900045e+00 1.58442855e-01 1.00315094e-01
1.08344689e-01 4.11381334e-01 2.83368081e-01 1.70334101e-01
1.61697179e-01 3.28322053e-02 1.63761124e-01 1.23384118e-01
-1.45728067e-01 1.32874891e-01 2.76316434e-01 6.31548986e-02
-3.82585913e-01 7.51756549e-01 -5.31093299e-01 -7.65899122e-01
1.19472109e-01 3.07311088e-01 -2.85402358e-01 -9.58946720e-02
8.66235346e-02 4.74767983e-01 -4.92588729e-01 4.49234426e-01
5.55135608e-01 -2.97053196e-02 -7.54990429e-02 -4.12987173e-01
-2.13441402e-01 -1.40883416e-01 -1.42435014e+00 2.03611636e+00
-4.69955146e-01 4.36571330e-01 -8.04471895e-02 -6.39560401e-01
9.46818709e-01 3.69236141e-01 4.23897117e-01 -7.44546473e-01
5.56843758e-01 3.10357034e-01 1.18187502e-01 -1.17841530e+00
4.55012262e-01 -4.83686961e-02 -2.87671406e-02 2.50412852e-01
-3.05302560e-01 5.24808057e-02 4.36603166e-02 -4.38098796e-02
6.35377765e-01 7.51095951e-01 6.11084104e-01 8.69906321e-02
7.28825092e-01 -2.71510273e-01 6.20329082e-01 1.85533524e-01
3.66867706e-02 4.14467961e-01 4.71572205e-02 -3.47220600e-01
-8.17700267e-01 -4.43092406e-01 3.53805751e-01 4.41884309e-01
2.80837029e-01 -4.11722571e-01 -9.65076566e-01 -4.94012982e-01
-4.89367634e-01 3.09868783e-01 -1.45717070e-01 -2.72955507e-01
-8.51545334e-01 -1.90106392e-01 4.34522390e-01 3.81441325e-01
1.28177536e+00 -1.37460458e+00 -8.40533853e-01 3.98228675e-01
-5.54029047e-01 -8.53929877e-01 -8.81626666e-01 -5.13039172e-01
-8.82418156e-01 -4.61322993e-01 -1.12900198e+00 -9.88120556e-01
1.94741011e-01 3.66143733e-01 3.48987848e-01 2.60193467e-01
-8.24195966e-02 1.67001382e-01 -5.71889997e-01 -2.95797825e-01
-3.01380873e-01 2.44300872e-01 1.41643316e-01 1.41343161e-01
3.79979074e-01 -6.40552819e-01 -5.29171288e-01 3.89980644e-01
-6.67680383e-01 4.15564746e-01 6.58965409e-01 3.31268996e-01
1.39631659e-01 4.37282622e-02 3.50374162e-01 -4.17272508e-01
7.12681353e-01 -3.20668854e-02 -3.72582674e-02 8.93719047e-02
-2.61258453e-01 -1.32105770e-02 5.91668010e-01 -7.35246480e-01
-1.09734833e+00 9.05379280e-02 -3.84279191e-01 -3.41201246e-01
-1.24811344e-01 2.33693019e-01 -5.19782364e-01 -1.06077887e-01
2.99918413e-01 2.03602865e-01 1.44018307e-01 -4.54373151e-01
7.27887675e-02 1.21379960e+00 4.13019121e-01 -1.04962282e-01
5.79943001e-01 1.36342928e-01 -8.17568675e-02 -1.16926253e+00
-2.62775775e-02 -2.21143275e-01 -6.34235978e-01 -8.01775813e-01
1.06132686e+00 -4.37130392e-01 -1.25259590e+00 6.67053163e-01
-1.49503219e+00 -1.04013957e-01 1.73121110e-01 6.73182189e-01
-6.10666871e-01 8.46118808e-01 -5.68065226e-01 -6.75461471e-01
-3.53586823e-01 -1.18663347e+00 8.20078015e-01 5.46252906e-01
-4.78113592e-01 -5.73889196e-01 -2.09479615e-01 3.19994599e-01
1.68969154e-01 4.95259427e-02 4.78566438e-01 6.22422397e-02
-5.83876312e-01 1.21037580e-01 2.58880970e-03 8.52442458e-02
4.20830250e-01 2.16410607e-01 -5.39149463e-01 1.99704617e-01
2.77058423e-01 2.54485816e-01 6.14623725e-01 3.60674143e-01
1.03492033e+00 -5.15628755e-01 -5.03617883e-01 6.87971592e-01
1.05678332e+00 9.95108068e-01 1.23162758e+00 5.89723229e-01
7.97475278e-01 6.61230385e-01 9.48378026e-01 3.11131686e-01
5.45315854e-02 1.05542374e+00 -3.18206325e-02 3.41571793e-02
-1.24976255e-01 -3.25677961e-01 6.14246786e-01 1.24574983e+00
-5.90937138e-01 4.01311442e-02 -7.07046330e-01 8.00178200e-02
-1.84177804e+00 -1.41129982e+00 -4.14697349e-01 2.09018278e+00
7.22172081e-01 2.22974509e-01 5.73365912e-02 5.31467021e-01
7.87190735e-01 -2.75520179e-02 4.49004769e-02 -5.24795175e-01
8.41925889e-02 -3.00685642e-03 2.17801094e-01 4.81185019e-01
-9.15501058e-01 8.56337070e-01 5.02719212e+00 1.14757669e+00
-1.56016552e+00 3.21655869e-02 2.66482413e-01 -1.09834950e-02
-1.25118524e-01 -1.51232377e-01 -8.35253060e-01 8.17182064e-01
4.17646110e-01 2.66561974e-02 4.11619872e-01 4.68007296e-01
7.19576299e-01 -9.88229290e-02 -5.47606051e-01 1.59644127e+00
1.25565842e-01 -1.17104614e+00 -7.12166028e-03 -5.84260821e-02
2.64402419e-01 -8.18295479e-01 -1.73595086e-01 2.63931602e-01
-8.13856661e-01 -7.62268364e-01 7.76363254e-01 8.99688184e-01
9.23795044e-01 -4.89510566e-01 6.98042989e-01 5.10468602e-01
-1.29441774e+00 -4.43970673e-02 -2.80884236e-01 -3.11518848e-01
4.11049068e-01 6.28071651e-02 -7.91238904e-01 4.23639506e-01
7.07256734e-01 5.84978282e-01 -3.74165028e-01 1.03891480e+00
6.50320500e-02 7.03694578e-03 -2.54169703e-01 -4.39307451e-01
1.25124007e-01 -4.05629694e-01 5.75317740e-01 1.00714219e+00
8.51228297e-01 1.35462210e-01 -8.93318281e-02 3.85652155e-01
1.50158271e-01 4.13363725e-01 -8.43043387e-01 -1.27296239e-01
6.10094927e-02 1.10130489e+00 -7.03848004e-01 -3.30380768e-01
-2.85485327e-01 1.50342977e+00 -3.91108513e-01 -8.61663371e-03
-1.09587491e+00 -7.88508475e-01 7.34240636e-02 5.73567927e-01
-8.83866660e-03 -3.52864832e-01 7.33961258e-03 -1.07127440e+00
1.00781672e-01 -1.22018278e+00 -1.23731025e-01 -9.42619503e-01
-7.13718116e-01 5.17876089e-01 -6.75776824e-02 -1.60086334e+00
-4.01306361e-01 -4.53843892e-01 -6.60864413e-01 6.39866948e-01
-6.37069583e-01 -1.25309861e+00 -5.92030585e-01 5.38367450e-01
1.02731264e+00 -2.22615495e-01 5.10056198e-01 6.19126499e-01
-4.35492337e-01 3.84544879e-01 -2.98945218e-01 -6.88552335e-02
8.25816572e-01 -6.06595993e-01 1.94046095e-01 6.92872226e-01
3.10362428e-02 7.97426224e-01 8.15847754e-01 -9.76089835e-01
-1.07597816e+00 -3.74629080e-01 9.00379241e-01 -1.70487851e-01
1.43025760e-02 -1.22391470e-01 -5.86682618e-01 5.80160558e-01
3.54051739e-01 -8.35405231e-01 2.50503361e-01 -4.39104617e-01
5.25323689e-01 -4.88603055e-01 -1.08292842e+00 1.08418941e+00
1.29710865e+00 -1.08503341e-03 -7.62635231e-01 -1.46448508e-01
6.43136322e-01 -2.21358985e-01 -4.97719526e-01 4.99049872e-01
1.17766631e+00 -1.06846642e+00 7.19839215e-01 -1.47485927e-01
2.61612982e-01 -5.57367623e-01 1.72724113e-01 -8.95424306e-01
-4.03295666e-01 -1.11016381e+00 1.92320690e-01 1.30033123e+00
1.54798165e-01 -4.08406049e-01 9.40376699e-01 3.26770574e-01
-2.47408636e-02 -6.22599185e-01 -5.88864446e-01 -6.64871871e-01
-8.32205653e-01 -3.12858909e-01 6.76713943e-01 8.94174278e-01
2.74302512e-01 5.59248805e-01 -7.40513682e-01 -2.83065408e-01
1.22548662e-01 -1.37746394e-01 9.73861873e-01 -1.13250005e+00
-4.28188652e-01 -3.73074621e-01 -5.09379327e-01 -1.30257714e+00
-2.75087714e-01 -4.54285771e-01 -3.21527481e-01 -1.63755691e+00
-3.72592025e-02 -5.81251793e-02 2.39965886e-01 2.81214565e-01
2.22279578e-01 1.96231812e-01 5.00163972e-01 2.39077538e-01
-2.60192901e-01 4.22812998e-01 1.71889472e+00 -1.32710218e-01
-6.09447777e-01 2.93002069e-01 -3.42686146e-01 8.81465077e-01
9.72935081e-01 -1.41941264e-01 -6.12642407e-01 -2.72159070e-01
2.33661622e-01 3.76711607e-01 1.19902715e-01 -1.33459949e+00
1.75713316e-01 -1.55069426e-01 5.84385574e-01 -6.11059189e-01
5.13008773e-01 -8.95459831e-01 5.79094470e-01 8.29624534e-01
2.41349012e-01 3.44811976e-01 -6.82330057e-02 -1.61044896e-02
-1.17575593e-01 -6.51498318e-01 4.82647628e-01 -7.04922795e-01
-9.59301531e-01 5.20263202e-02 -5.74093461e-01 -4.91932094e-01
1.11794484e+00 -8.49772453e-01 1.80271305e-02 -8.72230828e-01
-6.70107484e-01 -2.02968158e-02 3.47483963e-01 5.50703824e-01
7.90185928e-01 -1.45441604e+00 -7.62085468e-02 1.91106930e-01
-2.77678192e-01 -6.19186997e-01 5.69835782e-01 7.95823276e-01
-9.85549271e-01 3.91844064e-01 -8.97757530e-01 -2.08163872e-01
-1.64302230e+00 5.87629974e-01 2.70811528e-01 3.91155705e-02
-5.14318347e-01 3.77027661e-01 -9.99229327e-02 1.89190567e-03
1.38614193e-01 -4.28208321e-01 -7.83829272e-01 1.38862804e-01
4.68415767e-01 6.22938991e-01 -4.08346683e-01 -5.31433761e-01
-1.65413693e-01 9.25180733e-01 2.92939126e-01 -3.70653927e-01
8.56500506e-01 -3.68896037e-01 -8.30898359e-02 5.38927019e-01
1.05639040e+00 3.03224355e-01 -7.67083645e-01 3.86998922e-01
-2.82117397e-01 -5.84393620e-01 -3.71524096e-01 -5.34336388e-01
-8.57782900e-01 1.08992720e+00 7.25695193e-01 3.72307077e-02
1.14261949e+00 -5.99589646e-01 1.39307761e+00 2.63909310e-01
7.57340968e-01 -1.24457300e+00 2.64664348e-02 2.26761818e-01
8.90522599e-01 -7.52318799e-01 -1.14930145e-01 -4.24056679e-01
-7.38568008e-01 1.39827263e+00 7.44145751e-01 2.68529683e-01
3.42544824e-01 1.18108608e-01 1.23635426e-01 5.51686212e-02
-1.72441244e-01 -1.19097039e-01 1.83036536e-01 5.91010690e-01
4.48529691e-01 -1.58260599e-01 -1.03042281e+00 8.12353790e-01
-3.86644036e-01 4.72333580e-01 7.36219943e-01 7.22804666e-01
-6.48547828e-01 -1.49451041e+00 -5.24730802e-01 3.14527243e-01
-6.51599884e-01 2.73360103e-01 -1.97984591e-01 6.78775787e-01
5.65298975e-01 7.09464192e-01 -3.00636083e-01 -6.61920905e-01
3.47735703e-01 1.97354868e-01 7.15434492e-01 -2.38997772e-01
-4.36772645e-01 2.51228631e-01 4.71712053e-02 -6.04616284e-01
-4.16400045e-01 -4.47531700e-01 -1.47589874e+00 -3.23285818e-01
-8.17171410e-02 -7.20680282e-02 6.38559222e-01 8.05310428e-01
2.77897082e-02 5.14580905e-01 3.26855659e-01 -8.70933115e-01
-1.64670706e-01 -1.12484741e+00 -3.44004631e-01 3.44107896e-01
-1.80207208e-01 -5.89323223e-01 5.87444641e-02 1.89468667e-01] | [10.783562660217285, -0.7933686971664429] |
560515f5-f55a-4600-b5fc-892769722b28 | generalizing-interactive-backpropagating | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Lin_Generalizing_Interactive_Backpropagating_Refinement_for_Dense_Prediction_Networks_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Lin_Generalizing_Interactive_Backpropagating_Refinement_for_Dense_Prediction_Networks_CVPR_2022_paper.pdf | Generalizing Interactive Backpropagating Refinement for Dense Prediction Networks | As deep neural networks become the state-of-the-art approach in the field of computer vision for dense prediction tasks, many methods have been developed for automatic estimation of the target outputs given the visual inputs. Although the estimation accuracy of the proposed automatic methods continues to improve, interactive refinement is oftentimes necessary for further correction. Recently, feature backpropagating refinement scheme (f-BRS) has been proposed for the task of interactive segmentation, which enables efficient optimization of a small set of auxiliary variables inserted into the pretrained network to produce object segmentation that better aligns with user inputs. However, the proposed auxiliary variables only contain channel-wise scale and bias, limiting the optimization to global refinement only. In this work, in order to generalize backpropagating refinement for a wide range of dense prediction tasks, we introduce a set of G-BRS (Generalized Backpropagating Refinement Scheme) layers that enable both global and localized refinement for the following tasks: interactive segmentation, semantic segmentation, image matting and monocular depth estimation. Experiments on SBD, Cityscapes, Mapillary Vista, Composition-1k and NYU-Depth-V2 show that our method can successfully generalize and significantly improve performance of existing pretrained state-of-the-art models with only a few clicks. | ['Tony Martinez', 'Brian Price', 'Fanqing Lin'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['image-matting'] | ['computer-vision'] | [ 3.44215184e-01 2.90944695e-01 -5.70323355e-02 -6.54229164e-01
-5.93463898e-01 -1.16388485e-01 4.86916929e-01 -9.26156119e-02
-6.07789457e-01 5.08563459e-01 -1.74513862e-01 -5.68428747e-02
3.44746441e-01 -8.07495713e-01 -8.74535620e-01 -5.44298291e-01
2.34870911e-01 6.25678420e-01 8.23676586e-01 5.30251339e-02
1.91551611e-01 4.97930497e-01 -1.47186911e+00 3.07679981e-01
9.99956906e-01 1.37802017e+00 5.72589338e-01 6.52419329e-01
-2.13138133e-01 7.28106678e-01 -3.70584637e-01 -3.32368821e-01
5.09332418e-01 -2.17838511e-01 -9.64493155e-01 1.04951277e-01
8.44891369e-01 -5.60672879e-01 -2.02598080e-01 1.08688521e+00
3.95685792e-01 1.32485330e-01 6.57515407e-01 -1.04049778e+00
-4.92908061e-01 4.63400334e-01 -8.79193127e-01 1.10460654e-01
-1.15051471e-01 2.09078893e-01 8.45191360e-01 -9.42177951e-01
7.11314440e-01 1.28537750e+00 6.44593954e-01 6.37807250e-01
-1.17329288e+00 -6.95682228e-01 5.84427238e-01 3.11864197e-01
-1.30799997e+00 -8.67852420e-02 6.75961614e-01 -4.42820668e-01
1.17395306e+00 4.28334773e-02 8.36040199e-01 7.48177588e-01
-5.78337768e-03 1.10977912e+00 9.96433616e-01 -2.60777920e-01
1.81934148e-01 2.47409314e-01 1.45216407e-02 9.07170951e-01
4.97762077e-02 5.88984303e-02 -4.78882402e-01 4.77525711e-01
1.08723330e+00 -2.13443547e-01 -1.68969274e-01 -5.50009072e-01
-9.24834549e-01 8.82213771e-01 9.44626451e-01 -7.10071716e-03
-3.52170020e-01 1.94107696e-01 1.21937141e-01 -4.90536690e-02
8.59203398e-01 4.42975730e-01 -7.20478773e-01 9.67422500e-02
-1.38026810e+00 3.45873833e-01 4.83781934e-01 9.46737111e-01
1.10825634e+00 6.56287149e-02 -3.36944789e-01 8.18669498e-01
3.29970926e-01 3.73496860e-01 2.78277814e-01 -1.03244328e+00
6.20430768e-01 8.50905597e-01 -6.62843091e-03 -8.02238464e-01
-5.93082726e-01 -5.27436435e-01 -9.06658471e-01 4.55463767e-01
4.25617754e-01 -7.57609382e-02 -1.64171004e+00 1.39508057e+00
5.76400995e-01 1.93383172e-01 -2.16735408e-01 1.07305157e+00
9.54770148e-01 5.64475656e-01 3.00915837e-02 1.24113522e-01
7.56531656e-01 -1.41425550e+00 -2.93002844e-01 -6.23034298e-01
5.30417144e-01 -5.28490722e-01 9.68532681e-01 3.87321442e-01
-1.13303638e+00 -7.57757425e-01 -8.99603486e-01 -4.33480710e-01
-2.71234363e-01 -3.89925030e-04 7.21461892e-01 4.35176283e-01
-1.17896867e+00 5.95858335e-01 -9.92624879e-01 -2.30898678e-01
8.50905359e-01 4.33679044e-01 -2.23365232e-01 -9.86046344e-02
-9.04289722e-01 7.70211458e-01 5.20119369e-01 3.37557346e-01
-8.86044741e-01 -8.18691730e-01 -9.26498652e-01 -9.35693309e-02
3.31092626e-01 -7.49183178e-01 1.17187345e+00 -1.04042578e+00
-1.45863080e+00 9.90368903e-01 -9.83050745e-03 -6.33141637e-01
7.46190131e-01 -4.23851341e-01 2.23521322e-01 4.41705063e-02
1.56421125e-01 1.49765038e+00 8.67906809e-01 -1.19820094e+00
-1.04661703e+00 -5.00598252e-01 6.72976971e-02 4.97796297e-01
-1.02131911e-01 -3.96134853e-01 -1.02214217e+00 -5.98230422e-01
4.26708400e-01 -7.87418604e-01 -5.97768188e-01 1.91025823e-01
-4.97235686e-01 -5.20701781e-02 7.74104953e-01 -6.77549779e-01
9.66626644e-01 -1.90735304e+00 6.35739565e-01 1.81279644e-01
3.09684277e-01 1.45049006e-01 -6.13332354e-02 -3.82642567e-01
1.83091223e-01 -1.17254995e-01 -5.53821862e-01 -8.46730530e-01
-2.93053389e-01 2.73690492e-01 4.25765552e-02 3.65879029e-01
8.26587453e-02 1.03441787e+00 -5.85460246e-01 -4.58904564e-01
5.80265045e-01 3.95805508e-01 -8.39806020e-01 3.57452214e-01
-5.79236209e-01 5.71018636e-01 -2.73235768e-01 7.21112072e-01
8.86973083e-01 -3.37979674e-01 -3.25438857e-01 -3.33000213e-01
-1.71142653e-01 -3.98027040e-02 -1.24581599e+00 1.91993415e+00
-3.46579820e-01 5.43727338e-01 1.37861550e-01 -8.92697453e-01
8.04110348e-01 -2.09484458e-01 3.27275693e-01 -7.18221903e-01
2.52357453e-01 8.14927444e-02 -7.53472596e-02 -2.91598439e-01
5.85902512e-01 1.93359330e-01 2.51775384e-01 8.00979324e-03
2.15421781e-01 -4.97380495e-01 7.03032017e-02 1.21296234e-01
6.53800309e-01 3.43426317e-01 9.45137590e-02 -4.02441099e-02
5.13142049e-01 4.98057529e-02 6.13215446e-01 7.24541724e-01
-1.42502040e-01 1.16844535e+00 1.23818353e-01 -4.82442558e-01
-9.85688150e-01 -6.32097900e-01 -1.00183561e-01 1.04469037e+00
4.85987186e-01 -2.08319724e-01 -1.01812541e+00 -7.87523925e-01
-4.48882915e-02 5.51843762e-01 -8.23302329e-01 1.36052445e-01
-5.81562340e-01 -7.92343557e-01 3.21553290e-01 9.52675998e-01
1.01150727e+00 -1.25164700e+00 -5.86018920e-01 2.37262860e-01
-6.72755092e-02 -1.29175723e+00 -4.09314930e-01 3.34855705e-01
-1.04797530e+00 -9.27443206e-01 -1.07439244e+00 -6.67444944e-01
7.00134873e-01 3.59651819e-03 1.04260945e+00 -2.68340409e-02
-2.86545038e-01 4.14700024e-02 -1.70126259e-01 -2.48434156e-01
-1.49891958e-01 4.75982606e-01 -4.74429607e-01 -1.99508332e-02
1.34451538e-01 -3.63847375e-01 -8.70630980e-01 1.67340010e-01
-7.46490121e-01 7.95227289e-01 6.22883260e-01 6.34784877e-01
9.73043859e-01 -3.41966897e-01 7.38419592e-02 -1.27365911e+00
6.36316789e-03 -1.42124206e-01 -7.25666344e-01 7.22665265e-02
-7.09146857e-01 1.40524253e-01 3.44178021e-01 -2.96429157e-01
-1.14162171e+00 3.67509723e-01 -5.16174316e-01 -6.80490494e-01
-3.72336149e-01 2.66883582e-01 -1.95099697e-01 -3.71553749e-01
5.86369693e-01 2.34953701e-01 -4.02823478e-01 -5.89125812e-01
5.61053693e-01 3.81690264e-01 6.04329586e-01 -1.68717027e-01
7.81006753e-01 4.30409074e-01 -5.93311936e-02 -5.65068066e-01
-1.05439925e+00 -5.14283597e-01 -9.11920726e-01 -1.60918728e-01
1.23757720e+00 -1.05933535e+00 -2.70686835e-01 1.00289285e+00
-1.06097078e+00 -8.32919478e-01 -2.04238400e-01 1.87877864e-01
-5.02608299e-01 1.78234667e-01 -6.07610345e-01 -5.04497588e-01
-5.63613117e-01 -1.41403544e+00 1.26988268e+00 3.79972965e-01
-6.65238947e-02 -9.12186503e-01 -1.63922951e-01 5.77868879e-01
3.81601423e-01 2.06713572e-01 8.06785941e-01 -1.95673525e-01
-9.41130161e-01 7.67144784e-02 -6.01147830e-01 5.11248767e-01
2.85262354e-02 -2.55737513e-01 -1.08471036e+00 -1.76907152e-01
-2.66104043e-01 -2.77809262e-01 1.14790869e+00 7.96042621e-01
1.43606353e+00 -2.37873830e-02 -3.52609664e-01 1.29917359e+00
1.30638516e+00 -6.09475281e-03 6.84323311e-01 5.78152239e-01
1.32076943e+00 2.75359690e-01 7.08176076e-01 2.49856487e-01
4.98319358e-01 6.19002879e-01 7.48320222e-01 -5.29555500e-01
-3.70310545e-01 -1.01242885e-01 -1.72444344e-01 3.64803433e-01
-2.05193833e-01 -5.68265617e-02 -8.04680645e-01 5.46192586e-01
-1.76107717e+00 -4.13842976e-01 -1.67315066e-01 1.90816545e+00
8.84919345e-01 5.11746585e-01 2.00724583e-02 -4.06020395e-02
4.78742063e-01 2.04378754e-01 -9.08315182e-01 -1.72831729e-01
-5.75311258e-02 3.94482344e-01 8.22309732e-01 7.00026631e-01
-1.21371627e+00 1.40681279e+00 5.47226572e+00 8.44286203e-01
-1.37921989e+00 9.13760364e-02 1.03127742e+00 -4.09673667e-03
-1.18161224e-01 -2.32045084e-01 -1.13450098e+00 1.32670194e-01
2.88245946e-01 3.99127424e-01 2.33465835e-01 1.08839917e+00
6.44859970e-02 -4.09060508e-01 -1.17990279e+00 1.09773338e+00
-4.63355146e-03 -1.46244109e+00 1.75269201e-01 -2.02255160e-01
1.22491610e+00 2.75557518e-01 1.01448566e-01 3.50326419e-01
-6.46024477e-03 -1.04331541e+00 8.78727973e-01 3.20780635e-01
7.75442660e-01 -6.88851416e-01 7.85772264e-01 3.40798706e-01
-1.10571814e+00 4.44980264e-02 -2.80916214e-01 2.91413553e-02
1.45617306e-01 4.38493371e-01 -8.37287605e-01 2.05135509e-01
9.86557543e-01 7.85080791e-01 -8.58348429e-01 1.17072463e+00
-3.19864988e-01 3.61894399e-01 -4.00554329e-01 1.99217886e-01
5.52918196e-01 2.90552787e-02 2.49792561e-01 1.07693911e+00
9.06556249e-02 6.91440254e-02 1.53976873e-01 9.04222190e-01
-1.71691209e-01 7.84367397e-02 2.42194580e-03 3.60862613e-01
2.05673762e-02 1.20742857e+00 -1.06744432e+00 -3.43335390e-01
-2.83514410e-01 1.20484686e+00 5.71059465e-01 3.95647854e-01
-8.26614022e-01 -7.18116015e-02 5.74238241e-01 4.58442777e-01
6.32635057e-01 -1.86311081e-01 -7.75367796e-01 -1.01029444e+00
-2.42324397e-01 -6.70665026e-01 1.44436881e-01 -6.64318621e-01
-8.59466910e-01 7.81883359e-01 -5.10469750e-02 -7.58383870e-01
-2.08001271e-01 -4.87973303e-01 -3.90207827e-01 9.55922842e-01
-1.63219023e+00 -1.24077797e+00 -7.20793068e-01 6.22065604e-01
1.11048055e+00 -9.56328865e-03 4.56196666e-01 2.43963420e-01
-5.02963662e-01 5.71922421e-01 -2.67596751e-01 -1.18544539e-02
6.25887632e-01 -1.39654362e+00 7.36224592e-01 8.94641995e-01
5.64428382e-02 5.37598915e-02 7.04871655e-01 -7.34140158e-01
-7.57198095e-01 -1.47464633e+00 3.54955733e-01 -6.32310659e-02
2.65585452e-01 -4.54541326e-01 -9.64585006e-01 5.72009563e-01
6.45679189e-03 1.39069721e-01 -3.47260125e-02 -1.08544983e-01
-1.01067889e-02 -4.48375307e-02 -1.07642591e+00 5.98821580e-01
1.07227373e+00 -1.71832159e-01 -1.94944799e-01 2.89090335e-01
7.79850781e-01 -9.87782717e-01 -4.97624576e-01 5.76116920e-01
3.05238783e-01 -1.12710786e+00 9.48160052e-01 -4.14883085e-02
5.12409687e-01 -3.36466044e-01 -6.26593456e-02 -1.07935083e+00
-2.70225346e-01 -2.25893676e-01 -2.57202297e-01 8.97128165e-01
5.87503314e-01 -1.64736077e-01 1.34061944e+00 6.98520184e-01
-3.30445409e-01 -1.09627497e+00 -7.44084179e-01 -2.02186301e-01
-9.70352292e-02 -6.40516281e-01 2.75952160e-01 5.79092026e-01
-8.19720149e-01 3.03648740e-01 -3.39423686e-01 2.07966089e-01
6.58071160e-01 -5.76987155e-02 8.96469355e-01 -1.16992295e+00
-4.93767172e-01 -4.35224354e-01 -4.50119406e-01 -1.67913413e+00
6.45858468e-03 -6.86246216e-01 2.14392483e-01 -1.83266258e+00
9.89200845e-02 -5.55105209e-01 -4.18065675e-02 5.65420985e-01
-3.25874358e-01 6.38144493e-01 1.65242106e-01 7.73354480e-03
-5.78271389e-01 5.89935422e-01 1.50286949e+00 -1.97600633e-01
-5.14727890e-01 3.03852350e-01 -4.51666951e-01 8.65085840e-01
5.36188662e-01 -3.16497684e-01 -5.32542527e-01 -5.69185376e-01
9.66428816e-02 -1.34348720e-01 3.75061363e-01 -1.04521096e+00
3.29281390e-01 -9.73730236e-02 5.58328450e-01 -9.59716558e-01
4.89312798e-01 -6.85572684e-01 -4.05311622e-02 2.54869968e-01
-2.94263154e-01 -2.34126732e-01 3.58605564e-01 4.58848000e-01
-2.16828167e-01 -1.02656312e-01 1.00457764e+00 -2.53076583e-01
-1.22111499e+00 7.23649800e-01 -1.29143931e-02 1.61629096e-01
9.15957451e-01 -5.64316988e-01 3.99007015e-02 -1.71477616e-01
-8.55996251e-01 3.13582867e-01 3.46203834e-01 3.47469628e-01
8.37927222e-01 -9.57528770e-01 -6.11028910e-01 2.71887451e-01
-8.25389922e-02 7.66920507e-01 4.30904120e-01 7.57454693e-01
-8.71977866e-01 2.83494353e-01 -2.02058733e-01 -9.06062663e-01
-1.24316084e+00 1.85103029e-01 5.02980828e-01 -2.54156768e-01
-8.90807450e-01 1.48608100e+00 6.79255605e-01 -3.23657125e-01
5.40385008e-01 -7.15814531e-01 -1.76477924e-01 -1.54320255e-01
2.85531491e-01 1.23894475e-01 2.45972514e-01 -5.23641884e-01
-2.54003912e-01 6.32794142e-01 -4.04594958e-01 3.68962698e-02
1.43693483e+00 -3.04451644e-01 1.64350137e-01 3.56067508e-01
1.08119524e+00 -6.03738129e-01 -1.75126469e+00 -2.10990950e-01
-3.97088766e-01 -4.75519389e-01 3.90591443e-01 -7.73278594e-01
-1.53252530e+00 1.01658297e+00 7.64784634e-01 -1.88933358e-01
1.17447412e+00 1.99801568e-02 8.27505410e-01 5.32755516e-02
2.83762008e-01 -1.10931969e+00 -3.42813134e-02 5.72682381e-01
7.58875430e-01 -1.44055665e+00 -1.42162964e-01 -5.55088580e-01
-6.01209521e-01 9.13551867e-01 1.18810165e+00 -1.44446000e-01
5.52149236e-01 3.17980617e-01 6.97262511e-02 -7.74908215e-02
-3.57239038e-01 -2.62170643e-01 5.14380813e-01 5.96623242e-01
3.14700335e-01 -1.07809454e-01 4.97655347e-02 3.54637295e-01
-2.34578520e-01 -1.76896587e-01 2.77050555e-01 5.74538231e-01
-5.17565429e-01 -7.64944971e-01 -1.95671976e-01 5.24732828e-01
-2.93723106e-01 -3.83836001e-01 -1.87021255e-01 9.23002779e-01
3.76320094e-01 4.36716586e-01 1.92229584e-01 -2.19382688e-01
3.69737327e-01 -2.77970850e-01 5.33044457e-01 -7.84513295e-01
-5.70514798e-01 1.51622742e-01 -2.07660079e-01 -7.48288453e-01
-4.59652811e-01 -4.90207672e-01 -1.26814473e+00 -7.98838772e-03
-4.63344187e-01 -3.70358855e-01 5.72855949e-01 1.17260623e+00
5.68653233e-02 7.07269192e-01 1.12481400e-01 -1.28270543e+00
-4.99585867e-02 -1.11914313e+00 -4.72056180e-01 2.23386094e-01
1.91896558e-01 -5.77749133e-01 -1.45646065e-01 9.12655964e-02] | [9.526975631713867, 0.009137987159192562] |
f37604a2-cd9f-4572-9e8a-0d0c960b4b74 | local-relighting-of-real-scenes | 2207.02774 | null | https://arxiv.org/abs/2207.02774v1 | https://arxiv.org/pdf/2207.02774v1.pdf | Local Relighting of Real Scenes | We introduce the task of local relighting, which changes a photograph of a scene by switching on and off the light sources that are visible within the image. This new task differs from the traditional image relighting problem, as it introduces the challenge of detecting light sources and inferring the pattern of light that emanates from them. We propose an approach for local relighting that trains a model without supervision of any novel image dataset by using synthetically generated image pairs from another model. Concretely, we collect paired training images from a stylespace-manipulated GAN; then we use these images to train a conditional image-to-image model. To benchmark local relighting, we introduce Lonoff, a collection of 306 precisely aligned images taken in indoor spaces with different combinations of lights switched on. We show that our method significantly outperforms baseline methods based on GAN inversion. Finally, we demonstrate extensions of our method that control different light sources separately. We invite the community to tackle this new task of local relighting. | ['David Bau', 'Rohit Kumar', 'Shahin Mahdizadehaghdam', 'Antonio Torralba', 'Agata Lapedriza', 'Ali Jahanian', 'Audrey Cui'] | 2022-07-06 | null | null | null | null | ['image-relighting'] | ['computer-vision'] | [ 8.95174026e-01 -1.31727681e-01 3.51424754e-01 -4.57642645e-01
-7.16997683e-01 -7.15456545e-01 7.96085835e-01 -5.69209158e-01
-1.34913996e-01 8.50370228e-01 1.61507219e-01 -1.10887110e-01
5.18921137e-01 -9.62596416e-01 -1.35783851e+00 -8.33528399e-01
6.25127792e-01 2.24123344e-01 6.09377883e-02 -1.83894157e-01
2.73623586e-01 3.97798926e-01 -1.52272356e+00 2.04046756e-01
7.68456638e-01 6.86292112e-01 2.85270631e-01 9.48973298e-01
2.61588812e-01 9.61977780e-01 -6.40311420e-01 -8.84727836e-02
6.45754158e-01 -8.94619346e-01 -6.35985434e-01 4.07965243e-01
1.07590091e+00 -5.63676655e-01 -3.18333238e-01 7.61021733e-01
3.98655534e-01 1.70707107e-01 4.68998790e-01 -1.20364892e+00
-7.47326851e-01 -5.50052226e-02 -6.74223959e-01 -8.72332975e-02
5.72595298e-01 3.52114141e-01 6.68941319e-01 -6.57041669e-01
9.08651888e-01 1.02895308e+00 5.71896732e-01 4.04463530e-01
-1.77614403e+00 -6.57162488e-01 -1.98323518e-01 -5.68040013e-02
-1.04276955e+00 -7.26791918e-01 8.84326160e-01 -5.69834590e-01
5.62058747e-01 3.09146225e-01 6.06822610e-01 1.23950887e+00
7.64643550e-02 3.80429178e-01 1.62169874e+00 -6.96644783e-01
2.59480655e-01 -1.63083836e-01 -3.94940734e-01 7.32054174e-01
-2.69947708e-01 3.43392998e-01 -7.59827673e-01 1.98822498e-01
9.48070586e-01 -7.75037780e-02 -5.42138815e-01 -4.59054857e-01
-1.52281952e+00 4.97770011e-01 6.65187240e-01 -3.90279456e-03
-6.98502660e-02 3.88849437e-01 -3.77258092e-01 1.50492758e-01
4.25741374e-01 4.78488654e-01 -1.27880603e-01 2.77107120e-01
-9.45998788e-01 2.42077813e-01 4.54676837e-01 1.14989603e+00
1.35411918e+00 1.53599363e-02 -3.04795682e-01 6.53104782e-01
-9.94601622e-02 8.30088377e-01 -1.28658935e-01 -1.41521645e+00
3.32675904e-01 1.03206523e-01 3.94804060e-01 -7.22204089e-01
3.13946232e-02 -1.52909294e-01 -6.94772542e-01 5.09994149e-01
2.78963268e-01 -1.06921248e-01 -1.15542829e+00 1.62432873e+00
3.31291020e-01 4.61724013e-01 -2.58207947e-01 8.03464532e-01
5.67129910e-01 9.31457400e-01 -5.13644516e-01 -4.16743793e-02
9.89150763e-01 -1.26029754e+00 -6.63478911e-01 -3.67734760e-01
8.39255154e-02 -8.24581265e-01 1.31946099e+00 4.61234212e-01
-1.12661755e+00 -5.40711462e-01 -1.07064509e+00 -6.04049087e-01
-2.33521760e-01 4.14098464e-02 3.54699105e-01 4.32637393e-01
-1.24346781e+00 3.07475150e-01 -5.61571896e-01 -3.13351959e-01
2.58061349e-01 -5.25067821e-02 -2.36241296e-01 -3.84381503e-01
-5.97955942e-01 6.12844825e-01 6.82726353e-02 2.63543520e-02
-1.30057740e+00 -8.24758947e-01 -1.05420256e+00 -3.56919885e-01
1.75194949e-01 -8.65912378e-01 1.16772807e+00 -9.89474356e-01
-1.75550973e+00 1.13514078e+00 -4.14365232e-01 -1.95322558e-01
5.63093483e-01 -8.13427418e-02 -7.86003694e-02 -3.24505661e-03
4.02217120e-01 6.53402150e-01 1.10556602e+00 -1.89679754e+00
-4.60578799e-01 -8.79589841e-02 1.28689989e-01 1.08441412e-01
3.67464155e-01 -2.16132477e-01 -5.19945741e-01 -6.35771334e-01
-1.61002934e-01 -9.68214810e-01 -2.53058970e-02 9.22884569e-02
-8.84755433e-01 6.34635746e-01 9.01060104e-01 -5.13363898e-01
4.03382480e-01 -1.98936248e+00 2.71298259e-01 -4.63506160e-03
4.69721518e-02 -1.40241683e-01 -3.47866774e-01 5.71113527e-01
-8.24326202e-02 -1.56769201e-01 -6.56453431e-01 -8.19521844e-01
-1.89059168e-01 4.74549770e-01 -6.49253190e-01 4.21876341e-01
-4.56314608e-02 9.24384177e-01 -1.05737257e+00 -2.80514032e-01
6.93452299e-01 6.39677942e-01 -5.23067474e-01 5.33256829e-01
-3.74330193e-01 1.17014110e+00 1.22532405e-01 4.54966754e-01
9.99995649e-01 2.10314989e-02 -4.76631038e-02 -2.41316766e-01
-3.27818632e-01 1.01441488e-01 -7.86248088e-01 2.10818005e+00
-8.95566046e-01 8.09174895e-01 -6.98929578e-02 -6.27941310e-01
7.71862984e-01 -1.84128985e-01 2.23505855e-01 -8.96102250e-01
2.01672949e-02 1.36712277e-02 -8.52355957e-01 -4.12585884e-01
3.64376754e-01 -2.60348290e-01 4.90235351e-02 4.55952764e-01
-6.51247352e-02 -1.00661540e+00 -6.00298494e-03 1.31821901e-01
1.09915161e+00 5.51829398e-01 -2.37445645e-02 1.49121910e-01
2.21377313e-01 -2.84493446e-01 3.58539641e-01 8.77614856e-01
2.38699228e-01 1.24141407e+00 1.21477783e-01 -4.34832036e-01
-1.11218643e+00 -1.42268050e+00 -4.08067293e-02 8.00703466e-01
2.71426022e-01 -2.80006886e-01 -8.32536161e-01 -5.24563909e-01
-2.66307086e-01 1.07360446e+00 -7.98189700e-01 1.90735891e-01
-5.44582546e-01 -4.37658787e-01 8.33196715e-02 -2.10541636e-02
9.82410371e-01 -1.01773202e+00 -7.44367838e-01 -2.83536911e-01
-6.63527489e-01 -1.37283134e+00 -4.99135196e-01 1.57317862e-01
-3.16317677e-01 -9.58954096e-01 -6.22686505e-01 -7.46730030e-01
7.77554333e-01 5.66493154e-01 1.48918426e+00 -3.30784824e-03
-5.34450173e-01 6.15132928e-01 -2.99620628e-01 -3.48977715e-01
-4.87005711e-01 -1.80914029e-01 -4.77433354e-01 3.91280025e-01
-3.53685558e-01 -6.90472662e-01 -8.30836475e-01 4.25136119e-01
-1.13019013e+00 5.86690664e-01 1.54348582e-01 7.41858363e-01
7.64837146e-01 -5.23259155e-02 -1.06477477e-01 -8.31244111e-01
5.00753187e-02 -3.17082144e-02 -7.92815328e-01 1.69850007e-01
-1.35168836e-01 2.55742241e-02 5.67038774e-01 1.31317988e-01
-1.46390212e+00 2.98503429e-01 1.41916096e-01 -2.17203781e-01
-2.59403676e-01 -3.62730086e-01 -2.27042779e-01 -4.39703375e-01
7.60945201e-01 3.40801597e-01 -4.72708136e-01 -1.96884885e-01
6.67367160e-01 3.21239680e-01 9.01415229e-01 -6.83494210e-01
1.32863820e+00 1.06410050e+00 2.05626637e-01 -1.00154567e+00
-1.04391432e+00 -1.37764081e-01 -6.78642511e-01 -2.02076256e-01
1.08925247e+00 -8.76729786e-01 -5.32265306e-01 6.89297616e-01
-1.24620199e+00 -1.16060019e+00 -6.25242531e-01 9.05035157e-03
-9.20791566e-01 9.97703299e-02 -3.89460802e-01 -4.37864214e-01
2.27556869e-01 -1.11066127e+00 1.67893028e+00 1.40168503e-01
2.02764019e-01 -8.87036026e-01 4.06197518e-01 6.09244108e-01
3.66653502e-01 6.36150718e-01 5.88599801e-01 7.60905683e-01
-1.11379457e+00 2.35458553e-01 -2.54539937e-01 4.88755733e-01
5.46948850e-01 -4.96525280e-02 -1.36611331e+00 -2.26761386e-01
8.15456584e-02 -5.33494532e-01 8.84461641e-01 3.80176783e-01
1.32562673e+00 -2.28777453e-01 -2.51271486e-01 1.23530400e+00
1.67895758e+00 1.00206062e-01 1.00552332e+00 5.01991451e-01
1.02339351e+00 5.59279501e-01 3.20535362e-01 2.52742589e-01
3.59416276e-01 8.65761995e-01 6.70213580e-01 -6.40402079e-01
-5.60986519e-01 -5.71816981e-01 3.11324418e-01 1.53651834e-01
-1.89652294e-01 -5.24720550e-01 -4.96034771e-01 4.12946880e-01
-1.51320398e+00 -8.93772006e-01 -7.91254360e-03 2.35474467e+00
8.52434158e-01 -4.58220840e-01 -2.59558052e-01 -6.22124001e-02
5.58192194e-01 4.86768663e-01 -5.26898026e-01 -1.22960873e-01
-3.26524496e-01 3.57170075e-01 7.32906461e-01 1.02848208e+00
-8.53923023e-01 1.01787019e+00 6.90515804e+00 4.24900740e-01
-1.09506810e+00 1.19864643e-01 8.04698288e-01 -8.39734674e-02
-6.44114077e-01 1.72145382e-01 -3.43243361e-01 3.73104006e-01
5.00235856e-01 3.90634060e-01 1.12538326e+00 2.21432269e-01
3.75956327e-01 -4.90415961e-01 -1.29231977e+00 1.19712341e+00
4.44995940e-01 -1.26580560e+00 6.07829317e-02 -1.61333494e-02
1.23030043e+00 4.24908362e-02 2.22001284e-01 -3.09582472e-01
6.27209067e-01 -9.50762451e-01 7.97844648e-01 6.85874820e-01
1.00398290e+00 -2.98119247e-01 -4.70612943e-03 8.01968575e-02
-8.66998792e-01 2.47609094e-01 -2.25643322e-01 3.96772549e-02
4.48977262e-01 7.04860985e-01 -8.18525374e-01 5.31530797e-01
6.96069717e-01 9.24222291e-01 -6.36300802e-01 6.57113492e-01
-8.33930135e-01 4.99893039e-01 -2.68145621e-01 7.80203938e-01
-2.71983534e-01 -7.03233659e-01 4.02281970e-01 9.67815876e-01
5.39874315e-01 -4.77761105e-02 -8.25866684e-02 1.29706669e+00
-3.17310125e-01 -4.42093998e-01 -1.02349603e+00 6.42578542e-01
1.25331506e-01 1.20432830e+00 -5.11868894e-01 -1.90496862e-01
-2.91286290e-01 1.75743401e+00 1.24762371e-01 7.98269808e-01
-8.86072695e-01 -3.67833346e-01 2.84773827e-01 2.96611220e-01
2.13392228e-01 -2.26726592e-01 -8.49483535e-02 -1.39159703e+00
8.36383402e-02 -7.02908218e-01 -7.77473599e-02 -1.79467130e+00
-1.08751011e+00 3.53978246e-01 -1.98804960e-02 -1.08806026e+00
-9.39540043e-02 -3.58600110e-01 -5.94877124e-01 7.73593009e-01
-1.79217947e+00 -1.58485210e+00 -9.68065143e-01 8.33945870e-01
5.04825592e-01 5.21496952e-01 6.17436707e-01 7.94795528e-02
-2.16615841e-01 8.30733255e-02 2.53215849e-01 1.00851562e-02
1.04689074e+00 -1.40784431e+00 5.54317951e-01 1.21353257e+00
2.68333972e-01 2.52248615e-01 8.67169082e-01 -3.85729253e-01
-1.26490211e+00 -1.34178507e+00 5.43949604e-01 -7.25313187e-01
2.26252899e-01 -7.62019157e-01 -3.10322195e-01 1.28698730e+00
7.41818190e-01 2.84193903e-01 2.03261524e-01 -4.86124635e-01
-4.11690563e-01 -4.08647716e-01 -1.24160218e+00 7.15986252e-01
1.39541698e+00 -5.99579215e-01 -1.85491070e-01 5.88177681e-01
7.43654430e-01 -4.94529754e-01 -3.15730035e-01 -3.31846550e-02
3.35283041e-01 -1.53030062e+00 1.25513613e+00 1.02258563e-01
6.15971625e-01 -5.96369505e-01 -3.63534749e-01 -1.72209275e+00
-9.02777240e-02 -8.92534792e-01 4.43125784e-01 1.25775564e+00
5.96595835e-03 -8.95594358e-01 5.32445550e-01 2.67504364e-01
-1.24119870e-01 1.21446950e-02 -6.63725495e-01 -6.83132470e-01
-4.28891331e-02 -2.35487729e-01 7.11362839e-01 7.92889595e-01
-6.67573810e-01 3.70576710e-01 -7.36955822e-01 1.40521601e-01
1.03036332e+00 3.25934410e-01 1.40901399e+00 -6.55431926e-01
-4.61757421e-01 1.64773509e-01 -1.19470745e-01 -1.26975977e+00
7.12877065e-02 -6.49471104e-01 4.79454398e-01 -1.70925331e+00
1.60503060e-01 -4.68576252e-01 2.14477137e-01 5.05718291e-01
1.54530839e-03 8.39754343e-01 6.67316541e-02 1.21508271e-01
-2.92422324e-01 6.26709938e-01 1.38938439e+00 -3.39889944e-01
-8.32870379e-02 -4.94922668e-01 -4.48199689e-01 5.38678586e-01
6.59446657e-01 -3.76902789e-01 -5.68623424e-01 -8.34103286e-01
3.83442670e-01 -1.38784677e-01 8.44391525e-01 -1.10128057e+00
-6.21629283e-02 -2.78614998e-01 5.02877295e-01 -4.05913413e-01
5.53072989e-01 -7.39150167e-01 3.77785414e-01 1.01247154e-01
-2.82372266e-01 -1.79262578e-01 -1.19083412e-01 5.17277062e-01
-3.82761681e-03 1.25642061e-01 9.04912770e-01 -2.16180071e-01
-4.22196209e-01 2.74220973e-01 -2.74322838e-01 1.13318712e-01
1.06968606e+00 -5.03861047e-02 -5.13494253e-01 -5.22377551e-01
-3.45382541e-01 -2.63514042e-01 1.11267102e+00 1.79275542e-01
5.97216129e-01 -1.34499681e+00 -5.22408783e-01 6.07893467e-01
3.26823354e-01 4.48179454e-01 1.59992844e-01 5.29805779e-01
-9.69808042e-01 -1.51672652e-02 -1.26846492e-01 -7.12695897e-01
-1.07802439e+00 5.81253290e-01 5.55946708e-01 2.23255247e-01
-6.85905397e-01 7.77579725e-01 7.77549982e-01 -5.53717434e-01
-2.91054845e-01 -5.66214383e-01 5.03102720e-01 -5.09347916e-01
4.37418044e-01 2.28420898e-01 8.75830054e-02 -5.39149165e-01
-1.20880567e-01 1.04086626e+00 4.51116890e-01 -4.53347325e-01
1.28674245e+00 -4.46197540e-01 -3.64805400e-01 6.06901765e-01
1.19835949e+00 4.28051412e-01 -1.62897837e+00 -2.16536373e-01
-8.51189971e-01 -1.10599875e+00 3.56909558e-02 -9.26180780e-01
-1.12426388e+00 8.36276114e-01 5.72778404e-01 -8.55545774e-02
1.42700708e+00 -7.66505748e-02 7.45961905e-01 2.68097579e-01
6.33091271e-01 -9.01884794e-01 2.28686839e-01 2.35760778e-01
1.03431273e+00 -1.32357860e+00 -9.76172611e-02 -3.88388008e-01
-3.18297535e-01 9.12112474e-01 2.53377140e-01 -1.56586021e-01
3.65879238e-01 3.01212907e-01 3.18168074e-01 -2.19847158e-01
-2.06916243e-01 9.81508940e-03 -6.21780679e-02 8.58005822e-01
1.75228298e-01 -2.49655508e-02 4.25769478e-01 -5.81246138e-01
-3.74799430e-01 1.97816566e-01 7.36884415e-01 9.16856408e-01
-8.12600926e-02 -1.09105277e+00 -6.56462491e-01 -1.48724154e-01
8.58234465e-02 -2.32733950e-01 -4.07209426e-01 4.56552029e-01
4.45136189e-01 1.12546277e+00 1.12920348e-02 -1.62905410e-01
1.89875767e-01 -3.50854397e-01 9.27069902e-01 -6.09667122e-01
-7.14394376e-02 -1.95975661e-01 -1.04419284e-01 -8.86124372e-01
-7.46364474e-01 -5.63961267e-01 -7.36112893e-01 -1.60291314e-01
3.04174162e-02 -1.92671180e-01 7.30966926e-01 7.38824248e-01
2.74368942e-01 5.33113360e-01 8.70297253e-01 -1.42700672e+00
3.16345900e-01 -5.85749984e-01 -6.34021401e-01 5.27659535e-01
9.21340644e-01 -5.64285398e-01 -6.15440071e-01 6.49448276e-01] | [9.83217716217041, -2.894676446914673] |
afb395e4-6749-49a1-8412-90f647a99e9e | knee-osteoarthritis-severity-prediction-using | 2106.14292 | null | https://arxiv.org/abs/2106.14292v1 | https://arxiv.org/pdf/2106.14292v1.pdf | Knee Osteoarthritis Severity Prediction using an Attentive Multi-Scale Deep Convolutional Neural Network | Knee Osteoarthritis (OA) is a destructive joint disease identified by joint stiffness, pain, and functional disability concerning millions of lives across the globe. It is generally assessed by evaluating physical symptoms, medical history, and other joint screening tests like radiographs, Magnetic Resonance Imaging (MRI), and Computed Tomography (CT) scans. Unfortunately, the conventional methods are very subjective, which forms a barrier in detecting the disease progression at an early stage. This paper presents a deep learning-based framework, namely OsteoHRNet, that automatically assesses the Knee OA severity in terms of Kellgren and Lawrence (KL) grade classification from X-rays. As a primary novelty, the proposed approach is built upon one of the most recent deep models, called the High-Resolution Network (HRNet), to capture the multi-scale features of knee X-rays. In addition, we have also incorporated an attention mechanism to filter out the counterproductive features and boost the performance further. Our proposed model has achieved the best multiclass accuracy of 71.74% and MAE of 0.311 on the baseline cohort of the OAI dataset, which is a remarkable gain over the existing best-published works. We have also employed the Gradient-based Class Activation Maps (Grad-CAMs) visualization to justify the proposed network learning. | ['Palash Ghosh', 'Arijit Sur', 'Sibaji Gaj', 'Prasen Kumar Sharma', 'Rohit Kumar Jain'] | 2021-06-27 | null | null | null | null | ['severity-prediction'] | ['computer-vision'] | [-2.53673047e-01 -1.35210723e-01 -4.33970690e-01 -4.14233916e-02
-7.68311262e-01 2.92474717e-01 2.69312024e-01 -9.64785963e-02
-4.86020595e-01 8.64994407e-01 5.03235877e-01 8.13907534e-02
-5.61469138e-01 -7.47440696e-01 -1.52494714e-01 -5.56559443e-01
-6.30412161e-01 5.40964603e-01 3.18262309e-01 -2.27144003e-01
6.84048086e-02 4.09191132e-01 -1.39190531e+00 2.17063040e-01
8.16681683e-01 1.31460822e+00 -6.71688281e-03 2.11366475e-01
3.92506838e-01 7.79305756e-01 -4.98023778e-01 4.30737175e-02
-4.87755612e-02 -1.23047136e-01 -5.44619560e-01 -1.10250629e-01
2.07234800e-01 -6.99419677e-01 -6.86945081e-01 7.76823461e-01
7.51698732e-01 -1.66111320e-01 6.20414317e-01 -7.79527366e-01
-1.00698721e+00 -4.01397282e-03 -5.44161618e-01 4.77777511e-01
-3.20544321e-04 3.09118360e-01 7.38737404e-01 -1.05180669e+00
6.56321108e-01 1.12961817e+00 5.94063163e-01 5.56948602e-01
-6.62006557e-01 -5.77543497e-01 -2.59646028e-01 4.90601450e-01
-1.07668042e+00 2.35498399e-01 5.98411143e-01 -4.55675215e-01
5.19742310e-01 3.37700218e-01 1.20280516e+00 1.17934442e+00
1.03406382e+00 8.02388430e-01 1.55151343e+00 -6.57059252e-02
2.71173269e-01 -6.67051375e-01 2.46194631e-01 6.50548995e-01
2.77462363e-01 2.24589676e-01 -3.03702533e-01 -2.66615540e-01
1.19523728e+00 4.46765363e-01 -3.29884410e-01 -4.49757695e-01
-1.24568450e+00 7.57056952e-01 1.05045331e+00 2.16783285e-01
-5.88561893e-01 2.50884056e-01 4.07579005e-01 1.15030386e-01
3.12670022e-01 4.02959347e-01 -2.35557601e-01 3.48443016e-02
-6.42113686e-01 3.33721191e-01 2.00992659e-01 4.03767359e-03
5.20001538e-02 1.46061063e-01 -5.80516346e-02 1.16737616e+00
5.22509634e-01 2.85132289e-01 1.20268118e+00 -7.97934115e-01
-1.02892574e-02 9.14969921e-01 -2.89550364e-01 -9.00525689e-01
-5.40908277e-01 -9.50810373e-01 -1.15105653e+00 6.17341101e-01
4.27085534e-02 2.36398652e-01 -1.41665721e+00 1.44893372e+00
6.20817281e-02 -1.02527082e-01 -4.24480438e-01 1.55851305e+00
9.26295519e-01 1.04793288e-01 1.78528473e-01 -7.82138929e-02
1.76019013e+00 -1.04855871e+00 -7.38363624e-01 -3.29119772e-01
4.24142867e-01 -5.06809473e-01 1.38395810e+00 6.00712180e-01
-8.36038351e-01 -4.13777232e-01 -1.21895564e+00 -2.39044484e-02
-2.84797758e-01 4.22744274e-01 9.13532495e-01 2.21638113e-01
-8.41716170e-01 5.87222219e-01 -9.82110500e-01 -3.12788874e-01
3.87757152e-01 4.04819101e-01 -5.30764818e-01 -9.60705876e-02
-1.45918775e+00 8.34779024e-01 -1.02482893e-01 5.53901970e-01
-9.34891224e-01 -4.43997025e-01 -3.70228022e-01 -4.11921069e-02
3.84324551e-01 -9.67609346e-01 8.34545135e-01 -5.90304971e-01
-1.33058941e+00 6.63685203e-01 4.59370285e-01 -2.15597481e-01
5.11923730e-01 -1.01884735e+00 -5.95386982e-01 3.14171374e-01
7.74145573e-02 5.14193416e-01 3.78741682e-01 -8.73727977e-01
-5.50312735e-02 -8.46394300e-01 -7.92112276e-02 2.31683418e-01
-4.48025525e-01 -1.45536810e-01 -4.85817164e-01 -1.03191864e+00
3.95390809e-01 -1.03273261e+00 -4.89051968e-01 2.08224371e-01
-4.58093911e-01 -3.11179698e-01 5.87246418e-01 -9.79817510e-01
1.32025945e+00 -2.05503392e+00 1.40761971e-01 2.22801909e-01
5.14070213e-01 1.80908725e-01 1.57219455e-01 1.52995735e-01
-2.72090703e-01 1.25423402e-01 -1.43956244e-01 2.49185070e-01
-3.07587177e-01 1.03869215e-01 2.48791933e-01 3.30304950e-01
1.83801100e-01 9.42721069e-01 -5.76125145e-01 -4.23905104e-01
1.23364374e-01 4.60403293e-01 -3.82374763e-01 9.41270515e-02
1.73080444e-01 3.80648583e-01 -5.35161376e-01 1.01040840e+00
2.54440933e-01 -5.85489213e-01 6.27769977e-02 -3.17148417e-01
2.32408807e-01 5.09741977e-02 -1.05410802e+00 1.73659396e+00
-1.23708643e-01 2.40919262e-01 -5.07807657e-02 -8.00322592e-01
7.71649420e-01 3.41447711e-01 7.54836380e-01 -9.51099157e-01
1.60022601e-01 5.37523508e-01 3.40548038e-01 -6.52783871e-01
-1.97813094e-01 1.58931434e-01 -3.52054685e-02 1.21582925e-01
-2.60670751e-01 4.85191673e-01 -2.30183732e-02 3.52380015e-02
1.30115688e+00 -2.10897923e-02 3.33746821e-01 -2.38326997e-01
4.28326398e-01 -1.46041483e-01 4.86760795e-01 5.99619925e-01
-3.23621929e-01 8.01421523e-01 4.46947306e-01 -7.26852655e-01
-9.31294918e-01 -1.49185860e+00 -2.07908452e-01 5.16252458e-01
4.39367965e-02 -2.66042233e-01 -3.02023947e-01 -4.20770437e-01
2.34737232e-01 -3.18935037e-01 -7.98580766e-01 -4.68841583e-01
-7.24535227e-01 -1.01557982e+00 2.91655898e-01 8.41784418e-01
7.14575469e-01 -9.82993305e-01 -3.96353811e-01 9.45532173e-02
2.48749837e-01 -3.97857815e-01 -3.40702683e-02 5.04444875e-02
-1.45197237e+00 -1.14461768e+00 -1.30041337e+00 -9.39779699e-01
3.90056014e-01 -2.02730000e-01 7.07223237e-01 2.78768241e-01
-7.38069713e-01 6.12560622e-02 5.02980016e-02 1.74663886e-01
1.72055230e-01 1.24811493e-01 1.99662328e-01 -2.06105277e-01
1.12848930e-01 -6.38206065e-01 -1.39841974e+00 2.17164502e-01
-7.78188765e-01 -2.82131936e-02 1.41484189e+00 8.55113506e-01
9.27681386e-01 -2.10904136e-01 7.88832009e-01 -5.23261845e-01
1.03472185e+00 -4.81785029e-01 1.98909596e-01 1.08781435e-01
-9.53412294e-01 -2.55472720e-01 2.32875779e-01 -3.32128674e-01
-7.68603742e-01 -3.51507217e-01 -1.66957006e-01 -2.33734354e-01
-2.80047506e-02 8.28594267e-01 1.65970907e-01 2.56916955e-02
4.86596107e-01 1.89751033e-02 1.90201610e-01 -8.57320428e-01
-1.20796375e-01 7.98421443e-01 7.29527414e-01 -3.67519945e-01
2.48174474e-01 3.14790487e-01 8.20732117e-02 -4.16692555e-01
-7.48347640e-01 -3.73023450e-01 -3.44364613e-01 -4.10110116e-01
8.62410367e-01 -1.12345886e+00 -6.02170408e-01 5.98429799e-01
-5.12286544e-01 -5.75057194e-02 9.33869779e-02 1.06403351e+00
-2.58210719e-01 4.60959941e-01 -1.17752588e+00 -2.68946350e-01
-6.45472169e-01 -1.28331828e+00 1.10125566e+00 1.26586705e-01
-2.43280351e-01 -4.82622951e-01 1.87767014e-01 4.49504733e-01
5.88809609e-01 3.91136378e-01 1.33403695e+00 -3.89811933e-01
-5.15220642e-01 -3.04417759e-01 -3.76533240e-01 5.42661130e-01
1.65662169e-01 -4.17522460e-01 -3.87530029e-01 -1.66143507e-01
1.01877615e-01 -5.22602379e-01 8.76722634e-01 4.60100681e-01
1.41546154e+00 1.20693512e-01 -1.88449651e-01 2.88307071e-01
1.48931706e+00 2.18405977e-01 8.56437683e-01 8.74155819e-01
5.82748532e-01 5.57361916e-03 5.12134671e-01 2.09288895e-01
6.90016672e-02 6.99767828e-01 6.08730614e-01 -3.21822375e-01
-4.79607880e-01 1.31596893e-01 3.60973142e-02 1.19553280e+00
-5.26923895e-01 2.57863402e-01 -1.09506440e+00 5.08242488e-01
-1.73495448e+00 -3.94950271e-01 -2.39645377e-01 1.82107949e+00
7.87190795e-01 4.75722700e-01 -1.55944556e-01 7.76583105e-02
5.70039451e-01 -3.41524296e-02 -5.90089560e-01 7.01103136e-02
7.60834888e-02 6.03370965e-01 2.53974497e-01 -1.23678707e-01
-1.01556432e+00 2.84154058e-01 6.40236187e+00 6.53483331e-01
-1.29142380e+00 1.77411348e-01 4.60049987e-01 -1.89552084e-01
1.01176478e-01 -2.40971759e-01 -6.22819439e-02 5.73041379e-01
6.94195211e-01 2.43076086e-01 6.78953603e-02 9.94608760e-01
2.48931915e-01 -2.76350081e-01 -7.50932336e-01 6.50279462e-01
-1.18421458e-01 -1.20871222e+00 2.88575262e-01 3.16120028e-01
2.81512380e-01 1.32830739e-01 2.38337502e-01 4.24783796e-01
3.20931189e-02 -9.75610375e-01 6.95129260e-02 8.47380221e-01
6.72666490e-01 -5.58898091e-01 1.13548541e+00 -2.64624298e-01
-1.02597058e+00 -2.45373651e-01 4.56086211e-02 -1.94793239e-01
-4.14623693e-02 6.33269787e-01 -3.79071176e-01 4.63576406e-01
1.06759977e+00 5.79563081e-01 -8.37761760e-01 1.45949876e+00
-2.39021361e-01 7.11079478e-01 -1.26320675e-01 1.68186665e-01
3.23172390e-01 -5.79148978e-02 4.07880843e-01 4.95291829e-01
3.78689587e-01 -1.81486279e-01 3.75878721e-01 4.34313476e-01
-1.46420542e-02 1.66870639e-01 -2.31813654e-01 1.43913209e-01
3.19541525e-03 1.04731441e+00 -6.19440913e-01 -3.14160466e-01
-3.80398512e-01 4.78825986e-01 3.92345944e-03 2.92191595e-01
-5.64781904e-01 -2.93219566e-01 3.35035801e-01 2.54308283e-01
-1.51206076e-01 -2.57897228e-01 -3.02892834e-01 -1.08254158e+00
2.43536666e-01 -9.69970822e-01 5.36700189e-01 -1.11037576e+00
-1.39862978e+00 3.65179420e-01 -1.55888394e-01 -1.30330276e+00
-7.38006234e-02 -7.99305558e-01 -3.88296694e-01 6.38372540e-01
-1.18508506e+00 -9.00859654e-01 -5.32843471e-01 4.04715896e-01
5.23103058e-01 -2.14556769e-01 6.79622114e-01 7.40823507e-01
-5.99242151e-01 3.45162824e-02 2.00048745e-01 4.17927146e-01
7.23102152e-01 -1.29096127e+00 -1.84706092e-01 3.45462799e-01
-6.17237508e-01 9.01514232e-01 2.99961090e-01 -9.45445001e-01
-1.18649435e+00 -7.64357030e-01 4.18805808e-01 9.19134915e-02
9.33736205e-01 6.94906861e-02 -9.44224179e-01 4.50673342e-01
-3.31014879e-02 1.92093477e-01 7.42237091e-01 -3.15491343e-03
-5.02568809e-03 -1.30255848e-01 -7.07317770e-01 6.41016066e-01
9.32067513e-01 -2.31067240e-01 -8.85908425e-01 3.41640890e-01
4.38686788e-01 -2.41128832e-01 -1.44224191e+00 1.07855988e+00
9.30081367e-01 -7.76188016e-01 1.19214880e+00 -7.40726948e-01
7.55875766e-01 -1.15247391e-01 -3.72632686e-03 -9.95556891e-01
-4.98690724e-01 2.65879750e-01 -3.34656447e-01 6.75428391e-01
-1.39814749e-01 -4.37423021e-01 8.63311350e-01 9.10461023e-02
-4.70260978e-01 -1.51119483e+00 -1.02693129e+00 -8.10437560e-01
-3.55643667e-02 -5.89569509e-02 3.32061708e-01 6.90336943e-01
-4.29883063e-01 1.74046770e-01 -3.73652041e-01 -1.10374786e-01
5.12430608e-01 -1.00455405e-02 3.66167784e-01 -1.50967169e+00
-1.80087939e-01 -4.29225653e-01 -7.88101852e-01 -5.32064199e-01
-5.67481041e-01 -8.59757721e-01 -3.90043288e-01 -1.92176449e+00
3.86950403e-01 -1.75545290e-01 -9.89683449e-01 1.81161091e-01
-3.53265628e-02 5.73349178e-01 -1.74000353e-01 7.57682800e-01
-4.04554695e-01 6.73461676e-01 1.74360275e+00 -2.34742492e-01
-3.87661420e-02 -1.30225524e-01 -4.34060097e-01 8.16418231e-01
7.60211229e-01 -3.49843085e-01 -2.29443654e-01 -3.54646742e-01
2.02648640e-01 5.63509092e-02 7.91928530e-01 -1.46892846e+00
-9.89787728e-02 6.05378412e-02 8.41083109e-01 -6.66042507e-01
4.48765069e-01 -5.30498326e-01 1.06293961e-01 1.10694790e+00
-3.17102611e-01 1.00582443e-01 -2.32683018e-01 7.22410619e-01
-5.52774310e-01 1.83407396e-01 5.16640842e-01 -4.13085878e-01
-9.16422665e-01 3.81844491e-01 -4.38114911e-01 -2.36600980e-01
7.56029904e-01 -1.94347546e-01 -4.23776001e-01 6.98175356e-02
-1.16088355e+00 1.21629938e-01 1.33729875e-01 5.13284504e-01
8.85496616e-01 -1.83038843e+00 -4.01484162e-01 -5.98777644e-02
1.33592859e-01 -2.36165807e-01 3.01483512e-01 1.29266644e+00
-6.72827542e-01 4.25054520e-01 -6.90949082e-01 -7.52360940e-01
-1.03617871e+00 1.07057668e-01 3.50069165e-01 -8.69532228e-01
-9.75987732e-01 4.19240355e-01 1.37286142e-01 -1.39011681e-01
3.51061165e-01 -2.58971542e-01 -4.14379030e-01 -1.89269602e-01
1.75246909e-01 4.38627362e-01 1.73738658e-01 -1.01705953e-01
-4.49762076e-01 5.72118223e-01 -3.66028249e-01 4.64679077e-02
1.54264879e+00 1.41761974e-01 -3.27464670e-01 4.81343508e-01
1.07540393e+00 -2.42059350e-01 -7.80054390e-01 -9.65946987e-02
-5.74296899e-02 -2.93272614e-01 3.50883901e-01 -1.02861071e+00
-1.31141984e+00 8.20798159e-01 1.45682108e+00 -2.64655519e-02
9.19861257e-01 2.44015470e-01 1.22460532e+00 4.74070132e-01
3.26924950e-01 -1.01777887e+00 6.16670310e-01 -1.94899142e-02
1.13350308e+00 -1.05937302e+00 2.99899101e-01 -1.88261092e-01
-2.43699297e-01 1.21262538e+00 8.96182656e-01 -6.55450940e-01
6.66239381e-01 -1.17496334e-01 2.42246062e-01 -6.95891201e-01
-4.80606556e-01 8.49458948e-02 4.42613155e-01 3.07694674e-01
5.32537818e-01 7.86776990e-02 -9.53421891e-01 7.88308084e-01
1.06087796e-01 5.43409228e-01 1.30582705e-01 1.12645400e+00
-6.23708904e-01 -7.71950305e-01 -3.24389070e-01 1.04368949e+00
-6.81229174e-01 1.45959809e-01 -4.89739478e-01 1.33103645e+00
-1.45005397e-02 2.19558179e-01 -4.22234014e-02 -3.39422226e-01
2.06231982e-01 5.57574676e-04 1.40858486e-01 -4.71687168e-01
-3.99060637e-01 2.14553043e-01 3.29846889e-02 -7.64868855e-01
-5.09858012e-01 -1.69838667e-01 -1.24298835e+00 3.03281099e-01
-1.18053332e-01 -1.48800001e-01 4.24107969e-01 7.85701394e-01
2.66615272e-01 1.04591990e+00 2.46690109e-01 -4.09343272e-01
-5.37273407e-01 -1.10171342e+00 -6.97473586e-01 4.06742126e-01
9.86966193e-02 -1.21785533e+00 -3.51811141e-01 -4.06356126e-01] | [14.626235008239746, -1.8331093788146973] |
d1f46ff6-6720-45db-a35f-7692f6966f82 | capsnet-for-medical-image-segmentation | 2203.08948 | null | https://arxiv.org/abs/2203.08948v1 | https://arxiv.org/pdf/2203.08948v1.pdf | CapsNet for Medical Image Segmentation | Convolutional Neural Networks (CNNs) have been successful in solving tasks in computer vision including medical image segmentation due to their ability to automatically extract features from unstructured data. However, CNNs are sensitive to rotation and affine transformation and their success relies on huge-scale labeled datasets capturing various input variations. This network paradigm has posed challenges at scale because acquiring annotated data for medical segmentation is expensive, and strict privacy regulations. Furthermore, visual representation learning with CNNs has its own flaws, e.g., it is arguable that the pooling layer in traditional CNNs tends to discard positional information and CNNs tend to fail on input images that differ in orientations and sizes. Capsule network (CapsNet) is a recent new architecture that has achieved better robustness in representation learning by replacing pooling layers with dynamic routing and convolutional strides, which has shown potential results on popular tasks such as classification, recognition, segmentation, and natural language processing. Different from CNNs, which result in scalar outputs, CapsNet returns vector outputs, which aim to preserve the part-whole relationships. In this work, we first introduce the limitations of CNNs and fundamentals of CapsNet. We then provide recent developments of CapsNet for the task of medical image segmentation. We finally discuss various effective network architectures to implement a CapsNet for both 2D images and 3D volumetric medical image segmentation. | ['Ngan Le', 'Khoa Luu', 'Hien Nguyen', 'Kyle Quinn', 'Viet-Khoa Vo-Ho', 'Minh Tran'] | 2022-03-16 | null | null | null | null | ['volumetric-medical-image-segmentation'] | ['medical'] | [ 1.65332586e-01 2.65499264e-01 -4.44055974e-01 -5.80310345e-01
-1.37620389e-01 -5.96537590e-01 1.01631917e-01 1.40265509e-01
-5.36941350e-01 4.51950699e-01 4.07719091e-02 -2.50084460e-01
1.29723445e-01 -8.00873518e-01 -5.25938511e-01 -5.38705707e-01
-2.30976149e-01 2.08465621e-01 3.07526350e-01 -1.42913550e-01
-6.59571439e-02 9.46737111e-01 -9.89369810e-01 4.77561414e-01
5.61363101e-01 1.23961234e+00 -7.76588097e-02 3.92139673e-01
-2.76148379e-01 7.54202366e-01 -7.02353060e-01 -1.22352429e-01
3.18285644e-01 -1.72490180e-01 -1.07629120e+00 -5.29490598e-02
3.18891913e-01 -1.60886794e-01 -4.77474391e-01 1.06487405e+00
4.92354333e-01 -1.89401954e-01 5.02418041e-01 -1.23895895e+00
-8.14659595e-01 5.28792322e-01 -4.10710007e-01 1.05815858e-01
-7.03847185e-02 1.13036022e-01 5.97755432e-01 -5.56612790e-01
7.41335690e-01 1.12557435e+00 1.02307165e+00 7.96390176e-01
-1.00162840e+00 -5.24265409e-01 3.40673365e-02 -2.46587455e-01
-1.28373706e+00 -3.32927220e-02 6.28557563e-01 -4.40428197e-01
9.11359012e-01 4.84269202e-01 8.90528977e-01 9.99988496e-01
4.95953113e-01 1.11596584e+00 8.37086439e-01 1.23224698e-01
1.90651238e-01 6.24744641e-03 2.22658202e-01 6.41924858e-01
3.32527906e-01 -1.23557195e-01 -1.63674708e-02 6.26779348e-02
1.07087481e+00 3.28161269e-01 -4.02603716e-01 -5.60030341e-01
-1.17251110e+00 9.49555874e-01 1.17950344e+00 5.68389356e-01
-1.56019300e-01 2.55043238e-01 6.45589948e-01 1.39121443e-01
2.11308271e-01 7.42023170e-01 -6.49317861e-01 4.63420004e-01
-1.05269027e+00 1.12045079e-01 5.38510203e-01 1.03900170e+00
4.26236540e-01 1.72932908e-01 -4.11342919e-01 8.47947121e-01
1.62664836e-03 9.69221666e-02 8.39802504e-01 -6.32143378e-01
1.68350473e-01 9.06842828e-01 -4.94008005e-01 -1.27859771e+00
-1.03566849e+00 -4.72297966e-01 -1.45127153e+00 2.11896926e-01
3.33532691e-01 -9.56721082e-02 -1.61104488e+00 1.40933108e+00
-1.02563515e-01 -1.42919585e-01 -3.02116480e-02 9.13726211e-01
1.39805388e+00 3.20201665e-01 1.04161158e-01 1.30749360e-01
1.30230749e+00 -9.04151022e-01 -7.47558415e-01 -3.87602150e-01
4.91046280e-01 -5.66599011e-01 4.76450264e-01 1.49076849e-01
-1.00090432e+00 -6.24006212e-01 -1.09575009e+00 -2.63509423e-01
-7.84130394e-01 -1.81738921e-02 8.78922999e-01 7.87298620e-01
-1.15025139e+00 6.27868056e-01 -1.03723419e+00 -3.58663708e-01
1.26982379e+00 8.17493916e-01 -5.33112943e-01 -2.52853017e-02
-9.58875954e-01 6.90113306e-01 6.43368244e-01 4.19405043e-01
-6.44306362e-01 -6.82353318e-01 -1.13388228e+00 7.66539127e-02
1.95030898e-01 -5.45387924e-01 9.18794990e-01 -1.15940893e+00
-1.22446036e+00 9.08321619e-01 2.94152945e-01 -6.16289854e-01
5.72512448e-01 3.84675600e-02 -1.54412821e-01 2.48388097e-01
-6.24380149e-02 1.25821722e+00 6.68445885e-01 -9.81338441e-01
-1.59731656e-01 -2.70398825e-01 -1.61634058e-01 -1.38484210e-01
-3.21748883e-01 -1.01051681e-01 -6.34786785e-01 -7.81819642e-01
5.16132593e-01 -8.63643408e-01 -8.67306232e-01 4.13921148e-01
-6.85374439e-01 3.13293538e-03 1.01831102e+00 -3.33581895e-01
7.61215687e-01 -2.14273000e+00 -1.46118611e-01 3.23743731e-01
5.81624985e-01 6.66469336e-01 -1.81193165e-02 -1.54519141e-01
-2.08224013e-01 5.30570447e-01 -4.15439874e-01 -1.17069818e-01
-4.46759909e-01 3.81408930e-01 7.51664415e-02 5.71639299e-01
4.40113932e-01 1.43309021e+00 -7.70940781e-01 -7.86451459e-01
4.40366030e-01 6.01036012e-01 -4.89718258e-01 -2.03176755e-02
-9.68150869e-02 4.81006891e-01 -5.51197827e-01 1.04485321e+00
9.20704365e-01 -4.87635523e-01 1.80664614e-01 -3.74173105e-01
3.65419500e-02 -3.07738036e-01 -8.28041494e-01 1.72482800e+00
4.61446531e-02 7.52631009e-01 1.44635513e-01 -1.29152095e+00
8.41706097e-01 3.90291274e-01 8.11846733e-01 -6.28381491e-01
5.39150119e-01 1.14202783e-01 -2.98253745e-02 -4.78331059e-01
4.71918404e-01 1.02762312e-01 -1.23705298e-01 -9.58367065e-02
2.15678319e-01 -3.50199759e-01 -4.11591083e-02 -8.09405372e-02
9.18997109e-01 -1.76698491e-01 1.79705352e-01 -2.44570076e-01
3.98361892e-01 1.10676266e-01 7.56998718e-01 6.87237620e-01
-5.24896920e-01 1.25006294e+00 7.69546807e-01 -1.01281428e+00
-7.70132422e-01 -8.35444093e-01 -4.76853490e-01 3.81796926e-01
1.67526230e-01 -1.99101374e-01 -6.56782687e-01 -8.12634706e-01
-7.09241927e-02 -1.76536933e-01 -9.02224839e-01 -1.14225753e-01
-8.17598045e-01 -9.50689733e-01 7.42522120e-01 9.09982443e-01
7.29940474e-01 -1.33256698e+00 -8.40561390e-01 1.58629492e-01
1.46848127e-01 -1.16868484e+00 -1.86596543e-01 4.81111199e-01
-1.09827709e+00 -1.36826003e+00 -9.73143160e-01 -1.22740614e+00
1.06439316e+00 -5.49623296e-02 1.06137049e+00 2.65707046e-01
-7.77721167e-01 1.38104130e-02 -1.39599189e-01 -5.14698982e-01
-5.81965968e-03 4.70313221e-01 -4.04198676e-01 -2.67566681e-01
1.37986034e-01 -2.34308064e-01 -8.99577320e-01 2.28176624e-01
-1.18519211e+00 -4.35357504e-02 7.23250866e-01 9.55278695e-01
7.49190390e-01 -2.70382643e-01 4.20454532e-01 -1.10228467e+00
5.89245081e-01 -3.85024637e-01 -3.50257903e-01 2.29925111e-01
-2.73582309e-01 -2.44816318e-01 7.70480931e-01 -2.74680614e-01
-4.55056608e-01 4.04048115e-01 -2.76978672e-01 -5.81550479e-01
-2.76776880e-01 3.57821256e-01 2.70013243e-01 -3.06925684e-01
5.70425510e-01 -5.35978153e-02 3.00093412e-01 -2.69622862e-01
2.34424859e-01 4.46815938e-01 4.71891850e-01 -2.31369615e-01
3.71336132e-01 7.17471719e-01 -3.57188634e-03 -8.07444334e-01
-6.63158178e-01 -3.48345757e-01 -8.81605566e-01 -1.58752073e-02
1.35448098e+00 -6.56926751e-01 -7.26932347e-01 6.46348834e-01
-1.08228302e+00 -1.23130269e-01 -4.02174979e-01 3.50773931e-01
-2.79947072e-01 1.73676878e-01 -8.27875137e-01 -1.41244411e-01
-5.57863891e-01 -1.49400210e+00 7.21655428e-01 6.39161170e-01
-2.04070345e-01 -1.04697013e+00 -3.70040208e-01 1.03597231e-01
7.49093115e-01 7.87356198e-01 8.06665063e-01 -7.70564437e-01
-5.11599600e-01 -3.20197612e-01 -4.41944033e-01 6.08771861e-01
2.24801093e-01 -5.51412441e-02 -9.51509476e-01 -5.32872498e-01
-2.11233258e-01 -4.46177393e-01 9.22860086e-01 8.69905829e-01
1.81606066e+00 -5.82468249e-02 -5.88189363e-01 1.08805835e+00
1.44152868e+00 2.14672431e-01 8.13986063e-01 1.67211175e-01
7.83112288e-01 4.68312263e-01 2.60959379e-02 -7.94492140e-02
-1.24585554e-01 1.78045273e-01 7.93877363e-01 -8.22840333e-01
-1.67497471e-01 2.27437750e-01 -3.13413948e-01 5.73904514e-01
4.04103883e-02 -2.66571939e-02 -8.60065520e-01 7.56339908e-01
-1.68594289e+00 -5.61816633e-01 3.41301933e-02 1.74909568e+00
6.13123775e-01 1.34097755e-01 -2.12069198e-01 -1.89296067e-01
4.96602207e-01 2.24529386e-01 -6.60201490e-01 -4.56628680e-01
-1.65102050e-01 6.87184095e-01 8.36841822e-01 -1.22707002e-01
-1.46685052e+00 9.43612218e-01 6.91640091e+00 4.90936518e-01
-1.63520408e+00 -1.49387717e-01 1.04981494e+00 8.16740096e-02
9.29771513e-02 -5.27057111e-01 -3.16689938e-01 2.26815045e-01
3.80447388e-01 4.22278762e-01 -1.16663478e-01 1.00862026e+00
-2.57691205e-01 1.19803905e-01 -9.82051194e-01 9.89686847e-01
1.03168702e-02 -1.75592351e+00 1.21244103e-01 -1.00688219e-01
8.25598419e-01 3.90078127e-01 2.72038370e-01 1.82199106e-01
1.26211569e-01 -1.73512089e+00 3.88610214e-01 2.53409177e-01
8.30802917e-01 -8.10930431e-01 1.14250529e+00 -7.39566684e-02
-1.08949852e+00 1.31504959e-03 -5.68903387e-01 3.10408831e-01
-3.92478891e-02 2.63851285e-01 -8.93254519e-01 4.77577060e-01
1.00382185e+00 7.87574649e-01 -6.80300295e-01 1.34096050e+00
7.35914037e-02 2.40281776e-01 -1.65240988e-01 -4.05177288e-02
6.19169474e-01 1.74225811e-02 1.30221754e-01 1.39801717e+00
1.63492896e-02 -9.69314128e-02 3.36012036e-01 1.01309788e+00
-2.97419399e-01 8.53544548e-02 -5.82205355e-01 -4.70138490e-02
-1.41136860e-02 1.39367008e+00 -1.40537560e+00 -1.80939421e-01
-3.47332060e-01 7.73119211e-01 1.67060539e-01 3.00951809e-01
-6.67423546e-01 -4.64403331e-01 8.12558174e-01 1.00114970e-02
4.77360100e-01 -1.72792420e-01 -6.46982968e-01 -8.40970099e-01
-1.98482409e-01 -6.53189063e-01 4.49125469e-01 -4.40087676e-01
-1.12085342e+00 9.88169670e-01 -1.62165597e-01 -1.20876479e+00
1.70225680e-01 -1.16979229e+00 -4.93167907e-01 5.63371599e-01
-1.47045207e+00 -1.25447857e+00 -3.31495672e-01 7.63569951e-01
4.73672152e-01 -1.81075353e-02 8.26460719e-01 3.31967533e-01
-5.40849149e-01 5.56242704e-01 -1.74461812e-01 9.81887102e-01
4.80791241e-01 -1.12741494e+00 2.88572103e-01 4.70448375e-01
-1.15930453e-01 7.83929706e-01 1.66688696e-01 -4.46839601e-01
-1.05333662e+00 -1.39769125e+00 4.37498271e-01 -7.99561590e-02
7.60262758e-02 -3.46915513e-01 -8.04896533e-01 8.25493336e-01
3.15951824e-01 8.04438770e-01 6.49053216e-01 -2.24875420e-01
-1.55374452e-01 -2.23252922e-02 -1.52984571e+00 5.09516001e-01
6.63056195e-01 -2.28091463e-01 -3.62313688e-01 4.48268950e-01
6.98864937e-01 -8.85032833e-01 -8.53763819e-01 5.53849578e-01
4.02835280e-01 -8.38470936e-01 1.24565303e+00 -7.61286199e-01
3.35747957e-01 -2.99394101e-01 2.66951740e-01 -1.07567620e+00
-2.55291104e-01 -3.21399778e-01 4.59712207e-01 6.24959886e-01
3.51908922e-01 -5.79376459e-01 1.04482102e+00 5.69929063e-01
-3.92780244e-01 -1.03951013e+00 -8.97917747e-01 -5.63486218e-01
3.93531501e-01 -3.27615649e-01 6.34984851e-01 1.12312961e+00
-2.94703901e-01 -1.00559324e-01 -1.88492581e-01 3.87337022e-02
2.58662909e-01 -7.04639703e-02 2.54426509e-01 -1.13475585e+00
3.58486623e-01 -5.51890731e-01 -8.40655208e-01 -8.30193579e-01
-2.83542108e-02 -1.00819051e+00 -4.63754237e-02 -1.68298769e+00
-1.27141505e-01 -5.50594866e-01 -4.36858237e-01 8.70122850e-01
3.04438919e-01 7.65281379e-01 1.73269227e-01 3.94417532e-02
-5.08357763e-01 8.45518261e-02 1.69657457e+00 -5.42313516e-01
-3.28152388e-01 -5.82921393e-02 -5.76236069e-01 7.37610400e-01
1.03723419e+00 -4.32352662e-01 -2.77048826e-01 -4.63723779e-01
8.51167515e-02 -3.26861084e-01 4.44596916e-01 -1.08288181e+00
3.55543882e-01 3.89210470e-02 9.25234258e-01 -7.22305477e-01
1.25780120e-01 -9.18963313e-01 -1.10365022e-02 7.35600471e-01
-1.87036067e-01 1.93325337e-02 3.37294132e-01 1.76639006e-01
-4.66149956e-01 -1.22412533e-01 9.31771874e-01 -6.89756155e-01
-7.09589362e-01 6.52153850e-01 -3.99057955e-01 6.48594499e-02
1.18203056e+00 -4.07891631e-01 -1.62982166e-01 7.85269439e-02
-9.45546031e-01 3.46321404e-01 1.97868228e-01 5.57649910e-01
7.82242239e-01 -1.22875285e+00 -3.48755598e-01 3.52170765e-01
-4.49534133e-02 6.73544943e-01 2.16152072e-01 8.41920078e-01
-1.13850951e+00 5.10499418e-01 -5.85546017e-01 -8.73552263e-01
-1.07965326e+00 5.58463752e-01 6.32040620e-01 -2.79994905e-01
-7.69094467e-01 9.47150171e-01 1.21658497e-01 -6.23301327e-01
4.15144593e-01 -9.68102753e-01 -4.86255348e-01 4.34354134e-02
4.04483676e-01 -1.07339405e-01 2.88385510e-01 -4.20189917e-01
-4.76878524e-01 4.51271623e-01 -4.07472223e-01 5.66012084e-01
1.49635315e+00 4.43994701e-01 -3.58932257e-01 -6.84997588e-02
1.44144726e+00 -5.07143199e-01 -1.07166600e+00 -4.15601768e-02
-8.26396048e-02 -2.09019527e-01 -1.77329574e-02 -7.36714661e-01
-1.87615812e+00 9.05821860e-01 7.21389472e-01 2.45328292e-01
9.58512664e-01 1.01884276e-01 9.58921790e-01 2.15035468e-01
1.20750383e-01 -1.16512537e+00 2.10670847e-02 4.63769257e-01
8.59621942e-01 -1.27664113e+00 3.46636549e-02 -4.29132491e-01
-4.81359392e-01 1.47463655e+00 8.21166337e-01 -2.42551416e-01
8.75122845e-01 4.07555670e-01 4.00117040e-01 -6.14296079e-01
6.70709368e-03 -7.22570270e-02 3.14291656e-01 7.81892061e-01
5.40095627e-01 1.15882933e-01 -1.36726856e-01 3.91272634e-01
-1.13189146e-01 1.86620615e-02 2.79397100e-01 1.07993388e+00
8.39606486e-03 -9.20890450e-01 -3.62527847e-01 5.54038167e-01
-6.94587111e-01 9.95365009e-02 -4.47915941e-01 1.06011808e+00
4.12135839e-01 4.24189180e-01 3.22657496e-01 -1.45810843e-01
2.88663596e-01 -2.90195882e-01 1.42976061e-01 -5.57069719e-01
-1.09842849e+00 -1.10188566e-01 -4.80969876e-01 -7.94910133e-01
-4.49007481e-01 -3.48896921e-01 -1.49429750e+00 9.33092833e-02
1.70017984e-02 -1.36582315e-01 5.72838724e-01 5.69288194e-01
2.21290112e-01 9.32844698e-01 2.81812906e-01 -7.52797008e-01
-2.03189328e-01 -6.87816501e-01 -4.00434822e-01 3.77273560e-01
5.32664895e-01 -2.82968789e-01 6.98089898e-02 -1.21201314e-01] | [14.69187068939209, -2.627533197402954] |
d443fec6-f341-4f39-a0d5-1b2291b15482 | seq-u-net-a-one-dimensional-causal-u-net-for | 1911.06393 | null | https://arxiv.org/abs/1911.06393v1 | https://arxiv.org/pdf/1911.06393v1.pdf | Seq-U-Net: A One-Dimensional Causal U-Net for Efficient Sequence Modelling | Convolutional neural networks (CNNs) with dilated filters such as the Wavenet or the Temporal Convolutional Network (TCN) have shown good results in a variety of sequence modelling tasks. However, efficiently modelling long-term dependencies in these sequences is still challenging. Although the receptive field of these models grows exponentially with the number of layers, computing the convolutions over very long sequences of features in each layer is time and memory-intensive, prohibiting the use of longer receptive fields in practice. To increase efficiency, we make use of the "slow feature" hypothesis stating that many features of interest are slowly varying over time. For this, we use a U-Net architecture that computes features at multiple time-scales and adapt it to our auto-regressive scenario by making convolutions causal. We apply our model ("Seq-U-Net") to a variety of tasks including language and audio generation. In comparison to TCN and Wavenet, our network consistently saves memory and computation time, with speed-ups for training and inference of over 4x in the audio generation experiment in particular, while achieving a comparable performance in all tasks. | ['Daniel Stoller', 'Simon Dixon', 'Sebastian Ewert', 'Mi Tian'] | 2019-11-14 | null | null | null | null | ['audio-generation', 'music-modeling'] | ['audio', 'music'] | [ 2.02210054e-01 -3.29707444e-01 3.47201079e-01 -3.78169745e-01
-2.80440629e-01 -4.97260839e-01 8.17125142e-01 -1.27341077e-01
-6.31580114e-01 6.64332569e-01 3.86551768e-01 -4.88315016e-01
5.34859076e-02 -7.63679266e-01 -8.34779680e-01 -5.97111642e-01
-4.49388444e-01 -1.26394883e-01 2.78038949e-01 -2.94140071e-01
5.35191270e-03 2.96056688e-01 -1.50551486e+00 6.23624384e-01
1.82527512e-01 1.13506413e+00 3.88323575e-01 1.28074503e+00
1.06548540e-01 1.03166699e+00 -5.27892053e-01 -1.73363343e-01
1.00096717e-01 -3.35380912e-01 -1.06490111e+00 -3.69555026e-01
1.89407289e-01 -3.32812637e-01 -6.02293491e-01 5.05648077e-01
7.76066899e-01 6.93211854e-01 3.47727150e-01 -1.03131831e+00
-5.01558423e-01 7.71234274e-01 -1.10453516e-01 5.78301370e-01
1.22971565e-01 1.50399208e-01 1.13924539e+00 -6.97566509e-01
4.16191995e-01 1.32303834e+00 1.01264799e+00 6.36538684e-01
-1.18018579e+00 -6.28396213e-01 3.09949547e-01 1.27482712e-01
-1.18563640e+00 -6.94649637e-01 2.40679830e-01 -4.62880880e-01
1.79958797e+00 2.94638246e-01 4.86195058e-01 1.30395639e+00
3.35432380e-01 7.34072089e-01 5.52861333e-01 -3.03020239e-01
8.80356953e-02 -4.14093524e-01 -2.61830807e-01 4.72519338e-01
-6.55195236e-01 3.64801139e-01 -7.85873413e-01 -5.50723635e-02
1.02104211e+00 -1.96905192e-02 -1.12949498e-01 5.32256544e-01
-1.36604059e+00 6.98846042e-01 5.52064300e-01 4.27776784e-01
-4.93739486e-01 8.78512681e-01 7.12674916e-01 6.38174117e-01
8.27872574e-01 3.77316296e-01 -8.06728542e-01 -4.64499801e-01
-1.08375275e+00 4.21154737e-01 5.73684096e-01 7.79754996e-01
4.41061080e-01 3.71646613e-01 -1.32285759e-01 1.08633268e+00
-2.25659743e-01 -4.05985080e-02 8.42938900e-01 -8.84435833e-01
2.69694924e-01 -1.11251026e-01 -2.11619705e-01 -6.05719209e-01
-5.82257152e-01 -6.24244213e-01 -1.16580331e+00 8.65495726e-02
3.49985719e-01 -5.12561977e-01 -1.05588675e+00 1.93988192e+00
-1.37143269e-01 7.16252744e-01 -2.74179243e-02 7.62534022e-01
5.41751444e-01 1.16319776e+00 2.24156708e-01 -1.46485139e-02
1.22328949e+00 -9.42028403e-01 -6.25012219e-01 -3.24932307e-01
6.42459393e-01 -9.08530951e-01 7.33789444e-01 3.61249179e-01
-1.09635842e+00 -9.31287408e-01 -8.18352938e-01 -2.35562652e-01
-3.70100945e-01 -1.88775256e-01 9.83215570e-01 1.58969909e-01
-1.39439988e+00 1.15681159e+00 -9.38028514e-01 -3.32777619e-01
1.31529063e-01 5.39359152e-01 -2.60176897e-01 3.53187233e-01
-1.56759667e+00 8.46759021e-01 5.40982842e-01 1.66189775e-01
-7.88338184e-01 -9.73237157e-01 -7.70531595e-01 2.15672791e-01
-1.79470882e-01 -8.37885082e-01 1.56308472e+00 -1.03000772e+00
-1.56060147e+00 1.65205330e-01 -2.29525790e-01 -8.19465637e-01
3.22678983e-01 -4.43928480e-01 -4.25700873e-01 -2.78093845e-01
-4.39741403e-01 9.62129533e-01 8.46442938e-01 -4.05988783e-01
-7.14813232e-01 2.41130829e-01 7.41870180e-02 9.49104056e-02
-3.43955725e-01 2.74747401e-01 -1.25245243e-01 -9.82648373e-01
-1.46143645e-01 -8.94555986e-01 -4.82059360e-01 -2.71462321e-01
-4.08437336e-03 -2.70035297e-01 7.88108170e-01 -7.91291118e-01
1.33418190e+00 -2.24304867e+00 1.39662296e-01 3.22193466e-02
-2.42585670e-02 2.58002967e-01 -3.36156517e-01 5.32336950e-01
-2.48647422e-01 5.26234694e-02 -4.52609137e-02 -3.67451370e-01
-2.06366539e-01 4.97869104e-01 -6.33928180e-01 1.27306178e-01
5.30233383e-01 9.28965032e-01 -9.51685548e-01 -1.04525268e-01
5.54865301e-02 7.53504574e-01 -8.47051799e-01 1.44521251e-01
-4.06290770e-01 4.90677416e-01 -8.17414299e-02 2.54401807e-02
1.60467505e-01 -1.77990124e-01 -2.76837330e-02 1.26022309e-01
-3.39907318e-01 8.27457190e-01 -9.41523194e-01 1.70441484e+00
-9.81211543e-01 1.06370401e+00 -3.53100419e-01 -9.48227704e-01
6.77535534e-01 8.78246427e-01 4.15875345e-01 -5.98029196e-01
-1.16800480e-01 2.32678711e-01 1.89459279e-01 -4.11129564e-01
5.45716286e-01 -4.08382952e-01 6.08604215e-02 5.34492433e-01
4.74329650e-01 9.77016836e-02 2.60893852e-01 -6.80252612e-02
1.41829753e+00 3.63609850e-01 -2.43101772e-02 -1.88036431e-02
1.57564983e-01 -5.17885625e-01 5.37464201e-01 6.88766062e-01
3.15155864e-01 6.24895692e-01 3.33534867e-01 -7.76738763e-01
-1.30436718e+00 -7.54888892e-01 5.21379188e-02 1.54932034e+00
-6.76437378e-01 -6.40373707e-01 -5.49331665e-01 -1.04117371e-01
-1.54327303e-01 5.31143725e-01 -4.51919764e-01 -1.21455356e-01
-7.80969203e-01 -7.02180207e-01 9.25326586e-01 9.30995941e-01
3.56199861e-01 -1.49451137e+00 -7.70477057e-01 9.37243760e-01
-2.98189044e-01 -1.04318595e+00 -4.81032342e-01 5.80985367e-01
-9.19740200e-01 -4.38952684e-01 -7.08404720e-01 -6.88783407e-01
1.93725899e-01 -3.08556467e-01 1.21846032e+00 -1.08435564e-01
-3.14582139e-01 -2.13720817e-02 -3.52368027e-01 -4.22922790e-01
-4.17510062e-01 1.85563847e-01 1.07376680e-01 -1.35903448e-01
2.04298738e-02 -9.79457736e-01 -6.21258259e-01 -1.14770234e-01
-9.69973028e-01 1.46713421e-01 5.24137676e-01 1.05001140e+00
2.53359169e-01 2.18964443e-01 6.26401305e-01 -7.16532767e-01
7.50168741e-01 -5.15607357e-01 -3.28138143e-01 -1.12472273e-01
-2.06298679e-01 2.53732264e-01 8.11715007e-01 -8.31430554e-01
-9.51644480e-01 -4.04586270e-02 -4.70266730e-01 -5.36198258e-01
1.28362000e-01 7.48573899e-01 5.39994657e-01 2.88041472e-01
7.06386566e-01 2.38411397e-01 -1.83924675e-01 -5.37096143e-01
2.32003674e-01 4.57151622e-01 5.47833979e-01 -4.27577436e-01
5.06603837e-01 1.00843906e-01 -6.58019483e-02 -9.28552032e-01
-7.78335810e-01 -3.70705724e-01 -6.06299877e-01 -1.82684094e-01
6.97896063e-01 -9.58881319e-01 -7.40525484e-01 5.73210657e-01
-1.51980960e+00 -7.32517481e-01 -3.02746356e-01 7.31967807e-01
-7.11551309e-01 -1.27207607e-01 -1.08473504e+00 -8.05814028e-01
-3.32709879e-01 -7.55677938e-01 8.97098124e-01 -1.86337411e-01
-6.28267646e-01 -1.31743217e+00 5.99542148e-02 -4.59265739e-01
7.92385221e-01 1.14597648e-01 1.00838196e+00 -3.04980189e-01
-3.95027936e-01 -3.41903903e-02 -4.40431060e-04 6.29252970e-01
-1.75720841e-01 -1.02669567e-01 -1.39151728e+00 -1.99898988e-01
-1.65668771e-01 -3.34215254e-01 1.14500940e+00 4.93320197e-01
1.42277288e+00 -3.74064207e-01 -1.05249375e-01 6.97623968e-01
1.05807209e+00 2.20924482e-01 6.74146652e-01 -5.11796633e-03
5.05535483e-01 4.62026060e-01 1.46551743e-01 6.40699208e-01
1.74171086e-02 5.67464948e-01 1.91607490e-01 -2.15968713e-01
-2.12058246e-01 -1.66658521e-01 6.16145074e-01 1.04280913e+00
-5.66737235e-01 -2.74433047e-01 -7.74059713e-01 7.19489872e-01
-2.02506495e+00 -1.26364815e+00 6.16299845e-02 2.00372243e+00
9.92000163e-01 1.49044126e-01 2.75942590e-03 2.93016851e-01
3.36646646e-01 3.05660367e-01 -2.63787538e-01 -7.47146785e-01
1.87763311e-02 7.49372423e-01 3.06686133e-01 4.65411365e-01
-1.19714975e+00 8.13914359e-01 7.10886097e+00 7.27179706e-01
-1.47251105e+00 1.36365533e-01 6.35646820e-01 -2.99504369e-01
-2.32007466e-02 -3.00364979e-02 -6.50977015e-01 2.95651525e-01
1.56807351e+00 5.53397760e-02 6.83226764e-01 4.39492792e-01
2.72568315e-01 2.06275150e-01 -1.38028753e+00 8.40549231e-01
-3.85030031e-01 -1.40935302e+00 -8.65400732e-02 -1.45926341e-01
6.42591119e-01 3.58066797e-01 1.38068467e-01 4.75011706e-01
5.35737574e-01 -1.41794693e+00 8.69362950e-01 5.13047755e-01
9.58653390e-01 -7.85429060e-01 5.04116178e-01 4.24174547e-01
-1.32754564e+00 -1.48359984e-01 -4.54000264e-01 -5.07858157e-01
1.20557122e-01 6.93539798e-01 -8.84414256e-01 2.44584307e-01
8.34226012e-01 7.77061224e-01 -1.39657691e-01 8.12821746e-01
4.92901029e-03 8.44983518e-01 -3.84714633e-01 -7.17990752e-03
5.95798969e-01 2.44305298e-01 1.38488144e-01 1.74364305e+00
6.18251562e-01 -1.81614086e-02 -1.11909732e-01 6.75588131e-01
-7.19831511e-02 -2.91094750e-01 -4.49321508e-01 -1.77316174e-01
2.05660433e-01 9.80518997e-01 -3.86510968e-01 -3.05103272e-01
-4.30845916e-01 9.83720601e-01 3.15155625e-01 5.36905885e-01
-6.84900045e-01 -5.40565908e-01 9.48684454e-01 7.21980482e-02
5.05422354e-01 -5.08150101e-01 -1.75012350e-02 -9.05200303e-01
-7.53972754e-02 -5.01756907e-01 2.66008794e-01 -7.20570028e-01
-1.10040164e+00 8.56322169e-01 -1.11545131e-01 -1.05383945e+00
-8.83377850e-01 -6.25828326e-01 -5.91580570e-01 1.18539739e+00
-1.52717412e+00 -9.41413105e-01 2.45795518e-01 6.53543115e-01
7.13659763e-01 9.58210155e-02 1.03488564e+00 4.04172510e-01
-5.36790974e-02 3.86917979e-01 -3.43827866e-02 2.03822300e-01
4.82046515e-01 -1.08394110e+00 1.04434931e+00 6.81490004e-01
2.16728941e-01 7.24154890e-01 6.54934704e-01 -3.77167434e-01
-9.68355536e-01 -1.29208267e+00 1.34048510e+00 -1.58367366e-01
8.32399011e-01 -5.91471434e-01 -9.14791107e-01 9.21554387e-01
3.35071117e-01 2.76675254e-01 5.70945263e-01 3.91254216e-01
-6.09131753e-01 1.84083715e-01 -3.79194021e-01 6.24300778e-01
1.27599251e+00 -8.48663330e-01 -1.55500710e-01 3.26708764e-01
7.40330398e-01 -4.56279874e-01 -9.40355659e-01 3.49404514e-01
7.41918743e-01 -7.82552660e-01 9.33429241e-01 -8.17092001e-01
5.28011262e-01 -8.50390643e-02 2.43665408e-02 -1.64475632e+00
-5.81456125e-01 -9.27000880e-01 -2.18946949e-01 9.05335546e-01
5.00669718e-01 -4.70937043e-01 4.10776526e-01 1.39351323e-01
-2.66938716e-01 -7.12523103e-01 -8.99774134e-01 -7.51780152e-01
1.09193534e-01 -8.88820767e-01 4.90892291e-01 8.01227808e-01
-5.04768863e-02 3.77678752e-01 -5.78092754e-01 -5.59372343e-02
-9.82002541e-02 -2.58862019e-01 2.89575845e-01 -1.09506035e+00
-5.97854137e-01 -4.59254384e-01 -3.74636412e-01 -1.18140495e+00
2.11681426e-01 -8.89221251e-01 1.47119224e-01 -1.19909954e+00
-1.73087031e-01 -5.45610905e-01 -5.27420163e-01 6.69946909e-01
-6.77900612e-02 2.72435158e-01 1.03955269e-01 -3.44658317e-03
-1.75179154e-01 3.17532569e-01 9.91635799e-01 3.11816763e-02
-2.27609165e-02 1.10298090e-01 -1.21360444e-01 6.42806590e-01
7.59149909e-01 -2.87386924e-01 -5.09525776e-01 -6.93226218e-01
4.87836182e-01 2.13407323e-01 4.65954453e-01 -1.16220212e+00
2.67859638e-01 -3.07980292e-02 6.45587027e-01 -3.69840890e-01
6.48463309e-01 -4.76185828e-01 3.18280727e-01 3.72861028e-01
-6.99220002e-01 2.21731409e-01 3.52106959e-01 3.29121530e-01
-4.55214232e-01 -6.12891559e-03 5.77342212e-01 -2.28967130e-01
-7.92327881e-01 3.75637978e-01 -6.68543339e-01 -1.75058737e-01
4.13660109e-01 -8.41557235e-03 9.77659225e-02 -6.79220676e-01
-8.15371037e-01 -7.91788250e-02 -2.55653918e-01 5.90841651e-01
6.18878186e-01 -1.31933510e+00 -8.26723158e-01 4.00817007e-01
-2.24332020e-01 9.32964757e-02 2.96305299e-01 6.58240736e-01
-3.37942421e-01 4.92209762e-01 -1.10695973e-01 -4.61397171e-01
-1.10312891e+00 2.47301161e-01 3.50860327e-01 -5.16676188e-01
-6.70551717e-01 1.28062558e+00 3.17594141e-01 -2.90440738e-01
2.27861598e-01 -4.90710825e-01 -1.29286796e-01 -2.23546904e-02
6.65396094e-01 2.89183527e-01 2.48774722e-01 -3.81545752e-01
-9.83998850e-02 2.09965885e-01 -1.40123874e-01 -3.21077347e-01
1.45965278e+00 1.30963653e-01 -2.80174389e-02 6.65369093e-01
1.26445556e+00 -4.57795173e-01 -1.58388650e+00 -3.33674699e-01
3.74383517e-02 3.50786177e-05 1.54128268e-01 -6.11753345e-01
-1.02213407e+00 1.15280414e+00 2.91510940e-01 2.24299744e-01
1.14793432e+00 -2.91339964e-01 9.73663330e-01 4.09637690e-01
8.13146904e-02 -1.06190038e+00 4.99564707e-02 1.24532688e+00
9.30069625e-01 -6.88535690e-01 -3.56858939e-01 -2.06288379e-02
-4.05746728e-01 1.41474485e+00 2.85254896e-01 -2.71332264e-01
7.42538989e-01 7.40303755e-01 -8.51743016e-03 9.75980610e-02
-1.50055563e+00 -3.42062749e-02 2.15849876e-01 4.39986467e-01
9.13800716e-01 1.63481180e-02 1.06528223e-01 1.58322081e-01
-5.71522415e-01 1.07900985e-01 2.26057649e-01 9.44981992e-01
-1.95474491e-01 -1.06745422e+00 -3.93737443e-02 3.35512102e-01
-7.00536311e-01 -4.65298265e-01 1.42755941e-01 5.02396584e-01
1.39545441e-01 9.35798049e-01 5.59988618e-01 -5.84848464e-01
1.72043413e-01 4.73344594e-01 5.23332775e-01 -5.36820471e-01
-1.02699268e+00 2.19740808e-01 2.44520918e-01 -5.96755087e-01
-5.08522809e-01 -5.30225694e-01 -1.17649972e+00 -5.00305533e-01
-1.76904112e-01 -1.64197654e-01 7.42090046e-01 9.34700429e-01
2.09976614e-01 1.13106644e+00 3.92171860e-01 -9.49965239e-01
-5.35724640e-01 -1.26211154e+00 -3.81868422e-01 3.45635176e-01
6.51829422e-01 -2.40379274e-01 -2.85884459e-02 5.09756982e-01] | [10.95485782623291, 6.541244983673096] |
7da4482e-1170-4537-b7e4-4f7c079f8b4c | efficient-unsupervised-sentence-compression-1 | 2205.08221 | null | https://arxiv.org/abs/2205.08221v1 | https://arxiv.org/pdf/2205.08221v1.pdf | Efficient Unsupervised Sentence Compression by Fine-tuning Transformers with Reinforcement Learning | Sentence compression reduces the length of text by removing non-essential content while preserving important facts and grammaticality. Unsupervised objective driven methods for sentence compression can be used to create customized models without the need for ground-truth training data, while allowing flexibility in the objective function(s) that are used for learning and inference. Recent unsupervised sentence compression approaches use custom objectives to guide discrete search; however, guided search is expensive at inference time. In this work, we explore the use of reinforcement learning to train effective sentence compression models that are also fast when generating predictions. In particular, we cast the task as binary sequence labelling and fine-tune a pre-trained transformer using a simple policy gradient approach. Our approach outperforms other unsupervised models while also being more efficient at inference time. | ['Georgiana Ifrim', 'Chris Hokamp', 'Demian Gholipour Ghalandari'] | 2022-05-17 | efficient-unsupervised-sentence-compression | https://aclanthology.org/2022.acl-long.90 | https://aclanthology.org/2022.acl-long.90.pdf | acl-2022-5 | ['sentence-compression', 'unsupervised-abstractive-sentence-compression'] | ['natural-language-processing', 'natural-language-processing'] | [ 6.67314887e-01 5.79833567e-01 -3.68948460e-01 -6.51458025e-01
-9.74022627e-01 -3.58870685e-01 5.64108551e-01 5.52429140e-01
-7.07431197e-01 1.08547771e+00 5.79559028e-01 -5.96133113e-01
1.62793901e-02 -9.46463287e-01 -8.37843478e-01 -2.78736383e-01
3.17051351e-01 8.64821792e-01 -1.85984448e-02 -3.26011151e-01
6.23663962e-01 3.35400961e-02 -1.50750339e+00 2.67405003e-01
1.09796560e+00 4.80673671e-01 6.75109446e-01 1.04610419e+00
-1.72403648e-01 9.68182147e-01 -6.98081851e-01 -6.54550612e-01
-1.31240875e-01 -7.49213755e-01 -1.03893101e+00 8.66234228e-02
1.72015399e-01 -3.28426152e-01 -3.09192687e-02 9.92815495e-01
5.50010324e-01 3.40509087e-01 5.81829906e-01 -4.55442458e-01
-2.46152371e-01 1.01300383e+00 -6.43262789e-02 1.27934322e-01
4.87186164e-01 2.51423895e-01 1.11089921e+00 -3.53533894e-01
7.75417864e-01 1.27628255e+00 4.28132474e-01 7.06199288e-01
-1.45753193e+00 -1.77083001e-01 5.80163226e-02 1.89144894e-01
-9.14660931e-01 -7.07363129e-01 6.80348992e-01 -3.29098776e-02
1.50974226e+00 5.17782867e-01 8.51262808e-01 9.46287870e-01
2.33635202e-01 7.55400598e-01 7.70442188e-01 -9.34622705e-01
6.07602298e-01 1.55965574e-02 -2.93999314e-01 8.07574272e-01
-2.99309511e-02 -1.28613934e-01 -3.66436869e-01 -4.61050533e-02
2.27730632e-01 -3.57067227e-01 -5.41452020e-02 -1.51626125e-01
-6.94879234e-01 1.12408566e+00 -2.54019313e-02 1.48259789e-01
-3.01658660e-01 3.37029487e-01 5.58553815e-01 3.80902350e-01
5.47367930e-01 8.76453221e-01 -4.23655838e-01 -6.41776979e-01
-1.37237751e+00 5.08915246e-01 9.81761038e-01 9.06611860e-01
7.39340007e-01 3.06776119e-03 -4.76611018e-01 9.37116385e-01
2.94863731e-01 1.03714213e-01 7.55548477e-01 -1.22569680e+00
6.74121261e-01 3.38442117e-01 -3.41571420e-02 -7.18330026e-01
-1.49698585e-01 -3.50830615e-01 -4.32628304e-01 -1.08628541e-01
-4.57631843e-03 -1.47026405e-01 -9.89587724e-01 1.67015398e+00
5.29700480e-02 -1.86391726e-01 1.58952460e-01 5.27751923e-01
2.07119435e-01 6.92786932e-01 1.01518348e-01 -3.39730531e-01
1.05620182e+00 -9.66569483e-01 -6.52057528e-01 -4.52312529e-01
9.69342053e-01 -5.57416260e-01 1.19624865e+00 2.85555869e-01
-1.51592684e+00 -2.27100596e-01 -1.13957775e+00 -2.43569508e-01
-1.94391608e-01 -1.78593814e-01 5.23693442e-01 7.60337949e-01
-1.05216587e+00 1.08603191e+00 -8.96061957e-01 -1.32134095e-01
4.76062119e-01 5.57494044e-01 1.27054736e-01 -1.64627418e-01
-1.16451812e+00 1.12675881e+00 8.35623085e-01 -3.98804307e-01
-7.58986235e-01 -4.78482753e-01 -1.01589429e+00 3.82054567e-01
4.72438693e-01 -8.99575233e-01 1.55698562e+00 -8.68979871e-01
-1.93022203e+00 4.69724238e-01 -1.81938171e-01 -9.63272214e-01
4.28465813e-01 -6.88864887e-02 3.15098604e-03 3.26650053e-01
-8.64291564e-02 9.56984580e-01 8.90467167e-01 -7.82684505e-01
-4.47697103e-01 6.23738952e-02 2.36371487e-01 4.05687332e-01
-4.33189899e-01 -8.88345316e-02 -4.55444187e-01 -6.60485327e-01
-4.09943134e-01 -6.23120070e-01 -5.79355359e-01 -4.60613459e-01
-5.10998309e-01 -1.64938644e-01 5.22601545e-01 -8.13521564e-01
1.47701788e+00 -1.61386549e+00 2.57123679e-01 2.14137986e-01
-1.27425641e-01 2.48529151e-01 -1.85116351e-01 6.61434233e-01
4.45026428e-01 2.93784887e-01 -6.58351719e-01 -7.37216949e-01
1.58098772e-01 4.65003908e-01 -2.12809354e-01 -1.35805577e-01
3.42670798e-01 1.01408982e+00 -1.00702345e+00 -8.46406281e-01
2.29973286e-01 2.28134602e-01 -1.16294956e+00 2.49672383e-01
-9.57022190e-01 2.56739706e-01 -2.17377946e-01 2.30850935e-01
1.14043415e-01 -7.72366151e-02 4.23209310e-01 2.45631710e-01
2.04170376e-01 1.03342438e+00 -8.80899131e-01 1.89737177e+00
-7.39591658e-01 4.95762825e-01 -3.21716666e-01 -1.33430219e+00
9.43029165e-01 3.57593745e-02 1.34288788e-01 -6.92318141e-01
6.88157231e-02 6.32633939e-02 1.49517460e-02 -5.18576384e-01
8.24563742e-01 -2.70911783e-01 -7.53300637e-02 7.71047175e-01
1.24327213e-01 -6.69679105e-01 6.98991060e-01 2.58050680e-01
1.01089334e+00 2.64397591e-01 7.84252957e-02 -1.53717071e-01
4.62026089e-01 2.70380965e-03 4.27144438e-01 8.66388917e-01
4.19039875e-01 5.21960914e-01 5.80415010e-01 -1.38364404e-01
-1.53757393e+00 -4.85326856e-01 7.93061927e-02 1.10105193e+00
-3.60706896e-01 -7.87424028e-01 -1.10651755e+00 -7.19147503e-01
-2.45088741e-01 1.43939674e+00 -2.11286187e-01 -4.39389348e-01
-7.65398562e-01 -5.65552056e-01 5.31878769e-01 4.19245362e-01
1.22325279e-01 -1.34175718e+00 -9.01486278e-01 4.61729616e-01
-3.69977742e-01 -9.74806190e-01 -5.12903690e-01 4.34382677e-01
-1.19411564e+00 -6.06614053e-01 -3.52556199e-01 -6.91599131e-01
8.06960285e-01 -5.35325468e-01 1.36824596e+00 3.12872201e-01
-1.94501564e-01 -1.60438195e-02 -3.09356034e-01 -5.04691422e-01
-7.95630872e-01 6.08907878e-01 -3.31468135e-01 -5.27620196e-01
2.72198558e-01 -4.81039524e-01 -3.34837526e-01 -3.38451147e-01
-9.42010283e-01 2.04608694e-01 4.59360033e-01 1.01725340e+00
6.62379622e-01 3.82108577e-02 4.87665892e-01 -1.24125242e+00
1.06217241e+00 -1.26117826e-01 -4.69349623e-01 2.24624142e-01
-1.00475657e+00 7.15416849e-01 9.55823123e-01 -5.65824956e-02
-1.06324685e+00 1.17780745e-01 -4.51083273e-01 -1.29753187e-01
3.47721726e-02 6.68360949e-01 6.85686916e-02 3.76017720e-01
6.58513129e-01 3.52801472e-01 1.01324856e-01 -4.43839967e-01
3.20145637e-01 6.68641269e-01 3.34014058e-01 -6.22758210e-01
4.22255397e-01 -1.04210488e-02 -1.29869282e-01 -6.58351362e-01
-8.39216650e-01 2.16462091e-02 -5.79206824e-01 1.43726796e-01
4.52979803e-01 -5.40500104e-01 -2.72277296e-01 -1.70322031e-01
-8.52214277e-01 -6.33696616e-01 -6.53976798e-01 3.60888809e-01
-1.04846787e+00 5.69939613e-01 -5.79800308e-01 -7.16275096e-01
-7.11728156e-01 -1.07721257e+00 1.07969606e+00 6.39462471e-02
-6.11260414e-01 -1.12639380e+00 6.55682385e-02 4.26891387e-01
4.18541312e-01 -1.92070395e-01 1.15297544e+00 -4.72885817e-01
-4.53427821e-01 -1.80239573e-01 1.77707642e-01 3.47675174e-01
-1.46756724e-01 -3.96521837e-02 -6.59796357e-01 -3.44910204e-01
-5.14190309e-02 -6.73184097e-01 9.48033810e-01 4.19776917e-01
1.56047380e+00 -8.56749713e-01 -8.70565921e-02 6.34603858e-01
1.26407158e+00 -1.21159017e-01 7.74810791e-01 5.20756185e-01
2.61364549e-01 5.97392082e-01 5.90091884e-01 4.61352795e-01
4.56271917e-01 5.75031221e-01 2.24604577e-01 2.66752273e-01
3.43358852e-02 -6.46015108e-01 3.58095735e-01 6.60069346e-01
3.63742709e-02 -3.08445603e-01 -6.16388023e-01 4.55124259e-01
-1.76669061e+00 -1.15359187e+00 5.67521513e-01 2.11740065e+00
1.49645758e+00 5.80713809e-01 1.00003690e-01 3.14044893e-01
4.53653902e-01 6.29229844e-02 -4.49316442e-01 -1.06098807e+00
1.40727907e-01 5.83902955e-01 5.04382551e-01 1.07878292e+00
-7.87576377e-01 1.12600553e+00 6.65280533e+00 8.40468764e-01
-9.37021732e-01 -1.74426049e-01 7.56618321e-01 -3.97926390e-01
-7.26528227e-01 3.58169168e-01 -7.60934472e-01 6.00334823e-01
1.35184157e+00 -6.77005053e-02 6.68259859e-01 6.31694496e-01
3.84155631e-01 -2.50669658e-01 -9.82404768e-01 5.01994014e-01
3.97118367e-02 -1.53409970e+00 6.64246455e-02 -1.96818471e-01
5.40671289e-01 -2.20272943e-01 -3.87354910e-01 3.29754770e-01
4.28953350e-01 -1.10759473e+00 6.26021922e-01 7.06341743e-01
7.75749922e-01 -1.08232141e+00 6.09790504e-01 6.51411474e-01
-3.87663871e-01 -1.25536010e-01 -5.02440751e-01 -2.70332932e-01
2.57838368e-01 5.28181195e-01 -1.17627835e+00 1.08739510e-01
2.61246473e-01 5.23322821e-01 -5.56928933e-01 8.51319730e-01
-4.43366855e-01 7.61905015e-01 -3.92714471e-01 -4.93532330e-01
2.60876894e-01 -1.05945431e-01 4.35700536e-01 1.57748044e+00
1.23829722e-01 -1.97197765e-01 1.85437620e-01 8.19512665e-01
-7.08128884e-02 5.48903197e-02 -3.66917223e-01 -2.81898022e-01
4.16412413e-01 8.13380897e-01 -6.87803626e-01 -3.87162864e-01
7.78875202e-02 1.04747546e+00 5.68663538e-01 3.02418135e-03
-4.85111743e-01 -4.96564656e-01 4.75680567e-02 3.65926959e-02
6.48453474e-01 -1.98744282e-01 -4.67967063e-01 -1.10929465e+00
-4.62994091e-02 -9.79665816e-01 1.71917424e-01 -5.37718177e-01
-6.15822971e-01 3.13384384e-01 1.33828729e-01 -6.81537449e-01
-1.05688298e+00 -1.37418091e-01 -5.47456503e-01 8.25677216e-01
-1.41609013e+00 -6.76578760e-01 3.59738588e-01 2.31540978e-01
7.60581136e-01 -1.30232885e-01 1.01862109e+00 -8.34588110e-02
-4.49690342e-01 8.29437375e-01 1.61371469e-01 -9.95716676e-02
2.84013927e-01 -1.42721033e+00 5.72980523e-01 7.50406384e-01
1.83731273e-01 5.13412118e-01 1.04898095e+00 -7.33586788e-01
-1.24627829e+00 -9.47960258e-01 1.40155506e+00 -6.44472986e-02
2.66797453e-01 -4.78836149e-01 -6.20583415e-01 3.97703588e-01
2.86055028e-01 -6.68857098e-01 5.29976845e-01 1.05625473e-01
2.62258984e-02 6.19037561e-02 -1.21767509e+00 6.55530214e-01
8.17230940e-01 -4.89609599e-01 -5.11369288e-01 4.79185641e-01
7.45868921e-01 -4.55889016e-01 -7.43349433e-01 1.57428607e-01
2.11390585e-01 -7.39459157e-01 6.91539228e-01 -6.47670031e-01
9.35916841e-01 1.14635661e-01 2.09360593e-03 -1.36936593e+00
-2.92335421e-01 -7.24686325e-01 -5.62662721e-01 1.14340007e+00
6.82069361e-01 -2.68987805e-01 1.07009232e+00 6.09986067e-01
-2.01213852e-01 -1.07073092e+00 -7.62819409e-01 -4.03873861e-01
1.90652162e-02 -4.85372633e-01 6.33769453e-01 4.66661960e-01
3.94020170e-01 6.11610830e-01 -3.49669963e-01 -5.14408708e-01
3.98859978e-01 8.17492232e-02 4.94938523e-01 -7.14277744e-01
-8.45833838e-01 -6.16142154e-01 -1.15514718e-01 -1.12286711e+00
1.58269078e-01 -9.88268971e-01 2.16155812e-01 -1.49148405e+00
1.41112909e-01 -3.89856786e-01 1.05118215e-01 5.55530012e-01
-2.32243612e-01 -3.95221151e-02 2.68074840e-01 1.41042629e-02
-6.02730393e-01 7.41515040e-01 1.13321519e+00 -2.75408834e-01
-1.44286573e-01 7.33202770e-02 -8.58685493e-01 4.28650677e-01
1.15671611e+00 -5.48878551e-01 -6.69652104e-01 -5.81246138e-01
3.70162129e-01 1.56370416e-01 -6.70082048e-02 -8.65888834e-01
2.24961653e-01 -2.14945022e-02 2.34739095e-01 -4.11438316e-01
3.06820184e-01 -3.74734014e-01 -3.79913688e-01 5.89934051e-01
-8.80707383e-01 -1.66842137e-02 2.38983594e-02 5.15179336e-01
-1.29912004e-01 -1.02870858e+00 7.16839015e-01 -4.30301607e-01
-1.60194933e-01 -3.23778056e-02 -5.33707917e-01 1.85276285e-01
6.34014428e-01 -1.85330346e-01 1.26609117e-01 -6.50501192e-01
-5.48416555e-01 2.86718786e-01 5.59463620e-01 1.42531067e-01
6.31257415e-01 -7.61074066e-01 -7.20658898e-01 1.97162062e-01
-3.41703862e-01 2.19009221e-01 -1.87675789e-01 2.35102728e-01
-5.80547094e-01 6.26917303e-01 -4.55124713e-02 -3.16328973e-01
-1.06309688e+00 4.19768065e-01 3.37676227e-01 -7.02698469e-01
-7.03534007e-01 6.53203964e-01 -7.30028868e-01 -5.37644327e-01
3.95004511e-01 -1.70227647e-01 -2.20844239e-01 -1.18816353e-01
4.15954798e-01 1.15972780e-01 2.61122316e-01 -9.91624296e-02
6.44485280e-02 8.86045992e-02 -4.39853728e-01 -4.19813722e-01
1.50265551e+00 -1.13555983e-01 -1.13869131e-01 1.40271589e-01
1.07337415e+00 -1.33197978e-01 -1.19773066e+00 -4.80333567e-02
2.83433080e-01 -3.34477037e-01 2.08021328e-01 -9.06028986e-01
-6.10864878e-01 7.02718675e-01 8.00839216e-02 1.14358716e-01
1.14719200e+00 -2.88255662e-01 9.11506891e-01 6.44442379e-01
1.59684435e-01 -1.46162498e+00 -4.78171669e-02 6.70332432e-01
5.72694719e-01 -8.14521492e-01 1.72915876e-01 -1.60062030e-01
-6.18362308e-01 1.29552162e+00 4.03772622e-01 -1.67617217e-01
2.71849900e-01 3.22335809e-01 -3.85680377e-01 5.24065457e-02
-1.03411210e+00 2.14785133e-02 2.42177099e-02 5.27069271e-01
5.77197731e-01 -8.21008980e-02 -8.48010361e-01 1.20757051e-01
-8.44757497e-01 8.12312067e-02 5.56536496e-01 1.11503172e+00
-6.14534199e-01 -1.70294130e+00 -3.68034132e-02 9.26989973e-01
-6.77860260e-01 -3.76473814e-01 -2.14955777e-01 1.55061245e-01
-1.41836479e-01 9.52520132e-01 1.01558067e-01 -2.73321927e-01
-2.62634992e-03 2.30260596e-01 6.56052887e-01 -9.11532521e-01
-4.98119920e-01 5.42159565e-03 5.48393548e-01 -2.84985870e-01
-2.41824090e-01 -8.31551254e-01 -1.20828605e+00 -2.42843926e-01
-4.12184834e-01 3.25991720e-01 6.32236004e-01 1.26856637e+00
4.07010853e-01 4.90722060e-01 5.64858913e-01 -6.96324646e-01
-8.79592180e-01 -9.99496460e-01 2.66664531e-02 3.19627136e-01
1.74956739e-01 -1.43480510e-01 5.55646047e-02 1.38321608e-01] | [12.117677688598633, 9.266711235046387] |
022eda6f-1a50-4f96-b43e-e55e4d95dc85 | x-scitldr-cross-lingual-extreme-summarization | 2205.15051 | null | https://arxiv.org/abs/2205.15051v1 | https://arxiv.org/pdf/2205.15051v1.pdf | X-SCITLDR: Cross-Lingual Extreme Summarization of Scholarly Documents | The number of scientific publications nowadays is rapidly increasing, causing information overload for researchers and making it hard for scholars to keep up to date with current trends and lines of work. Consequently, recent work on applying text mining technologies for scholarly publications has investigated the application of automatic text summarization technologies, including extreme summarization, for this domain. However, previous work has concentrated only on monolingual settings, primarily in English. In this paper, we fill this research gap and present an abstractive cross-lingual summarization dataset for four different languages in the scholarly domain, which enables us to train and evaluate models that process English papers and generate summaries in German, Italian, Chinese and Japanese. We present our new X-SCITLDR dataset for multilingual summarization and thoroughly benchmark different models based on a state-of-the-art multilingual pre-trained model, including a two-stage `summarize and translate' approach and a direct cross-lingual model. We additionally explore the benefits of intermediate-stage training using English monolingual summarization and machine translation as intermediate tasks and analyze performance in zero- and few-shot scenarios. | ['Simone Paolo Ponzetto', 'Kai Eckert', 'Niklas Friedrich', 'Tommaso Green', 'Sotaro Takeshita'] | 2022-05-30 | null | null | null | null | ['extreme-summarization'] | ['natural-language-processing'] | [ 1.44095331e-01 3.64621915e-02 -4.53944981e-01 -4.47786674e-02
-1.51440167e+00 -6.89668477e-01 7.51573384e-01 5.65153301e-01
-4.31383699e-01 1.23664391e+00 7.37618685e-01 -5.05215466e-01
2.64495939e-01 -2.55614221e-01 -5.99189520e-01 -1.31964639e-01
4.45822746e-01 7.96868563e-01 -3.00298259e-02 -3.48537385e-01
7.46513069e-01 2.26480410e-01 -1.17659688e+00 5.57888210e-01
1.63237596e+00 1.00848295e-01 3.31987351e-01 9.25827742e-01
-4.68338519e-01 6.61302149e-01 -1.18891501e+00 -7.45094955e-01
-2.90051699e-01 -8.49823594e-01 -1.11299229e+00 -4.10892954e-03
7.39084423e-01 1.84267730e-01 -1.30183563e-01 9.67285275e-01
1.04132962e+00 -3.99799831e-02 6.92599475e-01 -4.99364287e-01
-8.09221029e-01 1.29311049e+00 -7.16526091e-01 3.50651443e-01
7.79802144e-01 -2.94840306e-01 1.14753878e+00 -5.92194498e-01
1.16807282e+00 1.18782485e+00 5.73267579e-01 4.00038421e-01
-1.10208035e+00 -2.66064465e-01 -5.60434759e-02 3.51248085e-01
-7.74568260e-01 -7.74390578e-01 7.41038203e-01 -1.93870328e-02
1.46979928e+00 4.28043276e-01 5.11483610e-01 1.44597363e+00
8.03853154e-01 1.20126271e+00 9.62360203e-01 -7.35682309e-01
4.87592556e-02 1.57067508e-01 2.82200158e-01 2.40319252e-01
6.01063490e-01 -9.07775640e-01 -6.32108212e-01 1.63949598e-02
-1.20411068e-01 -6.23247087e-01 -2.51705498e-01 3.24799806e-01
-1.45454597e+00 6.27959967e-01 -3.41979116e-01 6.32975996e-01
-4.14196819e-01 -4.00175601e-01 1.03470325e+00 5.13393402e-01
8.56695771e-01 9.32457268e-01 -2.86539882e-01 -4.31480110e-01
-1.66159379e+00 3.52964580e-01 1.38198972e+00 1.38466597e+00
2.78001547e-01 9.30895880e-02 -5.43980181e-01 1.11566734e+00
-1.84316054e-01 5.40215552e-01 8.79519582e-01 -8.01990628e-01
9.68357146e-01 4.30053830e-01 -3.50858361e-01 -5.78045428e-01
-1.71652675e-01 -5.13995826e-01 -8.64259064e-01 -7.28333592e-01
-1.67889789e-01 -3.70465100e-01 -3.78693372e-01 1.12807024e+00
-2.04281762e-01 -3.45119417e-01 6.31615698e-01 3.54394227e-01
1.28235888e+00 1.11294067e+00 -2.61327147e-01 -8.94375980e-01
1.21784830e+00 -1.35709298e+00 -1.05820334e+00 -1.80150896e-01
8.63722444e-01 -1.33736718e+00 8.56557131e-01 4.38635886e-01
-1.53418159e+00 -4.10763830e-01 -1.05058467e+00 -5.92426300e-01
-4.17739213e-01 6.01976216e-01 1.99028775e-01 2.64302492e-01
-1.00470114e+00 7.69797623e-01 -7.96390891e-01 -9.70446050e-01
2.20041767e-01 -2.42062986e-01 -1.92405254e-01 -8.28008056e-02
-1.03450572e+00 1.23587167e+00 4.63407844e-01 -2.49446049e-01
-3.63510251e-01 -7.89037466e-01 -6.69700444e-01 -1.16967559e-02
2.12656617e-01 -8.30658734e-01 1.47290933e+00 -2.47722179e-01
-1.67948544e+00 8.55412483e-01 -4.16018158e-01 -6.25463307e-01
6.40350461e-01 -5.64433992e-01 -2.79737771e-01 9.98726115e-02
5.60331047e-01 3.50391328e-01 1.47041783e-01 -1.07492459e+00
-5.68010509e-01 -3.05188417e-01 -5.68299413e-01 4.35402691e-01
-3.32142204e-01 3.80362302e-01 -5.62616169e-01 -7.67106891e-01
-5.26274502e-01 -6.55149877e-01 -1.20996900e-01 -1.12473845e+00
-1.11065269e+00 -4.57054585e-01 6.89528704e-01 -1.27083468e+00
1.64099574e+00 -1.24589682e+00 5.47956347e-01 -4.83773291e-01
-3.03155065e-01 2.82191902e-01 -2.37223059e-01 1.03588080e+00
2.46222720e-01 2.41091177e-01 -3.07562768e-01 -8.02229226e-01
4.98420894e-02 3.73893753e-02 -5.11710107e-01 1.35420382e-01
1.26085594e-01 1.04062104e+00 -1.06237125e+00 -8.25407743e-01
-7.86373019e-02 1.69986203e-01 -1.94092512e-01 1.30717218e-01
-2.73569316e-01 3.44324946e-01 -5.25160432e-01 5.93172610e-01
2.22259983e-01 3.15535545e-01 1.02632232e-01 -1.29521303e-02
-6.89560533e-01 6.81853652e-01 -5.57696521e-01 2.27941585e+00
-5.53336978e-01 1.00081730e+00 -2.87246674e-01 -1.01200807e+00
8.50402176e-01 4.51917589e-01 2.48608261e-01 -7.46755779e-01
3.18096370e-01 6.19756043e-01 -3.52775514e-01 -5.62978148e-01
1.14895904e+00 2.44028449e-01 -5.07248282e-01 6.72931969e-01
2.67125368e-01 -6.49406612e-01 1.18720734e+00 5.71566045e-01
9.84665811e-01 3.74716371e-01 5.39547443e-01 -4.55193013e-01
6.39254212e-01 4.41600263e-01 1.56721011e-01 8.74290705e-01
3.15192133e-01 6.31765187e-01 5.87596297e-01 1.23753786e-01
-1.28505862e+00 -6.67487621e-01 -1.48895398e-01 9.64113593e-01
-3.70137930e-01 -6.28178000e-01 -1.05459750e+00 -5.40409267e-01
-3.07917088e-01 1.45134592e+00 -4.33006547e-02 -7.13249519e-02
-6.99133456e-01 -9.18351948e-01 8.09479535e-01 6.16952740e-02
5.74440658e-01 -1.14127719e+00 -4.34555322e-01 4.85412776e-01
-5.91643512e-01 -1.22518671e+00 -4.45811301e-01 3.34932134e-02
-1.00886512e+00 -5.35459161e-01 -9.25284445e-01 -9.40372825e-01
1.18134178e-01 1.64665043e-01 1.47566867e+00 -5.95676184e-01
-3.10317874e-01 2.93534577e-01 -5.06222546e-01 -8.22417557e-01
-1.00548482e+00 9.74581003e-01 -1.60765648e-01 -6.38707519e-01
2.02957347e-01 -4.11304921e-01 -3.29399034e-02 -4.69376445e-01
-7.11960316e-01 3.03289145e-01 1.06245220e+00 5.61990023e-01
4.91698891e-01 -4.39170450e-01 1.14227343e+00 -1.14997113e+00
1.43527687e+00 -5.39581954e-01 3.21544670e-02 7.55003512e-01
-5.23010671e-01 2.19292268e-01 7.95194387e-01 -1.15905993e-01
-1.39961958e+00 -6.06690109e-01 -2.32860968e-01 1.01756550e-01
4.83689122e-02 1.11545384e+00 -4.73060608e-02 5.79205871e-01
6.27409697e-01 4.63411719e-01 -2.53894418e-01 -6.37584448e-01
5.94244301e-01 1.02757502e+00 7.12587535e-01 -7.06609607e-01
3.59952360e-01 -2.08216608e-02 -3.04776400e-01 -1.29684877e+00
-1.15445149e+00 -5.14969528e-01 -8.47556114e-01 -1.63869619e-01
7.30182886e-01 -8.49299669e-01 5.84586710e-02 2.70054638e-01
-1.57112992e+00 -3.62103581e-02 -5.80069780e-01 3.77277285e-01
-5.27413189e-01 7.12321937e-01 -8.84142101e-01 -1.90115914e-01
-1.16592872e+00 -1.01005113e+00 1.31786048e+00 3.14595938e-01
-5.89484096e-01 -1.19387400e+00 6.05516195e-01 4.71703053e-01
2.74753273e-01 1.36858776e-01 8.09909999e-01 -9.08025086e-01
-3.72633636e-02 -2.73530155e-01 3.00707947e-02 3.35390717e-01
2.07903460e-02 1.36088207e-01 -5.59288621e-01 -4.25534964e-01
9.39202867e-03 -5.38116038e-01 1.03201616e+00 3.96982461e-01
7.45329082e-01 -4.42094117e-01 -2.63579190e-01 1.96391791e-01
1.05255890e+00 -2.04790235e-01 5.62140346e-01 5.83994508e-01
7.31147587e-01 5.49577713e-01 4.43528682e-01 2.32429028e-01
7.09844708e-01 3.90461266e-01 -5.12073755e-01 4.08851095e-02
-3.45126152e-01 -7.32490495e-02 5.79039812e-01 2.06500864e+00
-3.00592750e-01 -5.71878910e-01 -7.27877975e-01 6.76096618e-01
-1.94165576e+00 -1.12645364e+00 -2.69397974e-01 1.81043732e+00
1.48303652e+00 -1.01241507e-02 -2.70534456e-01 -3.15239608e-01
4.14704621e-01 2.38883466e-01 -3.29733491e-01 -1.07587838e+00
-5.57067633e-01 2.15023711e-01 2.87699103e-01 2.65049607e-01
-8.79876196e-01 1.21262383e+00 5.90938568e+00 1.05128884e+00
-1.06117904e+00 -9.43745486e-03 5.09767532e-01 -2.61229813e-01
-2.66203791e-01 -2.19857916e-02 -9.07499611e-01 2.35921130e-01
1.39862061e+00 -9.66148973e-01 4.14973162e-02 6.54145896e-01
4.51850921e-01 -1.37862161e-01 -1.23769248e+00 6.63319170e-01
7.25129128e-01 -1.48621738e+00 5.88545859e-01 -2.21539095e-01
1.35787177e+00 2.35118300e-01 -3.60511810e-01 6.44830108e-01
2.45289564e-01 -6.65437400e-01 6.40714705e-01 5.66580474e-01
6.53370142e-01 -8.19492519e-01 8.75467360e-01 4.64795440e-01
-6.58985496e-01 5.09059310e-01 -5.43248653e-01 2.93424278e-01
3.65218788e-01 5.37367761e-01 -7.17445731e-01 1.28956687e+00
3.71422321e-01 1.29881537e+00 -7.52625823e-01 8.14178348e-01
-2.92669713e-01 7.34202504e-01 4.90641817e-02 -2.07041502e-01
2.92297184e-01 -3.82881731e-01 1.05546379e+00 2.03062272e+00
6.27071917e-01 -3.65582377e-01 1.20639503e-01 6.44635677e-01
-5.48287749e-01 6.75378919e-01 -5.76370060e-01 -2.61947900e-01
2.64140934e-01 1.27423179e+00 -5.44160306e-01 -6.76600277e-01
-3.41680735e-01 1.16236150e+00 3.07474136e-01 2.82584041e-01
-4.67184007e-01 -7.67862618e-01 -2.11579636e-01 -3.17217797e-01
1.53001379e-02 -1.63981974e-01 -4.39032078e-01 -1.67201614e+00
5.75950667e-02 -1.21999896e+00 2.47452810e-01 -5.08716106e-01
-1.18590641e+00 5.78099430e-01 6.13128766e-02 -1.02278984e+00
-4.54481930e-01 -6.95949197e-02 -1.03247035e+00 7.68950880e-01
-1.62385571e+00 -1.18699861e+00 2.72307545e-01 -7.72286803e-02
1.31533933e+00 -4.46779728e-01 7.56485522e-01 2.13755250e-01
-9.27394271e-01 4.74488974e-01 6.24173939e-01 -2.60317594e-01
1.18570817e+00 -1.33103180e+00 4.77832258e-01 9.90930736e-01
8.83181319e-02 7.10962296e-01 9.66570973e-01 -7.83344746e-01
-1.68662786e+00 -1.24699056e+00 1.66427934e+00 -5.52206814e-01
9.05307412e-01 -7.01573566e-02 -8.98093402e-01 6.97917581e-01
1.18958187e+00 -1.14753699e+00 5.42047381e-01 5.28564192e-02
3.32170248e-01 3.31217237e-02 -6.10511184e-01 9.59478676e-01
7.20961273e-01 -1.97166741e-01 -1.13646054e+00 5.86134851e-01
6.74107313e-01 -4.10509676e-01 -1.18384147e+00 2.21297711e-01
2.46067286e-01 -5.56231081e-01 6.45625591e-01 -5.45634151e-01
1.03908527e+00 1.76483825e-01 4.13078487e-01 -1.83384895e+00
1.03104867e-01 -8.18708420e-01 -1.09387994e-01 1.68140960e+00
6.31580889e-01 -3.20584446e-01 3.17206830e-01 -1.85540706e-01
-8.78264427e-01 -6.65256500e-01 -8.52312267e-01 -4.64332134e-01
6.38134480e-01 -3.74977812e-02 -1.41697880e-02 9.12789941e-01
4.29645360e-01 1.19457638e+00 -1.74187183e-01 -6.08655035e-01
5.89807212e-01 3.93582195e-01 8.31828177e-01 -1.09343660e+00
1.32019088e-01 -9.24604535e-01 1.62980527e-01 -1.02654397e+00
4.70085800e-01 -1.26691282e+00 -7.32036866e-03 -2.26258969e+00
6.29890025e-01 2.99518049e-01 2.81766534e-01 1.37140259e-01
-3.65669757e-01 2.03207694e-03 -2.95186136e-03 4.18146491e-01
-1.04821861e+00 6.79928184e-01 1.23706746e+00 -3.37043494e-01
-3.63931984e-01 -1.67480409e-01 -9.01453435e-01 3.56975704e-01
7.72899270e-01 -2.27172390e-01 -2.00030863e-01 -5.24917543e-01
2.02939272e-01 1.54496044e-01 -4.20942515e-01 -9.24440503e-01
5.39771795e-01 2.16754049e-01 1.43370986e-01 -1.00466096e+00
-2.82456189e-01 6.46791682e-02 -2.90857434e-01 2.06636161e-01
-6.89179957e-01 3.07034314e-01 5.08545339e-01 1.60774037e-01
-4.14610714e-01 -4.68566954e-01 3.98321629e-01 -2.35985979e-01
-2.52736062e-01 -2.71436602e-01 -7.42515445e-01 4.03985500e-01
7.84728587e-01 1.49092093e-01 -5.59222817e-01 -2.36214653e-01
-2.18118519e-01 5.40808558e-01 3.56651396e-01 5.46682119e-01
3.08903694e-01 -8.84686589e-01 -1.46893382e+00 -4.33525354e-01
-1.12965722e-02 -1.46542981e-01 1.37331367e-01 9.45617676e-01
-6.71053708e-01 8.88648033e-01 -1.80530861e-01 -4.33600605e-01
-1.20070517e+00 2.54486322e-01 -2.59531051e-01 -9.25450087e-01
-5.80087662e-01 2.29633376e-01 -5.75378954e-01 -5.09833515e-01
-4.39474359e-03 -3.00812781e-01 -5.42434454e-01 5.37483871e-01
5.08949220e-01 7.95056045e-01 4.43061382e-01 -5.59714079e-01
1.16558485e-01 4.62844759e-01 -5.52027345e-01 -2.07076192e-01
1.39866829e+00 -2.10191220e-01 -6.07569158e-01 8.44917119e-01
1.14671302e+00 3.05863947e-01 -2.73943484e-01 -1.77445874e-01
3.69349718e-01 3.90965283e-01 3.26741748e-02 -8.83909285e-01
-4.28984731e-01 8.29533577e-01 -3.23851198e-01 1.77666724e-01
7.59027660e-01 1.65877566e-02 8.84111702e-01 8.53509188e-01
-2.12599766e-02 -1.59865522e+00 -6.97792247e-02 8.89743984e-01
1.09285605e+00 -1.06753075e+00 5.78323364e-01 8.08320493e-02
-8.38311076e-01 1.33585906e+00 1.19411454e-01 9.45015922e-02
-1.20372005e-01 1.20010205e-01 -4.66822423e-02 -5.59215546e-02
-9.00592864e-01 2.82829672e-01 4.95387793e-01 -3.85438800e-02
9.72885489e-01 -9.59806815e-02 -9.42706287e-01 3.89010847e-01
-7.46285737e-01 -1.43058896e-01 9.17667329e-01 1.12852776e+00
-4.10209030e-01 -1.17143822e+00 -2.07426012e-01 6.12522304e-01
-9.16041791e-01 -3.69712055e-01 -7.96205342e-01 7.40100026e-01
-5.56716383e-01 9.67060685e-01 -2.70351432e-02 1.97880521e-01
5.29883385e-01 1.06024712e-01 5.24468720e-01 -9.35935318e-01
-7.39394665e-01 2.89569110e-01 5.97223818e-01 7.23530650e-02
-4.58440870e-01 -1.14597869e+00 -8.85240436e-01 -4.76541787e-01
-1.46274701e-01 6.01227999e-01 7.95864642e-01 8.87896061e-01
4.74414527e-01 9.25514162e-01 2.56499320e-01 -1.04594064e+00
-6.02016926e-01 -1.54198575e+00 -1.30706966e-01 6.65412545e-02
-1.15139760e-01 2.41175026e-01 -1.03067681e-01 4.09680039e-01] | [12.405743598937988, 9.55396556854248] |
acc16a74-f180-49e4-bb4c-4a7d373aa5be | low-rank-quaternion-matrix-completion-based | 2211.12793 | null | https://arxiv.org/abs/2211.12793v1 | https://arxiv.org/pdf/2211.12793v1.pdf | Low Rank Quaternion Matrix Completion Based on Quaternion QR Decomposition and Sparse Regularizer | Matrix completion is one of the most challenging problems in computer vision. Recently, quaternion representations of color images have achieved competitive performance in many fields. Because it treats the color image as a whole, the coupling information between the three channels of the color image is better utilized. Due to this, low-rank quaternion matrix completion (LRQMC) algorithms have gained considerable attention from researchers. In contrast to the traditional quaternion matrix completion algorithms based on quaternion singular value decomposition (QSVD), we propose a novel method based on quaternion Qatar Riyal decomposition (QQR). In the first part of the paper, a novel method for calculating an approximate QSVD based on iterative QQR is proposed (CQSVD-QQR), whose computational complexity is lower than that of QSVD. The largest $r \ (r>0)$ singular values of a given quaternion matrix can be computed by using CQSVD-QQR. Then, we propose a new quaternion matrix completion method based on CQSVD-QQR which combines low-rank and sparse priors of color images. Experimental results on color images and color medical images demonstrate that our model outperforms those state-of-the-art methods. | ['LiZhi Liu', 'Jifei Miao', 'Kit Ian Kou', 'Liqiao Yang', 'Juan Han'] | 2022-11-23 | null | null | null | null | ['matrix-completion'] | ['methodology'] | [-2.28018016e-01 -5.03633559e-01 2.53894925e-01 1.17653802e-01
-5.63901544e-01 -3.46556120e-02 1.59404978e-01 -8.23983178e-02
-7.35389233e-01 5.40376484e-01 -5.79006299e-02 -1.11934826e-01
3.68216373e-02 -5.20544767e-01 -4.48989600e-01 -7.96446919e-01
-4.30211332e-03 -1.33289158e-01 1.75724164e-01 -6.30156755e-01
1.79213792e-01 2.63813585e-01 -1.02230012e+00 -2.07180277e-01
8.50980997e-01 6.45361841e-01 7.09317699e-02 4.09955382e-01
2.40687966e-01 5.81502497e-01 -3.92264068e-01 -5.20847678e-01
3.35421503e-01 -4.53822047e-01 -4.19037908e-01 4.15912509e-01
7.53092691e-02 -4.20634866e-01 -5.82682312e-01 1.41240180e+00
3.30720186e-01 1.49546891e-01 5.07416785e-01 -1.26092958e+00
-8.49354267e-01 4.90232222e-02 -1.23372936e+00 -3.43117677e-02
4.28656697e-01 -3.48694056e-01 9.67643082e-01 -1.44775045e+00
5.57052433e-01 1.45368183e+00 4.43539381e-01 -7.96619207e-02
-1.21026123e+00 -6.05622947e-01 2.63638627e-02 6.71170592e-01
-1.88229489e+00 2.19117582e-01 7.37445831e-01 -1.39915958e-01
2.76499629e-01 3.36630285e-01 8.90950382e-01 1.34833634e-01
1.90079376e-01 7.43856490e-01 1.32249486e+00 -3.76471758e-01
1.74067363e-01 -2.58831978e-01 -4.20063525e-01 9.88896370e-01
4.79836136e-01 -3.13919604e-01 -4.19120580e-01 -4.57688756e-02
1.08112299e+00 2.51492679e-01 -4.20023799e-01 -4.21827942e-01
-1.87738895e+00 8.19837213e-01 5.29676974e-01 9.12701115e-02
-5.83239317e-01 2.56912764e-02 1.69718470e-02 -3.84545214e-02
2.14228377e-01 -1.04359224e-01 1.89121753e-01 -5.29037183e-03
-7.45195866e-01 -1.12082161e-01 6.06781721e-01 1.00058472e+00
1.01271951e+00 2.38221586e-01 2.39436492e-01 8.19617867e-01
6.66068733e-01 1.09778249e+00 2.67342627e-01 -8.49085867e-01
4.68533516e-01 4.64050621e-01 3.32606494e-01 -1.56056917e+00
-2.64025807e-01 -1.88713074e-01 -1.42795908e+00 2.19272580e-02
1.18238002e-01 -5.87248765e-02 -5.91968894e-01 1.18492270e+00
4.64719057e-01 3.30645204e-01 1.43875018e-01 1.35792959e+00
7.12898135e-01 1.12699425e+00 -2.33115181e-01 -5.16881824e-01
1.38928425e+00 -3.80792856e-01 -9.74220037e-01 2.37010881e-01
2.70332061e-02 -1.06806266e+00 6.40677035e-01 7.98801959e-01
-8.78229976e-01 -4.67795074e-01 -1.34014630e+00 -1.96008151e-03
4.10654657e-02 4.52379823e-01 5.08489370e-01 5.61745822e-01
-7.44075060e-01 2.73942679e-01 -9.40832496e-01 -4.18328531e-02
-2.05865622e-01 1.95002094e-01 -5.79353452e-01 -3.32824379e-01
-1.17471492e+00 8.55365276e-01 2.57631540e-01 6.06079340e-01
-4.16745871e-01 -8.94032419e-02 -8.86036932e-01 -4.14794743e-01
1.93606913e-01 -2.14851588e-01 8.47818077e-01 -6.03930354e-01
-1.59791863e+00 4.52374101e-01 -1.27527729e-01 1.34009257e-01
3.42498243e-01 -2.04003796e-01 -4.14896935e-01 5.83030283e-01
1.41532138e-01 2.40532175e-01 1.23486090e+00 -1.31016612e+00
-4.96238500e-01 -2.53419220e-01 -1.88940376e-01 3.23259443e-01
-1.46644399e-01 -1.68816194e-01 -1.03220057e+00 -7.80601799e-01
1.08108997e+00 -1.21629965e+00 -4.72839713e-01 -1.78709120e-01
-3.43071789e-01 -1.46104470e-01 2.32054278e-01 -8.91617537e-01
1.32239306e+00 -2.28209305e+00 6.14035785e-01 4.58861142e-01
1.11301072e-01 3.85206372e-01 -1.20077699e-01 7.98879862e-01
-2.44931534e-01 -2.34693617e-01 -2.25213155e-01 -5.23277149e-02
-5.48280440e-02 3.40534955e-01 -2.00852200e-01 1.11139596e+00
3.20854336e-01 3.04958940e-01 -1.02739096e+00 -7.61592329e-01
2.77948886e-01 6.78126335e-01 -4.32474017e-01 3.67262214e-02
5.20571530e-01 2.21383870e-01 -4.96952146e-01 6.31696820e-01
1.31586456e+00 -1.37444541e-01 3.36927265e-01 -8.76392365e-01
-4.32236075e-01 -4.82183874e-01 -1.93893802e+00 1.52171421e+00
-1.53336346e-01 1.98682696e-01 2.06836507e-01 -8.25233400e-01
8.29899013e-01 2.00135827e-01 6.64192379e-01 -7.53956974e-01
-1.75269619e-02 2.38457277e-01 -2.09692851e-01 -2.43333727e-01
7.95976102e-01 -4.04996067e-01 8.45012441e-02 1.59176826e-01
-2.74389356e-01 -2.34004781e-01 7.21103370e-01 5.19182861e-01
6.41307890e-01 2.54639834e-01 5.07449687e-01 2.39774734e-02
9.92183983e-01 -2.03698605e-01 1.05022562e+00 5.85859977e-02
-2.55081058e-01 8.47563863e-01 4.40912157e-01 -1.78739727e-02
-8.76228034e-01 -1.17553604e+00 -9.70129967e-02 4.94409084e-01
7.23170042e-01 -6.67559385e-01 -4.66059983e-01 1.91562474e-02
-1.89187229e-01 9.80836377e-02 -3.29937041e-01 1.13187224e-01
-6.32451177e-01 -9.51984763e-01 4.84924763e-02 2.52144396e-01
9.00396168e-01 -2.78676450e-01 -2.67160177e-01 3.29748988e-01
-4.16090459e-01 -1.05612683e+00 -4.48187321e-01 -5.07181227e-01
-8.35518837e-01 -1.07282269e+00 -1.17986631e+00 -6.54842615e-01
1.10000682e+00 1.02418613e+00 6.75931156e-01 1.75525978e-01
-2.42389515e-01 7.15574682e-01 -7.92652786e-01 -1.54689148e-01
-6.65620416e-02 -6.41393125e-01 2.61132538e-01 6.09094620e-01
4.91559170e-02 -1.37336671e-01 -9.82860506e-01 4.15313184e-01
-1.31565118e+00 1.13586456e-01 1.06691504e+00 8.50357831e-01
9.75882411e-01 2.82293675e-03 2.42575899e-01 -5.61229765e-01
5.93781590e-01 -1.75733835e-01 -8.17613006e-01 2.90978789e-01
-3.33473235e-01 3.12788449e-02 6.26115918e-01 -1.65564284e-01
-9.54010725e-01 2.50059634e-01 -9.20537487e-02 -4.35978323e-01
5.95644176e-01 8.36741149e-01 4.17747311e-02 -1.20289072e-01
2.51904190e-01 4.40478206e-01 2.33468294e-01 -4.21814889e-01
7.94463158e-01 4.53036129e-01 3.65379512e-01 -1.93182841e-01
1.25734115e+00 8.10550749e-01 4.17244673e-01 -1.15132952e+00
-4.26604658e-01 -8.05966198e-01 -6.37341380e-01 -2.10959688e-01
8.36356997e-01 -1.26164293e+00 -1.11795521e+00 4.69492614e-01
-9.47724342e-01 6.82025433e-01 1.68636322e-01 1.22118199e+00
-3.41571003e-01 1.19588161e+00 -5.63424349e-01 -7.04422534e-01
-6.87553212e-02 -1.14546263e+00 8.05146635e-01 2.77433068e-01
4.43017244e-01 -6.71860397e-01 1.82447001e-01 8.71026441e-02
-7.43336231e-02 2.04956323e-01 4.57224429e-01 2.88202465e-01
-7.53801763e-01 -2.37731576e-01 -4.85694051e-01 6.13175929e-01
1.52262509e-01 1.03870057e-01 -9.08833072e-02 -6.05678737e-01
-3.59989330e-02 -1.41844794e-01 6.06257141e-01 4.03758287e-02
4.76137102e-01 -1.56187505e-01 1.00881130e-01 5.50172985e-01
1.73963785e+00 4.48186100e-02 8.66118789e-01 5.02297461e-01
7.65985012e-01 4.19565663e-02 1.18884790e+00 9.40680802e-01
6.37098193e-01 4.50374335e-01 4.60527956e-01 -3.65489632e-01
2.33962476e-01 -2.80265566e-02 3.94525111e-01 1.58751035e+00
-5.16919613e-01 5.18952727e-01 -5.78224659e-01 2.21059635e-01
-1.89013255e+00 -7.18295157e-01 -4.77125436e-01 2.46426463e+00
7.37940252e-01 -3.09347332e-01 2.56171115e-02 3.22931290e-01
7.46622980e-01 1.75020188e-01 -3.79837215e-01 -1.72556326e-01
-9.46732461e-02 9.14299712e-02 5.97705007e-01 2.38363311e-01
-1.06701875e+00 5.72951853e-01 5.41955614e+00 6.94900692e-01
-9.26586509e-01 -1.90170467e-01 2.83335540e-02 5.84625959e-01
-2.34807685e-01 1.49684221e-01 -4.21685517e-01 1.41052589e-01
2.53433645e-01 -7.01165572e-02 5.78562319e-01 4.77411211e-01
1.95412323e-01 -3.93741816e-01 -5.45191467e-01 1.60806310e+00
3.19580704e-01 -8.69530439e-01 2.61642188e-01 -1.27146915e-01
7.36152172e-01 -3.23264122e-01 2.05784068e-01 1.50622323e-01
-4.68403585e-02 -4.80606616e-01 5.53429067e-01 5.85304797e-01
7.63247550e-01 -8.52418482e-01 6.23131871e-01 5.31170219e-02
-1.53527665e+00 3.30615431e-01 -8.95354271e-01 2.33889326e-01
-2.42478549e-02 5.97019315e-01 -6.51606560e-01 1.09474468e+00
5.12368023e-01 1.03400207e+00 -6.07318342e-01 1.08102214e+00
-5.01424074e-01 4.92406845e-01 -2.35385150e-01 -5.54110706e-02
1.95087180e-01 -1.18888581e+00 4.83626515e-01 9.78133738e-01
3.74166220e-01 6.52686775e-01 1.51999891e-01 4.71240819e-01
2.07054168e-01 5.78666210e-01 -5.76494336e-02 -1.47196889e-01
1.80605307e-01 1.50965106e+00 -8.00357163e-01 -1.74795896e-01
-7.15874553e-01 1.13115799e+00 -9.98239666e-02 5.32108486e-01
-7.33127415e-01 -7.34006047e-01 3.84282380e-01 -4.32895094e-01
4.98652637e-01 -8.32541108e-01 3.09617788e-01 -1.58327556e+00
-6.81622773e-02 -1.04423273e+00 3.01729113e-01 -7.48650074e-01
-1.06824422e+00 2.30249465e-01 -1.28543898e-01 -2.16578913e+00
2.94525586e-02 -6.60852551e-01 -1.38088346e-01 9.65369105e-01
-1.71008492e+00 -1.06007195e+00 -2.73454875e-01 9.32904363e-01
-1.74312517e-02 2.07661223e-02 6.75106943e-01 2.62647808e-01
-6.83315217e-01 1.16711490e-01 6.27532125e-01 2.44200960e-01
7.39562452e-01 -1.38831592e+00 -5.19446321e-02 1.15481210e+00
2.07821578e-01 9.58384633e-01 6.52772188e-01 -5.85543811e-01
-2.19164252e+00 -7.85569906e-01 4.20426220e-01 2.99614280e-01
5.59690177e-01 2.93550819e-01 -6.17603719e-01 4.73124355e-01
2.60792345e-01 -5.43115586e-02 8.70222092e-01 -3.09487969e-01
-3.41296315e-01 -5.32617569e-01 -6.30939364e-01 7.12937236e-01
3.38171631e-01 -4.66949821e-01 -4.96763676e-01 3.39394987e-01
1.52460501e-01 -3.31098914e-01 -1.24528587e+00 1.89142138e-01
3.75366807e-01 -1.04214442e+00 1.14124978e+00 -7.84065947e-02
-9.36039072e-03 -1.17530847e+00 -3.39078993e-01 -1.32439148e+00
-4.31390524e-01 -6.67951822e-01 2.99292266e-01 8.61631215e-01
-2.31740326e-01 -4.88757133e-01 3.65176767e-01 2.13734627e-01
1.58335865e-01 -4.60284829e-01 -1.05693471e+00 -5.55013120e-01
-3.16613406e-01 -1.20938249e-01 1.47122949e-01 6.43646479e-01
9.21995938e-02 2.66045660e-01 -5.86940467e-01 5.11765778e-01
9.99429762e-01 3.49144131e-01 7.90127993e-01 -7.93019593e-01
-8.40705186e-02 1.43384069e-01 -5.73802233e-01 -1.02088594e+00
-3.28731596e-01 -5.63231349e-01 1.00701898e-01 -1.82170129e+00
1.62429631e-01 -4.66599651e-02 -2.34958604e-01 2.19935738e-02
-4.93251652e-01 6.56908751e-01 6.09084785e-01 4.52751398e-01
-9.30801809e-01 9.09383118e-01 1.61100340e+00 -1.32951021e-01
1.75877646e-01 -2.58511990e-01 -3.72797579e-01 8.16763818e-01
2.78662205e-01 -9.35752392e-02 -3.11473638e-01 -1.66263029e-01
6.53317630e-01 2.73568094e-01 8.74917060e-02 -8.93973827e-01
1.32837728e-01 -1.65552273e-01 4.92534012e-01 -1.00576746e+00
5.77402890e-01 -6.95602417e-01 7.04562888e-02 6.96572185e-01
3.99881542e-01 2.73573607e-01 -2.32096940e-01 8.16011012e-01
-6.42609239e-01 -3.87422405e-02 7.66573131e-01 -2.49612346e-01
-9.58720446e-01 3.46256375e-01 -5.86281717e-01 -1.67211205e-01
9.11866963e-01 -1.12076320e-01 -1.09997205e-01 -4.67350900e-01
-6.69995129e-01 1.63576081e-01 1.82880357e-01 4.28949408e-02
1.30567741e+00 -1.61676610e+00 -7.87425637e-01 8.90869126e-02
7.56950155e-02 -1.14204764e-01 3.52513224e-01 9.57234919e-01
-1.09672976e+00 3.39498460e-01 -3.25560838e-01 -6.33047521e-01
-1.09947598e+00 4.11277771e-01 -1.36437342e-01 -4.47121337e-02
-3.30956191e-01 6.43776357e-01 5.48390895e-02 -7.18214959e-02
-1.30393282e-01 -1.56959295e-01 -4.55419213e-01 2.02457860e-01
5.33409059e-01 6.07462168e-01 1.35425082e-03 -1.25981569e+00
-4.25335050e-01 8.64754617e-01 -2.22186521e-02 -4.05135512e-01
1.29845870e+00 -3.96534771e-01 -6.73562884e-01 3.03387254e-01
1.29997766e+00 1.53819919e-01 -1.02564859e+00 -2.74990737e-01
-3.87396872e-01 -7.39923954e-01 5.04772812e-02 -1.01113789e-01
-9.66310263e-01 7.65190423e-01 6.47307098e-01 2.76445597e-02
1.41578531e+00 -6.87073588e-01 7.37643301e-01 5.83123386e-01
6.24030352e-01 -1.16965830e+00 3.91338557e-01 3.04337174e-01
1.08565605e+00 -1.23466063e+00 8.03400755e-01 -6.28792226e-01
-8.44674945e-01 1.34662545e+00 2.94339538e-01 -4.10040706e-01
7.47542799e-01 -7.10396588e-01 1.35698780e-01 2.38586068e-01
-1.38756320e-01 -3.72553796e-01 4.09859568e-01 3.12157780e-01
4.06518638e-01 1.16221376e-01 -9.34184313e-01 1.54788256e-01
1.64633617e-01 -1.19268276e-01 8.38615239e-01 1.13802505e+00
-2.97586977e-01 -1.25171506e+00 -1.02456093e+00 -9.15801898e-02
-3.05758893e-01 -4.43861783e-02 9.16776881e-02 6.16730869e-01
-4.95547839e-02 1.09314251e+00 -4.00403351e-01 -4.05149400e-01
5.53683102e-01 -7.06357300e-01 5.71112096e-01 -3.65755171e-01
1.24858342e-01 4.64472115e-01 -3.34068686e-01 -5.78024685e-01
-8.70545506e-01 -7.30330348e-01 -1.62901473e+00 -6.88267648e-02
-2.28634611e-01 4.38350528e-01 8.43763232e-01 5.00424027e-01
-5.89147955e-02 8.14507753e-02 1.03530276e+00 -8.39535892e-01
-7.51694500e-01 -6.36008799e-01 -1.27423215e+00 7.40201652e-01
1.68677434e-01 -6.89023674e-01 -1.31446078e-01 2.02953443e-01] | [10.823116302490234, -1.6898629665374756] |
8e1bedbb-eec4-4ff6-a09b-ed6cf88f4cca | neural-laplace-control-for-continuous-time | 2302.12604 | null | https://arxiv.org/abs/2302.12604v2 | https://arxiv.org/pdf/2302.12604v2.pdf | Neural Laplace Control for Continuous-time Delayed Systems | Many real-world offline reinforcement learning (RL) problems involve continuous-time environments with delays. Such environments are characterized by two distinctive features: firstly, the state x(t) is observed at irregular time intervals, and secondly, the current action a(t) only affects the future state x(t + g) with an unknown delay g > 0. A prime example of such an environment is satellite control where the communication link between earth and a satellite causes irregular observations and delays. Existing offline RL algorithms have achieved success in environments with irregularly observed states in time or known delays. However, environments involving both irregular observations in time and unknown delays remains an open and challenging problem. To this end, we propose Neural Laplace Control, a continuous-time model-based offline RL method that combines a Neural Laplace dynamics model with a model predictive control (MPC) planner--and is able to learn from an offline dataset sampled with irregular time intervals from an environment that has a inherent unknown constant delay. We show experimentally on continuous-time delayed environments it is able to achieve near expert policy performance. | ['Mihaela van der Schaar', 'Hao Sun', 'Zhaozhi Qian', 'Alihan Hüyük', 'Samuel Holt'] | 2023-02-24 | null | null | null | null | ['offline-rl'] | ['playing-games'] | [ 1.51218340e-01 2.41922781e-01 -2.24779144e-01 2.14972079e-01
-3.83727401e-01 -7.10290253e-01 7.64873207e-01 5.51131189e-01
-5.75306773e-01 1.21774805e+00 -4.04034436e-01 -4.78113353e-01
-5.29696107e-01 -7.54513502e-01 -9.15451348e-01 -8.78749728e-01
-9.79902387e-01 7.94967651e-01 8.58249515e-02 -3.15425247e-01
-6.46390468e-02 3.01821470e-01 -1.24349475e+00 -6.25761926e-01
8.72825623e-01 9.17772174e-01 2.89090931e-01 8.61040831e-01
3.41305971e-01 7.99504220e-01 -5.24060905e-01 6.59407735e-01
5.03772378e-01 -7.60055184e-02 -4.19164419e-01 1.64838135e-01
-4.98485506e-01 -3.86446625e-01 -5.28420925e-01 8.70473504e-01
2.42445096e-01 6.09460175e-01 4.27727014e-01 -1.23984563e+00
-3.63984965e-02 2.99552321e-01 -2.65120953e-01 1.67251900e-01
1.77613661e-01 5.52448988e-01 4.02107298e-01 3.09248567e-01
3.72592479e-01 1.11147594e+00 5.03495574e-01 4.69935149e-01
-1.47071648e+00 -1.74631223e-01 6.84078276e-01 1.55846938e-01
-1.03118634e+00 7.32533336e-02 4.93185222e-01 -3.01613927e-01
9.08359468e-01 -8.87210146e-02 8.59976590e-01 9.71038401e-01
6.50762022e-01 6.18720114e-01 1.29504454e+00 -8.43838602e-02
9.38775063e-01 -2.64505327e-01 -5.60480654e-01 3.63712043e-01
-1.06451333e-01 7.62693346e-01 9.66949984e-02 -4.62516770e-02
8.06478024e-01 -1.86626043e-03 -2.65207052e-01 -2.23644748e-02
-1.33360183e+00 6.53887868e-01 2.77590394e-01 -1.60820372e-02
-8.70153606e-01 2.68804818e-01 4.00410682e-01 1.06233120e+00
3.18087757e-01 5.96443594e-01 -7.01675713e-01 -2.22058728e-01
-4.26099598e-01 5.02025723e-01 1.08566308e+00 9.56398964e-01
5.34773469e-01 7.02831805e-01 -5.70399277e-02 8.11262876e-02
-5.24028875e-02 8.21358442e-01 3.26344222e-01 -1.11913037e+00
4.36692417e-01 -1.81177661e-01 1.00085390e+00 -7.76876152e-01
-6.49206519e-01 -5.64798236e-01 -7.88965106e-01 5.00765085e-01
6.33724034e-01 -7.42320836e-01 -6.60171866e-01 1.84846270e+00
4.88769174e-01 5.81102014e-01 3.13805610e-01 9.10788357e-01
-5.35843492e-01 9.54093933e-01 -2.88317621e-01 -8.01021874e-01
8.27500522e-01 -5.30787826e-01 -9.03191209e-01 -2.73914397e-01
2.54081547e-01 -2.64259160e-01 6.73394024e-01 5.92541873e-01
-1.05291283e+00 -2.32002497e-01 -8.01488161e-01 8.17016721e-01
-2.84966379e-01 -8.18696767e-02 1.13522261e-01 -7.65603855e-02
-9.85891521e-01 8.19056511e-01 -1.19710267e+00 -2.00245425e-01
-4.60391849e-01 5.84855914e-01 2.39567924e-02 2.33801886e-01
-1.44517875e+00 9.95156884e-01 4.60444540e-01 2.89624780e-01
-1.53118205e+00 -5.91090381e-01 -5.60011804e-01 -2.04125553e-01
1.23484862e+00 -3.80811661e-01 1.80590343e+00 -1.07063556e+00
-1.88104272e+00 -7.41818249e-02 1.50730059e-01 -8.47069085e-01
7.76662886e-01 3.69153246e-02 -4.85473484e-01 1.90295309e-01
-1.38536409e-01 -1.94784440e-02 1.08488405e+00 -1.06934035e+00
-8.20816040e-01 -2.86141902e-01 2.82445073e-01 4.62645382e-01
2.00407490e-01 -6.17747843e-01 2.69746423e-01 -4.43915486e-01
-2.01104283e-01 -1.35530865e+00 -6.33744597e-01 -7.97943100e-02
-2.59207990e-02 2.31743548e-02 9.47949469e-01 -4.42461103e-01
8.29583883e-01 -1.78770053e+00 2.89424419e-01 2.01718464e-01
-2.12051600e-01 1.82345897e-01 -1.30631953e-01 9.35682237e-01
2.19681710e-01 -2.47648939e-01 -1.53397426e-01 -7.20627978e-02
-5.02652153e-02 7.26681769e-01 -6.31677508e-01 6.25428498e-01
1.43781096e-01 2.01405123e-01 -1.30375862e+00 1.39511362e-01
2.50097543e-01 8.51578116e-02 -2.05159232e-01 3.17757726e-01
-1.09674311e+00 1.01008534e+00 -8.95261168e-01 2.72586435e-01
2.75472045e-01 5.61528280e-02 4.30361241e-01 6.27877295e-01
-7.46487260e-01 -1.02211013e-01 -1.43190610e+00 1.31643367e+00
-7.22469151e-01 5.77985823e-01 6.31945670e-01 -1.13987529e+00
8.09668660e-01 6.36234343e-01 6.44674659e-01 -8.94940853e-01
2.60572821e-01 1.91741422e-01 1.94393506e-04 -6.73392057e-01
4.55015004e-01 -2.53652632e-01 1.87435016e-01 3.64465445e-01
-3.75237823e-01 -5.20310342e-01 -4.86788861e-02 -1.20330229e-01
1.51092517e+00 6.60988828e-03 1.70937583e-01 -1.37469649e-01
3.93455297e-01 2.39796430e-01 8.61280203e-01 7.75310993e-01
-2.26222992e-01 -1.63074285e-01 7.00759470e-01 -4.18628544e-01
-1.01583326e+00 -7.27463484e-01 3.31663430e-01 5.85266709e-01
3.24401379e-01 1.83119029e-01 -1.30601481e-01 -1.98959529e-01
1.14760615e-01 8.73550534e-01 -7.51667738e-01 -1.26654297e-01
-6.27671897e-01 -2.16580719e-01 -6.60912842e-02 2.36565500e-01
2.74429172e-01 -9.33311939e-01 -7.92090237e-01 1.00620663e+00
8.87208432e-02 -1.03449368e+00 -2.87959963e-01 5.23199141e-01
-9.19061482e-01 -1.04377985e+00 -4.11403686e-01 -4.16089714e-01
6.77793741e-01 -7.44952559e-02 7.74864197e-01 -2.05399409e-01
-1.13111347e-01 9.69267428e-01 -1.77988812e-01 -6.15191519e-01
-5.15065551e-01 -3.05356592e-01 3.70622724e-01 1.99632302e-01
-5.63607037e-01 -4.09109831e-01 -4.54523742e-01 2.47203484e-01
-1.29418695e+00 -3.00275326e-01 1.26038820e-01 9.84403014e-01
8.32695007e-01 7.10197091e-01 6.61300540e-01 -4.33519185e-01
5.87445915e-01 -5.50364494e-01 -1.31791663e+00 1.93896145e-01
-4.53627110e-01 1.75157323e-01 1.25207281e+00 -8.98895085e-01
-8.83084297e-01 1.50473684e-01 3.04102123e-01 -4.07468498e-01
-1.16938397e-01 6.67912304e-01 2.79362440e-01 1.70180455e-01
2.76478797e-01 6.30459726e-01 4.61968362e-01 -2.89747477e-01
-1.05753923e-02 3.15084219e-01 3.95970464e-01 -1.01625097e+00
8.15750837e-01 4.35604304e-01 4.20392901e-01 -9.42827940e-01
-3.67899179e-01 -9.65934247e-02 -2.29706913e-01 -3.52488428e-01
4.67195481e-01 -1.05884969e+00 -1.09847832e+00 6.18749976e-01
-7.50697494e-01 -1.14078712e+00 -5.15031457e-01 6.58071637e-01
-1.11303508e+00 -3.45917456e-02 -5.37244081e-01 -1.33086848e+00
2.85291076e-01 -8.08454216e-01 6.67641401e-01 3.71355921e-01
2.84277111e-01 -1.45551288e+00 3.00114065e-01 -4.87289697e-01
5.89431405e-01 5.74652731e-01 7.01995730e-01 -3.67535859e-01
-5.31052649e-01 -1.23867027e-01 5.35567522e-01 1.75101578e-01
-2.17932053e-02 -1.59056440e-01 -3.15363854e-01 -9.06911254e-01
1.09466761e-01 -4.61567312e-01 1.27042428e-01 5.05035043e-01
9.40352201e-01 -7.56872535e-01 -2.45861709e-01 6.71809390e-02
1.65413761e+00 8.36842597e-01 2.03204732e-02 5.93770504e-01
-4.83477451e-02 5.07678866e-01 1.13693035e+00 1.02859449e+00
4.79697406e-01 5.87925375e-01 1.02623558e+00 9.87524912e-02
8.04597318e-01 -2.44959086e-01 4.47897136e-01 3.05569887e-01
-2.92452896e-04 -2.42995620e-01 -7.72596300e-01 8.14455211e-01
-2.00060511e+00 -8.88242364e-01 1.10552058e-01 2.79241228e+00
7.44388759e-01 1.27937943e-01 1.01704612e-01 -1.98914446e-02
5.62135577e-01 3.20198871e-02 -1.07354629e+00 -3.06298912e-01
8.15870091e-02 -1.99510023e-01 8.62638235e-01 5.92244804e-01
-9.00241852e-01 5.21418810e-01 5.27543402e+00 3.55671138e-01
-1.46240485e+00 -2.88113207e-01 2.69388288e-01 -8.29017907e-02
1.16378613e-01 6.08449765e-02 -4.98307794e-01 7.00408995e-01
1.19051135e+00 -4.13229853e-01 9.12812531e-01 6.00760579e-01
1.04695606e+00 -2.16189504e-01 -1.01392055e+00 4.23074543e-01
-6.67846143e-01 -8.01881015e-01 -6.81769133e-01 5.31120375e-02
8.24656844e-01 8.16803128e-02 1.11700475e-01 6.22070312e-01
5.82183123e-01 -5.32504439e-01 6.29112661e-01 6.99882150e-01
5.80382109e-01 -9.59138334e-01 3.64299953e-01 9.88505542e-01
-1.14337254e+00 -7.00953424e-01 -2.13523790e-01 -3.68263602e-01
2.95481592e-01 3.96761298e-01 -5.80792129e-01 5.62993050e-01
2.18901381e-01 6.17812812e-01 2.31550768e-01 1.18384278e+00
-2.38671944e-01 6.91454947e-01 -6.00996673e-01 -1.89923063e-01
7.32606828e-01 -4.28509444e-01 9.56426978e-01 5.15145123e-01
4.23389375e-01 3.18627119e-01 7.96049476e-01 5.00781238e-01
6.65777981e-01 -5.17291307e-01 -8.86425197e-01 -7.39604831e-02
4.37900186e-01 7.76767910e-01 -5.10331213e-01 -1.93899974e-01
-1.48028180e-01 6.04541838e-01 -2.93016620e-02 6.90097451e-01
-8.90482426e-01 -3.17784816e-01 7.64245629e-01 -1.87881906e-02
1.45170406e-01 -8.54923427e-01 5.40385306e-01 -1.01746178e+00
-2.22456008e-01 -8.65438581e-01 1.18932702e-01 -4.62112457e-01
-9.77525771e-01 3.88948739e-01 -2.62216538e-01 -1.63660610e+00
-5.60219288e-01 -2.37749755e-01 -6.05078518e-01 5.64627588e-01
-1.68345475e+00 -5.67071438e-01 2.54474998e-01 7.20806718e-01
7.29053855e-01 -4.73327450e-02 6.34066164e-01 -2.68630058e-01
-5.95933974e-01 -3.35974455e-01 8.42326105e-01 -3.52060467e-01
3.92769098e-01 -1.59787309e+00 6.56996146e-02 3.50039512e-01
-6.53550565e-01 9.12090912e-02 1.29784906e+00 -6.09073102e-01
-2.00359058e+00 -1.41299546e+00 2.68563598e-01 2.88321376e-01
1.11069286e+00 5.28006628e-02 -9.55807090e-01 7.93541312e-01
1.11078285e-01 1.20912150e-01 -1.64385304e-01 -5.53777218e-01
3.72572035e-01 -2.11139172e-01 -1.09461653e+00 4.51355129e-01
3.89142185e-01 -2.59169072e-01 -2.25544810e-01 4.90025908e-01
6.23158813e-01 -6.93760812e-01 -7.32086122e-01 3.24432850e-01
2.09024474e-01 -3.99543285e-01 5.34640968e-01 -6.94220543e-01
-1.16028696e-01 -3.01645875e-01 4.63859737e-02 -2.11282849e+00
1.22738229e-02 -1.30703402e+00 -2.50557154e-01 5.84188282e-01
1.70710608e-01 -1.10257518e+00 3.49752128e-01 5.11969388e-01
2.72562522e-02 -5.68250954e-01 -1.30297577e+00 -1.34390903e+00
4.26819324e-02 -1.92470118e-01 3.64202410e-01 7.42690623e-01
-1.28505662e-01 -1.69532254e-01 -5.46766043e-01 6.97450340e-01
6.26309752e-01 -2.61993445e-02 6.13340080e-01 -8.47861946e-01
-6.26016498e-01 5.57110868e-02 3.30741741e-02 -1.10416663e+00
4.21816140e-01 1.19757377e-01 6.55642331e-01 -1.28915465e+00
-5.64170420e-01 -7.73185790e-01 -9.92603153e-02 2.58487314e-01
1.61479011e-01 -7.94759035e-01 1.12793058e-01 -7.41574317e-02
-5.36403716e-01 9.58877265e-01 1.27905023e+00 -2.21027240e-01
-5.26675761e-01 6.63262248e-01 2.87009716e-01 4.12458807e-01
9.70744610e-01 -4.13980335e-01 -7.89493501e-01 -4.86395657e-01
1.60076275e-01 1.22909069e+00 1.99869648e-01 -1.13726580e+00
2.50307679e-01 -8.16715539e-01 -3.66651475e-01 -4.05112565e-01
3.32800090e-01 -1.19697464e+00 2.74925083e-01 9.99700189e-01
-4.80971843e-01 2.27756575e-01 3.28657746e-01 1.39828217e+00
-1.65211424e-01 6.61129057e-02 6.83709502e-01 -1.76551014e-01
-6.55201495e-01 5.53841233e-01 -8.63963187e-01 1.82540286e-02
1.20572174e+00 2.64984101e-01 -3.15637082e-01 -8.63931537e-01
-1.17587054e+00 8.76640022e-01 2.15868130e-01 2.84525245e-01
1.62115514e-01 -7.47132599e-01 -4.94362682e-01 1.07531443e-01
-5.18277526e-01 -3.45733277e-02 2.66227007e-01 9.60470557e-01
-1.32817134e-01 3.15312058e-01 -1.46303564e-01 -4.55471188e-01
-7.99063921e-01 6.84198141e-01 5.63345730e-01 -3.01181972e-01
-5.86453021e-01 1.70872182e-01 -2.49103516e-01 -3.37909639e-01
3.42939973e-01 -6.17292166e-01 -4.85179983e-02 9.09769908e-02
5.80766797e-01 4.35150206e-01 -9.03465301e-02 -1.19016752e-01
2.34243959e-01 2.60021865e-01 3.34767073e-01 -3.62913519e-01
1.47266972e+00 -2.87849456e-01 2.89066851e-01 8.79515290e-01
6.65745258e-01 -6.91118956e-01 -2.15546513e+00 -4.29734260e-01
-2.90626306e-02 -2.90373921e-01 3.20215411e-02 -7.90632367e-01
-8.88708293e-01 4.63722616e-01 3.68151516e-01 7.97783673e-01
1.07575989e+00 -7.21566617e-01 5.47477305e-01 6.14763856e-01
8.82596195e-01 -1.39492321e+00 -2.51267441e-02 1.03637123e+00
9.53933656e-01 -1.00628078e+00 -4.42906737e-01 2.33539268e-01
-7.56957293e-01 1.20256186e+00 5.08145690e-01 -4.46639448e-01
6.38304293e-01 4.83447820e-01 -8.98904130e-02 4.29461390e-01
-1.41904974e+00 -2.97154158e-01 -4.27824199e-01 6.66203439e-01
-2.40994349e-01 3.45589936e-01 -2.44310528e-01 1.18163135e-02
4.43386048e-01 1.49313256e-01 9.50280070e-01 1.30469656e+00
-5.41215539e-01 -8.30886424e-01 -6.51520908e-01 1.21988952e-01
-2.08778143e-01 6.21165633e-01 9.93234366e-02 9.64501143e-01
-2.22635463e-01 9.92139459e-01 1.39345586e-01 6.66903406e-02
3.72768193e-01 -1.78126767e-01 2.09312364e-01 -4.18943375e-01
-4.77123916e-01 3.26840907e-01 4.90465760e-02 -7.25289404e-01
3.38984095e-02 -8.14489186e-01 -1.55539417e+00 -2.04611242e-01
-1.34781823e-01 2.69597143e-01 8.93576622e-01 1.09309328e+00
2.86989480e-01 6.60142004e-01 1.15030360e+00 -1.03270888e+00
-1.34620547e+00 -6.67590320e-01 -1.07075489e+00 -3.67788076e-01
1.08432388e+00 -7.24767029e-01 -6.38138771e-01 -1.37463465e-01] | [4.548150062561035, 2.2391626834869385] |
273cca4a-1486-4733-9b88-18b30ca49b30 | counterfactual-multihop-qa-a-cause-effect | 2210.07138 | null | https://arxiv.org/abs/2210.07138v1 | https://arxiv.org/pdf/2210.07138v1.pdf | Counterfactual Multihop QA: A Cause-Effect Approach for Reducing Disconnected Reasoning | Multi-hop QA requires reasoning over multiple supporting facts to answer the question. However, the existing QA models always rely on shortcuts, e.g., providing the true answer by only one fact, rather than multi-hop reasoning, which is referred as $\textit{disconnected reasoning}$ problem. To alleviate this issue, we propose a novel counterfactual multihop QA, a causal-effect approach that enables to reduce the disconnected reasoning. It builds upon explicitly modeling of causality: 1) the direct causal effects of disconnected reasoning and 2) the causal effect of true multi-hop reasoning from the total causal effect. With the causal graph, a counterfactual inference is proposed to disentangle the disconnected reasoning from the total causal effect, which provides us a new perspective and technology to learn a QA model that exploits the true multi-hop reasoning instead of shortcuts. Extensive experiments have conducted on the benchmark HotpotQA dataset, which demonstrate that the proposed method can achieve notable improvement on reducing disconnected reasoning. For example, our method achieves 5.8% higher points of its Supp$_s$ score on HotpotQA through true multihop reasoning. The code is available at supplementary material. | ['Hanjiang Lai', 'Qinkang Gong', 'Wangzhen Guo'] | 2022-10-13 | null | null | null | null | ['counterfactual-inference'] | ['miscellaneous'] | [-4.81827632e-02 5.94413221e-01 -4.94615674e-01 -5.03071487e-01
-1.10249507e+00 -5.20394683e-01 3.06861132e-01 8.24678838e-02
5.40434420e-02 1.27542830e+00 4.68315274e-01 -6.78940833e-01
-6.28774762e-01 -1.41437733e+00 -9.20441031e-01 -5.22900224e-01
1.79426745e-01 4.42817360e-01 3.71787697e-01 -5.39782941e-01
1.67285070e-01 -2.67391264e-01 -1.18873525e+00 5.18489301e-01
1.39099705e+00 7.96257615e-01 -3.14359188e-01 2.46119991e-01
-3.93564761e-01 1.33574152e+00 -6.87032819e-01 -9.57916796e-01
2.17421383e-01 -5.61097205e-01 -1.13009632e+00 -7.11807549e-01
4.56293911e-01 -5.29007614e-01 -3.74020249e-01 1.07950962e+00
2.18400225e-01 -1.36141166e-01 2.67822832e-01 -1.68303657e+00
-6.34292305e-01 1.15278029e+00 -9.54195559e-01 2.88516790e-01
7.88897514e-01 2.95675129e-01 1.59832215e+00 -1.89788759e-01
3.86602044e-01 1.57201147e+00 5.04707277e-01 1.85818136e-01
-9.34094489e-01 -9.37987626e-01 3.48839790e-01 6.51757240e-01
-1.13464093e+00 -9.08301324e-02 7.94374824e-01 4.56799567e-02
5.94018519e-01 4.18112874e-01 4.62669045e-01 7.92165041e-01
3.03153276e-01 7.41553068e-01 1.35818267e+00 -2.36481130e-01
8.44053626e-02 -2.10060731e-01 4.82266426e-01 9.23359931e-01
2.94780284e-01 1.55554950e-01 -7.18262196e-01 -3.73206913e-01
2.46237174e-01 -1.19352035e-01 -2.33913258e-01 -8.90050009e-02
-1.12400103e+00 8.49073291e-01 8.23452234e-01 -2.86348164e-01
-2.42387310e-01 2.68370062e-01 1.43896282e-01 3.79430979e-01
-1.25902206e-01 3.32390904e-01 -7.03882337e-01 1.59377992e-01
-5.80926955e-01 4.20595348e-01 7.40215003e-01 8.33230436e-01
9.96660233e-01 -5.24756134e-01 -3.90469521e-01 2.82897085e-01
3.80875826e-01 7.54406571e-01 -1.30727366e-01 -1.18833935e+00
9.16371584e-01 1.10424006e+00 3.01079720e-01 -1.11561644e+00
-3.14018309e-01 -2.37883538e-01 -6.29795134e-01 -1.23879984e-01
3.76495123e-01 -4.02465969e-01 -6.21808469e-01 1.99892879e+00
6.40727997e-01 3.48469853e-01 2.23749474e-01 9.26573992e-01
8.47325504e-01 4.54404086e-01 1.17822655e-01 -2.97340006e-01
1.33861840e+00 -8.97602677e-01 -8.08323264e-01 4.62379009e-02
6.23212636e-01 -4.19094175e-01 1.34727550e+00 4.14932251e-01
-9.47265446e-01 -8.60325396e-02 -1.00709236e+00 -7.96409845e-02
-2.17817545e-01 -3.12812269e-01 9.56847668e-01 5.92841446e-01
-5.15316010e-01 1.62035823e-01 -2.69016713e-01 1.26028687e-01
5.38811386e-01 2.24776357e-01 -2.74073035e-02 -6.97615921e-01
-1.97359300e+00 7.34957814e-01 3.18220794e-01 3.52824330e-02
-9.43445086e-01 -1.07412446e+00 -5.22469461e-01 2.63458759e-01
1.34374976e+00 -9.15660679e-01 1.06134951e+00 -4.55522597e-01
-1.25918519e+00 3.05552185e-01 -2.39027128e-01 -4.81427312e-01
7.83056974e-01 -3.72429103e-01 -4.87399936e-01 2.72279501e-01
5.36918461e-01 2.78672695e-01 3.58000487e-01 -1.18936551e+00
-7.27341831e-01 -4.50842977e-01 8.91532600e-01 1.50503844e-01
9.21940058e-02 -4.74612892e-01 -1.39572993e-01 -1.22831628e-01
6.09396957e-02 -6.03859127e-01 -9.52621922e-03 -3.01461250e-01
-7.69307077e-01 -4.99708384e-01 7.88918674e-01 -4.33506101e-01
1.17698157e+00 -1.70078528e+00 -1.04046084e-01 5.98435663e-02
4.13034171e-01 -2.08967198e-02 -1.49856238e-02 5.58916450e-01
1.69628546e-01 2.29631290e-01 -4.42125142e-01 6.26584828e-01
4.58243489e-02 2.07211852e-01 -6.57032549e-01 4.51625101e-02
1.43366084e-01 9.09794986e-01 -1.18740046e+00 -6.82120919e-01
-8.86342525e-02 -8.06905478e-02 -5.31306565e-01 -2.09490210e-02
-6.05075896e-01 1.98384732e-01 -6.71757698e-01 5.90272486e-01
1.06593132e+00 -2.92773038e-01 3.42986017e-01 -3.15545589e-01
9.95807871e-02 6.23714924e-01 -1.13499892e+00 1.75405312e+00
-2.98296094e-01 5.12402877e-02 -1.15327701e-01 -8.25153530e-01
6.57738328e-01 2.02126414e-01 3.29275638e-01 -9.25374925e-01
-4.18620259e-02 1.46312580e-01 8.40477720e-02 -7.36827135e-01
6.33208901e-02 -4.11645859e-01 -2.69861668e-01 4.80468512e-01
-3.20801139e-01 -2.04455648e-02 2.93818980e-01 7.60128975e-01
1.35964942e+00 5.75053394e-02 9.55369473e-02 -2.22543031e-01
6.22092664e-01 4.81690019e-01 9.16016161e-01 7.47511625e-01
-2.99979568e-01 1.03222236e-01 1.09030449e+00 -2.50102639e-01
-2.09194183e-01 -1.47058988e+00 2.28397295e-01 6.05743587e-01
7.34579504e-01 -4.82162625e-01 -3.70584965e-01 -1.45131648e+00
1.25934973e-01 1.07319927e+00 -5.89852273e-01 -3.27081174e-01
-5.14818549e-01 -6.65439129e-01 8.77582908e-01 2.99800307e-01
1.06461370e+00 -6.57135010e-01 -4.08100009e-01 -8.91671628e-02
-9.64609802e-01 -7.74761736e-01 -1.12830065e-02 -3.52708399e-01
-5.81541121e-01 -1.74465585e+00 -2.49105133e-02 -7.08602443e-02
3.27334851e-01 4.62879062e-01 1.18558919e+00 4.49706525e-01
1.04787491e-01 1.55502632e-01 -2.46402875e-01 -3.09802860e-01
4.20951247e-02 -1.80302665e-01 -2.58299619e-01 -3.36110175e-01
4.27476317e-01 -4.53240871e-01 -8.16588223e-01 3.13471138e-01
-7.44798720e-01 6.61877394e-02 8.18668902e-01 8.46099436e-01
4.40753877e-01 4.96291310e-01 1.04833281e+00 -1.25666571e+00
5.58503211e-01 -9.65354502e-01 -3.11994672e-01 6.81634724e-01
-7.63171852e-01 2.37542197e-01 8.83285046e-01 7.70630911e-02
-1.62745035e+00 -5.09386659e-01 -6.84744492e-02 -9.55370441e-02
-2.10987199e-02 6.78504825e-01 -5.56849301e-01 4.24288243e-01
5.06337583e-01 -7.42609650e-02 -4.28607851e-01 -8.18826631e-02
7.91439056e-01 3.32374275e-01 4.82693642e-01 -9.36428249e-01
8.44785154e-01 7.85458565e-01 1.64474189e-01 -1.04103662e-01
-1.33179343e+00 -1.94796562e-01 -2.33666241e-01 -4.64807749e-02
7.36605406e-01 -7.88321435e-01 -1.18744993e+00 -8.28527734e-02
-1.08899045e+00 -1.23106711e-01 -1.60605937e-01 2.93881655e-01
-3.82931411e-01 2.95356274e-01 -3.60953957e-01 -6.65672243e-01
-1.77480325e-01 -8.24189246e-01 6.91708446e-01 3.95143330e-01
1.73987627e-01 -7.79162824e-01 2.56746765e-02 9.31911170e-01
-1.48837745e-01 3.12159300e-01 1.31820905e+00 -4.29896146e-01
-9.75115359e-01 -1.64523721e-02 -6.18130028e-01 -1.12765118e-01
6.37694374e-02 -1.44519463e-01 -8.21923614e-01 9.75851715e-03
-3.01145483e-02 -4.46765602e-01 8.14290524e-01 7.53438398e-02
1.02921307e+00 -5.31201541e-01 -4.30440664e-01 2.57318513e-03
1.51620698e+00 -4.57514301e-02 7.58661985e-01 2.36005131e-02
5.70312142e-01 6.67871952e-01 9.90161598e-01 1.91976905e-01
9.25219953e-01 2.92663217e-01 6.67022705e-01 1.33367598e-01
-8.01793635e-02 -6.22917593e-01 7.57226273e-02 2.72328734e-01
7.50294551e-02 -2.28326812e-01 -9.44840133e-01 5.59672952e-01
-2.00632858e+00 -1.02695096e+00 -6.55930638e-01 1.96528757e+00
7.45376885e-01 2.69149005e-01 -4.01642499e-03 2.92633355e-01
7.01340795e-01 1.01495005e-01 -7.81279802e-01 -1.56856641e-01
-8.90592933e-02 -7.16924965e-02 1.66600630e-01 6.72135174e-01
-4.50766593e-01 7.38141239e-01 5.49792814e+00 7.83501625e-01
-6.46840692e-01 1.34097546e-01 3.62364471e-01 -1.14662103e-01
-9.53143895e-01 5.11296809e-01 -6.00799680e-01 5.49916029e-01
5.33989727e-01 -3.59489322e-01 2.86557138e-01 3.91347945e-01
2.37198427e-01 -4.01343137e-01 -9.28548753e-01 3.68904561e-01
-2.69909620e-01 -1.36436009e+00 4.74429220e-01 -1.01477169e-01
6.73754811e-01 -4.87041771e-01 -2.18301937e-01 5.76600015e-01
8.78044486e-01 -8.79588783e-01 4.58642453e-01 5.47961771e-01
6.49261355e-01 -8.64308953e-01 9.03362572e-01 5.50855517e-01
-1.20200515e+00 -3.97362322e-01 -1.11703388e-01 -3.61087024e-01
1.27309576e-01 9.65549111e-01 -5.20877600e-01 1.43308818e+00
7.62811303e-01 4.88980204e-01 -2.35259086e-01 7.97334611e-01
-7.94294119e-01 6.97911561e-01 -1.16431214e-01 3.51508241e-03
1.30555034e-01 -2.78146192e-02 4.41779941e-01 7.25555182e-01
8.33593979e-02 5.91198385e-01 -6.62603155e-02 9.97078061e-01
-2.90647000e-01 -2.56816715e-01 -3.87359560e-01 2.47540802e-01
6.70351148e-01 1.07383323e+00 -4.14596498e-01 -3.73365909e-01
-4.65132892e-01 5.06872296e-01 5.81932545e-01 3.52997750e-01
-1.22728884e+00 -5.49225986e-01 3.44988734e-01 -4.98757185e-03
2.66921204e-02 2.01354712e-01 -4.54934418e-01 -1.16252697e+00
1.87898308e-01 -8.14632177e-01 1.06070948e+00 -1.00754893e+00
-1.48744583e+00 1.03881925e-01 1.93413898e-01 -8.76018345e-01
2.57449448e-01 -1.95152625e-01 -9.78977561e-01 8.50966275e-01
-1.99912870e+00 -1.14163196e+00 -4.46471721e-01 7.06680357e-01
2.03373432e-01 4.06555146e-01 5.34611762e-01 1.54803559e-01
-5.36585569e-01 5.96966207e-01 -5.64094722e-01 -1.53938949e-01
8.19979966e-01 -1.29883528e+00 -2.89288044e-01 8.46362770e-01
-8.19352791e-02 7.31264710e-01 6.58944905e-01 -7.85622239e-01
-1.46620536e+00 -8.07351291e-01 9.27119076e-01 -6.28234744e-01
6.22656524e-01 1.79074407e-01 -8.33087623e-01 6.86036170e-01
4.15382296e-01 -3.22944671e-01 7.09019721e-01 4.58382308e-01
-7.73240507e-01 -4.50779378e-01 -1.33122611e+00 6.46321714e-01
1.16054165e+00 -2.01993838e-01 -1.12746239e+00 2.68114358e-01
1.23135090e+00 -1.63143665e-01 -6.27414763e-01 5.90210080e-01
3.94165933e-01 -1.28194547e+00 9.97426987e-01 -6.85835838e-01
9.67742801e-01 -5.23132503e-01 -1.14156686e-01 -1.10630822e+00
-9.20043364e-02 -4.85600531e-01 -3.78741235e-01 1.21552515e+00
6.31970942e-01 -6.62879348e-01 7.37184465e-01 4.80251551e-01
2.38570701e-02 -9.34441805e-01 -9.88823771e-01 -5.80659389e-01
1.27182394e-01 -3.35387468e-01 1.08351660e+00 1.24373436e+00
1.41137674e-01 4.98428047e-01 -2.92057782e-01 6.19268239e-01
7.68897831e-01 5.15628040e-01 5.65356910e-01 -1.11849117e+00
-3.28520596e-01 -5.08737452e-02 1.66669995e-01 -1.06768656e+00
6.89690188e-02 -7.09981680e-01 -5.97439446e-02 -2.01804709e+00
3.11219245e-01 -6.38478577e-01 -2.81064123e-01 5.32557487e-01
-6.14342928e-01 -2.65810430e-01 -2.13416517e-02 5.98441437e-02
-8.50398242e-01 5.62256277e-01 1.55653739e+00 -2.91134238e-01
1.31756589e-01 -8.62278640e-02 -1.13232398e+00 6.55999124e-01
6.71632290e-01 -5.29041827e-01 -8.63715291e-01 -5.78193307e-01
6.56264305e-01 6.07587934e-01 6.76390290e-01 -7.09216237e-01
4.01590914e-01 -5.91114223e-01 -1.02599360e-01 -5.44839561e-01
2.15815932e-01 -7.32453763e-01 -3.06693047e-01 6.30996883e-01
-4.08122629e-01 -9.23668742e-02 7.54271597e-02 1.02782321e+00
-1.12385616e-01 1.08890966e-01 2.18598038e-01 -1.78482920e-01
-5.65102816e-01 9.60727781e-02 1.55527219e-01 4.72041637e-01
1.05539596e+00 3.16961586e-01 -1.08797741e+00 -3.07418138e-01
-4.53800529e-01 8.20668519e-01 -1.23239540e-01 1.27422020e-01
6.19869590e-01 -1.19838417e+00 -8.13061297e-01 -3.34130913e-01
-6.99242949e-02 1.86040908e-01 6.68553114e-01 1.08013964e+00
-1.61767825e-02 3.50223064e-01 -1.00684211e-01 -2.28771716e-01
-9.73157585e-01 6.96435452e-01 2.87035346e-01 -5.27125120e-01
-3.02791148e-01 6.95900142e-01 1.04577161e-01 -5.45540154e-01
-1.94001213e-01 -1.48610607e-01 -1.25331670e-01 -1.18127339e-01
3.21345389e-01 8.44407439e-01 -1.20876797e-01 -4.50807363e-02
-5.17796218e-01 2.95558989e-01 2.03887727e-02 -8.73472989e-02
9.12730455e-01 -2.88940877e-01 -3.86168957e-01 2.65502423e-01
6.51812494e-01 3.41825366e-01 -9.51145291e-01 -1.38147727e-01
-2.27167532e-01 -7.32872605e-01 -1.81759849e-01 -1.47249162e+00
-9.31276321e-01 9.64478016e-01 -3.67300846e-02 5.53735256e-01
1.24079728e+00 2.78036781e-02 1.04979467e+00 3.98669422e-01
5.77912033e-01 -7.47665465e-01 2.89087254e-03 1.91635430e-01
8.33569109e-01 -1.27711391e+00 1.14793718e-01 -9.11718965e-01
-5.10583401e-01 7.71940112e-01 1.01062882e+00 -3.64676416e-02
4.47119117e-01 1.15111191e-03 8.41175243e-02 -6.31886482e-01
-9.08388734e-01 -3.93256173e-02 -9.54161137e-02 3.18386018e-01
1.83608055e-01 2.44679406e-01 -5.89233518e-01 9.96619344e-01
-2.79165298e-01 5.96197098e-02 5.65730810e-01 6.98681176e-01
-2.07633540e-01 -8.63281548e-01 -2.95218229e-01 5.75384974e-01
-1.03652768e-01 -1.36220962e-01 -6.06186271e-01 9.14963245e-01
3.86922389e-01 1.53244245e+00 -3.63559932e-01 -5.22849739e-01
4.48030353e-01 -2.61510313e-02 2.81950057e-01 -4.12583977e-01
-3.73058617e-01 -6.20314002e-01 1.54738232e-01 -7.05757797e-01
-4.41032171e-01 -3.18109274e-01 -1.69085932e+00 -7.80287266e-01
-3.69654864e-01 5.06223321e-01 7.96435401e-03 1.38373792e+00
4.29986745e-01 7.09991217e-01 4.87239897e-01 3.09229612e-01
-6.05025828e-01 -5.19532263e-01 -2.41782889e-01 2.68638283e-01
2.97602803e-01 -8.22449088e-01 -3.65739822e-01 -3.91575009e-01] | [9.951370239257812, 7.842447757720947] |
1b42c65f-17de-4f93-892b-31d54c6beef3 | causal-augmentation-for-causal-sentence | null | null | https://aclanthology.org/2021.cinlp-1.1 | https://aclanthology.org/2021.cinlp-1.1.pdf | Causal Augmentation for Causal Sentence Classification | Scarcity of annotated causal texts leads to poor robustness when training state-of-the-art language models for causal sentence classification. In particular, we found that models misclassify on augmented sentences that have been negated or strengthened with respect to its causal meaning. This is worrying since minor linguistic differences in causal sentences can have disparate meanings. Therefore, we propose the generation of counterfactual causal sentences by creating contrast sets (Gardner et al., 2020) to be included during model training. We experimented on two model architectures and predicted on two out-of-domain corpora. While our strengthening schemes proved useful in improving model performance, for negation, regular edits were insufficient. Thus, we also introduce heuristics like shortening or multiplying root words of a sentence. By including a mixture of edits when training, we achieved performance improvements beyond the baseline across both models, and within and out of corpus’ domain, suggesting that our proposed augmentation can also help models generalize. | ['Roger Zimmermann', 'Soujanya Poria', 'See-Kiong Ng', 'Devamanyu Hazarika', 'Fiona Anting Tan'] | null | null | null | null | emnlp-cinlp-2021-11 | ['sentence-classification'] | ['natural-language-processing'] | [ 5.36572158e-01 6.28514051e-01 -2.93834984e-01 -8.11847985e-01
-6.65937006e-01 -6.55201077e-01 1.21330094e+00 5.40009916e-01
-5.12595952e-01 1.28878248e+00 8.37183118e-01 -5.89573264e-01
-7.16883838e-02 -6.52962327e-01 -9.49951768e-01 -2.31754750e-01
-2.52971619e-01 2.96344161e-01 1.17932022e-01 -3.49929631e-01
4.66762513e-01 4.96239252e-02 -1.16555822e+00 7.91559458e-01
9.12591457e-01 1.61409035e-01 1.54824093e-01 5.60554147e-01
-5.20089082e-02 1.32472062e+00 -9.40846920e-01 -8.83776069e-01
-1.42655775e-01 -6.12215400e-01 -8.87439370e-01 -4.57122564e-01
4.82812405e-01 -2.50820369e-01 -1.34427264e-01 7.01784313e-01
3.55268925e-01 7.45256664e-03 8.73858750e-01 -1.19070828e+00
-1.02327120e+00 1.56503034e+00 -4.41794157e-01 3.41428071e-01
5.47441423e-01 1.01084113e-01 1.36016500e+00 -5.90442181e-01
8.49549711e-01 1.72179413e+00 8.25371623e-01 8.62201810e-01
-1.40126956e+00 -6.39436126e-01 5.25522768e-01 2.52259761e-01
-6.59298182e-01 -4.54647541e-01 8.03940654e-01 -3.48161221e-01
1.42034709e+00 4.29921031e-01 3.84058118e-01 1.68712914e+00
2.68540055e-01 6.39029860e-01 1.00135219e+00 -4.84810740e-01
1.69045076e-01 -7.36254454e-02 -1.64782941e-01 3.44946414e-01
3.02944928e-01 1.22047253e-01 -5.25312841e-01 -4.83859181e-01
2.04663381e-01 -6.92519963e-01 -2.67306983e-01 1.73856199e-01
-1.29728401e+00 1.08741117e+00 3.04824084e-01 2.72151381e-01
-3.63505512e-01 5.67382097e-01 5.13517022e-01 1.71706378e-01
7.19304979e-01 9.56818461e-01 -8.69081438e-01 -1.30051717e-01
-7.39603877e-01 6.68035686e-01 6.23247325e-01 5.46453118e-01
-5.35763688e-02 -2.71651208e-01 -3.54780823e-01 8.39436769e-01
1.68676421e-01 1.79014638e-01 3.00512582e-01 -9.69709039e-01
7.15669096e-01 5.14727294e-01 1.03499420e-01 -9.94910061e-01
-4.80535835e-01 -2.77929842e-01 -5.75752258e-01 -2.57856876e-01
5.49451053e-01 -3.75250578e-01 -7.11673319e-01 2.23738480e+00
7.63463005e-02 1.91291869e-01 1.15989417e-01 8.65550578e-01
5.57074249e-01 4.72213835e-01 7.87950397e-01 -4.50187415e-01
9.84626055e-01 -5.15139341e-01 -7.83807755e-01 -5.87010264e-01
1.04725373e+00 -7.47215867e-01 1.17967761e+00 9.62550342e-02
-1.00063324e+00 -2.05838144e-01 -1.01122880e+00 -7.43783936e-02
-1.22403182e-01 -3.51887345e-01 1.10016525e+00 5.17697513e-01
-6.77884042e-01 9.93842721e-01 -5.21029711e-01 -2.29755029e-01
3.88579935e-01 -5.21636866e-02 -3.25441122e-01 1.19517677e-01
-1.93552971e+00 1.25187385e+00 4.07207310e-01 -3.34131867e-02
-6.53484821e-01 -1.10042179e+00 -8.77061725e-01 -1.43052384e-01
2.87326634e-01 -8.83327186e-01 1.36583138e+00 -7.35067368e-01
-8.04273903e-01 7.55290985e-01 -2.36418322e-01 -7.37105489e-01
6.10719264e-01 -4.07909453e-01 -4.12209451e-01 -2.97176987e-01
3.62271488e-01 5.88610172e-01 5.67368507e-01 -1.20445466e+00
-5.52756786e-01 -8.39885548e-02 2.86179274e-01 1.05332412e-01
-1.42669857e-01 4.14147824e-01 3.96599561e-01 -9.07669127e-01
-3.78573120e-01 -5.52277327e-01 -2.43479446e-01 -5.16546786e-01
-5.62248290e-01 -5.02512097e-01 4.60228652e-01 -5.94525218e-01
1.43452823e+00 -1.77671516e+00 -1.58563275e-02 -1.71758369e-01
-1.07611425e-01 -5.29861823e-02 -2.68274963e-01 3.82436752e-01
-4.65059549e-01 9.02927279e-01 -4.39456910e-01 -3.34700376e-01
-1.10273272e-01 3.30705285e-01 -7.27635026e-01 1.65549845e-01
6.50584757e-01 7.26107061e-01 -1.26173210e+00 -5.33096850e-01
-1.40845403e-01 1.44732475e-01 -8.24462414e-01 -2.89083403e-02
-6.11955583e-01 3.34114224e-01 -1.38299257e-01 2.13314876e-01
3.48693341e-01 1.09348468e-01 3.86410505e-01 6.65212199e-02
5.70903234e-02 1.25765729e+00 -9.31605935e-01 1.42298913e+00
-5.62468946e-01 5.22298872e-01 -4.61336792e-01 -9.33361530e-01
5.02865970e-01 4.95453805e-01 7.67184943e-02 -4.93649989e-01
-8.74236450e-02 2.30390981e-01 5.13650656e-01 -5.44831693e-01
4.53047633e-01 -4.89060700e-01 -2.24446014e-01 6.17437541e-01
-1.76878795e-01 -3.43094051e-01 6.09220207e-01 3.98147106e-01
1.17783380e+00 1.37374163e-01 3.15521598e-01 -2.90755272e-01
2.65994668e-01 2.31536984e-01 6.74295485e-01 9.52064455e-01
-3.98637280e-02 4.46007788e-01 9.23864305e-01 -1.60961598e-01
-1.04546249e+00 -8.42486024e-01 -2.48243690e-01 9.88609970e-01
-4.14893925e-01 -5.47198772e-01 -5.86169958e-01 -1.09276712e+00
1.15879506e-01 1.65543544e+00 -8.46630037e-01 -3.02613050e-01
-8.57403696e-01 -1.08834696e+00 8.91238272e-01 6.21912062e-01
1.02116190e-01 -1.12499928e+00 -6.30789995e-01 4.34758693e-01
-2.98299998e-01 -7.86199808e-01 -2.38212466e-01 2.85654277e-01
-6.93755507e-01 -1.16186821e+00 -2.60686159e-01 -1.55495942e-01
2.67771870e-01 -3.03226143e-01 1.39969254e+00 1.28467157e-01
1.42050266e-01 -1.38544858e-01 -2.66661465e-01 -7.12943614e-01
-9.72798765e-01 -9.34110433e-02 1.56753156e-02 -5.88870287e-01
1.22706033e-01 -5.52863419e-01 -1.40113920e-01 -1.35407180e-01
-5.68760395e-01 9.75048989e-02 2.63340145e-01 9.65238929e-01
-6.25476167e-02 -1.59777507e-01 8.85165453e-01 -1.35968697e+00
1.10533977e+00 -5.11691630e-01 -4.39350121e-02 1.54899329e-01
-5.91102004e-01 4.30131912e-01 7.41727591e-01 -4.86888736e-01
-1.55625546e+00 -5.06295264e-01 -9.90033969e-02 2.68782109e-01
-1.23878904e-01 7.67729163e-01 6.28306493e-02 7.86386430e-01
1.02422023e+00 -5.37306428e-01 -3.37679803e-01 -2.07109138e-01
5.91876328e-01 3.47604871e-01 3.43773067e-01 -7.57753611e-01
5.61871529e-01 7.81131536e-02 -1.12586908e-01 -3.02993655e-01
-1.09248006e+00 1.06857911e-01 -4.07215148e-01 6.53946847e-02
6.59269214e-01 -8.35954309e-01 -3.62793446e-01 1.84663728e-01
-1.67199540e+00 -6.41654909e-01 -1.31610408e-01 4.02888954e-01
-2.04567343e-01 1.01534806e-01 -6.20979190e-01 -8.88074934e-01
3.39207463e-02 -6.62124693e-01 7.27422595e-01 -2.03560337e-01
-1.04632378e+00 -1.23108780e+00 4.53538187e-02 1.38082623e-01
2.56502360e-01 3.80129665e-01 1.36453497e+00 -8.21765721e-01
1.06759354e-01 -1.93569809e-03 -9.98120382e-03 1.88750565e-01
2.11971879e-01 2.56709754e-01 -9.20757651e-01 3.00626010e-01
-9.78414938e-02 -4.52048868e-01 1.07194519e+00 2.01543450e-01
1.03607404e+00 -6.35294974e-01 -4.72784489e-01 -9.58662480e-02
8.63500237e-01 1.64353445e-01 4.83238369e-01 3.64416361e-01
4.89454716e-01 8.06091130e-01 5.88356912e-01 2.77061045e-01
4.48148221e-01 4.05708581e-01 2.52210081e-01 1.63462702e-02
-6.28871024e-02 -5.21413863e-01 4.63248253e-01 1.73423141e-01
-5.11733294e-02 -4.40537423e-01 -9.24660087e-01 8.67073476e-01
-1.70048916e+00 -1.28311884e+00 -8.02444100e-01 1.80617034e+00
1.53682745e+00 6.16227031e-01 -2.53579050e-01 1.36993796e-01
5.06584525e-01 3.61342996e-01 -2.28761554e-01 -7.47403085e-01
-3.05210680e-01 1.49159849e-01 4.07387674e-01 6.52764499e-01
-1.04763591e+00 1.01791501e+00 6.46878672e+00 5.50289690e-01
-1.03635800e+00 1.46042109e-01 8.05615604e-01 -2.29945421e-01
-8.62911403e-01 3.58074486e-01 -5.40639341e-01 4.56335038e-01
1.04424703e+00 -1.80528536e-01 1.44285083e-01 4.97826755e-01
5.98018646e-01 -2.93699980e-01 -1.40742421e+00 7.86090717e-02
-1.03647336e-01 -1.55512238e+00 2.10384309e-01 -3.63289535e-01
7.88019955e-01 -2.08267555e-01 -3.53339344e-01 3.27171713e-01
5.96163750e-01 -1.05097246e+00 1.10361099e+00 2.54185081e-01
4.24554467e-01 -5.39263427e-01 8.62933338e-01 2.96766639e-01
-5.30140817e-01 3.86988604e-03 -1.18783019e-01 -6.93658054e-01
4.55922931e-01 8.61847401e-01 -1.17602241e+00 4.15524781e-01
5.02215147e-01 5.95206499e-01 -8.41483772e-01 3.26330006e-01
-8.99535596e-01 1.09136236e+00 -1.08060345e-01 -2.75103062e-01
-3.64091694e-02 4.80677098e-01 4.86406505e-01 1.47728837e+00
-8.62310976e-02 2.30879590e-01 -2.71163911e-01 1.08279192e+00
-1.32924825e-01 -1.23519704e-01 -6.51943386e-01 -1.94444254e-01
6.80613101e-01 7.89536893e-01 -2.75653481e-01 -4.35296625e-01
-2.64017612e-01 7.07712412e-01 4.20451045e-01 2.54326284e-01
-1.13548994e+00 -1.62930805e-02 5.44406533e-01 1.48378938e-01
-2.43395805e-01 7.38249421e-02 -9.27677572e-01 -9.63381231e-01
2.28405241e-02 -6.72393024e-01 4.48387504e-01 -8.69242847e-01
-1.42639005e+00 1.36110425e-01 3.21691781e-01 -5.57518899e-01
-5.09586334e-01 -3.94736350e-01 -8.42083097e-01 8.60815704e-01
-1.18567932e+00 -1.02747607e+00 3.77573222e-01 5.92189208e-02
4.59623158e-01 2.58035153e-01 7.87220418e-01 1.48781031e-01
-3.11553031e-01 5.40199816e-01 -6.14259601e-01 1.44394627e-02
1.20583165e+00 -1.42601657e+00 6.95875168e-01 9.41397667e-01
-2.30468325e-02 1.31965351e+00 1.26113749e+00 -1.00201869e+00
-7.02451885e-01 -1.14436460e+00 1.68945837e+00 -5.68971574e-01
9.53227401e-01 -4.38389182e-01 -1.00797749e+00 7.25829780e-01
5.56412756e-01 -6.15159333e-01 4.65269089e-01 5.75334430e-01
-5.99920809e-01 2.82699227e-01 -1.05608988e+00 9.26367462e-01
1.39137864e+00 -4.86442536e-01 -1.10822451e+00 4.18079913e-01
1.03465652e+00 -1.17891997e-01 -5.40765285e-01 5.00860572e-01
2.33548775e-01 -7.04575300e-01 6.97377443e-01 -1.11154187e+00
1.27693629e+00 -1.52965531e-01 1.51529703e-02 -1.74618304e+00
-2.07607120e-01 -3.38970125e-01 1.91542819e-01 1.65994191e+00
1.14968753e+00 -5.08219898e-01 2.54590869e-01 8.69518757e-01
-3.31036419e-01 -5.92315793e-01 -7.17887759e-01 -4.61531699e-01
3.57393950e-01 -8.90144289e-01 4.44018304e-01 1.39586568e+00
4.65455741e-01 7.08760858e-01 -1.97358459e-01 -6.52457103e-02
3.91903162e-01 -2.48363569e-01 4.07902181e-01 -1.06654501e+00
-2.36951023e-01 -4.60073084e-01 2.20066309e-01 -4.49452817e-01
7.07829714e-01 -9.42105114e-01 -2.31701229e-02 -1.63285863e+00
3.30240041e-01 -4.57568705e-01 -4.29294594e-02 8.55636835e-01
-7.46410072e-01 -4.24105003e-02 1.14248902e-01 -1.07155934e-01
2.27867365e-02 4.87148583e-01 1.08613956e+00 -2.34469801e-01
-2.09540930e-02 -3.45375091e-01 -1.00743353e+00 8.48238766e-01
8.98018301e-01 -7.11514294e-01 -4.10289794e-01 -5.73794663e-01
5.93544841e-01 -8.93996060e-02 7.77209878e-01 -3.03472579e-01
-4.57273908e-02 -5.63913882e-01 1.81805506e-01 -1.28527761e-01
-1.19361646e-01 -2.06728861e-01 6.02243980e-03 5.34312069e-01
-1.03278553e+00 1.07972778e-01 4.30543512e-01 4.25732762e-01
-7.53003582e-02 -2.63396144e-01 4.03995812e-01 -2.32144877e-01
-4.59926695e-01 -6.62414372e-01 -3.61440867e-01 2.82790571e-01
6.48618042e-01 2.98491478e-01 -7.25942135e-01 -3.04290116e-01
-2.73258269e-01 1.48457676e-01 2.59430468e-01 6.79592550e-01
3.46529335e-01 -1.04445994e+00 -1.00675094e+00 -3.68506581e-01
-3.33909281e-02 -2.48814449e-01 3.00296526e-02 8.17494869e-01
-1.88148499e-01 5.36149800e-01 2.30823070e-01 -1.51474282e-01
-1.03308487e+00 4.99654561e-01 1.38130367e-01 -2.90220141e-01
-1.01826757e-01 1.05246854e+00 -4.97313682e-03 -4.81145650e-01
-9.25930142e-02 -5.91670454e-01 -1.20876238e-01 2.15909153e-01
3.12028438e-01 3.81341912e-02 7.48829767e-02 -1.02752388e-01
-5.94118893e-01 -1.39770627e-01 -1.41176268e-01 -4.26198661e-01
1.29848313e+00 1.37326986e-01 -3.59641612e-01 6.86951458e-01
8.26780856e-01 3.33538860e-01 -9.09883201e-01 3.16401362e-01
3.57597917e-01 -2.66005605e-01 -1.30314216e-01 -1.32249904e+00
-3.22627008e-01 5.17782867e-01 -9.85269472e-02 2.02471346e-01
7.72417247e-01 1.20604798e-01 3.60172838e-01 2.03457475e-01
2.19851539e-01 -9.85567272e-01 -2.77527452e-01 6.59382701e-01
1.34159303e+00 -1.19361162e+00 1.82328582e-01 -5.41236460e-01
-6.50296271e-01 8.35136116e-01 4.92375940e-01 5.70813008e-02
6.79975376e-02 2.95575827e-01 -1.69471856e-02 -1.28630042e-01
-1.36085641e+00 1.73510402e-01 5.41132540e-02 3.65406483e-01
1.25204170e+00 1.58138424e-01 -1.14951873e+00 6.36750996e-01
-5.65727293e-01 -2.72591442e-01 8.72838676e-01 7.10855067e-01
4.18155901e-02 -1.14187074e+00 -3.24307352e-01 5.08644819e-01
-5.80331326e-01 -6.98078752e-01 -7.27878034e-01 9.96915400e-01
1.90116882e-01 1.27327645e+00 1.90171711e-02 -2.66863257e-01
3.60278159e-01 2.43648276e-01 4.90062356e-01 -7.27101564e-01
-7.75232136e-01 -3.41156870e-01 9.21854079e-01 -3.63529652e-01
-5.60858846e-01 -9.95186269e-01 -1.43885398e+00 -2.01156378e-01
-2.16627404e-01 1.15425266e-01 4.62037683e-01 1.18162048e+00
1.25826627e-01 7.57960737e-01 2.01644555e-01 -3.31813604e-01
-7.86539376e-01 -1.42696881e+00 -1.22153796e-02 6.53669477e-01
1.77071050e-01 -7.21525371e-01 -6.51131094e-01 3.56113642e-01] | [9.943744659423828, 8.109054565429688] |
96230801-e78b-4385-9b2e-5e2f227ff6b2 | multiple-riemannian-manifold-valued | 1908.0195 | null | https://arxiv.org/abs/1908.01950v1 | https://arxiv.org/pdf/1908.01950v1.pdf | Multiple Riemannian Manifold-valued Descriptors based Image Set Classification with Multi-Kernel Metric Learning | The importance of wild video based image set recognition is becoming monotonically increasing. However, the contents of these collected videos are often complicated, and how to efficiently perform set modeling and feature extraction is a big challenge for set-based classification algorithms. In recent years, some proposed image set classification methods have made a considerable advance by modeling the original image set with covariance matrix, linear subspace, or Gaussian distribution. As a matter of fact, most of them just adopt a single geometric model to describe each given image set, which may lose some other useful information for classification. To tackle this problem, we propose a novel algorithm to model each image set from a multi-geometric perspective. Specifically, the covariance matrix, linear subspace, and Gaussian distribution are applied for set representation simultaneously. In order to fuse these multiple heterogeneous Riemannian manifoldvalued features, the well-equipped Riemannian kernel functions are first utilized to map them into high dimensional Hilbert spaces. Then, a multi-kernel metric learning framework is devised to embed the learned hybrid kernels into a lower dimensional common subspace for classification. We conduct experiments on four widely used datasets corresponding to four different classification tasks: video-based face recognition, set-based object categorization, video-based emotion recognition, and dynamic scene classification, to evaluate the classification performance of the proposed algorithm. Extensive experimental results justify its superiority over the state-of-the-art. | ['Xiao-Jun Wu', 'Rui Wang', 'Josef Kittler'] | 2019-08-06 | null | null | null | null | ['object-categorization'] | ['computer-vision'] | [-1.85561981e-02 -7.16509044e-01 4.48606648e-02 -4.69664574e-01
-4.97416437e-01 -4.01182353e-01 2.86977530e-01 -2.75131553e-01
-2.17308462e-01 2.04818204e-01 -2.16146678e-01 2.22745419e-01
-6.04377866e-01 -5.37133217e-01 -4.02834207e-01 -1.14563227e+00
1.30770132e-01 -1.56502426e-01 -1.60714149e-01 -1.86270759e-01
1.81449622e-01 3.54447424e-01 -1.54733109e+00 7.11509138e-02
7.89473057e-01 1.14166617e+00 1.17784023e-01 3.32814455e-02
-1.00800835e-01 6.33482456e-01 -2.09330246e-01 -3.46629024e-01
1.15819059e-01 -3.78688455e-01 -3.77661645e-01 6.98665023e-01
1.77830338e-01 -7.14294799e-03 -4.78785843e-01 1.38239706e+00
1.48974106e-01 3.96104991e-01 7.09808409e-01 -1.47654188e+00
-7.60238469e-01 7.82168061e-02 -4.98676211e-01 3.85131314e-02
1.24423325e-01 -4.01242683e-03 7.25740433e-01 -1.14771116e+00
3.33178282e-01 1.22060096e+00 1.06282689e-01 4.52944249e-01
-8.49529147e-01 -6.73437476e-01 1.65373087e-01 5.98273039e-01
-1.69827032e+00 -3.24186057e-01 1.24938011e+00 -5.44864893e-01
1.51856020e-01 4.16110098e-01 6.48555458e-01 6.71487272e-01
-1.16024844e-01 6.10366881e-01 1.03788161e+00 -5.83268180e-02
1.04088485e-01 3.71729314e-01 2.56908417e-01 8.41369748e-01
1.83752134e-01 -3.97463441e-01 -3.99806947e-01 -3.17809451e-03
4.57812726e-01 6.47020161e-01 -7.46861219e-01 -7.34111965e-01
-1.24640310e+00 6.99160695e-01 4.21652049e-01 4.86794204e-01
-3.18095982e-01 -2.74734050e-01 2.76371539e-01 2.11747885e-01
5.28099835e-01 -1.68067724e-01 -5.89127280e-02 5.45899048e-02
-4.90044832e-01 -7.54626002e-03 5.22570312e-01 7.14676082e-01
8.95740628e-01 -6.18356317e-02 1.01759508e-01 1.01323247e+00
5.48729956e-01 6.09177232e-01 5.12855530e-01 -5.14270782e-01
4.13274437e-01 9.28639591e-01 -9.42913890e-02 -1.68364275e+00
-2.35652432e-01 -2.79766113e-01 -1.13890803e+00 -1.51157156e-01
2.96181440e-01 1.92346573e-01 -2.66713053e-01 1.57089102e+00
5.11764526e-01 5.69255471e-01 6.02486692e-02 1.22212422e+00
5.80901682e-01 5.90917408e-01 -3.05862218e-01 -2.88840562e-01
1.19104791e+00 -5.73558211e-01 -6.14638090e-01 3.87595326e-01
8.41229498e-01 -5.09024441e-01 8.98598969e-01 2.69030482e-01
-5.38007319e-01 -4.79812443e-01 -1.05325103e+00 3.51992756e-01
-2.73313433e-01 2.95325220e-01 5.18016994e-01 7.65106559e-01
-6.09833717e-01 3.84637564e-01 -6.52671337e-01 -1.69687882e-01
5.39483726e-01 2.65192509e-01 -6.87311232e-01 -5.02394438e-01
-9.19733882e-01 4.73564357e-01 3.65948439e-01 4.69353914e-01
-5.35270333e-01 -3.59728456e-01 -9.62658703e-01 -2.55620219e-02
3.94219637e-01 -3.68401170e-01 3.85722011e-01 -8.45860898e-01
-1.36919069e+00 7.18956828e-01 1.28966510e-01 1.05151534e-01
1.20360315e-01 -2.95231808e-02 -6.45690799e-01 3.33822787e-01
-2.33987257e-01 -4.54704575e-02 1.20764887e+00 -1.12944555e+00
-2.44737908e-01 -8.46585095e-01 1.37354597e-01 3.86552781e-01
-1.12165415e+00 6.12573884e-02 -3.15526426e-01 -4.54967201e-01
3.10733467e-01 -9.82662916e-01 1.32993415e-01 -1.01534920e-02
-8.43488052e-02 -1.27898097e-01 1.14761567e+00 -4.47451890e-01
1.34777009e+00 -2.43654561e+00 8.77336860e-01 1.72081783e-01
2.63781697e-01 2.19321340e-01 -1.85896099e-01 7.64827654e-02
-1.70903593e-01 -9.54216719e-03 -4.65143651e-01 -2.49358967e-01
-2.64227569e-01 4.66485694e-02 -1.48752913e-01 7.84635246e-01
2.51414716e-01 4.73069906e-01 -9.06064868e-01 -5.48233688e-01
5.76360643e-01 6.35814130e-01 -5.04536390e-01 2.34219357e-01
2.34499395e-01 6.93311572e-01 -7.43947864e-01 5.05501807e-01
8.97621632e-01 -1.03321746e-01 8.03066697e-03 -5.40397584e-01
9.48807746e-02 -6.47785485e-01 -1.42242253e+00 1.77411580e+00
-3.75734121e-01 1.78176850e-01 7.45645389e-02 -1.60937202e+00
9.33400750e-01 1.30630463e-01 8.11567545e-01 -1.72287926e-01
2.75118172e-01 1.81501299e-01 1.65635213e-01 -6.97550476e-01
6.63187355e-02 5.63493520e-02 8.52296427e-02 1.99046314e-01
1.87727064e-01 5.46541065e-02 -9.84197389e-03 3.14644538e-02
7.85779893e-01 -2.26740036e-02 -3.57484706e-02 -3.00721943e-01
1.24421859e+00 -4.11329299e-01 5.64482868e-01 -6.92638159e-02
-1.40188575e-01 5.39754510e-01 1.18203737e-01 -2.39717126e-01
-4.87919241e-01 -6.66002333e-01 -2.62146115e-01 6.08250618e-01
4.97170240e-01 -3.98908675e-01 -9.68004823e-01 -7.12430954e-01
-1.72358453e-01 2.43333936e-01 -4.77753252e-01 -6.40098691e-01
-4.09308463e-01 -8.76695335e-01 7.70430490e-02 4.12553400e-02
8.19874704e-01 -6.25090778e-01 -7.82490820e-02 3.53011377e-02
-9.24310759e-02 -1.04174995e+00 -6.83596611e-01 -5.66923738e-01
-7.36586928e-01 -1.27990699e+00 -7.87769973e-01 -7.49788165e-01
7.41073668e-01 8.98500621e-01 3.69035900e-01 2.45171458e-01
-4.15814757e-01 8.64011049e-01 -4.99233007e-01 -1.37334317e-01
1.28163487e-01 -3.62980902e-01 5.12878835e-01 1.18334293e+00
3.34868342e-01 -4.16607291e-01 -6.02889419e-01 5.58845580e-01
-1.25540900e+00 -9.49125811e-02 4.11138088e-01 1.00568080e+00
3.32414538e-01 3.72794122e-01 3.60555977e-01 -2.90111810e-01
2.94663727e-01 -6.45604312e-01 -3.98438603e-01 2.65476435e-01
-3.31193134e-02 -1.74830317e-01 6.40159905e-01 -5.35072148e-01
-8.51372182e-01 -1.13898531e-01 3.84784997e-01 -1.03371119e+00
5.29256426e-02 7.19889700e-01 -6.32387280e-01 -3.32692266e-01
1.04367979e-01 5.83238423e-01 4.42089856e-01 -4.56857860e-01
3.04968923e-01 7.82057643e-01 4.91596162e-02 -3.85177135e-01
1.02802384e+00 6.13667667e-01 1.47806227e-01 -1.11804438e+00
-8.19110334e-01 -5.96133769e-01 -9.02377427e-01 -5.79934418e-01
9.17851090e-01 -6.55279338e-01 -6.83514118e-01 7.87726581e-01
-9.25531983e-01 2.50305802e-01 2.11871356e-01 6.65256500e-01
-4.15096045e-01 6.90854013e-01 -3.61600220e-01 -7.99944401e-01
-1.31255195e-01 -1.23180282e+00 9.44872379e-01 2.76193708e-01
5.83762109e-01 -1.03835797e+00 -2.30609700e-01 4.62185681e-01
5.25593795e-02 2.55038828e-01 8.37708950e-01 -3.75783265e-01
-6.08188450e-01 -2.33533159e-01 -2.34526813e-01 7.06092238e-01
4.92256403e-01 -3.87946032e-02 -5.84516168e-01 -3.66760194e-01
6.33743227e-01 -9.40694511e-02 6.91919148e-01 -5.33832498e-02
1.51289463e+00 -2.16150656e-01 -3.12866956e-01 6.93494201e-01
1.39323461e+00 1.98147446e-01 2.89968342e-01 8.19335207e-02
1.08193719e+00 6.44664049e-01 8.21391642e-01 6.00568354e-01
2.84025908e-01 7.23192096e-01 4.12809104e-01 3.46432447e-01
4.71250236e-01 1.44163534e-01 4.50585544e-01 1.26697099e+00
-8.57882798e-02 2.54297275e-02 -6.25591516e-01 7.28400471e-03
-1.97344065e+00 -1.14026904e+00 1.37854338e-01 2.49290657e+00
2.96230704e-01 -1.52752578e-01 3.73873040e-02 3.96650225e-01
8.80301952e-01 1.84681743e-01 -4.15610909e-01 3.28889430e-01
-1.07053652e-01 -3.07832092e-01 -7.60773420e-02 -1.26039639e-01
-1.21430469e+00 6.60110116e-01 4.00241375e+00 1.06400859e+00
-1.30712378e+00 7.79268593e-02 5.18781304e-01 1.08896852e-01
-4.76145260e-02 -7.89670646e-02 -4.72668797e-01 7.06985295e-01
5.15375912e-01 -2.69285738e-01 6.15822554e-01 8.88021350e-01
1.15075409e-01 2.57430196e-01 -9.66585577e-01 1.72064281e+00
4.85705793e-01 -1.04652274e+00 3.10122252e-01 2.82376796e-01
3.29380959e-01 -5.88766873e-01 3.58922780e-01 4.43696737e-01
-5.72338402e-01 -8.32848132e-01 4.54140544e-01 9.14268911e-01
5.35710692e-01 -7.96736479e-01 5.04400551e-01 4.07286942e-01
-1.19412911e+00 -1.78405121e-01 -6.01877213e-01 2.57869661e-02
-1.55176222e-01 4.95020241e-01 -1.36496276e-02 8.92735004e-01
6.57050610e-01 1.30317307e+00 -6.03419662e-01 9.80749071e-01
5.27160943e-01 4.66210395e-01 -1.78480804e-01 -1.72901824e-01
2.33113796e-01 -1.01721239e+00 7.25541234e-01 5.55349112e-01
4.42146540e-01 4.59650278e-01 2.96779960e-01 6.32094800e-01
-1.10339850e-01 5.43521523e-01 -7.87637830e-01 -2.57961601e-01
9.51076001e-02 1.75279140e+00 -7.10260212e-01 -1.95995629e-01
-5.80975890e-01 1.07431865e+00 3.53945196e-01 2.29024813e-01
-9.14688110e-01 -4.61269379e-01 9.71043169e-01 -3.24745595e-01
7.75766522e-02 -5.85401475e-01 3.91104251e-01 -1.76976359e+00
2.33878791e-01 -7.50942349e-01 1.80274025e-01 -4.72093403e-01
-1.09768248e+00 3.96378279e-01 -5.15295491e-02 -1.81939852e+00
1.95133224e-01 -8.96676481e-01 -5.79844713e-01 4.22626346e-01
-1.10363984e+00 -1.00882804e+00 -5.69516361e-01 8.74254882e-01
3.53871137e-01 -2.73717016e-01 6.65332317e-01 5.59580207e-01
-1.05923104e+00 3.92789274e-01 5.03303468e-01 3.50239605e-01
4.94806349e-01 -9.58867431e-01 -7.39639282e-01 4.47530478e-01
2.87410229e-01 6.20401442e-01 1.78674795e-02 -3.77344377e-02
-2.17460084e+00 -1.19499671e+00 -1.13811092e-02 -4.33693081e-01
6.54749513e-01 -4.21983689e-01 -1.12170887e+00 3.69238585e-01
-3.16701263e-01 3.48202229e-01 8.35275948e-01 -1.48056760e-01
-3.20690781e-01 -5.74200213e-01 -1.00811636e+00 6.32154346e-01
1.03058422e+00 -5.58767080e-01 -2.20518455e-01 7.30746269e-01
4.65726882e-01 -8.42969418e-02 -1.21791327e+00 3.37724388e-01
3.03810984e-01 -9.41234469e-01 7.86581933e-01 -6.06579006e-01
9.02343541e-02 -4.89816010e-01 -4.66480106e-01 -1.30831635e+00
-2.01295286e-01 -3.04453522e-01 5.94269559e-02 1.48707259e+00
-3.82712334e-01 -6.81300581e-01 5.08648455e-01 4.45394993e-01
-3.90367657e-02 -8.07865620e-01 -8.71236920e-01 -7.37517238e-01
-2.55393714e-01 -4.57196385e-01 5.58404326e-01 1.04132092e+00
6.59006611e-02 3.23340774e-01 -3.19059700e-01 1.53421208e-01
9.30075645e-01 2.68116534e-01 6.84815764e-01 -1.10281169e+00
8.28111451e-03 -5.69726288e-01 -1.03951597e+00 -6.12863958e-01
4.55643356e-01 -1.06913698e+00 -3.59363198e-01 -9.98373330e-01
2.97300637e-01 -4.09403950e-01 -6.15710437e-01 6.35016859e-02
-1.70604095e-01 1.86848536e-01 2.66318440e-01 3.96038383e-01
-7.48635650e-01 1.15021515e+00 1.12313092e+00 -3.13875943e-01
-3.78331263e-03 -8.94680545e-02 -3.45502436e-01 6.74382865e-01
4.87720937e-01 -1.21155217e-01 -4.71721083e-01 -3.13776374e-01
-2.91256994e-01 4.56164144e-02 4.90753174e-01 -1.24603391e+00
1.42148048e-01 -2.44235143e-01 2.10970670e-01 -1.13116197e-01
6.01119220e-01 -1.01019716e+00 9.94899869e-03 1.64538503e-01
7.52864257e-02 -4.21736181e-01 -6.11086655e-03 8.03308785e-01
-4.83375192e-01 -1.44077376e-01 8.68108094e-01 1.36174589e-01
-7.92025924e-01 8.74968529e-01 5.09089381e-02 -2.06154436e-01
1.30053961e+00 -1.88049823e-01 3.05281311e-01 -1.45848632e-01
-6.40071273e-01 1.31239414e-01 4.11162049e-01 7.34987378e-01
9.87011611e-01 -1.59463418e+00 -5.27676642e-01 3.11294615e-01
4.22692150e-01 -1.88121542e-01 7.77262270e-01 1.04752731e+00
-1.19671337e-01 1.94944978e-01 -1.64722562e-01 -9.04703736e-01
-1.27856088e+00 6.74901366e-01 4.76291329e-01 2.55698174e-01
-3.83754432e-01 5.84043443e-01 3.66537988e-01 -4.75988269e-01
-1.37683049e-01 4.23821136e-02 -4.69480485e-01 7.77394474e-02
6.45674825e-01 3.13844681e-01 -1.27383292e-01 -1.15999162e+00
-5.05034149e-01 1.05375266e+00 9.84499156e-02 1.73171833e-01
1.18411291e+00 -2.22740918e-01 -3.33196074e-01 5.36198378e-01
1.74190819e+00 -4.02241349e-01 -8.20790768e-01 -4.84656096e-01
-6.74418509e-02 -6.86943650e-01 2.08462313e-01 1.64933339e-01
-1.25465333e+00 1.09433138e+00 7.18611479e-01 6.23759031e-02
1.17247784e+00 -2.28200793e-01 4.28100616e-01 4.98681426e-01
7.47643948e-01 -8.16181779e-01 2.68714249e-01 2.11909041e-01
1.04668260e+00 -1.41361988e+00 -1.85913727e-01 -6.29222870e-01
-5.64761519e-01 1.20643246e+00 6.37540162e-01 -1.53539434e-01
1.22406805e+00 -5.38322270e-01 -3.22953045e-01 -1.35702878e-01
-2.57162988e-01 -3.56195122e-02 5.11255443e-01 2.45213866e-01
1.11672260e-01 8.26466829e-03 -1.19456090e-01 6.56693041e-01
1.90293193e-01 -1.04239859e-01 2.95991272e-01 6.77651227e-01
-3.45647424e-01 -8.65892887e-01 -4.80217189e-01 5.51017404e-01
-1.72935680e-01 2.93494046e-01 6.06743693e-02 5.33407986e-01
-3.57683077e-02 9.96605098e-01 -1.02991872e-01 -7.93408215e-01
1.26377985e-01 -8.81930664e-02 6.12800717e-01 -6.02047026e-01
3.89007367e-02 -1.65178284e-01 -6.01543546e-01 -4.68858033e-01
-5.82370520e-01 -7.97626734e-01 -1.08462131e+00 4.21991795e-02
-4.40494001e-01 3.05553049e-01 6.94219410e-01 8.96488011e-01
3.82156074e-01 1.78929120e-02 1.16727626e+00 -9.57050979e-01
-6.89145088e-01 -8.81567240e-01 -9.92690444e-01 8.64213109e-01
-5.46684600e-02 -1.20982623e+00 -5.42643011e-01 -3.15952562e-02] | [7.946136474609375, 4.074167728424072] |
b26bbd1b-8351-4a2d-ad31-54ba14676806 | opental-towards-open-set-temporal-action | 2203.05114 | null | https://arxiv.org/abs/2203.05114v1 | https://arxiv.org/pdf/2203.05114v1.pdf | OpenTAL: Towards Open Set Temporal Action Localization | Temporal Action Localization (TAL) has experienced remarkable success under the supervised learning paradigm. However, existing TAL methods are rooted in the closed set assumption, which cannot handle the inevitable unknown actions in open-world scenarios. In this paper, we, for the first time, step toward the Open Set TAL (OSTAL) problem and propose a general framework OpenTAL based on Evidential Deep Learning (EDL). Specifically, the OpenTAL consists of uncertainty-aware action classification, actionness prediction, and temporal location regression. With the proposed importance-balanced EDL method, classification uncertainty is learned by collecting categorical evidence majorly from important samples. To distinguish the unknown actions from background video frames, the actionness is learned by the positive-unlabeled learning. The classification uncertainty is further calibrated by leveraging the guidance from the temporal localization quality. The OpenTAL is general to enable existing TAL models for open set scenarios, and experimental results on THUMOS14 and ActivityNet1.3 benchmarks show the effectiveness of our method. The code and pre-trained models are released at https://www.rit.edu/actionlab/opental. | ['Yu Kong', 'Qi Yu', 'Wentao Bao'] | 2022-03-10 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Bao_OpenTAL_Towards_Open_Set_Temporal_Action_Localization_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Bao_OpenTAL_Towards_Open_Set_Temporal_Action_Localization_CVPR_2022_paper.pdf | cvpr-2022-1 | ['action-localization'] | ['computer-vision'] | [ 2.36894011e-01 1.20398059e-01 -8.24123383e-01 -4.04418051e-01
-1.02847850e+00 -2.35675290e-01 6.43271327e-01 -3.29303980e-01
-7.59817883e-02 8.57302725e-01 3.94110143e-01 2.95625310e-02
-3.40779573e-01 -2.83437312e-01 -7.53942430e-01 -7.52403140e-01
-1.46447226e-01 2.58776367e-01 3.55583102e-01 1.89851269e-01
1.99943185e-01 -1.38331711e-01 -1.35526931e+00 4.64861542e-01
8.68973851e-01 1.63791752e+00 -1.13541521e-01 3.41148913e-01
-1.25328021e-04 1.50344229e+00 -1.86899319e-01 -9.06683803e-02
4.22117501e-01 -3.30842257e-01 -6.77332997e-01 2.98163116e-01
3.11629534e-01 -6.17696285e-01 -4.68835503e-01 8.85252357e-01
3.17756385e-01 2.60759503e-01 5.22121489e-01 -1.87750459e+00
-6.68668628e-01 8.56688619e-01 -6.32885039e-01 3.72568101e-01
1.92825839e-01 4.75526690e-01 1.02508855e+00 -9.76925015e-01
3.30464929e-01 1.41351211e+00 5.63015103e-01 5.03240585e-01
-7.00249374e-01 -6.28212869e-01 7.60172486e-01 7.06225455e-01
-1.11042488e+00 -5.50171375e-01 6.50780261e-01 -4.44330752e-01
4.82911199e-01 -6.46152571e-02 5.15078187e-01 1.44239390e+00
1.77239358e-01 1.42526853e+00 1.20569694e+00 -8.46266448e-02
4.61168379e-01 -3.84448081e-01 2.37128690e-01 5.91611505e-01
3.31529826e-02 2.32368663e-01 -6.33535206e-01 9.66022909e-02
7.58871257e-01 4.85693902e-01 -2.30040163e-01 -5.55139959e-01
-1.28675354e+00 4.13270235e-01 3.10111016e-01 3.11943181e-02
-2.76715875e-01 5.11750102e-01 5.40555120e-01 1.29860669e-01
5.87577701e-01 -2.64041498e-02 -6.71167195e-01 -5.46311796e-01
-6.25433981e-01 -1.60541713e-01 3.11235756e-01 1.06860828e+00
5.79989076e-01 7.20487684e-02 -6.03463531e-01 5.45186460e-01
4.60003585e-01 3.64040613e-01 5.23338139e-01 -1.40028608e+00
5.40364504e-01 5.46958148e-01 1.68462560e-01 -5.70524037e-01
-1.07662395e-01 -3.16208482e-01 -5.59044123e-01 2.09143296e-01
3.31076115e-01 -2.60105461e-01 -9.46621060e-01 1.75384092e+00
2.85512924e-01 8.38694990e-01 8.68138596e-02 7.44097114e-01
4.49957728e-01 5.06233096e-01 8.25055037e-03 -3.90738487e-01
7.90875673e-01 -1.36279380e+00 -1.00123382e+00 -2.50443041e-01
7.74568737e-01 -5.77190556e-02 1.07805145e+00 5.61673224e-01
-7.20407009e-01 -4.92932916e-01 -1.00162780e+00 1.34689942e-01
-2.84079630e-02 2.36479864e-01 6.52075827e-01 2.72945523e-01
-4.75990742e-01 5.25409162e-01 -1.18562770e+00 -1.79646462e-01
1.08884990e+00 3.43160741e-02 -2.08309680e-01 -3.30760866e-01
-1.18495309e+00 5.06737292e-01 5.47423542e-01 2.45102033e-01
-1.43479884e+00 -4.83634323e-01 -9.11424816e-01 -2.13068828e-01
1.24343526e+00 -2.06581220e-01 1.50256085e+00 -1.16797459e+00
-1.50798011e+00 2.34532565e-01 1.76002637e-01 -7.67437577e-01
7.93789029e-01 -5.83472192e-01 -4.35150117e-01 8.32454637e-02
4.38137442e-01 4.85666931e-01 9.52805161e-01 -1.10252285e+00
-7.84139454e-01 -2.76914150e-01 2.87468463e-01 2.31040612e-01
-2.53063470e-01 -3.42434287e-01 -4.19441640e-01 -6.44623697e-01
1.46906851e-02 -9.48900700e-01 -2.79453337e-01 3.25250596e-01
-2.28593528e-01 -3.21906805e-01 7.59987354e-01 -5.17153025e-01
1.33072710e+00 -2.27119517e+00 -7.66066648e-03 -1.97079062e-01
2.02399582e-01 4.35786881e-02 9.06876288e-03 2.51545221e-01
-8.42779502e-03 -4.19421494e-02 -2.46222630e-01 -2.54860431e-01
3.44495565e-01 3.83249998e-01 -4.12824512e-01 4.34363425e-01
1.82204574e-01 9.21534181e-01 -1.18476963e+00 -7.73931980e-01
3.14604849e-01 -6.62887935e-03 -3.98038238e-01 1.10680070e-02
-6.61188900e-01 6.27443552e-01 -8.15192938e-01 1.02811038e+00
4.26102191e-01 -3.78357649e-01 -4.85987812e-02 -1.15847871e-01
4.22082916e-02 -1.42346602e-02 -1.30193889e+00 2.00618649e+00
-1.55096218e-01 4.35320020e-01 -3.61501247e-01 -9.11719859e-01
4.89212215e-01 2.99060464e-01 7.84786463e-01 -4.81785029e-01
1.45993724e-01 1.19476251e-01 -7.42015913e-02 -5.87887943e-01
1.01608917e-01 1.98352233e-01 -9.81315598e-02 3.51731747e-01
1.56999156e-01 3.16327333e-01 1.18294254e-01 3.08081716e-01
1.22322381e+00 8.92669559e-01 3.59328926e-01 8.38783756e-02
4.29628640e-01 -1.04532823e-01 1.16326594e+00 6.94814265e-01
-8.72725010e-01 3.43335658e-01 5.91476917e-01 -2.86164582e-01
-3.01319778e-01 -1.04360592e+00 -1.25021696e-01 9.89832401e-01
4.13686484e-01 -4.75105822e-01 -5.93693018e-01 -1.24362481e+00
-1.18772633e-01 6.08609736e-01 -7.38823473e-01 -2.52019137e-01
-3.47482800e-01 -3.97115380e-01 3.74812901e-01 9.10594523e-01
8.54237258e-01 -1.03110743e+00 -3.28656882e-01 8.66902322e-02
-3.41570824e-01 -1.20188558e+00 -4.94997412e-01 5.35464138e-02
-7.85449743e-01 -1.18994212e+00 -3.91314834e-01 -1.77554026e-01
3.36405396e-01 2.10457310e-01 8.61734271e-01 -2.65578568e-01
-5.10364771e-02 6.82624280e-01 -5.93906820e-01 -4.12813276e-01
1.05167001e-01 -3.77064675e-01 4.06347036e-01 4.73542929e-01
4.59204257e-01 -5.70789397e-01 -6.63562357e-01 5.16328990e-01
-7.10667670e-01 -9.13587958e-02 8.01619172e-01 7.66470850e-01
9.43276525e-01 1.95826218e-01 7.66951442e-01 -4.44273710e-01
9.73864086e-03 -7.94002414e-01 -3.30010414e-01 3.49478066e-01
-6.47118986e-01 1.36270657e-01 1.70751274e-01 -5.46181917e-01
-1.28730309e+00 -2.04678234e-02 1.69449896e-01 -9.13560390e-01
-1.70624688e-01 4.44697469e-01 -5.59156775e-01 2.48570710e-01
3.84100407e-01 1.88009530e-01 -2.33240262e-01 -3.10798973e-01
4.25224543e-01 5.08152723e-01 4.41563010e-01 -7.94386923e-01
4.49884683e-01 7.44689524e-01 -1.70453668e-01 -2.01835409e-01
-1.40902293e+00 -3.10707867e-01 -6.39329135e-01 -5.50281584e-01
8.47361863e-01 -1.16968441e+00 -3.28060418e-01 5.51363170e-01
-6.38889551e-01 -6.06744647e-01 -6.74606204e-01 7.14515746e-01
-8.96582663e-01 5.42319417e-01 -3.69848460e-01 -9.85029638e-01
9.64624211e-02 -1.07275045e+00 1.18263662e+00 8.08726400e-02
2.11459026e-01 -8.17656577e-01 3.09583563e-02 6.57150626e-01
-1.24846458e-01 2.22757384e-01 1.56978697e-01 -7.37450302e-01
-1.07751203e+00 -1.14381336e-01 -2.55579185e-02 5.79864383e-01
2.21174240e-01 -8.06938782e-02 -1.04709864e+00 -1.59632824e-02
-1.59787561e-03 -7.78753877e-01 1.15597534e+00 5.44867456e-01
1.55790639e+00 -1.28327414e-01 -2.41631180e-01 4.29137588e-01
1.08358073e+00 1.96535960e-01 6.09112680e-01 2.91918784e-01
6.73525870e-01 2.49573365e-01 1.46852565e+00 8.16216707e-01
4.30074692e-01 4.84283715e-01 9.16718185e-01 3.43044639e-01
5.39418831e-02 -4.83208269e-01 8.06899607e-01 3.73865336e-01
-6.09442852e-02 -3.52272153e-01 -7.02366233e-01 4.80011046e-01
-2.46624398e+00 -1.28016686e+00 2.98580583e-02 2.00525999e+00
6.77718401e-01 3.42808366e-01 -9.78217926e-03 1.01830803e-01
6.03244841e-01 4.72069353e-01 -9.51642394e-01 3.77575248e-01
-2.67940480e-02 -4.27892238e-01 3.35182905e-01 2.45560750e-01
-1.41360998e+00 8.60089421e-01 5.02234936e+00 1.14574850e+00
-5.95355809e-01 3.36758554e-01 6.59592450e-01 -2.92443633e-01
4.31389958e-02 1.93671286e-01 -7.94802904e-01 6.80224538e-01
6.06583893e-01 -5.31453826e-02 1.08203858e-01 1.01760268e+00
3.46967250e-01 -1.98320001e-01 -1.30559254e+00 1.04785705e+00
4.20915969e-02 -1.17413354e+00 -1.80553406e-01 -1.70977842e-02
7.70771861e-01 4.16903347e-02 1.35725647e-01 8.79134536e-01
4.68939811e-01 -6.81001127e-01 7.76505232e-01 8.11273217e-01
5.54477155e-01 -3.93122166e-01 5.29724181e-01 5.69094837e-01
-1.31324720e+00 -6.64611399e-01 -1.47199839e-01 -2.31162757e-01
3.02143008e-01 5.12674928e-01 -2.75687158e-01 6.34396374e-01
8.49713624e-01 1.55008626e+00 -4.22195613e-01 8.78716171e-01
-4.94136482e-01 8.14031780e-01 -6.74332976e-02 2.93017209e-01
4.06962425e-01 -1.42713606e-01 5.09262979e-01 6.06810212e-01
1.71747863e-01 9.08497125e-02 6.42019212e-01 6.29133046e-01
-3.94248292e-02 -2.68192232e-01 -4.68566626e-01 -2.22969741e-01
3.63436759e-01 9.86567140e-01 -4.28624243e-01 -3.27165544e-01
-4.30227906e-01 8.66904557e-01 2.34024405e-01 2.92871892e-01
-1.35468471e+00 1.63997024e-01 6.14063501e-01 -1.89793587e-01
2.35593423e-01 -2.50067376e-02 5.19327372e-02 -1.39882612e+00
2.79612303e-01 -7.96494603e-01 6.86875165e-01 -9.26771581e-01
-1.37511706e+00 2.74567306e-01 2.51761466e-01 -1.83032203e+00
-2.59888619e-02 -5.44390202e-01 -5.93719840e-01 1.26759440e-01
-1.50425375e+00 -1.13539410e+00 -3.09169769e-01 7.25355268e-01
1.01949847e+00 -1.55799925e-01 3.37954342e-01 2.45159492e-01
-9.64289963e-01 2.76169956e-01 5.29568195e-02 9.08814091e-03
8.38107586e-01 -1.25235403e+00 -1.11816078e-01 8.78795385e-01
1.68059498e-01 1.05814405e-01 3.71673852e-01 -7.80417085e-01
-1.10070455e+00 -1.41360974e+00 1.84608027e-01 -7.77779043e-01
1.14261758e+00 -1.78043962e-01 -6.59191966e-01 1.10095477e+00
1.65834632e-02 5.33449650e-01 6.54816329e-01 -9.54116806e-02
-3.84955198e-01 -1.29044324e-01 -8.37836325e-01 5.60519934e-01
1.44954979e+00 -3.24448228e-01 -6.33192778e-01 5.93623936e-01
9.80716109e-01 -2.58330971e-01 -8.95140946e-01 5.12901604e-01
4.52990890e-01 -8.83970618e-01 7.58530438e-01 -7.50933230e-01
4.78927821e-01 -4.18439686e-01 -4.45088089e-01 -1.05028546e+00
-1.78552464e-01 -5.91057897e-01 -6.90861404e-01 1.23665750e+00
1.69330478e-01 -4.91971374e-01 7.30744481e-01 4.41818297e-01
-4.70207870e-01 -8.51286590e-01 -1.11064005e+00 -1.13090289e+00
-2.34184325e-01 -8.10487032e-01 3.90353173e-01 7.73445845e-01
-3.64260525e-02 2.52962977e-01 -7.35013247e-01 2.38138467e-01
7.50991940e-01 5.57117797e-02 6.57555223e-01 -1.12757266e+00
-5.56381643e-01 -4.03761072e-03 -4.12789315e-01 -1.39434886e+00
3.60782832e-01 -4.86403286e-01 2.57931203e-01 -1.49971187e+00
2.22538099e-01 -5.19622028e-01 -7.35562921e-01 6.94023967e-01
-1.86461806e-01 -1.52640000e-01 1.11827180e-01 3.37782681e-01
-1.52725828e+00 1.13230479e+00 1.18248403e+00 -2.14543968e-01
-3.63673866e-02 1.58923179e-01 -5.29866219e-01 1.15511799e+00
8.98583591e-01 -3.82324278e-01 -9.52113450e-01 -3.70111465e-01
4.92527557e-04 1.14272729e-01 5.73128104e-01 -1.23183048e+00
-1.67849492e-02 -6.20413542e-01 1.01343855e-01 -7.64735520e-01
6.26639545e-01 -1.00463605e+00 -2.59748161e-01 2.59810895e-01
-5.02639830e-01 -4.90704119e-01 -9.25753191e-02 1.29335809e+00
-7.90486857e-02 2.95624528e-02 5.47387540e-01 -1.72806196e-02
-1.24485195e+00 9.31369126e-01 -2.81724110e-02 3.83438200e-01
1.35221243e+00 -2.01334059e-01 -5.05584180e-01 -3.83999556e-01
-9.33353961e-01 7.12441266e-01 1.32307723e-01 4.13422346e-01
6.43763781e-01 -1.58231568e+00 -5.00385463e-01 -2.08825022e-01
4.76919085e-01 3.12350076e-02 4.97559428e-01 1.35089195e+00
1.73748657e-01 1.61214784e-01 -6.44871071e-02 -7.48486638e-01
-8.89484406e-01 7.55033016e-01 4.22671497e-01 -3.09239805e-01
-7.02173412e-01 7.72402406e-01 3.72280002e-01 -2.73127019e-01
5.54761469e-01 -4.00163710e-01 -1.97811976e-01 -2.22762302e-01
5.78437626e-01 5.46831310e-01 -3.43758762e-01 -3.98688287e-01
-3.60549480e-01 1.61239237e-01 9.57818702e-03 -1.43814057e-01
1.16346478e+00 -2.77236402e-01 2.93759584e-01 9.67972934e-01
7.21264899e-01 -2.86065936e-01 -2.04175186e+00 -4.80108798e-01
6.55051693e-02 -6.63049459e-01 1.25879705e-01 -7.60181785e-01
-9.88380313e-01 8.34040940e-01 7.12400615e-01 -1.65553197e-01
1.10572529e+00 -9.74615104e-03 6.57425046e-01 4.62960750e-01
6.76062465e-01 -1.23050666e+00 6.09314978e-01 3.84554446e-01
8.25405896e-01 -1.67089260e+00 -3.97203788e-02 -2.93860763e-01
-9.68093216e-01 6.88280821e-01 1.06690407e+00 3.10075795e-03
8.14640939e-01 7.50129968e-02 -2.02291355e-01 -9.14128423e-02
-1.06999946e+00 -3.61361623e-01 4.06310819e-02 6.03407145e-01
-7.97184557e-02 -3.05094998e-02 -9.00149345e-02 1.04897833e+00
7.24728763e-01 4.95724648e-01 4.05974418e-01 1.18485785e+00
-5.64176619e-01 -8.00097048e-01 -1.82198912e-01 4.59830046e-01
-2.98502833e-01 4.17166054e-02 -2.83516645e-01 6.37724161e-01
4.05703902e-01 9.86504197e-01 -1.65862381e-01 -4.56349075e-01
5.81892431e-02 -3.28918546e-02 3.08443487e-01 -5.71318150e-01
1.19109742e-01 1.86813734e-02 9.83215347e-02 -9.70134735e-01
-6.84727371e-01 -9.18966353e-01 -1.43394768e+00 1.77333698e-01
-4.09784704e-01 -1.06672728e-02 8.31436217e-02 1.16147625e+00
3.24114054e-01 6.98717594e-01 7.24578202e-01 -8.03997934e-01
-9.15197372e-01 -1.02723670e+00 -6.50329530e-01 1.74433872e-01
3.02013010e-01 -1.12920940e+00 -5.49957633e-01 1.12765640e-01] | [8.561345100402832, 0.7451103925704956] |
c703ea97-a5c2-4304-b355-d21d5c1984c1 | joint-bayesian-inference-of-graphical | 2305.19366 | null | https://arxiv.org/abs/2305.19366v1 | https://arxiv.org/pdf/2305.19366v1.pdf | Joint Bayesian Inference of Graphical Structure and Parameters with a Single Generative Flow Network | Generative Flow Networks (GFlowNets), a class of generative models over discrete and structured sample spaces, have been previously applied to the problem of inferring the marginal posterior distribution over the directed acyclic graph (DAG) of a Bayesian Network, given a dataset of observations. Based on recent advances extending this framework to non-discrete sample spaces, we propose in this paper to approximate the joint posterior over not only the structure of a Bayesian Network, but also the parameters of its conditional probability distributions. We use a single GFlowNet whose sampling policy follows a two-phase process: the DAG is first generated sequentially one edge at a time, and then the corresponding parameters are picked once the full structure is known. Since the parameters are included in the posterior distribution, this leaves more flexibility for the local probability models of the Bayesian Network, making our approach applicable even to non-linear models parametrized by neural networks. We show that our method, called JSP-GFN, offers an accurate approximation of the joint posterior, while comparing favorably against existing methods on both simulated and real data. | ['Yoshua Bengio', 'Laurent Charlin', 'Nikolay Malkin', 'Jithendaraa Subramanian', 'Mizu Nishikawa-Toomey', 'Tristan Deleu'] | 2023-05-30 | null | null | null | null | ['bayesian-inference'] | ['methodology'] | [ 1.30465090e-01 3.86714756e-01 -9.49577689e-02 -3.83381337e-01
-2.83363461e-01 -4.92714345e-01 1.06303811e+00 -1.16932757e-01
-2.64570475e-01 8.17583263e-01 2.75109589e-01 -2.12734684e-01
-5.37606359e-01 -1.13849902e+00 -7.43583679e-01 -7.59425044e-01
-3.90619814e-01 1.18923044e+00 4.44831222e-01 5.59058785e-01
7.11471215e-02 6.78788841e-01 -1.14160907e+00 -3.96167219e-01
5.71156263e-01 4.33467209e-01 -2.13012025e-02 1.00560880e+00
-1.06956050e-01 6.72128797e-01 -4.66875553e-01 -4.28135574e-01
3.64658260e-03 -3.69792193e-01 -7.77600050e-01 1.35778993e-01
1.93230972e-01 -4.80228782e-01 -5.18909097e-01 1.20346463e+00
3.84641960e-02 2.60351688e-01 1.24248707e+00 -1.45147634e+00
7.33078420e-02 8.35475743e-01 -3.13282579e-01 9.06506702e-02
-2.53833067e-02 6.09597452e-02 1.05019653e+00 -3.91347200e-01
6.73642337e-01 1.64978731e+00 5.04771471e-01 6.11910596e-02
-1.85605788e+00 -4.49496478e-01 2.93369263e-01 3.59137915e-02
-1.37860107e+00 -2.88845122e-01 6.84297323e-01 -6.10855401e-01
3.32366318e-01 -1.80876747e-01 7.31711686e-01 1.21467745e+00
3.41288775e-01 6.93822682e-01 8.61946046e-01 -3.24624479e-01
7.76350558e-01 -1.77431643e-01 1.85159206e-01 6.60856187e-01
5.35255134e-01 2.00806290e-01 -5.22654116e-01 -6.46220505e-01
9.71344411e-01 1.95594817e-01 2.19107904e-02 -7.44451582e-01
-9.14732337e-01 8.48107040e-01 7.29634315e-02 -4.84144054e-02
-3.78932923e-01 4.73192334e-01 -2.37652157e-02 -1.73609123e-01
3.92320454e-01 -2.75462329e-01 -1.21223293e-02 -1.03482105e-01
-1.06109190e+00 5.02956748e-01 1.47716534e+00 9.27549183e-01
1.06565928e+00 -1.05169050e-01 -2.25652918e-01 4.45371389e-01
8.39801908e-01 4.74600285e-01 -3.41924250e-01 -1.15766513e+00
2.92732537e-01 1.60187557e-01 3.08329672e-01 -8.61760259e-01
-7.76968971e-02 -4.56590563e-01 -8.75299394e-01 1.06557831e-01
8.21441233e-01 -4.49191362e-01 -1.05779612e+00 2.01008415e+00
3.61471444e-01 3.44505042e-01 -2.20587566e-01 3.65032434e-01
6.84544593e-02 7.79214025e-01 9.40958932e-02 -8.38297307e-02
1.03302169e+00 -2.96090811e-01 -2.68098295e-01 -2.30428398e-01
-1.49958193e-01 -4.05258566e-01 3.64040405e-01 4.25898135e-01
-7.92827487e-01 -2.38679767e-01 -7.64126837e-01 3.97418499e-01
-4.26922515e-02 -5.99895827e-02 3.59218836e-01 7.42075562e-01
-1.16113555e+00 8.24693382e-01 -1.37195361e+00 -3.98631096e-01
3.96326512e-01 2.09429681e-01 6.72552139e-02 -4.72303927e-01
-9.28498566e-01 3.84607613e-01 5.04035830e-01 2.65540481e-01
-1.41148245e+00 -4.55528796e-01 -5.74690402e-01 3.76399040e-01
6.57578170e-01 -7.31624544e-01 1.16283989e+00 -3.61276656e-01
-1.49648380e+00 -7.01628625e-02 -3.83038342e-01 -4.18440819e-01
6.44734800e-01 8.52924436e-02 -1.55245895e-02 1.98339060e-01
-1.00160781e-02 5.52961111e-01 9.71539855e-01 -1.11528146e+00
-3.81007403e-01 -2.19413459e-01 3.72837931e-02 -1.12727910e-01
2.23690167e-01 -4.47364748e-01 -5.69169044e-01 -2.64492810e-01
1.14483923e-01 -9.75617588e-01 -4.55598056e-01 7.13452511e-03
-8.02644372e-01 -4.49669868e-01 5.42659104e-01 -4.45327371e-01
9.51305747e-01 -1.96795344e+00 2.19266281e-01 8.07393789e-01
2.54919589e-01 -2.70266831e-01 1.57654792e-01 7.06653118e-01
3.73523265e-01 -7.82764796e-03 -5.08663774e-01 -3.94673556e-01
2.06675395e-01 5.58795810e-01 -1.96843207e-01 6.91503525e-01
1.40419796e-01 3.54147404e-01 -9.77695286e-01 -4.47935343e-01
1.34536669e-01 4.93252873e-01 -5.14633060e-01 2.21303612e-01
-4.27801579e-01 1.70185193e-01 -6.88077688e-01 1.49197318e-03
6.43106997e-01 -3.46698403e-01 5.97023368e-01 1.01577908e-01
1.95815966e-01 1.87887415e-01 -1.71508527e+00 1.23916709e+00
-1.67107448e-01 5.03905058e-01 1.08113371e-01 -9.13784564e-01
9.02790725e-01 3.03008556e-01 3.79504710e-01 2.39826947e-01
-2.88506337e-02 -1.42289683e-01 1.97318316e-01 -9.00539234e-02
2.20920354e-01 -1.78906113e-01 8.84141549e-02 6.97070062e-01
4.37206239e-01 3.79599370e-02 6.11673176e-01 5.95224500e-01
1.02064300e+00 4.12902206e-01 6.70376793e-02 -4.87934500e-01
3.67254205e-02 -4.48543906e-01 6.70588195e-01 1.34350419e+00
3.12044263e-01 4.59432960e-01 1.22030973e+00 -1.98169723e-01
-1.06418800e+00 -1.49064589e+00 -1.71640068e-01 5.07895291e-01
-2.13332221e-01 -4.77943391e-01 -8.17127883e-01 -5.14755309e-01
5.50935902e-02 6.46498382e-01 -6.09601378e-01 4.64669131e-02
-2.73487628e-01 -1.01407456e+00 1.62826911e-01 3.83619964e-01
3.01866859e-01 -7.26278663e-01 -4.13033158e-01 5.61837375e-01
4.24011704e-03 -1.10570467e+00 -2.38680303e-01 1.32861629e-01
-1.11601436e+00 -1.22000313e+00 -4.22174096e-01 -4.02591676e-02
7.56785393e-01 -4.73014712e-01 1.01446688e+00 -4.64199513e-01
-5.10502085e-02 4.56697971e-01 2.27258712e-01 -1.41739726e-01
-6.12549305e-01 2.47302920e-01 -1.87006056e-01 4.02834564e-01
6.78611025e-02 -8.25259030e-01 -3.77038300e-01 1.75552368e-01
-1.00583315e+00 -2.92659570e-02 4.28836882e-01 4.99157995e-01
3.94738048e-01 1.90458119e-01 2.89430439e-01 -1.10636973e+00
4.75538403e-01 -6.77315652e-01 -1.01337290e+00 1.77240670e-01
-5.41569471e-01 4.51530606e-01 4.33635682e-01 -3.84178042e-01
-1.25637746e+00 1.00342010e-03 1.99717909e-01 -1.82653829e-01
-4.50026929e-01 5.51831305e-01 -3.03521961e-01 3.66200924e-01
3.01032543e-01 -5.64256720e-02 1.02913044e-01 -6.89864159e-01
3.38435173e-01 2.65465617e-01 5.43199360e-01 -8.81474614e-01
6.21804893e-01 6.12444818e-01 6.55017257e-01 -8.60938549e-01
-5.89533925e-01 -1.55422404e-01 -7.12595642e-01 -2.92457521e-01
5.66746235e-01 -6.27907157e-01 -7.61816323e-01 4.55223858e-01
-1.01866972e+00 -3.94828826e-01 -2.89630562e-01 7.01438487e-01
-5.42211592e-01 2.89186418e-01 -5.09857118e-01 -1.14260244e+00
3.79908860e-01 -9.76538777e-01 7.03104854e-01 2.87249029e-01
-1.30647600e-01 -1.35437405e+00 4.93144989e-01 -3.09025377e-01
1.93185776e-01 1.53934166e-01 9.96558785e-01 -5.95907271e-01
-9.53303933e-01 -2.80347139e-01 -1.02370322e-01 2.56374627e-01
-2.36317720e-02 6.07307196e-01 -7.17013419e-01 -2.35093325e-01
-3.09103489e-01 2.53097206e-01 9.27862585e-01 7.94321895e-01
9.23131108e-01 -3.27955067e-01 -5.77392042e-01 2.84854412e-01
1.36995506e+00 -8.77614468e-02 5.35018682e-01 -2.37369552e-01
4.89349514e-01 5.22128046e-01 -5.95419295e-02 5.93542099e-01
3.92158866e-01 3.66845042e-01 4.41898674e-01 3.10340405e-01
7.50874877e-02 -6.29345238e-01 2.61679351e-01 2.66034544e-01
1.08036257e-01 -5.78567266e-01 -8.80099118e-01 7.27549970e-01
-1.82991970e+00 -8.78746808e-01 4.00209166e-02 2.30325961e+00
5.84803402e-01 3.90130430e-01 4.18838114e-01 -1.82010621e-01
9.79070127e-01 2.85014510e-01 -6.08263016e-01 -1.30004687e-02
2.84755230e-01 1.65502638e-01 5.28674960e-01 8.43962669e-01
-7.47331440e-01 3.41360360e-01 7.07906866e+00 6.41271770e-01
-4.82453197e-01 -1.73863962e-01 5.69497347e-01 1.44517824e-01
-4.53755170e-01 6.24736667e-01 -1.09637904e+00 5.76500356e-01
1.23694599e+00 -3.37104611e-02 3.96327794e-01 5.03116906e-01
8.66420493e-02 -5.10755360e-01 -1.18110061e+00 2.09534779e-01
-3.38922620e-01 -1.10715318e+00 3.96471983e-03 4.86255080e-01
6.56693578e-01 1.65187210e-01 -3.05669099e-01 -1.10868558e-01
1.26131785e+00 -6.77828252e-01 6.07422173e-01 9.95624363e-01
3.24341387e-01 -8.52747917e-01 4.71963108e-01 5.47198534e-01
-8.61772299e-01 1.40686527e-01 -2.55555153e-01 1.46911055e-01
3.20586234e-01 1.12925184e+00 -1.17601418e+00 4.07774717e-01
3.90257210e-01 6.93295836e-01 -2.01302290e-01 1.24922526e+00
-3.75719726e-01 1.18985856e+00 -9.24377143e-01 -1.05403408e-01
1.24540009e-01 -4.72024471e-01 7.47165382e-01 1.03421676e+00
1.94899842e-01 -4.10541534e-01 3.61501217e-01 1.24971783e+00
3.15223932e-02 -4.92780238e-01 -3.30494285e-01 -1.22123115e-01
6.44675016e-01 1.08566451e+00 -1.14143097e+00 -3.12440008e-01
-1.08722635e-01 1.51695758e-01 2.74442166e-01 8.43355536e-01
-3.66622359e-01 -2.45775744e-01 3.97960186e-01 2.72916965e-02
6.69300914e-01 -3.30112934e-01 2.13816375e-01 -1.02394676e+00
-1.71279833e-01 -2.98139244e-01 4.49712396e-01 -4.60184455e-01
-1.33228374e+00 1.78194389e-01 7.77759492e-01 -6.07285261e-01
-9.04460073e-01 -4.11993325e-01 -6.40330017e-01 1.00936794e+00
-1.18426323e+00 -6.41322613e-01 1.24726959e-01 2.94343084e-01
6.85034618e-02 1.51988000e-01 5.29144347e-01 -1.86896592e-01
-5.52346230e-01 -2.02350095e-01 3.57532054e-01 2.48013467e-01
1.64357319e-01 -1.42741156e+00 6.06058240e-01 1.03094399e+00
2.82595366e-01 4.77942765e-01 9.66503501e-01 -9.48460877e-01
-1.01620996e+00 -8.37059379e-01 5.63714504e-01 -1.90204516e-01
6.57985747e-01 -6.15015030e-01 -8.02806556e-01 9.03124630e-01
-2.48692315e-02 7.15790763e-02 3.08002681e-01 3.50981265e-01
-1.82645053e-01 -1.26574248e-01 -9.99436200e-01 3.52476060e-01
6.87428057e-01 -2.16893211e-01 -6.57626316e-02 8.15057680e-02
2.17078328e-01 -9.03880522e-02 -7.96027064e-01 6.89746737e-02
5.19326627e-01 -1.00555539e+00 6.72849894e-01 -5.36773086e-01
9.65065658e-02 -4.37513471e-01 -4.74030189e-02 -1.31649470e+00
-3.26741189e-01 -8.08352113e-01 -3.36126417e-01 1.25220227e+00
3.04526508e-01 -7.09812522e-01 7.63585806e-01 4.70069736e-01
3.88698965e-01 -3.29103768e-01 -1.07193410e+00 -6.10992134e-01
-8.95595700e-02 -4.55952853e-01 6.91677034e-01 2.29983285e-01
-5.28246999e-01 2.07777992e-01 -3.65424901e-01 3.63953859e-01
1.23234737e+00 -2.41103657e-02 7.07966506e-01 -1.78979886e+00
-6.18795812e-01 -1.87458113e-01 -3.54081124e-01 -1.14441180e+00
2.22426653e-01 -5.40235162e-01 1.92837253e-01 -1.45666933e+00
4.10297513e-01 -3.46581489e-01 1.67889938e-01 1.27275333e-01
1.89459871e-03 -3.52945566e-01 -2.22711526e-02 1.32869139e-01
-3.56419206e-01 4.11206961e-01 9.16807115e-01 2.38531202e-01
4.89474051e-02 5.02578974e-01 -3.40193957e-01 7.48272061e-01
4.14105892e-01 -7.47344971e-01 -6.32049918e-01 -6.87047541e-02
2.13537380e-01 3.56026322e-01 5.90968788e-01 -7.51811743e-01
3.61713469e-01 -2.13561952e-01 3.20145011e-01 -6.50432646e-01
1.84181556e-01 -6.61496222e-01 5.71164668e-01 4.35173213e-01
-3.64686608e-01 -2.24808306e-01 -3.07458848e-01 1.10882056e+00
8.84770527e-02 -5.40061474e-01 6.12505972e-01 -7.50976354e-02
-6.65189251e-02 5.23358464e-01 -6.87602341e-01 1.99355751e-01
4.51012284e-01 1.71636477e-01 -1.51837602e-01 -7.28229165e-01
-1.10197854e+00 1.60181731e-01 2.37960413e-01 -1.20907612e-01
1.92450508e-01 -1.08671498e+00 -6.24616683e-01 2.77113408e-01
-3.21964324e-01 3.54972899e-01 1.37680322e-01 5.74418783e-01
-2.63988018e-01 2.05723599e-01 9.02230069e-02 -8.49545836e-01
-6.31803453e-01 1.11199036e-01 4.23760235e-01 -4.02205139e-01
-6.40327096e-01 3.97741765e-01 -1.17888320e-02 -2.90310770e-01
2.31615335e-01 -2.74268448e-01 3.14959586e-02 -6.85349703e-02
2.87006289e-01 6.71157420e-01 -2.90916115e-01 -2.54273921e-01
-3.54037397e-02 1.37291372e-01 -1.95594013e-01 -5.77950895e-01
1.34642601e+00 -1.09290041e-01 -2.16165736e-01 6.94252491e-01
8.44090879e-01 -3.09665442e-01 -1.88949752e+00 -5.47700465e-01
-1.06145509e-01 -3.79418194e-01 3.41470800e-02 -4.39328521e-01
-9.70748425e-01 7.68851340e-01 6.05773330e-02 5.44258893e-01
4.92251962e-01 2.65765369e-01 -2.94377804e-02 1.72287017e-01
3.00204068e-01 -7.32714355e-01 -1.85878232e-01 3.75392646e-01
2.18314141e-01 -5.19858837e-01 1.05181091e-01 -3.64510119e-01
-4.44512554e-02 1.17056823e+00 9.63535160e-02 -4.81703520e-01
1.01921415e+00 3.88434887e-01 -6.21716440e-01 -7.24049062e-02
-8.44434679e-01 1.74120739e-01 1.69757813e-01 7.05703676e-01
3.47335194e-03 7.34165013e-02 3.01059067e-01 3.92281683e-03
-5.84237613e-02 1.59160703e-01 8.18131149e-01 7.25531042e-01
-3.61148864e-01 -1.18327796e+00 -3.01017344e-01 6.19954228e-01
-3.30103070e-01 1.96854129e-01 -3.40576395e-02 7.47178078e-01
-2.30756819e-01 6.15260720e-01 4.22368526e-01 2.98122019e-01
-5.99454949e-03 1.35875836e-01 5.80765903e-01 -5.22990227e-01
2.62445450e-01 3.73533964e-01 1.23934038e-01 -3.69325966e-01
-4.33412045e-01 -1.19931901e+00 -7.00339377e-01 -3.85838509e-01
-1.85903564e-01 3.30275685e-01 6.12373710e-01 1.20651722e+00
2.06791729e-01 4.21676010e-01 4.10668224e-01 -7.65852153e-01
-7.77536392e-01 -1.06136250e+00 -9.05931950e-01 -2.09186390e-01
3.63562346e-01 -7.56331444e-01 -4.94813263e-01 7.27380300e-03] | [6.925859451293945, 4.280847072601318] |
66def7bd-3e10-4cca-aa14-956b02ee78e4 | guaranteed-non-convex-optimization-submodular | 1606.05615 | null | https://arxiv.org/abs/1606.05615v5 | https://arxiv.org/pdf/1606.05615v5.pdf | Guaranteed Non-convex Optimization: Submodular Maximization over Continuous Domains | Submodular continuous functions are a category of (generally) non-convex/non-concave functions with a wide spectrum of applications. We characterize these functions and demonstrate that they can be maximized efficiently with approximation guarantees. Specifically, i) We introduce the weak DR property that gives a unified characterization of submodularity for all set, integer-lattice and continuous functions; ii) for maximizing monotone DR-submodular continuous functions under general down-closed convex constraints, we propose a Frank-Wolfe variant with $(1-1/e)$ approximation guarantee, and sub-linear convergence rate; iii) for maximizing general non-monotone submodular continuous functions subject to box constraints, we propose a DoubleGreedy algorithm with $1/3$ approximation guarantee. Submodular continuous functions naturally find applications in various real-world settings, including influence and revenue maximization with continuous assignments, sensor energy management, multi-resolution data summarization, facility location, etc. Experimental results show that the proposed algorithms efficiently generate superior solutions compared to baseline algorithms. | ['Baharan Mirzasoleiman', 'Andreas Krause', 'Joachim M. Buhmann', 'Andrew An Bian'] | 2016-06-17 | null | null | null | null | ['data-summarization'] | ['miscellaneous'] | [ 1.51432574e-01 3.32819611e-01 -5.45387506e-01 -4.87327397e-01
-9.34133530e-01 -1.04529858e+00 -4.62667346e-01 2.59336293e-01
-5.46373315e-02 1.19609261e+00 2.31376722e-01 1.16255119e-01
-8.78717244e-01 -8.77172232e-01 -1.14997172e+00 -8.99511278e-01
-4.86997247e-01 8.05642068e-01 -2.97159255e-01 -3.20941448e-01
3.80519144e-02 4.59273726e-01 -1.11923444e+00 -2.28018820e-01
1.08496737e+00 1.23693419e+00 5.13912499e-01 2.91157395e-01
1.84112772e-01 4.30203468e-01 -3.97732556e-01 5.75085059e-02
7.07282066e-01 -3.57732959e-02 -6.54183030e-01 4.86828476e-01
2.82578558e-01 -3.93857956e-01 -3.01502526e-01 1.22019696e+00
2.86824495e-01 9.61459056e-02 3.75003189e-01 -1.56669831e+00
-8.02066982e-01 8.48234117e-01 -1.16606748e+00 -2.56208908e-02
2.15089560e-01 -3.87340754e-01 1.43256950e+00 -6.48697138e-01
6.60423517e-01 1.32821500e+00 3.83573443e-01 2.69030988e-01
-1.17307532e+00 -5.05316019e-01 3.93566251e-01 -1.21528767e-01
-1.42685604e+00 -1.53895468e-01 4.03394610e-01 1.50247008e-01
7.54175305e-01 9.33575749e-01 3.16656411e-01 -8.01529661e-02
1.45659789e-01 9.11970973e-01 5.92697978e-01 1.20183118e-01
2.28218645e-01 1.54084012e-01 1.03888065e-01 5.00249445e-01
9.07885313e-01 -5.68803251e-01 -1.87056839e-01 -4.72299695e-01
3.24000508e-01 3.22367966e-01 -7.05595553e-01 -6.26869977e-01
-1.08353388e+00 8.71057272e-01 3.69680464e-01 -2.90254187e-02
-5.70617557e-01 6.60465002e-01 1.66376621e-01 6.30257726e-01
1.96134210e-01 3.50349456e-01 -5.93116403e-01 2.37928063e-01
-8.80174339e-01 6.50000989e-01 8.89332831e-01 1.75707364e+00
5.89346766e-01 -5.26525192e-02 -4.54519242e-01 7.08100379e-01
-2.24355701e-03 1.01804936e+00 -3.69378865e-01 -1.11359894e+00
9.86630738e-01 5.86246550e-01 5.86983621e-01 -1.06376994e+00
-6.70593202e-01 -2.94378161e-01 -8.72879028e-01 -4.21950877e-01
9.55519034e-04 -2.26916403e-01 -3.91970903e-01 1.78049505e+00
3.19683582e-01 -6.73659265e-01 -8.83536860e-02 8.57919157e-01
5.33667743e-01 1.14017069e+00 -7.25147486e-01 -1.29239190e+00
9.88714337e-01 -7.66319513e-01 -1.00914812e+00 1.56501383e-01
4.43115264e-01 -2.48891890e-01 7.00352073e-01 5.66777229e-01
-1.60545528e+00 3.89915079e-01 -1.00411820e+00 -1.77141968e-02
-3.36759835e-02 -3.05481017e-01 7.26800025e-01 7.34978139e-01
-1.05047107e+00 -3.11479475e-02 -3.39140058e-01 -1.36999995e-01
4.08475876e-01 8.14428270e-01 -4.96641435e-02 -5.06312132e-01
-4.25501376e-01 3.12443286e-01 3.99804831e-01 -1.97572280e-02
-8.94867063e-01 -8.37884784e-01 -7.18411386e-01 2.24940568e-01
8.17787647e-01 -5.14776409e-01 1.00044191e+00 -2.24116161e-01
-7.78793573e-01 6.98590875e-01 -2.28526756e-01 -2.59676516e-01
4.07635331e-01 8.39193836e-02 -9.86134782e-02 8.36521611e-02
9.06370133e-02 2.80904591e-01 3.55921656e-01 -1.46651149e+00
-8.02773476e-01 -9.23153758e-01 4.35273707e-01 5.89889467e-01
-7.53289282e-01 -2.15767965e-01 3.33269797e-02 -1.64216489e-01
2.74994791e-01 -5.35385013e-01 -5.10956883e-01 -2.80468427e-02
-4.82729048e-01 -2.46598616e-01 8.09553862e-01 -4.99190003e-01
1.36703575e+00 -1.94700503e+00 4.00371611e-01 5.52691877e-01
3.17357808e-01 -8.42747390e-01 -6.73755482e-02 6.08943522e-01
3.89551908e-01 -3.92788500e-02 -4.95501608e-01 -1.09464586e-01
3.95474017e-01 4.77811873e-01 -1.36243045e-01 9.35152709e-01
-5.87139308e-01 9.80083108e-01 -9.00992930e-01 -8.10444877e-02
-3.46254915e-01 -3.48986715e-01 -4.96075362e-01 -1.08439274e-01
-4.43375766e-01 -2.71308661e-01 -5.76532066e-01 1.21944308e+00
1.22656119e+00 -2.72467107e-01 5.06976545e-01 1.72785670e-01
4.94377455e-03 -5.83635747e-01 -1.60335290e+00 1.50151002e+00
-4.56706852e-01 1.32295460e-01 1.07319391e+00 -1.34391069e+00
9.94007647e-01 -5.48484400e-02 1.01341081e+00 -4.35323328e-01
7.58386105e-02 5.09001970e-01 -6.63863420e-01 -2.81494766e-01
8.30758870e-01 -1.95621192e-01 -3.99830341e-01 5.09427667e-01
-4.56822455e-01 -8.21411833e-02 3.60146940e-01 2.31852114e-01
1.16067445e+00 -7.89656579e-01 3.75579774e-01 -9.60586190e-01
4.42731172e-01 -6.99199513e-02 7.24150002e-01 6.38708055e-01
2.44301766e-01 5.88455081e-01 6.60662830e-01 -1.15538567e-01
-6.93305016e-01 -1.06825030e+00 -8.32825750e-02 1.34340656e+00
6.75356984e-01 1.39480889e-01 -4.07528341e-01 -3.77008945e-01
8.13372731e-01 6.97032392e-01 -4.28722620e-01 2.76425779e-01
-5.12306631e-01 -1.11936653e+00 -1.84732422e-01 3.62684876e-01
1.88386068e-01 -5.23226261e-01 -2.01854974e-01 3.45582485e-01
-6.17656037e-02 -9.56994772e-01 -1.35973299e+00 3.95115674e-01
-1.13757801e+00 -8.80468905e-01 -7.26112664e-01 -1.00663519e+00
1.18158722e+00 9.21830654e-01 9.91734684e-01 -1.91237509e-01
-1.96368366e-01 8.16563249e-01 -1.01903774e-01 -3.76009643e-01
3.48048180e-01 -3.02114010e-01 2.28872940e-01 6.23402279e-03
-7.79337361e-02 -4.57618386e-01 -5.34654200e-01 4.43479091e-01
-1.14016187e+00 -6.67501986e-01 3.45687091e-01 4.10664558e-01
1.25528765e+00 2.79347926e-01 9.42089319e-01 -9.08294320e-01
8.76911819e-01 -9.06455815e-01 -1.17690420e+00 5.79420924e-01
-7.36218214e-01 1.13518544e-01 7.18783200e-01 -4.00400609e-02
-7.74715126e-01 8.67576078e-02 5.21971464e-01 -1.71434641e-01
8.72275293e-01 3.67130309e-01 -5.95515311e-01 -3.91876400e-02
2.85224229e-01 5.36988497e-01 -7.49797374e-02 -2.41728425e-01
4.52507615e-01 7.43211031e-01 5.28887570e-01 -8.48168612e-01
8.83404911e-01 9.30555820e-01 3.50702047e-01 -6.80215538e-01
-7.58007944e-01 -7.36955106e-01 5.64729162e-02 1.00799881e-01
3.72481227e-01 -8.95484507e-01 -1.11911559e+00 -4.63779420e-02
-1.00397050e+00 1.19995706e-01 -5.71374595e-01 -7.65380859e-02
-7.80104816e-01 3.63243073e-01 -8.10799468e-03 -1.25208688e+00
-7.49044776e-01 -6.86523318e-01 8.70356083e-01 2.30422303e-01
4.51748163e-01 -7.93695271e-01 -1.75927207e-01 5.04812539e-01
3.04782838e-01 6.60456896e-01 7.57410407e-01 -2.31737465e-01
-1.05715990e+00 3.02630272e-02 -2.63774574e-01 2.49216139e-01
1.78259164e-01 -7.09044993e-01 -2.53824770e-01 -8.85874629e-01
-9.49247926e-02 -3.31408739e-01 5.71936905e-01 9.01853263e-01
1.64591205e+00 -9.57303643e-01 -5.10344803e-01 8.71550262e-01
1.65933728e+00 1.96566448e-01 3.70582163e-01 -2.67865565e-02
3.23151439e-01 4.59810585e-01 1.02195787e+00 1.16745114e+00
7.00396001e-01 2.93517739e-01 1.02083850e+00 2.51428485e-01
8.31791461e-01 1.27586396e-02 3.70047420e-01 5.66615224e-01
2.14783207e-01 -6.39421642e-01 2.64894566e-03 1.08016109e+00
-2.22980118e+00 -9.48417246e-01 -3.55786204e-01 2.41972184e+00
6.77143097e-01 -2.86210835e-01 4.85244185e-01 -4.89556082e-02
8.66310000e-01 2.60038167e-01 -8.82242739e-01 -6.81773365e-01
-5.64185262e-01 -1.85250878e-01 1.36476719e+00 4.12276864e-01
-5.32293737e-01 1.48262098e-01 6.43352270e+00 7.43338764e-01
-4.11471158e-01 2.81409770e-01 6.92060828e-01 -7.86211193e-01
-1.23202586e+00 -5.52149117e-01 -9.34437037e-01 3.62447828e-01
2.76578248e-01 -9.14056420e-01 8.93105388e-01 9.67562735e-01
4.01524454e-01 -5.05065545e-02 -1.19645429e+00 1.33093810e+00
-2.70266384e-02 -1.50107169e+00 -3.53743136e-01 5.77740610e-01
1.50346720e+00 -1.64084256e-01 -5.10091372e-02 -3.92856821e-02
1.19439192e-01 -1.13114142e+00 4.72062230e-01 4.12594788e-02
8.31349909e-01 -1.26289093e+00 5.29859066e-01 4.42079872e-01
-1.37495208e+00 -7.27606833e-01 -8.89465332e-01 2.22389817e-01
4.83536690e-01 9.31066990e-01 -4.80936021e-01 8.23713124e-01
6.71240985e-01 3.56014937e-01 5.15988290e-01 1.16661191e+00
5.30947864e-01 -7.69253867e-03 -8.12951148e-01 -4.19216275e-01
2.98643708e-01 -4.44267482e-01 6.92424834e-01 1.02457738e+00
5.05193293e-01 4.23296541e-01 2.87747920e-01 8.86874020e-01
-5.58152974e-01 2.76102751e-01 -6.37686253e-01 -8.06259960e-02
6.42356098e-01 1.44334972e+00 -4.91274923e-01 3.27289343e-01
-2.53218144e-01 7.13553548e-01 -1.58968419e-02 1.90502465e-01
-1.00240421e+00 -4.17751372e-01 8.41361403e-01 4.12580132e-01
4.17053014e-01 -2.57417679e-01 -7.17057109e-01 -6.21874869e-01
6.39983773e-01 -3.54204178e-01 8.55139434e-01 -1.36547565e-01
-1.06620622e+00 -5.62583581e-02 2.16345817e-01 -7.63285637e-01
2.79013067e-01 -4.09034759e-01 -4.29385364e-01 3.71671796e-01
-1.67261493e+00 -6.70032322e-01 -4.44800556e-01 7.24409103e-01
2.04064339e-01 -1.43769190e-01 1.99831054e-01 3.55145961e-01
-9.91399959e-02 5.52456081e-01 7.52674758e-01 -8.69079173e-01
2.76359431e-02 -1.53093290e+00 -5.04463792e-01 5.34334898e-01
-4.71529096e-01 2.14680761e-01 7.34694541e-01 -3.17804426e-01
-2.28113866e+00 -1.22034764e+00 5.69348216e-01 -2.21703853e-02
4.09366190e-01 -3.78128976e-01 -2.92798787e-01 6.62086904e-01
-4.86739278e-02 1.35551333e-01 4.23240483e-01 -2.25892827e-01
2.96167374e-01 -9.01272237e-01 -1.89095652e+00 7.41675869e-02
1.57365704e+00 3.62911344e-01 7.85710067e-02 9.96275842e-01
1.10753369e+00 -4.25261348e-01 -1.00396633e+00 4.33309019e-01
2.71710098e-01 -4.31488603e-01 9.06130612e-01 -3.43460798e-01
2.05708435e-03 -3.38473380e-01 -9.06381845e-01 -9.21728969e-01
-4.24241632e-01 -1.21942449e+00 -6.42085075e-01 8.38852167e-01
4.25177008e-01 -7.84624338e-01 8.29359412e-01 4.28098321e-01
-1.91097423e-01 -1.15131366e+00 -1.10450304e+00 -1.18530965e+00
-8.16026703e-02 4.85602543e-02 7.12315619e-01 6.71640217e-01
2.27796689e-01 -1.30087808e-01 -5.61358929e-01 3.70410472e-01
9.61968541e-01 6.23080373e-01 5.76650977e-01 -9.07591522e-01
-4.05618638e-01 -2.93841977e-02 1.13279326e-03 -1.74438405e+00
-2.45268434e-01 -1.11001110e+00 8.55596066e-02 -1.84932375e+00
6.88145459e-01 -6.01340234e-01 -2.38444611e-01 4.00485337e-01
4.25814718e-01 -1.08867466e-01 2.93307635e-03 -1.75217450e-01
-1.05282354e+00 6.98614061e-01 1.47094750e+00 -2.55269080e-01
-3.58673394e-01 2.47973904e-01 -1.36361825e+00 2.86843516e-02
4.33510154e-01 -3.94753218e-01 -7.63541639e-01 -4.89991546e-01
4.71515328e-01 5.36923945e-01 -2.76673347e-01 -3.06727648e-01
8.07789713e-02 -8.86831522e-01 -2.55277663e-01 -9.56978858e-01
1.66482717e-01 -1.24858797e+00 2.33805299e-01 5.12578487e-01
1.72614887e-01 1.77770957e-01 -1.34890839e-01 7.16761887e-01
5.01027405e-02 -4.74793166e-01 8.21765840e-01 -6.96045021e-03
-2.95634151e-01 7.88109362e-01 2.45675221e-01 3.49884540e-01
1.75803387e+00 -3.95463467e-01 -4.64118838e-01 -8.11641812e-01
-3.44710946e-01 1.21023345e+00 1.79400027e-01 1.82196081e-01
6.39193892e-01 -1.30131865e+00 -9.17828918e-01 -5.21598577e-01
-4.93939072e-02 5.03097832e-01 2.89596319e-01 9.77858424e-01
-5.43807328e-01 6.96631074e-01 1.88015357e-01 -5.03032386e-01
-1.02252138e+00 4.30073738e-01 1.84015319e-01 -4.08179075e-01
-3.04671079e-01 8.79593372e-01 3.01043123e-01 -5.24546504e-01
4.20537055e-01 -3.75833899e-01 4.47918385e-01 1.18301652e-01
3.65050077e-01 9.07290936e-01 -9.74434614e-02 8.92464891e-02
-4.96787816e-01 2.73635298e-01 -3.29370871e-02 3.00104171e-01
1.97872198e+00 -3.42539608e-01 -6.00250125e-01 -7.96655864e-02
1.26433218e+00 5.34568131e-02 -9.60651636e-01 -2.47960806e-01
-1.59290120e-01 -5.87862551e-01 -1.65896967e-01 -6.02566719e-01
-1.31315231e+00 -1.84202462e-01 9.00008455e-02 5.56192219e-01
1.54343832e+00 1.90852493e-01 1.06911862e+00 6.49472117e-01
1.01024342e+00 -1.41722059e+00 -2.14073151e-01 6.52754605e-02
1.33580124e+00 -8.81911039e-01 4.13303882e-01 -7.04844296e-01
-4.04368103e-01 1.09218216e+00 4.79288608e-01 -1.79744884e-01
6.30703211e-01 5.21338761e-01 -9.01232719e-01 -1.94363087e-01
-7.29913890e-01 1.22198716e-01 2.46582497e-02 2.90659636e-01
-7.49669448e-02 4.84995216e-01 -9.02880549e-01 8.88605475e-01
-1.20437473e-01 -1.81980118e-01 7.94477224e-01 8.32940459e-01
-1.07412791e+00 -6.46447480e-01 -6.06530428e-01 8.15787435e-01
-1.96754098e-01 2.28138193e-01 -7.23392814e-02 5.28889656e-01
-2.91848212e-01 9.99491692e-01 -2.69915685e-02 1.40509352e-01
3.93993586e-01 -7.63635695e-01 7.34091938e-01 -4.54927593e-01
-2.30972633e-01 -1.52714059e-01 1.06427604e-02 -4.15678024e-01
-1.95269976e-02 -8.19105089e-01 -1.61720610e+00 -4.71360117e-01
-2.81464219e-01 1.38200507e-01 5.49881995e-01 5.23618042e-01
1.64060265e-01 2.18301535e-01 1.21214855e+00 -1.69466466e-01
-1.44027102e+00 -5.64560294e-01 -1.29698420e+00 -2.72728540e-02
2.60975152e-01 -3.13115388e-01 -1.59014314e-01 -4.90062833e-01] | [6.579762935638428, 4.918125629425049] |
c8c08455-99ee-4f8d-ab0b-18432391b931 | active-source-free-domain-adaptation | 2205.10711 | null | https://arxiv.org/abs/2205.10711v1 | https://arxiv.org/pdf/2205.10711v1.pdf | Active Source Free Domain Adaptation | Source free domain adaptation (SFDA) aims to transfer a trained source model to the unlabeled target domain without accessing the source data. However, the SFDA setting faces an effect bottleneck due to the absence of source data and target supervised information, as evidenced by the limited performance gains of newest SFDA methods. In this paper, for the first time, we introduce a more practical scenario called active source free domain adaptation (ASFDA) that permits actively selecting a few target data to be labeled by experts. To achieve that, we first find that those satisfying the properties of neighbor-chaotic, individual-different, and target-like are the best points to select, and we define them as the minimum happy (MH) points. We then propose minimum happy points learning (MHPL) to actively explore and exploit MH points. We design three unique strategies: neighbor ambient uncertainty, neighbor diversity relaxation, and one-shot querying, to explore the MH points. Further, to fully exploit MH points in the learning process, we design a neighbor focal loss that assigns the weighted neighbor purity to the cross-entropy loss of MH points to make the model focus more on them. Extensive experiments verify that MHPL remarkably exceeds the various types of baselines and achieves significant performance gains at a small cost of labeling. | ['Yilong Yin', 'Zhiyan Zhang', 'Zhongyi Han', 'Fan Wang'] | 2022-05-22 | null | null | null | null | ['source-free-domain-adaptation'] | ['computer-vision'] | [-1.96863934e-02 1.63049594e-01 -7.51212895e-01 -4.44924682e-01
-1.26805031e+00 -6.34598076e-01 6.03816986e-01 -2.27932930e-02
-1.60085693e-01 8.36903691e-01 2.18032226e-01 4.14886922e-02
-1.89459994e-01 -5.76340973e-01 -6.69724703e-01 -9.37704563e-01
1.89579621e-01 6.67614639e-01 3.33068430e-01 -2.04427149e-02
8.09590518e-02 3.80456835e-01 -1.17670047e+00 4.50246211e-04
1.10121357e+00 1.05466270e+00 2.02640682e-01 -1.41795874e-01
-2.47323886e-01 5.43976784e-01 -4.34167832e-01 -2.92546391e-01
3.71780813e-01 -5.96693218e-01 -6.68585956e-01 1.74098089e-01
8.83930698e-02 -4.58796710e-01 2.58155260e-02 9.77934599e-01
6.09858274e-01 4.46236044e-01 7.50314415e-01 -1.41827250e+00
-7.62578964e-01 6.42294586e-01 -6.93059564e-01 1.93535477e-01
1.25449076e-01 1.20941602e-01 9.22362566e-01 -1.26395535e+00
5.97405255e-01 1.22998250e+00 4.43716496e-01 7.44158030e-01
-1.23559129e+00 -8.07202995e-01 4.83047366e-01 1.09933056e-02
-1.34985423e+00 -7.19317436e-01 1.09657097e+00 -2.47710139e-01
3.47608447e-01 -9.43503976e-02 1.37994349e-01 1.30817235e+00
-3.72346073e-01 9.94239867e-01 8.75383914e-01 -5.82713842e-01
6.97595954e-01 4.81929064e-01 1.38746902e-01 4.76242065e-01
3.59975249e-02 1.08606569e-01 -5.82463384e-01 -3.89649600e-01
6.42689645e-01 2.04947740e-02 -3.13779235e-01 -9.53580976e-01
-1.29045904e+00 1.03874254e+00 4.50448185e-01 8.51019025e-02
-2.66730517e-01 -5.25723875e-01 1.60472468e-01 3.22413653e-01
5.74986577e-01 5.03614604e-01 -6.72410488e-01 1.89085037e-01
-6.79276586e-01 3.64682749e-02 5.47508001e-01 1.20585525e+00
1.01914895e+00 -2.25127771e-01 -2.78401494e-01 1.06983912e+00
3.11125457e-01 4.60699201e-01 6.29644275e-01 -1.01051748e+00
6.10826313e-01 7.64731824e-01 1.94440931e-01 -6.07786655e-01
6.94532767e-02 -5.22038341e-01 -6.07014060e-01 8.02976713e-02
3.02226633e-01 -2.10440904e-01 -9.36667502e-01 1.99395311e+00
7.06516504e-01 2.64903963e-01 2.43851289e-01 1.01003349e+00
4.73026067e-01 7.01849461e-01 1.23027787e-01 -3.69603604e-01
8.69935334e-01 -1.07506335e+00 -3.42163444e-01 -4.63629454e-01
7.42146432e-01 -3.57221693e-01 1.33603585e+00 1.33306921e-01
-7.23293483e-01 -4.77810353e-01 -1.04383028e+00 1.54207826e-01
-3.10293704e-01 5.75641654e-02 2.66882986e-01 3.79830569e-01
-5.67189932e-01 4.32858139e-01 -6.55502558e-01 -2.39477456e-01
6.92624986e-01 1.98542610e-01 -2.65499979e-01 -2.90931016e-01
-1.30276763e+00 5.68797350e-01 4.78223801e-01 -4.32933718e-01
-1.16493607e+00 -9.51452494e-01 -8.42871964e-01 -1.53493032e-01
7.27687955e-01 -5.21226048e-01 1.18444574e+00 -1.14223456e+00
-1.51730192e+00 7.83500791e-01 -3.49637538e-01 -3.98774773e-01
4.90904540e-01 -1.40427545e-01 -4.82575208e-01 4.96249534e-02
3.68495286e-01 8.63424659e-01 8.12197149e-01 -1.50036097e+00
-7.60267735e-01 -4.18523133e-01 -6.56556105e-03 4.55190867e-01
-6.39452457e-01 -2.60203034e-01 -4.48631406e-01 -5.79954445e-01
1.76318601e-01 -7.96261966e-01 -1.75622046e-01 6.86529502e-02
-3.95326048e-01 -4.62968349e-01 8.32356811e-01 -2.00806096e-01
1.16865611e+00 -2.43250036e+00 3.37913968e-02 2.92405456e-01
2.16232300e-01 4.07972455e-01 -2.59265512e-01 1.63608372e-01
8.86039734e-02 -6.46377876e-02 -4.76878136e-01 -4.74030435e-01
-8.46321732e-02 2.13075861e-01 -4.67748940e-01 2.40220502e-01
2.04777166e-01 7.41031647e-01 -1.13814843e+00 -5.41263103e-01
8.42135474e-02 2.61431992e-01 -4.06908631e-01 3.00614268e-01
-2.82415330e-01 5.27924418e-01 -8.56264770e-01 8.29644263e-01
7.88984597e-01 -4.75142479e-01 -9.46225598e-02 -5.01031950e-02
1.48535579e-01 2.71661937e-01 -1.10514402e+00 1.86056280e+00
-2.65358001e-01 9.44419485e-03 -3.54199223e-02 -9.07125652e-01
1.11478651e+00 1.27483368e-01 5.13818502e-01 -6.21696174e-01
-1.40922844e-01 4.67295200e-01 -2.92699069e-01 -3.42991427e-02
1.00764237e-01 -7.41278529e-02 -1.87850371e-01 3.28604996e-01
1.96201935e-01 3.33421379e-01 -1.34802759e-01 2.42776915e-01
8.23007703e-01 1.70770571e-01 4.63619560e-01 -1.49538472e-01
4.38348353e-01 1.42457902e-01 9.06693161e-01 5.39707124e-01
-6.64647639e-01 7.15290070e-01 3.90945911e-01 -5.25723957e-02
-8.84723365e-01 -1.28856611e+00 8.46650172e-03 1.27845705e+00
4.20547545e-01 1.03035577e-01 -6.84496224e-01 -1.25571072e+00
-4.09525223e-02 9.36480224e-01 -5.85220575e-01 -5.37188828e-01
-5.55184662e-01 -4.65295881e-01 2.11921617e-01 4.67614174e-01
5.62321305e-01 -8.57052028e-01 -1.68611422e-01 1.77348658e-01
-1.99288502e-01 -7.59389341e-01 -8.01747680e-01 3.65501106e-01
-8.45910311e-01 -7.14883268e-01 -1.02674627e+00 -8.94650996e-01
6.52077615e-01 4.57069606e-01 1.03017449e+00 -5.55131137e-01
4.71097231e-01 -5.81996813e-02 -4.96344537e-01 -1.95063263e-01
-2.13302240e-01 3.28481436e-01 9.90240201e-02 4.68625352e-02
7.68409014e-01 -5.64081967e-01 -5.79125762e-01 5.28296530e-01
-6.22456610e-01 -2.87115484e-01 7.76278913e-01 8.32345486e-01
1.02230418e+00 -2.69709062e-02 1.01268184e+00 -1.01450121e+00
4.70290869e-01 -9.35258448e-01 -4.06805128e-01 3.33900094e-01
-8.42683494e-01 2.48750485e-02 7.67653584e-01 -6.32715464e-01
-1.10796714e+00 2.75196493e-01 2.00917348e-01 -9.24434721e-01
-2.82665759e-01 2.85100490e-01 -7.38254547e-01 1.61500499e-01
8.44833791e-01 1.82486847e-01 -3.99627462e-02 -5.63816249e-01
3.71886969e-01 7.07770288e-01 3.04926127e-01 -6.36957169e-01
7.28565216e-01 3.72549593e-01 -4.82358754e-01 -4.21719164e-01
-1.31155336e+00 -7.11647153e-01 -6.25124633e-01 2.22916290e-01
3.96233022e-01 -1.02951300e+00 7.34114125e-02 1.75606653e-01
-7.45073438e-01 -2.98545390e-01 -7.32698798e-01 3.79059345e-01
-5.09883821e-01 2.44409814e-01 1.17170596e-02 -6.57938540e-01
-2.02228740e-01 -9.72146094e-01 1.06603205e+00 3.08297664e-01
-1.03746139e-01 -1.00938797e+00 1.73875913e-01 1.70816258e-01
1.76526755e-01 1.43652096e-01 8.06162715e-01 -1.15555632e+00
-4.35264349e-01 1.96823776e-02 -1.08766474e-01 4.21648771e-01
3.43437701e-01 -6.05212450e-01 -8.88196766e-01 -3.63132477e-01
5.05983084e-02 -5.13885379e-01 8.28693390e-01 3.49441946e-01
1.07078111e+00 -3.14176321e-01 -6.24139309e-01 5.69372416e-01
1.23253632e+00 4.08739001e-01 2.99709320e-01 3.32529038e-01
6.31444037e-01 5.50874770e-01 8.79554808e-01 3.96668762e-01
4.14144635e-01 6.10733330e-01 3.04453313e-01 -1.79920439e-02
-1.36755005e-01 -6.60562813e-01 5.18355370e-01 6.51542842e-01
6.19790137e-01 -3.74687374e-01 -8.51760864e-01 8.21907699e-01
-1.79733264e+00 -6.60328329e-01 4.40329224e-01 2.45863366e+00
1.02356350e+00 1.94365725e-01 3.28211159e-01 -6.75510019e-02
8.93779457e-01 1.20359831e-01 -1.12854576e+00 1.41713992e-01
-1.80341706e-01 -1.48761839e-01 3.49489659e-01 3.57408166e-01
-1.37157297e+00 9.13293242e-01 5.54325199e+00 1.14415371e+00
-9.54164088e-01 2.03212783e-01 7.49825716e-01 -1.36654779e-01
-3.66365105e-01 4.74145822e-02 -9.81169701e-01 6.58145428e-01
7.97050416e-01 -1.77077904e-01 2.99044847e-01 1.21825254e+00
1.39727043e-02 2.97639340e-01 -1.17863643e+00 6.09449565e-01
-5.42191155e-02 -9.94552612e-01 1.46567270e-01 3.69110629e-02
8.13169837e-01 5.65732419e-02 2.80072410e-02 4.99972910e-01
5.51838279e-01 -5.94824970e-01 4.77208167e-01 2.47053131e-01
7.67718017e-01 -9.38154697e-01 4.79846925e-01 6.83835089e-01
-9.62136447e-01 -3.03979844e-01 -5.16981781e-01 4.92368281e-01
1.94172971e-02 7.40509868e-01 -7.17582226e-01 4.13382888e-01
5.09249866e-01 8.45930278e-01 -2.80156970e-01 9.64324534e-01
-9.83289406e-02 7.12596059e-01 -3.27163011e-01 1.48802847e-01
3.41022551e-01 -1.05674274e-01 6.46000326e-01 8.11461866e-01
4.84227777e-01 1.00193642e-01 4.52253461e-01 9.33146060e-01
-2.15091735e-01 1.44688919e-01 -5.80384493e-01 7.32318088e-02
1.12216890e+00 9.57571208e-01 -3.90931219e-01 -2.30731025e-01
-2.53620058e-01 1.01857007e+00 3.96588624e-01 4.01121110e-01
-6.15073919e-01 -3.77312362e-01 6.08338952e-01 2.03951031e-01
3.56461257e-01 1.42881587e-01 -1.83399305e-01 -1.17167580e+00
1.53650209e-01 -8.05020154e-01 6.33758664e-01 -4.22075748e-01
-1.78242505e+00 4.32366878e-01 4.68619242e-02 -1.62295318e+00
-1.57887444e-01 -1.94705516e-01 -5.70615053e-01 8.75004232e-01
-1.67772138e+00 -1.08835626e+00 -1.52254980e-02 8.42914701e-01
7.57705212e-01 -4.06480283e-01 6.80771351e-01 1.97610900e-01
-4.89528149e-01 9.47277844e-01 4.31099415e-01 8.93339291e-02
9.30416286e-01 -1.22380614e+00 1.57441795e-01 7.18877375e-01
1.36679383e-02 5.91976166e-01 2.84353286e-01 -5.28609455e-01
-9.54388797e-01 -1.41466534e+00 8.11089098e-01 -4.20973778e-01
3.71723890e-01 -3.88105631e-01 -1.17323053e+00 6.29188001e-01
-1.34001553e-01 1.55879393e-01 6.23422742e-01 6.03138171e-02
-5.41608751e-01 -2.64760435e-01 -1.44887626e+00 4.46081161e-01
1.02914417e+00 -2.70718604e-01 -6.86385393e-01 3.42392743e-01
1.09900093e+00 -1.66924044e-01 -7.38737762e-01 4.75164264e-01
1.64741054e-02 -8.43374252e-01 1.10218787e+00 -3.24013025e-01
2.11789876e-01 -3.97040844e-01 -1.76545963e-01 -1.59597552e+00
-4.87161666e-01 -4.58995908e-01 -3.25894922e-01 1.66075087e+00
6.13508463e-01 -7.56493092e-01 9.00038183e-01 4.18591052e-01
-1.92440912e-01 -7.15029776e-01 -9.87983763e-01 -9.78212059e-01
4.10300285e-01 6.63586706e-02 8.90748143e-01 1.23162937e+00
-1.07959196e-01 4.31610972e-01 -3.41177613e-01 2.01261163e-01
8.13738763e-01 2.90994704e-01 5.42915404e-01 -1.35563481e+00
-2.70891428e-01 -3.27226162e-01 1.14182845e-01 -1.42186916e+00
1.96016476e-01 -7.84544528e-01 1.61576197e-01 -1.20556557e+00
1.91744298e-01 -9.55930531e-01 -7.01493561e-01 5.67074716e-01
-2.89006889e-01 -1.76877230e-01 -2.18082275e-02 7.31687963e-01
-8.52217138e-01 7.51069725e-01 1.24437881e+00 4.50781994e-02
-5.15014052e-01 1.70721680e-01 -1.10393143e+00 5.48514545e-01
7.80792177e-01 -6.17941558e-01 -8.50830853e-01 -3.52695614e-01
-3.00548106e-01 -3.33704948e-02 6.56364262e-02 -9.01777506e-01
9.91077125e-02 -4.42194045e-01 4.33388114e-01 -5.39363325e-01
3.10113966e-01 -8.52145851e-01 -2.67104030e-01 2.32675541e-02
-7.42574215e-01 -5.57765424e-01 -2.24663720e-01 7.24970102e-01
-2.34954491e-01 -1.39061794e-01 1.08263040e+00 -3.32735330e-02
-9.49426174e-01 5.23218989e-01 1.92965388e-01 3.76961559e-01
1.24756980e+00 -1.81107074e-01 -3.30470979e-01 -2.10810244e-01
-5.83479583e-01 5.27604222e-01 5.29332995e-01 4.22876120e-01
4.79999989e-01 -1.56949866e+00 -6.03288352e-01 1.86741307e-01
5.75094759e-01 4.11091655e-01 7.87533820e-02 5.48196435e-01
2.44716853e-01 2.48327568e-01 4.95783947e-02 -6.11416459e-01
-6.09758079e-01 7.65006006e-01 3.51797640e-01 -2.35131189e-01
-6.23258650e-01 9.92742658e-01 4.65120554e-01 -8.26694965e-01
3.49912196e-01 1.09988309e-01 -2.01074466e-01 3.99408070e-03
5.02050638e-01 4.11441922e-01 -4.80041057e-02 -4.93527949e-01
-3.63787085e-01 3.70015442e-01 -2.72846580e-01 6.05069883e-02
1.06263661e+00 -4.51831937e-01 3.90519887e-01 4.30192381e-01
1.32528532e+00 3.37604843e-02 -1.72229826e+00 -7.95012355e-01
1.34585708e-01 -5.16448081e-01 3.66402580e-03 -1.07581520e+00
-1.07107592e+00 7.31771588e-01 6.74670756e-01 6.24660887e-02
1.24591923e+00 2.50981569e-01 9.80514765e-01 2.92416453e-01
3.31746936e-01 -1.14098728e+00 1.33720100e-01 3.54664534e-01
4.99395579e-01 -1.46431506e+00 -3.63984793e-01 -3.30429733e-01
-1.00142574e+00 6.42486155e-01 9.71889555e-01 -6.51146248e-02
6.22692049e-01 -8.58115032e-02 9.38959494e-02 7.89343119e-02
-7.40837514e-01 -1.95308998e-01 1.26702592e-01 8.76940846e-01
1.65210962e-02 -1.04716318e-02 6.14814833e-02 8.00922096e-01
2.24607185e-01 3.75970677e-02 9.17821527e-02 8.03107440e-01
-6.84510767e-01 -1.15395463e+00 -2.29006693e-01 3.95214975e-01
-1.21282563e-01 8.99063498e-02 -5.07125258e-01 6.25788569e-01
2.23802313e-01 9.12647724e-01 -1.47249058e-01 -2.47892067e-01
4.18028653e-01 1.89258605e-01 -3.86048034e-02 -6.01452708e-01
-1.51267633e-01 1.01458803e-01 -2.71181583e-01 -2.79526234e-01
-2.96655387e-01 -6.08545542e-01 -1.34597921e+00 4.27587144e-02
-3.85859877e-01 3.61582220e-01 3.46627384e-01 9.45678473e-01
6.97679520e-01 1.22726932e-01 1.08414984e+00 -4.49218124e-01
-9.02828395e-01 -7.22602546e-01 -6.35407746e-01 3.86391908e-01
4.54210818e-01 -8.83983970e-01 -5.98442018e-01 -9.19563398e-02] | [10.344001770019531, 3.101552963256836] |
b0216dfb-2b0c-4b53-999f-d2a1349c96f0 | tupa-at-mrp-2019-a-multi-task-baseline-system | null | null | https://aclanthology.org/K19-2002 | https://aclanthology.org/K19-2002.pdf | TUPA at MRP 2019: A Multi-Task Baseline System | This paper describes the TUPA system submission to the shared task on Cross-Framework Meaning Representation Parsing (MRP) at the 2019 Conference for Computational Language Learning (CoNLL). Because it was prepared by one of the task co-organizers, TUPA provides a baseline point of comparison and is not considered in the official ranking of participating systems. While originally developed for UCCA only, TUPA has been generalized to support all MRP frameworks included in the task, and trained using multi-task learning to parse them all with a shared model. It is a transition-based parser with a BiLSTM encoder, augmented with BERT contextualized embeddings. | ['Daniel Hershcovich', 'Ofir Arviv'] | 2019-11-01 | null | null | null | conll-2019-11 | ['ucca-parsing'] | ['natural-language-processing'] | [ 4.7898299e-01 5.8247954e-01 -3.1645378e-01 -3.8282838e-01
-1.4751624e+00 -6.3894677e-01 6.1059588e-01 1.9379853e-01
-4.8311278e-01 5.3806180e-01 5.7622308e-01 -6.8271106e-01
2.0127875e-01 -3.6527613e-01 -6.2643951e-01 -2.7767128e-01
5.0763838e-02 5.9220588e-01 1.4302313e-01 -7.4695848e-02
-9.1707855e-02 -9.5136419e-02 -1.1494145e+00 9.9613124e-01
2.6208326e-01 6.9691062e-01 4.0078434e-01 9.7141725e-01
-4.8309514e-01 1.0113955e+00 -5.2245051e-01 -6.3879037e-01
-1.3372028e-01 -1.9985950e-01 -1.4691091e+00 -6.4941537e-01
5.1528120e-01 1.7127255e-01 1.8387939e-01 5.6081259e-01
2.6794541e-01 1.4220177e-02 2.8980783e-01 -1.0344712e+00
-8.0783480e-01 1.1073562e+00 -1.9435863e-01 3.5522026e-01
5.4395396e-01 -6.8913841e-01 1.7679992e+00 -6.9654506e-01
8.0445927e-01 1.5815611e+00 6.1497009e-01 9.6809292e-01
-1.1920373e+00 -4.7355634e-01 4.0080768e-01 -4.4514947e-02
-6.3777608e-01 -4.4694117e-01 5.6997925e-01 -2.2347888e-01
2.0249140e+00 -2.4266539e-02 1.6235596e-01 1.3044925e+00
3.8599017e-01 1.2159541e+00 8.7606460e-01 -8.5794139e-01
6.6589236e-02 -2.7055433e-01 6.3538414e-01 5.6483960e-01
2.5223926e-01 -3.2542008e-01 -4.4848731e-01 -3.8320276e-01
5.4610008e-01 -6.5256202e-01 2.6313403e-01 -4.4468021e-01
-1.0411481e+00 9.4646746e-01 -5.0194528e-02 8.5901105e-01
-4.6160683e-02 5.7033730e-01 8.4261471e-01 4.3302795e-01
4.5782274e-01 2.1961609e-01 -1.0960286e+00 -5.1737314e-01
-6.4723068e-01 3.5073152e-01 1.0257934e+00 9.1136038e-01
3.5564098e-01 -3.7775807e-02 7.1722530e-02 1.0095055e+00
3.7204516e-01 5.6272529e-02 6.2021959e-01 -1.0732582e+00
8.3110070e-01 4.3300062e-01 -2.8187886e-01 -3.1228930e-01
-3.2881689e-01 6.3468650e-02 -2.1960458e-01 -2.1434560e-01
4.4216341e-01 -3.3489698e-01 -6.3013881e-01 1.6846608e+00
-1.8442550e-01 6.2469305e-03 5.6349087e-01 8.9624882e-02
9.9842614e-01 4.2476630e-01 7.9504848e-01 4.0096369e-02
1.4898089e+00 -1.0779916e+00 -7.6301759e-01 -4.7561947e-01
1.2531884e+00 -8.3558792e-01 8.7043113e-01 4.5289001e-01
-9.9898010e-01 -6.9145060e-01 -1.0200670e+00 -5.3873312e-01
-5.9122527e-01 9.2640415e-02 9.4277865e-01 7.9604298e-01
-1.2583379e+00 3.7923911e-01 -9.9556321e-01 -5.3612310e-01
2.1954997e-01 4.7812577e-02 -4.7982389e-01 -1.8639825e-01
-1.3242322e+00 1.0382143e+00 6.7053682e-01 -3.0285120e-01
-5.5822420e-01 -4.8669776e-01 -1.0864893e+00 -1.9138327e-01
5.5644117e-02 -5.6069034e-01 1.5290217e+00 -7.3277533e-01
-1.4009343e+00 1.2803118e+00 -3.3505902e-01 -7.4673611e-01
-2.7877957e-02 -7.0695776e-01 -5.0923461e-01 -1.7452632e-01
4.0134338e-01 1.0043287e+00 3.5471234e-01 -1.0523684e+00
-7.8480041e-01 -1.4848660e-01 4.8091474e-01 9.9700138e-02
2.5125530e-01 7.1309739e-01 -2.0147389e-01 -4.3462104e-01
-2.3556324e-02 -7.6401782e-01 -2.4152693e-01 -1.0375488e+00
-1.5451188e-01 -7.6115185e-01 8.4699339e-01 -8.2290912e-01
1.1737832e+00 -1.9162302e+00 8.5335635e-02 -6.3188607e-01
-1.5612435e-01 9.7256705e-02 -2.8275362e-01 5.6095010e-01
-5.0357908e-01 4.8199281e-01 -4.5672199e-01 -7.8856397e-01
1.2817474e-01 6.7775577e-01 -4.0029925e-01 -1.7176044e-01
4.5270440e-01 8.6956483e-01 -1.1424590e+00 -5.8099389e-01
2.4250239e-01 4.5609662e-01 -4.2933679e-01 -1.2124503e-02
-3.5229182e-01 2.3899943e-01 -6.3076198e-01 4.2400724e-01
4.2256624e-01 4.5639977e-02 6.1450577e-01 2.8182494e-02
-1.7107269e-01 8.1032532e-01 -8.5983670e-01 2.3121591e+00
-8.8210607e-01 3.7971759e-01 1.2030933e-02 -1.0414528e+00
8.3169985e-01 8.4750462e-01 4.0752095e-01 -2.5704190e-01
-2.5436035e-01 3.5809758e-01 -1.7667253e-03 -3.2300975e-02
6.7094415e-01 -1.9909552e-01 -7.5999373e-01 4.3390465e-01
6.7316538e-01 -3.8144243e-01 2.2981735e-01 2.0440754e-01
1.4683932e+00 8.4244150e-01 5.7581329e-01 -3.5977983e-01
7.2418642e-01 5.1190648e-02 6.1392945e-01 4.9061480e-01
-2.8874898e-01 5.5845368e-01 6.8918276e-01 -6.0583997e-01
-8.0807900e-01 -1.0072603e+00 -3.7756830e-01 1.6229296e+00
-4.9205896e-01 -8.8284034e-01 -7.2896975e-01 -1.1605657e+00
-4.8938128e-01 1.1020533e+00 -4.7153711e-01 5.3664196e-01
-8.4067702e-01 -5.6887656e-01 9.3389624e-01 6.3257468e-01
3.4634236e-01 -1.4284402e+00 -7.2746140e-01 5.1781839e-01
-2.3584355e-01 -1.4917245e+00 -3.3245079e-02 8.3176553e-01
-8.2591611e-01 -1.3156916e+00 -3.0376628e-01 -1.2467681e+00
-6.7898192e-02 -1.9008197e-01 1.6141448e+00 -1.6257837e-01
1.9746091e-02 6.6939712e-01 -6.6390252e-01 -5.5017036e-01
-5.5187082e-01 3.3246329e-01 -6.8403238e-01 -6.5185255e-01
6.2965894e-01 -2.5645536e-01 1.9978435e-01 -3.5535523e-01
-6.5293354e-01 1.3237740e-01 3.2017952e-01 9.2786252e-01
7.0047051e-01 -4.0858239e-01 7.4876273e-01 -1.3174014e+00
7.0060766e-01 -4.0407345e-01 -2.2459109e-01 5.3968334e-01
-7.5437009e-02 5.6457687e-02 4.2013288e-01 2.9168934e-01
-1.5695732e+00 3.3423632e-01 -4.7191510e-01 2.0943113e-01
-6.2920183e-01 3.0311394e-01 -5.2225637e-01 3.6034679e-01
9.7783908e-02 -1.6641816e-01 -7.8203601e-01 -7.8687668e-01
7.7575940e-01 4.3081796e-01 4.5109522e-01 -1.0455071e+00
1.3022655e-01 -7.2757035e-02 -1.3600944e-01 -8.0246156e-01
-1.0987029e+00 -4.7576153e-01 -1.0912447e+00 4.5028797e-01
1.4541724e+00 -1.1460305e+00 3.2230014e-01 7.8783385e-02
-1.8653266e+00 -3.7925825e-01 -3.4952018e-01 1.7062052e-01
-6.6368765e-01 3.7950462e-01 -7.4634969e-01 -6.4744228e-01
-4.9082777e-01 -1.0033208e+00 1.0637732e+00 -3.0855787e-01
-3.7863731e-01 -1.5186396e+00 4.5883611e-01 3.5845959e-01
1.0472078e-01 3.5798019e-01 1.2786434e+00 -1.1672909e+00
1.5426530e-01 -4.9532153e-02 -1.5143606e-01 6.3637990e-01
9.7870417e-02 -4.5133706e-02 -1.3168792e+00 4.9074519e-02
-1.2683371e-01 -4.9924409e-01 1.0385610e+00 5.1325977e-01
8.3985132e-01 3.4085041e-01 -3.4928852e-01 1.7323858e-01
1.3847897e+00 1.7491056e-01 4.2121500e-01 5.7937056e-01
6.3160729e-01 7.3647326e-01 3.5633168e-01 -9.3385056e-02
7.6570559e-01 4.0403131e-01 3.6296538e-01 2.2252667e-01
-3.4593585e-01 -2.3900367e-01 6.6389829e-01 1.1236609e+00
7.1836412e-02 -1.9363531e-01 -1.0949427e+00 6.3810301e-01
-1.8085246e+00 -5.7532924e-01 -4.1996989e-01 1.9014635e+00
4.7397265e-01 2.4267289e-01 -2.8934065e-01 -3.1884667e-01
5.4234976e-01 6.1909682e-01 7.6740861e-02 -1.3195983e+00
-1.7187914e-01 8.9387733e-01 1.7243488e-01 5.6500477e-01
-1.3876287e+00 1.4134603e+00 7.5242715e+00 5.1583278e-01
-2.5504351e-01 7.8794813e-01 6.0500681e-01 4.0692922e-01
-2.9225439e-01 4.8923707e-01 -8.9226121e-01 -9.9709667e-02
1.5875297e+00 -8.7497741e-02 1.2514279e-02 1.0202888e+00
-3.9841089e-01 8.6522691e-02 -1.3196641e+00 4.9487373e-01
1.4217111e-02 -1.0792577e+00 3.1394757e-02 -1.8058640e-01
4.0917084e-01 6.7998677e-01 -4.1224408e-01 9.3119639e-01
9.2165697e-01 -9.2638707e-01 4.7000921e-01 -7.6045714e-02
7.7455467e-01 -7.4224621e-01 9.9382114e-01 -9.4446748e-02
-1.6082833e+00 7.4519373e-02 -5.2396369e-01 -2.1021003e-01
4.4775426e-01 1.2236435e-01 -1.9605179e-01 9.6170408e-01
4.7351906e-01 7.1529931e-01 -4.4789922e-01 3.0192834e-01
-5.5949229e-01 8.5277402e-01 -3.4622751e-02 3.2490295e-01
3.6553296e-01 1.4351022e-01 4.5186296e-01 1.6396345e+00
-1.5869522e-02 -2.4220096e-01 4.8255852e-01 3.5762602e-01
-9.7278744e-02 2.4595881e-01 -5.9853077e-01 -1.7186713e-01
4.0353483e-01 1.1948594e+00 -6.2378037e-01 -5.1289940e-01
-7.7285057e-01 8.7697661e-01 4.5901936e-01 -1.3697165e-01
-5.8864677e-01 -2.1399014e-01 5.5585909e-01 -3.8961306e-01
2.9032463e-01 -2.3003542e-01 -1.3764688e-01 -1.1850241e+00
-2.9598272e-01 -4.6999419e-01 8.3982909e-01 -6.2319309e-01
-1.2649866e+00 1.0650059e+00 3.8070610e-01 -5.6855994e-01
-4.4347748e-01 -9.8986000e-01 -7.5504136e-01 9.3394345e-01
-1.5846581e+00 -1.6858476e+00 5.8614892e-01 3.6477298e-01
8.5907549e-01 -2.8569043e-01 1.6098385e+00 7.2329611e-02
-5.2416515e-01 4.6371165e-01 -3.4734607e-01 9.6528791e-02
5.7309598e-01 -1.7622430e+00 9.4997948e-01 8.6725497e-01
4.8407194e-01 5.8219492e-01 4.3092319e-01 -4.4017360e-01
-1.1514494e+00 -1.0232209e+00 1.7296270e+00 -7.4793261e-01
1.0884302e+00 -3.6287814e-01 -7.5953591e-01 1.3274471e+00
6.9655108e-01 4.6191294e-02 9.9471289e-01 5.2393872e-01
-3.6400038e-01 3.9586562e-01 -8.6219192e-01 -8.1336580e-02
7.9439533e-01 -5.0256127e-01 -1.3976958e+00 3.2249776e-01
1.1275724e+00 -4.2245179e-01 -1.0283835e+00 -5.8132466e-03
4.5662424e-01 -5.7875186e-01 7.6990247e-01 -6.3762003e-01
6.3138187e-01 2.1927238e-01 -6.7776126e-01 -9.7811973e-01
-2.1376853e-01 -3.8866958e-01 9.9223651e-02 1.6182945e+00
5.1585513e-01 -7.4606591e-01 5.7382774e-01 3.1835774e-01
-6.5779805e-01 -4.1855451e-01 -1.4335256e+00 -5.9338820e-01
8.5022330e-01 -1.0599779e+00 3.9116460e-01 7.2289050e-01
8.7281957e-02 7.0611691e-01 -4.3210145e-02 -1.6602349e-01
5.6527776e-01 -1.6466513e-01 5.1689547e-01 -1.2072686e+00
-4.0796354e-01 -3.0499765e-01 -7.9868883e-02 -8.5303855e-01
6.7989635e-01 -1.3687192e+00 -7.8979306e-02 -1.9236526e+00
4.3044038e-02 -2.6098946e-01 -8.2162929e-01 9.0945750e-01
-2.3662081e-02 -2.6618761e-01 4.3285662e-01 3.3278234e-02
-8.1050324e-01 4.8982717e-02 7.6866674e-01 9.6614502e-02
8.5924454e-02 -8.3299913e-02 -9.3717444e-01 8.4207410e-01
6.5905625e-01 -7.3566628e-01 -2.5697502e-01 -7.6832277e-01
1.3386275e-01 4.4563673e-03 5.3365119e-02 -1.0184296e+00
-4.1221112e-01 1.4588451e-01 2.4763092e-02 -6.6068834e-01
2.0736223e-01 -5.2049673e-01 -2.3207825e-01 3.0466706e-01
-4.7916055e-01 2.7359924e-01 5.2513295e-01 1.3802785e-01
-4.5117071e-01 -5.6384736e-01 3.1003150e-01 -5.6263870e-01
-9.4504160e-01 -1.6484591e-01 -4.8664778e-01 2.3599601e-01
5.9959120e-01 -1.4342461e-02 -1.3036133e-01 2.3400538e-01
-9.2136252e-01 1.8821250e-01 2.4804631e-03 6.1606675e-01
2.5158265e-01 -1.0385787e+00 -6.8040270e-01 -1.7650136e-01
1.6585435e-01 -4.8796106e-02 1.5644360e-01 1.3758087e-01
-1.8030252e-01 9.0833008e-01 4.4310577e-03 -2.7152199e-01
-1.2649959e+00 1.0426120e-01 -6.2216446e-02 -1.3473730e+00
-7.3603356e-01 8.4217042e-01 1.1041149e-01 -9.2184985e-01
-9.1694199e-02 -6.2151915e-01 -5.6564033e-01 1.2688225e-02
5.6607884e-01 2.9171264e-02 1.8487443e-01 -6.6135734e-01
-4.5893461e-01 2.9403263e-01 -4.1259535e-02 -1.1074546e-01
1.5402569e+00 1.2752163e-01 -1.2953313e-01 8.6877525e-01
1.1718538e+00 -1.3587216e-01 -8.4834081e-01 -6.6675968e-02
7.9904151e-01 2.9303092e-01 4.7477156e-02 -1.0430422e+00
-5.9264350e-01 1.1688805e+00 3.8517162e-01 -6.5980405e-02
7.9268926e-01 7.7043153e-02 9.2625350e-01 5.0605279e-01
5.1635474e-01 -9.6084213e-01 -2.8052130e-01 1.1594707e+00
9.0841556e-01 -8.9790398e-01 -2.0124745e-01 -2.7459928e-01
-7.4935573e-01 1.4294614e+00 4.0081373e-01 -3.7336656e-01
5.9345233e-01 4.8192236e-01 -1.2564315e-02 7.0557736e-02
-1.1993394e+00 -9.0377010e-02 -1.2403861e-01 8.9257908e-01
1.1855611e+00 3.6802992e-01 -6.0477221e-01 1.1663930e+00
-2.8017068e-01 -6.4173095e-02 4.3714702e-01 1.1566082e+00
-4.1889438e-01 -1.9517375e+00 1.6383484e-01 1.7782307e-01
-9.2316079e-01 -4.1031471e-01 -2.6924071e-01 1.0826520e+00
3.5616204e-01 9.7503835e-01 3.5735302e-02 -1.2889221e-01
3.6768165e-01 6.2759614e-01 5.8802807e-01 -1.1672966e+00
-1.0020188e+00 -7.3802747e-02 8.8174832e-01 -5.0264895e-01
-5.7660067e-01 -8.9038002e-01 -1.3687148e+00 3.3529410e-01
9.4677418e-02 3.7458256e-01 7.1415091e-01 1.0255226e+00
7.4728958e-02 7.0382899e-01 -2.7099729e-04 -8.4056258e-01
-1.9991827e-01 -1.1761367e+00 -3.7940931e-01 9.0562589e-02
-1.7691018e-01 -4.7766933e-01 -2.2495188e-01 2.0246148e-01] | [10.361757278442383, 9.494482040405273] |
821d5fb3-4731-4e72-97d1-c15fc4294168 | improving-simultaneous-machine-translation | 2212.01188 | null | https://arxiv.org/abs/2212.01188v1 | https://arxiv.org/pdf/2212.01188v1.pdf | Improving Simultaneous Machine Translation with Monolingual Data | Simultaneous machine translation (SiMT) is usually done via sequence-level knowledge distillation (Seq-KD) from a full-sentence neural machine translation (NMT) model. However, there is still a significant performance gap between NMT and SiMT. In this work, we propose to leverage monolingual data to improve SiMT, which trains a SiMT student on the combination of bilingual data and external monolingual data distilled by Seq-KD. Preliminary experiments on En-Zh and En-Ja news domain corpora demonstrate that monolingual data can significantly improve translation quality (e.g., +3.15 BLEU on En-Zh). Inspired by the behavior of human simultaneous interpreters, we propose a novel monolingual sampling strategy for SiMT, considering both chunk length and monotonicity. Experimental results show that our sampling strategy consistently outperforms the random sampling strategy (and other conventional typical NMT monolingual sampling strategies) by avoiding the key problem of SiMT -- hallucination, and has better scalability. We achieve +0.72 BLEU improvements on average against random sampling on En-Zh and En-Ja. Data and codes can be found at https://github.com/hexuandeng/Mono4SiMT. | ['Min Zhang', 'DaCheng Tao', 'Meishan Zhang', 'Xuebo Liu', 'Liang Ding', 'Hexuan Deng'] | 2022-12-02 | null | null | null | null | ['nmt'] | ['computer-code'] | [-4.53547947e-03 9.51221958e-02 -5.93075395e-01 -1.67975813e-01
-1.61748743e+00 -6.69016302e-01 7.23298311e-01 -2.47256413e-01
-5.63164294e-01 1.32235634e+00 5.07918239e-01 -8.59717488e-01
3.12271178e-01 -3.95462692e-01 -1.04664564e+00 -3.92923176e-01
4.77918029e-01 9.97802258e-01 -3.16255510e-01 -3.78163248e-01
-2.23019067e-03 -2.50306129e-01 -6.65015519e-01 3.82900029e-01
1.58659351e+00 2.26962984e-01 5.92404008e-01 4.02800500e-01
-1.87388316e-01 7.34531760e-01 -6.00459695e-01 -5.69863737e-01
3.57551783e-01 -9.57526684e-01 -1.10607123e+00 -3.91984642e-01
5.37512779e-01 -2.38663763e-01 -1.85333803e-01 1.00669539e+00
7.96818852e-01 -2.20202759e-01 2.47791722e-01 -7.14975238e-01
-8.22739065e-01 1.24909282e+00 -4.63026643e-01 1.20103158e-01
4.32869405e-01 2.72731423e-01 1.10387576e+00 -1.09227443e+00
9.67245340e-01 1.24406111e+00 7.82430768e-01 5.87062240e-01
-1.34175622e+00 -7.70216882e-01 -2.14022309e-01 2.13171780e-01
-1.30763245e+00 -6.81107104e-01 1.81764707e-01 -8.99065956e-02
1.24032295e+00 3.03220868e-01 5.49599946e-01 1.37707961e+00
2.95114726e-01 1.24240792e+00 1.53743589e+00 -5.86854398e-01
-7.32946321e-02 2.58178741e-01 -2.44145095e-01 5.41953862e-01
-7.47871399e-02 4.57797274e-02 -9.24876630e-01 -1.06398329e-01
6.20858610e-01 -5.74861825e-01 -4.21681672e-01 1.19973615e-01
-1.83592212e+00 7.00147688e-01 3.38419043e-02 3.59874070e-01
-3.66694599e-01 -2.09675953e-01 4.55942720e-01 7.90949643e-01
6.41880035e-01 7.02947438e-01 -6.06855094e-01 -5.40289342e-01
-1.22401536e+00 1.18641682e-01 8.38606954e-01 1.28186893e+00
7.43025184e-01 7.28772208e-02 -2.54731894e-01 1.05377269e+00
-1.78752616e-01 1.01511955e+00 7.41932213e-01 -8.53080094e-01
9.28559065e-01 1.17736638e-01 -4.92145196e-02 -2.09417343e-01
4.57948521e-02 -5.30218005e-01 -7.92570114e-01 -5.61239541e-01
4.63878244e-01 -3.16778958e-01 -7.18970597e-01 1.86657858e+00
9.16402042e-02 -1.24383159e-01 3.62813562e-01 9.40284252e-01
5.51166594e-01 9.15393114e-01 -3.47859144e-01 -5.15145540e-01
1.04312456e+00 -1.23236620e+00 -8.81531715e-01 -4.19335246e-01
9.42216218e-01 -1.21341538e+00 1.36822629e+00 1.88516617e-01
-1.42791307e+00 -3.28372002e-01 -7.83660471e-01 -1.67537346e-01
8.25804770e-02 2.03217626e-01 3.81101578e-01 2.94289082e-01
-1.25524950e+00 5.57609379e-01 -7.51706839e-01 -6.45192921e-01
7.33285397e-02 2.34199494e-01 -2.79904395e-01 -3.88680071e-01
-1.51126087e+00 1.21430421e+00 3.73309582e-01 8.74595810e-03
-8.65655780e-01 -7.52450883e-01 -6.99310422e-01 -4.43431824e-01
3.98426056e-01 -8.25065434e-01 1.69665849e+00 -9.60541666e-01
-1.79084742e+00 6.80288911e-01 -4.96816427e-01 -4.17905033e-01
8.10180366e-01 -3.33629876e-01 -1.94722444e-01 -1.03456013e-01
5.09018779e-01 5.98652065e-01 3.24843347e-01 -9.69834507e-01
-4.22291696e-01 -7.41838515e-02 -3.40958565e-01 7.09254622e-01
-1.42003283e-01 1.83013737e-01 -5.53624034e-01 -6.47309184e-01
1.31205255e-02 -9.85031903e-01 -4.55982983e-02 -7.86659479e-01
-5.79053521e-01 -1.68835044e-01 1.79519176e-01 -1.17771804e+00
1.16416979e+00 -1.72498465e+00 4.02022421e-01 -1.92415431e-01
-9.77699906e-02 3.03871185e-01 -4.31024760e-01 7.42698133e-01
2.56759048e-01 -3.28870863e-02 -2.20648110e-01 -4.06031668e-01
-8.12776759e-02 3.48853648e-01 -2.38429114e-01 3.01332951e-01
-6.50004223e-02 1.28182471e+00 -1.11600888e+00 -4.02528793e-01
-1.58525109e-01 8.46631676e-02 -3.32637280e-01 1.52832523e-01
-3.77218068e-01 8.15439165e-01 -1.95685611e-03 6.73115551e-01
4.78212029e-01 -2.16679230e-01 7.17753708e-01 2.90463507e-01
-1.46783665e-01 9.81659830e-01 -5.76957941e-01 2.11513567e+00
-6.40011311e-01 6.66130602e-01 -1.05230637e-01 -6.04758263e-01
7.60239303e-01 5.32316804e-01 -2.28967480e-02 -1.11170387e+00
-8.88280869e-02 8.29433322e-01 1.57679543e-01 -2.27399603e-01
7.19472945e-01 -3.73757929e-01 -1.65485829e-01 7.50445545e-01
2.87891150e-01 -7.25755095e-02 2.57249385e-01 3.10399503e-01
7.87543416e-01 2.22429425e-01 3.43436092e-01 -5.34369707e-01
1.37083322e-01 3.92376393e-01 8.68864894e-01 7.47382641e-01
-1.06701776e-02 3.92207295e-01 3.35155725e-01 -8.16338137e-02
-1.40494597e+00 -1.05027223e+00 9.62536111e-02 9.57543492e-01
-1.35413468e-01 -4.68651325e-01 -9.78980482e-01 -5.64539254e-01
-1.98499098e-01 1.00824642e+00 -2.24018440e-01 9.35521349e-02
-1.04756486e+00 -8.05469215e-01 7.84261584e-01 3.58041912e-01
5.37235975e-01 -9.06961024e-01 1.51841998e-01 2.14227691e-01
-9.29814219e-01 -1.20002651e+00 -9.24963713e-01 1.24946553e-02
-9.39475477e-01 -4.60702270e-01 -8.60569119e-01 -7.52531469e-01
3.58178288e-01 3.23086262e-01 1.34722817e+00 -2.96101779e-01
3.42022806e-01 -1.64825425e-01 -3.34062576e-01 -2.81201005e-02
-8.85147870e-01 6.26907885e-01 3.54506642e-01 -4.75336760e-01
5.10089576e-01 -4.76418704e-01 -2.41499916e-01 2.64901429e-01
-4.41908389e-01 6.16981208e-01 9.68264222e-01 1.10824573e+00
4.97068435e-01 -7.20622599e-01 5.77789664e-01 -9.41267312e-01
8.22074294e-01 -4.87483740e-01 -3.08768272e-01 4.47064131e-01
-8.98274601e-01 1.39124975e-01 7.64388978e-01 -5.12848020e-01
-9.84033227e-01 -6.70524597e-01 -2.37124767e-02 -2.44467676e-01
1.35902107e-01 7.68216729e-01 -8.92796665e-02 3.42082947e-01
6.87197208e-01 8.22978020e-01 1.62024051e-01 -5.62311292e-01
4.39595968e-01 9.62178171e-01 4.55947101e-01 -7.64864445e-01
5.77357769e-01 -1.06991276e-01 -7.39063203e-01 -6.15539312e-01
-7.57305562e-01 -2.74995446e-01 -4.17248726e-01 5.11855036e-02
5.06983638e-01 -1.24753582e+00 -1.58744305e-01 4.15728807e-01
-1.24901295e+00 -8.25740695e-01 -1.05179414e-01 8.48960936e-01
-6.79067314e-01 2.74859816e-01 -1.16266143e+00 -3.54796916e-01
-6.76753759e-01 -1.31417429e+00 9.61977541e-01 -1.35559440e-01
-4.54847604e-01 -1.01346028e+00 3.06244135e-01 7.84746349e-01
5.04350722e-01 -3.66444439e-01 8.48503292e-01 -6.32046342e-01
-6.40806794e-01 3.38719279e-01 -1.33511081e-01 4.12695199e-01
-8.54112953e-02 -5.99856555e-01 -5.11017919e-01 -6.09058678e-01
-5.05960584e-02 -5.05115390e-01 5.96077025e-01 1.98499635e-01
2.77284682e-01 -5.87562621e-01 -3.55118774e-02 6.34054840e-01
1.35460210e+00 3.14694010e-02 4.36877608e-01 4.86170709e-01
6.05000138e-01 3.24724108e-01 5.86879671e-01 6.71339035e-02
7.01194048e-01 7.11680472e-01 -3.00282001e-01 -1.30152134e-02
-3.90875250e-01 -6.13472700e-01 9.24423277e-01 1.87107527e+00
-9.73823816e-02 -1.06067389e-01 -1.13120151e+00 6.72491431e-01
-1.90482187e+00 -7.97995508e-01 -1.27321690e-01 2.20378494e+00
1.53473973e+00 -7.67632276e-02 7.27466196e-02 -4.62721288e-01
6.39243662e-01 -1.02993719e-01 -4.85790253e-01 -5.47333241e-01
-5.08782148e-01 1.72877267e-01 6.49943769e-01 9.40434515e-01
-3.92984629e-01 1.46953022e+00 5.91690683e+00 1.13886595e+00
-1.02265513e+00 6.16119981e-01 3.34055334e-01 -2.86541134e-01
-5.50919235e-01 1.06417790e-01 -9.14947510e-01 4.58581060e-01
1.28845620e+00 -4.65168685e-01 8.77348781e-01 2.71368742e-01
2.95011401e-01 2.13858783e-02 -1.03916752e+00 7.17058957e-01
1.77733183e-01 -1.29852450e+00 1.13910399e-01 1.43335655e-01
1.07910359e+00 6.07510269e-01 -6.15930893e-02 6.29450321e-01
6.12193644e-01 -7.90996253e-01 8.39437068e-01 1.64480835e-01
1.10058808e+00 -6.13375902e-01 7.97206521e-01 8.04457366e-01
-6.11527324e-01 3.13644856e-01 -3.60795021e-01 -1.09749012e-01
1.74987793e-01 6.15401685e-01 -1.20519066e+00 1.00393784e+00
3.90423179e-01 7.95130789e-01 -1.46205485e-01 5.91375709e-01
-5.79974473e-01 9.60872889e-01 -1.88726425e-01 1.69033483e-02
3.70889872e-01 -3.59615177e-01 6.23455584e-01 1.37244797e+00
5.17838478e-01 -3.61561686e-01 1.03586905e-01 6.90062940e-01
-3.80131215e-01 4.31985855e-01 -5.35192966e-01 -8.62591788e-02
6.69476211e-01 7.67466307e-01 -5.87674603e-02 -6.71117127e-01
-4.29101676e-01 1.40061128e+00 6.12060785e-01 4.49478984e-01
-7.18583524e-01 -1.02627933e-01 6.31952345e-01 -8.78178775e-02
7.92345777e-02 -3.70321900e-01 -2.73271561e-01 -1.55702698e+00
2.80398846e-01 -1.53920543e+00 -6.20763935e-02 -5.82959414e-01
-1.12607217e+00 8.39638412e-01 -9.49544236e-02 -1.23392022e+00
-5.76474369e-01 -1.83302373e-01 -1.95608437e-01 1.21717811e+00
-1.43575633e+00 -1.24568856e+00 3.90465230e-01 3.84833902e-01
9.06560004e-01 -2.15015441e-01 8.15656900e-01 3.18692356e-01
-5.80750942e-01 9.06649351e-01 6.48869336e-01 9.85813960e-02
1.14198935e+00 -1.27116442e+00 7.79983282e-01 9.25216138e-01
1.45727605e-01 9.12899911e-01 6.37495160e-01 -8.94951224e-01
-1.58328259e+00 -1.08533931e+00 1.70405316e+00 -5.73073864e-01
7.94726551e-01 -4.59774464e-01 -6.97944880e-01 9.36615348e-01
8.67083669e-01 -7.43456900e-01 7.58314073e-01 3.71339440e-01
-4.52114195e-01 1.38058454e-01 -7.36539364e-01 8.74564648e-01
1.11581743e+00 -6.85719550e-01 -7.29987562e-01 5.59839725e-01
8.05525899e-01 -5.33409536e-01 -9.77016449e-01 2.48620689e-01
4.12796766e-01 -6.66306078e-01 5.49300253e-01 -4.80847716e-01
5.00246942e-01 -1.48830846e-01 -2.00220615e-01 -1.80397570e+00
-1.57758579e-01 -1.07025278e+00 -3.67090032e-02 9.55818474e-01
8.06897104e-01 -7.37794638e-01 4.10940975e-01 -4.99692932e-02
-2.83825874e-01 -6.79606140e-01 -1.00514293e+00 -1.25546420e+00
6.59274578e-01 -2.43026197e-01 5.54602981e-01 1.29876518e+00
3.65833849e-01 8.19247425e-01 -7.79933929e-01 -1.67993397e-01
6.71249092e-01 2.09595546e-01 8.17717075e-01 -5.62234759e-01
-6.18070722e-01 -3.80288631e-01 4.31489557e-01 -1.48341811e+00
1.90158829e-01 -1.39606047e+00 1.13667242e-01 -1.57885826e+00
4.48487788e-01 -1.61331192e-01 8.83828029e-02 5.56181073e-01
-4.03383166e-01 2.01239586e-01 2.92317301e-01 5.88113606e-01
-4.69649851e-01 5.92906296e-01 1.58326960e+00 3.62549946e-02
-1.17470473e-01 -3.33883703e-01 -7.00875163e-01 1.83738127e-01
1.06034386e+00 -4.36594129e-01 -1.82053804e-01 -1.02841985e+00
1.85958549e-01 4.18860346e-01 -2.18684420e-01 -5.71405411e-01
2.45280251e-01 -1.39183342e-01 -3.16477306e-02 -3.98358941e-01
3.21814194e-02 -1.68151855e-01 1.52154520e-01 4.55234438e-01
-3.83887917e-01 2.96998769e-01 1.65862799e-01 6.35640696e-02
-2.58365840e-01 7.87987709e-02 4.60557550e-01 -4.02875751e-01
-3.06021065e-01 -6.67894781e-02 -4.34744447e-01 2.81063050e-01
5.52505076e-01 5.61457239e-02 -5.87769508e-01 -3.07976544e-01
-3.71424079e-01 3.91306669e-01 5.08573234e-01 3.50250065e-01
3.34096402e-01 -1.44097316e+00 -1.33019423e+00 1.88929841e-01
5.60477488e-02 -2.29042873e-01 -1.41221820e-03 1.45170903e+00
-3.42042804e-01 6.70301139e-01 -2.89938878e-02 -4.59017903e-01
-1.07059431e+00 1.99997365e-01 1.06465444e-01 -4.96044964e-01
-4.21511382e-01 7.26114988e-01 -2.29594782e-01 -1.09536064e+00
-7.40025491e-02 -1.52555138e-01 5.64777195e-01 -2.03348890e-01
3.67894679e-01 3.27970207e-01 1.63060412e-01 -5.02020001e-01
-1.25782952e-01 1.26431122e-01 -3.77735734e-01 -5.31112850e-01
1.04044783e+00 -3.10646594e-01 -3.92725140e-01 5.49125910e-01
1.08977056e+00 2.62285531e-01 -6.95720732e-01 -7.15550780e-01
1.17522962e-02 -4.38180536e-01 -2.58268803e-01 -1.16932523e+00
-4.65570778e-01 7.18385696e-01 4.34267148e-02 -3.63809913e-01
8.97009492e-01 -4.46672505e-03 1.20547283e+00 4.81080025e-01
5.56267500e-01 -1.16203344e+00 -3.54189336e-01 8.76818895e-01
6.58989310e-01 -1.21092618e+00 -3.49549949e-01 -1.81459598e-02
-7.71455050e-01 7.24703968e-01 5.26755869e-01 3.09521556e-01
-1.05242454e-01 1.89923599e-01 5.00753462e-01 3.18605721e-01
-1.04273796e+00 -3.31506394e-02 -1.31756244e-02 1.69521376e-01
7.89828479e-01 4.26179469e-01 -6.21780872e-01 2.09020093e-01
-5.86423278e-01 7.16642216e-02 4.18392628e-01 7.64723063e-01
-3.41095924e-01 -1.53571236e+00 -3.78159344e-01 2.19255298e-01
-3.99359196e-01 -7.58604825e-01 -3.99884582e-01 7.26142824e-01
-1.69015393e-01 8.04635882e-01 -2.83724368e-01 -5.04007638e-01
9.78173465e-02 3.06887895e-01 5.94802082e-01 -6.04125500e-01
-7.39073932e-01 3.61085296e-01 4.20956403e-01 -4.98018563e-01
-2.41889983e-01 -9.75879848e-01 -8.68040144e-01 -6.93729997e-01
-1.39411256e-01 5.18681705e-01 4.60618794e-01 9.64311898e-01
4.02199894e-01 1.99761778e-01 4.70499903e-01 -4.21151668e-01
-8.79152894e-01 -1.43961477e+00 -2.39726529e-01 9.97614563e-02
1.84360698e-01 -1.32840434e-02 -2.01660618e-01 -1.28738075e-01] | [11.599058151245117, 10.285199165344238] |
86564557-7b6f-494c-94be-48a0fcb5ddaa | the-curse-of-dimensionality-in-operator | 2306.15924 | null | https://arxiv.org/abs/2306.15924v1 | https://arxiv.org/pdf/2306.15924v1.pdf | The curse of dimensionality in operator learning | Neural operator architectures employ neural networks to approximate operators mapping between Banach spaces of functions; they may be used to accelerate model evaluations via emulation, or to discover models from data. Consequently, the methodology has received increasing attention over recent years, giving rise to the rapidly growing field of operator learning. The first contribution of this paper is to prove that for general classes of operators which are characterized only by their $C^r$- or Lipschitz-regularity, operator learning suffers from a curse of dimensionality, defined precisely here in terms of representations of the infinite-dimensional input and output function spaces. The result is applicable to a wide variety of existing neural operators, including PCA-Net, DeepONet and the FNO. The second contribution of the paper is to prove that the general curse of dimensionality can be overcome for solution operators defined by the Hamilton-Jacobi equation; this is achieved by leveraging additional structure in the underlying solution operator, going beyond regularity. To this end, a novel neural operator architecture is introduced, termed HJ-Net, which explicitly takes into account characteristic information of the underlying Hamiltonian system. Error and complexity estimates are derived for HJ-Net which show that this architecture can provably beat the curse of dimensionality related to the infinite-dimensional input and output function spaces. | ['Andrew M. Stuart', 'Samuel Lanthaler'] | 2023-06-28 | null | null | null | null | ['operator-learning'] | ['miscellaneous'] | [ 1.97441027e-01 3.77757519e-01 4.56115194e-02 1.31633624e-01
-1.85867772e-01 -1.47391200e-01 1.54469088e-01 -1.69358198e-02
-5.07133424e-01 6.51770949e-01 -8.75245333e-02 -3.22530806e-01
-5.82405388e-01 -6.08888328e-01 -7.71054804e-01 -7.67363787e-01
-4.05813038e-01 3.75439785e-02 -2.10287273e-01 -3.73822689e-01
-8.40613022e-02 6.99807227e-01 -1.37388599e+00 -3.02216977e-01
9.21183765e-01 1.34064257e+00 -1.33835897e-01 5.04553080e-01
1.99349457e-03 8.25757146e-01 -1.07847536e-02 -4.22225669e-02
4.81331021e-01 -6.79256022e-01 -9.47073519e-01 -3.66529115e-02
-1.26540894e-03 1.08359300e-01 -5.37517011e-01 1.46998239e+00
4.82308328e-01 3.55812818e-01 7.25933909e-01 -1.09615564e+00
-8.17041874e-01 4.54203814e-01 9.93484408e-02 2.22334117e-02
-7.56487343e-03 -6.97236061e-02 1.07295156e+00 -1.02818882e+00
4.37788516e-01 9.71720159e-01 1.07893503e+00 6.15598023e-01
-1.61338806e+00 -1.46396443e-01 -3.13409567e-01 2.18966901e-01
-1.52200699e+00 3.18871811e-02 1.05915415e+00 -6.26463830e-01
7.41116703e-01 3.25955570e-01 6.99477732e-01 8.11638474e-01
3.28018904e-01 7.33114839e-01 1.08393037e+00 -6.24544442e-01
4.89761174e-01 2.66534239e-01 1.59628004e-01 1.09398925e+00
-4.49892208e-02 2.27036148e-01 -1.43400490e-01 -4.66635674e-02
9.02079582e-01 -1.84933484e-01 -8.90116990e-01 -6.81963861e-01
-1.13144445e+00 1.28631246e+00 6.18184626e-01 7.24926710e-01
-5.17682195e-01 2.91586220e-02 6.08044088e-01 4.80704367e-01
4.18462902e-01 9.10006940e-01 -1.81248695e-01 -2.65671331e-02
-5.88455141e-01 3.06676090e-01 1.05936170e+00 8.52557182e-01
7.50228941e-01 2.52935767e-01 1.41603732e-02 4.47303981e-01
-3.62623185e-02 3.16891670e-01 5.86712480e-01 -1.19875193e+00
1.99289963e-01 6.20725989e-01 6.93974495e-02 -1.06029987e+00
-7.79872000e-01 -4.65104997e-01 -1.45416355e+00 3.48064601e-01
4.23116624e-01 -1.57330468e-01 -4.15440232e-01 1.96406269e+00
1.04226813e-01 2.38187723e-02 8.96880478e-02 8.90240371e-01
7.76324943e-02 5.22625804e-01 -3.91579151e-01 -4.83593553e-01
9.10943747e-01 -5.62960923e-01 -9.32248533e-01 3.56207311e-01
7.75457740e-01 -1.35585845e-01 1.14225805e+00 2.99205601e-01
-1.35397935e+00 -5.41493475e-01 -1.18126869e+00 3.86552922e-02
-4.93173182e-01 -1.01010323e-01 5.18778145e-01 2.38042340e-01
-1.31300235e+00 1.14793468e+00 -6.36462450e-01 -1.87062517e-01
3.67188424e-01 4.86005515e-01 -3.43263298e-01 3.23280036e-01
-1.54223263e+00 1.04607689e+00 6.35731995e-01 6.37867510e-01
-3.25882047e-01 -9.48670268e-01 -8.16968918e-01 2.06393808e-01
1.41242370e-01 -6.10059619e-01 1.07803607e+00 -1.06069064e+00
-1.58175874e+00 6.90699816e-01 1.23226002e-01 -7.45455563e-01
7.32455850e-01 1.22896507e-01 -3.13923389e-01 2.13368222e-01
-1.36182547e-01 2.50675380e-01 9.43165839e-01 -7.57037640e-01
-1.49748266e-01 -2.40280405e-01 2.12890446e-01 -2.48060912e-01
-7.12324083e-01 -3.13638091e-01 3.07575732e-01 -6.66182041e-01
6.46782070e-02 -9.11207199e-01 -3.98545682e-01 1.16208248e-01
-2.00301245e-01 -3.60416412e-01 7.45979309e-01 -4.27510142e-01
1.42829466e+00 -2.05721045e+00 9.12047446e-01 3.20324481e-01
3.21090698e-01 3.37726355e-01 4.95396256e-02 4.38691318e-01
-4.84794319e-01 6.46857619e-02 -8.34671557e-01 -1.02448404e-01
3.06822330e-01 3.11558545e-01 -3.56960475e-01 8.36716235e-01
4.75176752e-01 9.73716855e-01 -7.21250832e-01 -2.38142774e-01
1.76035866e-01 5.10967970e-01 -5.29205918e-01 3.02390680e-02
-1.59185410e-01 4.80786353e-01 -3.09761196e-01 6.70287311e-02
4.66735482e-01 -2.31577903e-01 -3.57826054e-01 -1.69777378e-01
-2.90732503e-01 -2.06294090e-01 -1.23731160e+00 1.66396177e+00
-6.24551713e-01 6.25053883e-01 4.85884160e-01 -1.87095642e+00
6.59021497e-01 6.37237549e-01 7.87878871e-01 -4.68209445e-01
4.06484932e-01 6.64531052e-01 6.56673834e-02 -6.23678505e-01
-5.81697077e-02 -6.77771330e-01 7.03552365e-02 2.98857480e-01
2.40545515e-02 5.23614623e-02 2.55626291e-01 -3.09344172e-01
1.09034812e+00 -2.99036205e-01 2.19993398e-01 -8.15294862e-01
1.17809868e+00 -2.68577844e-01 2.57318556e-01 6.16793454e-01
-2.64178276e-01 1.54172018e-01 7.62728512e-01 -6.67723656e-01
-1.01897621e+00 -1.08752263e+00 -6.56953514e-01 5.24366677e-01
-3.43091100e-01 8.73332247e-02 -8.59098136e-01 -3.95013034e-01
1.50482124e-02 3.49366009e-01 -9.43115234e-01 -5.81006408e-01
-7.37161815e-01 -3.30601305e-01 5.77075243e-01 5.15866995e-01
6.31121874e-01 -1.35330629e+00 -6.78557277e-01 2.77649224e-01
5.92318811e-02 -9.34632719e-01 -4.12601918e-01 5.20886958e-01
-8.96452248e-01 -8.64483774e-01 -1.17480254e+00 -8.95162165e-01
5.33740997e-01 -5.53064823e-01 7.36747444e-01 -2.52835751e-01
-2.69003302e-01 6.57619417e-01 6.71275184e-02 -3.45379084e-01
-4.84232038e-01 1.49855688e-01 5.01877367e-01 3.59053850e-01
1.19295031e-01 -9.11053717e-01 -2.61423707e-01 1.74494982e-01
-1.16267478e+00 -1.18622772e-01 3.91263962e-01 1.12920606e+00
5.37706017e-01 3.13499123e-01 6.57784522e-01 -5.27607560e-01
7.81232715e-01 -3.29275072e-01 -8.60645890e-01 -5.31775504e-02
-5.62300801e-01 5.91647565e-01 1.12231159e+00 -5.27817547e-01
-5.55081427e-01 8.86214599e-02 -1.58742964e-01 -7.33586729e-01
3.06668043e-01 8.25546324e-01 6.46621408e-03 -3.54320854e-01
8.05505633e-01 2.06435308e-01 2.00586766e-01 -3.64187002e-01
1.85807616e-01 5.06038010e-01 6.97272301e-01 -4.77193266e-01
8.11838150e-01 4.60750580e-01 6.55497253e-01 -9.85715806e-01
-9.50848699e-01 -4.45332259e-01 -7.59091794e-01 -9.87009630e-02
9.42763925e-01 -2.60826081e-01 -1.08832550e+00 3.29067200e-01
-1.33045208e+00 -4.27534401e-01 -8.25550199e-01 6.00378156e-01
-1.03497660e+00 2.41506264e-01 -8.41387272e-01 -1.06673348e+00
-2.15630159e-01 -1.16876340e+00 4.58124280e-01 -2.43885264e-01
2.43493076e-02 -1.47408509e+00 3.48571278e-02 -3.96928936e-01
5.72459698e-01 4.40233707e-01 1.03614128e+00 -1.85557693e-01
-1.31022960e-01 -3.58903915e-01 -1.13120884e-01 9.34159219e-01
-1.63627148e-01 -7.22705185e-01 -7.51113057e-01 -2.81459242e-01
7.99543738e-01 -1.73638508e-01 8.56783688e-01 4.10245121e-01
1.27279222e+00 -4.56375748e-01 1.93259701e-01 6.88434958e-01
1.38422310e+00 -1.18097989e-02 2.75512636e-01 -1.56680904e-02
6.73111260e-01 6.88353837e-01 -2.13101089e-01 1.94792569e-01
-2.50952423e-01 5.85584164e-01 3.54490370e-01 3.03809028e-02
3.96992832e-01 6.53355196e-02 3.05766791e-01 1.01409936e+00
-2.57772535e-01 4.22006607e-01 -7.60074019e-01 3.54301840e-01
-2.01878238e+00 -9.33119595e-01 -1.41074732e-01 2.19042277e+00
7.05616355e-01 9.08349976e-02 2.77648807e-01 6.70426965e-01
5.81918299e-01 1.11981984e-02 -6.39183044e-01 -7.12616205e-01
-2.90072560e-01 4.86448139e-01 5.13762832e-01 7.05768108e-01
-9.48545814e-01 3.51877898e-01 5.92453623e+00 5.45938432e-01
-1.19113612e+00 3.77828293e-02 2.12200671e-01 4.00470644e-01
1.17005119e-02 -4.07923520e-01 -2.51429945e-01 2.37417549e-01
1.12844718e+00 -3.12989920e-01 7.77619958e-01 8.65085304e-01
2.61671692e-01 4.58483219e-01 -1.32621539e+00 1.08500433e+00
-1.16011143e-01 -1.45228946e+00 -2.48405844e-01 4.25353527e-01
7.14634657e-01 -2.04999492e-01 1.55733839e-01 4.15546894e-01
-3.63883287e-01 -1.21282327e+00 6.14020586e-01 6.55768156e-01
8.09716702e-01 -8.68235528e-01 8.22089374e-01 3.74895632e-01
-1.16539896e+00 -4.14606273e-01 -4.47085381e-01 -4.27576780e-01
1.82730436e-01 4.25344110e-01 -2.93766081e-01 4.05888408e-01
3.33096921e-01 8.44762623e-01 -7.77334049e-02 7.89616406e-01
2.24519968e-01 2.67214775e-01 -1.91679850e-01 -1.32314399e-01
5.21397889e-01 -5.37130773e-01 6.02203906e-01 1.01028955e+00
4.15906131e-01 5.28998971e-02 7.94946402e-02 1.27310073e+00
-1.63335100e-01 2.06757993e-01 -8.46740365e-01 -2.61948347e-01
-2.69594222e-01 1.02136517e+00 -3.98671627e-01 -3.95233333e-02
-2.30239168e-01 1.11954641e+00 4.91693676e-01 5.47350824e-01
-7.65719414e-01 -6.74991310e-01 4.86486673e-01 6.62078336e-02
2.71284908e-01 -1.32138550e-01 -2.38724038e-01 -1.09972894e+00
3.54329705e-01 -4.34103429e-01 2.69239753e-01 -2.44308382e-01
-1.17347741e+00 4.30511773e-01 -9.99525422e-04 -1.02447820e+00
-2.42249578e-01 -1.23382664e+00 -4.80057329e-01 1.04450500e+00
-1.16715229e+00 -5.30709982e-01 1.09040089e-01 7.26356149e-01
-6.69379625e-03 -1.39699783e-02 1.08040643e+00 3.74410838e-01
-4.41944212e-01 4.40854043e-01 3.04521173e-01 2.54071712e-01
-1.76431745e-01 -1.71811414e+00 -9.84574631e-02 6.08675420e-01
1.16052190e-02 4.85632867e-01 7.89806187e-01 1.60483997e-02
-1.54694164e+00 -9.74217296e-01 8.83272469e-01 -2.16052637e-01
1.03643119e+00 -4.73057896e-01 -1.16320121e+00 5.74631810e-01
-1.40123963e-01 5.56918263e-01 2.58372635e-01 -9.76319313e-02
-1.12973705e-01 -1.78353921e-01 -1.00426424e+00 6.91397130e-01
1.00869882e+00 -8.68323624e-01 -3.46612364e-01 4.07330930e-01
7.54590154e-01 -2.57643551e-01 -1.09650135e+00 3.67126584e-01
2.47701213e-01 -8.03485036e-01 8.60413313e-01 -8.78575504e-01
3.65038007e-01 -3.72094736e-02 -9.97365415e-02 -1.22940373e+00
-3.19506854e-01 -1.05749583e+00 -6.38973117e-01 4.67479110e-01
3.29790533e-01 -7.97461867e-01 4.04981554e-01 3.46393287e-01
-4.20507759e-01 -1.18459606e+00 -1.45923388e+00 -9.90867853e-01
5.64771295e-01 -6.61522448e-01 6.63787946e-02 9.39400733e-01
5.45383275e-01 3.91735792e-01 -3.65245074e-01 -4.63818610e-02
4.78111655e-01 -2.46275708e-01 1.53034851e-01 -1.34306955e+00
-4.72525269e-01 -9.97441888e-01 -6.50854468e-01 -8.71670425e-01
6.51823759e-01 -1.25773895e+00 1.34413302e-01 -1.06859624e+00
-5.10917425e-01 -1.54268831e-01 -5.97292244e-01 4.77546342e-02
2.43910804e-01 1.46841556e-01 7.07844794e-02 -5.34905344e-02
-3.98663878e-01 8.29731166e-01 1.33183944e+00 -3.61566506e-02
-2.90923834e-01 1.24034867e-01 -2.94854194e-01 9.13962901e-01
6.21766031e-01 -7.49088544e-03 -3.78778607e-01 1.46985734e-02
4.12286222e-01 1.18887849e-01 5.79875886e-01 -1.43507433e+00
2.15052024e-01 2.82292306e-01 -2.07880400e-02 4.63588610e-02
2.24115506e-01 -9.37810302e-01 -2.13871136e-01 7.85164893e-01
-6.41409576e-01 2.13787854e-01 7.25733861e-02 7.55795717e-01
-2.54198849e-01 -4.28741634e-01 8.48790467e-01 -2.92103011e-02
-2.45531276e-01 3.52734298e-01 -3.66326541e-01 3.41758162e-01
8.19978774e-01 1.86388846e-02 4.60694075e-01 -2.33936310e-01
-9.38172817e-01 -9.23239738e-02 2.73761973e-02 -3.96999083e-02
3.76627326e-01 -1.63454485e+00 -3.99487972e-01 3.67078900e-01
-1.47606470e-02 -7.82052279e-02 1.67833269e-01 1.33103633e+00
-4.10041124e-01 6.21021628e-01 -2.28705574e-02 -5.73786438e-01
-4.52185571e-01 9.18340623e-01 7.71184504e-01 -4.14343178e-01
-9.51515853e-01 5.97031713e-01 4.77859825e-02 -4.17042285e-01
4.42337930e-01 -7.52221942e-01 1.79844141e-01 -2.10488319e-01
4.91337746e-01 4.38336849e-01 -9.47538242e-02 -6.86643004e-01
-2.79149432e-02 3.77052724e-01 4.17158216e-01 -1.55181736e-01
1.10950053e+00 1.84613258e-01 -3.27532470e-01 9.83135939e-01
1.84355378e+00 -4.08117175e-01 -1.13441217e+00 -3.80036354e-01
2.90841162e-01 3.27050924e-01 7.79342279e-02 -1.94791064e-01
-8.95434558e-01 1.04956663e+00 5.72877049e-01 7.94515431e-01
1.21757984e+00 -1.44490585e-01 8.14868450e-01 6.46190405e-01
1.68019295e-01 -1.26509345e+00 -1.19615443e-01 7.74551511e-01
1.18801105e+00 -1.07902575e+00 -5.95210254e-01 -5.64258061e-02
1.47855142e-03 1.34543765e+00 3.43798473e-02 -4.79942828e-01
9.99267101e-01 -4.36687768e-02 -4.60882425e-01 -1.19669579e-01
-7.47899860e-02 -1.61746413e-01 4.99385685e-01 3.84948194e-01
3.54229182e-01 -1.14010118e-01 -4.34851468e-01 4.70310569e-01
-1.60084084e-01 7.32860267e-02 4.07295525e-01 5.69205225e-01
-2.74850905e-01 -7.70819426e-01 -1.86609551e-01 3.14207166e-01
-3.23791713e-01 4.94535454e-02 -1.17196620e-01 9.35488641e-01
1.10497445e-01 4.09655333e-01 -1.54172286e-01 -1.24212041e-01
4.18149114e-01 3.97937417e-01 4.04307276e-01 -2.16056839e-01
-2.10201219e-01 -4.65463191e-01 -4.48106617e-01 -5.69412827e-01
-4.28257823e-01 -3.94804388e-01 -1.20377636e+00 -1.19063705e-01
-1.34717241e-01 4.45622176e-01 4.12597567e-01 1.02408254e+00
1.95233617e-02 6.08541191e-01 5.44056833e-01 -9.84590590e-01
-1.20360887e+00 -9.50704038e-01 -8.53469849e-01 3.35841030e-01
8.41652274e-01 -6.81389809e-01 -8.37967217e-01 -2.53690004e-01] | [7.526030540466309, 3.7129886150360107] |
193b167a-e1d3-4989-bf9b-faf60f3a77e3 | ref-rotation-equivariant-features-for-local | 2203.05206 | null | https://arxiv.org/abs/2203.05206v1 | https://arxiv.org/pdf/2203.05206v1.pdf | ReF -- Rotation Equivariant Features for Local Feature Matching | Sparse local feature matching is pivotal for many computer vision and robotics tasks. To improve their invariance to challenging appearance conditions and viewing angles, and hence their usefulness, existing learning-based methods have primarily focused on data augmentation-based training. In this work, we propose an alternative, complementary approach that centers on inducing bias in the model architecture itself to generate `rotation-specific' features using Steerable E2-CNNs, that are then group-pooled to achieve rotation-invariant local features. We demonstrate that this high performance, rotation-specific coverage from the steerable CNNs can be expanded to all rotation angles by combining it with augmentation-trained standard CNNs which have broader coverage but are often inaccurate, thus creating a state-of-the-art rotation-robust local feature matcher. We benchmark our proposed methods against existing techniques on HPatches and a newly proposed UrbanScenes3D-Air dataset for visual place recognition. Furthermore, we present a detailed analysis of the performance effects of ensembling, robust estimation, network architecture variations, and the use of rotation priors. | ['K. Madhava Krishna', 'Sourav Garg', 'Michael Milford', 'Avneesh Mishra', 'Kinal Mehta', 'Abhishek Peri'] | 2022-03-10 | null | null | null | null | ['visual-place-recognition'] | ['computer-vision'] | [ 1.19602151e-01 -9.35091749e-02 -1.93172559e-01 -5.44021130e-01
-5.25464892e-01 -4.50558275e-01 1.01439619e+00 -2.78032601e-01
-4.04379517e-01 5.29854536e-01 3.11967254e-01 -9.76303741e-02
-9.08600315e-02 -6.40158832e-01 -1.00345993e+00 -6.71424031e-01
4.72672936e-03 2.24011704e-01 2.10152194e-01 -5.09376347e-01
1.13840364e-01 1.15082467e+00 -1.87999773e+00 3.46146338e-02
6.15950763e-01 9.49093223e-01 8.09908062e-02 3.64881426e-01
4.67254013e-01 6.29274905e-01 -2.69983530e-01 1.56779200e-01
6.61685348e-01 3.08601502e-02 -5.60679257e-01 -5.89755736e-03
9.94189382e-01 -3.74529123e-01 -5.68362534e-01 7.96383619e-01
5.17730832e-01 4.12210464e-01 5.46235502e-01 -1.06659222e+00
-6.74046993e-01 2.29102775e-01 -4.86655772e-01 -5.66408746e-02
2.64256358e-01 5.27323894e-02 8.85122538e-01 -1.06775463e+00
8.51295531e-01 1.15227664e+00 8.87417078e-01 4.21519190e-01
-1.11799765e+00 -6.27400875e-01 2.48318046e-01 2.45519400e-01
-1.54078686e+00 -8.96011233e-01 1.01552093e+00 -3.07038724e-01
1.20806015e+00 1.96924686e-01 5.37405849e-01 1.17241478e+00
-1.16304673e-01 6.39116406e-01 9.55581307e-01 -5.48047900e-01
1.57741532e-01 -1.34192467e-01 -2.67464995e-01 6.64006174e-01
1.99001446e-01 5.31341016e-01 -5.03854871e-01 1.05438888e-01
1.12215507e+00 1.68055549e-01 -3.67424935e-01 -1.11491954e+00
-1.50648201e+00 7.31841922e-01 1.21295691e+00 1.70504823e-01
-4.31276590e-01 3.22127342e-01 1.59666896e-01 2.03002125e-01
4.29989189e-01 5.49312651e-01 -5.31900942e-01 3.61156762e-01
-9.31239605e-01 5.06031871e-01 3.81951332e-01 1.14277411e+00
1.09055197e+00 4.06229556e-01 -4.35380824e-02 9.16719019e-01
4.67537224e-01 4.25456256e-01 5.00498652e-01 -6.92166567e-01
3.67575198e-01 5.49493790e-01 1.41971469e-01 -1.33189261e+00
-7.19462216e-01 -6.82333112e-01 -9.17325258e-01 2.18407691e-01
2.26930201e-01 5.00864647e-02 -1.32647955e+00 1.83608067e+00
2.56832093e-01 1.41636744e-01 -1.97366662e-02 1.15916395e+00
8.51033509e-01 3.62280279e-01 -2.48667896e-01 4.98302042e-01
9.24282014e-01 -1.17994905e+00 -3.41896772e-01 -4.91458744e-01
5.47775686e-01 -7.14306295e-01 6.21258080e-01 -3.86959054e-02
-5.51863432e-01 -7.61590600e-01 -1.29692030e+00 -1.95721000e-01
-5.22206008e-01 4.34729129e-01 7.00402141e-01 3.67945880e-01
-1.29944038e+00 5.94847381e-01 -7.28553891e-01 -7.36474514e-01
4.84716803e-01 3.91188473e-01 -7.52262294e-01 -2.94924915e-01
-7.52494276e-01 1.17704368e+00 1.35726810e-01 3.86412114e-01
-8.81436050e-01 -4.65697557e-01 -1.29436791e+00 -1.48664355e-01
8.72433484e-02 -7.36985147e-01 9.17154372e-01 -7.59871840e-01
-1.56100750e+00 8.15345943e-01 -1.06565103e-01 -5.01332283e-01
3.06537747e-01 -3.45468730e-01 -2.23618239e-01 -1.21321753e-01
-7.83312693e-02 1.14726925e+00 9.63846207e-01 -1.38615584e+00
-3.20165515e-01 -5.05668342e-01 -7.32801929e-02 3.66251707e-01
-6.33280501e-02 -1.28034204e-01 -2.11292177e-01 -6.13690436e-01
4.93264288e-01 -1.21287107e+00 -5.88370621e-01 8.92214030e-02
-3.13796341e-01 1.95114881e-01 8.33403170e-01 -5.23589969e-01
5.41915417e-01 -2.18942285e+00 1.57061577e-01 4.22367841e-01
3.60232592e-02 2.35565051e-01 -4.88021642e-01 2.58645058e-01
-3.03629667e-01 -2.03241616e-01 -1.25014121e-02 -3.11869502e-01
-1.62379786e-01 1.23322792e-01 -2.24503517e-01 1.00459743e+00
3.47574443e-01 1.03811419e+00 -7.45623171e-01 1.68736745e-02
6.55179441e-01 6.95151329e-01 -5.48723936e-01 4.79608513e-02
6.51967674e-02 5.51102161e-01 -3.22343439e-01 7.46149063e-01
7.10568666e-01 1.34182319e-01 -1.81178600e-01 -4.88921642e-01
-4.27010298e-01 1.71810299e-01 -1.17418742e+00 2.03529596e+00
-6.41795635e-01 8.33959579e-01 3.49878147e-02 -1.12351573e+00
1.35377312e+00 -4.60267738e-02 3.48220348e-01 -6.55304909e-01
2.89239526e-01 2.92133242e-01 -1.19086079e-01 4.79273461e-02
7.12272108e-01 2.58713543e-01 6.37063757e-02 -1.54143780e-01
4.00602788e-01 -1.86761558e-01 -2.98996061e-01 -3.33429754e-01
9.34070826e-01 5.94984055e-01 3.36528897e-01 -2.99990475e-01
4.06354725e-01 -5.56818061e-02 3.02285641e-01 7.02234626e-01
-1.88791141e-01 1.06959260e+00 -3.19516063e-01 -7.66720891e-01
-1.33761418e+00 -6.53909683e-01 -2.08406180e-01 8.94350290e-01
1.92321226e-01 -5.70384823e-02 -3.29873025e-01 -4.90483552e-01
6.68651536e-02 2.82609373e-01 -7.04375923e-01 -2.85899520e-01
-6.58899784e-01 -4.08133328e-01 5.01005709e-01 6.85191751e-01
7.74509072e-01 -1.02334964e+00 -7.06596851e-01 1.31178617e-01
-6.34787604e-02 -1.19346261e+00 -7.07894266e-02 4.39509451e-01
-7.19458103e-01 -8.26235592e-01 -8.47786605e-01 -7.65630186e-01
8.05089772e-01 5.50947309e-01 8.25584888e-01 -1.83754966e-01
-1.93988219e-01 3.33345652e-01 -5.92661917e-01 -2.80790776e-01
1.47316918e-01 3.68415624e-01 1.16881721e-01 7.57871345e-02
1.11762479e-01 -6.85268044e-01 -5.47298491e-01 4.65932935e-01
-6.17670715e-01 1.01760052e-01 8.52961898e-01 9.71307516e-01
4.33631867e-01 -7.56542385e-01 3.27881962e-01 -4.24787909e-01
1.47791535e-01 -2.01878071e-01 -5.82904160e-01 3.13085839e-02
-2.48510167e-01 2.71732956e-01 5.77915609e-01 -2.74920881e-01
-8.90998781e-01 5.08298635e-01 -1.05682135e-01 -5.51736414e-01
-5.24707556e-01 4.00788516e-01 -1.18339576e-01 -7.55583882e-01
1.09499454e+00 2.05129668e-01 -6.74786270e-02 -3.85462672e-01
7.31292188e-01 3.77260208e-01 7.07071304e-01 -4.11853701e-01
1.17496061e+00 5.96497834e-01 1.81824610e-01 -9.57661033e-01
-5.63589036e-01 -7.78173506e-01 -7.80036211e-01 -6.58106282e-02
6.52532637e-01 -1.17443156e+00 -2.07966596e-01 7.05503464e-01
-1.12852716e+00 -2.70549715e-01 -2.91005194e-01 5.43163180e-01
-6.69090450e-01 4.91643474e-02 -1.49295673e-01 -4.07623529e-01
-2.56488889e-01 -1.12808847e+00 1.23539019e+00 2.74270177e-01
-6.06292300e-03 -7.28053629e-01 2.80014068e-01 2.90515218e-02
8.28446627e-01 4.61616874e-01 2.99720258e-01 -8.10015798e-01
-7.21639097e-01 -4.27487552e-01 -3.60683978e-01 2.58552969e-01
-5.13050370e-02 -1.80062026e-01 -1.18866479e+00 -4.39262480e-01
-5.17524123e-01 -5.09180129e-01 1.00728524e+00 3.99283797e-01
8.35018754e-01 -1.98763028e-01 -4.36512321e-01 1.07299626e+00
1.40813136e+00 -1.57989368e-01 7.34833717e-01 6.67472184e-01
9.40893650e-01 4.00409669e-01 5.36058128e-01 2.99679339e-01
4.24996287e-01 9.99050617e-01 6.67724788e-01 -4.33851212e-01
-2.66962767e-01 -2.56931007e-01 2.04877466e-01 4.37337339e-01
-4.61628199e-01 2.74751242e-02 -8.36960196e-01 8.39547455e-01
-2.03136349e+00 -8.86544228e-01 1.86916307e-01 2.25896621e+00
4.24231142e-01 -3.50129336e-01 -1.59283385e-01 -1.00308605e-01
7.05653012e-01 5.00562787e-01 -3.05791944e-01 -1.62824810e-01
-1.99253380e-01 3.62559915e-01 7.94377804e-01 3.68253410e-01
-1.38021815e+00 1.23554897e+00 6.26329374e+00 5.85840702e-01
-1.52499247e+00 -2.33407244e-01 3.29901099e-01 2.49926776e-01
-1.24731466e-01 9.99658480e-02 -7.31173813e-01 -3.13187480e-01
6.08833671e-01 5.50164759e-01 5.02390325e-01 1.14795816e+00
1.55918784e-02 1.28005549e-01 -1.06545389e+00 9.61804688e-01
3.10542643e-01 -1.32985497e+00 -1.28127754e-01 -6.10312484e-02
9.82788265e-01 6.16579473e-01 2.07035569e-03 3.38526130e-01
2.51973748e-01 -1.02553630e+00 6.88678384e-01 3.87833208e-01
6.60162508e-01 -6.71800196e-01 6.36091411e-01 7.83569515e-02
-1.14385533e+00 -6.95353746e-02 -5.51864564e-01 -1.27378181e-02
-1.52887136e-01 3.82474899e-01 -9.77949679e-01 8.48338366e-01
7.68711805e-01 9.81042624e-01 -8.03307176e-01 1.24725091e+00
-2.02287555e-01 9.63061899e-02 -5.67133188e-01 1.05964899e-01
2.62272835e-01 1.31018488e-02 5.33978224e-01 1.09221101e+00
3.72699499e-01 -4.44575012e-01 1.13970242e-01 8.42870057e-01
7.51458283e-04 5.84445372e-02 -1.12578487e+00 5.03708363e-01
5.22328794e-01 1.54092419e+00 -4.56662029e-01 -1.71032488e-01
-3.97720635e-01 8.07340324e-01 5.73208630e-01 4.74756449e-01
-5.50170600e-01 -4.35319006e-01 6.33965731e-01 -1.23505304e-02
6.48239493e-01 -4.62045044e-01 -1.03863090e-01 -1.35239685e+00
-2.47189798e-03 -9.04115558e-01 -2.39904389e-01 -8.74723017e-01
-7.88395047e-01 6.37284458e-01 -6.68465300e-03 -1.31668997e+00
-4.03676152e-01 -6.68321013e-01 -4.92453307e-01 7.84334481e-01
-1.90750134e+00 -1.73316097e+00 -6.98732674e-01 7.41665542e-01
2.51164079e-01 -2.66472220e-01 7.69186020e-01 2.40242049e-01
-2.84299284e-01 5.74784279e-01 9.53640118e-02 1.71401247e-01
9.07453179e-01 -8.91297817e-01 5.40735781e-01 1.08602500e+00
2.34132424e-01 6.85250103e-01 6.82980657e-01 -2.64132023e-01
-1.19297934e+00 -1.33805716e+00 5.68691134e-01 -2.23224312e-01
3.38629007e-01 -5.37277758e-01 -5.75164795e-01 8.39879811e-01
8.88079181e-02 5.75394213e-01 2.02009693e-01 2.38382936e-01
-6.79569185e-01 -3.62848967e-01 -1.17377818e+00 7.17956424e-01
1.22732770e+00 -5.65029502e-01 -4.43871915e-01 9.29793045e-02
5.50050735e-01 -5.28047025e-01 -6.15083933e-01 7.63363719e-01
5.98135948e-01 -9.71809506e-01 1.24120593e+00 -4.53711599e-01
2.24309310e-01 -4.34586614e-01 -3.44481975e-01 -1.58444333e+00
-5.28829396e-01 -3.87859017e-01 1.45533115e-01 9.48789477e-01
2.63992369e-01 -6.65458381e-01 6.95756137e-01 1.90868303e-01
-5.59044361e-01 -4.71417069e-01 -9.87509847e-01 -6.80227697e-01
-1.31697923e-01 -1.78250343e-01 5.62232673e-01 1.15764451e+00
-2.08271533e-01 1.26618564e-01 -5.78681648e-01 2.74425387e-01
2.80213267e-01 1.00489082e-02 1.10438943e+00 -1.10201705e+00
3.56206782e-02 -2.71557331e-01 -9.59282756e-01 -1.17239404e+00
1.65007934e-01 -7.69703865e-01 4.35614467e-01 -1.27746511e+00
-1.22328207e-01 -5.15266240e-01 -1.61030874e-01 7.95058370e-01
2.26781711e-01 4.97328609e-01 1.12824745e-01 1.83877498e-01
-5.37147224e-01 8.42681408e-01 1.21176827e+00 6.40714541e-02
-2.56984979e-01 -1.48597270e-01 -4.73612040e-01 5.65090835e-01
7.55738556e-01 -1.43507197e-01 -2.94314951e-01 -4.12911475e-01
3.06616902e-01 -4.17074442e-01 7.26120353e-01 -1.34977818e+00
2.08691120e-01 -6.09199852e-02 6.11685991e-01 -6.69775486e-01
4.40164864e-01 -9.26876724e-01 -3.07027418e-02 2.36768156e-01
-1.92137867e-01 1.21422999e-01 3.36206734e-01 4.60238367e-01
-3.44405860e-01 1.11425892e-01 7.87433207e-01 -8.10369700e-02
-8.78828287e-01 3.68219078e-01 -2.39444360e-01 -4.71379757e-01
8.47736180e-01 -3.04847062e-01 -4.74116981e-01 -3.54355872e-01
-4.59569752e-01 -1.47851795e-01 6.95742726e-01 5.64028025e-01
8.13586593e-01 -1.50982583e+00 -5.82101762e-01 6.21249497e-01
4.15116668e-01 4.51333165e-01 3.67752276e-02 7.64011323e-01
-6.66786969e-01 6.36562824e-01 -6.26940191e-01 -6.71625912e-01
-8.76839459e-01 4.73900646e-01 4.65612710e-01 3.70259248e-02
-5.10839105e-01 6.89546764e-01 6.31895959e-02 -1.06366003e+00
1.81898102e-02 -4.57220763e-01 -3.97340536e-01 -1.92932412e-01
6.89289719e-02 1.31067157e-01 3.98907751e-01 -8.98427427e-01
-5.58853865e-01 7.52453864e-01 3.29127908e-02 2.98593659e-02
1.61050141e+00 -2.57590674e-02 6.45831376e-02 -8.34425017e-02
1.17069376e+00 -3.32025439e-02 -1.30207598e+00 -3.88843685e-01
-2.81959444e-01 -4.98223245e-01 2.37595394e-01 -4.78255391e-01
-1.14653635e+00 6.41832471e-01 7.58667767e-01 -3.51086169e-01
9.07025695e-01 -1.27807707e-01 1.92895636e-01 8.16591263e-01
4.60280478e-01 -7.96465099e-01 -1.08150057e-01 7.67286897e-01
1.21958017e+00 -1.31854403e+00 9.33848768e-02 -3.01138997e-01
-3.65060806e-01 1.12600589e+00 6.70304120e-01 -5.79704165e-01
6.25154674e-01 -1.61679223e-01 1.52775496e-01 -1.41952485e-01
-2.65593737e-01 -3.57688755e-01 6.08009219e-01 1.00370932e+00
2.20991954e-01 -1.77712530e-01 2.29653910e-01 6.07051924e-02
-1.90476924e-01 -1.81489930e-01 2.69993782e-01 9.89652038e-01
-3.25387388e-01 -8.75170827e-01 -4.68145370e-01 5.01379222e-02
9.39227641e-02 -1.54926836e-01 -3.03295642e-01 9.11437869e-01
-9.20758992e-02 4.95732158e-01 3.35163921e-02 -5.41442335e-01
3.05708528e-01 -1.23408601e-01 6.71354413e-01 -4.54701543e-01
-3.06476474e-01 -2.01106623e-01 1.68585315e-01 -7.27261901e-01
-7.82472730e-01 -7.38090217e-01 -7.16150880e-01 -1.08956657e-01
-5.79838037e-01 -4.49013710e-01 9.49739635e-01 9.89534199e-01
6.61281407e-01 3.40423942e-01 6.11360312e-01 -1.65973139e+00
-5.33159912e-01 -1.05656040e+00 -2.16886073e-01 3.66776019e-01
5.55999339e-01 -9.30566013e-01 -1.77029893e-01 -1.80597037e-01] | [7.795804023742676, -1.972265362739563] |
2e64de5f-c3c1-4efd-bde6-7c5913b544f6 | qursim-a-corpus-for-evaluation-of-relatedness | null | null | https://aclanthology.org/L12-1051 | https://aclanthology.org/L12-1051.pdf | QurSim: A corpus for evaluation of relatedness in short texts | This paper presents a large corpus created from the original Quranic text, where semantically similar or related verses are linked together. This corpus will be a valuable evaluation resource for computational linguists investigating similarity and relatedness in short texts. Furthermore, this dataset can be used for evaluation of paraphrase analysis and machine translation tasks. Our dataset is characterised by: (1) superior quality of relatedness assignment; as we have incorporated relations marked by well-known domain experts, this dataset could thus be considered a gold standard corpus for various evaluation tasks, (2) the size of our dataset; over 7,600 pairs of related verses are collected from scholarly sources with several levels of degree of relatedness. This dataset could be extended to over 13,500 pairs of related verses observing the commutative property of strongly related pairs. This dataset was incorporated into online query pages where users can visualize for a given verse a network of all directly and indirectly related verses. Empirical experiments showed that only 33{\%} of related pairs shared root words, emphasising the need to go beyond common lexical matching methods, and incorporate -in addition- semantic, domain knowledge, and other corpus-based approaches. | ['Abdul-Baquee Sharaf', 'Eric Atwell'] | 2012-05-01 | null | null | null | lrec-2012-5 | ['text-clustering'] | ['natural-language-processing'] | [-7.75703117e-02 5.25246840e-03 -3.68557900e-01 -4.45300192e-02
-7.77937949e-01 -1.05623853e+00 8.76049101e-01 6.85098112e-01
-4.72152084e-01 7.83659220e-01 7.31904626e-01 -1.14481449e-01
-5.76828778e-01 -8.71275306e-01 -1.87174052e-01 -3.31579328e-01
7.36805424e-02 8.04935098e-01 3.53693575e-01 -1.08777773e+00
7.70423770e-01 5.02666354e-01 -1.54487479e+00 2.03383476e-01
9.97283340e-01 5.07967114e-01 2.93311685e-01 1.21501714e-01
-2.32687265e-01 7.55015254e-01 -7.75544882e-01 -9.68976438e-01
4.00923081e-02 -7.72612691e-01 -1.22630143e+00 -5.55912435e-01
5.90390503e-01 4.56822842e-01 -3.14602554e-01 1.05406678e+00
5.56241095e-01 3.73403043e-01 5.34623504e-01 -1.18259096e+00
-6.10436738e-01 9.40122604e-01 -6.10310212e-02 5.73064685e-01
9.62123096e-01 -4.58907992e-01 1.68738961e+00 -8.14250886e-01
1.25119197e+00 1.02530241e+00 3.79267663e-01 1.53593123e-01
-8.42465937e-01 -5.48062265e-01 -8.61792624e-01 7.85787940e-01
-1.37760615e+00 -1.03255257e-01 8.92407060e-01 -3.71252686e-01
9.04643357e-01 4.57400382e-01 6.69137836e-01 1.09142649e+00
-1.11779660e-01 2.42498726e-01 1.08314741e+00 -6.64975464e-01
-1.60191625e-01 3.35483074e-01 1.58712193e-01 3.40711206e-01
1.49250366e-02 -2.85461396e-01 -7.68932879e-01 -2.14282334e-01
4.20711070e-01 -5.26348889e-01 -3.82809609e-01 -1.35077223e-01
-1.57953501e+00 7.43268073e-01 4.89931047e-01 1.11099052e+00
-1.54076383e-01 -1.51236266e-01 7.07354665e-01 7.38068223e-01
8.73513967e-02 1.12440288e+00 -5.96219338e-02 -4.41483140e-01
-7.44896531e-01 4.27857846e-01 1.03919733e+00 9.63387609e-01
7.81597376e-01 -6.43818021e-01 1.39581650e-01 9.73905444e-01
-1.09665059e-01 1.36657238e-01 7.43449628e-01 -1.05855632e+00
7.60852158e-01 7.59166598e-01 -2.60090232e-01 -1.49013495e+00
-2.35320121e-01 -3.47494632e-01 -4.54580635e-01 -4.15646285e-01
5.48501313e-01 3.61132413e-01 1.81549206e-01 1.66234827e+00
5.42311631e-02 -2.92194784e-01 8.65850896e-02 8.51402342e-01
1.07196701e+00 6.04494095e-01 -2.83494264e-01 -3.15022141e-01
1.58382154e+00 -6.56311750e-01 -7.77413309e-01 1.58919424e-01
8.40976954e-01 -1.35878634e+00 1.23688245e+00 1.60750002e-01
-1.22684634e+00 -5.98077655e-01 -1.08999074e+00 -2.52790123e-01
-6.08906746e-01 -4.67254639e-01 3.78540218e-01 4.51694041e-01
-8.14786136e-01 1.17464018e+00 -1.83106765e-01 -8.95838022e-01
8.64164531e-02 -1.36599867e-02 -3.68028462e-01 -8.56555626e-03
-1.64352369e+00 1.64682817e+00 5.74839950e-01 -3.72356474e-01
-1.10396847e-01 -6.29014432e-01 -6.35589063e-01 -5.63602597e-02
4.06054437e-01 -4.76457328e-01 5.29295683e-01 -6.69197202e-01
-9.29760098e-01 1.31918132e+00 8.33691284e-02 -1.19930282e-01
2.32115462e-01 -3.83179821e-02 -7.67573655e-01 5.17862499e-01
4.01000530e-01 1.79497704e-01 1.20107993e-01 -8.94577742e-01
-2.67275512e-01 -2.43192002e-01 2.37545818e-01 4.84543860e-01
-7.01497138e-01 5.72149932e-01 -3.32637668e-01 -1.02430832e+00
2.07609177e-01 -8.66657794e-01 3.30368876e-01 -1.21242605e-01
1.59897618e-02 -6.53366387e-01 5.73684394e-01 -8.39583218e-01
1.54136026e+00 -1.66235280e+00 5.10191321e-01 3.38634163e-01
1.19802222e-01 -2.30168607e-02 -2.65329689e-01 1.30556440e+00
-6.87856004e-02 1.05033211e-01 -1.90257519e-01 2.56628394e-01
3.12778614e-02 2.63233244e-01 -9.85179916e-02 3.20834279e-01
-2.04848677e-01 9.60139871e-01 -1.21164751e+00 -9.66381371e-01
2.88488176e-02 1.24984242e-01 -6.68413341e-02 4.01696824e-02
1.03493407e-01 9.10632312e-02 -3.87465209e-01 5.49189270e-01
-5.46598844e-02 -1.03002854e-01 2.54618555e-01 -4.28465933e-01
1.43268391e-01 7.86525965e-01 -8.67915750e-01 2.18022013e+00
-5.01751125e-01 1.09667838e+00 -5.92999518e-01 -8.84778619e-01
1.23152590e+00 3.95615131e-01 5.35098016e-01 -9.16514993e-01
1.27244651e-01 4.37748611e-01 1.52785435e-01 -7.12282121e-01
8.69087160e-01 -2.31084064e-01 -2.63902605e-01 6.69671476e-01
1.33816496e-01 -5.42058110e-01 8.11079204e-01 5.54250002e-01
1.00805819e+00 2.46654063e-01 5.18447280e-01 -6.69860899e-01
6.94143951e-01 4.17620540e-01 1.46940738e-01 2.58361995e-01
7.54420608e-02 2.58375853e-01 4.69740421e-01 -2.36388713e-01
-1.17258251e+00 -9.90050912e-01 -4.63685930e-01 1.15664744e+00
4.68433052e-01 -8.51187885e-01 -3.71588737e-01 -2.95102984e-01
-2.90748656e-01 7.66711295e-01 -4.78767395e-01 6.53118314e-03
-9.26719606e-01 -4.00685310e-01 9.27616000e-01 2.36576825e-01
1.74227521e-01 -1.11938465e+00 -4.15413052e-01 2.00364202e-01
-8.41402292e-01 -9.27825809e-01 -2.58999795e-01 -1.67882130e-01
-6.26802981e-01 -1.28379118e+00 -5.82510591e-01 -9.04432058e-01
9.34195518e-03 7.66498670e-02 1.64358282e+00 4.71337378e-01
-7.18025491e-02 -4.83040586e-02 -7.44311690e-01 8.55509490e-02
-8.22106481e-01 2.15011671e-01 -4.87132519e-02 -5.46422541e-01
4.64873374e-01 -9.01952624e-01 -2.60316283e-01 6.24187231e-01
-7.13611245e-01 -2.94425011e-01 1.66717753e-01 6.60664260e-01
7.61533976e-02 -4.05025929e-01 6.66903198e-01 -6.31009936e-01
8.86649489e-01 -6.67591393e-01 -1.44972458e-01 4.41061705e-01
-5.39443612e-01 -9.40532051e-03 6.71853244e-01 -3.24890345e-01
-7.32547164e-01 -6.88010037e-01 -8.71246606e-02 1.91377133e-01
2.55657673e-01 6.89927936e-01 -6.46768957e-02 -6.53948784e-02
1.03127980e+00 -5.62471412e-02 -1.01734668e-01 -5.58700502e-01
5.30900836e-01 8.30088973e-01 6.55238688e-01 -8.67097974e-01
7.97964454e-01 1.09224349e-01 2.62274444e-01 -6.84787989e-01
-6.14480257e-01 -8.51973295e-01 -8.01352143e-01 -2.75022924e-01
4.70859706e-01 -6.77333057e-01 -7.07396567e-01 -3.53928447e-01
-9.99143958e-01 2.29133174e-01 -2.83655465e-01 5.04189074e-01
-5.70727706e-01 7.96711743e-01 -6.44362628e-01 4.62569483e-02
-2.79020756e-01 -6.84478104e-01 6.38348579e-01 1.52770072e-01
-9.75039303e-01 -1.29997015e+00 3.33808213e-01 7.58104503e-01
2.19205767e-01 3.70987296e-01 1.28874588e+00 -1.09713030e+00
-8.93631577e-03 -1.16841510e-01 -6.28767759e-02 1.06768608e-01
2.29814053e-01 -5.00682592e-02 -4.04540867e-01 -2.21389994e-01
-1.75962195e-01 -6.75032079e-01 2.20016226e-01 -5.31490445e-01
4.64115590e-01 -1.73891306e-01 -1.48616061e-01 -6.04117801e-03
1.43299377e+00 -8.12424049e-02 7.41288066e-01 7.68383682e-01
5.55682778e-01 1.10013652e+00 7.26889789e-01 2.38651291e-01
3.71988565e-01 1.05307007e+00 2.08277479e-01 3.26657623e-01
-3.53846818e-01 -2.26592451e-01 1.02287546e-01 1.42381620e+00
-4.45086420e-01 -2.73400486e-01 -1.02216971e+00 7.71503508e-01
-1.64175522e+00 -1.36369753e+00 -5.36055863e-01 1.98572850e+00
1.09774470e+00 2.51953509e-02 2.49156550e-01 3.14521223e-01
6.62509561e-01 1.37492418e-01 1.62341222e-01 -4.45998222e-01
-4.81710941e-01 4.97455984e-01 2.27700826e-02 3.16288620e-01
-4.54465896e-01 9.10868764e-01 6.03240871e+00 1.26311374e+00
-6.85715616e-01 -2.99868826e-03 -8.55671540e-02 1.18385695e-01
-6.45395041e-01 3.34922671e-01 -4.74094659e-01 3.65068555e-01
7.06283927e-01 -6.09848320e-01 2.76239783e-01 4.33703274e-01
-7.66962953e-03 9.84158274e-03 -1.14670432e+00 8.43946159e-01
4.29266036e-01 -1.67020357e+00 6.71190694e-02 -3.00970115e-02
6.79122388e-01 -5.40919900e-02 -5.62526703e-01 5.80842793e-03
6.47286624e-02 -9.95335400e-01 5.99857211e-01 3.86983484e-01
6.37312710e-01 -7.39546061e-01 8.09428155e-01 2.64521599e-01
-1.13912702e+00 3.53904337e-01 -3.89936388e-01 -1.44311294e-01
1.12462237e-01 2.27347715e-03 -4.74705309e-01 9.56246972e-01
5.57239532e-01 1.03154290e+00 -9.20405984e-01 7.89717793e-01
-3.73864263e-01 3.67576420e-01 -3.33722621e-01 -6.69152200e-01
2.93900430e-01 -4.59443986e-01 6.89867616e-01 1.28587365e+00
2.73827255e-01 -4.16136570e-02 -2.84910083e-01 7.56754696e-01
-1.86569080e-01 7.25388467e-01 -4.19738919e-01 -8.72460660e-03
1.01559091e+00 1.26891398e+00 -7.67861664e-01 -2.54147202e-01
-2.95497209e-01 9.90119934e-01 3.32097322e-01 -1.48359776e-01
-6.98834717e-01 -8.17194998e-01 3.64940375e-01 1.65460438e-01
-1.50840819e-01 -1.03363164e-01 -1.61398664e-01 -9.79599893e-01
1.62629202e-01 -1.01353049e+00 7.23502457e-01 -9.75794375e-01
-1.66448367e+00 8.54700387e-01 1.82218090e-01 -1.34643281e+00
-3.33456665e-01 -3.19094241e-01 -3.74976277e-01 9.48072612e-01
-1.09381652e+00 -9.72048163e-01 -2.11827978e-01 6.88933551e-01
3.69728357e-01 -2.37733260e-01 9.32744920e-01 4.68892545e-01
-3.55767123e-02 3.93590629e-01 7.90377483e-02 2.72762477e-01
8.71631265e-01 -1.11439598e+00 1.21181756e-01 6.18227422e-01
6.89122021e-01 7.40013123e-01 8.80010605e-01 -4.15401191e-01
-1.07634008e+00 -3.93498600e-01 1.68516648e+00 -7.09370673e-01
1.26690269e+00 -7.42614567e-02 -1.08314359e+00 2.45906129e-01
6.92005336e-01 -4.15509492e-01 1.03882098e+00 9.64425057e-02
-4.83086526e-01 -1.19982706e-02 -8.98508966e-01 6.28216803e-01
1.38099182e+00 -6.64249241e-01 -1.40035045e+00 6.85130656e-01
5.37597895e-01 -2.09098592e-01 -1.49963939e+00 1.64467245e-01
3.99983823e-01 -9.02012289e-01 9.49202836e-01 -6.56383395e-01
8.91164839e-01 -1.73936188e-01 -2.95793235e-01 -1.15521181e+00
-1.16728023e-01 -8.28442812e-01 2.08936766e-01 1.48580897e+00
4.02418941e-01 -1.46339163e-01 5.49509645e-01 2.53217578e-01
-1.42697334e-01 -4.74026293e-01 -1.11456215e+00 -9.83182907e-01
4.84560013e-01 -2.08382890e-01 3.82093877e-01 1.49164760e+00
7.22337484e-01 7.28108048e-01 -1.51373306e-02 -5.20299733e-01
3.08666855e-01 4.28766698e-01 4.53717113e-01 -1.16330028e+00
-2.16157168e-01 -8.60013068e-01 -5.63714266e-01 -8.56191933e-01
1.28713459e-01 -1.33783519e+00 -4.92894292e-01 -1.52228987e+00
1.13141224e-01 -5.67538142e-01 1.31160706e-01 2.54503250e-01
-9.47757438e-03 7.45547056e-01 1.56516075e-01 7.21017063e-01
-2.91352063e-01 3.61299038e-01 1.33978844e+00 2.13506952e-01
2.54039336e-02 -3.98251981e-01 -5.55243552e-01 5.62381625e-01
6.10403776e-01 -6.83728933e-01 -2.30160534e-01 -1.46856219e-01
8.47754538e-01 1.39770761e-01 1.53056324e-01 -6.05925977e-01
3.34061801e-01 -2.91523308e-01 -2.96173513e-01 -4.46278870e-01
2.73988456e-01 -8.74767840e-01 3.01504493e-01 4.35010076e-01
-4.82941061e-01 4.06247914e-01 -1.70305535e-01 8.13200995e-02
-4.86857086e-01 -7.20025659e-01 4.96468663e-01 -2.12171495e-01
-7.71027148e-01 -3.87381554e-01 -2.29884565e-01 5.72425842e-01
9.72877502e-01 -5.05839646e-01 -3.57798100e-01 -3.08755547e-01
-5.21281242e-01 -6.96931854e-02 5.11160553e-01 6.37312770e-01
2.57723004e-01 -1.51941419e+00 -1.08533788e+00 -6.86125040e-01
6.11926675e-01 -6.58088148e-01 -2.40224838e-01 5.84741950e-01
-7.65609801e-01 1.29055798e-01 -5.05972207e-01 -2.44891644e-01
-1.40805900e+00 4.33177859e-01 -8.71497020e-02 -2.00788692e-01
-5.41592002e-01 5.77419102e-01 -6.99697852e-01 -3.17222744e-01
-7.98428506e-02 2.29304597e-01 -4.76412147e-01 6.08276606e-01
2.28907615e-01 6.31713808e-01 1.23677298e-01 -1.09891307e+00
-4.54013348e-01 7.42257059e-01 3.72733444e-01 -1.81835607e-01
1.38893533e+00 -3.85726720e-01 -6.35583460e-01 5.80330670e-01
1.37099326e+00 3.37638021e-01 -1.99982896e-02 -5.01854181e-01
5.16429126e-01 -5.99652469e-01 -4.66512561e-01 -7.66487122e-01
-5.15330315e-01 6.24034286e-01 -2.04849377e-01 2.68488795e-01
9.73230302e-01 4.29314584e-01 7.71071672e-01 6.95311546e-01
2.86936164e-01 -9.31994379e-01 3.45285892e-01 3.51905704e-01
9.91133273e-01 -9.73729312e-01 2.80530542e-01 -4.78869885e-01
-6.31567657e-01 1.47546804e+00 1.81286588e-01 -1.01466693e-01
1.88912839e-01 -5.04685864e-02 9.65270028e-03 -4.01454836e-01
-4.33085114e-01 -2.11652339e-01 6.80111945e-01 4.21129555e-01
7.39941895e-01 -3.93949181e-01 -1.22457623e+00 2.11331517e-01
-8.42690051e-01 -4.09968823e-01 6.20274007e-01 7.05488563e-01
-2.92712152e-01 -1.55892265e+00 -2.52010047e-01 1.03851013e-01
-3.50662857e-01 -5.25388479e-01 -5.90049386e-01 1.04533505e+00
8.32701996e-02 9.82088089e-01 7.95537010e-02 -1.74159527e-01
2.17659250e-01 -1.80072173e-01 7.30823755e-01 -4.17062104e-01
-1.17959321e+00 -5.03286302e-01 5.74039698e-01 -3.12190503e-01
-7.86572635e-01 -5.93999445e-01 -1.03916156e+00 -6.60193324e-01
-4.12526011e-01 5.62677085e-01 6.13719821e-01 1.08802855e+00
4.17547533e-03 8.76085311e-02 5.50154805e-01 -2.40983337e-01
-3.44521046e-01 -1.11097372e+00 -4.14301604e-01 1.03908920e+00
-5.23345709e-01 -3.61395150e-01 -2.94617206e-01 2.75553703e-01] | [10.868712425231934, 9.353446006774902] |
ee860e6f-c3ab-4aa5-89aa-b6fc33ca7a85 | a-general-framework-for-information | 1904.03296 | null | http://arxiv.org/abs/1904.03296v1 | http://arxiv.org/pdf/1904.03296v1.pdf | A General Framework for Information Extraction using Dynamic Span Graphs | We introduce a general framework for several information extraction tasks
that share span representations using dynamically constructed span graphs. The
graphs are constructed by selecting the most confident entity spans and linking
these nodes with confidence-weighted relation types and coreferences. The
dynamic span graph allows coreference and relation type confidences to
propagate through the graph to iteratively refine the span representations.
This is unlike previous multi-task frameworks for information extraction in
which the only interaction between tasks is in the shared first-layer LSTM. Our
framework significantly outperforms the state-of-the-art on multiple
information extraction tasks across multiple datasets reflecting different
domains. We further observe that the span enumeration approach is good at
detecting nested span entities, with significant F1 score improvement on the
ACE dataset. | ['Mari Ostendorf', 'Yi Luan', 'Hannaneh Hajishirzi', 'Dave Wadden', 'Amy Shah', 'Luheng He'] | 2019-04-05 | a-general-framework-for-information-1 | https://aclanthology.org/N19-1308 | https://aclanthology.org/N19-1308.pdf | naacl-2019-6 | ['joint-entity-and-relation-extraction'] | ['natural-language-processing'] | [ 1.04594275e-01 9.83974636e-01 -6.28211379e-01 -3.03186655e-01
-1.02975321e+00 -7.36900508e-01 4.09716368e-01 7.26009429e-01
-4.11858439e-01 1.05736208e+00 5.36469340e-01 -6.97934031e-02
-4.20473695e-01 -9.26509380e-01 -7.52307773e-01 1.21929727e-01
-6.89449370e-01 7.33476162e-01 2.69418478e-01 -7.46759698e-02
-4.58520651e-02 2.94636905e-01 -9.16309416e-01 6.40895307e-01
6.88936293e-01 1.02385700e+00 -3.25125456e-01 3.83952081e-01
-2.22668558e-01 7.67466843e-01 -4.72766012e-01 -8.46705556e-01
9.74103808e-02 2.09831491e-01 -1.43391728e+00 -6.49828017e-01
4.84323591e-01 -8.97254571e-02 -4.72119033e-01 7.99962759e-01
1.73496246e-01 2.31705785e-01 4.50155228e-01 -1.15795112e+00
-4.71352041e-01 1.78060806e+00 -6.64815545e-01 4.04546648e-01
7.37472653e-01 -4.64261532e-01 1.78956294e+00 -7.96795130e-01
1.00490010e+00 1.26286030e+00 9.14264858e-01 1.63732350e-01
-1.53227723e+00 -7.82637656e-01 5.36010981e-01 2.28616968e-01
-1.31310439e+00 -3.52206975e-01 5.40424228e-01 -1.08458482e-01
1.78382301e+00 1.18990645e-01 3.41731757e-01 1.18540013e+00
-1.53274537e-04 7.48190403e-01 5.22990286e-01 -4.89766449e-01
-3.36956382e-01 -2.88046569e-01 6.76148474e-01 8.65301728e-01
6.90767169e-01 -7.82564953e-02 -8.54665637e-01 -3.66229862e-01
3.82590115e-01 -4.33976859e-01 -1.18055701e-01 4.76080589e-02
-1.23581982e+00 7.85917759e-01 6.00360870e-01 4.14708644e-01
-4.55375463e-01 1.95736930e-01 5.53018093e-01 3.60538542e-01
4.28392857e-01 8.89541030e-01 -9.15249228e-01 2.99808443e-01
-1.20627749e+00 3.52280438e-01 1.22939932e+00 1.29266191e+00
6.03702605e-01 -3.81953418e-01 -5.60251057e-01 6.16587579e-01
2.11622059e-01 -1.08380280e-01 -5.82309961e-02 -1.07080126e+00
9.59599793e-01 6.24679625e-01 -3.46785448e-02 -8.06856096e-01
-9.28759336e-01 -5.95191419e-01 -4.30222690e-01 -2.81191170e-01
4.41889942e-01 -6.07188284e-01 -8.11977923e-01 2.13426089e+00
1.68848932e-01 1.53052747e-01 6.63247108e-02 3.37725520e-01
1.08144045e+00 2.32197344e-01 5.63364685e-01 -2.28902280e-01
1.70816934e+00 -8.44594002e-01 -7.10199535e-01 -4.78871554e-01
6.99756265e-01 -3.16461086e-01 2.49416962e-01 1.47682667e-01
-1.13555801e+00 -3.27805161e-01 -1.25100994e+00 -3.52392524e-01
-5.18020451e-01 -2.83362240e-01 9.35535073e-01 -5.96144870e-02
-8.59439194e-01 9.27032471e-01 -5.78963339e-01 -2.35298991e-01
3.09692651e-01 2.06982017e-01 -6.97308958e-01 3.14170629e-01
-1.95890057e+00 1.44078124e+00 1.16891396e+00 -1.10232189e-01
-5.73704123e-01 -9.10416543e-01 -1.36773443e+00 5.15467644e-01
6.97460175e-01 -8.92325759e-01 1.25142705e+00 -1.49367034e-01
-5.86956918e-01 9.33700383e-01 -1.75411448e-01 -1.05718541e+00
2.26787463e-01 -6.36081874e-01 -6.78878665e-01 7.98083320e-02
2.55351245e-01 8.37605357e-01 2.53863752e-01 -1.09160507e+00
-7.16330528e-01 -1.88510388e-01 8.21306631e-02 9.40419659e-02
1.76130593e-01 1.29619524e-01 -3.02525043e-01 -3.15165162e-01
2.58317832e-02 -6.01491868e-01 -1.68516263e-01 -4.87555891e-01
-8.22589576e-01 -8.08417499e-01 4.34463352e-01 -9.05228674e-01
1.73186731e+00 -1.42138731e+00 1.13865778e-01 3.27544451e-01
5.25729179e-01 -1.35506228e-01 -1.21848144e-01 6.87009454e-01
-2.30574548e-01 2.97426701e-01 2.18216865e-03 -3.15398365e-01
8.30919296e-02 -6.09263368e-02 -2.85729915e-01 2.19607074e-02
4.26823437e-01 1.01608372e+00 -1.00678444e+00 -8.65030408e-01
-3.81314039e-01 1.75707266e-01 -2.95771271e-01 8.62425193e-03
-4.81585592e-01 -1.58345670e-01 -3.32752407e-01 3.77582163e-01
3.57990861e-01 -6.10179007e-01 6.67583704e-01 -5.07081032e-01
3.19021530e-02 1.10091400e+00 -1.02268720e+00 1.99767375e+00
-2.55497783e-01 4.71864074e-01 -1.08566210e-01 -7.86290646e-01
8.75040889e-01 5.96646905e-01 4.86873269e-01 -2.03344628e-01
-2.13797882e-01 -5.84194763e-03 1.27829770e-02 -4.12344843e-01
6.88354611e-01 2.58966833e-01 -5.28699458e-01 4.80962276e-01
6.10086799e-01 3.28744471e-01 7.19817758e-01 7.49160647e-01
1.23812115e+00 2.56817222e-01 6.29862607e-01 -1.95684761e-01
2.72736549e-01 -2.23786980e-01 8.92735898e-01 7.06859887e-01
1.37965590e-01 6.29520565e-02 8.37840497e-01 -4.62692350e-01
-7.41398573e-01 -1.16842997e+00 -1.67938843e-01 1.23738837e+00
-2.56648302e-01 -8.41192842e-01 -4.46975917e-01 -1.29943085e+00
3.30827624e-01 9.74856973e-01 -8.52998376e-01 8.69562402e-02
-7.45636284e-01 -1.39588386e-01 1.02300096e+00 8.67292285e-01
4.73475128e-01 -1.27333581e+00 -3.86200070e-01 5.24026275e-01
-5.64158499e-01 -1.36197770e+00 -3.21265042e-01 4.72285360e-01
-6.37639165e-01 -1.40918291e+00 -3.17751504e-02 -9.55353200e-01
1.51881814e-01 -5.59289217e-01 1.79601443e+00 4.41539427e-03
7.33704641e-02 -9.25248787e-02 -1.09448649e-01 -2.33190402e-01
-2.05447767e-02 7.50078440e-01 -2.55158037e-01 -7.22092271e-01
7.16008186e-01 -6.55954361e-01 -4.25942391e-01 -1.83355033e-01
-3.72774899e-01 4.63713035e-02 3.99357110e-01 7.65344977e-01
2.79358357e-01 -4.04440850e-01 9.37409878e-01 -1.49136174e+00
6.54212892e-01 -7.94399917e-01 -2.36189276e-01 6.86907709e-01
-8.43715131e-01 4.57393378e-01 1.83465868e-01 -7.11096600e-02
-1.08290708e+00 -2.07327992e-01 1.06447041e-02 -2.94908974e-02
-6.17260747e-02 9.73757327e-01 9.60092917e-02 5.12863457e-01
6.70110226e-01 -5.58592498e-01 -6.52018130e-01 -4.55228269e-01
6.57594323e-01 1.66273087e-01 6.58378720e-01 -9.26105678e-01
4.86496180e-01 -3.92932966e-02 3.02088615e-02 -2.00803712e-01
-1.46664262e+00 -2.13410571e-01 -8.04147720e-01 2.62525737e-01
6.71174943e-01 -9.34805512e-01 -7.06032395e-01 -1.45518526e-01
-1.48196435e+00 3.66060473e-02 -3.37789714e-01 3.18628430e-01
-3.39725703e-01 3.48049223e-01 -1.21325970e+00 -6.06553793e-01
-7.99317241e-01 -4.68889147e-01 9.17563260e-01 3.27406049e-01
-9.45858896e-01 -1.08337760e+00 1.26642063e-01 4.19162102e-02
6.72679469e-02 6.11181498e-01 1.19621789e+00 -1.15655446e+00
-2.61895239e-01 -2.37683654e-02 -4.47133064e-01 -5.83439708e-01
-7.61472955e-02 -9.50782597e-02 -9.67413664e-01 -6.36570230e-02
-7.00534642e-01 -5.75313330e-01 1.27011883e+00 2.25639701e-01
9.79849875e-01 -4.93081361e-01 -1.01906204e+00 5.46503067e-01
1.21187532e+00 -7.73110464e-02 2.15476587e-01 3.10657740e-01
3.85331839e-01 6.35152876e-01 4.71619576e-01 2.25295126e-01
6.23917341e-01 3.20888966e-01 1.15884528e-01 4.74218056e-02
-1.39480159e-02 -7.49954283e-01 -7.43754301e-03 2.70183653e-01
-5.93998022e-02 -2.31967136e-01 -7.82921195e-01 7.39880145e-01
-1.96123946e+00 -1.10371125e+00 2.23075170e-02 1.81275916e+00
1.40721881e+00 6.30003035e-01 8.14300850e-02 2.38293670e-02
6.88291192e-01 4.71750587e-01 -5.47398150e-01 -4.44490194e-01
-1.26232766e-02 4.55742806e-01 4.55684304e-01 8.30591440e-01
-1.35721529e+00 1.13178599e+00 6.86816740e+00 4.28651780e-01
-3.77885371e-01 -1.39442578e-01 4.03476477e-01 -1.38643667e-01
-4.76770639e-01 1.55958995e-01 -1.14173782e+00 8.37012082e-02
9.50970054e-01 -2.80084461e-01 2.06844330e-01 6.21109486e-01
-4.67638642e-01 1.54751716e-02 -1.63375914e+00 3.84035230e-01
-2.47358292e-01 -1.42466533e+00 -1.03360236e-01 -2.86264718e-01
4.61002141e-01 2.41254270e-01 -4.18368131e-01 5.69324493e-01
1.04648066e+00 -1.12617147e+00 3.67408752e-01 5.81426203e-01
1.19434571e+00 -7.22997427e-01 5.45989811e-01 9.08557177e-02
-1.56804347e+00 1.09734982e-02 -8.53940845e-02 2.72439420e-01
5.26562572e-01 7.54047453e-01 -9.28534865e-01 8.76038134e-01
6.39148653e-01 5.78289688e-01 -5.08204877e-01 8.22598994e-01
-6.45102978e-01 4.36200202e-01 -3.90913039e-01 2.81719059e-01
8.72179717e-02 1.80245697e-01 6.45383537e-01 1.82730305e+00
9.69638973e-02 -1.50683776e-01 2.36817479e-01 1.06164992e+00
-5.59218526e-01 -2.62221158e-01 -6.73449993e-01 -7.54686669e-02
1.10930073e+00 1.46915770e+00 -5.39248228e-01 -5.06187320e-01
-4.19030696e-01 4.88136649e-01 1.16153026e+00 4.27949011e-01
-6.36372805e-01 -7.65063047e-01 4.75492358e-01 -1.72581047e-01
3.87541860e-01 -2.55994916e-01 -4.70161706e-01 -9.50055540e-01
-2.67733663e-01 -5.66223860e-01 1.40639675e+00 -4.54920053e-01
-1.26198101e+00 7.08653808e-01 3.46763968e-01 -3.90029132e-01
-6.67030513e-01 -2.80800045e-01 -6.08403146e-01 8.22732627e-01
-1.36156881e+00 -1.43523288e+00 7.52556771e-02 3.16552341e-01
2.04181775e-01 1.94741756e-01 1.21712768e+00 -7.07333162e-02
-3.70942116e-01 8.09404969e-01 -7.39562452e-01 6.91391349e-01
7.42310524e-01 -1.46913266e+00 8.20806086e-01 8.87445450e-01
4.76300240e-01 9.03516412e-01 7.33596981e-01 -1.08067989e+00
-6.23239934e-01 -8.53964627e-01 1.49823034e+00 -4.26896930e-01
6.73700869e-01 -3.70551586e-01 -9.93585229e-01 1.42655432e+00
7.62849510e-01 -1.41562074e-01 7.01443732e-01 1.04991794e+00
-1.07002294e+00 1.25311971e-01 -1.18712842e+00 2.26438656e-01
1.52033794e+00 -4.46737766e-01 -1.08505762e+00 1.78636521e-01
1.26573443e+00 -3.97222489e-01 -1.37257624e+00 8.05352211e-01
5.13502479e-01 -5.42130888e-01 8.19483221e-01 -1.08856761e+00
3.64373833e-01 1.71733961e-01 1.88935980e-01 -1.37948751e+00
-6.56641483e-01 -7.57609427e-01 -1.06549931e+00 1.34133244e+00
1.29266226e+00 -4.03508693e-01 6.55112684e-01 7.29152381e-01
6.46684840e-02 -8.86641920e-01 -6.25808537e-01 -4.18481320e-01
-2.27342565e-02 -2.43147224e-01 7.87466049e-01 9.66847718e-01
7.20846951e-01 9.38394785e-01 -6.25533313e-02 1.90656364e-01
7.28475094e-01 5.86757362e-01 1.09190652e-02 -1.65425074e+00
-4.08100337e-01 -3.62087101e-01 3.08572859e-01 -6.67796433e-01
4.73702252e-01 -1.10879207e+00 -1.72300622e-01 -1.83727944e+00
1.69534937e-01 -3.69495571e-01 -6.46402776e-01 8.73510957e-01
-4.31410879e-01 -2.91772723e-01 1.38859143e-02 -3.12965751e-01
-9.43165064e-01 -4.08952124e-02 6.34289265e-01 -1.19116388e-01
-1.57165498e-01 -2.14485034e-01 -1.09287882e+00 7.23860919e-01
7.57622540e-01 -5.07912993e-01 -2.68431365e-01 -5.35568357e-01
6.26427650e-01 4.26693231e-01 -2.40866423e-01 -7.82149553e-01
4.14323032e-01 1.09028727e-01 7.58525372e-01 -7.58487642e-01
2.97507048e-01 -3.98647636e-01 9.07135382e-02 3.36695254e-01
-9.39178884e-01 8.69608298e-02 3.56433302e-01 3.72391403e-01
9.64630488e-03 -1.14368081e-01 3.94499809e-01 -2.37344012e-01
-4.79079694e-01 3.51468295e-01 9.72926803e-03 5.89590609e-01
6.91053331e-01 2.37532452e-01 -7.18912780e-01 -3.30281258e-01
-1.02150655e+00 7.10094690e-01 -1.91383347e-01 4.80994612e-01
4.73154694e-01 -1.17338824e+00 -8.77729714e-01 -1.57851093e-02
1.92524232e-02 3.32048237e-01 -1.77812472e-01 4.69364375e-01
1.44596070e-01 5.69556475e-01 -3.21613178e-02 -1.13605184e-03
-1.23999691e+00 6.99083686e-01 1.79272786e-01 -1.26908958e+00
-6.27637446e-01 1.10983920e+00 -3.05425823e-01 -2.28716373e-01
3.76800597e-01 -4.91342157e-01 -5.75624764e-01 4.23544168e-01
4.60932910e-01 3.94982040e-01 7.07185864e-02 -1.22553326e-01
-5.26456475e-01 9.20284167e-02 -5.24801016e-01 -2.05591887e-01
1.40837872e+00 1.81104504e-02 -1.30379096e-01 2.20245391e-01
7.13678479e-01 1.07216820e-01 -9.26822126e-01 -6.04965806e-01
9.77989316e-01 1.83759496e-01 -4.54261422e-01 -1.03235900e+00
-7.98763454e-01 3.20505381e-01 -4.19261336e-01 4.55961615e-01
5.95234990e-01 3.33809316e-01 8.56744707e-01 5.23798883e-01
4.58930254e-01 -1.16293299e+00 -4.01908159e-01 6.75665081e-01
1.03370893e+00 -9.13561404e-01 1.85673982e-01 -6.09213889e-01
-6.23846471e-01 1.03783166e+00 9.18586195e-01 -2.22171828e-01
6.80711985e-01 7.31624246e-01 -4.53232616e-01 -4.83690739e-01
-1.27216339e+00 -2.99833387e-01 5.99966049e-01 4.83432621e-01
9.46928799e-01 7.60361478e-02 -4.61244345e-01 1.01661563e+00
-3.78007561e-01 -1.94268897e-01 -6.17126338e-02 5.07691443e-01
-4.72896725e-01 -9.54105616e-01 1.66514054e-01 7.22758710e-01
-7.74807870e-01 -3.77997994e-01 -5.36470413e-01 8.01996648e-01
2.28116717e-02 1.10542643e+00 1.40217483e-01 -3.77936482e-01
2.06211701e-01 7.08902180e-01 6.71393812e-01 -7.33315110e-01
-7.96236038e-01 -3.64648223e-01 1.06078100e+00 -7.80978799e-01
-1.86353683e-01 -5.93266189e-01 -1.50013995e+00 -9.53100845e-02
-4.34268236e-01 5.42047560e-01 -1.62993511e-03 9.64520752e-01
5.14387429e-01 5.65476358e-01 1.31676510e-01 -3.47207218e-01
-4.06130195e-01 -1.42484462e+00 -2.96177626e-01 4.24482942e-01
3.11628968e-01 -7.07300544e-01 4.00910489e-02 -3.46180588e-01] | [9.33401870727539, 8.968188285827637] |
6e8372cc-71be-4ae6-ac05-251f8cfaa642 | active-learning-with-gaussian-processes-for | 1901.06803 | null | http://arxiv.org/abs/1901.06803v1 | http://arxiv.org/pdf/1901.06803v1.pdf | Active Learning with Gaussian Processes for High Throughput Phenotyping | A looming question that must be solved before robotic plant phenotyping
capabilities can have significant impact to crop improvement programs is
scalability. High Throughput Phenotyping (HTP) uses robotic technologies to
analyze crops in order to determine species with favorable traits, however, the
current practices rely on exhaustive coverage and data collection from the
entire crop field being monitored under the breeding experiment. This works
well in relatively small agricultural fields but can not be scaled to the
larger ones, thus limiting the progress of genetics research. In this work, we
propose an active learning algorithm to enable an autonomous system to collect
the most informative samples in order to accurately learn the distribution of
phenotypes in the field with the help of a Gaussian Process model. We
demonstrate the superior performance of our proposed algorithm compared to the
current practices on sorghum phenotype data collection. | ['Katia Sycara', 'Sumit Kumar', 'George Kantor', 'Wenhao Luo'] | 2019-01-21 | null | null | null | null | ['plant-phenotyping'] | ['computer-vision'] | [ 2.96213269e-01 1.43852979e-01 -3.42469096e-01 -1.89428285e-01
4.48205806e-02 -1.02848208e+00 -3.42966676e-01 5.34131050e-01
2.86944360e-02 8.01585138e-01 -4.69819635e-01 -5.43297827e-01
-5.82730711e-01 -1.15582454e+00 -4.65904176e-01 -1.00034904e+00
-1.90926984e-01 7.27111518e-01 2.06966162e-01 -1.83676526e-01
-2.54586083e-03 6.60270870e-01 -1.64782643e+00 -2.09739208e-01
1.38963282e+00 6.70365393e-01 1.11445773e+00 7.53797829e-01
-1.55006543e-01 1.38229683e-01 -3.69036496e-01 3.23003858e-01
2.57204771e-01 -8.40830728e-02 -4.82639402e-01 -2.83372719e-02
-3.55174750e-01 -4.55089718e-01 4.56355155e-01 9.74400103e-01
6.30992532e-01 -1.84033334e-01 4.62650210e-01 -1.23053098e+00
-6.86674893e-01 9.30100143e-01 -7.48356819e-01 -4.72154975e-01
-7.09088705e-03 8.05518106e-02 6.45064235e-01 -2.11992070e-01
6.89661860e-01 9.78893518e-01 5.09073734e-01 -3.64013277e-02
-1.32048559e+00 -3.13443691e-01 3.45544852e-02 2.06237927e-01
-1.09213269e+00 -1.21952951e-01 5.02378404e-01 -3.37507963e-01
4.34647739e-01 6.99293092e-02 1.01790571e+00 5.59052944e-01
2.73988366e-01 8.12245011e-01 5.35075963e-01 -3.00558358e-01
8.95424724e-01 -1.46448746e-01 1.84972718e-01 4.54868466e-01
7.75508523e-01 6.35436624e-02 -2.27827847e-01 -2.25793108e-01
5.54304957e-01 -8.04956555e-02 -2.45247483e-02 -1.04912317e+00
-6.83265567e-01 7.60943115e-01 4.49398369e-01 5.54771647e-02
-9.73710775e-01 -1.76533133e-01 2.01968774e-01 -1.08868301e-01
1.40125796e-01 8.25316429e-01 -8.81633222e-01 -2.18104422e-01
-6.69561148e-01 1.44994199e-01 9.36275721e-01 1.23152399e+00
6.70880377e-01 -6.43816665e-02 9.50360373e-02 6.05707049e-01
3.41918558e-01 6.69519663e-01 -1.55167254e-02 -1.07251120e+00
-2.80350804e-01 9.41339493e-01 4.51828957e-01 -8.45024467e-01
-5.14222205e-01 -5.47523834e-02 -5.04190505e-01 3.73944849e-01
3.83693159e-01 -5.38796782e-01 -6.22390389e-01 1.32668817e+00
4.07959610e-01 -3.32225621e-01 1.93722248e-01 4.62236583e-01
4.92358446e-01 6.21832907e-01 2.23246306e-01 -2.67800003e-01
1.00181508e+00 -1.50571227e-01 -6.52277291e-01 -1.84853002e-02
8.17157626e-01 -7.62435079e-01 7.62989759e-01 5.81705034e-01
-5.30211926e-01 -4.91911530e-01 -1.13703120e+00 4.39451903e-01
-6.55238509e-01 7.92411149e-01 1.34624636e+00 8.87810409e-01
-8.75388324e-01 6.41367316e-01 -1.19612348e+00 -8.87637496e-01
5.81969202e-01 3.24423283e-01 -3.27953845e-01 -1.56962067e-01
-4.51062322e-01 9.19157922e-01 7.79127717e-01 6.09164834e-01
-1.05324638e+00 -7.92289436e-01 -4.68561232e-01 1.66118309e-01
6.39000714e-01 -1.98376596e-01 8.02446902e-01 -4.53703135e-01
-2.03854227e+00 5.86020410e-01 3.66621800e-02 -2.82031447e-01
-8.23703557e-02 -4.71350789e-01 2.60744914e-02 -6.16464950e-02
-2.20843464e-01 8.28700602e-01 3.54380608e-01 -1.01837289e+00
-7.34539270e-01 -6.68983757e-01 -6.12927899e-02 -1.13264648e-02
-3.11544210e-01 -1.05183661e-01 2.75976449e-01 8.65398124e-02
5.46095490e-01 -1.41916156e+00 -4.27974194e-01 3.57700437e-01
-2.21579045e-01 2.80854464e-01 1.35950565e+00 -7.28090525e-01
3.14741641e-01 -1.93832326e+00 1.53105572e-01 -3.03967088e-01
-2.19858944e-01 5.20018876e-01 -1.80919021e-01 5.19675970e-01
2.77636498e-01 -1.70401767e-01 -1.54097289e-01 5.87730110e-01
-2.68669128e-01 3.83657277e-01 -5.33136129e-02 3.51674050e-01
4.45481002e-01 6.25438869e-01 -9.45523202e-01 -6.35473132e-02
3.66554648e-01 4.61737067e-02 -2.54416615e-01 3.75391275e-01
-5.03309131e-01 4.56222385e-01 -7.42695749e-01 9.85246658e-01
1.31127012e+00 1.38629884e-01 3.87225568e-01 1.01748221e-01
-6.22750938e-01 -5.43986082e-01 -1.07914305e+00 1.65700245e+00
-7.81943724e-02 1.99586555e-01 4.61614788e-01 -1.20224881e+00
1.43409705e+00 2.03413785e-01 7.83957601e-01 2.41872191e-01
6.74963603e-03 1.66126609e-01 3.42696965e-01 -7.17111051e-01
4.79208291e-01 6.93196535e-01 2.07289428e-01 2.06094936e-01
2.94556230e-01 -3.10050964e-01 3.78416598e-01 -1.34802967e-01
1.03314960e+00 8.00379515e-01 3.67936164e-01 -6.98732495e-01
6.60989583e-02 6.64983869e-01 9.83812511e-01 6.55877650e-01
-5.60025990e-01 1.20704025e-01 6.83999956e-01 -8.17462280e-02
-9.22096610e-01 -8.55022669e-01 -1.76309615e-01 1.09565055e+00
2.72816837e-01 2.18196899e-01 -4.70891267e-01 -4.87816989e-01
2.64992505e-01 8.12434793e-01 -2.68188775e-01 -1.72652006e-01
4.39997464e-02 -1.52234280e+00 5.15629232e-01 4.27669376e-01
5.61705053e-01 -1.08421433e+00 -1.06466568e+00 4.31977540e-01
1.30127221e-01 -8.33935857e-01 7.50088513e-01 7.48738587e-01
-8.89390349e-01 -8.63105118e-01 -6.66760147e-01 -7.06988454e-01
5.09879649e-01 3.88145626e-01 5.95248401e-01 -5.33392668e-01
-7.50990868e-01 -2.61964202e-01 -7.74783969e-01 -1.20721412e+00
-5.11639714e-01 4.64495718e-01 -2.44778603e-01 -4.77948070e-01
4.36007172e-01 -6.60374284e-01 -2.91007727e-01 2.35297874e-01
-4.46795791e-01 -1.03046089e-01 9.01341915e-01 7.72091329e-01
3.73463005e-01 4.03098315e-01 7.19519734e-01 -7.88543046e-01
2.21331045e-01 -5.28281629e-01 -1.17922640e+00 9.24184322e-01
-2.37060636e-01 -7.41662979e-02 5.26960790e-01 -5.03229380e-01
-1.20285487e+00 6.91085935e-01 3.16821545e-01 4.72297192e-01
-6.91446304e-01 6.65421367e-01 -5.95609426e-01 -1.07871160e-01
5.22036433e-01 -7.58056641e-02 2.19536602e-01 -3.22557926e-01
4.90141779e-01 5.77763021e-01 3.50195199e-01 -6.22107983e-01
5.04307389e-01 1.66268140e-01 3.91983211e-01 -1.04642117e+00
-4.08395827e-01 -5.39344609e-01 -1.11744678e+00 -1.45435527e-01
5.87158024e-01 -5.47316253e-01 -9.73584771e-01 5.47622979e-01
-7.19855189e-01 -4.77901608e-01 -2.45400965e-01 9.87049341e-01
-7.00035095e-01 -1.22891106e-01 -1.82282194e-01 -1.00870013e+00
-3.76006454e-01 -1.10881186e+00 8.56042922e-01 6.56740367e-01
-9.03383493e-02 -5.79803169e-01 1.46522552e-01 -1.09835871e-01
6.68407798e-01 3.17776442e-01 1.13218331e+00 -2.84493148e-01
-5.16241491e-01 -4.76648152e-01 -3.03719401e-01 9.34465900e-02
4.22854304e-01 7.27143764e-01 -9.35908377e-01 -6.50503263e-02
-2.02756241e-01 -4.92651761e-01 6.55390918e-02 8.81891906e-01
9.49698389e-01 6.05471551e-01 -4.35482740e-01 5.47158539e-01
1.48610961e+00 6.11304760e-01 6.61229432e-01 7.55408630e-02
3.25278282e-01 9.24206197e-01 1.32731092e+00 4.63664860e-01
-2.09331751e-01 2.49135479e-01 5.40429235e-01 -6.83011860e-02
2.89256036e-01 -8.80878791e-02 9.11655575e-02 2.35506147e-01
-1.29228473e-01 -3.90882939e-01 -1.17239952e+00 4.70477641e-01
-2.14444160e+00 -8.19082558e-01 -2.22336709e-01 2.15596962e+00
5.07687569e-01 -2.18151331e-01 -2.53551662e-01 1.67631075e-01
7.61901498e-01 -3.36027503e-01 -9.37541246e-01 -2.24530354e-01
-3.30283225e-01 4.08254527e-02 1.00037539e+00 3.72359785e-03
-1.12093651e+00 1.11765146e+00 6.84409952e+00 3.12437326e-01
-1.18675208e+00 -3.89834642e-01 3.52810174e-01 4.97219712e-01
3.25948238e-01 6.96403861e-01 -8.73906612e-01 -4.54164818e-02
6.22489631e-01 -6.80587590e-02 3.02163571e-01 1.19087303e+00
5.53653598e-01 -6.29217148e-01 -7.05725789e-01 3.35392743e-01
-4.62533921e-01 -7.89394915e-01 -4.19951409e-01 6.09674677e-02
6.00948095e-01 -3.04640085e-01 -2.44195566e-01 4.94590029e-02
1.01950562e+00 -6.09003603e-01 1.82507351e-01 4.07299846e-01
3.07305396e-01 -5.46691358e-01 7.53456473e-01 5.71204960e-01
-9.00536060e-01 -3.55917811e-01 -1.03060758e+00 -9.85163227e-02
2.06500441e-01 8.75172675e-01 -1.50342715e+00 8.26804459e-01
5.97619712e-01 5.59683442e-01 -5.93605936e-01 1.35161221e+00
-9.91001576e-02 9.82837796e-01 -5.02118230e-01 -2.80147642e-01
-3.49022076e-02 -4.89293665e-01 3.75554860e-01 4.70632672e-01
8.36494446e-01 -1.63062587e-01 2.84911871e-01 9.76567984e-01
4.94299471e-01 2.41096407e-01 -5.78045428e-01 -4.41046625e-01
6.94465935e-01 1.49752772e+00 -1.13589871e+00 1.48229346e-01
1.87120046e-02 8.78218830e-01 1.30145460e-01 1.31551549e-01
-3.59614462e-01 -4.84887958e-01 1.85319260e-01 -2.42298484e-01
3.76082093e-01 -4.98770356e-01 -5.00967383e-01 -3.87354434e-01
-3.46442163e-01 -5.57379544e-01 -1.21089086e-01 -9.21165586e-01
-8.71415317e-01 -2.68533409e-01 -4.18751389e-02 -7.08195806e-01
-1.45602286e-01 -8.45273316e-01 -4.86750424e-01 1.01830077e+00
-9.31708276e-01 -1.24521136e+00 -6.77428126e-01 -2.13734433e-01
2.37461507e-01 -3.74924213e-01 1.51085293e+00 -1.75158963e-01
-7.92063415e-01 4.70363721e-02 7.05981135e-01 -4.27528888e-01
6.16211236e-01 -1.14022338e+00 1.01242959e-01 9.29482639e-01
-4.10190403e-01 2.84769177e-01 7.44064510e-01 -1.10305715e+00
-1.98762417e+00 -9.85932648e-01 2.53352493e-01 1.03073224e-01
5.64812541e-01 -3.15970004e-01 -6.30327404e-01 4.15448219e-01
-1.66615963e-01 -1.59340993e-01 5.33599555e-01 4.64909405e-01
3.86357009e-01 -3.08583140e-01 -1.34233570e+00 1.70790076e-01
6.45436227e-01 1.26331419e-01 2.04670161e-01 1.70874119e-01
4.89041448e-01 5.51839881e-02 -8.95500362e-01 6.02060556e-01
6.99632525e-01 -1.54211894e-01 6.18019938e-01 -4.72963452e-01
1.43932298e-01 -3.63782018e-01 -2.64388025e-01 -1.43925583e+00
-5.94448864e-01 -6.79909527e-01 2.30589271e-01 1.49696946e+00
3.74176413e-01 -5.72607815e-01 9.23725665e-01 3.90669674e-01
1.20148599e-01 -2.19334513e-01 -5.61831333e-02 -6.67974949e-01
-8.33897144e-02 1.26451463e-01 9.20727372e-01 7.59735763e-01
-1.22439705e-01 -1.91739753e-01 -1.26098186e-01 7.17368782e-01
5.82908750e-01 1.59208670e-01 8.98480833e-01 -1.62605679e+00
-1.54523775e-01 3.92225012e-02 -6.54817522e-01 -3.69601011e-01
-4.92075160e-02 -5.04788160e-01 6.00592911e-01 -1.30290246e+00
3.22496474e-01 -1.04265392e+00 2.30311289e-01 3.42513740e-01
-4.69906181e-01 -6.72779202e-01 -2.45509252e-01 -2.01579973e-01
9.71140712e-02 2.10508868e-01 9.08804238e-01 1.07848294e-01
-6.65761709e-01 2.36495361e-01 -7.09659219e-01 5.81078589e-01
1.07833838e+00 -4.80722368e-01 -8.86288881e-01 -4.77813214e-01
8.33420306e-02 9.51994956e-02 8.69405642e-02 -9.36299086e-01
3.60865518e-02 -5.48313081e-01 5.48439145e-01 -9.90101755e-01
1.09738901e-01 -7.28929520e-01 4.38687742e-01 5.14847696e-01
-4.01071995e-01 -1.03645129e-02 2.20992073e-01 4.24037367e-01
2.11698174e-01 -7.01278985e-01 5.35685658e-01 4.45660651e-02
-8.41178775e-01 1.99894145e-01 -5.54499269e-01 -6.63679004e-01
1.38927960e+00 1.27627254e-01 -4.54006612e-01 -3.27051413e-04
-5.72712839e-01 2.17367426e-01 4.36975926e-01 3.70037496e-01
1.66107431e-01 -7.00369418e-01 -8.98677707e-01 8.83446261e-02
1.36945948e-01 -6.83647618e-02 2.52980471e-01 4.71044481e-01
-1.35090029e+00 4.16506261e-01 -9.47405338e-01 -9.03376460e-01
-1.17225611e+00 6.83141291e-01 -1.47177085e-01 -6.24617524e-02
-3.63545656e-01 5.52011013e-01 -1.25446677e-01 -6.71943545e-01
1.64167047e-01 -2.04494759e-01 -4.35699701e-01 4.48932908e-02
2.74482787e-01 3.80411685e-01 7.47150108e-02 -1.40720233e-01
-8.91302302e-02 8.26399401e-02 2.17061549e-01 -4.11724970e-02
1.61625612e+00 1.96358282e-02 -1.63734794e-01 7.21313477e-01
3.22833180e-01 -3.82304192e-01 -1.21023703e+00 3.35329980e-01
3.05293560e-01 -4.49058354e-01 2.00370923e-01 -8.39565516e-01
-7.35349000e-01 6.05941594e-01 1.12366688e+00 3.22818041e-01
1.09170818e+00 -4.45492774e-01 -4.65971185e-03 7.81939924e-01
6.11742020e-01 -1.21251380e+00 -6.71793342e-01 1.94882527e-01
5.46296597e-01 -1.31413543e+00 -1.78321972e-02 -8.64595413e-01
-4.84167784e-01 1.08880186e+00 3.36667150e-01 -1.34841725e-01
7.48331726e-01 7.40488112e-01 -5.42188324e-02 7.86990002e-02
-6.99931979e-01 -2.16636181e-01 -6.54238939e-01 1.33034420e+00
6.54262185e-01 7.16764390e-01 -3.07616770e-01 2.48401642e-01
-9.12460983e-02 3.87809217e-01 6.23339057e-01 1.39261210e+00
-6.44213259e-01 -1.20527279e+00 -5.96334934e-01 4.96850938e-01
6.05618395e-02 3.76618952e-01 -4.91894901e-01 3.69835943e-01
2.57692844e-01 9.31422293e-01 -2.67485589e-01 3.90597582e-02
1.06649823e-01 -7.08964420e-03 5.42114556e-01 -6.06034279e-01
5.56344464e-02 -1.48853108e-01 -4.72756922e-02 -2.98125505e-01
-1.71890229e-01 -9.58296418e-01 -7.28729606e-01 -3.03865254e-01
-7.40807593e-01 -1.06942333e-01 1.28802466e+00 4.08846617e-01
5.72072148e-01 4.77207243e-01 6.55154169e-01 -5.88790238e-01
-5.16348302e-01 -1.15610588e+00 -1.00292456e+00 -2.55940586e-01
-3.36909235e-01 -8.42107713e-01 2.77490109e-01 9.62089971e-02] | [9.119287490844727, -1.5927984714508057] |
ea855aa2-82b4-4273-9aee-05e8087abc90 | park-detect-towards-efficient-multi-task | 2302.13263 | null | https://arxiv.org/abs/2302.13263v1 | https://arxiv.org/pdf/2302.13263v1.pdf | PaRK-Detect: Towards Efficient Multi-Task Satellite Imagery Road Extraction via Patch-Wise Keypoints Detection | Automatically extracting roads from satellite imagery is a fundamental yet challenging computer vision task in the field of remote sensing. Pixel-wise semantic segmentation-based approaches and graph-based approaches are two prevailing schemes. However, prior works show the imperfections that semantic segmentation-based approaches yield road graphs with low connectivity, while graph-based methods with iterative exploring paradigms and smaller receptive fields focus more on local information and are also time-consuming. In this paper, we propose a new scheme for multi-task satellite imagery road extraction, Patch-wise Road Keypoints Detection (PaRK-Detect). Building on top of D-LinkNet architecture and adopting the structure of keypoint detection, our framework predicts the position of patch-wise road keypoints and the adjacent relationships between them to construct road graphs in a single pass. Meanwhile, the multi-task framework also performs pixel-wise semantic segmentation and generates road segmentation masks. We evaluate our approach against the existing state-of-the-art methods on DeepGlobe, Massachusetts Roads, and RoadTracer datasets and achieve competitive or better results. We also demonstrate a considerable outperformance in terms of inference speed. | ['Ming Wu', 'Chuang Zhang', 'Junli Yang', 'Zhenglin Xian', 'Wanfeng Zheng', 'Shenwei Xie'] | 2023-02-26 | null | null | null | null | ['road-segementation', 'keypoint-detection'] | ['computer-vision', 'computer-vision'] | [ 4.59758013e-01 1.10008689e-02 -9.94201973e-02 -3.62382531e-01
-7.88095713e-01 -5.56493521e-01 5.65543830e-01 -2.42603421e-02
-3.86763722e-01 5.66613734e-01 -1.79884449e-01 -7.24111557e-01
-3.95969301e-01 -1.45898044e+00 -7.96759605e-01 -4.57748622e-01
-2.53069490e-01 4.28726196e-01 8.15326333e-01 -2.74303496e-01
3.35997224e-01 8.03915620e-01 -1.50596023e+00 -1.90326050e-01
1.26242387e+00 7.36082554e-01 3.57709408e-01 7.41317630e-01
-2.02947780e-01 3.51724237e-01 1.80262715e-01 -9.45325717e-02
4.26124603e-01 -1.99967548e-02 -9.84957457e-01 8.93696174e-02
8.06679428e-01 -5.05661368e-01 -3.98958951e-01 1.13573885e+00
3.10573816e-01 1.43766388e-01 5.09557605e-01 -1.02200520e+00
-3.15910697e-01 4.46292251e-01 -1.03738379e+00 3.27439189e-01
-2.10098833e-01 1.94110833e-02 1.28241670e+00 -8.47293317e-01
4.89411861e-01 1.24780059e+00 8.75018239e-01 -1.93527892e-01
-1.08757651e+00 -5.37159681e-01 5.50388157e-01 2.00943172e-01
-1.80682218e+00 -1.85869053e-01 7.75775731e-01 -4.17560130e-01
8.63718688e-01 1.80460081e-01 4.93297368e-01 1.41747385e-01
-1.04460597e-01 6.84013486e-01 1.11620152e+00 2.88766548e-02
-3.38596776e-02 -3.81763905e-01 2.28245154e-01 8.61914039e-01
2.28079125e-01 -6.83943257e-02 3.79513390e-02 1.61705524e-01
1.09327066e+00 1.12247944e-01 -1.44266337e-01 -1.97974727e-01
-1.08709776e+00 8.76248300e-01 1.04321361e+00 2.84831226e-02
-7.49413490e-01 2.63433725e-01 2.65503563e-02 -1.53752893e-01
5.72351754e-01 -3.84648368e-02 -2.30909288e-01 5.33810914e-01
-1.38037729e+00 3.75732809e-01 4.81880933e-01 6.80471957e-01
1.34873366e+00 -1.22248717e-01 1.64008856e-01 7.78459132e-01
3.41352969e-01 5.84519804e-01 -3.45741749e-01 -7.85452127e-01
7.13115215e-01 9.11958456e-01 -6.30055293e-02 -1.44202363e+00
-6.84972465e-01 -4.04526263e-01 -7.14722514e-01 2.67978162e-01
3.19132209e-01 -1.35844037e-01 -1.41878796e+00 1.12824750e+00
4.69849259e-01 3.95791888e-01 -8.86552632e-02 9.76838648e-01
9.10298288e-01 6.86448514e-01 2.25063056e-01 4.57472354e-01
1.34149194e+00 -9.94978786e-01 -2.38000676e-01 -6.14517033e-01
4.07126337e-01 -4.03934240e-01 8.81015539e-01 1.36459025e-03
-5.79648852e-01 -5.13903499e-01 -8.19954634e-01 -8.28534886e-02
-6.91815734e-01 3.08081627e-01 7.90367424e-01 3.74905527e-01
-1.27743638e+00 4.43475902e-01 -5.52966058e-01 -4.96934265e-01
7.97000170e-01 4.90532257e-02 -1.81497678e-01 -1.94603667e-01
-1.20485926e+00 8.98686528e-01 3.85696083e-01 5.64830244e-01
-9.75884974e-01 -5.82989335e-01 -9.20595050e-01 -3.76489013e-02
5.63868046e-01 -4.73662078e-01 6.71985745e-01 -5.07809877e-01
-8.53023708e-01 9.32225347e-01 -1.09994650e-01 -5.41463614e-01
5.33459306e-01 -6.30398169e-02 -2.09443957e-01 4.13279772e-01
5.76491833e-01 1.07202172e+00 4.94324267e-01 -1.30905664e+00
-1.21298182e+00 -3.58661532e-01 4.13697511e-02 4.71712887e-01
3.52586746e-01 -2.73514718e-01 -5.63280761e-01 -3.22305620e-01
5.65917373e-01 -8.02116632e-01 -7.62671530e-01 1.00032225e-01
-7.73406148e-01 -2.64990509e-01 9.96787906e-01 -8.04246366e-01
1.26615369e+00 -1.93353403e+00 -3.03438395e-01 5.72493911e-01
4.39072698e-01 2.21172944e-01 -1.74139768e-01 2.08803624e-01
2.54077435e-01 4.17744964e-01 -6.95197940e-01 3.60834152e-01
-2.41850585e-01 2.39720464e-01 -1.42485082e-01 3.49851459e-01
3.49128306e-01 1.11357868e+00 -8.85355771e-01 -6.85268283e-01
4.18181390e-01 3.49032104e-01 -1.47665247e-01 -1.76714420e-01
-1.69251725e-01 2.45935991e-01 -9.18779075e-01 9.42800760e-01
9.96258378e-01 -2.12190270e-01 -1.26990955e-02 -3.61175418e-01
-5.21259248e-01 1.95662290e-01 -1.22375941e+00 1.35484326e+00
-2.16758832e-01 6.29232526e-01 1.69449061e-01 -1.20072103e+00
1.08630145e+00 -2.65874892e-01 3.09565157e-01 -7.81670332e-01
-3.35927606e-01 1.57067403e-01 -2.28072852e-01 -4.22180891e-01
8.12480927e-01 2.75085270e-01 1.32145897e-01 8.33892971e-02
-5.04008174e-01 -3.43941480e-01 1.05215870e-01 2.48503417e-01
8.60813558e-01 4.36211377e-01 9.13508907e-02 -2.48523965e-01
3.11289698e-01 6.27729475e-01 3.67029756e-01 7.51638830e-01
-1.56322584e-01 6.44856691e-01 3.50024134e-01 -5.81103027e-01
-9.70611453e-01 -9.52110648e-01 -1.07896566e-01 9.14015770e-01
4.14367735e-01 3.41260992e-02 -6.62943006e-01 -7.13085830e-01
2.64842659e-01 3.96151036e-01 -4.35281277e-01 6.51060820e-01
-6.60278141e-01 -9.75713313e-01 7.44661272e-01 4.93607789e-01
1.00169516e+00 -8.14847231e-01 -5.35565376e-01 4.63909000e-01
-3.21476102e-01 -1.34009635e+00 4.80981693e-02 -6.37860000e-02
-9.07027543e-01 -1.20146334e+00 -6.38280153e-01 -6.65507674e-01
7.32197821e-01 9.63244438e-01 9.39644694e-01 1.28073409e-01
-2.90734023e-01 -1.18553832e-01 -1.26323223e-01 -1.76425353e-02
1.60928965e-01 3.13624144e-01 -8.98819327e-01 -9.94020328e-02
1.21819809e-01 -5.15661776e-01 -8.77831221e-01 5.47699273e-01
-6.29105210e-01 1.32729545e-01 9.35374260e-01 3.28478992e-01
9.73627508e-01 3.71002764e-01 3.92092228e-01 -8.71022522e-01
1.65958390e-01 -5.94879270e-01 -7.67623425e-01 2.62435555e-01
-5.38013935e-01 -2.33670652e-01 3.98175120e-02 2.69218951e-01
-9.38581288e-01 4.73871380e-01 -1.06622063e-01 -7.63305724e-02
-3.03834438e-01 7.09007204e-01 -1.05814107e-01 -3.79000932e-01
5.40962458e-01 2.94431388e-01 -1.10045463e-01 -2.92465627e-01
7.65142262e-01 6.35438263e-01 6.05053127e-01 -1.97848991e-01
9.42280531e-01 8.74812245e-01 5.31422906e-02 -1.05158293e+00
-8.10785353e-01 -7.35280812e-01 -7.91922033e-01 -4.74627495e-01
9.55438435e-01 -1.25476658e+00 -3.35327804e-01 6.16406262e-01
-8.51142585e-01 -4.78503317e-01 2.60852754e-01 1.76297769e-01
-2.69942820e-01 3.53592247e-01 -1.78896621e-01 -7.39596784e-01
-4.97316957e-01 -1.02408504e+00 1.34117913e+00 3.94555151e-01
3.35866362e-01 -8.85186732e-01 -1.70717672e-01 4.78512853e-01
3.94639343e-01 5.29203296e-01 7.04515398e-01 -3.38184506e-01
-1.13665080e+00 2.91086943e-03 -1.00258374e+00 8.04412179e-03
-2.88967863e-02 1.62748858e-01 -8.10738325e-01 2.63616323e-01
-9.29317594e-01 1.38302833e-01 1.37244153e+00 7.66631126e-01
9.89543319e-01 -8.52790102e-02 -7.32577324e-01 6.59397244e-01
1.74655974e+00 -1.06658556e-01 9.50197756e-01 5.52115619e-01
1.22865069e+00 9.01141167e-01 7.26703584e-01 1.53203560e-02
1.09609199e+00 3.03370118e-01 8.87738526e-01 -7.09594846e-01
-1.60889447e-01 -3.68978262e-01 -1.50308579e-01 -4.49911430e-02
-2.40824640e-01 -1.82633907e-01 -1.33303821e+00 1.10686803e+00
-1.94849646e+00 -9.59252119e-01 -8.90088618e-01 1.81546295e+00
3.26138198e-01 1.57653987e-01 1.79434046e-01 -8.04041326e-02
1.05512416e+00 3.72676969e-01 -5.80969751e-01 9.92638543e-02
-2.85677969e-01 1.21337175e-01 1.39902461e+00 6.08207524e-01
-1.54935253e+00 1.67698824e+00 5.89941502e+00 7.92170048e-01
-9.81713414e-01 -2.40041129e-02 6.37671471e-01 5.28304636e-01
-3.36824089e-01 2.77819157e-01 -9.56261039e-01 -4.47051376e-02
4.63398069e-01 4.32882279e-01 1.69592291e-01 7.72018611e-01
4.64413047e-01 -6.68379128e-01 -2.10707888e-01 5.73151767e-01
-3.19587469e-01 -1.42562735e+00 2.13074580e-01 1.73546746e-01
7.17057109e-01 7.01530516e-01 -3.06150377e-01 -8.67736116e-02
5.59903204e-01 -9.74068105e-01 5.85016251e-01 4.84454334e-01
6.10598505e-01 -6.51639223e-01 5.08911848e-01 1.98645458e-01
-1.80443776e+00 1.31226843e-02 -3.00028294e-01 2.04635244e-02
3.29335600e-01 8.04751098e-01 -8.50602865e-01 8.81857693e-01
8.47456515e-01 1.01208448e+00 -6.53151751e-01 1.22651815e+00
-5.57108223e-01 6.75189137e-01 -4.72746760e-01 2.75021315e-01
8.50577950e-01 -5.75891554e-01 3.85674864e-01 1.31061172e+00
1.83235973e-01 1.54508963e-01 5.30594409e-01 9.12884712e-01
1.59568295e-01 -5.33560887e-02 -7.92239606e-01 1.78433567e-01
5.68684042e-01 1.62123382e+00 -1.22361326e+00 -3.40810984e-01
-3.95159990e-01 5.87603092e-01 1.23373993e-01 4.87742215e-01
-7.45762765e-01 -6.41717494e-01 6.13240361e-01 3.36574733e-01
5.07685304e-01 -5.85892677e-01 -5.37339628e-01 -5.90997756e-01
-8.23808536e-02 -3.86646777e-01 2.81279117e-01 -8.01444888e-01
-7.95989037e-01 3.37603271e-01 -5.53185157e-02 -1.01864922e+00
3.05898309e-01 -2.70432204e-01 -8.13899755e-01 9.65297043e-01
-2.47818446e+00 -1.62240934e+00 -7.43917167e-01 4.87147719e-01
5.41357577e-01 4.51145947e-01 2.57998794e-01 2.08791316e-01
-4.62792993e-01 -1.75429538e-01 -7.67764971e-02 4.34372008e-01
9.92042273e-02 -1.11767530e+00 1.05139565e+00 1.22489679e+00
-9.91603509e-02 7.30932504e-02 1.59647986e-01 -9.26951826e-01
-9.96458471e-01 -1.52238429e+00 8.34359705e-01 1.64855609e-03
8.35132599e-01 1.74755510e-02 -1.08064437e+00 4.45734501e-01
-2.59900659e-01 2.40131430e-02 -1.05713559e-02 -1.02111205e-01
-9.12939757e-02 -1.74499705e-01 -9.54207063e-01 5.16052723e-01
1.39406395e+00 -5.24536252e-01 -4.55111444e-01 3.22836071e-01
3.87065738e-01 -1.75708279e-01 -5.29911399e-01 5.01009643e-01
4.30065066e-01 -8.82034719e-01 1.17062545e+00 7.23319054e-02
3.15118730e-01 -7.62673914e-01 6.53357506e-02 -1.15568376e+00
-2.78363764e-01 -2.48581499e-01 6.65748775e-01 9.68990207e-01
6.37169003e-01 -7.42687285e-01 7.32780337e-01 1.78436842e-02
-3.38108033e-01 -4.21143144e-01 -7.22496152e-01 -5.97815156e-01
-1.39373004e-01 -3.56685370e-01 5.52301168e-01 8.29615116e-01
-7.88053036e-01 2.94502586e-01 -4.63119671e-02 7.08240449e-01
8.23942482e-01 3.51535141e-01 8.82280350e-01 -1.50735724e+00
5.89167595e-01 -5.64285934e-01 -3.94172162e-01 -1.20491433e+00
1.32137865e-01 -8.47507834e-01 2.09097996e-01 -2.24317670e+00
-1.06087774e-01 -8.26303959e-01 4.63827401e-02 8.25300992e-01
-3.67994517e-01 3.39195907e-01 -1.87747940e-01 2.99268216e-01
-5.15146911e-01 2.54846156e-01 1.23655057e+00 -2.64658451e-01
-3.74357551e-01 -4.90515530e-02 -5.21242559e-01 6.76161647e-01
8.67094278e-01 -3.54460984e-01 -4.46626216e-01 -5.66718936e-01
3.45321059e-01 -1.99992001e-01 8.63268971e-01 -9.17138755e-01
3.45866591e-01 -1.95790738e-01 1.10966172e-02 -1.21305180e+00
-7.91557804e-02 -5.76091349e-01 1.09275103e-01 3.00354391e-01
9.07841474e-02 -1.51194856e-01 1.46721244e-01 7.63338208e-01
-1.70884371e-01 1.45622700e-01 6.44136786e-01 -2.99242228e-01
-1.45555389e+00 5.61771154e-01 -3.27343643e-01 -2.67743114e-02
1.01281822e+00 -6.27511382e-01 -4.07783926e-01 -4.06524949e-02
-5.92418551e-01 7.42870271e-01 1.81118786e-01 1.61143154e-01
7.49657154e-01 -6.48899317e-01 -1.00824225e+00 1.95310041e-02
1.46274060e-01 3.79103780e-01 2.73744076e-01 9.28784847e-01
-8.83681238e-01 3.03358108e-01 -1.32053569e-01 -8.05305004e-01
-1.08529854e+00 -6.36304077e-03 4.44758892e-01 -1.22399665e-01
-9.06780481e-01 6.91477478e-01 2.27932483e-01 -5.46296477e-01
-3.48123729e-01 -3.63499433e-01 -4.52636600e-01 3.89302492e-01
5.37646860e-02 4.66233611e-01 8.56556185e-03 -8.43469203e-01
-5.30012429e-01 1.12798083e+00 1.38810933e-01 -2.89598275e-02
1.40521288e+00 -3.66751611e-01 -4.53940518e-02 -1.32951856e-01
6.84195399e-01 -4.14992929e-01 -1.45215690e+00 -3.63400400e-01
2.24069193e-01 -4.48191792e-01 5.84766746e-01 -7.43021846e-01
-1.28250694e+00 9.09287632e-01 4.94421095e-01 2.07909122e-01
1.00813544e+00 9.88074988e-02 8.14593554e-01 3.06350708e-01
3.08983862e-01 -1.18767452e+00 -5.47042608e-01 4.62366402e-01
4.98681277e-01 -1.46079981e+00 1.64641425e-01 -9.12035525e-01
-6.26188099e-01 1.03179681e+00 4.12007987e-01 -1.96067989e-01
7.30860710e-01 -3.52450460e-02 -1.36113842e-03 -6.45242691e-01
1.36748217e-02 -1.17317212e+00 2.19776303e-01 6.91631198e-01
-8.55610818e-02 4.52377051e-01 -1.13836393e-01 -1.33565366e-01
1.09351717e-01 -1.21710278e-01 2.16080770e-01 7.49641597e-01
-1.00730503e+00 -6.11262918e-01 -3.44817787e-01 5.78620136e-01
-8.57546702e-02 -4.51572835e-01 -3.13183665e-01 9.57757175e-01
-6.68082461e-02 9.13595319e-01 1.95810392e-01 -2.86377072e-01
4.26132858e-01 -3.37458491e-01 -5.54633401e-02 -4.53798771e-01
-3.09592158e-01 7.69250169e-02 3.51672232e-01 -5.80995739e-01
-6.88822269e-01 -5.36080778e-01 -1.54438019e+00 -2.37265021e-01
-3.74590456e-01 -6.92275465e-02 8.64485681e-01 1.03404582e+00
5.61226070e-01 2.87085831e-01 7.32630730e-01 -9.61782217e-01
9.43451375e-02 -6.03768468e-01 -5.57086706e-01 -1.12300724e-01
2.63772577e-01 -5.54737449e-01 -1.80749625e-01 -7.30708390e-02] | [8.980910301208496, -1.4865748882293701] |
73e820bf-44ec-4ec9-82e8-53a6fde6bb21 | efficient-regional-memory-network-for-video | 2103.12934 | null | https://arxiv.org/abs/2103.12934v2 | https://arxiv.org/pdf/2103.12934v2.pdf | Efficient Regional Memory Network for Video Object Segmentation | Recently, several Space-Time Memory based networks have shown that the object cues (e.g. video frames as well as the segmented object masks) from the past frames are useful for segmenting objects in the current frame. However, these methods exploit the information from the memory by global-to-global matching between the current and past frames, which lead to mismatching to similar objects and high computational complexity. To address these problems, we propose a novel local-to-local matching solution for semi-supervised VOS, namely Regional Memory Network (RMNet). In RMNet, the precise regional memory is constructed by memorizing local regions where the target objects appear in the past frames. For the current query frame, the query regions are tracked and predicted based on the optical flow estimated from the previous frame. The proposed local-to-local matching effectively alleviates the ambiguity of similar objects in both memory and query frames, which allows the information to be passed from the regional memory to the query region efficiently and effectively. Experimental results indicate that the proposed RMNet performs favorably against state-of-the-art methods on the DAVIS and YouTube-VOS datasets. | ['Wenxiu Sun', 'Shengping Zhang', 'Shangchen Zhou', 'Hongxun Yao', 'Haozhe Xie'] | 2021-03-24 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Xie_Efficient_Regional_Memory_Network_for_Video_Object_Segmentation_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Xie_Efficient_Regional_Memory_Network_for_Video_Object_Segmentation_CVPR_2021_paper.pdf | cvpr-2021-1 | ['one-shot-visual-object-segmentation'] | ['computer-vision'] | [-1.26718611e-01 -4.81045008e-01 -5.70664227e-01 -2.49864861e-01
-3.26442838e-01 -1.66467890e-01 2.02147007e-01 -2.83814445e-02
-6.12828195e-01 6.46788359e-01 3.15582976e-02 4.68871623e-01
5.96140735e-02 -7.64926970e-01 -6.03745580e-01 -6.23603642e-01
-4.91666794e-02 9.32104141e-03 1.07073236e+00 1.93769440e-01
4.70746934e-01 5.48565149e-01 -1.69893897e+00 4.24962670e-01
5.18384337e-01 1.33535993e+00 8.67928267e-01 4.71798241e-01
-3.96570355e-01 9.69581902e-01 -5.86728275e-01 2.87367970e-01
3.82952720e-01 -4.33708102e-01 -8.11452806e-01 2.03073621e-01
8.51720095e-01 -6.26416028e-01 -8.56792152e-01 1.03256559e+00
3.05332333e-01 6.37570143e-01 2.29468998e-02 -1.22312117e+00
-3.38370055e-01 1.07709214e-01 -3.79254580e-01 7.77494013e-01
2.76164919e-01 2.18705520e-01 6.77656531e-01 -1.13163698e+00
1.13207233e+00 1.35683823e+00 1.28345087e-01 3.79282027e-01
-6.98328853e-01 -8.85437191e-01 5.48744142e-01 6.35692060e-01
-1.60758913e+00 -6.55574918e-01 8.69395137e-01 -4.13649559e-01
7.65701652e-01 -5.40403649e-03 8.44218910e-01 2.77348697e-01
1.86130375e-01 1.08443379e+00 5.17016649e-01 2.13820245e-02
2.41054278e-02 -7.99165964e-02 -8.14141482e-02 9.30420160e-01
-1.45652786e-01 2.58828849e-01 -9.55886006e-01 2.52004853e-03
1.23042023e+00 3.23749572e-01 -5.58949709e-01 -5.29403925e-01
-1.48415017e+00 5.12840271e-01 6.27342224e-01 3.91061991e-01
-4.99425560e-01 8.35660994e-02 3.73342365e-01 1.30224258e-01
3.89621466e-01 -2.48621069e-02 -2.95781344e-01 1.27605870e-01
-1.30927145e+00 7.39388540e-02 4.83328432e-01 9.49567378e-01
1.01941884e+00 1.99447125e-01 -2.50322908e-01 5.04904628e-01
5.02557039e-01 3.70702028e-01 4.70219523e-01 -1.20698535e+00
7.55988777e-01 5.18593013e-01 3.13406795e-01 -1.52438378e+00
-1.18594207e-01 -3.64869446e-01 -6.67394578e-01 -8.38881284e-02
4.45225120e-01 2.14686766e-01 -9.87885535e-01 1.56910062e+00
4.74955380e-01 9.09252703e-01 1.19885504e-01 1.32434714e+00
1.15969610e+00 9.12082732e-01 1.88555762e-01 -5.25300741e-01
8.56150150e-01 -1.29060650e+00 -8.07485998e-01 -4.25943553e-01
3.97916257e-01 -8.46114993e-01 4.19566423e-01 -1.74000427e-01
-1.20499861e+00 -1.14213300e+00 -9.27258015e-01 1.27648547e-01
-1.38961285e-01 6.20012134e-02 1.88230470e-01 -1.73124541e-02
-9.54964757e-01 7.22353160e-01 -8.86729896e-01 -2.12961406e-01
4.71796066e-01 2.29219854e-01 -2.61801720e-01 -2.75422186e-01
-1.17247415e+00 5.47624111e-01 7.92282641e-01 3.92359406e-01
-1.11234319e+00 -4.77528542e-01 -1.03529370e+00 4.99325171e-02
5.30346096e-01 -5.66224694e-01 8.04900348e-01 -1.12834299e+00
-1.04846561e+00 5.51456034e-01 -6.81426466e-01 -4.17089820e-01
4.46847111e-01 -5.21114506e-02 -3.16788375e-01 6.51562095e-01
2.60761619e-01 1.12213945e+00 9.79429185e-01 -9.88228679e-01
-1.02800894e+00 -1.07764557e-01 -1.04880199e-01 3.59083354e-01
-3.08274273e-02 -7.79396743e-02 -9.04664457e-01 -7.81607866e-01
5.85252523e-01 -7.02661991e-01 -2.20440421e-02 4.18945789e-01
-6.15804829e-02 -3.05100322e-01 1.33288431e+00 -6.95283055e-01
1.31215537e+00 -2.22080517e+00 1.34450287e-01 1.70973297e-02
1.42344490e-01 5.92988670e-01 -2.59698391e-01 -9.28615108e-02
6.76231533e-02 -2.34998643e-01 9.74433795e-02 -2.86116511e-01
-5.83014727e-01 4.05608177e-01 -2.75380522e-01 6.01058066e-01
-5.86108752e-02 9.63642180e-01 -1.13638020e+00 -1.08647323e+00
5.37594378e-01 3.94400299e-01 -2.05084518e-01 5.35620153e-01
-8.55698958e-02 7.08387792e-01 -4.49568331e-01 7.24506080e-01
5.70395470e-01 -2.22720236e-01 -2.53461540e-01 -3.57235789e-01
-1.52068451e-01 2.07165867e-01 -1.45689499e+00 1.86836874e+00
5.51129654e-02 8.20759892e-01 -1.01880019e-03 -8.37213874e-01
8.93482983e-01 2.84703493e-01 6.37789965e-01 -8.75296652e-01
-3.16946656e-02 2.53798217e-01 -2.86472499e-01 -6.64217651e-01
5.39652050e-01 2.76465863e-01 4.43197817e-01 2.14639992e-01
9.97510254e-02 3.61446112e-01 5.19050658e-01 1.91704035e-01
3.95038038e-01 2.60268509e-01 1.60000116e-01 2.43557002e-02
8.42953444e-01 -2.24023014e-02 1.21976209e+00 7.51206160e-01
-7.33226001e-01 6.84748948e-01 -8.90569463e-02 -7.52888799e-01
-6.76558435e-01 -8.73132169e-01 1.89876333e-01 8.91100347e-01
9.44552600e-01 -3.47136438e-01 -4.20447290e-01 -5.83732307e-01
-1.26880646e-01 1.00894764e-01 -2.57411182e-01 -1.25187531e-01
-1.01046252e+00 1.86773874e-02 3.29004824e-02 7.00684488e-01
1.00986862e+00 -1.34751654e+00 -8.71540189e-01 5.02323329e-01
-5.55257022e-01 -1.28329229e+00 -1.00126791e+00 -4.14570689e-01
-1.27794719e+00 -1.03937185e+00 -9.97851431e-01 -1.03748524e+00
5.92598259e-01 7.74497807e-01 8.99524987e-01 4.72675294e-01
-9.25701410e-02 2.58452028e-01 -7.81258792e-02 2.10166082e-01
-1.86816335e-01 -3.10172856e-01 -1.28533378e-01 3.27365458e-01
7.18631819e-02 -2.51118690e-01 -9.52729702e-01 7.86444247e-01
-8.94319177e-01 2.29538366e-01 4.24795747e-01 6.28306031e-01
9.42396402e-01 -9.86992344e-02 5.16590357e-01 -3.40809405e-01
-2.18476638e-01 -2.86266923e-01 -5.35897791e-01 3.41243207e-01
-3.06191504e-01 -1.63858280e-01 1.49256408e-01 -6.10466182e-01
-1.02783561e+00 2.77096093e-01 3.22509855e-01 -9.33950961e-01
-1.42757550e-01 3.30076516e-01 -1.42539963e-01 -2.39703268e-01
8.83896053e-02 5.73269486e-01 -8.86225402e-02 -3.75536621e-01
1.20855212e-01 3.64699543e-01 7.59064317e-01 -3.16334009e-01
4.34533447e-01 7.12499797e-01 -1.14494585e-01 -6.34739876e-01
-7.43977547e-01 -9.12125647e-01 -9.10121500e-01 -5.13293505e-01
9.50912356e-01 -1.09348738e+00 -4.10118282e-01 6.19773924e-01
-1.41907883e+00 -4.07730639e-02 -2.51860350e-01 5.17836511e-01
-4.45415318e-01 6.10154092e-01 -7.46106505e-01 -5.80677330e-01
-1.62161022e-01 -1.17397785e+00 8.14806759e-01 6.79020226e-01
-1.66508500e-02 -1.00237560e+00 -2.88814247e-01 1.47864699e-01
3.36280078e-01 2.82782763e-02 4.08243448e-01 -3.22987080e-01
-1.34957862e+00 1.93734244e-01 -4.53648925e-01 1.89561263e-01
1.60402209e-01 -2.02089086e-01 -6.76649809e-01 -4.58351672e-01
8.48156735e-02 2.25020811e-01 1.05505443e+00 4.75576490e-01
7.18217373e-01 -1.62381083e-01 -5.41971028e-01 6.16915584e-01
1.24706459e+00 5.16393840e-01 5.18793941e-01 3.25305521e-01
1.01562560e+00 5.08439541e-01 1.13202584e+00 2.18335941e-01
3.67709190e-01 7.07540393e-01 3.96260023e-01 -1.86812095e-02
-4.50653195e-01 -2.73149818e-01 3.10531884e-01 7.26361096e-01
1.87660053e-01 -7.62897655e-02 -5.79639316e-01 9.79510248e-01
-2.22957015e+00 -1.07748008e+00 1.17510326e-01 2.24474573e+00
5.30213118e-01 9.86121669e-02 -8.06554556e-02 -2.53081858e-01
1.11198223e+00 6.89964890e-01 -7.26466060e-01 2.57861286e-01
-2.84254581e-01 -3.37196022e-01 2.81787515e-01 4.18759376e-01
-1.15467274e+00 1.09036303e+00 5.78343201e+00 8.26069117e-01
-1.28280985e+00 2.33159199e-01 5.97921789e-01 -1.37139827e-01
2.62654871e-01 1.51324674e-01 -9.25382912e-01 4.63768870e-01
4.69110638e-01 -5.95212728e-02 2.09169447e-01 7.17772663e-01
3.39669317e-01 -5.68003178e-01 -1.11572683e+00 1.29024291e+00
2.86034077e-01 -1.58601427e+00 2.04429001e-01 -2.14123338e-01
9.09782112e-01 -3.67680229e-02 -7.01329261e-02 -2.71743238e-02
-5.67405999e-01 -5.82000196e-01 9.28304851e-01 7.10226834e-01
4.28392529e-01 -6.15584731e-01 6.35609567e-01 3.28905910e-01
-1.85265648e+00 -3.79318520e-02 -5.94485164e-01 4.73315679e-02
3.58456194e-01 2.90783793e-01 -5.54434538e-01 3.91656786e-01
8.10083687e-01 1.15220094e+00 -5.00609457e-01 1.31203055e+00
1.08026654e-01 2.30565608e-01 -2.25589022e-01 3.88513803e-01
3.29544246e-01 -4.09341790e-02 8.32990646e-01 1.00402188e+00
2.39097357e-01 1.11617468e-01 6.94987178e-01 9.57100630e-01
1.35325611e-01 7.27576613e-02 -2.57554710e-01 1.14637703e-01
4.89112973e-01 9.06296670e-01 -9.35911298e-01 -8.19958210e-01
-4.30562317e-01 9.42098796e-01 1.70887470e-01 6.56646609e-01
-5.57701290e-01 -3.54049385e-01 5.75354397e-01 7.11937696e-02
5.86647630e-01 -4.44751471e-01 1.91651821e-01 -1.21586204e+00
1.22801997e-01 -3.11814398e-01 6.48339272e-01 -9.21698332e-01
-8.46369386e-01 4.54745203e-01 -1.08712234e-01 -1.47878408e+00
-2.27660105e-01 -1.25744920e-02 -5.68494260e-01 8.69838655e-01
-1.85483563e+00 -8.07936847e-01 -5.64131916e-01 8.50322425e-01
7.21565187e-01 -7.39840642e-02 1.42688408e-01 4.97420162e-01
-3.19363832e-01 1.62580922e-01 -2.62096941e-01 6.30820990e-01
7.93673813e-01 -6.42408371e-01 1.43952042e-01 1.10038209e+00
3.74939233e-01 4.84749258e-01 3.48986745e-01 -9.79115009e-01
-1.00429225e+00 -1.15932345e+00 9.69104826e-01 -3.73657458e-02
2.40670756e-01 2.26636231e-01 -1.25666797e+00 5.26481807e-01
-5.14762290e-02 5.74922442e-01 1.73341166e-02 -8.20754230e-01
-5.80361113e-02 -2.17567354e-01 -9.05469358e-01 2.91897178e-01
1.07479441e+00 -7.58121133e-01 -5.41567385e-01 8.09271932e-02
7.39769757e-01 -7.89685249e-01 -6.45831168e-01 3.48545879e-01
4.42405462e-01 -1.04276586e+00 1.06507111e+00 -3.91765088e-01
1.56866327e-01 -7.81622469e-01 -1.66859701e-01 -8.16889524e-01
-1.04052395e-01 -4.59235728e-01 -3.68735939e-01 1.14193630e+00
-2.02383831e-01 -3.47826153e-01 8.68867457e-01 6.73818111e-01
-5.27674183e-02 -6.41163826e-01 -1.17509162e+00 -7.06141233e-01
-5.52932382e-01 -2.03396752e-01 4.27166522e-01 7.89596021e-01
-4.48314935e-01 -1.96298003e-01 -4.84831154e-01 2.74239570e-01
5.43381512e-01 6.58202529e-01 6.50179207e-01 -1.09281611e+00
-8.00394937e-02 -1.90438882e-01 -8.14471424e-01 -1.76119924e+00
3.01466674e-01 -6.64484560e-01 1.36213347e-01 -1.41671073e+00
2.05879882e-01 -3.78659487e-01 -5.13316214e-01 2.36832067e-01
-3.75442386e-01 3.16772580e-01 6.21082067e-01 6.97361469e-01
-1.05162954e+00 4.31211650e-01 1.62651002e+00 -2.27030739e-01
-5.86318493e-01 -1.04913197e-01 6.23924993e-02 7.12054431e-01
4.65684593e-01 -5.71554363e-01 -1.67997211e-01 -3.88700396e-01
-3.41980815e-01 5.22994816e-01 5.71728349e-01 -1.20283127e+00
8.51657271e-01 -2.70704567e-01 5.32276392e-01 -9.87481296e-01
5.25970757e-01 -7.53944099e-01 1.33438379e-01 5.10235965e-01
-2.13794619e-01 1.58277765e-01 2.67716572e-02 8.75913203e-01
-7.01057494e-01 -1.66511595e-01 7.39908993e-01 -3.28011394e-01
-1.50703263e+00 7.04295456e-01 -2.76744273e-02 7.69528896e-02
1.02620494e+00 -5.38004518e-01 -1.48940518e-01 -3.90688092e-01
-8.81131172e-01 3.84709537e-01 2.83398658e-01 6.76387906e-01
1.05229676e+00 -1.30229223e+00 -3.44428867e-01 3.79367977e-01
-1.92784145e-02 2.42824137e-01 5.38783848e-01 9.10999060e-01
-3.37103605e-01 5.15455544e-01 -4.03788596e-01 -1.03795874e+00
-1.44084775e+00 7.28189170e-01 5.10033906e-01 1.70934752e-01
-6.42832696e-01 7.27884352e-01 4.51556474e-01 1.97009757e-01
4.25909191e-01 -3.13920289e-01 -2.76088834e-01 1.09095164e-01
6.49397194e-01 4.70200747e-01 -3.54793280e-01 -1.21537769e+00
-3.11517447e-01 9.25450265e-01 -1.79701485e-02 7.18540400e-02
9.42151368e-01 -6.83713973e-01 -1.56462952e-01 5.18104613e-01
1.28293920e+00 -3.91081661e-01 -1.62231505e+00 -9.07819450e-01
-6.92265108e-02 -1.13184071e+00 7.47245625e-02 -1.62781537e-01
-1.71420360e+00 8.54248941e-01 7.77775884e-01 -3.59058142e-01
1.05569601e+00 -1.26806468e-01 1.26452231e+00 1.49916351e-01
4.84381378e-01 -1.06236422e+00 2.39661410e-01 3.69156331e-01
4.69216406e-01 -1.21095741e+00 4.93649468e-02 -4.89405751e-01
-4.35950726e-01 1.19650888e+00 9.11137819e-01 -7.58501738e-02
6.22807503e-01 -2.58421659e-01 1.65257692e-01 1.79429371e-02
-4.75563616e-01 -1.72371879e-01 4.85871077e-01 4.33542073e-01
-2.38913093e-02 -4.04311776e-01 5.88838607e-02 2.04827376e-02
5.16545057e-01 2.90475916e-02 2.62453824e-01 1.08576620e+00
-6.51885509e-01 -7.94998705e-01 -5.05136967e-01 1.90117493e-01
-2.50191450e-01 1.04675584e-01 -1.93577248e-03 6.40440583e-01
1.80555344e-01 9.55341578e-01 3.82646739e-01 -1.55965328e-01
7.69020766e-02 -1.80336051e-02 2.25122720e-01 -3.94325733e-01
-3.63489747e-01 3.73477668e-01 -3.25034767e-01 -9.89452243e-01
-9.56947446e-01 -5.01879513e-01 -1.53077531e+00 -1.84912220e-01
-3.47588867e-01 1.98216382e-02 1.45181254e-01 1.10890245e+00
6.30888462e-01 3.47702295e-01 3.35161448e-01 -1.24362409e+00
1.30780235e-01 -5.85393071e-01 -4.37854767e-01 5.30099750e-01
6.04601562e-01 -7.73767233e-01 -1.73818290e-01 1.69803396e-01] | [9.28734016418457, -0.2322259396314621] |
dbf3e1ab-2e11-4b0b-a306-55275bec8d89 | image-differential-invariants | 1911.05327 | null | https://arxiv.org/abs/1911.05327v2 | https://arxiv.org/pdf/1911.05327v2.pdf | Rotation Differential Invariants of Images Generated by Two Fundamental Differential Operators | In this paper, we design two fundamental differential operators for the derivation of rotation differential invariants of images. Each differential invariant obtained by using the new method can be expressed as a homogeneous polynomial of image partial derivatives, which preserve their values when the image is rotated by arbitrary angles. We produce all possible instances of homogeneous invariants up to the given order and degree, and discuss the independence of them in detail. As far as we know, no previous papers have published so many explicit forms of high-order rotation differential invariants of images. In the experimental part, texture classification and image patch verification are carried out on popular real databases. These rotation differential invariants are used as image feature vector. We mainly evaluate the effects of various factors on the performance of them. The experimental results also validate that they have better performance than some commonly used image features in some cases. | ['Hanlin Mo', 'Hua Li'] | 2019-11-13 | null | null | null | null | ['texture-classification'] | ['computer-vision'] | [-9.06277969e-02 -3.65529180e-01 -3.68191093e-01 -2.47612983e-01
2.69802734e-02 -4.14599776e-01 4.70793009e-01 -3.85235727e-01
-3.44881833e-01 3.59180897e-01 -3.62040013e-01 -1.22969776e-01
-3.61852229e-01 -4.63962615e-01 -1.89373538e-01 -9.14778233e-01
-4.02544349e-01 1.31394908e-01 4.62624401e-01 -3.76912832e-01
4.74161565e-01 1.13856816e+00 -1.51759648e+00 -2.89074183e-01
6.06544614e-01 7.26027966e-01 -3.22633237e-01 7.42681324e-01
3.84415835e-01 5.18903136e-01 -6.58397317e-01 -2.05995917e-01
5.01897991e-01 -3.59360725e-01 -1.10250521e+00 5.67313671e-01
1.14308335e-01 -1.84252620e-01 -4.76068199e-01 1.13106120e+00
3.88674200e-01 -1.05220377e-01 9.52431738e-01 -1.09755445e+00
-9.46093559e-01 2.40626037e-01 -6.25955582e-01 4.24023002e-01
5.69787562e-01 -3.81646901e-01 6.24989748e-01 -7.42290199e-01
8.28287542e-01 1.12731290e+00 6.50935531e-01 1.57680675e-01
-1.22157180e+00 -9.69728380e-02 -4.09388900e-01 4.75483984e-01
-1.64826155e+00 -4.32634279e-02 1.04506886e+00 -3.00204545e-01
4.63105738e-01 5.94121575e-01 4.70187634e-01 3.30273300e-01
7.64217079e-01 6.12057924e-01 1.23167181e+00 -6.62720740e-01
-3.30516815e-01 2.78614879e-01 4.47693735e-01 9.11793709e-01
3.95085514e-01 -2.41456181e-01 1.97891951e-01 -4.52205874e-02
1.22826648e+00 -8.04442540e-02 -3.45536321e-01 -5.60508609e-01
-1.17944336e+00 5.10897040e-01 3.06050926e-01 8.61946821e-01
-9.74834338e-02 -4.35152292e-01 1.78539708e-01 5.69481194e-01
2.36700088e-01 3.68031234e-01 -2.46489853e-01 3.11545402e-01
-3.65200102e-01 -4.42920811e-02 7.79168725e-01 8.50737333e-01
9.11821604e-01 -7.82192200e-02 1.48890493e-02 6.65861070e-01
-1.86595209e-02 6.88136578e-01 6.15582526e-01 -5.73293269e-01
-5.06608605e-01 4.28576529e-01 -8.75708759e-02 -1.47283316e+00
-4.43217427e-01 -9.46247298e-03 -7.89876044e-01 2.58689731e-01
1.70357093e-01 1.84292465e-01 -6.87058151e-01 1.02074313e+00
3.25805932e-01 -1.45181969e-01 1.38972953e-01 8.80748808e-01
7.21374214e-01 4.80229855e-01 -3.58570278e-01 -3.31069708e-01
1.36667562e+00 -6.10153854e-01 -8.90872121e-01 7.77495861e-01
1.85944036e-01 -1.35568058e+00 4.12128627e-01 3.64967853e-01
-6.40743017e-01 -7.67486215e-01 -1.24012208e+00 -1.00533850e-02
-3.03662628e-01 4.30220127e-01 8.12647939e-01 6.01473451e-01
-1.08250034e+00 6.79491699e-01 -6.86006188e-01 -4.02603567e-01
-3.51625770e-01 4.56094712e-01 -7.00947464e-01 4.74990070e-01
-1.03457260e+00 1.05747664e+00 -2.12316643e-02 2.60887057e-01
-2.19448715e-01 -2.85654962e-01 -7.39230335e-01 -4.02424902e-01
-3.69610190e-01 -2.22202569e-01 1.09689426e+00 -1.17601073e+00
-1.50388634e+00 1.14241540e+00 -1.67490527e-01 3.88068496e-03
4.22669530e-01 2.74365097e-01 -8.44111562e-01 4.09450203e-01
-9.73036289e-02 1.25347212e-01 9.16211665e-01 -1.15496504e+00
-2.42594197e-01 -3.28030080e-01 1.94208488e-01 -4.84577194e-02
7.34206736e-02 1.11402467e-01 -3.98399979e-01 -4.30104792e-01
5.81522167e-01 -1.20171762e+00 -1.77606985e-01 -1.29116371e-01
-2.68930852e-01 -2.26813912e-01 1.15667999e+00 -6.88981354e-01
1.03412855e+00 -2.27660418e+00 2.08184943e-01 4.92578477e-01
-2.14029744e-01 4.22603428e-01 5.65573871e-02 2.50539541e-01
-4.72581744e-01 -2.23676991e-02 2.09598586e-01 3.12164932e-01
-2.63841569e-01 4.06835586e-01 -9.63148475e-02 9.76251960e-01
3.25044453e-01 3.81999642e-01 -5.19539535e-01 -7.36760378e-01
2.94536680e-01 9.27815676e-01 -2.03827932e-01 -8.27941075e-02
5.67479372e-01 4.42993701e-01 -8.29816401e-01 6.57087147e-01
1.14415574e+00 3.19531381e-01 3.84056494e-02 -6.83251143e-01
-3.57969910e-01 -5.20619512e-01 -1.42763937e+00 7.47938514e-01
-6.40681088e-02 8.47566485e-01 -3.76888514e-01 -1.12109661e+00
1.15447199e+00 2.88065016e-01 8.62498283e-01 -3.98894399e-01
2.18270689e-01 1.95723012e-01 9.77928191e-02 -8.10971975e-01
5.53501248e-01 7.35018253e-02 2.11809471e-01 9.97303799e-02
1.69552416e-01 -1.34548798e-01 4.35746282e-01 -1.62658229e-01
5.63137293e-01 -2.42731459e-02 3.89298111e-01 -6.54665172e-01
1.22736573e+00 -4.55201194e-02 3.44555050e-01 2.54475683e-01
-2.78434604e-01 5.26711047e-01 7.46899307e-01 -6.49989367e-01
-1.20485723e+00 -6.78647578e-01 -8.50783825e-01 3.04118574e-01
5.02730370e-01 1.16806157e-01 -7.28934407e-01 -2.04172060e-01
3.36057134e-02 -2.70797968e-01 -4.77877080e-01 -3.14342566e-02
-5.76984823e-01 -8.88396859e-01 6.25704288e-01 1.32668957e-01
8.02076817e-01 -6.84816360e-01 -3.37383270e-01 -2.11904615e-01
2.49107406e-01 -1.24183559e+00 -2.97331512e-01 -2.64361113e-01
-9.02886808e-01 -1.32019353e+00 -8.50420058e-01 -1.16178918e+00
1.08145809e+00 6.32396579e-01 6.78027153e-01 1.89624339e-01
-6.51005268e-01 6.66573584e-01 -4.41149592e-01 -2.10905686e-01
-4.62795943e-01 -3.58502492e-02 1.58984348e-01 3.26818377e-01
2.27913484e-01 -4.19179857e-01 -3.92023414e-01 8.04165483e-01
-1.07359290e+00 -5.46933055e-01 7.75217235e-01 6.29488111e-01
6.73204839e-01 2.44198382e-01 1.50750186e-02 -5.56976378e-01
5.37415087e-01 5.49685434e-02 -8.30181539e-01 3.41343254e-01
-5.08792758e-01 5.89390278e-01 3.98415744e-01 -5.81206441e-01
-9.76238847e-01 2.19226882e-01 1.87325045e-01 -1.20685980e-01
5.64808585e-03 2.82228768e-01 1.17409222e-01 -8.76275480e-01
6.36453986e-01 1.63840532e-01 2.62696266e-01 -5.50073981e-01
4.44787070e-02 6.42902792e-01 6.28224909e-01 -5.71988046e-01
9.50657129e-01 6.22021377e-01 6.96762562e-01 -1.48923516e+00
-4.13373351e-01 -7.09340394e-01 -1.00093508e+00 -1.99725226e-01
8.29742789e-01 -4.25068170e-01 -6.45475745e-01 9.17799890e-01
-9.99159694e-01 4.16862071e-01 8.27401876e-02 6.34069562e-01
-4.65485066e-01 8.45652759e-01 -4.68247712e-01 -5.93814552e-01
-9.57436711e-02 -1.37054789e+00 8.29619348e-01 4.35151845e-01
3.56912702e-01 -1.22094858e+00 3.11973751e-01 -3.54175389e-01
3.41464400e-01 4.14622396e-01 7.11586475e-01 -2.86253452e-01
-4.76146370e-01 -4.98664796e-01 3.62337902e-02 6.27366841e-01
2.84674883e-01 8.36543024e-01 -5.85696578e-01 -1.12289809e-01
2.40309328e-01 4.37373698e-01 5.56888998e-01 2.73616165e-01
7.31007040e-01 -2.21638139e-02 -3.43791366e-01 8.37230086e-01
1.56588280e+00 3.63663703e-01 9.27674055e-01 3.17650169e-01
3.93377006e-01 8.92282128e-02 6.55798554e-01 1.08223006e-01
-3.91609333e-02 7.78944731e-01 -1.22614102e-02 -3.69537771e-01
2.14310307e-02 3.80722731e-01 4.04566735e-01 9.14726555e-01
-8.55833530e-01 4.65026319e-01 -5.19546926e-01 3.89966458e-01
-1.51340795e+00 -1.21961188e+00 -5.41837096e-01 2.23159981e+00
7.11130679e-01 -2.72679210e-01 -9.30044055e-02 5.14113545e-01
9.60280538e-01 2.06155419e-01 -7.98161328e-02 -7.13986754e-01
-5.03169954e-01 4.57721770e-01 7.28946030e-01 6.31449580e-01
-1.42871535e+00 7.16432095e-01 7.61281109e+00 3.63805264e-01
-1.59084415e+00 -2.70096272e-01 1.28330126e-01 8.35595369e-01
-1.08770125e-01 2.89152890e-01 -6.31250024e-01 -1.01930931e-01
2.34102860e-01 -4.19780046e-01 5.31567782e-02 9.18319881e-01
-2.54083544e-01 -2.16956288e-01 -6.85087860e-01 1.06014776e+00
2.39383206e-01 -6.06079817e-01 1.05084457e-01 -9.47791189e-02
9.24849570e-01 -4.72487092e-01 2.50113100e-01 -2.79986024e-01
-4.08111840e-01 -6.28716350e-01 1.91029444e-01 8.94325137e-01
5.26212871e-01 -9.47377980e-01 9.00147021e-01 -2.41294056e-01
-1.11143339e+00 4.05877769e-01 -8.00089777e-01 8.08258802e-02
-1.85699120e-01 5.75676799e-01 -6.08786702e-01 9.09053326e-01
2.15777919e-01 6.10482275e-01 -8.59606922e-01 1.10788643e+00
-3.07973087e-01 9.05462727e-03 -1.55778736e-01 -1.61371589e-01
3.32643725e-02 -6.26322150e-01 6.68675184e-01 1.04480004e+00
2.64661908e-01 2.03986049e-01 -8.93664882e-02 3.56312841e-01
5.36537766e-01 4.47628826e-01 -7.57922351e-01 2.03753054e-01
-9.93182436e-02 1.43650424e+00 -6.18422449e-01 -1.58587620e-01
-3.45404297e-01 1.13385594e+00 -3.32898766e-01 4.52394873e-01
-7.79163063e-01 -7.57611632e-01 8.14877510e-01 -1.49428621e-01
2.72481710e-01 -4.72730041e-01 2.93978781e-01 -1.37392831e+00
-6.30954141e-03 -6.64419711e-01 1.78819150e-01 -5.78116298e-01
-8.91381025e-01 6.87687159e-01 3.10074896e-01 -1.33452713e+00
-3.65378708e-01 -1.16727209e+00 -3.85850072e-01 7.99208760e-01
-1.17285132e+00 -9.23076034e-01 -2.09207967e-01 7.79352725e-01
-7.52802268e-02 -2.57571101e-01 9.30985332e-01 2.58609205e-01
-5.03169596e-01 5.46367407e-01 2.95745432e-01 3.81739616e-01
7.51340151e-01 -9.56315875e-01 -5.76630533e-02 8.64397109e-01
1.34103730e-01 5.72473586e-01 9.38019812e-01 -1.39171809e-01
-1.58715570e+00 -1.65773481e-01 8.15514445e-01 -2.67912030e-01
4.77467895e-01 3.69828761e-01 -7.96332300e-01 8.44911218e-01
1.30961299e-01 2.41264924e-01 3.10735583e-01 -3.60521466e-01
-2.16315195e-01 -3.20660204e-01 -1.15578604e+00 3.86552513e-01
6.34964764e-01 -4.65923071e-01 -5.33096492e-01 3.22273612e-01
2.09938720e-01 -4.79676366e-01 -1.24172544e+00 5.29244661e-01
7.08692729e-01 -1.04324639e+00 9.91771460e-01 -5.21991551e-01
-3.60348560e-02 -7.00001180e-01 1.06361806e-01 -9.95572448e-01
-4.21990484e-01 -6.55256093e-01 5.20797014e-01 8.56753349e-01
7.88676664e-02 -9.18525577e-01 2.97367685e-02 1.61579937e-01
2.22269759e-01 -4.89346892e-01 -7.69807696e-01 -9.17083621e-01
-3.20891857e-01 1.17615506e-01 5.06040514e-01 1.17472315e+00
1.24408014e-01 1.18082695e-01 -2.84703642e-01 3.51570815e-01
4.70988184e-01 1.32769421e-01 8.60041022e-01 -1.22835732e+00
-7.79997604e-03 -2.75502861e-01 -1.62944007e+00 -6.96819186e-01
2.93197870e-01 -6.80983067e-01 -4.95361507e-01 -9.64086711e-01
2.05811843e-01 -2.81110972e-01 -1.33820415e-01 1.79383829e-01
1.79205000e-01 2.83348680e-01 -2.31345192e-01 2.77906656e-01
-1.49317250e-01 -2.80770436e-02 1.59179556e+00 3.21344770e-02
-8.94051492e-02 6.45384490e-02 -2.66916335e-01 9.08408463e-01
7.70780921e-01 -1.27519920e-01 -2.43714258e-01 -1.26907840e-01
-2.61257917e-01 -3.21729183e-01 1.90165147e-01 -1.08609343e+00
-4.68499921e-02 -2.87091404e-01 3.52729678e-01 -3.33791047e-01
-6.17079735e-02 -8.47226739e-01 4.63709384e-01 6.27868772e-01
1.63212910e-01 5.81091285e-01 -1.71278909e-01 -2.15912402e-01
-6.20022237e-01 -5.13381124e-01 1.12350142e+00 1.58022687e-01
-1.03253734e+00 1.55703023e-01 -2.41333216e-01 -5.23681164e-01
1.11718810e+00 -1.93361625e-01 -2.72406012e-01 -2.81108797e-01
-5.44828415e-01 -5.76300621e-01 6.33600652e-01 3.02948326e-01
3.87942225e-01 -1.51915371e+00 -4.73539144e-01 5.70224464e-01
-1.29528016e-01 -4.02024090e-01 1.76373288e-01 1.20587063e+00
-1.31352139e+00 5.33196867e-01 -7.73434877e-01 -7.33642101e-01
-1.52094781e+00 9.10668969e-01 6.88517570e-01 9.66418814e-03
-3.60061258e-01 1.69491664e-01 -1.98827818e-01 6.51059523e-02
-2.56270975e-01 -3.87293011e-01 -3.90380591e-01 -1.29761353e-01
5.14983773e-01 4.54508066e-01 -1.12392381e-02 -1.32349670e+00
-3.91900808e-01 1.31076932e+00 1.79244995e-01 -1.02705918e-01
1.02333868e+00 1.97993070e-01 -8.00997555e-01 1.63879782e-01
1.59284449e+00 2.71534830e-01 -5.28341889e-01 -1.32809877e-01
-2.83786774e-01 -6.00222528e-01 -2.97443777e-01 8.14735796e-03
-1.17371452e+00 3.41441602e-01 7.85942554e-01 5.42074502e-01
1.32421911e+00 -1.33927122e-01 2.44851589e-01 4.98041183e-01
3.02394539e-01 -8.32245409e-01 -2.25456506e-01 6.49670660e-01
9.99703288e-01 -8.52867424e-01 4.09172505e-01 -7.14184225e-01
-4.57496017e-01 1.58744109e+00 2.50527114e-01 -7.35813260e-01
9.09613550e-01 8.06538388e-02 3.31740946e-01 1.34781361e-01
-1.34737715e-01 -3.68726492e-01 6.85212731e-01 7.49929965e-01
6.75712466e-01 5.87849803e-02 -1.19077277e+00 -1.50911823e-01
-9.49873403e-02 -2.91682214e-01 6.57793939e-01 8.16903889e-01
-4.13761407e-01 -1.31445336e+00 -7.02382982e-01 -3.67901534e-01
-6.73309863e-01 5.80386281e-01 -3.64165038e-01 1.27803290e+00
-4.72628772e-02 4.43638355e-01 7.31801391e-02 -3.66866171e-01
5.07004797e-01 -1.35572463e-01 8.95276964e-01 1.99021310e-01
1.85700301e-02 -1.33104160e-01 -2.22612172e-01 -2.75654346e-01
-9.90027726e-01 -7.28662312e-01 -1.18520188e+00 -4.00560766e-01
-2.47242376e-01 2.33407900e-01 8.31505835e-01 6.47830486e-01
7.70727694e-02 1.34460345e-01 1.11082220e+00 -7.98014581e-01
-8.32234085e-01 -8.77388895e-01 -8.27081501e-01 6.05756938e-01
3.75065088e-01 -6.40030980e-01 -4.91626978e-01 3.06450039e-01] | [9.5503511428833, -1.671940803527832] |
4d10f362-936c-4b6b-b15e-671387fee453 | swin-unet-unet-like-pure-transformer-for | 2105.05537 | null | https://arxiv.org/abs/2105.05537v1 | https://arxiv.org/pdf/2105.05537v1.pdf | Swin-Unet: Unet-like Pure Transformer for Medical Image Segmentation | In the past few years, convolutional neural networks (CNNs) have achieved milestones in medical image analysis. Especially, the deep neural networks based on U-shaped architecture and skip-connections have been widely applied in a variety of medical image tasks. However, although CNN has achieved excellent performance, it cannot learn global and long-range semantic information interaction well due to the locality of the convolution operation. In this paper, we propose Swin-Unet, which is an Unet-like pure Transformer for medical image segmentation. The tokenized image patches are fed into the Transformer-based U-shaped Encoder-Decoder architecture with skip-connections for local-global semantic feature learning. Specifically, we use hierarchical Swin Transformer with shifted windows as the encoder to extract context features. And a symmetric Swin Transformer-based decoder with patch expanding layer is designed to perform the up-sampling operation to restore the spatial resolution of the feature maps. Under the direct down-sampling and up-sampling of the inputs and outputs by 4x, experiments on multi-organ and cardiac segmentation tasks demonstrate that the pure Transformer-based U-shaped Encoder-Decoder network outperforms those methods with full-convolution or the combination of transformer and convolution. The codes and trained models will be publicly available at https://github.com/HuCaoFighting/Swin-Unet. | ['Manning Wang', 'Qi Tian', 'Xiaopeng Zhang', 'Dongsheng Jiang', 'Joy Chen', 'Yueyue Wang', 'Hu Cao'] | 2021-05-12 | null | null | null | null | ['cardiac-segmentation'] | ['medical'] | [ 1.01556897e-01 -1.62021741e-02 -5.15922531e-02 -4.47152436e-01
-5.84899247e-01 -1.28132492e-01 8.92042965e-02 -9.30353999e-02
-2.98915803e-01 4.47784573e-01 1.70467213e-01 -3.09594780e-01
4.81139794e-02 -9.51786995e-01 -7.10051179e-01 -7.68446565e-01
1.64063185e-01 -3.44895981e-02 5.88490248e-01 -6.07900135e-02
-8.74930173e-02 1.22573525e-01 -9.57400501e-01 7.00928152e-01
7.29095757e-01 1.37399161e+00 5.74477255e-01 3.58061880e-01
-2.90291488e-01 6.66834593e-01 -2.74791509e-01 -1.00290887e-01
1.56507552e-01 -6.43826187e-01 -7.36823916e-01 -1.74901277e-01
4.32792399e-03 -3.48613769e-01 -5.37767768e-01 9.66753066e-01
7.91779816e-01 -2.36415535e-01 2.23392278e-01 -6.99199915e-01
-9.08180475e-01 4.62188184e-01 -5.25991738e-01 4.52570409e-01
-4.16611247e-02 3.68941426e-02 6.46544516e-01 -8.50683808e-01
3.08276206e-01 1.04908931e+00 9.41173851e-01 4.81215239e-01
-8.56735170e-01 -7.38867819e-01 -1.19361363e-01 3.45814884e-01
-1.43570125e+00 5.96839897e-02 6.04133844e-01 -2.22526312e-01
9.57662106e-01 1.02862686e-01 7.38636255e-01 5.59372067e-01
5.97237945e-01 8.03286552e-01 9.01542246e-01 -8.45719203e-02
-2.35259548e-01 -1.02070943e-01 -9.71453935e-02 9.75092113e-01
-2.64085889e-01 2.05286462e-02 -2.85954505e-01 1.96448252e-01
1.35658228e+00 5.05573094e-01 -4.02471274e-01 3.01475767e-02
-1.38498652e+00 7.88494885e-01 1.12898576e+00 6.37013495e-01
-2.86692917e-01 1.17626488e-01 6.88021004e-01 3.26169759e-01
4.36541706e-01 2.00454772e-01 -6.14836156e-01 1.74571663e-01
-9.03963625e-01 -1.54516324e-01 4.89151292e-02 9.26359773e-01
6.24152064e-01 1.70815196e-02 -4.72613961e-01 1.00174606e+00
-9.91497561e-02 1.95944190e-01 9.32532072e-01 -5.07713974e-01
2.64860332e-01 7.80668318e-01 -5.10669410e-01 -6.86303139e-01
-5.85000813e-01 -8.37969482e-01 -1.33999574e+00 -1.36795267e-01
1.46980152e-01 -2.01562613e-01 -1.21598339e+00 1.45231366e+00
2.90220290e-01 3.45501989e-01 5.04934043e-02 1.03705895e+00
1.22013736e+00 8.13471317e-01 -3.54916528e-02 1.02896743e-01
1.80217028e+00 -1.13112819e+00 -7.48850465e-01 -6.19489588e-02
7.13976681e-01 -8.31364572e-01 1.06478167e+00 3.91870178e-02
-9.41150665e-01 -7.83899248e-01 -1.03191316e+00 -5.57634234e-01
-2.95720249e-01 2.88167417e-01 1.70156077e-01 1.93455145e-01
-9.88713920e-01 5.99192023e-01 -9.30109739e-01 -2.09254563e-01
9.27950859e-01 2.41891474e-01 -3.79894078e-01 -1.51936129e-01
-1.36458457e+00 7.06507802e-01 2.52179414e-01 2.99330801e-01
-7.91711628e-01 -7.68910527e-01 -8.13023210e-01 2.52197981e-01
5.98980710e-02 -7.26932108e-01 1.16421676e+00 -9.28554416e-01
-1.17951143e+00 6.77705884e-01 -1.14174291e-01 -4.38969016e-01
2.95222908e-01 4.08193655e-02 -2.03772455e-01 2.52511591e-01
3.21917832e-01 7.19789028e-01 5.21286488e-01 -4.89318401e-01
-5.90089440e-01 -3.87524456e-01 -1.87128171e-01 2.28399515e-01
-3.77341121e-01 1.32806391e-01 -4.32043195e-01 -9.77430642e-01
3.92365932e-01 -5.93478084e-01 -2.71522075e-01 2.42611691e-01
-4.37114865e-01 -2.94231959e-02 1.06844151e+00 -7.66537786e-01
1.24464834e+00 -2.39012098e+00 -1.48018837e-01 -7.45096877e-02
2.50967562e-01 2.92834312e-01 -7.14136660e-02 2.72224873e-01
-3.13377917e-01 3.79791521e-02 -4.35164005e-01 -3.10726855e-02
-5.29914439e-01 2.16256082e-01 3.24519277e-02 2.85429120e-01
2.06997484e-01 1.11733067e+00 -8.04074943e-01 -5.92411995e-01
2.57069618e-01 6.09786332e-01 -5.03402591e-01 1.27199829e-01
6.64316863e-02 6.25308931e-01 -6.02895439e-01 6.36582077e-01
6.72962070e-01 -4.38094765e-01 -9.73746032e-02 -4.52494711e-01
-1.45542428e-01 2.40469709e-01 -5.44217944e-01 1.92624116e+00
-4.49526697e-01 5.54241061e-01 3.59471999e-02 -1.32555473e+00
8.82160783e-01 6.06416166e-01 5.27128518e-01 -1.06142819e+00
2.80412078e-01 3.97602290e-01 1.01605512e-01 -6.55718207e-01
-5.60744330e-02 -4.79547560e-01 8.83537009e-02 1.29886314e-01
1.58329114e-01 2.95671910e-01 -1.11464463e-01 -1.46162525e-01
9.92297173e-01 -2.71822382e-02 1.99886218e-01 -4.42587912e-01
6.56534374e-01 -1.10454559e-01 7.52014756e-01 1.98952138e-01
1.33495545e-03 9.95536804e-01 5.53753495e-01 -6.72217190e-01
-1.00989830e+00 -8.92355323e-01 -4.94357854e-01 7.98583090e-01
1.69679567e-01 -3.29907060e-01 -8.07844043e-01 -5.51037073e-01
-3.62391442e-01 1.09566025e-01 -6.47456408e-01 -2.78857440e-01
-7.15961933e-01 -8.57296109e-01 6.91886902e-01 8.75169814e-01
1.10289574e+00 -1.15740287e+00 -7.57762909e-01 4.19433236e-01
-3.27829242e-01 -1.03650510e+00 -9.12473261e-01 2.83522516e-01
-1.09518695e+00 -1.06034410e+00 -1.13804460e+00 -1.23802102e+00
8.11751246e-01 1.63641334e-01 6.61470473e-01 1.70142367e-01
-5.88668704e-01 -3.96239579e-01 -3.79182339e-01 -1.31383240e-01
1.57856941e-01 1.95282117e-01 -7.04939842e-01 -2.22332235e-02
2.05107927e-01 -6.25334918e-01 -1.06370544e+00 4.54806417e-01
-9.34238136e-01 3.85131985e-01 7.22184956e-01 1.09894753e+00
7.28285432e-01 -6.22944497e-02 5.13747275e-01 -8.24169934e-01
4.05325323e-01 -3.39150459e-01 -2.03307569e-01 1.00593090e-01
-2.92136908e-01 9.78693832e-03 8.70649815e-01 -2.23626897e-01
-8.02737534e-01 -1.93428531e-01 -6.81729317e-01 -6.55094564e-01
1.45759091e-01 5.34643471e-01 -3.60063575e-02 6.38442487e-02
5.09557068e-01 5.74849486e-01 1.00188412e-01 -5.00409722e-01
-1.18881755e-01 5.91538310e-01 4.42875803e-01 -3.42363060e-01
3.47576827e-01 4.26713377e-01 -2.22642601e-01 -5.75875938e-01
-8.79116774e-01 -1.70413047e-01 -4.30647433e-01 7.00390562e-02
1.31851459e+00 -8.80675733e-01 -3.41555983e-01 7.31757224e-01
-1.11671996e+00 -3.62091392e-01 -3.03759724e-01 4.92937148e-01
-2.98776597e-01 1.55897409e-01 -9.80716586e-01 -1.51581168e-02
-7.00731158e-01 -1.46436679e+00 9.43343818e-01 4.91199374e-01
2.07104087e-01 -9.12740767e-01 -4.39721376e-01 1.32643312e-01
6.85316324e-01 1.53464586e-01 1.04125261e+00 -4.55882192e-01
-5.82157373e-01 -4.22560982e-03 -5.69762588e-01 5.55547118e-01
2.59451568e-01 -5.61735511e-01 -7.20522523e-01 -2.08191141e-01
7.37072676e-02 -1.48353562e-01 9.69839394e-01 7.83167481e-01
1.62173510e+00 -1.39747754e-01 -3.83245528e-01 1.10511339e+00
1.50739026e+00 4.12049443e-01 8.25615466e-01 2.44775504e-01
6.56289279e-01 1.51460171e-01 2.96433270e-01 2.34751701e-01
3.28562587e-01 3.28133285e-01 3.48592371e-01 -6.82597280e-01
-3.17457348e-01 -2.27489367e-01 -1.01900123e-01 7.69556761e-01
-4.19821478e-02 1.79193541e-01 -8.10919821e-01 5.17167687e-01
-1.68617678e+00 -7.37106979e-01 1.02381958e-02 1.89084077e+00
9.32814240e-01 7.68150836e-02 -1.88120574e-01 -6.19558729e-02
7.87852407e-01 1.67766720e-01 -6.23975456e-01 -2.18962789e-01
-7.21178995e-03 4.30321217e-01 5.48814595e-01 2.97840774e-01
-9.29402590e-01 7.22475708e-01 5.38687658e+00 1.02995718e+00
-1.42909241e+00 4.93239313e-01 8.88173580e-01 8.90158191e-02
-1.03486240e-01 -1.91191286e-01 -5.73367774e-01 4.61821049e-01
4.71993595e-01 1.61489844e-01 1.07003108e-03 5.85246503e-01
1.36956379e-01 1.97579950e-01 -6.54921174e-01 9.22170341e-01
-2.00229123e-01 -1.52795041e+00 -2.16782615e-02 -1.63396776e-01
4.47342724e-01 2.32313588e-01 1.23457283e-01 1.37188286e-01
-2.06933513e-01 -1.03820109e+00 6.08189523e-01 2.85275549e-01
1.28336096e+00 -6.44911766e-01 9.53586459e-01 2.68862516e-01
-1.55625761e+00 -1.99415475e-01 -4.38681453e-01 4.76374775e-02
4.89596277e-02 7.08103180e-01 -4.89138454e-01 5.30699790e-01
9.39688981e-01 9.79340017e-01 -2.77059704e-01 9.51576352e-01
-2.49153213e-03 5.09185612e-01 -2.04375654e-01 1.56395018e-01
5.01300514e-01 -1.51312277e-01 1.93599254e-01 1.26282394e+00
5.61963320e-01 2.78467059e-01 5.38504049e-02 8.08303356e-01
-1.56781912e-01 6.20679185e-02 -2.26091623e-01 2.10262373e-01
2.67688841e-01 1.22075987e+00 -8.62004697e-01 -4.07697618e-01
-5.43585360e-01 9.21762109e-01 7.64105543e-02 2.94485986e-01
-9.48284924e-01 -7.02274978e-01 5.27879775e-01 4.58410650e-01
5.54166675e-01 1.74498722e-01 -3.28809738e-01 -1.01816344e+00
6.03967942e-02 -6.03336811e-01 3.44148993e-01 -7.75252640e-01
-1.03233016e+00 9.88140881e-01 -1.96510509e-01 -1.29155672e+00
3.61155778e-01 -5.72037697e-01 -8.09348226e-01 9.63477671e-01
-1.64634418e+00 -1.19605672e+00 -4.23781246e-01 8.73622894e-01
6.05588615e-01 -7.87876919e-02 7.40673602e-01 5.65274179e-01
-5.37511408e-01 5.11118889e-01 1.03420474e-01 5.82005501e-01
5.90917051e-01 -8.97893071e-01 1.81054443e-01 5.32416224e-01
-3.35991234e-01 6.33520842e-01 1.80686072e-01 -5.22950351e-01
-7.85297096e-01 -1.26929414e+00 6.89053118e-01 2.54838139e-01
2.44806454e-01 -2.54704922e-01 -8.82641375e-01 7.49956310e-01
5.16378999e-01 5.71614981e-01 4.38095152e-01 -2.39680618e-01
-2.22536698e-01 -3.81210595e-01 -1.12356889e+00 1.47221267e-01
9.20437813e-01 -4.14743364e-01 -4.67229456e-01 3.35367560e-01
7.55352020e-01 -7.68436909e-01 -1.03062236e+00 5.36397576e-01
5.00239670e-01 -1.08847773e+00 1.02031946e+00 -8.45488757e-02
8.01466167e-01 -3.08920532e-01 3.54031473e-02 -1.13764930e+00
-4.44569170e-01 -1.88820437e-01 5.72935879e-01 7.28006124e-01
3.61996800e-01 -8.85972857e-01 6.90899372e-01 -2.32720509e-01
-7.01790869e-01 -1.50027788e+00 -1.17322612e+00 -3.07505816e-01
2.77272135e-01 -2.07957476e-02 5.83195508e-01 7.10238218e-01
-1.00004546e-01 3.13519001e-01 -1.68712392e-01 -1.30363762e-01
3.33961487e-01 2.45500237e-01 3.67238373e-02 -7.47519672e-01
-1.33492559e-01 -3.78457218e-01 -3.98838103e-01 -1.42796302e+00
-3.93391967e-01 -1.03768539e+00 -1.36748999e-01 -1.74369180e+00
2.80430645e-01 -4.64989245e-01 -5.13382852e-01 7.82635033e-01
-1.14595443e-01 4.71025854e-01 -1.29832178e-02 1.30171210e-01
-1.83387369e-01 5.09597957e-01 1.88406289e+00 -1.85255662e-01
9.48631391e-02 -2.97935214e-02 -5.75443268e-01 5.52802801e-01
9.57254231e-01 -3.81286860e-01 -4.86507624e-01 -6.95196867e-01
-1.05740868e-01 2.41994143e-01 5.56282938e-01 -1.06050694e+00
3.00529420e-01 2.80973762e-01 6.88983500e-01 -6.38639271e-01
1.90903425e-01 -7.35162914e-01 -2.37844065e-02 8.35037768e-01
-3.84889930e-01 3.04713160e-01 1.67135298e-01 1.47890329e-01
-6.44418955e-01 -2.71855611e-02 9.50405896e-01 -4.99287426e-01
-4.51962620e-01 7.96185017e-01 -1.82983711e-01 6.00353368e-02
9.15305674e-01 -3.31182837e-01 -1.93532065e-01 3.70138958e-02
-8.31132352e-01 2.21591666e-01 6.64720908e-02 2.09686384e-01
9.79088783e-01 -1.36587048e+00 -6.53753102e-01 5.43716908e-01
-1.31476924e-01 4.13068831e-01 8.29838574e-01 1.23689926e+00
-8.21736693e-01 4.94617134e-01 -4.35915142e-01 -6.83195412e-01
-1.12712967e+00 3.65124851e-01 7.69512594e-01 -4.07199591e-01
-1.16065979e+00 9.86721933e-01 7.00406253e-01 -3.86844695e-01
-5.15350290e-02 -5.19218564e-01 -3.76377515e-02 -2.99230993e-01
4.17310685e-01 -1.18856564e-01 9.16999802e-02 -3.93097013e-01
-3.07615936e-01 6.50885582e-01 -1.93564102e-01 2.74386019e-01
1.38430762e+00 4.38360265e-03 -3.68236810e-01 1.07414685e-01
1.54768825e+00 -4.91617918e-01 -1.14330709e+00 -4.35252279e-01
-5.12031138e-01 -2.46278659e-01 1.69130534e-01 -6.35139704e-01
-1.68422019e+00 1.25033760e+00 6.80658042e-01 -2.86269933e-02
1.45136774e+00 -1.16780382e-02 1.40606117e+00 -1.82952225e-01
4.99231219e-02 -7.05979466e-01 3.68364044e-02 3.75871330e-01
7.44308293e-01 -9.11048710e-01 -2.59592861e-01 -3.92903507e-01
-4.94285434e-01 1.22669935e+00 7.03159392e-01 -3.12164873e-01
9.61046040e-01 6.13718331e-01 1.32223845e-01 -2.69684225e-01
-6.18862569e-01 -4.40625697e-02 1.54225761e-02 3.81352454e-01
7.02524960e-01 -5.69266528e-02 -3.51610005e-01 7.31644630e-01
-1.10231087e-01 1.75193086e-01 1.17830358e-01 8.37297142e-01
-3.80211800e-01 -9.17017281e-01 -2.10664153e-01 7.34493911e-01
-5.87230027e-01 -3.78292412e-01 3.18585157e-01 5.44439137e-01
5.39362907e-01 4.40708667e-01 1.47394612e-01 -4.56238836e-01
3.53905708e-01 -1.62219703e-01 3.69652450e-01 -5.41877449e-01
-8.10819566e-01 4.09905702e-01 -2.99872369e-01 -5.42770684e-01
-2.71842808e-01 -4.48413998e-01 -1.61995471e+00 -2.89881639e-02
-1.69478357e-01 1.17795743e-01 2.30215237e-01 8.60629380e-01
3.43130231e-01 1.10971570e+00 4.03154820e-01 -4.16710794e-01
-3.26657146e-01 -1.06817782e+00 -5.75069845e-01 8.35226774e-02
4.65495855e-01 -3.55957657e-01 3.23593132e-02 4.25366266e-03] | [14.549463272094727, -2.5909957885742188] |
46a26cf1-24fc-4046-aa71-766db36f87a0 | mrn-a-locally-and-globally-mention-based | null | null | https://aclanthology.org/2021.findings-acl.117 | https://aclanthology.org/2021.findings-acl.117.pdf | MRN: A Locally and Globally Mention-Based Reasoning Network for Document-Level Relation Extraction | null | ['Donghong Ji', 'Yafeng Ren', 'Hao Fei', 'Fei Li', 'Kang Xu', 'Jingye Li'] | null | null | null | null | findings-acl-2021-8 | ['document-level-relation-extraction'] | ['natural-language-processing'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.357705593109131, 3.7358553409576416] |
e46f73ca-26bc-44e9-8e04-85d8a56071b0 | logical-tasks-for-measuring-extrapolation-and | 2211.07727 | null | https://arxiv.org/abs/2211.07727v1 | https://arxiv.org/pdf/2211.07727v1.pdf | Logical Tasks for Measuring Extrapolation and Rule Comprehension | Logical reasoning is essential in a variety of human activities. A representative example of a logical task is mathematics. Recent large-scale models trained on large datasets have been successful in various fields, but their reasoning ability in arithmetic tasks is limited, which we reproduce experimentally. Here, we recast this limitation as not unique to mathematics but common to tasks that require logical operations. We then propose a new set of tasks, termed logical tasks, which will be the next challenge to address. This higher point of view helps the development of inductive biases that have broad impact beyond the solution of individual tasks. We define and characterize logical tasks and discuss system requirements for their solution. Furthermore, we discuss the relevance of logical tasks to concepts such as extrapolation, explainability, and inductive bias. Finally, we provide directions for solving logical tasks. | ['Ryota Kanai', 'Ippei Fujisawa'] | 2022-11-14 | null | null | null | null | ['logical-reasoning'] | ['reasoning'] | [ 1.29481256e-01 4.05861676e-01 -7.07569271e-02 -5.82177579e-01
-2.19988599e-01 -4.85483766e-01 7.48125136e-01 2.71265626e-01
-1.22152641e-01 8.71678948e-01 1.78951502e-01 -7.03865647e-01
-6.55147135e-01 -1.01380694e+00 -9.17264640e-01 -2.19227687e-01
-7.89881200e-02 6.09800994e-01 -8.99201259e-02 -3.32940310e-01
7.53627419e-01 4.65522528e-01 -1.30065262e+00 5.93112767e-01
1.17577648e+00 8.20873082e-01 -1.58908680e-01 3.96930844e-01
-1.44671187e-01 1.58732665e+00 -6.34325504e-01 -5.05123138e-01
-1.31121427e-01 -3.45898628e-01 -1.29869604e+00 -5.06535411e-01
5.97673893e-01 -3.26236755e-01 -9.69785079e-03 1.02676475e+00
1.64892018e-01 2.61288166e-01 8.66654932e-01 -1.65369022e+00
-9.59295273e-01 1.05075943e+00 1.30579352e-01 3.57857614e-04
5.64595282e-01 2.22037052e-04 1.16954875e+00 -7.70982325e-01
4.29436058e-01 1.52537763e+00 8.15926313e-01 4.75428075e-01
-1.25416446e+00 -6.09633029e-01 4.08105701e-01 5.94007671e-01
-1.15512288e+00 -3.58930588e-01 7.01607823e-01 -4.68300849e-01
1.03362155e+00 2.33426318e-01 7.43663192e-01 9.36799407e-01
3.41801077e-01 8.10362399e-01 1.40741932e+00 -4.50331122e-01
4.44793880e-01 -8.83474723e-02 5.08838058e-01 6.72299922e-01
5.86455166e-01 -1.21006288e-01 -7.02024817e-01 3.76124606e-02
7.81269610e-01 -3.87875289e-01 -1.87324852e-01 -3.50414544e-01
-1.34889197e+00 8.30773115e-01 3.19286197e-01 2.73593187e-01
-1.79625198e-01 5.33102691e-01 1.48907647e-01 4.94736224e-01
9.89225134e-02 1.05767429e+00 -4.12805647e-01 2.28216246e-01
-7.04397559e-01 7.65329301e-01 1.03711069e+00 9.82442498e-01
3.32998812e-01 -1.43305257e-01 -1.73670724e-01 1.93205550e-01
2.88980186e-01 3.18161368e-01 4.72815381e-03 -1.30422831e+00
4.06925350e-01 5.71645677e-01 4.77345884e-02 -8.76185834e-01
-6.75535142e-01 -3.68287414e-01 -6.39136732e-01 2.09580630e-01
7.20727444e-01 2.81195432e-01 -3.26036036e-01 1.88151133e+00
3.39374617e-02 3.54509540e-02 -1.01635214e-02 5.44653773e-01
7.87683666e-01 3.16797197e-01 4.19980884e-01 -8.73069465e-03
1.26460719e+00 -6.71077490e-01 -8.43535006e-01 -2.62155622e-01
7.78567016e-01 -5.13192177e-01 1.17462218e+00 6.91264153e-01
-1.47575319e+00 -4.23111647e-01 -9.76535976e-01 -5.48223972e-01
-6.33218586e-01 -1.23207845e-01 1.32933986e+00 5.21950781e-01
-1.02660942e+00 6.32431328e-01 -4.12648708e-01 -2.36268893e-01
4.85345095e-01 2.54069090e-01 7.51324371e-02 -2.44713575e-02
-1.50949240e+00 1.56520069e+00 4.66437221e-01 7.49768913e-02
-6.69228017e-01 -9.12208140e-01 -9.16993558e-01 3.76485288e-01
4.21921253e-01 -9.44892108e-01 1.29577672e+00 -5.04515886e-01
-1.00797355e+00 9.70993698e-01 -3.10313076e-01 -6.42862856e-01
5.69777906e-01 -2.85292059e-01 -1.27932459e-01 -9.55063924e-02
-2.25510914e-02 6.51582181e-01 4.82418835e-01 -1.14936042e+00
-1.97896972e-01 -1.03761248e-01 5.89076757e-01 9.36721340e-02
-4.74734977e-02 6.84521198e-02 3.85905027e-01 -6.15572631e-01
2.63195246e-01 -5.43794096e-01 2.14440413e-02 1.58853382e-01
-2.80755877e-01 -6.60396397e-01 2.39853695e-01 -2.83952802e-01
1.40809691e+00 -1.74549890e+00 3.09483498e-01 1.53589904e-01
4.30372745e-01 -2.33298615e-01 4.54401933e-02 2.39529356e-01
-2.26215795e-01 4.44018036e-01 -9.70383584e-02 1.00152723e-01
5.92599273e-01 -1.62308998e-02 -6.31848276e-01 1.35106668e-02
4.32869136e-01 1.32790279e+00 -1.02734673e+00 -7.17226803e-01
1.16498679e-01 -1.32423893e-01 -6.51035130e-01 6.29721656e-02
-6.95955038e-01 2.35476792e-02 -3.06450278e-01 6.31811798e-01
5.60689449e-01 -5.33532858e-01 3.00474018e-01 -2.53246650e-02
1.51316926e-01 8.29775512e-01 -1.12145686e+00 1.36010396e+00
-4.19896811e-01 6.52883530e-01 -3.04645181e-01 -1.08350074e+00
6.69805348e-01 6.14318140e-02 -7.69429281e-02 -3.93162429e-01
1.04757965e-01 1.34613112e-01 4.47370768e-01 -6.93964064e-01
5.18295109e-01 -4.13745493e-01 -1.38093129e-01 7.70583391e-01
-7.78886825e-02 -8.55452836e-01 4.10813481e-01 3.57181281e-01
1.01519907e+00 3.47239166e-01 5.87592542e-01 -4.96698827e-01
4.98313725e-01 1.61423236e-01 2.07536802e-01 9.82704461e-01
-1.87751636e-01 6.12500459e-02 8.15361261e-01 -5.59702218e-01
-8.59726429e-01 -1.18373048e+00 -3.41475874e-01 1.16273844e+00
1.24738842e-01 -4.35750365e-01 -5.32898068e-01 -4.62920487e-01
2.03099847e-01 1.15416384e+00 -5.44789612e-01 -2.66063809e-01
-6.19389653e-01 -5.69063842e-01 6.99648499e-01 6.82594836e-01
4.13526118e-01 -1.19101357e+00 -5.89406371e-01 -4.20188084e-02
-3.78742903e-01 -1.10842705e+00 4.67821985e-01 3.33301395e-01
-1.16039002e+00 -1.33684540e+00 8.26030746e-02 -7.03557432e-01
6.61998391e-01 -4.22020443e-03 1.66877437e+00 4.24482852e-01
8.44533518e-02 3.95117730e-01 -8.28856155e-02 -7.55472839e-01
-3.09402704e-01 -1.62259750e-02 -8.72646794e-02 -8.08687866e-01
6.57474935e-01 -5.00660956e-01 -2.36007734e-03 1.02253050e-01
-6.96483195e-01 2.97748595e-01 4.33096230e-01 6.73941493e-01
1.28125370e-01 2.53210694e-01 6.10478818e-01 -9.29118037e-01
1.00064266e+00 -4.80829418e-01 -4.40837890e-01 5.87721527e-01
-4.72973675e-01 2.27901205e-01 6.30649269e-01 -4.01372641e-01
-1.11515701e+00 -4.69056934e-01 1.80795699e-01 1.62486687e-01
-1.53205525e-02 6.19993627e-01 -7.48073906e-02 -1.39872795e-02
6.77484751e-01 -8.69137943e-02 -2.61521995e-01 -1.66453123e-02
3.70686710e-01 -1.83907934e-02 1.34870097e-01 -1.49320090e+00
7.68319786e-01 3.04507822e-01 5.19105196e-01 -6.53153479e-01
-1.25115967e+00 3.16413373e-01 -5.08022845e-01 -1.44520715e-01
6.56527877e-01 -8.30304623e-01 -1.15018868e+00 2.77496397e-01
-1.36395657e+00 -7.98422337e-01 -3.73367667e-01 4.30528849e-01
-8.21296453e-01 2.02152893e-01 -6.32537365e-01 -8.59169066e-01
1.15083404e-01 -9.22575235e-01 7.15946496e-01 3.78361046e-02
-9.82752204e-01 -1.33956981e+00 -4.09927964e-01 4.39248383e-01
3.58990520e-01 2.33303811e-02 1.64364159e+00 -6.40674651e-01
-9.92638052e-01 1.93837121e-01 -3.96913320e-01 1.83455780e-01
-4.41822201e-01 -5.30306846e-02 -8.29878211e-01 3.15339327e-01
1.55779123e-01 -9.25082505e-01 8.83116961e-01 3.87948066e-01
1.58090341e+00 3.94597761e-02 -1.90160260e-01 4.68734086e-01
1.05787003e+00 -4.55715246e-02 7.96191216e-01 4.06688541e-01
3.79566014e-01 7.94802964e-01 6.85621202e-01 1.95419982e-01
7.58034110e-01 2.36758932e-01 9.04688761e-02 2.49410033e-01
-8.87049828e-03 -2.99817830e-01 1.54092610e-01 5.47069848e-01
-4.17642444e-01 9.69998015e-04 -1.15350485e+00 2.86023051e-01
-1.89094198e+00 -1.18582630e+00 -2.94387937e-01 1.87523544e+00
1.09882665e+00 4.07910854e-01 -2.66409814e-01 4.00142789e-01
3.37593853e-01 8.54758769e-02 -3.16289008e-01 -6.41764581e-01
-9.17233229e-02 5.92228115e-01 -1.62073210e-01 7.32411623e-01
-8.80857408e-01 1.16154349e+00 8.04996395e+00 5.18604636e-01
-6.25723898e-01 -1.35788590e-01 5.99978566e-01 2.50848513e-02
-5.81734180e-01 1.90215498e-01 -6.57470822e-01 4.96863462e-02
7.29500055e-01 -1.35248005e-01 5.93130589e-01 8.84149253e-01
-1.82569902e-02 -3.79547626e-01 -1.96149242e+00 4.71946567e-01
1.11205034e-01 -1.15592229e+00 1.84601694e-01 -2.25424424e-01
6.73249006e-01 -7.02197909e-01 8.99505243e-02 6.25035763e-01
5.85985601e-01 -1.53223026e+00 8.91608238e-01 4.92835730e-01
5.39347589e-01 -3.82970661e-01 5.97526431e-01 3.36661518e-01
-8.07075083e-01 -1.15400836e-01 -3.45647693e-01 -1.00689471e+00
-4.22336459e-02 5.90830564e-01 -4.21747833e-01 1.28320888e-01
4.26272988e-01 5.66904247e-01 -6.47672892e-01 7.12249041e-01
-9.95122552e-01 1.85910329e-01 -7.61864260e-02 -3.91518384e-01
-1.65870696e-01 -5.62691092e-02 -2.11234726e-02 1.05377424e+00
1.56005502e-01 2.64407128e-01 -1.92584962e-01 1.40693820e+00
8.06351751e-03 -3.59525770e-01 -7.10000336e-01 -6.75306171e-02
7.70883322e-01 7.49657035e-01 -4.95132118e-01 -5.02954245e-01
-2.47410983e-01 4.53293025e-01 6.55569732e-01 3.93003553e-01
-1.13917160e+00 -1.98189795e-01 7.03656912e-01 -1.83071829e-02
-3.42991680e-01 -3.49443436e-01 -1.09644902e+00 -1.35280204e+00
-9.30522289e-03 -1.03624809e+00 2.99284816e-01 -1.10230434e+00
-1.33878565e+00 -2.00948745e-01 4.56148773e-01 -6.55900538e-01
-2.97785729e-01 -1.10222936e+00 -5.88166296e-01 7.80914187e-01
-1.56658769e+00 -9.82748508e-01 -5.15940905e-01 3.72130066e-01
2.92414010e-01 1.14993028e-01 8.88675570e-01 -2.27749377e-01
-1.85157418e-01 2.95909435e-01 -4.94108498e-01 1.42014831e-01
6.89746201e-01 -1.38578987e+00 3.69750589e-01 6.01177573e-01
-2.22334549e-01 1.16182899e+00 5.82144260e-01 -5.53129792e-01
-1.44582152e+00 -6.98950171e-01 1.52492654e+00 -8.18600118e-01
9.33682263e-01 -3.59197915e-01 -7.65486538e-01 1.19361305e+00
-7.58962557e-02 -3.59020084e-01 5.70719779e-01 6.96218610e-01
-6.02673709e-01 1.50060683e-01 -1.11488724e+00 8.43945205e-01
1.42309916e+00 -6.78931057e-01 -1.30559719e+00 4.55676377e-01
5.56368649e-01 -5.21467566e-01 -8.06629777e-01 3.81533653e-01
7.21865952e-01 -1.05263090e+00 1.32134473e+00 -9.54808652e-01
1.06575739e+00 -1.46480441e-01 1.23572710e-03 -1.15689385e+00
-5.55132151e-01 -1.85112283e-01 -3.35648447e-01 9.02272284e-01
2.98118234e-01 -7.08715618e-01 5.46763420e-01 9.48940396e-01
4.40467857e-02 -6.38520241e-01 -4.46171343e-01 -8.12462807e-01
6.66448951e-01 -7.80341387e-01 7.19327748e-01 1.31411684e+00
6.42533183e-01 3.46871853e-01 1.23484895e-01 -1.74189761e-01
7.48164356e-01 3.40439290e-01 6.33660257e-01 -1.54275453e+00
-4.16667126e-02 -7.94824600e-01 -1.12760760e-01 -1.00807261e+00
7.29447484e-01 -1.20851636e+00 -1.52871430e-01 -1.74017274e+00
2.23280475e-01 -6.94997072e-01 -1.12888731e-01 3.78452033e-01
-2.41673112e-01 -1.80472329e-01 2.76125610e-01 -1.74597859e-01
-6.90952659e-01 1.93359703e-01 1.31338036e+00 -2.02736914e-01
2.17748970e-01 -2.66291350e-01 -1.03959846e+00 9.86936271e-01
1.01585937e+00 -2.11482748e-01 -4.62402165e-01 -6.38039172e-01
8.45248520e-01 -2.73837715e-01 5.71457922e-01 -1.03579843e+00
2.23110959e-01 -7.31542110e-01 6.50455773e-01 -3.56053472e-01
2.52804101e-01 -7.08530009e-01 -2.18292654e-01 5.31192005e-01
-6.94205165e-01 9.30926725e-02 1.62657067e-01 5.30524831e-03
-3.85308191e-02 -3.56628776e-01 6.32238209e-01 -3.77318859e-01
-6.94641232e-01 -4.83510882e-01 -3.40059698e-01 3.27239633e-01
9.91428256e-01 1.28312223e-02 -7.69066811e-01 -2.95637846e-01
-6.71575665e-01 5.44840097e-01 3.12469095e-01 8.31744522e-02
6.71595633e-01 -1.28490102e+00 -5.79626262e-01 -8.06969479e-02
-9.26300064e-02 2.26167873e-01 -8.16888884e-02 8.64369035e-01
-6.02292836e-01 7.39044487e-01 -3.21959376e-01 -2.00401306e-01
-9.46836531e-01 6.98968112e-01 2.25416079e-01 -4.44618732e-01
-1.55034825e-01 8.53749692e-01 2.79173732e-01 -5.05676687e-01
4.10555720e-01 -9.14325118e-01 -5.21291345e-02 -3.91092077e-02
4.79708165e-01 4.96790111e-01 -2.36181244e-01 1.40154719e-01
-4.45279568e-01 5.63129127e-01 2.40023777e-01 6.79985955e-02
1.12257910e+00 2.35798389e-01 -6.73085093e-01 6.76125526e-01
3.79200369e-01 -3.11096400e-01 -5.45013368e-01 -1.47250384e-01
2.25350067e-01 -2.12056175e-01 -3.25346112e-01 -1.03266454e+00
-2.01108620e-01 1.20781159e+00 -3.31374466e-01 1.99708790e-01
9.75422263e-01 2.73474716e-02 1.77587181e-01 1.00038517e+00
4.76909578e-01 -1.14503539e+00 8.23484138e-02 1.01078534e+00
9.80699599e-01 -1.36998570e+00 5.14349461e-01 -6.96720541e-01
-3.52338314e-01 1.30562711e+00 8.06922495e-01 -1.04515284e-01
5.38009107e-01 4.13424134e-01 -3.90895128e-01 -3.40584815e-01
-9.86432254e-01 -1.66855663e-01 3.68689932e-02 6.05177462e-01
9.65578556e-01 1.05090469e-01 -4.86644715e-01 6.77988172e-01
-5.81779122e-01 4.11549807e-01 4.38708037e-01 1.07502222e+00
-4.59456116e-01 -9.03278887e-01 -6.28432930e-01 3.60679746e-01
-5.73736541e-02 -3.96757931e-01 -6.20382071e-01 9.85992134e-01
2.47275710e-01 8.30515921e-01 2.01940045e-01 -1.08212046e-01
1.22454502e-02 4.45399612e-01 1.18612230e+00 -5.36898017e-01
-6.26771867e-01 -7.45648265e-01 3.57232600e-01 -4.40524429e-01
-3.10272992e-01 -5.74369788e-01 -1.28386319e+00 -8.94232571e-01
8.49517807e-02 -1.99727342e-02 3.19633216e-01 1.11805475e+00
-1.85684696e-01 7.05410600e-01 -2.45232597e-01 -5.80271423e-01
-8.26027334e-01 -8.77468228e-01 -3.38639885e-01 3.70589495e-01
-1.05704218e-02 -7.11361349e-01 -4.12314773e-01 1.29221261e-01] | [9.416400909423828, 7.257997989654541] |
1724840a-c545-4cb3-935e-65a9b8257bba | beyond-statistical-similarity-rethinking | 2302.02913 | null | https://arxiv.org/abs/2302.02913v3 | https://arxiv.org/pdf/2302.02913v3.pdf | Beyond Statistical Similarity: Rethinking Metrics for Deep Generative Models in Engineering Design | Deep generative models, such as Variational Autoencoders (VAEs), Generative Adversarial Networks (GANs), Diffusion Models, and Transformers, have shown great promise in a variety of applications, including image and speech synthesis, natural language processing, and drug discovery. However, when applied to engineering design problems, evaluating the performance of these models can be challenging, as traditional statistical metrics based on likelihood may not fully capture the requirements of engineering applications. This paper doubles as a review and a practical guide to evaluation metrics for deep generative models (DGMs) in engineering design. We first summarize well-accepted `classic' evaluation metrics for deep generative models grounded in machine learning theory and typical computer science applications. Using case studies, we then highlight why these metrics seldom translate well to design problems but see frequent use due to the lack of established alternatives. Next, we curate a set of design-specific metrics which have been proposed across different research communities and can be used for evaluating deep generative models. These metrics focus on unique requirements in design and engineering, such as constraint satisfaction, functional performance, novelty, and conditioning. We structure our review and discussion as a set of practical selection criteria and usage guidelines. Throughout our discussion, we apply the metrics to models trained on simple 2-dimensional example problems. Finally, to illustrate the selection process and classic usage of the presented metrics, we evaluate three deep generative models on a multifaceted bicycle frame design problem considering performance target achievement, design novelty, and geometric constraints. We publicly release the code for the datasets, models, and metrics used throughout the paper at decode.mit.edu/projects/metrics/. | ['Faez Ahmed', 'Dan Gutfreund', 'Akash Srivastava', 'Lyle Regenwetter'] | 2023-02-06 | null | null | null | null | ['speech-synthesis'] | ['speech'] | [ 1.09597392e-01 -7.17855478e-03 -1.21663019e-01 1.21939415e-02
-4.77004915e-01 -6.16173029e-01 6.21743083e-01 -3.89384985e-01
2.20553741e-01 8.76197934e-01 1.14284150e-01 -2.94538945e-01
-5.44838071e-01 -7.25212514e-01 -5.12697935e-01 -7.38700509e-01
1.83076844e-01 2.92457551e-01 -2.68287599e-01 -1.62636355e-01
1.09235235e-01 6.54658079e-01 -1.38316703e+00 3.58434208e-02
6.83910191e-01 1.04229391e+00 -7.68357813e-02 4.45852965e-01
2.47003406e-01 5.63373744e-01 -8.32848012e-01 -4.80410010e-01
-6.96242154e-02 -8.07769060e-01 -4.77194190e-01 1.42323785e-02
-1.95245042e-01 -1.16667025e-01 -2.83934534e-01 6.62902534e-01
9.81716871e-01 -1.60171855e-02 1.18189681e+00 -1.35226142e+00
-1.15129149e+00 4.27940130e-01 4.62439544e-02 -9.82328877e-02
3.42204422e-01 3.46078962e-01 1.02331018e+00 -9.83133435e-01
4.91601646e-01 9.24143076e-01 7.47400522e-01 7.95022309e-01
-1.15986443e+00 -4.24917847e-01 -7.26113841e-02 4.89203222e-02
-1.41610777e+00 -4.35763568e-01 9.35040891e-01 -7.26307094e-01
1.02151680e+00 3.47752869e-01 6.77760899e-01 1.34787476e+00
5.51489115e-01 7.27926016e-01 8.38106871e-01 -3.19506168e-01
5.72836339e-01 6.63244724e-02 -2.58645028e-01 4.63289648e-01
1.30999073e-01 4.99820173e-01 -3.46384943e-01 -1.13801651e-01
9.71222937e-01 -1.70954540e-01 -1.70773089e-01 -4.33343083e-01
-8.31516385e-01 1.04349709e+00 1.21761322e-01 4.70373839e-01
-4.02939528e-01 5.54598033e-01 1.66845396e-01 2.31475055e-01
2.15287149e-01 6.52642548e-01 -2.01457903e-01 -2.89486915e-01
-8.15114260e-01 5.17290533e-01 7.21942782e-01 1.15653765e+00
3.44585419e-01 7.23742247e-01 -2.46664494e-01 9.24280584e-01
5.23345709e-01 4.05417860e-01 2.04646587e-01 -8.91966939e-01
-2.27473184e-01 5.19280508e-02 -5.32761700e-02 -9.19141114e-01
-3.33703458e-01 -6.91414952e-01 -8.50698113e-01 2.17875823e-01
-4.88729142e-02 -3.99664015e-01 -8.93021822e-01 1.66981912e+00
5.81642240e-02 -2.80618042e-01 -1.59513444e-01 7.75289595e-01
1.15656650e+00 4.48632747e-01 -8.86065662e-02 -2.82830417e-01
9.96505201e-01 -5.70625007e-01 -7.29422569e-01 3.15064117e-02
1.81849003e-01 -1.03361356e+00 8.91300380e-01 4.41789150e-01
-1.18972325e+00 -5.42805493e-01 -1.08018136e+00 2.65545338e-01
-3.73780102e-01 8.14964026e-02 7.42614985e-01 1.18337047e+00
-9.77557540e-01 6.62253678e-01 -7.34567404e-01 -1.79638892e-01
4.56119299e-01 2.87622660e-01 3.44192952e-01 9.68603864e-02
-1.14694929e+00 9.09808338e-01 -1.82204098e-01 1.02387629e-01
-1.41942811e+00 -9.30774868e-01 -7.48623550e-01 -3.38616110e-02
3.02753210e-01 -1.05257142e+00 1.15136003e+00 -5.01965582e-01
-1.84749603e+00 4.30872381e-01 3.42138678e-01 -1.33622408e-01
4.34369475e-01 -2.93561727e-01 -4.74409014e-01 -4.78622705e-01
-3.09450448e-01 3.92812937e-01 7.10534811e-01 -1.26254225e+00
-9.27466005e-02 1.22140795e-01 1.82921037e-01 -3.01713318e-01
-6.54948950e-02 -8.36603791e-02 -7.39235431e-02 -9.46668267e-01
-3.39278847e-01 -9.21406090e-01 -3.68973464e-01 -2.20750645e-01
-6.97571099e-01 -1.18120819e-01 6.46902263e-01 -3.36120307e-01
1.66221571e+00 -1.93492389e+00 4.82726902e-01 2.86245286e-01
8.27108696e-02 1.40018657e-01 9.03667882e-02 8.12781036e-01
-1.42332926e-01 1.96878538e-01 -4.11364526e-01 -4.30850834e-02
3.99519682e-01 1.05474554e-01 -9.95754078e-02 4.55995530e-01
3.46479058e-01 1.32978022e+00 -4.99537706e-01 2.21010707e-02
4.13951755e-01 7.26164103e-01 -7.87690878e-01 5.68863870e-05
-4.02077228e-01 3.70119691e-01 -5.17004073e-01 7.14334011e-01
1.34095863e-01 -1.71570599e-01 1.21341668e-01 -2.86935389e-01
1.61467895e-01 8.73273835e-02 -1.11073184e+00 1.47576809e+00
-6.69321120e-01 6.09193921e-01 -3.20397466e-01 -9.99590397e-01
9.62738931e-01 3.50769132e-01 6.66612923e-01 -5.74269533e-01
3.25464070e-01 2.84532934e-01 1.56581745e-01 -3.32389176e-01
1.93615377e-01 -2.95835286e-01 -3.04770738e-01 4.86120164e-01
1.13780767e-01 -5.35840631e-01 1.30647436e-01 -2.00340644e-01
1.02468419e+00 2.31118023e-01 1.46792337e-01 -2.92059809e-01
3.59320492e-01 -3.37478042e-01 3.49191815e-01 3.60569537e-01
1.40475288e-01 7.40158916e-01 5.83861887e-01 -2.29461759e-01
-1.32865334e+00 -1.05774498e+00 -1.62270412e-01 6.31971419e-01
-2.29096666e-01 -5.45750916e-01 -7.23345935e-01 -2.14757994e-01
-6.62011513e-03 8.83978724e-01 -7.49436200e-01 -4.42624450e-01
-3.81674200e-01 -9.75018442e-01 4.70460445e-01 7.07771599e-01
7.36657977e-02 -9.32114482e-01 -6.06915891e-01 3.49248677e-01
3.55502307e-01 -7.97867835e-01 -2.85314500e-01 2.79909838e-02
-5.39095640e-01 -9.04237866e-01 -8.89268339e-01 -5.42045832e-01
2.80993164e-01 -4.15527940e-01 1.25622964e+00 -2.59045437e-02
-5.79620719e-01 6.67795956e-01 -2.06646487e-01 -4.00652587e-01
-7.40577459e-01 -6.21275492e-02 1.11608423e-01 -3.14646333e-01
1.73339415e-02 -7.84104407e-01 -6.99625611e-01 5.61832905e-01
-9.66279745e-01 -1.46275371e-01 5.52759230e-01 1.01340222e+00
5.90263128e-01 -4.40791585e-02 8.77204835e-01 -5.50552666e-01
9.75653708e-01 -3.75437975e-01 -4.22389537e-01 1.85238495e-01
-8.55551541e-01 3.96564342e-02 3.27196598e-01 -4.30550635e-01
-7.34766245e-01 -3.35834652e-01 -5.37446797e-01 -4.89318430e-01
4.63813879e-02 7.38895595e-01 -4.00409549e-01 -2.30584876e-03
7.51974404e-01 8.31014141e-02 4.59634745e-03 -3.54565769e-01
5.08572698e-01 4.08992141e-01 1.25525564e-01 -7.73593903e-01
6.76103532e-01 1.42317349e-02 1.92908332e-01 -8.33074629e-01
-2.17831597e-01 2.51903057e-01 -1.65549517e-01 -3.52947295e-01
6.86337411e-01 -4.55280423e-01 -7.74430811e-01 2.62733281e-01
-1.02758551e+00 -2.49194294e-01 -6.60631716e-01 2.85407007e-01
-1.10144484e+00 2.70714555e-02 -5.64365268e-01 -8.19218636e-01
-2.99987525e-01 -1.50531256e+00 9.00879443e-01 1.92906797e-01
-5.22691905e-01 -1.23723483e+00 3.70490067e-02 -7.31427073e-02
7.61902809e-01 7.05767393e-01 1.23758698e+00 -2.30209514e-01
-5.69446325e-01 -1.31879300e-01 4.51991051e-01 5.38032949e-01
2.13869840e-01 3.41229200e-01 -8.30767393e-01 -5.76888397e-02
-2.45869830e-02 -6.08841889e-02 3.96759659e-01 9.15042102e-01
1.00591183e+00 -1.60507113e-01 -2.27452740e-01 4.98647362e-01
1.41379678e+00 6.23944044e-01 8.99304807e-01 -1.10296952e-02
5.60509801e-01 4.48654234e-01 2.05601051e-01 6.15816772e-01
-2.18075305e-01 7.79555261e-01 3.40604514e-01 -3.73511552e-03
-2.55308956e-01 -4.93632965e-02 3.45757723e-01 7.41348863e-01
-3.63461554e-01 -5.39139152e-01 -6.19334996e-01 3.12332690e-01
-1.57688129e+00 -8.07841957e-01 9.20542330e-02 2.00123429e+00
5.77659905e-01 1.18639663e-01 3.26208115e-01 2.00715825e-01
5.54491043e-01 -9.20420140e-02 -6.15157902e-01 -5.71917534e-01
-5.69177791e-02 6.29051208e-01 2.04745904e-01 2.80781537e-01
-7.09143102e-01 5.86490035e-01 7.56061172e+00 1.11800528e+00
-1.10305548e+00 1.90075308e-01 8.38613451e-01 -2.61082292e-01
-7.82965660e-01 -1.70149088e-01 -5.58023214e-01 3.05521846e-01
1.00121868e+00 -3.58448744e-01 3.38100672e-01 8.74652624e-01
1.88576415e-01 3.23017597e-01 -1.45695233e+00 9.54938054e-01
-1.54776707e-01 -1.66185737e+00 -7.72647262e-02 3.36974561e-01
9.43027079e-01 -4.19892907e-01 6.31260395e-01 1.55652285e-01
1.90341696e-01 -1.44353831e+00 8.69665444e-01 5.31940877e-01
1.03507411e+00 -9.12583232e-01 4.47457105e-01 -1.47328079e-01
-8.58092427e-01 1.05109503e-02 -8.49689767e-02 1.03636928e-01
4.89581347e-01 6.99130952e-01 -3.90375376e-01 6.17289722e-01
3.16877127e-01 5.92534661e-01 2.41120812e-02 9.14270282e-01
-2.09453717e-01 5.19997716e-01 -2.59350896e-01 -4.08439726e-01
1.44683868e-01 -1.08192042e-01 5.68231642e-01 9.37406063e-01
5.41959584e-01 -8.26523006e-02 -3.41277063e-01 1.64996159e+00
2.29004592e-01 -5.71169294e-02 -5.79838932e-01 -3.93460035e-01
3.97447050e-01 7.95608997e-01 -4.61038738e-01 1.14675783e-01
-3.33661109e-01 4.56290334e-01 -6.14096045e-01 5.81351757e-01
-1.42977667e+00 -2.64829993e-01 1.01242769e+00 2.22207516e-01
1.58144295e-01 -3.51266742e-01 -4.51372653e-01 -6.02187276e-01
-1.68106169e-01 -1.08109343e+00 -7.74661377e-02 -6.34289861e-01
-1.16044545e+00 5.56333184e-01 2.93803781e-01 -1.29510128e+00
-6.79459453e-01 -7.99167752e-01 -4.73145485e-01 7.81731129e-01
-8.79029512e-01 -1.03702235e+00 -2.93028336e-02 8.08831006e-02
6.18725002e-01 -4.88325477e-01 7.77023792e-01 5.76066911e-01
-7.39191532e-01 6.63193047e-01 1.72601759e-01 -2.13553265e-01
1.22563548e-01 -8.84199977e-01 6.08691514e-01 5.94385564e-01
6.10223301e-02 6.39802873e-01 9.57865536e-01 -4.67843145e-01
-1.58220518e+00 -7.69935548e-01 2.12185025e-01 -3.62514943e-01
4.80543137e-01 -2.38639995e-01 -5.06512642e-01 4.35161024e-01
2.27367222e-01 -5.03025532e-01 8.01067591e-01 3.24423588e-03
1.95476532e-01 1.22034132e-01 -9.99905527e-01 6.53754890e-01
1.16202509e+00 -3.06107193e-01 -9.60481390e-02 3.22182655e-01
5.57507038e-01 -4.92454797e-01 -1.36760080e+00 5.40085495e-01
7.34858096e-01 -9.30379927e-01 1.12123251e+00 -6.33417904e-01
6.86496198e-01 -2.03830197e-01 -2.79295921e-01 -1.31799066e+00
-5.55570841e-01 -9.61552799e-01 -2.51672059e-01 1.21257210e+00
5.57012022e-01 -3.82032603e-01 6.57203078e-01 6.64162815e-01
-5.02316117e-01 -1.33222508e+00 -8.36107552e-01 -8.20622861e-01
3.25515002e-01 -6.94317997e-01 8.16788733e-01 6.43857002e-01
-2.84490526e-01 4.20556068e-01 -4.77758795e-01 -3.14627558e-01
3.00635070e-01 -6.84032440e-02 6.58409476e-01 -9.32907760e-01
-4.25505966e-01 -9.27129805e-01 -5.93440115e-01 -9.10413563e-01
-2.34160095e-01 -6.43134892e-01 -1.63456336e-01 -1.63427007e+00
-1.31236494e-01 -3.54233682e-01 4.31352807e-03 1.04781143e-01
2.93110818e-01 1.43159315e-01 9.36424360e-03 -1.32206842e-01
8.42727870e-02 9.03098464e-01 1.29282606e+00 -2.55573720e-01
-6.58653527e-02 1.10060208e-01 -8.18567872e-01 4.37408268e-01
6.18965268e-01 -8.78492370e-02 -6.04340494e-01 -1.30646601e-01
4.37898338e-01 -9.05394033e-02 3.98355186e-01 -1.05765212e+00
-1.83102280e-01 -3.01700652e-01 3.61238092e-01 -1.94195941e-01
3.64961505e-01 -6.93639815e-01 9.30088043e-01 5.41295230e-01
-9.78862941e-02 6.39270693e-02 2.85001248e-01 3.21439594e-01
-3.45515050e-02 -2.41778255e-01 8.43991756e-01 1.38714880e-01
-5.41208982e-01 3.00318509e-01 -4.52593148e-01 -1.30213737e-01
1.05135810e+00 -3.45426202e-01 -7.07616806e-02 -5.81583381e-01
-9.61818278e-01 -1.09154105e-01 3.43047470e-01 5.72559476e-01
8.17742646e-01 -1.61758184e+00 -7.10286200e-01 7.72672668e-02
-8.95574316e-02 -1.81875393e-01 3.03364515e-01 6.82784617e-01
-6.19779408e-01 5.88813305e-01 -2.08525747e-01 -6.44986808e-01
-7.25366831e-01 6.46774113e-01 6.27638876e-01 -5.44781722e-02
-2.52666593e-01 6.74848557e-01 4.26524401e-01 7.82519355e-02
-1.23840887e-02 -5.63512266e-01 1.42346725e-01 -1.91158131e-01
6.74221218e-02 4.00727630e-01 1.94310263e-01 -3.41259390e-01
-4.16995943e-01 6.65178657e-01 3.90500367e-01 -1.13049923e-02
1.44500887e+00 8.88395011e-02 2.36844987e-01 2.45939329e-01
9.66747999e-01 -3.16836119e-01 -1.03297770e+00 1.79383278e-01
-3.36517692e-01 1.67212337e-02 6.40151650e-02 -6.98473275e-01
-1.39886999e+00 8.21466148e-01 6.60321772e-01 3.17478865e-01
1.13736618e+00 1.17762983e-01 6.67376220e-01 -1.75906628e-01
2.76271641e-01 -1.01512051e+00 3.59603941e-01 3.66624117e-01
1.23456931e+00 -4.89849895e-01 2.98174494e-03 -3.40976536e-01
-4.72310871e-01 8.58854353e-01 3.78426909e-01 -9.47468206e-02
1.03036869e+00 5.07547081e-01 -4.18523997e-01 -3.06667477e-01
-6.25498235e-01 -1.96313597e-02 5.28472245e-01 7.67128646e-01
7.80887008e-01 1.07704513e-02 -4.20025021e-01 7.80576050e-01
-3.85568678e-01 -4.01223190e-02 2.58532852e-01 8.98005247e-01
-2.24061221e-01 -1.28218055e+00 -2.50023156e-01 4.03520405e-01
-3.49356681e-01 -8.03899914e-02 -2.68522024e-01 7.81342924e-01
1.46656990e-01 7.97517240e-01 -1.70251563e-01 -6.13041222e-01
6.10724628e-01 -9.38039795e-02 8.20364654e-01 -5.02464771e-01
-4.36253250e-01 1.95795089e-01 6.44627884e-02 -2.57694036e-01
-4.32612240e-01 -5.54306686e-01 -7.93205917e-01 -6.43993735e-01
-4.71618116e-01 -2.17940356e-03 7.49489307e-01 8.95953894e-01
4.35622066e-01 1.12073123e+00 4.23487872e-01 -6.68899894e-01
-3.50824922e-01 -8.64521444e-01 -5.15759230e-01 -1.36120860e-02
-1.83727499e-02 -1.06510282e+00 -2.37885490e-02 1.20523691e-01] | [5.833241939544678, 3.3213672637939453] |
5532328a-d07b-4274-8d44-e036aa0202c6 | nowcasting-the-2022-mpox-outbreak-in-england | 2302.09076 | null | https://arxiv.org/abs/2302.09076v1 | https://arxiv.org/pdf/2302.09076v1.pdf | Nowcasting the 2022 mpox outbreak in England | In May 2022, a cluster of mpox cases were detected in the UK that could not be traced to recent travel history from an endemic region. Over the coming months, the outbreak grew, with over 3000 total cases reported in the UK, and similar outbreaks occurring worldwide. These outbreaks appeared linked to sexual contact networks between gay, bisexual and other men who have sex with men. Following the COVID-19 pandemic, local health systems were strained, and therefore effective surveillance for mpox was essential for managing public health policy. However, the mpox outbreak in the UK was characterised by substantial delays in the reporting of the symptom onset date and specimen collection date for confirmed positive cases. These delays led to substantial backfilling in the epidemic curve, making it challenging to interpret the epidemic trajectory in real-time. Many nowcasting models exist to tackle this challenge in epidemiological data, but these lacked sufficient flexibility. We have developed a novel nowcasting model using generalised additive models to correct the mpox epidemic curve in England, and provide real-time characteristics of the state of the epidemic, including the real-time growth rate. This model benefited from close collaboration with individuals involved in collecting and processing the data, enabling temporal changes in the reporting structure to be built into the model, which improved the robustness of the nowcasts generated. | ['Thomas Ward', 'Charlie Turner', 'Rob Paton', 'Owen Jones', 'Julie Day', 'Fergus Cumming', 'Rachel Christie', 'Sam Abbott', 'Christopher E. Overton'] | 2023-02-17 | null | null | null | null | ['additive-models'] | ['methodology'] | [ 1.93856955e-01 4.65963706e-02 5.29034296e-03 -1.75670177e-01
-4.42614913e-01 -6.05681539e-01 9.55768287e-01 8.48001063e-01
-7.04583049e-01 7.27136433e-01 6.32241666e-01 -4.97895598e-01
-6.21421874e-01 -9.37215686e-01 -3.16246986e-01 -5.00812471e-01
-8.89794707e-01 8.63145709e-01 2.50324845e-01 -3.02476227e-01
-1.35154977e-01 2.98880160e-01 -7.79632628e-01 4.08870913e-02
6.16413116e-01 2.18899220e-01 2.72266567e-01 9.32159901e-01
8.03874135e-02 3.75568122e-01 -7.58216619e-01 -2.53851473e-01
1.69177413e-01 -2.70483404e-01 -3.39971185e-01 -3.95300359e-01
-5.43468118e-01 -5.42024970e-01 -3.07670116e-01 3.14796090e-01
2.81396091e-01 -7.22210035e-02 5.65362096e-01 -1.30993974e+00
-1.24813192e-01 4.03587781e-02 -5.35727978e-01 6.09203339e-01
6.10446036e-01 2.85304964e-01 2.00593308e-01 -1.84254095e-01
1.09982026e+00 1.10909188e+00 1.25461864e+00 1.28337011e-01
-1.25232232e+00 -6.43129945e-01 -1.70704901e-01 -2.53355056e-01
-1.23566568e+00 -2.18485236e-01 1.69219859e-02 -4.37168926e-01
1.12479854e+00 6.11976802e-01 1.06785262e+00 1.00321651e+00
5.29646873e-01 -3.41955543e-01 1.07272410e+00 1.71396390e-01
-1.65547773e-01 -2.25706324e-02 -4.91873622e-01 7.85601735e-02
2.75207192e-01 4.43094224e-01 -1.51860505e-01 -8.86788130e-01
6.87286019e-01 5.80204606e-01 -1.72696427e-01 2.20959723e-01
-1.16533542e+00 9.84387875e-01 5.11885226e-01 1.26856253e-01
-8.18766296e-01 -7.40218610e-02 4.71930057e-01 2.73835033e-01
8.43104482e-01 -2.75635272e-02 -4.83617812e-01 -4.31539059e-01
-9.20538366e-01 4.15774316e-01 5.70602477e-01 2.62745410e-01
1.16328925e-01 -3.46406221e-01 2.69089937e-01 4.88515556e-01
4.52731818e-01 1.09079742e+00 -4.07615274e-01 -9.24642920e-01
5.84140360e-01 5.07473528e-01 3.77186179e-01 -1.48777270e+00
-8.44853997e-01 -8.38116258e-02 -8.47961724e-01 -2.58977085e-01
6.67454004e-01 -6.01365030e-01 -9.70102966e-01 1.52432680e+00
4.34919953e-01 3.23303938e-01 -6.22978905e-06 6.32501304e-01
-1.74530260e-02 1.18611288e+00 3.34722966e-01 -5.13895988e-01
1.26599586e+00 2.41303563e-01 -7.51710832e-01 -4.96246777e-02
5.86133063e-01 -8.27945828e-01 -1.67718425e-01 -4.33183797e-02
-9.40148175e-01 1.20120890e-01 -4.76578653e-01 8.48840237e-01
-4.67942446e-01 -9.74168479e-01 6.05666578e-01 6.03068888e-01
-1.13766181e+00 4.22039449e-01 -1.08415806e+00 -1.23023510e+00
1.50357574e-01 3.89328063e-01 -3.69714797e-01 -2.48792470e-01
-1.50681829e+00 8.57424021e-01 2.37463549e-01 4.68499959e-01
-2.76359797e-01 -9.30479050e-01 -7.81487226e-01 -2.72275388e-01
1.62129968e-01 -5.17841280e-01 8.53602648e-01 -1.95714742e-01
-3.82909358e-01 4.45365906e-01 -4.85319309e-02 -3.74956071e-01
5.65444291e-01 5.01525939e-01 -1.00453055e+00 3.16542871e-02
2.64583379e-01 2.11657569e-01 1.50270388e-01 -9.41350520e-01
-9.60073054e-01 -6.02159619e-01 -3.81227493e-01 5.15076444e-02
5.47507942e-01 6.44560754e-01 -2.31561184e-01 -5.96676171e-01
-4.81884658e-01 -8.24601889e-01 -5.57545543e-01 -4.51640844e-01
3.43932360e-01 2.03858495e-01 6.82142496e-01 -1.17938638e+00
1.10996783e+00 -1.97183204e+00 -4.91388112e-01 4.88326699e-01
8.64872113e-02 3.31362277e-01 1.22993156e-01 1.36330366e+00
4.04710323e-01 -2.32745428e-02 -3.20508748e-01 1.36543840e-01
-1.95633098e-01 6.86362505e-01 -3.65782380e-01 8.38149607e-01
2.80996859e-01 6.10313237e-01 -1.48521292e+00 -8.48021582e-02
1.28294855e-01 7.27067292e-01 -3.47548068e-01 1.90804020e-01
2.27821469e-01 5.15608132e-01 -1.06520392e-01 2.80859590e-01
8.84358287e-01 -1.89183634e-02 4.74524617e-01 8.41012180e-01
-3.95576000e-01 8.83757770e-02 -9.30782735e-01 6.64987504e-01
-2.29463652e-01 4.04933274e-01 7.35900104e-01 -4.57711846e-01
6.51953518e-01 6.48447216e-01 7.20279932e-01 -6.84500754e-01
-2.05832303e-01 2.29278460e-01 1.47836700e-01 -4.59938169e-01
6.49881423e-01 -2.09045514e-01 -9.06153172e-02 7.99608588e-01
-5.59801817e-01 1.22596428e-01 1.55645072e-01 3.61756414e-01
1.09640360e+00 -2.64824182e-01 6.21196590e-02 -2.55409777e-01
-2.42784008e-01 5.81963122e-01 6.22483194e-01 5.42483985e-01
-7.37016127e-02 2.16541499e-01 4.79320258e-01 -5.37681639e-01
-9.94729936e-01 -1.26402390e+00 -3.62454683e-01 7.07058609e-01
-2.98018515e-01 -4.16376948e-01 -2.73352653e-01 -6.84598014e-02
3.94538008e-02 5.00484228e-01 -8.77541065e-01 5.81662767e-02
-7.36436903e-01 -1.38111687e+00 5.69908977e-01 1.11918353e-01
2.86450654e-01 -8.53146791e-01 -7.53331482e-01 8.30338001e-01
-1.74259529e-01 -6.73732579e-01 -3.63662392e-01 -2.47501656e-02
-6.59055889e-01 -1.38744783e+00 -1.11163223e+00 -3.08711201e-01
8.47884595e-01 -1.17119374e-02 6.59311831e-01 8.79999548e-02
-5.44054627e-01 5.07752657e-01 9.47903693e-02 -6.52695954e-01
-6.70398355e-01 -4.55530703e-01 1.26940355e-01 -3.42952818e-01
2.14486033e-01 5.61413318e-02 -8.32910597e-01 1.99356019e-01
-1.04742157e+00 -3.94129366e-01 9.14879143e-02 3.51762384e-01
-1.02558911e-01 1.79191809e-02 9.47863400e-01 -8.62635612e-01
8.43608379e-01 -1.19428957e+00 -3.91233653e-01 -1.59053609e-01
-4.83373910e-01 -6.49477243e-01 2.05634281e-01 -2.44855449e-01
-1.08500135e+00 -6.26723230e-01 -1.12455912e-01 4.39734727e-01
-3.83504815e-02 8.12369347e-01 8.65855694e-01 5.82791686e-01
1.94291085e-01 8.18811655e-02 4.45588738e-01 -4.21242833e-01
-1.31150141e-01 9.32096243e-01 4.40443814e-01 1.31959110e-01
6.26276553e-01 6.99375510e-01 -1.43542081e-01 -1.19601083e+00
2.18284041e-01 -8.26287150e-01 -1.36808693e-01 -2.50795484e-01
7.30994582e-01 -7.31717229e-01 -7.55576849e-01 6.70555353e-01
-1.15546608e+00 -4.89615500e-01 1.33365884e-01 5.88074148e-01
-1.38157699e-03 -9.67265442e-02 -6.48243427e-01 -1.06064856e+00
-1.16741378e-02 -6.46339953e-01 6.42507553e-01 -1.52118087e-01
-7.44250476e-01 -1.60791337e+00 9.21217382e-01 -6.34332076e-02
9.69876528e-01 6.90497041e-01 7.66468763e-01 -6.04305685e-01
-1.24092102e-01 -2.97963351e-01 -5.58496177e-01 -4.52421486e-01
5.10506392e-01 4.63981787e-03 -2.83204257e-01 -5.97839177e-01
-1.89146698e-01 3.37301761e-01 3.76029134e-01 3.22881699e-01
-1.40827432e-01 -7.25472927e-01 -9.35763299e-01 2.10584328e-01
1.10154688e+00 5.68965554e-01 3.25540245e-01 1.69524372e-01
7.54792942e-03 1.05097997e+00 3.74860168e-01 5.20783603e-01
5.78378320e-01 6.38291895e-01 2.37007856e-01 -2.64985323e-01
3.60222936e-01 -6.90015405e-02 2.07092658e-01 6.63047910e-01
-4.76610661e-01 -1.42920122e-01 -1.28698349e+00 8.73567283e-01
-1.49624312e+00 -1.16029227e+00 -5.72787821e-01 2.38771200e+00
6.65123582e-01 -9.20218602e-03 4.39322919e-01 -4.66531932e-01
5.44336498e-01 1.90679118e-01 3.83545309e-02 -7.93470800e-01
3.41147840e-01 -2.35088214e-01 9.29282188e-01 7.29022324e-01
-6.55855238e-01 1.57023445e-02 7.54169464e+00 1.42700896e-01
-1.00351965e+00 8.29051808e-02 4.90760803e-01 4.90988083e-02
-2.45285317e-01 -1.29766315e-02 -4.40343380e-01 6.73451185e-01
1.54803288e+00 -2.96327591e-01 3.29569340e-01 -2.35469505e-01
8.78043175e-01 -2.86347985e-01 -2.12013736e-01 2.53821820e-01
-1.51178226e-01 -1.34864438e+00 -5.87980807e-01 6.58986270e-01
6.73918605e-01 4.30734098e-01 -2.83709437e-01 -2.35554371e-02
4.98386025e-01 -8.58202875e-01 -9.46716368e-02 7.03032553e-01
7.34651089e-01 -1.03145266e+00 9.36918318e-01 3.82734507e-01
-1.10291183e+00 1.03566796e-01 6.77597076e-02 -1.55822381e-01
8.77197087e-01 2.89858401e-01 -1.35524356e+00 3.54148358e-01
6.72229767e-01 2.99476594e-01 -2.02825904e-01 9.23022389e-01
4.23313230e-01 6.95918083e-01 -7.09050059e-01 2.67191619e-01
5.90374172e-01 -2.97595441e-01 6.52309179e-01 1.17676961e+00
4.37594026e-01 1.47313863e-01 -1.95737332e-01 3.85360628e-01
3.26528192e-01 -3.69025230e-01 -8.49971414e-01 -1.60668939e-01
1.77108735e-01 8.19393158e-01 -8.14613461e-01 -4.41170074e-02
-4.13791984e-01 7.24019051e-01 -2.69265473e-01 5.80376983e-01
-7.39971936e-01 -4.21421081e-01 9.14022982e-01 8.75085473e-01
1.09945893e-01 -2.06676230e-01 5.68606794e-01 -3.84901255e-01
-3.49203110e-01 -5.03897071e-01 5.45236826e-01 -3.19823891e-01
-9.39875245e-01 4.57879126e-01 4.11406815e-01 -5.33237934e-01
-9.05317783e-01 -7.03179045e-03 -7.22625494e-01 1.03601742e+00
-1.03856921e+00 -9.53036249e-01 3.87370229e-01 2.00036630e-01
2.66219527e-02 4.19612736e-01 9.47920084e-01 2.04938143e-01
-1.32833317e-01 -1.73506990e-01 7.34204173e-01 -6.30230084e-02
6.10449255e-01 -9.74944174e-01 6.89000487e-01 3.63389522e-01
-6.31752789e-01 1.05822277e+00 8.19642663e-01 -1.43566322e+00
-1.09907770e+00 -1.35414433e+00 1.59128845e+00 -4.36578184e-01
1.08759522e+00 -5.09816110e-01 -7.77738333e-01 9.13294375e-01
4.28362340e-01 -5.10132372e-01 6.39850736e-01 -7.22975954e-02
1.96865216e-01 1.59968212e-01 -1.34531498e+00 4.06329453e-01
5.79008520e-01 -2.61212140e-01 -5.55089116e-01 4.13968682e-01
5.08501172e-01 1.28086498e-02 -9.08759534e-01 2.06446648e-01
5.41326582e-01 -5.35601497e-01 8.72694314e-01 -4.83798444e-01
-2.89208770e-01 8.43842253e-02 3.92115742e-01 -1.56316781e+00
-2.38832578e-01 -9.56976771e-01 4.10689563e-01 1.16547573e+00
2.89352804e-01 -1.15458739e+00 3.18351835e-01 6.27763033e-01
3.43520552e-01 -4.53814477e-01 -1.23826516e+00 -7.10000336e-01
-2.20032841e-01 -2.97693163e-01 6.35358810e-01 8.51233840e-01
-1.43794388e-01 -3.08040172e-01 -4.44495499e-01 3.54529947e-01
6.04589581e-01 -2.55988330e-01 4.28424448e-01 -1.19120646e+00
2.56446563e-02 1.05731308e-01 -2.86910266e-01 -2.18091056e-01
-9.12550271e-01 -5.17943203e-01 -2.98620015e-01 -1.73937261e+00
1.72285914e-01 -6.70524001e-01 2.05910802e-01 2.18360588e-01
3.99083458e-02 3.72147858e-01 1.73511833e-01 2.33659998e-01
-3.25537100e-02 -4.20500152e-02 8.56121123e-01 1.94106977e-02
-4.77622807e-01 2.03575104e-01 -2.64003485e-01 5.28762579e-01
8.41449201e-01 -7.68761277e-01 -3.08301061e-01 -3.37584823e-01
7.28601396e-01 3.89616728e-01 6.43004000e-01 -5.73752403e-01
2.10068688e-01 -3.62853199e-01 3.90786350e-01 -8.97625446e-01
3.72616142e-01 -1.01803291e+00 1.07890368e+00 1.11013436e+00
2.46627718e-01 5.41695356e-01 4.56657529e-01 7.09046662e-01
2.44631842e-01 1.23656094e-01 1.71827763e-01 -3.89219671e-02
-3.54783945e-02 1.68868884e-01 -1.09461081e+00 1.60457358e-01
1.15946031e+00 -1.38864368e-01 -6.24249220e-01 -4.60713655e-01
-6.75010204e-01 6.01039112e-01 5.78122020e-01 3.21696937e-01
3.58296245e-01 -9.59538460e-01 -1.11280262e+00 2.88316250e-01
-1.55410618e-01 -3.69850695e-01 4.62289423e-01 1.09608066e+00
-9.80404556e-01 8.85971129e-01 -1.81852043e-01 -2.01953381e-01
-9.15535808e-01 6.51411057e-01 2.66862772e-02 -4.98583585e-01
-4.91995931e-01 3.89039479e-02 -2.59678155e-01 -7.97967553e-01
-9.23274606e-02 -1.49329320e-01 -1.05872964e-02 4.31858689e-01
9.14028704e-01 6.83665872e-01 -2.73489475e-01 -1.03346670e+00
-6.71020269e-01 9.09342915e-02 1.50460824e-02 -3.05715412e-01
1.72165024e+00 -4.00986254e-01 -2.39991397e-02 4.07384992e-01
1.18053353e+00 2.09617447e-02 -1.35417783e+00 1.95875674e-01
-1.25290543e-01 -3.11781645e-01 -2.94704854e-01 -8.55049133e-01
-6.81571364e-01 2.85872579e-01 3.20074797e-01 6.82899773e-01
7.77324617e-01 2.05871537e-02 1.06158304e+00 -3.37918162e-01
3.53889197e-01 -6.39962077e-01 -8.84861767e-01 9.01730806e-02
8.03498566e-01 -8.39487970e-01 4.78217155e-02 -6.75250888e-02
-2.82978266e-01 6.75077736e-01 -3.62161726e-01 1.51807889e-01
7.61882424e-01 3.57132316e-01 2.09971189e-01 -5.42961478e-01
-7.98085690e-01 2.18054086e-01 -1.75392911e-01 1.31904364e+00
-4.02685367e-02 1.90775841e-01 -4.41076428e-01 1.36960715e-01
2.54469484e-01 1.76954150e-01 4.90197539e-01 1.09313548e+00
-3.31073254e-02 -9.68779147e-01 -8.12803566e-01 7.48304784e-01
-5.75641632e-01 -1.54767530e-02 -2.74109125e-01 1.18851924e+00
1.68334976e-01 9.35692251e-01 6.72607780e-01 2.08701134e-01
2.88801402e-01 -2.32261330e-01 4.94680740e-02 -1.50222838e-01
-7.95291245e-01 1.62721917e-01 2.87739635e-01 -4.05610025e-01
-3.78032327e-01 -9.32668984e-01 -1.15741134e+00 -9.79951084e-01
8.59022811e-02 4.73764479e-01 5.73203444e-01 6.21524513e-01
3.30396563e-01 2.36919880e-01 5.67909062e-01 -7.06033349e-01
-1.28960639e-01 -8.62751365e-01 -7.29326844e-01 2.07820624e-01
5.93365312e-01 -1.92767143e-01 -4.82875347e-01 -1.72172129e-01] | [5.954069137573242, 4.386684894561768] |
60e3c5be-b782-463d-be27-f6312ee47be1 | advancements-in-arabic-grammatical-error | 2305.14734 | null | https://arxiv.org/abs/2305.14734v1 | https://arxiv.org/pdf/2305.14734v1.pdf | Advancements in Arabic Grammatical Error Detection and Correction: An Empirical Investigation | Grammatical error correction (GEC) is a well-explored problem in English with many existing models and datasets. However, research on GEC in morphologically rich languages has been limited due to challenges such as data scarcity and language complexity. In this paper, we present the first results on Arabic GEC by using two newly developed Transformer-based pretrained sequence-to-sequence models. We address the task of multi-class Arabic grammatical error detection (GED) and present the first results on multi-class Arabic GED. We show that using GED information as auxiliary input in GEC models improves GEC performance across three datasets spanning different genres. Moreover, we also investigate the use of contextual morphological preprocessing in aiding GEC systems. Our models achieve state-of-the-art results on two Arabic GEC shared tasks datasets and establish a strong benchmark on a newly created dataset. | ['Nizar Habash', 'Christian Khairallah', 'Go Inoue', 'Bashar Alhafni'] | 2023-05-24 | null | null | null | null | ['grammatical-error-detection', 'grammatical-error-correction'] | ['natural-language-processing', 'natural-language-processing'] | [ 9.65224206e-02 -3.33042651e-01 4.73977685e-01 -4.61098403e-01
-1.00535500e+00 -7.09806442e-01 1.31753981e-01 5.58422387e-01
-7.84660578e-01 4.99957561e-01 6.38453737e-02 -4.29365098e-01
3.70666683e-01 -5.65874517e-01 -9.41366315e-01 -2.50832647e-01
-2.42208734e-01 8.29770863e-01 9.78557095e-02 -1.03305364e+00
5.40163159e-01 1.86513830e-02 -1.28104997e+00 7.67751813e-01
1.62543547e+00 4.55831856e-01 1.70732781e-01 6.70766532e-01
1.70675457e-01 2.83946276e-01 -8.31545055e-01 -1.15950882e+00
-1.77465767e-01 -6.32756948e-01 -8.61520767e-01 -2.21410736e-01
3.68163794e-01 8.01382735e-02 6.25568092e-01 1.07028306e+00
6.07498407e-01 9.09430254e-03 6.97308183e-01 -8.59784067e-01
-1.21888995e+00 1.08901107e+00 -3.35919738e-01 4.31173714e-03
4.86569703e-01 -9.83970761e-02 1.07198644e+00 -1.42113221e+00
9.31171536e-01 1.35464466e+00 1.07357979e+00 8.43670607e-01
-8.02602112e-01 -8.70746747e-02 1.79649353e-01 6.67165756e-01
-1.14580464e+00 -5.02324104e-01 4.07390118e-01 2.66981274e-01
1.70306432e+00 1.83827087e-01 2.15782255e-01 8.96944642e-01
-1.45394877e-01 1.02712119e+00 1.17579257e+00 -1.15669692e+00
-6.94956705e-02 -2.03642040e-01 1.96881890e-01 1.12575853e+00
4.59497385e-02 -3.34496647e-01 -5.24668396e-01 4.68914926e-01
-1.88157201e-01 -8.87207448e-01 -3.94958138e-01 6.56778991e-01
-9.53709006e-01 8.76636267e-01 1.62410751e-01 3.19384575e-01
1.56008989e-01 -2.43249424e-02 6.55845821e-01 7.14556336e-01
6.01850808e-01 3.51652354e-01 -7.40463614e-01 -4.87091064e-01
-7.03143358e-01 2.45324120e-01 9.76875663e-01 1.13906658e+00
1.65899277e-01 3.28451514e-01 2.32624352e-01 1.20324779e+00
4.21786696e-01 7.68313825e-01 4.60730731e-01 -3.87631536e-01
1.07899523e+00 5.49387455e-01 -1.93767726e-01 -9.86213923e-01
-5.13901651e-01 -2.32887894e-01 -5.02307177e-01 -1.52872577e-01
8.13028812e-01 4.28098552e-02 -6.73265576e-01 1.49416339e+00
-2.00615618e-02 -5.19707918e-01 6.92223236e-02 8.07389498e-01
9.08508658e-01 6.39797270e-01 1.57905906e-01 6.46404549e-02
1.26494777e+00 -1.08367908e+00 -8.10639977e-01 -4.33885604e-01
1.21050501e+00 -1.04559112e+00 1.56302142e+00 6.61980093e-01
-1.11605334e+00 -1.94805831e-01 -9.86041307e-01 -5.17263949e-01
-8.56987178e-01 7.91921437e-01 1.60786837e-01 1.11860740e+00
-8.11754525e-01 6.27277195e-01 -9.00439382e-01 -5.15105426e-01
1.01294951e-03 -2.48297770e-02 -4.08786982e-01 -4.55532044e-01
-1.16237617e+00 1.44003999e+00 6.97740078e-01 3.28419447e-01
-5.51467240e-01 -4.70110476e-01 -8.97214115e-01 -2.54335821e-01
1.43703669e-01 3.92555520e-02 1.06013989e+00 -7.42891550e-01
-1.34945524e+00 1.22919810e+00 -1.10500805e-01 -4.97284532e-01
2.51453221e-01 -5.38790643e-01 -8.96282673e-01 -1.98714435e-01
-2.53457308e-01 6.00158632e-01 6.01131678e-01 -1.23928368e+00
-6.88968241e-01 -4.90047455e-01 -3.35275084e-01 1.68052077e-01
-2.73847319e-02 8.23435485e-01 -2.89424688e-01 -1.09818137e+00
5.15905656e-02 -1.02370775e+00 2.79564261e-01 -5.32358229e-01
-1.63220599e-01 -4.32733119e-01 5.91039658e-01 -1.54178786e+00
1.72421646e+00 -1.87174916e+00 5.61140358e-01 -3.30167562e-02
-5.78275442e-01 5.25400639e-01 -5.83557189e-01 3.96118343e-01
4.04226109e-02 3.85019064e-01 -7.74607480e-01 -8.42392802e-01
1.90959543e-01 3.91719103e-01 -8.19589850e-03 2.06683427e-01
9.50769663e-01 1.08326983e+00 -8.94956112e-01 -2.18824446e-01
-3.54685068e-01 1.31528392e-01 -7.97591329e-01 -2.27351248e-01
-4.48426187e-01 1.89158455e-01 5.60380816e-01 1.24973905e+00
8.53592277e-01 2.36942142e-01 4.35888827e-01 1.39928728e-01
-1.48016764e-02 6.63652420e-01 -1.09700918e+00 1.76652920e+00
-5.31928957e-01 9.30232778e-02 -1.95874125e-01 -9.65258479e-01
7.57786155e-01 6.36004284e-02 -4.34349746e-01 -7.29028106e-01
3.30455273e-01 1.14095914e+00 3.68616670e-01 -9.91206467e-02
1.09733772e+00 2.54450720e-02 -3.74684870e-01 5.62219799e-01
5.21328628e-01 -2.15849072e-01 7.46707678e-01 9.83129144e-02
6.46304071e-01 3.93319726e-01 2.90412128e-01 -3.39048296e-01
5.39468765e-01 2.46916562e-01 3.99596214e-01 4.86987293e-01
-1.23039179e-01 7.43924081e-01 1.81642473e-01 -1.68416157e-01
-7.03275442e-01 -6.40099704e-01 3.92160565e-02 1.47639883e+00
-2.84736842e-01 -9.04999197e-01 -1.13814712e+00 -1.21978843e+00
-2.04092428e-01 9.12524641e-01 -5.42239428e-01 -6.55225152e-03
-1.28466320e+00 -1.37223256e+00 1.20825195e+00 4.70313430e-01
3.17786217e-01 -1.25676131e+00 -1.07015833e-01 6.50762141e-01
-7.51512170e-01 -1.16257691e+00 -5.08989930e-01 1.46807596e-01
-8.17796350e-01 -1.20175219e+00 -3.57856035e-01 -1.08723998e+00
4.29850757e-01 -3.11555505e-01 1.73019683e+00 6.03273392e-01
3.61807421e-02 5.91058284e-02 -1.12113535e+00 -8.39048207e-01
-9.18384492e-01 4.07609820e-01 -8.97603631e-02 -5.27425945e-01
5.72779894e-01 1.80389836e-01 2.64020622e-01 1.40132546e-01
-5.74639380e-01 -3.61869127e-01 6.40152618e-02 8.53634596e-01
4.89550889e-01 -4.43932861e-01 8.40531588e-01 -1.13944566e+00
4.87897158e-01 -3.69848371e-01 -4.43996131e-01 6.95073187e-01
-4.90427047e-01 5.48878312e-02 6.03926599e-01 -4.48569655e-02
-1.11848438e+00 -2.97124416e-01 -7.12520540e-01 6.42787039e-01
1.45882785e-01 9.82206821e-01 -4.53575939e-01 -6.23887107e-02
6.92678273e-01 1.28974512e-01 -1.64293930e-01 -8.30808342e-01
4.90083963e-01 6.45432651e-01 5.67889869e-01 -1.04060113e+00
1.84547037e-01 -2.22400770e-01 -2.11710304e-01 -5.61540246e-01
-7.67292559e-01 2.12099135e-01 -7.85364211e-01 7.13940663e-03
4.20340389e-01 -9.15099919e-01 -3.32529247e-01 1.11632085e+00
-1.44836199e+00 -3.36966753e-01 1.16840407e-01 -1.51085094e-01
-2.73535699e-01 6.08406782e-01 -1.11630535e+00 -3.68404269e-01
-6.24824464e-01 -1.05676234e+00 1.13590407e+00 -2.83541381e-01
5.35075143e-02 -1.19912636e+00 -2.79908013e-02 1.27138481e-01
4.17188317e-01 2.78335158e-02 1.33705282e+00 -5.27661264e-01
-3.96417379e-01 2.14886874e-01 -4.50443244e-03 5.30523419e-01
-1.50389314e-01 5.62455691e-02 -5.97613513e-01 -4.53721285e-01
-4.19556230e-01 -4.82980788e-01 9.97731149e-01 -6.34500310e-02
8.54169846e-01 -1.44616485e-01 3.18642706e-01 6.42363489e-01
1.44962060e+00 -4.40235473e-02 9.75128412e-01 5.79135537e-01
6.26748145e-01 5.45391917e-01 8.52901816e-01 3.17882180e-01
7.58487463e-01 4.61404085e-01 5.64037502e-01 4.56513822e-01
-4.78797913e-01 -7.49829784e-02 7.22072482e-01 1.63978839e+00
-3.90361130e-01 -8.44133019e-01 -1.20803082e+00 8.13780546e-01
-1.57039380e+00 -3.09120834e-01 -7.81908810e-01 1.77344763e+00
1.22447300e+00 -3.66214722e-01 -1.17086411e-01 3.97367328e-01
5.39208174e-01 -4.68394846e-01 2.31007174e-01 -9.28214610e-01
-9.10040855e-01 7.59631753e-01 1.94172204e-01 4.33883011e-01
-1.31821060e+00 1.53009927e+00 5.80865288e+00 9.64275718e-01
-6.59136236e-01 4.97619808e-01 3.73039246e-01 2.92878777e-01
-2.13543624e-01 -1.02795355e-01 -1.27690279e+00 6.05974078e-01
1.12161839e+00 5.07634759e-01 4.10094261e-01 3.75956148e-01
-2.41857350e-01 -1.27589941e-01 -8.91644955e-01 7.96757400e-01
6.46867454e-01 -1.06109178e+00 2.10600480e-01 -4.25750464e-01
8.34399819e-01 2.10586097e-02 -1.55712366e-01 5.70004344e-01
4.01525438e-01 -1.08930898e+00 1.15739095e+00 2.08117545e-01
1.05236554e+00 -9.75927114e-01 8.93101871e-01 -4.11177762e-02
-8.37276816e-01 6.42614812e-02 -4.54115570e-01 -3.83188650e-02
2.43550763e-01 2.83098996e-01 -4.07258928e-01 7.94189870e-01
8.42421532e-01 1.13695967e+00 -1.41142344e+00 8.29785824e-01
-7.75718510e-01 1.05620003e+00 -2.91797876e-01 -5.31191118e-02
2.29955137e-01 -2.77799308e-01 5.95972657e-01 1.73589730e+00
5.94336629e-01 -2.16583982e-02 -1.63084120e-01 3.98807138e-01
-2.30225205e-01 5.65738618e-01 -8.82117748e-02 -1.60842109e-02
4.20224190e-01 9.04807031e-01 -5.68231761e-01 -2.38365754e-02
-3.48937303e-01 1.36913586e+00 8.83388638e-01 -2.82740235e-01
-6.92525268e-01 -6.65387034e-01 3.29330742e-01 -3.06832314e-01
4.39215928e-01 -3.96845102e-01 -4.53835934e-01 -1.40590835e+00
1.42881781e-01 -1.52721143e+00 7.57195890e-01 -4.71929610e-01
-1.47008932e+00 6.38994634e-01 -6.01549029e-01 -1.05109417e+00
-1.37203872e-01 -1.25080943e+00 -3.19754153e-01 8.94184768e-01
-1.84658766e+00 -1.79031479e+00 1.10020459e-01 5.23043394e-01
5.56951523e-01 -5.07322669e-01 1.18236125e+00 6.92746282e-01
-4.82767850e-01 1.15684021e+00 3.98752093e-01 3.48713636e-01
1.23722243e+00 -1.65517139e+00 7.58509755e-01 1.51315689e+00
2.52407253e-01 5.11478126e-01 1.60812676e-01 -8.07169557e-01
-1.26141858e+00 -1.14086795e+00 1.60671616e+00 -9.76517439e-01
6.65594041e-01 -4.25081223e-01 -1.27340138e+00 7.87367225e-01
2.22751290e-01 -1.38441667e-01 4.74898189e-01 3.29873174e-01
-4.53203857e-01 4.56756502e-01 -1.00971699e+00 3.30104947e-01
1.22889328e+00 -4.24425453e-01 -5.54477930e-01 1.30428031e-01
5.57244480e-01 -7.90520847e-01 -8.20898175e-01 3.22116196e-01
1.96979329e-01 -4.26123321e-01 6.04384601e-01 -1.06304944e+00
7.48664260e-01 -3.63296121e-01 -5.33238232e-01 -2.13381243e+00
-1.46048004e-02 -3.80126178e-01 -6.32750690e-02 1.56449842e+00
7.44669557e-01 -2.29168579e-01 1.50003925e-01 -2.67310530e-01
-9.34536695e-01 -2.86002189e-01 -7.39375949e-01 -9.66126084e-01
7.27139831e-01 -7.00814664e-01 6.84911489e-01 1.13269973e+00
2.57506985e-02 1.03690192e-01 -1.78829640e-01 2.96463877e-01
2.06604987e-01 -1.52108938e-01 8.09483230e-02 -9.76337135e-01
-1.05977230e-01 -2.18683138e-01 -1.08086288e-01 -3.97525698e-01
3.02289218e-01 -1.39583385e+00 3.53647560e-01 -1.14473498e+00
-3.23926985e-01 -6.89207852e-01 1.44644424e-01 7.91132212e-01
-6.38785601e-01 5.71336150e-01 5.60562074e-01 -2.08024740e-01
-4.60882127e-01 5.23867607e-01 6.84804142e-01 7.04848382e-04
4.43300232e-02 -6.16457224e-01 -5.00657260e-01 5.80367506e-01
1.05417526e+00 -6.36794925e-01 3.34855884e-01 -1.27324188e+00
8.83924246e-01 -5.58190882e-01 2.41343137e-02 -7.42892504e-01
-8.56335536e-02 2.84304172e-01 4.72565293e-02 -3.96918982e-01
-2.76667982e-01 -3.44999284e-01 -5.48226118e-01 3.46869290e-01
1.92519709e-01 7.53445745e-01 4.65466887e-01 3.62457223e-02
-3.42706740e-01 -7.01872826e-01 8.37969720e-01 -1.72134578e-01
-1.11113644e+00 -2.30047241e-01 -4.70921725e-01 7.58595765e-01
1.76020399e-01 1.70845166e-01 -7.19487846e-01 1.76259682e-01
-6.64058447e-01 1.14071943e-01 3.53639185e-01 6.88982069e-01
6.40608311e-01 -1.20998764e+00 -1.37110496e+00 2.80245006e-01
6.26456201e-01 -5.01111507e-01 -1.76453769e-01 7.30882108e-01
-8.20864916e-01 1.72605991e-01 -3.10719252e-01 -2.29272559e-01
-1.51194990e+00 1.15945645e-01 3.30093682e-01 -4.68333781e-01
-4.49425913e-03 1.07272530e+00 -9.78598416e-01 -1.43413448e+00
-1.24408007e-01 -2.54698366e-01 -3.31411630e-01 1.62970468e-01
5.21617413e-01 6.81211412e-01 1.10958350e+00 -8.95661116e-01
-5.51652312e-01 3.30058008e-01 -2.25307599e-01 4.62614931e-02
1.37992668e+00 -2.53098339e-01 -5.99755287e-01 -7.48789590e-03
7.47718871e-01 1.55560821e-01 -4.97682542e-01 -1.98610127e-01
6.78583622e-01 -2.92964816e-01 -3.82357746e-01 -1.61129653e+00
-7.41144598e-01 1.05553961e+00 4.75606352e-01 -6.00230455e-01
1.10164654e+00 -3.19334090e-01 7.60279775e-01 5.92088878e-01
5.33779085e-01 -1.58502126e+00 -1.90645736e-02 1.81697202e+00
1.12907052e+00 -1.56091905e+00 -5.14934957e-01 -6.83673501e-01
-8.73390257e-01 1.11087215e+00 6.29871786e-01 -8.46272036e-02
5.37951708e-01 1.17597111e-01 5.65273613e-02 -1.06550246e-01
-4.32934523e-01 -4.21707064e-01 4.92903888e-01 8.77175570e-01
9.37791288e-01 2.21181810e-01 -9.72566068e-01 1.17871809e+00
-5.00849485e-01 -3.70659381e-01 8.90208840e-01 1.07104576e+00
-1.29202485e-01 -1.70880115e+00 -1.50430530e-01 1.55889153e-01
-6.43370390e-01 -7.93461323e-01 -3.15326601e-01 1.03463852e+00
3.24845940e-01 1.08126020e+00 -2.57865116e-02 6.20932616e-02
4.68826652e-01 4.29240972e-01 9.85312462e-01 -7.64364004e-01
-1.00661588e+00 -3.56243759e-01 5.44251919e-01 -4.15996045e-01
-3.18798572e-01 -9.09305334e-01 -1.34282160e+00 -3.87740582e-01
-4.00683880e-01 -1.23506762e-01 8.00291419e-01 1.00463545e+00
3.30898732e-01 2.34147966e-01 -4.04054783e-02 -4.51338440e-01
-4.29757893e-01 -1.37934566e+00 -6.00840271e-01 5.59971213e-01
-1.46708891e-01 -2.71487415e-01 -8.01477507e-02 3.76520395e-01] | [11.072563171386719, 10.721488952636719] |
256bc072-7859-41a8-b0ff-15bf2a183ef2 | improving-speech-emotion-recognition | 2305.14402 | null | https://arxiv.org/abs/2305.14402v1 | https://arxiv.org/pdf/2305.14402v1.pdf | Improving Speech Emotion Recognition Performance using Differentiable Architecture Search | Speech Emotion Recognition (SER) is a critical enabler of emotion-aware communication in human-computer interactions. Deep Learning (DL) has improved the performance of SER models by improving model complexity. However, designing DL architectures requires prior experience and experimental evaluations. Encouragingly, Neural Architecture Search (NAS) allows automatic search for an optimum DL model. In particular, Differentiable Architecture Search (DARTS) is an efficient method of using NAS to search for optimised models. In this paper, we propose DARTS for a joint CNN and LSTM architecture for improving SER performance. Our choice of the CNN LSTM coupling is inspired by results showing that similar models offer improved performance. While SER researchers have considered CNNs and RNNs separately, the viability of using DARTs jointly for CNN and LSTM still needs exploration. Experimenting with the IEMOCAP dataset, we demonstrate that our approach outperforms best-reported results using DARTS for SER. | ['Björn Schuller', 'Berrak Sisman', 'Sara Khalifa', 'Rajib Rana', 'Thejan Rajapakshe'] | 2023-05-23 | null | null | null | null | ['architecture-search', 'speech-emotion-recognition'] | ['methodology', 'speech'] | [ 2.71330010e-02 1.12806924e-01 -1.47685325e-02 -3.63007396e-01
-6.76645041e-01 -3.82411003e-01 4.20508802e-01 -2.18827873e-01
-4.68855321e-01 2.82335222e-01 3.60178798e-01 -3.72326732e-01
2.26790622e-01 -7.53166005e-02 -4.70238417e-01 -3.19531947e-01
-2.79639155e-01 1.41590521e-01 -4.66164798e-01 -3.33129853e-01
5.14188036e-02 6.42205596e-01 -1.37180710e+00 5.61628938e-01
4.39262122e-01 1.41211331e+00 1.08429961e-01 1.02984512e+00
-2.42681295e-01 1.16018188e+00 -7.72277296e-01 -3.50679278e-01
-1.02933124e-01 -5.71659207e-01 -1.04269218e+00 -2.77531564e-01
-9.11920518e-03 6.64707646e-02 -6.75659180e-02 6.29381537e-01
7.10849583e-01 3.95534039e-01 1.95844293e-01 -1.40081775e+00
-3.78563285e-01 7.68775403e-01 3.71078998e-02 3.55838418e-01
5.09366468e-02 1.03167124e-01 1.24458516e+00 -9.28074896e-01
2.67864078e-01 1.38205814e+00 6.24283373e-01 1.00564241e+00
-1.04966068e+00 -5.69613159e-01 4.32852447e-01 1.72884256e-01
-1.28598559e+00 -9.89423156e-01 8.35568309e-01 1.30724952e-01
1.96073663e+00 4.31520522e-01 7.82439649e-01 1.46936274e+00
2.33339414e-01 1.29727030e+00 7.61086941e-01 -4.83251333e-01
4.23592657e-01 3.32015544e-01 -6.20071106e-02 5.63470364e-01
-5.11823416e-01 1.17577419e-01 -9.00343180e-01 -3.22382569e-01
5.21129549e-01 -4.19959754e-01 -9.04182345e-02 4.57366966e-02
-8.26797187e-01 6.10607743e-01 4.22105908e-01 6.47127032e-01
-6.76659524e-01 5.24425507e-01 7.83968985e-01 6.70737565e-01
6.61778867e-01 1.07362914e+00 -8.82474244e-01 -7.33908832e-01
-8.44320297e-01 -1.80332795e-01 9.81852889e-01 6.68653369e-01
2.33743981e-01 6.09165967e-01 6.79050460e-02 1.19388235e+00
1.28741115e-01 -2.18318671e-01 5.15532255e-01 -1.15964305e+00
2.02580646e-01 3.91819626e-01 -2.17487663e-01 -8.26017201e-01
-5.50873041e-01 -8.51375699e-01 -6.81755364e-01 -9.60169286e-02
-1.72113329e-01 -4.23698545e-01 -6.08317316e-01 1.83225441e+00
-2.32580036e-01 4.11904901e-01 4.94676620e-01 7.96851575e-01
7.00330853e-01 7.93652236e-01 3.74452382e-01 -7.98321441e-02
1.07424140e+00 -1.17396331e+00 -7.75804818e-01 -5.61534882e-01
8.98458004e-01 -3.54924768e-01 9.83897567e-01 7.65050650e-01
-1.24897933e+00 -2.29653373e-01 -9.82956171e-01 1.61454663e-01
-1.87464744e-01 2.74051338e-01 7.81276882e-01 6.63175702e-01
-1.63916409e+00 4.18998122e-01 -9.19241667e-01 -5.79343140e-01
3.34046930e-01 6.74448192e-01 -1.75868813e-02 6.00086212e-01
-1.27335012e+00 1.12654376e+00 2.78505772e-01 4.52631712e-01
-8.87833297e-01 -3.65577936e-01 -6.76764369e-01 3.78608227e-01
1.95095554e-01 -6.79491103e-01 1.66880834e+00 -1.54894471e+00
-2.10899448e+00 5.03911078e-01 -4.16704625e-01 -1.05006742e+00
-1.50689781e-01 -2.45688751e-01 -6.54586554e-01 1.11655168e-01
-9.24655020e-01 1.09170330e+00 8.59825015e-01 -1.15274882e+00
-5.32662928e-01 4.47477773e-02 3.37700732e-02 3.76145422e-01
-9.33569193e-01 3.26191843e-01 -3.05653095e-01 -3.83386731e-01
-1.87961236e-01 -8.79923165e-01 -4.11776066e-01 -3.59847367e-01
-1.57205686e-01 -3.40429366e-01 7.39944041e-01 -6.02807641e-01
1.53839719e+00 -2.08499551e+00 1.38212755e-01 2.18808889e-01
2.16860622e-01 3.15725893e-01 -4.62913394e-01 1.84271529e-01
-1.80006608e-01 5.25296569e-01 1.50704280e-01 -8.24294984e-01
6.58007413e-02 2.67686784e-01 -1.97839200e-01 -4.05601859e-02
4.48903650e-01 1.16269052e+00 -4.34952110e-01 -2.68467367e-01
1.33525357e-01 8.31879795e-01 -7.55352199e-01 2.44758993e-01
-2.34823644e-01 1.62340075e-01 -2.39184663e-01 6.70346260e-01
3.93641293e-02 -3.80622298e-01 2.05963314e-01 -1.78105637e-01
-6.75114170e-02 6.02651417e-01 -5.39131641e-01 1.53630579e+00
-9.31411684e-01 1.01843548e+00 3.15536380e-01 -8.34403396e-01
1.09738064e+00 7.16091394e-01 3.84312540e-01 -9.64032352e-01
4.57118630e-01 1.34446025e-01 2.07354859e-01 -3.59223425e-01
5.26554763e-01 -5.89733310e-02 7.29716048e-02 4.01234984e-01
7.76707381e-02 1.67615160e-01 -5.07357955e-01 -6.17388822e-02
1.24971592e+00 -2.79758543e-01 1.20301887e-01 -1.45394742e-01
2.90383607e-01 -2.77079433e-01 4.02026832e-01 5.58846951e-01
-4.88369465e-01 2.23190576e-01 3.40068758e-01 -4.34653223e-01
-8.44831169e-01 -5.74268222e-01 2.76895314e-01 1.31687713e+00
-2.58756965e-01 -3.88457924e-01 -9.20661092e-01 -5.45059681e-01
-6.84088826e-01 1.00186980e+00 -5.31828761e-01 -2.89959639e-01
-7.17570066e-01 -5.90965986e-01 8.54480565e-01 5.21656573e-01
4.71540987e-01 -1.53131461e+00 -9.08797264e-01 4.60729718e-01
-3.19566190e-01 -1.23909533e+00 -3.35468203e-01 5.84115088e-01
-8.97206783e-01 -3.11767846e-01 -4.88377810e-01 -7.08659351e-01
2.27094457e-01 -7.13882670e-02 1.21862495e+00 2.64610261e-01
1.41414357e-02 5.61318398e-01 -3.55864793e-01 -3.93598944e-01
-5.39218485e-01 5.61943173e-01 -9.66740120e-03 -1.16412774e-01
2.61442572e-01 -5.82822978e-01 -4.57981765e-01 1.44300357e-01
-5.87435305e-01 1.05921485e-01 5.70249617e-01 7.47519195e-01
1.45396456e-01 -7.30401576e-02 8.34403694e-01 -3.05020481e-01
1.21453261e+00 -3.93878430e-01 2.13143658e-02 1.89703315e-01
-8.00303638e-01 2.50070393e-01 3.29443187e-01 -4.76796299e-01
-1.01478827e+00 -8.46691802e-02 -4.66595858e-01 -7.24674582e-01
-3.24337393e-01 7.16354430e-01 1.21460326e-01 -8.09271783e-02
5.14088750e-01 -3.70167121e-02 -2.95836627e-02 -3.91042084e-01
-3.20324004e-02 6.74625874e-01 1.81246519e-01 -4.96144980e-01
-2.02534914e-01 -8.43760222e-02 -5.28951466e-01 -9.07045424e-01
-4.45040435e-01 -1.13367870e-01 -9.14944801e-03 -3.60009938e-01
8.37590277e-01 -8.65740001e-01 -9.97425199e-01 1.23452723e-01
-1.33696127e+00 -6.13831401e-01 5.90012670e-02 3.29156250e-01
-4.45555508e-01 -1.59473553e-01 -8.05479527e-01 -1.36486185e+00
-7.62447953e-01 -1.24014986e+00 8.48963678e-01 1.00475319e-01
-7.96667457e-01 -1.37329268e+00 -2.45360568e-01 2.02018380e-01
8.49232852e-01 -1.34628341e-02 8.46461236e-01 -1.02563024e+00
-1.60649166e-01 -6.75731301e-02 2.02193260e-01 5.87907672e-01
-2.79823095e-01 -1.10439762e-01 -1.42875969e+00 -1.96813270e-01
1.70800626e-01 -4.57856268e-01 6.19381070e-01 4.08708334e-01
1.39512670e+00 -5.37411273e-01 -2.03171283e-01 3.84025127e-01
9.30542469e-01 6.24119818e-01 6.28542960e-01 5.41751266e-01
5.83866537e-01 5.31800866e-01 2.43518040e-01 4.01402384e-01
5.27803600e-01 5.75275540e-01 4.59402025e-01 -3.54845405e-01
1.38162270e-01 5.95352240e-03 8.26142192e-01 1.10710502e+00
1.83263928e-01 -6.34989679e-01 -1.02742493e+00 6.10969365e-01
-1.79058063e+00 -7.86011696e-01 4.27073270e-01 1.70256186e+00
5.27046144e-01 1.97601140e-01 1.29407510e-01 6.03353269e-02
4.79407161e-01 6.00287616e-02 -6.56872690e-01 -1.27359724e+00
-1.68412715e-01 4.38653082e-01 1.70899928e-01 4.99924481e-01
-7.99176037e-01 1.21507430e+00 6.81421566e+00 7.42643833e-01
-1.39455485e+00 1.76636770e-01 1.03780532e+00 -5.67670941e-01
-2.86511630e-01 -2.88924813e-01 -5.13153553e-01 4.49291021e-02
1.55899453e+00 7.79417828e-02 7.42460489e-01 8.54468465e-01
3.23158622e-01 3.04426134e-01 -1.13850498e+00 1.26854551e+00
4.35567833e-02 -1.32659817e+00 -1.45726100e-01 4.17650342e-02
3.57320547e-01 2.27987215e-01 2.53337771e-01 5.68506658e-01
1.56913206e-01 -1.29339135e+00 9.13743496e-01 3.54536027e-01
3.05405706e-01 -1.07484627e+00 5.49138129e-01 2.18076885e-01
-1.10503364e+00 -3.80847216e-01 9.03566480e-02 -2.04063840e-02
-3.94059531e-02 -4.48634326e-02 -1.13205624e+00 1.41755536e-01
8.11181128e-01 5.95604539e-01 -3.92732888e-01 6.77884519e-01
1.69974193e-01 9.12360847e-01 -2.57791698e-01 -3.83488476e-01
5.68385720e-01 3.24794948e-01 5.56982100e-01 1.53229451e+00
3.06298465e-01 -8.05020053e-03 -9.37326849e-02 8.00529838e-01
-1.07858598e-01 5.81117906e-03 -5.43586016e-01 -4.95500922e-01
6.10901833e-01 1.22573304e+00 -5.13440907e-01 -1.46933958e-01
6.58429265e-02 1.00846398e+00 4.02545333e-01 6.03881836e-01
-7.11812794e-01 -6.56417087e-02 1.04891753e+00 -3.74350905e-01
1.17615685e-01 -3.22665900e-01 -5.21821618e-01 -7.18911886e-01
-2.73770362e-01 -1.21769822e+00 3.56594354e-01 -8.64460766e-01
-1.03930247e+00 1.28707027e+00 -2.97859639e-01 -9.46889877e-01
-5.63328505e-01 -4.27936465e-01 -6.31797969e-01 6.21004999e-01
-1.32657981e+00 -1.02605140e+00 1.12095788e-01 4.19247746e-01
9.56775486e-01 -2.77579337e-01 1.05595684e+00 7.44902864e-02
-8.86861324e-01 9.76722658e-01 -5.12111604e-01 -1.42940328e-01
2.90416241e-01 -9.39948916e-01 6.35191679e-01 5.30221403e-01
1.65750518e-01 6.51445031e-01 9.06124473e-01 -2.66691953e-01
-1.52877116e+00 -7.21184731e-01 7.53194153e-01 -3.65362823e-01
5.38980663e-01 -5.17040610e-01 -7.83175290e-01 6.46713734e-01
7.53046334e-01 -4.82886583e-01 6.30728662e-01 3.95126671e-01
-2.67028902e-02 1.30077645e-01 -9.31077719e-01 8.92137289e-01
9.68375444e-01 -6.75417423e-01 -2.51606125e-02 -2.73186713e-01
9.55744088e-01 -3.66385967e-01 -9.99733984e-01 3.54900151e-01
5.18071353e-01 -8.54800761e-01 8.22780550e-01 -5.13222277e-01
1.28637731e-01 1.98617890e-01 -2.31556892e-01 -1.57779658e+00
1.09617636e-02 -9.01253104e-01 -3.83447140e-01 1.05473197e+00
9.05350804e-01 -6.25228345e-01 7.06229210e-01 1.04783833e+00
-6.46489918e-01 -1.15964484e+00 -1.03896153e+00 -6.61889076e-01
-8.82743075e-02 -9.97685492e-01 5.64474404e-01 9.28045392e-01
1.75257385e-01 3.78719032e-01 -4.11860377e-01 1.11743510e-01
4.83189337e-02 -6.24951243e-01 4.49533135e-01 -9.23315287e-01
-2.74656922e-01 -9.36246037e-01 -6.75431415e-02 -9.75610912e-01
5.67759454e-01 -7.64966786e-01 1.32196888e-01 -1.42577863e+00
-2.30433911e-01 -3.38644207e-01 -6.61368608e-01 8.87803078e-01
3.05521101e-01 -1.78012460e-01 2.70324588e-01 -7.74039924e-02
-7.74223387e-01 7.78838813e-01 8.29061210e-01 -5.24381967e-03
-5.66790462e-01 -3.21142375e-01 -8.18143487e-01 6.24789774e-01
1.29812181e+00 -2.84449637e-01 -4.69488442e-01 -5.20650446e-01
3.93726319e-01 9.29693207e-02 3.58461380e-01 -9.82239068e-01
4.52525884e-01 1.33658707e-01 2.31858030e-01 -1.72799692e-01
7.99922705e-01 -7.57919729e-01 3.13286111e-02 2.24419162e-01
-8.27180624e-01 2.96978593e-01 7.90611982e-01 1.94260105e-01
-2.25783125e-01 8.76110699e-03 5.71110070e-01 -4.16666362e-03
-8.63936007e-01 9.96733382e-02 -7.69592106e-01 -1.94442108e-01
3.91514570e-01 -3.06980729e-01 -1.23518199e-01 -7.16935039e-01
-9.10859108e-01 2.95811236e-01 -3.35405841e-02 8.43314409e-01
1.02763462e+00 -1.22479749e+00 -2.94250578e-01 8.60108659e-02
-7.13176429e-02 -5.92748821e-01 6.23043366e-02 8.90191615e-01
2.90137604e-02 4.59740907e-01 3.18692103e-02 -4.25791919e-01
-1.46197617e+00 2.57644197e-03 9.16518629e-01 -2.83554167e-01
-3.30991805e-01 1.12497735e+00 -1.12664133e-01 -4.50736552e-01
6.62205517e-01 -3.15476447e-01 -1.37308612e-01 -1.80474415e-01
3.58029753e-01 2.54343033e-01 2.82882094e-01 -2.56420493e-01
-4.68622774e-01 -1.44777372e-01 -2.41071388e-01 -5.09724557e-01
1.41098690e+00 -2.72894084e-01 -5.61262146e-02 6.27418399e-01
1.33884108e+00 -6.69625580e-01 -1.13742924e+00 -3.53229381e-02
4.21135604e-01 2.12869748e-01 5.97010553e-01 -1.01675129e+00
-1.06368661e+00 9.22715425e-01 7.39897013e-01 2.22600058e-01
1.33870935e+00 -5.69505244e-02 9.99536395e-01 6.79891109e-01
2.40684107e-01 -1.30816007e+00 3.86743814e-01 8.95763755e-01
1.02705026e+00 -9.79642212e-01 -7.00493693e-01 2.44674161e-01
-1.09304178e+00 1.11471617e+00 9.31984961e-01 2.02337757e-01
5.24062157e-01 5.13679802e-01 2.32300788e-01 -3.94688874e-01
-1.55257678e+00 -4.28327620e-02 2.93982804e-01 1.14647880e-01
7.51051486e-01 -5.97060435e-02 3.19628596e-01 7.09080637e-01
-3.74730915e-01 -1.36582628e-01 5.95506132e-02 8.96978974e-01
-3.20564806e-01 -1.06372440e+00 -2.83546299e-02 2.57414669e-01
-4.84154433e-01 -4.44624245e-01 -6.73363447e-01 5.16364515e-01
-1.88902199e-01 1.25572133e+00 2.39351690e-01 -8.84243131e-01
3.40540498e-01 3.97268623e-01 2.45233238e-01 -3.53039682e-01
-1.32757080e+00 1.98008791e-01 5.72178483e-01 -7.28046477e-01
-2.35267311e-01 -4.45117295e-01 -1.44663429e+00 -1.81743920e-01
-2.88507462e-01 2.69973099e-01 9.14832115e-01 1.08199060e+00
8.15156579e-01 8.00048828e-01 5.71947575e-01 -8.15990210e-01
-1.01534911e-01 -7.54788816e-01 -8.33240002e-02 -3.03189784e-01
4.22109336e-01 -2.11930513e-01 -3.37524593e-01 -2.69980758e-01] | [14.117558479309082, 6.089045524597168] |
cea6d348-df1f-49e3-8525-0b0d9ac37008 | a-large-scale-japanese-dataset-for-aspect | null | null | https://aclanthology.org/2022.lrec-1.758 | https://aclanthology.org/2022.lrec-1.758.pdf | A Large-Scale Japanese Dataset for Aspect-based Sentiment Analysis | There has been significant progress in the field of sentiment analysis. However, aspect-based sentiment analysis (ABSA) has not been explored in the Japanese language even though it has a huge scope in many natural language processing applications such as 1) tracking sentiment towards products, movies, politicians etc; 2) improving customer relation models. The main reason behind this is that there is no standard Japanese dataset available for ABSA task. In this paper, we present the first standard Japanese dataset for the hotel reviews domain. The proposed dataset contains 53,192 review sentences with seven aspect categories and two polarity labels. We perform experiments on this dataset using popular ABSA approaches and report error analysis. Our experiments show that contextual models such as BERT works very well for the ABSA task in the Japanese language and also show the need to focus on other NLP tasks for better performance through our error analysis. | ['Ikuko Hardaway', 'Sudha Bhingardive', 'Gautam Kumar', 'Koji Murakami', 'Yuki Nakayama'] | null | null | null | null | lrec-2022-6 | ['aspect-based-sentiment-analysis'] | ['natural-language-processing'] | [ 3.92062450e-03 -1.23785459e-01 -1.94110379e-01 -8.48595619e-01
-7.74760246e-01 -5.53710163e-01 5.41866839e-01 3.47314298e-01
-5.41991234e-01 6.59490943e-01 3.39610636e-01 -5.49669445e-01
1.06136329e-01 -6.66221082e-01 -3.23543698e-01 -6.38669431e-01
4.52280939e-01 4.62879360e-01 3.14188540e-01 -6.42093956e-01
5.90366185e-01 7.70397335e-02 -1.13228965e+00 4.95010465e-01
6.29587352e-01 1.00979567e+00 1.19680114e-01 4.45714593e-01
-4.24985319e-01 9.82646644e-01 -7.26870418e-01 -1.06612670e+00
-1.42414672e-02 -3.58591646e-01 -1.04346049e+00 1.14922345e-01
2.83855051e-02 4.04065788e-01 6.21692359e-01 1.08770514e+00
2.97835588e-01 5.67089096e-02 5.01648903e-01 -1.04491270e+00
-7.34433055e-01 8.41179550e-01 -6.63821995e-01 1.30305663e-01
2.78597295e-01 -5.09340107e-01 1.44000995e+00 -8.64671588e-01
6.44296944e-01 1.08283591e+00 5.88896096e-01 2.38384232e-01
-4.92935687e-01 -3.40713322e-01 4.26750898e-01 1.44229725e-01
-8.20625365e-01 5.44356219e-02 9.30429280e-01 -2.74802178e-01
1.35856760e+00 2.88774580e-01 6.60982370e-01 7.51569033e-01
4.46294934e-01 9.81493533e-01 1.44414735e+00 -6.92239285e-01
9.34850797e-02 7.62748599e-01 7.68464148e-01 3.20033461e-01
3.55026901e-01 -8.08528960e-01 -3.88004243e-01 1.35316938e-01
-1.14522159e-01 -2.07219005e-01 2.62512475e-01 -7.04048052e-02
-1.04720104e+00 1.11332154e+00 -1.11882407e-02 5.72922885e-01
-2.87940353e-01 -3.28357577e-01 6.79762423e-01 4.91127312e-01
8.18185806e-01 8.23802054e-01 -1.23935008e+00 -2.67175049e-01
-4.82499689e-01 3.72718930e-01 1.09164631e+00 1.03433216e+00
5.58578789e-01 -1.44657418e-01 2.94729650e-01 9.42219317e-01
3.97701293e-01 6.90552711e-01 5.30347705e-01 -5.88986933e-01
7.38644898e-01 8.85409057e-01 1.42535970e-01 -1.17322409e+00
-4.97486532e-01 -1.46299958e-01 -3.35027307e-01 -3.52813043e-02
2.52763778e-01 -2.85747051e-01 -6.93451703e-01 1.12932312e+00
1.17734790e-01 -8.59104931e-01 4.48712856e-01 6.61152482e-01
9.88786519e-01 8.34687889e-01 1.81100483e-03 -3.43052477e-01
1.94960439e+00 -1.36681855e+00 -1.01762021e+00 -4.77049619e-01
1.12513626e+00 -1.36390078e+00 1.19985569e+00 7.20897973e-01
-8.46697867e-01 -3.08445454e-01 -7.89292753e-01 -1.75543591e-01
-9.78669643e-01 2.76230603e-01 9.91427362e-01 8.60521197e-01
-7.27663517e-01 -1.49869293e-01 -5.45592606e-01 -6.99329317e-01
9.19642076e-02 2.98049420e-01 -2.56744057e-01 -2.04536989e-02
-1.19685018e+00 1.09643507e+00 -5.57912790e-05 2.69333214e-01
-6.26694039e-02 -4.99868616e-02 -1.13772237e+00 -2.44255021e-01
4.94123489e-01 -3.09393823e-01 1.41116571e+00 -1.19506681e+00
-1.46307862e+00 9.72747207e-01 -4.83615845e-01 -3.59613329e-01
-2.47504309e-01 -4.95404899e-01 -6.88186407e-01 -2.58109897e-01
3.38654608e-01 3.04436713e-01 3.21284473e-01 -1.07768714e+00
-9.23891366e-01 -5.44957161e-01 3.69384378e-01 2.43204430e-01
-3.31253797e-01 6.93802536e-01 -3.64813536e-01 -6.66819572e-01
-1.59643322e-01 -1.19981551e+00 -4.87521857e-01 -8.83492231e-01
-4.71879929e-01 -4.87897187e-01 6.56668484e-01 -4.11249220e-01
1.45723689e+00 -1.85739660e+00 -2.43013054e-01 1.65141508e-01
-4.13200080e-01 1.73042372e-01 8.91742781e-02 5.38634300e-01
-7.82758519e-02 2.50464469e-01 -2.54166007e-01 -4.50137138e-01
-1.50239334e-01 2.24809110e-01 -5.53355694e-01 -2.72502955e-02
1.26110926e-01 7.94770300e-01 -7.66778886e-01 -5.54455042e-01
-1.12408683e-01 2.22592086e-01 -4.44657475e-01 -1.31384656e-01
-6.95954785e-02 1.69498354e-01 -8.00236225e-01 9.56610203e-01
4.01290804e-01 -1.43224880e-01 1.06630892e-01 -9.18928683e-02
-4.47503626e-01 7.87166297e-01 -8.80249679e-01 1.37057209e+00
-6.37621462e-01 7.37674654e-01 -1.86183050e-01 -1.02243054e+00
1.03020728e+00 2.47048512e-01 2.50430375e-01 -6.01121664e-01
4.25895065e-01 3.01254243e-01 1.95589721e-01 -6.59364522e-01
1.15811145e+00 -3.80632192e-01 -5.87664485e-01 3.22778493e-01
-2.22979262e-01 -5.61051130e-01 7.34586060e-01 1.32489324e-01
5.42674899e-01 1.37871522e-02 4.47670013e-01 -5.27499914e-01
7.89872587e-01 6.72546029e-01 4.76247609e-01 3.98434341e-01
-2.37188280e-01 5.67585886e-01 7.88075686e-01 -5.81301332e-01
-7.93546855e-01 -1.24422699e-01 -2.86484629e-01 1.04722297e+00
-2.63292324e-02 -7.56699026e-01 -4.95993793e-01 -9.59597468e-01
-4.70436960e-01 7.76265085e-01 -6.11067891e-01 5.07095397e-01
-4.69735265e-01 -1.21133232e+00 -8.15269500e-02 4.32510048e-01
3.80770087e-01 -1.38697863e+00 -2.08307773e-01 1.57129869e-01
-1.49373904e-01 -1.33992386e+00 -3.09336483e-01 3.86941910e-01
-8.74514163e-01 -9.33857143e-01 -3.92676294e-01 -1.16540277e+00
6.28448904e-01 3.72688860e-01 1.24358845e+00 -4.23221529e-01
1.45504281e-01 1.96281672e-01 -8.37368429e-01 -1.04240525e+00
-2.56528288e-01 4.60928172e-01 -7.03352988e-02 -2.33748749e-01
1.02504778e+00 6.56448901e-02 -3.16637218e-01 4.13826287e-01
-7.35317528e-01 -1.62983134e-01 5.61776161e-01 4.58211035e-01
4.83758688e-01 2.41774529e-01 7.51879156e-01 -1.69972157e+00
9.26162958e-01 -2.01955482e-01 -5.59715807e-01 9.78792682e-02
-8.50422204e-01 -2.21303165e-01 5.78024924e-01 1.03989266e-01
-1.17473233e+00 -1.54738918e-01 -4.10364747e-01 7.11811960e-01
-1.80477545e-01 1.01777577e+00 -9.25436392e-02 3.65637273e-01
3.44988227e-01 -1.87839821e-01 -3.27998549e-01 -3.28204185e-01
7.33455494e-02 8.56700897e-01 -2.25848988e-01 -1.98551685e-01
2.97123730e-01 3.60598743e-01 -1.56144023e-01 -8.71941447e-01
-1.46453929e+00 -9.26918685e-01 -4.43666339e-01 1.27920657e-01
1.05623329e+00 -9.80351031e-01 -4.76901740e-01 4.39221978e-01
-1.08262885e+00 1.26586467e-01 -1.60855845e-01 7.03524053e-01
-2.15530381e-01 3.00756186e-01 -6.45007491e-01 -8.90524924e-01
-5.94118893e-01 -1.50280333e+00 9.40053582e-01 2.73847431e-01
-4.55309182e-01 -1.01340353e+00 2.48953536e-01 8.60089004e-01
3.12650204e-01 -3.33859831e-01 8.15860331e-01 -9.09383357e-01
-1.89034402e-01 -5.26283383e-01 1.52085036e-01 6.26736462e-01
2.14142486e-01 1.73600689e-01 -9.23089921e-01 1.93468168e-01
2.81973958e-01 -4.26607877e-01 6.83316231e-01 4.54186648e-01
5.98849595e-01 1.51975239e-02 -1.47849405e-02 -1.66727498e-01
1.21667683e+00 5.96795261e-01 6.46139979e-01 8.27178717e-01
5.78865111e-01 1.04358268e+00 1.29076898e+00 1.21939331e-01
6.92128181e-01 3.42731595e-01 2.04041615e-01 -1.86369084e-02
3.51794571e-01 2.11044699e-01 6.25670731e-01 1.73566628e+00
-4.33449633e-02 -4.34622854e-01 -7.71004856e-01 9.00125682e-01
-1.82296860e+00 -6.95095658e-01 -6.58534408e-01 1.78539157e+00
7.03785539e-01 3.76216769e-01 -7.74777448e-03 1.07425667e-01
2.64533848e-01 2.85132736e-01 1.21169098e-01 -1.22915959e+00
-3.75122964e-01 -5.86432731e-03 2.44595915e-01 5.31445742e-01
-1.41215098e+00 1.17260635e+00 5.66433954e+00 5.67839622e-01
-7.81436026e-01 3.98077443e-02 8.29247177e-01 3.90246481e-01
-3.38013887e-01 2.06577435e-01 -1.14275599e+00 1.22248389e-01
7.81963825e-01 2.64831543e-01 -1.69593826e-01 1.27781522e+00
1.62579015e-01 -4.76244748e-01 -5.19590974e-01 6.36737585e-01
4.09560561e-01 -9.15201366e-01 1.20889224e-01 -9.00002196e-02
1.01978028e+00 -4.22325321e-02 -4.41622585e-02 5.55459321e-01
5.84041588e-02 -6.45198107e-01 3.63859981e-01 6.97832704e-02
4.02207449e-02 -8.98978710e-01 1.54812610e+00 8.60966220e-02
-8.74006331e-01 2.92415202e-01 -5.87805033e-01 -1.93881929e-01
4.46272135e-01 7.50952542e-01 -6.95891976e-01 5.24114668e-01
9.00484800e-01 1.00056601e+00 -6.86691046e-01 6.16161406e-01
-3.56105894e-01 7.67061532e-01 -2.99044456e-02 -5.93172073e-01
5.58263540e-01 -6.20988667e-01 3.24116498e-01 1.32054186e+00
2.13129729e-01 -5.93831316e-02 -1.04132563e-01 -5.21742254e-02
-1.21932859e-02 7.14156926e-01 -7.70213485e-01 -2.22294509e-01
-3.30107182e-01 1.51122332e+00 -1.16837585e+00 -3.06168705e-01
-9.78615046e-01 6.16878748e-01 -7.20565841e-02 1.54985040e-01
-5.12736559e-01 -7.00914323e-01 4.35365647e-01 -1.25450850e-01
3.91988605e-01 -3.43312845e-02 -6.05525374e-01 -1.31024301e+00
1.56051554e-02 -1.00665987e+00 3.69278878e-01 -1.00338268e+00
-1.34010863e+00 1.04587090e+00 -1.94736645e-01 -1.13613701e+00
-1.11174069e-01 -1.17137313e+00 -4.35918659e-01 6.43292367e-01
-1.77174425e+00 -1.15580177e+00 3.09072316e-01 2.95182794e-01
9.49023008e-01 -2.04164073e-01 8.54650259e-01 2.98876941e-01
-5.41341305e-01 6.61738366e-02 -1.51395187e-01 -9.63601342e-04
1.06332314e+00 -1.41425467e+00 3.15575212e-01 7.91665196e-01
1.79032519e-01 9.00660872e-01 8.80549073e-01 -5.83787084e-01
-1.22484744e+00 -5.81949592e-01 1.88639486e+00 -9.42767441e-01
9.74141359e-01 -2.97333032e-01 -5.94422936e-01 7.81235993e-01
6.59137249e-01 -4.99799818e-01 1.06666374e+00 6.78717494e-01
-2.43408512e-02 -3.71485323e-01 -6.72554731e-01 4.96242106e-01
2.65743881e-01 -3.73923033e-01 -7.74673879e-01 3.82095933e-01
7.58854091e-01 -1.05781786e-01 -6.66532397e-01 4.18542475e-01
3.36360335e-01 -7.11478651e-01 5.18689871e-01 -6.33346558e-01
7.37142801e-01 -4.80905205e-01 -2.31781468e-01 -1.23223460e+00
8.17414671e-02 -6.96323514e-02 5.13047934e-01 1.38967192e+00
1.15123737e+00 -5.54708481e-01 6.69309199e-01 7.43157566e-01
-2.94299662e-01 -9.98031557e-01 -4.27652448e-01 -2.40272462e-01
-6.10487759e-02 -9.41630721e-01 3.59305471e-01 9.71845746e-01
1.67926177e-01 9.27631736e-01 -3.48878860e-01 -9.67844501e-02
-9.02452841e-02 6.57115936e-01 7.60605991e-01 -9.46216285e-01
6.35376945e-02 -2.96804816e-01 -9.42321047e-02 -8.35105777e-01
-6.44582091e-03 -5.01043439e-01 3.34773213e-02 -1.73261094e+00
2.43991107e-01 -3.89189184e-01 -2.48601779e-01 3.09100479e-01
-3.02986950e-01 4.75627869e-01 9.06551331e-02 -9.80434380e-03
-9.19126093e-01 3.36467594e-01 1.28128052e+00 1.40645914e-02
-1.78708881e-01 3.17979962e-01 -1.22474730e+00 1.00795519e+00
9.12529290e-01 -6.22994602e-01 -3.38920116e-01 -3.97862256e-01
1.12550879e+00 -4.34241533e-01 -5.77518165e-01 -2.57550180e-01
2.81768236e-02 -1.37488812e-01 2.53054556e-02 -9.14692104e-01
1.65750936e-01 -1.06732917e+00 -4.64154243e-01 8.96080732e-02
-2.20603153e-01 6.13762558e-01 4.25001010e-02 3.97718638e-01
-6.87334359e-01 -6.23416483e-01 2.87223935e-01 -1.90415710e-01
-7.21105933e-01 -2.20009983e-01 -7.05562532e-01 5.51868454e-02
8.75264406e-01 5.59556112e-02 -3.14558059e-01 -2.60978431e-01
-5.49372077e-01 2.66198516e-01 2.83052802e-01 5.29562533e-01
3.03542465e-01 -8.85589778e-01 -3.23314905e-01 -9.56614539e-02
4.51469719e-01 1.94338113e-02 2.56619006e-02 8.65410805e-01
-6.49808466e-01 9.38854635e-01 2.57921398e-01 -2.87475765e-01
-1.30015123e+00 6.67398870e-01 -2.65421987e-01 -7.48644412e-01
-1.03457803e-02 6.53509915e-01 2.09040105e-01 -6.77305162e-01
-6.90262243e-02 -6.24260783e-01 -1.07341337e+00 4.93849725e-01
5.14696538e-01 5.59337996e-02 3.23141813e-01 -9.30086792e-01
-4.88937289e-01 7.62260914e-01 -4.23577219e-01 -4.73196991e-02
1.35861516e+00 -4.48363394e-01 -4.64749694e-01 9.39968705e-01
1.04955113e+00 5.16939640e-01 -2.47090772e-01 1.37606356e-02
4.04256284e-01 -2.10084692e-01 -6.55270517e-02 -8.98127437e-01
-9.29037094e-01 7.73956954e-01 -6.25739768e-02 6.26080811e-01
1.10023475e+00 9.66058001e-02 8.35474372e-01 6.40811503e-01
4.03335899e-01 -1.42939579e+00 -2.14258969e-01 9.44238663e-01
6.93256199e-01 -1.70559442e+00 1.64213225e-01 -4.82137680e-01
-1.48760545e+00 9.98992562e-01 4.60039675e-01 5.67867130e-04
8.59094620e-01 1.37677714e-01 7.16476381e-01 -5.60385168e-01
-7.05871165e-01 -2.62170970e-01 2.04033926e-01 1.78828254e-01
8.98585141e-01 4.27872948e-02 -9.08848703e-01 9.96216834e-01
-3.44268024e-01 -2.70606220e-01 8.73810470e-01 1.06234729e+00
-2.92205334e-01 -1.39219499e+00 -1.42412454e-01 4.52524036e-01
-1.24039674e+00 -4.27168041e-01 -5.16925931e-01 9.29653049e-01
-1.41321942e-01 1.26539588e+00 -3.14828336e-01 -6.00919360e-03
5.18122077e-01 6.63455129e-02 -8.92121531e-03 -8.90195310e-01
-8.56488943e-01 2.76601791e-01 6.24399602e-01 -2.20795020e-01
-1.17576373e+00 -8.57967377e-01 -1.00444508e+00 2.77085863e-02
-5.93472600e-01 6.88496947e-01 1.12043142e+00 1.03994083e+00
5.77349588e-02 3.17039013e-01 6.10550582e-01 -1.24363989e-01
1.75029993e-01 -1.19667065e+00 -6.57326996e-01 1.98898181e-01
7.73730502e-02 -1.75958097e-01 -3.55804712e-01 1.19556203e-01] | [11.341278076171875, 6.772342681884766] |
cd3da91e-d7bc-42f8-8898-ee5c355b0e7a | learning-combinatorial-prompts-for-universal | 2303.06338 | null | https://arxiv.org/abs/2303.06338v2 | https://arxiv.org/pdf/2303.06338v2.pdf | Learning Combinatorial Prompts for Universal Controllable Image Captioning | Controllable Image Captioning (CIC) -- generating natural language descriptions about images under the guidance of given control signals -- is one of the most promising directions towards next-generation captioning systems. Till now, various kinds of control signals for CIC have been proposed, ranging from content-related control to structure-related control. However, due to the format and target gaps of different control signals, all existing CIC works (or architectures) only focus on one certain control signal, and overlook the human-like combinatorial ability. By ``combinatorial", we mean that our humans can easily meet multiple needs (or constraints) simultaneously when generating descriptions. To this end, we propose a novel prompt-based framework for CIC by learning Combinatorial Prompts, dubbed as ComPro. Specifically, we directly utilize a pretrained language model GPT-2 as our language model, which can help to bridge the gap between different signal-specific CIC architectures. Then, we reformulate the CIC as a prompt-guide sentence generation problem, and propose a new lightweight prompt generation network to generate the combinatorial prompts for different kinds of control signals. For different control signals, we further design a new mask attention mechanism to realize the prompt-based CIC. Due to its simplicity, our ComPro can easily be extended to more complex combined control signals by concatenating these prompts. Extensive experiments on two prevalent CIC benchmarks have verified the effectiveness and efficiency of our ComPro on both single and combined control signals. | ['Long Chen', 'Jian Shao', 'Fei Gao', 'Lei Chen', 'Jun Xiao', 'Zhen Wang'] | 2023-03-11 | null | null | null | null | ['controllable-image-captioning'] | ['computer-vision'] | [ 5.39256990e-01 -1.76468298e-01 -2.42831498e-01 -4.71338391e-01
-6.90295219e-01 -5.41154802e-01 7.70775318e-01 -3.64331543e-01
-4.33319807e-02 4.98084158e-01 6.17723763e-01 -2.96149760e-01
1.50946364e-01 -7.38223612e-01 -8.29224110e-01 -6.28478885e-01
5.85705638e-01 1.24258481e-01 2.20464319e-01 -5.38646042e-01
1.81295767e-01 5.30061917e-03 -1.10480535e+00 4.19459105e-01
1.07286441e+00 8.38153481e-01 6.48319066e-01 2.93884128e-01
-3.27382892e-01 8.15316677e-01 -6.79850399e-01 -1.92159131e-01
1.84445037e-03 -7.68412888e-01 -4.29243743e-01 3.56358767e-01
1.44155785e-01 -2.28721574e-01 -4.51310128e-01 9.99469459e-01
6.38162076e-01 -1.14513308e-01 4.96253103e-01 -1.39202988e+00
-1.36862552e+00 6.89947188e-01 -5.20340145e-01 -1.39417589e-01
6.51913702e-01 7.09866881e-01 9.42544580e-01 -6.55619323e-01
3.75814766e-01 1.49553752e+00 2.15814248e-01 9.77571905e-01
-1.24502909e+00 -7.06504107e-01 7.55908310e-01 1.61879078e-01
-1.21940219e+00 -3.38466704e-01 9.55988526e-01 -2.28646919e-01
5.24675548e-01 2.58007586e-01 4.19637054e-01 1.31823778e+00
-2.51657873e-01 1.26667738e+00 9.77954268e-01 -2.10139364e-01
1.26318410e-02 -3.70838083e-02 -1.46535918e-01 4.10553992e-01
-6.21141717e-02 -1.54854357e-01 -1.00033246e-01 2.20391482e-01
9.36815798e-01 -1.78074002e-01 -7.42795348e-01 -9.17655900e-02
-1.71808350e+00 7.10933864e-01 5.07652819e-01 7.00496808e-02
-7.79202580e-02 2.46806949e-01 4.61302459e-01 1.29029706e-01
3.81227732e-02 6.13181114e-01 -2.06872702e-01 1.37903824e-01
-4.40979809e-01 3.68230313e-01 4.44721133e-01 1.47360146e+00
5.45568526e-01 1.60192147e-01 -1.08813095e+00 8.34604144e-01
1.73396274e-01 5.78711033e-01 5.85287571e-01 -4.87036586e-01
8.93175125e-01 6.18203759e-01 3.29072773e-01 -1.08991647e+00
-2.16973156e-01 -3.35815698e-01 -1.08178651e+00 -5.66196084e-01
7.11689563e-03 -3.47381294e-01 -1.02694798e+00 2.15791774e+00
-2.06697006e-02 2.92936653e-01 3.32414061e-02 1.24273252e+00
9.57692921e-01 1.17574191e+00 7.46786743e-02 -2.83740640e-01
1.41393018e+00 -1.22921383e+00 -7.53648341e-01 -2.77123779e-01
2.12084860e-01 -5.20510614e-01 1.70222723e+00 5.84003627e-02
-7.09456921e-01 -6.85859859e-01 -1.01695204e+00 -1.46453276e-01
3.56067307e-02 2.52375484e-01 5.40272772e-01 2.03341156e-01
-9.85922933e-01 -1.46918014e-01 -3.70936483e-01 1.12824682e-02
2.90878534e-01 1.74510479e-01 -5.87528571e-02 -2.45305389e-01
-1.41791296e+00 3.73866349e-01 4.01974976e-01 2.48259589e-01
-9.35091257e-01 -5.96223533e-01 -9.87784505e-01 2.26708144e-01
7.67764211e-01 -7.44342148e-01 1.39443243e+00 -8.84183884e-01
-1.51684070e+00 3.97111088e-01 -1.71481043e-01 -1.15350634e-01
3.68942976e-01 -1.97784360e-02 -4.42300141e-01 1.10019453e-01
2.91117191e-01 1.10846150e+00 7.92141616e-01 -1.33137095e+00
-3.52557510e-01 3.21755558e-01 4.72356796e-01 2.48859227e-01
-4.05774891e-01 7.13296533e-02 -8.83284390e-01 -9.41902637e-01
-2.23408625e-01 -8.25470865e-01 -3.98448557e-01 -4.78635170e-02
-7.02050328e-01 -4.10192251e-01 5.53449929e-01 -1.31577581e-01
1.35558796e+00 -2.17847085e+00 2.61075914e-01 -2.73719788e-01
7.73082152e-02 3.34481150e-01 -6.46302462e-01 3.70415449e-01
-9.75233167e-02 3.64331216e-01 -3.59346569e-01 -2.18252569e-01
2.09010497e-01 1.79484040e-01 -5.50405264e-01 -2.71847337e-01
7.15595245e-01 1.21964681e+00 -1.15198052e+00 -5.36316872e-01
6.88199550e-02 2.30770543e-01 -5.62533438e-01 4.41861272e-01
-9.02203381e-01 4.92159158e-01 -8.16872001e-01 4.61632490e-01
6.58275604e-01 -4.40133601e-01 -1.16054609e-01 -4.15300190e-01
-9.26422700e-02 1.92515522e-01 -8.61189604e-01 1.66566336e+00
-6.44547462e-01 2.38442630e-01 -2.45747101e-02 -8.47236753e-01
1.09320414e+00 3.26838404e-01 3.03905942e-02 -7.95194566e-01
-1.67132530e-03 1.63642526e-01 1.20479716e-02 -7.07660317e-01
3.82340848e-01 -1.40998080e-01 -3.95493984e-01 2.32139230e-01
-2.38356173e-01 -1.71640962e-01 3.52307111e-01 2.11837813e-01
8.41138601e-01 2.28773788e-01 3.15095745e-02 -6.47251531e-02
8.65296602e-01 -2.05347925e-01 8.23083937e-01 6.61623657e-01
-1.79701850e-01 1.14611995e+00 9.23434138e-01 -3.02202046e-01
-8.92823398e-01 -7.47927547e-01 3.54786307e-01 8.24072182e-01
4.85831141e-01 -3.03939581e-01 -7.39310145e-01 -7.30805576e-01
-3.19138438e-01 7.03155696e-01 -3.79297525e-01 -2.18148693e-01
-7.11238384e-01 -5.43296456e-01 4.14622188e-01 5.78459263e-01
9.23592687e-01 -1.38731897e+00 -2.13721275e-01 3.12791318e-01
-5.73806643e-01 -1.37104952e+00 -1.27214718e+00 -2.66856074e-01
-1.60833985e-01 -6.46810591e-01 -1.04297519e+00 -1.16155565e+00
7.69849658e-01 4.24110770e-01 1.00654590e+00 2.18560636e-01
2.45144933e-01 9.30189118e-02 -4.95690256e-01 -1.92376181e-01
-3.15285444e-01 1.93036675e-01 -2.36460045e-01 3.82100493e-01
6.67294115e-02 -4.50745046e-01 -7.50318408e-01 5.31225920e-01
-1.22896492e+00 6.57674193e-01 8.71201873e-01 9.90599275e-01
5.28495967e-01 -3.07277769e-01 9.41524506e-01 -6.29375577e-01
1.03141224e+00 -4.66662288e-01 -5.76784134e-01 5.33016264e-01
-1.65811956e-01 3.39693367e-01 1.10664880e+00 -8.53093207e-01
-1.02220106e+00 7.80271692e-03 -9.36869159e-02 -5.77189267e-01
-2.34828249e-01 6.99109793e-01 -8.12400877e-01 3.26845169e-01
2.99465984e-01 7.15796590e-01 -1.50370792e-01 -1.94780976e-01
4.56864923e-01 7.42216587e-01 7.47744560e-01 -9.37858939e-01
7.17311740e-01 3.99620123e-02 -1.87754631e-01 -2.60629654e-01
-8.99750650e-01 -1.99154206e-02 3.69990282e-02 3.11713312e-02
1.04308236e+00 -8.20703447e-01 -6.75916433e-01 5.48598766e-01
-1.52618086e+00 -2.77266592e-01 1.26505688e-01 6.10458851e-02
-5.06261945e-01 3.47657055e-01 -5.72346091e-01 -5.54346859e-01
-3.73093218e-01 -1.67411959e+00 1.20237565e+00 4.98093873e-01
1.23026423e-01 -6.03576183e-01 -3.65128219e-01 1.71504363e-01
5.84012866e-01 2.42813647e-01 1.00506032e+00 -3.17976475e-01
-7.71164179e-01 5.05446307e-02 -6.00704670e-01 3.23080778e-01
2.42685124e-01 -2.33858928e-01 -5.52612782e-01 3.75041887e-02
-7.95021746e-03 -3.76312822e-01 6.66495085e-01 1.69518255e-02
1.40208888e+00 -5.26001215e-01 -1.80818513e-01 4.49223906e-01
1.28706217e+00 3.21950793e-01 7.97736883e-01 1.94585919e-01
8.13321829e-01 5.06271124e-01 6.80161774e-01 2.12721363e-01
6.46676660e-01 8.33315730e-01 4.65239733e-01 -1.56469688e-01
-1.16548106e-01 -8.30248415e-01 4.81388360e-01 8.61957490e-01
3.68753999e-01 -5.58040082e-01 -6.19444191e-01 5.66328049e-01
-1.94322038e+00 -8.07516515e-01 -8.95681977e-03 1.76437116e+00
1.07000053e+00 2.25344807e-01 -1.39059380e-01 -2.00349152e-01
1.04292870e+00 3.83697271e-01 -6.50744796e-01 -1.25082463e-01
-1.31050572e-01 -1.92680076e-01 7.67796412e-02 1.77081153e-01
-1.02622461e+00 9.68686938e-01 4.98280573e+00 1.15226781e+00
-1.45025551e+00 -2.38861784e-01 7.95180559e-01 1.16592586e-01
-8.12572479e-01 8.17303173e-03 -7.53926635e-01 9.87145603e-01
3.85315686e-01 -3.03042382e-01 3.94693047e-01 7.63117373e-01
6.30175173e-01 3.72566640e-01 -1.09907866e+00 1.15510666e+00
1.21091954e-01 -1.41597176e+00 6.17777824e-01 -3.30567509e-01
7.38373160e-01 -4.10311103e-01 1.50550291e-01 5.80630839e-01
-1.62827242e-02 -9.21363056e-01 9.33387399e-01 3.48330885e-01
1.10247684e+00 -4.46198523e-01 6.03132248e-01 3.40686679e-01
-1.30559862e+00 -1.06683001e-01 -2.08935127e-01 -6.97509274e-02
4.00983542e-01 3.52463514e-01 -4.04541194e-01 6.97344244e-01
5.95109444e-03 6.49436474e-01 -5.25204599e-01 9.57019508e-01
-6.48759007e-01 5.27963936e-01 -1.17660671e-01 -4.27646756e-01
4.35558200e-01 -4.40582186e-02 3.45920384e-01 1.03498125e+00
3.59843016e-01 2.76519120e-01 4.55364108e-01 1.22697186e+00
-1.79136768e-01 1.20729385e-02 -3.36584270e-01 -1.09514147e-01
7.34264135e-01 1.18414009e+00 -4.62102324e-01 -3.27433497e-01
-4.88611698e-01 1.00473392e+00 5.70255369e-02 4.43836600e-01
-1.12372112e+00 -6.22107327e-01 3.26397717e-01 8.66988748e-02
2.78002441e-01 -2.06128418e-01 -7.51993284e-02 -1.49244356e+00
3.44912291e-01 -1.18936384e+00 7.52894357e-02 -1.21872020e+00
-1.29916298e+00 7.41574228e-01 2.41126984e-01 -1.45316422e+00
-7.41666332e-02 -4.71092463e-01 -1.03018630e+00 7.55055189e-01
-1.66929758e+00 -1.27895832e+00 -4.18316156e-01 5.06515265e-01
8.54803205e-01 7.55303875e-02 4.50376600e-01 2.64026552e-01
-7.72147715e-01 7.13972032e-01 -6.41989470e-01 5.72072268e-02
5.92765033e-01 -1.10949099e+00 6.05860353e-01 9.57778096e-01
-1.06829472e-01 7.05332041e-01 5.58507383e-01 -3.80611748e-01
-1.48270738e+00 -1.28708351e+00 7.52792478e-01 -4.07878682e-02
4.60776359e-01 -6.08069599e-01 -8.89613748e-01 4.28382218e-01
4.13503945e-01 -8.67489353e-02 2.18177140e-01 -4.76047426e-01
-2.98089415e-01 -2.32946128e-01 -6.87968850e-01 1.01837254e+00
1.17610586e+00 -2.92475671e-01 -5.02740264e-01 3.32361639e-01
1.54566526e+00 -4.48457211e-01 -3.16792428e-01 5.29106379e-01
1.37203470e-01 -6.89866364e-01 6.99088037e-01 -2.94833243e-01
7.67520964e-01 -8.07133794e-01 5.57591580e-02 -1.24798155e+00
-4.21446115e-01 -8.94109130e-01 3.34864080e-01 1.61728132e+00
4.92393196e-01 -4.07458186e-01 4.99929160e-01 5.28936923e-01
-6.22440755e-01 -9.41770852e-01 -5.06878793e-01 -7.20468938e-01
-1.35355443e-01 -4.15166348e-01 1.10299170e+00 7.76986241e-01
-1.45252958e-01 6.98184967e-01 -6.19528055e-01 3.85551415e-02
3.49869691e-02 2.88982660e-01 8.71584058e-01 -5.43201804e-01
-4.03530210e-01 -4.46092278e-01 -1.11682460e-01 -1.85805869e+00
5.76721020e-02 -7.71119237e-01 3.50627482e-01 -1.64522088e+00
1.87512204e-01 -5.51635861e-01 -3.86177860e-02 7.12099195e-01
-5.97301960e-01 -4.18329574e-02 5.22566617e-01 1.53769851e-01
-6.14880383e-01 9.56210196e-01 1.85111701e+00 -2.55758792e-01
-3.17669436e-02 -3.58322740e-01 -1.21534288e+00 3.70395541e-01
6.05856001e-01 7.35775055e-03 -8.46283853e-01 -8.46530020e-01
1.24513187e-01 2.38049537e-01 3.38898927e-01 -8.40105951e-01
2.06957877e-01 -5.80083966e-01 -1.60816893e-01 -4.04853851e-01
1.43952528e-02 -4.94173020e-01 -1.16790242e-01 1.59153521e-01
-5.75780928e-01 1.52180478e-01 1.04312925e-02 5.07468104e-01
-4.11755919e-01 5.42525807e-03 4.38556701e-01 -1.48637578e-01
-8.18587065e-01 5.52631736e-01 -7.11712539e-02 2.51275510e-01
9.42085624e-01 1.90330788e-01 -5.11865139e-01 -6.32549942e-01
-2.72460550e-01 8.74558210e-01 3.04033399e-01 7.42639124e-01
7.39743590e-01 -1.64551568e+00 -8.59587908e-01 2.42580444e-01
4.11664546e-01 4.01681304e-01 2.68905729e-01 6.16253674e-01
-3.10538530e-01 4.30514723e-01 8.69668946e-02 -6.63035393e-01
-6.22238576e-01 8.38031471e-01 1.88370824e-01 -2.58197516e-01
-4.51878875e-01 6.82938159e-01 6.68311834e-01 -2.62278914e-01
5.43140396e-02 -4.99226153e-01 -1.83675900e-01 -4.00721371e-01
6.00520372e-01 -4.17292804e-01 -4.11532432e-01 -3.66486758e-01
-5.43282107e-02 4.73033875e-01 -1.87319517e-01 -7.00551644e-02
9.89741027e-01 -1.50596589e-01 -6.70760199e-02 6.46347851e-02
1.01748943e+00 -2.80199051e-01 -1.40338147e+00 -9.95099023e-02
-2.10617676e-01 -3.60578984e-01 -3.36847961e-01 -8.01694632e-01
-1.14803851e+00 9.25129294e-01 -5.93873067e-03 1.13578357e-01
1.40844393e+00 -1.04773611e-01 9.11551356e-01 1.67689115e-01
3.11794758e-01 -6.11837327e-01 3.52477729e-01 3.80336642e-01
1.28492451e+00 -1.08102214e+00 -4.81094003e-01 -5.62689245e-01
-8.73854697e-01 8.85013402e-01 1.13147616e+00 -4.11818502e-03
2.31407750e-02 2.12657023e-02 1.06460918e-02 7.96642303e-02
-1.00613606e+00 -1.49591967e-01 1.20145179e-01 6.96139395e-01
3.14811945e-01 -4.94358130e-02 -6.25923514e-01 9.02713180e-01
1.60096120e-02 2.63452232e-01 6.51419222e-01 5.33390939e-01
-1.21443957e-01 -1.23294246e+00 -2.35067055e-01 3.08952332e-01
-4.76884395e-02 -3.33256334e-01 -2.38599211e-01 5.93669415e-01
2.47563601e-01 1.07984257e+00 -1.17301211e-01 -4.74213719e-01
5.35830796e-01 -2.88801014e-01 1.36604920e-01 -7.76701033e-01
-2.85411716e-01 1.49050653e-01 -3.24059613e-02 -3.54595006e-01
-3.60075474e-01 -3.52063358e-01 -1.17278087e+00 1.30157843e-01
-3.64525199e-01 1.80007502e-01 1.48592740e-01 8.23121488e-01
5.41223586e-01 6.18799567e-01 8.40193570e-01 -7.22939789e-01
-6.46286190e-01 -1.01228285e+00 -3.53586264e-02 5.08583248e-01
2.09570095e-01 -5.00910282e-01 -9.96267423e-02 5.27480394e-02] | [10.883955001831055, 0.9230377078056335] |
77542786-3730-42e6-92b3-f0de35ca6b47 | nfresnet-multi-scale-and-u-shaped-networks | 2212.05909 | null | https://arxiv.org/abs/2212.05909v1 | https://arxiv.org/pdf/2212.05909v1.pdf | NFResNet: Multi-scale and U-shaped Networks for Deblurring | Multi-Scale and U-shaped Networks are widely used in various image restoration problems, including deblurring. Keeping in mind the wide range of applications, we present a comparison of these architectures and their effects on image deblurring. We also introduce a new block called as NFResblock. It consists of a Fast Fourier Transformation layer and a series of modified Non-Linear Activation Free Blocks. Based on these architectures and additions, we introduce NFResnet and NFResnet+, which are modified multi-scale and U-Net architectures, respectively. We also use three different loss functions to train these architectures: Charbonnier Loss, Edge Loss, and Frequency Reconstruction Loss. Extensive experiments on the Deep Video Deblurring dataset, along with ablation studies for each component, have been presented in this paper. The proposed architectures achieve a considerable increase in Peak Signal to Noise (PSNR) ratio and Structural Similarity Index (SSIM) value. | ['Aarya Makwana', 'Esha Pahwa', 'Preyansh Agrawal', 'Tanish Mittal'] | 2022-12-12 | null | null | null | null | ['deblurring'] | ['computer-vision'] | [ 3.77994686e-01 -4.97662634e-01 1.31242722e-01 -8.49645659e-02
-5.35154581e-01 -4.07482907e-02 4.14513707e-01 -5.20780623e-01
-3.67242128e-01 6.96525395e-01 5.83467305e-01 -1.04217075e-01
1.20248817e-01 -4.15543079e-01 -7.53418505e-01 -9.21680689e-01
-2.42279708e-01 -7.05644190e-01 3.15485567e-01 -2.77671248e-01
2.91939974e-01 5.15742123e-01 -1.06814909e+00 2.12036744e-01
1.00709748e+00 1.05062366e+00 2.53887266e-01 6.12172961e-01
5.57882309e-01 1.00195372e+00 -5.14315486e-01 -2.78876871e-01
3.09345841e-01 -5.90210855e-01 -6.61810338e-01 -7.29093775e-02
7.13141382e-01 -9.30394292e-01 -1.10601151e+00 1.22179890e+00
9.76803958e-01 4.37124968e-01 5.37157297e-01 -7.24830568e-01
-1.22552109e+00 4.61447120e-01 -9.25430238e-01 8.99773717e-01
1.00947879e-01 3.05221118e-02 4.82389688e-01 -9.02303100e-01
4.09083009e-01 1.17136478e+00 1.12910736e+00 4.25578535e-01
-1.09929466e+00 -6.70767963e-01 -2.14214578e-01 8.41239631e-01
-1.34305918e+00 -5.63011885e-01 8.37032795e-01 -1.38852432e-01
9.11680758e-01 2.17332557e-01 3.43407989e-01 9.86997128e-01
5.81256032e-01 7.76376963e-01 1.06530797e+00 -3.95936489e-01
4.00120858e-03 -5.14822483e-01 1.38933837e-01 3.00155610e-01
1.49825215e-01 4.17035580e-01 -2.47764990e-01 8.12606514e-02
1.19077182e+00 -5.99537529e-02 -1.08231604e+00 -1.54123574e-01
-1.20372009e+00 4.49037135e-01 8.30742538e-01 3.32852840e-01
-4.08050537e-01 4.35350537e-01 4.36970770e-01 4.83776480e-01
5.21243930e-01 1.46756753e-01 -1.73341885e-01 3.27223659e-01
-1.06275654e+00 2.34399736e-02 1.05155118e-01 5.87920725e-01
4.75039244e-01 4.36124891e-01 -1.94662899e-01 1.26100993e+00
1.44069731e-01 2.95095742e-01 9.42942381e-01 -8.80109251e-01
1.25300691e-01 -1.05031624e-01 1.02484629e-01 -1.27412510e+00
-3.29289675e-01 -6.06779754e-01 -1.62461698e+00 1.94347441e-01
-1.57061021e-03 -1.32399768e-01 -9.99817371e-01 1.67396295e+00
-2.20870301e-02 8.52977991e-01 -2.75317137e-03 1.32683921e+00
8.42454910e-01 6.83015525e-01 -1.49966851e-01 -2.15458080e-01
1.16714001e+00 -1.20358336e+00 -1.07615137e+00 -1.40725523e-01
2.52826661e-01 -7.74710238e-01 7.98317730e-01 3.15073937e-01
-1.33799815e+00 -7.00532436e-01 -1.33977222e+00 -3.52983713e-01
6.76867664e-02 1.97076395e-01 3.24216008e-01 5.24903595e-01
-1.50518954e+00 1.05749273e+00 -7.74178624e-01 -1.32079542e-01
4.63320315e-01 1.94425121e-01 -2.65443802e-01 -1.48829192e-01
-1.42371082e+00 9.81994808e-01 4.03304733e-02 4.50496972e-01
-1.06563270e+00 -5.84705174e-01 -8.45177352e-01 1.44308686e-01
-3.59140448e-02 -7.60572135e-01 8.53264213e-01 -1.00896871e+00
-1.45496953e+00 6.41160309e-01 1.91407830e-01 -5.15222251e-01
4.45385545e-01 -4.49169546e-01 -7.01297045e-01 1.90331295e-01
-1.97613969e-01 6.01166010e-01 1.19901860e+00 -1.01900196e+00
-2.11503819e-01 -1.62049741e-01 -1.72360942e-01 1.33734882e-01
-2.83724308e-01 3.35864842e-01 -2.51439452e-01 -1.60952365e+00
9.91826206e-02 -6.24137342e-01 -1.42011076e-01 7.68917501e-02
-1.39845580e-01 3.02401811e-01 9.20684814e-01 -1.41307008e+00
1.43125105e+00 -2.18092322e+00 3.85738820e-01 -2.60843039e-01
2.63734877e-01 4.88943905e-01 -7.43933797e-01 -3.08114439e-02
-7.12019861e-01 7.11722523e-02 -5.24594605e-01 -8.23017359e-02
-5.41740596e-01 -2.43382394e-01 -2.99582809e-01 9.66946304e-01
1.02469176e-02 8.77902031e-01 -6.27136528e-01 1.40195385e-01
2.26312190e-01 8.95747304e-01 -5.45695245e-01 3.53715420e-02
3.26821417e-01 2.85777539e-01 -7.86467455e-03 4.65208143e-01
1.08795154e+00 -2.09369555e-01 -2.01184273e-01 -6.39010787e-01
1.07340127e-01 -2.17787623e-02 -9.61845100e-01 1.76762891e+00
-4.96921927e-01 1.01789010e+00 3.34100634e-01 -9.12581682e-01
5.08235455e-01 4.39880282e-01 3.30984592e-01 -6.28506541e-01
2.63254315e-01 2.92431146e-01 -4.06865366e-02 -4.99884188e-01
6.17885411e-01 -7.03526810e-02 6.02196872e-01 3.08462501e-01
2.71453802e-02 2.23500833e-01 2.42152438e-02 -4.93304851e-03
9.84812975e-01 -8.37281998e-03 1.54280514e-01 -5.24372935e-01
8.50690782e-01 -5.68967104e-01 4.11189348e-01 4.52693552e-01
-3.93776864e-01 1.11432445e+00 2.43541777e-01 -5.87460279e-01
-1.24465251e+00 -8.40011179e-01 -1.03524499e-01 6.49367034e-01
5.86546004e-01 -1.65222481e-01 -7.50592113e-01 -4.38485563e-01
-5.00873625e-01 4.60681736e-01 -5.30409813e-01 -3.01901281e-01
-9.59428549e-01 -1.15281546e+00 4.55177188e-01 4.44422662e-01
9.50754583e-01 -1.04438567e+00 -1.89194039e-01 1.31798998e-01
-5.57835102e-01 -9.80096161e-01 -1.11718464e+00 -2.29613110e-01
-1.04108334e+00 -1.02056563e+00 -1.49547839e+00 -1.07611167e+00
5.92605054e-01 7.78245211e-01 9.98079360e-01 9.65520963e-02
-1.42065883e-01 -2.29552034e-02 -4.09731388e-01 4.26921636e-01
-5.04412711e-01 -2.37517685e-01 -6.36103749e-02 -6.40788898e-02
-3.14473242e-01 -9.95277703e-01 -1.16347361e+00 6.57958269e-01
-1.36283600e+00 -9.06284992e-03 6.49141848e-01 1.13027000e+00
3.41982841e-01 1.10965706e-01 6.86688066e-01 -1.68685228e-01
8.52884412e-01 -2.21238092e-01 -3.66312891e-01 1.40916720e-01
-5.50743818e-01 -2.36412883e-01 5.01116872e-01 -6.10901654e-01
-8.82343352e-01 -3.89244944e-01 -2.37992138e-01 -7.02337861e-01
1.99004531e-01 3.67451102e-01 -5.69143193e-03 -5.31609654e-01
8.18648577e-01 5.10889649e-01 1.05261572e-01 -5.95948756e-01
3.57840121e-01 7.46815026e-01 7.99772561e-01 -3.53323221e-02
6.25915825e-01 3.19125175e-01 -2.64398217e-01 -8.70222986e-01
-3.51771623e-01 -1.84079960e-01 -1.98856413e-01 -1.39738053e-01
6.74736381e-01 -1.15276659e+00 -3.49877775e-01 1.37922561e+00
-1.32389390e+00 -2.92626381e-01 -5.77417202e-02 5.18610358e-01
-3.78224164e-01 1.03049564e+00 -1.05267835e+00 -9.83040258e-02
-6.78446233e-01 -1.31281722e+00 5.48321128e-01 3.30464542e-01
3.85836929e-01 -9.28780496e-01 -8.85168314e-02 2.00665355e-01
1.01257718e+00 -6.19348232e-03 7.40505874e-01 -1.11228041e-01
-3.93288165e-01 -1.04791038e-01 -6.90379381e-01 1.04170990e+00
2.33146071e-01 -6.67722106e-01 -7.34938920e-01 -8.62478435e-01
3.97176147e-01 -1.62202179e-01 1.23186159e+00 8.58470380e-01
1.27964509e+00 -3.27441216e-01 -4.90647592e-02 9.76502955e-01
1.33741021e+00 9.35164317e-02 1.33225644e+00 4.42833155e-01
6.67302132e-01 4.64601777e-02 -2.83888727e-01 6.71482906e-02
1.11518309e-01 8.34421873e-01 4.17992145e-01 -2.85263926e-01
-9.16917562e-01 3.41821849e-01 5.96394718e-01 9.34657693e-01
-1.69102624e-01 -4.03464019e-01 -5.43468595e-01 5.21413982e-01
-1.56004941e+00 -9.94808555e-01 -2.74442360e-02 1.99812746e+00
9.39277589e-01 -3.84881049e-01 -2.74885476e-01 -7.41421729e-02
1.14057958e+00 5.13317585e-01 -7.04157352e-01 3.39959264e-02
-5.33903539e-01 5.90004101e-02 6.14925504e-01 6.54238343e-01
-1.32675791e+00 7.86719739e-01 6.67158651e+00 1.14697647e+00
-1.26912928e+00 3.18883955e-01 1.05727828e+00 2.25197241e-01
-3.74631025e-02 -3.42137516e-01 -1.62438750e-01 3.80572319e-01
7.69871950e-01 1.04407690e-01 8.66230845e-01 3.18482786e-01
3.43127429e-01 -1.78011395e-02 -6.64860725e-01 1.16434836e+00
2.46343896e-01 -1.37396622e+00 1.05843686e-01 -4.11529094e-01
9.77059960e-01 1.85967281e-01 7.82680213e-02 -1.09008156e-01
8.28394368e-02 -1.09512925e+00 5.92401862e-01 2.75079668e-01
1.11813939e+00 -5.07900834e-01 7.80247569e-01 -1.50972307e-01
-8.95722389e-01 -5.52631989e-02 -3.40018868e-01 2.60662109e-01
3.62108648e-01 6.23655200e-01 1.88906103e-01 6.87734008e-01
8.16705823e-01 1.04874837e+00 -2.59069413e-01 1.42922473e+00
-2.49756753e-01 5.09858191e-01 -2.73150653e-02 6.15225554e-01
-4.09598351e-02 -2.03546152e-01 8.30737352e-01 1.11997890e+00
5.71668923e-01 1.26320362e-01 -5.65533698e-01 7.50505030e-01
-2.95278281e-01 -2.27523327e-01 1.61649019e-01 4.81527537e-01
9.85388756e-02 1.04574239e+00 -4.32426423e-01 -3.25160325e-01
-3.40926170e-01 1.46420109e+00 -2.88622171e-01 6.13396108e-01
-8.72891545e-01 -6.59593165e-01 8.63972425e-01 1.06268495e-01
3.08134407e-01 -1.57273158e-01 -5.36826029e-02 -1.48334336e+00
-5.95951974e-02 -1.13656139e+00 -1.12304715e-02 -1.01088619e+00
-1.33787954e+00 8.30968440e-01 -1.59077138e-01 -1.32054174e+00
1.76676199e-01 -3.27064484e-01 -5.15447021e-01 1.19739056e+00
-2.00260997e+00 -7.83358574e-01 -4.79730725e-01 5.03378928e-01
6.27945721e-01 -1.67568102e-01 2.99052924e-01 7.30786562e-01
-7.32246459e-01 6.75105691e-01 4.67839718e-01 2.18117431e-01
8.86208892e-01 -6.78969145e-01 8.70159566e-01 1.21851480e+00
-4.27017093e-01 4.85190958e-01 7.30249107e-01 -5.68922937e-01
-1.15724361e+00 -1.10233283e+00 3.73172432e-01 3.05146664e-01
5.52913129e-01 2.79130816e-01 -1.20875573e+00 5.15553832e-01
5.46711683e-01 3.90272230e-01 7.35364854e-02 -6.41801953e-01
-3.81600797e-01 -2.23200731e-02 -1.28794384e+00 5.77666283e-01
9.78883445e-01 -3.19810569e-01 -3.08666199e-01 1.59419358e-01
6.65116608e-01 -5.28329194e-01 -7.09265590e-01 6.76199138e-01
4.51819241e-01 -1.33189094e+00 1.32453012e+00 -2.61896998e-01
7.80285954e-01 -3.20434064e-01 -2.85478998e-02 -1.63207817e+00
-8.23621511e-01 -8.31348777e-01 -1.99913874e-01 7.71852612e-01
-1.00162297e-01 -6.64350271e-01 4.48451072e-01 -1.15334112e-02
-4.56007957e-01 -6.67492986e-01 -1.00857866e+00 -7.78897285e-01
3.00628841e-01 6.59097731e-02 2.66183704e-01 9.81787503e-01
-4.22432691e-01 1.06629781e-01 -9.45019424e-01 2.35425368e-01
6.83371902e-01 -3.92147392e-01 1.79624274e-01 -5.79515100e-01
-1.27692610e-01 -5.51880717e-01 -3.62818897e-01 -1.66485405e+00
-5.01663126e-02 -6.67193234e-01 -3.66233289e-02 -1.47916770e+00
2.72439450e-01 -1.08481333e-01 -4.58252192e-01 2.14441434e-01
-3.24082881e-01 6.47275865e-01 -7.19583035e-02 5.41149020e-01
-2.53652688e-02 9.54142988e-01 1.52168846e+00 -2.62528270e-01
-1.78518027e-01 -1.50518939e-01 -5.26286244e-01 6.76246822e-01
6.79006577e-01 -8.32575560e-02 -2.63916731e-01 -8.33400905e-01
-6.84636384e-02 2.26466835e-01 4.66831267e-01 -1.15952373e+00
1.30785629e-01 1.87886134e-01 4.11395818e-01 -2.52227157e-01
1.87081963e-01 -7.02505946e-01 2.74596632e-01 4.08660442e-01
-3.40230078e-01 2.14732900e-01 3.35481077e-01 5.15465677e-01
-4.13109660e-01 -3.11630480e-02 1.38592815e+00 1.09628402e-01
-6.60487354e-01 1.79204419e-01 -4.00053680e-01 -3.13637525e-01
8.02702487e-01 -3.54286402e-01 -6.04363620e-01 -5.48831701e-01
-6.11803472e-01 -1.55427724e-01 3.53636354e-01 4.54424530e-01
8.97022128e-01 -1.50796425e+00 -1.10875845e+00 3.36028725e-01
-5.27564466e-01 -3.95869344e-01 8.62209797e-01 1.16934085e+00
-8.55747819e-01 2.08878815e-01 -6.17100477e-01 -8.23898017e-02
-1.09646499e+00 4.72156197e-01 7.54314065e-01 -1.54807732e-01
-8.67390871e-01 8.97421420e-01 4.80315208e-01 4.81791086e-02
1.71723679e-01 -2.73484260e-01 -1.38739765e-01 -2.51262456e-01
9.12657261e-01 8.12307894e-01 1.14561498e-01 -7.50576615e-01
-1.52074650e-01 6.18216157e-01 -2.26579502e-01 2.16139406e-01
1.37070823e+00 -5.46150029e-01 -5.03233075e-01 -4.28658128e-01
1.27395678e+00 -1.35673836e-01 -1.49551404e+00 -3.43415320e-01
-4.29882586e-01 -6.26060665e-01 6.47617579e-01 -9.50335503e-01
-1.73292410e+00 7.21018970e-01 1.19188833e+00 1.57551689e-03
1.72574532e+00 -3.96301389e-01 1.28911376e+00 -2.64900714e-01
-6.55365065e-02 -4.79223847e-01 1.63130388e-01 4.22204882e-01
1.44565821e+00 -1.02045214e+00 1.10600002e-01 -2.76858628e-01
-4.14116122e-02 1.09556615e+00 4.71159518e-01 -4.40663695e-01
6.15176797e-01 2.31792152e-01 1.68271065e-01 3.37666005e-01
-2.93330401e-01 3.52515668e-01 3.45300376e-01 5.08032143e-01
4.61803466e-01 -4.41751122e-01 -5.84897161e-01 3.34216326e-01
2.48699963e-01 2.57639468e-01 8.01225483e-01 3.56387675e-01
-3.83321643e-01 -7.34771550e-01 -6.35498285e-01 2.51204342e-01
-7.74961948e-01 -4.85980958e-01 2.14095756e-01 2.92338997e-01
-1.17240384e-01 8.37634087e-01 -1.32704809e-01 -5.37834823e-01
1.78577989e-01 -6.44891322e-01 4.43079114e-01 1.47678137e-01
-5.12163758e-01 1.72389075e-01 -2.68645525e-01 -5.81811607e-01
-5.52032173e-01 -2.50346810e-01 -7.30220497e-01 -5.19442916e-01
-5.57537675e-01 -2.97452480e-01 4.85004574e-01 4.41022843e-01
5.43545008e-01 7.50174165e-01 5.20360231e-01 -1.27291787e+00
-6.77230775e-01 -1.25724137e+00 -5.88991642e-01 1.61610276e-01
8.81583333e-01 -3.88991475e-01 -4.95921075e-01 3.63274485e-01] | [11.44458293914795, -2.4874396324157715] |
66e2ade7-aea3-437a-b09b-e80e7cb0414b | syntax-aware-hybrid-prompt-model-for-few-shot | 2306.01312 | null | https://arxiv.org/abs/2306.01312v1 | https://arxiv.org/pdf/2306.01312v1.pdf | Syntax-aware Hybrid prompt model for Few-shot multi-modal sentiment analysis | Multimodal Sentiment Analysis (MSA) has been a popular topic in natural language processing nowadays, at both sentence and aspect level. However, the existing approaches almost require large-size labeled datasets, which bring about large consumption of time and resources. Therefore, it is practical to explore the method for few-shot sentiment analysis in cross-modalities. Previous works generally execute on textual modality, using the prompt-based methods, mainly two types: hand-crafted prompts and learnable prompts. The existing approach in few-shot multi-modality sentiment analysis task has utilized both methods, separately. We further design a hybrid pattern that can combine one or more fixed hand-crafted prompts and learnable prompts and utilize the attention mechanisms to optimize the prompt encoder. The experiments on both sentence-level and aspect-level datasets prove that we get a significant outperformance. | ['Zikai Zhou'] | 2023-06-02 | null | null | null | null | ['multimodal-sentiment-analysis', 'sentiment-analysis', 'multimodal-sentiment-analysis'] | ['computer-vision', 'natural-language-processing', 'natural-language-processing'] | [ 2.06905231e-01 -1.89592123e-01 -1.86659947e-01 -5.81809163e-01
-1.22890401e+00 -5.59618473e-01 5.85643768e-01 1.29616439e-01
-6.43802881e-01 4.63343322e-01 4.19466525e-01 -2.34759733e-01
2.38300219e-01 -5.87314367e-01 -3.32782924e-01 -6.71148300e-01
5.62246323e-01 -3.27836201e-02 2.46761620e-01 -5.50961614e-01
4.10568833e-01 -4.20735508e-01 -1.50010872e+00 5.31995714e-01
7.12269545e-01 8.77700865e-01 2.03227520e-01 6.74646080e-01
-7.53314734e-01 9.85831380e-01 -3.70975792e-01 -6.69641495e-01
3.18796672e-02 -5.34929752e-01 -7.35653162e-01 -6.13773800e-03
-2.21512526e-01 -8.17733333e-02 1.78712919e-01 1.01013994e+00
9.69442487e-01 3.22794825e-01 4.04905498e-01 -1.25655234e+00
-6.52963936e-01 7.43509293e-01 -1.04141092e+00 2.85674274e-01
6.39036477e-01 2.23309040e-01 1.09798336e+00 -8.62900972e-01
3.41002285e-01 1.23769987e+00 4.63448524e-01 6.01194918e-01
-5.94839573e-01 -4.75973189e-01 4.93135959e-01 3.42806220e-01
-9.32995081e-01 -5.71909428e-01 1.07396078e+00 -2.75234580e-01
8.73426795e-01 5.92529625e-02 2.53306508e-01 1.20083332e+00
9.31639299e-02 1.17470443e+00 1.12783098e+00 -7.31328905e-01
1.11392118e-01 2.86432594e-01 4.08499300e-01 5.53465486e-01
-3.60757709e-01 -4.29030329e-01 -7.34736502e-01 7.29106441e-02
1.39549300e-01 2.78615206e-01 -1.23365745e-01 1.84703767e-01
-1.26837826e+00 7.74085402e-01 -1.67951092e-01 3.64428759e-01
-2.80731261e-01 -2.46242628e-01 9.71794724e-01 3.89103413e-01
5.64701080e-01 2.47098342e-01 -8.19849491e-01 -6.08892322e-01
-7.94844389e-01 -1.09076284e-01 5.32262266e-01 1.00427377e+00
7.26747215e-01 -1.37183219e-01 -5.29867351e-01 8.85932565e-01
3.91370118e-01 5.97475648e-01 8.06904554e-01 -1.64213598e-01
9.07455623e-01 9.25232708e-01 6.55676275e-02 -1.02035213e+00
-5.18475294e-01 1.59791753e-01 -5.56279898e-01 -3.31564456e-01
8.15922320e-02 -7.04235137e-01 -7.59937644e-01 1.75106919e+00
5.23591816e-01 -7.44392201e-02 2.31847271e-01 1.02316737e+00
1.15188003e+00 8.70154560e-01 3.07378054e-01 -4.13936913e-01
1.90972281e+00 -1.33662498e+00 -1.12558234e+00 -1.39103040e-01
7.88806677e-01 -1.00114179e+00 1.80912709e+00 3.27099949e-01
-9.27470863e-01 -4.68782604e-01 -1.01540875e+00 -2.95673370e-01
-6.33448422e-01 1.69717744e-01 5.47410131e-01 7.36583352e-01
-5.99812210e-01 1.81730941e-01 -6.07515633e-01 -3.31006676e-01
2.85944402e-01 1.67767450e-01 -2.32026175e-01 -1.08929933e-03
-1.42528129e+00 6.84020400e-01 4.47890908e-02 2.32530445e-01
-6.33268356e-01 -4.04232234e-01 -8.21136653e-01 1.22259922e-01
7.19899893e-01 -3.73296916e-01 1.52138627e+00 -1.21837831e+00
-1.77583158e+00 6.40461862e-01 -3.86628062e-01 2.13611126e-01
2.63555720e-02 -2.62860894e-01 -5.77564955e-01 1.52962789e-01
1.08653679e-01 3.86708647e-01 8.35782588e-01 -9.14065778e-01
-7.19415903e-01 -3.21630746e-01 4.75636929e-01 4.10665631e-01
-1.01298916e+00 5.48003614e-01 -4.26231593e-01 -4.21339482e-01
-2.67675668e-01 -5.20263433e-01 -2.80814588e-01 -5.44064760e-01
-5.43610811e-01 -4.79185432e-01 7.96052337e-01 -6.17639780e-01
1.47838855e+00 -2.11833668e+00 1.64877772e-01 -3.46941501e-01
-1.47695333e-01 1.72483891e-01 -2.99664259e-01 6.37918890e-01
1.91139672e-02 6.78719953e-02 -7.64890537e-02 -7.13393152e-01
2.08050162e-01 -1.33407205e-01 -3.24463010e-01 2.85015106e-02
2.35427707e-01 9.21970844e-01 -1.01433420e+00 -7.85772622e-01
1.00706637e-01 1.90622658e-01 -2.31507167e-01 3.87241453e-01
-2.46762395e-01 4.57644850e-01 -7.73875356e-01 8.90688539e-01
5.00851274e-01 -2.67267168e-01 -1.57594323e-01 -3.54947895e-01
-2.17351899e-01 7.26073086e-02 -1.00352800e+00 1.99984777e+00
-6.94009185e-01 2.53125042e-01 -1.72701493e-01 -8.55015457e-01
7.25268960e-01 7.59502649e-01 3.67755115e-01 -6.44982338e-01
5.86178184e-01 6.40952811e-02 -2.83314168e-01 -1.18119228e+00
6.77674353e-01 -4.28994954e-01 -4.47152525e-01 5.98406374e-01
2.26094380e-01 3.07834536e-01 3.22149575e-01 2.72523820e-01
1.07612860e+00 2.92734265e-01 2.47504458e-01 6.63529634e-02
9.03009415e-01 7.66251609e-02 3.87278527e-01 4.86654550e-01
-2.77571172e-01 7.08628118e-01 8.60798359e-01 -3.89753014e-01
-8.26149821e-01 -2.14967325e-01 3.16370726e-01 1.60853899e+00
3.38457346e-01 -6.15997672e-01 -7.02536702e-01 -9.36467946e-01
-6.20637655e-01 5.12653172e-01 -5.74939609e-01 -1.63050830e-01
-1.32783547e-01 -8.66435349e-01 3.51620525e-01 5.29743373e-01
3.98934871e-01 -1.26513088e+00 -5.52458704e-01 2.25077555e-01
-4.07128066e-01 -1.18498969e+00 -5.39981306e-01 1.72817767e-01
-6.18797660e-01 -8.13496351e-01 -7.59788334e-01 -6.76765203e-01
6.30488634e-01 4.34098065e-01 7.80380607e-01 -3.59312855e-02
3.47595438e-02 4.99274403e-01 -1.04654527e+00 -6.26203477e-01
2.90571928e-01 2.68800706e-01 -1.38568923e-01 4.22336698e-01
8.03549111e-01 -3.97053838e-01 -5.45390487e-01 3.95596996e-02
-9.33506846e-01 1.22083738e-01 7.31133461e-01 8.75713587e-01
3.03285331e-01 -1.92201763e-01 8.73074293e-01 -1.04650450e+00
1.00155592e+00 -5.85577250e-01 -2.45326445e-01 4.72220272e-01
-4.07678753e-01 -6.99832216e-02 9.47052717e-01 -4.80515897e-01
-1.37780881e+00 8.99185836e-02 -2.34229267e-01 -4.31586623e-01
-2.72638351e-01 9.22376633e-01 -3.61657649e-01 3.20567787e-01
2.71149784e-01 2.31686458e-01 -1.93729252e-01 -3.76813918e-01
4.72640127e-01 9.51767921e-01 8.64851347e-04 -6.03540301e-01
5.23766994e-01 2.88444459e-01 -4.97019529e-01 -5.08523464e-01
-1.03748429e+00 -6.97768569e-01 -3.39081794e-01 -3.86967391e-01
1.13718796e+00 -8.47427905e-01 -8.25777709e-01 3.61115366e-01
-1.27908313e+00 1.58746522e-02 -4.86514978e-02 3.97228271e-01
-3.54057401e-01 3.10083568e-01 -4.30305511e-01 -1.00498164e+00
-6.36328340e-01 -1.24630940e+00 1.48408127e+00 5.78302622e-01
-2.68433746e-02 -8.73217702e-01 2.25270972e-01 4.88262892e-01
3.15559268e-01 -3.76618467e-02 8.39897573e-01 -8.42087030e-01
-2.31124178e-01 -4.54526365e-01 -1.39811277e-01 3.86558548e-02
8.35469551e-03 -1.13835521e-02 -1.26242101e+00 -3.68444510e-02
1.34248927e-01 -7.46886432e-01 6.06821656e-01 1.28179893e-01
1.10784864e+00 -2.56898731e-01 -1.21169597e-01 1.16807468e-01
1.39116538e+00 2.68200219e-01 5.19905925e-01 2.93322116e-01
6.43173933e-01 7.24418819e-01 9.30759072e-01 5.90498149e-01
6.80458784e-01 2.90415525e-01 1.16735466e-01 -6.23993650e-02
3.28294277e-01 -1.50126770e-01 4.62690204e-01 1.31020808e+00
-1.67640373e-01 -4.06633407e-01 -7.21067429e-01 6.03850782e-01
-2.24242544e+00 -1.00436985e+00 1.41816124e-01 1.64646411e+00
9.40658033e-01 1.06753513e-01 1.00677274e-01 2.31515393e-01
7.13650882e-01 3.18277031e-01 -3.27560902e-01 -4.54649925e-01
-1.12787876e-02 -9.20721367e-02 -4.15944345e-02 2.74932355e-01
-1.05122173e+00 9.02135968e-01 5.35048532e+00 1.03934145e+00
-1.20696187e+00 3.45217407e-01 5.75864494e-01 -3.23907673e-01
-4.93731380e-01 7.06904978e-02 -8.75626624e-01 7.24914014e-01
9.28975403e-01 -8.94166157e-02 1.60834298e-01 7.44914532e-01
3.69329035e-01 -1.81252062e-01 -8.73257816e-01 1.04882884e+00
2.70372331e-01 -9.20340955e-01 -1.44736245e-01 -5.14880896e-01
5.87031901e-01 -3.86337966e-01 -9.56890583e-02 7.01787889e-01
-1.22128904e-01 -5.83411634e-01 5.00649929e-01 5.18986166e-01
5.41392148e-01 -7.02124417e-01 1.06663013e+00 4.44377244e-01
-1.27931118e+00 2.38939617e-02 -3.00171643e-01 -2.34638125e-01
6.39792919e-01 3.98962170e-01 -2.45322391e-01 7.76795149e-01
5.61173439e-01 8.12948346e-01 -4.96997565e-01 6.55533552e-01
-3.12142998e-01 6.17345631e-01 1.07078468e-02 -5.61160982e-01
3.34475130e-01 -1.95762232e-01 2.33189911e-01 1.11877561e+00
1.80868343e-01 3.13506782e-01 2.37726495e-01 1.88059300e-01
6.83915839e-02 5.70008159e-01 -3.66187781e-01 -3.21399927e-01
2.35910162e-01 1.85422409e+00 -8.08293104e-01 -4.54517663e-01
-1.02615714e+00 6.82842970e-01 1.70863286e-01 2.83716023e-01
-9.52331781e-01 -8.32087815e-01 -5.26236854e-02 -4.60247248e-01
3.11380148e-01 7.02845007e-02 -2.74119228e-01 -1.44855595e+00
1.43647254e-01 -1.08857703e+00 5.18877804e-01 -9.57365811e-01
-1.53117990e+00 6.99874043e-01 -1.67124003e-01 -1.42513263e+00
-3.06038596e-02 -5.91497540e-01 -8.48964393e-01 7.06806660e-01
-1.64215744e+00 -1.36534500e+00 -1.60972536e-01 7.23138452e-01
9.03301537e-01 -1.38224706e-01 8.44258606e-01 6.82617426e-01
-9.55443919e-01 6.04902446e-01 -2.87278295e-01 -2.96408832e-02
9.00220454e-01 -9.58350360e-01 -2.71735102e-01 8.12197566e-01
-1.25059649e-01 6.52731895e-01 6.81202769e-01 -3.28499556e-01
-1.73074651e+00 -5.41436911e-01 9.73106146e-01 -4.48579729e-01
7.45232403e-01 -3.20747674e-01 -7.26801395e-01 6.19826913e-01
7.71708429e-01 -2.69251704e-01 1.04477406e+00 3.68481934e-01
-2.55257413e-02 -2.55415142e-01 -9.46766436e-01 6.38862550e-01
5.64336300e-01 -5.34820974e-01 -7.55956054e-01 3.39838058e-01
1.01497054e+00 -2.96756268e-01 -6.54584706e-01 2.99726993e-01
4.63986427e-01 -6.53893650e-01 5.18021822e-01 -8.41360688e-01
9.89237249e-01 -4.02713120e-01 -6.83813542e-03 -1.18045735e+00
8.50871652e-02 -4.62136179e-01 -7.66036138e-02 1.79722476e+00
6.36813879e-01 -4.08057213e-01 5.17088234e-01 7.60332882e-01
-1.84885502e-01 -9.12472248e-01 -5.73712289e-01 -2.72628009e-01
-4.68610793e-01 -3.94171208e-01 6.14603281e-01 1.00870895e+00
5.85911334e-01 1.01979387e+00 -6.45911217e-01 -8.18180665e-02
1.46051915e-02 2.88607627e-01 8.05307269e-01 -7.25835979e-01
-1.28139570e-01 -3.33652973e-01 -3.64015345e-03 -1.01025558e+00
-4.90197055e-02 -4.54885602e-01 2.74811149e-01 -1.40163386e+00
3.81174594e-01 -7.39291534e-02 -4.94536847e-01 6.82116151e-01
-6.33132994e-01 -7.93768093e-02 4.39035818e-02 -5.10951057e-02
-1.17111337e+00 7.13679135e-01 1.35866082e+00 -2.70024687e-02
-1.85910389e-01 -2.15744019e-01 -9.84201252e-01 6.78992391e-01
7.48972237e-01 -3.95228863e-01 -6.50159061e-01 -6.04586303e-01
7.05110848e-01 1.22292168e-01 -2.37307131e-01 -5.57071745e-01
4.87856746e-01 -4.41997319e-01 -2.81766243e-02 -8.37316215e-01
3.73727620e-01 -8.12847018e-01 -5.17047346e-01 -5.84428012e-02
-4.40565139e-01 3.16042453e-01 1.58048004e-01 5.93699753e-01
-5.51597595e-01 -5.13357401e-01 3.03068131e-01 -2.74008870e-01
-9.29613352e-01 3.22621644e-01 -3.20556939e-01 8.63267407e-02
9.78759766e-01 4.72964346e-02 -3.92542452e-01 -3.66470218e-01
-2.98317045e-01 4.79639471e-01 -1.12045981e-01 6.19542062e-01
4.61805791e-01 -1.26163256e+00 -2.75256693e-01 5.57421148e-03
3.89027447e-01 -1.48361340e-01 6.32082582e-01 1.14771128e+00
-5.45187518e-02 2.40887314e-01 -3.52509767e-02 -3.38968009e-01
-1.11475575e+00 7.58309543e-01 -1.28865376e-01 -2.38086179e-01
-1.71864048e-01 9.03368771e-01 -1.34741053e-01 -5.43530226e-01
9.38201845e-02 2.32605282e-02 -8.05216610e-01 5.34877956e-01
7.88609564e-01 8.33371282e-02 -7.40984082e-02 -5.28057933e-01
-2.56032228e-01 7.23706782e-01 -8.07882994e-02 -3.13956618e-01
1.19797862e+00 -4.99984056e-01 -2.27161452e-01 9.30741370e-01
1.12104881e+00 2.13598944e-02 -8.59037399e-01 -3.52784991e-01
2.24177036e-02 -3.72591972e-01 -2.99784038e-02 -7.63755977e-01
-9.50920701e-01 1.12698996e+00 3.81290644e-01 4.64798927e-01
1.25879216e+00 -8.67267549e-02 1.08626449e+00 4.22367662e-01
3.76795560e-01 -1.27100575e+00 3.09372067e-01 4.54057068e-01
4.51497078e-01 -1.58867395e+00 -1.77027613e-01 -2.31016606e-01
-1.05286717e+00 1.08616328e+00 9.73387837e-01 2.10220978e-01
4.98928189e-01 1.84906185e-01 5.14732063e-01 -2.25700334e-01
-9.89657223e-01 -1.86192796e-01 1.28217161e-01 1.89892113e-01
7.41858304e-01 -1.82775840e-01 -7.31626034e-01 1.39118099e+00
1.60828039e-01 1.77741036e-01 4.20598924e-01 1.26490605e+00
-4.07505691e-01 -1.18211949e+00 -3.11261117e-01 4.50030327e-01
-5.54246008e-01 -3.53549391e-01 -2.30044592e-02 2.92652875e-01
1.41483903e-01 1.20608759e+00 -1.90618411e-01 -5.04994273e-01
4.85916018e-01 3.05902988e-01 2.33418554e-01 -6.73758686e-01
-7.73461342e-01 6.65410829e-04 3.83841507e-02 -3.89896959e-01
-8.90873015e-01 -5.41522861e-01 -1.02462363e+00 -1.18513919e-01
-7.10155427e-01 1.72850028e-01 5.07702708e-01 1.31873131e+00
3.10767174e-01 6.49880826e-01 8.00272465e-01 -6.74070299e-01
-4.55871493e-01 -1.21735418e+00 -4.23075855e-01 3.15157950e-01
2.59422600e-01 -5.33264577e-01 -2.33332455e-01 2.65311480e-01] | [12.984673500061035, 5.494215488433838] |
829e319c-34cb-42b3-9745-69f988165688 | improving-diffusion-based-image-translation | 2306.04396 | null | https://arxiv.org/abs/2306.04396v1 | https://arxiv.org/pdf/2306.04396v1.pdf | Improving Diffusion-based Image Translation using Asymmetric Gradient Guidance | Diffusion models have shown significant progress in image translation tasks recently. However, due to their stochastic nature, there's often a trade-off between style transformation and content preservation. Current strategies aim to disentangle style and content, preserving the source image's structure while successfully transitioning from a source to a target domain under text or one-shot image conditions. Yet, these methods often require computationally intense fine-tuning of diffusion models or additional neural networks. To address these challenges, here we present an approach that guides the reverse process of diffusion sampling by applying asymmetric gradient guidance. This results in quicker and more stable image manipulation for both text-guided and image-guided image translation. Our model's adaptability allows it to be implemented with both image- and latent-diffusion models. Experiments show that our method outperforms various state-of-the-art models in image translation tasks. | ['Jong Chul Ye', 'Gihyun Kwon'] | 2023-06-07 | null | null | null | null | ['image-manipulation'] | ['computer-vision'] | [ 6.96267009e-01 -1.06038839e-01 -3.38270158e-01 -2.27050737e-01
-6.39353335e-01 -7.22076595e-01 9.83272612e-01 -1.60259247e-01
-4.59061861e-01 5.15413821e-01 2.30182499e-01 -3.92987877e-01
3.15895259e-01 -6.25502884e-01 -4.98290718e-01 -5.92187464e-01
5.26537180e-01 4.08850104e-01 2.81116396e-01 -1.61579043e-01
4.34197366e-01 4.43694174e-01 -1.10956383e+00 2.97537684e-01
8.59948158e-01 6.15954518e-01 5.26710093e-01 7.27657735e-01
-3.96920383e-01 7.93977380e-01 -3.94764870e-01 -4.64895040e-01
3.06494445e-01 -9.36276555e-01 -8.56328309e-01 2.56031334e-01
6.27347708e-01 -3.82394493e-01 -3.26444954e-01 1.35506368e+00
5.84331274e-01 3.80005711e-03 6.98925376e-01 -1.03852391e+00
-1.30772853e+00 4.00776982e-01 -7.24903524e-01 2.46709377e-01
3.12399715e-01 3.89877647e-01 6.31858051e-01 -7.33135819e-01
1.11587310e+00 1.27138829e+00 2.86982566e-01 7.13414311e-01
-1.75850248e+00 -5.34796000e-01 1.25246540e-01 -4.00028452e-02
-9.32180285e-01 -6.85824156e-01 8.63830745e-01 -3.89164299e-01
7.31028438e-01 1.26532882e-01 7.03026056e-01 1.36219859e+00
4.23743725e-01 7.75210798e-01 1.52053034e+00 -5.16061723e-01
1.06935173e-01 3.25071126e-01 -5.31908989e-01 6.96376026e-01
-1.85473040e-02 1.30775213e-01 -6.40033364e-01 3.58587354e-02
1.02937984e+00 -2.54058242e-01 -2.88803607e-01 -6.97039664e-01
-1.55272758e+00 6.72942758e-01 2.08243325e-01 3.04055333e-01
-3.41180682e-01 1.13762923e-01 2.42258713e-01 7.15137839e-01
6.17770791e-01 5.47081590e-01 5.72959147e-02 -1.75942212e-01
-1.27928245e+00 3.12252253e-01 6.52203858e-01 1.00888503e+00
6.51597917e-01 5.19554727e-02 -3.97661626e-01 8.55263114e-01
4.85739931e-02 6.84910059e-01 5.57815194e-01 -9.84244764e-01
3.81269246e-01 3.46689761e-01 9.97796804e-02 -1.08267844e+00
8.34953338e-02 -2.44754374e-01 -8.42269540e-01 5.09241045e-01
4.99574631e-01 1.16912536e-01 -1.16159582e+00 1.87856567e+00
1.87596008e-01 -1.63152352e-01 -1.05375484e-01 8.69121194e-01
3.34762096e-01 6.08753562e-01 1.05458431e-01 -1.79490656e-01
1.20315790e+00 -1.22810519e+00 -7.62485325e-01 -3.23349118e-01
4.16268587e-01 -1.08692718e+00 1.42127657e+00 6.34305552e-02
-1.51088870e+00 -4.51665401e-01 -9.21647310e-01 -3.40272844e-01
-2.46094808e-01 2.79716775e-02 2.21022412e-01 5.85930109e-01
-1.29268539e+00 5.43634832e-01 -7.19684839e-01 -5.83442628e-01
5.57077765e-01 2.61421502e-01 -3.23257208e-01 -1.40299261e-01
-8.40771258e-01 9.40154374e-01 4.41264361e-02 -2.71643728e-01
-7.77584136e-01 -6.41939938e-01 -5.69364488e-01 -6.81155100e-02
2.07910731e-01 -1.05424821e+00 1.26666987e+00 -1.45237982e+00
-1.99424505e+00 1.16649151e+00 -1.90643460e-01 -2.86555022e-01
9.47544932e-01 -6.77833483e-02 -1.28992066e-01 1.94570050e-01
1.34485826e-01 1.07336426e+00 1.37636435e+00 -1.19936180e+00
-5.17531097e-01 -1.73583224e-01 -9.62350443e-02 3.99379253e-01
-4.96955991e-01 1.50740370e-01 -6.38588786e-01 -9.05631185e-01
-1.20760864e-02 -1.15848029e+00 -1.32352382e-01 5.39903104e-01
-2.42294610e-01 3.31345677e-01 1.01176453e+00 -4.73065704e-01
1.08905923e+00 -2.08490705e+00 5.69553018e-01 -5.92722557e-02
2.95152366e-01 3.18986773e-01 -5.06113589e-01 3.00911158e-01
1.44904152e-01 -1.03997551e-02 -3.28527868e-01 -5.76312184e-01
-1.38062447e-01 2.64087990e-02 -4.09453839e-01 3.33543658e-01
7.03398362e-02 1.14799953e+00 -9.77393746e-01 -5.87127328e-01
2.15682425e-02 4.36770260e-01 -5.87159276e-01 1.98478967e-01
-3.08568358e-01 7.12522626e-01 -2.41169944e-01 3.65281165e-01
5.76503158e-01 -2.78608918e-01 1.77996621e-01 -1.35133132e-01
-7.64666796e-02 1.19271830e-01 -8.79785836e-01 1.99916387e+00
-4.70844954e-01 8.50744903e-01 2.08397761e-01 -6.26460791e-01
7.60131896e-01 8.17957222e-02 2.52416432e-01 -1.06351399e+00
3.61259095e-02 3.29604983e-01 -6.80293292e-02 -3.25624466e-01
6.53392613e-01 -2.44330212e-01 3.28525960e-01 9.13414240e-01
-1.64624959e-01 -4.90448266e-01 1.96994543e-01 3.25112104e-01
7.18081772e-01 3.89488637e-01 -9.67043489e-02 -3.55225772e-01
3.69097441e-01 9.18829143e-02 2.64900506e-01 8.23597372e-01
-2.19766527e-01 8.05472493e-01 3.03827763e-01 -3.14932853e-01
-1.37859797e+00 -9.72055912e-01 3.91182333e-01 1.06037474e+00
1.91596821e-01 -5.95837906e-02 -1.04486144e+00 -6.38294160e-01
-2.63414264e-01 7.42552519e-01 -5.87959290e-01 -2.77846783e-01
-6.73958361e-01 -5.05307972e-01 5.35557210e-01 1.61930770e-01
5.55565774e-01 -1.07177806e+00 -6.74950838e-01 3.12801957e-01
-2.85191089e-01 -9.78298008e-01 -1.04320812e+00 -1.62016734e-01
-9.51477587e-01 -5.16792297e-01 -1.31194699e+00 -8.50114763e-01
8.73453617e-01 6.26864195e-01 1.07922637e+00 -1.13160066e-01
-1.14342988e-01 3.89208615e-01 -1.51468232e-01 -8.88586566e-02
-9.14623022e-01 3.89452875e-01 -1.44582927e-01 6.53695539e-02
7.98319560e-03 -5.99220395e-01 -8.02313566e-01 3.85918349e-01
-1.31712627e+00 4.14035618e-01 6.69419944e-01 8.81291628e-01
5.44181764e-01 -2.66314715e-01 2.07103819e-01 -9.16828752e-01
1.11170793e+00 -2.60710008e-02 -6.31825149e-01 3.73627305e-01
-9.35221195e-01 3.09530735e-01 4.34675515e-01 -8.41746688e-01
-1.22057617e+00 2.27532443e-02 2.05420032e-01 -4.88466561e-01
8.05599466e-02 2.54694492e-01 1.05656244e-01 -2.72978097e-01
7.70969331e-01 5.19720614e-01 2.96282798e-01 -2.94144690e-01
6.97236776e-01 4.91007745e-01 4.60592806e-01 -6.11990511e-01
8.12324524e-01 7.72395730e-01 -2.62679309e-01 -7.16128886e-01
-4.96327877e-01 7.40920603e-02 -6.95828736e-01 -6.04379214e-02
8.44053030e-01 -6.58474743e-01 -1.14511572e-01 6.89973295e-01
-1.16399503e+00 -6.50686145e-01 -3.70456010e-01 1.33894935e-01
-7.76807487e-01 3.89858663e-01 -5.72036505e-01 -3.46938014e-01
-3.61982524e-01 -1.48051858e+00 9.92476225e-01 6.97920993e-02
-4.66630131e-01 -1.05827236e+00 7.16895238e-02 1.55414984e-01
9.40338075e-01 -1.24024138e-01 1.02699935e+00 3.04895155e-02
-7.55658746e-01 5.90640493e-02 -4.30563271e-01 1.53123900e-01
3.08282942e-01 -8.36910903e-02 -6.08392835e-01 -4.58248973e-01
-1.29417442e-02 -3.15429837e-01 7.89733529e-01 3.12325001e-01
6.65059686e-01 -3.09305012e-01 -2.46102527e-01 8.20142090e-01
1.18950450e+00 4.99575809e-02 6.76190972e-01 5.78436017e-01
6.13181710e-01 5.35265923e-01 3.72859120e-01 -4.94450666e-02
3.74899447e-01 7.75640130e-01 -3.98221798e-02 -3.44130844e-01
-6.62524879e-01 -4.68488902e-01 4.00697201e-01 6.08906388e-01
1.37041241e-01 -3.28029513e-01 -7.00342476e-01 5.37771702e-01
-1.68252397e+00 -1.02435148e+00 1.84784785e-01 2.08043242e+00
1.03672576e+00 1.97081506e-01 1.16334192e-01 -3.58835042e-01
4.37051743e-01 3.70846301e-01 -7.46492147e-01 -5.29753089e-01
-2.98644811e-01 -2.20050305e-01 5.05894363e-01 5.89843094e-01
-8.11479330e-01 1.31261086e+00 6.92216253e+00 7.88148999e-01
-1.57106459e+00 1.94741249e-01 8.35254252e-01 -1.36362955e-01
-5.19852996e-01 -1.64654832e-02 -5.19578218e-01 3.42751265e-01
6.35889590e-01 -2.41735697e-01 7.03773141e-01 4.99897927e-01
3.13893259e-01 -1.50766283e-01 -9.63060498e-01 9.44676995e-01
1.17788494e-01 -1.54369342e+00 3.37995946e-01 1.35778740e-01
9.43250120e-01 -4.48826998e-02 4.68021065e-01 -2.02224016e-01
4.30264533e-01 -7.71250606e-01 1.12268519e+00 3.94375235e-01
1.13199794e+00 -3.38008046e-01 6.74793422e-02 2.31264323e-01
-7.66101599e-01 1.62898049e-01 -2.72527426e-01 2.43368059e-01
2.44715542e-01 3.80222291e-01 -3.16634595e-01 1.36898026e-01
3.97106677e-01 6.02202296e-01 -4.65518266e-01 6.20157719e-01
-2.26488203e-01 2.72028357e-01 -3.95338461e-02 1.04963280e-01
2.88722634e-01 -4.66916561e-01 6.23432398e-01 1.22355545e+00
4.29902643e-01 -1.32507965e-01 7.05060810e-02 1.15781736e+00
-9.28014368e-02 1.45408615e-01 -7.79404342e-01 -2.41476908e-01
1.61997557e-01 9.85711396e-01 -1.07092798e+00 -4.30365413e-01
-3.36094230e-01 1.72284245e+00 3.55701029e-01 5.61256349e-01
-7.14317381e-01 -6.84061870e-02 5.54991305e-01 2.34521315e-01
2.55033910e-01 -4.95596409e-01 -4.12680060e-01 -1.32479334e+00
-1.21477962e-01 -1.10517490e+00 -1.65563315e-01 -8.53769958e-01
-1.07089806e+00 8.24597478e-01 -6.27879575e-02 -9.52363253e-01
-4.51463275e-02 -2.72071153e-01 -4.24030840e-01 1.00039756e+00
-1.54570448e+00 -1.32262850e+00 -3.28698158e-02 6.09805048e-01
7.07769275e-01 -8.40292498e-02 6.50587201e-01 2.54536062e-01
-2.54298955e-01 5.67295969e-01 2.52647460e-01 -1.75349414e-01
1.02652752e+00 -1.04584324e+00 6.99483037e-01 9.03184354e-01
1.45689443e-01 6.32869303e-01 8.78881156e-01 -7.73965299e-01
-1.45375264e+00 -8.53412449e-01 9.20689225e-01 -3.62757027e-01
6.37620807e-01 -3.34109426e-01 -8.17018270e-01 3.98640096e-01
5.76435924e-01 -2.59301603e-01 3.33278060e-01 -2.70774633e-01
-5.18402874e-01 1.10753290e-01 -1.02342498e+00 1.14408529e+00
1.17592263e+00 -7.38477468e-01 -2.51424998e-01 2.10191473e-01
4.68071669e-01 -5.03206372e-01 -5.02614319e-01 -1.57055810e-01
6.37830079e-01 -9.08625126e-01 9.37069356e-01 -3.91872317e-01
7.02770114e-01 -2.35778958e-01 4.25872207e-02 -1.48495817e+00
-5.47873616e-01 -9.85431254e-01 9.14004222e-02 1.06819010e+00
4.77950156e-01 -4.89016175e-01 6.80777073e-01 5.97328484e-01
3.18306506e-01 -4.24588859e-01 -6.08663082e-01 -7.94901788e-01
2.43066281e-01 -3.13328728e-02 4.20823246e-01 9.30283964e-01
-2.71169990e-01 4.16944593e-01 -5.62299550e-01 -3.27874601e-01
5.81384122e-01 2.27589950e-01 7.78910518e-01 -7.73596525e-01
-2.40171224e-01 -9.41645145e-01 -6.18369617e-02 -1.30009663e+00
3.89089389e-03 -7.49192953e-01 -3.49827521e-02 -1.48831785e+00
2.22638279e-01 -3.69499534e-01 -7.10317194e-02 3.44356447e-01
-1.22538865e-01 4.94857877e-01 3.80580008e-01 6.11098766e-01
-3.05688262e-01 5.01922071e-01 1.71431446e+00 -3.00886214e-01
-3.61607313e-01 -2.67336369e-01 -7.76514590e-01 3.79009664e-01
7.99928188e-01 -5.27753234e-01 -7.44028568e-01 -8.25273812e-01
2.47185767e-01 -1.18730199e-02 1.39837056e-01 -6.56389713e-01
3.13971937e-01 -3.65941197e-01 1.75250113e-01 -1.18922934e-01
1.65659055e-01 -6.63955033e-01 1.68616727e-01 5.03316224e-01
-7.24875629e-01 4.04759824e-01 8.80709067e-02 6.90786600e-01
-8.93138275e-02 -5.05186617e-02 1.16960967e+00 -2.45058119e-01
-4.75718558e-01 3.07995111e-01 -5.26453495e-01 -6.05499893e-02
8.84893656e-01 -4.01843637e-01 -2.07921892e-01 -5.76952815e-01
-5.53676367e-01 -2.39714935e-01 1.03423357e+00 6.61845624e-01
5.85866213e-01 -1.28949189e+00 -6.91757917e-01 4.56919342e-01
1.99704038e-04 -3.79768699e-01 6.60636872e-02 9.38813925e-01
-6.30231738e-01 1.99312866e-01 -4.18519706e-01 -6.26985312e-01
-1.27631414e+00 6.32099390e-01 2.73076534e-01 -3.21061790e-01
-7.73371220e-01 7.20801115e-01 2.93852627e-01 -2.29982331e-01
8.53319243e-02 -1.20413713e-01 2.87115008e-01 -1.09792426e-01
5.50802290e-01 1.06684841e-01 -2.05072671e-01 -6.84639573e-01
6.89627677e-02 6.44871116e-01 -4.43759650e-01 -6.17921174e-01
1.12288618e+00 -6.90029323e-01 -1.64254993e-01 2.07153529e-01
1.08447552e+00 -1.08125405e-02 -1.56735945e+00 -4.74671334e-01
-9.77702662e-02 -6.95625305e-01 1.72662273e-01 -8.64544272e-01
-1.10114491e+00 9.30072725e-01 7.26595044e-01 5.07092811e-02
1.19022214e+00 -2.46811047e-01 9.33392346e-01 1.70612097e-01
2.61459321e-01 -1.14384770e+00 2.93493599e-01 3.36305290e-01
8.46600413e-01 -1.25627005e+00 -8.76384303e-02 -1.60552040e-01
-7.49731421e-01 9.45348203e-01 2.97210276e-01 6.87173977e-02
3.83291155e-01 2.48315096e-01 4.50025022e-01 -6.39429092e-02
-6.48728371e-01 1.18351437e-01 3.36506426e-01 6.98534966e-01
4.68473405e-01 -1.37883529e-01 -1.63068324e-01 -4.50833082e-01
1.49094518e-02 1.70418441e-01 4.61439818e-01 9.92597342e-01
-3.27660143e-01 -1.40681338e+00 -3.14956009e-01 1.32917300e-01
-3.68329376e-01 -2.54914939e-01 -4.83959287e-01 4.69590038e-01
-3.24065685e-01 6.99911594e-01 -2.18396932e-01 -1.16848471e-02
1.21906750e-01 5.77065675e-03 7.69571960e-01 -4.11413312e-01
-4.78317231e-01 3.44803959e-01 -2.81004488e-01 -5.78457296e-01
-4.85290706e-01 -6.33625090e-01 -6.92943692e-01 -4.72022831e-01
-1.10276476e-01 -2.53543347e-01 7.65651524e-01 7.68678188e-01
6.51962817e-01 3.74864399e-01 3.36696148e-01 -8.53454411e-01
-4.98340726e-01 -6.37619078e-01 -3.48751336e-01 6.14986360e-01
1.96777537e-01 -2.96648830e-01 7.69182593e-02 4.19333279e-01] | [11.383232116699219, -0.3142123222351074] |
4db3cb4c-a2b1-46d8-9166-1751cd44cb45 | prior-guided-dropout-for-robust-visual | null | null | http://openaccess.thecvf.com/content_ICCV_2019/html/Huang_Prior_Guided_Dropout_for_Robust_Visual_Localization_in_Dynamic_Environments_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Huang_Prior_Guided_Dropout_for_Robust_Visual_Localization_in_Dynamic_Environments_ICCV_2019_paper.pdf | Prior Guided Dropout for Robust Visual Localization in Dynamic Environments | Camera localization from monocular images has been a long-standing problem, but its robustness in dynamic environments is still not adequately addressed. Compared with classic geometric approaches, modern CNN-based methods (e.g. PoseNet) have manifested the reliability against illumination or viewpoint variations, but they still have the following limitations. First, foreground moving objects are not explicitly handled, which results in poor performance and instability in dynamic environments. Second, the output for each image is a point estimate without uncertainty quantification. In this paper, we propose a framework which can be generally applied to existing CNN-based pose regressors to improve their robustness in dynamic environments. The key idea is a prior guided dropout module coupled with a self-attention module which can guide CNNs to ignore foreground objects during both training and inference. Additionally, the dropout module enables the pose regressor to output multiple hypotheses from which the uncertainty of pose estimates can be quantified and leveraged in the following uncertainty-aware pose-graph optimization to improve the robustness further. We achieve an average accuracy of 9.98m/3.63deg on RobotCar dataset, which outperforms the state-of-the-art method by 62.97%/47.08%. The source code of our implementation is available at https://github.com/zju3dv/RVL-dynamic.
| [' Guofeng Zhang', ' Hujun Bao', ' Xiaowei Zhou', ' Jianping Shi', ' Yan Xu', 'Zhaoyang Huang'] | 2019-10-01 | null | null | null | iccv-2019-10 | ['camera-localization'] | ['computer-vision'] | [-1.49228215e-01 -1.01804480e-01 7.94001855e-03 -2.67845809e-01
-6.14987671e-01 -4.96778607e-01 3.67030650e-01 -4.26890671e-01
-5.77280819e-01 6.10248327e-01 -2.87175745e-01 -1.09669760e-01
2.14309722e-01 -5.70019960e-01 -1.22865200e+00 -8.60488296e-01
3.77060264e-01 1.48294449e-01 4.78921682e-01 1.32654130e-01
1.83219388e-01 4.97478038e-01 -1.32019663e+00 -3.42983305e-01
1.00923944e+00 1.16911662e+00 3.62271518e-01 4.33293909e-01
2.08553255e-01 5.68824410e-01 -5.43693125e-01 -2.72017062e-01
2.03891978e-01 -2.63936762e-02 -2.57204413e-01 -4.75173257e-02
8.26296329e-01 -5.67394674e-01 -7.37176538e-01 1.38753343e+00
4.79558319e-01 2.07050622e-01 3.84289920e-01 -1.16555059e+00
-7.40433395e-01 2.14511588e-01 -7.55435169e-01 5.66710485e-03
8.16864967e-02 4.44400549e-01 5.98086953e-01 -1.02945566e+00
4.95014876e-01 1.33224332e+00 6.08646810e-01 4.75925416e-01
-9.68113005e-01 -8.67727518e-01 6.38937235e-01 3.47609073e-01
-1.59088540e+00 -4.00270283e-01 7.75752604e-01 -3.49375874e-01
8.20085764e-01 -1.89560473e-01 5.73075235e-01 1.22658777e+00
3.42345297e-01 9.11157429e-01 7.48490512e-01 1.52912587e-01
5.96807711e-02 -6.14160001e-02 -9.88375172e-02 8.96994829e-01
4.67335194e-01 1.86480731e-01 -3.62283647e-01 3.83303940e-01
9.83001173e-01 1.64442271e-01 -4.10428703e-01 -7.07965672e-01
-1.14862764e+00 6.58756614e-01 1.04307151e+00 -1.90390751e-01
-6.98759183e-02 5.16014814e-01 1.33005217e-01 -1.20824158e-01
4.74401802e-01 3.09886336e-01 -4.63578999e-01 7.14426339e-02
-7.18146920e-01 3.60988557e-01 4.56324756e-01 1.38397443e+00
7.79643655e-01 3.13417196e-01 1.48919970e-01 5.57973325e-01
6.50498390e-01 7.79088676e-01 -2.87314504e-02 -8.88164103e-01
6.20798945e-01 5.63544869e-01 9.87255052e-02 -1.25822616e+00
-5.26879787e-01 -7.58016229e-01 -7.15046644e-01 3.20280910e-01
5.19671023e-01 -1.46361724e-01 -1.04124331e+00 1.72887695e+00
3.21787238e-01 3.14698488e-01 -2.50898868e-01 1.28271568e+00
8.87801707e-01 5.04270792e-01 -2.32708737e-01 2.55633205e-01
1.05774319e+00 -1.27178323e+00 -6.49867415e-01 -5.43895245e-01
2.34304443e-01 -6.91016078e-01 8.86624634e-01 3.90007168e-01
-8.20877373e-01 -5.22298932e-01 -1.29313254e+00 -2.36119553e-01
-3.40560734e-01 3.15692514e-01 4.70143497e-01 2.99751639e-01
-9.77091312e-01 3.92305255e-01 -1.12464356e+00 -2.48957634e-01
5.33557713e-01 4.12256747e-01 -2.37811938e-01 -2.48373359e-01
-9.02535141e-01 9.73259628e-01 2.82321513e-01 4.48595017e-01
-1.06112790e+00 -5.47169149e-01 -1.15408361e+00 -1.77531540e-01
7.65139043e-01 -7.45620430e-01 1.12704194e+00 -6.21693790e-01
-1.65383470e+00 4.96800303e-01 -6.19798414e-02 -3.70582104e-01
8.37184787e-01 -7.32838452e-01 -5.81805781e-02 1.01909088e-02
7.79940635e-02 8.94063830e-01 8.98259163e-01 -1.33395553e+00
-5.22683144e-01 -4.55296338e-01 2.51224279e-01 3.49370331e-01
8.69659781e-02 -2.48490542e-01 -1.00621760e+00 -4.75328594e-01
4.10217762e-01 -1.19500601e+00 -1.58951715e-01 3.07893604e-01
-5.55378079e-01 1.48794442e-01 8.20394933e-01 -5.15922725e-01
8.76010299e-01 -1.99223804e+00 2.14129284e-01 -2.85592943e-01
1.03546314e-01 2.17503399e-01 -2.58586742e-02 1.00360718e-02
2.71389157e-01 -6.11739829e-02 -1.95584700e-01 -6.98578537e-01
-5.97529188e-02 2.77792178e-02 -1.59540236e-01 8.98955107e-01
5.53545415e-01 1.11342728e+00 -8.94889116e-01 -2.40017056e-01
7.68833935e-01 8.32384348e-01 -4.58128273e-01 8.20531249e-02
-2.77325004e-01 8.19884062e-01 -4.05742586e-01 8.80254924e-01
1.00531900e+00 -2.27177382e-01 -3.62100631e-01 -3.07570308e-01
-2.87570566e-01 2.22042710e-01 -1.28219521e+00 1.89030313e+00
-3.06210935e-01 7.98154533e-01 1.60580277e-01 -6.01107121e-01
7.67120540e-01 -4.19539437e-02 1.34705216e-01 -4.83930558e-01
4.54840839e-01 9.90372673e-02 -5.49909286e-02 -3.20935220e-01
5.23937881e-01 4.28719491e-01 4.33647409e-02 -3.13265502e-01
3.66679765e-02 -2.19629064e-01 -6.07572086e-02 3.87746766e-02
8.66035461e-01 6.46498263e-01 -1.31552488e-01 -1.03750482e-01
5.25328159e-01 -1.13913886e-01 8.78673792e-01 6.41516030e-01
-2.34097824e-01 1.07468653e+00 2.47645319e-01 -2.98156679e-01
-8.19673002e-01 -1.03295517e+00 -2.35586494e-01 6.02150440e-01
6.90224171e-01 2.05892250e-02 -5.74643135e-01 -5.05029857e-01
3.14610526e-02 5.00461519e-01 -3.67234558e-01 -2.37263456e-01
-6.73077047e-01 -6.49328828e-01 3.06820899e-01 7.62138069e-01
6.67591572e-01 -7.73027122e-01 -6.30217135e-01 2.16132820e-01
-7.32690245e-02 -1.44515288e+00 -4.21128273e-01 1.33203954e-01
-8.53950918e-01 -1.12258852e+00 -7.03213096e-01 -5.93067408e-01
7.03507006e-01 4.14617747e-01 9.27127182e-01 -7.60331452e-02
-1.74229667e-01 1.77586511e-01 -2.24355131e-01 -4.48811233e-01
1.57463387e-01 2.68929273e-01 1.96046919e-01 -6.93747401e-02
8.78512487e-02 -4.39210206e-01 -7.86063194e-01 4.33549911e-01
-6.72110260e-01 6.27672523e-02 6.55944824e-01 8.10100734e-01
6.32016063e-01 -3.34354967e-01 2.40896493e-01 -5.93539178e-01
-9.84306037e-02 -3.64437044e-01 -1.15943205e+00 -1.05565622e-01
-4.58462954e-01 -7.06214607e-02 5.46572983e-01 -5.48015356e-01
-9.16690469e-01 3.51685107e-01 -6.81427717e-02 -8.17166388e-01
-1.17073514e-01 3.80127519e-01 -4.55648750e-01 -3.93864065e-01
3.27960968e-01 1.48314729e-01 -5.53189516e-02 -3.95602405e-01
3.17366838e-01 1.52867928e-01 6.94718659e-01 -3.12167823e-01
1.08570468e+00 5.05497038e-01 -3.41920368e-02 -7.09225059e-01
-8.95330369e-01 -3.83001626e-01 -6.16339743e-01 -3.20087880e-01
9.37765241e-01 -1.37755346e+00 -8.35546672e-01 7.63660014e-01
-1.23158956e+00 -3.57803851e-01 3.29955012e-01 5.97971320e-01
-3.70806932e-01 2.91571409e-01 -5.64675748e-01 -7.56358981e-01
-8.32433030e-02 -1.43144548e+00 1.17583299e+00 5.23020148e-01
3.59913945e-01 -7.48989105e-01 -4.57976371e-01 2.45419621e-01
4.09508616e-01 3.86717141e-01 2.79718697e-01 -1.72557846e-01
-1.28678823e+00 -2.37626791e-01 -4.63976175e-01 3.47951442e-01
-2.00821504e-01 9.83177200e-02 -1.05191207e+00 -4.74460661e-01
-5.95436618e-03 -2.81891227e-01 1.00440025e+00 5.90929627e-01
1.07103109e+00 -2.90693138e-02 -3.52304757e-01 1.15059102e+00
1.42193830e+00 1.25793427e-01 6.82491124e-01 5.32125890e-01
1.16195869e+00 2.25813955e-01 7.48070717e-01 1.79092109e-01
5.48800588e-01 7.17906296e-01 9.97242749e-01 -2.26114932e-02
-1.08236291e-01 -1.96798757e-01 4.08842981e-01 6.02575779e-01
2.04043135e-01 -3.45543861e-01 -8.05618882e-01 3.53332639e-01
-2.12948394e+00 -5.68705797e-01 -2.73007482e-01 2.01145029e+00
4.48559791e-01 3.28641057e-01 -3.60375404e-01 -3.10922623e-01
6.27394676e-01 3.64660352e-01 -9.46258008e-01 2.95269877e-01
-5.67520820e-02 -3.25627327e-01 8.87120366e-01 5.75040460e-01
-1.31176412e+00 1.12711275e+00 4.84625244e+00 4.15844023e-01
-1.26303017e+00 -7.12095350e-02 5.19659102e-01 -2.09176064e-01
2.33077392e-01 -7.87483081e-02 -1.11210120e+00 4.62767571e-01
4.55270678e-01 2.19516158e-01 2.79464424e-01 1.02251327e+00
1.40210077e-01 -3.14486980e-01 -1.02127552e+00 1.15727043e+00
1.50758192e-01 -1.13075793e+00 -3.11132371e-01 1.38256764e-02
8.11930358e-01 4.25100386e-01 3.49862725e-01 2.88901716e-01
7.30179325e-02 -9.83082414e-01 1.00332475e+00 4.85184431e-01
6.65961444e-01 -7.99584210e-01 9.01242912e-01 3.70446652e-01
-1.22691488e+00 -2.80800276e-02 -6.25331700e-01 -4.67767790e-02
2.63025224e-01 6.51395500e-01 -5.67811072e-01 7.36392498e-01
8.59190226e-01 9.50316489e-01 -6.77249551e-01 1.27861738e+00
-5.72655678e-01 3.40426534e-01 -4.86516654e-01 4.45755422e-02
3.41252595e-01 -3.02338660e-01 6.54686630e-01 9.56489503e-01
2.25454181e-01 -1.95910990e-01 2.69795477e-01 1.18551230e+00
-8.05773586e-02 -3.31268966e-01 -5.22081077e-01 3.86666805e-01
5.43837428e-01 1.35497475e+00 -6.15667820e-01 -5.68909124e-02
-5.04843652e-01 9.13731933e-01 4.17487383e-01 4.03366983e-01
-1.21561778e+00 -3.32004011e-01 7.54704714e-01 -6.47138730e-02
4.38970178e-01 -5.55005193e-01 -3.18071902e-01 -1.44353914e+00
3.30921173e-01 -7.16985524e-01 -8.04035217e-02 -8.50814998e-01
-1.08029819e+00 6.32160962e-01 -5.69309369e-02 -1.19349682e+00
5.21997213e-02 -8.73172879e-01 -4.73419845e-01 7.68312633e-01
-1.67817116e+00 -1.20123446e+00 -6.47642732e-01 3.28033119e-01
5.33493221e-01 1.01475291e-01 2.83918649e-01 4.46453571e-01
-9.69591618e-01 5.64861417e-01 3.11847404e-02 3.15044701e-01
7.39251494e-01 -1.25901699e+00 5.80786169e-01 1.20288384e+00
-1.03240855e-01 6.55861616e-01 6.54074967e-01 -6.10684276e-01
-1.66702592e+00 -1.44099033e+00 3.81131023e-01 -6.66594326e-01
6.72969222e-01 -5.91820180e-01 -9.28713441e-01 6.40099525e-01
-3.97239216e-02 3.28880221e-01 -1.16622038e-01 -2.08724678e-01
-2.97566682e-01 -1.13734849e-01 -8.01074922e-01 8.05974126e-01
1.21096373e+00 -4.09784138e-01 -1.88359931e-01 1.06391959e-01
8.99695516e-01 -9.60208297e-01 -5.46716392e-01 5.33730924e-01
3.35549861e-01 -9.54959154e-01 9.74713147e-01 4.22036126e-02
1.44141287e-01 -7.15925694e-01 -9.48090628e-02 -1.07913637e+00
-2.29591310e-01 -5.52266359e-01 -3.51700157e-01 1.08774543e+00
2.98027575e-01 -7.40411282e-01 7.97688544e-01 5.17195821e-01
-4.53931659e-01 -8.44119549e-01 -8.71515691e-01 -9.06019092e-01
-3.59575190e-02 -4.73580986e-01 4.38179255e-01 5.65705776e-01
-6.31140709e-01 3.00779641e-01 -4.19890583e-01 6.46333456e-01
8.09112132e-01 -8.20801035e-02 9.11957204e-01 -9.56578910e-01
-7.46653602e-02 -3.66613179e-01 -4.65253532e-01 -1.50937760e+00
2.02764198e-01 -5.87948740e-01 3.88362974e-01 -1.47864652e+00
-3.58799845e-02 -2.39479065e-01 -8.16294849e-02 2.03139916e-01
-2.72234261e-01 1.95236921e-01 2.74723440e-01 5.14755286e-02
-7.19624519e-01 7.75116980e-01 1.26796675e+00 -1.14856884e-01
-1.64552748e-01 4.05452177e-02 -4.28878456e-01 8.15053105e-01
8.48941863e-01 -4.13672358e-01 -3.58366638e-01 -7.24844635e-01
1.08661197e-01 -1.47193447e-01 7.92797506e-01 -1.31279278e+00
4.40213710e-01 1.04175046e-01 6.35965049e-01 -9.56606686e-01
4.88047421e-01 -8.10087681e-01 5.60186319e-02 4.66369122e-01
9.53357071e-02 2.25760892e-01 2.41404861e-01 7.77846992e-01
-1.41222849e-01 9.50087886e-03 8.91531229e-01 -2.37687640e-02
-8.46170664e-01 6.81846261e-01 -4.35312577e-02 6.70973361e-02
8.20188224e-01 -2.03135371e-01 -3.97923470e-01 -3.59123468e-01
-2.61440128e-01 4.03143704e-01 7.40173161e-01 6.70851529e-01
7.62211442e-01 -1.23474801e+00 -5.31616032e-01 8.80793035e-02
1.23059765e-01 7.18047857e-01 1.76021039e-01 1.03062010e+00
-7.00380564e-01 3.84258389e-01 3.42592262e-02 -9.61226702e-01
-8.33962262e-01 5.96203327e-01 4.40167338e-01 1.33001566e-01
-7.33474195e-01 9.46353197e-01 4.03787404e-01 -4.03745830e-01
6.79590046e-01 -5.20046949e-01 7.40524158e-02 -2.00595811e-01
3.62347811e-01 2.01987863e-01 -3.03287711e-02 -6.39797151e-01
-6.16913855e-01 7.68456936e-01 -7.96634331e-02 8.24469402e-02
1.31084144e+00 -2.96766639e-01 1.41388446e-01 3.11458796e-01
1.35052502e+00 -3.01730245e-01 -1.94561124e+00 -2.09914044e-01
-1.00407712e-01 -5.38238764e-01 1.25898406e-01 -6.15447283e-01
-1.34963357e+00 9.17839110e-01 5.72157323e-01 -4.63008672e-01
8.07457387e-01 -1.11336924e-01 5.59925258e-01 4.21326518e-01
4.59344149e-01 -9.41208482e-01 2.12559015e-01 9.85837519e-01
9.13597882e-01 -1.60778940e+00 6.56783953e-02 -5.08157015e-01
-5.39337039e-01 9.14021909e-01 1.04168129e+00 -4.45646644e-01
5.16577780e-01 3.02804619e-01 2.18811288e-01 -9.32981893e-02
-5.16060352e-01 -2.78939426e-01 4.63878214e-01 5.87890685e-01
2.43618995e-01 -2.36813888e-01 2.18120426e-01 3.92142117e-01
-9.70357656e-03 -3.94846410e-01 3.54879975e-01 7.64733434e-01
-3.52020860e-01 -5.10669708e-01 -3.98027331e-01 4.53971922e-02
-3.09221804e-01 -5.71978018e-02 -2.16497123e-01 9.37136889e-01
3.67966294e-02 8.67230475e-01 -1.68432062e-03 -3.90964240e-01
4.26913798e-01 -2.52997279e-01 3.85621279e-01 -6.19082630e-01
-3.21813166e-01 2.23962620e-01 -1.88388422e-01 -8.82432282e-01
-2.23026872e-01 -6.67486966e-01 -1.20117903e+00 -2.03620121e-01
-6.34791672e-01 -4.38410908e-01 7.68823743e-01 7.05837727e-01
3.88118267e-01 7.56670058e-01 3.73971164e-01 -1.32072663e+00
-5.41640162e-01 -8.60680819e-01 -1.21968940e-01 4.62988429e-02
5.08040905e-01 -1.00330162e+00 -3.92105222e-01 -2.23993614e-01] | [7.978815078735352, -2.15317702293396] |
926c5294-0686-4b73-b73b-e876cbc47050 | 3dn-3d-deformation-network | 1903.03322 | null | http://arxiv.org/abs/1903.03322v1 | http://arxiv.org/pdf/1903.03322v1.pdf | 3DN: 3D Deformation Network | Applications in virtual and augmented reality create a demand for rapid
creation and easy access to large sets of 3D models. An effective way to
address this demand is to edit or deform existing 3D models based on a
reference, e.g., a 2D image which is very easy to acquire. Given such a source
3D model and a target which can be a 2D image, 3D model, or a point cloud
acquired as a depth scan, we introduce 3DN, an end-to-end network that deforms
the source model to resemble the target. Our method infers per-vertex offset
displacements while keeping the mesh connectivity of the source model fixed. We
present a training strategy which uses a novel differentiable operation, mesh
sampling operator, to generalize our method across source and target models
with varying mesh densities. Mesh sampling operator can be seamlessly
integrated into the network to handle meshes with different topologies.
Qualitative and quantitative results show that our method generates higher
quality results compared to the state-of-the art learning-based methods for 3D
shape generation. Code is available at github.com/laughtervv/3DN. | ['Weiyue Wang', 'Duygu Ceylan', 'Ulrich Neumann', 'Radomir Mech'] | 2019-03-08 | 3dn-3d-deformation-network-1 | http://openaccess.thecvf.com/content_CVPR_2019/html/Wang_3DN_3D_Deformation_Network_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Wang_3DN_3D_Deformation_Network_CVPR_2019_paper.pdf | cvpr-2019-6 | ['3d-shape-generation'] | ['computer-vision'] | [ 3.30500811e-01 4.25820053e-01 2.86920458e-01 -2.51362622e-01
-6.39411271e-01 -5.52381754e-01 4.35650885e-01 -3.12836505e-02
1.67035669e-01 4.78341848e-01 -1.18765414e-01 -2.35989764e-01
9.49333310e-02 -1.15226972e+00 -1.04597688e+00 -1.10665821e-01
4.66528125e-02 7.87560940e-01 3.62776339e-01 -2.73213178e-01
2.58205712e-01 1.17541337e+00 -1.52436185e+00 -4.49231863e-02
6.29688501e-01 9.79955196e-01 3.44648540e-01 5.31253159e-01
-3.30636322e-01 -1.68994278e-01 -3.58981401e-01 -3.23915929e-01
6.45887077e-01 -1.98524445e-01 -7.12635934e-01 3.58439833e-01
7.44093955e-01 -1.76098779e-01 -8.19616243e-02 1.02277398e+00
7.07886815e-01 -3.45136859e-02 5.50523818e-01 -1.03759313e+00
-6.44215405e-01 8.29939917e-02 -6.43858790e-01 -4.02084887e-01
5.53298056e-01 -2.14436844e-01 3.87985438e-01 -1.21843469e+00
1.12583697e+00 1.34227681e+00 8.13555598e-01 5.66223860e-01
-1.34684587e+00 -4.13266987e-01 1.15382440e-01 -3.90185028e-01
-1.37644958e+00 -2.86926657e-01 1.17288864e+00 -4.49903995e-01
5.20125926e-01 3.37712288e-01 9.64795232e-01 8.74309897e-01
3.27649981e-01 3.64440501e-01 7.77571678e-01 -5.15848935e-01
1.72674671e-01 1.38671517e-01 -4.85657424e-01 7.03007996e-01
-3.03218924e-02 1.80888280e-01 -2.50445575e-01 -3.02799463e-01
1.58406544e+00 1.17455507e-02 -3.93237174e-01 -1.02988565e+00
-1.24871886e+00 5.17725825e-01 5.64395249e-01 -9.88141969e-02
-4.20051366e-01 8.09105113e-02 1.68199092e-01 3.09946120e-01
7.89805174e-01 3.68608713e-01 -6.40977144e-01 2.17778049e-03
-6.32272124e-01 4.46337700e-01 7.30537415e-01 1.09170640e+00
9.06551659e-01 1.08099490e-01 3.67277771e-01 8.12237024e-01
4.37725872e-01 4.52901661e-01 -5.63013367e-02 -1.23190200e+00
1.68511420e-01 7.21708894e-01 1.73641160e-01 -9.70950842e-01
-2.02460229e-01 -2.13467583e-01 -9.83735204e-01 7.24096715e-01
-1.60961635e-02 1.22712836e-01 -1.17280638e+00 1.41690564e+00
9.87712085e-01 2.96871662e-01 -3.28172475e-01 8.47287297e-01
9.80903625e-01 4.65999186e-01 -6.92419410e-01 -5.40967379e-03
8.23110461e-01 -6.29027247e-01 -3.79918933e-01 9.32717696e-02
3.63481939e-01 -8.24240685e-01 1.13884616e+00 2.32800186e-01
-1.48538160e+00 -4.64808613e-01 -9.53444779e-01 -1.35867313e-01
-3.47073674e-01 -2.73790151e-01 2.42966026e-01 3.53862464e-01
-1.26452684e+00 9.61104929e-01 -8.27754498e-01 -3.07016522e-01
5.34581006e-01 2.90026993e-01 -4.03942674e-01 1.98751613e-02
-8.62135053e-01 9.61251140e-01 8.90006274e-02 7.44710937e-02
-5.57868540e-01 -1.10690475e+00 -1.00475454e+00 -3.12736988e-01
2.87075996e-01 -1.08276904e+00 1.28028834e+00 -7.76747167e-01
-1.74924624e+00 1.20614338e+00 -2.88789906e-02 8.24720506e-03
7.67345905e-01 1.75373703e-01 -6.81427494e-03 -1.16433119e-02
6.22645728e-02 6.90576494e-01 8.22648644e-01 -1.92760849e+00
-2.43070260e-01 -3.79105031e-01 1.97318792e-01 2.48989686e-01
2.63017744e-01 -3.79625946e-01 -5.11933982e-01 -6.55992985e-01
5.31347215e-01 -7.17161655e-01 -3.51201504e-01 8.58538687e-01
-4.79019701e-01 1.60253480e-01 1.15043056e+00 -5.55119812e-01
7.33405292e-01 -1.93401027e+00 3.78695965e-01 5.30106068e-01
3.31720144e-01 1.42283738e-01 -1.50955081e-01 4.24285561e-01
-2.15793908e-01 1.93778694e-01 -5.96551359e-01 -6.28039062e-01
-4.25457358e-02 1.79862157e-01 -1.15135023e-02 3.96231979e-01
1.14897884e-01 6.66992426e-01 -9.81058240e-01 -3.98386240e-01
5.42055905e-01 9.47279692e-01 -4.54665512e-01 1.31550029e-01
-2.83372462e-01 7.14567780e-01 -2.74778783e-01 6.77203476e-01
9.86056805e-01 -2.75301605e-01 -8.10680464e-02 -3.12200874e-01
-1.57908827e-01 9.32322592e-02 -1.57111251e+00 2.25459862e+00
-6.94842875e-01 1.93134338e-01 3.97192627e-01 -6.85887158e-01
1.16854107e+00 3.04172665e-01 4.22243595e-01 -4.65044558e-01
1.35436878e-01 3.49093646e-01 -4.24157172e-01 -1.50545120e-01
3.71431530e-01 -1.22991703e-01 1.45850271e-01 3.85048747e-01
-2.89406747e-01 -8.52098167e-01 -2.52759457e-01 9.47176963e-02
6.75958097e-01 5.17767549e-01 9.62963030e-02 -9.10563394e-02
3.96835864e-01 -1.80154264e-01 4.00887579e-01 2.98270196e-01
4.42192614e-01 1.01850092e+00 2.25623980e-01 -6.66501939e-01
-1.27811003e+00 -1.41116655e+00 -3.70007575e-01 3.07465434e-01
3.03931564e-01 -1.83058262e-01 -5.05214930e-01 -4.39757854e-01
2.10636720e-01 5.56362927e-01 -7.09070086e-01 7.95637295e-02
-7.52601445e-01 -1.47226620e-02 -8.70608911e-02 4.44292784e-01
2.59892732e-01 -9.34091926e-01 -5.74288368e-01 1.25388831e-01
2.32836083e-01 -7.67108262e-01 -6.92132890e-01 -1.54318884e-01
-1.17269218e+00 -9.03033137e-01 -7.05746353e-01 -9.39193547e-01
9.96832311e-01 1.22618742e-01 1.31092799e+00 2.96190023e-01
-1.40295878e-01 4.57280397e-01 -1.73572600e-01 -3.00788164e-01
-7.78913856e-01 -2.70009246e-02 -4.43737209e-02 -1.45945698e-01
-5.60155094e-01 -1.15318263e+00 -4.59087849e-01 2.94049948e-01
-1.13631225e+00 5.75214267e-01 1.43628925e-01 5.59789777e-01
1.20885992e+00 -3.06169301e-01 3.43952626e-01 -7.38689959e-01
3.57115537e-01 -1.81637362e-01 -7.45973885e-01 -1.23246806e-02
-3.42328250e-01 -1.00536376e-01 4.50633526e-01 -5.51254690e-01
-7.69947588e-01 3.11411232e-01 -3.45622063e-01 -9.07893598e-01
-9.23070312e-02 5.20141065e-01 -3.34272206e-01 -2.70660669e-01
6.52012467e-01 -2.41317954e-02 4.06480491e-01 -7.69942582e-01
4.79977131e-01 4.13235277e-01 5.10372996e-01 -4.81274396e-01
1.05570972e+00 6.66365504e-01 2.39869431e-01 -7.46656835e-01
-5.40798962e-01 -7.14170337e-02 -1.14438689e+00 -2.21140712e-01
4.50954229e-01 -5.58607697e-01 -4.28194016e-01 5.27845263e-01
-1.49674213e+00 -5.90026081e-01 -5.76892257e-01 4.74846270e-03
-7.04480410e-01 1.49585515e-01 -3.27441543e-01 -5.03494322e-01
-4.00088072e-01 -1.00387061e+00 1.43522632e+00 1.62012456e-03
-2.15078175e-01 -1.16491151e+00 1.65363014e-01 -1.96834758e-01
3.33591580e-01 9.57965672e-01 9.26865399e-01 2.89611481e-02
-7.40962923e-01 -2.54079878e-01 -8.89435932e-02 2.15774372e-01
5.79448879e-01 2.88478434e-01 -6.47517681e-01 -1.58070937e-01
-2.42509663e-01 -2.99088284e-02 1.81739002e-01 4.25772786e-01
1.28414702e+00 -2.35582575e-01 -4.75077301e-01 8.19218159e-01
1.45203912e+00 1.07935324e-01 7.02104628e-01 1.34672925e-01
1.01940012e+00 3.96245480e-01 2.66893059e-01 3.59039426e-01
4.34458703e-01 9.25099611e-01 7.04267204e-01 -3.44175279e-01
-2.18178704e-01 -4.11845475e-01 -2.20261157e-01 8.73974681e-01
-7.57415295e-02 6.04434907e-02 -9.33322251e-01 3.24254364e-01
-1.67928183e+00 -6.41241610e-01 -1.70871913e-01 2.60214472e+00
8.04717422e-01 1.81747019e-01 -4.97107431e-02 7.11462796e-02
7.81683505e-01 3.20141278e-02 -7.66021729e-01 -3.76600653e-01
2.01258585e-01 3.71898770e-01 1.59385443e-01 7.44919062e-01
-7.25637496e-01 6.85726345e-01 5.70432806e+00 5.59122860e-01
-1.39452755e+00 -1.08736865e-02 3.41050178e-01 -1.19555734e-01
-7.22777486e-01 -2.05385655e-01 -3.44388515e-01 2.55369127e-01
4.88800049e-01 -3.91824573e-01 3.32512885e-01 7.58179784e-01
3.85237068e-01 1.09173410e-01 -1.16240776e+00 1.04713428e+00
1.43272895e-02 -1.90302372e+00 3.10253292e-01 2.53656328e-01
8.99889886e-01 -8.56202021e-02 -2.46419609e-01 -1.89909905e-01
3.04985076e-01 -8.74141097e-01 9.49740589e-01 8.50204885e-01
1.27350426e+00 -6.77664280e-01 3.16719174e-01 4.45755810e-01
-1.36090171e+00 6.66433096e-01 -1.54666170e-01 1.47105247e-01
5.67747951e-01 6.18133128e-01 -7.01065063e-01 6.81409657e-01
6.04894161e-01 7.16812134e-01 -2.31190756e-01 1.18067741e+00
9.57637355e-02 -1.04273029e-01 -4.32517469e-01 3.90499532e-01
-2.14411899e-01 -3.24550986e-01 7.77933776e-01 6.53028429e-01
6.03029013e-01 1.57002900e-02 2.77900577e-01 1.17807531e+00
-2.62719661e-01 9.73078236e-03 -1.01052153e+00 3.80517751e-01
7.18297184e-01 1.18060255e+00 -8.87232482e-01 -2.97321290e-01
-1.55866563e-01 1.09162259e+00 2.18118116e-01 9.09339041e-02
-6.43563867e-01 -3.27602029e-01 5.01963735e-01 6.24699414e-01
2.52134860e-01 -3.52543563e-01 -3.53487343e-01 -9.12974656e-01
2.19000369e-01 -6.42955422e-01 -2.42228076e-01 -1.13320351e+00
-1.33289516e+00 7.89243817e-01 1.53386131e-01 -1.51393652e+00
-1.33138031e-01 -4.01076287e-01 -6.28853142e-01 1.02525008e+00
-1.21575534e+00 -1.16239929e+00 -6.38543844e-01 3.11949164e-01
5.07991195e-01 2.70196676e-01 9.08624053e-01 2.40312397e-01
-7.76640251e-02 2.39005908e-01 -7.09269792e-02 -1.21693112e-01
3.94631892e-01 -1.14003134e+00 8.78629625e-01 3.88832629e-01
-7.96172395e-02 2.57584244e-01 4.58486885e-01 -8.40958238e-01
-1.42762899e+00 -1.14160526e+00 6.16553187e-01 -6.27182662e-01
2.33469397e-01 -4.50896382e-01 -1.18421209e+00 6.99494064e-01
-9.09596831e-02 2.12516069e-01 6.42870888e-02 -3.56648684e-01
-5.13991341e-02 6.57419786e-02 -1.49512351e+00 6.45294845e-01
1.51487505e+00 -2.87105918e-01 -2.96181709e-01 2.28995353e-01
9.58179593e-01 -1.42988122e+00 -1.02074540e+00 4.76825863e-01
5.28791726e-01 -9.09626424e-01 1.13508022e+00 -2.10923612e-01
5.01711071e-01 -4.02341694e-01 -3.09215933e-02 -1.70023918e+00
-1.55185714e-01 -7.60334313e-01 -2.33470634e-01 7.91898310e-01
3.77630055e-01 -7.18343556e-01 8.08343887e-01 5.62951386e-01
-5.38878918e-01 -1.17299294e+00 -1.14180005e+00 -7.18080163e-01
2.00879291e-01 -3.93731177e-01 9.63334441e-01 1.09249413e+00
-5.26813865e-01 8.22797120e-02 2.93240771e-02 3.39108586e-01
6.94498658e-01 2.45105758e-01 9.97456491e-01 -1.64790916e+00
1.94382757e-01 -3.47927690e-01 -4.24169421e-01 -1.15356147e+00
1.59532931e-02 -9.43632364e-01 -1.04379393e-01 -2.03179932e+00
-4.55269575e-01 -8.30539405e-01 5.32901943e-01 2.85099834e-01
3.47485781e-01 2.42872089e-01 -1.50653347e-02 7.44385570e-02
1.63496003e-01 6.86836779e-01 1.88933957e+00 1.89849913e-01
-6.63314819e-01 5.17640412e-02 -4.65112776e-01 8.01240265e-01
6.14468157e-01 -3.23690027e-01 -4.37830418e-01 -5.13932109e-01
3.05255651e-01 3.66843134e-01 5.72208107e-01 -7.50956237e-01
-6.76673427e-02 -1.23492233e-01 4.06915247e-01 -8.55838954e-01
6.13447964e-01 -9.33531821e-01 7.03874648e-01 5.52224368e-02
-2.88204905e-02 2.73402184e-01 3.71085912e-01 3.22969407e-01
2.45019987e-01 -8.91149640e-02 7.41734982e-01 -3.65614206e-01
-1.32129684e-01 7.50300229e-01 2.05156907e-01 -1.51115581e-01
8.70793521e-01 -7.55184650e-01 1.03142731e-01 -3.59905183e-01
-9.19682562e-01 6.93629533e-02 9.66862321e-01 4.63257194e-01
1.05113995e+00 -1.79420221e+00 -6.65000618e-01 4.17911977e-01
-1.54679686e-01 8.66427839e-01 1.83095858e-01 5.43399692e-01
-7.25782752e-01 -2.47827157e-01 -1.35413751e-01 -7.19105184e-01
-1.03788292e+00 2.67918974e-01 8.37487102e-01 5.05339168e-02
-9.91978824e-01 6.06616199e-01 -4.04934324e-02 -1.08992934e+00
1.20471619e-01 -4.96087164e-01 2.37056077e-01 -2.58320481e-01
2.05055982e-01 3.15069050e-01 2.62524784e-01 -6.96996331e-01
-2.28868619e-01 8.38981807e-01 2.16051057e-01 -9.82888341e-02
1.50142074e+00 -7.19521344e-02 -8.11589360e-02 5.83012760e-01
1.21608198e+00 4.92389388e-02 -1.42017961e+00 -1.62012413e-01
-6.06276572e-01 -7.31046379e-01 4.03784066e-02 -5.83464444e-01
-1.34125984e+00 5.28912425e-01 5.06039083e-01 3.52981508e-01
7.43907034e-01 2.26971895e-01 9.11700130e-01 -3.55243161e-02
7.34912694e-01 -7.18353927e-01 7.94323161e-02 4.33497041e-01
1.54301679e+00 -9.78905916e-01 2.95916907e-02 -7.16082335e-01
-1.09420709e-01 1.11338079e+00 5.59310317e-01 -2.43711874e-01
1.10706854e+00 2.62657195e-01 1.69486791e-01 -5.20552456e-01
-3.13555151e-01 2.92329341e-01 3.58265400e-01 7.83626437e-01
1.46682575e-01 -6.25609383e-02 1.51349485e-01 -7.19667003e-02
-4.21633720e-01 3.22434716e-02 5.82501650e-01 1.07271111e+00
-2.89951116e-02 -1.26448333e+00 -5.33648729e-01 5.07757008e-01
9.54893231e-02 1.42273784e-01 -2.24007234e-01 7.76475966e-01
-2.11818912e-03 2.71879792e-01 4.43375021e-01 -2.89189845e-01
7.86406577e-01 -1.38686568e-01 7.17586458e-01 -9.31005836e-01
-2.17526361e-01 -2.49091582e-03 -8.59149545e-02 -5.40702879e-01
-5.06846786e-01 -5.00188231e-01 -1.35645068e+00 -4.04224038e-01
-2.37508178e-01 -2.73212284e-01 7.50100434e-01 5.30623496e-01
7.55931675e-01 2.60848224e-01 8.20783198e-01 -1.78594112e+00
-1.55966327e-01 -6.86599433e-01 -4.90370750e-01 3.26310515e-01
3.75420213e-01 -8.47221315e-01 -3.43107641e-01 2.57723927e-01] | [8.78897762298584, -3.574474811553955] |
ba4b4a7a-3974-43a4-9e7f-9abc51cf3265 | exploiting-method-names-to-improve-code | 2103.11448 | null | https://arxiv.org/abs/2103.11448v2 | https://arxiv.org/pdf/2103.11448v2.pdf | Exploiting Method Names to Improve Code Summarization: A Deliberation Multi-Task Learning Approach | Code summaries are brief natural language descriptions of source code pieces. The main purpose of code summarization is to assist developers in understanding code and to reduce documentation workload. In this paper, we design a novel multi-task learning (MTL) approach for code summarization through mining the relationship between method code summaries and method names. More specifically, since a method's name can be considered as a shorter version of its code summary, we first introduce the tasks of generation and informativeness prediction of method names as two auxiliary training objectives for code summarization. A novel two-pass deliberation mechanism is then incorporated into our MTL architecture to generate more consistent intermediate states fed into a summary decoder, especially when informative method names do not exist. To evaluate our deliberation MTL approach, we carried out a large-scale experiment on two existing datasets for Java and Python. The experiment results show that our technique can be easily applied to many state-of-the-art neural models for code summarization and improve their performance. Meanwhile, our approach shows significant superiority when generating summaries for methods with non-informative names. | ['Shikun Zhang', 'Jinan Sun', 'Wei Ye', 'Rui Xie'] | 2021-03-21 | null | null | null | null | ['code-summarization'] | ['computer-code'] | [ 3.72708619e-01 1.55691102e-01 -5.20334542e-01 -4.44660664e-01
-9.45449889e-01 -3.47191006e-01 3.11058581e-01 3.90988886e-01
-7.28113949e-03 5.82003295e-01 6.88453853e-01 -3.62300158e-01
2.60976821e-01 -3.94327730e-01 -7.38647521e-01 -2.13444501e-01
1.95828229e-01 -5.93005531e-02 2.12163180e-01 8.30237120e-02
7.76964366e-01 -1.18365027e-01 -1.49650371e+00 8.22298706e-01
1.41347253e+00 1.77237481e-01 5.66957176e-01 7.27457166e-01
-7.18971848e-01 1.20961487e+00 -1.01875579e+00 -4.19212133e-01
-2.99286187e-01 -6.72782958e-01 -1.07074857e+00 -5.96885867e-02
3.68231446e-01 -2.61235833e-01 1.51032656e-01 1.16226614e+00
5.77847362e-01 1.04574978e-01 5.31364501e-01 -1.00812435e+00
-6.74357533e-01 1.41189170e+00 -9.09366131e-01 2.03805536e-01
5.44700563e-01 2.95773316e-02 1.09272063e+00 -6.37241960e-01
5.02859950e-01 1.04669297e+00 7.56768227e-01 7.78267562e-01
-1.11795270e+00 -4.19873923e-01 2.10933208e-01 7.40607232e-02
-8.49773288e-01 -5.91926396e-01 8.52554202e-01 -5.13355076e-01
1.49377441e+00 3.35865170e-01 2.96906948e-01 8.43636513e-01
4.44070995e-01 1.06049645e+00 2.40338519e-01 -3.75396520e-01
1.30038559e-01 1.83560356e-01 4.93280739e-01 8.14794958e-01
6.06561780e-01 -7.48035848e-01 -3.90202790e-01 -4.64201838e-01
9.57426950e-02 4.05441448e-02 -1.38912857e-01 7.46440217e-02
-1.30136335e+00 7.58723915e-01 5.43295518e-02 4.13916737e-01
-3.07060748e-01 3.85039240e-01 8.99910450e-01 1.57258958e-02
3.12367618e-01 6.45894170e-01 -4.73484993e-01 -4.42838758e-01
-1.13425231e+00 3.27059507e-01 9.56348956e-01 1.28443837e+00
8.89217556e-01 1.54330298e-01 -7.14154124e-01 1.11375725e+00
2.38669083e-01 1.86357632e-01 9.43874061e-01 -9.21184123e-01
9.14027631e-01 1.03082764e+00 -7.92633072e-02 -9.04828250e-01
-2.30621785e-01 -3.51157606e-01 -6.55960083e-01 -2.80397862e-01
-2.75409877e-01 -2.43268326e-01 -4.09160078e-01 1.55690253e+00
-1.82135344e-01 -2.38725662e-01 2.13849738e-01 1.55061945e-01
1.24986267e+00 8.53354037e-01 -8.57586861e-02 -4.77201194e-01
1.26799798e+00 -1.32077968e+00 -6.23775423e-01 -5.23191094e-01
9.39300656e-01 -7.80132890e-01 1.01248443e+00 4.44742516e-02
-1.05664849e+00 -5.40935636e-01 -8.24936390e-01 -1.65581688e-01
2.06862703e-01 6.74444854e-01 5.83149910e-01 2.95223922e-01
-9.04889107e-01 5.54336429e-01 -9.73644435e-01 -3.10304582e-01
4.41949785e-01 2.04195548e-02 1.01078965e-01 2.32743979e-01
-5.03946245e-01 7.08661973e-01 7.13357091e-01 -2.94575632e-01
-8.08308125e-01 -5.90709567e-01 -1.09061790e+00 4.64288265e-01
4.13585871e-01 -6.95178986e-01 1.77667975e+00 -8.69292855e-01
-1.27028358e+00 4.78341132e-01 -5.44606447e-01 -2.57438511e-01
-7.54279550e-03 -2.56491631e-01 -1.02424823e-01 -2.34551877e-01
5.17920196e-01 3.08969915e-01 4.39766079e-01 -1.29907286e+00
-7.47803569e-01 -2.86930092e-02 4.97101620e-02 -6.85916170e-02
-4.50181156e-01 1.63143948e-01 -4.63771820e-01 -7.25709021e-01
-3.49399030e-01 -6.45561874e-01 -3.31487626e-01 -6.92999959e-01
-8.24838936e-01 -5.62353790e-01 3.57291430e-01 -1.02173901e+00
1.96556067e+00 -2.01779485e+00 2.18172580e-01 -3.55430990e-01
2.35768288e-01 4.15961802e-01 -3.35046083e-01 6.44358933e-01
8.52243528e-02 2.44576439e-01 -5.28068185e-01 -3.78450781e-01
-8.21402203e-03 -2.04722926e-01 -4.10570621e-01 -2.45579332e-02
2.19837815e-01 8.20302963e-01 -9.19168293e-01 -7.18955457e-01
-3.32755625e-01 -7.79172033e-02 -9.92985368e-01 3.82939816e-01
-5.77256203e-01 1.01832747e-01 -8.14601898e-01 5.02260804e-01
5.34247577e-01 -3.82127702e-01 1.31323636e-02 2.43128780e-02
-2.50877559e-01 6.30935252e-01 -7.51266479e-01 2.03660560e+00
-8.28188360e-01 6.15143239e-01 -5.01936078e-01 -9.08583581e-01
9.71553862e-01 1.22876488e-01 1.31468028e-01 -5.31892180e-01
-1.48984328e-01 1.72792152e-01 1.35043776e-02 -9.68043983e-01
7.15581596e-01 4.49890882e-01 -4.34034079e-01 9.34973598e-01
-7.13459849e-02 -8.91365483e-02 7.12309361e-01 3.90060961e-01
1.33619189e+00 3.26359689e-01 8.30845952e-01 -2.07812309e-01
7.96678543e-01 4.53262255e-02 9.02561069e-01 9.35399473e-01
1.93221360e-01 3.49948853e-01 9.27351534e-01 -3.82489145e-01
-9.38192070e-01 -5.25760710e-01 4.06800300e-01 1.20200336e+00
-1.21559180e-01 -9.42214847e-01 -9.23936188e-01 -9.59852040e-01
-2.28500247e-01 1.21180439e+00 -4.34420794e-01 -3.60665798e-01
-8.80145669e-01 -7.69508898e-01 7.85926819e-01 5.21336257e-01
6.14620626e-01 -1.33852303e+00 -8.74148309e-01 3.70665848e-01
-3.77420276e-01 -5.62477529e-01 -8.17463040e-01 2.31006537e-02
-9.07159150e-01 -9.57710683e-01 -5.34974933e-01 -9.32928383e-01
6.55992627e-01 1.92959517e-01 1.19895613e+00 2.66643077e-01
7.35693499e-02 -6.79963231e-02 -4.25843716e-01 -3.25336486e-01
-1.08437371e+00 5.24468005e-01 -5.13237894e-01 -5.51496863e-01
3.82117480e-01 -6.30966187e-01 -2.63751417e-01 -1.45260647e-01
-8.69810283e-01 4.63200629e-01 9.66099203e-01 6.93445981e-01
1.83498815e-01 -3.41370434e-01 7.98453867e-01 -1.25170124e+00
1.00579369e+00 -5.82410038e-01 -4.47768420e-01 6.70851886e-01
-5.07496834e-01 6.92906559e-01 1.00592256e+00 -4.06271607e-01
-1.45719826e+00 1.14118196e-01 -1.78761676e-01 3.14860046e-01
1.06620893e-01 9.18840528e-01 -9.46511421e-03 3.88688713e-01
7.90855110e-01 7.24143803e-01 -1.64214611e-01 -7.55041420e-01
2.48417273e-01 9.95759904e-01 3.18944186e-01 -6.39328122e-01
5.28612494e-01 -3.03625558e-02 -5.66169858e-01 -4.83063519e-01
-8.07450175e-01 -3.32353055e-01 -3.36031258e-01 1.67286724e-01
5.72132230e-01 -7.04420745e-01 -4.97087479e-01 3.49104404e-01
-1.76522231e+00 -9.85109955e-02 1.74615514e-02 9.98189077e-02
-5.36418378e-01 7.63676226e-01 -5.81667006e-01 -4.44927692e-01
-8.72404397e-01 -1.45188618e+00 1.02258146e+00 4.93270189e-01
-4.76833850e-01 -8.01943839e-01 3.59191567e-01 1.63691059e-01
5.24048865e-01 -1.53200538e-03 1.31745636e+00 -1.03523719e+00
-4.15184170e-01 1.14426650e-01 -1.30701751e-01 3.34860623e-01
3.61047953e-01 3.20594639e-01 -5.81859469e-01 -2.67066121e-01
1.02183176e-02 -2.34393835e-01 9.66886818e-01 1.57528803e-01
1.31903780e+00 -8.62386644e-01 -4.84532893e-01 5.20393252e-01
1.31304550e+00 3.22591990e-01 3.52229863e-01 3.36787045e-01
9.20694172e-01 3.54978383e-01 4.13004071e-01 6.97473943e-01
6.25375330e-01 6.00907266e-01 2.29556456e-01 3.93188745e-01
-2.31940717e-01 -1.43975735e-01 7.17222869e-01 1.42605817e+00
2.82911003e-01 -1.53329998e-01 -1.12206626e+00 7.61526108e-01
-1.88958991e+00 -1.08370984e+00 -3.17772597e-01 1.99389708e+00
1.37618566e+00 -7.56775290e-02 3.56118865e-02 -3.52448046e-01
8.60122740e-01 1.81854352e-01 -6.46325648e-01 -6.91708207e-01
3.63062114e-01 -2.73051947e-01 7.88890012e-03 1.67298287e-01
-8.66013408e-01 7.49776423e-01 5.50098324e+00 9.31325078e-01
-8.46241415e-01 8.08264837e-02 2.65200287e-01 4.98770811e-02
-7.52603412e-01 2.20915481e-01 -9.25878525e-01 7.45398641e-01
8.60747457e-01 -9.02967513e-01 3.11623156e-01 1.17854130e+00
1.62143245e-01 -2.30056241e-01 -1.41062951e+00 7.30609477e-01
4.24047321e-01 -1.55113816e+00 4.61892337e-01 -4.00109202e-01
1.08089566e+00 -1.38409287e-02 -4.66952652e-01 6.30948603e-01
2.37901926e-01 -5.33477843e-01 7.57328331e-01 3.85867327e-01
5.00241756e-01 -6.56993270e-01 7.71572769e-01 6.58096969e-01
-1.12619495e+00 -1.82285860e-01 -4.80738491e-01 5.40242307e-02
-5.63439243e-02 5.91836929e-01 -9.16234314e-01 6.28970444e-01
3.14313143e-01 1.03852332e+00 -9.63447034e-01 1.30371213e+00
-2.40363821e-01 4.44945186e-01 2.77301013e-01 -4.06925857e-01
4.79491465e-02 2.09556386e-01 5.70194304e-01 1.65005279e+00
3.30938905e-01 -2.71637112e-01 1.65114447e-01 1.25246871e+00
-2.97466695e-01 1.57699317e-01 -5.52821040e-01 -3.11085284e-01
5.62880576e-01 1.07893944e+00 -5.99538624e-01 -5.82883000e-01
-7.05390632e-01 8.12079608e-01 4.00090963e-01 2.71225870e-01
-9.30976748e-01 -1.03343272e+00 3.69772762e-01 -4.31531221e-01
3.01143855e-01 1.76556855e-01 -4.87621784e-01 -1.37750244e+00
3.98292720e-01 -1.01876366e+00 2.51329482e-01 -6.32104397e-01
-7.97286212e-01 7.47621000e-01 3.64571102e-02 -1.23702598e+00
-4.60606635e-01 1.09004125e-01 -1.10100329e+00 6.68537796e-01
-1.32799280e+00 -9.20926094e-01 -3.45559567e-01 -9.96319652e-02
1.06961560e+00 -4.73684341e-01 7.14861572e-01 2.44769976e-01
-8.30422997e-01 6.59974277e-01 2.13841677e-01 1.37703136e-01
4.55025911e-01 -1.19666457e+00 8.53998303e-01 1.41415536e+00
-1.29232213e-01 1.18056870e+00 6.94021225e-01 -9.90782201e-01
-1.27499092e+00 -1.36203754e+00 1.03456450e+00 -2.07480043e-01
4.01590407e-01 -1.21701017e-01 -1.05813730e+00 7.95227170e-01
3.77066076e-01 -6.43972278e-01 6.44022644e-01 -3.75681855e-02
-2.29169235e-01 -1.91567838e-02 -6.29803121e-01 4.82412189e-01
7.07557559e-01 -4.28234160e-01 -9.98420238e-01 3.49893808e-01
9.90054250e-01 -4.09423977e-01 -5.17674387e-01 1.69147521e-01
3.36102039e-01 -8.84489715e-01 4.29360926e-01 -4.44980055e-01
1.14181030e+00 -2.83690065e-01 2.43164808e-01 -1.37373924e+00
-1.06848277e-01 -7.09039092e-01 -2.55257696e-01 1.73219645e+00
4.72517163e-01 -2.21467450e-01 4.93201196e-01 1.52041882e-01
-5.40591836e-01 -6.71878278e-01 -5.09860933e-01 -5.39042473e-01
-2.23526716e-01 -1.04975186e-01 7.16670871e-01 7.03004539e-01
3.17353189e-01 4.87132311e-01 -2.30423853e-01 -2.20323339e-01
3.90490532e-01 5.12796581e-01 8.64556611e-01 -9.26802814e-01
-5.10162055e-01 -6.89427376e-01 1.26473576e-01 -1.18985045e+00
5.22195280e-01 -1.19222295e+00 3.66456568e-01 -1.88630700e+00
1.05534565e+00 5.29888608e-02 1.03842206e-01 7.41908371e-01
-6.28523707e-01 -4.73053783e-01 4.39797938e-02 3.19696605e-01
-1.00706863e+00 5.99960208e-01 8.07729244e-01 -4.29717451e-01
-4.34330046e-01 4.24483120e-01 -1.03326476e+00 5.80877721e-01
6.71399534e-01 -8.80893111e-01 -4.76857811e-01 -8.42504621e-01
3.47357094e-01 3.56839001e-01 -2.32270688e-01 -8.74685585e-01
3.55245918e-01 -1.58911988e-01 -2.33746901e-01 -6.03360474e-01
-4.89275187e-01 -2.31618375e-01 1.82109214e-02 6.94426000e-01
-7.30838478e-01 2.80510515e-01 3.32971036e-01 2.63363272e-01
-2.36046821e-01 -7.84983218e-01 5.75966358e-01 -2.92169303e-01
-7.43581235e-01 -8.29131007e-02 -5.23509264e-01 3.19175243e-01
8.58904541e-01 -1.14694953e-01 -7.01930165e-01 7.53344446e-02
-1.14877904e-02 3.40890676e-01 6.29467010e-01 5.51441371e-01
6.46938980e-01 -1.15478444e+00 -9.38268661e-01 6.62749931e-02
4.11478668e-01 7.42077678e-02 1.45120814e-01 6.40608311e-01
-7.05852807e-01 4.99089658e-01 -2.67864704e-01 -3.55973512e-01
-1.25849831e+00 3.94244760e-01 6.77814856e-02 -3.73569906e-01
-5.75965464e-01 7.32231677e-01 1.11491360e-01 -4.05515403e-01
2.43142679e-01 -6.16895676e-01 -4.41338748e-01 -8.19338411e-02
7.80332685e-01 3.92166555e-01 8.31036195e-02 -2.66586274e-01
-5.31411350e-01 3.90977472e-01 -5.05621433e-01 3.81529272e-01
1.67564392e+00 6.42217845e-02 -6.88639820e-01 5.47716320e-01
1.28238463e+00 2.35321596e-01 -9.88160133e-01 -2.47586414e-01
5.45272171e-01 -2.40449876e-01 -3.66486013e-01 -7.89743543e-01
-9.18359101e-01 6.92290425e-01 -8.38905722e-02 7.56327882e-02
1.01225948e+00 1.35597512e-01 7.51037717e-01 8.02412450e-01
2.23311514e-01 -7.73628414e-01 2.25595042e-01 5.67435861e-01
8.53894651e-01 -1.04290569e+00 1.80778608e-01 -1.46601781e-01
-7.29202449e-01 1.32120502e+00 8.90462995e-01 1.92452416e-01
-2.36678168e-01 1.91679269e-01 -3.00140709e-01 -1.91212758e-01
-9.72848713e-01 2.47016832e-01 2.30685115e-01 2.80586660e-01
8.45359921e-01 -2.97094017e-01 -6.18054926e-01 9.04128075e-01
-5.27810818e-03 -8.22478756e-02 1.15560174e+00 1.02978432e+00
-6.70495033e-01 -1.07585049e+00 1.05082750e-01 8.99848521e-01
-5.83369374e-01 -4.12654817e-01 -2.82349259e-01 2.81507671e-01
-1.54700130e-01 8.52060318e-01 -2.67278880e-01 -3.55727285e-01
3.21662217e-01 -1.29525410e-02 1.04164205e-01 -1.35808313e+00
-9.12564397e-01 -3.40852886e-01 1.56125203e-01 -5.34147799e-01
-3.33984464e-01 -7.01017022e-01 -1.35863853e+00 -1.67505026e-01
-4.10943210e-01 4.64187145e-01 5.52255213e-01 9.00286376e-01
6.05176568e-01 9.69178796e-01 5.96723616e-01 -6.19307876e-01
-6.90491140e-01 -1.09261692e+00 -6.12977566e-03 2.72984117e-01
4.59408134e-01 -2.36573696e-01 -1.47675619e-01 4.40822214e-01] | [7.628077983856201, 7.941091537475586] |
63a7e85f-588c-4fa9-a813-989e767fded2 | unsupervised-foreign-object-detection-based | 2104.05326 | null | https://arxiv.org/abs/2104.05326v1 | https://arxiv.org/pdf/2104.05326v1.pdf | Unsupervised foreign object detection based on dual-energy absorptiometry in the food industry | X-ray imaging is a widely used technique for non-destructive inspection of agricultural food products. One application of X-ray imaging is the autonomous, in-line detection of foreign objects in food samples. Examples of such inclusions are bone fragments in meat products, plastic and metal debris in fish, fruit infestations. This article presents a processing methodology for unsupervised foreign object detection based on dual-energy X-ray absorptiometry (DEXA). A foreign object is defined as a fragment of material with different X-ray attenuation properties than those belonging to the food product. A novel thickness correction model is introduced as a pre-processing technique for DEXA data. The aim of the model is to homogenize regions in the image that belong to the food product and enhance contrast where the foreign object is present. In this way, the segmentation of the foreign object is more robust to noise and lack of contrast. The proposed methodology was applied to a dataset of 488 samples of meat products. The samples were acquired from a conveyor belt in a food processing factory. Approximately 60\% of the samples contain foreign objects of different types and sizes, while the rest of the samples are void of foreign objects. The results show that samples without foreign objects are correctly identified in 97% of cases, the overall accuracy of foreign object detection reaches 95%. | ['Kees Joost Batenburg', 'Tristan van Leeuwen', 'Robert van Liere', 'Vladyslav Andriiashen'] | 2021-04-12 | null | null | null | null | ['line-detection'] | ['computer-vision'] | [ 2.93798357e-01 -4.11638543e-02 1.87332078e-03 -1.41632214e-01
-2.10808650e-01 -3.30165356e-01 -1.71194226e-02 8.19247961e-01
-7.34178424e-02 8.43922049e-02 -5.01842976e-01 1.82345137e-02
-2.19849810e-01 -1.19952750e+00 -8.13611329e-01 -9.56910968e-01
-5.91904717e-03 7.86995173e-01 6.22014105e-01 -8.06964859e-02
-4.68237959e-02 6.48853242e-01 -1.53319526e+00 4.70590472e-01
7.22879469e-01 1.01893961e+00 7.79131114e-01 3.94616097e-01
-3.86418216e-02 2.85968989e-01 -4.38570917e-01 -6.94946432e-03
3.28785747e-01 -5.90039670e-01 -6.51906848e-01 7.69210935e-01
-1.06374472e-01 -2.92954117e-01 2.60916054e-01 1.36652601e+00
-1.69218741e-02 6.00055829e-02 1.06765485e+00 -7.69200742e-01
-5.65901637e-01 4.94924217e-01 -8.68092120e-01 3.76657695e-01
2.02294022e-01 1.18489087e-01 2.90691465e-01 -6.55598164e-01
6.67988002e-01 1.17702675e+00 3.39060068e-01 1.25131637e-01
-9.10963416e-01 -4.86916184e-01 -2.90968508e-01 3.12670380e-01
-1.02386141e+00 2.85793453e-01 6.48008883e-01 -5.93005598e-01
4.84160125e-01 4.28554982e-01 6.41064167e-01 2.05680370e-01
9.13494468e-01 6.90524936e-01 1.14305758e+00 -6.71817362e-01
1.81250513e-01 2.29372054e-01 4.18193996e-01 5.82047045e-01
5.98188519e-01 1.54010087e-01 9.94008109e-02 -7.13171437e-03
4.33330923e-01 3.31778049e-01 -1.08890891e-01 -2.76639998e-01
-7.42038071e-01 7.61499345e-01 3.32582474e-01 6.07851624e-01
-7.99693108e-01 -5.87457538e-01 4.49271590e-01 1.98897555e-01
7.54504979e-01 -9.11432132e-03 -4.22731280e-01 5.00195801e-01
-4.49667841e-01 1.31163478e-01 5.41611791e-01 6.51233196e-01
4.78847384e-01 -2.04161093e-01 5.35647810e-01 7.20113039e-01
5.28792799e-01 9.31203485e-01 4.36641902e-01 -2.24133492e-01
-5.26172034e-02 8.43430102e-01 -1.93192169e-01 -1.06410444e+00
-2.35253960e-01 -1.97518840e-01 -4.25709963e-01 6.23733103e-01
5.87847173e-01 2.27451280e-01 -1.10797930e+00 7.10005879e-01
1.03877640e+00 -4.49762017e-01 -7.45729683e-03 1.29729247e+00
8.99496734e-01 1.08629656e+00 4.77590784e-03 -2.52443880e-01
2.13177896e+00 -5.65733969e-01 -8.03975046e-01 7.88345337e-02
1.64788991e-01 -1.24059927e+00 6.31995618e-01 8.09185505e-01
-1.16542375e+00 -6.55675888e-01 -1.15340030e+00 6.05541170e-01
-3.03613275e-01 3.88464391e-01 4.04083669e-01 4.88358229e-01
3.33864205e-02 5.21466613e-01 -9.46129739e-01 -6.30378902e-01
4.26279418e-02 3.46080333e-01 -3.99270236e-01 -1.02685556e-01
-4.06316161e-01 6.61018729e-01 5.70835829e-01 3.33137810e-01
-6.43516302e-01 -7.41453528e-01 -6.79864883e-01 -1.71673030e-01
6.31255865e-01 -8.93759280e-02 9.12296951e-01 -1.25918198e+00
-1.26178992e+00 1.12627900e+00 4.56240416e-01 -9.70000476e-02
3.79002541e-01 -3.50472987e-01 -4.81910616e-01 6.74159229e-01
3.26885611e-01 -6.33344501e-02 7.13664055e-01 -1.69825828e+00
-7.60508597e-01 -7.61119068e-01 -2.66654998e-01 -1.73734620e-01
3.79072845e-01 3.80546272e-01 -7.43252784e-03 -6.40312970e-01
8.05984378e-01 -6.71239734e-01 8.58601257e-02 2.96707265e-02
-3.94122243e-01 -1.57343000e-01 1.29980886e+00 -8.85436356e-01
3.94660026e-01 -2.24969006e+00 -4.20205772e-01 5.41230261e-01
-1.26508564e-01 1.68164417e-01 2.40510717e-01 2.72227407e-01
-2.38491207e-01 -4.80044723e-01 -3.97660702e-01 5.69864213e-01
-3.08105081e-01 1.38803571e-01 2.32906014e-01 8.50529671e-01
1.11617029e-01 1.74164668e-01 -7.98942327e-01 -6.00716710e-01
2.97807932e-01 2.92978883e-01 8.67685750e-02 1.87542904e-02
-3.42156917e-01 3.19606364e-01 -5.49884021e-01 1.19603097e+00
1.11115324e+00 2.25923374e-01 -8.84734467e-02 -6.04075372e-01
-2.63239175e-01 -4.76475328e-01 -1.55002093e+00 9.27618265e-01
4.62470017e-02 7.60988984e-03 8.76774669e-01 -9.68951941e-01
1.03560650e+00 3.37205052e-01 7.77566135e-01 -6.97937310e-01
5.89372039e-01 2.86772996e-01 3.09685785e-02 -1.04580867e+00
1.82281435e-01 -4.48243797e-01 6.24286056e-01 4.69139040e-01
-2.64709026e-01 -2.27282360e-01 5.31809568e-01 -3.02250892e-01
5.31919181e-01 -1.65483952e-02 2.02354416e-01 -7.61893213e-01
4.37663615e-01 4.65217263e-01 2.56645977e-01 2.36480266e-01
2.36883685e-01 4.58332747e-01 -1.62031278e-01 -5.01586378e-01
-8.80033612e-01 -1.31501722e+00 -6.08990729e-01 8.54176641e-01
4.28068787e-01 4.86742407e-01 -9.89440918e-01 -4.15373415e-01
1.57451779e-01 2.25668088e-01 -5.00356436e-01 -1.32548168e-01
-7.34208465e-01 -1.10778296e+00 -2.24546254e-01 1.51084095e-01
7.50578046e-01 -1.34601593e+00 -1.03050280e+00 3.47125977e-01
-7.56634539e-03 -7.33967781e-01 -5.06913960e-02 3.60330969e-01
-1.21143425e+00 -1.45955622e+00 -6.06246293e-01 -1.12665451e+00
1.00744712e+00 3.25135738e-01 1.02692139e+00 3.09931844e-01
-8.47758770e-01 2.36054078e-01 -8.22544873e-01 -5.90930760e-01
-1.10373259e+00 -4.88642514e-01 -1.61244527e-01 1.04472131e-01
5.40202320e-01 3.03970575e-01 -6.03747249e-01 5.31237483e-01
-1.33176196e+00 -5.24220288e-01 5.22766888e-01 7.02867806e-01
9.55728590e-01 8.83713186e-01 -5.51096648e-02 -9.94915605e-01
1.66214049e-01 -5.43899417e-01 -6.01669490e-01 -4.97700050e-02
-2.80500859e-01 -4.67339456e-01 4.39708203e-01 -6.72425687e-01
-1.07805765e+00 -2.39921376e-01 -2.08029762e-01 8.63827094e-02
-7.11697876e-01 4.77711707e-01 -2.46541664e-01 1.15651153e-01
6.12884760e-01 -1.39022395e-01 3.10794234e-01 -7.32849121e-01
-4.67074513e-01 5.28635740e-01 6.65911913e-01 -3.61968160e-01
3.21984202e-01 5.40820062e-01 2.30345130e-02 -1.19059169e+00
-1.77213520e-01 -6.38373077e-01 -4.66789812e-01 -3.63061100e-01
1.01825047e+00 -4.01292503e-01 -6.57388806e-01 6.15464270e-01
-7.30375230e-01 -3.69734764e-02 -4.47611660e-01 1.19347060e+00
-3.02881211e-01 4.39088643e-01 -1.10259497e+00 -6.12937450e-01
-4.86487925e-01 -1.25941539e+00 7.92835832e-01 1.53571323e-01
1.43663839e-01 -6.12708032e-01 -3.25167358e-01 3.51524830e-01
-1.08087942e-01 5.07951438e-01 1.25250602e+00 -4.50545639e-01
-3.19758296e-01 -4.03676212e-01 2.62475252e-01 4.53040272e-01
3.90436560e-01 3.10008079e-01 -5.36491096e-01 -2.74669975e-01
7.24871635e-01 1.21820427e-01 6.03015840e-01 8.96665931e-01
4.36774909e-01 -3.59441247e-03 -5.09935141e-01 2.12361455e-01
1.72871256e+00 8.00099254e-01 4.07265604e-01 4.17251319e-01
1.71890095e-01 1.00587964e+00 1.36523998e+00 2.64769107e-01
-3.80966067e-01 4.57290441e-01 6.94967151e-01 -4.76769298e-01
-7.87611082e-02 3.48334551e-01 2.55577415e-01 6.29271090e-01
-2.75299251e-01 -2.94480890e-01 -4.79605526e-01 4.81196344e-01
-1.06615782e+00 -8.99543703e-01 -9.39548671e-01 1.80432534e+00
4.89726216e-01 4.81562614e-02 2.73598582e-01 7.76044428e-01
8.89383733e-01 -8.41101289e-01 -3.16035122e-01 -2.92439640e-01
1.78927220e-02 3.95336330e-01 4.97716486e-01 4.31200355e-01
-1.04646850e+00 3.06134224e-01 5.91324091e+00 7.59925902e-01
-1.16087258e+00 1.47957325e-01 4.29235280e-01 4.40293103e-01
7.56858587e-02 -4.99420404e-01 -3.90929163e-01 6.13430142e-01
8.46284926e-02 5.78112483e-01 -2.79544830e-01 7.51466215e-01
1.64929077e-01 -1.08518672e+00 -6.77698195e-01 5.16493917e-01
1.39632255e-01 -6.62469029e-01 -1.62441090e-01 -1.19269326e-01
4.16168332e-01 -3.63522172e-01 -9.38890427e-02 -5.61448753e-01
-7.49595091e-02 -4.29639935e-01 1.22401261e+00 2.02372849e-01
4.02691603e-01 -8.32849741e-01 1.03866708e+00 1.84099421e-01
-1.07424331e+00 9.70835146e-03 -3.78754944e-01 1.53773069e-01
2.22076133e-01 1.00057495e+00 -6.02718592e-01 5.15242994e-01
1.06786942e+00 -5.07992133e-02 -2.79384404e-01 9.90880311e-01
1.80030957e-01 5.98628998e-01 -6.59708500e-01 1.24574162e-01
2.70269692e-01 -8.61932635e-01 4.61522430e-01 1.13676000e+00
3.98659557e-01 4.37367976e-01 2.31543168e-01 9.41915929e-01
5.98280966e-01 6.23265266e-01 -3.79236102e-01 1.95432410e-01
-6.58946335e-02 1.02265847e+00 -1.71209514e+00 -3.27133685e-01
-3.36900234e-01 7.89548516e-01 -7.77960837e-01 -6.81741685e-02
-6.76172376e-01 -1.34044826e-01 -9.48256776e-02 6.67671025e-01
3.89950246e-01 -2.50239447e-02 -1.08670294e-01 -3.63296866e-01
1.85135394e-01 -8.36242616e-01 4.79465842e-01 -6.95331931e-01
-1.30398500e+00 4.74197894e-01 3.84601444e-01 -1.13564956e+00
8.61866586e-03 -9.45932746e-01 -3.86055917e-01 5.82301378e-01
-8.70399594e-01 -1.31224680e+00 -3.81024122e-01 2.58583754e-01
7.33859777e-01 -3.34716681e-03 5.46639621e-01 2.71948546e-01
-1.77229851e-01 -5.88925928e-02 4.11162972e-01 -1.74110174e-01
7.79103935e-02 -1.08568549e+00 -5.32606483e-01 7.43548393e-01
-3.36456180e-01 2.21028164e-01 1.18316138e+00 -1.03871059e+00
-1.37836671e+00 -7.22148776e-01 4.22614157e-01 7.26776272e-02
1.62675530e-01 -3.00675146e-02 -9.78598058e-01 3.82571340e-01
1.41500071e-01 1.25659168e-01 6.00521982e-01 -6.28850996e-01
4.22171831e-01 -1.91031396e-03 -1.78897238e+00 -3.42525691e-01
2.20163658e-01 1.65380523e-01 -4.86334443e-01 4.53921884e-01
-2.34535318e-02 -5.16544163e-01 -1.00505602e+00 4.13240045e-01
4.66085881e-01 -1.05489850e+00 1.11817491e+00 -2.19214633e-02
3.56699079e-01 -4.29641545e-01 9.85972881e-02 -8.79077077e-01
-3.24723810e-01 2.27464035e-01 3.40991139e-01 1.13884032e+00
-7.95138627e-02 -3.33096266e-01 7.24507451e-01 -5.80481775e-02
-4.51460510e-01 -3.54931742e-01 -5.96788168e-01 -7.72546411e-01
-1.00701042e-01 3.78187597e-02 4.51863140e-01 7.69611180e-01
-2.85811871e-01 -2.09750518e-01 3.23416054e-01 3.83689940e-01
6.96191788e-01 3.63512486e-01 2.99335152e-01 -1.31411648e+00
-3.87134552e-01 2.45366199e-03 -5.31523705e-01 -4.14560825e-01
-3.01386327e-01 -7.27288544e-01 1.25360727e-01 -1.51011360e+00
1.98091000e-01 -5.93117774e-01 2.83190042e-01 6.67144135e-02
-1.46608585e-02 5.54245234e-01 7.20968693e-02 2.34604701e-01
3.15095603e-01 -9.67462510e-02 1.44050872e+00 -4.10095751e-01
-3.04112464e-01 4.37097132e-01 3.07076145e-03 9.78998303e-01
6.90932691e-01 -4.42135125e-01 -1.14638649e-01 -2.50936568e-01
-2.02186123e-01 -5.63464537e-02 3.41532022e-01 -8.83868456e-01
-2.85454333e-01 -1.61456853e-01 4.74566579e-01 -1.11705422e+00
2.60631349e-02 -1.53410792e+00 7.10242212e-01 1.07896304e+00
3.83854747e-01 -5.00104972e-04 1.44883916e-01 1.49807602e-01
-2.17847213e-01 -9.52500224e-01 1.12821817e+00 -5.38590074e-01
-4.61202353e-01 -1.52411118e-01 -5.53449571e-01 -6.93574190e-01
1.54390109e+00 -5.32152891e-01 1.02374166e-01 4.96827453e-01
-8.90495300e-01 -3.20777923e-01 4.25808191e-01 1.65281929e-02
5.23772955e-01 -1.08985305e+00 -7.31100023e-01 6.56760156e-01
4.11273204e-02 1.18435793e-01 3.46673071e-01 7.69702733e-01
-1.28663301e+00 -4.22014147e-02 -4.45242316e-01 -7.20128894e-01
-1.72395086e+00 7.74593890e-01 2.50596315e-01 1.35942414e-01
-9.95298028e-01 4.92531538e-01 2.33780131e-01 8.59427229e-02
-3.08659613e-01 -9.88724947e-01 -3.98734659e-01 9.77727994e-02
6.04360104e-01 7.86784470e-01 4.71349210e-01 -9.45507109e-01
-8.53049979e-02 8.28802466e-01 7.66680622e-03 3.49657565e-01
1.48022974e+00 -8.73882920e-02 -3.58710498e-01 5.22282779e-01
8.63268256e-01 1.09278716e-01 -8.65938544e-01 -2.72656921e-02
-3.06602120e-01 -5.29564142e-01 -6.98635448e-03 -8.03784311e-01
-1.33981204e+00 5.32622457e-01 1.10872996e+00 5.67288935e-01
1.24603748e+00 3.82824242e-01 6.56222045e-01 -1.72461063e-01
3.20770323e-01 -1.26541650e+00 -4.76482091e-03 -8.33006725e-02
6.84248030e-01 -1.05721939e+00 1.96789876e-01 -1.14468646e+00
-2.37863317e-01 1.34608483e+00 4.50505912e-01 -4.51225281e-01
7.39070773e-01 5.45125782e-01 1.66761339e-01 -7.86450744e-01
7.03851506e-02 -8.10631439e-02 -5.20689087e-03 9.56058741e-01
1.79814041e-01 2.01324046e-01 -5.84174335e-01 5.97040057e-01
5.95146827e-02 -1.17889822e-01 2.61724621e-01 1.31648910e+00
-7.36784756e-01 -5.30670226e-01 -1.21798599e+00 5.24454474e-01
-8.72024775e-01 5.44258833e-01 2.00385526e-01 8.53575170e-01
7.27204680e-01 1.00704968e+00 3.31883430e-01 2.98352242e-01
5.16691327e-01 -3.37857544e-01 7.71532178e-01 -4.01872069e-01
-8.26625824e-01 6.10802174e-01 -3.79546285e-01 -5.19901626e-02
-5.44870138e-01 -7.67360747e-01 -1.66105354e+00 -1.61839217e-01
-8.10188234e-01 1.41723156e-01 9.59938407e-01 7.96187222e-01
-6.68329775e-01 5.11959851e-01 5.77014089e-01 -4.05906081e-01
-3.37775588e-01 -9.90305781e-01 -1.38660550e+00 6.76171780e-01
9.26737338e-02 -8.51148963e-01 -3.29913616e-01 7.54580140e-01] | [7.379330635070801, 1.7867361307144165] |
7ee7aad1-697d-41e6-b107-be8ef96b49a2 | intrinsic-relationship-reasoning-for-small | 2009.00833 | null | https://arxiv.org/abs/2009.00833v1 | https://arxiv.org/pdf/2009.00833v1.pdf | Intrinsic Relationship Reasoning for Small Object Detection | The small objects in images and videos are usually not independent individuals. Instead, they more or less present some semantic and spatial layout relationships with each other. Modeling and inferring such intrinsic relationships can thereby be beneficial for small object detection. In this paper, we propose a novel context reasoning approach for small object detection which models and infers the intrinsic semantic and spatial layout relationships between objects. Specifically, we first construct a semantic module to model the sparse semantic relationships based on the initial regional features, and a spatial layout module to model the sparse spatial layout relationships based on their position and shape information, respectively. Both of them are then fed into a context reasoning module for integrating the contextual information with respect to the objects and their relationships, which is further fused with the original regional visual features for classification and regression. Experimental results reveal that the proposed approach can effectively boost the small object detection performance. | ['Lin Ma', 'Yonghong Tian', 'Kui Fu', 'Kai Mu', 'Jia Li'] | 2020-09-02 | null | null | null | null | ['small-object-detection'] | ['computer-vision'] | [ 6.86886385e-02 -2.37066790e-01 -7.62946308e-02 -4.24154460e-01
1.84610542e-02 -3.23103130e-01 3.12704355e-01 5.19269407e-01
9.66560915e-02 2.99689740e-01 2.21895456e-01 2.83698589e-01
-3.65358263e-01 -8.19962919e-01 -6.36504233e-01 -6.45974398e-01
-8.81552026e-02 1.77091673e-01 6.49301171e-01 6.67438805e-02
3.00450534e-01 3.63732785e-01 -1.79956782e+00 3.90590787e-01
9.79431510e-01 1.32026911e+00 5.28878808e-01 2.94430047e-01
-3.65312845e-01 1.09112811e+00 -5.68351448e-01 6.01382144e-02
-8.15476105e-03 -3.63008231e-01 -2.65406787e-01 5.17040193e-01
5.09260774e-01 -2.40549952e-01 -3.05515975e-01 1.09423912e+00
2.64582992e-01 2.82074422e-01 4.80644047e-01 -1.03930807e+00
-6.94698155e-01 4.04851526e-01 -9.89220440e-01 4.23661143e-01
4.41554338e-01 -1.08839776e-02 9.38602448e-01 -1.04316223e+00
3.60440165e-01 1.43427157e+00 3.98210734e-01 -6.06911629e-02
-8.72985482e-01 -7.76827455e-01 7.13366628e-01 5.07563591e-01
-1.78935349e+00 -2.91477382e-01 9.74971950e-01 -3.38170499e-01
4.60713297e-01 2.88448423e-01 9.64241624e-01 5.91353059e-01
-4.29379284e-01 9.62450564e-01 7.93521106e-01 -3.08473408e-01
1.01921365e-01 4.04890060e-01 1.15312465e-01 7.65906394e-01
4.79101539e-01 -2.88147330e-01 -4.80859250e-01 8.00458342e-02
8.14959705e-01 4.39868629e-01 -8.11122432e-02 -5.25681555e-01
-1.18673587e+00 5.22759378e-01 8.74372542e-01 4.19492811e-01
-1.79632410e-01 -7.34314620e-02 1.42834457e-02 -3.00459653e-01
2.33478025e-01 -1.03105359e-01 -7.43684024e-02 4.97012109e-01
-8.55063498e-01 6.95992187e-02 4.02888477e-01 1.33636725e+00
9.88342464e-01 -3.79969150e-01 -2.23532245e-01 9.04992700e-01
5.63903987e-01 4.79339987e-01 2.03362241e-01 -7.13701844e-01
6.91069543e-01 1.21114099e+00 7.48158619e-02 -1.75536275e+00
-3.18227381e-01 -6.00521624e-01 -6.45443976e-01 -2.92546093e-01
1.88472942e-01 2.30311438e-01 -4.66900826e-01 1.61558425e+00
7.04872608e-01 6.81643724e-01 -1.43596038e-01 9.87026155e-01
1.19510221e+00 5.27936876e-01 1.77681103e-01 -1.38118312e-01
1.68576610e+00 -1.03931510e+00 -6.34549081e-01 -5.38161039e-01
3.80685061e-01 -7.09793746e-01 7.79980063e-01 -7.28853745e-03
-8.81386697e-01 -8.12113822e-01 -9.60919619e-01 -9.47265401e-02
-3.51402342e-01 6.76865041e-01 5.78718603e-01 3.20699364e-01
-5.45176208e-01 9.76838097e-02 -2.96568960e-01 -3.43910575e-01
6.64393961e-01 2.33273491e-01 -1.89653009e-01 -2.22205862e-01
-8.26766074e-01 3.39446545e-01 6.72071815e-01 4.55589861e-01
-6.46228909e-01 -4.73006248e-01 -9.40905333e-01 2.78962731e-01
7.36249387e-01 -6.52375698e-01 5.47234654e-01 -9.03160870e-01
-5.55865288e-01 6.16865218e-01 -3.46903741e-01 9.20768455e-02
1.99456826e-01 1.16517954e-01 -5.04649580e-01 3.56205761e-01
3.66135299e-01 6.29421115e-01 9.63121772e-01 -1.56843328e+00
-1.12447250e+00 -6.31344259e-01 2.55696326e-01 5.92885196e-01
-6.44106746e-01 1.16258308e-01 -7.17911899e-01 -7.94002175e-01
4.82399970e-01 -4.82735366e-01 -8.43333304e-02 2.70148873e-01
-4.26109493e-01 -2.96325475e-01 1.06311727e+00 -6.85490966e-01
1.32876968e+00 -2.29624224e+00 2.30873957e-01 5.03240347e-01
3.13318610e-01 4.68386412e-02 -7.70409629e-02 1.81356762e-02
5.68613261e-02 -2.83086568e-01 -5.51733859e-02 -7.71096870e-02
-2.79185861e-01 1.23115949e-01 -1.45058215e-01 2.55348325e-01
9.05278623e-02 9.14483428e-01 -1.11164045e+00 -8.61681700e-01
4.67660636e-01 4.45481300e-01 -2.86396027e-01 2.06808224e-01
-1.67962804e-01 3.82717133e-01 -7.94744670e-01 8.70425463e-01
9.05027032e-01 -2.40545705e-01 1.95082426e-01 -4.98800963e-01
1.86346486e-01 -2.53484964e-01 -1.52293825e+00 1.52411032e+00
-1.83131695e-01 3.46209764e-01 1.27430543e-01 -1.18440962e+00
1.07472610e+00 -5.89529760e-02 3.83956432e-01 -6.35001183e-01
5.55156618e-02 -3.97039726e-02 -1.80445135e-01 -7.28495657e-01
3.68636727e-01 1.83093220e-01 6.08117320e-02 2.62052137e-02
-1.68718472e-01 2.68845588e-01 2.55907327e-01 2.94879943e-01
5.70811987e-01 5.12106642e-02 3.41501981e-01 -1.72494039e-01
9.01024461e-01 -2.33714625e-01 6.57382548e-01 4.83549893e-01
-7.22826943e-02 4.69495773e-01 2.53659695e-01 -3.41444612e-01
-6.74922287e-01 -1.17048657e+00 -3.05700824e-02 1.00187159e+00
9.21747744e-01 -6.57092154e-01 -6.40108824e-01 -6.46675169e-01
1.05962798e-01 3.12875301e-01 -7.84988761e-01 -1.51469573e-01
-4.52889532e-01 -4.90004331e-01 1.79785937e-02 9.43883121e-01
6.19163573e-01 -9.81496096e-01 -3.58990043e-01 4.01395140e-03
-2.86492229e-01 -1.25699461e+00 -5.32414436e-01 -2.71656901e-01
-7.74042130e-01 -1.20171213e+00 -4.43886399e-01 -1.02542818e+00
1.18555403e+00 9.00334001e-01 6.36997759e-01 5.34867227e-01
-5.18362999e-01 2.64523506e-01 -3.92931372e-01 -3.31500173e-01
3.15822721e-01 -3.40017349e-01 4.07558819e-03 3.92620951e-01
3.84936422e-01 -5.94322443e-01 -8.85654211e-01 5.85033476e-01
-6.76845551e-01 1.55912384e-01 7.63505757e-01 5.89352131e-01
6.39651358e-01 5.12316942e-01 2.32599929e-01 -6.18579805e-01
1.98475823e-01 -5.66388309e-01 -3.34563643e-01 4.96927351e-01
2.23340057e-02 -3.78081590e-01 4.26139385e-01 -5.23437560e-01
-1.23225701e+00 -6.32416010e-02 4.66483802e-01 -6.78305686e-01
-3.89766395e-01 3.33505541e-01 -5.65240264e-01 -1.07114360e-01
2.60951787e-01 4.52915311e-01 -3.43188405e-01 -5.88045835e-01
2.99252778e-01 5.59866607e-01 3.14053148e-01 -6.64957464e-01
8.72394025e-01 6.54551923e-01 8.42411220e-02 -7.55192339e-01
-1.08666956e+00 -6.52139425e-01 -1.04197872e+00 -2.26357937e-01
8.76868069e-01 -1.17755902e+00 -5.52688539e-01 2.60519505e-01
-8.26104105e-01 2.53552347e-01 -2.65083104e-01 3.29473287e-01
-1.94341481e-01 6.25250518e-01 -4.28104013e-01 -7.07702398e-01
1.89246476e-01 -8.54230642e-01 1.17554581e+00 6.88121080e-01
7.50833675e-02 -9.34370816e-01 -6.36879325e-01 5.99418700e-01
3.87239046e-02 5.60493469e-02 8.47477973e-01 -3.51257592e-01
-9.50819910e-01 -1.53870046e-01 -8.35215986e-01 -6.06840104e-02
3.21548998e-01 -8.22950676e-02 -7.91691542e-01 -9.08774659e-02
-7.84226656e-02 1.55705750e-01 6.72260523e-01 3.34743977e-01
1.45754206e+00 -2.77061343e-01 -7.07719982e-01 5.77766955e-01
1.24298680e+00 1.80657208e-01 2.29839072e-01 -3.16151492e-02
1.23309362e+00 9.88408148e-01 1.07505631e+00 4.91755277e-01
4.42280501e-01 6.08578265e-01 2.86625892e-01 -4.62202467e-02
-2.82313913e-01 -5.08216560e-01 7.11111277e-02 7.49975920e-01
-1.50674090e-01 1.79880291e-01 -6.64786100e-01 4.16497737e-01
-2.11023927e+00 -1.06561434e+00 -2.12732792e-01 1.94105899e+00
3.60955685e-01 -1.37707135e-02 6.47764951e-02 -1.07610315e-01
9.96880114e-01 1.10399768e-01 -5.57898521e-01 3.22159231e-01
-3.16021264e-01 -4.53431875e-01 1.26690581e-01 -7.16418251e-02
-9.91991639e-01 8.87462020e-01 5.50874805e+00 1.25724089e+00
-5.71633101e-01 5.31856455e-02 5.51713407e-01 -9.79568884e-02
-2.75456637e-01 1.22010380e-01 -8.42392683e-01 6.03281140e-01
3.27581614e-02 -5.53392153e-03 3.17295521e-01 8.57193530e-01
2.07996666e-01 -3.44079524e-01 -9.32353258e-01 1.00163579e+00
2.64936596e-01 -1.03165174e+00 3.35750818e-01 -1.81239862e-02
6.90337300e-01 -7.88414717e-01 1.14727020e-03 -1.30544195e-03
-8.58724043e-02 -9.59948480e-01 9.63628531e-01 6.29050434e-01
4.74382371e-01 -8.15935314e-01 5.47335207e-01 5.69372892e-01
-1.92862630e+00 -5.71568429e-01 -6.48646116e-01 -8.40448439e-02
-2.14884773e-01 4.18846309e-01 -4.69542623e-01 6.30831957e-01
8.39473367e-01 1.18614626e+00 -9.11285937e-01 1.05273807e+00
-4.49186921e-01 1.40295446e-01 -2.56894737e-01 -9.22143683e-02
-4.37057018e-02 -4.49043453e-01 4.66642559e-01 9.58570302e-01
2.78587312e-01 2.51697063e-01 4.98592645e-01 1.10998869e+00
2.56383240e-01 2.88829684e-01 -3.53163749e-01 2.79880345e-01
7.28075922e-01 1.36875963e+00 -1.07762718e+00 -5.79676807e-01
-7.63810754e-01 7.95333862e-01 4.07619447e-01 4.19298768e-01
-8.76878202e-01 -3.44443828e-01 4.87546921e-01 1.43968523e-01
6.14555776e-01 -9.92314741e-02 -2.51265317e-01 -1.28282166e+00
4.12329555e-01 -4.54237163e-01 5.43501675e-01 -9.92941320e-01
-1.19333744e+00 7.88050368e-02 2.01465651e-01 -1.34335828e+00
4.08640295e-01 -4.30113435e-01 -7.12964654e-01 6.96653426e-01
-1.45551789e+00 -1.56843102e+00 -7.19219148e-01 7.08861172e-01
6.11818314e-01 -9.82712060e-02 3.92492294e-01 2.66052365e-01
-7.64554083e-01 2.82080173e-01 -1.14411160e-01 2.51052171e-01
2.84894019e-01 -9.10091043e-01 -2.50088245e-01 8.42969179e-01
2.74094224e-01 9.19202566e-01 2.59081960e-01 -8.11860859e-01
-1.13709593e+00 -1.18439364e+00 5.21280050e-01 -2.91844189e-01
4.26621258e-01 -5.21104395e-01 -9.03919578e-01 5.55418849e-01
-3.02326173e-01 2.83532828e-01 5.47479808e-01 1.96008489e-01
-4.03978795e-01 -3.35325330e-01 -1.00886977e+00 5.26100338e-01
1.53436863e+00 -5.58775067e-01 -7.61176705e-01 2.56133556e-01
6.47105634e-01 -2.02437177e-01 -6.68531418e-01 4.85602021e-01
4.66190696e-01 -1.03643632e+00 1.49135661e+00 -2.14843139e-01
2.70346910e-01 -8.02118063e-01 -2.50741512e-01 -9.77374017e-01
-5.72450101e-01 2.42333636e-01 -1.55761644e-01 1.50068521e+00
-6.92060515e-02 -2.64251471e-01 6.69623733e-01 2.86368638e-01
1.66608542e-01 -5.07700741e-01 -6.35726511e-01 -5.64773738e-01
-5.83486617e-01 -3.00461888e-01 7.76989341e-01 9.82619703e-01
-4.78892289e-02 3.65291089e-01 -1.31783679e-01 4.60775882e-01
8.43647599e-01 7.66080439e-01 5.48356473e-01 -1.23791885e+00
-1.97462097e-01 -4.53771144e-01 -6.97914660e-01 -1.14482379e+00
-5.77067845e-02 -6.94543481e-01 -1.20212607e-01 -1.44604683e+00
6.85947359e-01 -7.14642525e-01 -4.85659480e-01 2.94096202e-01
-5.06771803e-01 2.70989239e-01 1.80243831e-02 2.63689071e-01
-1.02506471e+00 6.12945914e-01 1.20669937e+00 -2.80770689e-01
-7.03926086e-02 -1.01097651e-01 -8.35149407e-01 8.17267418e-01
4.38007861e-01 -9.15191621e-02 -5.98977864e-01 -3.40931296e-01
-9.73614454e-02 -8.20314214e-02 7.94410467e-01 -1.28911674e+00
3.78547847e-01 -3.19725722e-01 1.01190519e+00 -7.66484976e-01
3.40571880e-01 -1.10169971e+00 -1.03886850e-01 2.52081752e-01
-1.50208861e-01 -4.73741323e-01 4.94291447e-02 9.21203792e-01
-2.77211100e-01 -2.33069524e-01 4.75058913e-01 -2.63850689e-01
-1.00938463e+00 3.78894717e-01 5.30151166e-02 -1.28168806e-01
1.42802846e+00 -4.77012128e-01 -1.02824308e-01 -9.99881253e-02
-8.79679263e-01 4.98270899e-01 4.03121352e-01 5.35970271e-01
9.29890573e-01 -1.42876673e+00 -4.36584383e-01 4.90752995e-01
4.43668962e-01 1.94053754e-01 6.74795508e-01 7.85545886e-01
-2.09893420e-01 2.60234535e-01 -1.93098202e-01 -8.00046504e-01
-1.57511675e+00 9.72263098e-01 1.36168718e-01 3.32901239e-01
-4.99892056e-01 9.40521419e-01 8.85047376e-01 -1.89520791e-01
2.00065091e-01 -3.88336271e-01 -6.24667704e-01 2.49115646e-01
6.74390435e-01 2.55437106e-01 -4.74750072e-01 -1.02239382e+00
-4.71510708e-01 1.12223959e+00 9.78403986e-02 5.19809961e-01
1.22808492e+00 -5.54596841e-01 -3.12111586e-01 3.67988944e-01
9.18635190e-01 1.88517824e-01 -1.23648643e+00 -7.41238177e-01
-3.61071452e-02 -1.04553044e+00 -1.96230084e-01 -4.26400781e-01
-1.22304535e+00 1.01201046e+00 4.02207017e-01 5.76274060e-02
1.23779547e+00 4.56744939e-01 4.44331855e-01 1.02546364e-01
4.13916022e-01 -1.02910542e+00 3.61198962e-01 -1.22389654e-02
6.13892913e-01 -1.11783421e+00 2.34143123e-01 -1.14076757e+00
-4.14027035e-01 1.01620746e+00 9.33451712e-01 -1.13053963e-01
5.50210416e-01 1.06369801e-01 -4.45745021e-01 -3.01420659e-01
-3.94465804e-01 -4.29122657e-01 5.57240903e-01 8.64210308e-01
6.17605001e-02 9.35698226e-02 6.48539737e-02 8.39511693e-01
2.61078000e-01 -1.88091710e-01 -2.03061640e-01 6.68477714e-01
-7.14113057e-01 -7.26532340e-01 -5.90127409e-01 3.97610843e-01
1.84079424e-01 -1.05719939e-02 -3.79887700e-01 4.14350152e-01
7.00984001e-01 9.28657115e-01 3.59368056e-01 -3.68838012e-01
3.92592937e-01 -5.03000796e-01 4.90411878e-01 -6.69861794e-01
-1.93625212e-01 1.21047519e-01 -1.48064792e-01 -4.46630388e-01
-4.41559345e-01 -7.82641828e-01 -1.37854469e+00 -1.58111863e-02
-3.81532311e-01 4.36492823e-03 2.08586439e-01 8.87161791e-01
2.77410954e-01 5.65655768e-01 5.79192936e-01 -6.71072543e-01
-5.77934645e-02 -7.02499151e-01 -8.29604566e-01 6.65027201e-01
1.35328576e-01 -9.28709745e-01 -8.87837484e-02 -2.50399653e-02] | [10.0711669921875, 1.6962300539016724] |
1ee84cdf-ba70-46cd-a623-9aa88d3ecf8c | automatic-road-crack-detection-using-random | null | null | https://ieeexplore.ieee.org/document/7471507/similar#similar | https://ieeexplore.ieee.org/document/7471507/similar#similar | Automatic Road Crack Detection Using Random Structured Forests | Cracks are a growing threat to road conditions and
have drawn much attention to the construction of intelligent
transportation systems. However, as the key part of an intelli-
gent transportation system, automatic road crack detection has
been challenged because of the intense inhomogeneity along the
cracks, the topology complexity of cracks, the inference of noises
with similar texture to the cracks, and so on. In this paper , we
propose CrackForest, a novel road crack detection framework
based on random structured forests, to address these issues. Our
contributions are shown as follows: 1) apply the integral channel
features to redefine the tokens that constitute a crack and get
better representation of the cracks with intensity inhomogeneity;
2) introduce random structured forests to generate a high-
performance crack detector, which can identify arbitrarily com-
plex cracks; and 3) propose a new crack descriptor to characterize
cracks and discern them from noises effectively. In addition, our
method is faster and easier to parallel. Experimental results prove
the state-of-the-art detection precision of CrackForest compared
with competing methods. | ['and Zhensong Chen', 'Fan Meng', 'Zhiquan Qi', 'Limeng Cui', 'Yong Shi'] | 2016-05-18 | null | null | null | ieee-transactions-on-intelligent-17 | ['crack-segmentation'] | ['computer-vision'] | [ 1.19499221e-01 -4.25676167e-01 8.47844779e-02 9.35539417e-03
-4.81310457e-01 -3.70960444e-01 2.81491816e-01 -6.59305006e-02
5.23293298e-03 4.70210016e-01 3.44753452e-02 -2.83095688e-01
7.64781311e-02 -1.35448897e+00 -5.62628210e-01 -7.59357750e-01
3.41559172e-01 2.83873022e-01 1.06831491e+00 -3.09742957e-01
7.54251301e-01 6.82628810e-01 -1.91737008e+00 2.72005618e-01
1.22889066e+00 8.38294387e-01 2.79159337e-01 5.08376241e-01
-2.33151510e-01 3.25252742e-01 3.65119316e-02 4.99713719e-02
-2.40949579e-02 -4.86305691e-02 -5.86438715e-01 1.40078172e-01
3.83810282e-01 -3.42063904e-01 -3.98236185e-01 9.72784877e-01
4.13762182e-01 -2.39510357e-01 1.00213373e+00 -1.01255143e+00
-1.79754272e-01 3.76233876e-01 -9.01047826e-01 1.98837072e-01
3.11875671e-01 5.39573766e-02 8.10440540e-01 -1.31426203e+00
4.78537202e-01 1.25744462e+00 8.99011970e-01 1.76806644e-01
-7.64332414e-01 -8.32112849e-01 8.02390724e-02 5.05511820e-01
-1.45415199e+00 -2.54041702e-01 1.06624556e+00 -5.58297813e-01
2.44029269e-01 2.20948756e-01 5.98537087e-01 7.31761396e-01
2.79502198e-02 6.52474642e-01 1.07614279e+00 -3.65780771e-01
4.91899252e-02 -2.15623841e-01 3.94234806e-01 1.18121731e+00
5.10434091e-01 8.75962339e-03 -1.07334174e-01 -4.71200533e-02
6.38822496e-01 5.87244593e-02 -2.46535495e-01 -2.34893769e-01
-9.16977823e-01 8.64519954e-01 5.25292814e-01 1.72439083e-01
-1.42937869e-01 1.70541659e-01 1.50216147e-01 -1.32087037e-01
8.30654427e-02 -2.63974577e-01 1.77150995e-01 1.42312184e-01
-7.06193209e-01 3.83523285e-01 3.48424643e-01 5.91155231e-01
1.15146387e+00 -1.91569135e-01 1.86433747e-01 8.16048682e-01
5.82563758e-01 9.11525190e-01 -2.34140661e-02 -7.74158716e-01
6.84696376e-01 7.86215305e-01 -4.60095517e-02 -1.35885870e+00
-4.78424042e-01 -1.25134557e-01 -9.57864404e-01 3.38641822e-01
4.77034271e-01 5.40478267e-02 -9.21375215e-01 7.19377100e-01
4.78681862e-01 3.10890019e-01 -3.28674734e-01 6.72670007e-01
7.97785699e-01 5.39754748e-01 -4.56933856e-01 2.33313948e-01
1.33008587e+00 -8.67155910e-01 -4.40808892e-01 -4.72012013e-02
2.93814987e-01 -8.20101023e-01 1.34148824e+00 3.47797334e-01
-5.27119100e-01 -4.91961390e-01 -1.08814788e+00 2.23414466e-01
-4.48713094e-01 1.59547791e-01 3.35448772e-01 7.28867054e-01
-4.72303957e-01 8.25817361e-02 -7.50922740e-01 -2.73916483e-01
5.85645199e-01 -2.24284977e-01 1.29616559e-01 -4.55033690e-01
-1.17290127e+00 6.92930400e-01 -8.81658271e-02 5.02380669e-01
-7.14673817e-01 -3.07633758e-01 -7.16730595e-01 -2.24992082e-01
4.99410719e-01 -1.76863775e-01 7.07277358e-01 -7.59444684e-02
-1.15000832e+00 3.92470032e-01 -3.19197744e-01 2.01024935e-01
4.55394655e-01 -3.40545699e-02 -3.95513415e-01 2.52979517e-01
3.78756464e-01 1.58462711e-02 8.45027566e-01 -1.66666961e+00
-7.87705719e-01 -3.03258508e-01 -3.63664776e-01 -1.39213711e-01
4.28083465e-02 -2.50854611e-01 -2.91730762e-01 -3.49175960e-01
5.83602011e-01 -8.07911932e-01 -3.48416179e-01 1.62315995e-01
-8.78994584e-01 -2.64602751e-01 1.28212464e+00 -3.28633785e-01
1.34245241e+00 -1.97196269e+00 -3.91356498e-01 7.62610137e-01
4.55033958e-01 1.95593014e-01 1.67394236e-01 6.58273518e-01
2.61955887e-01 2.17050448e-01 -7.18952417e-01 2.29092374e-01
-3.35021764e-01 3.03299814e-01 -2.75689960e-01 4.48492795e-01
2.85952479e-01 2.72006929e-01 -7.40476251e-01 -7.86548257e-01
1.33017719e-01 1.46915019e-01 -2.16059640e-01 -1.33942276e-01
1.76963776e-01 7.63411447e-02 -8.45990419e-01 1.08394170e+00
1.14221990e+00 1.90247819e-01 -6.36781037e-01 -3.72006655e-01
-6.97599769e-01 -3.01053703e-01 -1.52185786e+00 9.53955948e-01
-7.89893419e-02 3.99764031e-01 1.88218340e-01 -7.75769472e-01
1.03122568e+00 4.69425023e-02 2.22342521e-01 -3.91946793e-01
-3.39206867e-02 3.79972100e-01 -4.75479156e-01 -1.10063195e+00
3.90226960e-01 1.11892432e-01 1.52390942e-01 2.25681931e-01
-8.90218079e-01 -4.46753412e-01 1.50020257e-01 1.26837537e-01
1.01068509e+00 -2.41043568e-01 -6.44612312e-01 -2.50823468e-01
5.84135175e-01 1.47104293e-01 3.02094966e-01 5.43962419e-01
-1.25135809e-01 7.02804625e-01 2.24752501e-01 -6.37083173e-01
-6.76235378e-01 -1.14768004e+00 -2.54642248e-01 4.11967218e-01
7.12767482e-01 1.73275217e-01 -6.55593455e-01 -6.91335917e-01
3.64016145e-01 7.50180706e-02 -6.16145730e-01 2.43047878e-01
-7.02742875e-01 -8.75046670e-01 7.78810382e-01 5.03437817e-01
8.27281713e-01 -6.48373783e-01 -5.49771845e-01 3.57575044e-02
-6.11840546e-01 -9.05903637e-01 -5.40503442e-01 -2.69952983e-01
-6.32174611e-01 -1.55547571e+00 -4.59434450e-01 -9.56890702e-01
6.92458630e-01 7.72994101e-01 5.84862530e-01 5.55253506e-01
-6.73788249e-01 1.09796055e-01 -4.05596077e-01 -2.36904368e-01
-5.34294322e-02 3.09463274e-02 -3.82707685e-01 3.77986252e-01
-1.43272549e-01 -3.62856001e-01 -7.46215820e-01 6.09425843e-01
-7.85760164e-01 -1.63862437e-01 6.58863246e-01 6.27915323e-01
6.72594130e-01 6.21149957e-01 3.36974710e-01 -9.52797830e-01
5.92399836e-01 -5.71030498e-01 -4.79073942e-01 8.67203102e-02
-7.67962456e-01 6.16286416e-03 1.16654471e-01 -1.78983107e-01
-1.17416549e+00 2.93384522e-01 -2.86618501e-01 3.36699001e-02
-1.42224178e-01 5.54616988e-01 -9.85624865e-02 -2.90788442e-01
5.89562476e-01 2.14654252e-01 9.80181247e-02 -3.87262464e-01
3.56855005e-01 8.69255006e-01 5.30941248e-01 -6.81707382e-01
1.26901829e+00 9.67173874e-01 1.82461306e-01 -1.38811505e+00
-4.60875660e-01 -6.69776499e-01 -6.05676413e-01 -7.69466043e-01
7.71883070e-01 -5.42288303e-01 -5.92111349e-01 1.03357804e+00
-1.18192255e+00 2.62611210e-02 1.68772385e-01 2.11859450e-01
-5.45541681e-02 9.49017644e-01 -6.35029852e-01 -9.72869456e-01
-1.23417445e-01 -1.17564464e+00 9.51736510e-01 3.79462481e-01
4.94682461e-01 -5.15849590e-01 2.65619308e-01 5.30754507e-01
5.27450681e-01 3.82723927e-01 9.96317327e-01 4.49946284e-01
-9.04962420e-01 -1.41791835e-01 -6.74175084e-01 6.50140271e-02
8.51335824e-02 8.66580725e-01 -9.53572810e-01 1.65156916e-01
-4.70960617e-01 -2.67653428e-02 1.34494746e+00 1.49895266e-01
7.54308522e-01 2.26522669e-01 -7.42466271e-01 1.57117210e-02
1.42559361e+00 -1.19206280e-01 9.40805852e-01 2.61342585e-01
7.73320735e-01 7.76157260e-01 5.90219021e-01 3.07755798e-01
6.92566276e-01 3.82303715e-01 8.64791155e-01 -3.17858219e-01
-2.74225354e-01 -3.58271420e-01 1.28551394e-01 8.35495651e-01
-3.69199663e-01 3.80713940e-02 -1.02144504e+00 8.06478798e-01
-1.64489067e+00 -1.14308214e+00 -1.23250079e+00 1.95209587e+00
4.21597421e-01 2.52158880e-01 2.16038510e-01 5.99863410e-01
1.12303269e+00 3.56472731e-02 -2.72083431e-01 8.16354379e-02
-1.35768026e-01 -3.56465615e-02 6.44585252e-01 5.27878404e-01
-9.61219549e-01 7.73546875e-01 6.07064629e+00 9.32863891e-01
-1.01497984e+00 -1.09214634e-01 1.74604654e-01 8.93440366e-01
-5.19518733e-01 2.68296525e-03 -8.07367027e-01 5.21046340e-01
4.00786586e-02 5.18184662e-01 1.85218588e-01 5.86647928e-01
2.68099040e-01 -2.99480349e-01 -1.79008573e-01 6.79908097e-01
-1.98702261e-01 -1.12459552e+00 -3.99924256e-02 -1.05219865e-02
5.54535031e-01 1.66379154e-01 -6.02384470e-02 -1.59467191e-01
1.52723089e-01 -7.88047910e-01 8.21169078e-01 6.30235136e-01
7.64391363e-01 -6.26870096e-01 6.59657478e-01 3.16260338e-01
-1.73684919e+00 -1.89644635e-01 -4.87193584e-01 -4.95120212e-02
1.21758573e-01 1.10247684e+00 -5.94089866e-01 5.30323327e-01
8.01546156e-01 7.05097079e-01 -5.28892338e-01 1.28468752e+00
-1.63710728e-01 8.30385029e-01 -5.63926041e-01 -1.36798725e-01
1.31336033e-01 -1.34973451e-01 2.59970248e-01 1.26506805e+00
3.62333417e-01 -4.09499928e-02 4.10107136e-01 7.43020415e-01
2.94353753e-01 1.08087733e-01 -6.84917092e-01 4.30437475e-01
8.99630189e-01 1.07146025e+00 -1.02789855e+00 -4.03573401e-02
-5.35541415e-01 3.88289213e-01 1.27164572e-02 1.41133979e-01
-8.32958817e-01 -8.31755340e-01 3.42123657e-01 6.08453870e-01
3.20180058e-01 -3.29283208e-01 -5.21658540e-01 -8.92454803e-01
-3.86339538e-02 -4.71403837e-01 2.76279207e-02 -4.69046175e-01
-1.36766362e+00 4.03128147e-01 -2.94182152e-01 -1.24360442e+00
7.00972736e-01 -4.50033993e-01 -9.60440218e-01 4.68046576e-01
-1.94612145e+00 -1.37299788e+00 -8.62990677e-01 7.77513206e-01
5.81191063e-01 4.55130011e-01 2.70073324e-01 5.95060647e-01
-6.43983483e-01 2.62418568e-01 1.31490976e-01 3.82474542e-01
4.16567922e-01 -8.77199233e-01 1.68723792e-01 9.84131694e-01
-2.51058310e-01 8.19848627e-02 4.18736726e-01 -8.45597863e-01
-1.23595536e+00 -1.15865862e+00 6.92504048e-01 -2.39566281e-01
6.87826157e-01 -5.10837547e-02 -9.18033361e-01 2.52519120e-02
-1.22414380e-01 1.90188162e-04 1.23134730e-02 -1.98744237e-01
-4.83253360e-01 -3.20079565e-01 -1.08217669e+00 3.01543832e-01
8.88599336e-01 -4.42836821e-01 -3.22768718e-01 7.56337047e-02
2.57930547e-01 -2.28245363e-01 -5.25290549e-01 4.97365654e-01
9.01158810e-01 -1.18281388e+00 9.05830145e-01 3.08469683e-01
3.25841963e-01 -6.14018917e-01 6.71750233e-02 -8.79749894e-01
-3.41480792e-01 -1.38408631e-01 4.31699008e-01 1.08584905e+00
4.69907761e-01 -8.79442871e-01 1.02939749e+00 -2.49349903e-02
-3.41053098e-01 -7.70564616e-01 -9.94318128e-01 -4.80724186e-01
-9.49004218e-02 -4.86732900e-01 7.38233328e-01 7.98094094e-01
-2.06812978e-01 2.13952065e-01 -1.24784693e-01 6.72506988e-01
8.47274840e-01 3.31043482e-01 6.50261879e-01 -1.82436883e+00
3.13719600e-01 -5.53435206e-01 -2.12812692e-01 -1.13527894e+00
-3.80305529e-01 -3.20780665e-01 5.53248763e-01 -1.60596418e+00
1.75996915e-01 -1.12297404e+00 8.79875273e-02 2.87503153e-01
-2.22338527e-01 1.95402920e-01 -3.32424849e-01 3.99746537e-01
-2.48370603e-01 3.67390573e-01 1.43647802e+00 -4.82804894e-01
7.99908713e-02 4.86678511e-01 -2.77322054e-01 7.36381114e-01
7.25999296e-01 -4.69633818e-01 -2.43447468e-01 -3.20782930e-01
2.44832844e-01 1.72766112e-02 3.37348044e-01 -1.32707202e+00
2.10199654e-01 -2.39998519e-01 -5.66666126e-02 -1.09259403e+00
4.21911851e-02 -9.44389224e-01 -6.43592924e-02 7.75232375e-01
2.48643517e-01 4.08259444e-02 -1.82233512e-01 9.49130297e-01
-1.27359733e-01 -3.64120722e-01 9.08601344e-01 2.46755369e-02
-5.10562122e-01 1.38203397e-01 -8.92835975e-01 3.93435620e-02
1.03620994e+00 -5.16114116e-01 -6.37529254e-01 -1.12019755e-01
-1.83475852e-01 2.96171039e-01 5.04755020e-01 1.07748963e-01
9.67167199e-01 -1.08469677e+00 -7.92576075e-01 4.29343671e-01
1.55895606e-01 2.61460125e-01 2.15584978e-01 5.88310838e-01
-1.00207424e+00 -1.69119909e-02 -4.32152674e-02 -6.71024203e-01
-1.21960223e+00 2.61212617e-01 1.62278458e-01 -1.46335453e-01
-5.38951755e-01 7.64161289e-01 -9.65881273e-02 -3.68652403e-01
3.76820490e-02 -8.13404620e-01 -3.78669143e-01 3.46549451e-02
1.64919600e-01 1.01167679e+00 9.43339840e-02 -6.94424033e-01
-3.05307150e-01 1.41959167e+00 1.90048143e-01 2.50014484e-01
1.02727878e+00 -2.46974543e-01 -2.95670688e-01 2.53984869e-01
5.71683705e-01 4.51188207e-01 -9.56255198e-01 -9.07789394e-02
-1.14288673e-01 -4.99897897e-01 -1.07041923e-02 -4.20274645e-01
-1.38383877e+00 1.04671574e+00 7.12662637e-01 4.91033792e-01
8.59143138e-01 -9.55325291e-02 1.25129533e+00 2.01613918e-01
5.41532993e-01 -1.16403401e+00 2.49218792e-01 4.11375403e-01
5.37633002e-01 -9.80821490e-01 -7.84214735e-02 -1.03106904e+00
-3.32153171e-01 1.32686853e+00 5.00038087e-01 -1.14737310e-01
1.00810587e+00 3.48266482e-01 1.14181995e-01 -5.15048802e-01
-9.05126741e-04 -4.62923974e-01 -2.08853573e-01 6.93906903e-01
-3.05990338e-01 -4.49809767e-02 -1.26069590e-01 2.18722612e-01
4.55822885e-01 -9.16557014e-02 6.45635009e-01 9.98780847e-01
-1.23610330e+00 -1.01490068e+00 -7.97697127e-01 5.11805952e-01
1.28564075e-01 9.30716544e-02 -1.06422052e-01 7.08933532e-01
4.70407039e-01 1.21435571e+00 -3.68610412e-01 -6.91266596e-01
5.25325775e-01 -3.00387293e-01 9.64645147e-02 -3.25885862e-01
5.82818054e-02 -2.79099286e-01 -8.43147188e-02 -3.34168792e-01
-2.98235953e-01 -6.67824030e-01 -1.45565748e+00 -9.99079496e-02
-8.14508796e-01 2.56651193e-01 6.50434256e-01 8.95976663e-01
2.64948428e-01 9.43208933e-02 1.00718594e+00 -7.98816085e-01
-3.70439678e-01 -7.13160157e-01 -7.38720655e-01 5.33269569e-02
3.54453892e-01 -1.09350431e+00 -4.30588186e-01 7.76740238e-02] | [7.501223564147949, 1.3390380144119263] |
be8626a4-e288-4f30-8027-84aff949aef0 | siamese-contrastive-embedding-network-for-1 | 2206.14475 | null | https://arxiv.org/abs/2206.14475v1 | https://arxiv.org/pdf/2206.14475v1.pdf | Siamese Contrastive Embedding Network for Compositional Zero-Shot Learning | Compositional Zero-Shot Learning (CZSL) aims to recognize unseen compositions formed from seen state and object during training. Since the same state may be various in the visual appearance while entangled with different objects, CZSL is still a challenging task. Some methods recognize state and object with two trained classifiers, ignoring the impact of the interaction between object and state; the other methods try to learn the joint representation of the state-object compositions, leading to the domain gap between seen and unseen composition sets. In this paper, we propose a novel Siamese Contrastive Embedding Network (SCEN) (Code: https://github.com/XDUxyLi/SCEN-master) for unseen composition recognition. Considering the entanglement between state and object, we embed the visual feature into a Siamese Contrastive Space to capture prototypes of them separately, alleviating the interaction between state and object. In addition, we design a State Transition Module (STM) to increase the diversity of training compositions, improving the robustness of the recognition model. Extensive experiments indicate that our method significantly outperforms the state-of-the-art approaches on three challenging benchmark datasets, including the recent proposed C-QGA dataset. | ['Muli Yang', 'Cheng Deng', 'Kun Wei', 'Xu Yang', 'Xiangyu Li'] | 2022-06-29 | siamese-contrastive-embedding-network-for | http://openaccess.thecvf.com//content/CVPR2022/html/Li_Siamese_Contrastive_Embedding_Network_for_Compositional_Zero-Shot_Learning_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Li_Siamese_Contrastive_Embedding_Network_for_Compositional_Zero-Shot_Learning_CVPR_2022_paper.pdf | cvpr-2022-1 | ['compositional-zero-shot-learning'] | ['computer-vision'] | [ 1.23374298e-01 -4.18148428e-01 -1.68498382e-01 9.31394622e-02
-3.94961566e-01 -5.90742826e-01 8.27175677e-01 -2.53256977e-01
-5.07002473e-02 3.13784331e-01 2.48367414e-01 1.29303992e-01
7.02084675e-02 -5.15256643e-01 -7.54923820e-01 -1.23588502e+00
3.14361751e-01 5.01397789e-01 2.14367554e-01 -6.32072613e-02
-3.42218764e-02 2.31206313e-01 -1.64334321e+00 3.25689316e-01
6.80981815e-01 1.07016528e+00 1.17962167e-01 4.25156265e-01
-2.39811555e-01 1.02560198e+00 -4.75784361e-01 -3.88275325e-01
3.90971243e-01 -7.16258347e-01 -3.69339913e-01 2.53629923e-01
6.83063626e-01 -1.29830718e-01 -6.98097527e-01 1.38003838e+00
3.74316841e-01 2.26040661e-01 7.19791353e-01 -1.57409120e+00
-1.19841027e+00 4.40575063e-01 -1.37567371e-01 9.94651020e-02
2.39881679e-01 6.48485065e-01 1.21319270e+00 -8.14180613e-01
8.13584507e-01 1.10531688e+00 -4.99134436e-02 8.86210620e-01
-1.30296457e+00 -8.78969133e-01 2.58589149e-01 8.24590206e-01
-1.37913001e+00 -5.55114388e-01 1.20020080e+00 -5.58138490e-01
6.24681890e-01 3.17491621e-01 7.94508398e-01 1.52267230e+00
-8.37057605e-02 1.04437089e+00 9.82922435e-01 -1.59717336e-01
4.44943368e-01 1.67466551e-01 2.13185534e-01 1.00553346e+00
2.06958711e-01 3.84113491e-01 -7.46252060e-01 -1.41705483e-01
2.48532042e-01 2.22761124e-01 -5.22340000e-01 -1.13357711e+00
-1.31539285e+00 4.13316429e-01 6.42864227e-01 1.58002332e-01
9.53048542e-02 4.28428948e-02 1.37279361e-01 2.92924285e-01
-4.37912829e-02 4.29148644e-01 2.53458414e-02 2.00517830e-02
-5.02743900e-01 -5.31419776e-02 8.38006318e-01 8.30512285e-01
8.53988230e-01 2.61766136e-01 -3.00515503e-01 3.91777635e-01
3.16916138e-01 3.46786082e-01 6.63762331e-01 -4.25972402e-01
3.98167312e-01 9.63655412e-01 -1.61135644e-01 -4.79905605e-01
1.85874447e-01 -4.63027894e-01 -8.87347519e-01 2.28044346e-01
1.78819686e-01 2.39140257e-01 -1.09963000e+00 1.69867563e+00
3.86689812e-01 6.35054290e-01 2.05600634e-01 1.22361708e+00
9.28626120e-01 8.09765995e-01 -6.49941117e-02 1.07391573e-01
1.20031536e+00 -1.26704109e+00 -6.27050042e-01 -1.09174982e-01
1.97730392e-01 -4.30211246e-01 9.09174204e-01 2.05370739e-01
-7.39155412e-01 -5.53306460e-01 -1.41673684e+00 -2.07609646e-02
-4.59810436e-01 2.05659643e-01 3.89535218e-01 3.08918417e-01
-3.77331555e-01 5.80584109e-01 -9.32185829e-01 3.85272987e-02
4.06731874e-01 2.83016384e-01 -4.74879444e-01 -2.04357892e-01
-1.01747608e+00 7.29888916e-01 5.94318986e-01 1.13845624e-01
-1.41140902e+00 -6.30867481e-01 -9.52279747e-01 2.97653884e-01
5.93171895e-01 -5.26959479e-01 7.49036789e-01 -1.04969645e+00
-1.68593299e+00 4.29459333e-01 -3.96597497e-02 1.02264486e-01
3.66789639e-01 3.28124911e-01 -5.17832518e-01 1.05249099e-01
-3.92896026e-01 2.74901390e-01 1.06484663e+00 -1.35365510e+00
-1.77441299e-01 -4.05141681e-01 -1.40140980e-01 2.20149800e-01
-3.03834975e-01 -2.37106293e-01 -2.97103584e-01 -2.82329172e-01
3.19567919e-01 -8.87355745e-01 2.57940531e-01 2.69934833e-01
-4.58490044e-01 -3.28580827e-01 1.01027286e+00 -2.74689168e-01
8.44524860e-01 -2.54651451e+00 7.11044729e-01 -9.21131521e-02
3.44746739e-01 3.78839225e-01 -4.73468900e-01 4.58384812e-01
-2.16543645e-01 -1.80176809e-01 -2.66710192e-01 -4.32313293e-01
3.99638087e-01 4.47506666e-01 -4.07870799e-01 8.02828908e-01
3.73375088e-01 1.06299412e+00 -9.73181129e-01 -3.59141260e-01
2.28623480e-01 3.19705218e-01 -2.69043744e-01 4.33529139e-01
-2.85636753e-01 5.08385658e-01 -7.34888241e-02 7.62808681e-01
5.84393442e-01 -3.93338829e-01 3.01686555e-01 -4.61299539e-01
1.11466542e-01 1.19086221e-01 -1.32817435e+00 1.74294281e+00
-2.41150297e-02 2.51172572e-01 -1.55796498e-01 -9.27811861e-01
6.15826547e-01 3.02972257e-01 2.38105327e-01 -5.90047300e-01
2.52965301e-01 1.00854427e-01 4.28211600e-01 -4.96730566e-01
6.00526258e-02 -1.85658783e-01 -8.08650106e-02 4.15677249e-01
7.78023839e-01 2.68520135e-02 1.01292841e-01 2.19768032e-01
9.15227056e-01 7.91123360e-02 2.14372784e-01 -5.87041900e-02
4.73797858e-01 -4.04117405e-01 7.99624801e-01 4.77684945e-01
-5.68656683e-01 4.34678227e-01 6.31648600e-01 -3.49328905e-01
-1.05272269e+00 -1.48013604e+00 9.35010612e-02 7.79494166e-01
8.21811557e-01 -3.89224231e-01 -3.19776684e-01 -8.64597857e-01
1.78786740e-02 4.81331497e-01 -6.47329926e-01 -6.32355332e-01
-3.76408577e-01 -4.32768136e-01 2.10289314e-01 2.55421519e-01
4.89632159e-01 -9.03410912e-01 -2.00894132e-01 -9.21826810e-02
1.03576988e-01 -1.01846385e+00 -6.99300885e-01 1.58888489e-01
-3.56850058e-01 -1.00848651e+00 -4.06091541e-01 -8.26603234e-01
7.73452282e-01 2.41468847e-02 5.80311656e-01 -1.48290917e-01
-3.79739493e-01 1.02125384e-01 -3.15161109e-01 1.96642876e-01
-4.60412353e-01 -1.19374909e-01 3.44828479e-02 5.87923646e-01
1.31711379e-01 -5.20268381e-01 -5.49656570e-01 2.05924675e-01
-1.05634391e+00 2.48943537e-01 5.59317291e-01 1.07984197e+00
3.17409933e-01 -1.61285549e-01 -4.64563631e-02 -4.56826359e-01
1.60047308e-01 -5.57930470e-01 -5.63673913e-01 6.54151857e-01
-4.19978470e-01 5.72256386e-01 8.65902126e-01 -9.83813643e-01
-9.37905073e-01 2.09406614e-01 3.24010670e-01 -1.12049842e+00
-6.38830438e-02 2.71950453e-03 -5.37160575e-01 -7.75005072e-02
3.95787627e-01 6.43569827e-01 2.31621742e-01 -4.99660313e-01
3.85990113e-01 4.46601450e-01 5.01870930e-01 -6.16652966e-01
9.64297771e-01 4.80056942e-01 9.65714920e-03 -5.11340797e-01
-7.40939021e-01 -4.63742346e-01 -3.76578480e-01 -2.72396058e-02
7.51806200e-01 -7.79372513e-01 -8.22183132e-01 5.27408302e-01
-9.23003852e-01 -2.37027571e-01 -4.91743177e-01 3.22855324e-01
-2.52105087e-01 5.47543705e-01 -6.82308435e-01 -5.89768529e-01
-9.56261829e-02 -1.40018833e+00 8.52345347e-01 2.57878721e-01
1.66093022e-01 -6.73768342e-01 2.67899781e-01 3.60435218e-01
1.41007766e-01 1.11174189e-01 1.03656113e+00 -4.83655840e-01
-1.10883093e+00 -1.68732647e-02 -1.02659889e-01 4.17137891e-01
4.61905487e-02 1.63570076e-01 -8.04415286e-01 -4.67911959e-01
-1.42464088e-02 -3.99652213e-01 1.12206829e+00 -4.95250106e-01
9.03472841e-01 -4.47483301e-01 -8.91085118e-02 6.17939234e-01
1.41588557e+00 2.06598476e-01 4.82264608e-01 -1.71778917e-01
9.70109224e-01 3.12265515e-01 -1.83835998e-03 2.03321084e-01
2.57135063e-01 6.82942033e-01 4.74974662e-01 4.04935956e-01
-4.59016234e-01 -5.83352089e-01 7.53360450e-01 1.17530203e+00
2.31985196e-01 -3.40315908e-01 -5.88499546e-01 4.09891874e-01
-1.94965398e+00 -1.03996694e+00 3.75771284e-01 1.85778224e+00
5.27165174e-01 -2.07635328e-01 -2.40428656e-01 -4.20880206e-02
5.94864786e-01 4.72658247e-01 -8.71750951e-01 -1.01508317e-03
-2.32842609e-01 3.04912120e-01 9.19487998e-02 2.67858624e-01
-8.28666389e-01 9.03429747e-01 4.81482887e+00 9.38550830e-01
-1.39037180e+00 3.93997908e-01 4.56689894e-02 -3.40555668e-01
-3.83290887e-01 2.83735245e-01 -6.07458293e-01 7.68089652e-01
5.14946461e-01 9.06886440e-03 6.85515344e-01 6.58929408e-01
-5.23838580e-01 1.15736358e-01 -1.24317849e+00 9.02054429e-01
4.40851301e-01 -1.24969292e+00 2.33249113e-01 -5.35252877e-02
9.81570780e-01 -1.13698892e-01 3.58895361e-01 4.91417944e-01
1.25307649e-01 -5.53963065e-01 9.74200010e-01 6.02199435e-01
6.25070810e-01 -2.80372113e-01 4.06673193e-01 3.34822297e-01
-1.18027639e+00 -3.68787855e-01 -1.81016937e-01 1.58646315e-01
-1.75706357e-01 1.26515940e-01 -4.14295793e-01 5.99482656e-01
2.88291335e-01 7.91006684e-01 -6.32960796e-01 9.16121125e-01
-4.44461852e-01 3.94372344e-01 3.47812325e-02 -2.81745017e-01
1.49929136e-01 -4.68396127e-01 8.79298389e-01 5.20208716e-01
1.57453045e-01 -1.13136292e-01 2.11377725e-01 1.44386053e+00
-1.95591614e-01 -3.27570915e-01 -3.12035382e-01 -4.90354449e-01
1.45484000e-01 1.11144567e+00 -2.61391878e-01 -5.82422495e-01
-2.59773880e-01 1.27518165e+00 6.38272643e-01 5.68207204e-01
-8.93637419e-01 -1.59413368e-01 9.66984987e-01 -3.66034687e-01
6.28933549e-01 -2.72182345e-01 2.31396079e-01 -1.89405060e+00
1.65557057e-01 -9.69952941e-01 2.30798230e-01 -5.26797652e-01
-1.39514065e+00 3.92974108e-01 -2.56828129e-01 -1.43020046e+00
3.50005120e-01 -6.82329059e-01 -6.80160463e-01 5.61645269e-01
-1.34485841e+00 -1.11555505e+00 -3.51266295e-01 4.76271063e-01
3.18169147e-01 -3.46323609e-01 7.26459444e-01 3.47847968e-01
-1.02834892e+00 5.60203850e-01 2.49072328e-01 1.53832883e-01
5.45002580e-01 -1.08194435e+00 1.38607845e-01 8.31857264e-01
6.22896135e-01 5.90637505e-01 4.08352286e-01 -5.14461696e-01
-1.94048977e+00 -8.94275367e-01 3.76347750e-01 -3.72726619e-01
7.90527225e-01 -8.98251891e-01 -8.88198614e-01 3.98558825e-01
2.01206073e-01 5.06346107e-01 6.01555645e-01 -2.10702643e-01
-8.60302866e-01 -3.05995047e-01 -7.78160155e-01 8.26082468e-01
1.15399897e+00 -9.63000596e-01 -5.43385625e-01 3.51846144e-02
6.52834833e-01 -2.48673081e-01 -5.33508062e-01 2.02140346e-01
6.61548376e-01 -8.19164455e-01 6.64771795e-01 -7.15407133e-01
3.57923180e-01 -6.82720423e-01 -1.61947563e-01 -1.42737341e+00
-4.28763449e-01 -4.38284427e-01 -7.68567502e-01 9.14849520e-01
1.19520679e-01 -6.78397059e-01 6.74301207e-01 4.72595930e-01
-8.74030292e-02 -6.98842645e-01 -1.05037439e+00 -1.09634578e+00
-2.42331371e-01 1.52029231e-01 7.77601898e-01 8.45433474e-01
2.62080245e-02 4.20582175e-01 -3.47820729e-01 2.59932190e-01
7.54906952e-01 5.66616297e-01 5.26058793e-01 -8.39856982e-01
-5.47339082e-01 -4.42593575e-01 -8.38793099e-01 -8.06812227e-01
5.40539086e-01 -1.17203486e+00 -2.20477600e-02 -9.81169999e-01
5.51916897e-01 -1.85811862e-01 -6.90837383e-01 3.90721560e-01
-1.88069627e-01 9.23286602e-02 5.08629203e-01 3.03886116e-01
-8.05882633e-01 1.16093493e+00 1.50495362e+00 -7.09640145e-01
2.01825291e-01 -5.35545528e-01 -3.54154378e-01 7.73703605e-02
4.39679831e-01 -5.04017413e-01 -3.25672925e-01 -2.91219383e-01
-1.53003052e-01 -1.42238259e-01 6.42906606e-01 -1.10341322e+00
4.15594041e-01 -1.39242277e-01 1.70644715e-01 -1.78443879e-01
7.59645700e-01 -8.12719822e-01 5.01499474e-01 6.08490765e-01
-2.75621057e-01 -4.86604899e-01 -1.73080131e-01 7.80378759e-01
-2.18863666e-01 -3.44484359e-01 8.32248390e-01 -7.03077242e-02
-7.35498667e-01 6.83195949e-01 2.31252052e-02 -1.39923289e-01
1.22695017e+00 1.80332421e-03 -7.07526147e-01 3.40196192e-02
-6.27360940e-01 1.28640413e-01 6.56860113e-01 4.87987578e-01
5.67688167e-01 -1.49471581e+00 -2.33963758e-01 6.15728438e-01
5.05296588e-01 -2.32854754e-01 6.14235878e-01 6.50250852e-01
-2.09467337e-01 1.01412103e-01 -4.00656998e-01 -4.67284530e-01
-1.10631931e+00 8.12638640e-01 4.23904359e-01 -3.47546451e-02
-4.55748260e-01 8.51313174e-01 4.01023030e-01 -4.18432504e-01
1.92156583e-01 -2.63524264e-01 2.03111976e-01 -7.53850862e-03
3.60705733e-01 2.98887670e-01 -3.49860549e-01 -7.17687607e-01
-2.34872654e-01 4.10751790e-01 -1.95319295e-01 2.46000327e-02
9.41260219e-01 3.19923073e-01 -1.50209889e-01 7.02744722e-01
1.66138244e+00 -3.08834165e-01 -1.48592579e+00 -4.66988504e-01
-2.10009858e-01 -4.60677356e-01 -1.16805814e-01 -5.64360797e-01
-1.02357554e+00 1.01257527e+00 8.30846786e-01 -5.24695814e-02
7.46331990e-01 2.77818948e-01 7.47757256e-01 2.84721464e-01
1.15431190e-01 -7.80794978e-01 4.85786557e-01 2.21922308e-01
7.52756774e-01 -1.33633661e+00 -9.94322002e-02 -7.96851814e-02
-7.03491330e-01 9.18840945e-01 8.08247566e-01 -2.18688026e-01
4.65528458e-01 -3.31125148e-02 -1.93181112e-01 -2.80185968e-01
-9.71113622e-01 -3.51495862e-01 5.31153083e-01 1.28611982e-01
-3.43168795e-01 2.87294477e-01 1.83062881e-01 4.34216052e-01
1.32793739e-01 -4.09671456e-01 2.55538076e-01 8.88633609e-01
8.52520298e-03 -1.24636257e+00 3.71052064e-02 1.42458424e-01
2.01247931e-01 1.54662207e-01 -4.96135056e-01 3.44496757e-01
4.45529431e-01 5.00668287e-01 2.36365706e-01 -5.27075291e-01
2.99646109e-02 3.05155039e-01 7.55105674e-01 -6.63662016e-01
-3.94419521e-01 -1.92143962e-01 -3.42011422e-01 -5.25652885e-01
-2.30026096e-01 -5.85267603e-01 -9.13911283e-01 -7.79002458e-02
-3.84157896e-01 2.12992877e-02 5.63438535e-01 7.23145664e-01
4.81323957e-01 4.47546810e-01 6.84780002e-01 -6.18175626e-01
-9.20514345e-01 -7.32867181e-01 -6.87318683e-01 7.50928521e-01
6.42081738e-01 -8.99823010e-01 -5.41263998e-01 -1.49637312e-01] | [10.217557907104492, 2.302574872970581] |
8f3d6532-0894-4600-a0b1-a76566245e78 | formulation-graphs-for-mapping-structure | 2307.03811 | null | https://arxiv.org/abs/2307.03811v1 | https://arxiv.org/pdf/2307.03811v1.pdf | Formulation Graphs for Mapping Structure-Composition of Battery Electrolytes to Device Performance | Advanced computational methods are being actively sought for addressing the challenges associated with discovery and development of new combinatorial material such as formulations. A widely adopted approach involves domain informed high-throughput screening of individual components that can be combined into a formulation. This manages to accelerate the discovery of new compounds for a target application but still leave the process of identifying the right 'formulation' from the shortlisted chemical space largely a laboratory experiment-driven process. We report a deep learning model, Formulation Graph Convolution Network (F-GCN), that can map structure-composition relationship of the individual components to the property of liquid formulation as whole. Multiple GCNs are assembled in parallel that featurize formulation constituents domain-intuitively on the fly. The resulting molecular descriptors are scaled based on respective constituent's molar percentage in the formulation, followed by formalizing into a combined descriptor that represents a complete formulation to an external learning architecture. The use case of proposed formulation learning model is demonstrated for battery electrolytes by training and testing it on two exemplary datasets representing electrolyte formulations vs battery performance -- one dataset is sourced from literature about Li/Cu half-cells, while the other is obtained by lab-experiments related to lithium-iodide full-cell chemistry. The model is shown to predict the performance metrics like Coulombic Efficiency (CE) and specific capacity of new electrolyte formulations with lowest reported errors. The best performing F-GCN model uses molecular descriptors derived from molecular graphs that are informed with HOMO-LUMO and electric moment properties of the molecules using a knowledge transfer technique. | ['Young-Hye La', 'Daniele Congiu', 'Linda Sundberg', 'Khanh Nugyuen', 'Andy Tek', 'Dmitry Zubarev', 'Maxwell Giammona', 'Vidushi Sharma'] | 2023-07-07 | null | null | null | null | ['transfer-learning'] | ['miscellaneous'] | [ 3.73723537e-01 -3.37443024e-01 -3.51178348e-01 -3.73751372e-01
-6.56875014e-01 -7.85496414e-01 5.15089631e-01 9.85814810e-01
-3.83556187e-01 1.32670355e+00 -1.36512116e-01 -4.16596383e-01
-5.05325258e-01 -9.77818489e-01 -9.80277479e-01 -1.19330561e+00
-1.35237500e-01 7.91083872e-01 -3.29448611e-01 -3.32063973e-01
5.66194475e-01 8.30324411e-01 -1.37313008e+00 3.64940763e-01
1.09303391e+00 1.35181856e+00 2.46059328e-01 4.21489090e-01
-3.52602541e-01 2.77158469e-01 -2.79489726e-01 -5.71545064e-02
6.38533980e-02 -1.89422503e-01 -3.73566896e-01 -2.88963526e-01
2.36138195e-01 2.25932032e-01 4.87812189e-03 8.36438239e-01
6.97934568e-01 -1.01718707e-02 1.14525199e+00 -1.00940132e+00
-5.87661088e-01 4.48612869e-01 2.22156018e-01 -1.31995350e-01
4.07521993e-01 2.20105991e-01 8.49711835e-01 -8.63328755e-01
7.61330605e-01 9.39170063e-01 7.00826645e-01 1.77566290e-01
-1.30881000e+00 -8.21816802e-01 -1.57745883e-01 1.51773393e-01
-1.41611278e+00 -1.93999425e-01 5.84983289e-01 -4.83816803e-01
1.44540501e+00 3.10040414e-01 6.55358672e-01 5.14695048e-01
4.72365201e-01 3.58351022e-01 1.10076940e+00 -2.18546122e-01
6.32802963e-01 3.32788229e-01 3.16640139e-01 4.16511804e-01
7.01536775e-01 2.93417126e-02 -6.77022934e-01 -4.89931218e-02
5.35599776e-02 -1.78249940e-01 -2.31950760e-01 -7.59970427e-01
-5.81666648e-01 7.93641031e-01 6.97875857e-01 3.53070110e-01
-5.80070794e-01 7.94212595e-02 4.12110656e-01 8.42650533e-02
1.44308209e-01 6.65451884e-01 -4.87345695e-01 2.10565835e-01
-7.77978837e-01 7.11010814e-01 1.19582999e+00 8.69961858e-01
1.06483829e+00 1.67920947e-01 -1.65662199e-01 1.85947612e-01
3.99361461e-01 4.72675174e-01 2.01273307e-01 -1.52005687e-01
3.74433935e-01 9.75785613e-01 3.14579785e-01 -3.87794793e-01
-5.33472300e-01 -4.70221192e-01 -4.73298520e-01 9.36724991e-02
8.35997537e-02 -1.68050706e-01 -1.00467575e+00 1.35856330e+00
3.18899125e-01 -2.20010191e-01 3.52286786e-01 5.64326584e-01
1.08011556e+00 6.74071968e-01 5.67250013e-01 -2.08007038e-01
9.44922686e-01 -4.99880552e-01 -5.39265037e-01 3.29799026e-01
5.71221411e-01 -1.83827311e-01 3.65165442e-01 4.76517320e-01
-9.58439171e-01 -4.20287877e-01 -1.54461300e+00 4.04972956e-02
-1.30975473e+00 1.14229538e-01 6.71163619e-01 8.06237817e-01
-8.82486820e-01 1.13484371e+00 -4.78160441e-01 -1.96442455e-01
5.25383532e-01 1.35884178e+00 -4.47017342e-01 -4.08800261e-04
-1.15293264e+00 1.06255448e+00 7.69658267e-01 2.98183322e-01
-1.20528769e+00 -9.67185616e-01 -6.67515576e-01 9.91288349e-02
2.86326613e-02 -4.80079830e-01 6.30237401e-01 -6.51517510e-01
-1.36759555e+00 4.22501832e-01 2.49256263e-03 -5.16120791e-01
1.46028534e-01 3.61922793e-02 -5.80210268e-01 -1.44001648e-01
-3.37988399e-02 6.12320960e-01 3.94001722e-01 -1.47256064e+00
-3.42030078e-01 -3.47187996e-01 -6.14203811e-02 2.28204742e-01
-3.49190533e-01 -5.47238767e-01 6.70855865e-02 1.20952010e-01
-3.28847408e-01 -6.11263812e-01 -4.38412353e-02 -2.93661594e-01
-3.79262030e-01 -2.60426491e-01 5.39160788e-01 -4.98100847e-01
1.04649603e+00 -1.65314150e+00 4.07921433e-01 6.76938176e-01
2.29387134e-01 5.08626997e-01 -6.57105148e-02 1.00104630e+00
-4.26471651e-01 7.94108436e-02 -2.27635637e-01 1.07426375e-01
2.00873777e-01 -3.20133157e-02 8.66986513e-02 5.73862016e-01
4.85289246e-01 1.13898838e+00 -8.91739249e-01 7.18008056e-02
2.70405829e-01 7.33526289e-01 -2.48257458e-01 1.18696563e-01
-6.42172813e-01 3.54447871e-01 -5.53262889e-01 7.57219076e-01
1.01849818e+00 -3.65651101e-02 3.70179296e-01 -6.04530871e-01
-3.57518584e-01 2.79984266e-01 -1.05644488e+00 1.45319474e+00
-1.49534121e-01 1.67365059e-01 -2.96316445e-01 -1.04027033e+00
1.12350023e+00 1.67626180e-02 4.08525556e-01 -1.01106822e+00
3.03984195e-01 6.11594796e-01 1.40620098e-01 -4.44021493e-01
3.09964627e-01 -3.79846334e-01 1.59491926e-01 -9.10258666e-02
3.14088076e-01 3.18052694e-02 5.04317403e-01 -8.09208080e-02
6.75640643e-01 3.92485410e-01 1.81306645e-01 -6.94325984e-01
8.26067746e-01 1.42114952e-01 -2.07632687e-02 4.22743917e-01
3.15755397e-01 -3.54105458e-02 4.01990741e-01 -3.59559596e-01
-1.10401702e+00 -9.99441922e-01 -4.00978297e-01 7.17496634e-01
2.23428845e-01 -3.66494000e-01 -6.60707653e-01 -1.07984647e-01
4.34062421e-01 6.52958751e-01 -7.01902688e-01 -2.45722473e-01
-9.82616097e-02 -8.08179379e-01 3.68991882e-01 5.37764847e-01
2.05783546e-01 -9.68022466e-01 -2.73656815e-01 5.09129047e-01
7.17806816e-01 -6.44519866e-01 2.00836957e-01 1.05092394e+00
-7.59722829e-01 -1.05858421e+00 -4.92476493e-01 -5.12333274e-01
4.50023264e-01 -5.30739605e-01 8.46295118e-01 -3.83555633e-03
-5.09452581e-01 1.15956508e-01 4.31486927e-02 -7.65921235e-01
-4.21840191e-01 6.21666349e-02 1.02237411e-01 6.02126122e-02
4.63915348e-01 -5.53196013e-01 -8.57375443e-01 -2.59141833e-01
-8.64903629e-01 -3.51520628e-01 5.83255470e-01 6.03229582e-01
9.36088502e-01 6.38588816e-02 7.24562824e-01 -7.35305548e-01
8.30208480e-01 -6.22278988e-01 -5.94831705e-01 5.57942748e-01
-7.85119772e-01 3.74085844e-01 6.44759953e-01 -3.98176640e-01
-5.96327722e-01 1.23397537e-01 -2.24356521e-02 -1.27993643e-01
-1.26971211e-02 7.18838453e-01 -5.95332205e-01 -4.99559313e-01
4.97331917e-01 3.20460886e-01 -1.25502780e-01 -2.97480464e-01
1.27900898e-01 3.96227866e-01 4.60873306e-01 -8.94554675e-01
4.53438699e-01 4.62855324e-02 5.12209594e-01 -7.34184325e-01
-1.12014942e-01 -4.41351444e-01 -5.65477192e-01 -7.93424398e-02
8.00213516e-01 -9.80782509e-01 -1.27695453e+00 1.41902417e-01
-8.80866587e-01 -2.68639386e-01 4.71861596e-04 2.21630543e-01
-1.80890843e-01 -1.06946364e-01 -1.75086170e-01 -7.57951438e-01
-7.99325049e-01 -1.38710558e+00 9.73725438e-01 4.74963158e-01
1.60713252e-02 -1.25181687e+00 -1.62372105e-02 9.33604315e-03
5.48896074e-01 5.48106313e-01 1.33791280e+00 -1.05455136e+00
-7.34404147e-01 -3.82378787e-01 -2.10933656e-01 2.84010917e-01
1.68927610e-01 -2.96055257e-01 -1.07561636e+00 -4.42347229e-01
-4.16272670e-01 -2.18557626e-01 7.90822625e-01 3.04066569e-01
9.47145402e-01 2.54787594e-01 -5.96005201e-01 4.57643002e-01
2.03650951e+00 6.27738953e-01 7.93028235e-01 2.78828472e-01
8.47314596e-01 3.64709347e-02 2.73477584e-01 3.82212341e-01
1.08804792e-01 2.71687835e-01 4.57255602e-01 -5.20570725e-02
2.94201057e-02 -2.45440915e-01 2.37048462e-01 2.30477959e-01
-1.62265256e-01 -6.92827523e-01 -7.12533116e-01 1.72635153e-01
-1.41769278e+00 -3.95648867e-01 -2.79662311e-01 2.28823280e+00
6.08157635e-01 1.47504076e-01 1.42488614e-01 6.91274479e-02
3.41783762e-01 -5.91612041e-01 -9.76160109e-01 -7.55999982e-01
-2.16013223e-01 8.48010540e-01 8.55423927e-01 4.10361290e-01
-7.76177704e-01 6.70049787e-01 5.92785358e+00 8.54268312e-01
-1.46809125e+00 -3.73687536e-01 5.00940979e-01 1.27538338e-01
-5.34487605e-01 8.69548693e-02 -1.16440845e+00 2.25111693e-01
1.23427033e+00 -2.91749597e-01 4.84475464e-01 4.03954446e-01
1.55711994e-01 -3.26491177e-01 -1.58804083e+00 9.06541944e-01
-1.50420383e-01 -1.81936812e+00 4.03516859e-01 2.01894134e-01
6.22752964e-01 -1.84839875e-01 1.02009006e-01 1.72741711e-01
-3.42093199e-01 -1.32396853e+00 6.54770911e-01 8.24137211e-01
9.75490987e-01 -8.16425741e-01 5.81286252e-01 -1.32411286e-01
-1.31234014e+00 -2.79177099e-01 -2.08066389e-01 5.44536635e-02
-2.80356497e-01 8.56132135e-02 -1.03078508e+00 8.57958615e-01
2.24303126e-01 6.78785741e-01 -6.56781137e-01 1.19608223e+00
4.92626339e-01 2.17607930e-01 -2.76590377e-01 -5.97137094e-01
4.52543646e-01 -6.06627762e-01 2.97243185e-02 1.26028109e+00
3.61097574e-01 -6.54770508e-02 -9.34186578e-02 1.30546725e+00
-2.99736142e-01 3.79531413e-01 -6.00014389e-01 -5.25168777e-01
3.16962391e-01 1.21476340e+00 -7.05649972e-01 -3.64814758e-01
3.63719016e-02 4.56465662e-01 9.23283026e-02 4.47205901e-01
-5.54574013e-01 -2.91505367e-01 3.80538404e-01 2.19346359e-01
5.61477542e-01 9.98587087e-02 -1.56920016e-01 -4.33558345e-01
-3.34958017e-01 -7.19518483e-01 -6.75670356e-02 -7.14904785e-01
-1.17627788e+00 4.38291997e-01 1.30685866e-01 -8.56312513e-01
2.84036845e-01 -1.14994335e+00 -5.41655779e-01 1.32639468e+00
-1.63082325e+00 -1.03161836e+00 -2.46332020e-01 3.81625831e-01
1.05419330e-01 -3.07417542e-01 1.05747235e+00 5.11193991e-01
-4.19743448e-01 4.47665095e-01 5.96316755e-01 -4.85831410e-01
4.84383881e-01 -1.35804284e+00 -2.25545332e-01 -4.60132919e-02
-3.12926024e-01 7.68569529e-01 5.98393083e-01 -9.08492327e-01
-1.98525953e+00 -9.78024781e-01 6.53541982e-01 -3.37904066e-01
6.36425138e-01 -5.62006354e-01 -8.47262561e-01 -2.67211888e-02
2.59091079e-01 -2.54939497e-01 8.52430224e-01 -3.82908136e-01
3.46974917e-02 -3.46714497e-01 -1.12911355e+00 2.30892390e-01
6.30648851e-01 -6.18100464e-01 -2.35076174e-02 5.22218347e-01
3.74255538e-01 -4.23936397e-01 -1.21273541e+00 3.48898798e-01
5.94687223e-01 -6.36653841e-01 8.85122895e-01 -7.76663423e-01
1.71715200e-01 -4.43623066e-01 -3.48257154e-01 -1.03012013e+00
4.66489382e-02 -4.47324604e-01 -8.76570418e-02 9.12572205e-01
6.67866886e-01 -3.72501403e-01 6.39443219e-01 8.24089050e-01
-2.58827269e-01 -9.82580781e-01 -7.71113157e-01 -5.62662303e-01
2.54980743e-01 -1.37532413e-01 8.94619465e-01 4.17594552e-01
-2.51731783e-01 5.02857029e-01 1.10059619e-01 1.33711115e-01
3.74741137e-01 -6.47960305e-02 3.95522445e-01 -1.28693783e+00
1.26361817e-01 -3.11528891e-01 -5.45560122e-01 -3.72409821e-01
1.20354891e-01 -1.29300654e+00 -2.12005049e-01 -1.59744978e+00
1.23907141e-01 -4.86400962e-01 -7.26595819e-01 1.80941775e-01
2.68572688e-01 -9.86430272e-02 -7.15927109e-02 -3.74195874e-01
-5.67397535e-01 3.76179129e-01 1.05809093e+00 -5.22410870e-01
-4.48634088e-01 -2.81907886e-01 -7.67372191e-01 -4.17166390e-02
5.21457195e-01 -3.13709676e-01 -4.95531410e-01 1.85342461e-01
4.24729615e-01 1.13410093e-01 2.66762078e-01 -1.18441617e+00
2.51683444e-01 9.76597443e-02 7.76495457e-01 -6.43066287e-01
2.66984195e-01 -8.19821119e-01 6.44130886e-01 5.07136345e-01
-2.67854035e-01 -6.91886395e-02 6.54366493e-01 6.77767158e-01
2.93641686e-02 -2.99679250e-01 3.36157739e-01 3.38892974e-02
-7.04418302e-01 4.73772675e-01 -8.21655467e-02 -6.61628664e-01
8.82949948e-01 -5.44842124e-01 -1.93360518e-03 -4.25950140e-02
-8.36993992e-01 1.98323473e-01 1.37557670e-01 9.84872878e-02
6.01876676e-01 -1.17102432e+00 -3.17234427e-01 2.36276492e-01
1.50784537e-01 -1.25177830e-01 2.25436568e-01 6.40203118e-01
-7.08189487e-01 6.94222450e-01 -2.60527074e-01 -6.00034952e-01
-1.07772899e+00 6.79006577e-01 7.89672732e-01 -2.46913239e-01
-1.00323103e-01 6.17396176e-01 1.08050615e-01 -1.69288129e-01
2.27867365e-01 -5.20755529e-01 -2.90837705e-01 1.51300594e-01
1.54993385e-01 2.48255715e-01 5.52705884e-01 -5.42829037e-01
-4.49017763e-01 5.70661247e-01 1.56696253e-02 4.26707685e-01
1.88732874e+00 3.23560148e-01 -2.29839996e-01 2.89425403e-01
1.37289321e+00 -4.39930588e-01 -1.09342146e+00 2.58664310e-01
1.60360798e-01 1.96990356e-01 1.01313204e-01 -1.16270506e+00
-7.39641428e-01 7.48318672e-01 1.28269458e+00 -2.60632455e-01
9.52524424e-01 -3.70643020e-01 3.57998967e-01 7.85813034e-01
4.38619666e-02 -1.10054159e+00 -1.44743681e-01 1.70371845e-01
8.68058562e-01 -1.02061772e+00 3.60993743e-01 -8.22540522e-02
-2.41008162e-01 1.50029480e+00 4.00801927e-01 -1.30530894e-01
6.12205029e-01 1.77953124e-01 -4.95041877e-01 -6.68448448e-01
-4.36028808e-01 -4.43370640e-02 4.34312195e-01 6.46970689e-01
5.00447035e-01 2.10147560e-01 -4.50500697e-01 7.31834352e-01
2.29354426e-01 1.85549155e-01 4.93522733e-02 8.95902514e-01
-2.37956017e-01 -1.49289560e+00 -1.18848599e-01 6.22319758e-01
-2.35827416e-02 -2.72781223e-01 -4.37883854e-01 8.38957965e-01
6.98824406e-01 5.73623300e-01 -1.31438166e-01 -1.78347528e-01
4.17787790e-01 2.30768859e-01 7.89188325e-01 -3.72535259e-01
-7.88144708e-01 1.69128925e-01 1.79255739e-01 -6.63789436e-02
-4.45518225e-01 -3.48836601e-01 -1.50374508e+00 -7.98595995e-02
-5.93093395e-01 3.39609563e-01 1.17691386e+00 1.10051286e+00
3.35707366e-01 7.09164381e-01 3.23517710e-01 -1.22215307e+00
-1.46189809e-01 -5.46582699e-01 -8.41728806e-01 3.78103644e-01
3.30478996e-01 -8.51304114e-01 5.71282357e-02 -1.67630181e-01] | [5.118278980255127, 5.577910900115967] |
8680fd58-34c4-4dd8-a05b-8775fcf4804a | bidirectional-generative-framework-for-cross | 2305.09509 | null | https://arxiv.org/abs/2305.09509v1 | https://arxiv.org/pdf/2305.09509v1.pdf | Bidirectional Generative Framework for Cross-domain Aspect-based Sentiment Analysis | Cross-domain aspect-based sentiment analysis (ABSA) aims to perform various fine-grained sentiment analysis tasks on a target domain by transferring knowledge from a source domain. Since labeled data only exists in the source domain, a model is expected to bridge the domain gap for tackling cross-domain ABSA. Though domain adaptation methods have proven to be effective, most of them are based on a discriminative model, which needs to be specifically designed for different ABSA tasks. To offer a more general solution, we propose a unified bidirectional generative framework to tackle various cross-domain ABSA tasks. Specifically, our framework trains a generative model in both text-to-label and label-to-text directions. The former transforms each task into a unified format to learn domain-agnostic features, and the latter generates natural sentences from noisy labels for data augmentation, with which a more accurate model can be trained. To investigate the effectiveness and generality of our framework, we conduct extensive experiments on four cross-domain ABSA tasks and present new state-of-the-art results on all tasks. Our data and code are publicly available at \url{https://github.com/DAMO-NLP-SG/BGCA}. | ['Lidong Bing', 'Sinno Jialin Pan', 'Wenxuan Zhang', 'Yue Deng'] | 2023-05-16 | null | null | null | null | ['aspect-based-sentiment-analysis'] | ['natural-language-processing'] | [ 1.69116467e-01 -2.35748291e-01 -7.95159638e-02 -7.99593210e-01
-1.27888763e+00 -7.34582365e-01 6.21614337e-01 -1.65068373e-01
-1.10577092e-01 7.03923702e-01 1.65148169e-01 -6.82951137e-02
1.31552666e-01 -8.39600444e-01 -5.62865794e-01 -7.81655610e-01
6.20154977e-01 5.86898446e-01 -1.10937946e-01 -4.76819426e-01
-1.15273811e-01 -1.04674615e-01 -9.75591063e-01 5.13743460e-01
9.29758489e-01 1.06520939e+00 1.20197251e-01 1.93505749e-01
-2.89694399e-01 4.62397844e-01 -6.62332356e-01 -8.07432294e-01
9.86146256e-02 -5.26703596e-01 -8.60918760e-01 2.21459582e-01
-2.03192066e-02 1.17402516e-01 -1.44269527e-03 1.07988298e+00
5.45159280e-01 -3.47160101e-02 7.74355412e-01 -1.29341757e+00
-8.41954708e-01 3.78821433e-01 -6.88294291e-01 -2.25425899e-01
1.98320433e-01 6.05304129e-02 1.16438162e+00 -1.03523588e+00
5.85565269e-01 1.23600662e+00 5.39196968e-01 6.99328482e-01
-1.16893625e+00 -9.51152623e-01 5.48576176e-01 -1.40984610e-01
-1.09604800e+00 -2.63837278e-01 1.13952184e+00 -4.30526525e-01
4.89637285e-01 -9.18948427e-02 3.53616655e-01 1.59977555e+00
-1.84338957e-01 1.09018993e+00 1.38363695e+00 -2.99726635e-01
1.25484556e-01 2.58342683e-01 1.65016264e-01 3.56106728e-01
3.45063694e-02 -3.26613307e-01 -6.01489961e-01 -1.81150272e-01
3.15314204e-01 -1.18887313e-01 -2.29955316e-01 -4.14371938e-01
-1.07670820e+00 1.16592395e+00 2.34761849e-01 3.64629239e-01
-6.14125878e-02 -3.80985469e-01 7.13440299e-01 3.28427166e-01
9.84974444e-01 3.73067230e-01 -9.07499790e-01 -1.00124642e-01
-6.30214274e-01 2.77664244e-01 8.21623445e-01 1.00031722e+00
8.68340611e-01 -3.91481929e-02 -8.09792355e-02 1.27006423e+00
4.02157485e-01 6.33215606e-01 6.03138983e-01 -3.49365532e-01
6.86321020e-01 8.16120982e-01 -5.98830655e-02 -8.79351616e-01
-3.51417899e-01 -4.90889609e-01 -8.51773500e-01 -8.91290233e-02
5.16489148e-01 -3.04940939e-01 -7.94211686e-01 1.99765730e+00
4.30260897e-01 -6.84529692e-02 2.75053769e-01 8.75867367e-01
7.41352141e-01 8.49983573e-01 1.14114530e-01 1.43865064e-01
1.51780486e+00 -1.12965894e+00 -5.88464916e-01 -6.35993004e-01
9.07595277e-01 -8.87944281e-01 1.48275685e+00 3.51752758e-01
-6.86142087e-01 -5.74838817e-01 -1.01172698e+00 -2.39397869e-01
-6.18711591e-01 3.67911994e-01 2.27056846e-01 4.94215608e-01
-7.04413116e-01 3.59338224e-02 -5.72508574e-01 -2.79114723e-01
5.12977719e-01 3.97889540e-02 -4.01004821e-01 -4.48192656e-01
-1.48304355e+00 6.26806796e-01 3.51671249e-01 3.30989808e-02
-9.19121027e-01 -5.21556318e-01 -1.12965524e+00 -1.62394732e-01
2.73365408e-01 -8.41866553e-01 1.28901958e+00 -1.09489429e+00
-1.39064860e+00 1.23196125e+00 -2.47995973e-01 -2.24365398e-01
3.22077066e-01 -3.14049959e-01 -5.70810854e-01 -2.15450019e-01
5.15995681e-01 3.96210730e-01 9.29510057e-01 -1.39327157e+00
-5.51613688e-01 -7.10698664e-01 2.17353433e-01 1.02785289e-01
-6.29945993e-01 7.94457570e-02 -4.91247863e-01 -9.41015661e-01
-3.24344009e-01 -1.08787453e+00 -8.80579725e-02 -3.66834342e-01
-4.80261892e-01 -3.64879757e-01 8.65959346e-01 -6.12481356e-01
1.08448660e+00 -2.34886742e+00 1.55180961e-01 -2.01663107e-01
-1.23686660e-02 3.68095338e-01 -4.00921553e-01 4.14727807e-01
-1.43620893e-01 -8.79011974e-02 -4.62091923e-01 -7.06284523e-01
1.61426425e-01 2.36227065e-02 -4.31278735e-01 2.32980952e-01
4.32117820e-01 8.07243943e-01 -8.11225951e-01 -3.30853820e-01
-4.10494097e-02 5.46972930e-01 -4.01594698e-01 4.31259513e-01
-4.02993023e-01 6.32054865e-01 -8.31123054e-01 7.30290830e-01
9.02394056e-01 -4.14645463e-01 1.07340433e-01 -1.60721272e-01
2.32278541e-01 4.54024166e-01 -9.39315796e-01 1.98470402e+00
-8.70983720e-01 2.16839328e-01 1.55565128e-01 -1.28715551e+00
1.19818604e+00 2.45855868e-01 2.80216664e-01 -7.06105411e-01
3.31004441e-01 3.18278641e-01 -2.88729131e-01 -2.94713080e-01
3.38758737e-01 -5.80493033e-01 -6.86629474e-01 5.27614713e-01
2.97343612e-01 -2.10438296e-01 2.71277815e-01 1.42628267e-01
7.20186114e-01 2.17878655e-01 3.24095547e-01 -2.88955063e-01
7.43025541e-01 1.50232002e-01 7.53351092e-01 3.27858925e-01
-1.17880143e-01 7.37842321e-01 5.34048975e-01 -3.55968207e-01
-8.93661976e-01 -8.65854084e-01 -1.85245469e-01 1.38496101e+00
6.95012137e-02 -4.02888596e-01 -7.47354388e-01 -1.24131298e+00
-1.27120316e-01 7.11610436e-01 -7.44317532e-01 -2.44699433e-01
-2.17547074e-01 -9.49401140e-01 3.92611951e-01 4.86942500e-01
5.90559542e-01 -1.03137493e+00 2.27220505e-01 1.57316968e-01
-4.18092877e-01 -1.31920731e+00 -5.76256454e-01 1.91003487e-01
-6.31261051e-01 -9.23762023e-01 -7.77190685e-01 -9.86889362e-01
5.17399549e-01 2.49795109e-01 1.35334623e+00 -4.39242572e-01
1.69319451e-01 1.24740884e-01 -7.21976936e-01 -5.47031224e-01
-5.06282270e-01 4.56949711e-01 -1.09947793e-01 2.48970032e-01
7.85584748e-01 -4.93934512e-01 -4.75910008e-01 4.70343620e-01
-9.63216424e-01 2.23867986e-02 5.16034245e-01 1.00501478e+00
7.02671826e-01 -8.77108350e-02 9.35281396e-01 -1.36510706e+00
8.45130920e-01 -6.76869273e-01 -5.04255891e-01 8.34598020e-02
-3.83551300e-01 -1.11332849e-01 9.58567560e-01 -1.89709619e-01
-1.29509974e+00 5.86946458e-02 -4.34769422e-01 -3.03892970e-01
-4.07077223e-01 6.78589225e-01 -7.80191958e-01 4.33119953e-01
4.45666790e-01 3.72690976e-01 -1.98085427e-01 -6.66201472e-01
5.21372378e-01 8.60741615e-01 1.99859381e-01 -8.08593035e-01
7.08305836e-01 4.95442182e-01 -5.33470333e-01 -3.36478502e-01
-1.52487421e+00 -6.20527923e-01 -5.84679127e-01 1.03265077e-01
8.91720653e-01 -1.30961645e+00 2.41968571e-03 6.79559171e-01
-1.06702840e+00 -3.30421507e-01 -1.90550119e-01 9.33983698e-02
-4.61880565e-01 7.68792704e-02 -4.61572111e-01 -3.58724385e-01
-4.18094903e-01 -1.22638226e+00 1.39180672e+00 2.09330201e-01
-1.79238871e-01 -1.29899275e+00 4.54468012e-01 6.94745183e-01
1.15195319e-01 1.50850536e-02 8.14770162e-01 -1.04094863e+00
-5.97311147e-02 -1.82389870e-01 -3.12990248e-01 8.10359597e-01
3.73485416e-01 -3.52645755e-01 -1.26838052e+00 -4.45854574e-01
2.22386241e-01 -7.20938861e-01 7.78698266e-01 8.64309445e-02
1.12324846e+00 -1.11589894e-01 -2.59259552e-01 5.09115517e-01
1.33302367e+00 -1.56283490e-02 2.53062010e-01 4.73622859e-01
7.23704040e-01 6.07000947e-01 9.94488657e-01 5.22047162e-01
6.05563760e-01 6.12840354e-01 2.67091841e-01 -1.88704446e-01
-1.05039380e-01 -3.77343595e-01 3.62432957e-01 1.02931249e+00
4.42468315e-01 -4.97655213e-01 -8.71077657e-01 9.45658386e-01
-1.67541027e+00 -5.12980878e-01 2.43344326e-02 1.78044105e+00
1.12754357e+00 8.52214396e-02 1.73793644e-01 -1.26729375e-02
6.25415444e-01 4.43036407e-01 -6.21635020e-01 -3.54849964e-01
-2.04766393e-01 8.49369243e-02 -7.85566941e-02 1.62730142e-01
-1.40363729e+00 1.04285407e+00 4.85806036e+00 1.02527332e+00
-1.03146291e+00 4.31866974e-01 6.97675169e-01 2.79797968e-02
-4.38535869e-01 -9.22049284e-02 -9.41244662e-01 6.50181472e-01
7.48254836e-01 -2.28252530e-01 -1.07167363e-01 1.23423719e+00
-2.62543000e-02 3.68661582e-01 -9.34741378e-01 8.27427208e-01
1.96294356e-02 -9.40096974e-01 1.49436712e-01 1.09310346e-02
9.78017211e-01 7.81313330e-02 2.00058848e-01 6.69043958e-01
4.91791397e-01 -6.49387896e-01 4.49801177e-01 6.03355356e-02
8.07781339e-01 -8.39504063e-01 8.74560118e-01 2.81135499e-01
-1.02077115e+00 2.06296816e-01 -3.26310873e-01 8.39795172e-02
2.51018494e-01 9.98148263e-01 -5.94163656e-01 8.63956273e-01
7.15632498e-01 1.08889675e+00 -4.76960659e-01 4.30709809e-01
-3.72979939e-01 5.53998947e-01 2.02312637e-02 1.04360841e-01
4.18190658e-01 -3.78843963e-01 3.38774204e-01 1.39783847e+00
2.37498343e-01 -2.31892794e-01 1.52334765e-01 7.99465001e-01
-3.92140299e-01 2.97240436e-01 -6.96474612e-01 -1.23126633e-01
2.87141919e-01 1.48237467e+00 -3.83378386e-01 -2.45124295e-01
-8.85239363e-01 9.97460544e-01 4.71750140e-01 3.27971578e-01
-8.26401472e-01 -3.63423645e-01 1.03435314e+00 3.25334165e-03
2.99276620e-01 -5.58593720e-02 -3.88806701e-01 -1.53299975e+00
1.35205045e-01 -1.23448551e+00 5.49251914e-01 -6.32398963e-01
-2.03331375e+00 8.06355774e-01 -2.67832637e-01 -1.46140027e+00
-2.68723607e-01 -7.50008047e-01 -4.08582181e-01 1.02983403e+00
-1.71985650e+00 -1.50668466e+00 -2.73378819e-01 8.43748212e-01
7.11876273e-01 -2.61134237e-01 8.94838095e-01 5.42619944e-01
-6.00988567e-01 6.71536326e-01 2.64643103e-01 3.81261677e-01
1.22399938e+00 -1.17480075e+00 4.32311058e-01 8.22434187e-01
1.85462777e-02 4.28257674e-01 5.04460990e-01 -4.69672650e-01
-1.09549773e+00 -1.38067043e+00 7.93597460e-01 -6.41771793e-01
9.38298702e-01 -7.04252958e-01 -1.05604708e+00 8.13422203e-01
2.18402475e-01 8.68617650e-03 1.03628540e+00 3.97307903e-01
-5.68119466e-01 -1.78650990e-01 -7.92099476e-01 3.42085540e-01
7.82845616e-01 -6.84679627e-01 -6.19184017e-01 3.18302542e-01
6.71203196e-01 -3.18227947e-01 -7.98899174e-01 4.22863156e-01
1.97812602e-01 -8.04823935e-01 7.28217721e-01 -5.78997672e-01
7.02087522e-01 -2.99107105e-01 -1.94533795e-01 -1.70760286e+00
-2.85686497e-02 -1.64087519e-01 2.95511454e-01 1.70933676e+00
6.36881888e-01 -7.58109152e-01 5.95821023e-01 3.13770026e-01
-1.83893457e-01 -7.59719968e-01 -6.69208169e-01 -7.78245628e-01
4.86649841e-01 -5.64119637e-01 6.95058405e-01 1.25141156e+00
-1.50569022e-01 7.05737710e-01 -3.89367044e-01 -2.22664233e-03
4.11244690e-01 6.80929601e-01 9.29000854e-01 -1.07991052e+00
-1.40700787e-01 -2.86488801e-01 -7.64542865e-03 -1.12922430e+00
5.07856131e-01 -9.99770105e-01 9.48524103e-03 -1.54405153e+00
3.04513901e-01 -5.06728053e-01 -3.18245739e-01 5.43010950e-01
-3.45683992e-01 2.98479825e-01 1.11364260e-01 9.03235376e-02
-7.51419544e-01 9.68585014e-01 1.43225598e+00 -2.17981592e-01
1.34907672e-02 1.79489240e-01 -1.16199458e+00 7.46013999e-01
9.26812828e-01 -6.95470989e-01 -4.82749701e-01 -6.10621929e-01
3.16919029e-01 -2.12571517e-01 1.03110097e-01 -5.15043378e-01
-1.53135970e-01 1.50352987e-02 5.25881499e-02 -4.05746281e-01
3.96123081e-01 -6.54119790e-01 -3.77218664e-01 -9.50839520e-02
-2.27084041e-01 -1.24181964e-01 2.94782221e-01 5.75059116e-01
-8.32759321e-01 -9.20631662e-02 8.71066451e-01 1.48745542e-02
-6.50831997e-01 3.87654424e-01 4.62869322e-03 5.88293552e-01
8.86064768e-01 2.96696126e-01 -3.16293597e-01 -3.23681176e-01
-7.12979913e-01 2.74766803e-01 5.38416803e-01 6.57114685e-01
3.04289758e-01 -1.44194782e+00 -9.16796088e-01 1.70543313e-01
6.31947637e-01 2.75151014e-01 4.59360093e-01 5.39814413e-01
3.11340075e-02 4.48343366e-01 -4.66725491e-02 -4.52355206e-01
-9.89449024e-01 5.95074713e-01 1.93809494e-01 -6.85936570e-01
-1.76093340e-01 9.59417105e-01 8.70528460e-01 -8.88313830e-01
-2.65160054e-01 2.08820328e-02 -1.62994668e-01 2.44324654e-01
4.83159035e-01 -1.35325894e-01 1.63755432e-01 -7.30007410e-01
-3.62643152e-01 5.89609087e-01 -3.69096935e-01 1.15393095e-01
1.34515905e+00 -3.52266461e-01 4.38594259e-02 4.51455474e-01
1.47764897e+00 3.31905037e-02 -1.26628840e+00 -5.92477262e-01
-2.61952907e-01 -3.43509793e-01 -9.14555937e-02 -1.02056813e+00
-1.17230308e+00 1.04088104e+00 1.66218042e-01 1.93375543e-01
1.45565665e+00 2.88495362e-01 8.70727658e-01 -4.63223122e-02
2.02440202e-01 -1.06803548e+00 1.47756740e-01 6.34154916e-01
9.52971339e-01 -1.57655597e+00 -2.56370723e-01 -4.16688770e-01
-1.13577282e+00 8.51621985e-01 7.65860915e-01 4.84690256e-03
7.14087665e-01 2.61598974e-01 3.77694070e-01 -1.56273380e-01
-5.30896664e-01 -1.68483421e-01 2.21259445e-01 5.75464725e-01
5.13545930e-01 1.09969907e-01 -2.47534633e-01 1.36234891e+00
-2.18615159e-01 -1.69157758e-02 1.59617752e-01 7.97218740e-01
1.04426645e-01 -1.56040418e+00 -1.11404851e-01 1.31656975e-01
-4.84721452e-01 -7.24040344e-02 -3.78359407e-01 9.15713787e-01
2.37637497e-02 9.77501571e-01 -3.06454897e-01 -2.95814455e-01
5.48429370e-01 2.51832217e-01 4.62594032e-02 -7.88734615e-01
-4.06407028e-01 1.62957355e-01 2.00470299e-01 -3.74952495e-01
-5.94832778e-01 -7.90298879e-01 -8.79216611e-01 -9.40515250e-02
-1.24485783e-01 1.35490060e-01 6.19146466e-01 1.02006364e+00
5.24365783e-01 5.48899949e-01 7.89825320e-01 -4.24098253e-01
-3.33600730e-01 -1.12044442e+00 -7.66579926e-01 7.03673601e-01
1.95837602e-01 -5.58012605e-01 -3.09500992e-01 1.27024502e-01] | [11.409918785095215, 6.688283443450928] |
c89320be-12f3-4377-96c4-8c26ca0b17da | applade-adjustable-plug-and-play-audio | 2202.08028 | null | https://arxiv.org/abs/2202.08028v1 | https://arxiv.org/pdf/2202.08028v1.pdf | APPLADE: Adjustable Plug-and-play Audio Declipper Combining DNN with Sparse Optimization | In this paper, we propose an audio declipping method that takes advantages of both sparse optimization and deep learning. Since sparsity-based audio declipping methods have been developed upon constrained optimization, they are adjustable and well-studied in theory. However, they always uniformly promote sparsity and ignore the individual properties of a signal. Deep neural network (DNN)-based methods can learn the properties of target signals and use them for audio declipping. Still, they cannot perform well if the training data have mismatches and/or constraints in the time domain are not imposed. In the proposed method, we use a DNN in an optimization algorithm. It is inspired by an idea called plug-and-play (PnP) and enables us to promote sparsity based on the learned information of data, considering constraints in the time domain. Our experiments confirmed that the proposed method is stable and robust to mismatches between training and test data. | ['Yasuhiro Oikawa', 'Masahiro Yasuda', 'Kohei Yatabe', 'Tomoro Tanaka'] | 2022-02-16 | null | null | null | null | ['audio-declipping'] | ['audio'] | [ 1.70234248e-01 -2.61424780e-01 -4.06709731e-01 -2.09704831e-01
-3.48402977e-01 -1.63481340e-01 1.83260784e-01 -2.15682402e-01
-2.50984550e-01 6.28667653e-01 3.86760324e-01 1.35867625e-01
-3.72011930e-01 -5.62796414e-01 -6.23879611e-01 -8.12322497e-01
4.71422449e-02 9.07612741e-02 -4.41389084e-02 -1.56839028e-01
4.66882512e-02 2.67460614e-01 -1.67365491e+00 1.20070711e-01
8.18417251e-01 1.13082230e+00 2.18331248e-01 2.70738661e-01
-2.81894729e-02 6.58674240e-01 -5.78791142e-01 5.68278842e-02
4.08926427e-01 -4.57522631e-01 -5.02954572e-02 1.16270222e-01
2.01859713e-01 -9.54818130e-02 -6.54495299e-01 1.09708846e+00
8.19797516e-01 1.64636716e-01 2.35609084e-01 -1.22689331e+00
-3.42748076e-01 8.06033432e-01 -3.75740051e-01 1.34094283e-01
3.58628362e-01 -5.21135107e-02 8.97429407e-01 -9.35171902e-01
3.07558328e-01 8.56263995e-01 9.33679998e-01 2.97995657e-01
-1.01464176e+00 -8.99804354e-01 2.68186688e-01 3.82211685e-01
-1.42413199e+00 -6.28036916e-01 1.38708997e+00 -2.12608874e-01
4.52723145e-01 3.53464931e-01 9.91642475e-01 1.09407127e+00
-1.00810239e-02 9.62114573e-01 6.93052769e-01 -4.35685843e-01
3.20448846e-01 2.04827487e-02 -2.80946791e-01 3.12790334e-01
-1.46227807e-01 3.75371367e-01 -9.01977122e-01 -8.02105591e-02
6.16526365e-01 5.22120595e-02 -6.26741230e-01 -2.79622257e-01
-1.23263466e+00 7.94894755e-01 2.21199036e-01 4.34052259e-01
-3.31719875e-01 -1.21248603e-01 4.74900573e-01 3.83876175e-01
2.46353850e-01 4.94673938e-01 -2.81389832e-01 -1.71187803e-01
-1.27701974e+00 2.20973253e-01 7.21675336e-01 7.93079615e-01
5.09415150e-01 8.20479751e-01 -1.13360234e-01 9.60352063e-01
1.84670284e-01 3.42585683e-01 9.81607318e-01 -8.33116353e-01
4.67633128e-01 1.75290704e-01 -5.97708151e-02 -1.42797196e+00
-5.77522516e-01 -9.21701491e-01 -1.08396256e+00 -3.60228558e-04
2.91065812e-01 -3.46511930e-01 -6.90119982e-01 1.84349239e+00
2.99074769e-01 7.54160941e-01 3.38919200e-02 9.94140387e-01
7.63227165e-01 6.93660438e-01 -3.15374911e-01 -5.48751473e-01
7.94128060e-01 -7.55608499e-01 -1.06459653e+00 -9.85901728e-02
2.47549146e-01 -7.58002222e-01 1.09058678e+00 8.16623688e-01
-1.03530550e+00 -7.01698184e-01 -1.28069985e+00 2.85243094e-01
3.09669450e-02 1.55388460e-01 5.49919426e-01 5.71123719e-01
-8.00743282e-01 5.99976659e-01 -7.23840475e-01 7.31069446e-02
-7.12671652e-02 4.63138133e-01 -2.18660757e-01 2.33337656e-01
-1.45359659e+00 3.99049997e-01 4.18718606e-01 4.81540084e-01
-1.01693046e+00 -6.50976121e-01 -8.00484598e-01 2.83814788e-01
5.24868369e-01 -4.19136494e-01 1.06342387e+00 -1.12680924e+00
-1.98308003e+00 3.65845621e-01 1.17411770e-01 -5.38217604e-01
3.30312699e-01 -3.84778082e-01 -6.85302019e-01 1.55250221e-01
-2.43488908e-01 1.38510033e-01 1.47078991e+00 -7.95521498e-01
-5.25412321e-01 3.79739180e-02 -3.50950137e-02 1.67090312e-01
-6.23452783e-01 -3.32236290e-01 -3.44071090e-01 -1.07251191e+00
5.60329795e-01 -6.80707574e-01 -1.69294342e-01 -8.38437825e-02
-5.92399120e-01 2.37647474e-01 1.08264124e+00 -5.08133769e-01
1.53098083e+00 -2.51633668e+00 1.39128536e-01 4.05147403e-01
1.38568491e-01 4.11173642e-01 -2.22312436e-01 4.65033531e-01
-2.36421451e-01 -2.39020914e-01 -1.35052860e-01 -3.60765278e-01
-1.24619640e-02 1.57270178e-01 -3.16402674e-01 5.64376771e-01
-8.44758283e-03 3.59489322e-01 -7.35288024e-01 -5.02030551e-01
2.20388293e-01 4.54683214e-01 -9.48925853e-01 4.15745825e-01
-1.99858770e-01 6.87509060e-01 -3.23493510e-01 5.92954516e-01
6.59850478e-01 1.70088664e-01 2.71769643e-01 -4.38078254e-01
-2.30089143e-01 2.42219031e-01 -1.47357571e+00 1.65328193e+00
-5.48171699e-01 6.11285090e-01 4.30942833e-01 -1.56194901e+00
1.13687348e+00 6.38044834e-01 7.25924134e-01 -5.72086394e-01
1.28356650e-01 3.28625202e-01 -7.36403838e-03 -5.92192650e-01
2.55879611e-01 -1.87960729e-01 3.29051137e-01 1.87813520e-01
1.95963219e-01 -2.95847267e-01 7.55273551e-02 -2.56047159e-01
7.07705379e-01 -9.89008546e-02 2.02690572e-01 -1.27500683e-01
6.55405104e-01 -6.22433662e-01 1.08860278e+00 6.92398906e-01
6.33105487e-02 7.84379482e-01 5.44210374e-01 -3.68173867e-01
-9.11229551e-01 -6.42436981e-01 -2.57323742e-01 1.03012896e+00
8.60896185e-02 -5.05317748e-01 -4.30583119e-01 -1.96828440e-01
-1.05940223e-01 3.30922872e-01 -3.20102006e-01 -3.36644471e-01
-6.90962315e-01 -6.01123273e-01 4.92846072e-01 3.27728719e-01
4.61393863e-01 -7.87812114e-01 -2.58300632e-01 5.70245266e-01
-1.87439084e-01 -1.11467636e+00 -6.25259995e-01 2.72701651e-01
-9.84469175e-01 -8.69670987e-01 -7.34770238e-01 -9.58044291e-01
3.61361295e-01 3.96006256e-02 7.25708365e-01 -2.78310440e-02
2.39036292e-01 1.86815977e-01 -4.02565718e-01 -3.98756206e-01
-1.05851784e-01 1.88907564e-01 2.88541853e-01 4.98429328e-01
-4.54932749e-02 -1.35231781e+00 -3.51619363e-01 4.62983251e-01
-9.14111495e-01 -1.46159649e-01 6.52005315e-01 1.03506887e+00
6.78694844e-01 3.05798829e-01 7.08979785e-01 -6.87626719e-01
7.23764479e-01 -4.88561690e-01 -7.41681397e-01 -3.15122232e-02
-4.90785390e-01 2.90734116e-02 9.58603382e-01 -8.48478734e-01
-6.82150781e-01 1.43166602e-01 -3.79079044e-01 -8.61634195e-01
2.02909604e-01 9.60860491e-01 -4.28620577e-01 -1.55287415e-01
4.46105063e-01 4.72517133e-01 -5.24240658e-02 -6.79470241e-01
6.89653829e-02 5.52077889e-01 5.87335050e-01 -6.16707325e-01
7.03664064e-01 4.07486618e-01 -6.45557046e-02 -9.57749486e-01
-8.27103078e-01 -2.20100775e-01 -2.69464761e-01 -2.24631727e-01
3.65850389e-01 -8.23981643e-01 -6.81614220e-01 6.57041669e-01
-8.88522327e-01 -1.11280650e-01 -3.50543380e-01 9.68157947e-01
-4.43035811e-01 4.99219090e-01 -2.46903121e-01 -7.23985493e-01
-1.51996791e-01 -9.47194338e-01 6.38566375e-01 2.10865721e-01
4.65792380e-02 -8.44108939e-01 1.54407835e-02 -1.17312089e-01
6.19303882e-01 1.85629562e-01 6.53086305e-01 -6.41120732e-01
-4.12985235e-01 -3.55231255e-01 2.85088718e-01 5.14574885e-01
1.92155749e-01 5.44911139e-02 -8.73613000e-01 -3.06784064e-01
6.66971803e-01 -2.24632218e-01 6.51781201e-01 6.85266614e-01
1.41968036e+00 -6.91672444e-01 4.47088704e-02 1.08124685e+00
1.25298846e+00 5.74728614e-03 6.12388372e-01 2.42213562e-01
5.81759095e-01 4.04171079e-01 4.35778648e-01 9.25606072e-01
-8.74599144e-02 7.47236907e-01 7.01335013e-01 -2.02993751e-02
1.05802029e-01 -3.33038300e-01 4.24538702e-01 1.22515047e+00
-3.48465852e-02 -1.12764135e-01 -5.35500526e-01 5.16069472e-01
-1.84746850e+00 -9.91307199e-01 1.83207884e-01 2.16507435e+00
1.08978581e+00 3.93333972e-01 6.13826476e-02 6.94248438e-01
9.05313551e-01 4.62840319e-01 -5.89255750e-01 -9.93160978e-02
-2.58814365e-01 3.23655158e-01 2.09797189e-01 4.73614514e-01
-9.12908256e-01 4.76867497e-01 6.32942104e+00 9.92653489e-01
-1.65719986e+00 2.05885302e-02 2.84164175e-02 -2.87122130e-01
-3.65355045e-01 -1.86170012e-01 -6.77115381e-01 6.72672391e-01
5.91785669e-01 -1.59943715e-01 4.37551171e-01 8.62636447e-01
4.59222913e-01 2.92319536e-01 -1.12410331e+00 1.15756798e+00
-1.09968774e-01 -1.22148144e+00 -2.44419262e-01 -4.77549404e-01
6.37091458e-01 -2.47404143e-01 3.40770662e-01 2.99740314e-01
-4.28259939e-01 -1.02975583e+00 9.21099186e-01 5.01263738e-01
5.44272900e-01 -6.03999138e-01 7.36491442e-01 5.08687139e-01
-1.21194434e+00 -2.76491046e-01 -3.55743498e-01 -2.62428641e-01
1.58575907e-01 1.02518630e+00 -6.17551088e-01 4.98982608e-01
5.28120518e-01 1.02146304e+00 2.48726718e-02 1.28619385e+00
-5.36140621e-01 9.26912367e-01 -5.31568050e-01 -2.22467985e-02
1.60993740e-01 -2.82103539e-01 8.88170242e-01 9.85152125e-01
4.32858855e-01 -1.09300509e-01 3.17310691e-01 7.13959813e-01
1.01666942e-01 1.00132287e-01 -4.68342811e-01 -5.67085929e-02
6.19534135e-01 8.48430097e-01 -1.54234990e-01 1.85636766e-02
-3.78638446e-01 2.92353123e-01 -9.24138427e-02 5.28421700e-01
-8.55894506e-01 -5.54160357e-01 5.59551716e-01 1.44842818e-01
4.90144342e-01 -3.36705565e-01 -1.60647392e-01 -1.35431516e+00
1.87889919e-01 -1.16023397e+00 2.97598362e-01 -5.06604135e-01
-1.19849288e+00 5.04370928e-01 -1.65493265e-01 -1.68246210e+00
-2.60283828e-01 -1.88460618e-01 -7.34634459e-01 4.69509155e-01
-1.58027840e+00 -8.23803067e-01 -1.97328597e-01 9.92383957e-01
4.35693890e-01 -1.91760838e-01 5.90457737e-01 6.63323104e-01
-5.07269561e-01 6.41437829e-01 2.33926684e-01 2.03380641e-02
5.36171436e-01 -8.11782777e-01 -3.58755082e-01 8.37854326e-01
3.82076949e-01 5.70591927e-01 9.37339664e-01 -2.77242720e-01
-1.32787406e+00 -7.89909422e-01 5.95409453e-01 4.06458020e-01
6.38449728e-01 -4.07631367e-01 -9.67411458e-01 3.91062558e-01
1.54189691e-01 4.08216529e-02 7.06810892e-01 1.79389998e-01
-1.37087792e-01 -5.58433831e-01 -8.65910232e-01 3.75929952e-01
7.87383556e-01 -6.44281089e-01 -6.39196336e-01 3.74783814e-01
7.52489567e-01 -7.31376052e-01 -6.16406441e-01 4.84549433e-01
6.03667974e-01 -1.14263880e+00 9.50290024e-01 -4.89831805e-01
1.45043239e-01 -3.44723046e-01 -2.85589814e-01 -1.27745819e+00
-2.37286419e-01 -1.02752578e+00 -3.84309828e-01 1.07883668e+00
3.00647289e-01 -6.31266177e-01 8.59032393e-01 8.74188617e-02
-4.70248252e-01 -7.34392464e-01 -1.05923975e+00 -1.01854575e+00
-3.48481804e-01 -6.51415825e-01 7.98277020e-01 1.05269241e+00
1.89513013e-01 4.94006425e-02 -9.49840724e-01 4.64223385e-01
4.13766891e-01 1.06438115e-01 5.45059562e-01 -1.02755177e+00
-6.03156507e-01 -2.68506050e-01 -3.85366082e-01 -1.16469550e+00
1.77407861e-01 -5.75253427e-01 2.22267769e-02 -9.15397584e-01
-4.40944403e-01 -5.71363866e-01 -5.55077136e-01 3.91726881e-01
5.05614914e-02 6.57437444e-02 2.54265487e-01 3.17669809e-01
-2.49336287e-01 9.01744425e-01 1.14160228e+00 -3.27867627e-01
-5.49933195e-01 4.61076647e-01 -5.23127139e-01 7.72734940e-01
9.18020725e-01 -5.84089756e-01 -5.57635903e-01 -4.73383933e-01
2.90574759e-01 1.17465325e-01 2.55557802e-02 -1.32074654e+00
3.07426780e-01 -1.72838077e-01 -5.70319518e-02 -5.06586909e-01
5.90094745e-01 -1.02944505e+00 2.22414598e-01 3.74150962e-01
-3.68018329e-01 -1.80965319e-01 2.05833018e-01 5.25562942e-01
-7.49883890e-01 -3.37437302e-01 8.21473718e-01 1.23957731e-01
-4.32848960e-01 3.87218028e-01 -3.37025404e-01 1.25324696e-01
5.99204183e-01 -1.89673796e-01 3.26267421e-01 -8.46279502e-01
-8.50917280e-01 1.47224814e-01 -3.70358303e-02 8.78683627e-02
7.19093621e-01 -1.54097998e+00 -6.36039138e-01 5.25216699e-01
-2.69162685e-01 3.59587371e-02 1.45069212e-01 1.02256441e+00
-1.69741824e-01 3.47350448e-01 -2.33761311e-01 -6.94626570e-01
-9.56256270e-01 5.53173780e-01 2.33121917e-01 -1.74697340e-01
-5.79064071e-01 8.40997636e-01 1.15057170e-01 -3.39462906e-01
7.18041897e-01 -6.06331408e-01 -2.68382728e-01 1.16769008e-01
4.04928178e-01 1.31719694e-01 1.49788365e-01 -2.60283649e-01
-1.75517187e-01 6.65525258e-01 2.86628544e-01 -1.28416613e-01
1.52141416e+00 1.35374325e-03 5.71677648e-02 4.13938701e-01
1.18352556e+00 5.07582307e-01 -1.11696863e+00 -3.91503900e-01
-2.63314247e-01 -5.60811102e-01 1.46087527e-01 -2.10664332e-01
-1.48478842e+00 8.34377170e-01 3.08430105e-01 5.01395166e-01
1.47950935e+00 -5.12238443e-01 9.88830566e-01 3.10325176e-01
1.28319129e-01 -1.27968466e+00 2.90087342e-01 4.59844857e-01
8.72530818e-01 -8.22149336e-01 3.56209539e-02 -3.42794567e-01
-2.73709983e-01 1.31431878e+00 4.84264225e-01 -2.18489975e-01
8.76685977e-01 3.38989347e-01 1.90416933e-03 1.58100113e-01
-5.80281436e-01 6.23866729e-02 1.87284172e-01 8.21857929e-01
1.62825376e-01 -3.15823764e-01 -3.51114482e-01 8.84396434e-01
-3.31645250e-01 2.02475592e-01 4.11941290e-01 7.20858335e-01
-4.81673896e-01 -1.00510836e+00 -5.46612561e-01 2.80596167e-01
-4.02610779e-01 -1.50752172e-01 9.41095129e-02 6.96246743e-01
2.22225443e-01 8.05057824e-01 -2.55202204e-01 -6.25251472e-01
3.35867614e-01 -1.65284932e-01 2.30249181e-01 -5.24414182e-01
-4.42809492e-01 2.81590730e-01 -6.63167611e-02 -6.47355914e-01
-5.41179478e-01 -5.31023920e-01 -9.67127264e-01 -1.57895699e-01
-7.37089813e-01 4.50619221e-01 5.47155023e-01 8.98726344e-01
3.15312743e-01 5.16821802e-01 9.25888777e-01 -9.02855098e-01
-7.31965899e-01 -7.61561692e-01 -9.58475292e-01 1.42071992e-02
6.41419172e-01 -7.12366879e-01 -7.06666172e-01 -3.69375683e-02] | [15.41474723815918, 5.530331611633301] |
f24ef454-23c4-4353-a239-77834edc74d7 | winogavil-gamified-association-benchmark-to | 2207.12576 | null | https://arxiv.org/abs/2207.12576v2 | https://arxiv.org/pdf/2207.12576v2.pdf | WinoGAViL: Gamified Association Benchmark to Challenge Vision-and-Language Models | While vision-and-language models perform well on tasks such as visual question answering, they struggle when it comes to basic human commonsense reasoning skills. In this work, we introduce WinoGAViL: an online game of vision-and-language associations (e.g., between werewolves and a full moon), used as a dynamic evaluation benchmark. Inspired by the popular card game Codenames, a spymaster gives a textual cue related to several visual candidates, and another player tries to identify them. Human players are rewarded for creating associations that are challenging for a rival AI model but still solvable by other human players. We use the game to collect 3.5K instances, finding that they are intuitive for humans (>90% Jaccard index) but challenging for state-of-the-art AI models, where the best model (ViLT) achieves a score of 52%, succeeding mostly where the cue is visually salient. Our analysis as well as the feedback we collect from players indicate that the collected associations require diverse reasoning skills, including general knowledge, common sense, abstraction, and more. We release the dataset, the code and the interactive game, allowing future data collection that can be used to develop models with better association abilities. | ['Roy Schwartz', 'Gabriel Stanovsky', 'Mohit Bansal', 'Yuval Elovici', 'Ron Yosef', 'Nitzan Bitton Guetta', 'Yonatan Bitton'] | 2022-07-25 | null | null | null | null | ['visual-reasoning', 'general-knowledge', 'visual-reasoning', 'multimodal-association'] | ['computer-vision', 'miscellaneous', 'reasoning', 'time-series'] | [-3.08838665e-01 1.77344412e-01 6.25441596e-02 1.66762710e-01
-3.57703209e-01 -9.31413412e-01 6.12249374e-01 3.03017706e-01
-5.28685093e-01 4.65638995e-01 7.38263205e-02 -4.22683030e-01
-6.76942170e-02 -5.40773034e-01 -6.84967041e-01 -1.62041515e-01
-3.23654599e-02 8.57490182e-01 3.63163978e-01 -6.42238259e-01
4.40248400e-01 -1.80186909e-02 -1.81247437e+00 6.14557266e-01
8.27094913e-01 7.71874666e-01 3.66110295e-01 6.90798879e-01
-1.55279726e-01 1.38721776e+00 -5.85621953e-01 -1.06875551e+00
3.31645697e-01 -4.67248768e-01 -8.69581997e-01 -4.69365269e-01
6.89171910e-01 -1.63226336e-01 -1.60568178e-01 9.71988797e-01
3.25800627e-01 2.02225998e-01 6.77321136e-01 -1.53504300e+00
-1.22184300e+00 8.93193364e-01 -6.30222321e-01 3.25950831e-01
7.41534472e-01 8.14531088e-01 1.50221968e+00 -1.08755231e+00
8.61928642e-01 1.28346097e+00 4.68677402e-01 6.24291956e-01
-1.08746290e+00 -7.97107875e-01 6.09615780e-02 8.32657635e-01
-1.14563882e+00 -1.68753527e-02 5.55066943e-01 -6.94807827e-01
1.10044396e+00 3.15644741e-01 1.17053652e+00 9.59448099e-01
-3.22650880e-01 8.80417049e-01 1.23428893e+00 -4.21597183e-01
2.56524414e-01 7.89722651e-02 2.58217286e-02 8.05616856e-01
2.94464916e-01 2.11546898e-01 -9.58564818e-01 4.46897857e-02
7.05011129e-01 -3.24486673e-01 9.21092033e-02 -5.49686432e-01
-1.40194392e+00 7.98119187e-01 8.02043557e-01 1.94054797e-01
-5.99172652e-01 2.63486505e-01 5.62740341e-02 1.27791896e-01
-3.06564957e-01 1.13832843e+00 4.69248779e-02 -5.03710747e-01
-4.78793621e-01 7.00813174e-01 7.14081883e-01 7.86946714e-01
6.00456953e-01 -1.30870953e-01 -3.87963921e-01 7.91743159e-01
1.25290781e-01 4.92618561e-01 2.02957243e-01 -1.19435942e+00
2.74206787e-01 8.10873687e-01 7.34079108e-02 -1.32624960e+00
-4.62279975e-01 -3.26962262e-01 -3.41452032e-01 7.71668792e-01
9.36194539e-01 3.63003239e-02 -6.23780966e-01 1.78612947e+00
6.00681305e-02 -6.96733743e-02 -7.64491688e-03 1.19032097e+00
1.34559643e+00 3.71536821e-01 3.87905508e-01 4.03282911e-01
1.75979149e+00 -1.07668877e+00 -3.10966760e-01 -9.47679460e-01
4.45780098e-01 -5.10144770e-01 1.65419900e+00 5.99026799e-01
-1.30756688e+00 -4.49342817e-01 -8.52500379e-01 -3.47504020e-01
-5.30739725e-01 -1.33113533e-01 1.12883961e+00 2.22413361e-01
-9.62207258e-01 1.41723543e-01 -1.93193197e-01 -5.22705138e-01
5.22054732e-01 -5.84164588e-03 -2.99364388e-01 -1.46373957e-01
-1.26161778e+00 1.32317400e+00 2.47673258e-01 -1.39010787e-01
-8.41222048e-01 -6.27342284e-01 -7.64024913e-01 -5.38558653e-03
5.29751360e-01 -7.73003519e-01 1.16059232e+00 -9.83729005e-01
-9.36756432e-01 1.43283236e+00 3.48562330e-01 -5.62682152e-01
6.40413046e-01 -1.75864413e-01 -1.31385187e-02 1.65031612e-01
3.19909871e-01 1.01078892e+00 4.82168525e-01 -1.33602166e+00
-5.81134439e-01 -1.22283563e-01 8.43446493e-01 3.85567635e-01
-3.94630432e-02 1.44421682e-01 -5.65075338e-01 -5.33308208e-01
-2.83136278e-01 -9.72097993e-01 2.27280278e-02 2.58963674e-01
-3.93516511e-01 -2.73391753e-01 9.45057347e-02 -6.23321772e-01
9.71568882e-01 -1.99262810e+00 3.76209021e-01 7.43010268e-02
5.91746986e-01 -2.82949004e-02 -1.46586373e-01 3.93208772e-01
-7.54458979e-02 1.14606127e-01 1.12315416e-01 -3.38211991e-02
2.87161916e-01 -7.90015515e-03 -1.32680163e-01 -1.43120989e-01
7.61573091e-02 1.44347382e+00 -1.21445322e+00 -6.50895298e-01
4.86343578e-02 -4.48946655e-03 -6.29257441e-01 7.04171509e-03
-2.93734223e-01 3.03027537e-02 9.27455872e-02 6.59723401e-01
2.72557825e-01 -4.67046946e-01 -6.72022253e-02 -8.98669567e-03
1.13416918e-01 5.86815923e-02 -8.18136752e-01 1.52454638e+00
-1.45507246e-01 9.21543539e-01 -2.73482382e-01 -3.58833224e-01
8.01775396e-01 -3.80106717e-01 -1.46245986e-01 -8.98404956e-01
-1.06136367e-01 -1.72568381e-01 5.93009174e-01 -6.69080377e-01
6.96985960e-01 -1.02024443e-01 -1.44204721e-01 3.65165234e-01
-2.06653103e-01 -5.72330236e-01 5.41880846e-01 6.97593987e-01
1.03514862e+00 9.84107256e-02 3.90821099e-01 -6.51538223e-02
1.42381536e-02 5.77613771e-01 1.41972795e-01 1.15228772e+00
-3.14948708e-01 3.78226429e-01 8.83219540e-01 -5.11957407e-01
-1.05017018e+00 -1.16372991e+00 6.18163705e-01 1.50903976e+00
4.46526319e-01 -5.95974982e-01 -5.04975855e-01 -3.34877670e-01
1.66180041e-02 1.08818245e+00 -9.40725505e-01 -2.36488178e-01
-2.01339871e-01 -1.83616087e-01 5.21797180e-01 6.80211961e-01
3.57276231e-01 -1.56233239e+00 -1.03083837e+00 -2.00172201e-01
-3.01062465e-01 -9.78510261e-01 -6.51142150e-02 -3.55566368e-02
-8.94415006e-02 -1.23592293e+00 -5.18863797e-01 -7.46874988e-01
3.27470422e-01 2.59345621e-01 1.75920451e+00 5.13110816e-01
-4.57443833e-01 6.13057196e-01 -4.05768156e-01 -7.30884016e-01
-2.30696023e-01 -4.27321494e-01 -2.04334274e-01 -5.31677186e-01
6.30684793e-01 -2.80188113e-01 -5.79484522e-01 1.32756293e-01
-4.84703988e-01 5.40996969e-01 6.07827723e-01 8.06339145e-01
3.33945423e-01 -3.72383475e-01 7.34881237e-02 -4.34770405e-01
9.02962804e-01 -5.01500249e-01 -2.80009806e-01 2.65053004e-01
-2.00628102e-01 -2.25218505e-01 3.13459933e-01 -5.83099246e-01
-5.62597156e-01 -1.44364655e-01 4.65307862e-01 -4.41657215e-01
1.89030226e-02 5.22934854e-01 3.28146189e-01 -8.87899771e-02
1.32464921e+00 1.05165847e-01 -8.86595249e-02 8.04626048e-02
7.36347020e-01 5.90631515e-02 8.29174101e-01 -8.35775793e-01
9.25052941e-01 2.54878312e-01 -1.42609179e-01 -6.28048599e-01
-8.15108597e-01 -2.01906845e-01 -3.30378026e-01 -5.37680209e-01
1.05738521e+00 -9.13144827e-01 -1.41596794e+00 2.11643934e-01
-1.30582523e+00 -6.51963472e-01 -3.35124969e-01 2.96476781e-01
-5.20307243e-01 5.61039262e-02 -3.82618308e-01 -8.16536844e-01
-1.65939573e-02 -8.43239009e-01 5.35460591e-01 5.53105772e-01
-6.51920795e-01 -6.27209485e-01 2.42145499e-03 7.57938921e-01
3.88719022e-01 1.13951363e-01 1.07215452e+00 -5.78490674e-01
-5.72044730e-01 1.26260761e-02 -5.80449343e-01 -2.47377023e-01
-4.78614002e-01 6.29112497e-02 -7.18853235e-01 2.58279890e-01
-7.32451677e-01 -9.34147596e-01 9.68419909e-01 1.66863263e-01
1.15202785e+00 3.52130160e-02 -1.69310465e-01 1.57056093e-01
9.82023060e-01 1.89113602e-01 7.16929138e-01 7.45689213e-01
8.12755764e-01 7.23346531e-01 5.71116269e-01 3.97354126e-01
9.56691027e-01 6.90725386e-01 7.86887527e-01 -1.30639896e-01
-1.52902946e-01 -3.46798062e-01 1.16739184e-01 -1.82326641e-02
-5.52579165e-01 -6.00700304e-02 -1.45490098e+00 6.36584818e-01
-1.98203015e+00 -1.30929887e+00 -1.33099988e-01 1.98912930e+00
8.39486182e-01 3.62582147e-01 5.76328278e-01 -1.89026505e-01
4.89984840e-01 -1.29782911e-02 -5.95615149e-01 -5.21865606e-01
-3.49171102e-01 2.71221220e-01 6.64304346e-02 5.41822374e-01
-9.04965997e-01 1.44272900e+00 6.63031149e+00 7.75102079e-01
-6.52825654e-01 -3.10224652e-01 5.91357708e-01 -2.02643156e-01
-4.53351617e-01 2.01327205e-01 -2.83972770e-01 2.08313599e-01
2.92903543e-01 -2.32604265e-01 8.14058542e-01 8.84085774e-01
-1.88440904e-01 -4.23807174e-01 -1.02643847e+00 1.33143425e+00
3.35163146e-01 -1.36821544e+00 1.09190039e-01 -1.87671646e-01
2.24051535e-01 -2.63070166e-01 2.39570886e-01 6.61628783e-01
7.57186353e-01 -1.70566893e+00 9.89288926e-01 6.48108363e-01
4.88309830e-01 -4.85179424e-01 3.29443067e-01 1.94389105e-01
-9.38258410e-01 -2.03724906e-01 -3.46127927e-01 -7.94051528e-01
-3.00639272e-01 -6.02979772e-02 -9.14101303e-01 3.71550210e-02
9.33931589e-01 5.36502540e-01 -1.11983502e+00 1.18006146e+00
-5.30619800e-01 1.76968887e-01 -1.94245204e-01 -6.64032280e-01
1.33897543e-01 7.00049698e-02 4.89765525e-01 9.42456305e-01
5.36833964e-02 2.54713833e-01 1.90775797e-01 1.32388222e+00
1.30956948e-01 1.85322449e-01 -5.50700009e-01 -2.18652308e-01
6.16652250e-01 1.22608769e+00 -6.44165218e-01 -3.02620173e-01
-2.88305670e-01 7.78609574e-01 7.36479640e-01 2.77827114e-01
-9.66093004e-01 -2.19808370e-01 6.71061099e-01 2.26743281e-01
7.31468201e-02 -2.22181797e-01 -6.19713306e-01 -8.88025880e-01
-1.22495227e-01 -1.12877333e+00 5.06546676e-01 -1.62850916e+00
-1.49757469e+00 4.95515168e-01 -1.27648830e-01 -1.03076100e+00
-2.09966287e-01 -8.35075974e-01 -7.68709302e-01 6.38127685e-01
-9.88955021e-01 -1.32833099e+00 -8.06694269e-01 5.95610142e-01
2.35893220e-01 -2.64739275e-01 6.95847213e-01 -3.94246727e-01
-1.86667800e-01 4.41861212e-01 -7.14486241e-01 3.01447779e-01
7.48856246e-01 -1.53034878e+00 5.27101338e-01 6.61609232e-01
5.42693973e-01 5.22423387e-01 1.01390362e+00 -5.99387169e-01
-1.26146376e+00 -2.14826256e-01 5.72870433e-01 -8.07505310e-01
9.95377421e-01 -2.97076523e-01 -6.75643504e-01 5.10313034e-01
4.32127982e-01 -6.52752891e-02 7.45921314e-01 3.75809431e-01
-7.90394485e-01 3.17314088e-01 -7.83747613e-01 1.15280426e+00
1.29592299e+00 -5.60337186e-01 -1.06516492e+00 3.02413583e-01
4.65863824e-01 -5.30766249e-01 -2.78184593e-01 -2.34589800e-01
8.01159859e-01 -1.19211221e+00 1.28048062e+00 -1.00839996e+00
1.00803792e+00 -4.00796026e-01 -5.42617440e-02 -1.36925507e+00
-6.45432413e-01 -4.50628161e-01 6.60835207e-02 8.22912097e-01
6.31216526e-01 -1.22475766e-01 7.50824451e-01 9.72328603e-01
3.26180398e-01 -5.13472140e-01 -5.46914816e-01 -6.17114365e-01
7.65273720e-02 -7.49270260e-01 2.27287918e-01 1.10687661e+00
4.98760819e-01 4.19484615e-01 -3.18578035e-01 -1.17986001e-01
5.59695661e-01 1.76266566e-01 1.04874182e+00 -1.31510234e+00
-6.59018934e-01 -1.03200662e+00 -4.53202963e-01 -6.91176891e-01
-1.04945935e-02 -8.62432599e-01 -8.77158940e-02 -1.75913680e+00
4.44645435e-01 -2.91178435e-01 -7.70606101e-02 8.43863606e-01
-4.07283306e-01 5.72903335e-01 8.15424621e-01 8.02774131e-02
-8.62017453e-01 1.21391742e-02 1.35881758e+00 -3.38767052e-01
-1.70429289e-01 -6.91618741e-01 -1.19574893e+00 9.09594595e-01
6.02398336e-01 2.93409731e-03 -2.61059284e-01 -4.38177288e-01
9.81095433e-01 -2.68197864e-01 1.04525626e+00 -1.09371054e+00
3.78711849e-01 -4.60397989e-01 5.56423545e-01 -1.94651976e-01
7.53182590e-01 -5.33496976e-01 -9.24429744e-02 4.16563094e-01
-4.32489187e-01 2.25442886e-01 4.45780426e-01 2.35893682e-01
7.56404772e-02 -8.19055736e-02 4.80688542e-01 -4.11521435e-01
-1.23769367e+00 -1.05548829e-01 -2.23709553e-01 5.26663125e-01
1.14874923e+00 -4.28819895e-01 -8.25219512e-01 -9.54544246e-01
-7.54408360e-01 4.29779232e-01 6.67065799e-01 5.08767545e-01
7.13201523e-01 -1.27718472e+00 -9.60419655e-01 -3.42151225e-01
6.92673802e-01 -3.04275393e-01 2.88544655e-01 5.92366338e-01
-7.24411845e-01 -1.12473600e-01 -6.06884897e-01 -3.58803689e-01
-1.20392084e+00 7.08817124e-01 2.13696212e-01 -5.90542518e-02
-5.86959422e-01 1.35161471e+00 4.36389744e-01 -6.22059181e-02
2.60458030e-02 -4.29488197e-02 -5.30506432e-01 1.47755325e-01
7.05287457e-01 2.03220285e-02 -5.74827850e-01 -4.80401427e-01
-6.22070253e-01 5.87213278e-01 7.11867809e-02 -1.92364708e-01
1.09461725e+00 2.26938292e-01 -1.66480348e-01 4.49991107e-01
1.41706169e-01 1.15149409e-01 -1.01968765e+00 -1.03550091e-01
-1.15223773e-01 -5.72239518e-01 -4.42394227e-01 -1.27584934e+00
-7.48617113e-01 8.33574891e-01 2.70916820e-01 3.09527993e-01
6.79483235e-01 4.30521339e-01 9.29095596e-02 5.77269256e-01
3.15847307e-01 -9.27266419e-01 5.84423244e-01 7.02583611e-01
1.24106050e+00 -1.39450693e+00 -3.61054093e-02 -1.08328901e-01
-1.35001051e+00 9.18774843e-01 1.20495903e+00 -1.40436009e-01
7.79506341e-02 6.12992756e-02 9.81868505e-02 -4.76791084e-01
-1.07581544e+00 -6.01082861e-01 4.25355643e-01 9.79250550e-01
3.76764923e-01 2.95186490e-01 -3.81786413e-02 7.26337016e-01
-8.64366055e-01 -3.48450065e-01 5.79731345e-01 4.42905933e-01
-4.03582335e-01 -4.28372204e-01 -4.67959970e-01 4.01444912e-01
7.20545426e-02 -5.74449003e-01 -9.64460135e-01 8.89134109e-01
2.32437804e-01 9.33360696e-01 2.45014414e-01 -5.37112057e-01
3.35555017e-01 -1.68323666e-01 5.61773300e-01 -5.00004530e-01
-6.87248826e-01 -5.72055042e-01 3.34822685e-01 -5.81145406e-01
-1.91493645e-01 -5.39848447e-01 -1.32148933e+00 -4.68436122e-01
1.13663279e-01 -1.53986558e-01 2.82169104e-01 8.07149291e-01
2.07389459e-01 3.97822678e-01 -3.46518546e-01 -7.46041358e-01
1.77051991e-01 -8.23884368e-01 -2.91331619e-01 7.47462988e-01
-1.66121304e-01 -9.47683990e-01 -2.22221106e-01 -1.57190934e-01] | [10.74134349822998, 1.9822757244110107] |
88797494-0dc3-4003-90d5-9f1b8b9bda9a | query2doc-query-expansion-with-large-language | 2303.07678 | null | https://arxiv.org/abs/2303.07678v1 | https://arxiv.org/pdf/2303.07678v1.pdf | Query2doc: Query Expansion with Large Language Models | This paper introduces a simple yet effective query expansion approach, denoted as query2doc, to improve both sparse and dense retrieval systems. The proposed method first generates pseudo-documents by few-shot prompting large language models (LLMs), and then expands the query with generated pseudo-documents. LLMs are trained on web-scale text corpora and are adept at knowledge memorization. The pseudo-documents from LLMs often contain highly relevant information that can aid in query disambiguation and guide the retrievers. Experimental results demonstrate that query2doc boosts the performance of BM25 by 3% to 15% on ad-hoc IR datasets, such as MS-MARCO and TREC DL, without any model fine-tuning. Furthermore, our method also benefits state-of-the-art dense retrievers in terms of both in-domain and out-of-domain results. | ['Furu Wei', 'Nan Yang', 'Liang Wang'] | 2023-03-14 | null | null | null | null | ['memorization'] | ['natural-language-processing'] | [-1.61446646e-01 -1.44585773e-01 -5.28267443e-01 -5.15226722e-02
-1.69721806e+00 -5.92105746e-01 1.14944983e+00 4.35477585e-01
-8.56985927e-01 7.95272231e-01 5.96633375e-01 1.22047193e-01
-3.40278178e-01 -7.18109727e-01 -5.98638952e-01 -2.06426129e-01
-8.26191902e-02 1.17557180e+00 6.75872147e-01 -7.56594300e-01
3.49758834e-01 8.22201371e-02 -1.26984346e+00 6.54779077e-01
1.06346834e+00 7.80794084e-01 5.58902621e-01 4.45172936e-01
-5.29603541e-01 4.85997826e-01 -5.65112770e-01 -2.59721279e-01
1.08549548e-02 1.75773893e-02 -1.01696241e+00 -4.42246497e-01
3.13547969e-01 -3.43785793e-01 -6.17032945e-01 7.79527962e-01
8.20719063e-01 5.55284441e-01 6.92776561e-01 -2.18143106e-01
-1.05103695e+00 9.14154291e-01 -4.66493547e-01 4.20684606e-01
6.24536991e-01 -3.38556856e-01 1.16820693e+00 -1.46834838e+00
9.39308703e-01 1.52429330e+00 -1.21635966e-01 3.65674943e-01
-1.04619396e+00 -6.49338782e-01 -9.48956758e-02 1.41353980e-01
-1.97087443e+00 -5.16810238e-01 3.63230258e-01 6.19270690e-02
1.42497039e+00 2.33760670e-01 2.16263875e-01 1.00060940e+00
-2.26245314e-01 1.11585546e+00 4.53374803e-01 -8.48005354e-01
3.42225254e-01 2.45914608e-01 4.40037608e-01 2.29326621e-01
3.00542772e-01 -8.16427693e-02 -6.79556906e-01 -6.36961937e-01
5.56821764e-01 7.93142319e-02 -2.03333184e-01 5.97314052e-02
-9.70651686e-01 9.58054066e-01 5.09246469e-01 6.24748230e-01
-7.85223007e-01 -1.51883453e-01 2.07432613e-01 2.14280859e-01
6.22212291e-01 1.01619923e+00 -1.78412184e-01 1.74862981e-01
-1.06996894e+00 6.39481962e-01 6.99645400e-01 1.23540246e+00
9.59370136e-01 -6.13903403e-01 -7.74861991e-01 1.34757531e+00
2.09509432e-01 1.00565541e+00 9.70464408e-01 -4.94785190e-01
6.48669302e-01 5.27373731e-01 4.27763343e-01 -8.05953145e-01
-7.50864809e-03 -5.72928667e-01 -5.37598073e-01 -1.00113022e+00
-5.11067331e-01 1.69641733e-01 -1.28252673e+00 1.40750909e+00
1.01839468e-01 -1.05466925e-01 2.69132793e-01 8.68305206e-01
8.96436155e-01 9.80316818e-01 3.04306716e-01 -2.32560277e-01
1.42183590e+00 -1.06014395e+00 -6.58725739e-01 -6.90250039e-01
7.11623728e-01 -9.99232531e-01 1.19229484e+00 -8.43554661e-02
-1.04931509e+00 -4.35532033e-01 -6.74357772e-01 -2.48694852e-01
-6.07723355e-01 4.63644527e-02 5.94058633e-01 5.25386855e-02
-1.02736306e+00 1.44045654e-04 -5.53028345e-01 -6.34922028e-01
1.01852037e-01 1.26806730e-02 -3.79663482e-02 -6.93593502e-01
-1.81121457e+00 7.69905210e-01 7.56149888e-01 -5.39567053e-01
-1.08412266e+00 -8.34826589e-01 -5.72727919e-01 4.14014071e-01
5.28304279e-01 -6.34887278e-01 1.45900381e+00 -2.69585215e-02
-8.98175001e-01 7.09123373e-01 -4.89477128e-01 -6.72456682e-01
-1.46699965e-01 -6.60232842e-01 -6.88045621e-01 3.65560681e-01
3.93760204e-01 1.01931024e+00 5.93261778e-01 -1.10732305e+00
-6.10725760e-01 -2.55514860e-01 3.88132967e-02 5.82768381e-01
-4.69336778e-01 1.29078805e-01 -1.39562881e+00 -6.61708355e-01
3.98822129e-02 -7.75747299e-01 -3.04672927e-01 -8.24388206e-01
-3.93238753e-01 -7.95295656e-01 5.08691192e-01 -3.19375783e-01
1.81040943e+00 -1.87059987e+00 -2.11935580e-01 4.05949622e-01
4.35368195e-02 6.99135542e-01 -6.56278431e-01 8.67477834e-01
5.48070967e-01 1.39459983e-01 3.29362512e-01 -2.18284100e-01
1.50031686e-01 -5.43048084e-02 -9.41343188e-01 -2.99392372e-01
-2.47560114e-01 1.27761757e+00 -1.08695447e+00 -7.93933570e-01
-1.65933564e-01 4.93192345e-01 -3.15124393e-01 2.02891991e-01
-6.44678175e-01 -1.75073028e-01 -1.02800834e+00 7.18282104e-01
3.93080771e-01 -5.46390414e-01 -3.64528000e-02 1.45294785e-01
3.80312502e-01 5.60233176e-01 -6.16197944e-01 2.24886441e+00
-4.42058414e-01 2.20644698e-01 -4.13403183e-01 -4.99431878e-01
8.03523719e-01 2.23961756e-01 3.22877795e-01 -1.40347004e+00
-3.82234484e-01 3.41898352e-01 -6.99787140e-01 -3.43204200e-01
1.35054493e+00 2.98548222e-01 -2.32505098e-01 6.56679928e-01
1.09722257e-01 -9.28859413e-02 7.44900763e-01 1.07389212e+00
1.11278105e+00 -4.19554412e-01 1.65246427e-01 -5.50551452e-02
2.89433509e-01 2.93488741e-01 -4.19962453e-03 1.42642295e+00
5.61020792e-01 2.06457362e-01 -3.42041224e-01 1.21300861e-01
-7.78988659e-01 -1.00318408e+00 -5.41547388e-02 1.70858026e+00
2.43126124e-01 -7.85615385e-01 -2.75083303e-01 -5.01564324e-01
3.48238796e-01 8.72379422e-01 -3.12291443e-01 -4.75872368e-01
-3.63259763e-01 -5.37551939e-01 5.81234574e-01 3.68281484e-01
2.80914843e-01 -1.00229931e+00 8.13252330e-02 4.42202240e-01
-3.83931547e-01 -1.05354464e+00 -7.00113237e-01 -1.29386634e-01
-7.97640204e-01 -6.39766037e-01 -1.29231203e+00 -7.74486125e-01
5.22279203e-01 8.52975488e-01 1.54219568e+00 -4.55760844e-02
-3.50109667e-01 5.44968069e-01 -7.50946462e-01 -3.25015843e-01
-9.27930623e-02 6.15404069e-01 7.15039372e-02 -4.95999008e-01
1.01976478e+00 -2.16763243e-01 -8.25250268e-01 2.70337641e-01
-1.33831692e+00 -4.22497749e-01 9.45953846e-01 7.39755630e-01
8.13989103e-01 -2.09906682e-01 8.02345216e-01 -9.24001455e-01
1.39461637e+00 -6.02409720e-01 -5.05889833e-01 9.32599664e-01
-1.21609080e+00 3.50606680e-01 -4.92582992e-02 -5.26347220e-01
-1.24329925e+00 -3.52854580e-01 3.26214463e-01 -2.84038275e-01
3.64875823e-01 1.09906662e+00 6.15992367e-01 1.14765197e-01
1.27315390e+00 5.28229713e-01 -6.45673037e-01 -8.73583734e-01
6.82049572e-01 9.39194679e-01 3.48102093e-01 -7.09175646e-01
6.49743855e-01 2.19028994e-01 -6.00806713e-01 -6.98981047e-01
-1.10589099e+00 -1.40297663e+00 -2.87165970e-01 2.22472280e-01
3.22854519e-01 -1.41055477e+00 2.76929792e-02 -1.69233561e-01
-1.07821703e+00 4.65182215e-02 -1.65482968e-01 4.80535775e-01
1.65623240e-02 2.33859465e-01 -7.47038186e-01 -7.14559972e-01
-1.07037485e+00 -7.25645483e-01 1.55890024e+00 3.88709694e-01
-7.36384541e-02 -7.84885287e-01 4.75441277e-01 2.42737278e-01
7.54061878e-01 -8.95307779e-01 9.04787004e-01 -1.13179803e+00
-7.33946562e-01 -6.66139424e-01 -2.89055854e-01 -1.76463991e-01
-1.41474560e-01 -8.09364974e-01 -8.76611531e-01 -6.07555687e-01
-7.13269830e-01 -9.83283043e-01 1.32746637e+00 1.35126382e-01
7.72494674e-01 -2.93624371e-01 -8.62432003e-01 -6.06369153e-02
1.35193968e+00 1.14617839e-01 7.59832621e-01 2.18041390e-01
1.57908082e-01 2.92052597e-01 1.02796924e+00 4.64516908e-01
2.47159675e-01 8.93125594e-01 -2.96407282e-01 7.52098113e-02
-1.87321201e-01 -6.10952020e-01 8.12161937e-02 6.65764630e-01
2.52587497e-01 -3.54188442e-01 -8.94361973e-01 7.70485163e-01
-1.73398387e+00 -8.05382490e-01 6.46312892e-01 2.16147709e+00
1.42669713e+00 -6.53095320e-02 -2.96288341e-01 -6.51835680e-01
5.29870450e-01 2.26423025e-01 -5.24947107e-01 3.04129630e-01
-2.29732424e-01 6.18422210e-01 3.06479037e-01 7.33006239e-01
-7.90778339e-01 1.65329230e+00 6.09292316e+00 1.42003381e+00
-7.64098465e-01 -1.06989823e-01 2.13403016e-01 -4.54549879e-01
-4.86314476e-01 5.30360304e-02 -1.32040644e+00 8.23080987e-02
8.79511952e-01 -9.42930102e-01 4.53084022e-01 9.89885390e-01
-2.36523494e-01 -2.27888182e-01 -8.75775754e-01 9.03583527e-01
3.35179538e-01 -1.34898913e+00 7.24676549e-01 -1.42269388e-01
8.31375718e-01 4.47547108e-01 -4.74923998e-02 1.16559196e+00
5.30868590e-01 -7.66603053e-01 8.46057683e-02 6.31365716e-01
9.12956715e-01 -5.34966290e-01 6.34850144e-01 4.05555755e-01
-1.05196118e+00 1.28503993e-01 -9.53018606e-01 6.68097436e-01
1.20850809e-01 5.30690074e-01 -1.36071098e+00 4.73859996e-01
4.68810529e-01 3.61837149e-01 -8.50385606e-01 1.11333001e+00
-8.77140015e-02 4.42578554e-01 -3.89020294e-01 -3.52022290e-01
5.37712097e-01 3.15041840e-01 3.89089257e-01 1.39922667e+00
3.20065469e-01 3.38363826e-01 3.36635590e-01 7.10093975e-01
-3.90988886e-01 4.46836561e-01 -3.69753122e-01 -5.34554303e-01
1.06752837e+00 1.10861683e+00 -1.52359098e-01 -9.37494755e-01
-3.41985486e-02 9.11970794e-01 2.78756082e-01 9.65710402e-01
-1.82628617e-01 -7.51351893e-01 1.69287860e-01 -1.35222552e-02
1.82447270e-01 1.05562747e-01 4.39827442e-01 -1.29142761e+00
3.65125388e-02 -8.31048608e-01 7.80022383e-01 -9.16308343e-01
-1.20784414e+00 7.60552347e-01 2.75915831e-01 -1.01465976e+00
-9.19001400e-01 8.51130784e-02 7.27812722e-02 1.12917876e+00
-1.69578302e+00 -9.69340444e-01 -5.91589585e-02 5.88897228e-01
6.47300243e-01 -3.90638828e-01 1.16447294e+00 5.49391747e-01
5.16556129e-02 5.59020936e-01 5.30052245e-01 6.54122457e-02
1.10012865e+00 -8.93080890e-01 3.53958756e-01 4.96333420e-01
6.14615679e-01 1.29001987e+00 4.25803930e-01 -8.77298594e-01
-1.43242991e+00 -1.05698359e+00 1.26744568e+00 -3.69979143e-01
5.59415698e-01 -1.86809734e-01 -1.00507641e+00 3.94047529e-01
1.78089708e-01 -2.58893102e-01 6.20544076e-01 5.96263289e-01
-5.55191398e-01 -2.08050981e-01 -7.19990075e-01 4.87249464e-01
7.28435695e-01 -1.07398999e+00 -1.02102649e+00 8.13749552e-01
1.14470756e+00 -3.29480618e-01 -6.03237033e-01 3.17977816e-01
3.43616903e-01 -1.06037259e-01 1.55060959e+00 -9.17062342e-01
-3.74123119e-02 1.88035831e-01 -3.01104099e-01 -1.36680865e+00
-4.61311728e-01 -4.17662531e-01 -4.81755465e-01 1.03872001e+00
6.39245391e-01 -2.25833446e-01 5.48480451e-01 6.97022140e-01
1.97074115e-02 -5.39194703e-01 -5.23496509e-01 -8.00062597e-01
1.87842757e-03 -2.78291196e-01 6.04338944e-01 7.00675726e-01
2.83236682e-01 8.34048212e-01 -6.57990798e-02 -1.04316734e-01
2.79940665e-01 3.04688394e-01 6.14386976e-01 -1.13605058e+00
-1.09655194e-01 -2.09615991e-01 1.90144956e-01 -1.72946656e+00
9.50148609e-03 -1.09596109e+00 3.43239792e-02 -1.58708870e+00
5.59375167e-01 -5.28534770e-01 -6.21718943e-01 4.13534224e-01
-4.82028186e-01 -2.48941220e-02 -1.83073878e-01 7.00988710e-01
-1.47565639e+00 7.22629786e-01 1.08037210e+00 -4.51938689e-01
-4.61597085e-01 -2.80815631e-01 -8.32495391e-01 6.68286160e-02
1.98975623e-01 -7.16323853e-01 -7.31215656e-01 -7.42332518e-01
2.80257344e-01 1.62229314e-01 -7.75841475e-02 -5.49796164e-01
7.69548476e-01 -5.46964398e-03 2.81825632e-01 -8.62210155e-01
4.68963712e-01 -4.48839247e-01 -4.46206182e-01 1.36137381e-01
-1.01961386e+00 -4.58360501e-02 1.09346561e-01 8.80499005e-01
-5.54069400e-01 -3.05778980e-01 2.13174507e-01 -3.76465976e-01
-9.60802674e-01 3.98046046e-01 -5.27503490e-02 5.26330769e-01
2.85170853e-01 6.56756341e-01 -6.05784714e-01 -6.62261486e-01
-3.67211103e-01 6.43567026e-01 1.16263013e-02 6.06924951e-01
7.66653478e-01 -1.56517589e+00 -8.27698410e-01 -8.88570473e-02
9.07781959e-01 -2.13766117e-02 2.20845729e-01 2.66828895e-01
1.48481429e-02 1.58539581e+00 6.18926227e-01 -4.45738435e-01
-9.47876453e-01 5.90592265e-01 -2.57356673e-01 -8.51144552e-01
-3.84575993e-01 1.10147417e+00 7.49380589e-02 -1.51248202e-01
5.00519037e-01 5.77156097e-02 -3.08958918e-01 1.77200064e-01
1.28723145e+00 -9.32167098e-02 2.50046611e-01 -1.52755901e-01
-2.03693658e-01 2.45209768e-01 -9.18590069e-01 -5.58719635e-01
1.14105701e+00 -3.20354909e-01 -9.49224159e-02 -5.32243475e-02
1.25600028e+00 -1.04746126e-01 -3.01060408e-01 -1.29858446e+00
4.65386242e-01 -4.37054038e-01 3.85066330e-01 -1.23970163e+00
-4.87466872e-01 4.87231344e-01 4.78056252e-01 -2.92826705e-02
1.11030364e+00 4.83302295e-01 1.14617097e+00 1.38923752e+00
7.46544123e-01 -1.24379766e+00 3.40054750e-01 8.12252581e-01
1.00161242e+00 -1.24937570e+00 -3.68359387e-02 9.79059115e-02
-5.58712721e-01 5.74946344e-01 4.90194410e-01 2.26443991e-01
4.56251800e-01 -3.19613427e-01 6.76842183e-02 -4.84561145e-01
-1.06251836e+00 -5.46609223e-01 1.07358110e+00 2.24959567e-01
4.77474242e-01 -1.76100761e-01 -4.84757811e-01 5.76828301e-01
1.25587523e-01 1.61825061e-01 -4.11223948e-01 1.01278722e+00
-1.02873373e+00 -1.07073045e+00 -1.87258095e-01 4.81587291e-01
-3.43729943e-01 -8.01008880e-01 -5.70014656e-01 5.40031374e-01
-8.36731136e-01 9.72535193e-01 -1.20730795e-01 -2.99125135e-01
1.10540651e-01 3.65408659e-01 1.74174517e-01 -9.68467057e-01
-1.86005861e-01 4.85662103e-01 2.11253271e-01 -5.35854042e-01
-1.01522014e-01 5.09680472e-02 -1.16667986e+00 6.76515028e-02
-6.12683475e-01 9.72465873e-01 4.56917405e-01 7.86952555e-01
8.45770419e-01 7.51127973e-02 4.56114024e-01 -3.02624226e-01
-9.47408855e-01 -1.69969523e+00 -5.89129508e-01 4.62988406e-01
-5.07901013e-02 -4.84906912e-01 -5.00927903e-02 -3.05696845e-01] | [11.518173217773438, 7.665882110595703] |
f3c3b3b8-0290-4517-951e-40d145bb5164 | dynamic-character-graph-via-online-face | 2007.14913 | null | https://arxiv.org/abs/2007.14913v1 | https://arxiv.org/pdf/2007.14913v1.pdf | Dynamic Character Graph via Online Face Clustering for Movie Analysis | An effective approach to automated movie content analysis involves building a network (graph) of its characters. Existing work usually builds a static character graph to summarize the content using metadata, scripts or manual annotations. We propose an unsupervised approach to building a dynamic character graph that captures the temporal evolution of character interaction. We refer to this as the character interaction graph(CIG). Our approach has two components:(i) an online face clustering algorithm that discovers the characters in the video stream as they appear, and (ii) simultaneous creation of a CIG using the temporal dynamics of the resulting clusters. We demonstrate the usefulness of the CIG for two movie analysis tasks: narrative structure (acts) segmentation, and major character retrieval. Our evaluation on full-length movies containing more than 5000 face tracks shows that the proposed approach achieves superior performance for both the tasks. | ['Prakhar Kulshreshtha', 'Tanaya Guha'] | 2020-07-29 | null | null | null | null | ['face-clustering'] | ['computer-vision'] | [ 8.95754546e-02 -1.91516742e-01 -1.40600428e-01 -3.26166064e-01
-3.92846107e-01 -1.06777132e+00 8.89606416e-01 3.53306532e-01
-4.46665147e-03 1.02887705e-01 4.73275095e-01 1.82195693e-01
-9.05183479e-02 -6.08934999e-01 -2.93841422e-01 -4.93337035e-01
-3.02342236e-01 3.67394209e-01 5.82364082e-01 1.95410520e-01
5.27079821e-01 5.22537768e-01 -1.44919324e+00 5.05905330e-01
3.69819045e-01 9.18002665e-01 -6.03086464e-02 9.50798869e-01
-4.17530149e-01 1.42402267e+00 -5.82310379e-01 -8.51118982e-01
-1.10561503e-02 -8.29923391e-01 -9.20717299e-01 7.47215509e-01
3.55334967e-01 -8.51243511e-02 -4.51894969e-01 1.01348662e+00
-1.13376707e-01 2.57892638e-01 7.61234760e-01 -1.34028196e+00
-2.68924743e-01 9.15777206e-01 -5.07027686e-01 3.94567817e-01
7.41795421e-01 -3.21878105e-01 9.86227393e-01 -6.12115145e-01
1.18450189e+00 1.07020402e+00 7.31628001e-01 2.81036675e-01
-9.89340484e-01 -2.05244631e-01 1.47856250e-01 1.40834123e-01
-1.36684644e+00 -6.08014584e-01 9.08145130e-01 -9.12399650e-01
6.31334066e-01 1.56566396e-01 8.05170238e-01 7.84117043e-01
-2.05556706e-01 6.91035271e-01 6.92804873e-01 -4.35980409e-01
1.53384805e-01 4.52105962e-02 6.13406181e-01 8.65115881e-01
-3.07914883e-01 -7.89471507e-01 -8.00157249e-01 -4.55922067e-01
7.33598173e-01 -3.70148420e-01 6.58382624e-02 -1.12681687e-02
-6.73073113e-01 5.73214352e-01 -4.92429405e-01 2.49912053e-01
-3.75558913e-01 1.64946720e-01 7.40332901e-01 2.60735273e-01
5.26277244e-01 1.11075982e-01 2.83083200e-01 -6.22061491e-01
-1.13469422e+00 1.63087755e-01 1.10370982e+00 1.09685075e+00
4.38864142e-01 -1.54520065e-01 -1.26173317e-01 8.82232368e-01
1.28008544e-01 -3.40882957e-01 1.78826392e-01 -1.18406451e+00
1.89861655e-01 5.83040774e-01 -9.21330079e-02 -1.41821420e+00
1.45651465e-02 2.40130261e-01 -2.71341383e-01 -3.25958401e-01
3.58983755e-01 -1.55410469e-01 -5.25807738e-01 1.63077283e+00
4.35451180e-01 5.03630221e-01 -4.35129225e-01 2.74631083e-01
9.32633996e-01 6.09715700e-01 -5.58060743e-02 -5.83513737e-01
1.28077900e+00 -8.30243528e-01 -1.04835045e+00 4.94908541e-01
1.64052144e-01 -6.39926791e-01 5.52445531e-01 4.67179149e-01
-1.35262346e+00 -3.77084345e-01 -7.59333551e-01 2.88905472e-01
2.05069464e-02 1.26922071e-01 6.72048032e-01 6.55863881e-01
-1.21736538e+00 8.69907856e-01 -7.60861576e-01 -5.83885729e-01
4.15047199e-01 2.69937456e-01 -3.94375622e-01 2.60656118e-01
-5.10545075e-01 1.00703537e-02 2.18593687e-01 -3.56595576e-01
-9.62552607e-01 -3.23087722e-01 -6.30090952e-01 4.12925072e-02
4.50412989e-01 -3.74903977e-02 1.06305707e+00 -1.44899690e+00
-1.60251093e+00 1.05717599e+00 -2.50540107e-01 -1.69258595e-01
2.58514374e-01 -1.87416583e-01 -3.85672003e-01 8.14245939e-01
-1.89098567e-01 2.10617587e-01 1.06173909e+00 -1.32747924e+00
-7.92049885e-01 -1.89711243e-01 -2.29902416e-02 2.18040600e-01
-7.88340330e-01 7.13624835e-01 -1.40186465e+00 -8.62918437e-01
-1.58354605e-03 -9.17301774e-01 2.12304547e-01 -3.75129282e-01
-4.08510804e-01 -5.11093020e-01 8.00525546e-01 -8.93714786e-01
2.02334642e+00 -2.37403560e+00 4.06476378e-01 3.96341264e-01
4.63854522e-01 -2.46327609e-01 1.48357555e-01 8.62502694e-01
3.29102203e-02 1.47271931e-01 2.46842336e-02 -6.62510037e-01
-3.30755949e-01 -3.41455415e-02 -1.17229462e-01 5.29627740e-01
-2.99985379e-01 4.77140397e-01 -7.81737804e-01 -9.30071115e-01
-2.45812535e-01 1.77793235e-01 -4.87623245e-01 2.74648666e-01
-2.91682243e-01 1.46160305e-01 -2.37370700e-01 6.50595605e-01
1.85059920e-01 -2.04403937e-01 8.19580615e-01 -4.84388061e-02
-2.61836112e-01 -6.05490059e-02 -1.25937092e+00 1.42997324e+00
5.65427423e-01 1.07775593e+00 2.06260815e-01 -7.22957850e-01
7.50649869e-01 5.15717924e-01 1.05614018e+00 1.50248408e-01
2.71597922e-01 -4.28666592e-01 -2.33196393e-01 -6.87614322e-01
5.80401063e-01 4.39989060e-01 -1.20816603e-01 7.33891845e-01
4.36315030e-01 3.74230593e-01 7.31183648e-01 8.54445696e-01
1.39515710e+00 2.14710906e-02 9.17443186e-02 -2.45218113e-01
5.75484335e-01 1.75660655e-01 5.48429370e-01 5.71178913e-01
-1.95850372e-01 4.44207013e-01 1.16463971e+00 -2.58517236e-01
-1.01515090e+00 -9.06214118e-01 1.31218642e-01 1.15478611e+00
-3.76007855e-02 -1.26868939e+00 -1.37557840e+00 -5.85281730e-01
-2.44033188e-01 1.38455525e-01 -6.64019883e-01 3.08943659e-01
-6.94219053e-01 -4.59817767e-01 5.50818443e-01 2.54042000e-01
2.42599100e-01 -1.14392769e+00 -8.99248049e-02 1.12364829e-01
-2.38534629e-01 -1.22814381e+00 -6.65158451e-01 -5.27821541e-01
-7.31616914e-01 -1.20242703e+00 -6.81148618e-02 -9.31337237e-01
7.50640512e-01 1.74944237e-01 1.05712128e+00 2.79521883e-01
-2.35183835e-01 9.73796725e-01 -7.25843608e-01 1.77943021e-01
-5.50460517e-01 -2.21533164e-01 1.07061960e-01 5.87832272e-01
2.29718953e-01 -7.16371953e-01 -2.24268615e-01 2.59183854e-01
-8.28684926e-01 -1.78641737e-01 -1.23096868e-01 2.47767285e-01
4.56228375e-01 4.82563257e-01 2.52721995e-01 -1.34424114e+00
8.59393418e-01 -7.59582222e-01 -4.00742799e-01 3.39624494e-01
-4.21977490e-01 -5.30095160e-01 4.82484609e-01 -4.81677324e-01
-1.18424487e+00 4.57985193e-01 2.77364403e-01 -5.94921649e-01
-1.27717763e-01 5.32376349e-01 5.37999421e-02 8.23417529e-02
3.56399775e-01 2.20403180e-01 -5.82242571e-02 -6.61876738e-01
3.17013949e-01 6.79403245e-01 9.46520209e-01 -6.15547061e-01
6.44469321e-01 6.67319596e-01 -2.11570427e-01 -1.10250092e+00
-4.52955902e-01 -7.62401104e-01 -1.03421664e+00 -1.05206096e+00
1.20928538e+00 -7.46837199e-01 -1.00317252e+00 6.17884040e-01
-8.62808585e-01 -1.31248057e-01 -1.24237814e-03 1.47507116e-01
-5.62331855e-01 9.02876377e-01 -1.17957401e+00 -9.00903344e-01
-1.08896427e-01 -5.57116807e-01 5.56370318e-01 2.68042624e-01
-5.98691761e-01 -9.18053508e-01 4.05555099e-01 3.86926711e-01
-3.62596333e-01 5.05260706e-01 9.31491971e-01 -5.69995999e-01
-3.44499230e-01 -4.38547283e-01 1.62098229e-01 -5.43085486e-02
2.27530733e-01 9.60682750e-01 -6.41980529e-01 -1.11492440e-01
-1.45761356e-01 -9.00153890e-02 5.28209448e-01 4.13928717e-01
1.17633855e+00 -2.88880080e-01 -3.93911153e-01 5.30276597e-01
1.25767243e+00 5.70144415e-01 7.16217577e-01 -1.70075707e-02
8.89459789e-01 9.12237704e-01 4.04551208e-01 8.42941642e-01
2.63589263e-01 6.63147509e-01 2.08572727e-02 5.05890906e-01
-5.61993159e-02 -3.34184051e-01 5.49643397e-01 1.29059899e+00
-4.98023272e-01 -5.40904760e-01 -8.38135839e-01 5.65006733e-01
-2.18647885e+00 -1.40958285e+00 -3.98560971e-01 1.88121665e+00
6.41034424e-01 1.84858009e-01 7.33270168e-01 1.28941134e-01
1.00269639e+00 9.16409791e-02 -1.48115590e-01 -4.69234526e-01
3.70099247e-02 -1.62714958e-01 -7.78764263e-02 1.97411254e-01
-1.24100590e+00 1.10278809e+00 7.30451250e+00 8.45106781e-01
-4.95548427e-01 5.46863116e-02 5.27057767e-01 -7.38914236e-02
5.26402593e-02 9.74370316e-02 -6.68654144e-01 5.71173370e-01
8.52107346e-01 -1.70315057e-01 4.16667134e-01 8.53383064e-01
1.47305503e-01 -3.32345277e-01 -9.88248765e-01 1.07089901e+00
4.87746656e-01 -1.52524304e+00 3.60409431e-02 1.64828617e-02
6.74425304e-01 -5.54394782e-01 -3.29689920e-01 -2.70869523e-01
4.68442500e-01 -7.36008883e-01 9.19688344e-01 6.69674337e-01
8.66664231e-01 -8.60396624e-01 1.31490737e-01 1.34066880e-01
-1.66437864e+00 -8.63889828e-02 1.04586385e-01 1.91333100e-01
3.56282502e-01 1.00469328e-01 -4.71113950e-01 2.59636074e-01
6.81990921e-01 1.11536443e+00 -7.37833381e-01 8.96575391e-01
-3.09319533e-02 8.54548573e-01 8.02909676e-03 7.71471485e-02
-8.45571421e-03 -7.28968740e-01 7.11251020e-01 1.49967504e+00
-1.25802353e-01 4.43909764e-01 1.80465966e-01 3.88904989e-01
-1.10238254e-01 3.09976667e-01 -4.46993560e-01 -6.71072423e-01
5.71304023e-01 1.57126498e+00 -1.46469665e+00 -4.12836194e-01
-5.73544562e-01 1.15641212e+00 3.86906981e-01 3.86086367e-02
-7.69119084e-01 -1.21358007e-01 6.23266220e-01 2.87177175e-01
3.48224878e-01 -5.30181050e-01 1.53524652e-01 -1.00729680e+00
-2.57334888e-01 -8.45975697e-01 8.63345087e-01 -6.39240563e-01
-1.25559390e+00 8.07805836e-01 1.79856732e-01 -1.09990895e+00
-2.40046024e-01 -1.21367805e-01 -8.92833054e-01 8.58679861e-02
-3.69119316e-01 -9.36207294e-01 -3.79964411e-01 9.54145432e-01
8.32080960e-01 -3.51056546e-01 5.83119094e-01 4.25153613e-01
-9.79643703e-01 5.30736983e-01 -8.96650925e-03 5.68877995e-01
3.77956063e-01 -1.18460870e+00 2.17945486e-01 9.78162348e-01
4.80191946e-01 4.77999121e-01 4.96870637e-01 -1.13648319e+00
-1.39323151e+00 -6.61309719e-01 7.08576500e-01 -6.55763328e-01
8.04449022e-01 -6.88287735e-01 -5.59068501e-01 7.99919963e-01
3.71912122e-01 -5.98446906e-01 1.00113857e+00 -7.53504783e-02
-2.23062590e-01 2.80982196e-01 -8.27539265e-01 5.17255068e-01
1.24320316e+00 -5.70460856e-01 -3.02558601e-01 4.37430471e-01
3.53386551e-01 5.94531111e-02 -8.92255187e-01 -2.63681263e-01
6.35127366e-01 -8.75467658e-01 6.12423778e-01 -7.81950653e-01
7.09881842e-01 -1.14636980e-01 -6.90970430e-03 -5.78866005e-01
-5.28619945e-01 -1.21636713e+00 -4.19670820e-01 1.83351922e+00
-7.10093044e-03 2.07038388e-01 9.12445605e-01 6.07882857e-01
4.52895686e-02 -3.94816190e-01 -4.54395622e-01 -4.82043624e-01
-6.34449124e-01 -3.69400352e-01 1.06937654e-01 1.25698113e+00
2.94487417e-01 3.50461185e-01 -4.90398258e-01 -9.61197466e-02
6.82625115e-01 -2.83060856e-02 7.02923715e-01 -1.35701275e+00
-3.48437279e-01 -4.66799110e-01 -4.98145998e-01 -7.24673569e-01
2.49380767e-01 -7.27723658e-01 -1.92091465e-01 -1.13764918e+00
6.77714288e-01 -1.51717961e-01 1.84746817e-01 3.15818638e-02
2.17087194e-01 1.49811327e-01 1.68767497e-01 6.86534941e-01
-9.59876835e-01 -1.48042440e-02 4.83689874e-01 3.65162045e-01
-3.09670776e-01 -1.15996622e-01 -2.59900302e-01 1.05615270e+00
4.49488997e-01 -5.63702285e-01 -4.32877034e-01 -1.60574466e-02
3.00797254e-01 2.47916237e-01 -2.39962459e-01 -8.91058683e-01
4.94859219e-01 -8.87085944e-02 1.48470461e-01 -6.75787330e-01
2.94570774e-01 -4.89573449e-01 6.29855454e-01 4.56912443e-02
-3.16241801e-01 3.79683197e-01 -1.15900844e-01 7.34715521e-01
-3.78331214e-01 -4.71536964e-01 6.18818939e-01 -2.96329141e-01
-6.76360786e-01 3.69580716e-01 -9.79997337e-01 -4.42179032e-02
1.19243455e+00 -4.61839259e-01 -8.18281844e-02 -7.74941206e-01
-1.05291581e+00 1.96819417e-02 5.60246885e-01 3.41153145e-01
4.70447302e-01 -1.36497712e+00 -4.40239489e-01 -2.03551084e-01
-2.62142688e-01 -7.06865191e-01 1.88474178e-01 5.78766346e-01
-8.69836509e-01 -8.65028128e-02 -1.60975173e-01 -3.72749269e-01
-2.07884979e+00 5.42027175e-01 -4.71005887e-02 -9.68447626e-02
-6.39859080e-01 1.01433253e+00 -9.81701016e-02 4.12453771e-01
4.28392649e-01 4.47114736e-01 -9.44289505e-01 7.75481641e-01
5.83478928e-01 6.22566283e-01 -4.70118076e-01 -1.08529305e+00
-1.45543754e-01 6.41065300e-01 2.55971737e-02 -3.56619298e-01
1.48731208e+00 -3.47554475e-01 -6.50106430e-01 6.39753103e-01
1.00633216e+00 2.48445630e-01 -1.22502697e+00 -1.79996833e-01
4.24327433e-01 -5.69250643e-01 -2.97754496e-01 4.29097749e-02
-1.09891200e+00 2.13864133e-01 -2.99966428e-02 7.15938330e-01
1.21678412e+00 1.98034957e-01 7.42249966e-01 1.26283720e-01
7.94940218e-02 -1.35368311e+00 5.12855291e-01 4.06702459e-01
4.56822306e-01 -6.29177034e-01 1.21331021e-01 -8.94721270e-01
-8.15873802e-01 1.35496414e+00 3.94179761e-01 -1.61592141e-01
9.22779202e-01 3.81740987e-01 -2.96998650e-01 -7.82025397e-01
-8.74203146e-01 -1.35710714e-02 2.87287623e-01 2.09587142e-01
2.38401532e-01 -1.45854786e-01 -3.71352881e-01 7.24590242e-01
5.92187559e-03 -3.21674794e-01 7.69032001e-01 1.10592651e+00
-3.60962451e-01 -1.10435843e+00 -1.22781798e-01 4.98987079e-01
-9.33920383e-01 1.26993448e-01 -1.19516671e+00 4.46372986e-01
1.45729510e-02 1.04305422e+00 4.51126099e-01 -5.96024632e-01
5.43748811e-02 3.27142149e-01 5.40590167e-01 -8.22040319e-01
-8.95091951e-01 4.57129091e-01 4.17663395e-01 -5.84856629e-01
-6.68333650e-01 -1.23040771e+00 -1.04495692e+00 -5.15558302e-01
-5.97238727e-02 9.42503661e-02 4.31709021e-01 6.43993258e-01
1.21040963e-01 1.97103232e-01 9.78420317e-01 -5.05561352e-01
4.29348230e-01 -7.44935691e-01 -7.76416123e-01 8.38750124e-01
-3.13497216e-01 -4.80274141e-01 -2.65365928e-01 8.76714170e-01] | [10.583479881286621, 0.6939753890037537] |
3d8051bd-5891-4bfa-86bd-c103bbb1b50a | osvidcap-a-framework-for-the-simultaneous | null | null | https://ieeexplore.ieee.org/abstract/document/9552885 | https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9552885 | OSVidCap: A Framework for the Simultaneous Recognition and Description of Concurrent Actions in Videos in an Open-Set Scenario | Automatically understanding and describing the visual content of videos in natural language is a challenging task in computer vision. Existing approaches are often designed to describe single events in a closed-set setting. However, in real-world scenarios, concurrent activities and previously unseen actions may appear in a video. This work presents the OSVidCap, a novel open-set video captioning framework that recognizes and describes, in natural language, concurrent known actions and deal with unknown ones. The OSVidCap is based on the encoder-decoder framework and uses a detection-and-tracking-object-based mechanism followed by a background blurring method to focus on specific targets in a video. Additionally, we employ the TI3D Network with the Extreme Value Machine (EVM), which learns representations and recognizes unknown actions. We evaluate the proposed approach on the benchmark ActivityNet Captions dataset. Also, an enhanced version of the LIRIS human activity dataset was proposed by providing descriptions for each action. We also provide spatial, temporal, and caption annotations for existing unlabeled actions in the dataset - considered unknown actions in our experiments. Experimental results showed our method’s effectiveness in recognizing and describing concurrent actions in natural language and the strong ability to deal with detected unknown activities. Based on these results, we believe that the proposed approach can be potentially helpful for many real-world applications, including human behavior analysis, safety monitoring, and surveillance. | ['Heitor Silvério Lopes', 'André Eugênio Lazzaretti', 'Matheus Gutoski', 'Andrei De Souza Inácio'] | 2021-09-29 | null | null | null | ieee-access-2021-9 | ['open-set-video-captioning'] | ['computer-vision'] | [ 5.31748354e-01 -9.62441489e-02 -2.24001840e-01 -2.48547763e-01
-5.26674688e-01 -4.53606695e-01 8.57392848e-01 -2.32595325e-01
-3.01355869e-01 6.96969092e-01 4.52663243e-01 2.63584740e-02
2.37796292e-01 -3.08655709e-01 -1.00705099e+00 -7.13935196e-01
-3.26533139e-01 2.89356768e-01 6.07526302e-01 6.94689453e-02
-5.75292110e-02 3.75679791e-01 -1.62158334e+00 6.93525791e-01
1.15354955e-01 1.00472629e+00 3.48946184e-01 6.82724178e-01
9.32040066e-02 1.59284365e+00 -5.55559218e-01 -2.73075849e-01
2.61615276e-01 -5.05423605e-01 -5.63797772e-01 6.37838662e-01
5.70031881e-01 -5.82190812e-01 -6.90365314e-01 7.40035832e-01
1.09846629e-01 3.49203467e-01 5.28439403e-01 -1.73893070e+00
-5.98179221e-01 2.72004277e-01 -6.31434739e-01 5.17548442e-01
6.69603288e-01 4.80735868e-01 5.92347682e-01 -6.22102320e-01
5.08201480e-01 1.18104553e+00 2.98071623e-01 8.52285147e-01
-6.95797980e-01 -5.81364989e-01 5.03841817e-01 5.83703637e-01
-1.27546608e+00 -5.09313405e-01 6.34810507e-01 -5.82645297e-01
9.10661995e-01 1.36264071e-01 5.49132228e-01 1.60494149e+00
1.32706746e-01 1.07075608e+00 6.48580194e-01 -8.17448944e-02
1.86933547e-01 -5.00255171e-03 1.37450348e-03 6.21046722e-01
1.17739499e-01 1.83751076e-01 -4.70599502e-01 -1.92142185e-02
7.98603654e-01 3.36775661e-01 -4.20009702e-01 -5.27344286e-01
-1.57859111e+00 4.00434613e-01 2.13742420e-01 1.48112550e-01
-5.16480803e-01 4.68445182e-01 6.48806095e-01 -6.66012838e-02
3.60314667e-01 -5.53417504e-02 -3.40199202e-01 -2.35073313e-01
-6.96203768e-01 2.84294388e-03 5.96764743e-01 1.22157288e+00
4.51807797e-01 -3.83238718e-02 -7.46396720e-01 4.14970636e-01
9.52160805e-02 5.69764435e-01 3.51204783e-01 -9.54525828e-01
6.52625680e-01 3.67662162e-01 4.71016049e-01 -1.00783074e+00
-1.56497896e-01 -3.18364114e-01 -6.97276533e-01 -4.39094119e-02
1.76140636e-01 2.07914934e-02 -9.05263960e-01 1.78250563e+00
2.48146191e-01 9.61623371e-01 3.90525937e-01 1.13811719e+00
7.66646683e-01 9.47531044e-01 3.46629173e-01 -3.69646966e-01
1.36699307e+00 -1.16165757e+00 -1.00451589e+00 -4.50403631e-01
4.06188637e-01 -3.77252251e-01 7.19415188e-01 2.43500188e-01
-7.19428062e-01 -7.02305377e-01 -7.48365462e-01 1.58497185e-01
-1.38227135e-01 3.20215613e-01 3.28537971e-01 2.53891230e-01
-7.83421516e-01 -4.12383303e-02 -9.93358254e-01 -4.89982903e-01
5.21704257e-01 2.65660007e-02 -6.03223026e-01 -2.19916150e-01
-9.10844028e-01 7.47656941e-01 7.06182063e-01 1.74662873e-01
-1.65397406e+00 -1.34812251e-01 -1.23105145e+00 1.34242356e-01
8.96093905e-01 -5.11873066e-01 1.34923232e+00 -1.42254388e+00
-1.04595339e+00 5.93853056e-01 -1.61524430e-01 -8.31776559e-01
4.63104784e-01 -2.79539376e-01 -6.52263880e-01 3.57408255e-01
1.19633213e-01 6.85515046e-01 7.37852454e-01 -1.16355312e+00
-7.72248030e-01 -1.72189265e-01 3.32681388e-01 3.58379036e-01
-1.31346494e-01 1.44149348e-01 -8.00958693e-01 -6.10713661e-01
-4.80764031e-01 -8.66299272e-01 -5.04210964e-02 2.05200523e-01
-4.92777526e-02 -3.54598300e-03 1.13882899e+00 -7.88377106e-01
9.85939503e-01 -2.14585972e+00 5.38354144e-02 -3.22464913e-01
-2.85983104e-02 3.80363315e-01 -2.89051026e-01 3.06658179e-01
5.47341667e-02 -4.39805686e-01 -3.26369613e-01 -2.20042765e-01
-2.44789898e-01 3.66900355e-01 -2.86467999e-01 4.93515968e-01
3.30957800e-01 7.92672634e-01 -1.07843339e+00 -6.66229546e-01
4.59803611e-01 4.94894415e-01 -2.16790497e-01 4.71622050e-01
-6.28769338e-01 4.90948141e-01 -2.88625151e-01 6.14734590e-01
3.59137744e-01 -1.33874968e-01 5.55572771e-02 -1.45215988e-01
1.46138063e-02 -4.25282449e-01 -1.21807396e+00 1.81261969e+00
-3.42834353e-01 7.45003402e-01 -2.83022434e-01 -9.33136284e-01
5.44093549e-01 5.28272450e-01 6.23839915e-01 -5.06215990e-01
6.01961836e-02 -1.55658096e-01 -2.14639321e-01 -9.14990127e-01
2.75991470e-01 2.93500543e-01 -9.51989889e-02 2.20339432e-01
2.17235789e-01 4.75397289e-01 3.91487211e-01 2.53700882e-01
1.40201676e+00 4.83431816e-01 4.71661091e-01 3.59746903e-01
8.55401635e-01 1.13133766e-01 5.79495311e-01 6.81545556e-01
-5.34909964e-01 6.59026802e-01 3.02112520e-01 -5.92522264e-01
-8.79669309e-01 -1.05149174e+00 4.62061346e-01 1.10187113e+00
3.89070451e-01 -2.14159548e-01 -5.74341655e-01 -9.47436512e-01
-4.89897937e-01 7.00366497e-01 -5.63060880e-01 -1.89839929e-01
-4.72265482e-01 -3.64791512e-01 4.32012856e-01 7.55732417e-01
7.45274901e-01 -1.41666806e+00 -9.39046741e-01 1.92923531e-01
-5.58142304e-01 -1.66230989e+00 -8.66033256e-01 -1.61781326e-01
-3.74151111e-01 -1.44077325e+00 -8.32086742e-01 -9.74007010e-01
5.08398235e-01 5.17204881e-01 9.40867126e-01 -2.44619623e-01
-2.06988573e-01 8.07003379e-01 -5.18627763e-01 -2.82874018e-01
-4.63383257e-01 -5.99271238e-01 -3.87498215e-02 6.57886505e-01
4.84959394e-01 -2.74474360e-02 -4.87039715e-01 3.88882577e-01
-1.11372089e+00 2.58094996e-01 6.78858876e-01 5.59309125e-01
4.68890905e-01 -9.52113345e-02 3.76614660e-01 -4.17938858e-01
2.76482970e-01 -6.42499447e-01 -4.78885621e-01 5.31784654e-01
1.29056931e-01 4.18995321e-02 5.31624258e-01 -7.54123092e-01
-1.25986457e+00 6.52147114e-01 2.96888560e-01 -8.95479083e-01
-4.29393858e-01 7.60605037e-02 -3.03446621e-01 2.47982278e-01
4.70092624e-01 5.71561456e-01 -8.06534290e-02 -2.02114254e-01
3.28488380e-01 7.68194318e-01 9.03067887e-01 -2.36584976e-01
5.33104420e-01 6.39035642e-01 -1.64176509e-01 -7.11207092e-01
-8.86407256e-01 -8.61203313e-01 -4.03411120e-01 -6.01207376e-01
1.18050611e+00 -1.30520093e+00 -5.17946243e-01 4.57321286e-01
-1.42975390e+00 -1.37388706e-01 -7.02416748e-02 5.89768291e-01
-9.43530917e-01 5.93617141e-01 -3.87070984e-01 -9.39179003e-01
-2.21601129e-01 -1.06128478e+00 1.37967336e+00 1.02695376e-01
1.26317009e-01 -7.07802832e-01 7.98554625e-03 4.43813354e-01
8.55769888e-02 5.00813901e-01 2.83111393e-01 -6.82854176e-01
-6.61023676e-01 -4.95591126e-02 -1.61298379e-01 5.55913925e-01
1.45228446e-01 -2.83193886e-01 -7.98459768e-01 -1.29638225e-01
-1.24432020e-01 -3.11624229e-01 7.80071318e-01 2.56717533e-01
1.22994959e+00 -5.20231605e-01 -3.85876387e-01 1.79527655e-01
1.31247032e+00 4.36192155e-01 6.98883355e-01 6.44238666e-02
5.59739590e-01 4.30978358e-01 8.87834847e-01 6.42638803e-01
2.05464453e-01 8.67514074e-01 7.06101775e-01 -1.62313163e-01
-2.04366863e-01 -3.35223526e-01 8.21502566e-01 1.02467835e-01
-3.57778445e-02 -6.95998549e-01 -7.35681832e-01 7.60460734e-01
-2.24284673e+00 -1.44174242e+00 -2.45825592e-02 2.07255960e+00
3.46410930e-01 -6.41153753e-03 1.12007603e-01 -1.83686435e-01
1.07564867e+00 2.49072194e-01 -4.92257684e-01 -6.17544726e-02
-5.91823645e-02 -2.46025667e-01 5.13828814e-01 5.48417158e-02
-1.40175104e+00 7.89995015e-01 5.27424908e+00 4.78500724e-01
-7.63153195e-01 3.20321172e-01 3.86483967e-01 -7.40400702e-02
4.71973479e-01 -3.35074753e-01 -5.84804654e-01 5.80944717e-01
8.90243471e-01 3.74931023e-02 2.40337461e-01 8.58374238e-01
5.12862921e-01 -1.47189841e-01 -1.30887020e+00 1.25350416e+00
5.37976325e-01 -1.17979729e+00 2.21261382e-01 -1.96406722e-01
6.49768233e-01 -9.44021568e-02 -2.25060806e-01 4.93137151e-01
-1.52480062e-02 -7.22763777e-01 7.65136838e-01 6.60560310e-01
6.62581384e-01 -5.18924832e-01 8.58845234e-01 6.42479539e-01
-1.37042975e+00 -3.57050627e-01 -5.72133735e-02 -1.32240772e-01
4.08426285e-01 -9.21819061e-02 -8.17607522e-01 3.39278936e-01
6.09137475e-01 1.05798209e+00 -4.49699581e-01 1.09603226e+00
-1.95375130e-01 5.09455681e-01 -2.51744930e-02 -6.88565448e-02
4.45858210e-01 3.83099280e-02 5.71609259e-01 1.17843091e+00
2.46182755e-01 2.16355100e-01 4.34594333e-01 6.19108319e-01
1.68352038e-01 -2.16435656e-01 -8.65265131e-01 -3.42751630e-02
1.07276410e-01 8.98658872e-01 -5.46535194e-01 -5.34205735e-01
-8.04832637e-01 1.30887187e+00 -2.40212549e-02 4.02231306e-01
-1.48836231e+00 4.22248319e-02 6.03923023e-01 1.77875742e-01
4.32400852e-01 -1.48520008e-01 6.10018671e-01 -1.39043725e+00
2.19106674e-01 -8.24037969e-01 6.20857775e-01 -1.20303571e+00
-8.82396817e-01 6.14258111e-01 4.97140288e-01 -1.64144051e+00
-3.19367796e-01 -5.74994564e-01 -4.39569414e-01 1.51173621e-01
-1.23656654e+00 -1.25495446e+00 -5.80630422e-01 8.38037312e-01
1.21578300e+00 -1.57691583e-01 3.75579208e-01 3.20212811e-01
-5.30309379e-01 3.97111066e-02 -2.85175536e-02 2.44531989e-01
5.22429168e-01 -7.64598787e-01 1.87704846e-01 1.00204992e+00
4.65056360e-01 -1.23963863e-01 6.81589246e-01 -6.61336064e-01
-1.26387322e+00 -1.59256399e+00 5.74360609e-01 -4.12977695e-01
4.36287671e-01 -4.72299188e-01 -7.52811670e-01 1.10357523e+00
2.50608116e-01 3.31862152e-01 3.17219526e-01 -6.54796720e-01
-2.25657839e-02 -3.29713710e-02 -9.72386181e-01 4.40889359e-01
1.27627575e+00 -1.93587273e-01 -8.14239681e-01 5.77720106e-01
7.60934830e-01 -5.12113273e-01 -2.85840243e-01 3.93832386e-01
4.03122455e-01 -7.16488123e-01 9.44997728e-01 -7.89718032e-01
2.90591657e-01 -6.09129250e-01 -1.38907850e-01 -9.37638223e-01
-1.44911587e-01 -3.21365952e-01 -4.79868174e-01 1.01734006e+00
-3.55424583e-02 1.53386286e-02 6.71203196e-01 4.41188961e-01
-1.93060860e-01 -2.53670931e-01 -8.77935052e-01 -9.41191554e-01
-8.68572474e-01 -5.34624636e-01 2.97063828e-01 5.21174908e-01
-3.66155565e-01 1.52095810e-01 -9.91418183e-01 5.12536585e-01
5.62578797e-01 -2.20358461e-01 8.10731232e-01 -7.25641251e-01
-3.40475261e-01 2.51621604e-01 -9.44510102e-01 -1.06025028e+00
3.53824198e-01 -4.29617643e-01 2.76404589e-01 -1.43106830e+00
5.29823720e-01 3.21727455e-01 -3.87884915e-01 5.55853486e-01
1.40944898e-01 2.54257053e-01 3.08777034e-01 7.53407255e-02
-1.20853209e+00 6.38476729e-01 8.57674539e-01 -5.24723709e-01
-5.55853508e-02 1.05161086e-01 -1.21868305e-01 6.62445366e-01
6.18599832e-01 -4.22697604e-01 -5.93088686e-01 -4.56299990e-01
-4.94505495e-01 3.47608447e-01 8.22734952e-01 -1.43175316e+00
1.20134629e-01 -4.02439892e-01 2.34523222e-01 -6.20595157e-01
6.06566906e-01 -1.27013624e+00 3.74778509e-01 4.89713877e-01
-3.92121971e-01 -1.12705506e-01 1.88897967e-01 1.08029163e+00
-3.18604529e-01 6.12095483e-02 5.93445480e-01 -2.40337461e-01
-1.47451770e+00 4.49809521e-01 -4.25070375e-01 -1.37864277e-01
1.84162378e+00 -2.39493579e-01 -2.42784292e-01 -5.63477933e-01
-7.78784037e-01 3.95717680e-01 2.04362109e-01 8.40695739e-01
9.74609792e-01 -1.33709383e+00 -7.89058447e-01 4.66277674e-02
6.62030876e-01 -2.59954035e-01 4.18568522e-01 6.35613501e-01
-5.31898618e-01 4.19931442e-01 -4.00500625e-01 -7.16764033e-01
-1.44203782e+00 1.01347435e+00 3.62523347e-01 8.01067613e-03
-5.96064746e-01 3.42847168e-01 5.55518866e-01 1.60415336e-01
6.16061866e-01 -3.73876929e-01 -4.07660395e-01 -3.86924028e-01
7.87047923e-01 2.32513428e-01 -3.44850838e-01 -1.08268821e+00
-5.25633991e-01 2.90571928e-01 1.79208308e-01 -6.59740195e-02
1.04885292e+00 -2.26633370e-01 3.53687525e-01 3.73068184e-01
1.02707529e+00 -5.63675165e-01 -1.60210562e+00 -3.69383335e-01
-1.11532047e-01 -3.56379241e-01 -1.05617434e-01 -7.52442002e-01
-1.01214337e+00 7.10570872e-01 8.80831301e-01 -3.58771145e-01
1.26640809e+00 1.62397802e-01 6.31265283e-01 3.87075484e-01
4.49407637e-01 -8.40757251e-01 4.07243311e-01 2.00860903e-01
9.47714150e-01 -1.45135200e+00 -2.61761069e-01 -2.83201098e-01
-1.17288065e+00 1.06627440e+00 1.01556158e+00 2.71534026e-01
1.85074583e-01 1.22978315e-01 -1.37957320e-01 -3.43939401e-02
-1.01441658e+00 -3.86875063e-01 1.89529762e-01 8.44318330e-01
-3.45068648e-02 -9.08305719e-02 -8.28660205e-02 5.13425708e-01
8.09008777e-01 2.54112095e-01 7.68249512e-01 1.04526269e+00
-4.33718055e-01 -5.58510721e-01 -5.94066441e-01 1.54343322e-01
-2.52927125e-01 1.12530790e-01 -3.43999445e-01 7.14590907e-01
4.47755009e-01 1.10681951e+00 2.84452021e-01 -2.41266981e-01
2.27519363e-01 -6.44756714e-03 3.20637137e-01 -6.10888898e-01
-2.46959731e-01 -1.74564928e-01 6.96726292e-02 -7.36638367e-01
-8.46591532e-01 -6.42995358e-01 -1.13218296e+00 3.70241344e-01
-5.46149909e-02 4.13518697e-02 3.88171226e-01 8.59736443e-01
3.46283972e-01 4.72864032e-01 3.81530702e-01 -6.82147861e-01
-4.12226528e-01 -9.45196092e-01 -4.34622347e-01 7.20232368e-01
5.22786617e-01 -8.54836464e-01 -1.86560199e-01 6.46386743e-01] | [8.690948486328125, 0.6696100831031799] |
af356f70-0286-4366-86a7-10f70271c9b6 | adas-a-direct-adaptation-strategy-for-multi | 2203.06811 | null | https://arxiv.org/abs/2203.06811v1 | https://arxiv.org/pdf/2203.06811v1.pdf | ADAS: A Direct Adaptation Strategy for Multi-Target Domain Adaptive Semantic Segmentation | In this paper, we present a direct adaptation strategy (ADAS), which aims to directly adapt a single model to multiple target domains in a semantic segmentation task without pretrained domain-specific models. To do so, we design a multi-target domain transfer network (MTDT-Net) that aligns visual attributes across domains by transferring the domain distinctive features through a new target adaptive denormalization (TAD) module. Moreover, we propose a bi-directional adaptive region selection (BARS) that reduces the attribute ambiguity among the class labels by adaptively selecting the regions with consistent feature statistics. We show that our single MTDT-Net can synthesize visually pleasing domain transferred images with complex driving datasets, and BARS effectively filters out the unnecessary region of training images for each target domain. With the collaboration of MTDT-Net and BARS, our ADAS achieves state-of-the-art performance for multi-target domain adaptation (MTDA). To the best of our knowledge, our method is the first MTDA method that directly adapts to multiple domains in semantic segmentation. | ['Sunghoon Im', 'Minwoo Choi', 'Changjae Kim', 'Wonhyeok Choi', 'Seunghun Lee'] | 2022-03-14 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Lee_ADAS_A_Direct_Adaptation_Strategy_for_Multi-Target_Domain_Adaptive_Semantic_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Lee_ADAS_A_Direct_Adaptation_Strategy_for_Multi-Target_Domain_Adaptive_Semantic_CVPR_2022_paper.pdf | cvpr-2022-1 | ['multi-target-domain-adaptation'] | ['computer-vision'] | [ 4.73562300e-01 8.02325532e-02 -9.28375274e-02 -6.56648695e-01
-9.10814047e-01 -6.63556337e-01 4.26871330e-01 -1.80088162e-01
-4.19587702e-01 5.09345114e-01 -1.60042301e-01 9.21704574e-04
3.56798284e-02 -7.61148691e-01 -9.19455886e-01 -5.71103990e-01
5.66718996e-01 9.10419524e-01 7.71592557e-01 -2.54177362e-01
3.89032923e-02 4.06500965e-01 -1.29216325e+00 4.06336635e-01
1.24628127e+00 1.08365655e+00 5.82941413e-01 2.89826661e-01
-3.07108313e-01 3.23839873e-01 -6.23861969e-01 -1.83506295e-01
3.18061173e-01 -5.59118986e-01 -8.27850223e-01 3.57177109e-01
6.71542346e-01 -2.84698248e-01 -7.58200213e-02 1.04835534e+00
4.71083492e-01 2.18361452e-01 1.07143557e+00 -1.40115941e+00
-9.55391467e-01 3.92288238e-01 -6.97258294e-01 9.47073393e-04
-2.20012948e-01 1.57032400e-01 6.59949481e-01 -1.02000225e+00
7.73106217e-01 1.39852548e+00 5.29297173e-01 8.31909597e-01
-1.43978631e+00 -7.69823372e-01 5.65649927e-01 2.48154625e-01
-1.22109103e+00 -1.63666934e-01 1.03955436e+00 -4.26359475e-01
5.26445687e-01 -1.74816132e-01 4.95563358e-01 1.31142545e+00
-7.42763504e-02 1.06874502e+00 1.30944848e+00 -3.69567633e-01
4.87607270e-01 9.21694487e-02 -1.21196173e-01 4.11095023e-01
5.60140014e-02 -1.63547590e-01 -4.52589959e-01 1.86750069e-01
9.88717079e-01 -3.65077406e-01 1.36415124e-01 -9.83243465e-01
-1.09696066e+00 7.88567305e-01 4.22255039e-01 1.99851334e-01
-2.90521801e-01 -1.89669803e-01 4.37258214e-01 2.88010031e-01
5.49114704e-01 2.92334080e-01 -7.72719324e-01 3.51103038e-01
-7.13986635e-01 1.52749375e-01 2.59848356e-01 1.24478006e+00
9.21206713e-01 1.80541188e-01 -3.15645874e-01 1.19336641e+00
1.29062966e-01 5.48030555e-01 5.43949664e-01 -1.10012007e+00
3.39590371e-01 6.84131145e-01 -7.08121136e-02 -4.75381792e-01
-2.61779934e-01 -4.20593321e-01 -6.62725687e-01 4.89193350e-01
5.33887327e-01 -9.63053480e-02 -1.45660806e+00 1.99254978e+00
5.85906684e-01 5.70406914e-02 2.37286702e-01 1.02499831e+00
6.87040865e-01 4.33653265e-01 3.96643668e-01 1.67539671e-01
1.21158421e+00 -1.16180980e+00 -4.54042912e-01 -7.38236904e-01
2.06670374e-01 -5.83703518e-01 1.37225485e+00 1.48863643e-01
-9.28255379e-01 -8.04236948e-01 -9.59236026e-01 -1.29924968e-01
-4.56954658e-01 1.32593870e-01 3.69155824e-01 3.22034746e-01
-9.33928728e-01 2.22641826e-01 -6.86934710e-01 -6.04338765e-01
8.11417162e-01 3.14263076e-01 -4.26357418e-01 -2.07618490e-01
-1.04927170e+00 8.39752078e-01 5.46930254e-01 -4.52088714e-01
-1.12045562e+00 -7.61454642e-01 -9.73769784e-01 -1.47495031e-01
4.50554818e-01 -9.12759960e-01 1.35444331e+00 -1.62486911e+00
-1.64320827e+00 1.11564755e+00 -1.74304605e-01 -3.81303400e-01
4.15415794e-01 -8.70908946e-02 -3.66593748e-01 1.93053737e-01
4.99677956e-01 1.16580808e+00 1.19473970e+00 -1.55765522e+00
-6.13059938e-01 -4.94870484e-01 -2.93004423e-01 3.45180869e-01
-3.79908204e-01 -1.56649619e-01 -5.67388475e-01 -1.02182460e+00
5.53195924e-02 -8.84999931e-01 -2.26803049e-01 2.71526963e-01
-2.46251360e-01 -2.22287506e-01 1.04956365e+00 -4.58504379e-01
7.09567130e-01 -2.27733755e+00 4.22579169e-01 7.13894218e-02
1.32645503e-01 4.12582844e-01 -5.29040158e-01 -6.39013350e-02
-9.36948322e-03 -2.34192342e-01 -7.72604346e-01 -3.12602013e-01
3.76861989e-02 5.11080682e-01 -1.38687328e-01 1.25659168e-01
5.25851309e-01 8.92219901e-01 -8.05598617e-01 -7.63060749e-01
2.63678461e-01 2.05413938e-01 -4.38875854e-01 3.13172728e-01
-5.82865894e-01 7.10078359e-01 -6.50067389e-01 6.37001991e-01
9.60342526e-01 -2.49509260e-01 1.20401166e-01 -2.01973960e-01
2.45541006e-01 -1.11167535e-01 -9.52135563e-01 1.97370231e+00
-2.51744986e-01 4.35519278e-01 1.51818559e-01 -1.16674280e+00
1.17548287e+00 -7.54285306e-02 3.50395620e-01 -9.75760877e-01
8.71983320e-02 3.52975130e-01 -1.52380019e-01 -1.26328185e-01
2.47220501e-01 -1.24123745e-01 -3.80995005e-01 1.87830120e-01
3.82417202e-01 -2.49263376e-01 -4.53281961e-02 4.96983994e-03
7.44797170e-01 3.65534097e-01 2.55718142e-01 -2.97206581e-01
4.36286777e-01 3.66876006e-01 8.64702404e-01 6.68690324e-01
-3.86411577e-01 8.70825768e-01 1.89338237e-01 -4.37721491e-01
-1.26923490e+00 -1.48789024e+00 -2.86254864e-02 1.49032831e+00
3.40289593e-01 2.73981094e-01 -1.04738057e+00 -1.05338144e+00
1.67428821e-01 7.48948693e-01 -6.43204629e-01 -3.05969834e-01
-5.39634764e-01 -4.24245238e-01 3.88967991e-01 7.48017192e-01
9.02242482e-01 -1.16779089e+00 -2.96834081e-01 1.71071976e-01
-2.22480744e-01 -1.16817153e+00 -7.55458295e-01 3.21776509e-01
-8.95467699e-01 -8.85219097e-01 -9.92944181e-01 -1.23657668e+00
8.52876604e-01 1.68931082e-01 1.06846452e+00 -5.84890962e-01
1.18674316e-01 2.65757501e-01 -3.45361739e-01 -4.37488347e-01
-5.86843848e-01 2.74047166e-01 -1.12548634e-01 1.70567185e-01
4.74246800e-01 -6.27863526e-01 -5.69376767e-01 6.93692148e-01
-8.63072813e-01 2.39068881e-01 7.94276834e-01 8.01442087e-01
1.05197382e+00 -2.40397915e-01 7.05660462e-01 -1.09872246e+00
4.76533711e-01 -3.71977448e-01 -5.86166680e-01 2.57693022e-01
-5.57011664e-01 9.27457139e-02 7.83867359e-01 -6.62801266e-01
-1.34869218e+00 4.64725584e-01 7.95757025e-02 -6.92604065e-01
-5.76624811e-01 -4.93627340e-02 -4.81262207e-01 -1.07281990e-01
8.38537574e-01 2.90578395e-01 -3.68202291e-02 -6.47621095e-01
6.09619260e-01 6.03062272e-01 8.22178006e-01 -7.64907837e-01
8.88949633e-01 4.97941881e-01 -1.94228426e-01 -4.35747802e-01
-9.45034385e-01 -3.34391862e-01 -1.03256631e+00 -1.35440707e-01
1.08726263e+00 -1.06618440e+00 6.04110211e-02 9.15911734e-01
-1.01236999e+00 -7.58486450e-01 -3.78870815e-01 1.08107746e-01
-8.08569252e-01 2.44383514e-01 -3.48632336e-01 -2.16703549e-01
-1.41660020e-01 -1.04797602e+00 1.20700610e+00 2.74275333e-01
-1.59057274e-01 -9.21624541e-01 -6.41167462e-02 3.92962575e-01
3.27907085e-01 2.71923691e-01 8.06620121e-01 -7.99947083e-01
-4.40680981e-01 4.43669558e-01 -4.84835893e-01 4.80203271e-01
3.10268819e-01 -4.41896945e-01 -9.59524870e-01 -3.04907501e-01
-2.08297342e-01 -4.11061376e-01 1.00860369e+00 6.45557761e-01
1.33340001e+00 -4.75634774e-03 -4.09251243e-01 8.49954247e-01
1.19916105e+00 4.67379838e-01 4.55626726e-01 6.92803800e-01
8.93316209e-01 5.50171196e-01 9.80194509e-01 1.80482328e-01
4.47861701e-01 7.75493205e-01 3.19077373e-01 -5.62407255e-01
-5.62764645e-01 -3.49809349e-01 2.77364284e-01 2.71714896e-01
3.05481821e-01 -1.55907407e-01 -7.91456819e-01 8.91264319e-01
-1.96662343e+00 -4.68960226e-01 1.31741002e-01 1.86024535e+00
8.69971514e-01 1.64847374e-01 3.32189262e-01 -3.82885605e-01
1.04948866e+00 -6.62972480e-02 -1.20827997e+00 -3.29602957e-01
-2.86132336e-01 2.04300895e-01 5.95489800e-01 2.64164627e-01
-1.39880872e+00 1.35796678e+00 6.23842907e+00 1.17198575e+00
-8.59362960e-01 2.60935098e-01 5.64125419e-01 3.09027433e-01
-2.42383718e-01 -2.29513496e-01 -6.30486012e-01 4.65839446e-01
4.79662806e-01 -1.05383523e-01 2.84752458e-01 1.25675237e+00
-5.63853420e-02 1.80182606e-01 -8.72562289e-01 6.82063043e-01
3.78868543e-02 -1.05297744e+00 3.74421299e-01 -5.99803664e-02
9.45310831e-01 -1.12223245e-01 2.42719725e-01 2.44429573e-01
6.17184579e-01 -7.73059726e-01 9.04989362e-01 1.00116052e-01
9.31019187e-01 -7.05945194e-01 3.59391659e-01 2.11163275e-02
-1.19652736e+00 -1.24850318e-01 -4.19353306e-01 3.71804535e-01
1.20353788e-01 3.76962066e-01 -8.07354152e-01 3.67786884e-01
8.02406430e-01 8.60560000e-01 -5.72467148e-01 8.28937113e-01
-2.66865432e-01 4.76182103e-01 -1.56674787e-01 4.76821542e-01
2.33654305e-01 -1.39465734e-01 6.42102897e-01 1.09168136e+00
2.55479366e-01 -1.29243940e-01 2.54144639e-01 1.00906610e+00
-4.51939106e-02 -1.34880677e-01 -5.50531566e-01 2.44618416e-01
7.18502402e-01 1.04989696e+00 -7.56622195e-01 -4.75958079e-01
-2.08866373e-01 1.44136548e+00 4.22786027e-01 4.37737912e-01
-7.93821037e-01 -2.49601915e-01 8.71189177e-01 1.09657109e-01
6.94665670e-01 -4.54036556e-02 -6.24039590e-01 -9.27231014e-01
-5.29128462e-02 -8.49911094e-01 7.01700866e-01 -9.45569038e-01
-1.74364233e+00 6.64065421e-01 2.34825134e-01 -1.42812765e+00
-2.94626765e-02 -6.58918202e-01 -2.89784670e-01 7.73230672e-01
-1.66335523e+00 -1.44912064e+00 -2.88880587e-01 9.45521295e-01
8.38712513e-01 -3.96676928e-01 6.01321578e-01 2.27145940e-01
-3.42792869e-01 7.17645824e-01 2.07431972e-01 1.36288732e-01
1.12598872e+00 -1.45492208e+00 7.65334666e-01 8.26952100e-01
-3.13393474e-01 1.86633006e-01 5.02672613e-01 -6.49516463e-01
-6.74891233e-01 -1.54713559e+00 2.74155498e-01 -3.31178635e-01
4.06304449e-01 -4.52951401e-01 -1.21115887e+00 7.69180417e-01
1.28839925e-01 -6.75591536e-06 3.81036103e-01 -1.85276106e-01
-5.49348712e-01 -3.29896659e-01 -1.29998219e+00 5.18416345e-01
1.17191970e+00 -3.20612639e-01 -7.17241645e-01 1.52997658e-01
9.04237449e-01 -5.63676775e-01 -7.70612240e-01 4.65638667e-01
2.36176819e-01 -7.93038547e-01 1.21059990e+00 -3.61207247e-01
2.59963512e-01 -5.50754666e-01 -9.20782462e-02 -1.63387489e+00
-5.34850061e-01 -1.79955781e-01 2.78505057e-01 1.35551190e+00
3.66855264e-01 -6.84202015e-01 7.05307424e-01 3.02960157e-01
-5.08244574e-01 -2.14715049e-01 -1.07521260e+00 -1.01534176e+00
3.73925507e-01 -1.29706204e-01 7.30362475e-01 9.79576528e-01
-6.70868874e-01 4.01218832e-01 -1.81084648e-01 2.57821739e-01
8.32170129e-01 2.44137660e-01 6.84260547e-01 -1.34960914e+00
-2.00365171e-01 -3.40023100e-01 -1.59517184e-01 -1.16390491e+00
2.72984475e-01 -7.56997287e-01 2.10506752e-01 -1.68202877e+00
1.37147844e-01 -5.64869344e-01 -3.40301096e-01 7.24907458e-01
-1.60861149e-01 2.04671159e-01 1.39207348e-01 1.15154840e-01
-7.46676505e-01 7.16081679e-01 1.73853040e+00 -3.16674113e-01
-3.00229847e-01 -1.80479437e-01 -9.47139800e-01 6.67222321e-01
8.93092930e-01 -6.26732647e-01 -6.48535967e-01 -5.95607221e-01
-5.09571671e-01 -3.79008740e-01 3.35401177e-01 -9.77823198e-01
4.94691497e-03 -4.37521994e-01 6.91515923e-01 -5.10323167e-01
2.74386168e-01 -7.95596898e-01 -5.39009534e-02 8.33299607e-02
-1.74179807e-01 -1.95014954e-01 5.53900778e-01 6.54198825e-01
-2.27721080e-01 1.26066759e-01 1.27231753e+00 -1.47382185e-01
-1.38435447e+00 2.52509683e-01 -3.88992310e-01 2.71012008e-01
1.24331009e+00 -4.97287333e-01 -3.69728804e-01 -8.67024902e-03
-6.78634346e-01 3.66474599e-01 7.75505960e-01 5.70409119e-01
6.61726952e-01 -1.44215894e+00 -6.04004860e-01 3.21875244e-01
4.14303660e-01 4.84021127e-01 4.03461695e-01 2.24023640e-01
-1.65594384e-01 -1.03227878e-02 -6.90887153e-01 -8.04334402e-01
-1.22022319e+00 6.82208478e-01 4.58245307e-01 -8.68417919e-02
-5.36013246e-01 9.10052240e-01 7.88206220e-01 -7.17187941e-01
-1.48009405e-01 -6.00828677e-02 -2.05823794e-01 -6.36461154e-02
1.76560253e-01 5.93102351e-02 -1.25686508e-02 -6.62055790e-01
-4.04371858e-01 8.70866239e-01 -3.06557328e-01 -4.95283678e-02
1.21548736e+00 -3.09162349e-01 1.65408939e-01 2.27786690e-01
9.13561046e-01 -4.58833516e-01 -1.95432293e+00 -3.67640316e-01
-2.38293290e-01 -4.25478786e-01 -1.60168752e-01 -1.10215449e+00
-1.14205623e+00 8.22068930e-01 6.90147758e-01 -3.80926788e-01
1.61396897e+00 2.90139377e-01 8.75479937e-01 4.50982042e-02
2.59638220e-01 -1.38085163e+00 4.85718608e-01 4.77259755e-01
7.92825937e-01 -1.21102905e+00 -4.04323637e-01 -5.00983179e-01
-1.21184444e+00 8.49441588e-01 1.22577906e+00 -9.10115167e-02
2.13501200e-01 -3.57528999e-02 3.51769686e-01 -3.70288603e-02
-3.46613824e-01 -4.44830388e-01 3.53343576e-01 1.29567599e+00
-1.89578012e-01 7.42464960e-02 6.74245581e-02 6.24866307e-01
1.32557169e-01 -1.20520055e-01 2.13604823e-01 6.77070200e-01
-5.62097847e-01 -1.35450125e+00 -5.14612317e-01 1.17542051e-01
4.62473780e-02 1.18734621e-01 -5.38889647e-01 8.54757607e-01
3.19746852e-01 7.25063801e-01 1.88498616e-01 -3.54647398e-01
5.94747901e-01 1.02386788e-01 4.23804969e-01 -6.21963322e-01
-2.45903566e-01 1.80931300e-01 -2.04030380e-01 -3.17976356e-01
-4.08120096e-01 -7.01007128e-01 -1.25957096e+00 1.22695990e-01
2.32708260e-01 -1.66454673e-01 3.57353836e-01 8.40577543e-01
6.01848185e-01 7.09193170e-01 4.39639777e-01 -7.46465087e-01
-1.80432230e-01 -8.04637194e-01 -5.88842332e-01 6.45798683e-01
2.96064407e-01 -9.88085270e-01 1.94354984e-03 2.43271589e-01] | [9.740579605102539, 1.3624237775802612] |
705c1451-625c-46e1-9f3c-46d0250b8cec | node-centric-graph-learning-from-data-for | 2011.02179 | null | https://arxiv.org/abs/2011.02179v1 | https://arxiv.org/pdf/2011.02179v1.pdf | Node-Centric Graph Learning from Data for Brain State Identification | Data-driven graph learning models a network by determining the strength of connections between its nodes. The data refers to a graph signal which associates a value with each graph node. Existing graph learning methods either use simplified models for the graph signal, or they are prohibitively expensive in terms of computational and memory requirements. This is particularly true when the number of nodes is high or there are temporal changes in the network. In order to consider richer models with a reasonable computational tractability, we introduce a graph learning method based on representation learning on graphs. Representation learning generates an embedding for each graph node, taking the information from neighbouring nodes into account. Our graph learning method further modifies the embeddings to compute the graph similarity matrix. In this work, graph learning is used to examine brain networks for brain state identification. We infer time-varying brain graphs from an extensive dataset of intracranial electroencephalographic (iEEG) signals from ten patients. We then apply the graphs as input to a classifier to distinguish seizure vs. non-seizure brain states. Using the binary classification metric of area under the receiver operating characteristic curve (AUC), this approach yields an average of 9.13 percent improvement when compared to two widely used brain network modeling methods. | ['Stark C. Draper', 'Taufik A. Valiante', 'Roman Genov', 'David M. Groppe', 'Nafiseh Ghoroghchian'] | 2020-11-04 | null | null | null | null | ['graph-similarity'] | ['graphs'] | [ 4.17997241e-01 4.85548854e-01 3.62765715e-02 -3.11488926e-01
-1.68409407e-01 -3.20589781e-01 6.08129680e-01 7.12178111e-01
-1.75181434e-01 4.65188473e-01 1.78518176e-01 -4.15819436e-01
-6.17723703e-01 -8.81082833e-01 -2.26676762e-01 -5.98568797e-01
-9.24883068e-01 3.20204824e-01 1.39157623e-01 -9.13342014e-02
1.98581070e-01 6.21737003e-01 -9.91118610e-01 -3.24810930e-02
7.14690268e-01 8.02631199e-01 -9.85047501e-03 7.13018596e-01
-2.36362126e-02 8.28233778e-01 -4.94687825e-01 -8.94805938e-02
1.75441857e-02 -6.17649615e-01 -6.37175083e-01 1.74174115e-01
-5.72217964e-02 2.22016156e-01 -6.87306881e-01 1.10841823e+00
4.76130515e-01 -5.42608090e-02 8.87982249e-01 -1.53249419e+00
-2.86765069e-01 8.47133279e-01 -3.25468868e-01 6.31050706e-01
3.55790824e-01 -2.37471804e-01 9.79136109e-01 -4.51831192e-01
5.72060883e-01 8.92208040e-01 5.93121290e-01 2.51357615e-01
-1.61250377e+00 -4.43090022e-01 1.64026156e-01 5.20048916e-01
-1.26312590e+00 -1.94159746e-01 1.16443920e+00 -7.61859059e-01
1.02097058e+00 8.17834139e-02 1.20691943e+00 8.95216882e-01
7.52037942e-01 6.28043562e-02 1.24445379e+00 -4.06149685e-01
4.86855328e-01 -2.82000512e-01 4.70156729e-01 7.87836790e-01
4.10856903e-01 5.32459468e-02 -4.48120832e-01 -3.97590399e-01
5.13166308e-01 1.45160183e-02 -3.64372790e-01 -6.36885166e-01
-1.05473554e+00 9.18032765e-01 6.14432573e-01 4.92052227e-01
-5.87593615e-01 1.02816097e-01 3.72043848e-01 6.44681692e-01
5.66961527e-01 3.76125097e-01 -2.24170536e-02 1.15780205e-01
-9.21043277e-01 -2.47286126e-01 1.01932216e+00 5.24152219e-01
6.96059823e-01 1.20155759e-01 2.30566055e-01 5.82957029e-01
4.31163341e-01 -8.77824351e-02 5.42194009e-01 -1.98040918e-01
1.77607268e-01 9.88054872e-01 -7.34242320e-01 -1.30111992e+00
-9.10079718e-01 -3.39099258e-01 -9.87811089e-01 6.62271604e-02
2.11354807e-01 -1.79302216e-01 -8.37522089e-01 1.52814341e+00
-1.00330837e-01 4.78998661e-01 -1.22120261e-01 4.95143801e-01
7.05896080e-01 2.24439248e-01 -8.99130926e-02 -3.56337637e-01
1.24463320e+00 -2.87522078e-01 -6.65094256e-01 -2.28999689e-01
7.29839146e-01 -1.13616824e-01 5.42344868e-01 2.96303093e-01
-6.45372748e-01 1.97101310e-02 -1.11478150e+00 6.10064149e-01
-5.09508133e-01 -4.79968250e-01 5.40630937e-01 5.19693255e-01
-1.44731283e+00 7.55794942e-01 -1.04455483e+00 -4.34318393e-01
3.73584062e-01 6.38899505e-01 -5.33440411e-01 -4.70252987e-03
-1.17632556e+00 8.77553999e-01 4.22860533e-01 -2.40008384e-02
-6.30991459e-01 -4.99811232e-01 -1.07992530e+00 9.08587947e-02
7.43405223e-02 -3.70409608e-01 4.34974223e-01 -8.10542762e-01
-1.08883321e+00 7.20062256e-01 1.91478148e-01 -6.86247885e-01
3.41417909e-01 6.94983363e-01 -8.18810523e-01 3.52200121e-01
-1.42281353e-01 1.73624337e-01 1.07422686e+00 -7.58294165e-01
9.14930776e-02 -5.27801394e-01 -1.08773030e-01 -9.30106714e-02
-3.67813230e-01 -2.17856586e-01 -6.91673905e-02 -7.10049093e-01
4.47252035e-01 -9.90282834e-01 -2.84902364e-01 -3.10819536e-01
-5.66479564e-01 -2.15347260e-01 7.30261683e-01 -5.75572848e-01
1.36088085e+00 -2.03235841e+00 2.46378243e-01 6.94243670e-01
8.46821368e-01 -1.61569998e-01 -1.52212843e-01 6.66884124e-01
-5.97560823e-01 -6.50309846e-02 -4.69124138e-01 1.36009216e-01
-3.34850490e-01 1.83873530e-02 3.88465822e-01 7.83097208e-01
1.36446610e-01 8.56125474e-01 -1.06437027e+00 -3.45110506e-01
2.29237318e-01 6.22285724e-01 -4.53771353e-01 1.00295879e-01
3.51383716e-01 3.35171252e-01 -2.99803764e-01 2.24509597e-01
1.85189068e-01 -4.55022991e-01 5.95345676e-01 -2.87349313e-01
3.90810907e-01 1.73178777e-01 -8.85334551e-01 1.51789498e+00
-2.08389640e-01 8.75495195e-01 -2.99902439e-01 -1.58898628e+00
1.17271876e+00 3.21211457e-01 8.45290899e-01 -4.27834868e-01
2.89813191e-01 4.03210670e-02 6.24643981e-01 -3.51204008e-01
-3.63865674e-01 -1.02110296e-01 5.75593486e-02 6.08344555e-01
2.01014310e-01 -2.04989165e-01 1.89606920e-01 4.05446142e-01
1.76291215e+00 -5.67597747e-01 6.42371178e-01 -6.13811553e-01
5.61861694e-01 -3.43422353e-01 2.02054799e-01 1.44336462e-01
-1.99480951e-01 2.12428972e-01 1.13337183e+00 -3.58735353e-01
-5.60830951e-01 -1.01249468e+00 -1.32496744e-01 6.12369001e-01
-1.61525160e-01 -6.50913417e-01 -8.59452367e-01 -5.52149415e-01
2.31775013e-03 4.77575868e-01 -8.38326991e-01 -7.09045410e-01
-4.38332051e-01 -8.55877995e-01 1.67771578e-01 2.90380448e-01
-2.25056782e-02 -1.10493314e+00 -4.57410723e-01 2.94736147e-01
2.28253856e-01 -8.38793218e-01 -3.69747996e-01 4.45455045e-01
-1.10368299e+00 -1.47697210e+00 -2.94677973e-01 -1.03214288e+00
1.06977153e+00 -2.29143575e-01 8.55491221e-01 1.96058482e-01
-4.74870086e-01 5.98353505e-01 -3.53067875e-01 -2.88653255e-01
-4.12266403e-01 1.84833028e-04 2.26581424e-01 3.81333143e-01
3.95198882e-01 -9.63791251e-01 -5.18531919e-01 -2.49693682e-03
-7.22601116e-01 -2.52092630e-01 4.88881499e-01 9.38937187e-01
4.72314298e-01 2.45821357e-01 7.92059958e-01 -9.16946292e-01
1.08956313e+00 -7.04733551e-01 -5.77730775e-01 5.40835522e-02
-9.53663886e-01 2.44845346e-01 6.88646913e-01 -4.22776967e-01
-1.92782864e-01 2.09002137e-01 3.60484302e-01 -4.01692331e-01
2.22233564e-01 8.70609820e-01 -1.02158748e-02 -2.27048412e-01
6.12909257e-01 2.66470194e-01 4.04979318e-01 -1.74941555e-01
2.99718320e-01 5.15466630e-01 2.96672016e-01 1.10420242e-01
5.92368662e-01 1.39757708e-01 3.27435404e-01 -1.09577143e+00
-1.15400217e-01 -3.15451145e-01 -9.80931520e-01 -4.18811917e-01
6.35930181e-01 -3.91655833e-01 -5.79808176e-01 1.64319888e-01
-7.24147558e-01 -3.51359397e-01 -1.16404094e-01 7.29245245e-01
-5.48474371e-01 3.69138628e-01 -5.05513012e-01 -6.14656925e-01
-3.26578677e-01 -9.21844721e-01 5.01303017e-01 -3.47335160e-01
-4.63521093e-01 -1.59552979e+00 2.90442675e-01 -2.49612808e-01
2.60688841e-01 6.84329748e-01 1.33872688e+00 -9.14975524e-01
-1.27314568e-01 -4.43128735e-01 -1.43502370e-01 1.02839373e-01
5.85686743e-01 -3.24242204e-01 -5.99984825e-01 -6.16749406e-01
6.80764318e-02 1.90969840e-01 5.61289430e-01 5.76316595e-01
1.04049337e+00 -2.28383958e-01 -4.24609065e-01 4.71521527e-01
1.31790996e+00 2.55167514e-01 3.90996277e-01 9.97813642e-02
7.73015022e-01 8.49225283e-01 -4.34883088e-02 3.65659237e-01
2.16313422e-01 4.19403285e-01 4.67692524e-01 1.24553151e-01
-1.38798997e-01 -8.96721631e-02 2.93659121e-01 1.09906983e+00
-4.13661711e-02 -1.39795825e-01 -1.29401255e+00 4.83219594e-01
-1.65353513e+00 -6.65548742e-01 -1.93770453e-01 2.24238634e+00
3.50695223e-01 2.69397080e-01 1.93082809e-01 4.97197926e-01
8.49694133e-01 2.37900317e-01 -5.84395885e-01 -2.57271528e-01
1.94676310e-01 1.98040038e-01 4.12172407e-01 5.28289318e-01
-7.05710471e-01 5.20776510e-01 6.49042416e+00 -5.06753800e-03
-1.28206146e+00 8.21796129e-04 4.45473909e-01 4.77646105e-02
-2.35375851e-01 -1.55763607e-02 1.88021928e-01 5.39394855e-01
1.36581385e+00 -6.17168725e-01 6.79850757e-01 3.34069729e-01
2.49750540e-01 1.98804155e-01 -1.30395746e+00 1.28260756e+00
3.04303467e-01 -1.15650582e+00 -3.35615613e-02 2.61967719e-01
2.15609208e-01 2.45947763e-01 -3.50917459e-01 -8.41998458e-02
4.64512855e-02 -1.18112922e+00 2.38312647e-01 7.68653750e-01
8.09468746e-01 -6.69176459e-01 5.71740806e-01 2.15181500e-01
-1.30534661e+00 -7.13836700e-02 -7.91733488e-02 -1.59654960e-01
5.94810881e-02 6.41379833e-01 -1.07307184e+00 3.11065048e-01
3.37001979e-01 1.07404363e+00 -7.13351250e-01 9.82562721e-01
-2.17577100e-01 7.27179289e-01 1.04123624e-02 -1.04510501e-01
5.02988473e-02 -3.53410363e-01 7.41933346e-01 1.02433372e+00
2.06203759e-01 1.04206325e-02 3.45755480e-02 7.45876551e-01
-6.70025498e-02 1.68050513e-01 -1.14172447e+00 -2.40540087e-01
3.72166425e-01 1.34844923e+00 -1.24475443e+00 -1.91777974e-01
-3.53909343e-01 9.34426844e-01 5.31365931e-01 2.29088545e-01
-2.90565521e-01 -6.33406758e-01 3.27415466e-01 4.23006773e-01
-1.06676355e-01 -3.20003003e-01 4.09087054e-02 -1.01700890e+00
-1.40072614e-01 -5.40719330e-01 4.22087073e-01 -3.21259230e-01
-1.45317197e+00 9.11813855e-01 5.73927015e-02 -1.23593867e+00
-4.98009980e-01 -6.44459009e-01 -8.49812686e-01 6.27212286e-01
-1.24914861e+00 -7.32582331e-01 -4.24525201e-01 8.15201044e-01
6.27426133e-02 -3.85482877e-01 9.14358079e-01 1.10233009e-01
-3.82958323e-01 2.29073659e-01 -2.59374857e-01 2.94434637e-01
2.06183732e-01 -1.33792961e+00 4.53709722e-01 7.24677563e-01
4.06180531e-01 4.73062605e-01 4.35909688e-01 -6.80514812e-01
-1.58276904e+00 -1.13930404e+00 9.29267585e-01 -3.26253623e-01
1.25245965e+00 -6.34644806e-01 -1.04237938e+00 7.47361124e-01
-2.21313424e-02 4.66225475e-01 7.29231179e-01 -3.42813134e-02
-1.35825917e-01 -4.98039722e-02 -1.07668543e+00 4.06314999e-01
1.01722777e+00 -7.44535506e-01 -6.71739340e-01 4.90801483e-01
2.76635766e-01 2.08447695e-01 -1.20803761e+00 1.24151282e-01
2.96637505e-01 -7.74428427e-01 6.09770536e-01 -4.96506661e-01
-1.39182195e-01 -4.31880588e-04 2.06322312e-01 -1.91014314e+00
-4.41547126e-01 -7.75685310e-01 -1.95358425e-01 7.54960179e-01
3.84152532e-01 -1.05671358e+00 7.67812252e-01 4.87059951e-01
1.06840476e-01 -8.15055966e-01 -9.50537086e-01 -8.49674702e-01
-9.95668992e-02 -5.02688646e-01 5.85875094e-01 1.12803578e+00
7.77235925e-01 5.04502475e-01 9.87407267e-02 4.64196354e-02
7.42584229e-01 -2.12663904e-01 1.40511245e-01 -1.58335865e+00
-2.57815067e-02 -5.34171164e-01 -1.31279492e+00 -7.81314149e-02
4.15725797e-01 -1.54054117e+00 -3.89716864e-01 -1.65659809e+00
4.98065315e-02 -4.38342482e-01 -7.67039180e-01 4.81973052e-01
2.17669606e-01 3.94450948e-02 -2.04283111e-02 5.95701709e-02
-1.80941135e-01 2.19306827e-01 8.60164106e-01 -2.50184029e-01
-4.02280807e-01 -5.71488403e-02 -5.00220001e-01 7.52484381e-01
8.99903119e-01 -4.85555112e-01 -8.34354103e-01 -3.16334032e-02
6.36934936e-02 3.18469495e-01 3.45737994e-01 -9.77277637e-01
3.76231790e-01 4.79354382e-01 1.75155416e-01 -9.69886184e-02
1.80205807e-01 -9.22074139e-01 2.56448418e-01 8.54252934e-01
-2.76833117e-01 1.87237486e-01 3.29723372e-03 8.03772986e-01
-2.97133476e-01 2.86472943e-02 6.55428112e-01 1.43826917e-01
-5.39394915e-01 5.90205193e-01 -5.62788069e-01 -1.57517001e-01
1.17485142e+00 -3.34497303e-01 8.04333538e-02 -5.26094675e-01
-1.18290281e+00 -3.38168181e-02 1.80037364e-01 3.65615904e-01
9.67165351e-01 -1.29394817e+00 -6.71372294e-01 5.52201092e-01
2.91004062e-01 -5.86938798e-01 -1.11242443e-01 1.23123777e+00
-4.43873852e-01 2.06158087e-01 -4.20325547e-01 -5.85245669e-01
-1.40024853e+00 3.78145754e-01 4.71763372e-01 -1.69017643e-01
-9.56334114e-01 5.57260454e-01 -1.67683035e-01 -2.01477334e-01
-8.32762662e-03 -4.81254846e-01 -5.75879574e-01 4.94573146e-01
2.78970748e-01 4.55239207e-01 2.68839538e-01 -7.28500545e-01
-6.34950697e-01 4.01966602e-01 7.45995194e-02 8.94875079e-02
1.47516191e+00 2.47800723e-01 -3.84163678e-01 4.98904705e-01
1.44666946e+00 -3.64540726e-01 -8.67279053e-01 -1.80476829e-01
3.28016073e-01 -1.60419270e-01 3.93816024e-01 -3.74787480e-01
-1.24470031e+00 7.59898245e-01 7.68968284e-01 6.75877988e-01
1.08794725e+00 1.42202660e-01 2.24256441e-01 2.68391907e-01
5.90933442e-01 -6.22707665e-01 -4.12872173e-02 1.65834397e-01
8.56693566e-01 -1.10522556e+00 1.16581224e-01 -3.13990414e-01
-2.82038987e-01 1.39959192e+00 7.49826431e-02 -5.31919122e-01
1.28445065e+00 2.85166383e-01 -2.01295108e-01 -8.76378059e-01
-5.66587985e-01 8.24835226e-02 4.93608773e-01 7.10146070e-01
3.81168932e-01 3.49210769e-01 -4.02456462e-01 2.56037265e-01
-4.10464227e-01 -1.42626897e-01 5.79285085e-01 7.21653044e-01
-2.56800890e-01 -1.01617718e+00 1.18420105e-02 1.15318847e+00
-1.42525777e-01 -2.00542256e-01 -5.55359244e-01 6.37863636e-01
-3.67338479e-01 1.08184659e+00 1.60556540e-01 -5.73531389e-01
3.76923680e-01 2.32961755e-02 5.06903887e-01 -8.47671807e-01
-2.24175796e-01 -2.39069611e-01 8.01654346e-03 -6.61443889e-01
-3.95618409e-01 -7.37927318e-01 -1.36869192e+00 -1.03479281e-01
-2.73742318e-01 6.89380094e-02 5.90813816e-01 7.04577684e-01
4.45145726e-01 7.76258707e-01 7.02781618e-01 -7.64083266e-01
5.65391146e-02 -8.97757411e-01 -1.07461941e+00 3.87410462e-01
3.47822815e-01 -7.77854383e-01 -5.08143485e-01 -1.16517141e-01] | [12.30441951751709, 3.437166452407837] |
18717bb5-fbb5-49a5-8a1e-5433a46b1ff0 | bloom-a-176b-parameter-open-access | 2211.051 | null | https://arxiv.org/abs/2211.05100v4 | https://arxiv.org/pdf/2211.05100v4.pdf | BLOOM: A 176B-Parameter Open-Access Multilingual Language Model | Large language models (LLMs) have been shown to be able to perform new tasks based on a few demonstrations or natural language instructions. While these capabilities have led to widespread adoption, most LLMs are developed by resource-rich organizations and are frequently kept from the public. As a step towards democratizing this powerful technology, we present BLOOM, a 176B-parameter open-access language model designed and built thanks to a collaboration of hundreds of researchers. BLOOM is a decoder-only Transformer language model that was trained on the ROOTS corpus, a dataset comprising hundreds of sources in 46 natural and 13 programming languages (59 in total). We find that BLOOM achieves competitive performance on a wide variety of benchmarks, with stronger results after undergoing multitask prompted finetuning. To facilitate future research and applications using LLMs, we publicly release our models and code under the Responsible AI License. | ['Nikolaus Muellner', 'Nicholas Michio Broad', 'Nathan Dahlberg', 'Helena U. Vrabec', 'Gully Burns', 'Giyaseddin Bayrak', 'Gabriel Altay', 'Florian Fuhrimann', 'Alfredo Palasciano', 'Abhinav Ramesh Kashyap', 'Zach Nguyen', 'Yoyo Yang', 'Trieu Le', 'Tobi Oyebade', 'Ryan Hao', 'Rasmus Kromann', 'Ran An', 'Olanrewaju Samuel', 'Nour Fahmy', 'Nour Elkott', 'Nazneen Rajani', 'Nafis Abrar', 'Mykola Burynok', 'Muhammed Ghauri', 'Mike Qiu', 'Michael McKenna', 'Martha Akinlolu', 'Marissa Gerchick', 'Margot Mieskes', 'Maraim Elbadri', 'Mairon Samagaio', 'Livia Dutra', 'Jesse Passmore', 'Isar Nejadgholi', 'Irina Sedenko', 'Irene Solaiman', 'Indrani Bhattacharya', 'Hessie Jones', 'Habib Rezanejad', 'Frankline Ononiwu', 'Fatima Mirza', 'Ezinwanne Ozoani', 'Emi Baylor', 'Edward Tan', 'Duong A. Nguyen', 'Douwe Kiela', 'Davis David', 'David Lansky', 'Danish Contractor', 'Daniel McDuff', 'Maiko Takeuchi', 'Thanh Le', 'Shachar Mirkin', 'Omar Sanseviero', 'Tali Bers', 'Taewoon Kim', 'Davut Emre Taşar', 'Arun Raja', 'Zeerak Talat', 'Shamsuddeen Hassan Muhammad', 'Shamik Bose', 'BigScience Workshop', 'Chirag Jain', 'Chenxi Zhou', 'Caio Brito', 'Bo wang', 'Benjamin Beilharz', 'Ayush Singh', 'Antonio Miranda-Escalada', 'Anima Shukla', 'Alison Callahan', 'Sylvain Viguier', 'Sourav Roy', 'Silas Wang', 'Sarmad Shubber', 'Samira Alizadeh', 'Julio Bonis Sanz', 'Josh Seltzer', 'Carlos Muñoz Ferrandis', 'Bharat Saxena', 'Benjamin Ajibade', 'Bahareh Behroozi', 'Azadeh HajiHosseini', 'Aycha Tammour', 'Arezoo Abdollahi', 'Arash Aghagol', 'Antigona Unldreaj', 'Anthony Hevia', 'Ana Santos', 'Amy Faranak', 'Ammar Khan', 'Amir Feizpour', 'Amanda Pestana', 'Alice Rueda', 'Zdeněk Kasner', 'Zachary Bamberger', 'Yonatan Belinkov', 'Yada Pruksachatkun', 'Vladislav Mikhailov', 'Vitaly Protasov', 'Verena Rieser', 'Tomasz Limisiewicz', 'Tian Yun', 'Thomas Scialom', 'Tatiana Shavrina', 'Shani Pais', 'Sebastian Gehrmann', 'Ruochen Zhang', 'Rui Zhang', 'Oskar van der Wal', 'Omer Antverg', 'Oleg Serikov', 'Newton Cheng', 'Najoung Kim', 'Miruna Clinciu', 'Marine Carpuat', 'Liam Hazan', 'Ken Kawamura', 'Jungo Kasai', 'Jordan Clive', 'Jessica Zosa Forde', 'Jekaterina Novikova', 'Jan-Christoph Kalo', 'Hailey Schoelkopf', 'Genta Indra Winata', 'Eli Bogdanov', 'Ekaterina Voloshina', 'Ekaterina Taktasheva', 'Ehud Reiter', 'Deepak Tunuguntla', 'Dan Garrette', 'Charles Lovering', 'Aurélie Névéol', 'Arjun Subramonian', 'Anne-Laure Ligozat', 'Anastasia Cheveleva', 'Amanpreet Singh', 'Ahmed Baruwa', 'Tim Dettmers', 'Suraj Patil', 'Stéphane Requena', 'Shaden Smith', 'Sanchit Gandhi', 'Samyam Rajbhandari', 'Rémi Lacroix', 'Pierre François Lavallée', 'Pierre Cornette', 'Patrick von Platen', 'Nouamane Tazi', 'Nicolas Patry', 'Myriam Peyrounette', 'Mohammad Shoeybi', 'Minjia Zhang', 'Mayank Mishra', 'Max Ryabinin', 'Jeff Rasley', 'Jared Casper', 'Hatim Bourfoune', 'Deepak Narayanan', 'Conglong Li', 'Ofir Press', 'Jason Phang', 'Jaesung Tae', 'Hyung Won Chung', 'Adam Roberts', 'Hadar Tojarieh', 'Yallow Uri', 'Shaked Brody', 'Zhiqing Sun', 'Zheng-Xin Yong', 'Xiangru Tang', 'Vikas Raunak', 'Urmish Thakker', 'Trishala Neeraj', 'Thibault Fevry', 'Stephen H. Bach', 'Srulik Ben-David', 'Sheng Shen', 'Samuel Albanie', 'Ryan Teehan', 'Nihal Nayak', 'Matteo Manica', 'Maged S. Al-shaibani', 'M Saiful Bari', 'Lintang Sutawika', 'Leo Gao', 'Jos Rozen', 'Jason Alan Fries', 'Hendrik Strobelt', 'Harshit Pandey', 'Han Wang', 'Gunjan Chhablani', 'Eliza Szczechla', 'Debajyoti Datta', 'Arnaud Stiegler', 'Antoine Chaffin', 'Andrea Santilli', 'Abheesht Sharma', 'Wilson Y. Lee', 'Sabrina J. Mielke', 'Elizabeth Salesky', 'Chenglei Si', 'Benjamin Heinzerling', 'Zaid Alyafeai', 'Vrinda Prabhu', 'Violette Lepercq', 'Veronika Laippala', 'Vassilina Nikoulina', 'Valentin Danchev', 'Tristan Thrush', 'Timo Schick', 'Tiago Timponi Torrent', 'Sydney Zink', 'Suhas Pai', 'Stanislav Silberberg', 'Somaieh Nikpoor', 'Shayne Longpre', 'Sebastian Nagel', 'Sampo Pyysalo', 'Salomey Osei', 'Rui Ribeiro', 'Roberto Luis López', 'Rishi Bommasani', 'Rheza Harliman', 'Quentin Lhoest', 'Priscilla Amuok', 'Pierre Colombo', 'Peter Henderson', 'Paulo Villegas', 'Ona de Gibert', 'Omar Espejel', 'Olivier Nguyen', 'Nurulaqilla Khamis', 'Nora Kassner', 'Nishant Subramani', 'Mustafa Ghaleb', 'Mohammad A. Jauhar', 'Minh Chien Vu', 'Mike Tian-Jian Jiang', 'Mayank Singh', 'Maximin Coavoux', 'Max Huang', 'Mario Šaško', 'María Grandury', 'Maraim Masoud', 'Manuel Romero Muñoz', 'Manan Dey', 'Ludovic Tanguy', 'Loubna Ben allal', 'Long Phan', 'Leon Weber', 'Leandro von Werra', 'Kyle Lo', 'Kimbo Chen', 'Khalid Almubarak', 'Joydeep Bhattacharjee', 'Joseph Tobing', 'Jörg Frohberg', 'Jonathan Chang', 'Jian Zhu', 'Jesse Dodge', 'Jenny Chim', 'Javier de la Rosa', 'Itziar Gonzalez-Dios', 'Isaac Johnson', 'Idris Abdulmumin', 'Ian Yu', 'Hieu Tran', 'Hamza Benyamina', 'Hady Elsahar', 'Giada Pistilli', 'Germán Kruszewski', 'Gérard Dupont', 'Francesco De Toni', 'Eyal Bar Natan', 'Ethan Kim', 'Efrat Levkovizh', 'Eduardo González Ponferrada', 'Dragomir Radev', 'David Ifeoluwa Adelani', 'Daniel van Strien', 'Chris Emezue', 'Chenghao Mou', 'Natasha Seelam', 'Shanya Sharma', 'Colin Leong', 'Christopher Klamm', 'Ariel Kreisberg Nitzav', 'Adi Simhi', 'Aaron Gokaslan', 'Matthias Gallé', 'Christopher Akiki', 'Teven Le Scao', ':', 'Thomas Wolf', 'Younes Belkada', 'Mathilde Bras', 'Zifan Ye', 'Zhongli Xie', 'Zhe Tan', 'Yu Xu', 'Yingxin Xu', 'Yifan Xu', 'Yash Venkatraman', 'Yash Shailesh Bajaj', 'Yanis Labrak', 'Wojciech Kusa', 'Tomoya Kainuma', 'Théo Gigant', 'Tanmay Laud', 'Sushil Bharati', 'Stefan Schweter', 'Srishti Kumar', 'Sinee Sang-aroonsiri', 'Simon Ott', 'Sid Kiblawi', 'Shubhanshu Mishra', 'Shlok S Deshmukh', 'Samuele Garda', 'Samuel Cahyawijaya', 'Ruisi Su', 'Rosaline Su', 'Rodrigo Canalli', 'Robert Martin', 'Renata Eisenberg', 'Ramya Chandrasekhar', 'Patrick Haller', 'Pascale Fung', 'Myungsun Kang', 'Moritz Freidank', 'Minna Liu', 'Mina Mihaljcic', 'Michiel De Wolf', 'Michael Weinberg', 'Michael Cullan', 'Matthias Samwald', 'Mario Sänger', 'Marianna Nezhurina', 'Maria A Castillo', 'Marc Pàmies', 'Madeleine Hahn de Bykhovetz', 'Luisa Shinzato', 'Lu Liu', 'Lokesh Bulchandani', 'Karthik Rangasai Sivaraman', 'Jose David Posada', 'Jonas Golde', 'John Giorgi', 'Jihyun Kang', 'Ishani Dash', 'Imane Bello', 'Fabio Barth', 'Enrique Manjavacas', 'Dian Yu', 'Daniel Molano', 'Daniel León Periñán', 'Clémentine Fourrier', 'Chuxin Xu', 'Canwen Xu', 'Anna Rogers', 'Amit Alfassy', 'Alham Fikri Aji', 'Aitor Soroa', 'Colin Raffel', 'Margaret Mitchell', 'Julien Launay', 'Yacine Jernite', 'Hugo Laurençon', 'Victor Sanh', 'Pedro Ortiz Suarez', 'Samson Tan', 'Lucile Saulnier', 'Huu Nguyen', 'Iz Beltagy', 'Angelina McMillan-Major', 'Stas Bekman', 'Rachel Bawden', 'Olatunji Ruwase', 'Albert Villanova del Moral', 'Niklas Muennighoff', 'Benoît Sagot', 'Thomas Wang', 'Pawan Sasanka Ammanamanchi', 'Albert Webson', 'Stella Biderman', 'Alexander M. Rush', 'Jonathan Tow', 'François Yvon', 'Alexandra Sasha Luccioni', 'Roman Castagné', 'Daniel Hesslow', 'Suzana Ilić', 'Ellie Pavlick', 'Angela Fan'] | 2022-11-09 | null | null | null | null | ['multilingual-nlp'] | ['natural-language-processing'] | [ 2.98066717e-02 1.32226884e-01 -6.24141753e-01 -2.88979977e-01
-1.08707190e+00 -6.57280803e-01 8.54784727e-01 -4.75242995e-02
-5.36374927e-01 5.33224523e-01 4.71827328e-01 -7.43008494e-01
4.12053972e-01 -4.08134162e-01 -9.00333762e-01 -1.11243263e-01
-1.82780907e-01 7.58250833e-01 3.74049067e-01 -4.45738643e-01
2.89619058e-01 1.38739139e-01 -1.16489983e+00 6.43485725e-01
7.86086082e-01 4.82093990e-01 5.76783419e-01 5.75167716e-01
-2.97833264e-01 1.13233709e+00 -1.54930830e-01 -5.53740442e-01
1.92563400e-01 1.30545840e-01 -1.01906097e+00 -4.94603515e-01
3.73170644e-01 -3.42770159e-01 -2.86606431e-01 6.93099916e-01
4.51497972e-01 1.47457151e-02 1.52686983e-01 -1.15401697e+00
-7.41099238e-01 1.32736456e+00 -3.83713096e-01 3.03900898e-01
4.51664925e-01 3.60806584e-01 1.29843366e+00 -1.02498102e+00
5.84847927e-01 1.30871356e+00 5.86658537e-01 4.15561318e-01
-1.32852519e+00 -9.81245577e-01 6.39800951e-02 -6.90570623e-02
-1.25104058e+00 -9.21822309e-01 4.72968638e-01 -4.06495303e-01
1.85192680e+00 -1.56307578e-01 3.10901374e-01 1.35181797e+00
3.87677610e-01 1.13125479e+00 1.06059992e+00 -6.46657526e-01
-3.07145119e-01 1.45099536e-01 9.84728485e-02 1.01422477e+00
1.27971157e-01 -5.48432069e-03 -7.65269876e-01 -3.19297910e-01
6.36883199e-01 -3.50393683e-01 -3.90639566e-02 -3.03526998e-01
-1.51811922e+00 7.60466337e-01 1.00795977e-01 5.56066096e-01
1.99947488e-02 3.17043126e-01 6.37481034e-01 7.32163250e-01
3.60201031e-01 7.25187063e-01 -7.68533528e-01 -5.57850778e-01
-6.70991361e-01 3.69243592e-01 9.21734512e-01 1.08376610e+00
5.70155203e-01 1.72754470e-02 1.56976342e-01 8.68247151e-01
2.99296111e-01 3.94115895e-01 8.62932622e-01 -1.00680864e+00
6.85273290e-01 5.50045490e-01 -8.49474818e-02 -5.54115832e-01
-5.48511803e-01 -4.28538740e-01 -3.78885180e-01 -2.09138960e-01
5.40749967e-01 1.24871120e-01 -5.22564113e-01 1.85589159e+00
-2.22571224e-01 -1.18998043e-01 -1.97390076e-02 2.55658835e-01
6.73523903e-01 6.82201803e-01 9.55147147e-02 2.03444839e-01
1.06168711e+00 -1.11476874e+00 -2.99671441e-01 -6.12190425e-01
1.16761923e+00 -8.26974213e-01 1.59974980e+00 7.53026605e-01
-1.15068042e+00 -3.13217282e-01 -8.77941310e-01 -5.00371933e-01
-3.79562169e-01 -1.86680749e-01 7.99659073e-01 5.30962527e-01
-1.19392610e+00 1.81153595e-01 -9.93776381e-01 -4.55006182e-01
2.24338472e-01 2.11514562e-01 -3.32952797e-01 -1.48351595e-01
-1.18182921e+00 1.18588018e+00 3.84355158e-01 -4.05832142e-01
-1.09582841e+00 -5.26348293e-01 -7.76356101e-01 -7.91917965e-02
3.80374193e-01 -5.19503534e-01 1.59865475e+00 -7.17035890e-01
-1.47406864e+00 1.17848504e+00 -1.10267706e-01 -6.62193358e-01
3.42482090e-01 -3.69304031e-01 -3.42507988e-01 -3.42764109e-01
1.13796107e-01 6.58111751e-01 4.70488399e-01 -7.81732321e-01
-6.75656319e-01 4.22874168e-02 2.18564078e-01 -2.58128550e-02
-6.23096168e-01 5.39246380e-01 -4.27587122e-01 -6.86509371e-01
-3.88702393e-01 -8.99762094e-01 -8.22109058e-02 -4.74173784e-01
-3.13022405e-01 -5.50510943e-01 4.14083302e-01 -7.27223635e-01
1.31707573e+00 -2.02876925e+00 9.95607674e-02 -2.15907156e-01
3.75152260e-01 2.51215398e-01 -6.51949108e-01 7.21145630e-01
9.69036445e-02 3.39552552e-01 2.56071836e-02 -4.17462021e-01
2.52296388e-01 -4.48517650e-02 -5.58366835e-01 2.52259821e-01
-1.06335431e-01 1.03303576e+00 -9.61072445e-01 -2.87234426e-01
-1.08651243e-01 1.10452980e-01 -7.12435961e-01 1.01686485e-01
-5.13674796e-01 1.64704591e-01 -3.11879039e-01 6.45172954e-01
1.16498232e-01 -5.42611241e-01 1.04645275e-01 2.24921539e-01
-1.44290641e-01 1.01439953e+00 -6.84841156e-01 2.14057565e+00
-8.85633528e-01 7.44254827e-01 1.08155437e-01 -7.12433100e-01
7.91809738e-01 4.11308587e-01 2.58570701e-01 -6.77078128e-01
-3.57644111e-02 4.70079571e-01 2.15708882e-01 -3.39603662e-01
5.86995244e-01 3.76032405e-02 -3.94044667e-01 8.46244514e-01
2.18739435e-01 -3.46679658e-01 3.89155477e-01 4.43036079e-01
1.35028422e+00 6.48519546e-02 4.63985682e-01 -3.99917275e-01
1.90345719e-01 1.20621718e-01 2.30769575e-01 9.31887805e-01
-1.10522524e-01 -4.13103141e-02 2.47358397e-01 -5.48601210e-01
-1.16452885e+00 -9.53100562e-01 -1.31369717e-02 1.77482665e+00
-4.46134299e-01 -8.78180921e-01 -4.72035229e-01 -3.40510279e-01
-3.08951270e-02 8.84234369e-01 -1.58585146e-01 -1.10205412e-01
-7.08447874e-01 -3.49306375e-01 1.02409613e+00 4.33703303e-01
4.54982042e-01 -1.12191188e+00 -5.61763406e-01 2.55752742e-01
-3.37344646e-01 -1.24163866e+00 -4.65766579e-01 3.54250610e-01
-6.85656071e-01 -7.73577452e-01 -4.33144033e-01 -1.02109826e+00
6.99417889e-02 2.19396666e-01 1.53705275e+00 1.30914897e-01
5.16094342e-02 2.40954205e-01 -2.20208615e-01 -3.57320905e-01
-8.35579216e-01 7.71614730e-01 1.59036309e-01 -6.56657279e-01
4.43187475e-01 -6.79553628e-01 3.10716867e-01 1.15899771e-01
-7.00082958e-01 4.18019176e-01 6.97420657e-01 7.05573976e-01
4.72939834e-02 -3.38252485e-01 5.35926819e-01 -1.02106810e+00
8.43929172e-01 -4.80576485e-01 -8.12550545e-01 2.26340860e-01
-7.01058984e-01 2.19532251e-01 7.18981326e-01 -3.29829395e-01
-9.25383568e-01 -1.81602016e-01 -1.99098423e-01 8.38799253e-02
4.07800265e-02 7.50745535e-01 9.04768184e-02 -1.27886474e-01
8.21936727e-01 3.64592314e-01 -2.94503957e-01 -5.13713121e-01
4.92776871e-01 6.21605814e-01 3.77104700e-01 -1.10690475e+00
7.71282494e-01 -2.14586049e-01 -5.74062765e-01 -7.28359878e-01
-7.40453899e-01 -2.41845503e-01 -3.15210998e-01 2.26200566e-01
4.57495570e-01 -1.29770780e+00 -5.75426817e-01 3.84023756e-01
-1.12213862e+00 -1.02737391e+00 1.33857384e-01 2.81182379e-01
-4.94699240e-01 2.38621116e-01 -9.16306317e-01 -4.25465941e-01
-2.38958597e-01 -1.33186030e+00 9.66261566e-01 -1.06701642e-01
-5.12511373e-01 -1.02662206e+00 2.51063287e-01 4.15921926e-01
6.86321020e-01 -4.72023308e-01 1.28128242e+00 -8.15566599e-01
-7.32083976e-01 1.46638267e-02 -7.69990906e-02 2.81533748e-01
-3.95661928e-02 -4.80811633e-02 -8.40793014e-01 -4.78931785e-01
-3.20770979e-01 -9.52236831e-01 8.34005713e-01 -5.31883463e-02
9.92327809e-01 -2.86223739e-01 -4.03937519e-01 5.93080044e-01
1.13605952e+00 -2.75687482e-02 2.82226741e-01 5.51637530e-01
6.26717150e-01 2.65817970e-01 1.18753336e-01 2.63033152e-01
8.40435207e-01 4.61236268e-01 1.24318413e-01 1.41319618e-01
-5.51282950e-02 -5.69129050e-01 8.29614460e-01 1.32227886e+00
9.78772044e-02 -1.82340086e-01 -1.44442284e+00 5.72561920e-01
-1.38920224e+00 -5.98689735e-01 3.82419601e-02 1.94568861e+00
1.25639606e+00 5.94271541e-01 1.38189673e-01 -4.92422223e-01
5.65386228e-02 3.09863716e-01 -5.32321334e-01 -6.13988578e-01
-2.72776842e-01 3.02990079e-01 5.30069470e-01 5.77807307e-01
-7.23658741e-01 1.45785606e+00 7.55479383e+00 8.64946425e-01
-1.19584787e+00 3.19013208e-01 3.92724305e-01 -8.09147879e-02
-5.59671879e-01 1.38987392e-01 -8.61054480e-01 1.71018988e-01
1.34430933e+00 -7.13407159e-01 8.68060589e-01 8.09065998e-01
-7.22619286e-03 -3.55056934e-02 -1.34753895e+00 8.02812099e-01
-1.02329127e-01 -1.34451699e+00 -2.97707655e-02 -1.53141677e-01
6.70780420e-01 9.56907153e-01 2.18116224e-01 7.83681154e-01
1.01164711e+00 -1.16840398e+00 8.08447480e-01 1.92475021e-01
8.28150749e-01 -3.45614851e-01 1.49124950e-01 8.65521967e-01
-1.05663157e+00 -2.15892345e-01 -1.93706572e-01 -4.19429660e-01
-8.06393847e-02 2.97126353e-01 -8.95065725e-01 -5.17043471e-02
7.50874937e-01 6.92597806e-01 -6.83392227e-01 6.21145070e-01
-2.76803970e-01 8.85653377e-01 -3.60604197e-01 -1.05112813e-01
1.22937210e-01 3.43535453e-01 4.86907184e-01 1.45991266e+00
-9.84488651e-02 -2.82177508e-01 4.20049220e-01 9.03150678e-01
-3.63046259e-01 5.47499992e-02 -8.46498609e-01 -5.38616478e-01
5.70924044e-01 1.12848449e+00 -3.32204521e-01 -3.61700743e-01
-7.64246583e-01 5.77816248e-01 6.06597066e-01 2.74783611e-01
-7.03832507e-01 -2.79723048e-01 5.68759084e-01 1.39942303e-01
-1.68415874e-01 -6.93560362e-01 -1.81504577e-01 -1.45969868e+00
-2.00578406e-01 -1.55502236e+00 2.27543414e-01 -7.68454194e-01
-1.09346616e+00 6.54852092e-01 1.33269385e-01 -6.32319510e-01
-5.40690899e-01 -7.26915181e-01 -2.39975408e-01 8.04873645e-01
-1.41729808e+00 -1.20072997e+00 1.64996654e-01 3.73409390e-01
7.25368917e-01 -5.63914120e-01 1.00128937e+00 5.14520049e-01
-4.22522515e-01 7.81565309e-01 6.52349293e-02 1.88452184e-01
1.00790298e+00 -9.38361287e-01 1.03276551e+00 8.53784978e-01
3.43460053e-01 1.06623554e+00 6.52785003e-01 -4.54147935e-01
-1.55867624e+00 -8.08476985e-01 1.40835583e+00 -7.41446912e-01
1.15140271e+00 -7.47542799e-01 -7.29685366e-01 1.46509969e+00
4.73662823e-01 -3.84876370e-01 5.30901015e-01 3.00279707e-01
-6.42282963e-01 3.81325334e-02 -6.06427729e-01 5.75458825e-01
1.13708603e+00 -8.95415068e-01 -7.53154278e-01 4.78413403e-01
9.66427624e-01 -4.65345025e-01 -7.94169486e-01 2.20827326e-01
5.77349544e-01 -8.10038030e-01 7.17312753e-01 -6.84440553e-01
6.07519925e-01 1.49093404e-01 -4.24808860e-01 -1.28239846e+00
-3.81664544e-01 -9.28114116e-01 2.63820235e-02 1.09329021e+00
7.28475332e-01 -9.04227734e-01 3.82799208e-01 5.25899470e-01
-3.71589780e-01 -5.97291172e-01 -6.99377179e-01 -8.19508135e-01
5.24996340e-01 -8.41283798e-01 4.96545643e-01 1.02108932e+00
3.29428941e-01 6.51235640e-01 -2.49123722e-01 -3.77123863e-01
2.24625245e-01 1.04756178e-02 1.01531303e+00 -9.43831861e-01
-5.15182853e-01 -7.03012407e-01 2.84502119e-01 -1.38127112e+00
5.23255348e-01 -1.34708190e+00 -2.16552466e-02 -1.34182870e+00
3.51411343e-01 -7.51055837e-01 -1.20056815e-01 9.32058990e-01
1.90418482e-01 3.75292301e-02 1.22946255e-01 2.95176119e-01
-6.80900097e-01 3.22606266e-01 9.21018124e-01 -2.79700071e-01
-5.43819070e-02 -1.64194912e-01 -1.01508272e+00 9.29031968e-01
6.79295480e-01 -4.66791302e-01 -2.80857861e-01 -1.09164143e+00
7.25478888e-01 -1.56539410e-01 -6.12642951e-02 -8.94041121e-01
2.34748647e-01 -1.74883559e-01 -1.81141749e-01 -2.12407440e-01
1.87376097e-01 -4.32772964e-01 -1.08298562e-01 4.05527800e-01
-7.97406077e-01 5.33691823e-01 3.79815906e-01 2.15820060e-03
-8.96826088e-02 3.55938822e-02 5.92151821e-01 -3.52936804e-01
-9.43394005e-01 1.03323102e-01 -5.57678223e-01 3.18442076e-01
5.83296180e-01 2.91921377e-01 -5.22968590e-01 -3.98939967e-01
-1.59154683e-01 2.66596735e-01 6.11731231e-01 6.44640148e-01
2.70573974e-01 -1.06525445e+00 -7.98354089e-01 1.56768993e-01
3.06524485e-01 -1.99593186e-01 -3.61380309e-01 8.87395263e-01
-4.35980886e-01 7.84694850e-01 -4.70552184e-02 -2.87200958e-01
-9.15933609e-01 4.08984482e-01 2.18068764e-01 -6.21345401e-01
-6.45932436e-01 1.03080678e+00 1.94913939e-01 -7.89339364e-01
2.85929799e-01 -7.92326868e-01 1.52479753e-01 -4.86914366e-01
4.83596474e-01 2.30169460e-01 -1.39175607e-02 -4.51097578e-01
-3.61923635e-01 9.72324982e-02 -2.69015938e-01 -2.05481440e-01
1.37722909e+00 4.39633690e-02 -3.44309360e-01 8.01934719e-01
1.08464932e+00 3.14099133e-01 -8.02178264e-01 -6.28649890e-01
2.21293792e-01 -9.04087052e-02 -4.58259918e-02 -8.83539915e-01
-6.57852411e-01 8.69105756e-01 1.22018360e-01 -2.76817419e-02
9.80741203e-01 8.91647041e-02 1.14187610e+00 8.40992093e-01
7.69303620e-01 -9.64959443e-01 1.45848334e-01 1.15442133e+00
6.20484829e-01 -1.16368151e+00 -3.29835504e-01 7.28475824e-02
-4.01707709e-01 9.13497269e-01 5.45178711e-01 1.10726627e-02
5.18026829e-01 7.13842154e-01 -2.83771064e-02 9.60745066e-02
-1.33820760e+00 2.23743081e-01 3.01836133e-02 4.54527020e-01
9.43187177e-01 1.23102814e-01 -1.00776099e-01 7.78684497e-01
-4.37949449e-01 1.69653296e-01 4.60353076e-01 8.97522569e-01
-5.47056973e-01 -1.40266430e+00 -4.03637290e-02 4.64577019e-01
-4.70717281e-01 -6.66898072e-01 -2.37412572e-01 7.18001902e-01
-2.02562913e-01 8.15092802e-01 -2.47080028e-01 -4.81620967e-01
1.06430076e-01 2.33691439e-01 5.83127141e-01 -8.93349051e-01
-7.34141529e-01 -1.95227638e-01 3.98072213e-01 -6.01060808e-01
-1.78835616e-02 -4.55792814e-01 -1.19417214e+00 -5.95448613e-01
-6.47131726e-02 -9.11011919e-02 4.83463734e-01 1.03091204e+00
3.36387515e-01 2.35225320e-01 2.36243919e-01 -5.96296191e-01
-8.33942711e-01 -1.08061564e+00 -3.24161112e-01 -3.73968668e-02
2.85267144e-01 -4.50620592e-01 -1.31134078e-01 1.32237673e-01] | [10.632351875305176, 8.347243309020996] |
43ce812a-fd4c-4afe-864b-60377cb85487 | self-supervised-learning-for-organs-at-risk | 2305.02491 | null | https://arxiv.org/abs/2305.02491v1 | https://arxiv.org/pdf/2305.02491v1.pdf | Self-Supervised Learning for Organs At Risk and Tumor Segmentation with Uncertainty Quantification | In this study, our goal is to show the impact of self-supervised pre-training of transformers for organ at risk (OAR) and tumor segmentation as compared to costly fully-supervised learning. The proposed algorithm is called Monte Carlo Transformer based U-Net (MC-Swin-U). Unlike many other available models, our approach presents uncertainty quantification with Monte Carlo dropout strategy while generating its voxel-wise prediction. We test and validate the proposed model on both public and one private datasets and evaluate the gross tumor volume (GTV) as well as nearby risky organs' boundaries. We show that self-supervised pre-training approach improves the segmentation scores significantly while providing additional benefits for avoiding large-scale annotation costs. | ['Ulas Bagci', 'Damla Turgut', 'Mohamed Abazeed', 'Bulent Aydogan', 'Patrick Kelly', 'Justin Rineer', 'Curtis Lisle', 'Debesh Jha', 'Ilkin Isler'] | 2023-05-04 | null | null | null | null | ['tumor-segmentation'] | ['computer-vision'] | [-3.77167873e-02 7.74886072e-01 -5.67363203e-01 -3.90659809e-01
-1.60254264e+00 -4.82267946e-01 4.89331007e-01 5.46374083e-01
-2.92196363e-01 1.20401835e+00 4.47744250e-01 -6.37893021e-01
-2.90057026e-02 -9.40222263e-01 -9.63721454e-01 -5.65261304e-01
-1.37669519e-01 6.99744403e-01 3.77337903e-01 4.64451700e-01
-3.54803145e-01 2.05443978e-01 -3.51460606e-01 9.58039016e-02
1.34719002e+00 1.12381959e+00 -1.71357334e-01 3.98407221e-01
2.83299237e-01 8.53497922e-01 -1.41325846e-01 -6.73990130e-01
3.38288218e-01 -1.81817561e-01 -9.47828829e-01 -1.36113986e-01
2.45671436e-01 -3.51775587e-01 -1.34434119e-01 9.45136011e-01
5.13058245e-01 -2.31088966e-01 9.56657588e-01 -1.01078892e+00
-2.82603968e-02 1.20545924e+00 -4.39933091e-01 2.19660863e-01
-2.33639419e-01 4.65215355e-01 8.18029225e-01 -1.90432444e-01
6.49816096e-01 5.33007264e-01 1.25703466e+00 3.77319068e-01
-1.17015898e+00 -5.74582279e-01 -3.12876254e-01 -4.73978579e-01
-1.36020648e+00 7.28113800e-02 3.05884808e-01 -5.58357477e-01
8.87611628e-01 1.20271154e-01 6.33078933e-01 1.01011693e+00
7.11105525e-01 7.77211428e-01 1.05669844e+00 -1.97581440e-01
3.72186661e-01 3.73867124e-01 9.29471254e-02 8.97819579e-01
4.88705486e-01 3.92898500e-01 8.42662826e-02 -2.04125792e-01
7.75083065e-01 -1.33049667e-01 1.43174510e-02 -4.02293950e-01
-1.09419298e+00 7.10295916e-01 8.90603006e-01 1.68159798e-01
-3.33241284e-01 7.85285354e-01 6.45593822e-01 -2.83115894e-01
7.93854296e-01 -7.46824071e-02 -3.48136425e-01 9.22442228e-02
-1.26971686e+00 -2.01885298e-01 4.21917677e-01 9.79553044e-01
1.87310800e-01 -4.53594625e-02 -8.01931500e-01 2.89153844e-01
4.74617273e-01 1.80018753e-01 5.55451870e-01 -4.58441854e-01
2.18556404e-01 3.93360227e-01 -4.38487865e-02 2.15068996e-01
-4.58718896e-01 -9.90962446e-01 -9.24531281e-01 1.65454313e-01
7.33767748e-01 -6.30788267e-01 -1.43982720e+00 1.48839891e+00
3.87526393e-01 3.51122528e-01 -8.72488394e-02 4.76172209e-01
8.58016014e-01 1.21014886e-01 5.29398263e-01 1.50285631e-01
1.32892680e+00 -9.87846732e-01 -4.08279836e-01 1.01202451e-01
1.11708844e+00 -4.12765801e-01 7.92175591e-01 5.11295982e-02
-1.15945065e+00 1.70480847e-01 -5.91572583e-01 2.49705046e-01
-1.01895921e-01 3.19870710e-01 9.71130788e-01 1.37685394e+00
-1.02069259e+00 7.17934549e-01 -1.15963554e+00 -2.27903664e-01
1.26080441e+00 4.02496755e-01 6.28022011e-03 2.54381835e-01
-9.31390047e-01 1.08200192e+00 2.61672676e-01 -4.01204050e-01
-1.95181072e+00 -1.33399093e+00 -7.81919181e-01 1.74453914e-01
4.85783339e-01 -7.37609267e-01 1.41887426e+00 -5.12511909e-01
-1.31556356e+00 7.44089544e-01 2.33029559e-01 -1.17522097e+00
1.28334963e+00 2.35495403e-01 2.63908952e-01 -1.47179708e-01
7.31309131e-02 8.20141137e-01 2.71502435e-01 -1.03146195e+00
-3.90785694e-01 -2.84954488e-01 -2.82804847e-01 1.48476690e-01
2.38832980e-02 -3.68049264e-01 -1.30181730e-01 -6.95941746e-01
-2.26562649e-01 -6.60533905e-01 -7.09511697e-01 7.21647739e-02
-7.37302840e-01 1.49176151e-01 1.91298172e-01 -7.74524927e-01
7.42193222e-01 -1.66157413e+00 -6.06517136e-01 6.17298424e-01
2.16641083e-01 -1.89527556e-01 4.51145411e-01 -3.37501377e-01
-4.49352004e-02 3.25604111e-01 -5.12178183e-01 -3.83600324e-01
-1.82210013e-01 2.36296251e-01 3.29168081e-01 6.67209148e-01
-4.09507230e-02 1.32576537e+00 -9.17116523e-01 -1.11371255e+00
2.65592396e-01 3.71737003e-01 -7.17582881e-01 -2.30891168e-01
-3.21592093e-01 7.14288950e-01 -3.84723544e-01 9.71010089e-01
6.81175888e-01 -2.35866025e-01 1.24148168e-01 -1.58657402e-01
2.53276169e-01 3.30967486e-01 -6.39076412e-01 1.53247297e+00
-5.93086302e-01 5.97059019e-02 -2.82804251e-01 -6.33251369e-01
4.13278222e-01 4.22260195e-01 6.64278805e-01 -5.57358921e-01
5.31533718e-01 9.19632912e-02 3.94760072e-02 1.13786295e-01
-1.47467246e-02 -5.41370630e-01 -4.17730957e-02 1.03696562e-01
3.20120484e-01 -1.04169808e-01 -3.11426789e-01 3.34431887e-01
1.18617451e+00 2.31323838e-01 4.70686674e-01 -6.37489438e-01
1.33200929e-01 2.56093532e-01 5.74294269e-01 7.63881683e-01
-5.59259713e-01 5.59248865e-01 7.90287018e-01 1.70975298e-01
-8.68091166e-01 -1.50244176e+00 -6.15690708e-01 5.24280369e-01
-2.13520467e-01 -1.57958474e-02 -1.01083124e+00 -1.33847880e+00
2.66294479e-01 9.77834582e-01 -9.47374940e-01 6.06856979e-02
-1.81625217e-01 -1.12581229e+00 9.04372036e-01 8.99077117e-01
3.57587010e-01 -5.97705662e-01 -4.37052310e-01 9.13680270e-02
5.05097099e-02 -9.73220110e-01 -3.89128745e-01 5.70881426e-01
-1.49977231e+00 -9.56630707e-01 -1.07825947e+00 -2.39078566e-01
1.01811111e+00 -6.78144455e-01 1.26245558e+00 -3.69509906e-01
-5.47135592e-01 3.01932782e-01 -6.42322004e-02 -5.66638470e-01
-5.45979440e-01 1.27514556e-01 -4.39754993e-01 -4.07654315e-01
-2.02254578e-01 -3.15718532e-01 -6.55745745e-01 2.47653916e-01
-5.78046501e-01 -1.35365516e-01 6.98595226e-01 8.49427104e-01
9.94979322e-01 3.94076444e-02 2.97750622e-01 -1.30583990e+00
2.75326937e-01 -5.31584322e-01 -8.75463963e-01 4.43705857e-01
-8.22297573e-01 -4.81495745e-02 3.06502879e-01 -5.33384085e-02
-1.25420702e+00 3.16127956e-01 -2.08654761e-01 -2.47552991e-01
-5.51577993e-02 3.25781584e-01 1.33527458e-01 -3.94467622e-01
8.31062138e-01 -8.74001086e-02 -2.06139728e-01 -1.66031435e-01
2.58743376e-01 9.78772044e-02 2.42715433e-01 -6.10567868e-01
6.07485533e-01 6.19046450e-01 2.03209952e-01 -1.08915858e-01
-8.35286796e-01 -1.39682204e-01 -5.72770953e-01 -3.07560295e-01
8.17582011e-01 -9.77922082e-01 -5.58378041e-01 1.93090037e-01
-5.63768804e-01 -6.05578840e-01 -1.03376472e+00 6.80292070e-01
-4.96002346e-01 1.67994842e-01 -7.43063629e-01 -8.88357341e-01
-6.73096180e-01 -1.29029703e+00 9.18725669e-01 1.37036875e-01
1.90284787e-04 -1.25231874e+00 1.03686601e-01 3.73712242e-01
6.47110581e-01 5.37809312e-01 5.36884665e-01 -6.96074605e-01
-8.40332508e-01 -2.07710862e-01 -2.81737864e-01 2.19813392e-01
-1.48001254e-01 -2.15255484e-01 -1.09088767e+00 -1.49361908e-01
-3.53037745e-01 -4.09785330e-01 9.49540317e-01 1.21511126e+00
1.27969241e+00 -2.64621843e-02 -7.03370810e-01 6.78413987e-01
1.69694853e+00 -1.71610624e-01 7.35489309e-01 7.83921182e-02
5.07039964e-01 1.09878875e-01 5.32746196e-01 4.92532998e-01
1.57680064e-01 3.32386233e-02 8.06179762e-01 -2.74747849e-01
-2.14314729e-01 -4.10182893e-01 -9.87167656e-02 -5.91533538e-03
-3.38754021e-02 -1.32559225e-01 -1.35388386e+00 8.10410440e-01
-1.48628187e+00 -4.06309724e-01 -1.48468614e-01 2.27488732e+00
8.33265960e-01 5.25742292e-01 7.15701208e-02 -2.87288696e-01
5.67727923e-01 -2.55121022e-01 -5.51977277e-01 -8.48309547e-02
2.97257155e-01 4.41730171e-01 1.32617831e+00 5.91988802e-01
-1.10958898e+00 7.78973818e-01 7.18830872e+00 1.14640820e+00
-7.01207995e-01 6.87929749e-01 1.60096622e+00 -1.85746744e-01
-3.48118782e-01 1.35412648e-01 -7.79782891e-01 2.75147319e-01
1.06176388e+00 -1.77031979e-01 -2.18543276e-01 1.02957118e+00
1.52354360e-01 -6.05499208e-01 -1.12815416e+00 3.97811949e-01
-3.12319189e-01 -1.61055350e+00 -1.92623958e-01 1.81875557e-01
8.80685389e-01 3.91504556e-01 6.95374161e-02 2.93122649e-01
9.49398339e-01 -1.19105208e+00 4.93303984e-01 3.32682282e-01
1.10564125e+00 -7.12061644e-01 1.09127700e+00 1.99441299e-01
-9.20410156e-01 2.51114994e-01 -2.26583540e-01 5.53579628e-01
1.87601835e-01 1.11412203e+00 -1.45258498e+00 6.94943190e-01
5.24295330e-01 2.66829133e-01 -5.78473926e-01 1.41905272e+00
1.03380403e-03 1.12846637e+00 -6.74366117e-01 1.24620311e-01
4.00220186e-01 1.10774890e-01 2.44433656e-01 1.19194317e+00
3.31316113e-01 -1.61749542e-01 -2.56681442e-01 1.13312435e+00
-3.25944871e-01 8.79364908e-02 -1.92703113e-01 1.92760542e-01
1.32520199e-01 1.36126351e+00 -1.08201337e+00 -5.45225918e-01
8.17248318e-03 6.33005023e-01 -4.58247550e-02 -3.08024704e-01
-1.38467085e+00 2.33323604e-01 -1.92766652e-01 3.81078541e-01
1.54054016e-01 4.80480075e-01 -8.58463764e-01 -8.97006273e-01
-4.38820153e-01 -1.30204096e-01 7.51163244e-01 -4.56894636e-01
-1.28481698e+00 3.13366830e-01 3.44398394e-02 -1.28605747e+00
1.22724257e-01 -3.01646680e-01 -9.61144924e-01 8.04616272e-01
-1.62163520e+00 -1.51620007e+00 -3.10199052e-01 3.97957832e-01
1.39995813e-01 1.48242503e-01 5.61497569e-01 3.55592519e-01
-4.34425354e-01 1.06935716e+00 1.61378399e-01 2.74258375e-01
5.12497246e-01 -1.52142537e+00 -6.76112901e-03 5.41832983e-01
-3.76199365e-01 -1.53169157e-02 3.83607030e-01 -1.15811205e+00
-7.47289121e-01 -1.45503533e+00 4.60566819e-01 -4.42591965e-01
6.12077951e-01 -2.55176008e-01 -2.33852878e-01 9.71789122e-01
1.78416222e-01 4.38403547e-01 7.13303864e-01 -1.24118157e-01
2.01127499e-01 -5.34349866e-02 -1.96794450e+00 2.12350339e-01
8.11091423e-01 -1.49853539e-03 -2.23429203e-01 6.57160640e-01
5.66683233e-01 -7.79227078e-01 -1.28997529e+00 6.52469099e-01
2.00244665e-01 -8.61136079e-01 9.08767581e-01 -3.28200787e-01
4.86530691e-01 1.70811787e-01 1.21481009e-01 -1.24987650e+00
-3.74942981e-02 -3.85998249e-01 1.66526794e-01 1.18650532e+00
7.99375296e-01 -5.52930832e-01 1.53110397e+00 6.34507477e-01
-2.79909015e-01 -9.09086049e-01 -1.18655980e+00 -7.29108512e-01
6.11561775e-01 -4.87548143e-01 3.91670883e-01 8.96205127e-01
-1.16825156e-01 -3.57603490e-01 -1.95173044e-02 2.69498944e-01
1.21286845e+00 -4.28808749e-01 2.13104039e-01 -8.52165818e-01
-4.16120529e-01 -2.53655255e-01 -4.47246939e-01 -2.14516371e-01
-2.30536476e-01 -1.17748666e+00 -1.24004468e-01 -1.56246090e+00
5.26228309e-01 -6.02334857e-01 -3.59316766e-01 6.40159309e-01
9.85640734e-02 2.65502512e-01 -3.11067134e-01 -1.45243794e-01
-3.55240434e-01 5.35093546e-01 1.32058942e+00 -2.20177248e-01
1.73202425e-01 1.44329622e-01 -3.35143745e-01 8.22929263e-01
6.87470257e-01 -7.39823103e-01 -2.86893427e-01 1.38152391e-01
-3.02242666e-01 4.87768263e-01 5.78503430e-01 -1.18769491e+00
1.57880392e-02 6.05105460e-02 7.01799333e-01 -9.39913154e-01
-2.39388704e-01 -8.46658587e-01 7.53673837e-02 9.21248138e-01
-3.10257405e-01 -5.22102535e-01 3.44899803e-01 5.29327095e-01
1.05562545e-01 -4.02342439e-01 1.09034359e+00 -4.56853449e-01
9.27598104e-02 3.71601075e-01 -3.09077501e-01 1.78742081e-01
1.63801885e+00 -1.21406905e-01 -1.09543055e-01 -1.93654478e-01
-1.14595830e+00 5.32691360e-01 2.14110747e-01 -6.42925560e-01
2.01377288e-01 -1.00879693e+00 -6.15570605e-01 -2.97445565e-01
1.08522639e-01 2.98135966e-01 5.26526511e-01 1.12340283e+00
-7.63704538e-01 5.16338408e-01 -5.78907020e-02 -6.75851345e-01
-9.49195623e-01 2.00329885e-01 1.00117457e+00 -1.05619550e+00
-7.72319674e-01 1.14777935e+00 1.53535262e-01 -6.93120658e-01
4.41760063e-01 -8.11654806e-01 2.12191358e-01 -4.04825211e-01
-2.35004872e-01 3.38632882e-01 1.17229879e-01 -8.20469633e-02
-2.90929914e-01 -1.46109805e-01 -5.14104925e-02 -1.50404543e-01
1.16878307e+00 1.43239290e-01 2.16514781e-01 5.90706989e-02
8.95561993e-01 -1.27764687e-01 -1.36699140e+00 -1.11316644e-01
-8.50712880e-02 -1.62883371e-01 5.84218740e-01 -1.45949841e+00
-1.36402428e+00 7.01350212e-01 1.04364300e+00 -4.64617103e-01
8.53102207e-01 1.17945924e-01 7.33208537e-01 -1.18480034e-01
4.17378008e-01 -9.29749370e-01 -2.93359965e-01 -1.05832731e-02
1.13806121e-01 -1.33744550e+00 2.13929713e-01 -7.93061554e-01
-8.91916931e-01 4.31531787e-01 6.48255765e-01 -3.86750847e-01
9.56016600e-01 8.02414954e-01 -8.99130628e-02 -1.08109362e-01
-4.83868808e-01 -2.41099715e-01 2.12793484e-01 6.16398215e-01
5.04740179e-01 3.78125340e-01 -1.08172990e-01 7.15389967e-01
-9.64140370e-02 4.67975318e-01 4.45157886e-01 8.52346718e-01
-4.37312424e-02 -8.65704894e-01 -2.86757141e-01 1.07568467e+00
-8.61243784e-01 -2.78059781e-01 -1.33321047e-01 7.87612498e-01
1.43529382e-02 4.55655344e-02 -1.58150658e-01 3.67603339e-02
-3.32500152e-02 7.67344190e-03 6.63491905e-01 -4.23671722e-01
-1.07900012e+00 2.72528857e-01 1.97335795e-01 -5.08375823e-01
-3.90887409e-02 -6.04803085e-01 -1.39002049e+00 2.65691783e-02
-6.33203745e-01 4.28409688e-02 6.69710875e-01 7.75127769e-01
-9.47352722e-02 1.08543122e+00 3.37807804e-01 -3.56744647e-01
-8.07244480e-01 -8.75916898e-01 -6.84909105e-01 2.32708324e-02
-7.10094869e-02 -7.00920820e-01 -2.73822069e-01 -2.37403944e-01] | [14.671676635742188, -2.3034050464630127] |
46fbaa29-ab08-4516-a65b-82caebe7724f | interacting-hand-object-pose-estimation-via | 2211.08805 | null | https://arxiv.org/abs/2211.08805v1 | https://arxiv.org/pdf/2211.08805v1.pdf | Interacting Hand-Object Pose Estimation via Dense Mutual Attention | 3D hand-object pose estimation is the key to the success of many computer vision applications. The main focus of this task is to effectively model the interaction between the hand and an object. To this end, existing works either rely on interaction constraints in a computationally-expensive iterative optimization, or consider only a sparse correlation between sampled hand and object keypoints. In contrast, we propose a novel dense mutual attention mechanism that is able to model fine-grained dependencies between the hand and the object. Specifically, we first construct the hand and object graphs according to their mesh structures. For each hand node, we aggregate features from every object node by the learned attention and vice versa for each object node. Thanks to such dense mutual attention, our method is able to produce physically plausible poses with high quality and real-time inference speed. Extensive quantitative and qualitative experiments on large benchmark datasets show that our method outperforms state-of-the-art methods. The code is available at https://github.com/rongakowang/DenseMutualAttention.git. | ['Hongdong Li', 'Wei Mao', 'Rong Wang'] | 2022-11-16 | null | null | null | null | ['hand-object-pose'] | ['computer-vision'] | [-2.92480677e-01 1.00083221e-02 -9.70880315e-02 -2.40544267e-02
-4.94943082e-01 -4.07539010e-01 5.59315503e-01 -1.47326529e-01
-1.20145805e-01 4.62534726e-01 6.31235391e-02 1.81730330e-01
-3.49800646e-01 -7.18213737e-01 -9.30103898e-01 -6.77778542e-01
1.99471921e-01 1.05803359e+00 2.93693811e-01 6.53131083e-02
2.24293023e-01 7.46576250e-01 -1.50810349e+00 2.18892526e-02
9.07031596e-01 9.06779230e-01 6.30204022e-01 5.93684852e-01
1.33628562e-01 5.70719838e-01 -5.86422533e-02 -2.21555531e-01
2.06149772e-01 -9.44270194e-02 -9.79034305e-01 1.95404902e-01
5.03131151e-01 -4.40975875e-01 -6.29373252e-01 9.14020956e-01
5.07356882e-01 1.65105134e-01 6.83068931e-01 -1.05209494e+00
-5.67336261e-01 4.29441273e-01 -7.95539916e-01 -1.87641233e-01
3.63526553e-01 3.76350582e-01 1.35438502e+00 -9.96651173e-01
8.06473434e-01 1.26393306e+00 1.95119932e-01 3.71636093e-01
-1.35354829e+00 -5.86006045e-01 4.64740217e-01 4.67509359e-01
-1.48990047e+00 -2.33219057e-01 1.11493444e+00 -5.53636372e-01
7.00647175e-01 2.88842738e-01 9.50250924e-01 1.05773067e+00
1.86629176e-01 9.59801614e-01 7.75405407e-01 -3.92880976e-01
-2.20251409e-03 -1.74779609e-01 9.54803377e-02 8.53505015e-01
7.41304904e-02 1.54819321e-02 -5.16160429e-01 -5.46509549e-02
1.19875705e+00 2.08211258e-01 -3.70765150e-01 -5.98174393e-01
-1.28968716e+00 6.01421297e-01 9.31947052e-01 2.45004401e-01
-5.89094043e-01 4.45140272e-01 -1.61145523e-01 -2.83470541e-01
4.44735795e-01 1.99262485e-01 -2.83639103e-01 1.27078384e-01
-5.14846742e-01 6.31746769e-01 7.19536364e-01 8.08913827e-01
6.81133628e-01 -4.10891742e-01 -2.22602844e-01 4.73134369e-01
6.55459881e-01 5.04264653e-01 -2.43317857e-01 -9.38318133e-01
2.88282365e-01 5.56730747e-01 2.34729603e-01 -1.12559688e+00
-2.96825677e-01 -4.45487380e-01 -6.11268640e-01 5.11792660e-01
5.92501700e-01 2.27482527e-01 -1.01838458e+00 1.47082198e+00
6.24774456e-01 2.47926652e-01 -6.98121905e-01 1.15667629e+00
6.30065560e-01 3.31313223e-01 -4.60794531e-02 1.19103439e-01
1.26648521e+00 -1.11026549e+00 -6.23122334e-01 -2.76516587e-01
1.87780224e-02 -7.64012814e-01 1.14315212e+00 3.43594581e-01
-1.28672493e+00 -6.26685143e-01 -6.05963349e-01 -2.73014486e-01
-9.10464395e-03 2.05658786e-02 6.32406771e-01 -1.05620408e-02
-4.97866839e-01 7.07073629e-01 -1.19584525e+00 -1.24793835e-01
6.16234958e-01 4.29788768e-01 -3.34288508e-01 -1.75664127e-01
-5.50471544e-01 8.44484925e-01 8.96474421e-02 4.19726193e-01
-6.90811396e-01 -6.39094889e-01 -7.35773087e-01 -7.46403709e-02
6.22565091e-01 -9.23290312e-01 1.12076294e+00 -7.01411068e-01
-1.50520313e+00 7.20246196e-01 -1.62655100e-01 -1.31229702e-02
7.71633685e-01 -6.30851984e-01 5.72068930e-01 8.93008783e-02
-1.28121525e-01 6.58577263e-01 8.32656384e-01 -1.65223050e+00
-2.43843257e-01 -7.05847740e-01 1.72278002e-01 2.72893935e-01
-1.25096897e-02 -3.28909874e-01 -8.73555779e-01 -6.03493333e-01
1.98908523e-01 -9.67141688e-01 -2.01011494e-01 3.03263634e-01
-6.68568432e-01 -5.68757296e-01 9.08359826e-01 -7.40984499e-01
8.17976296e-01 -1.87468445e+00 8.84366393e-01 2.57266819e-01
5.58776498e-01 7.20914751e-02 -2.56756302e-02 1.89171255e-01
3.29434752e-01 -1.26921818e-01 -4.54333462e-02 -6.40991330e-01
4.97173592e-02 1.93401664e-01 -7.99144059e-02 6.74593270e-01
1.21073715e-01 1.28036177e+00 -8.79207432e-01 -6.26240790e-01
5.95064700e-01 8.95906568e-01 -8.43239486e-01 4.35452729e-01
-7.10467041e-01 8.63054097e-01 -7.20442474e-01 6.63453400e-01
4.58287776e-01 -5.17682314e-01 1.23439908e-01 -5.29189408e-01
2.78255455e-02 1.95057347e-01 -1.18420386e+00 1.90521717e+00
-5.05919516e-01 2.12948427e-01 1.49988323e-01 -7.03945100e-01
4.49918181e-01 2.05622807e-01 4.35831875e-01 -3.90390962e-01
4.66368586e-01 1.07363924e-01 2.07236297e-02 -3.50486934e-01
-5.00519201e-02 1.16965979e-01 4.51896131e-01 4.06190783e-01
-7.41119031e-03 -2.10912868e-01 -8.60336348e-02 7.26404041e-03
7.29704499e-01 4.03630465e-01 1.87669739e-01 -1.66702896e-01
3.25889200e-01 -2.82154232e-01 2.51169175e-01 4.83245105e-01
2.81278074e-01 6.55035555e-01 2.82846630e-01 -3.61530572e-01
-9.83626425e-01 -1.15981412e+00 -2.28022840e-02 8.10527861e-01
2.61589259e-01 -2.36868173e-01 -6.93755150e-01 -5.52289486e-01
2.98119277e-01 3.89904946e-01 -9.06831801e-01 1.62660718e-01
-6.05196714e-01 1.14446692e-03 -3.11138004e-01 7.89549649e-01
1.75229192e-01 -1.34754539e+00 -7.88409948e-01 1.96345016e-01
-1.98397741e-01 -8.56451154e-01 -5.21040857e-01 -2.60756761e-01
-6.82935297e-01 -1.17782378e+00 -7.91651309e-01 -5.87259591e-01
6.98399365e-01 1.60301127e-03 1.10928643e+00 3.82813722e-01
-7.36893594e-01 4.41168100e-01 -3.13064933e-01 -1.14282511e-01
4.44607176e-02 2.57395744e-01 -5.23023531e-02 1.18355304e-01
-1.58804625e-01 -8.83152366e-01 -7.41940141e-01 3.28409553e-01
-5.86886883e-01 2.72346199e-01 6.96469784e-01 7.73883343e-01
7.81547666e-01 -2.25538909e-01 1.46992421e-02 -4.39017445e-01
5.87695949e-02 -1.83968663e-01 -6.74391985e-01 2.11723745e-01
-1.93309262e-02 1.90005451e-01 1.78283662e-01 -5.16619802e-01
-9.72876549e-01 5.77710986e-01 -1.53182536e-01 -7.86432981e-01
-1.53385386e-01 2.55996764e-01 -4.35328692e-01 -9.79141891e-02
2.21908852e-01 5.30706085e-02 -7.45850615e-03 -8.37435365e-01
4.51157928e-01 3.41933250e-01 3.34192365e-01 -8.00688505e-01
7.62691200e-01 7.46052921e-01 4.68917079e-02 -8.96500945e-01
-8.33304882e-01 -1.55695707e-01 -9.91588831e-01 -4.38800544e-01
1.07026541e+00 -3.38413090e-01 -1.11862051e+00 5.30098855e-01
-1.37526226e+00 -7.28015482e-01 -1.73567384e-01 4.73376662e-01
-6.70375586e-01 2.08074927e-01 -4.79095578e-01 -9.57259178e-01
-2.54135728e-01 -1.32103288e+00 1.40387714e+00 -5.60113378e-02
-2.17501953e-01 -8.84159684e-01 -2.76266690e-02 4.84040827e-01
-3.69814187e-02 2.19035178e-01 7.47972071e-01 -9.03642401e-02
-1.06524730e+00 -9.91201699e-02 -3.79205942e-01 2.27784980e-02
2.06579909e-01 1.85524166e-01 -7.61771202e-01 -1.91455081e-01
-3.02973717e-01 -3.10982972e-01 7.64530003e-01 6.31668568e-01
1.30227506e+00 -2.98835993e-01 -5.53923309e-01 4.90700781e-01
1.20975447e+00 -2.73180723e-01 5.16441643e-01 -6.22425936e-02
1.16437554e+00 6.61540270e-01 5.01921892e-01 4.67337340e-01
3.79523396e-01 1.01316404e+00 7.71563888e-01 5.10513186e-02
-2.33195096e-01 -2.50986278e-01 -1.29390702e-01 4.41802382e-01
-5.69085240e-01 -2.41304904e-01 -1.02486837e+00 4.21083897e-01
-1.98925507e+00 -7.12185085e-01 -2.14917570e-01 2.22633910e+00
6.36602521e-01 1.07473172e-01 1.18046291e-01 1.03067718e-01
4.12048548e-01 9.41468403e-02 -6.43617332e-01 5.40777087e-01
4.73035395e-01 2.63274223e-01 5.63377924e-02 8.95207226e-01
-8.72982502e-01 1.06385124e+00 4.79332638e+00 6.15033686e-01
-9.09419119e-01 6.43331558e-02 2.92504638e-01 -2.84351915e-01
-2.14199752e-01 -1.35356337e-01 -6.86578631e-01 2.55442768e-01
1.37051931e-02 1.70178980e-01 5.78554213e-01 6.06718004e-01
4.95020375e-02 -6.87446073e-02 -1.39548242e+00 8.00639927e-01
-9.75942910e-02 -1.28205478e+00 8.13177899e-02 3.92107934e-01
5.65023661e-01 -1.04344301e-01 1.73301734e-02 -2.50724375e-01
1.69259384e-01 -1.05907893e+00 9.91122007e-01 8.59747052e-01
4.60413128e-01 -6.75538898e-01 4.33627278e-01 3.81818593e-01
-1.47721827e+00 1.25657335e-01 7.34369010e-02 -1.52252793e-01
2.56016165e-01 5.26778519e-01 -3.67992729e-01 5.13872862e-01
8.93247545e-01 6.48906946e-01 -2.72620827e-01 8.89079630e-01
-3.88854712e-01 2.61156499e-01 -5.21333516e-01 9.24630687e-02
-1.76600236e-02 -1.18817821e-01 7.56206214e-01 7.62256980e-01
-4.79660034e-02 1.86983287e-01 4.74824637e-01 1.19611073e+00
3.67862917e-02 -2.45743580e-02 -2.50269681e-01 3.78219038e-02
3.07889402e-01 1.12599897e+00 -8.38176727e-01 -2.45078802e-01
-1.70031190e-01 1.16989553e+00 6.88310862e-01 2.63730556e-01
-8.85967374e-01 -2.55974159e-02 6.95077300e-01 4.67326611e-01
4.29284990e-01 -5.53046465e-01 -3.76422912e-01 -9.91890132e-01
4.33931589e-01 -5.10625243e-01 2.11555529e-02 -7.92950928e-01
-1.30209744e+00 4.91770506e-01 -1.14320517e-01 -7.88097858e-01
1.27353370e-01 -6.49420559e-01 -3.53434443e-01 8.90047252e-01
-1.14656413e+00 -1.49828172e+00 -6.95324361e-01 7.01058447e-01
5.89360476e-01 5.05199015e-01 5.83685219e-01 8.17921534e-02
-3.76501411e-01 1.60418630e-01 -3.29843879e-01 2.28404373e-01
3.80500764e-01 -1.20415175e+00 3.63195211e-01 3.33998382e-01
3.84555340e-01 6.06344283e-01 6.41774476e-01 -8.32946062e-01
-1.67179942e+00 -7.70614982e-01 5.49453437e-01 -6.30969882e-01
6.03044271e-01 -4.67370570e-01 -9.52699959e-01 8.18155110e-01
3.63163091e-02 3.87801021e-01 -7.05511123e-02 1.51811689e-01
-3.62920970e-01 8.34333450e-02 -7.72013903e-01 5.91214001e-01
1.34013808e+00 -3.08875233e-01 -5.46630263e-01 5.82196355e-01
4.84655619e-01 -6.85968816e-01 -7.24954665e-01 3.68899763e-01
9.23299551e-01 -8.84371698e-01 1.15804088e+00 -4.49008197e-01
5.38685501e-01 -2.33264193e-01 3.07211112e-02 -1.02736306e+00
-4.34119433e-01 -3.48171711e-01 -5.59359252e-01 8.88775408e-01
5.95173240e-02 -4.94697690e-01 8.01683903e-01 5.68280458e-01
1.50027916e-01 -1.06476474e+00 -8.71087551e-01 -7.21096635e-01
5.56670576e-02 -3.57570350e-01 3.47449303e-01 4.92569178e-01
-2.43637741e-01 2.88742810e-01 -2.99564809e-01 4.09088552e-01
1.03117537e+00 4.22684878e-01 7.98000276e-01 -1.50012040e+00
-6.66314781e-01 -6.71141207e-01 -2.64934033e-01 -1.29006696e+00
2.86911070e-01 -7.00368106e-01 1.98732108e-01 -1.74203968e+00
4.79148537e-01 -5.78075945e-01 -4.65581892e-03 5.01405597e-01
-2.59142011e-01 3.36024165e-01 3.09868485e-01 1.74529016e-01
-3.27927619e-01 6.23247206e-01 1.62933850e+00 -3.08475774e-02
-3.09927523e-01 -2.75299922e-02 -2.07670763e-01 8.25792313e-01
6.03344798e-01 -2.75926411e-01 -1.52866412e-02 -5.74408948e-01
-1.40967704e-02 5.01054637e-02 1.01703179e+00 -8.12458038e-01
1.86005741e-01 -3.47353220e-01 2.19372958e-01 -7.18371809e-01
5.89382112e-01 -1.00263011e+00 2.30547413e-01 5.52916825e-01
-1.56581685e-01 -3.42158079e-01 -6.03373274e-02 5.56435764e-01
2.67911166e-01 6.39888644e-02 8.18176985e-01 -1.99101761e-01
-2.79262871e-01 7.50970244e-01 2.25639001e-01 -1.75440207e-01
1.07828224e+00 1.12034060e-01 2.38396019e-01 -2.28443652e-01
-8.59474778e-01 2.15001598e-01 6.11043632e-01 4.76575434e-01
5.91428101e-01 -1.24317336e+00 -7.62461364e-01 1.59074336e-01
-9.61235762e-02 5.77004611e-01 3.29969436e-01 1.00055945e+00
-2.72736996e-01 2.77120262e-01 -1.43642053e-01 -9.01697874e-01
-1.49295199e+00 4.51424599e-01 3.82149279e-01 -1.03602506e-01
-8.70854616e-01 1.00672889e+00 4.98346537e-01 -2.94178516e-01
3.94098014e-01 -3.27231914e-01 1.55576468e-01 -1.17882699e-01
4.10537899e-01 4.74090219e-01 -1.04516536e-01 -7.01712668e-01
-4.14884120e-01 9.94027555e-01 -1.26463547e-01 -1.58288196e-01
1.45303690e+00 2.01416120e-01 -1.73260704e-01 3.06593210e-01
1.05921757e+00 -2.48860642e-02 -1.63290191e+00 -2.92552024e-01
-3.65059286e-01 -8.10494184e-01 2.27445841e-01 -6.07966006e-01
-1.37607026e+00 9.53314960e-01 3.87523830e-01 -1.84133518e-02
7.37300932e-01 5.96737325e-01 5.61241925e-01 2.91069925e-01
5.20901203e-01 -6.43634975e-01 4.46372330e-01 2.23957777e-01
1.56590331e+00 -1.07752776e+00 1.21554680e-01 -7.00907409e-01
-2.58998752e-01 8.81350696e-01 6.76754653e-01 -4.20320809e-01
7.93641448e-01 1.95901781e-01 -2.97541708e-01 -3.42587203e-01
-3.89128238e-01 -3.03672910e-01 6.81839943e-01 3.28970701e-01
3.01465154e-01 7.34053329e-02 5.34516461e-02 3.83967221e-01
1.76752359e-02 5.64230327e-03 -3.11096847e-01 9.44756925e-01
-2.48763427e-01 -1.21773624e+00 -4.03256625e-01 3.71912003e-01
-4.08643708e-02 1.82404280e-01 -4.51303482e-01 7.50442028e-01
5.42710833e-02 4.95523185e-01 1.14725485e-01 -3.33886206e-01
3.92192572e-01 -1.00139052e-01 1.04121852e+00 -6.27769351e-01
-2.94318676e-01 3.66692543e-01 -4.31636333e-01 -9.04025018e-01
-3.12078893e-01 -8.29170763e-01 -1.24457312e+00 -2.91977555e-01
-4.21176434e-01 -2.59837419e-01 4.51997340e-01 9.94386137e-01
3.74488592e-01 6.14021063e-01 2.98223495e-01 -1.75149727e+00
-3.78595233e-01 -9.26520109e-01 -4.28222835e-01 3.93809319e-01
3.45762789e-01 -1.31264913e+00 -1.73641086e-01 1.82201868e-04] | [6.688773155212402, -1.0679092407226562] |
305114c9-1a5f-4a06-95c9-8f4cd6920469 | decentralised-approach-for-multi-agent-path | 2106.05188 | null | https://arxiv.org/abs/2106.05188v1 | https://arxiv.org/pdf/2106.05188v1.pdf | Decentralised Approach for Multi Agent Path Finding | Multi Agent Path Finding (MAPF) requires identification of conflict free paths for agents which could be point-sized or with dimensions. In this paper, we propose an approach for MAPF for spatially-extended agents. These find application in real world problems like Convoy Movement Problem, Train Scheduling etc. Our proposed approach, Decentralised Multi Agent Path Finding (DeMAPF), handles MAPF as a sequence of pathplanning and allocation problems which are solved by two sets of agents Travellers and Routers respectively, over multiple iterations. The approach being decentralised allows an agent to solve the problem pertinent to itself, without being aware of other agents in the same set. This allows the agents to be executed on independent machines, thereby leading to scalability to handle large sized problems. We prove, by comparison with other distributed approaches, that the approach leads to a faster convergence to a conflict-free solution, which may be suboptimal, with lesser memory requirement. | ['M. Narasimha Murty', 'Shyni Thomas'] | 2021-06-03 | null | null | null | null | ['multi-agent-path-finding'] | ['playing-games'] | [-2.16830239e-01 2.26341560e-01 2.29149029e-01 8.49555358e-02
-2.94583946e-01 -8.75027657e-01 5.78361273e-01 6.12615764e-01
-7.89825201e-01 1.53334427e+00 -3.05864125e-01 -2.68852830e-01
-1.01859438e+00 -1.11524045e+00 -3.07959765e-01 -6.15597069e-01
-7.32186735e-01 1.49501526e+00 1.01899779e+00 -3.94150347e-01
4.61006969e-01 7.50546932e-01 -1.18908155e+00 -2.87259281e-01
6.65377438e-01 1.78982601e-01 4.87882882e-01 9.31688607e-01
-6.63861036e-02 7.96429157e-01 -6.44488275e-01 1.73703536e-01
3.86530399e-01 -2.21793294e-01 -1.39505017e+00 7.88443461e-02
-1.90627262e-01 -1.82711884e-01 2.18579754e-01 8.09198201e-01
1.56915963e-01 7.20789433e-01 7.25173891e-01 -1.87916219e+00
3.70934576e-01 4.82892036e-01 -8.09765100e-01 3.75067860e-01
4.99841690e-01 -1.38928756e-01 6.49524570e-01 -9.64236632e-02
8.68198097e-01 1.22764695e+00 3.94229800e-01 1.62138477e-01
-9.00503576e-01 -2.39037778e-02 3.56140137e-01 5.80197752e-01
-1.31304586e+00 1.45495161e-02 3.28298844e-02 -6.71701655e-02
1.32968843e+00 5.09781301e-01 5.37025750e-01 -6.63586184e-02
4.23784256e-01 1.08929776e-01 1.11852515e+00 -4.48899776e-01
6.06138647e-01 5.81091195e-02 7.92934522e-02 6.42729044e-01
4.51210022e-01 -1.16460823e-01 -9.28798318e-02 -3.93265724e-01
7.07212746e-01 -3.49024624e-01 -1.08894847e-01 -3.90723616e-01
-1.32385314e+00 9.40839767e-01 2.99679130e-01 2.55751401e-01
-9.51780617e-01 1.99223831e-01 2.69676387e-01 5.64679563e-01
-6.24430813e-02 4.49736387e-01 -4.16262060e-01 -1.31003037e-01
-3.01051050e-01 7.29678392e-01 1.15922964e+00 9.56002414e-01
1.03716576e+00 -3.70660156e-01 3.97731394e-01 4.62282777e-01
3.90883774e-01 2.65597016e-01 -8.87951478e-02 -9.48148191e-01
4.55108672e-01 5.73425353e-01 4.84336287e-01 -9.84200537e-01
-1.01725054e+00 2.71266490e-01 -5.89006901e-01 8.53567421e-01
6.29226983e-01 -3.49021345e-01 -4.27400142e-01 1.48419821e+00
1.02904510e+00 1.07648559e-01 3.62293690e-01 9.30767715e-01
2.16073707e-01 1.08359146e+00 -6.47901520e-02 -4.86395687e-01
1.56617916e+00 -1.41093314e+00 -4.28197235e-01 3.48795913e-02
6.59409106e-01 -7.77260721e-01 1.96002916e-01 4.53791291e-01
-1.14281762e+00 2.68847179e-02 -8.72726262e-01 4.87623125e-01
-6.12982452e-01 -6.52013719e-01 5.27079999e-01 4.23513502e-01
-1.47203791e+00 3.57756644e-01 -7.29986668e-01 -7.92253792e-01
-2.35619918e-01 8.21659386e-01 -4.28773075e-01 -3.85338664e-02
-9.00729179e-01 1.16674209e+00 7.61377037e-01 -3.01553309e-02
-9.10910785e-01 -4.55034375e-02 -3.12364310e-01 1.14032999e-04
8.98924530e-01 -7.75175571e-01 1.09290898e+00 -6.22670174e-01
-1.47466218e+00 1.14768587e-01 1.05803430e-01 -3.56539160e-01
5.27805388e-01 4.47009563e-01 -1.78879142e-01 1.45622909e-01
3.90966654e-01 3.80048364e-01 3.97759080e-01 -1.24605119e+00
-1.20508671e+00 -2.13553175e-01 4.07618672e-01 6.41730070e-01
2.93495089e-01 1.77198648e-01 -5.05035035e-02 1.67921022e-01
-1.64977789e-01 -1.14678323e+00 -1.00649834e+00 -4.79617387e-01
-3.05999845e-01 -4.60153967e-01 6.40289605e-01 -4.03455878e-03
8.03215325e-01 -1.50692332e+00 3.17260683e-01 8.83807957e-01
2.10738003e-01 -4.74379957e-02 -2.36864403e-01 1.22155797e+00
3.68687958e-01 -2.38446727e-01 6.01348765e-02 -1.13134189e-02
1.27431870e-01 5.53223193e-01 3.78162175e-01 6.64182484e-01
-3.06339622e-01 2.09757090e-01 -8.83205891e-01 -6.79490566e-01
7.14923441e-02 -4.05522548e-02 -3.98821980e-01 -2.31307700e-01
-3.17982942e-01 2.07689568e-01 -7.79212296e-01 1.76034287e-01
8.60019624e-01 2.59069111e-02 4.54088300e-01 6.11683309e-01
-7.67782509e-01 -1.37976885e-01 -1.89313173e+00 1.38658273e+00
-1.95898652e-01 3.82582098e-01 3.53081793e-01 -9.52864885e-01
7.25120366e-01 3.99910450e-01 7.51048505e-01 -4.67005521e-01
1.62220951e-02 3.24910372e-01 7.67590404e-02 -2.32862189e-01
7.54398346e-01 1.28904179e-01 -3.66377719e-02 1.04576826e+00
-4.53411222e-01 2.99129218e-01 7.34281957e-01 2.51596004e-01
1.50608158e+00 -2.26331353e-01 4.86500829e-01 -5.42409718e-01
8.98808062e-01 8.47106516e-01 4.41362113e-01 9.05762494e-01
-2.56883800e-01 -4.31132704e-01 3.87650579e-01 -7.19388068e-01
-7.65430748e-01 -9.05698061e-01 4.99496430e-01 8.64890099e-01
7.74717271e-01 -2.80350506e-01 -5.85948527e-01 -6.05368078e-01
-1.82981804e-01 1.80827200e-01 -1.49271861e-01 5.71912766e-01
-1.11328125e+00 -6.23910904e-01 2.86614057e-02 -2.66074032e-01
3.65569025e-01 -1.13502419e+00 -1.26764023e+00 9.24574077e-01
-3.26837413e-02 -8.71661186e-01 -2.13391945e-01 -2.51647104e-02
-5.26229918e-01 -1.34801018e+00 -4.68624741e-01 -8.21429253e-01
8.82388890e-01 5.02831042e-01 7.09068298e-01 2.99669385e-01
-1.33683085e-01 3.82016182e-01 -4.94429141e-01 -4.13963884e-01
-3.71446818e-01 3.08314651e-01 1.40042901e-01 -2.42174581e-01
6.02401756e-02 -6.01424098e-01 -6.01608932e-01 6.06694281e-01
-6.89525902e-01 -8.70498419e-02 4.57931876e-01 2.03388214e-01
4.55104023e-01 8.80536854e-01 8.39289844e-01 -8.22353840e-01
9.11771297e-01 -7.32537448e-01 -9.68754292e-01 3.37384313e-01
-6.25697672e-01 -1.54603094e-01 7.82927036e-01 -9.57640633e-02
-8.58308792e-01 6.76768199e-02 3.92199636e-01 4.06698853e-01
-3.96533936e-01 5.04177630e-01 2.00338438e-01 -4.63013858e-01
3.75936896e-01 -3.61565240e-02 2.36498073e-01 -1.85946643e-01
2.74804950e-01 1.37470961e-01 1.12897657e-01 -6.02063656e-01
6.10929489e-01 2.95600563e-01 6.10364616e-01 -5.96561491e-01
4.70185876e-01 -5.76447070e-01 -5.21656692e-01 -5.93159378e-01
6.07384622e-01 -3.19233388e-01 -1.10684764e+00 1.66945457e-01
-1.33338785e+00 -4.05343860e-01 1.41463116e-01 4.59595352e-01
-7.72387743e-01 2.07684815e-01 -3.79868627e-01 -8.05540144e-01
-2.11260039e-02 -9.85661328e-01 3.01681280e-01 4.22009617e-01
-2.94019729e-01 -1.18256354e+00 6.94450557e-01 7.75042102e-02
3.80274057e-01 4.58070189e-01 8.78356516e-01 -8.22863758e-01
-1.00503683e+00 8.66565481e-02 -4.91987057e-02 -9.29364920e-01
-1.86980870e-02 -2.16634229e-01 4.41369526e-02 -5.89355886e-01
-4.86732394e-01 2.40234330e-01 1.54375657e-02 2.94237047e-01
-1.58430293e-01 -5.77879488e-01 -9.63995457e-01 -3.01008075e-01
1.95460355e+00 7.51395524e-01 3.72052759e-01 1.01651216e+00
1.58574671e-01 9.75852668e-01 1.00433874e+00 6.17227852e-01
7.43173838e-01 7.58342445e-01 4.90419298e-01 -1.22283041e-01
2.82507807e-01 4.45624262e-01 4.10739742e-02 3.65125388e-01
-4.15341169e-01 -7.59801209e-01 -1.15595937e+00 8.17129612e-01
-2.44866347e+00 -8.82497549e-01 -5.20607889e-01 1.80985987e+00
1.10596158e-01 -1.15960680e-01 6.99050128e-01 7.87773728e-02
7.80697107e-01 -2.38387182e-01 -2.03098431e-01 -9.59445000e-01
3.22416186e-01 -2.10971594e-01 9.01939690e-01 9.53229845e-01
-6.21926486e-01 7.56494045e-01 5.56554174e+00 4.79439974e-01
-4.60274965e-01 4.14496213e-01 -4.55755852e-02 7.08410665e-02
-5.21729104e-02 4.21664834e-01 -4.54451680e-01 2.85444587e-01
1.02199829e+00 -3.60956103e-01 7.15788364e-01 3.34521413e-01
5.24070919e-01 -8.50318730e-01 -7.26794243e-01 3.95399898e-01
-3.51845175e-01 -1.23432958e+00 -3.34795594e-01 4.38414216e-01
7.85119057e-01 -3.25945430e-02 -6.18815243e-01 -1.93159893e-01
7.64965057e-01 -6.96716785e-01 5.19925416e-01 2.03635529e-01
-1.46550372e-01 -1.44258010e+00 7.09736049e-01 8.22213352e-01
-1.47329676e+00 -1.19435370e-01 -3.14974189e-01 -4.42358524e-01
7.00391591e-01 -7.69237578e-02 -1.14014375e+00 1.05067158e+00
4.99819368e-01 -2.16416553e-01 1.56846926e-01 1.36936951e+00
2.72506803e-01 -3.25476617e-01 -6.28191710e-01 -5.27045488e-01
8.04649532e-01 -5.55188417e-01 8.34333599e-01 8.22588265e-01
1.58654615e-01 3.93037945e-01 6.53882682e-01 4.11964029e-01
8.66108477e-01 3.24671656e-01 -6.70968115e-01 5.39899290e-01
5.49030483e-01 1.07304692e+00 -1.49175942e+00 -1.65780976e-01
-2.78290629e-01 7.34378517e-01 4.40718561e-01 3.82616192e-01
-6.66024625e-01 -6.47816479e-01 6.17062390e-01 7.08683729e-02
8.62042680e-02 -4.28573459e-01 3.56570154e-01 -1.96088210e-01
-3.38839918e-01 -6.75795019e-01 6.61258519e-01 -3.33889931e-01
-7.63283908e-01 9.71923590e-01 5.11451244e-01 -9.60035801e-01
-5.64196050e-01 -2.31260523e-01 -7.72129834e-01 7.36101627e-01
-1.66239798e+00 -8.33369434e-01 -1.60819665e-03 9.49130237e-01
4.99577135e-01 -3.25416058e-01 8.30376148e-01 2.89679855e-01
-4.61548239e-01 -1.03277475e-01 1.02994166e-01 -5.13786137e-01
9.42253843e-02 -1.13611698e+00 -1.31748334e-01 8.70567322e-01
-1.30464062e-01 1.99315265e-01 8.57688248e-01 -7.24608123e-01
-1.33963764e+00 -7.05951989e-01 7.93794870e-01 2.78207138e-02
5.91199577e-01 2.49552950e-01 -5.35734355e-01 6.65797412e-01
7.08530128e-01 -5.50772011e-01 3.56150836e-01 -1.12521730e-01
6.54439986e-01 6.10851161e-02 -1.27068746e+00 5.11004269e-01
6.18756115e-01 4.65146512e-01 -1.29924193e-01 5.00235736e-01
1.18137754e-01 -2.81201541e-01 -5.02519786e-01 -1.37058005e-01
1.26523539e-01 -8.20346355e-01 6.42814100e-01 -4.69660759e-01
-3.44098717e-01 -9.27227497e-01 3.12035620e-01 -1.69113553e+00
-5.19327402e-01 -9.01251137e-01 5.46263218e-01 9.87246692e-01
6.08056307e-01 -1.26340902e+00 8.83434951e-01 5.29673398e-01
-8.30645785e-02 -3.38655561e-01 -1.41295183e+00 -8.27992260e-01
-1.95186362e-01 7.51610696e-02 9.73002255e-01 8.83782864e-01
3.61839533e-01 2.35196710e-01 -2.50624299e-01 7.79873133e-01
7.45829165e-01 2.85556763e-01 1.04799700e+00 -1.25456035e+00
-2.39635989e-01 -3.84455800e-01 -2.60350972e-01 -6.93092167e-01
1.01518929e-01 -5.72094619e-01 1.73792079e-01 -2.07741094e+00
-2.38624930e-01 -1.08400166e+00 5.54594398e-02 5.71626902e-01
4.61632878e-01 -1.45876393e-01 3.44159454e-01 3.30137521e-01
-8.87746990e-01 -1.75487801e-01 1.18043697e+00 7.71191940e-02
-4.89772081e-01 1.66931555e-01 -5.39056957e-02 5.54772854e-01
1.06412888e+00 -8.72741222e-01 -7.73930550e-01 -3.94335598e-01
2.94049352e-01 6.79966152e-01 6.84143677e-02 -9.58429515e-01
8.99265110e-01 -8.45088542e-01 -4.64482874e-01 -6.67824805e-01
3.10733855e-01 -1.27045715e+00 8.76237214e-01 9.31979179e-01
1.35140881e-01 8.68808210e-01 1.15223683e-01 6.53146565e-01
-1.36928633e-01 -7.13857174e-01 3.78788799e-01 -5.78668654e-01
-9.64167893e-01 1.33884206e-01 -1.15051746e+00 -4.93627459e-01
1.89950693e+00 -4.52970147e-01 -6.08674109e-01 -3.39691728e-01
-5.84714711e-01 9.37779427e-01 2.99440295e-01 -1.61388412e-01
4.94544864e-01 -1.01875317e+00 -7.82172978e-01 -3.07378590e-01
-4.52085525e-01 1.03146434e-01 3.84070039e-01 8.81382167e-01
-1.25294471e+00 4.45883155e-01 -6.84730053e-01 -1.82898760e-01
-1.44798994e+00 7.24773645e-01 2.28024200e-01 -5.07615685e-01
-6.75517857e-01 4.09027606e-01 -3.27844769e-01 -1.77921355e-01
-4.81119119e-02 8.66550207e-02 -5.56459963e-01 8.38966146e-02
4.74775553e-01 9.27368641e-01 -3.45453173e-01 -5.90755761e-01
-6.85969055e-01 7.12378502e-01 -1.60703331e-01 -5.50118387e-01
1.29492104e+00 -4.90884453e-01 -5.21612465e-01 -2.36460939e-01
4.72405344e-01 -1.02117918e-01 -7.58888662e-01 -8.42347369e-02
2.92972118e-01 -5.32958329e-01 -2.22208261e-01 -6.63152695e-01
-8.06730926e-01 -1.69110492e-01 1.64949372e-01 7.69092262e-01
1.06592166e+00 -4.21963632e-02 5.58176219e-01 5.20958602e-01
9.40441906e-01 -1.08160245e+00 -2.58527100e-01 5.95341027e-01
4.71430451e-01 -8.53677452e-01 1.86461732e-01 -6.49877846e-01
-6.26149356e-01 1.23263133e+00 8.54249299e-01 -1.04541592e-01
3.62555057e-01 4.17411953e-01 2.30114907e-03 -4.72893268e-01
-9.82275784e-01 -3.56590241e-01 -5.78287184e-01 8.73585701e-01
-6.63322747e-01 1.99444965e-01 -7.67952144e-01 -1.91288188e-01
-1.31359294e-01 -1.75606370e-01 9.87851918e-01 1.28291643e+00
-8.69955301e-01 -1.42504394e+00 -7.55363226e-01 -2.17431970e-03
8.83386731e-02 4.63607073e-01 -1.91374138e-01 1.21018159e+00
2.46752605e-01 1.20178711e+00 3.70645195e-01 2.69326478e-01
4.26617175e-01 -4.74513143e-01 4.57561731e-01 -3.88070732e-01
-6.16216183e-01 -5.02346717e-02 6.62177622e-01 -3.09563965e-01
-5.74223101e-01 -7.32635915e-01 -1.75648081e+00 -7.48588383e-01
-1.81074202e-01 7.63140678e-01 5.44138730e-01 9.50707853e-01
1.09355606e-01 3.68845224e-01 7.14571655e-01 -6.60948932e-01
-3.16200405e-03 -4.56270933e-01 -7.09230125e-01 -3.54959100e-01
1.83900341e-01 -6.98035538e-01 1.00430168e-01 -5.60809851e-01] | [4.976676940917969, 1.7382606267929077] |
c5025001-f985-4abf-a3f6-12cee205f04b | rerender-a-video-zero-shot-text-guided-video | 2306.07954 | null | https://arxiv.org/abs/2306.07954v1 | https://arxiv.org/pdf/2306.07954v1.pdf | Rerender A Video: Zero-Shot Text-Guided Video-to-Video Translation | Large text-to-image diffusion models have exhibited impressive proficiency in generating high-quality images. However, when applying these models to video domain, ensuring temporal consistency across video frames remains a formidable challenge. This paper proposes a novel zero-shot text-guided video-to-video translation framework to adapt image models to videos. The framework includes two parts: key frame translation and full video translation. The first part uses an adapted diffusion model to generate key frames, with hierarchical cross-frame constraints applied to enforce coherence in shapes, textures and colors. The second part propagates the key frames to other frames with temporal-aware patch matching and frame blending. Our framework achieves global style and local texture temporal consistency at a low cost (without re-training or optimization). The adaptation is compatible with existing image diffusion techniques, allowing our framework to take advantage of them, such as customizing a specific subject with LoRA, and introducing extra spatial guidance with ControlNet. Extensive experimental results demonstrate the effectiveness of our proposed framework over existing methods in rendering high-quality and temporally-coherent videos. | ['Chen Change Loy', 'Ziwei Liu', 'Yifan Zhou', 'Shuai Yang'] | 2023-06-13 | null | null | null | null | ['patch-matching'] | ['computer-vision'] | [ 2.87077546e-01 -3.34871978e-01 -6.92327917e-02 -1.93515703e-01
-6.08644366e-01 -4.85690087e-01 6.49783731e-01 -5.18100679e-01
-1.89601824e-01 5.77648103e-01 1.28470510e-01 4.13210429e-02
8.01849589e-02 -6.69911146e-01 -9.20803845e-01 -6.61525548e-01
5.25645949e-02 3.90157215e-02 7.50545144e-01 -2.52224624e-01
2.33812898e-01 4.95867282e-01 -1.47110868e+00 5.45630455e-01
7.48908997e-01 9.51815605e-01 4.89755124e-01 8.62277806e-01
-1.71072572e-01 1.01782560e+00 -2.99243093e-01 -3.46923351e-01
5.45094848e-01 -6.78130567e-01 -7.00598300e-01 5.22189081e-01
8.94400418e-01 -6.90308869e-01 -3.58377069e-01 8.30799043e-01
4.98069316e-01 3.47215950e-01 2.87739247e-01 -1.28555048e+00
-8.97038460e-01 2.10487261e-01 -6.14444554e-01 2.13411316e-01
5.39320350e-01 3.78599972e-01 5.27784288e-01 -1.00069308e+00
1.02475643e+00 1.36398590e+00 5.28098583e-01 7.19319642e-01
-1.22447515e+00 -5.51938891e-01 3.54234397e-01 2.85280347e-01
-1.27966619e+00 -7.61866570e-01 7.92665660e-01 -5.08762002e-01
8.76753569e-01 2.28640884e-01 8.85397255e-01 9.49868381e-01
3.73164505e-01 5.90834558e-01 1.12437022e+00 -3.60981196e-01
1.76932260e-01 -1.43390074e-01 -5.70316195e-01 7.50425458e-01
-2.87777007e-01 1.99254647e-01 -7.81589806e-01 -1.15232412e-02
1.34308863e+00 -4.27501835e-03 -4.77190197e-01 -5.28214395e-01
-1.49642062e+00 4.07863200e-01 2.45891646e-01 2.36551493e-01
-4.19583261e-01 4.31227028e-01 2.17240527e-01 3.13186586e-01
4.95224208e-01 8.51981901e-03 -2.04779238e-01 1.30969277e-02
-1.27846098e+00 4.25096482e-01 4.19614553e-01 1.28089499e+00
8.52806509e-01 3.83610666e-01 -4.68680114e-01 8.71154666e-01
6.30198643e-02 5.66537678e-01 4.61347312e-01 -1.55431616e+00
3.30763310e-01 9.15571451e-02 2.75439113e-01 -1.22207284e+00
4.83986512e-02 6.30161026e-03 -7.69288361e-01 3.76586825e-01
1.12141564e-01 -1.86724234e-02 -9.42196250e-01 1.65005302e+00
6.28043294e-01 5.09125948e-01 -2.90292889e-01 1.08625877e+00
4.63758528e-01 7.97642112e-01 8.55055898e-02 -3.96805465e-01
1.09139645e+00 -1.37342048e+00 -8.14375043e-01 8.11575800e-02
7.62174502e-02 -1.15760314e+00 1.05799890e+00 2.34214127e-01
-1.67304921e+00 -8.27463925e-01 -8.38843524e-01 -3.09905797e-01
9.50725004e-02 -1.16593063e-01 4.56974238e-01 4.05907542e-01
-1.48013473e+00 5.92828929e-01 -7.82445669e-01 -3.70473623e-01
2.84547508e-01 3.32765192e-01 -2.32541606e-01 -1.01958603e-01
-9.30243671e-01 4.44454134e-01 2.91730791e-01 -6.44979347e-03
-9.92518961e-01 -8.32977116e-01 -7.45130181e-01 -1.90359429e-01
3.58675569e-01 -1.14710677e+00 1.10262561e+00 -1.53413570e+00
-2.08552957e+00 4.91707414e-01 -2.82278925e-01 -1.73302397e-01
8.32583606e-01 -1.31142393e-01 -2.46250466e-01 4.63907957e-01
1.27723992e-01 1.19191420e+00 1.43859506e+00 -1.20754147e+00
-8.63131762e-01 1.60929605e-01 1.45122036e-01 4.40344959e-01
-4.29796934e-01 1.63305938e-01 -1.05620074e+00 -1.23606563e+00
-9.17417780e-02 -8.12602103e-01 -1.24043524e-01 6.43572986e-01
9.69359651e-02 1.62148669e-01 1.16212475e+00 -7.59234846e-01
1.13266981e+00 -2.03525543e+00 4.61334437e-01 -4.16847616e-02
1.19735517e-01 1.47857785e-01 -4.36208010e-01 2.63986111e-01
6.18229434e-02 -2.71935314e-01 -1.97823137e-01 -3.89573961e-01
-3.17717075e-01 2.10184008e-01 -2.29192629e-01 2.72044748e-01
1.74398527e-01 8.39236856e-01 -9.08448517e-01 -6.70529246e-01
4.94219333e-01 7.52856612e-01 -9.37471986e-01 3.43449473e-01
-4.06691730e-01 6.97874308e-01 -3.48868757e-01 5.80227792e-01
7.85059452e-01 -1.24740601e-01 9.95393917e-02 -4.46253479e-01
-1.52542025e-01 -1.54222086e-01 -1.28910267e+00 2.04775214e+00
-4.37209815e-01 6.03267610e-01 2.61213273e-01 -5.85840285e-01
5.74282825e-01 2.71568894e-01 6.98671401e-01 -7.42707431e-01
-1.61082447e-02 4.68583144e-02 -3.97583276e-01 -6.35270178e-01
7.68056154e-01 1.38088092e-01 4.66481239e-01 3.57891589e-01
2.56195422e-02 -2.33569607e-01 4.87230122e-01 3.99671853e-01
5.95477223e-01 7.26299345e-01 -1.57612070e-01 -2.88213253e-01
7.37495244e-01 -6.85064718e-02 5.74169219e-01 4.40975338e-01
3.57375853e-02 9.57827926e-01 5.88426851e-02 -5.35161078e-01
-1.32600665e+00 -9.55452919e-01 3.36063951e-01 1.03591561e+00
3.82056922e-01 -4.44093525e-01 -9.31490958e-01 -3.27751368e-01
-3.96740884e-01 2.80303061e-01 -5.52318454e-01 1.75555900e-01
-8.36656630e-01 -3.14319015e-01 6.83386698e-02 4.53029037e-01
7.21554399e-01 -9.21746433e-01 -5.74646533e-01 3.87617260e-01
-4.49501872e-01 -1.25638986e+00 -1.15700507e+00 -5.10305285e-01
-9.01691616e-01 -8.23096216e-01 -1.17957783e+00 -1.01180828e+00
7.03240395e-01 6.79040670e-01 9.96231019e-01 3.74286234e-01
-3.36825222e-01 5.86124361e-01 -3.45669150e-01 4.15302843e-01
-4.96946394e-01 -2.27891624e-01 -1.16953447e-01 3.92672867e-01
-4.02484685e-01 -6.36810422e-01 -1.08620274e+00 5.78365147e-01
-1.32506919e+00 4.15214479e-01 2.52940774e-01 7.34073937e-01
6.86440110e-01 -2.11746708e-01 8.02558288e-02 -4.77339715e-01
5.02115846e-01 1.32410964e-02 -6.03609741e-01 3.27029079e-01
-3.61282557e-01 -2.54465461e-01 6.39113903e-01 -7.33069420e-01
-1.21888840e+00 7.79572204e-02 1.35285661e-01 -7.36905873e-01
5.43246269e-02 -8.28869566e-02 5.67267761e-02 -4.51129019e-01
3.81957740e-01 3.60101849e-01 -1.42056141e-02 -2.18516901e-01
6.28612936e-01 2.64514029e-01 6.79816008e-01 -8.72836590e-01
8.14170837e-01 6.04808211e-01 -2.83435673e-01 -7.60585189e-01
-3.01619291e-01 -1.47416070e-01 -7.78519213e-01 -4.39968318e-01
1.02021241e+00 -9.25351202e-01 -4.80043799e-01 5.98494649e-01
-1.19190252e+00 -7.32364178e-01 -3.32025766e-01 3.25151652e-01
-8.33631039e-01 6.66333437e-01 -8.34409833e-01 -1.82179227e-01
-3.15550774e-01 -1.38507664e+00 1.13223445e+00 5.51391654e-02
-7.49165416e-02 -1.05330443e+00 -1.45320809e-02 2.18609616e-01
7.57509947e-01 8.67233276e-02 6.34726763e-01 4.82533276e-01
-1.05772150e+00 3.71637970e-01 -2.21841797e-01 3.27859730e-01
2.29297817e-01 2.98176646e-01 -5.08208930e-01 -5.05162776e-01
-1.42107099e-01 -1.79925084e-01 6.24215126e-01 5.04262984e-01
1.13897586e+00 -3.39112222e-01 -1.39004275e-01 9.20319855e-01
1.36962092e+00 2.04662964e-01 7.40053236e-01 3.38092178e-01
8.09626937e-01 3.90195936e-01 5.28603971e-01 2.38790512e-01
3.96833569e-01 1.16665161e+00 1.05809765e-02 -3.01789373e-01
-7.68466353e-01 -1.57030553e-01 5.52310824e-01 8.89531493e-01
-3.05431515e-01 -1.70374170e-01 -4.38894928e-01 4.65535045e-01
-1.97858059e+00 -1.20401704e+00 1.18795224e-01 2.00589466e+00
8.35867703e-01 -2.87695110e-01 2.08613411e-01 -2.83597589e-01
6.89898431e-01 2.27477163e-01 -3.97470564e-01 -1.15157761e-01
-1.40197888e-01 5.43347048e-03 2.13426590e-01 7.68160462e-01
-8.04729640e-01 1.22664762e+00 6.44227982e+00 1.06129801e+00
-1.26362252e+00 3.99643362e-01 7.97015548e-01 -1.79885209e-01
-3.38092953e-01 5.52303158e-03 -3.50628614e-01 3.58380079e-01
3.34409237e-01 -7.86848217e-02 7.55474269e-01 4.78574753e-01
4.85271037e-01 4.51963507e-02 -8.90029848e-01 1.01632428e+00
2.82104552e-01 -1.69400132e+00 4.08826172e-01 -2.61368155e-01
1.18331790e+00 -3.69911641e-01 1.30731896e-01 -1.50519624e-01
1.68761298e-01 -6.39621317e-01 1.25061536e+00 5.66047132e-01
1.08794987e+00 -5.54413438e-01 5.09488508e-02 -8.73042569e-02
-1.40026355e+00 1.47133917e-01 -4.15660232e-01 1.76510379e-01
4.64307487e-01 1.83284625e-01 -1.86519518e-01 5.24640501e-01
8.90994072e-01 9.13867414e-01 -4.75967050e-01 8.70424449e-01
-4.72784080e-02 1.34038955e-01 -1.53048322e-01 3.62812459e-01
1.17312729e-01 -4.25699919e-01 5.25554419e-01 1.17011344e+00
5.43603361e-01 1.88375667e-01 3.82883459e-01 7.79589832e-01
9.22191516e-02 2.38222867e-01 -3.74080092e-01 3.46397549e-01
2.99316347e-01 1.29553413e+00 -8.69007289e-01 -6.74902320e-01
-5.00879824e-01 1.63820541e+00 9.07172933e-02 6.98575735e-01
-1.05396163e+00 -6.80703484e-03 5.36814332e-01 3.40703696e-01
4.89476115e-01 -4.28472310e-01 8.94715115e-02 -1.34358096e+00
3.63336317e-02 -1.05529368e+00 8.46257061e-02 -1.09137619e+00
-1.11334896e+00 8.54503453e-01 1.61966607e-01 -1.43195999e+00
-2.18108356e-01 -2.92768866e-01 -3.28084916e-01 6.26611054e-01
-1.46310699e+00 -1.41117632e+00 -4.81314272e-01 1.26127660e+00
1.00867951e+00 -1.12876624e-01 3.57777506e-01 4.81735080e-01
-3.05138499e-01 4.89007890e-01 -1.38557687e-01 -1.93888485e-01
9.32220876e-01 -8.42352569e-01 3.84676158e-01 1.05923474e+00
-7.89091438e-02 5.26290417e-01 4.98145849e-01 -7.15685666e-01
-1.51081681e+00 -1.23625827e+00 4.71583098e-01 -1.59998044e-01
4.66455519e-01 -2.64436483e-01 -7.79334307e-01 4.76762801e-01
7.46389449e-01 1.21684596e-01 1.67214513e-01 -6.43383086e-01
-1.83852494e-01 -2.71928132e-01 -1.01981676e+00 1.00071359e+00
1.25057399e+00 -3.73446196e-01 -7.51676187e-02 3.90026510e-01
7.37943053e-01 -7.33035147e-01 -7.92307377e-01 1.54245019e-01
5.22049427e-01 -1.16968274e+00 1.18505919e+00 -1.43369019e-01
5.59565306e-01 -6.74067736e-01 -1.35072187e-01 -1.09408331e+00
-6.52805924e-01 -1.22457159e+00 -6.05209954e-02 1.21917951e+00
-6.04467355e-02 -2.78481811e-01 5.52987158e-01 5.75990617e-01
-9.47185233e-02 -4.15197283e-01 -7.17236876e-01 -6.20332241e-01
-1.03217721e-01 -2.49133751e-01 4.90788221e-01 9.63537395e-01
-3.64552021e-01 -3.93549837e-02 -8.15532744e-01 1.73836067e-01
6.73847139e-01 1.04462400e-01 9.97749686e-01 -5.01079857e-01
-5.39403915e-01 -5.10812521e-01 -2.29888052e-01 -1.40627515e+00
-1.28433213e-01 -6.11557603e-01 5.06917760e-03 -1.36197519e+00
1.20334737e-01 -2.92847514e-01 6.33042818e-03 3.67328644e-01
-1.26887143e-01 6.54461086e-01 4.67854857e-01 3.60227734e-01
-5.66119015e-01 5.47166765e-01 1.63791084e+00 -1.33410528e-01
-2.73437589e-01 -4.75064009e-01 -2.35888138e-01 5.84440112e-01
6.00231946e-01 -3.32659066e-01 -6.94075167e-01 -7.72673428e-01
-1.10426657e-01 3.36860567e-01 4.49530870e-01 -1.03524053e+00
3.13008070e-01 -5.09996712e-01 4.27822500e-01 -2.42001027e-01
3.61133784e-01 -8.74806464e-01 5.61810017e-01 3.58435482e-01
-2.70926565e-01 4.06023681e-01 2.20361158e-01 6.04007661e-01
-2.38311917e-01 9.98486057e-02 9.44161832e-01 -7.70803690e-02
-9.02132630e-01 5.36834300e-01 -3.33259016e-01 -1.61321610e-01
1.23759604e+00 -4.84196216e-01 -1.07618175e-01 -5.48506260e-01
-6.69359684e-01 2.71575451e-02 9.05282199e-01 7.45820820e-01
8.55596781e-01 -1.46727848e+00 -7.18198538e-01 5.21135032e-01
-2.40969673e-01 -2.89857060e-01 5.68141341e-01 7.80841827e-01
-9.06478465e-01 8.30607489e-02 -4.22010899e-01 -8.42801154e-01
-1.33202720e+00 6.55974388e-01 3.66832703e-01 1.11627080e-01
-8.65722775e-01 7.35002160e-01 5.09945631e-01 1.40714059e-02
1.55412003e-01 -2.55266786e-01 3.51453602e-01 -3.76419485e-01
7.16964722e-01 2.13039070e-01 -1.76656529e-01 -7.19366252e-01
-1.01089627e-01 1.16547406e+00 -1.63411334e-01 -3.54859471e-01
1.15666807e+00 -5.15841007e-01 -1.39833003e-01 -3.48467715e-02
9.80593443e-01 5.50991073e-02 -1.90266144e+00 -1.61121368e-01
-5.60868740e-01 -9.39697444e-01 8.79119560e-02 -6.32997155e-01
-1.49052262e+00 6.59178853e-01 6.39352560e-01 -2.17531309e-01
1.32621825e+00 -5.45516014e-01 1.01865387e+00 -1.60745442e-01
5.21838605e-01 -1.06110048e+00 4.82614964e-01 2.93071717e-01
9.85126317e-01 -8.52970064e-01 -5.66902512e-04 -5.13552248e-01
-5.56724489e-01 1.22905111e+00 5.62069058e-01 -1.64334074e-01
4.03591841e-01 3.52251440e-01 2.00470924e-01 1.61352709e-01
-8.36225271e-01 1.58672199e-01 3.97062659e-01 7.63907552e-01
4.37080234e-01 -4.18504924e-01 -3.98407698e-01 -3.27247202e-01
2.80047685e-01 3.21580917e-01 4.83152986e-01 7.82588542e-01
-2.30135515e-01 -1.26029348e+00 -4.24803585e-01 -1.99986309e-01
-3.61140847e-01 -1.68398649e-01 9.37521383e-02 6.13627791e-01
1.50478050e-01 9.15860116e-01 4.65350486e-02 -7.25464076e-02
2.30389476e-01 -1.57670349e-01 7.38974929e-01 -2.21276194e-01
-5.20309329e-01 6.76879525e-01 -2.77056485e-01 -9.46929097e-01
-9.70509470e-01 -5.34064651e-01 -9.63994265e-01 -4.85243648e-01
3.67054380e-02 -1.86680406e-01 3.47323507e-01 5.17512918e-01
7.04094827e-01 6.11595273e-01 5.23066580e-01 -1.31776798e+00
4.08549272e-02 -4.94497925e-01 -2.57576197e-01 6.57378197e-01
2.68667430e-01 -4.57624376e-01 6.51837364e-02 8.00292909e-01] | [11.02837085723877, -0.8771497011184692] |
c0daa635-5b41-470a-aab9-a69ccf88ee63 | wind-turbine-blade-surface-damage-detection | 2108.08636 | null | https://arxiv.org/abs/2108.08636v2 | https://arxiv.org/pdf/2108.08636v2.pdf | Wind Turbine Blade Surface Damage Detection based on Aerial Imagery and VGG16-RCNN Framework | In this manuscript, an image analytics based deep learning framework for wind turbine blade surface damage detection is proposed. Turbine blade(s) which carry approximately one-third of a turbine weight are susceptible to damage and can cause sudden malfunction of a grid-connected wind energy conversion system. The surface damage detection of wind turbine blade requires a large dataset so as to detect a type of damage at an early stage. Turbine blade images are captured via aerial imagery. Upon inspection, it is found that the image dataset was limited and hence image augmentation is applied to improve blade image dataset. The approach is modeled as a multi-class supervised learning problem and deep learning methods like Convolutional neural network (CNN), VGG16-RCNN and AlexNet are tested for determining the potential capability of turbine blade surface damage. | ['Harsh S. Dhiman', 'Lagan Sharma', 'Juhi Patel'] | 2021-08-19 | null | null | null | null | ['image-augmentation'] | ['computer-vision'] | [-3.16729635e-01 -1.93307057e-01 4.84044194e-01 4.37793016e-01
-4.85891700e-02 -9.03057337e-01 4.67643403e-02 -3.27620693e-02
1.66718096e-01 3.06380183e-01 1.34320538e-02 -3.47579777e-01
1.76667109e-01 -9.10891294e-01 -3.80080752e-02 -9.25526261e-01
-3.51501346e-01 -2.55248874e-01 -1.98155027e-02 -3.38683754e-01
1.91691756e-01 9.59026456e-01 -1.61721933e+00 2.42250755e-01
5.46992362e-01 8.95878553e-01 3.26476336e-01 1.10479271e+00
4.21823770e-01 2.50027716e-01 -1.10132289e+00 3.49825293e-01
7.78360248e-01 7.28788525e-02 -8.15476298e-01 3.46657306e-01
3.68052065e-01 -7.66885102e-01 9.62035805e-02 9.66192544e-01
8.77460837e-01 1.34700695e-02 7.85671175e-01 -9.78824854e-01
-4.46335018e-01 -2.91732818e-01 -5.19453347e-01 9.40533102e-01
-2.80070275e-01 5.42814374e-01 6.75143600e-01 -1.29402840e+00
1.45979419e-01 8.26369941e-01 7.69865513e-01 1.34956896e-01
-4.40518171e-01 -1.19587861e-01 -4.58798170e-01 9.33412388e-02
-1.10962808e+00 9.69968140e-02 8.25891674e-01 -8.10874462e-01
1.24187541e+00 -2.07735412e-02 5.67382514e-01 4.46437269e-01
6.82829618e-01 1.42027751e-01 8.04545641e-01 -2.23456159e-01
4.08051997e-01 -6.24204516e-01 -3.16799879e-01 7.66687274e-01
6.44721448e-01 4.19731468e-01 -1.46461189e-01 2.00748891e-01
8.28817368e-01 -1.43501088e-01 -1.14619859e-01 4.49922740e-01
-3.32332492e-01 6.76680207e-01 8.60492110e-01 1.78734884e-01
-6.03356123e-01 -1.82502717e-01 9.12303984e-01 4.08746600e-01
5.31410873e-01 6.63285315e-01 -8.04896235e-01 1.06694445e-01
-7.59858787e-01 -7.79393304e-05 2.74812788e-01 1.47259727e-01
3.13726544e-01 1.07140398e+00 2.86549360e-01 5.78159153e-01
2.37433299e-01 3.42493981e-01 4.86094117e-01 -4.19456422e-01
-5.93279600e-02 6.47489071e-01 1.47405222e-01 -9.90261018e-01
-4.17829633e-01 -3.23783696e-01 -1.03490376e+00 1.11163843e+00
-1.25435710e-01 -8.40684593e-01 -9.19494450e-01 4.64129508e-01
1.98866993e-01 2.66095698e-01 -7.22086057e-02 1.34557354e+00
8.68262410e-01 8.11707854e-01 1.07171409e-01 3.88318300e-02
1.51969576e+00 -7.54833162e-01 -5.94627559e-01 -1.03871725e-01
7.53296912e-01 -5.96388757e-01 9.18050289e-01 4.85928655e-01
-8.51839602e-01 -8.61421168e-01 -1.58536279e+00 1.11747198e-01
-7.53656149e-01 6.47978246e-01 4.73235995e-01 6.01062775e-01
-1.01142907e+00 5.54235697e-01 -6.63500428e-01 -2.02208579e-01
5.45444787e-01 2.34758660e-01 -2.69305319e-01 2.78887331e-01
-9.28003192e-01 9.67842460e-01 2.00615421e-01 4.96716380e-01
-1.52634859e+00 -7.47249365e-01 -6.96855307e-01 2.85875320e-01
-1.73097193e-01 -7.05499530e-01 1.12470174e+00 -5.89496493e-01
-1.24777448e+00 6.29091144e-01 4.61110473e-01 -3.30653369e-01
-7.88653567e-02 -3.74174744e-01 -4.50536519e-01 5.07571638e-01
-1.66902557e-01 1.81635603e-01 1.41612625e+00 -1.34645915e+00
-1.01627040e+00 -4.02710527e-01 9.86308008e-02 2.07602143e-01
-5.12562215e-01 6.99060783e-02 9.13795590e-01 -7.66860008e-01
-4.91330743e-01 -4.86012697e-01 -1.38811752e-01 -2.72103459e-01
-3.42470050e-01 -2.71566689e-01 1.59831405e+00 -9.70264375e-01
8.34455311e-01 -1.91718245e+00 -2.89134920e-01 -2.48410821e-01
1.42242834e-01 9.64241505e-01 -3.04620303e-02 7.70313323e-01
-3.18227679e-01 4.73021418e-01 -5.72565943e-02 1.03447340e-01
-5.43491602e-01 3.57105047e-01 -2.83843040e-01 5.69501817e-01
7.30281472e-01 3.47713321e-01 -7.30654657e-01 2.81110018e-01
4.73805547e-01 1.40622094e-01 -2.30403095e-02 4.94468063e-01
2.04882443e-01 1.13067627e-01 -2.25083709e-01 1.12097502e+00
8.10612619e-01 5.40149391e-01 -5.13354957e-01 -5.64143717e-01
-5.02528429e-01 -2.57042259e-01 -5.61507702e-01 9.82647181e-01
-7.01425314e-01 6.75494134e-01 4.32226002e-01 -9.78656530e-01
1.08847690e+00 7.38906801e-01 4.49622124e-01 7.89596736e-02
2.91479439e-01 8.19481388e-02 1.22419663e-01 -1.06129396e+00
4.92263734e-01 -3.35193664e-01 4.81882393e-01 -1.37960181e-01
1.98173220e-03 -2.40723506e-01 -1.51550159e-01 -4.35689092e-01
1.31406033e+00 -3.05029184e-01 -1.77112788e-01 -5.91102481e-01
4.32995886e-01 4.16063607e-01 6.04924798e-01 -4.07108031e-02
-2.89242297e-01 7.63434112e-01 1.84066251e-01 -9.71620142e-01
-1.30732274e+00 -7.30847478e-01 -4.42959294e-02 5.36507189e-01
-1.04946181e-01 3.58357318e-02 -9.21219230e-01 -6.23062015e-01
1.24913231e-01 2.14117974e-01 -3.36924136e-01 -2.86511213e-01
-2.68772185e-01 -8.67180705e-01 4.79548842e-01 9.85171497e-01
6.64050937e-01 -1.21859562e+00 -1.28932095e+00 -2.51178239e-02
6.12120867e-01 -7.29365230e-01 -9.18380916e-02 3.52827609e-01
-1.05784595e+00 -1.44302213e+00 -3.75276834e-01 -1.02027333e+00
1.00240088e+00 6.33893073e-01 9.89700198e-01 8.06400359e-01
-1.14083886e+00 2.25561783e-01 -6.60414219e-01 -7.28034258e-01
-1.00980967e-01 -1.17208749e-01 3.44072670e-01 -3.88996661e-01
-6.03702292e-02 -4.42320466e-01 -1.18159282e+00 -1.71730965e-01
-8.54000330e-01 -6.63004637e-01 5.51922858e-01 1.16651142e+00
-3.02198026e-02 9.66526568e-01 8.56010139e-01 -3.67065489e-01
9.84926403e-01 -6.81644022e-01 -5.23337185e-01 -4.76052403e-01
-6.40384793e-01 -7.96165586e-01 1.11571872e+00 1.30720869e-01
-9.94982660e-01 -5.09210974e-02 -1.26831606e-01 -6.70545757e-01
-5.04206717e-01 1.07117414e+00 3.60859074e-02 -1.57843605e-02
5.62142193e-01 -2.73172349e-01 1.00264467e-01 -7.89189219e-01
-3.01589724e-02 8.27812493e-01 7.12802529e-01 1.68780759e-01
1.08772647e+00 4.17353630e-01 1.68637723e-01 -1.44046736e+00
-3.32400441e-01 -7.30571866e-01 -1.09241974e+00 -5.63834846e-01
8.65881860e-01 -1.00523663e+00 -6.33649647e-01 8.53339612e-01
-1.06382823e+00 -3.25063884e-01 -3.22133631e-01 1.35059997e-01
2.69342482e-01 2.98120439e-01 -5.66661596e-01 -9.53922689e-01
-9.74004984e-01 -7.61548519e-01 1.16330278e+00 3.42997044e-01
3.53165329e-01 -1.24726975e+00 -1.84482679e-01 2.94332474e-01
2.53992677e-01 5.46777487e-01 9.47644293e-01 -1.26614079e-01
1.59042627e-01 -5.12451887e-01 3.84691358e-02 9.23828304e-01
7.63244748e-01 7.12119758e-01 -1.05352712e+00 -8.59815240e-01
1.91920429e-01 -2.99284220e-01 6.59052312e-01 5.62329352e-01
9.11686003e-01 -3.20570588e-01 1.46963492e-01 5.63923478e-01
1.76835430e+00 2.81181186e-01 5.35919368e-01 2.39577875e-01
7.22593725e-01 2.44863510e-01 6.10490203e-01 7.32841849e-01
-1.80579841e-01 6.35223240e-02 1.34956646e+00 -7.31586039e-01
-2.40865052e-01 2.48606861e-01 5.44091284e-01 8.55437219e-01
-3.18314999e-01 -3.37659210e-01 -8.87586415e-01 1.19263899e+00
-9.29255903e-01 -8.63468289e-01 -5.87796330e-01 1.87587631e+00
9.51277390e-02 -1.48885995e-01 -6.70101494e-02 5.83196461e-01
5.39810717e-01 1.74052447e-01 -4.45499748e-01 -9.77331638e-01
-1.22551695e-01 4.36456650e-01 5.99459171e-01 1.67791113e-01
-1.30512452e+00 6.60475552e-01 6.71445322e+00 2.65043646e-01
-1.36235750e+00 -6.62135482e-02 5.97050190e-01 3.12229544e-01
3.20221663e-01 -1.31090641e-01 -3.18514884e-01 4.42207336e-01
5.64239144e-01 -3.43874171e-02 5.66475615e-02 8.68307114e-01
6.78038538e-01 -2.83453286e-01 -1.69842854e-01 4.97491658e-01
-1.98625773e-01 -1.14778233e+00 -8.36124867e-02 2.35570461e-01
8.31370056e-01 1.84663445e-01 -2.31812328e-01 -3.27624381e-01
2.25356609e-01 -1.03229725e+00 6.33240938e-02 3.26777011e-01
7.43672550e-01 -7.20110655e-01 1.35103714e+00 2.20618635e-01
-1.52511930e+00 -5.04329145e-01 -8.89481366e-01 -7.41198063e-01
-6.11588806e-02 7.90458500e-01 -1.26190925e+00 7.64705300e-01
1.23832667e+00 7.39331305e-01 -6.10987425e-01 1.14337385e+00
-3.29303563e-01 1.19702864e+00 -1.50361791e-01 5.05065203e-01
3.34548026e-01 -1.44285679e-01 6.39174163e-01 7.28458524e-01
6.48627877e-01 1.02466151e-01 3.76566708e-01 5.40088952e-01
1.36493696e-02 -3.06251585e-01 -1.00416827e+00 1.27731457e-01
6.35964870e-01 2.04163194e+00 -6.78981125e-01 3.90554965e-02
-2.89983571e-01 7.09384859e-01 -2.59759396e-01 1.51411667e-01
-1.09485559e-01 -2.77259171e-01 1.12986147e+00 1.83378309e-01
5.17512798e-01 -6.54446900e-01 -6.70392334e-01 -3.28892529e-01
-2.19896317e-01 -2.63922334e-01 4.22682881e-01 -1.20732617e+00
-1.54808247e+00 3.76540422e-01 -5.00350595e-01 -1.38808036e+00
1.94083855e-01 -1.34839356e+00 -1.65262258e+00 1.09184873e+00
-1.55689800e+00 -1.30756497e+00 -5.51693738e-01 -4.36683595e-02
1.14994383e+00 -6.82052791e-01 8.33887160e-01 -1.31407291e-01
-9.79482412e-01 -2.23885566e-01 1.25379026e-01 2.51381963e-01
-3.72675136e-02 -1.81077325e+00 5.55447340e-01 1.21800613e+00
-4.58338410e-01 6.01883046e-02 6.50560021e-01 -9.56375897e-01
-1.76133800e+00 -1.51140130e+00 2.17849195e-01 -1.20583124e-01
5.46188891e-01 3.65191787e-01 -9.69072700e-01 3.21601816e-02
9.34779227e-01 5.02232373e-01 5.59754431e-01 -5.10338366e-01
4.74832833e-01 1.71727929e-02 -1.37828720e+00 1.62260979e-01
4.52240825e-01 -1.44464329e-01 -5.32145321e-01 4.58397061e-01
2.35636085e-02 -2.40917966e-01 -1.44208574e+00 5.08819103e-01
-3.74802649e-02 -6.38653874e-01 8.73056889e-01 -8.28978360e-01
6.74513936e-01 -6.89383805e-01 4.80886400e-01 -1.95418203e+00
-5.28151691e-01 -5.67925453e-01 -1.78201541e-01 1.32680619e+00
-1.91007927e-01 -2.39409789e-01 6.01024449e-01 -6.84626773e-02
-9.79523659e-01 -9.06359851e-01 -8.94910932e-01 -8.44412506e-01
3.89314502e-01 3.60040694e-01 4.56256688e-01 8.74267101e-01
-2.01286644e-01 1.56501368e-01 -1.93114430e-02 1.03258049e+00
1.65150568e-01 -2.66855538e-01 3.65848362e-01 -1.49907839e+00
8.20062518e-01 -2.26543710e-01 -5.14054894e-01 -3.40632588e-01
-1.40817165e-01 -4.48935270e-01 8.31428841e-02 -1.76601911e+00
-6.51120663e-01 6.67735934e-02 -1.90430835e-01 7.46112585e-01
-2.54497409e-01 2.94341087e-01 -2.16778085e-01 1.93459421e-01
6.32010877e-01 5.90058148e-01 1.38998163e+00 -3.31997305e-01
2.41977364e-01 3.64182182e-02 -2.53477514e-01 6.98485911e-01
1.20740259e+00 -6.78128749e-02 -3.28813046e-01 -5.63534558e-01
1.91920549e-01 -1.12864845e-01 6.71664119e-01 -1.03095686e+00
1.90275565e-01 3.81814241e-02 5.58220088e-01 -8.31937075e-01
-2.39956766e-01 -8.32827926e-01 -3.47421974e-01 8.19638729e-01
5.21623731e-01 6.39257610e-01 5.04221976e-01 9.14077684e-02
-4.46996629e-01 -5.56977332e-01 5.26640832e-01 -2.27173015e-01
-9.11940396e-01 2.42046401e-01 -1.04853272e+00 -4.61815029e-01
8.72463286e-01 -4.36297953e-01 -5.42950630e-01 2.60454863e-01
-2.35054865e-01 1.88218027e-01 2.87172318e-01 5.42899609e-01
9.18527484e-01 -1.01907802e+00 -8.41555774e-01 3.05703342e-01
-3.93713623e-01 3.16302210e-01 4.33906078e-01 5.04166424e-01
-1.22862053e+00 -9.61844400e-02 -6.70803964e-01 -2.76770741e-01
-1.21790421e+00 4.37957704e-01 5.81243217e-01 1.21843643e-01
-8.64148378e-01 8.23426783e-01 -2.65828222e-01 2.02354506e-01
-4.42794055e-01 -4.68400657e-01 -6.32196486e-01 1.19157344e-01
1.62039891e-01 8.44725311e-01 6.91128492e-01 -6.49027109e-01
-1.47298202e-01 4.27431613e-01 3.84212196e-01 7.24692047e-01
1.38189256e+00 -4.88758199e-02 -2.02510551e-01 1.38133587e-02
8.27182353e-01 -3.52658957e-01 -1.53338146e+00 7.61662006e-01
-1.91830531e-01 -5.11730373e-01 9.37231004e-01 -9.35480952e-01
-1.66834176e+00 1.18124545e+00 1.06142175e+00 7.70572305e-01
1.44601667e+00 -7.39185929e-01 9.26205873e-01 2.40723416e-01
3.06162778e-02 -1.23603845e+00 2.69775003e-01 4.83725995e-01
1.11271024e+00 -1.29131830e+00 1.38276145e-01 -3.23448896e-01
-4.08128142e-01 1.52981281e+00 9.61454988e-01 -4.64379966e-01
7.41281033e-01 6.62353635e-01 4.35423970e-01 -8.14743996e-01
-5.97466826e-01 -3.11290324e-01 -1.82983826e-03 1.02327228e+00
3.42771024e-01 8.23089257e-02 3.87508720e-02 3.25309366e-01
-1.73408864e-03 -2.48932853e-01 5.52072406e-01 1.00085640e+00
-7.04464197e-01 -6.27853036e-01 -5.33578634e-01 7.99236178e-01
-5.75314283e-01 -7.82003254e-02 -5.38854659e-01 4.43720043e-01
4.79530603e-01 1.17015004e+00 2.44863063e-01 -5.14460266e-01
3.39264303e-01 -1.59876838e-01 -1.61390528e-01 -8.08306932e-01
-1.11525166e+00 -4.06526864e-01 -9.35691223e-02 1.59835532e-01
-1.13177076e-01 -6.06839478e-01 -1.19445813e+00 6.03383267e-03
-6.77048862e-01 -9.02883857e-02 6.44014716e-01 9.74096060e-01
4.92756218e-01 1.00575149e+00 1.40091538e+00 -1.16685009e+00
-5.16973138e-01 -1.29114616e+00 -1.03923285e+00 2.59849221e-01
3.99057359e-01 -9.82207000e-01 -5.04210591e-01 6.89837635e-01] | [6.959629535675049, 2.190608501434326] |
fc73558a-23c6-4051-b5c5-0f2353041b58 | source-free-domain-adaptation-for-real-world | 2207.06644 | null | https://arxiv.org/abs/2207.06644v1 | https://arxiv.org/pdf/2207.06644v1.pdf | Source-Free Domain Adaptation for Real-world Image Dehazing | Deep learning-based source dehazing methods trained on synthetic datasets have achieved remarkable performance but suffer from dramatic performance degradation on real hazy images due to domain shift. Although certain Domain Adaptation (DA) dehazing methods have been presented, they inevitably require access to the source dataset to reduce the gap between the source synthetic and target real domains. To address these issues, we present a novel Source-Free Unsupervised Domain Adaptation (SFUDA) image dehazing paradigm, in which only a well-trained source model and an unlabeled target real hazy dataset are available. Specifically, we devise the Domain Representation Normalization (DRN) module to make the representation of real hazy domain features match that of the synthetic domain to bridge the gaps. With our plug-and-play DRN module, unlabeled real hazy images can adapt existing well-trained source networks. Besides, the unsupervised losses are applied to guide the learning of the DRN module, which consists of frequency losses and physical prior losses. Frequency losses provide structure and style constraints, while the prior loss explores the inherent statistic property of haze-free images. Equipped with our DRN module and unsupervised loss, existing source dehazing models are able to dehaze unlabeled real hazy images. Extensive experiments on multiple baselines demonstrate the validity and superiority of our method visually and quantitatively. | ['Feng Zhao', 'Man Zhou', 'Qi Zhu', 'Yajing Liu', 'Jie Huang', 'Hu Yu'] | 2022-07-14 | null | null | null | null | ['image-dehazing', 'source-free-domain-adaptation'] | ['computer-vision', 'computer-vision'] | [ 3.81972998e-01 -1.90988332e-01 1.62293464e-01 -3.41967314e-01
-6.09924197e-01 -1.35425761e-01 5.41008174e-01 -3.61493349e-01
-2.43054375e-01 7.76947975e-01 8.43304768e-02 1.05836079e-01
1.33175969e-01 -1.04556262e+00 -8.54250371e-01 -1.19209898e+00
5.56968033e-01 5.51627912e-02 5.50438821e-01 -5.32002687e-01
2.45057438e-02 1.28037035e-01 -1.56187260e+00 1.13902181e-01
1.38562500e+00 1.06495106e+00 3.92288119e-01 4.25199598e-01
-1.22041121e-01 8.80431652e-01 -8.90035629e-01 -4.86111678e-02
5.35854578e-01 -5.53046584e-01 -1.80170611e-01 2.00696871e-01
3.88418347e-01 -6.09332860e-01 -6.15241647e-01 1.30887139e+00
5.61568201e-01 2.95228094e-01 9.18067932e-01 -1.37366891e+00
-1.12891030e+00 8.71139020e-02 -4.96459633e-01 1.67009458e-01
-2.39701554e-01 3.06403607e-01 3.26967984e-01 -1.10329008e+00
3.67476016e-01 9.48306441e-01 3.82792085e-01 6.53478742e-01
-9.63159144e-01 -1.06543303e+00 -4.91202772e-02 2.71816492e-01
-1.45836449e+00 -4.78219837e-01 1.17210710e+00 -4.36292917e-01
3.84079725e-01 -2.90394723e-01 2.55833417e-01 1.10472941e+00
4.66301339e-03 6.31088614e-01 8.96568298e-01 -4.72002238e-01
3.84463549e-01 4.34119731e-01 -3.84005934e-01 2.41907790e-01
2.25885987e-01 3.40956450e-01 -4.02949810e-01 2.94879913e-01
7.88864732e-01 1.63770258e-01 -3.25462490e-01 -5.96300602e-01
-8.60096455e-01 8.56885135e-01 5.49038470e-01 -7.80091528e-03
-2.15117157e-01 -1.39042944e-01 -3.52517888e-02 5.73985696e-01
7.14046597e-01 2.79883295e-01 -8.18587765e-02 6.43494964e-01
-1.05317509e+00 1.60646260e-01 3.56274009e-01 1.21172023e+00
1.27279985e+00 5.31576753e-01 -3.92962024e-02 1.10265648e+00
2.17885971e-01 9.54494596e-01 5.59729218e-01 -8.45961154e-01
3.77298445e-01 4.77201015e-01 1.38013273e-01 -8.30754161e-01
2.54674435e-01 -4.08598542e-01 -9.90522206e-01 6.31878316e-01
2.51423985e-01 7.52576962e-02 -1.23107457e+00 1.44091654e+00
3.70293409e-01 3.74829352e-01 3.75366002e-01 9.76879060e-01
7.28488505e-01 1.00495005e+00 -5.30205406e-02 -2.56137289e-02
8.52383912e-01 -9.55111444e-01 -7.60295331e-01 -3.10632259e-01
1.21874996e-01 -7.54858494e-01 1.21242404e+00 3.15969408e-01
-9.84090865e-01 -6.36002898e-01 -1.49952853e+00 -1.57744795e-01
-6.22728646e-01 -6.18368350e-02 -7.90735185e-02 6.14640117e-01
-9.01616454e-01 1.60910651e-01 -4.52270597e-01 -1.35008067e-01
5.64631283e-01 1.46837711e-01 -3.28378119e-02 -4.51122016e-01
-1.50102615e+00 7.14168131e-01 4.30235833e-01 -1.47293702e-01
-1.56894827e+00 -9.59318697e-01 -9.05604422e-01 -9.45628807e-02
2.45713174e-01 -5.95921636e-01 9.18809295e-01 -1.31256831e+00
-1.69566751e+00 8.27253044e-01 2.21966892e-01 -4.75846291e-01
3.31428379e-01 -2.05146596e-01 -9.05993998e-01 4.68405336e-01
1.13623202e-01 6.40060604e-01 1.62276840e+00 -1.56886494e+00
-5.88568747e-01 3.72948796e-02 -1.06111325e-01 2.32075617e-01
-5.17287910e-01 -2.30240762e-01 -1.54418990e-01 -1.12872648e+00
-2.56165981e-01 -3.98955941e-01 3.00208684e-02 3.36618096e-01
-9.00020003e-02 2.31831446e-01 1.14247489e+00 -7.60499120e-01
9.58526611e-01 -2.70325160e+00 -1.51904538e-01 8.78466219e-02
2.84745842e-01 5.67952573e-01 -2.82956839e-01 1.90801397e-01
-9.15009379e-02 -3.01408082e-01 -6.38277113e-01 -1.12701036e-01
-1.69758752e-01 1.74806476e-01 -6.42940104e-01 5.57031214e-01
5.71906507e-01 5.68910837e-01 -8.33718538e-01 -5.14814198e-01
3.36860478e-01 5.50028384e-01 -6.25973046e-01 6.06966376e-01
-4.28721964e-01 5.20626605e-01 -1.87970117e-01 5.68283379e-01
1.17506623e+00 -8.22886825e-02 -5.18644452e-01 3.87310982e-02
5.36276363e-02 1.72690734e-01 -9.71896350e-01 1.61168289e+00
-4.86200005e-01 4.33601618e-01 2.52074063e-01 -1.07423198e+00
1.08349109e+00 1.90569773e-01 5.26684709e-02 -8.62017632e-01
-1.92578770e-02 3.37467968e-01 -4.05768991e-01 -4.38248962e-01
3.36689353e-01 -5.51988721e-01 2.94499457e-01 1.26125097e-01
2.13014141e-01 -4.11131650e-01 -2.78753936e-01 1.56428829e-01
6.84162736e-01 -1.49694989e-02 1.67851876e-02 -1.62060827e-01
7.21349895e-01 1.31090060e-01 5.93351841e-01 4.92551982e-01
-2.73481637e-01 1.07881558e+00 -3.87572893e-03 -1.68841720e-01
-1.27560771e+00 -1.51423681e+00 -7.86835104e-02 9.90498841e-01
5.52103877e-01 2.20137551e-01 -7.52111852e-01 -5.24954021e-01
-2.94493586e-01 7.76779950e-01 -6.02380574e-01 -6.71324253e-01
-5.21668971e-01 -6.56708539e-01 4.61892784e-01 1.70369953e-01
1.13390303e+00 -7.70290017e-01 -1.30301639e-01 2.13555694e-01
-1.93696156e-01 -1.09302747e+00 -5.74175179e-01 -1.36287147e-02
-5.59099853e-01 -8.17139447e-01 -1.06161654e+00 -1.03422189e+00
5.38184106e-01 7.00582564e-01 8.96587431e-01 -4.48227040e-02
-2.30287835e-02 1.91105217e-01 -4.50515121e-01 -7.73948252e-01
-5.97751081e-01 -1.85386077e-01 1.37425112e-02 4.27840054e-01
4.12528783e-01 -1.01206005e+00 -9.03541088e-01 4.66016054e-01
-1.53047955e+00 -1.57992989e-01 4.97677207e-01 8.81375253e-01
3.71067137e-01 5.89307606e-01 5.22115648e-01 -6.06104970e-01
9.33684483e-02 -7.02032208e-01 -7.28833795e-01 4.96004941e-04
-7.51986265e-01 -1.08008794e-01 7.76862025e-01 -6.25735223e-01
-1.51593328e+00 -1.61361232e-01 3.67120877e-02 -8.74110162e-01
-2.52972782e-01 7.94080123e-02 -6.28519177e-01 -2.39257365e-01
9.99055505e-01 6.55976772e-01 2.48181686e-01 -2.88591772e-01
3.10563296e-01 6.93521202e-01 9.14405107e-01 -3.88859510e-01
1.64148772e+00 8.73360038e-01 -4.09177035e-01 -9.46384311e-01
-7.46724665e-01 -4.31225896e-01 -4.44102317e-01 2.81904228e-02
8.83686662e-01 -1.54865777e+00 2.31291473e-01 8.35434675e-01
-9.69516873e-01 -5.95509648e-01 -4.59309369e-01 3.50916088e-01
-3.36325675e-01 4.05401766e-01 -3.10894430e-01 -6.47575498e-01
1.15484558e-02 -8.18474829e-01 9.42414343e-01 2.11133733e-01
5.04400969e-01 -1.03248060e+00 -1.14408597e-01 2.01508582e-01
7.25657880e-01 1.16787001e-01 1.08456504e+00 -3.09259802e-01
-8.15399289e-01 2.11670190e-01 -3.64605695e-01 1.03076804e+00
2.99471051e-01 -2.77979702e-01 -1.14552462e+00 -3.29270929e-01
4.26322997e-01 -1.27857596e-01 1.07669091e+00 3.10092747e-01
9.88076448e-01 -3.55716854e-01 7.68842082e-03 1.01581621e+00
1.49303973e+00 1.84041873e-01 9.94442344e-01 5.64565659e-01
7.12613463e-01 6.49800241e-01 5.07603884e-01 2.39181861e-01
1.61883965e-01 4.41153169e-01 6.12280011e-01 -3.39609623e-01
-5.83409846e-01 -4.04758722e-01 6.05415821e-01 5.79293430e-01
3.23810458e-01 -4.01402622e-01 -6.10114455e-01 8.28715801e-01
-1.38771427e+00 -6.84957862e-01 2.45756939e-01 2.09188843e+00
9.89067435e-01 7.79092386e-02 -1.15027884e-03 8.14360306e-02
7.25292087e-01 3.65794182e-01 -6.99900448e-01 1.48763016e-01
-5.06645799e-01 1.73335597e-01 5.41814983e-01 4.03456718e-01
-1.18988574e+00 1.02264452e+00 5.66382217e+00 1.16347182e+00
-1.29805768e+00 2.29991421e-01 2.52200544e-01 -6.30181581e-02
-5.24859726e-01 -2.36066043e-01 -4.95848894e-01 7.32861757e-01
8.38457167e-01 -1.35984033e-01 5.61716318e-01 7.98762143e-01
1.24985553e-01 2.81122774e-02 -7.34660804e-01 1.07478011e+00
9.81441662e-02 -1.28095126e+00 3.38299900e-01 8.03168397e-03
9.66412127e-01 -3.79736349e-02 4.52412039e-01 2.94675261e-01
4.18781340e-01 -9.20160949e-01 7.53864110e-01 2.93689668e-01
1.10742092e+00 -6.41362548e-01 5.79676509e-01 3.66490155e-01
-9.03604388e-01 -7.24646300e-02 -8.25495541e-01 1.13689207e-01
-1.11420132e-01 8.99984479e-01 -6.93709373e-01 5.96434176e-01
9.45985675e-01 8.55605900e-01 -4.89468426e-01 7.70636082e-01
-4.12739098e-01 8.61636817e-01 -1.68317258e-01 6.68845952e-01
1.44759357e-01 -2.62380719e-01 5.29592454e-01 1.00252426e+00
3.83295268e-01 2.29268566e-01 -2.15992123e-01 1.14220393e+00
-8.46440345e-02 -2.25676015e-01 -7.34714091e-01 1.40399501e-01
4.59309459e-01 6.71850860e-01 -1.66101918e-01 -2.97011197e-01
-4.63761538e-01 1.08812106e+00 -1.04560666e-01 7.70200312e-01
-9.16157901e-01 -6.38311744e-01 1.00343406e+00 3.22270095e-01
4.54618335e-01 -1.78343326e-01 -3.20864618e-01 -1.32872617e+00
9.16742906e-03 -8.87714982e-01 2.35909402e-01 -9.96370971e-01
-1.66586185e+00 3.43318313e-01 6.71736225e-02 -1.61201155e+00
2.19962925e-01 -5.60396969e-01 -7.31860816e-01 1.06749678e+00
-2.28465366e+00 -1.09776306e+00 -7.54847467e-01 1.10627818e+00
6.06134236e-01 -4.26508635e-01 3.23099047e-01 5.22075713e-01
-4.72864836e-01 7.43081868e-01 5.28332710e-01 1.86103910e-01
1.20541239e+00 -1.01127160e+00 1.32033914e-01 1.15084136e+00
-3.16681027e-01 4.07580256e-01 7.90298760e-01 -4.45608407e-01
-8.80008280e-01 -1.57281494e+00 2.54228354e-01 -4.85578924e-01
4.76930380e-01 -4.08231020e-01 -1.42978382e+00 4.19653594e-01
1.58387467e-01 1.35004982e-01 4.91349041e-01 -8.59751821e-01
-6.83130383e-01 -4.50976193e-01 -1.14055276e+00 4.46972847e-01
8.18478286e-01 -6.48345888e-01 -7.48576760e-01 1.95025340e-01
1.11772001e+00 -2.59071112e-01 -5.01142502e-01 4.33578551e-01
4.86057922e-02 -1.07061279e+00 1.25967312e+00 -1.45500660e-01
7.13015616e-01 -7.36809433e-01 -1.72603235e-01 -1.46788347e+00
-2.60353446e-01 -5.38550079e-01 -5.19818952e-03 1.36831760e+00
1.12479538e-01 -6.73077345e-01 7.30889618e-01 1.08664304e-01
-3.25380206e-01 -3.80459540e-02 -6.89082026e-01 -1.10069799e+00
4.97105420e-01 -1.47883326e-01 9.00564611e-01 1.01058352e+00
-6.90662622e-01 -2.40908447e-03 -5.35111070e-01 6.57568872e-01
9.09007609e-01 -1.61156878e-01 7.78941929e-01 -9.36931491e-01
7.34980032e-03 -1.24674030e-01 -1.38749436e-01 -1.10517573e+00
1.57784253e-01 -6.42965972e-01 2.87302613e-01 -1.23848689e+00
-2.17346832e-01 -4.53573734e-01 -5.27629018e-01 1.75992459e-01
-1.64432794e-01 2.56094337e-01 -1.91293508e-02 6.70912564e-01
-2.69529223e-01 1.07795060e+00 1.31303167e+00 -6.12478197e-01
-1.93951890e-01 -2.30814219e-01 -7.08077133e-01 5.46145320e-01
7.61789441e-01 -7.22979665e-01 -7.63307095e-01 -6.46501720e-01
-6.31700009e-02 -4.20857608e-01 5.51371634e-01 -1.27324331e+00
2.92132974e-01 -3.90238672e-01 4.34659302e-01 -2.96466649e-01
2.48000309e-01 -8.50313485e-01 -1.01036675e-01 3.90471108e-02
-1.11997411e-01 -6.90342247e-01 6.24498911e-02 8.46535802e-01
-6.17036879e-01 9.01882574e-02 1.34920263e+00 -3.10894586e-02
-9.78532612e-01 3.66701990e-01 -2.80043483e-01 1.70719221e-01
8.92599344e-01 -5.33811033e-01 -4.48347121e-01 -4.34899032e-01
-1.84178367e-01 1.83315381e-01 8.22477221e-01 3.09838563e-01
8.88005793e-01 -1.23088479e+00 -8.36005270e-01 6.07742310e-01
4.85319644e-01 5.71623564e-01 3.68564367e-01 4.84211296e-01
-6.45925581e-01 -6.74720854e-02 -3.94702822e-01 -3.41812551e-01
-5.23014784e-01 1.00272977e+00 4.48805630e-01 2.46764243e-01
-7.11951256e-01 8.82358849e-01 1.07916892e+00 -3.94709527e-01
4.55030203e-02 -1.72838159e-02 -1.77384994e-03 -2.98428893e-01
7.02352643e-01 2.81774014e-01 4.79249135e-02 -3.54302883e-01
-1.46768317e-01 3.90042335e-01 6.12228923e-02 9.65229422e-03
1.22599554e+00 -4.20704782e-01 -7.38024041e-02 1.13286316e-01
1.09158540e+00 2.67086811e-02 -1.80744100e+00 -5.96667171e-01
-3.13234895e-01 -5.95218062e-01 4.13331343e-03 -6.56042397e-01
-1.06794286e+00 1.20997477e+00 7.29889154e-01 -8.68180394e-02
1.65553451e+00 -1.70628533e-01 1.12900138e+00 1.27039388e-01
2.19976634e-01 -1.03818643e+00 4.44403797e-01 4.55704510e-01
6.88321531e-01 -1.27643633e+00 -2.18882754e-01 -5.79616904e-01
-5.38110018e-01 7.19780624e-01 7.66811848e-01 -3.24859202e-01
8.40945542e-01 -9.39974487e-02 3.80657941e-01 -5.89310862e-02
-3.81794691e-01 -3.22512984e-02 1.03324957e-01 1.15523493e+00
-2.67017365e-01 -3.65812927e-01 2.61069059e-01 5.12275279e-01
-1.05684064e-01 -6.58382103e-02 7.01476157e-01 6.68681681e-01
-6.90836251e-01 -8.90373468e-01 -6.32273614e-01 -7.09603280e-02
-1.23595238e-01 -2.71245480e-01 -4.05064672e-02 7.35178530e-01
3.61267537e-01 1.05856490e+00 -1.06540017e-01 -3.47706169e-01
3.01920652e-01 -1.32951438e-02 1.57019898e-01 -7.79008567e-01
-6.48213923e-02 3.71634439e-02 -4.68231112e-01 -2.38443896e-01
-4.24157232e-01 -9.57130194e-02 -1.03013325e+00 -2.06467241e-01
-1.23992205e-01 1.72463611e-01 3.41154993e-01 8.10300171e-01
3.44686031e-01 4.84839648e-01 8.58468950e-01 -7.91731775e-01
-4.46764290e-01 -8.08622479e-01 -1.09397316e+00 5.07488012e-01
8.71297181e-01 -8.53211701e-01 -7.30827272e-01 5.63079476e-01] | [10.938694953918457, -3.1066362857818604] |
3eda3762-97b4-4255-b264-7820e27d165f | spatiotemporal-representation-learning-on | null | null | https://openreview.net/forum?id=Jh9VxCkrEZn | https://openreview.net/pdf?id=Jh9VxCkrEZn | Spatiotemporal Representation Learning on Time Series with Dynamic Graph ODEs | Spatiotemporal representation learning on multivariate time series has received tremendous attention in forecasting traffic and energy data. Recent works either rely on complicated discrete neural architectures or graph priors, hindering their effectiveness and applications in the real world. In this paper, inspired by neural ordinary differential equations and graph structure learning, we propose a fully continuous model named Dynamic Graph ODE (DyG-ODE) to capture both long-range spatial and temporal dependencies to learn expressive representations on arbitrary multivariate time series data without being restricted by rigid preconditions (e.g., graph priors). For modeling the continuous dynamics of spatiotemporal clues, we design a simple yet powerful dynamic graph ODE by coupling the proposed spatial and temporal ODEs, which not only allows the model to obtain infinite spatial and temporal receptive fields but also reduces numerical errors and model complexity significantly. Our empirical evaluations demonstrate the superior effectiveness and efficiency of DyG-ODE on a number of benchmark datasets. | ['Shirui Pan', 'Bin Yang', 'Yu Zheng', 'Yuan-Fang Li', 'Ming Jin'] | 2021-09-29 | null | null | null | null | ['graph-structure-learning'] | ['graphs'] | [-1.53899819e-01 -2.51668483e-01 -1.32032380e-01 -7.33176023e-02
-4.90401611e-02 -4.52250123e-01 6.31483018e-01 -7.86981806e-02
-1.87616423e-01 5.79632938e-01 2.55927950e-01 -4.53072637e-01
-3.70486379e-01 -7.38989234e-01 -6.67329073e-01 -7.68409550e-01
-3.82122546e-01 -9.30502266e-02 3.28237653e-01 -1.91292256e-01
-1.16069712e-01 7.10243165e-01 -1.33783722e+00 -2.79131919e-01
1.03081810e+00 1.19249892e+00 2.12282047e-01 4.54935342e-01
7.02092424e-02 1.02395272e+00 -1.90053895e-01 -1.35115162e-01
2.42426097e-01 -3.07690173e-01 -1.17975928e-01 -1.41667873e-02
-3.13877314e-02 -6.24481626e-02 -1.21195841e+00 8.22861314e-01
4.79913235e-01 8.21682155e-01 6.60170376e-01 -1.35088861e+00
-1.05919170e+00 1.71478361e-01 -6.28820360e-01 4.44333583e-01
-7.62749016e-02 4.69874024e-01 8.67015421e-01 -5.05617559e-01
3.79698634e-01 9.83083606e-01 7.52267778e-01 3.58612567e-01
-1.26164389e+00 -6.64007246e-01 6.41559541e-01 4.96328354e-01
-1.43679261e+00 -2.47960821e-01 1.34541798e+00 -4.66333985e-01
1.00185633e+00 2.19146177e-01 7.76356637e-01 1.16894770e+00
2.97222316e-01 5.58303833e-01 8.98348153e-01 6.83182180e-02
3.78761262e-01 -5.13149381e-01 2.03351215e-01 8.04657400e-01
-9.92782414e-03 2.82628924e-01 -3.31774771e-01 -4.01054658e-02
1.03528917e+00 3.47217321e-01 -2.47596025e-01 -2.88676381e-01
-9.83358681e-01 7.45256424e-01 8.07143807e-01 3.21048915e-01
-5.92341959e-01 5.73127568e-01 3.05864781e-01 1.26368672e-01
5.32122493e-01 5.09507619e-02 2.19881814e-02 -1.07128099e-01
-6.31490648e-01 -1.85265634e-02 4.21589077e-01 9.14011836e-01
6.90898240e-01 4.08772886e-01 -3.65288407e-01 5.65195680e-01
1.46204215e-02 5.76740026e-01 2.22941920e-01 -7.07791805e-01
4.08661872e-01 7.55473852e-01 -5.96297812e-03 -1.44965231e+00
-5.77211618e-01 -4.70947564e-01 -1.44476497e+00 -9.53870118e-02
3.72518927e-01 -1.39399245e-01 -9.89829421e-01 2.01283360e+00
1.98947147e-01 9.97962832e-01 -2.92303205e-01 1.06683469e+00
6.95771098e-01 9.23584461e-01 1.45006672e-01 -3.16927075e-01
9.24498379e-01 -5.85587442e-01 -7.30193377e-01 -2.96040345e-03
5.70641279e-01 1.47043383e-02 1.15059471e+00 1.96093600e-02
-8.81541967e-01 -4.91917819e-01 -7.90843308e-01 -2.07655460e-01
-5.13938546e-01 -2.77441717e-03 8.43427479e-01 1.69442207e-01
-1.09820628e+00 4.93271440e-01 -1.20212030e+00 -7.28851408e-02
4.12693322e-01 3.58572215e-01 -5.82195595e-02 -1.30664362e-02
-1.29552460e+00 3.61394942e-01 6.37594759e-02 3.49537462e-01
-6.98237598e-01 -7.19558001e-01 -8.77449334e-01 3.08693916e-01
3.97713482e-01 -7.21302748e-01 6.03967488e-01 -4.53081667e-01
-1.66061115e+00 2.42973611e-01 -1.40178040e-01 -5.31795740e-01
5.30064762e-01 2.52427533e-02 -5.69902718e-01 2.01648176e-02
-2.44335771e-01 2.94220597e-01 9.33690965e-01 -7.79990256e-01
-9.68162864e-02 -2.32223019e-01 3.43664765e-01 3.17992531e-02
-5.30877113e-01 -4.56815660e-01 -4.74738300e-01 -1.02640331e+00
-3.47726159e-02 -9.30915952e-01 -5.27222097e-01 6.87061399e-02
-9.93045792e-02 -4.72229451e-01 9.18051541e-01 -6.45323575e-01
1.53798974e+00 -2.14008784e+00 2.56996155e-01 2.22890511e-01
5.73904037e-01 6.34645224e-02 -1.58632040e-01 3.39878619e-01
3.65716293e-02 -6.88390881e-02 -2.76254565e-01 -2.01810002e-01
1.98806435e-01 4.71951783e-01 -4.87152070e-01 5.42369783e-01
1.29613429e-01 1.19525325e+00 -9.88172233e-01 -2.75765300e-01
3.10465515e-01 8.19637835e-01 -6.45083427e-01 2.10946783e-01
-2.52808869e-01 7.38106072e-01 -7.88000405e-01 2.37986967e-01
4.96297300e-01 -5.81941187e-01 -1.06564060e-01 -2.24348083e-01
-1.06594034e-01 1.16031855e-01 -1.05769408e+00 1.70459378e+00
-5.58445692e-01 5.87511718e-01 -4.59004641e-02 -1.31747520e+00
7.98480034e-01 1.48150250e-01 7.51999915e-01 -1.21742451e+00
1.10634327e-01 -7.62846842e-02 -2.81517059e-01 -5.87501705e-01
2.72749424e-01 -1.03273448e-02 -6.38653338e-02 4.02529538e-01
-1.28293768e-01 2.38089159e-01 1.73638985e-01 1.63349032e-01
1.16268480e+00 1.32898077e-01 -7.90749043e-02 -3.87639552e-01
4.40876365e-01 -3.08754116e-01 7.14485943e-01 5.30422688e-01
1.92771573e-02 2.76180178e-01 6.59964204e-01 -5.81501245e-01
-7.31439471e-01 -9.76867318e-01 3.07307214e-01 1.03512561e+00
2.91344941e-01 -3.03026408e-01 -3.95999998e-01 -5.42549968e-01
-4.66943793e-02 5.55254877e-01 -6.34582877e-01 -3.69846433e-01
-8.94234955e-01 -5.87457061e-01 3.14752519e-01 7.33374655e-01
4.15618479e-01 -8.22936356e-01 -5.44175148e-01 2.37657309e-01
-1.17304087e-01 -1.10382283e+00 -7.47204661e-01 -2.02980228e-02
-7.73279965e-01 -7.83459127e-01 -7.39002109e-01 -5.02437711e-01
6.05909705e-01 2.99971819e-01 8.51965010e-01 1.87125225e-02
-2.45295510e-01 4.41582054e-01 -6.97478503e-02 9.27576870e-02
2.22059742e-01 1.51044559e-02 4.25997749e-02 4.74474430e-01
-9.80856940e-02 -1.16582859e+00 -8.32955003e-01 2.94275820e-01
-9.59146440e-01 2.50709176e-01 4.10828590e-01 7.56695747e-01
6.86029971e-01 4.83452044e-02 6.51219130e-01 -5.08241713e-01
6.66801155e-01 -6.23950839e-01 -8.09302092e-01 2.55845964e-01
-4.31172103e-01 1.86160296e-01 1.00709999e+00 -7.43503153e-01
-8.92218828e-01 -9.75347757e-02 3.78776491e-02 -1.04573882e+00
7.93329775e-02 7.49517024e-01 5.36118895e-02 -1.47967830e-01
5.14454961e-01 4.46048468e-01 -1.50848836e-01 -4.67388421e-01
4.68306661e-01 2.42243297e-02 5.45258999e-01 -7.34337807e-01
7.58949101e-01 5.27625978e-01 4.33972448e-01 -8.95718455e-01
-4.92646426e-01 -2.10574627e-01 -4.77024168e-01 -2.42624000e-01
8.90669048e-01 -7.54451334e-01 -9.70415950e-01 3.54167134e-01
-9.36097026e-01 -7.51383603e-01 -2.78664410e-01 6.33439898e-01
-5.09805322e-01 3.93654495e-01 -6.25032604e-01 -7.53832519e-01
7.01694563e-02 -7.55657554e-01 8.75220060e-01 1.72044799e-01
2.45094419e-01 -1.33412850e+00 2.37177256e-02 -2.81719506e-01
6.10874176e-01 7.38592207e-01 1.02338994e+00 -6.90807849e-02
-7.85719335e-01 -1.43861696e-01 -3.33647937e-01 -2.50286441e-02
2.87600383e-02 -2.78429568e-01 -6.85266614e-01 -2.15136588e-01
1.59806058e-01 9.56439599e-03 9.52771306e-01 5.81102669e-01
1.64080679e+00 -5.06931901e-01 -4.47964966e-01 9.96079147e-01
1.15543139e+00 1.24379791e-01 2.83015937e-01 -3.49283755e-01
1.02303767e+00 4.85134214e-01 5.75740151e-02 4.66684371e-01
5.17096996e-01 4.88763005e-01 4.30514187e-01 -1.29329830e-01
-6.97615817e-02 -3.90033215e-01 2.56100327e-01 1.04996622e+00
-5.00424564e-01 -4.85063553e-01 -9.16633904e-01 5.14037371e-01
-2.20455122e+00 -9.99972284e-01 -1.39045298e-01 2.00134110e+00
2.04452872e-01 -7.12801702e-03 2.30283961e-01 -1.69070348e-01
7.03528821e-01 5.47073245e-01 -1.02281654e+00 -1.76568814e-02
-1.41767919e-01 5.66380471e-02 4.16766822e-01 2.59586632e-01
-8.91365588e-01 6.66526079e-01 5.87861204e+00 7.57361174e-01
-1.39118719e+00 1.67546332e-01 6.26053512e-01 -1.33003056e-01
-5.65566301e-01 -1.74603194e-01 -2.66494751e-01 5.17806649e-01
8.91445696e-01 -3.57373714e-01 8.71032298e-01 5.29560328e-01
3.96668196e-01 4.06449467e-01 -9.42289472e-01 1.26376843e+00
-2.60036707e-01 -1.54709733e+00 2.87041888e-02 2.38093004e-01
7.50651896e-01 6.17061295e-02 1.97179720e-01 4.39441413e-01
3.24603260e-01 -1.07745337e+00 5.63721061e-01 7.60033846e-01
8.58833015e-01 -6.39314950e-01 1.55185208e-01 5.19986808e-01
-1.64988375e+00 -1.46620676e-01 -4.09889013e-01 -2.62159556e-01
3.30590338e-01 5.13942480e-01 -1.52579442e-01 6.56226754e-01
4.43473458e-01 1.19970071e+00 -5.38092136e-01 8.81072938e-01
1.41391912e-02 7.79305935e-01 -6.46885633e-01 -8.95369276e-02
4.58941638e-01 -5.40433228e-01 5.63924134e-01 1.11345434e+00
4.58820790e-01 4.46435720e-01 3.11979771e-01 1.06108332e+00
-5.69829904e-02 -1.21938728e-01 -8.79923880e-01 -2.64994115e-01
2.07212090e-01 1.07330191e+00 -7.51236975e-01 -1.59705088e-01
-6.57433987e-01 6.59327567e-01 5.87558270e-01 1.02946651e+00
-1.19876873e+00 -1.88124120e-01 6.07078075e-01 1.99343666e-01
2.75648624e-01 -7.10271299e-01 -1.82986453e-01 -1.37310052e+00
1.95658714e-01 -2.55646735e-01 6.10337555e-01 -5.18651903e-01
-1.44860935e+00 5.83741665e-01 1.39806077e-01 -1.27287686e+00
-2.22416684e-01 -5.68487942e-01 -8.55451107e-01 7.13077784e-01
-1.47101867e+00 -1.26064336e+00 -4.23980206e-01 1.14919603e+00
4.18258198e-02 1.22050405e-01 3.38122100e-01 3.76095384e-01
-7.87566125e-01 5.16230702e-01 -4.27197888e-02 6.77744299e-02
8.11326131e-02 -1.02817535e+00 5.14007390e-01 7.72648394e-01
1.66221216e-01 5.50957680e-01 4.61882144e-01 -4.79397982e-01
-1.73144925e+00 -1.31622243e+00 4.29378986e-01 -2.45144069e-01
9.84540880e-01 -6.14466727e-01 -1.05177557e+00 6.70492232e-01
-5.20478636e-02 5.65371633e-01 2.93900937e-01 -5.61106801e-02
-3.70015085e-01 -2.70366699e-01 -7.82617927e-01 8.25191855e-01
1.65085101e+00 -8.32286596e-01 -1.63504452e-01 3.40501726e-01
7.06395149e-01 -2.88194567e-01 -7.78887630e-01 4.73137528e-01
3.33631158e-01 -6.74812376e-01 9.23364699e-01 -6.97517991e-01
-7.90248439e-02 -3.53209347e-01 -2.18042210e-02 -1.14568698e+00
-6.14060640e-01 -1.00002909e+00 -7.05510676e-01 1.03747809e+00
-5.45972085e-04 -9.88443434e-01 5.02740622e-01 6.56802356e-01
-3.43707442e-01 -9.16039050e-01 -1.19146168e+00 -8.57044220e-01
3.50835621e-02 -6.91816986e-01 8.22880745e-01 1.05299115e+00
-2.45679244e-01 2.08825499e-01 -4.17315513e-01 2.02030778e-01
4.71521407e-01 2.60899931e-01 5.19706368e-01 -1.27887809e+00
-2.85291195e-01 -7.11298823e-01 -5.79452991e-01 -1.46056795e+00
4.00313020e-01 -9.74470615e-01 -1.99202165e-01 -1.41232157e+00
-1.89670384e-01 -4.79894072e-01 -6.87076330e-01 3.16220373e-01
-7.55594373e-02 -7.78153464e-02 1.48555590e-03 1.47809699e-01
-6.24460161e-01 1.01666784e+00 1.33742678e+00 -1.23205297e-01
-3.22098494e-01 -1.75469637e-01 -3.59981596e-01 5.11074185e-01
6.27017677e-01 -2.68978149e-01 -9.19247508e-01 -5.54909110e-01
2.83615410e-01 3.08087766e-01 6.21552706e-01 -1.06648433e+00
5.14700532e-01 -3.68471473e-01 1.80029899e-01 -3.73459578e-01
3.54293317e-01 -7.19385087e-01 3.04543644e-01 1.64933518e-01
-1.72308102e-01 3.76349628e-01 2.03458637e-01 1.11594069e+00
-1.41635105e-01 6.11467421e-01 5.24311543e-01 1.02801092e-01
-8.72118413e-01 9.27427948e-01 -1.56844720e-01 9.01569128e-02
1.12371457e+00 -1.83558494e-01 -3.10162634e-01 -5.23252189e-01
-9.50167954e-01 3.90757918e-01 3.30447525e-01 4.98782337e-01
4.94794935e-01 -1.63352346e+00 -3.19049805e-01 3.53709459e-01
1.34916697e-03 -7.14218989e-02 5.62821507e-01 1.15236831e+00
-2.90032387e-01 4.57358688e-01 -6.36513308e-02 -6.65703714e-01
-4.34679240e-01 9.21912611e-01 4.04320717e-01 -4.16463971e-01
-1.03049219e+00 5.12559652e-01 6.51005685e-01 -1.18315123e-01
2.72510827e-01 -7.73711562e-01 -1.59799665e-01 -2.14710414e-01
2.08228633e-01 4.48759675e-01 -2.98860729e-01 -6.28627717e-01
-2.81869709e-01 6.24266326e-01 5.04638731e-01 7.49797598e-02
1.43419647e+00 -1.05626658e-01 1.90951705e-01 5.37902474e-01
1.15596032e+00 -2.55590916e-01 -1.74078035e+00 -2.97226518e-01
-1.48477986e-01 -2.20009595e-01 1.95514515e-01 -2.24545941e-01
-1.32413650e+00 1.12869036e+00 4.61765736e-01 5.13061702e-01
1.27630162e+00 -1.30398825e-01 1.04727948e+00 5.07829964e-01
3.90310377e-01 -8.45295489e-01 1.82630584e-01 5.55315435e-01
9.88214910e-01 -8.75579357e-01 -3.07168186e-01 -2.83818811e-01
-5.12011766e-01 9.19291735e-01 5.26269317e-01 -5.21196604e-01
1.10954678e+00 1.23876827e-02 -4.72227156e-01 -3.07656884e-01
-8.18114758e-01 -3.26169908e-01 5.94179153e-01 5.78441203e-01
1.58902064e-01 5.41441627e-02 -1.27673477e-01 7.07858026e-01
1.78657487e-01 1.16163202e-01 -4.71937992e-02 5.87453425e-01
5.82364984e-02 -5.34164488e-01 2.01313034e-01 4.20816988e-01
-2.30011269e-01 1.64563581e-01 -6.92931786e-02 6.84552550e-01
-1.02184154e-01 7.96980560e-01 2.07790956e-01 -4.30482030e-01
3.56057525e-01 -1.26865745e-01 2.23085538e-01 -2.64693171e-01
-2.59784997e-01 8.93048644e-02 -2.38996208e-01 -7.70081699e-01
-2.60817051e-01 -4.09986109e-01 -1.25405157e+00 -4.67085481e-01
-5.10668121e-02 5.18766567e-02 2.67561227e-01 8.17424893e-01
8.56897116e-01 7.63790488e-01 6.00598574e-01 -9.59427118e-01
-2.53004342e-01 -7.68024206e-01 -4.84275639e-01 4.54916507e-01
6.25546575e-01 -9.60556924e-01 -2.61158228e-01 -8.18534940e-02] | [6.755388259887695, 2.7834079265594482] |
0511fedf-050e-4fce-ada4-35cf2b787322 | deep-bayesian-video-frame-interpolation | null | null | https://www.ecva.net/papers.php | https://www.ecva.net/papers/eccv_2022/papers_ECCV/papers/136750141.pdf | Deep Bayesian Video Frame Interpolation | Abstract. We present deep Bayesian video frame interpolation, a novel approach for upsampling a low frame-rate video temporally to its higher frame-rate counterpart. Our approach learns posterior distributions of optical flows and frames to be interpolated, which is optimized via learned gradient descent for fast convergence. Each learned step is a lightweight network manipulating gradients of the log-likelihood of estimated frames and flows. Such gradients, parameterized either explicitly or implicitly, model the fidelity of current estimations when matching real image and flow distributions to explain the input observations. With this approach we show new records on 8 of 10 benchmarks, using an architecture with half the parameters of the state-of-the-art model. | ['Jimmy S. Ren', 'Xijun Chen', 'Dongqing Zou', 'Xujie Xiang', 'Yu Zhang', 'ZHIYANG YU'] | 2022-10-23 | null | null | null | conference-2022-10 | ['video-frame-interpolation'] | ['computer-vision'] | [-3.70182358e-02 -3.66990594e-03 -5.58121085e-01 -6.01034462e-01
-5.34631431e-01 -1.89276814e-01 6.57989740e-01 -5.44996202e-01
-5.06900668e-01 1.17153430e+00 5.28968990e-01 -2.06018344e-01
3.99848551e-01 -5.14440179e-01 -1.23071086e+00 -3.41043621e-01
-6.28767610e-01 2.85504818e-01 4.51192409e-01 2.06492007e-01
3.00277650e-01 4.76167232e-01 -1.50501978e+00 6.82931960e-01
2.24643365e-01 9.55324948e-01 1.44501096e-02 1.45832598e+00
-1.36844888e-01 1.62845027e+00 -4.62628543e-01 -4.37158406e-01
5.41689336e-01 -4.87511963e-01 -8.47661734e-01 7.76113346e-02
1.02714884e+00 -1.20148015e+00 -1.11035502e+00 7.53198802e-01
-1.42858088e-01 3.69933695e-01 3.89557660e-01 -1.21094263e+00
-5.31023920e-01 5.35070181e-01 -3.68813485e-01 7.56370723e-01
4.94129807e-01 5.49771190e-01 5.85634768e-01 -8.48498702e-01
9.50109899e-01 1.67778456e+00 5.75693369e-01 8.41328621e-01
-1.45264614e+00 -5.75016439e-01 2.97521651e-01 3.21233660e-01
-1.10490668e+00 -8.07887077e-01 3.29782516e-01 -5.86107254e-01
1.05177224e+00 -1.97336942e-01 9.07728076e-01 1.09674203e+00
2.31908619e-01 6.41138911e-01 4.02215630e-01 -1.43395722e-04
1.01801597e-01 -6.87384419e-03 -4.34923202e-01 8.16327214e-01
-2.07919441e-02 3.83858979e-01 -9.48673666e-01 -7.84898549e-02
1.40502286e+00 -1.54778838e-01 -5.00302196e-01 -1.27439842e-01
-1.28324854e+00 4.42109227e-01 4.64669019e-01 -3.64161760e-01
-3.65381241e-01 1.00606370e+00 3.74505818e-01 3.55119072e-02
6.66057587e-01 -7.91677460e-02 -5.34467697e-01 -4.09214288e-01
-1.20531225e+00 4.09088314e-01 8.57581556e-01 1.08578920e+00
1.10623217e+00 4.68294144e-01 -4.20777589e-01 6.77916259e-02
5.77301681e-01 3.35904390e-01 1.10201418e-01 -1.91765952e+00
3.62825185e-01 -1.84323221e-01 4.91575539e-01 -7.71004140e-01
2.50778049e-01 -8.67197961e-02 -5.01006067e-01 5.02192318e-01
7.75525749e-01 -2.78265566e-01 -8.42725217e-01 1.53882754e+00
1.20805323e-01 1.28353798e+00 -1.29124388e-01 9.57782924e-01
2.42715985e-01 9.68339145e-01 2.12654054e-01 -2.20291033e-01
7.97325015e-01 -1.07291472e+00 -5.87385297e-01 -7.54929855e-02
4.45806980e-02 -7.62156904e-01 6.34877324e-01 3.17156166e-01
-1.57195520e+00 -8.43188524e-01 -9.09512579e-01 -2.53201485e-01
2.62944400e-01 -1.32305413e-01 6.36574090e-01 4.61477131e-01
-1.52578866e+00 1.10171402e+00 -1.15040255e+00 2.61800289e-01
7.60428429e-01 2.37276226e-01 -2.02342600e-01 7.58281583e-03
-9.41090584e-01 6.04557335e-01 4.73400533e-01 8.92892852e-03
-1.43910742e+00 -1.33531618e+00 -8.55227172e-01 1.32930130e-01
-1.99813187e-01 -9.65848386e-01 1.31939650e+00 -1.22058880e+00
-1.84856057e+00 6.29549742e-01 -7.40221977e-01 -1.05422509e+00
9.16399539e-01 -5.35754323e-01 -7.19002634e-02 4.96251285e-01
-1.77329987e-01 1.32156336e+00 1.21513617e+00 -1.08385825e+00
-1.02147841e+00 3.54720443e-01 2.89015949e-01 -6.24176227e-02
1.43124819e-01 -1.57300428e-01 -4.71760035e-01 -5.01679301e-01
-4.61116582e-01 -6.47653043e-01 -1.89945251e-01 7.90909708e-01
1.31672025e-01 6.56459704e-02 9.18929815e-01 -7.39890158e-01
9.45838451e-01 -1.93068790e+00 -2.17896700e-02 -2.07250863e-01
1.47637755e-01 1.40793040e-01 -5.84246516e-02 -1.63142949e-01
-3.30436323e-03 7.10332841e-02 -3.53475772e-02 -6.96131825e-01
-2.82243669e-01 3.62143010e-01 -6.75220609e-01 6.10964715e-01
4.41334903e-01 7.13377655e-01 -1.27070701e+00 -4.65360940e-01
5.99178731e-01 1.02941716e+00 -9.31496561e-01 5.48628330e-01
-5.00435352e-01 7.88780451e-01 2.02121790e-02 -1.64670423e-02
7.37157047e-01 -2.46497408e-01 -1.14009092e-02 -4.18683082e-01
-9.18759108e-02 4.25909489e-01 -1.15432203e+00 1.94215560e+00
-4.12066579e-01 1.16543221e+00 -1.21351086e-01 -4.68851238e-01
5.21440744e-01 3.99028420e-01 4.64521527e-01 -1.56622142e-01
-1.81253389e-01 2.35735923e-02 -5.02937019e-01 -4.80159342e-01
6.33213878e-01 1.11505300e-01 8.59289110e-01 2.76383311e-01
3.00171077e-01 -5.87320812e-02 5.31770289e-01 3.26064438e-01
6.74739957e-01 9.74892259e-01 -1.62049755e-01 -1.96730271e-01
6.14208937e-01 -2.71344811e-01 5.97968459e-01 8.66332114e-01
-3.40249389e-01 7.76578069e-01 5.75751543e-01 -1.16753352e+00
-1.38957965e+00 -1.22964561e+00 2.29694173e-02 9.57111537e-01
-2.19849087e-02 -3.65477502e-01 -6.11930251e-01 -6.60809696e-01
-2.62754653e-02 5.76807499e-01 -5.47173917e-01 1.43547669e-01
-1.00924361e+00 -2.28803232e-01 5.45154572e-01 5.52594483e-01
4.91899371e-01 -8.67310822e-01 -7.50652909e-01 3.01409811e-01
-4.73136343e-02 -1.48524129e+00 -7.00802326e-01 -4.67125833e-01
-1.02095091e+00 -1.10914600e+00 -5.98703623e-01 -4.31298494e-01
5.04548371e-01 1.89452656e-02 1.65154922e+00 4.14090037e-01
-1.01079300e-01 1.59352511e-01 2.02301174e-01 1.38053373e-01
-6.32219315e-01 -3.21218848e-01 -1.31423086e-01 1.02980465e-01
1.16429873e-01 -5.64451993e-01 -9.95619714e-01 -1.51669374e-02
-9.08708215e-01 2.69569010e-01 -1.24342188e-01 6.49838805e-01
5.58839917e-01 -6.40162110e-01 -4.96343710e-02 -6.56264782e-01
9.16280970e-02 -4.57126230e-01 -9.33435082e-01 -3.68753262e-02
-1.03976078e-01 3.15968037e-01 6.58054769e-01 -3.47341329e-01
-1.32252228e+00 1.74685374e-01 -9.56875980e-02 -9.57120597e-01
-1.38312638e-01 -1.78312674e-01 6.29520714e-01 -8.42841193e-02
6.18698239e-01 7.75519619e-03 -1.33377507e-01 -6.64516538e-02
4.88529742e-01 1.43958375e-01 9.50342238e-01 -8.00462067e-01
4.77092505e-01 1.04514670e+00 1.95362642e-01 -3.83278519e-01
-9.50250804e-01 -4.17507887e-02 -5.30158281e-01 -4.60639209e-01
8.44857991e-01 -1.24523902e+00 -9.18563664e-01 3.58552873e-01
-1.65817833e+00 -6.65852785e-01 -6.35002553e-01 9.61485267e-01
-8.52944374e-01 1.27002060e-01 -9.61386144e-01 -7.19059289e-01
1.25631660e-01 -1.37022102e+00 1.01656306e+00 3.85381013e-01
-6.04599155e-02 -1.25752914e+00 1.25412345e-01 -1.45518690e-01
5.73236406e-01 1.68036669e-01 2.28608131e-01 1.00475460e-01
-1.24294972e+00 2.82349050e-01 -4.55488980e-01 4.61126059e-01
8.32350925e-02 6.80763006e-01 -1.12996066e+00 -3.12035382e-01
-1.99284762e-01 -1.69071332e-01 9.54340577e-01 7.99411058e-01
1.35055411e+00 -4.43196416e-01 -3.64121199e-02 1.25608742e+00
1.52297270e+00 -1.37307227e-01 8.70069802e-01 1.69853047e-01
7.28550673e-01 1.65255740e-01 2.60071397e-01 7.80255198e-01
4.15931076e-01 3.63448530e-01 5.45271575e-01 3.01671773e-01
-5.85770488e-01 -4.38444644e-01 5.74510992e-01 2.24265799e-01
-3.91934365e-01 -2.23261446e-01 -5.51041007e-01 5.26196659e-01
-1.80927110e+00 -1.29367268e+00 3.24083935e-03 2.09355259e+00
8.92455339e-01 3.56634498e-01 1.27642438e-01 -4.61742461e-01
7.68684268e-01 4.57715511e-01 -6.25397801e-01 -2.94493645e-01
3.67659666e-02 2.21965313e-01 7.20609307e-01 1.08549428e+00
-9.77022588e-01 1.01338494e+00 7.67915821e+00 4.29027915e-01
-1.01515162e+00 -6.10765740e-02 9.05391455e-01 -3.21479917e-01
-3.40994686e-01 3.36753845e-01 -9.44992423e-01 5.98723948e-01
1.36193252e+00 -1.37935862e-01 9.49248374e-01 4.19677466e-01
5.01212060e-01 -2.02672780e-01 -1.49461663e+00 8.82379353e-01
-1.76408008e-01 -2.21206379e+00 4.08656329e-01 -1.44512683e-01
9.84597981e-01 1.14933558e-01 1.05756917e-05 7.35417679e-02
5.29012859e-01 -8.57045412e-01 1.03676212e+00 8.78515899e-01
8.63415360e-01 -5.07068694e-01 3.12791914e-01 -1.54063582e-01
-1.11695111e+00 1.07430525e-01 -4.29768682e-01 -3.36713403e-01
6.92496300e-01 3.87569636e-01 -6.19017303e-01 -2.15721643e-03
9.05320346e-01 1.13473070e+00 -1.04024984e-01 1.01472974e+00
-3.11072856e-01 6.12375975e-01 -2.81824201e-01 5.43441534e-01
2.48392001e-01 -7.85073712e-02 5.51320672e-01 1.53066790e+00
2.12689474e-01 -1.99843496e-01 1.48023203e-01 1.01819491e+00
-3.41366231e-01 -5.97115815e-01 -3.18426073e-01 4.39771324e-01
4.34464335e-01 7.72880077e-01 -4.05988663e-01 -7.97213316e-01
-4.40271974e-01 7.20229745e-01 2.58147866e-01 6.60537422e-01
-9.45265591e-01 3.24980058e-02 1.22048080e+00 2.09652588e-01
3.73964190e-01 -3.88506144e-01 6.21319078e-02 -1.39870656e+00
-1.80938169e-01 -3.88719410e-01 2.03848094e-01 -1.01637471e+00
-9.90274489e-01 4.79182214e-01 1.62876025e-01 -1.07342649e+00
-7.80525804e-01 -5.17215014e-01 -4.86367106e-01 9.64634597e-01
-1.93128419e+00 -5.60819447e-01 -4.10198122e-01 6.00030541e-01
9.01828289e-01 -7.02562183e-02 4.30390537e-01 2.22058594e-01
-2.62504518e-01 2.03271657e-01 -1.40242368e-01 1.79295003e-01
8.06589007e-01 -1.14079034e+00 8.33257973e-01 1.04826534e+00
-1.34338380e-03 3.04811090e-01 9.24490571e-01 -4.86613244e-01
-1.09773111e+00 -1.09500885e+00 5.91984570e-01 -3.06891382e-01
7.12104321e-01 -5.79490466e-03 -8.52444887e-01 1.10366738e+00
4.34974343e-01 7.92683661e-01 -3.45497541e-02 -6.66768849e-01
-3.89729798e-01 -2.32831106e-01 -1.18732047e+00 5.20656168e-01
1.03285325e+00 -6.67207181e-01 -1.35749772e-01 2.86066145e-01
8.46659362e-01 -9.06883657e-01 -8.93266737e-01 -3.89420614e-02
7.94180572e-01 -1.48809385e+00 1.21091962e+00 -7.95680463e-01
6.90654695e-01 -4.36541885e-01 -8.62868577e-02 -1.10549986e+00
-1.29466623e-01 -1.11082399e+00 -8.45943987e-01 6.11778677e-01
5.02239726e-02 -3.00557375e-01 1.17648065e+00 7.80015349e-01
-3.87595706e-02 -4.43566263e-01 -8.35271657e-01 -4.53148872e-01
-8.83634165e-02 -3.54363024e-01 6.80308878e-01 5.33376515e-01
-4.55160767e-01 -3.87622863e-01 -6.80137217e-01 3.21632959e-02
8.85978103e-01 -4.75187093e-01 7.83072948e-01 -5.83191752e-01
-3.51486415e-01 -3.01591307e-01 -5.44131875e-01 -1.53476477e+00
5.78967869e-01 -4.03321534e-01 9.87595618e-02 -1.14965940e+00
-7.75510445e-02 2.07182392e-02 -9.24056470e-02 -8.25564563e-02
-1.49518862e-01 3.44585985e-01 4.03448731e-01 2.16455027e-01
-5.92812598e-01 3.25695992e-01 1.19432521e+00 -3.48379686e-02
-1.35128498e-01 -1.88565716e-01 -3.00254440e-03 8.71002853e-01
6.07003450e-01 -4.73103434e-01 -4.71711308e-01 -9.94304121e-01
6.52042329e-02 3.46541315e-01 6.51508152e-01 -1.15351665e+00
4.00976926e-01 -2.57856995e-01 6.76838994e-01 -4.07496929e-01
4.40587282e-01 -5.76243043e-01 2.61343569e-01 4.63514686e-01
-7.34745026e-01 3.00191611e-01 1.90825149e-01 6.87138855e-01
-2.13721693e-01 -8.19030553e-02 8.65381360e-01 -3.36637914e-01
-9.68505442e-01 6.89221144e-01 -3.56950194e-01 2.93759733e-01
6.66724741e-01 -2.13707730e-01 -2.55958796e-01 -7.10756838e-01
-7.17974901e-01 -2.52837446e-02 4.02164161e-01 4.14899617e-01
8.51712704e-01 -1.19549656e+00 -1.05803597e+00 3.27081501e-01
-5.59489906e-01 1.33735105e-01 -4.26902575e-03 4.06769812e-01
-1.19674838e+00 1.34251609e-01 -3.32944930e-01 -8.20942461e-01
-6.50095344e-01 3.64676476e-01 6.57746375e-01 -1.36122465e-01
-6.12834871e-01 9.65503871e-01 4.19702008e-02 3.41601253e-01
3.60531539e-01 -3.20009172e-01 1.69081494e-01 -5.38793981e-01
9.95383143e-01 6.10388577e-01 -3.58307600e-01 -6.50456667e-01
-1.44115090e-01 3.93532366e-01 7.76668042e-02 -2.80596524e-01
1.09297347e+00 -4.33917403e-01 -4.39258926e-02 3.28478605e-01
1.36731529e+00 -3.52350235e-01 -2.64797997e+00 -5.64001128e-02
-2.86574751e-01 -1.06869161e+00 1.47094071e-01 -1.72843054e-01
-1.35518301e+00 7.96758413e-01 4.56367791e-01 -2.68777966e-01
6.91181660e-01 -3.94090623e-01 8.26124191e-01 2.55025148e-01
3.29405755e-01 -6.88014328e-01 -9.93477330e-02 5.95803738e-01
4.91363645e-01 -1.13487673e+00 1.28537267e-01 -1.50344744e-01
-2.72040218e-01 1.57067311e+00 5.84483147e-01 -7.52855420e-01
7.63862133e-01 5.02337635e-01 -3.68531719e-02 4.20230389e-01
-1.16312504e+00 3.06650370e-01 7.38632008e-02 6.55533373e-01
6.10198975e-01 -3.76889914e-01 3.32441241e-01 -6.07930005e-01
1.76913023e-01 6.01389647e-01 8.07013392e-01 5.93468189e-01
-2.83065170e-01 -7.07494497e-01 -2.54083276e-01 2.21213087e-01
-6.52454436e-01 -1.02143966e-01 6.47152960e-01 6.04632735e-01
2.32462725e-03 6.50900960e-01 6.66676402e-01 1.39796332e-01
4.58886847e-03 -2.54866093e-01 8.29737067e-01 -1.41736001e-01
-4.53510880e-01 -1.68651015e-01 1.98614951e-02 -1.03846812e+00
-8.88250113e-01 -5.69253206e-01 -1.22932613e+00 -7.36406088e-01
3.10439050e-01 -7.32614100e-02 5.26636720e-01 8.36528957e-01
3.09477299e-01 4.68761683e-01 4.84074265e-01 -1.64232266e+00
-2.62822151e-01 -5.17008483e-01 -7.26939738e-02 5.86582839e-01
8.23610604e-01 -1.92815125e-01 -5.28853834e-01 8.56331170e-01] | [10.633267402648926, -1.300772786140442] |
a93a6920-f90f-499e-879b-6ba6addb8fb3 | you-can-mask-more-for-extremely-low-bitrate | 2306.15561 | null | https://arxiv.org/abs/2306.15561v1 | https://arxiv.org/pdf/2306.15561v1.pdf | You Can Mask More For Extremely Low-Bitrate Image Compression | Learned image compression (LIC) methods have experienced significant progress during recent years. However, these methods are primarily dedicated to optimizing the rate-distortion (R-D) performance at medium and high bitrates (> 0.1 bits per pixel (bpp)), while research on extremely low bitrates is limited. Besides, existing methods fail to explicitly explore the image structure and texture components crucial for image compression, treating them equally alongside uninformative components in networks. This can cause severe perceptual quality degradation, especially under low-bitrate scenarios. In this work, inspired by the success of pre-trained masked autoencoders (MAE) in many downstream tasks, we propose to rethink its mask sampling strategy from structure and texture perspectives for high redundancy reduction and discriminative feature representation, further unleashing the potential of LIC methods. Therefore, we present a dual-adaptive masking approach (DA-Mask) that samples visible patches based on the structure and texture distributions of original images. We combine DA-Mask and pre-trained MAE in masked image modeling (MIM) as an initial compressor that abstracts informative semantic context and texture representations. Such a pipeline can well cooperate with LIC networks to achieve further secondary compression while preserving promising reconstruction quality. Consequently, we propose a simple yet effective masked compression model (MCM), the first framework that unifies MIM and LIC end-to-end for extremely low-bitrate image compression. Extensive experiments have demonstrated that our approach outperforms recent state-of-the-art methods in R-D performance, visual quality, and downstream applications, at very low bitrates. Our code is available at https://github.com/lianqi1008/MCM.git. | ['Yao Zhao', 'Weisi Lin', 'Meng Wang', 'Chunjie Zhang', 'Runmin Cong', 'Huihui Bai', 'Jiaxin Han', 'Feng Li', 'Anqi Li'] | 2023-06-27 | null | null | null | null | ['image-compression'] | ['computer-vision'] | [ 6.13313973e-01 3.11014932e-02 -3.87761503e-01 -7.89542273e-02
-6.89755380e-01 -1.60607472e-01 5.86587667e-01 -1.73512354e-01
-1.87236890e-01 4.77815509e-01 4.06149417e-01 -1.78568721e-01
-2.99391858e-02 -8.45596254e-01 -7.72782207e-01 -8.78385842e-01
-3.79322991e-02 -8.44351724e-02 1.59895316e-01 -8.16273093e-02
1.88076600e-01 4.51960385e-01 -1.87080908e+00 7.91827977e-01
7.21042395e-01 1.11297345e+00 7.20764160e-01 7.42662072e-01
1.14439093e-01 6.79105639e-01 -3.69438648e-01 -5.39252341e-01
4.05306399e-01 -6.45684123e-01 -4.02490973e-01 2.32227236e-01
2.50309974e-01 -5.77922523e-01 -5.76759875e-01 1.02852476e+00
5.38579702e-01 -2.54922509e-01 4.40482289e-01 -6.37658119e-01
-6.06624961e-01 5.37699521e-01 -5.49065769e-01 -6.58706799e-02
-5.24226017e-02 4.58643496e-01 8.28838944e-01 -9.18024659e-01
4.88784879e-01 1.04278553e+00 4.32837874e-01 5.28982818e-01
-1.33966219e+00 -4.00801152e-01 -4.26913574e-02 4.64583069e-01
-1.37141323e+00 -6.92372739e-01 7.65285194e-01 -4.97280955e-02
7.58340359e-01 4.86136407e-01 6.14375651e-01 1.09550798e+00
2.35883728e-01 6.75962508e-01 1.06366897e+00 -4.56220925e-01
2.64660120e-01 -4.79259603e-02 -7.22437501e-01 4.02170628e-01
1.38862669e-01 3.02146524e-01 -7.52359211e-01 3.59499365e-01
8.13253582e-01 -7.73376897e-02 -4.78226811e-01 -1.60375834e-01
-1.08608210e+00 6.10740602e-01 4.81505424e-01 4.72499728e-01
-4.15225565e-01 2.45743051e-01 1.12300418e-01 3.11749220e-01
5.82615793e-01 1.41994938e-01 -2.81445563e-01 5.08841919e-03
-1.24985361e+00 1.98697835e-01 4.01025176e-01 7.20109820e-01
7.41591275e-01 1.10118613e-01 -5.00144303e-01 1.08329594e+00
2.16926530e-01 2.59705096e-01 4.76169318e-01 -1.08969676e+00
3.67855489e-01 2.53759146e-01 -1.02148257e-01 -1.02032793e+00
-7.86644814e-04 -8.10258389e-01 -1.20799839e+00 3.51848066e-01
-5.11711426e-02 4.09981400e-01 -9.72887933e-01 1.46330726e+00
7.61896297e-02 1.06218010e-01 -4.99910153e-02 1.13112843e+00
6.58115089e-01 8.00518453e-01 -8.29511806e-02 -2.86013305e-01
1.35083604e+00 -9.76053953e-01 -7.39046931e-01 -2.32871398e-01
2.03938186e-01 -8.98398399e-01 1.14670348e+00 7.03661799e-01
-1.58960330e+00 -8.47411096e-01 -1.23163784e+00 -2.84228116e-01
-1.28401751e-02 3.04138005e-01 3.78460407e-01 7.68916309e-01
-1.17779076e+00 9.28871453e-01 -7.61198342e-01 1.90611094e-01
7.76205540e-01 3.19583476e-01 3.65371481e-02 -4.90818292e-01
-9.31989193e-01 6.50263429e-01 1.96594521e-01 1.32006153e-01
-1.20014095e+00 -5.74772239e-01 -5.87341607e-01 2.11869836e-01
3.56545001e-01 -7.71192610e-01 9.18378413e-01 -9.55491364e-01
-1.66692603e+00 7.94237435e-01 -1.91522077e-01 -7.04339087e-01
5.08084238e-01 -1.79514304e-01 -2.71507472e-01 4.22587126e-01
-3.24521780e-01 9.66087997e-01 1.30895209e+00 -1.45087647e+00
-3.97553474e-01 -1.89417109e-01 -2.92486191e-01 1.01544462e-01
-4.96741027e-01 -1.62935138e-01 -6.93590641e-01 -1.07229698e+00
1.04019910e-01 -4.97950166e-01 -1.46949157e-01 1.93168208e-01
-1.45554885e-01 3.24112445e-01 7.37016976e-01 -8.29883873e-01
1.32916260e+00 -2.35457039e+00 3.62854719e-01 -1.62866652e-01
1.97959945e-01 6.13439560e-01 -2.75109529e-01 3.04420650e-01
1.00694904e-02 9.09774005e-02 -6.06370151e-01 -7.70614386e-01
-1.50220409e-01 1.68617710e-01 -4.58520740e-01 4.02655631e-01
3.45568299e-01 1.07948852e+00 -4.98232424e-01 -2.48996794e-01
4.88922328e-01 8.44464719e-01 -7.62447834e-01 3.99971426e-01
-3.83707315e-01 5.24184048e-01 4.60392050e-03 7.62376130e-01
9.85376298e-01 -1.55235186e-01 1.74924761e-01 -5.08485138e-01
-9.92790833e-02 2.24179015e-01 -8.01092029e-01 1.83809447e+00
-6.67589426e-01 5.99071562e-01 2.34315604e-01 -1.07795131e+00
8.91388297e-01 2.01977462e-01 4.66331452e-01 -1.04943419e+00
6.05539456e-02 4.29725856e-01 -1.67347938e-01 -4.38758165e-01
4.33326304e-01 -2.61738986e-01 4.54745412e-01 1.37304207e-02
3.29268649e-02 -1.43770903e-01 -5.34898601e-02 -3.89645584e-02
8.83722067e-01 2.89633989e-01 1.22584336e-01 1.10578880e-01
5.76313436e-01 -4.53510910e-01 2.28054002e-01 4.75190669e-01
1.15222000e-02 1.17336607e+00 3.19067836e-01 -2.24003151e-01
-1.42925227e+00 -9.99426246e-01 -3.72710317e-01 8.46096814e-01
2.09486976e-01 -6.63027108e-01 -8.12575877e-01 -1.59162492e-01
-3.97003651e-01 5.55363894e-01 -3.12873572e-01 -3.01341385e-01
-6.31159961e-01 -7.61780620e-01 4.42196250e-01 4.77657318e-02
8.34565341e-01 -9.87736225e-01 -7.37138629e-01 2.40701810e-01
-3.34339112e-01 -1.18563211e+00 -1.82471693e-01 1.17652319e-01
-9.02202308e-01 -5.23851812e-01 -9.46024895e-01 -4.52403903e-01
3.83720487e-01 3.11423242e-01 1.03579938e+00 2.50975817e-01
-4.03812289e-01 -3.04604918e-02 -5.59119284e-01 -7.80854793e-03
-5.51180601e-01 -1.77366078e-01 -4.45724308e-01 3.34101021e-01
-1.36840284e-01 -1.00522137e+00 -1.18579447e+00 4.71044570e-01
-1.32419002e+00 5.47248125e-01 9.86159861e-01 7.27431476e-01
7.64482737e-01 1.53686106e-01 2.78289080e-01 -4.64903325e-01
2.46096164e-01 -3.77084732e-01 -3.71945649e-01 -1.84071995e-03
-7.13399112e-01 -6.27935529e-02 7.80391753e-01 -3.48459035e-01
-9.15261984e-01 -7.89455250e-02 -4.92787421e-01 -5.82204282e-01
-1.84628785e-01 2.97988266e-01 -3.52735043e-01 -4.13378179e-02
6.18759751e-01 6.82288468e-01 6.72777593e-02 -7.20919967e-01
3.34683329e-01 7.25467980e-01 6.80752695e-01 -2.61941075e-01
6.56229019e-01 7.23468006e-01 1.29468545e-01 -6.73363686e-01
-5.07005513e-01 -1.27341092e-01 -2.82451212e-01 -1.06487654e-01
7.21080244e-01 -1.19324589e+00 -4.80372608e-01 4.03425157e-01
-8.82348239e-01 -4.66762960e-01 -4.08144891e-01 2.94232100e-01
-7.66901255e-01 5.77511311e-01 -5.18094301e-01 -6.61836267e-01
-3.04350227e-01 -1.32394171e+00 1.11986601e+00 -1.87648237e-01
1.95471093e-01 -4.18249726e-01 -3.52363288e-01 5.28638601e-01
9.26989079e-01 3.73982303e-02 8.48026872e-01 1.19377464e-01
-1.00806248e+00 1.08893357e-01 -4.39020157e-01 6.60629511e-01
-1.23461396e-01 -4.27607119e-01 -1.29255986e+00 -3.64790112e-01
3.93669248e-01 -2.47011155e-01 1.35705686e+00 5.90980828e-01
1.75618386e+00 -5.85439086e-01 7.66762719e-02 1.09628522e+00
1.52972305e+00 -1.02545932e-01 1.22509098e+00 2.03233138e-01
5.51055193e-01 6.10866189e-01 3.15536320e-01 5.44193447e-01
3.95887233e-02 9.91524816e-01 7.11013138e-01 -1.86278149e-01
-9.13852453e-01 -3.38428885e-01 5.13178706e-01 6.37984693e-01
-1.89020932e-01 -5.07781029e-01 -4.09453034e-01 4.19779867e-01
-1.56708920e+00 -8.14107120e-01 1.26569927e-01 2.20342994e+00
1.01404476e+00 1.04951166e-01 -5.39237186e-02 3.27552468e-01
4.97453362e-01 5.11842608e-01 -4.40098137e-01 -2.34215081e-01
-4.83120143e-01 4.43706185e-01 5.12599468e-01 4.22863841e-01
-8.58886123e-01 6.15739644e-01 5.05454922e+00 1.42527294e+00
-1.18424356e+00 3.72832984e-01 1.16843843e+00 -2.93206632e-01
-5.34954548e-01 4.98357415e-02 -5.61085641e-01 6.97154045e-01
1.21876740e+00 2.97078967e-01 8.90963435e-01 6.20557010e-01
2.32373074e-01 -6.64804727e-02 -8.85190666e-01 1.16588175e+00
7.48257339e-02 -1.66329122e+00 2.50608981e-01 2.27321818e-01
6.67024314e-01 -7.65729928e-03 4.92965788e-01 -1.78595465e-02
-3.52170706e-01 -1.36510360e+00 8.87511015e-01 3.22693884e-01
1.26406133e+00 -5.91762602e-01 4.91090238e-01 3.19020480e-01
-9.64616597e-01 -2.26414174e-01 -6.84057713e-01 2.13125259e-01
1.90202206e-01 9.27324593e-01 -3.99224907e-01 6.51243329e-01
7.19167531e-01 8.21346879e-01 -4.59107339e-01 9.09805536e-01
-1.31924912e-01 5.55555999e-01 -1.46219164e-01 4.81256634e-01
-2.72975815e-03 3.65442447e-02 5.34634113e-01 1.22644508e+00
4.57660615e-01 1.09356463e-01 -3.22910190e-01 1.11331975e+00
-1.46585092e-01 -5.50378673e-02 -2.89669275e-01 1.63195074e-01
1.00615688e-01 9.95003819e-01 -5.95481336e-01 -2.61647493e-01
-2.81053394e-01 1.19094121e+00 -8.08336884e-02 2.29668170e-01
-7.27788806e-01 7.16904998e-02 5.25731206e-01 4.59545672e-01
7.80748904e-01 -1.95012391e-01 -5.12965918e-01 -1.16489661e+00
2.30638474e-01 -1.11791527e+00 -4.58553433e-03 -6.91324770e-01
-9.18635428e-01 5.24512768e-01 -8.85484889e-02 -1.63175714e+00
6.45157024e-02 -4.71913427e-01 -2.49195397e-01 7.95057237e-01
-1.98814678e+00 -1.07693458e+00 -3.29809606e-01 5.14241934e-01
8.96195710e-01 -3.02920286e-02 6.08440101e-01 5.60410082e-01
-3.81365389e-01 5.68413734e-01 9.08248201e-02 -4.38666135e-01
4.54731047e-01 -6.58450902e-01 3.46580356e-01 9.68042195e-01
1.32796645e-01 3.18710506e-01 5.93861938e-01 -4.80943441e-01
-1.58081937e+00 -1.18854308e+00 5.55743754e-01 1.14739597e-01
1.44498259e-01 -4.26168263e-01 -9.31706429e-01 2.56618578e-02
2.51274586e-01 1.41278759e-01 3.48241419e-01 -5.62310755e-01
-3.40482563e-01 -2.90621161e-01 -1.13908827e+00 6.62512004e-01
1.13156557e+00 -3.80015612e-01 1.42736271e-01 2.79876322e-01
8.61739874e-01 -2.15739712e-01 -6.74254417e-01 7.15538263e-01
3.87043118e-01 -1.57439363e+00 1.26697075e+00 2.10989460e-01
1.05112672e+00 -3.56270730e-01 -5.50445139e-01 -9.21452880e-01
-2.53306478e-01 -7.75892556e-01 -5.17433107e-01 9.39187169e-01
1.13603666e-01 -2.81528443e-01 6.99888289e-01 7.08021149e-02
-3.18315625e-01 -1.10162830e+00 -1.05977881e+00 -6.45919740e-01
-2.07997963e-01 -6.27323925e-01 5.77512562e-01 4.51672435e-01
-3.88616890e-01 -1.03152394e-01 -8.16093802e-01 -1.74045432e-02
6.42595291e-01 3.95449139e-02 5.77406168e-01 -5.51666856e-01
-7.46723711e-01 -6.77889526e-01 -3.28759998e-01 -1.43376625e+00
-2.98972696e-01 -8.33994627e-01 -3.17333899e-02 -1.25947881e+00
1.19050510e-01 -3.78802896e-01 -1.99966326e-01 2.67509431e-01
-1.67827494e-03 6.97960556e-01 3.03327471e-01 5.67280412e-01
-2.93088078e-01 8.73908758e-01 1.41307914e+00 -3.49119976e-02
5.06072119e-02 -1.59806967e-01 -6.24037325e-01 3.66436332e-01
6.91835940e-01 -3.61637592e-01 -3.65770787e-01 -6.03215158e-01
5.63091272e-03 5.54620326e-02 6.59253061e-01 -1.15499747e+00
7.30575696e-02 -3.63155990e-03 5.22494376e-01 -3.99162561e-01
5.35669744e-01 -7.53879368e-01 3.76843840e-01 5.16292751e-01
-3.07302922e-01 -4.23905581e-01 1.32280225e-02 5.22355199e-01
-4.09442335e-01 -1.25482649e-01 9.83757317e-01 -1.82231858e-01
-4.08481061e-01 2.34686315e-01 -5.36718331e-02 -4.53902185e-01
7.55529821e-01 -5.03621340e-01 -3.17774057e-01 -3.72203767e-01
-5.31431675e-01 -4.28862929e-01 5.98200262e-01 2.32016787e-01
9.76321161e-01 -1.06833375e+00 -9.49878514e-01 5.56126833e-01
-1.41611323e-01 4.54020686e-02 6.44675791e-01 8.44111025e-01
-6.53423011e-01 2.78222948e-01 -2.85716742e-01 -6.00804567e-01
-9.94483650e-01 6.58548772e-01 1.45034954e-01 -2.28594482e-01
-8.45137298e-01 7.62775362e-01 3.07688028e-01 2.68352240e-01
3.37666154e-01 -1.02338001e-01 2.78906655e-02 -3.44460517e-01
6.86724842e-01 1.83134660e-01 2.17485756e-01 -4.98465896e-01
1.02807179e-01 6.40951395e-01 5.95275946e-02 -2.25629639e-02
1.58988667e+00 -2.96119332e-01 -1.16078384e-01 -1.90957058e-02
1.29398966e+00 4.24697362e-02 -1.55870509e+00 -1.28895000e-01
-4.71795827e-01 -1.00275850e+00 3.22151572e-01 -6.88289165e-01
-1.32494473e+00 1.02013624e+00 8.98559570e-01 4.67131138e-02
1.73500514e+00 5.40058427e-02 9.95008826e-01 -2.20801860e-01
2.67555833e-01 -6.86208546e-01 2.97264814e-01 -1.46893233e-01
1.16341245e+00 -9.35149491e-01 2.20759735e-01 -4.86471087e-01
-4.89705086e-01 1.04550648e+00 1.79270938e-01 -1.66357160e-02
4.38778758e-01 2.48472795e-01 -2.45934382e-01 1.57526147e-03
-8.28655064e-01 -7.31257722e-02 3.43009382e-01 5.83668828e-01
3.45818847e-01 1.31703597e-02 -3.72685701e-01 3.52811515e-01
-1.55889690e-01 -2.90466342e-02 3.05948317e-01 6.06767774e-01
-4.77744281e-01 -1.26139283e+00 -3.88652533e-01 1.97857067e-01
-5.81637442e-01 -3.11227173e-01 1.01465851e-01 2.22903594e-01
4.89751071e-01 8.68326724e-01 -7.43694752e-02 -6.18224740e-01
1.26073584e-02 -3.33203077e-01 4.27365959e-01 -3.60482663e-01
-4.44451004e-01 4.31110054e-01 -1.62549257e-01 -7.86594689e-01
-4.82875794e-01 -2.51957804e-01 -6.38245165e-01 -3.81332070e-01
-4.98117432e-02 -3.90167654e-01 8.04839909e-01 5.74869394e-01
6.10370219e-01 5.18761992e-01 7.73948133e-01 -1.14808750e+00
-3.69817823e-01 -7.87124693e-01 -3.09466571e-01 3.66342545e-01
5.12430727e-01 -3.32296193e-01 -3.77306968e-01 2.52908856e-01] | [11.31421184539795, -1.6715164184570312] |
7de23d99-a3fb-48e1-8991-9d5a9cbdc196 | visually-aware-audio-captioning-with-adaptive | 2210.16428 | null | https://arxiv.org/abs/2210.16428v3 | https://arxiv.org/pdf/2210.16428v3.pdf | Visually-Aware Audio Captioning With Adaptive Audio-Visual Attention | Audio captioning aims to generate text descriptions of audio clips. In the real world, many objects produce similar sounds. How to accurately recognize ambiguous sounds is a major challenge for audio captioning. In this work, inspired by inherent human multimodal perception, we propose visually-aware audio captioning, which makes use of visual information to help the description of ambiguous sounding objects. Specifically, we introduce an off-the-shelf visual encoder to extract video features and incorporate the visual features into an audio captioning system. Furthermore, to better exploit complementary audio-visual contexts, we propose an audio-visual attention mechanism that adaptively integrates audio and visual context and removes the redundant information in the latent space. Experimental results on AudioCaps, the largest audio captioning dataset, show that our proposed method achieves state-of-the-art results on machine translation metrics. | ['Wenwu Wang', 'Volkan Kılıç', 'Mark D. Plumbley', 'Lilian H. Tang', 'Yu Zhang', 'Tom Ko', 'Shengchen Li', 'Jianyuan Sun', 'Qiuqiang Kong', 'Haohe Liu', 'Xinhao Mei', 'Qiushi Huang', 'Xubo Liu'] | 2022-10-28 | null | null | null | null | ['audio-captioning'] | ['audio'] | [ 5.14021814e-01 -2.14870319e-01 2.37792265e-02 -6.55644909e-02
-1.19754505e+00 -6.90628648e-01 3.82807702e-01 1.90592781e-02
-4.88983542e-02 5.32672644e-01 7.39418566e-01 1.19764367e-02
4.55249429e-01 -2.25968778e-01 -9.65309083e-01 -4.11818951e-01
1.88536435e-01 1.47136718e-01 -1.98996463e-03 -4.67317142e-02
1.49067104e-01 -5.06060570e-02 -1.86495066e+00 9.88069236e-01
4.51563030e-01 1.18002415e+00 5.44734299e-01 1.02002728e+00
-2.71421373e-01 7.05882668e-01 -5.12951732e-01 -2.39713222e-01
-1.15388744e-01 -6.98252439e-01 -5.31038463e-01 1.54837295e-01
7.22022772e-01 -3.24511200e-01 -3.71676028e-01 9.59144175e-01
6.48785949e-01 3.71718891e-02 5.18635988e-01 -1.57180703e+00
-1.09429109e+00 7.92510986e-01 -2.83495218e-01 3.10181975e-01
8.26656282e-01 3.16151708e-01 1.32107651e+00 -1.26430058e+00
3.71202290e-01 1.26534522e+00 1.66649491e-01 6.39069140e-01
-1.10531390e+00 -8.01519513e-01 2.44894356e-01 6.85135841e-01
-1.54907393e+00 -7.31735051e-01 1.15463531e+00 -5.43707848e-01
5.19678593e-01 5.15653610e-01 7.53732562e-01 1.44061112e+00
-2.73700207e-01 9.61083770e-01 5.72119951e-01 -5.14679313e-01
1.97186291e-01 6.36694729e-02 -4.41872507e-01 2.06801683e-01
-1.93392843e-01 -1.15384221e-01 -9.90649939e-01 -1.18729666e-01
5.72447658e-01 -1.98798001e-01 -6.21827483e-01 -3.27255696e-01
-1.45869052e+00 5.90316832e-01 1.55793160e-01 2.01104760e-01
-3.38790208e-01 5.36783218e-01 3.93154442e-01 -6.66532218e-02
5.49893379e-02 3.00153106e-01 1.09975941e-01 -5.09341717e-01
-7.71421790e-01 -3.40852118e-03 1.97817981e-01 1.01482522e+00
3.16167384e-01 1.26536235e-01 -6.58302367e-01 8.44411969e-01
4.65360582e-01 7.15691626e-01 6.13768160e-01 -9.46318448e-01
8.74264598e-01 1.92187428e-01 2.51364976e-01 -9.39420640e-01
3.94876376e-02 -2.73405015e-01 -4.56156522e-01 -3.16284895e-01
-9.84477773e-02 7.68555254e-02 -8.39103758e-01 1.70326388e+00
7.74084777e-02 5.43966711e-01 6.33541169e-03 1.39762163e+00
1.15151834e+00 9.71738636e-01 9.48587880e-02 -2.37494648e-01
1.50482023e+00 -1.07724082e+00 -9.72107768e-01 -3.19438070e-01
-7.00640231e-02 -9.82528687e-01 1.49932241e+00 1.53539941e-01
-1.12714088e+00 -8.52645159e-01 -1.03612065e+00 -5.11605218e-02
6.25819117e-02 1.62276596e-01 2.93787599e-01 2.36057281e-01
-7.64393508e-01 -2.71972585e-02 -6.41604245e-01 -5.53712733e-02
2.77592421e-01 6.86345203e-03 -2.06299618e-01 1.03272773e-01
-1.28809536e+00 1.92670584e-01 4.05129761e-01 8.82433578e-02
-1.31291938e+00 -4.47751820e-01 -1.01038575e+00 2.79900283e-01
3.49680960e-01 -6.72528923e-01 1.53909993e+00 -1.34044278e+00
-1.52445447e+00 3.52554828e-01 -3.47500265e-01 -1.53365493e-01
2.03219250e-01 -2.49668986e-01 -5.16871333e-01 7.57959962e-01
-1.49253821e-02 1.13136089e+00 1.13218760e+00 -1.53108811e+00
-3.61509591e-01 8.16119313e-02 -2.45724376e-02 5.17305374e-01
-6.48538470e-01 1.51695311e-01 -6.32183075e-01 -8.68656814e-01
-8.50062743e-02 -8.46767485e-01 2.59511828e-01 1.16372630e-01
-3.77445608e-01 9.19004828e-02 7.13174522e-01 -7.23988831e-01
1.18772972e+00 -2.45708513e+00 2.50892669e-01 -1.76593587e-01
-3.60006955e-03 3.85477245e-02 -5.01801491e-01 4.14043695e-01
3.24567519e-02 -2.32848264e-02 -1.04345180e-01 -4.22026306e-01
1.89863995e-01 -7.38887638e-02 -6.64463401e-01 1.10030267e-02
4.18364435e-01 9.16273892e-01 -1.08918703e+00 -8.56750429e-01
4.58559878e-02 8.52365196e-01 -5.74861884e-01 3.98750991e-01
-3.49525750e-01 4.37300354e-01 -3.06810796e-01 7.38552332e-01
3.91816050e-01 -2.34928682e-01 -1.37425855e-01 -9.25724134e-02
4.85767983e-02 2.60942400e-01 -1.05415881e+00 2.01512194e+00
-4.01409179e-01 1.05384791e+00 1.37631064e-02 -5.73325157e-01
8.06976676e-01 7.40130067e-01 1.87543347e-01 -6.48254454e-01
6.34469241e-02 4.12048101e-02 -2.65158415e-01 -7.33744919e-01
5.54812074e-01 1.09521016e-01 -2.56398749e-02 1.58691540e-01
6.21296056e-02 -3.86976590e-03 -2.02665254e-02 1.36918813e-01
7.99944818e-01 1.08860619e-01 9.95299965e-02 5.54903746e-01
5.67127526e-01 -2.23913938e-01 3.07818174e-01 4.90226239e-01
-3.23921770e-01 1.24180675e+00 2.43417978e-01 -1.00092568e-01
-9.16559815e-01 -1.28831136e+00 2.47608095e-01 1.51413989e+00
2.54304141e-01 -5.67902923e-01 -7.26080418e-01 -3.93941730e-01
-3.59196901e-01 5.33188641e-01 -5.19144356e-01 -1.00875728e-01
-3.35846454e-01 -1.30728900e-01 4.71510679e-01 6.50708318e-01
1.18194513e-01 -1.23088074e+00 -5.15111208e-01 2.77845293e-01
-8.90578687e-01 -1.38237810e+00 -9.46220040e-01 -4.38434809e-01
-2.66184270e-01 -5.88046134e-01 -9.86191034e-01 -1.03042257e+00
3.37089002e-01 5.00935435e-01 9.64594066e-01 -2.23460063e-01
-2.66928971e-01 6.58033550e-01 -6.21718228e-01 -4.16915983e-01
-4.61038858e-01 -1.74584806e-01 -4.39688610e-03 4.44006771e-01
1.57204852e-01 -5.94294488e-01 -6.40638590e-01 2.44775832e-01
-8.34644854e-01 4.32479471e-01 6.45241737e-01 6.77671552e-01
8.43142152e-01 -6.47890270e-01 6.62372053e-01 1.16736338e-01
5.95362902e-01 -4.06195313e-01 -2.08445638e-01 2.12552488e-01
1.35586455e-01 -1.77974522e-01 7.35094070e-01 -9.75935698e-01
-8.09331715e-01 4.15464163e-01 2.40838021e-01 -9.61628973e-01
-2.27623314e-01 3.70333046e-01 -3.22278410e-01 2.11951271e-01
4.24206883e-01 4.76361513e-01 -3.93644780e-01 -5.87927341e-01
5.61220229e-01 1.18187857e+00 9.65514362e-01 -3.37501317e-01
6.41377032e-01 4.06486094e-01 -2.40552068e-01 -8.09222460e-01
-6.06542766e-01 -4.77408260e-01 -2.71247178e-01 -5.16573846e-01
1.03386092e+00 -1.24607575e+00 -6.63966775e-01 -9.52040404e-02
-1.45164371e+00 7.32726380e-02 -7.26901740e-02 6.28019035e-01
-8.85167062e-01 3.55823308e-01 -3.25276375e-01 -9.36042547e-01
-3.22270989e-01 -1.27396941e+00 1.44705832e+00 1.94090098e-01
-2.44508848e-01 -1.63954943e-01 2.83956885e-01 5.57500601e-01
1.58667475e-01 -4.48515303e-02 5.35264671e-01 -5.87878942e-01
-7.48089552e-01 1.69941768e-01 -2.91871428e-01 6.37053102e-02
-7.89384916e-02 -1.50229916e-01 -1.38257897e+00 -1.11797430e-01
-4.27091956e-01 -3.86965215e-01 7.76455402e-01 1.67172343e-01
1.32015908e+00 -4.34597820e-01 -8.59616399e-02 4.31246519e-01
1.02253532e+00 3.16168010e-01 5.59123337e-01 7.19733536e-02
8.61822248e-01 5.28193891e-01 6.43004775e-01 6.04237914e-01
3.35378140e-01 1.10784650e+00 6.50683224e-01 7.41467923e-02
-4.34557199e-01 -7.39700854e-01 6.99798167e-01 1.06427300e+00
2.67049700e-01 -2.95576096e-01 -7.25076199e-01 7.14476466e-01
-1.80317330e+00 -9.99298632e-01 2.88566470e-01 1.83219910e+00
9.14136887e-01 -7.10779354e-02 6.68582246e-02 2.43262768e-01
1.03977728e+00 1.27627421e-02 -2.73984283e-01 -3.54814261e-01
-5.32133803e-02 -7.41684064e-02 -1.58136621e-01 3.45546991e-01
-1.15245831e+00 7.14886963e-01 5.96589231e+00 7.36968815e-01
-1.16792679e+00 8.36230516e-02 1.54796615e-01 -4.73318130e-01
-4.91349071e-01 -2.56398767e-01 -3.75225365e-01 6.52916372e-01
1.00024450e+00 -9.95552763e-02 6.01520300e-01 6.45708859e-01
4.54124421e-01 3.89563143e-01 -1.30194473e+00 1.37173772e+00
5.17801881e-01 -1.20468092e+00 4.40881193e-01 -2.76885122e-01
4.54121143e-01 -3.15825850e-01 3.78281981e-01 1.06833182e-01
-3.75316113e-01 -9.91748273e-01 1.24110079e+00 5.14299273e-01
9.05756891e-01 -7.21582830e-01 4.44115371e-01 -1.81283519e-01
-1.46263075e+00 -7.99224898e-02 -1.55666783e-01 -7.67924488e-02
3.52554828e-01 2.64274399e-03 -1.09247804e+00 2.31130198e-01
8.27562332e-01 4.98756289e-01 -6.54900610e-01 1.44048369e+00
-1.91798523e-01 7.44459927e-01 1.29145046e-03 -8.29194859e-02
-3.77090531e-03 4.08611596e-01 8.54664505e-01 1.40292633e+00
5.77087402e-01 -1.39981747e-01 8.85137618e-02 7.85482883e-01
-2.22018033e-01 3.59044075e-01 -5.04828870e-01 -4.96809453e-01
6.99000597e-01 9.62614655e-01 -4.18192446e-01 -1.69162542e-01
-5.01636028e-01 1.07292700e+00 -1.05201058e-01 3.47575277e-01
-1.08531392e+00 -5.82101226e-01 6.51238501e-01 -1.26425430e-01
6.08105898e-01 -9.34737325e-02 9.46903527e-02 -1.09161091e+00
3.57369393e-01 -8.49333882e-01 1.74309209e-01 -1.50649405e+00
-9.87697124e-01 7.15615511e-01 -1.25393495e-01 -1.77478373e+00
-3.53612751e-01 -3.94596308e-01 -3.64904046e-01 4.53825951e-01
-1.36100805e+00 -1.27630687e+00 -4.32545096e-01 5.03511608e-01
8.83459628e-01 -9.12125409e-02 7.37439513e-01 3.52607161e-01
-1.54108137e-01 7.23264515e-01 -5.52280284e-02 2.45954379e-01
9.74944353e-01 -8.28756869e-01 3.23146701e-01 6.99270725e-01
7.52118587e-01 2.09458366e-01 9.53531086e-01 -4.00400996e-01
-1.40109968e+00 -1.12139344e+00 7.97568500e-01 -1.88628033e-01
7.30831206e-01 -6.58782423e-01 -9.47623074e-01 5.14682889e-01
4.02009696e-01 3.62685248e-02 1.01510394e+00 -5.10049105e-01
-5.46068788e-01 -5.62015548e-03 -5.56763470e-01 6.88684881e-01
1.02280259e+00 -9.65150058e-01 -9.13714290e-01 9.75335985e-02
1.24640512e+00 -3.03386778e-01 -3.10232133e-01 5.43242022e-02
7.23673105e-01 -4.36334550e-01 1.13832939e+00 -5.84778666e-01
6.16243362e-01 -6.03363872e-01 -4.15328711e-01 -1.23258030e+00
-1.41020104e-01 -8.77577901e-01 -2.13564664e-01 1.48273373e+00
3.50269616e-01 1.54217452e-01 4.31978881e-01 -7.58925900e-02
-3.47679108e-01 -1.41840413e-01 -1.05766678e+00 -7.59006977e-01
-4.98163670e-01 -5.84960997e-01 6.25286043e-01 7.37138569e-01
3.29755664e-01 4.42099184e-01 -7.43212759e-01 2.59171396e-01
3.98376971e-01 2.68886387e-01 6.08628333e-01 -1.07687759e+00
-5.18039763e-01 -2.15296000e-01 -6.21732533e-01 -1.00728500e+00
1.94367275e-01 -7.73773730e-01 4.44886655e-01 -1.44433951e+00
3.99635404e-01 3.71970952e-01 -4.45401579e-01 4.58243728e-01
-1.39700830e-01 7.94555664e-01 5.90440631e-01 1.88008621e-01
-1.01070666e+00 8.70245576e-01 1.34615469e+00 -5.01303494e-01
-3.61208797e-01 -4.21469659e-01 -8.16186965e-01 4.22469616e-01
5.19952476e-01 -4.27459210e-01 -3.72833848e-01 -5.73804557e-01
2.83255160e-01 3.22957397e-01 5.64439356e-01 -1.14996612e+00
1.24368027e-01 -3.01747173e-01 3.21319163e-01 -7.32083082e-01
9.01514351e-01 -7.57481337e-01 4.70046364e-02 1.81148984e-02
-7.76194036e-01 5.51020317e-02 3.86239558e-01 7.70150781e-01
-5.31867743e-01 -8.75338912e-02 3.50943446e-01 1.49385750e-01
-4.90527242e-01 5.03227673e-02 -6.44652724e-01 1.78417966e-01
7.90580809e-01 8.09228793e-02 -3.73522907e-01 -7.72887349e-01
-8.07068706e-01 1.04035601e-01 2.78601259e-01 8.49973440e-01
9.50071812e-01 -1.94489694e+00 -8.46496761e-01 -1.78162232e-02
6.09873891e-01 -4.28471208e-01 3.38345528e-01 3.59586984e-01
-1.80865943e-01 5.04392326e-01 -3.16443592e-01 -7.69869506e-01
-1.44744432e+00 7.05422461e-01 -9.67216268e-02 5.25456190e-01
-4.73935843e-01 7.79661357e-01 4.05752599e-01 4.98652905e-01
5.03173649e-01 -2.98189580e-01 -4.51836139e-01 1.05306022e-01
9.67644155e-01 2.76388954e-02 -1.92474753e-01 -9.46241558e-01
-3.54789644e-01 5.57148337e-01 2.24018872e-01 -6.74398005e-01
9.43695128e-01 -4.84672040e-01 3.25186580e-01 7.25114286e-01
1.22381973e+00 7.63865262e-02 -1.26376903e+00 -1.45912439e-01
-3.83733422e-01 -5.56171775e-01 8.00444707e-02 -8.09004366e-01
-7.58931518e-01 1.38051283e+00 7.10584223e-01 1.42414436e-01
1.19320273e+00 1.48358256e-01 8.36996675e-01 2.66031921e-01
-1.12335281e-02 -8.75356436e-01 5.67132175e-01 2.71314859e-01
1.28384531e+00 -1.12091136e+00 -5.36391675e-01 -3.71563613e-01
-9.48003590e-01 1.03382254e+00 6.26444757e-01 4.15891230e-01
2.41252258e-02 -2.73905098e-02 2.52517194e-01 2.91486800e-01
-9.11115825e-01 -2.70116627e-01 6.40117824e-01 7.68933773e-01
2.71322310e-01 8.54558125e-03 1.33914024e-01 1.08250713e+00
-2.99455494e-01 -8.80825073e-02 6.02963984e-01 6.91117167e-01
-5.41556239e-01 -7.32477427e-01 -7.01511085e-01 -2.51736641e-01
-4.27645564e-01 -3.30465943e-01 -7.60362864e-01 1.74937874e-01
-6.30910620e-02 1.13522506e+00 2.83653796e-01 -4.91442263e-01
1.75161108e-01 3.36202979e-01 4.26737756e-01 -4.30049092e-01
-2.64768153e-01 4.29148555e-01 -1.34516820e-01 -4.57619816e-01
-3.38530898e-01 -4.75669086e-01 -1.31711781e+00 5.20774484e-01
-7.34563544e-02 3.19870025e-01 7.73964286e-01 6.71163261e-01
6.01338923e-01 6.37552023e-01 5.56924701e-01 -1.15201163e+00
-2.33477652e-01 -9.85235810e-01 -1.71545073e-01 4.99487460e-01
6.73419356e-01 -6.95278406e-01 -5.04772246e-01 4.16308403e-01] | [15.193921089172363, 4.909128665924072] |
a7443b8a-f758-4080-93f5-95d3cf1b9f98 | lifting-2d-human-pose-to-3d-with-domain | 2111.11969 | null | https://arxiv.org/abs/2111.11969v1 | https://arxiv.org/pdf/2111.11969v1.pdf | Lifting 2D Human Pose to 3D with Domain Adapted 3D Body Concept | Lifting the 2D human pose to the 3D pose is an important yet challenging task. Existing 3D pose estimation suffers from 1) the inherent ambiguity between the 2D and 3D data, and 2) the lack of well labeled 2D-3D pose pairs in the wild. Human beings are able to imagine the human 3D pose from a 2D image or a set of 2D body key-points with the least ambiguity, which should be attributed to the prior knowledge of the human body that we have acquired in our mind. Inspired by this, we propose a new framework that leverages the labeled 3D human poses to learn a 3D concept of the human body to reduce the ambiguity. To have consensus on the body concept from 2D pose, our key insight is to treat the 2D human pose and the 3D human pose as two different domains. By adapting the two domains, the body knowledge learned from 3D poses is applied to 2D poses and guides the 2D pose encoder to generate informative 3D "imagination" as embedding in pose lifting. Benefiting from the domain adaptation perspective, the proposed framework unifies the supervised and semi-supervised 3D pose estimation in a principled framework. Extensive experiments demonstrate that the proposed approach can achieve state-of-the-art performance on standard benchmarks. More importantly, it is validated that the explicitly learned 3D body concept effectively alleviates the 2D-3D ambiguity in 2D pose lifting, improves the generalization, and enables the network to exploit the abundant unlabeled 2D data. | ['Yunhui Liu', 'Ziwei Liu', 'Qiang Nie'] | 2021-11-23 | null | null | null | null | ['3d-pose-estimation'] | ['computer-vision'] | [ 1.74649451e-02 3.98681343e-01 -4.55724657e-01 -2.76474506e-01
-3.77898335e-01 -5.53384662e-01 4.23633635e-01 -4.76787746e-01
-2.63681948e-01 4.53831702e-01 6.10227406e-01 2.44957358e-01
-5.27540110e-02 -5.50853908e-01 -7.78670609e-01 -5.07935703e-01
4.76252548e-02 6.99775696e-01 -2.22239375e-01 -4.57976371e-01
-2.66168803e-01 4.15317506e-01 -1.41368496e+00 -3.35850984e-01
6.46153390e-01 1.14398038e+00 -9.55087021e-02 8.02410319e-02
1.59995273e-01 6.71578497e-02 -4.79759634e-01 -2.78547138e-01
8.14084351e-01 -3.63800347e-01 -4.88220930e-01 4.84185666e-01
5.43888807e-01 -4.94167060e-01 -3.88532460e-01 9.84587371e-01
5.94890416e-01 -1.56441014e-02 6.46974742e-01 -1.13372135e+00
-5.21196008e-01 -3.06640323e-02 -7.45487034e-01 -3.33430409e-01
8.51737916e-01 1.72001570e-01 8.77207994e-01 -9.19455647e-01
7.66138494e-01 1.50023556e+00 6.92025900e-01 7.21951067e-01
-1.15840542e+00 -5.68648338e-01 2.27771237e-01 -3.12642515e-01
-1.44714236e+00 5.63275330e-02 1.30270052e+00 -6.61734402e-01
2.73953527e-01 -1.91850308e-02 1.30251348e+00 1.43812096e+00
1.41827434e-01 9.09125447e-01 1.19596624e+00 -3.94200176e-01
-2.87372191e-02 -2.19496340e-01 -2.03254491e-01 7.66959131e-01
4.19893324e-01 2.79808819e-01 -6.47610426e-01 1.44703284e-01
1.11710346e+00 1.91666797e-01 -2.27929652e-01 -1.07348871e+00
-1.27754140e+00 6.69730306e-01 6.29593194e-01 -2.04582617e-01
-4.43153560e-01 -9.40502286e-02 3.51590812e-01 -5.66233136e-02
5.03201902e-01 3.42765242e-01 -5.32023609e-01 1.28251314e-01
-4.41507876e-01 4.72749352e-01 6.64982438e-01 1.05101871e+00
7.02015162e-01 -9.03125033e-02 4.15528417e-02 4.98872668e-01
6.02530599e-01 9.28948164e-01 2.98839062e-01 -7.56711185e-01
6.54733539e-01 1.03520870e+00 1.20968059e-01 -1.00376189e+00
-4.51927960e-01 -5.35971224e-01 -7.82558441e-01 1.03583150e-01
4.15387481e-01 -1.01034613e-02 -1.01062799e+00 2.02041578e+00
7.32665598e-01 -3.62979084e-01 -8.26837346e-02 1.41433430e+00
7.03848660e-01 1.41861022e-01 -1.63455740e-01 1.00010768e-01
1.49768794e+00 -7.43703544e-01 -5.61609209e-01 -6.30106509e-01
1.82426482e-01 -3.07640642e-01 1.02375484e+00 1.64796799e-01
-8.68504286e-01 -8.40302646e-01 -1.23870194e+00 -1.92416281e-01
-2.20414981e-01 2.13447928e-01 5.49836338e-01 6.29816949e-01
-3.24516118e-01 3.21591973e-01 -7.94036210e-01 -4.48865652e-01
2.16402158e-01 4.93881077e-01 -8.13051701e-01 1.70071349e-02
-1.43283355e+00 1.08638740e+00 5.49018502e-01 3.75504017e-01
-7.84300625e-01 -5.02804458e-01 -1.29228067e+00 -4.32304621e-01
6.77956223e-01 -1.09400332e+00 8.49203348e-01 -5.14027655e-01
-1.45586300e+00 1.38124645e+00 1.13545954e-01 -2.85613954e-01
8.58376741e-01 -8.63241792e-01 5.05748950e-02 2.95477569e-01
1.74497694e-01 6.81336522e-01 1.07052755e+00 -1.32645166e+00
-4.25560512e-02 -1.05015683e+00 7.19095021e-02 6.87261462e-01
-1.59138948e-01 -8.74615192e-01 -5.88228643e-01 -8.45693231e-01
6.07654750e-01 -1.17460907e+00 1.93901006e-02 2.85173416e-01
-4.86805767e-01 -3.97200659e-02 5.09028912e-01 -7.18446076e-01
7.01195955e-01 -1.94046307e+00 7.20198035e-01 1.92009553e-01
2.67997235e-01 1.57933548e-01 1.43795460e-01 2.57899433e-01
-3.36051285e-02 -1.77529961e-01 -2.36017890e-02 -3.41357619e-01
2.62138426e-01 4.73846734e-01 -1.43251166e-01 6.99790061e-01
3.03774118e-01 1.08944798e+00 -1.11724639e+00 -6.08492076e-01
4.20834988e-01 5.31738162e-01 -5.45164347e-01 4.91281271e-01
3.48715559e-02 9.87308443e-01 -7.48406410e-01 5.88683128e-01
6.30494595e-01 -2.07486570e-01 1.91361174e-01 -7.17507839e-01
4.67212111e-01 1.91066712e-01 -1.14484394e+00 2.33544493e+00
-1.20733798e-01 -1.93304092e-01 -1.26308739e-01 -9.76185560e-01
1.09997225e+00 4.13925320e-01 6.94434941e-01 -5.73938906e-01
2.01242656e-01 2.18227148e-01 -2.52936900e-01 -5.24939179e-01
-7.41177350e-02 -5.47620714e-01 -4.12018776e-01 1.71279207e-01
4.15153116e-01 -4.58299369e-01 -2.11945966e-01 -2.44204208e-01
5.29813826e-01 8.78745973e-01 7.04001486e-01 -1.14742450e-01
5.31166136e-01 -2.53370404e-01 7.00910568e-01 2.50011474e-01
-3.46977532e-01 4.99255091e-01 2.83683360e-01 -6.34547293e-01
-9.86164093e-01 -1.64392245e+00 1.72456041e-01 6.29684091e-01
4.37224180e-01 -2.32784063e-01 -6.93626046e-01 -9.61663961e-01
3.95666361e-01 8.36251080e-02 -8.93109024e-01 -3.55240613e-01
-7.01644361e-01 -2.14034364e-01 4.17518079e-01 6.13718808e-01
6.20330215e-01 -4.93596166e-01 -8.01504672e-01 -2.08094135e-01
-3.26788038e-01 -1.23431504e+00 -5.91813207e-01 1.35607809e-01
-9.77632046e-01 -1.11583018e+00 -9.11164463e-01 -6.21848106e-01
7.66437173e-01 7.60333659e-03 9.87545669e-01 -3.23770940e-01
-9.07437578e-02 6.05393589e-01 -4.19631660e-01 -4.44906741e-01
-1.08460806e-01 -7.24106729e-02 5.98962665e-01 -4.44146581e-02
4.32866693e-01 -7.29589760e-01 -6.50624275e-01 4.66875196e-01
-4.52932984e-01 2.37388879e-01 7.84325004e-01 9.47911859e-01
7.44469702e-01 -2.07687810e-01 3.85636061e-01 -5.46580017e-01
1.88449308e-01 -1.48391649e-01 -9.71072838e-02 3.87252085e-02
-3.90239030e-01 1.78547353e-01 2.98169225e-01 -5.34198046e-01
-1.00820374e+00 4.50739294e-01 -7.86663666e-02 -6.75687373e-01
-3.30443770e-01 2.26689950e-01 -5.19726634e-01 9.49138999e-02
6.32072985e-01 1.52781427e-01 4.60471958e-01 -7.81281829e-01
4.86021668e-01 1.97841257e-01 6.50361657e-01 -9.65717256e-01
1.21161509e+00 7.22351015e-01 3.75991076e-01 -4.22838986e-01
-1.23275828e+00 -4.86722022e-01 -1.27332675e+00 -2.98451573e-01
1.00363910e+00 -1.25202298e+00 -7.33263493e-01 2.99887538e-01
-9.61633623e-01 1.25850171e-01 -4.20570910e-01 5.71897566e-01
-8.67777705e-01 6.34319425e-01 -3.57875317e-01 -7.52238989e-01
-4.30158079e-01 -1.12733924e+00 1.66552460e+00 -5.05011529e-02
-5.02517879e-01 -7.67199457e-01 -4.16232273e-02 4.96279776e-01
-3.41861904e-01 7.96725929e-01 6.78574681e-01 -4.64779347e-01
-1.26845255e-01 -4.81868863e-01 1.38889939e-01 5.11682451e-01
2.56647259e-01 -8.38401556e-01 -7.11422861e-01 -4.29832458e-01
2.96237767e-01 -6.67774618e-01 2.90059149e-01 2.39936426e-01
5.85325599e-01 -1.04809023e-01 -1.02263160e-01 6.29418075e-01
1.00426638e+00 -3.68131876e-01 2.05567986e-01 7.29536638e-02
8.66718888e-01 9.51310813e-01 7.68232644e-01 4.80838716e-01
4.68476981e-01 8.62821639e-01 5.20460725e-01 2.03879308e-02
-1.34591311e-01 -9.70635593e-01 2.30024427e-01 8.27752292e-01
-3.31375659e-01 3.68862331e-01 -8.44865322e-01 2.09270597e-01
-1.69572091e+00 -5.27557850e-01 3.60650986e-01 2.30219316e+00
8.93764079e-01 2.14253798e-01 2.12406829e-01 2.21996471e-01
5.88695943e-01 2.07507297e-01 -8.20651054e-01 3.93376440e-01
1.10457316e-01 4.71049324e-02 2.58801162e-01 1.41570777e-01
-1.06459844e+00 6.85988069e-01 5.38487387e+00 4.32100207e-01
-1.05475664e+00 -1.82947919e-01 7.62196332e-02 -1.06118448e-01
1.32595040e-02 -1.92438111e-01 -8.51023138e-01 3.24759573e-01
6.97015449e-02 5.71153685e-02 1.64127320e-01 8.84387016e-01
5.01946695e-02 3.14501107e-01 -1.64017200e+00 1.26011705e+00
3.05178642e-01 -5.83027363e-01 3.83275956e-01 2.78306007e-01
5.36344707e-01 -4.87913400e-01 4.78113405e-02 3.26581836e-01
-2.37533331e-01 -9.19602871e-01 1.05351484e+00 4.32072520e-01
9.66538846e-01 -4.91198927e-01 5.64757049e-01 7.53163099e-01
-1.28842008e+00 5.76247796e-02 -2.13913992e-02 -2.63282388e-01
2.40131453e-01 3.75114292e-01 -6.77438498e-01 8.11573148e-01
6.07586443e-01 8.22084904e-01 -3.14526170e-01 4.49120253e-01
-6.44208848e-01 1.18648792e-02 -4.01932240e-01 2.42608622e-01
-3.31143588e-02 -2.32457265e-01 6.95119917e-01 6.26095951e-01
2.24355802e-01 6.50467277e-02 5.29937387e-01 8.95548701e-01
2.57383585e-02 -1.43150426e-02 -7.45193243e-01 1.73143789e-01
2.79946953e-01 1.00130665e+00 -3.91841710e-01 -9.27443206e-02
-8.41531828e-02 1.07860470e+00 1.01054519e-01 2.71433979e-01
-7.45083570e-01 5.22274431e-03 5.80639243e-01 1.48040310e-01
1.51732579e-01 -3.84841979e-01 -1.42452046e-01 -1.38252461e+00
3.44902217e-01 -8.85645270e-01 2.91266590e-01 -7.69189417e-01
-1.39981151e+00 2.23517537e-01 4.13965821e-01 -1.56084180e+00
-4.72700387e-01 -9.26136434e-01 4.54898924e-02 8.32854629e-01
-1.17672372e+00 -1.36044908e+00 -4.13123071e-01 5.75531960e-01
3.51831138e-01 1.96358442e-01 7.55124569e-01 1.51177540e-01
-3.99794430e-02 5.87919712e-01 -7.04909682e-01 4.61126089e-01
8.32266152e-01 -1.24120021e+00 3.64021212e-01 4.18637544e-01
1.58963382e-01 8.34878802e-01 5.89605391e-01 -6.82686269e-01
-1.84138942e+00 -6.43716276e-01 6.02998912e-01 -8.47953200e-01
1.83250040e-01 -6.33468807e-01 -4.25519347e-01 6.72204792e-01
-5.34670055e-01 1.43506423e-01 6.74228668e-01 1.04519680e-01
-5.56106806e-01 -9.55807343e-02 -1.13592064e+00 5.61634064e-01
1.57365286e+00 -5.50266385e-01 -1.23094201e+00 2.30079368e-01
7.13465154e-01 -7.59977818e-01 -1.16786873e+00 7.45492041e-01
9.10164893e-01 -6.52191639e-01 1.42349601e+00 -6.91309690e-01
4.11322147e-01 -3.54890555e-01 -3.72976065e-01 -1.16776431e+00
-4.45689000e-02 -3.56642574e-01 -5.09516537e-01 6.54824972e-01
-1.80091903e-01 -4.47510153e-01 1.03933764e+00 3.00482005e-01
8.59156922e-02 -8.77134144e-01 -1.11713850e+00 -8.70522738e-01
8.04855302e-02 -1.72468439e-01 5.76931059e-01 7.18439639e-01
-1.44621968e-01 5.55442810e-01 -7.19192088e-01 2.78200507e-01
9.16654110e-01 3.68870080e-01 1.22238696e+00 -1.57486284e+00
-3.58365655e-01 -7.79203931e-03 -7.94891357e-01 -1.79103947e+00
2.29937762e-01 -7.43918061e-01 4.86478321e-02 -1.14271462e+00
1.93345740e-01 1.20858848e-01 -3.19328494e-02 3.69442254e-01
-1.33104697e-01 2.88562119e-01 3.43540639e-01 2.48639539e-01
-3.30958903e-01 8.49188268e-01 1.79279864e+00 -6.30948171e-02
-2.82784794e-02 -1.34540632e-01 -5.79293668e-01 9.81369197e-01
2.40681916e-01 -2.27445140e-01 -5.53343356e-01 -4.00165528e-01
2.27279618e-01 8.13772678e-02 6.99914992e-01 -9.20410275e-01
-1.28381327e-01 2.49466510e-03 7.40638793e-01 -6.79005325e-01
7.04945207e-01 -9.78664756e-01 -3.62461247e-02 5.35415947e-01
-1.22779340e-01 -3.05464149e-01 -1.62448972e-01 9.04244483e-01
-1.16922520e-01 1.80503055e-01 6.04989409e-01 -5.14206529e-01
-6.83611929e-01 4.78062958e-01 4.46244091e-01 3.80095631e-01
9.28331673e-01 -6.09085917e-01 2.78164446e-01 -3.83745760e-01
-8.98081839e-01 2.25645766e-01 6.78201735e-01 5.81381261e-01
5.74647844e-01 -1.67805374e+00 -4.78302240e-01 5.87964058e-01
3.41559947e-01 4.11827505e-01 2.38223761e-01 9.06986237e-01
-2.68430203e-01 3.32863092e-01 -4.68227565e-01 -9.84094560e-01
-9.30242598e-01 5.81767440e-01 3.39558065e-01 -2.49602139e-01
-7.22166836e-01 5.93753576e-01 4.51920033e-01 -8.93234730e-01
3.13992560e-01 -2.22126409e-01 9.96343568e-02 -6.11771867e-02
7.90087059e-02 1.98203206e-01 -2.52414286e-01 -1.11219966e+00
-4.32790548e-01 1.20469022e+00 2.13409469e-01 -2.74331383e-02
1.13114464e+00 -2.02513590e-01 3.46369863e-01 4.67980355e-01
1.38803661e+00 -2.74835601e-02 -1.51556718e+00 -4.31051731e-01
-3.44044209e-01 -5.56591213e-01 -4.88630950e-01 -6.38712525e-01
-8.34753811e-01 1.06429148e+00 4.83047783e-01 -3.98248672e-01
1.01212895e+00 1.14259616e-01 8.46885741e-01 4.02115643e-01
7.19890177e-01 -1.08850670e+00 4.37388003e-01 3.69432479e-01
1.04164982e+00 -1.36144912e+00 2.93254167e-01 -6.13103986e-01
-6.11707687e-01 9.97551203e-01 6.67680502e-01 -2.50069737e-01
5.77924192e-01 -2.13780046e-01 9.82586667e-02 -3.43125790e-01
-4.25600037e-02 -1.58786565e-01 8.94709766e-01 7.61951745e-01
2.91134417e-01 1.72983259e-01 -2.53327370e-01 7.70691693e-01
-2.82990187e-01 -1.72102824e-03 -3.23797226e-01 9.46838200e-01
-1.39164433e-01 -1.03488719e+00 -5.35239637e-01 -5.34435622e-02
-1.18603207e-01 4.94611353e-01 -5.45504689e-01 1.12024486e+00
4.68158066e-01 3.80608767e-01 -2.73369938e-01 -5.08970737e-01
7.13208258e-01 3.28306437e-01 8.94235849e-01 -8.04467857e-01
-4.22380306e-02 1.87419921e-01 -2.50660151e-01 -6.89444780e-01
-5.84025264e-01 -2.59137005e-01 -1.19523621e+00 1.60940260e-01
-1.45623386e-01 -6.53879568e-02 4.06708241e-01 9.36174095e-01
3.33604068e-01 6.81296960e-02 2.80568451e-01 -1.37320137e+00
-1.08717489e+00 -8.15173447e-01 -7.11197495e-01 1.09283185e+00
3.20134997e-01 -1.34640336e+00 -2.93716490e-01 7.33613372e-02] | [7.001377582550049, -1.0002002716064453] |
9daf56d0-e4cc-477e-aa1f-101659cb16e4 | one-shot-high-fidelity-talking-head-synthesis | 2304.05097 | null | https://arxiv.org/abs/2304.05097v1 | https://arxiv.org/pdf/2304.05097v1.pdf | One-Shot High-Fidelity Talking-Head Synthesis with Deformable Neural Radiance Field | Talking head generation aims to generate faces that maintain the identity information of the source image and imitate the motion of the driving image. Most pioneering methods rely primarily on 2D representations and thus will inevitably suffer from face distortion when large head rotations are encountered. Recent works instead employ explicit 3D structural representations or implicit neural rendering to improve performance under large pose changes. Nevertheless, the fidelity of identity and expression is not so desirable, especially for novel-view synthesis. In this paper, we propose HiDe-NeRF, which achieves high-fidelity and free-view talking-head synthesis. Drawing on the recently proposed Deformable Neural Radiance Fields, HiDe-NeRF represents the 3D dynamic scene into a canonical appearance field and an implicit deformation field, where the former comprises the canonical source face and the latter models the driving pose and expression. In particular, we improve fidelity from two aspects: (i) to enhance identity expressiveness, we design a generalized appearance module that leverages multi-scale volume features to preserve face shape and details; (ii) to improve expression preciseness, we propose a lightweight deformation module that explicitly decouples the pose and expression to enable precise expression modeling. Extensive experiments demonstrate that our proposed approach can generate better results than previous works. Project page: https://www.waytron.net/hidenerf/ | ['Xuelong Li', 'Liefeng Bo', 'Zhongjian Wang', 'Bang Zhang', 'Mulin Chen', 'Zhigang Wang', 'Bin Zhao', 'Dong Wang', 'Longhao Zhang', 'Weichuang Li'] | 2023-04-11 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Li_One-Shot_High-Fidelity_Talking-Head_Synthesis_With_Deformable_Neural_Radiance_Field_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Li_One-Shot_High-Fidelity_Talking-Head_Synthesis_With_Deformable_Neural_Radiance_Field_CVPR_2023_paper.pdf | cvpr-2023-1 | ['talking-head-generation', 'neural-rendering'] | ['computer-vision', 'computer-vision'] | [-1.65890772e-02 2.54077226e-01 1.21823981e-01 -7.58049786e-01
-4.24678892e-01 -4.83581930e-01 6.29511535e-01 -7.24405706e-01
1.85186133e-01 4.98879701e-01 5.94856679e-01 3.42442185e-01
2.42343456e-01 -5.67577899e-01 -7.06917822e-01 -7.82948494e-01
4.40909803e-01 1.36211589e-01 -3.85540396e-01 -4.12562042e-01
-2.38307729e-01 6.95377946e-01 -1.65216517e+00 3.71236242e-02
7.48599291e-01 1.09565842e+00 -1.88454509e-01 3.67130518e-01
1.31652474e-01 7.42401183e-01 -4.28758591e-01 -6.65773153e-01
3.64792943e-01 -5.03792286e-01 -4.97372895e-01 4.09429669e-01
7.42284775e-01 -5.38523257e-01 -5.38867593e-01 9.53978598e-01
8.25938284e-01 2.65793025e-01 4.49049175e-01 -1.35359335e+00
-7.79571593e-01 -2.72637606e-02 -7.07058728e-01 -3.90167207e-01
6.54833555e-01 3.15017074e-01 5.31532407e-01 -9.24685180e-01
8.14918756e-01 1.49280822e+00 6.69761598e-01 9.21823025e-01
-1.29266334e+00 -8.77085567e-01 3.83842498e-01 -1.38107156e-02
-1.56968963e+00 -9.92162943e-01 1.13561594e+00 -2.83680588e-01
4.54870164e-01 2.98858434e-01 6.96249902e-01 1.32839799e+00
3.01960148e-02 5.96695960e-01 1.07767558e+00 -5.28894253e-02
-7.12306518e-03 7.59437075e-03 -3.32919002e-01 7.70330906e-01
-2.67631292e-01 -1.90331880e-03 -5.57000637e-01 -9.03184712e-02
9.26982522e-01 -1.17576040e-01 -7.14148343e-01 -5.65318048e-01
-9.03273523e-01 7.25850284e-01 3.69083673e-01 6.35187551e-02
-2.91001767e-01 9.83039662e-02 1.76329985e-01 -2.68346351e-02
4.96228397e-01 2.24374570e-02 4.69940566e-02 4.41934774e-03
-9.26482677e-01 5.81452489e-01 5.86522043e-01 9.04339731e-01
7.96580791e-01 3.03884834e-01 -1.96052164e-01 9.64866638e-01
3.81181747e-01 5.56901455e-01 3.59856784e-01 -1.18917310e+00
1.74863875e-01 4.67713594e-01 -8.06118473e-02 -1.25986743e+00
-3.64026248e-01 -3.18251222e-01 -8.46632481e-01 2.39873499e-01
8.84620994e-02 -1.06263526e-01 -9.04805779e-01 2.19390416e+00
6.91329062e-01 1.36286736e-01 -1.76922455e-01 1.11814940e+00
9.83954012e-01 5.80005646e-01 -9.01450813e-02 -2.68188596e-01
1.30712175e+00 -7.70741940e-01 -1.08156097e+00 -1.57337308e-01
3.34330738e-01 -6.50317609e-01 8.71691644e-01 -9.84268356e-03
-1.42302680e+00 -4.35234636e-01 -9.76461887e-01 -3.50880653e-01
7.73737058e-02 5.98555505e-02 5.09729505e-01 5.95453203e-01
-1.32515073e+00 2.71505773e-01 -5.39199948e-01 -6.40976802e-02
3.38701576e-01 4.91535872e-01 -6.55907571e-01 7.88250789e-02
-1.13914323e+00 7.57731378e-01 -2.88832098e-01 2.16425762e-01
-5.71510553e-01 -7.62618482e-01 -1.13648653e+00 -5.28913029e-02
6.01525158e-02 -1.02020741e+00 1.17765915e+00 -1.20727205e+00
-1.98564863e+00 1.06164706e+00 -3.78804475e-01 6.68753982e-02
7.54260004e-01 -3.19777466e-02 -3.27124596e-01 1.29763350e-01
-1.56764090e-01 7.77091742e-01 1.00187767e+00 -1.49638438e+00
5.25808558e-02 -7.37161636e-01 1.05287805e-01 4.78186429e-01
-3.06363881e-01 -1.08568110e-01 -6.57150030e-01 -9.11492050e-01
2.42046323e-02 -1.10253835e+00 1.78030014e-01 4.18072999e-01
-2.92265445e-01 1.68889701e-01 1.00673723e+00 -7.59943902e-01
9.37405229e-01 -2.18687344e+00 3.38314891e-01 8.46918449e-02
3.72047514e-01 7.66932294e-02 -1.88266989e-02 2.75517558e-03
-3.12597901e-01 -1.31513000e-01 -1.91251025e-01 -8.10391724e-01
8.90329331e-02 1.78193316e-01 -2.49225110e-01 7.26456404e-01
1.30357236e-01 1.03628826e+00 -5.41554928e-01 -3.96187484e-01
1.04220800e-01 1.12006128e+00 -9.10011828e-01 3.53185803e-01
6.22735173e-02 8.64078224e-01 -3.71010512e-01 4.59071428e-01
1.08756161e+00 -1.06676817e-02 1.40856311e-01 -5.78390002e-01
7.42764249e-02 5.15780486e-02 -1.15301096e+00 1.80646718e+00
-6.24483705e-01 3.36319566e-01 7.09912002e-01 -5.88969767e-01
9.27575290e-01 4.23367739e-01 5.64466536e-01 -7.08475232e-01
3.14504772e-01 -6.76772892e-02 -3.82886320e-01 -3.52715582e-01
4.21874940e-01 -5.20428896e-01 1.34582641e-02 1.52155653e-01
3.98538560e-02 -3.73367369e-01 -3.32689881e-01 3.83063592e-02
5.80823421e-01 5.00581264e-01 -6.41024932e-02 -1.41793892e-01
6.38686359e-01 -8.05885494e-01 7.76084244e-01 -8.76870528e-02
-1.72128454e-01 9.52881098e-01 4.03591394e-01 -2.86683291e-01
-8.86610508e-01 -1.11166894e+00 -9.75791290e-02 8.94436777e-01
1.76359683e-01 -3.42399836e-01 -1.08397019e+00 -3.95704329e-01
-9.66339111e-02 6.08070850e-01 -7.33088911e-01 -2.63021559e-01
-7.82185555e-01 -5.06923199e-01 5.57587326e-01 4.28978115e-01
4.97265190e-01 -7.26330638e-01 -2.77460665e-01 -1.53965279e-01
-4.47335958e-01 -1.22930539e+00 -1.02762651e+00 -5.59100926e-01
-4.16449606e-01 -6.31232023e-01 -8.86819959e-01 -6.03844464e-01
8.23615849e-01 5.72647564e-02 7.28137672e-01 -6.30732030e-02
-1.64634734e-01 3.86867762e-01 1.26871523e-02 -7.91852996e-02
-3.44983548e-01 -4.19773310e-01 3.45139682e-01 5.84774256e-01
-1.86632052e-01 -1.00715613e+00 -7.37137735e-01 2.55763561e-01
-8.21214080e-01 3.10197085e-01 1.78520173e-01 7.32914329e-01
4.06150371e-01 -4.46195930e-01 3.46624732e-01 -6.03961349e-01
3.43662381e-01 -1.14946172e-01 -3.47366780e-01 4.92984615e-02
-2.30892316e-01 3.09836864e-02 5.92077136e-01 -4.52553838e-01
-1.51470971e+00 1.64059520e-01 -3.56782585e-01 -8.30481946e-01
5.00742607e-02 -2.05501512e-01 -7.31126010e-01 -2.19333380e-01
3.39533746e-01 3.13729048e-01 2.80742526e-01 -4.17995781e-01
4.65605915e-01 4.78881955e-01 7.09284604e-01 -7.28656411e-01
8.70098650e-01 7.46817946e-01 -6.44894615e-02 -7.75759280e-01
-4.37306345e-01 5.21356016e-02 -5.77946782e-01 -3.79100025e-01
7.75757194e-01 -1.12282789e+00 -1.01899636e+00 6.04363739e-01
-1.20111346e+00 -1.84122145e-01 -1.13049939e-01 2.44369283e-01
-7.03816533e-01 3.22840810e-01 -5.51365018e-01 -7.23177910e-01
-4.48284984e-01 -1.29142451e+00 1.49071598e+00 2.68811911e-01
-1.95487872e-01 -8.13356876e-01 2.15561246e-03 5.32470703e-01
4.39264625e-01 6.55261576e-01 5.77004611e-01 2.80120522e-01
-4.14261937e-01 6.68547302e-02 -2.94751879e-02 2.28757128e-01
1.23079382e-01 -3.67761813e-02 -1.23759985e+00 -3.58118683e-01
2.28993222e-01 -2.94411987e-01 4.35867906e-01 2.02957138e-01
1.04004741e+00 -5.44775665e-01 -1.08957216e-01 1.14656723e+00
9.89902020e-01 5.08102961e-02 6.84372485e-01 -2.09550992e-01
9.88591254e-01 9.72840786e-01 1.23320244e-01 5.96054196e-01
6.57517850e-01 1.19855154e+00 1.96458459e-01 -1.67001247e-01
-3.65462512e-01 -5.07615805e-01 5.65792322e-01 8.99379969e-01
-1.48099408e-01 6.82787895e-02 -4.90981162e-01 3.43794554e-01
-1.62277949e+00 -9.56553042e-01 3.68902534e-01 2.03305197e+00
8.65713596e-01 -3.39934289e-01 8.77369046e-02 -1.61517233e-01
6.61442339e-01 3.00460368e-01 -6.17102265e-01 -3.92177761e-01
-2.49754712e-01 1.50701880e-01 -2.39825738e-03 5.75402737e-01
-7.29819059e-01 7.16388047e-01 5.25132895e+00 5.94968915e-01
-1.48835158e+00 2.10510865e-01 6.73262000e-01 -4.09264356e-01
-6.03219211e-01 -2.37708926e-01 -7.53287315e-01 2.83850908e-01
5.54353178e-01 -2.55835474e-01 3.10545802e-01 8.11464190e-01
4.12002802e-01 4.17900085e-01 -9.51124430e-01 1.22305584e+00
3.98623228e-01 -1.07546329e+00 2.12284952e-01 1.76521614e-01
6.01323187e-01 -5.22933841e-01 2.99046576e-01 1.72115400e-01
-1.57591790e-01 -1.12393260e+00 1.09713638e+00 6.23380661e-01
1.11606717e+00 -8.73651564e-01 1.96028292e-01 7.39058480e-02
-1.22674668e+00 2.21406877e-01 6.53366596e-02 1.04780816e-01
3.87172967e-01 2.70988882e-01 -4.13340181e-01 5.00759661e-01
5.65307975e-01 5.95274508e-01 -2.73640960e-01 4.02105957e-01
-1.10222138e-01 6.84428513e-02 -2.42749915e-01 4.17578578e-01
-1.50082847e-02 -2.67126411e-01 7.40877330e-01 1.03973877e+00
2.19763786e-01 2.31679186e-01 3.76943573e-02 1.08560109e+00
-2.90759057e-01 2.29027972e-01 -6.54733062e-01 3.34306508e-01
2.25389093e-01 1.29366517e+00 -2.04449877e-01 -2.96152383e-02
-3.15174818e-01 1.18957698e+00 2.87745744e-01 4.13034320e-01
-1.04843032e+00 -4.04542983e-02 1.24299395e+00 4.60379034e-01
2.47902751e-01 -7.88546205e-02 -8.45852867e-02 -1.48907673e+00
2.61345625e-01 -8.72553349e-01 -1.69504046e-01 -8.79375100e-01
-8.81987154e-01 7.22113311e-01 -3.68525200e-02 -9.57116961e-01
-3.68939281e-01 -2.94620752e-01 -5.49912870e-01 9.25874710e-01
-1.20904672e+00 -1.75074100e+00 -6.51620209e-01 8.13891590e-01
3.69353056e-01 2.46226400e-01 7.98590302e-01 4.24810648e-01
-6.65282071e-01 1.09304941e+00 -2.86389858e-01 5.50838076e-02
8.20590615e-01 -7.32163250e-01 3.88641417e-01 4.64670241e-01
-1.75407872e-01 7.59652615e-01 5.94802320e-01 -3.87716860e-01
-1.60088909e+00 -1.07287955e+00 5.46259463e-01 -3.10505688e-01
1.88587129e-01 -7.14908540e-01 -8.77104580e-01 7.40753949e-01
-1.03454432e-02 2.47670680e-01 5.29202998e-01 -3.07398617e-01
-5.70878327e-01 -2.68558770e-01 -1.35439873e+00 7.49566019e-01
1.30952120e+00 -6.29329324e-01 -2.63969243e-01 -3.31895985e-02
5.83228707e-01 -6.55385792e-01 -9.44136918e-01 5.83424985e-01
8.22080612e-01 -1.06179702e+00 9.17738199e-01 -1.69421270e-01
2.81165421e-01 -3.63225996e-01 -1.88011393e-01 -1.18095505e+00
-2.59328961e-01 -1.03145528e+00 -1.79277897e-01 1.41514361e+00
-4.41768542e-02 -5.45021594e-01 7.74823189e-01 8.25252712e-01
-4.70641665e-02 -8.45186949e-01 -1.01547825e+00 -4.90103215e-01
4.86077480e-02 -1.67976603e-01 8.66833508e-01 1.02472305e+00
-1.91955000e-01 3.94992232e-01 -6.35135114e-01 -6.00910522e-02
5.45902371e-01 1.33305550e-01 9.31961417e-01 -8.15411270e-01
-2.09645361e-01 -3.43157560e-01 -5.26364744e-01 -1.15119684e+00
5.96648216e-01 -7.64247894e-01 -6.17085397e-02 -1.05763221e+00
8.08984786e-02 -1.87523246e-01 3.20034862e-01 4.18783098e-01
-6.26973156e-03 4.90723878e-01 3.17779273e-01 6.79432824e-02
-1.26291111e-01 1.04306316e+00 1.52471411e+00 1.11964576e-01
-1.20705687e-01 -1.97137967e-01 -7.49522269e-01 8.49181473e-01
5.20728588e-01 4.01963778e-02 -5.54126263e-01 -4.78557527e-01
-1.88350111e-01 1.26183182e-01 4.67790276e-01 -6.95950091e-01
-6.60808757e-03 -3.81308533e-02 5.44257104e-01 -1.48682803e-01
8.55601013e-01 -5.85031092e-01 5.89464307e-01 9.48765576e-02
-5.52782863e-02 1.00093633e-01 1.36707470e-01 3.63822997e-01
-1.28256097e-01 3.89098734e-01 1.16830456e+00 3.09848543e-02
-4.22112525e-01 5.62510312e-01 5.34114800e-02 -4.77397032e-02
9.25719857e-01 -1.95515513e-01 1.07022524e-01 -9.10057127e-01
-6.49219751e-01 9.54334810e-03 9.84947681e-01 6.51382208e-01
6.81713641e-01 -1.60845685e+00 -7.63148725e-01 4.81031090e-01
9.64402631e-02 5.36667816e-02 6.13110840e-01 8.95749986e-01
-4.37241048e-01 6.24055564e-02 -1.95227459e-01 -4.54701155e-01
-1.26158857e+00 3.51752669e-01 5.68285942e-01 7.11178482e-02
-7.02218354e-01 8.39132786e-01 9.03902709e-01 -5.92072129e-01
9.51334909e-02 1.27105236e-01 -7.54018947e-02 -4.80005424e-03
6.61938787e-01 9.77059603e-02 -5.28347492e-02 -1.53156853e+00
-4.78090227e-01 9.47199881e-01 -5.33380499e-03 -8.21346119e-02
1.28802872e+00 -2.99620956e-01 -8.42551365e-02 5.57430983e-02
1.55377674e+00 2.69924819e-01 -1.51988447e+00 2.04539392e-03
-6.68737054e-01 -5.86482346e-01 1.03541084e-01 -4.54213202e-01
-1.43485546e+00 7.49979019e-01 5.50571382e-01 -6.10387027e-01
1.36031210e+00 -1.49815843e-01 9.93080616e-01 -1.80276021e-01
3.53616476e-01 -8.05098057e-01 -5.50976582e-03 2.08167195e-01
1.31352031e+00 -8.36292744e-01 -1.28318667e-01 -6.17868960e-01
-7.19301522e-01 9.55497503e-01 6.96043015e-01 1.28554836e-01
5.52557290e-01 4.08160448e-01 1.03226021e-01 -1.03228867e-01
-4.87802744e-01 1.99141949e-01 3.84001821e-01 6.45586193e-01
5.00188768e-01 9.40964499e-04 8.57681409e-03 6.25560403e-01
-4.98259038e-01 -2.60471441e-02 2.34345704e-01 6.59862697e-01
1.87228188e-01 -1.01284182e+00 -5.54325759e-01 -9.03241560e-02
-3.48308057e-01 1.85933754e-01 -4.72715050e-01 5.26014626e-01
2.65536577e-01 6.56003296e-01 7.84429759e-02 -4.17885542e-01
6.25445783e-01 9.09264758e-02 7.18805969e-01 -2.76517481e-01
-4.55521524e-01 1.99583560e-01 5.20637026e-03 -8.21935892e-01
-3.28688353e-01 -5.69578409e-01 -1.23371601e+00 -6.95186794e-01
-1.21082366e-01 -1.17320612e-01 6.48019433e-01 5.54999292e-01
6.31903708e-01 3.23729545e-01 8.33801031e-01 -1.28513253e+00
-4.06507730e-01 -7.42327750e-01 -5.35554767e-01 7.87753403e-01
4.22389448e-01 -9.36951518e-01 -3.36654335e-01 1.80514187e-01] | [12.8928861618042, -0.3339753746986389] |
dd575ad3-ec39-461e-a4cc-170ea266fafc | rhythm-controllable-attention-with-high | 2306.02593 | null | https://arxiv.org/abs/2306.02593v1 | https://arxiv.org/pdf/2306.02593v1.pdf | Rhythm-controllable Attention with High Robustness for Long Sentence Speech Synthesis | Regressive Text-to-Speech (TTS) system utilizes attention mechanism to generate alignment between text and acoustic feature sequence. Alignment determines synthesis robustness (e.g, the occurence of skipping, repeating, and collapse) and rhythm via duration control. However, current attention algorithms used in speech synthesis cannot control rhythm using external duration information to generate natural speech while ensuring robustness. In this study, we propose Rhythm-controllable Attention (RC-Attention) based on Tracotron2, which improves robustness and naturalness simultaneously. Proposed attention adopts a trainable scalar learned from four kinds of information to achieve rhythm control, which makes rhythm control more robust and natural, even when synthesized sentences are extremely longer than training corpus. We use word errors counting and AB preference test to measure robustness of proposed method and naturalness of synthesized speech, respectively. Results shows that RC-Attention has the lowest word error rate of nearly 0.6%, compared with 11.8% for baseline system. Moreover, nearly 60% subjects prefer to the speech synthesized with RC-Attention to that with Forward Attention, because the former has more natural rhythm. | ['Binghuai Lin', 'Jiaen Liang', 'Jianqing Sun', 'Ya Li', 'Qi Luo', 'Jinlong Xue', 'Yukang Jia', 'Yayue Deng', 'Dengfeng Ke'] | 2023-06-05 | null | null | null | null | ['speech-synthesis'] | ['speech'] | [ 1.68283656e-01 2.01143082e-02 1.32871866e-01 -5.27812019e-02
-7.25494146e-01 -4.57526863e-01 4.20588911e-01 -3.62800866e-01
-2.41613105e-01 6.41693294e-01 7.25889146e-01 -3.14192802e-01
2.43347332e-01 -4.45869982e-01 -3.64859104e-01 -8.07482123e-01
4.93925601e-01 -1.12585418e-01 1.99832007e-01 -4.75008935e-01
3.92549008e-01 1.80666164e-01 -1.33040798e+00 7.19110519e-02
8.09871197e-01 5.30566931e-01 6.39030516e-01 1.07692289e+00
7.16745332e-02 4.03941035e-01 -1.10216057e+00 7.80952349e-02
1.82252936e-02 -7.72519052e-01 -3.67144793e-01 -1.90339446e-01
1.70406178e-01 -1.82304010e-01 -3.51370662e-01 8.47855747e-01
1.00253069e+00 4.47303295e-01 6.46319747e-01 -9.34001267e-01
-1.37086785e+00 8.94229710e-01 -2.52831846e-01 5.38721681e-01
3.95798475e-01 4.90048796e-01 8.92138183e-01 -6.95632756e-01
1.48875251e-01 1.35247505e+00 2.94115514e-01 9.44140673e-01
-9.82949615e-01 -6.59224153e-01 2.14426011e-01 -5.20878285e-02
-1.22963595e+00 -9.19811904e-01 6.40732050e-01 -2.06701279e-01
1.27082205e+00 6.75363064e-01 3.54986012e-01 1.20290196e+00
4.22665566e-01 3.50690186e-01 9.09512043e-01 -5.80120802e-01
-1.36235533e-02 7.60773942e-02 -1.54866233e-01 7.24280253e-02
-1.62922084e-01 2.35167891e-01 -5.48990607e-01 3.61829668e-01
6.30842149e-01 -6.22150004e-01 -5.15518904e-01 6.64748073e-01
-1.41392219e+00 4.78802979e-01 7.68723786e-02 4.71519023e-01
-2.65179157e-01 9.52297747e-02 3.81198674e-01 4.33921069e-01
7.98779801e-02 5.37945390e-01 -4.81984049e-01 -3.91445220e-01
-5.28045118e-01 3.54861915e-02 2.27326304e-01 1.02413678e+00
-2.39250604e-02 9.95888293e-01 -7.63100863e-01 1.01474738e+00
3.16278845e-01 1.00939274e+00 1.16369414e+00 -5.81081569e-01
6.70326173e-01 -1.54117849e-02 8.47388506e-02 -8.42416406e-01
-2.49398753e-01 -4.10924196e-01 -6.50646269e-01 -8.01660269e-02
-3.59853432e-02 -3.09989542e-01 -9.98675168e-01 2.01244116e+00
-4.26403023e-02 -2.20158339e-01 3.13980043e-01 9.27939951e-01
8.35663259e-01 1.15094531e+00 -4.75968309e-02 -6.81656420e-01
1.30597079e+00 -9.00321960e-01 -1.31276309e+00 -1.82325915e-01
1.09079510e-01 -1.18041277e+00 1.69175136e+00 3.37659389e-01
-1.01713157e+00 -9.04293120e-01 -1.19925034e+00 4.41468991e-02
4.07521017e-02 1.65097877e-01 -7.53677189e-02 8.05595815e-01
-9.08053100e-01 2.62876451e-01 -4.00245398e-01 -1.93090811e-01
-3.66801620e-01 2.82030821e-01 -1.00139715e-01 7.44014919e-01
-1.49917054e+00 7.81890213e-01 2.54815459e-01 -6.42529726e-02
-6.67101145e-01 -4.28761452e-01 -6.53684676e-01 7.35266507e-02
1.23618931e-01 -3.25945139e-01 1.40190649e+00 -8.48040223e-01
-2.26936269e+00 1.91516608e-01 -6.24382086e-02 -2.19593719e-01
2.39896968e-01 -4.83631998e-01 -8.55565012e-01 -9.15773660e-02
-1.51620045e-01 7.02819169e-01 1.09093952e+00 -7.82741129e-01
-3.79134536e-01 2.16920406e-01 -3.89541000e-01 4.45238173e-01
-3.77171904e-01 1.08303756e-01 -2.10229307e-01 -1.11606586e+00
-6.82912841e-02 -1.09411192e+00 1.50711849e-01 -6.69934750e-01
-5.58677673e-01 -3.53744388e-01 7.98259020e-01 -8.69725525e-01
1.80560386e+00 -2.10927320e+00 1.17905535e-01 -1.60305306e-01
-3.32751364e-01 4.83537078e-01 -2.95412987e-01 4.34134841e-01
-1.20124966e-01 3.47752362e-01 -1.81679726e-01 -3.86150517e-02
-9.60727185e-02 -1.95459038e-01 -4.92271096e-01 1.40125617e-01
3.10619265e-01 5.63424289e-01 -5.68866253e-01 -3.15644443e-01
1.58339992e-01 4.42602396e-01 -6.17985129e-01 3.85812193e-01
-1.59766048e-01 6.88223600e-01 -1.19833067e-01 3.42540115e-01
2.81888455e-01 4.46497679e-01 -1.56846434e-01 -7.03260303e-02
-4.38933879e-01 5.90455592e-01 -1.05628562e+00 1.14011705e+00
-8.80922675e-01 8.67245197e-01 -3.82027060e-01 -3.72680932e-01
1.27943671e+00 8.27546895e-01 -1.79629609e-01 -1.00610197e+00
2.16172665e-01 2.78827138e-02 6.04153395e-01 -7.81833410e-01
6.28799677e-01 1.04835533e-01 -2.81993654e-02 4.47204083e-01
-3.48841846e-01 -4.06164557e-01 1.89238153e-02 -8.43588784e-02
7.36218750e-01 1.46099761e-01 2.46111587e-01 -2.21610114e-01
6.36614740e-01 -7.41210461e-01 5.36274612e-01 5.66990912e-01
-1.54373467e-01 8.83195400e-01 2.96458662e-01 -5.33203930e-02
-1.19306171e+00 -9.48431611e-01 1.01549745e-01 1.20564032e+00
-1.78244695e-01 -2.17052341e-01 -9.03088272e-01 -1.62616923e-01
-4.55229700e-01 1.11026192e+00 -4.16550696e-01 -3.53413641e-01
-8.44138384e-01 -3.20125312e-01 5.92256606e-01 4.22139496e-01
2.38230944e-01 -1.68740559e+00 -3.98373455e-01 2.73570031e-01
-5.22222996e-01 -9.56242204e-01 -1.29069436e+00 -9.66750383e-02
-4.63119239e-01 -3.50005865e-01 -7.26593137e-01 -7.01586425e-01
3.07780653e-01 1.11647189e-01 6.83323264e-01 2.36704219e-02
3.92147228e-02 -2.30712250e-01 -5.10132909e-01 -3.85237038e-01
-6.35780871e-01 2.10204065e-01 4.13564980e-01 7.27231130e-02
-8.04172307e-02 -5.03142834e-01 -5.23999393e-01 3.45114231e-01
-6.63969219e-01 8.13977942e-02 5.12465119e-01 6.20634854e-01
2.28150994e-01 -2.99573869e-01 1.13593900e+00 -2.51500785e-01
1.14378572e+00 -1.83090910e-01 -4.41382021e-01 1.30305141e-01
-5.99705398e-01 2.07078874e-01 9.60136831e-01 -8.57841611e-01
-9.72947776e-01 -2.82650590e-01 -2.70525992e-01 -2.36433342e-01
6.20127097e-02 7.03071281e-02 -3.20046067e-01 4.66256201e-01
7.32177496e-01 4.19801742e-01 -1.60037056e-01 -7.92111605e-02
3.01369190e-01 1.20376217e+00 5.00204504e-01 -3.64574373e-01
6.50838077e-01 -3.56382668e-01 -5.15858769e-01 -1.13949156e+00
-2.92992920e-01 8.49306211e-02 -1.33499101e-01 -1.97993800e-01
9.47474539e-01 -7.17738092e-01 -7.79729068e-01 3.14521551e-01
-1.29256678e+00 -3.50210696e-01 3.58746760e-02 7.69223034e-01
-6.43237531e-01 1.72737911e-01 -5.50146759e-01 -1.15208566e+00
-6.74573243e-01 -1.28688371e+00 9.87241507e-01 2.41839975e-01
-5.60306907e-01 -4.49837893e-01 2.57063024e-02 1.60388514e-01
6.86669946e-01 -1.03550928e-03 7.28997886e-01 -1.65952548e-01
-2.61785299e-01 9.96866822e-02 1.72860011e-01 5.05107880e-01
4.38058019e-01 3.45651805e-01 -1.06606269e+00 -1.31275713e-01
-6.72674105e-02 1.57350264e-02 4.55995113e-01 4.75031614e-01
8.66641998e-01 -6.53477490e-01 2.69935936e-01 2.49694690e-01
8.14843535e-01 9.21815693e-01 9.84450877e-01 6.73109442e-02
5.14916718e-01 5.52186131e-01 4.83722687e-01 3.39641333e-01
8.84840935e-02 8.89137924e-01 -8.77563804e-02 1.80046454e-01
-5.74187577e-01 -2.39410907e-01 7.33901978e-01 1.63331461e+00
-1.21225035e-02 -7.19777584e-01 -5.72047174e-01 4.48081851e-01
-1.39465785e+00 -1.00411415e+00 -8.59067142e-02 2.32929921e+00
1.17697144e+00 4.53443050e-01 6.91824332e-02 4.36618805e-01
1.02833140e+00 1.04735531e-01 -3.37439418e-01 -9.26441550e-01
-1.42774001e-01 1.07416138e-01 1.39432997e-01 8.72067273e-01
-5.95105469e-01 1.09235334e+00 6.17924833e+00 9.14902151e-01
-1.49394834e+00 -7.87845924e-02 4.61645991e-01 -2.03999788e-01
-5.87294757e-01 -2.98646390e-01 -7.38021970e-01 7.46270716e-01
1.32796657e+00 -2.58483022e-01 5.97508729e-01 4.30210054e-01
8.96729708e-01 3.25216055e-01 -7.99526811e-01 8.56336474e-01
-9.70002450e-03 -8.64701807e-01 2.20059291e-01 -2.72910208e-01
6.39805436e-01 -4.43095624e-01 4.39705491e-01 3.13512683e-01
-1.49986401e-01 -1.13010633e+00 9.92539465e-01 5.17110407e-01
1.02721488e+00 -8.04005325e-01 4.12671387e-01 9.56140235e-02
-1.15531695e+00 2.45325528e-02 -1.71248183e-01 4.12796438e-02
1.11135229e-01 2.25800470e-01 -9.40309227e-01 2.09258199e-01
3.13689023e-01 5.11991866e-02 -1.19893201e-01 4.89815503e-01
-3.60651821e-01 9.42903578e-01 -9.49319527e-02 -3.95599365e-01
2.10201461e-02 1.02170780e-01 7.65879333e-01 1.35797679e+00
5.33030450e-01 5.09909764e-02 -2.78682262e-01 4.87083912e-01
1.27399862e-01 3.64458203e-01 -5.37542582e-01 -3.36873502e-01
8.86760354e-01 7.98793614e-01 -3.78114671e-01 -9.78648216e-02
-1.03374682e-01 7.82209694e-01 -5.10836579e-02 4.49365646e-01
-1.05930841e+00 -9.05205011e-01 4.65882123e-01 -1.11148775e-01
1.37023613e-01 -3.24754000e-01 -2.41441131e-01 -6.60295844e-01
-6.09227829e-02 -1.06881249e+00 -2.47786611e-01 -1.18057680e+00
-8.16125154e-01 1.19758749e+00 -1.65555179e-01 -1.33867395e+00
-3.95102382e-01 -1.38287380e-01 -7.96001613e-01 1.07091475e+00
-9.81874168e-01 -7.71323085e-01 -1.67157669e-02 2.82912612e-01
1.22295678e+00 -3.82973045e-01 6.67261600e-01 2.55022407e-01
-8.02753448e-01 1.08947742e+00 -3.50640982e-01 -2.16104805e-01
7.88288593e-01 -1.10130477e+00 7.42948234e-01 1.04264963e+00
7.04361126e-02 7.74459481e-01 9.60019708e-01 -6.14281416e-01
-1.09220493e+00 -9.27895546e-01 1.09582973e+00 -2.87271142e-01
2.94696212e-01 -3.10952961e-01 -7.18222558e-01 3.10946077e-01
5.98595083e-01 -4.11057979e-01 3.84914726e-01 -3.11685264e-01
-1.96080774e-01 -2.74019718e-01 -7.99227893e-01 1.17511976e+00
8.35135102e-01 -3.84246081e-01 -6.74814582e-01 6.26179017e-03
1.45854914e+00 -3.87130767e-01 -5.71421444e-01 2.74663955e-01
6.77233875e-01 -7.41662025e-01 5.18640876e-01 -3.02041501e-01
2.92209417e-01 -5.24705529e-01 -3.34042341e-01 -1.49570644e+00
-4.53152716e-01 -1.23144937e+00 2.03434959e-01 1.44885945e+00
7.19178021e-01 -5.00617325e-01 4.69785668e-02 4.89451401e-02
-6.45299494e-01 -3.02941769e-01 -6.34181261e-01 -9.09095347e-01
1.66004850e-03 -3.54982287e-01 6.91240668e-01 7.13647544e-01
-1.11819342e-01 5.64370275e-01 -7.04407454e-01 2.62243390e-01
-1.38457730e-01 -4.39427286e-01 6.26519978e-01 -5.26083112e-01
-2.80700624e-01 -4.91976619e-01 -9.86214448e-03 -8.18919539e-01
-8.43370110e-02 -4.85246956e-01 5.30530453e-01 -1.12151206e+00
-3.75064254e-01 -7.77993500e-02 -2.02945426e-01 2.57147789e-01
-4.10292983e-01 9.26677883e-02 4.03004438e-01 -6.51838705e-02
-4.95027155e-02 9.87112284e-01 1.48827875e+00 -2.89541576e-03
-7.21019566e-01 4.46796976e-02 -4.96034622e-01 3.10394228e-01
1.16542923e+00 -3.41918170e-01 -7.28313625e-01 -4.98797774e-01
-1.35891780e-01 3.53086799e-01 -3.82206470e-01 -1.08391666e+00
6.79051727e-02 -4.00841892e-01 1.20654739e-01 -3.17613959e-01
2.58294344e-01 -3.38567913e-01 3.56793664e-02 5.25800467e-01
-6.29382312e-01 5.47175586e-01 3.81005138e-01 3.01403612e-01
-1.48460697e-02 -1.35185644e-01 7.61223018e-01 1.21918261e-01
-1.59301579e-01 2.76519847e-03 -6.75187886e-01 1.97077870e-01
6.90770447e-01 -2.07071349e-01 -1.53095350e-01 -5.70447743e-01
-6.05976701e-01 -1.75532192e-01 -2.64892846e-01 8.83218467e-01
6.26297355e-01 -1.49427080e+00 -1.01400578e+00 3.44098598e-01
1.25617921e-01 -4.16355938e-01 3.11589062e-01 4.08253282e-01
-4.79059547e-01 5.94474673e-01 -2.54314244e-02 -4.41247970e-01
-1.24193203e+00 3.60229433e-01 1.70696631e-01 3.39264780e-01
-5.11627018e-01 7.41359353e-01 1.67652354e-01 -1.84618130e-01
4.50392008e-01 -5.84601641e-01 -3.85476291e-01 -2.45998904e-01
6.00432932e-01 3.01956683e-01 5.56835346e-02 -4.46432382e-01
-3.13252777e-01 7.20629632e-01 2.91142128e-02 -5.65726161e-01
6.73253000e-01 -4.11963642e-01 2.94804454e-01 5.83427012e-01
7.27236271e-01 6.34746909e-01 -1.23930275e+00 2.38745689e-01
-1.96012855e-01 -2.33017161e-01 1.41693335e-02 -8.93333316e-01
-8.52015197e-01 7.78775156e-01 5.73510528e-01 3.74311805e-01
9.82167304e-01 -5.04548669e-01 9.08168972e-01 1.55950010e-01
-2.25638777e-01 -1.21034718e+00 5.05551338e-01 6.84216440e-01
1.59949958e+00 -9.72218513e-01 -4.63065863e-01 3.99595052e-02
-1.01692128e+00 1.11597753e+00 9.67116237e-01 9.83475074e-02
3.29534799e-01 2.42804021e-01 2.44605690e-01 4.75182354e-01
-1.03359413e+00 -8.43414292e-02 2.84523398e-01 7.33339131e-01
7.22034454e-01 1.12033509e-01 -3.90175343e-01 3.46535355e-01
-8.64404321e-01 -4.57151085e-01 6.29012644e-01 2.99043953e-01
-7.18052924e-01 -8.04922402e-01 -5.72771430e-01 5.88154830e-02
-5.50531447e-01 -4.40703243e-01 -2.38338903e-01 5.36202371e-01
7.73826241e-02 1.45699728e+00 2.26649508e-01 -6.65581882e-01
4.40691084e-01 -3.24406661e-02 1.89371109e-01 -5.52567422e-01
-5.90130746e-01 4.59432632e-01 5.82612725e-03 -5.00479676e-02
-8.30275007e-03 -4.92885113e-01 -1.28730154e+00 -1.57122567e-01
-4.30097520e-01 2.44159639e-01 6.63074374e-01 8.18998098e-01
1.58409402e-01 1.13201618e+00 9.48680341e-01 -3.63712639e-01
-5.29003561e-01 -1.47562397e+00 -1.74458817e-01 1.36540279e-01
6.56838894e-01 -3.24924439e-01 -5.21450996e-01 1.06346078e-01] | [14.95374584197998, 6.629697799682617] |
dce898a0-f44f-4fa4-9446-d98ffb774f8a | three-branches-detecting-actions-with-richer | 1908.04519 | null | https://arxiv.org/abs/1908.04519v1 | https://arxiv.org/pdf/1908.04519v1.pdf | Three Branches: Detecting Actions With Richer Features | We present our three branch solutions for International Challenge on Activity Recognition at CVPR2019. This model seeks to fuse richer information of global video clip, short human attention and long-term human activity into a unified model. We have participated in two tasks: Task A, the Kinetics challenge and Task B, spatio-temporal action localization challenge. For Kinetics, we achieve 21.59% error rate. For the AVA challenge, our final model obtains 32.49% mAP on the test sets, which outperforms all submissions to the AVA challenge at CVPR 2018 for more than 10% mAP. As the future work, we will introduce human activity knowledge, which is a new dataset including key information of human activity. | ['Jiajun Tang', 'Cewu Lu', 'Jin Xia'] | 2019-08-13 | null | null | null | null | ['spatio-temporal-action-localization'] | ['computer-vision'] | [ 9.49022099e-02 -3.17192852e-01 -4.37354267e-01 1.53375477e-01
-9.87789452e-01 -5.15205562e-01 7.40569353e-01 -3.29118371e-01
-7.52979100e-01 7.13555872e-01 8.68978620e-01 4.27817583e-01
2.46328026e-01 -1.05744578e-01 -6.87708020e-01 -6.59689307e-01
-4.45994020e-01 -4.22928818e-02 4.08775628e-01 1.11565232e-01
9.26276147e-02 2.30574787e-01 -1.23850131e+00 7.68380821e-01
4.27046657e-01 1.19076657e+00 -6.16669804e-02 9.51285005e-01
5.65768480e-01 1.51305342e+00 -7.70154655e-01 9.54340585e-03
1.78111687e-01 -5.20078599e-01 -1.08775818e+00 -2.41509184e-01
6.29524469e-01 -9.53290910e-02 -9.96486545e-01 5.60550630e-01
5.58486879e-01 4.94675100e-01 5.66655397e-01 -1.47793567e+00
-4.68719780e-01 4.46248531e-01 -5.72773516e-01 1.22161329e+00
9.11881208e-01 6.39974773e-01 8.75804484e-01 -8.54433656e-01
7.89131582e-01 7.26674318e-01 7.28891075e-01 5.69746733e-01
-6.44765854e-01 -4.04814571e-01 3.42667460e-01 8.41008008e-01
-1.52103674e+00 -4.95330721e-01 3.70713413e-01 -6.88157678e-01
1.60106575e+00 2.60373056e-01 1.02669680e+00 1.95287383e+00
7.58667216e-02 1.57909715e+00 6.90572023e-01 1.59684226e-01
4.10923548e-02 -5.21420717e-01 5.33768535e-02 5.10703444e-01
-2.16093913e-01 -2.29006141e-01 -1.18751216e+00 1.08847663e-01
7.07243562e-01 -1.91733122e-01 -4.56340760e-01 5.04812747e-02
-1.68889701e+00 3.17784011e-01 3.10734034e-01 4.06455159e-01
-6.17583573e-01 4.01775181e-01 5.57631910e-01 -3.85403149e-02
5.56850612e-01 5.82065582e-01 -5.05615532e-01 -1.18409848e+00
-7.62331784e-01 3.59865457e-01 7.33224094e-01 8.10666561e-01
-1.29161999e-02 -6.73217326e-02 -1.01813889e+00 7.14134812e-01
1.55624449e-01 5.43390930e-01 6.23646557e-01 -1.29973733e+00
7.80471623e-01 3.23913872e-01 2.60664523e-01 -8.73793602e-01
-3.40766221e-01 -2.65685618e-01 -7.15832710e-01 -4.01967943e-01
4.97871339e-01 -1.18297152e-01 -8.01527500e-01 1.70348203e+00
-5.78448409e-03 9.91521537e-01 -8.68611876e-03 8.76986563e-01
8.06307614e-01 8.38947713e-01 3.46069276e-01 -3.33458245e-01
1.36766708e+00 -1.82480276e+00 -8.03041577e-01 -2.71406114e-01
7.52385139e-01 -7.05345869e-02 9.66479838e-01 5.54419637e-01
-1.23186362e+00 -6.87279761e-01 -1.06714523e+00 1.81462523e-02
-3.03516746e-01 2.14842752e-01 5.68586349e-01 1.03595316e-01
-1.01909804e+00 4.38126445e-01 -1.17091000e+00 -6.31450295e-01
8.12248409e-01 -1.35868311e-01 -5.56752026e-01 3.21166873e-01
-1.16803241e+00 8.35568964e-01 1.33681759e-01 1.18963800e-01
-1.38692987e+00 -9.55314994e-01 -8.34393919e-01 -1.22326687e-01
4.76291358e-01 -5.30584812e-01 1.35383284e+00 -5.95567584e-01
-1.29106498e+00 1.02526224e+00 -3.37365896e-01 -9.38254237e-01
9.02278781e-01 -8.08189213e-01 -5.57675183e-01 2.08840176e-01
2.05690354e-01 4.62261289e-01 1.99411303e-01 -2.38666669e-01
-7.88478315e-01 -2.26864696e-01 -6.65996298e-02 3.66483927e-01
-9.08153951e-02 2.36398295e-01 -9.16457415e-01 -7.29692042e-01
-5.33528566e-01 -9.57110345e-01 2.57738531e-02 -3.67322385e-01
1.34714603e-01 -6.55592322e-01 7.14789152e-01 -9.24161136e-01
1.47921205e+00 -2.00839996e+00 3.55675995e-01 -2.47792482e-01
3.13333124e-01 4.28437710e-01 -2.50210166e-01 3.51499647e-01
2.67785583e-02 3.04614031e-03 8.72763395e-02 -3.61698955e-01
-4.19005454e-02 -6.05940856e-02 -2.56989568e-01 5.77383339e-01
1.30046885e-02 1.49346185e+00 -1.08430517e+00 -4.02940691e-01
1.11667529e-01 3.38081568e-01 -1.92684636e-01 2.49680713e-01
-2.77266074e-02 7.27557182e-01 -3.06845427e-01 6.32906795e-01
1.65106580e-01 -5.88868320e-01 -1.30754769e-01 -1.87327653e-01
-1.44758904e-02 1.98897645e-01 -7.59217978e-01 2.30745435e+00
3.03512234e-02 9.51370955e-01 -4.35266912e-01 -8.55239332e-01
2.26232260e-01 4.69634593e-01 1.20567071e+00 -9.82225478e-01
-4.84566092e-02 -4.37785149e-01 -2.86399454e-01 -7.86845386e-01
2.60633051e-01 8.29996407e-01 -2.06501395e-01 4.29789215e-01
4.54510272e-01 7.13250220e-01 5.94519436e-01 3.83864254e-01
1.79569125e+00 5.67431629e-01 4.83173132e-01 -9.14110802e-03
6.15078032e-01 -1.81844279e-01 6.49350524e-01 1.07268500e+00
-1.07079899e+00 8.19941521e-01 5.35630584e-01 -6.12288892e-01
-6.81970477e-01 -1.08121681e+00 3.93829614e-01 1.30525529e+00
1.07713923e-01 -8.31450224e-01 -7.75119364e-01 -1.02222443e+00
-5.08439362e-01 6.84537366e-02 -1.09093797e+00 -6.41381592e-02
-9.24021840e-01 -7.72780418e-01 1.10430789e+00 1.06379104e+00
1.05994010e+00 -1.42988026e+00 -4.99566019e-01 -7.14766383e-02
-8.71138990e-01 -1.56665432e+00 -8.75520349e-01 -3.14172238e-01
-4.45725948e-01 -1.41636026e+00 -9.22738194e-01 -4.49059874e-01
-2.25120589e-01 1.77644596e-01 1.09134090e+00 -3.45778435e-01
-4.27632123e-01 7.72226572e-01 -4.96156186e-01 -2.33249098e-01
3.24094445e-01 1.78252727e-01 5.98950945e-02 1.93491414e-01
5.88380694e-01 -4.86950815e-01 -8.48611891e-01 4.75833863e-01
-4.81940657e-02 4.11632508e-02 4.22463238e-01 3.37034911e-01
6.99814558e-01 -5.86799145e-01 2.92642444e-01 -3.94245803e-01
3.79672915e-01 -7.28230953e-01 -1.12914637e-01 3.82228881e-01
-1.56453341e-01 -2.55858690e-01 -3.84763852e-02 -5.69494963e-01
-1.08377326e+00 1.43741131e-01 -1.31271839e-01 -5.11238575e-01
-2.29533583e-01 4.10928071e-01 -1.92746714e-01 1.87803432e-01
9.73055363e-01 5.37500083e-01 -4.79286492e-01 -4.79806095e-01
1.39244720e-01 2.70690709e-01 1.03599763e+00 -3.01987797e-01
1.31740138e-01 4.58714396e-01 -3.08177829e-01 -7.89589822e-01
-1.14126837e+00 -1.04005790e+00 -8.37279737e-01 -5.60350060e-01
1.42440546e+00 -1.42544842e+00 -7.45358229e-01 9.50608432e-01
-1.18179822e+00 -7.81723619e-01 -2.40000144e-01 5.20057619e-01
-8.81775081e-01 3.01127553e-01 -5.87182164e-01 -6.49295747e-01
-3.21383715e-01 -6.02431834e-01 9.89645422e-01 1.83732972e-01
-5.34338892e-01 -8.90811443e-01 6.87295735e-01 9.94888425e-01
2.22929358e-01 4.64884818e-01 -4.54337358e-01 -7.55296767e-01
-4.70594138e-01 -2.94288278e-01 1.24348151e-02 2.27745861e-01
-2.00766072e-01 -2.44631842e-01 -1.00855935e+00 -1.21494606e-01
-2.91901797e-01 -5.69771528e-01 1.12843645e+00 4.11410779e-01
1.35855806e+00 -1.57837331e-01 -4.99331713e-01 6.24924958e-01
6.40260994e-01 2.03997582e-01 1.10873616e+00 1.87127694e-01
8.25682998e-01 -2.15762779e-02 8.60217631e-01 5.30472457e-01
4.24204588e-01 1.05586600e+00 5.64006492e-02 3.00875813e-01
-3.80929977e-01 -3.77312213e-01 6.99600816e-01 7.00493515e-01
-1.03162360e+00 -4.89294440e-01 -1.14713478e+00 7.31014609e-01
-2.48876071e+00 -1.65493345e+00 -2.43779182e-01 1.66989326e+00
5.07169425e-01 -1.44532099e-01 6.30340874e-01 -2.60368496e-01
2.36004993e-01 7.31769979e-01 -3.93979192e-01 2.89581716e-01
-3.09540838e-01 -8.96351337e-02 4.47275549e-01 2.32094511e-01
-1.76293576e+00 9.85829890e-01 7.36569357e+00 9.80354369e-01
-5.56848407e-01 6.04612291e-01 8.89615789e-02 -5.84914327e-01
6.28915787e-01 -3.61759275e-01 -7.02719748e-01 7.37552047e-01
1.26446140e+00 -2.09310979e-01 3.58985096e-01 7.11145222e-01
4.30994868e-01 -3.31550598e-01 -1.06350625e+00 1.40889275e+00
5.45453250e-01 -1.38432527e+00 -2.14574561e-01 5.64686246e-02
9.53072190e-01 5.79578996e-01 -2.70420939e-01 7.26228952e-01
2.23128960e-01 -1.29105878e+00 5.52077651e-01 9.66857612e-01
3.67426723e-01 -3.51890862e-01 7.67842412e-01 4.42372888e-01
-1.52461195e+00 -6.37779608e-02 1.22253209e-01 -2.38131776e-01
4.98158246e-01 -1.88338459e-01 -9.44892764e-02 3.06638956e-01
1.11167943e+00 1.56670499e+00 -9.20209825e-01 1.21272409e+00
-3.70838046e-01 7.75502026e-01 -6.46396540e-04 7.78328478e-02
2.89120734e-01 3.07576090e-01 5.07115364e-01 1.50798416e+00
5.43994531e-02 1.69134215e-01 3.24748397e-01 2.96221465e-01
-2.51769543e-01 -2.36213267e-01 -4.89807755e-01 -3.96732002e-01
1.09909885e-01 1.03068697e+00 -2.87754238e-01 -5.35098255e-01
-3.78474921e-01 1.32998204e+00 5.30534267e-01 5.69200993e-01
-1.43120158e+00 -1.20890448e-02 7.66109943e-01 -4.08599377e-02
5.75143211e-02 -2.01112613e-01 1.01055264e-01 -1.54286695e+00
1.30920663e-01 -8.45443010e-01 7.57624388e-01 -9.80552614e-01
-9.85903502e-01 2.90931404e-01 1.40787840e-01 -1.06113207e+00
-2.36992404e-01 -3.69121432e-01 -6.42542422e-01 5.19069135e-01
-7.38904178e-01 -1.26753557e+00 -7.01577723e-01 8.59214127e-01
8.10289562e-01 -3.95901859e-01 5.44758737e-01 5.70218325e-01
-7.16549754e-01 6.12637162e-01 -4.11199570e-01 4.89054531e-01
7.68403471e-01 -1.12150395e+00 6.97076380e-01 9.94788766e-01
5.60396969e-01 1.56247675e-01 4.43282247e-01 -6.12180948e-01
-1.13241017e+00 -1.13321507e+00 9.68959272e-01 -1.31298947e+00
7.98110247e-01 -4.04120624e-01 -7.27223814e-01 1.15865970e+00
3.51101160e-01 3.12706113e-01 6.53798640e-01 9.96299684e-02
-1.92505613e-01 2.89263517e-01 -5.02919018e-01 4.09467041e-01
1.84402800e+00 -6.48246169e-01 -5.59555531e-01 4.12762910e-01
5.17558396e-01 -3.85061353e-01 -1.10956144e+00 5.21829069e-01
8.49619269e-01 -5.84675908e-01 1.06293404e+00 -1.10657072e+00
2.93500572e-01 -3.68911177e-01 -4.49149944e-02 -1.02881861e+00
-5.60005069e-01 -6.81419551e-01 -1.07222939e+00 6.48225665e-01
2.53194779e-01 2.87214341e-03 9.48930502e-01 1.78736985e-01
-2.33637020e-01 -6.45302415e-01 -9.76507783e-01 -1.02905560e+00
-3.35569680e-01 -6.24837041e-01 -6.78039864e-02 8.08623552e-01
1.46454066e-01 3.60594600e-01 -9.98111725e-01 -1.30908042e-01
2.14224353e-01 -4.74235684e-01 1.11309922e+00 -7.65131652e-01
-3.93297255e-01 -4.83533591e-01 -7.03207314e-01 -1.35343218e+00
8.92689377e-02 -6.26629949e-01 2.09961366e-03 -1.53158379e+00
7.16027260e-01 6.61581039e-01 -7.04799831e-01 6.42575920e-01
-2.02182624e-02 6.80907369e-01 1.98616460e-01 2.00548425e-01
-1.68881989e+00 5.46964824e-01 9.88747835e-01 -3.67787212e-01
-1.58402845e-01 -1.33077994e-01 -4.78739321e-01 6.65530980e-01
6.60273671e-01 -2.02183113e-01 -4.09719914e-01 -3.93827885e-01
3.30870263e-02 -1.33486792e-01 5.63582897e-01 -1.42935252e+00
4.51457024e-01 -4.68881816e-01 4.61486727e-01 -6.50262773e-01
5.08923054e-01 -1.61644861e-01 6.21472895e-02 2.93470621e-01
-4.35558617e-01 -1.34122968e-01 1.14688084e-01 9.02701259e-01
-3.34221870e-01 4.92940098e-01 5.00543594e-01 -7.36591369e-02
-1.18572307e+00 7.33334184e-01 -6.15023017e-01 5.44498205e-01
1.37569177e+00 -1.86446577e-01 -6.07403100e-01 -4.27237093e-01
-1.28904414e+00 4.95588690e-01 -9.39146578e-02 9.51048017e-01
3.08256954e-01 -1.58129632e+00 -8.71747911e-01 -3.14452797e-01
5.16281188e-01 -6.52239203e-01 7.63296247e-01 1.46617746e+00
-4.48612332e-01 6.86954319e-01 -3.47927183e-01 -5.99169791e-01
-1.41243589e+00 3.95719737e-01 4.45709527e-01 -5.70864677e-01
-5.95201373e-01 8.68824542e-01 2.66930480e-02 2.46974215e-01
4.45594519e-01 -2.02680826e-02 -5.44046938e-01 1.13106698e-01
1.15632308e+00 8.12261879e-01 -1.91997245e-01 -8.52106988e-01
-7.01862454e-01 3.13103914e-01 1.77701086e-01 -1.20275572e-01
1.22530258e+00 -4.04020697e-02 4.26986277e-01 5.64662993e-01
1.12724090e+00 -1.98993206e-01 -1.65527737e+00 -1.27080724e-01
8.94420743e-02 -5.17576456e-01 -2.82393694e-01 -1.10237694e+00
-8.83874059e-01 9.18055236e-01 6.54463589e-01 -1.34363323e-01
8.36411953e-01 3.13349873e-01 7.95399487e-01 4.37095076e-01
2.00683579e-01 -1.41107559e+00 6.08234942e-01 7.86938071e-01
1.29595280e+00 -1.25586116e+00 -8.49740133e-02 8.29126537e-02
-9.97905314e-01 4.24770176e-01 1.02000320e+00 -1.28638700e-01
4.48646128e-01 4.30434160e-02 -1.21798329e-01 -2.91453928e-01
-1.05882359e+00 -3.98346096e-01 6.81021452e-01 7.39953816e-01
2.08275735e-01 -6.55973032e-02 -2.92438567e-01 9.36781466e-01
2.70501435e-01 2.49946132e-01 1.77046895e-01 8.39069664e-01
-2.22946346e-01 -3.75940293e-01 2.75464296e-01 1.97598979e-01
-5.52275658e-01 2.81568825e-01 -6.53731048e-01 6.49854720e-01
1.07959345e-01 7.92982161e-01 8.72713402e-02 -4.75539237e-01
5.58240414e-01 2.36058220e-01 6.14412963e-01 -3.21174204e-01
-4.76071119e-01 -2.40398616e-01 3.87389064e-01 -1.57431078e+00
-7.45880187e-01 -9.26670492e-01 -9.19671893e-01 -3.27049971e-01
4.60727870e-01 -5.62390685e-03 2.20995173e-01 1.03158903e+00
5.98775387e-01 7.15063214e-01 -1.83222778e-02 -8.48560154e-01
1.79488640e-02 -1.33209312e+00 -3.93411696e-01 5.15979528e-01
3.27944383e-02 -7.03980327e-01 -3.03197473e-01 3.35736156e-01] | [8.321216583251953, 0.4723582863807678] |
4762f01a-a16c-467f-85a1-920f67917965 | learning-disentangling-and-fusing-networks | 1712.04646 | null | http://arxiv.org/abs/1712.04646v1 | http://arxiv.org/pdf/1712.04646v1.pdf | Learning Disentangling and Fusing Networks for Face Completion Under Structured Occlusions | Face completion aims to generate semantically new pixels for missing facial
components. It is a challenging generative task due to large variations of face
appearance. This paper studies generative face completion under structured
occlusions. We treat the face completion and corruption as disentangling and
fusing processes of clean faces and occlusions, and propose a jointly
disentangling and fusing Generative Adversarial Network (DF-GAN). First, three
domains are constructed, corresponding to the distributions of occluded faces,
clean faces and structured occlusions. The disentangling and fusing processes
are formulated as the transformations between the three domains. Then the
disentangling and fusing networks are built to learn the transformations from
unpaired data, where the encoder-decoder structure is adopted and allows DF-GAN
to simulate structure occlusions by modifying the latent representations.
Finally, the disentangling and fusing processes are unified into a dual
learning framework along with an adversarial strategy. The proposed method is
evaluated on Meshface verification problem. Experimental results on four
Meshface databases demonstrate the effectiveness of our proposed method for the
face completion under structured occlusions. | ['Ran He', 'Yibo Hu', 'Zhihang Li'] | 2017-12-13 | null | null | null | null | ['facial-inpainting'] | ['computer-vision'] | [ 3.69703323e-01 2.44099304e-01 3.79986018e-01 -4.89633113e-01
-7.84489036e-01 -5.30314922e-01 7.80125022e-01 -8.84134710e-01
1.94723368e-01 7.51125813e-01 2.99016148e-01 1.63671657e-01
1.21504582e-01 -8.14268947e-01 -1.00827241e+00 -1.17672503e+00
3.66476834e-01 4.29196060e-01 -6.94555581e-01 2.79951058e-02
-4.06620920e-01 4.61177707e-01 -1.48575628e+00 2.36531645e-01
9.91385937e-01 6.70372427e-01 -1.66117370e-01 5.32712877e-01
-1.15036294e-02 6.17711544e-01 -6.72460198e-01 -6.16818249e-01
4.56557572e-01 -4.63378161e-01 -2.52596557e-01 5.26138544e-01
5.98215461e-01 -5.91676235e-01 -6.09335184e-01 1.20521700e+00
4.49225187e-01 -9.95839015e-02 7.59003580e-01 -1.60896409e+00
-1.11872768e+00 2.69520879e-01 -7.39949405e-01 -4.17574018e-01
3.40211630e-01 2.60134667e-01 6.69083655e-01 -1.05913067e+00
5.00204384e-01 1.85445583e+00 2.96137959e-01 8.86898696e-01
-1.23193574e+00 -1.12804723e+00 2.61162251e-01 -1.92986637e-01
-1.47234225e+00 -7.18012631e-01 9.76803482e-01 -6.09341025e-01
1.09329380e-01 1.97208911e-01 4.00754303e-01 1.26701963e+00
2.75255125e-02 7.18197405e-01 1.07513559e+00 -1.19792126e-01
6.29776791e-02 -8.58784770e-04 -3.36830407e-01 8.03975582e-01
3.28097165e-01 2.48055339e-01 -1.74385473e-01 -2.26174071e-01
1.06976116e+00 2.64006495e-01 -3.90233755e-01 -2.79799461e-01
-8.02640259e-01 7.91768551e-01 2.28640839e-01 -2.90126204e-01
-3.74169528e-01 4.46390696e-02 3.10871936e-02 2.93745816e-01
5.35374582e-01 -2.44397327e-01 1.31585393e-02 6.37658060e-01
-7.89792240e-01 4.56083119e-01 6.56420112e-01 1.38168752e+00
7.99681425e-01 5.16119063e-01 -4.08956409e-01 9.17497516e-01
6.44162118e-01 8.97528529e-01 5.90033978e-02 -8.69462550e-01
6.15599334e-01 3.81588042e-01 -3.48994657e-02 -8.89777243e-01
2.79282629e-01 -4.66922402e-01 -1.14519310e+00 5.10743678e-01
1.10756271e-01 -4.36920822e-01 -1.25156987e+00 2.18929648e+00
5.02955616e-01 4.78838801e-01 1.81767181e-01 6.54220641e-01
1.03089833e+00 7.08682895e-01 -5.05257808e-02 -2.63959020e-01
1.38738751e+00 -8.97358716e-01 -1.15147352e+00 -7.71075860e-02
-1.35122672e-01 -7.43138611e-01 6.20639682e-01 2.60576427e-01
-1.26923215e+00 -7.31282830e-01 -1.11782384e+00 -2.34133258e-01
1.49863213e-01 3.49177122e-01 3.76062214e-01 5.54209352e-01
-8.68670166e-01 1.83532834e-01 -8.06980312e-01 2.99102783e-01
9.09247756e-01 5.65836132e-01 -8.20603549e-01 -4.03301626e-01
-1.00190258e+00 4.82839555e-01 -5.68818450e-02 4.95973408e-01
-1.46975756e+00 -7.58885682e-01 -1.12355173e+00 9.79986861e-02
2.38258690e-01 -9.99568284e-01 8.21295679e-01 -1.10259616e+00
-1.48454523e+00 9.43387210e-01 -2.35917270e-01 1.05409540e-01
6.81387544e-01 -1.12840220e-01 -3.75407100e-01 -2.84456462e-01
1.09624222e-01 4.72228616e-01 1.40663755e+00 -1.61892509e+00
-2.76815176e-01 -5.68084717e-01 -2.03599080e-01 1.92726105e-01
-2.14184858e-02 -1.84812441e-01 -5.36731184e-01 -8.74750912e-01
-7.10043998e-04 -7.86830127e-01 6.89721853e-02 1.51163831e-01
-5.53370059e-01 1.64538696e-01 1.04955685e+00 -1.10253620e+00
8.80003095e-01 -2.26596141e+00 7.07395673e-01 1.46230400e-01
3.82091016e-01 1.39386013e-01 -5.90565622e-01 1.09611027e-01
-3.79229158e-01 -8.63614306e-02 -4.23587173e-01 -9.07596648e-01
4.44212742e-02 4.73997623e-01 -3.56666833e-01 5.81419408e-01
3.78448606e-01 8.79043698e-01 -6.73612475e-01 -2.54765302e-01
1.11927100e-01 9.64613795e-01 -5.89523733e-01 6.35994613e-01
-8.89262557e-02 8.46318483e-01 -3.80016387e-01 6.90798640e-01
1.32993567e+00 3.61036062e-01 3.11908592e-02 -2.96185642e-01
3.56645137e-01 -2.49746054e-01 -1.24114370e+00 1.70853543e+00
-3.57266277e-01 5.92765212e-02 4.54982758e-01 -8.01199555e-01
1.03038585e+00 5.16915619e-01 9.90009606e-02 -2.10731208e-01
2.75265336e-01 1.66470394e-01 -4.09208946e-02 -4.76166248e-01
-8.98876712e-02 -4.02709484e-01 9.25942808e-02 2.94219881e-01
4.49612945e-01 -2.18894988e-01 -1.56178653e-01 2.76748575e-02
7.24926353e-01 3.87452245e-01 -8.74127001e-02 -7.71326721e-02
7.70246685e-01 -8.92319083e-01 1.00225675e+00 9.50454101e-02
1.02076434e-01 7.52058268e-01 7.01142013e-01 -3.36029589e-01
-9.81599927e-01 -1.31704259e+00 1.13149926e-01 5.01601100e-01
-8.61879960e-02 -2.13788468e-02 -9.67251778e-01 -8.94923747e-01
1.11874603e-02 4.55958843e-01 -9.23617899e-01 -3.52641433e-01
-5.89790881e-01 -5.82664430e-01 4.78629321e-01 3.81602198e-01
7.01659501e-01 -9.28582191e-01 4.44282502e-01 -1.47002935e-01
-1.31840900e-01 -1.06684577e+00 -7.61706412e-01 -2.72270769e-01
-5.27661562e-01 -1.03997838e+00 -8.67558777e-01 -9.73320663e-01
1.05142725e+00 7.61364028e-02 8.28638375e-01 6.88701496e-02
-3.04580986e-01 5.94027676e-02 1.54659599e-01 -1.77753568e-01
-5.71845353e-01 -4.18550670e-01 3.46486084e-02 6.47020042e-01
-1.48344949e-01 -1.04231775e+00 -6.15175903e-01 2.09094271e-01
-1.08179295e+00 2.33980432e-01 5.05479455e-01 1.13401031e+00
6.03490114e-01 7.60360882e-02 4.80815023e-01 -1.08452320e+00
4.03432816e-01 -5.68938434e-01 -6.86755657e-01 3.18762392e-01
-1.89320922e-01 1.25643596e-01 5.30370772e-01 -4.66720462e-01
-1.48692679e+00 1.99202895e-01 -2.75728792e-01 -9.48800325e-01
-1.04429260e-01 -8.59957412e-02 -1.29699969e+00 -6.74891397e-02
1.77549720e-01 2.61867315e-01 1.18155077e-01 -3.79079819e-01
6.41911685e-01 3.98667932e-01 7.11451888e-01 -8.47488403e-01
1.30798888e+00 5.68203330e-01 6.86464906e-02 -4.81476754e-01
-5.14840484e-01 3.67359042e-01 -5.10224640e-01 -1.17967591e-01
1.04556072e+00 -1.14269304e+00 -5.43270826e-01 8.99194419e-01
-1.19042885e+00 1.59066558e-01 -3.47853869e-01 8.67828578e-02
-5.38502455e-01 3.44372123e-01 -5.24458706e-01 -6.79965913e-01
-2.27346942e-01 -1.52578723e+00 1.27401316e+00 3.18756908e-01
3.00925821e-01 -9.14154828e-01 -5.24235740e-02 4.44210529e-01
-1.96675286e-02 8.78495574e-01 7.51016557e-01 -2.79614389e-01
-7.32109189e-01 -2.18882263e-01 -8.19780082e-02 7.46287704e-01
5.92263877e-01 -2.25527212e-02 -1.14298236e+00 -4.82253134e-01
3.70867968e-01 -2.54296154e-01 7.00057983e-01 1.30669266e-01
1.16080296e+00 -6.13944590e-01 -2.22442299e-01 1.19050062e+00
1.18377531e+00 3.81681800e-01 8.08036268e-01 -4.81243074e-01
1.03754902e+00 6.62001491e-01 1.34914488e-01 3.42874795e-01
1.29052728e-01 4.55381721e-01 5.55390298e-01 -2.66596377e-01
-3.32662404e-01 -6.45233035e-01 2.31651500e-01 6.86347544e-01
-4.84714583e-02 -2.56601632e-01 -4.57920402e-01 2.34334618e-01
-1.57514203e+00 -9.29982364e-01 2.33791739e-01 2.01048994e+00
7.34419584e-01 -4.03510123e-01 -3.58053446e-01 6.70037046e-02
9.47784960e-01 3.02951485e-01 -7.16322541e-01 7.21853077e-02
-2.03422174e-01 4.85862195e-01 -3.30429785e-02 8.20452332e-01
-8.85332465e-01 7.52038777e-01 5.13565540e+00 8.88069034e-01
-8.13549280e-01 3.48530561e-01 7.17983842e-01 5.10632060e-02
-6.52948618e-01 3.02687157e-02 -7.33986974e-01 5.47320127e-01
1.42917499e-01 2.36558154e-01 6.37458265e-01 5.25112867e-01
-3.01468540e-02 5.00540078e-01 -1.19087565e+00 1.10430014e+00
3.91493291e-01 -1.03672075e+00 4.07541543e-01 1.58889830e-01
1.02021170e+00 -6.15126014e-01 3.84705216e-01 1.93348125e-01
5.53277552e-01 -1.28028440e+00 7.17287660e-01 6.90081358e-01
1.17155480e+00 -7.36166835e-01 4.03893173e-01 1.78438187e-01
-1.14357996e+00 4.66076061e-02 -5.92055954e-02 3.02575171e-01
3.21415007e-01 4.83363986e-01 -2.38695398e-01 8.57718289e-01
1.33801416e-01 7.50347376e-01 -1.62242249e-01 4.13292408e-01
-6.72266304e-01 2.24138960e-01 4.21826020e-02 8.19616795e-01
-3.76307517e-01 -5.96900463e-01 5.83091795e-01 6.42142892e-01
3.41803670e-01 1.62852332e-01 1.58127341e-02 1.21238816e+00
-5.09837568e-01 -3.52759659e-01 -6.16634846e-01 9.23550725e-02
4.75409955e-01 1.30320966e+00 -1.67860195e-01 -3.12511444e-01
-2.55662590e-01 1.30801475e+00 3.48423481e-01 7.07760453e-01
-1.07661796e+00 -1.73985735e-01 9.18331623e-01 -6.30020052e-02
3.27229165e-02 3.97043787e-02 -2.69691437e-01 -1.52845550e+00
1.85351253e-01 -1.01147938e+00 1.00374341e-01 -7.25087106e-01
-1.36569095e+00 7.74353027e-01 5.01502119e-02 -1.08689415e+00
9.55683086e-03 -3.63355070e-01 -9.36339736e-01 1.31765378e+00
-1.54082739e+00 -1.66575968e+00 -5.59038401e-01 8.19431365e-01
5.18308878e-01 -3.37176979e-01 6.70307577e-01 5.01221776e-01
-8.02520752e-01 8.42235804e-01 -7.73570761e-02 2.72031605e-01
5.31093478e-01 -9.28520322e-01 2.23062009e-01 9.65242267e-01
-1.06627144e-01 6.30540013e-01 4.31039184e-01 -6.71857059e-01
-1.43653393e+00 -1.42159235e+00 3.98559570e-01 -2.35123351e-01
4.17722166e-02 -8.91104519e-01 -9.25353467e-01 1.01542449e+00
4.04158950e-01 3.88954669e-01 5.34882307e-01 -3.46418083e-01
-7.17283964e-01 -2.99201310e-01 -1.36535227e+00 5.02249002e-01
1.19532108e+00 -6.29403889e-01 -4.45945978e-01 3.44451427e-01
8.17503393e-01 -5.08700013e-01 -6.14287794e-01 4.72674698e-01
5.95227897e-01 -6.16661847e-01 1.02147985e+00 -5.34563422e-01
6.51987791e-01 -5.31367838e-01 -2.02796474e-01 -1.27653420e+00
-2.44080573e-01 -8.75275075e-01 -1.01039894e-01 1.84761369e+00
4.85548042e-02 -7.26862371e-01 5.82262695e-01 3.72952044e-01
-1.16573282e-01 -5.63824594e-01 -9.26233470e-01 -5.78218162e-01
2.85585791e-01 1.53328627e-01 1.07427180e+00 8.56876910e-01
-8.18546116e-01 3.73098552e-01 -7.81452656e-01 3.24206650e-01
9.11789656e-01 -1.29504651e-01 9.16784763e-01 -9.63824332e-01
-4.03639674e-01 -9.66057461e-03 -9.56934392e-02 -8.36502731e-01
5.86926639e-01 -9.26709473e-01 5.11370189e-02 -1.24136329e+00
2.67986000e-01 -3.12491804e-01 1.22755617e-01 3.89647305e-01
-3.30165476e-01 1.22943260e-01 1.03502475e-01 6.38827682e-02
1.81279004e-01 1.15371716e+00 1.66471303e+00 -3.86851192e-01
4.53766547e-02 -1.26924828e-01 -7.72112131e-01 6.21294379e-01
3.25772315e-01 -2.71453589e-01 -7.76262760e-01 -6.97209299e-01
-1.53204471e-01 3.82383019e-01 4.37213033e-01 -8.28742206e-01
-9.96308923e-02 1.25177866e-02 5.75118005e-01 -2.52716422e-01
6.03064179e-01 -8.89037132e-01 6.62493944e-01 3.43559444e-01
-1.34476036e-01 -9.93261188e-02 9.29050967e-02 6.92516804e-01
-2.95484781e-01 2.23999754e-01 9.17036235e-01 5.91298973e-04
1.02183491e-01 8.24670732e-01 1.60967201e-01 4.98499610e-02
1.22827888e+00 -1.44500092e-01 -5.55419959e-02 -2.60974318e-01
-1.14906895e+00 1.57338023e-01 2.44683981e-01 5.23345530e-01
8.13121855e-01 -1.76494455e+00 -1.08842647e+00 1.03759885e+00
-2.42920622e-01 3.81919593e-01 6.57770276e-01 3.42525125e-01
-2.53003299e-01 -2.06702471e-01 -3.67358029e-01 -3.85265142e-01
-1.37635660e+00 6.42386436e-01 4.39159125e-01 -2.69293875e-01
-2.44454503e-01 1.01793277e+00 9.44028080e-01 -5.11038244e-01
1.84519619e-01 -9.65896100e-02 -1.09935746e-01 -9.14463773e-02
4.18892950e-01 2.04035729e-01 -1.82436988e-01 -8.32016349e-01
-5.37279248e-02 6.90452754e-01 -1.28719211e-01 -1.34994775e-01
1.23387015e+00 -9.36186612e-02 -3.66986245e-01 -1.06600672e-01
1.26836765e+00 1.51430815e-01 -1.60887241e+00 -1.01088904e-01
-9.11010742e-01 -6.51767135e-01 -3.24632883e-01 -4.10517633e-01
-1.56540418e+00 1.15580571e+00 5.28405666e-01 -4.34039593e-01
1.40126598e+00 -2.18005538e-01 6.54781699e-01 -4.88810420e-01
1.64442971e-01 -3.52479428e-01 7.12954924e-02 2.50205696e-01
1.32475090e+00 -9.42588389e-01 -2.79964507e-01 -8.23524714e-01
-3.30686957e-01 7.66967595e-01 8.62594366e-01 -4.65253919e-01
8.21968257e-01 4.85153854e-01 -3.26800019e-01 -1.14437468e-01
-6.82373166e-01 2.49664173e-01 2.96938449e-01 7.67163754e-01
7.66228214e-02 1.24616526e-01 -1.17541835e-01 9.09351170e-01
-8.82825926e-02 -2.18878444e-02 1.49622202e-01 5.50055504e-01
2.56576777e-01 -1.38713872e+00 -5.53531587e-01 1.10839896e-01
-3.51097673e-01 -1.48440683e-02 -1.71242297e-01 5.94202697e-01
8.10294151e-01 7.98262358e-01 8.69175885e-03 -4.23515707e-01
2.52339900e-01 1.52980257e-02 6.71692252e-01 -6.60887957e-01
-4.17514056e-01 1.32935509e-01 -2.61223733e-01 -3.77045959e-01
-1.73752561e-01 -6.49494648e-01 -8.81303906e-01 -2.34267682e-01
-3.31526995e-01 1.48967683e-01 2.51838744e-01 7.65469313e-01
4.02448565e-01 8.11477482e-01 9.57142770e-01 -7.35243142e-01
-5.37362456e-01 -8.91528070e-01 -6.93063080e-01 6.32354081e-01
5.59408009e-01 -7.99290419e-01 -4.30936784e-01 3.98876756e-01] | [12.855669975280762, 0.07993996143341064] |
4cf25f2f-12aa-40fc-8c6a-65e092bf5b45 | when-does-bottom-up-beat-top-down-in | 2306.00833 | null | https://arxiv.org/abs/2306.00833v1 | https://arxiv.org/pdf/2306.00833v1.pdf | When Does Bottom-up Beat Top-down in Hierarchical Community Detection? | Hierarchical clustering of networks consists in finding a tree of communities, such that lower levels of the hierarchy reveal finer-grained community structures. There are two main classes of algorithms tackling this problem. Divisive ($\textit{top-down}$) algorithms recursively partition the nodes into two communities, until a stopping rule indicates that no further split is needed. In contrast, agglomerative ($\textit{bottom-up}$) algorithms first identify the smallest community structure and then repeatedly merge the communities using a $\textit{linkage}$ method. In this article, we establish theoretical guarantees for the recovery of the hierarchical tree and community structure of a Hierarchical Stochastic Block Model by a bottom-up algorithm. We also establish that this bottom-up algorithm attains the information-theoretic threshold for exact recovery at intermediate levels of the hierarchy. Notably, these recovery conditions are less restrictive compared to those existing for top-down algorithms. This shows that bottom-up algorithms extend the feasible region for achieving exact recovery at intermediate levels. Numerical experiments on both synthetic and real data sets confirm the superiority of bottom-up algorithms over top-down algorithms. We also observe that top-down algorithms can produce dendrograms with inversions. These findings contribute to a better understanding of hierarchical clustering techniques and their applications in network analysis. | ['Patrick Thiran', 'Matthias Grossglauser', 'Daichi Kuroda', 'Maximilien Dreveton'] | 2023-06-01 | null | null | null | null | ['stochastic-block-model', 'community-detection'] | ['graphs', 'graphs'] | [ 2.78614640e-01 2.60932803e-01 -6.87583238e-02 2.43063532e-02
-3.22872192e-01 -7.99230218e-01 1.19276129e-01 6.19507968e-01
-6.33689016e-02 4.55454350e-01 -8.93638507e-02 -4.49229211e-01
-7.58371949e-01 -1.23210418e+00 -4.78440225e-01 -9.47426260e-01
-7.43076026e-01 8.26889813e-01 5.35090923e-01 -4.08904813e-02
3.95433545e-01 3.70143116e-01 -1.44408941e+00 2.37611100e-01
1.00042975e+00 4.04194176e-01 8.21104273e-02 6.86919868e-01
-6.97776824e-02 3.76864165e-01 -1.74686313e-01 -2.03642800e-01
3.78294945e-01 -4.81044859e-01 -9.52397883e-01 3.80882949e-01
-2.48874575e-01 4.86534201e-02 -7.61658102e-02 1.30803895e+00
1.39781877e-01 -3.15147668e-01 6.90658033e-01 -1.45546043e+00
-1.99848101e-01 1.12331629e+00 -1.38438570e+00 1.38485685e-01
3.05826545e-01 -4.63568270e-01 1.26587069e+00 -6.57666922e-01
5.57605565e-01 1.14706790e+00 8.84277105e-01 -7.05012307e-02
-1.73630655e+00 -8.26466858e-01 1.50319085e-01 -1.75493196e-01
-1.87310398e+00 -1.54062048e-01 5.65027773e-01 -7.83264160e-01
3.24544251e-01 3.60868275e-01 6.85770571e-01 1.43404290e-01
-1.41444460e-01 3.93135786e-01 1.19743967e+00 -4.68388259e-01
3.71040612e-01 -1.54197276e-01 4.22226787e-01 6.70566380e-01
1.04237711e+00 -2.82795906e-01 -4.11985368e-01 -5.95466614e-01
7.92000353e-01 1.35486573e-02 -1.18743189e-01 -8.61888468e-01
-1.09670138e+00 9.11803961e-01 5.19365311e-01 5.11727810e-01
-3.81795108e-01 1.28839165e-01 2.20882073e-01 3.25423568e-01
3.17559808e-01 5.02383336e-02 6.87753186e-02 5.28108478e-01
-1.41713560e+00 -1.39355794e-01 7.26027727e-01 1.23877382e+00
1.03713787e+00 -4.17792946e-01 3.05322766e-01 4.93482590e-01
3.58686507e-01 1.51731774e-01 -2.60629207e-01 -1.13112068e+00
4.51091230e-01 7.78442860e-01 8.52887891e-03 -1.36551869e+00
-3.71465683e-01 -5.65842211e-01 -1.48777354e+00 -1.03215791e-01
6.22758567e-01 1.10259661e-02 -5.05004704e-01 1.68930805e+00
3.73253047e-01 -6.48021773e-02 -7.18538240e-02 4.18620974e-01
1.27056018e-01 5.59783995e-01 -3.30278695e-01 -7.34160602e-01
1.27573740e+00 -4.25128847e-01 -3.75587225e-01 2.51702726e-01
4.43294019e-01 -3.74421060e-01 4.82458949e-01 3.55620533e-01
-1.32952559e+00 -3.10847521e-01 -8.87826025e-01 6.46698534e-01
1.65256828e-01 -3.00709188e-01 4.37509775e-01 8.79459023e-01
-1.53063762e+00 6.08578086e-01 -9.04238224e-01 -5.62709332e-01
8.97356793e-02 4.65392262e-01 -2.27310017e-01 -1.17639080e-01
-7.60619640e-01 -3.29914875e-02 4.65476841e-01 3.18160504e-01
-6.68112397e-01 -4.20208901e-01 -5.03260911e-01 2.99928218e-01
4.78964806e-01 -6.84895039e-01 4.72815275e-01 -6.68403447e-01
-5.57135582e-01 8.33076596e-01 -3.44987452e-01 -4.55174983e-01
4.97967422e-01 6.42189085e-01 1.42218247e-01 6.15661621e-01
6.02093160e-01 3.08221459e-01 5.91598451e-01 -1.71695137e+00
-6.76143646e-01 -5.52928090e-01 -1.72766298e-01 3.49658504e-02
-4.19186056e-01 -5.17122746e-02 -4.67665374e-01 -3.01328242e-01
1.00055254e+00 -9.72982228e-01 -5.13861299e-01 -3.79147619e-01
-7.71010041e-01 -2.51733273e-01 2.42558658e-01 -2.35860586e-01
1.57591438e+00 -1.83868241e+00 2.52349615e-01 1.04293478e+00
8.86447728e-01 -4.68011975e-01 2.50645220e-01 1.06425154e+00
-1.39323890e-01 6.26078427e-01 -5.41041374e-01 -1.54824869e-03
-2.12443694e-01 9.06411633e-02 -4.07208689e-02 7.16388166e-01
-4.30017948e-01 1.61258534e-01 -7.97236979e-01 -7.45334208e-01
-1.71469852e-01 -6.40892684e-02 -8.71120989e-01 -4.18661147e-01
5.69692910e-01 2.60102123e-01 -3.60047698e-01 4.06125903e-01
8.57033312e-01 -7.20541298e-01 1.02157688e+00 1.60156384e-01
-3.48878205e-01 -1.30883843e-01 -1.64250016e+00 8.66563737e-01
2.05905169e-01 2.73360610e-01 7.50752330e-01 -1.28105581e+00
7.03953624e-01 2.84037471e-01 8.20919037e-01 8.01153481e-02
-6.49067760e-02 2.25822881e-01 2.22351328e-01 6.05277084e-02
3.13636541e-01 -4.28625524e-01 -1.87584370e-01 6.71478570e-01
-5.67866027e-01 1.21603236e-01 6.06793463e-01 7.45570838e-01
1.35315824e+00 -6.39703512e-01 2.97518998e-01 -8.03100884e-01
4.04149264e-01 1.13797732e-01 6.65585577e-01 1.11649895e+00
-2.69556135e-01 4.27365839e-01 8.87245595e-01 2.18710989e-01
-1.16241860e+00 -1.11269844e+00 -7.59901106e-02 9.19164479e-01
3.38898063e-01 -5.51315665e-01 -1.07210851e+00 -1.29739046e-01
8.92943218e-02 -4.83195111e-02 -8.05133283e-01 3.33238184e-01
-3.01243633e-01 -9.48480010e-01 4.25748140e-01 3.09280694e-01
3.55289996e-01 -5.03018439e-01 -3.58930737e-01 2.47975901e-01
-5.42651594e-01 -8.95295978e-01 -2.91670680e-01 1.19659342e-01
-1.30642736e+00 -1.22433841e+00 -5.60788751e-01 -9.74507987e-01
1.19336700e+00 7.66456544e-01 7.55231023e-01 6.02493584e-01
9.59968492e-02 1.86746240e-01 -3.91430527e-01 4.47634131e-01
-5.18320441e-01 1.86096817e-01 1.17163464e-01 2.01363992e-02
1.07654840e-01 -1.03074110e+00 -6.55052423e-01 4.27710980e-01
-7.83955097e-01 5.58899902e-02 4.57908243e-01 5.97858131e-01
4.83248860e-01 1.11763144e+00 3.69424373e-01 -7.57566690e-01
6.18327141e-01 -5.42110622e-01 -7.28586972e-01 7.36601055e-02
-6.89226985e-01 2.01210398e-02 4.86458331e-01 1.64256245e-01
-6.68487787e-01 1.22304589e-01 3.38722378e-01 -1.31926894e-01
1.36069059e-01 6.64782941e-01 -4.72314619e-02 3.42714489e-01
3.50577831e-01 2.02424452e-01 7.57108349e-03 -3.63433838e-01
1.81103513e-01 5.81185699e-01 3.54846299e-01 -7.39964247e-01
1.08598590e+00 1.05083239e+00 2.24365324e-01 -1.01872396e+00
-3.45552474e-01 -7.99633801e-01 -1.14546895e+00 -3.47180545e-01
5.08679330e-01 -7.45157003e-01 -1.07222581e+00 2.38467410e-01
-6.73677027e-01 -5.27826957e-02 1.01297602e-01 1.45601243e-01
-5.70466042e-01 8.43001068e-01 -8.11688483e-01 -1.13407636e+00
2.82552429e-02 -8.49650621e-01 6.30669832e-01 -2.59341896e-01
-2.08820120e-01 -8.46809208e-01 -7.81081021e-02 3.44467223e-01
-2.74508893e-01 3.66662979e-01 1.05390608e+00 -3.43757987e-01
-6.95128024e-01 6.72056824e-02 -3.89364123e-01 -2.66832620e-01
6.10084683e-02 2.42823482e-01 -3.11018944e-01 -7.63716459e-01
-2.07855821e-01 3.01159263e-01 8.60194087e-01 5.51690578e-01
7.11560071e-01 -5.65138578e-01 -8.27768981e-01 2.59263724e-01
1.57256222e+00 6.42199516e-02 4.33420420e-01 3.14038098e-01
5.22293746e-01 1.00549579e+00 1.58796981e-01 6.02667987e-01
4.59520817e-01 2.46619925e-01 3.26722026e-01 -3.36936861e-02
3.59014302e-01 -3.04236978e-01 -6.95163086e-02 1.22163832e+00
-1.28945887e-01 -7.56671578e-02 -1.22748733e+00 9.97388363e-01
-1.80102205e+00 -1.24711323e+00 -7.61134624e-01 2.46974635e+00
1.00930977e+00 2.81177729e-01 5.44950008e-01 7.02410877e-01
1.51258802e+00 -2.53727883e-01 -2.57517666e-01 -1.63951829e-01
1.23114223e-02 1.88045353e-02 4.78333712e-01 5.86037993e-01
-8.96422982e-01 5.76031685e-01 6.48732758e+00 7.04008758e-01
-3.07865560e-01 -6.22849353e-02 5.17989337e-01 2.67904162e-01
-4.34241980e-01 3.72680724e-01 -6.52178407e-01 2.53648967e-01
7.38027155e-01 -5.13802111e-01 2.65414268e-01 5.90746403e-01
4.48270172e-01 -2.42212445e-01 -8.94983709e-01 4.47171032e-01
-5.60659230e-01 -1.17040420e+00 -1.32745475e-01 7.00308025e-01
9.30163920e-01 -3.08297068e-01 -3.21126103e-01 -2.71625161e-01
1.01257873e+00 -7.18252659e-01 5.90276420e-01 -6.62185326e-02
7.05928445e-01 -1.04534447e+00 3.44250023e-01 6.57042980e-01
-1.80709505e+00 -2.51199275e-01 -2.56194949e-01 -2.14652225e-01
1.94860682e-01 9.79372859e-01 -6.68943703e-01 8.28846216e-01
9.32502806e-01 5.28171420e-01 -2.86828786e-01 9.52874303e-01
9.53719243e-02 6.37796104e-01 -5.37105978e-01 2.22766548e-01
1.53560191e-01 -6.16636932e-01 6.05137646e-01 1.10453522e+00
2.64194578e-01 2.59644806e-01 3.08865219e-01 6.93373621e-01
-9.42209065e-02 -5.22771245e-03 -4.86563832e-01 8.24154615e-02
8.79694641e-01 8.69415760e-01 -1.62428689e+00 -3.74162734e-01
4.89373244e-02 6.72400355e-01 2.08765432e-01 2.80046999e-01
-4.73879427e-01 -5.04936755e-01 2.62557417e-01 7.78766096e-01
4.98383373e-01 -4.82270867e-01 -5.20102680e-01 -9.10449266e-01
-2.43760511e-01 -7.51321197e-01 7.06740379e-01 -3.34486574e-01
-1.31010771e+00 2.13069484e-01 3.80668551e-01 -1.05344558e+00
6.03226572e-02 4.17376198e-02 -3.04455638e-01 5.33739388e-01
-8.21774900e-01 -5.12220681e-01 -1.43746093e-01 5.19363105e-01
8.83055702e-02 5.61653852e-01 3.37934434e-01 1.25543922e-01
-4.64541018e-01 1.67169645e-01 6.04792178e-01 3.40333998e-01
-3.77003625e-02 -1.30839884e+00 -5.77388182e-02 1.15361190e+00
-2.69003749e-01 1.15351558e+00 7.96782076e-01 -8.82701099e-01
-9.35069323e-01 -7.51971900e-01 9.22272265e-01 7.41572082e-02
9.41095650e-01 -4.47469652e-01 -7.71141112e-01 5.70940137e-01
2.88668036e-01 -5.99616706e-01 7.72102416e-01 2.97829241e-01
-3.34328085e-01 -7.14579299e-02 -1.17871416e+00 5.70617497e-01
1.23252356e+00 -2.35616222e-01 -2.35375613e-01 8.10559765e-02
3.40373635e-01 4.97030407e-01 -1.16537750e+00 4.52832520e-01
5.80333948e-01 -1.34121525e+00 9.46422100e-01 -9.76791978e-02
3.85367960e-01 -5.46249747e-01 -1.13867626e-01 -8.96161556e-01
-5.96264780e-01 -8.86502326e-01 3.57316196e-01 1.19920969e+00
2.67013490e-01 -6.62930191e-01 1.00285614e+00 1.96328722e-02
4.30324882e-01 -4.48823661e-01 -9.77606535e-01 -9.30195272e-01
4.69895273e-01 -3.87218110e-02 2.63604403e-01 1.17168236e+00
3.47318381e-01 3.53334188e-01 1.47301741e-02 3.93235207e-01
1.46800315e+00 4.45970118e-01 5.66788852e-01 -1.73434150e+00
-1.58469230e-01 -6.11460686e-01 -1.92175820e-01 -9.63355839e-01
-1.22664683e-01 -8.74157965e-01 -2.83976402e-02 -1.94476092e+00
8.84513021e-01 -9.26829815e-01 2.12974064e-02 2.83537745e-01
-3.16482559e-02 3.67748439e-01 3.64032164e-02 7.88005829e-01
-4.86364871e-01 7.69578889e-02 9.79213715e-01 1.03768036e-02
-3.64611834e-01 1.72296143e-03 -9.77227986e-01 7.43499637e-01
6.93088472e-01 -6.79318905e-01 -4.54742998e-01 1.15602463e-01
4.80642647e-01 4.83476311e-01 9.18427184e-02 -7.93555796e-01
5.64336956e-01 -7.03305081e-02 -6.96449727e-02 -1.06604052e+00
-5.15553653e-02 -6.88244820e-01 5.16801417e-01 9.05895412e-01
-2.98376918e-01 9.87601131e-02 -1.95983723e-01 9.26918805e-01
7.75174648e-02 -1.72680438e-01 8.72533023e-01 -1.04385354e-01
1.12181075e-01 -6.05008788e-02 -8.94558132e-01 3.54610831e-02
1.08320236e+00 -7.03093886e-01 -3.66162248e-02 -5.33338547e-01
-1.05498540e+00 5.10411441e-01 6.25815749e-01 -3.44214946e-01
2.85613298e-01 -9.84477639e-01 -9.51790392e-01 -8.71346667e-02
-1.17088914e-01 6.85101002e-02 6.21962361e-02 1.16465950e+00
-5.89580536e-01 3.37162614e-01 7.15093836e-02 -9.94092345e-01
-1.63610387e+00 7.68698335e-01 1.39036611e-01 -4.68394667e-01
-4.97481972e-01 6.42740548e-01 6.01879537e-01 -2.25898772e-01
-1.21566884e-01 -4.70421948e-02 8.01457614e-02 3.58060718e-01
1.07269339e-01 6.82242334e-01 -2.84558713e-01 -7.23233283e-01
-4.31338251e-01 6.63057685e-01 7.33126253e-02 -3.97198975e-01
1.30419242e+00 -7.10085034e-01 -8.02576959e-01 2.64646977e-01
8.63237977e-01 2.40248159e-01 -8.55385542e-01 1.11305146e-02
3.75868231e-01 -3.57423425e-01 -3.58759731e-01 4.95158248e-02
-9.24223721e-01 6.50548637e-01 -1.31341472e-01 1.18003225e+00
1.21475208e+00 2.38804802e-01 1.19404264e-01 1.89818695e-01
5.92225313e-01 -8.88513982e-01 5.59577206e-03 1.18582264e-01
3.93065751e-01 -4.23011869e-01 2.23355234e-01 -1.07796001e+00
-7.98266977e-02 7.39318311e-01 1.09277308e-01 -1.00748256e-01
7.93021321e-01 2.53426045e-01 -7.07588255e-01 -3.59750092e-01
-7.08113074e-01 -4.20991361e-01 -4.24922854e-01 5.63296616e-01
2.27504432e-01 1.68555260e-01 -4.80960548e-01 3.48361909e-01
-2.99777925e-01 -3.73697460e-01 8.77441168e-01 1.00571990e+00
-9.23003674e-01 -1.00795841e+00 -7.50695884e-01 3.37524772e-01
-2.69569784e-01 -1.86127394e-01 -5.75244248e-01 9.05580580e-01
-1.39988080e-01 1.52444708e+00 1.12054378e-01 -1.69022352e-01
2.80603170e-02 -2.87747860e-01 4.50791329e-01 -6.20191336e-01
-4.38887566e-01 6.69622719e-01 1.93410125e-02 -5.31119108e-02
-5.83737195e-01 -8.70230436e-01 -1.28527272e+00 -8.90071273e-01
-6.56338930e-01 7.81731963e-01 8.18025693e-02 4.04837787e-01
2.38540396e-01 2.97355920e-01 1.03786469e+00 -5.14024734e-01
-1.65458143e-01 -6.30737066e-01 -9.73472536e-01 2.33600056e-03
-2.43640076e-02 -4.94356155e-01 -7.31661975e-01 1.88684881e-01] | [6.936427593231201, 5.119372844696045] |
6de12f31-1ed1-4481-adf4-5e3614389d39 | virtual-sparse-convolution-for-multimodal-3d | 2303.02314 | null | https://arxiv.org/abs/2303.02314v1 | https://arxiv.org/pdf/2303.02314v1.pdf | Virtual Sparse Convolution for Multimodal 3D Object Detection | Recently, virtual/pseudo-point-based 3D object detection that seamlessly fuses RGB images and LiDAR data by depth completion has gained great attention. However, virtual points generated from an image are very dense, introducing a huge amount of redundant computation during detection. Meanwhile, noises brought by inaccurate depth completion significantly degrade detection precision. This paper proposes a fast yet effective backbone, termed VirConvNet, based on a new operator VirConv (Virtual Sparse Convolution), for virtual-point-based 3D object detection. VirConv consists of two key designs: (1) StVD (Stochastic Voxel Discard) and (2) NRConv (Noise-Resistant Submanifold Convolution). StVD alleviates the computation problem by discarding large amounts of nearby redundant voxels. NRConv tackles the noise problem by encoding voxel features in both 2D image and 3D LiDAR space. By integrating VirConv, we first develop an efficient pipeline VirConv-L based on an early fusion design. Then, we build a high-precision pipeline VirConv-T based on a transformed refinement scheme. Finally, we develop a semi-supervised pipeline VirConv-S based on a pseudo-label framework. On the KITTI car 3D detection test leaderboard, our VirConv-L achieves 85% AP with a fast running speed of 56ms. Our VirConv-T and VirConv-S attains a high-precision of 86.3% and 87.2% AP, and currently rank 2nd and 1st, respectively. The code is available at https://github.com/hailanyi/VirConv. | ['Cheng Wang', 'Xin Li', 'Shaoshuai Shi', 'Chenglu Wen', 'Hai Wu'] | 2023-03-04 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Wu_Virtual_Sparse_Convolution_for_Multimodal_3D_Object_Detection_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Wu_Virtual_Sparse_Convolution_for_Multimodal_3D_Object_Detection_CVPR_2023_paper.pdf | cvpr-2023-1 | ['depth-completion'] | ['computer-vision'] | [-1.16604269e-01 -2.33970582e-01 2.80721337e-01 -4.86294366e-02
-1.15955174e+00 -3.82471085e-01 4.97737259e-01 4.94058318e-02
-4.92141247e-01 8.69467184e-02 -4.08047467e-01 -1.53618783e-01
1.98852405e-01 -9.52785373e-01 -7.76788533e-01 -5.40365756e-01
2.34855562e-01 6.75772727e-01 7.46073008e-01 5.42296446e-04
8.79468694e-02 8.21970820e-01 -1.84080589e+00 -7.80886114e-02
7.48457789e-01 1.20248401e+00 4.88155663e-01 4.23566341e-01
-2.96888381e-01 3.17242481e-02 -3.36712569e-01 -2.39952981e-01
5.01470208e-01 8.55617374e-02 -2.42820591e-01 -2.61587780e-02
5.74718475e-01 -4.22364771e-01 -1.77295163e-01 1.11393070e+00
5.17075419e-01 -7.03993291e-02 6.61993265e-01 -1.27368569e+00
-4.09571268e-02 1.44239858e-01 -1.02992392e+00 -2.50097752e-01
1.87632024e-01 3.20376754e-01 6.75340295e-01 -1.47221613e+00
5.24484098e-01 1.42518544e+00 8.97319317e-01 2.91698307e-01
-1.35497832e+00 -8.65869701e-01 -5.67094423e-02 3.03448597e-03
-1.89861619e+00 -6.18408546e-02 7.22075105e-01 -6.04091525e-01
8.40383410e-01 3.64350468e-01 7.70297289e-01 8.06147516e-01
5.22638299e-02 7.64127076e-01 9.94686782e-01 -6.89250901e-02
3.32982928e-01 -1.71419173e-01 1.27311975e-01 8.40787053e-01
4.58108127e-01 2.99390346e-01 -3.97287548e-01 -2.98992902e-01
8.55919302e-01 2.85371840e-01 -1.66551337e-01 -5.34767628e-01
-1.08905315e+00 8.24289918e-01 6.05705440e-01 -2.89101183e-01
-3.57608885e-01 1.74137518e-01 1.90134630e-01 -2.09122717e-01
4.52850074e-01 -4.12999354e-02 -2.76150048e-01 1.92551062e-01
-8.98802578e-01 2.69939333e-01 3.55361223e-01 1.06839204e+00
1.11362231e+00 -5.56969419e-02 -1.26132101e-01 6.48868442e-01
4.80573922e-01 1.14864910e+00 -9.95505154e-02 -1.14200246e+00
1.86742797e-01 8.23879302e-01 -2.60699242e-02 -6.71833992e-01
-4.98112917e-01 -4.40280437e-01 -8.45482051e-01 7.82502353e-01
7.67450929e-02 2.40001664e-01 -1.13737357e+00 1.16092074e+00
7.81999648e-01 3.03948522e-01 -1.55391112e-01 1.18654871e+00
9.75494742e-01 4.53942120e-01 -3.02340031e-01 -1.67008378e-02
1.40589440e+00 -6.40416265e-01 -1.75804093e-01 -6.98865429e-02
4.88342375e-01 -8.01880062e-01 9.19852138e-01 4.20896292e-01
-8.32820296e-01 -6.22233987e-01 -1.02963018e+00 -1.89266056e-01
-2.46457532e-02 2.02989951e-01 4.78428632e-01 3.78142267e-01
-8.70505929e-01 3.81637961e-01 -9.38076377e-01 -1.23785198e-01
5.95385790e-01 3.20089072e-01 -3.26675385e-01 -3.15662950e-01
-4.68377203e-01 5.16113818e-01 2.26521611e-01 6.07045256e-02
-9.30453002e-01 -7.37432182e-01 -9.65022445e-01 -3.09757918e-01
5.87267876e-01 -8.74541640e-01 1.15495348e+00 -3.89000438e-02
-1.13756776e+00 9.48374629e-01 -2.56865501e-01 -3.24587524e-01
7.16857612e-01 -3.42194259e-01 -3.61338742e-02 8.44208524e-02
3.58143479e-01 6.47196293e-01 8.21101844e-01 -1.37061322e+00
-6.97513700e-01 -6.33056462e-01 -5.14097512e-01 1.73657462e-01
1.98129728e-01 -3.53178978e-01 -1.07554317e+00 -3.20960671e-01
7.31975794e-01 -8.97915602e-01 -3.24027807e-01 3.23057860e-01
-3.85819048e-01 -3.58901858e-01 9.47077870e-01 -1.72544211e-01
5.88865101e-01 -2.36062312e+00 -3.48799601e-02 2.62547165e-01
7.22869456e-01 4.06476200e-01 -1.76140845e-01 4.68491837e-02
2.71183610e-01 -2.19637409e-01 -3.82568926e-01 -8.43518674e-01
-1.68744221e-01 3.08731556e-01 -2.33947441e-01 6.57839119e-01
1.58759326e-01 7.80234814e-01 -9.48139131e-01 -4.91368890e-01
7.09487677e-01 7.79531181e-01 -4.76564974e-01 1.02137960e-01
-1.39229864e-01 2.43627548e-01 -4.09735173e-01 1.04320800e+00
1.35510445e+00 -1.10497750e-01 -4.24196929e-01 -5.05399346e-01
-5.46258748e-01 4.75885570e-02 -1.40388107e+00 1.71785259e+00
-6.94582164e-02 2.75307685e-01 4.05026227e-01 -4.57752645e-01
1.11494160e+00 -1.86693102e-01 6.07719064e-01 -6.01333857e-01
3.89089137e-01 3.80625874e-01 -7.30880857e-01 -2.15674490e-02
6.06452823e-01 1.03302412e-01 5.35611212e-02 1.65783867e-01
-1.39825186e-02 -5.07493436e-01 -9.81241167e-02 3.83395046e-01
1.26549637e+00 3.49283427e-01 1.59787849e-01 3.82519257e-03
4.87207144e-01 1.98871955e-01 7.12475240e-01 7.31002033e-01
-1.36939108e-01 8.78925085e-01 1.23406567e-01 -4.71019596e-02
-7.17497349e-01 -1.49798667e+00 -3.75470370e-01 3.33603621e-01
5.85646927e-01 -5.97676456e-01 -5.32637775e-01 -5.34684718e-01
4.96982843e-01 5.49257278e-01 -3.18209827e-01 -4.17368934e-02
-3.78057212e-01 -4.44453090e-01 3.09909910e-01 3.91355366e-01
4.05192584e-01 -4.91213053e-01 -7.12641835e-01 4.15270068e-02
-3.99830304e-02 -1.13985717e+00 -2.73056686e-01 1.97000757e-01
-1.01403701e+00 -1.11328769e+00 -4.87737745e-01 -3.69785577e-01
5.28837860e-01 8.93231213e-01 8.22660685e-01 -1.39637087e-02
-4.78297740e-01 4.16538388e-01 -4.39377308e-01 -4.83969003e-01
-1.67900845e-02 -2.31923327e-01 3.29408705e-01 -1.16643965e-01
4.94055271e-01 -4.97974545e-01 -5.16406000e-01 4.90315199e-01
-5.09329140e-01 2.80862123e-01 6.16636753e-01 5.44236124e-01
1.29259443e+00 -3.10182840e-01 -3.17853868e-01 -3.72185618e-01
-1.63986102e-01 -3.09682935e-01 -1.12842941e+00 -3.27162057e-01
-5.41614473e-01 -6.97540417e-02 3.63073088e-02 -2.96343118e-01
-6.85391665e-01 5.93205452e-01 -4.39857244e-01 -1.19359982e+00
-4.51615633e-04 -1.11463224e-03 -1.23395346e-01 -2.80312270e-01
6.84705138e-01 1.84928343e-01 1.04966804e-01 -8.37706387e-01
3.80013555e-01 6.92360103e-01 7.66619503e-01 -3.15931082e-01
1.26626873e+00 9.11469579e-01 1.12423129e-01 -9.71773982e-01
-6.79650068e-01 -6.87854052e-01 -6.57420814e-01 -3.48039418e-01
8.15602660e-01 -1.32818496e+00 -8.39666784e-01 4.70034808e-01
-1.35804427e+00 -2.91403383e-01 -4.96530682e-01 6.07662022e-01
-2.17275113e-01 4.71163005e-01 -4.04692292e-01 -9.78635788e-01
-4.17506725e-01 -1.21653843e+00 1.58244669e+00 -4.58396832e-03
1.00653283e-01 -1.23558395e-01 -8.91938880e-02 4.03456539e-01
-1.25286534e-01 3.79648924e-01 1.30829111e-01 -6.77393898e-02
-1.04602170e+00 -2.25737244e-01 -6.32484555e-01 2.93339700e-01
-1.85777962e-01 -1.16761453e-01 -1.13435650e+00 -1.97077900e-01
1.84200611e-03 -4.05145958e-02 1.03829372e+00 2.24168435e-01
8.80162656e-01 4.00831580e-01 -5.20825446e-01 1.02187490e+00
1.45197546e+00 -1.11808710e-01 4.26717520e-01 1.60540622e-02
1.00716579e+00 2.50367701e-01 9.25975621e-01 4.85306591e-01
4.58465487e-01 8.20290864e-01 9.73325253e-01 -1.35289088e-01
-4.41495001e-01 -2.28651330e-01 3.96399945e-01 7.05707014e-01
-1.80399165e-01 2.17331111e-01 -1.00151443e+00 2.73925811e-01
-1.90524983e+00 -5.41501999e-01 -9.38573122e-01 2.29992366e+00
3.55374783e-01 3.12587678e-01 -1.07410559e-02 1.57248259e-01
5.24874806e-01 -7.02072233e-02 -6.99375451e-01 2.62292981e-01
-1.66964114e-01 2.08754137e-01 8.19661677e-01 5.79041660e-01
-1.05940604e+00 9.41897988e-01 4.52874899e+00 8.66434932e-01
-9.10840273e-01 4.44162518e-01 -2.50150729e-02 -1.44246340e-01
-1.55087709e-01 -2.71374900e-02 -1.19634855e+00 3.17397118e-01
3.30726236e-01 2.96955526e-01 9.64790508e-02 1.12477064e+00
3.74752544e-02 -2.79418766e-01 -7.69090116e-01 1.49224770e+00
8.35771412e-02 -1.37720025e+00 -6.34704605e-02 2.23662227e-01
3.93964618e-01 7.18962550e-01 -2.57819086e-01 1.94509670e-01
1.91630736e-01 -5.75821817e-01 1.09557950e+00 2.76933283e-01
9.81982112e-01 -6.68628454e-01 6.71273112e-01 4.80182648e-01
-1.64537239e+00 1.87571675e-01 -5.24537265e-01 3.99990790e-02
3.25196028e-01 1.13629866e+00 -5.69883227e-01 6.00530207e-01
9.28031266e-01 6.63410544e-01 -4.74630654e-01 1.14156246e+00
-4.16878819e-01 3.64240497e-01 -6.60496712e-01 2.36369103e-01
8.80451724e-02 -3.19255918e-01 8.82310331e-01 1.04784071e+00
4.47060704e-01 3.40424150e-01 4.68652189e-01 9.27215517e-01
1.19693667e-01 -1.25154197e-01 -5.22058547e-01 6.14810109e-01
5.66648901e-01 1.49297166e+00 -8.00082803e-01 -2.57659465e-01
-2.87960947e-01 1.09358776e+00 2.28313789e-01 6.23538531e-02
-9.04772401e-01 -2.61465073e-01 9.94432092e-01 2.70887792e-01
5.54836631e-01 -5.72586358e-01 -3.83946657e-01 -1.10034776e+00
-5.60008883e-02 -2.98611253e-01 1.14505164e-01 -8.60624194e-01
-1.00989962e+00 5.41183770e-01 -1.02568358e-01 -1.35936272e+00
2.46890247e-01 -6.96804106e-01 -3.11584681e-01 8.45817566e-01
-1.49790788e+00 -1.11762846e+00 -8.17191541e-01 6.42533898e-01
3.85959506e-01 1.46986768e-01 5.65959990e-01 4.20074672e-01
-5.25960386e-01 2.31178001e-01 -1.94322839e-01 -2.47525349e-02
4.98289436e-01 -9.98537958e-01 5.54457188e-01 7.98195779e-01
5.86501956e-02 2.54227787e-01 3.36075753e-01 -8.79447281e-01
-1.79916298e+00 -1.45610297e+00 5.33783197e-01 -5.71065485e-01
2.51295716e-01 -5.76621056e-01 -9.82533395e-01 3.79306674e-01
-5.74634492e-01 3.33302498e-01 1.76083013e-01 -2.35858306e-01
-4.95462805e-01 -2.50967622e-01 -1.21207166e+00 3.42319757e-01
1.40785146e+00 -4.97224212e-01 -3.40111166e-01 3.28478068e-01
9.85119998e-01 -7.76614487e-01 -5.52139223e-01 5.65834045e-01
3.50669056e-01 -1.11399126e+00 1.28931022e+00 3.37827712e-01
-1.82204977e-01 -9.19061780e-01 -4.84057128e-01 -6.69096291e-01
-4.15478349e-01 -4.46062684e-01 -3.86426508e-01 9.86322999e-01
-1.35346934e-01 -6.35580540e-01 8.28510404e-01 4.76421341e-02
-5.33189118e-01 -5.68965316e-01 -1.23966348e+00 -9.35399890e-01
-4.54883158e-01 -1.14964151e+00 4.73163605e-01 5.76428771e-01
-7.37620533e-01 3.93415809e-01 5.43082058e-02 5.24070680e-01
1.27441990e+00 2.75534362e-01 1.15747607e+00 -1.61188185e+00
-5.75627610e-02 -4.00884002e-01 -5.90346873e-01 -1.15645313e+00
-2.79131263e-01 -9.84561801e-01 7.79156238e-02 -1.51937997e+00
6.51976094e-02 -5.04465699e-01 -2.48750057e-02 5.40540695e-01
1.13225803e-01 7.71611810e-01 2.54995584e-01 3.42580736e-01
-4.48136955e-01 6.59650862e-01 9.94810581e-01 3.22030410e-02
-3.27390254e-01 -1.06559224e-01 -2.84784019e-01 9.78065014e-01
6.26308441e-01 -5.30544877e-01 -6.73006177e-02 -4.27194029e-01
-7.31610209e-02 -1.25733763e-01 8.32558155e-01 -1.19925725e+00
1.92826658e-01 3.87290679e-02 3.77538413e-01 -1.39847136e+00
8.40007484e-01 -6.58677757e-01 2.29618877e-01 6.35179758e-01
6.36664510e-01 -1.13092877e-01 2.12955222e-01 5.42090893e-01
8.61171931e-02 1.14629060e-01 9.35424447e-01 5.69410734e-02
-9.05473173e-01 5.78117549e-01 -1.08929239e-01 -2.13904306e-01
1.13012230e+00 -4.03158963e-01 -4.83285934e-02 1.49608001e-01
-3.56558800e-01 3.76926303e-01 6.08918667e-01 2.40213275e-01
1.08908534e+00 -1.46366787e+00 -7.00213850e-01 5.03965855e-01
3.13256741e-01 6.51186109e-01 3.21102917e-01 9.77009296e-01
-5.09089291e-01 3.49445269e-02 2.77421355e-01 -1.27001393e+00
-1.41027880e+00 3.39591116e-01 7.24880174e-02 1.45624161e-01
-1.07777250e+00 1.10879111e+00 3.25436324e-01 -5.30687213e-01
1.70855999e-01 -4.03065085e-01 2.56994098e-01 1.10078245e-01
5.81344545e-01 5.36305904e-01 1.42804161e-01 -8.03292513e-01
-6.29722238e-01 9.51248050e-01 1.23029716e-01 -7.79463053e-02
1.17208016e+00 8.28963369e-02 -2.63167871e-03 2.25219533e-01
1.08771276e+00 -4.48203310e-02 -1.45304310e+00 -3.33951205e-01
-2.82543421e-01 -5.89605749e-01 4.63619262e-01 -3.73619884e-01
-1.03629637e+00 8.90982926e-01 8.88443768e-01 -4.12779570e-01
9.53732014e-01 3.29978496e-01 8.52841616e-01 2.62343615e-01
7.26269901e-01 -7.40044594e-01 -1.01828538e-02 7.54234791e-01
8.42213750e-01 -1.27431715e+00 2.85154402e-01 -6.93977594e-01
-1.82936847e-01 8.30880225e-01 6.33973062e-01 -1.27601266e-01
5.83152294e-01 3.59414488e-01 7.64435679e-02 -4.33132946e-01
-1.74554899e-01 -6.06939316e-01 1.91815242e-01 6.60112560e-01
-2.59731114e-01 1.56527117e-01 1.17282784e-02 5.08350194e-01
-6.18437082e-02 -1.92385226e-01 1.54009625e-01 8.68673444e-01
-6.00968122e-01 -8.47005785e-01 -8.21182847e-01 3.93247008e-01
3.07852745e-01 7.57054165e-02 -1.21180795e-01 7.29530990e-01
5.53783119e-01 6.25612020e-01 3.07617396e-01 -8.56417596e-01
5.82944334e-01 -3.80662233e-01 3.74358833e-01 -7.78398395e-01
-1.84611276e-01 5.29144764e-01 -3.66184831e-01 -1.04114795e+00
-4.63533141e-02 -8.70611787e-01 -1.61527467e+00 -3.15670848e-01
-3.66643727e-01 -6.61400929e-02 9.41956818e-01 5.53554833e-01
5.18007517e-01 3.64680558e-01 4.89851058e-01 -1.38529265e+00
-5.09052217e-01 -6.22545183e-01 -6.10876858e-01 3.88577282e-02
2.89421946e-01 -1.00207102e+00 -3.73582393e-01 -4.57766145e-01] | [7.803004741668701, -2.687835454940796] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.