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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0cc47a98-cf1f-4bdb-a86e-1f17e8ce01fc | synthesizing-affective-neurophysiological | 2306.03112 | null | https://arxiv.org/abs/2306.03112v1 | https://arxiv.org/pdf/2306.03112v1.pdf | Synthesizing Affective Neurophysiological Signals Using Generative Models: A Review Paper | The integration of emotional intelligence in machines is an important step in advancing human-computer interaction. This demands the development of reliable end-to-end emotion recognition systems. However, the scarcity of public affective datasets presents a challenge. In this literature review, we emphasize the use of generative models to address this issue in neurophysiological signals, particularly Electroencephalogram (EEG) and Functional Near-Infrared Spectroscopy (fNIRS). We provide a comprehensive analysis of different generative models used in the field, examining their input formulation, deployment strategies, and methodologies for evaluating the quality of synthesized data. This review serves as a comprehensive overview, offering insights into the advantages, challenges, and promising future directions in the application of generative models in emotion recognition systems. Through this review, we aim to facilitate the progression of neurophysiological data augmentation, thereby supporting the development of more efficient and reliable emotion recognition systems. | ['Mark Billinghurst', 'Gonzalo Maso Talou', 'Vanessa Tang', 'Alireza F. Nia'] | 2023-06-05 | null | null | null | null | ['eeg', 'emotional-intelligence', 'eeg'] | ['methodology', 'natural-language-processing', 'time-series'] | [ 3.00412357e-01 -2.63978153e-01 2.81376511e-01 -5.89267313e-01
-3.68434787e-01 -4.34776783e-01 1.31771997e-01 -3.79584753e-03
-3.21439147e-01 9.10521865e-01 8.98443684e-02 3.32236320e-01
-3.62270772e-01 -3.16638440e-01 -1.78588420e-01 -8.73240113e-01
-5.49794957e-02 3.84157263e-02 -9.63683426e-01 -2.26164192e-01
1.38359413e-01 5.43688238e-01 -1.57167184e+00 4.09760326e-01
1.03796184e+00 1.32841980e+00 1.59078211e-01 3.10861826e-01
2.30321556e-01 3.89325023e-01 -8.74566436e-01 -5.15342951e-01
-5.45330942e-01 -8.35024893e-01 -4.18128759e-01 9.94432159e-03
-3.76288980e-01 9.80350897e-02 -6.15275204e-02 1.01304495e+00
1.02053964e+00 3.81702483e-01 7.35428274e-01 -1.50325227e+00
-9.80408907e-01 3.00495118e-01 8.31097066e-02 4.45607275e-01
3.74049783e-01 -2.03485757e-01 5.46503305e-01 -1.04361403e+00
4.13535863e-01 5.42793155e-01 4.27779824e-01 8.82205725e-01
-1.07276118e+00 -7.28005886e-01 1.50655910e-01 5.37554085e-01
-1.19304240e+00 -4.81996328e-01 1.01439500e+00 -3.33015412e-01
1.43452978e+00 2.94186860e-01 1.20468712e+00 1.53215945e+00
4.80670512e-01 5.58159292e-01 1.43968058e+00 -1.44600064e-01
4.54127789e-01 5.57467163e-01 1.95449695e-01 9.35948193e-02
9.96746719e-02 3.27207334e-03 -1.05271578e+00 -2.45334476e-01
4.58523035e-01 -1.02323562e-01 -1.47467166e-01 -2.28421781e-02
-5.71542680e-01 6.11148834e-01 2.77246654e-01 7.53558934e-01
-9.97008085e-01 -3.42061184e-02 5.69164693e-01 1.15502462e-01
7.43712246e-01 5.46531260e-01 -2.05466732e-01 -4.68466014e-01
-1.02113104e+00 -2.08266959e-01 4.30330932e-01 3.90269369e-01
2.68842876e-01 7.29539514e-01 8.99415016e-02 1.06983805e+00
1.16818413e-01 4.79361176e-01 4.41241443e-01 -5.91821432e-01
-3.51424009e-01 2.53765643e-01 -2.38337889e-01 -8.44273686e-01
-7.00482130e-01 -5.16815007e-01 -9.00277913e-01 -2.25501489e-02
-6.37063920e-01 -4.30587143e-01 -2.94680595e-01 1.72664309e+00
-2.30766892e-01 -1.15795908e-02 1.16913043e-01 9.80858803e-01
1.04844487e+00 3.27621281e-01 5.23978949e-01 -5.68143427e-01
1.38324285e+00 -4.27104503e-01 -1.17871559e+00 -3.66723984e-01
3.97555046e-02 -2.75740445e-01 4.99881506e-01 5.43434501e-01
-1.00149906e+00 -3.62586945e-01 -7.77512193e-01 1.61805183e-01
-4.10774350e-01 3.15628141e-01 1.14895284e+00 1.02703726e+00
-9.67737794e-01 2.05132112e-01 -9.27833021e-01 -4.23501760e-01
4.87938732e-01 5.00430584e-01 -4.42197323e-01 3.84409815e-01
-1.39051175e+00 1.08472145e+00 -3.93134058e-02 3.66069019e-01
-5.40858924e-01 -5.13742268e-01 -6.91688001e-01 -1.39000937e-02
-3.34617138e-01 -8.30053389e-01 6.82338119e-01 -1.27043903e+00
-1.69370651e+00 9.21945751e-01 -2.84926593e-01 -1.47586569e-01
-3.62920970e-01 -1.90051734e-01 -6.92297637e-01 2.26472318e-01
-3.77881110e-01 5.83425581e-01 3.69411916e-01 -8.79850268e-01
1.06085874e-02 -7.51605928e-01 -5.53058207e-01 3.74056160e-01
-5.39176822e-01 4.31633979e-01 7.55303651e-02 -4.58499253e-01
-2.17416897e-01 -7.12315917e-01 -1.17499515e-01 -4.71670836e-01
8.68722738e-04 -1.91533089e-01 2.17926413e-01 -3.52213979e-01
1.05862224e+00 -2.21454859e+00 2.10910097e-01 2.38452226e-01
1.76395103e-01 8.59526098e-02 -5.45589030e-02 4.02031630e-01
-4.47971076e-01 -7.43639171e-02 -9.37404558e-02 -2.68591464e-01
-2.22960487e-02 -1.42011866e-01 -2.48921961e-01 4.79505301e-01
4.89633143e-01 1.39701521e+00 -8.28782856e-01 2.28756085e-01
4.84490365e-01 1.04136002e+00 -4.07758355e-01 1.32066637e-01
3.86310428e-01 6.65331364e-01 -3.12302798e-01 6.37397408e-01
1.89430490e-01 -5.61167374e-02 1.99053392e-01 -2.31924877e-01
-2.06323251e-01 2.62689203e-01 -6.43126130e-01 1.61428559e+00
-3.49218309e-01 8.83047760e-01 -1.36325538e-01 -1.06729376e+00
1.21757686e+00 6.83058739e-01 9.14362669e-01 -9.62918580e-01
6.16934180e-01 2.10179225e-01 1.76258534e-01 -5.58750510e-01
1.40392840e-01 -5.06511986e-01 -3.37578803e-02 4.17967230e-01
4.26040739e-01 -1.34790808e-01 -7.52521530e-02 -1.88175097e-01
6.49930239e-01 -3.67171019e-02 2.85557389e-01 -8.73357430e-02
1.58421174e-01 -4.53660131e-01 2.25631580e-01 3.92259896e-01
-3.43706578e-01 1.62319824e-01 3.08244467e-01 -2.44985014e-01
-5.11486053e-01 -8.00718009e-01 -2.71993846e-01 9.64519560e-01
-1.92279071e-01 -3.17868650e-01 -8.72284949e-01 6.82396665e-02
-4.86570209e-01 7.53836036e-01 -9.00725782e-01 -7.79309928e-01
4.72598851e-01 -1.51154113e+00 5.14573991e-01 6.98340237e-01
2.74981052e-01 -1.43648970e+00 -8.38566303e-01 3.25925648e-01
-3.90439719e-01 -1.12962484e+00 5.05660176e-01 5.34860551e-01
-8.47099781e-01 -6.23338580e-01 -4.57928121e-01 -4.84700799e-01
4.87903029e-01 -8.24280754e-02 8.84190917e-01 -3.94401371e-01
-5.33593476e-01 7.53372788e-01 -1.79361731e-01 -8.10265064e-01
2.39703402e-01 -1.36170432e-01 4.29345846e-01 -1.35802358e-01
8.13014209e-01 -8.48896325e-01 -5.64021230e-01 -9.78569984e-02
-8.32804441e-01 -1.75986305e-01 4.38092709e-01 7.65712261e-01
6.63731098e-01 -1.04885355e-01 1.27440655e+00 -5.29724658e-01
1.36925471e+00 -7.32425869e-01 1.67460553e-02 1.00080647e-01
-9.00753319e-01 -5.33632219e-01 4.24168587e-01 -4.05431837e-01
-1.08692348e+00 -3.17207456e-01 -4.43965286e-01 -1.94274142e-01
-3.21641773e-01 7.43588984e-01 -7.10032731e-02 -2.90417522e-01
8.08815241e-01 2.15157449e-01 -1.12886705e-01 1.10234199e-02
1.39454082e-01 8.59675467e-01 4.21549231e-01 -3.94543469e-01
-3.32436979e-01 1.47194982e-01 -2.87904888e-01 -1.06216800e+00
-4.22964305e-01 -1.91394448e-01 -1.82393000e-01 -6.14856482e-01
6.64365470e-01 -7.87964880e-01 -5.40512979e-01 5.32978952e-01
-7.78163373e-01 -9.33475569e-02 -2.01214328e-01 7.16327012e-01
-9.01690602e-01 -1.15515098e-01 -6.94377720e-01 -1.21288753e+00
-1.03811967e+00 -1.16113877e+00 8.80872071e-01 5.27423441e-01
-7.09396720e-01 -8.53860974e-01 3.06943834e-01 2.96723038e-01
4.80586827e-01 1.10008538e-01 7.13216543e-01 -6.30382359e-01
2.36772448e-01 -2.82104462e-01 2.42924020e-01 3.85643214e-01
2.04794094e-01 9.10105482e-02 -1.41845953e+00 3.83747816e-01
4.25466090e-01 -6.00916147e-01 3.14418703e-01 6.62276268e-01
1.11984241e+00 3.20675641e-01 -6.87615946e-02 4.88487750e-01
1.01532483e+00 5.65796614e-01 9.07531559e-01 -8.81643742e-02
9.28004086e-02 5.65103948e-01 3.42361867e-01 6.90697670e-01
2.86190599e-01 2.90521204e-01 -9.50032473e-03 -1.84581012e-01
1.29489392e-01 2.50302017e-01 2.46904850e-01 8.28571022e-01
-4.82279181e-01 -1.47005990e-01 -5.17818093e-01 2.22121730e-01
-1.37092257e+00 -1.03295493e+00 8.34168941e-02 1.91103661e+00
5.66884637e-01 -4.61955249e-01 9.80124250e-03 1.67624518e-01
3.58269632e-01 -3.37690115e-01 -8.33381176e-01 -7.62544274e-01
-3.83358419e-01 6.08618379e-01 -5.34559608e-01 -1.81871504e-02
-6.87397003e-01 8.74827802e-01 7.35309124e+00 2.48626664e-01
-1.58716059e+00 9.90338027e-02 6.61885560e-01 -4.73243713e-01
-2.80615538e-01 -7.47353673e-01 -1.64874732e-01 5.24107635e-01
1.48718023e+00 -3.10076892e-01 8.41459632e-01 6.76968277e-01
4.97256666e-01 -3.00392359e-01 -8.28599811e-01 1.33388400e+00
4.25647974e-01 -1.03998506e+00 -4.13253099e-01 -2.18456671e-01
5.12108266e-01 2.07480237e-01 3.21706980e-01 2.48874351e-01
-6.80455625e-01 -9.48187232e-01 3.67338121e-01 8.38441968e-01
9.38403606e-01 -9.18507516e-01 7.33298540e-01 -3.32906991e-02
-5.29850185e-01 1.53560296e-01 -7.78175965e-02 -1.83211133e-01
1.71373650e-01 7.65283346e-01 -1.01853363e-01 3.03408444e-01
7.80261755e-01 6.78356051e-01 -1.49125757e-03 8.53766143e-01
-2.24768937e-01 3.32147151e-01 -9.84034613e-02 -3.70698631e-01
-2.58010834e-01 -4.40640748e-01 2.19263062e-01 1.18518293e+00
2.99159080e-01 6.26453996e-01 -4.45028692e-01 1.24538720e+00
1.27903819e-02 2.35242173e-01 -5.45941889e-01 -6.46577716e-01
3.24337333e-01 1.53679669e+00 -7.07460761e-01 -8.07100981e-02
-2.70389140e-01 1.18201399e+00 2.12329283e-01 4.46409941e-01
-8.29384744e-01 -4.13944602e-01 8.00914764e-01 -5.98888516e-01
-7.27835715e-01 -9.18985531e-02 -5.18315673e-01 -1.07009876e+00
-1.47250623e-01 -9.41455364e-01 7.85214975e-02 -1.30773687e+00
-1.23791802e+00 9.52612519e-01 -2.71794517e-02 -5.20583928e-01
-3.83567244e-01 -5.25998354e-01 -4.38299179e-01 1.01528406e+00
-1.04127610e+00 -6.43875539e-01 -4.01754200e-01 4.58368897e-01
2.66585827e-01 1.78090885e-01 1.43534422e+00 4.18693364e-01
-7.75951564e-01 2.63129771e-01 4.62908670e-02 -4.66338187e-01
6.46136701e-01 -5.75105786e-01 -1.16531275e-01 2.92117178e-01
1.73952252e-01 6.97474658e-01 6.16005182e-01 -4.49237943e-01
-1.30883086e+00 -6.26281619e-01 5.42628884e-01 -3.07343304e-01
5.06535769e-01 -3.79391640e-01 -7.80551732e-01 4.90885139e-01
2.80761898e-01 -4.50166374e-01 1.65455496e+00 2.81608880e-01
2.42889956e-01 1.51368203e-02 -1.25890338e+00 4.75771189e-01
7.13923931e-01 -8.16812336e-01 -2.68766791e-01 1.21373557e-01
-2.64303029e-01 -1.33900642e-01 -1.04730105e+00 3.91192526e-01
7.81386554e-01 -9.87970293e-01 7.29006827e-01 -6.52939200e-01
1.70453072e-01 3.35924923e-01 6.83864951e-02 -1.73389030e+00
-4.44180191e-01 -3.88164788e-01 -7.55479932e-02 9.89039660e-01
1.13608204e-01 -6.48714721e-01 4.50845003e-01 1.34087253e+00
-1.98234394e-01 -1.07463539e+00 -7.46958971e-01 -1.61866099e-01
-1.29925951e-01 -8.48175883e-01 2.00601965e-01 6.11159086e-01
8.79752755e-01 4.25799400e-01 -2.17247516e-01 -2.95324713e-01
1.62939310e-01 -1.31469935e-01 7.18375593e-02 -1.06509316e+00
3.90086591e-01 -6.32988513e-01 -6.66017592e-01 -2.00620741e-02
4.33737218e-01 -8.04573774e-01 -1.51513219e-01 -1.55524230e+00
4.09658074e-01 3.90790291e-02 -6.67087913e-01 5.05705357e-01
-1.71042651e-01 6.03621066e-01 -1.51573196e-01 -3.29795957e-01
-4.38147604e-01 1.07849514e+00 6.94260657e-01 3.11501861e-01
-2.44122252e-01 -1.77203208e-01 -1.25473738e+00 6.55955791e-01
1.24113166e+00 -2.62258083e-01 -6.71468973e-01 -2.10356507e-02
4.48775530e-01 -8.11703727e-02 1.84407815e-01 -9.87070262e-01
1.49192996e-02 -1.16143085e-03 8.18117619e-01 -1.47637859e-01
7.08523273e-01 -7.26485074e-01 4.19417322e-01 6.03720918e-02
-1.49844691e-01 1.83560327e-01 5.79460800e-01 9.67254415e-02
-9.82718691e-02 -9.49229226e-02 7.03014910e-01 9.42111760e-02
-6.32042646e-01 2.45917588e-01 -9.87785280e-01 -3.22001100e-01
1.04297018e+00 -1.78225785e-01 -1.99240059e-01 -5.13954759e-01
-9.55932975e-01 -1.38441429e-01 -8.97916853e-02 5.48741579e-01
9.59929407e-01 -1.22674787e+00 -2.03026280e-01 3.70402575e-01
9.60711911e-02 -9.15379047e-01 8.23635995e-01 1.27085137e+00
2.29052544e-01 5.04701257e-01 -7.56097138e-01 -1.64196357e-01
-1.17220986e+00 3.19962561e-01 6.41919076e-01 3.11581194e-01
-1.41590208e-01 6.98073089e-01 7.04950690e-02 6.45538196e-02
4.81793657e-02 9.01101828e-02 -4.84844446e-01 2.27787971e-01
6.24863446e-01 2.66770989e-01 4.92381126e-01 -3.95056307e-01
-5.60013831e-01 -3.24093737e-02 3.18876117e-01 -2.24086776e-01
1.63554323e+00 -2.88240820e-01 -4.91538972e-01 6.77223802e-01
6.81946635e-01 -4.11624253e-01 -6.69467330e-01 4.11277175e-01
-5.32486141e-01 1.60328194e-01 2.83972144e-01 -1.20349741e+00
-1.12270510e+00 9.68777776e-01 8.45258832e-01 -1.47992015e-01
1.60135412e+00 -1.38173569e-02 3.16147774e-01 2.00846374e-01
5.71883142e-01 -1.28969848e+00 7.09812790e-02 2.77304381e-01
1.06500137e+00 -8.56959343e-01 -2.13424087e-01 -1.39881596e-01
-1.01094079e+00 9.55914021e-01 5.35657585e-01 -3.16560902e-02
7.71342099e-01 3.79042655e-01 2.98617706e-02 -5.90442359e-01
-9.27747786e-01 -1.68928534e-01 4.51334864e-01 7.80459166e-01
9.34515595e-01 1.90574214e-01 -5.54106355e-01 1.49625933e+00
-2.28551552e-01 4.09451991e-01 8.55984464e-02 5.96368313e-01
-4.46004458e-02 -9.33950424e-01 -1.38075858e-01 7.27546573e-01
-5.51231921e-01 -2.28090808e-01 -6.93312168e-01 8.53317082e-02
-3.54017206e-02 1.16614783e+00 -9.27201137e-02 -4.19075489e-01
3.29272419e-01 7.00257778e-01 7.99567878e-01 -5.03240764e-01
-8.28610241e-01 2.70053476e-01 7.92570412e-02 -4.90504146e-01
-7.24916339e-01 -7.70149589e-01 -1.21041715e+00 5.03957458e-02
-3.76367569e-01 1.97993040e-01 1.21594965e+00 9.01698709e-01
9.01899278e-01 9.01647747e-01 2.88213670e-01 -7.75694966e-01
3.29671502e-01 -1.16181171e+00 -7.98585832e-01 5.20957746e-02
-1.87393352e-01 -8.98563504e-01 -1.53600663e-01 2.73685157e-02] | [13.19323444366455, 3.419835329055786] |
4373669e-cb0e-4d1b-b539-ae009bf2f8fe | semimultipose-a-semi-supervised-multi-animal | 2204.07072 | null | https://arxiv.org/abs/2204.07072v1 | https://arxiv.org/pdf/2204.07072v1.pdf | SemiMultiPose: A Semi-supervised Multi-animal Pose Estimation Framework | Multi-animal pose estimation is essential for studying animals' social behaviors in neuroscience and neuroethology. Advanced approaches have been proposed to support multi-animal estimation and achieve state-of-the-art performance. However, these models rarely exploit unlabeled data during training even though real world applications have exponentially more unlabeled frames than labeled frames. Manually adding dense annotations for a large number of images or videos is costly and labor-intensive, especially for multiple instances. Given these deficiencies, we propose a novel semi-supervised architecture for multi-animal pose estimation, leveraging the abundant structures pervasive in unlabeled frames in behavior videos to enhance training, which is critical for sparsely-labeled problems. The resulting algorithm will provide superior multi-animal pose estimation results on three animal experiments compared to the state-of-the-art baseline and exhibits more predictive power in sparsely-labeled data regimes. | ['Anqi Wu', 'Liam Paninski', 'Andres Bendesky', 'Christoph Gebhardt', 'Ari Blau'] | 2022-04-14 | null | null | null | null | ['animal-pose-estimation'] | ['computer-vision'] | [ 1.09244630e-01 -2.73931593e-01 -3.44583213e-01 -6.95517004e-01
-6.60693347e-01 -4.60088104e-01 1.02003366e-01 -1.88800409e-01
-8.96384358e-01 8.78365755e-01 -1.64119437e-01 5.57167768e-01
1.78995207e-01 4.23273072e-02 -1.03085589e+00 -4.72279429e-01
-3.91455710e-01 6.64932609e-01 1.78174958e-01 7.51486942e-02
-8.30657780e-02 4.57072318e-01 -1.50631523e+00 2.43777543e-01
2.85619944e-01 6.96514726e-01 4.26056594e-01 6.28279686e-01
3.41906190e-01 5.36029518e-01 -2.39728168e-01 -5.14829278e-01
2.34941110e-01 -3.48321736e-01 -5.21221340e-01 1.05826944e-01
9.68534768e-01 -7.72567511e-01 -2.49987170e-01 9.71664011e-01
3.18469584e-01 4.41480987e-02 5.96302569e-01 -1.28996885e+00
-1.79333225e-01 4.08327013e-01 -1.10071230e+00 6.60131395e-01
2.13178307e-01 7.75670856e-02 8.53043199e-01 -6.13657892e-01
8.03891778e-01 1.24146080e+00 6.11423016e-01 8.36336315e-01
-1.56098652e+00 -9.22087014e-01 1.61918789e-01 3.24541628e-01
-1.07282925e+00 -5.14544308e-01 7.10579336e-01 -5.08262873e-01
4.80690211e-01 -1.65336598e-02 9.25320566e-01 1.48586309e+00
2.16850102e-01 9.34344947e-01 1.05165958e+00 1.83623537e-01
2.40455917e-03 -2.13707224e-01 -1.53343916e-01 7.75261641e-01
3.48614067e-01 1.90758526e-01 -6.81150734e-01 -1.80643544e-01
1.03110719e+00 2.94492215e-01 2.04026476e-01 -6.42271042e-01
-1.35305774e+00 8.87349606e-01 4.11022723e-01 1.04936741e-01
-2.95375109e-01 3.94036591e-01 2.88058162e-01 3.23501900e-02
5.00397146e-01 5.33815742e-01 -4.96316940e-01 -1.00203604e-01
-1.30653906e+00 6.49415374e-01 4.37868446e-01 1.01726282e+00
6.67343616e-01 9.22907367e-02 -4.26559895e-02 1.04126024e+00
4.45920169e-01 5.32213032e-01 4.67019886e-01 -1.19497645e+00
2.21904263e-01 4.30446684e-01 -8.68159235e-02 -8.21748316e-01
-8.28193784e-01 -5.81344962e-01 -5.51672816e-01 -8.76777470e-02
5.48122585e-01 -2.02301040e-01 -8.78489256e-01 1.89719474e+00
7.38272607e-01 4.48179781e-01 -4.88503665e-01 1.02940154e+00
9.02457535e-01 3.26463610e-01 2.90378690e-01 -2.11954385e-01
1.36380148e+00 -1.14676249e+00 -5.03156006e-01 -7.31333137e-01
5.13372719e-01 -6.18643820e-01 4.34537202e-01 1.50931850e-01
-9.53419507e-01 -3.62883329e-01 -9.44008887e-01 4.63349409e-02
6.78168237e-02 3.25556189e-01 9.83005822e-01 4.00719583e-01
-5.66173494e-01 6.21771872e-01 -1.15632451e+00 -2.26312995e-01
1.01628196e+00 4.78844464e-01 -7.69643843e-01 -3.21590342e-02
-4.75546032e-01 1.08435571e+00 -1.73593462e-02 2.25942969e-01
-1.42848992e+00 -7.85915077e-01 -9.94757354e-01 -3.55581492e-01
4.78729427e-01 -3.95251900e-01 1.17280221e+00 -6.04985476e-01
-1.19443977e+00 1.19926834e+00 1.12048976e-01 -6.32426381e-01
6.65165901e-01 -5.01828313e-01 -5.08266613e-02 5.11340261e-01
2.82287627e-01 1.36589265e+00 9.56857860e-01 -1.16018355e+00
-4.19310480e-01 -8.14949334e-01 1.02762729e-01 1.88068211e-01
-1.65656865e-01 1.69375449e-01 -3.99039239e-01 -8.91483009e-01
1.34538546e-01 -1.44060969e+00 -5.68136990e-01 5.74121416e-01
9.16072279e-02 1.79500788e-01 9.56075013e-01 -5.45065939e-01
6.44415200e-01 -1.83909941e+00 4.79790896e-01 -4.83022690e-01
4.56970334e-01 3.09591591e-01 -3.78041238e-01 3.03337667e-02
7.17231706e-02 -4.88394737e-01 -1.20685630e-01 -5.47953248e-01
-3.40484470e-01 4.11186308e-01 2.09700376e-01 1.19556642e+00
1.75349787e-01 8.59906375e-01 -8.33880424e-01 -9.50791895e-01
3.19240421e-01 2.93893605e-01 -1.11869574e+00 3.44012707e-01
-1.07158251e-01 9.66001630e-01 -5.45937538e-01 7.88226008e-01
5.25060058e-01 -3.68212432e-01 2.35479563e-01 -3.34849983e-01
2.36170366e-01 -4.83280540e-01 -8.28374982e-01 1.95006979e+00
-1.45951897e-01 4.06704575e-01 2.60221004e-01 -1.14417315e+00
2.33570546e-01 3.43960002e-02 1.01203179e+00 -1.77415580e-01
4.33337033e-01 5.26101589e-02 1.96791768e-01 -4.89910334e-01
2.64989227e-01 -1.34280622e-01 8.09918121e-02 2.05454782e-01
6.65652931e-01 -3.37468326e-01 5.52650988e-01 6.41706958e-02
9.12841260e-01 7.31087744e-01 4.38611209e-01 -1.36837766e-01
2.27265313e-01 -7.02245981e-02 7.98447251e-01 4.43246067e-01
-6.60519123e-01 4.48963672e-01 8.17153230e-02 -5.91457367e-01
-1.03628147e+00 -5.72084665e-01 -3.90558630e-01 1.50952625e+00
2.48276189e-01 -2.33406663e-01 -8.18996608e-01 -7.03308702e-01
7.60490522e-02 5.67301102e-02 -7.55020857e-01 3.07602622e-02
-6.78464532e-01 -8.71443331e-01 3.64852548e-01 5.84706724e-01
1.16750106e-01 -1.08377314e+00 -8.63661230e-01 2.51405060e-01
-2.43360534e-01 -1.72340250e+00 -3.55681568e-01 2.20651180e-01
-6.69225335e-01 -8.58249485e-01 -9.13774550e-01 -7.55607128e-01
7.03329027e-01 3.85380924e-01 7.72882342e-01 -3.91710401e-02
-6.80175543e-01 6.37286454e-02 -3.53719950e-01 -3.43699664e-01
-8.10893103e-02 -2.69282937e-01 4.37395304e-01 -2.42438242e-01
-5.04852235e-02 -7.31927454e-01 -7.10017741e-01 6.63570225e-01
-7.39892662e-01 6.54107407e-02 7.12169766e-01 9.51791167e-01
7.43247151e-01 -6.93641722e-01 5.73058367e-01 -8.43794048e-01
-3.48175973e-01 -4.53166902e-01 -7.37410665e-01 2.49024481e-02
2.22232729e-01 -1.59658059e-01 6.52492344e-01 -8.73899758e-01
-5.81488192e-01 4.09788430e-01 -1.02324210e-01 -6.56921983e-01
-2.45789349e-01 2.37581074e-01 3.57063949e-01 -8.21480811e-01
3.74530256e-01 -2.42270738e-01 5.66876344e-02 -4.05482590e-01
1.43776640e-01 1.96555838e-01 4.54221904e-01 -4.80819434e-01
6.18520677e-01 6.28782690e-01 3.27610731e-01 -9.90919888e-01
-1.20723939e+00 -7.41158009e-01 -8.11456442e-01 -5.15665531e-01
9.51365650e-01 -1.17084754e+00 -7.67361045e-01 4.37363029e-01
-9.71186578e-01 -2.15304598e-01 1.44008428e-01 1.00506818e+00
-8.19766819e-01 5.05259097e-01 -6.94643915e-01 -6.71439588e-01
-1.51166841e-01 -1.50687599e+00 1.40868843e+00 1.17259525e-01
-3.63122165e-01 -5.14183521e-01 7.08358660e-02 9.82248247e-01
6.38157278e-02 2.13291019e-01 3.80973071e-01 -7.83363044e-01
-3.84694517e-01 -3.51522088e-01 -1.07498452e-01 1.68709978e-02
-3.64121169e-01 -1.69761628e-01 -7.79641867e-01 -6.34085000e-01
-4.33385856e-02 -9.83813763e-01 5.06261766e-01 5.99166453e-01
1.27849388e+00 -5.82404695e-02 -3.91792715e-01 7.35297441e-01
1.00175309e+00 -1.37863100e-01 -1.22421242e-01 -9.95196588e-03
8.06031287e-01 7.66600728e-01 7.55843282e-01 4.58050191e-01
4.35876459e-01 1.02342522e+00 5.95052958e-01 1.36766788e-02
-2.60457508e-02 -1.44395962e-01 2.55670696e-01 7.62156427e-01
-1.49168193e-01 8.25863555e-02 -6.44230127e-01 4.24076676e-01
-1.76064706e+00 -1.10402369e+00 1.70220703e-01 1.83405745e+00
7.64919758e-01 -2.08025053e-01 4.96211112e-01 -3.73655707e-01
6.72561884e-01 2.19056964e-01 -8.91266704e-01 4.14772302e-01
-7.84083828e-03 -1.44025743e-01 6.20543659e-01 -1.79458037e-01
-1.40962696e+00 8.83255839e-01 7.19916821e+00 8.09078932e-01
-9.14087176e-01 5.73907308e-02 5.16443551e-01 -5.02939820e-01
3.39822620e-01 -3.98923576e-01 -9.76567090e-01 3.80245149e-01
6.61405027e-01 2.78490305e-01 4.24478173e-01 1.18274379e+00
3.51444408e-02 -2.96860129e-01 -1.36101162e+00 1.30369771e+00
4.36595589e-01 -1.12004292e+00 -1.92207038e-01 1.56752080e-01
8.90235484e-01 3.23634744e-01 1.86339170e-02 1.27121344e-01
2.70430624e-01 -7.39661455e-01 8.90254557e-01 -1.65887065e-02
5.47390223e-01 -4.69985366e-01 5.72799683e-01 5.74045420e-01
-1.09646523e+00 -1.13249980e-01 -4.74356741e-01 2.51320869e-01
2.58823156e-01 6.79943636e-02 -4.84893024e-01 2.23636925e-01
8.70755672e-01 8.84736657e-01 -8.16376925e-01 1.16266942e+00
2.07431033e-01 4.11689639e-01 -5.76722026e-01 -1.85271408e-02
3.06852788e-01 -2.10258961e-01 6.13730490e-01 8.63403678e-01
9.46332440e-02 9.78420898e-02 4.69729245e-01 4.09531623e-01
-1.17187761e-01 9.11759511e-02 -3.91013622e-01 -2.78306037e-01
1.19100891e-01 1.56147206e+00 -1.12005377e+00 -2.31668368e-01
-6.24012470e-01 7.36530304e-01 6.55647695e-01 -9.26078334e-02
-1.20916200e+00 4.77466822e-01 5.60867548e-01 7.18811080e-02
3.83845657e-01 -2.69242138e-01 1.48440495e-01 -1.32307422e+00
-2.08171606e-01 -9.69880283e-01 2.72475213e-01 -5.07580757e-01
-1.31572270e+00 4.54871565e-01 4.59944308e-01 -1.53433418e+00
-2.78779447e-01 -5.68015575e-01 1.74366180e-02 -1.29965395e-01
-1.19555938e+00 -1.64208984e+00 -1.22413635e-02 3.32650006e-01
7.48438776e-01 -2.14013264e-01 6.34561121e-01 6.34529233e-01
-5.32337189e-01 4.98426497e-01 1.59236100e-02 -3.20181027e-02
6.63745284e-01 -7.65278280e-01 -4.04222272e-02 7.93849289e-01
5.73496521e-01 5.02233446e-01 9.46139693e-01 -6.61996186e-01
-1.51713049e+00 -9.28583801e-01 1.11392610e-01 -4.05912638e-01
7.74380326e-01 -3.54200274e-01 -5.01739144e-01 8.42248797e-01
-2.64791548e-01 5.53974628e-01 9.75124538e-01 -1.50524070e-02
-3.78699563e-02 1.40131474e-01 -1.15254164e+00 6.37423992e-01
1.15863574e+00 -3.58859956e-01 -4.90872413e-01 6.26808286e-01
4.15381998e-01 -6.37096882e-01 -7.82519698e-01 5.93627095e-01
1.03464544e+00 -6.09748781e-01 1.15897930e+00 -7.24496245e-01
7.09511697e-01 -3.10374945e-01 -1.49004981e-01 -1.15292001e+00
-2.78871149e-01 -1.62381873e-01 -1.31484404e-01 7.72459209e-01
1.59401506e-01 5.91549044e-03 1.11380017e+00 2.60336548e-01
-2.45938660e-03 -7.53303111e-01 -9.96204078e-01 -6.79698110e-01
-1.49375275e-01 -1.39755964e-01 7.80917332e-02 7.85027444e-01
-3.00274730e-01 3.83008242e-01 -9.95306432e-01 -1.05868690e-02
1.04250133e+00 3.28466892e-01 9.99509990e-01 -1.13609517e+00
-2.40785286e-01 -3.41982208e-02 -9.43494141e-01 -1.14467227e+00
5.15228033e-01 -7.43319154e-01 2.66537696e-01 -1.00014782e+00
6.48263037e-01 -2.11409088e-02 1.05035596e-01 5.30480206e-01
-1.50236741e-01 8.76389086e-01 3.48778479e-02 1.68088898e-01
-1.10761356e+00 5.75187743e-01 1.68279099e+00 -4.82362928e-03
2.83653826e-01 -9.33770537e-02 -2.71685064e-01 1.17429388e+00
2.44988352e-01 -6.41430318e-01 -2.30016738e-01 -3.93946946e-01
-1.50716826e-01 2.79099375e-01 5.50177097e-01 -1.20166779e+00
6.58912808e-02 -2.83480227e-01 5.94205558e-01 -6.90089524e-01
8.09206009e-01 -1.02590406e+00 3.79758507e-01 3.39659035e-01
-3.81894559e-01 -1.27056450e-01 9.14919972e-02 7.82680213e-01
-5.81405610e-02 -1.76291466e-01 1.18878043e+00 -3.63991082e-01
-5.27306437e-01 7.19674766e-01 -2.75287420e-01 2.42450565e-01
1.34413791e+00 -6.14059269e-02 -9.96246040e-02 -4.46198583e-02
-7.53248811e-01 3.40032101e-01 3.27033877e-01 3.80143642e-01
3.68396878e-01 -1.28616917e+00 -5.86297929e-01 1.74211577e-01
1.76638559e-01 -4.77906018e-02 6.67468727e-01 1.03337026e+00
-7.38507271e-01 1.92674965e-01 -7.19426274e-01 -8.07612777e-01
-1.61290395e+00 3.69156837e-01 -4.40599322e-02 -2.27223948e-01
-3.96620363e-01 1.02679944e+00 4.01762933e-01 -4.30047244e-01
1.50928944e-01 -1.90225497e-01 -3.70076925e-01 2.47792989e-01
4.13605779e-01 3.64539623e-01 -1.96945712e-01 -1.15368819e+00
-4.19152379e-01 7.15902269e-01 -3.50067556e-01 1.88204885e-01
1.84882545e+00 -8.19643289e-02 1.92805439e-01 3.55009675e-01
1.32119429e+00 -3.83103579e-01 -1.55942810e+00 -1.50899038e-01
-2.27675483e-01 -6.47150934e-01 5.36004454e-03 -1.62846252e-01
-1.15696144e+00 8.27708900e-01 3.53593469e-01 -4.95055109e-01
3.29525501e-01 9.93860886e-02 8.30204666e-01 5.05659282e-01
7.79093862e-01 -1.23951268e+00 3.84368986e-01 6.16931170e-02
8.24359953e-01 -1.72869885e+00 6.28211737e-01 -4.45809186e-01
-9.37965631e-01 5.80088019e-01 1.05264831e+00 -2.20288664e-01
5.68453670e-01 4.19131696e-01 -1.50773644e-01 -3.48020554e-01
-6.40986741e-01 -1.19068533e-01 3.99353057e-01 5.20348251e-01
3.02516669e-01 -7.52641931e-02 -1.64863572e-01 3.22312117e-01
-2.00793982e-01 -2.67309640e-02 1.51348099e-01 9.87250030e-01
-1.79252028e-01 -8.04114461e-01 -2.00602472e-01 7.32458174e-01
-9.21842396e-01 9.93623137e-02 -3.97318900e-02 5.44834018e-01
9.19271782e-02 6.02729142e-01 -1.87374115e-01 -6.74204454e-02
-5.74647710e-02 -4.12297517e-01 9.89871264e-01 -7.94906557e-01
-5.59305549e-01 4.89918858e-01 4.46824990e-02 -8.02542210e-01
-1.06184232e+00 -8.50154400e-01 -8.38972867e-01 -1.43232808e-01
-5.40041983e-01 -1.11054093e-01 5.41855276e-01 1.02590680e+00
-4.20295484e-02 3.07529569e-01 3.65306109e-01 -1.51679909e+00
-3.68073851e-01 -1.01895559e+00 -6.10205650e-01 4.89065349e-01
3.65064532e-01 -1.20769382e+00 -2.64926642e-01 3.03396612e-01] | [7.604117393493652, -0.9115826487541199] |
90f54964-a243-46e6-a7d5-3e6a05398bb7 | coarse-to-fine-person-re-identification-with | null | null | http://openaccess.thecvf.com//content/CVPR2021/html/Zhang_Coarse-To-Fine_Person_Re-Identification_With_Auxiliary-Domain_Classification_and_Second-Order_Information_Bottleneck_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Zhang_Coarse-To-Fine_Person_Re-Identification_With_Auxiliary-Domain_Classification_and_Second-Order_Information_Bottleneck_CVPR_2021_paper.pdf | Coarse-To-Fine Person Re-Identification With Auxiliary-Domain Classification and Second-Order Information Bottleneck | Person re-identification (Re-ID) is to retrieve a particular person captured by different cameras, which is of great significance for security surveillance and pedestrian behavior analysis. However, due to the large intra-class variation of a person across cameras, e.g., occlusions, illuminations, viewpoints, and poses, Re-ID is still a challenging task in the field of computer vision. In this paper, to attack the issues concerning with intra-class variation, we propose a coarse-to-fine Re-ID framework with the incorporation of auxiliary-domain classification (ADC) and second-order information bottleneck (2O-IB). In particular, as an auxiliary task, ADC is introduced to extract the coarse-grained essential features to distinguish a person from miscellaneous backgrounds, which leads to the effective coarse- and fine-grained feature representations for Re-ID. On the other hand, to cope with the redundancy, irrelevance, and noise contained in the Re-ID features caused by intra-class variations, we integrate 2O-IB into the network to compress and optimize the features, without increasing additional computation overhead during inference. Experimental results demonstrate that our proposed method significantly reduces the neural network output variance of intra-class person images and achieves the superior performance to state-of-the-art methods. | ['Yongcheng Zhou', 'Wenxi Liu', 'Yuzhen Niu', 'Yueming Gao', 'Anguo Zhang'] | 2021-06-19 | null | null | null | cvpr-2021-1 | ['miscellaneous'] | ['miscellaneous'] | [ 1.08119287e-02 -7.41216004e-01 2.21513584e-01 -4.16951150e-01
-1.96479827e-01 -3.16599250e-01 3.89871508e-01 -3.86641733e-02
-5.31639099e-01 6.00595653e-01 1.50925443e-01 3.37383717e-01
-1.88481942e-01 -6.15379870e-01 -3.53506237e-01 -7.60817885e-01
3.02260727e-01 1.47273391e-01 2.65544653e-01 9.79756266e-02
6.92873374e-02 5.82141578e-01 -1.73706973e+00 5.25423698e-02
8.64773035e-01 1.02445114e+00 2.72600390e-02 3.83063227e-01
4.06890698e-02 3.46157044e-01 -7.71977127e-01 -6.48032427e-01
2.70672709e-01 -3.25773627e-01 -1.69559404e-01 3.65053296e-01
5.94389498e-01 -5.28266191e-01 -5.84195971e-01 1.39931154e+00
6.19779229e-01 3.48275989e-01 5.31247199e-01 -1.28157640e+00
-5.07810831e-01 -1.59897372e-01 -6.66777015e-01 4.35317665e-01
2.79972374e-01 2.68810838e-01 2.80618578e-01 -8.23524535e-01
1.55468091e-01 1.62519562e+00 7.56851971e-01 5.85571289e-01
-1.02571070e+00 -8.69928300e-01 2.61950701e-01 6.03206396e-01
-1.71552432e+00 -4.38476264e-01 8.48440289e-01 -4.94139612e-01
2.92037964e-01 3.23687643e-01 6.10566854e-01 9.76966023e-01
-1.65633872e-01 5.00562072e-01 9.20998335e-01 -8.31342712e-02
-3.39384214e-03 3.09027255e-01 3.86624634e-01 5.18275678e-01
7.53442228e-01 1.03450313e-01 -2.80632675e-01 -1.25954696e-03
7.44518936e-01 4.91272688e-01 -3.56237769e-01 -8.59550834e-02
-9.70991075e-01 3.84441286e-01 2.73525029e-01 -2.66239177e-02
-1.60232753e-01 -3.47432107e-01 4.91687596e-01 4.24021482e-02
7.84860328e-02 -1.13114521e-01 -2.40272775e-01 -9.99941528e-02
-5.75862885e-01 2.24478438e-01 5.37199676e-01 8.95156384e-01
7.27488160e-01 -1.27632618e-01 -4.41904247e-01 9.65312362e-01
1.68491989e-01 7.48067081e-01 5.91833115e-01 -5.96212983e-01
7.13608503e-01 7.94240117e-01 3.00799489e-01 -1.49347198e+00
-3.70550752e-01 -5.70211768e-01 -1.52321553e+00 -7.41226673e-02
4.95482892e-01 -5.28565049e-02 -5.75167596e-01 1.65181446e+00
4.65758264e-01 3.76561671e-01 -1.51121961e-02 1.13833582e+00
8.76215458e-01 4.43141580e-01 4.58605327e-02 -3.59797448e-01
1.66599286e+00 -8.52558970e-01 -5.45242131e-01 -3.30544978e-01
-9.84660313e-02 -5.80145776e-01 5.36574841e-01 1.68891430e-01
-7.45864749e-01 -1.08816826e+00 -8.64941359e-01 1.65801316e-01
-2.25857124e-01 3.98342311e-01 1.30408898e-01 8.11911345e-01
-4.70818937e-01 1.94869652e-01 -3.65939498e-01 -2.08138928e-01
4.58461374e-01 3.70110869e-01 -4.49334323e-01 -2.80201286e-01
-1.21642709e+00 5.49492061e-01 4.62718129e-01 3.99249762e-01
-5.82968891e-01 -6.27909303e-01 -7.12949872e-01 9.69807133e-02
4.75344658e-01 -5.16899347e-01 6.69938207e-01 -9.05783713e-01
-1.27950168e+00 5.46257854e-01 -5.14397800e-01 -8.88626873e-02
5.68972051e-01 -8.29631910e-02 -7.86367416e-01 2.67374534e-02
3.91186811e-02 1.96288779e-01 1.05173278e+00 -1.03153419e+00
-8.12323034e-01 -7.76709914e-01 -5.22179976e-02 2.63337761e-01
-5.59505820e-01 5.27220406e-02 -8.59960437e-01 -8.01801980e-01
-8.38379785e-02 -9.94866788e-01 2.60196999e-02 2.66858912e-03
-2.62032390e-01 -2.69762605e-01 7.62397408e-01 -8.91296148e-01
1.21872211e+00 -2.46583724e+00 7.44550526e-02 1.51804194e-01
2.35442549e-01 6.77745998e-01 3.86826359e-02 -1.03615128e-01
6.54938966e-02 -1.84759066e-01 7.42911845e-02 -2.83675879e-01
-1.18988007e-01 -6.48721755e-02 2.02065751e-01 4.57653493e-01
1.42353371e-01 4.59855795e-01 -6.38403058e-01 -5.83506107e-01
4.91083175e-01 5.84778368e-01 -3.12537223e-01 2.65816271e-01
4.61629748e-01 6.86285436e-01 -6.33322775e-01 6.56184852e-01
1.17531717e+00 -5.66937290e-02 -1.97444037e-01 -6.36876583e-01
-2.23013689e-03 -3.33043754e-01 -1.65374565e+00 1.16430938e+00
-9.02064145e-02 2.28097215e-01 8.66503865e-02 -1.02271056e+00
9.77281272e-01 8.44362453e-02 2.53506482e-01 -7.04036415e-01
1.51241437e-01 1.22946510e-02 -1.02010287e-01 -4.54387009e-01
2.56811172e-01 3.24238926e-01 -6.26201928e-02 -1.20306509e-02
-2.68937916e-01 7.47634351e-01 1.95353016e-01 -2.51332641e-01
5.47022104e-01 -2.85342216e-01 4.07810003e-01 -1.62862897e-01
1.26010919e+00 -5.45521021e-01 9.35645461e-01 6.15908921e-01
-5.43411314e-01 4.69299197e-01 2.17097215e-02 -7.49016047e-01
-8.78989220e-01 -8.53046477e-01 -1.19869202e-01 7.70748556e-01
6.66867733e-01 -6.59042448e-02 -7.78143048e-01 -3.77463818e-01
7.04373792e-02 1.44724503e-01 -3.78964335e-01 -3.13303411e-01
-5.98489642e-01 -1.07004595e+00 5.70691407e-01 5.17784595e-01
1.17935753e+00 -5.79419315e-01 -4.80184585e-01 1.85370535e-01
-3.73252273e-01 -1.27785528e+00 -8.41668069e-01 -4.37432468e-01
-6.00623548e-01 -1.11304247e+00 -9.94699895e-01 -7.24458933e-01
7.59356856e-01 6.52389228e-01 7.95232177e-01 6.21351376e-02
-3.60654712e-01 2.23305702e-01 -1.67876214e-01 -6.16810843e-02
9.27732512e-02 -1.99711338e-01 2.32175454e-01 5.87305367e-01
7.77415991e-01 -3.17902476e-01 -7.68012106e-01 6.52670622e-01
-6.32885695e-01 -6.73411340e-02 4.45598990e-01 9.52263117e-01
4.80371833e-01 5.86714208e-01 3.30636770e-01 -2.67288059e-01
4.64845151e-01 -1.79915592e-01 -7.25178838e-01 3.20656806e-01
-3.32394242e-01 -2.07226142e-01 8.14271092e-01 -6.46324217e-01
-1.18872225e+00 -6.31873757e-02 1.23175696e-01 -3.75757843e-01
-3.75064999e-01 4.27408703e-03 -7.59594202e-01 -1.01045206e-01
2.73319513e-01 4.66949701e-01 -6.05650768e-02 -5.50153196e-01
-1.48738131e-01 9.10250127e-01 7.02327490e-01 -4.18266535e-01
8.82310748e-01 4.16611314e-01 2.16742903e-01 -8.93550634e-01
-7.94389665e-01 -5.91723740e-01 -5.69663465e-01 -1.28467456e-01
9.38613474e-01 -1.20361233e+00 -1.01940632e+00 1.00917065e+00
-1.16181076e+00 4.08724964e-01 3.67320776e-02 5.46186626e-01
1.17066152e-01 6.78693175e-01 -4.43688422e-01 -8.36992800e-01
-3.63837987e-01 -1.31059384e+00 9.35990453e-01 8.30464363e-01
2.20178381e-01 -6.53892279e-01 -4.05306607e-01 4.16851282e-01
1.63087755e-01 9.64534208e-02 4.56802636e-01 -4.59186435e-01
-6.29102528e-01 -2.17407137e-01 -6.37674212e-01 4.44220454e-01
3.88884574e-01 -4.26677018e-01 -8.74179721e-01 -4.98968840e-01
-4.72919531e-02 4.98221591e-02 6.01215899e-01 2.25324780e-01
1.36186671e+00 -4.19468790e-01 -3.87629032e-01 8.83037448e-01
1.14763844e+00 3.27122867e-01 4.99363095e-01 3.95110071e-01
8.89887631e-01 5.75881243e-01 6.09954536e-01 7.60825217e-01
5.68980217e-01 9.06152070e-01 7.80477524e-02 1.61906853e-01
-2.70884484e-01 -4.63151932e-02 1.56447604e-01 4.78586942e-01
-2.65700161e-01 -1.87441751e-01 -4.59443539e-01 3.50370735e-01
-1.77342224e+00 -1.17479289e+00 -4.79495935e-02 2.48149419e+00
5.71480453e-01 -1.00404605e-01 2.46737868e-01 2.00239405e-01
1.34397531e+00 -2.48646550e-03 -8.09747994e-01 2.25732744e-01
-1.58479631e-01 -4.58124131e-01 4.70398515e-01 1.28200263e-01
-1.26167941e+00 4.91748393e-01 4.73760080e+00 1.05957484e+00
-9.32442129e-01 -4.83943261e-02 5.70977211e-01 1.54700384e-01
3.65050703e-01 -4.44754183e-01 -1.39617860e+00 1.07674539e+00
3.86468023e-01 -2.87077110e-03 7.17149675e-01 7.42131054e-01
5.08459024e-02 5.59384786e-02 -9.50376868e-01 1.67327631e+00
4.19165999e-01 -7.18006492e-01 -2.46779118e-02 -8.44132453e-02
5.54670453e-01 -5.72243690e-01 -7.29579572e-03 1.46819681e-01
-2.04222322e-01 -6.78825319e-01 5.66196799e-01 5.53602815e-01
8.64014924e-01 -9.87245202e-01 9.93878126e-01 3.77627909e-01
-1.59273839e+00 -3.33976477e-01 -6.62893116e-01 4.92225401e-02
2.05810126e-02 6.18284762e-01 -2.02193618e-01 6.11461699e-01
1.02820992e+00 7.92317271e-01 -7.83504546e-01 1.00942564e+00
1.21972345e-01 1.14609137e-01 -1.60655707e-01 1.35992289e-01
-2.50846505e-01 -2.30446205e-01 5.90545356e-01 1.18478763e+00
3.43542248e-01 2.83758491e-01 4.53975171e-01 4.76251394e-01
5.45696244e-02 -3.08697641e-01 -1.04833059e-01 4.10599828e-01
6.51205420e-01 1.01461649e+00 -2.86670119e-01 -5.57188392e-01
-4.68830734e-01 1.27631438e+00 1.38871670e-01 4.32015091e-01
-8.89942169e-01 -4.20673698e-01 8.84377003e-01 -5.96758872e-02
4.33879167e-01 -7.65669122e-02 7.81278461e-02 -1.38432133e+00
3.63261610e-01 -1.00363445e+00 4.91479456e-01 -2.99635082e-01
-1.62804210e+00 4.44111347e-01 7.74157792e-02 -1.38332772e+00
3.48186190e-03 -5.55500686e-01 -3.49009484e-01 1.06927931e+00
-1.56974316e+00 -1.13722467e+00 -9.62646842e-01 1.00290513e+00
4.95085090e-01 -4.08142477e-01 4.28957134e-01 7.64439285e-01
-9.52265382e-01 1.01555324e+00 1.59105763e-01 4.61534619e-01
7.50543773e-01 -6.58523142e-01 1.84106767e-01 9.87603962e-01
-5.02545178e-01 6.11203849e-01 4.01930094e-01 -6.04818404e-01
-1.44448054e+00 -1.40455425e+00 5.62142730e-01 -1.24426596e-01
1.19822703e-01 -1.03960566e-01 -7.70584702e-01 3.46810609e-01
-5.17868698e-01 2.93244183e-01 5.96537113e-01 -8.68992805e-02
-3.85791779e-01 -6.92210197e-01 -1.19902539e+00 5.46359777e-01
1.17459011e+00 -4.80433971e-01 -4.76484984e-01 3.35478690e-03
4.00643200e-01 -3.22577626e-01 -7.75838971e-01 3.29363078e-01
7.08953857e-01 -1.02986574e+00 1.45954633e+00 -1.86305836e-01
-1.76913649e-01 -6.46676302e-01 -8.77473578e-02 -1.07201517e+00
-6.94134414e-01 -2.80525416e-01 -3.83749008e-02 1.57432806e+00
-4.28898543e-01 -9.41431940e-01 5.34616590e-01 7.89299130e-01
2.47955680e-01 -1.48069382e-01 -1.04287839e+00 -9.15648699e-01
-4.55899209e-01 5.69704771e-02 9.15670514e-01 5.70032060e-01
-5.48944175e-01 9.30829197e-02 -7.73760021e-01 5.48087120e-01
1.02212048e+00 7.97857493e-02 1.03553033e+00 -1.56016791e+00
-2.30387285e-01 -2.15571761e-01 -7.53521442e-01 -1.21099675e+00
-1.89616401e-02 -3.98090035e-01 -5.91922104e-02 -1.09315860e+00
5.16177475e-01 -4.76712286e-01 -3.86594206e-01 7.14358911e-02
-5.96718788e-01 2.69298375e-01 3.96607846e-01 4.59547341e-01
-6.53985679e-01 5.79925776e-01 1.14055133e+00 -3.87920290e-01
-1.71400189e-01 1.38790533e-01 -6.98334038e-01 8.09800208e-01
5.44831693e-01 -3.46054465e-01 -2.42724583e-01 -5.41420102e-01
-3.59615773e-01 -1.02725998e-02 9.28608179e-01 -1.37191916e+00
4.43016410e-01 -1.07913129e-01 9.43426549e-01 -6.52660429e-01
5.26911020e-01 -1.08787775e+00 2.43610755e-01 4.87271428e-01
2.96059530e-02 2.62067541e-02 9.81627554e-02 7.83315718e-01
-3.03385347e-01 -1.82682976e-01 8.22655022e-01 -2.09680155e-01
-8.04351211e-01 5.42424977e-01 -8.17538947e-02 1.26682892e-01
9.56665039e-01 -5.05859375e-01 -3.82046878e-01 -1.12225756e-01
-3.17240477e-01 1.76913172e-01 4.39058721e-01 4.19366181e-01
5.90915501e-01 -1.39042652e+00 -7.26425231e-01 5.12560725e-01
1.37424067e-01 5.84975369e-02 7.91168571e-01 6.15488589e-01
-5.80564253e-02 3.80044401e-01 -3.73649865e-01 -6.33277714e-01
-1.44220877e+00 6.93658710e-01 2.59776741e-01 -2.99540937e-01
-4.98124063e-01 5.86745083e-01 4.18992698e-01 -1.58927888e-01
2.61201560e-01 9.72229913e-02 -4.40665334e-01 9.97150019e-02
9.29633915e-01 7.84773767e-01 -2.03031376e-01 -9.64343846e-01
-6.08986199e-01 9.91315663e-01 -7.77205154e-02 4.03106123e-01
8.38667452e-01 -4.92667288e-01 -3.84428687e-02 9.25844386e-02
1.08452725e+00 -1.09633937e-01 -1.44261932e+00 -3.65394890e-01
-4.05301005e-01 -8.46423030e-01 -1.88323021e-01 -5.47569692e-01
-1.02459276e+00 8.65767241e-01 9.68082547e-01 -1.30296677e-01
1.24773300e+00 -4.76398736e-01 8.95517349e-01 4.17570949e-01
4.58424270e-01 -1.11946356e+00 3.00952289e-02 2.60269403e-01
5.30545056e-01 -1.29988706e+00 1.07708320e-01 -5.30474782e-01
-5.80460548e-01 1.00482023e+00 7.80921817e-01 1.44731313e-01
5.54065049e-01 -1.37795150e-01 -2.56830186e-01 3.57200921e-01
6.75318688e-02 -7.72752464e-02 4.53763843e-01 7.85792828e-01
-1.63639665e-01 4.98320349e-02 4.10844013e-02 8.32454085e-01
1.86788112e-01 2.03177676e-01 3.40978689e-02 4.66785461e-01
-2.83811450e-01 -8.47300947e-01 -8.05146039e-01 3.37367117e-01
-2.64835060e-01 8.97980779e-02 6.85086921e-02 5.08289933e-01
6.11072123e-01 1.21674991e+00 9.34437066e-02 -4.06319588e-01
3.49242985e-01 -3.45957786e-01 1.97196543e-01 -8.82018656e-02
-3.35154742e-01 -1.18061066e-01 -1.87711775e-01 -3.44354779e-01
-5.02596259e-01 -6.72918320e-01 -6.63857341e-01 -4.25655246e-01
-1.82641700e-01 -5.98170869e-02 2.73801118e-01 9.34570014e-01
5.19081473e-01 5.46203732e-01 6.60777390e-01 -8.12986314e-01
-6.76288366e-01 -7.97624588e-01 -4.41357046e-01 8.07729244e-01
4.85983938e-01 -8.16765606e-01 -3.40092242e-01 7.14391991e-02] | [14.723139762878418, 0.9724113941192627] |
43ce75b1-0ab0-4801-b463-9bb4288688c9 | contrastive-training-improves-zero-shot | 2210.05613 | null | https://arxiv.org/abs/2210.05613v1 | https://arxiv.org/pdf/2210.05613v1.pdf | Contrastive Training Improves Zero-Shot Classification of Semi-structured Documents | We investigate semi-structured document classification in a zero-shot setting. Classification of semi-structured documents is more challenging than that of standard unstructured documents, as positional, layout, and style information play a vital role in interpreting such documents. The standard classification setting where categories are fixed during both training and testing falls short in dynamic environments where new document categories could potentially emerge. We focus exclusively on the zero-shot setting where inference is done on new unseen classes. To address this task, we propose a matching-based approach that relies on a pairwise contrastive objective for both pretraining and fine-tuning. Our results show a significant boost in Macro F$_1$ from the proposed pretraining step in both supervised and unsupervised zero-shot settings. | ['Miguel Ballesteros', 'Sunil Mallya', 'Graham Horwood', 'Shuai Wang', 'Yogarshi Vyas', 'Muhammad Khalifa'] | 2022-10-11 | null | null | null | null | ['document-classification'] | ['natural-language-processing'] | [ 6.57755792e-01 -3.58116813e-02 2.68810112e-02 -8.35943401e-01
-7.43729532e-01 -7.48875380e-01 8.64496112e-01 4.19183671e-01
-4.66621906e-01 5.05132556e-01 1.32715791e-01 -2.86278009e-01
-2.67509550e-01 -6.16703212e-01 -7.01936066e-01 -5.94118476e-01
1.08272575e-01 7.73786545e-01 3.55981916e-01 -1.66859180e-01
4.13421482e-01 2.24540867e-02 -1.91306865e+00 4.96802837e-01
7.37958252e-01 8.75565290e-01 2.89162070e-01 9.43857908e-01
-3.89828026e-01 5.04892051e-01 -6.60595179e-01 -3.93545598e-01
1.16087899e-01 -3.84804130e-01 -8.68569732e-01 5.08994818e-01
6.78551853e-01 -5.62651455e-02 1.75028414e-01 1.01029217e+00
5.06197095e-01 6.15457654e-01 8.05648804e-01 -8.55515540e-01
-4.78314757e-01 8.71059418e-01 -4.93225098e-01 1.74191728e-01
2.80636102e-01 -1.13248274e-01 1.23869014e+00 -1.09688604e+00
7.20075965e-01 1.18242788e+00 3.36683661e-01 5.48516989e-01
-1.47687805e+00 -3.57951015e-01 4.70491499e-01 1.22721322e-01
-1.22702384e+00 -4.99499530e-01 6.96285844e-01 -6.16029263e-01
7.11587846e-01 2.39373401e-01 2.55071282e-01 1.18357897e+00
-3.71620916e-02 1.06134057e+00 8.34171116e-01 -8.80212665e-01
5.20079255e-01 3.09762269e-01 7.57334113e-01 3.65559071e-01
2.73984462e-01 -1.82284161e-01 -4.08121765e-01 5.45824841e-02
1.03113316e-01 5.89987785e-02 -2.52004899e-02 -7.09819496e-01
-9.82431591e-01 8.68193686e-01 -2.84185857e-02 2.60027975e-01
3.30901682e-01 -1.75014421e-01 3.77348840e-01 2.98005968e-01
4.64475334e-01 6.70051396e-01 -2.79220223e-01 -3.42879027e-01
-1.29224539e+00 1.88836277e-01 7.39003539e-01 1.12479734e+00
5.76443553e-01 -2.74196535e-01 -4.62193727e-01 1.11079729e+00
5.06130904e-02 1.74464434e-01 6.59551978e-01 -5.20661771e-01
6.35049939e-01 5.05004704e-01 2.76618376e-02 -6.27074778e-01
-1.95609704e-01 -4.78710175e-01 -6.52017772e-01 -8.32351148e-02
3.56369078e-01 5.78846559e-02 -1.38302958e+00 1.49127781e+00
2.83115625e-01 -1.49876565e-01 -1.24262534e-01 3.83633912e-01
5.71359515e-01 3.94943625e-01 -1.60826802e-01 -1.96818411e-01
1.08863175e+00 -1.16052771e+00 -6.33024693e-01 -3.76278520e-01
5.98658919e-01 -6.03219390e-01 1.49061489e+00 4.64052111e-01
-9.32405651e-01 -7.28013813e-01 -1.26999152e+00 -1.03695869e-01
-7.58159220e-01 -2.71087319e-01 3.62434894e-01 7.26738393e-01
-7.50718594e-01 7.76064396e-01 -6.08425319e-01 -4.18880969e-01
4.38225985e-01 3.84346753e-01 -2.01997370e-01 -3.31730038e-01
-1.00826502e+00 3.91599417e-01 4.29830462e-01 -6.73187003e-02
-6.15926862e-01 -3.97508323e-01 -8.41357112e-01 3.30590427e-01
6.41226947e-01 -2.93897152e-01 1.40758336e+00 -6.40489697e-01
-1.26919830e+00 7.84854412e-01 -1.68516561e-01 -2.24939302e-01
5.57163715e-01 -3.28930579e-02 -1.04177773e-01 -1.50414988e-01
8.40752125e-02 3.39323401e-01 9.36727524e-01 -1.29111826e+00
-6.80448413e-01 -4.94423121e-01 -3.15482356e-03 1.12543344e-01
-6.04613006e-01 -1.17749088e-01 -5.56793690e-01 -7.76646137e-01
1.06795438e-01 -7.61170864e-01 -3.11139375e-02 -1.76459432e-01
-5.18779933e-01 -1.91649318e-01 9.93058860e-01 -1.65070713e-01
1.30543447e+00 -2.03151011e+00 1.02518357e-01 2.99186945e-01
3.23612168e-02 3.54086637e-01 -1.19772837e-01 1.67057291e-01
1.11087069e-01 1.95407659e-01 -2.63347268e-01 -6.74702108e-01
1.38801172e-01 -1.14891329e-03 -3.59057903e-01 -4.25022468e-02
1.05255060e-02 8.68832409e-01 -9.59504187e-01 -5.29863000e-01
8.03324953e-02 1.20970465e-01 -6.43926024e-01 2.56187171e-01
-2.73929209e-01 8.32256377e-02 -1.44403413e-01 7.36870229e-01
4.88919109e-01 -4.56312567e-01 4.31381226e-01 3.56745958e-01
6.39809370e-02 6.10385137e-03 -1.37749779e+00 1.59974933e+00
-4.92013335e-01 5.91892600e-01 -2.01275572e-01 -1.19102347e+00
8.15508187e-01 -2.57234331e-02 -4.79473099e-02 -3.76809269e-01
2.35909313e-01 -2.10228905e-01 5.05824462e-02 -2.15761483e-01
7.27780282e-01 -9.86768827e-02 -2.34492660e-01 6.62367642e-01
3.30763072e-01 -8.85945186e-02 5.43572664e-01 3.12407106e-01
1.08421969e+00 9.14058164e-02 3.72622102e-01 -5.10091662e-01
1.11597776e-01 -2.27013484e-01 2.64439046e-01 1.30231106e+00
9.75580700e-03 8.52083385e-01 4.87577349e-01 -2.00513363e-01
-9.40822005e-01 -1.04680312e+00 -3.35883915e-01 1.61992455e+00
1.40996948e-01 -3.69128555e-01 -7.13639081e-01 -9.12667632e-01
-8.38786364e-04 7.01749086e-01 -9.00413692e-01 -3.23329777e-01
-8.71298909e-02 -6.49425626e-01 1.10359922e-01 7.44012535e-01
1.66886449e-01 -7.33735919e-01 -6.02705956e-01 2.05321535e-01
9.51925144e-02 -9.43135321e-01 -5.23475528e-01 8.44231129e-01
-7.56832123e-01 -9.93376255e-01 -6.26311779e-01 -8.82310569e-01
8.64224076e-01 4.18663889e-01 9.95362520e-01 1.06966570e-01
-3.52131575e-01 2.24684536e-01 -6.18797660e-01 -4.52180713e-01
-2.09919900e-01 3.62663180e-01 1.74486134e-02 1.52648896e-01
4.58304167e-01 -3.66389483e-01 -2.84550905e-01 4.49689150e-01
-9.51855600e-01 1.00972630e-01 3.81458521e-01 1.32471371e+00
3.14534128e-01 2.80523986e-01 2.08023742e-01 -1.56502056e+00
5.50203323e-01 -3.33220422e-01 -5.06917059e-01 6.29554749e-01
-7.77440846e-01 2.23117054e-01 8.02825034e-01 -5.43132901e-01
-1.26954353e+00 2.24043075e-02 1.87616453e-01 -1.61584049e-01
-2.60408401e-01 3.87718350e-01 -3.68896067e-01 1.65694237e-01
7.46509671e-01 6.37151971e-02 -4.31872517e-01 -5.70987225e-01
1.51519239e-01 1.05175018e+00 4.39269960e-01 -7.33893871e-01
9.50659096e-01 3.57738376e-01 -3.31103712e-01 -9.95832145e-01
-1.07737172e+00 -7.33241558e-01 -1.19843650e+00 1.21282656e-02
5.49949944e-01 -6.01203740e-01 -1.88686341e-01 3.10494572e-01
-8.34237158e-01 -5.39746106e-01 -5.22008479e-01 8.67134631e-02
-3.44444513e-01 4.03769463e-01 -2.26363584e-01 -7.74332464e-01
-1.45193830e-01 -1.10842276e+00 1.22660279e+00 1.24770589e-01
-3.86523813e-01 -1.22847021e+00 1.35344677e-02 2.67034441e-01
2.59530067e-01 -5.50522357e-02 1.05132616e+00 -9.85907495e-01
-2.77822465e-01 -5.00597000e-01 -1.11984707e-01 1.82642385e-01
1.78379551e-01 4.15757764e-03 -1.14601552e+00 -3.89442384e-01
-1.69296995e-01 -6.62164807e-01 9.86282170e-01 1.66730404e-01
1.36453652e+00 -3.55094485e-02 -3.13497484e-01 5.63627124e-01
1.13775027e+00 3.23867530e-01 3.34959418e-01 1.94955572e-01
7.00878620e-01 7.12955475e-01 7.94214845e-01 6.13194883e-01
4.56262119e-02 7.35007942e-01 -7.29996562e-02 1.71767727e-01
7.01641068e-02 -2.83016920e-01 -1.43730357e-01 5.17638862e-01
1.81356058e-01 -7.65065730e-01 -1.14129293e+00 5.31161785e-01
-1.86342561e+00 -1.02173960e+00 2.61438012e-01 2.42815018e+00
9.22499001e-01 6.24342084e-01 -1.89003587e-01 4.36163843e-01
8.14240277e-01 1.86392486e-01 -4.72697318e-01 -2.39417374e-01
-6.33743107e-02 2.90028661e-01 1.05828783e-02 5.48912048e-01
-1.24676085e+00 9.07747984e-01 6.74050093e+00 7.29299843e-01
-9.14959371e-01 -1.38243631e-01 7.04147816e-01 -2.54833728e-01
-3.22723448e-01 -5.00896871e-02 -1.13774514e+00 4.37798142e-01
8.66034269e-01 -9.26618874e-02 2.28260517e-01 9.88725781e-01
-6.20296635e-02 -2.25924492e-01 -1.44908404e+00 8.50318491e-01
3.80850464e-01 -1.19989681e+00 1.88113987e-01 4.88903224e-02
8.56338084e-01 -4.06486005e-01 1.78606093e-01 5.95601261e-01
4.79964614e-01 -9.03181553e-01 7.06708848e-01 2.52514899e-01
9.55481470e-01 -5.63230455e-01 5.57051182e-01 3.49782348e-01
-1.00890625e+00 -1.17570214e-01 -3.41318816e-01 -1.91512015e-02
-3.90261807e-03 4.48554188e-01 -7.89459646e-01 8.00472200e-02
4.98729110e-01 6.84525549e-01 -9.24044788e-01 7.89884567e-01
-2.13644058e-01 5.48934221e-01 -8.67344216e-02 -1.27838269e-01
6.69011325e-02 2.39206210e-01 3.07813406e-01 1.33794963e+00
1.57542184e-01 4.41191671e-03 3.08318377e-01 4.76500005e-01
1.95204057e-02 4.22062092e-02 -4.49785829e-01 -1.71022385e-01
3.85293156e-01 1.14102840e+00 -1.14932990e+00 -5.45948505e-01
-3.47355336e-01 1.02084255e+00 4.56253111e-01 1.92816257e-01
-4.37780082e-01 -8.17400873e-01 3.72434378e-01 2.58215547e-01
6.34215176e-01 -8.49349722e-02 -2.93808579e-01 -1.45549476e+00
-8.84198919e-02 -6.28537059e-01 5.35320282e-01 -4.56578910e-01
-1.14068127e+00 5.11207402e-01 1.12233438e-01 -1.25620949e+00
-5.53190410e-01 -7.30544031e-01 -6.60084188e-01 3.18886280e-01
-1.13945210e+00 -9.10470247e-01 -3.46138805e-01 3.17123532e-01
1.02966893e+00 -1.70807987e-01 8.38107109e-01 -7.57318288e-02
-5.69109440e-01 8.84652376e-01 6.56463802e-01 1.32841408e-01
8.13131571e-01 -1.70332813e+00 4.67159122e-01 1.15324974e+00
3.85414571e-01 8.57026458e-01 8.11004162e-01 -4.54932809e-01
-1.16033077e+00 -1.03684723e+00 8.06379616e-01 -7.02632964e-01
5.70770264e-01 -9.85569775e-01 -9.43897367e-01 5.43704450e-01
-1.39102817e-01 1.28435150e-01 1.05628288e+00 5.62851131e-01
-4.84466106e-01 1.20142147e-01 -7.72533894e-01 5.13910890e-01
1.14949524e+00 -5.88408113e-01 -6.73298538e-01 2.28802174e-01
5.10330617e-01 -2.43115649e-01 -5.47127962e-01 2.11394683e-01
6.67083442e-01 -6.78593576e-01 6.69703066e-01 -7.83345282e-01
6.06603622e-01 2.21236702e-02 -3.32888693e-01 -1.29518354e+00
-3.74606520e-01 -5.77077568e-01 -2.73518175e-01 1.29080880e+00
5.78463197e-01 -1.00627899e-01 9.39683437e-01 7.18426228e-01
-3.55528854e-03 -5.45395434e-01 -4.88872916e-01 -9.46823418e-01
-1.06570356e-01 -3.89126658e-01 3.07548165e-01 8.70565236e-01
2.61676610e-01 8.07818890e-01 -4.07909542e-01 -3.52141619e-01
6.82231605e-01 3.58496666e-01 9.08557177e-01 -1.66102040e+00
-7.12922871e-01 -2.52486318e-01 -3.93007785e-01 -1.12276864e+00
8.27401504e-02 -8.13427627e-01 5.98315895e-01 -1.32504559e+00
6.26355648e-01 -2.79096961e-01 -3.66491705e-01 3.42124343e-01
-4.10316169e-01 1.51705801e-01 6.30274490e-02 3.70624177e-02
-1.00372183e+00 3.42522681e-01 8.23229194e-01 -3.89595330e-01
-3.38959396e-01 1.77781001e-01 -6.84791088e-01 5.17836392e-01
4.60179240e-01 -3.00951481e-01 -7.38379955e-01 -4.31467861e-01
3.08747427e-03 -3.06166202e-01 -1.88785225e-01 -9.95134175e-01
2.49494150e-01 -2.16392636e-01 5.06665409e-01 -5.65842152e-01
2.75381237e-01 -6.72088683e-01 -5.06698668e-01 1.31898358e-01
-8.12268198e-01 -3.06058973e-01 -2.26834528e-02 8.90156567e-01
-2.32227579e-01 -5.12470722e-01 8.42063248e-01 -1.50086388e-01
-7.45557070e-01 2.78705001e-01 -2.83810526e-01 3.93927008e-01
9.37134266e-01 -6.52432680e-01 -3.01752180e-01 -2.88300902e-01
-9.37554002e-01 1.90715030e-01 5.08787930e-01 5.90025246e-01
4.40281123e-01 -9.78213668e-01 -2.43689343e-01 4.05515105e-01
5.83226264e-01 1.81639373e-01 1.48902863e-01 4.45903391e-01
-6.51923120e-02 3.61380965e-01 -1.32506806e-02 -5.90646565e-01
-1.41824400e+00 6.21633053e-01 -4.67500240e-02 -4.41276312e-01
-3.47097993e-01 1.11292624e+00 3.86997789e-01 -3.71518224e-01
6.90896451e-01 -1.70222029e-01 -2.10476592e-01 4.21356648e-01
7.70818889e-01 1.33433685e-01 3.43140393e-01 -4.55820799e-01
-8.84627551e-02 2.26973340e-01 -7.20928609e-01 -3.71715039e-01
1.31203687e+00 -1.17343426e-01 3.60334188e-01 9.62215841e-01
1.22135377e+00 -2.47222319e-01 -1.23734188e+00 -4.47845906e-01
4.07535404e-01 -6.11540020e-01 4.38649580e-02 -6.74989223e-01
-3.45676512e-01 1.18502688e+00 4.86213386e-01 4.64318469e-02
6.83121026e-01 4.30841930e-02 4.83824492e-01 7.28668213e-01
2.71313399e-01 -1.53650665e+00 5.12796223e-01 8.06078553e-01
4.16524172e-01 -1.56345141e+00 -5.23393936e-02 -2.84611642e-01
-4.86533910e-01 1.09287357e+00 7.49496520e-01 3.90668869e-01
7.53256261e-01 2.44116977e-01 -1.55066222e-01 -2.34245006e-02
-9.35723543e-01 -1.17963627e-01 5.00861526e-01 4.96791899e-01
5.57714164e-01 -8.46944228e-02 9.38806385e-02 7.45137036e-01
-3.05609673e-01 -3.35424393e-01 4.85630900e-01 1.48452437e+00
-6.75926089e-01 -1.15934992e+00 -2.40172893e-01 8.02569747e-01
-1.90485045e-01 -1.32286176e-01 -4.95434314e-01 5.60117781e-01
-3.68342064e-02 8.21117222e-01 3.69382232e-01 -5.48328221e-01
2.49787286e-01 4.23724115e-01 4.30690378e-01 -1.35682857e+00
-2.41529346e-01 -1.70732867e-02 -1.34422898e-01 -1.40106738e-01
-1.31487444e-01 -7.68613398e-01 -9.15881515e-01 6.98578209e-02
-5.36985695e-01 2.90999025e-01 5.11973977e-01 1.10927713e+00
4.33025183e-03 5.77877522e-01 6.95818067e-01 -7.73995876e-01
-7.62067020e-01 -9.87141848e-01 -6.78261817e-01 5.67633808e-01
3.19142580e-01 -8.42840195e-01 -4.68895555e-01 1.98391318e-01] | [10.004424095153809, 3.4123926162719727] |
71fda749-5205-4114-b3e0-4da16586b5e8 | implicit-distributional-reinforcement | 2007.06159 | null | https://arxiv.org/abs/2007.06159v2 | https://arxiv.org/pdf/2007.06159v2.pdf | Implicit Distributional Reinforcement Learning | To improve the sample efficiency of policy-gradient based reinforcement learning algorithms, we propose implicit distributional actor-critic (IDAC) that consists of a distributional critic, built on two deep generator networks (DGNs), and a semi-implicit actor (SIA), powered by a flexible policy distribution. We adopt a distributional perspective on the discounted cumulative return and model it with a state-action-dependent implicit distribution, which is approximated by the DGNs that take state-action pairs and random noises as their input. Moreover, we use the SIA to provide a semi-implicit policy distribution, which mixes the policy parameters with a reparameterizable distribution that is not constrained by an analytic density function. In this way, the policy's marginal distribution is implicit, providing the potential to model complex properties such as covariance structure and skewness, but its parameter and entropy can still be estimated. We incorporate these features with an off-policy algorithm framework to solve problems with continuous action space and compare IDAC with state-of-the-art algorithms on representative OpenAI Gym environments. We observe that IDAC outperforms these baselines in most tasks. Python code is provided. | ['Mingyuan Zhou', 'Zhendong Wang', 'Yuguang Yue'] | 2020-07-13 | null | http://proceedings.neurips.cc/paper/2020/hash/4f20f7f5d2e7a1b640ebc8244428558c-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/4f20f7f5d2e7a1b640ebc8244428558c-Paper.pdf | neurips-2020-12 | ['distributional-reinforcement-learning'] | ['methodology'] | [-3.36326838e-01 3.55221927e-01 -3.85105282e-01 -9.50819701e-02
-7.52450764e-01 -6.55257344e-01 8.74272168e-01 -5.37182130e-02
-8.31183434e-01 1.15169311e+00 4.50871855e-01 -2.30668902e-01
-1.17651664e-01 -7.36782193e-01 -8.40540469e-01 -1.03067172e+00
-1.14965297e-01 7.40425169e-01 -1.00147754e-01 -2.01797128e-01
-3.41545530e-02 2.06443220e-01 -1.36221671e+00 -1.71509013e-01
1.21039116e+00 7.66800702e-01 1.55880257e-01 6.79096639e-01
1.34553146e-02 9.59644735e-01 -7.18215942e-01 -3.03265333e-01
3.44327718e-01 -5.43403864e-01 -4.99567300e-01 -4.92106259e-01
-4.51216381e-03 -6.01582527e-01 -3.96318585e-01 1.06308389e+00
7.69629002e-01 5.16365170e-01 1.05797052e+00 -1.24772680e+00
-7.94928551e-01 7.29605258e-01 -3.21148813e-01 -1.20013200e-01
-1.45179546e-02 8.02217603e-01 8.97654057e-01 -2.77588189e-01
5.31459928e-01 1.52038312e+00 5.32635808e-01 8.31938386e-01
-1.72599018e+00 -3.28063458e-01 3.76239359e-01 -1.49016216e-01
-9.61786032e-01 -9.70457774e-03 6.47296548e-01 -3.52545142e-01
5.61830819e-01 -4.44564670e-02 9.07854497e-01 1.69682884e+00
2.65782058e-01 1.10895085e+00 1.31580150e+00 -1.33722365e-01
9.97392654e-01 2.05119550e-02 -3.48259091e-01 3.45412940e-01
3.28067914e-02 5.32063484e-01 -1.47785395e-01 -4.40320432e-01
8.39153647e-01 -8.58231857e-02 -1.57281101e-01 -8.56022120e-01
-8.36972415e-01 1.04251218e+00 4.12782609e-01 -1.58264801e-01
-5.77701330e-01 6.83101118e-01 5.59810102e-01 1.46233514e-01
3.70679051e-01 4.82112646e-01 -3.29621285e-01 -5.56328416e-01
-6.22074783e-01 1.00282311e+00 9.17035162e-01 7.10566282e-01
5.48173487e-01 3.45180005e-01 -8.33481312e-01 7.23383904e-01
3.42070132e-01 8.44415367e-01 6.85338616e-01 -1.27873898e+00
2.50505000e-01 1.16355702e-01 5.24856567e-01 -2.90857017e-01
-2.83674866e-01 -5.25571406e-01 -6.00115955e-01 6.42876267e-01
7.81598449e-01 -5.15509367e-01 -9.54141498e-01 2.19642472e+00
3.79497677e-01 7.55828097e-02 -2.08576828e-01 9.53900754e-01
1.06698558e-01 4.25248146e-01 3.92950296e-01 2.71652061e-02
9.57623780e-01 -8.51717710e-01 -5.90279162e-01 -8.71045589e-02
5.57556987e-01 3.18250321e-02 1.56398106e+00 2.41156831e-01
-1.27710533e+00 -2.20192373e-01 -7.48018742e-01 1.37520179e-01
-1.51946247e-01 2.10623160e-01 3.82626414e-01 3.74385059e-01
-1.03480792e+00 9.28818882e-01 -1.20806992e+00 1.04981087e-01
4.64848906e-01 1.66883931e-01 1.94095314e-01 3.47562641e-01
-1.19088924e+00 1.07044888e+00 3.89493406e-01 -2.28247672e-01
-1.44148397e+00 -9.11476433e-01 -8.08645606e-01 7.44698048e-02
4.07512695e-01 -8.57640922e-01 1.52987230e+00 -9.31235969e-01
-2.30852103e+00 2.22890839e-01 4.41438556e-01 -6.96833968e-01
9.73244309e-01 -2.03163311e-01 1.54790685e-01 1.20137893e-01
-1.22723922e-01 5.74262202e-01 1.16611207e+00 -1.04067695e+00
-2.46915534e-01 -1.79953068e-01 1.21780656e-01 5.18139660e-01
-1.11776423e-02 -5.52008152e-01 2.03416988e-01 -7.18239665e-01
-6.96510792e-01 -1.13152254e+00 -3.77434283e-01 1.61781251e-01
-3.17434788e-01 -2.80708551e-01 3.35892022e-01 -5.26541829e-01
1.12678885e+00 -2.05314612e+00 3.11895639e-01 1.37395293e-01
-5.92635982e-02 3.55978638e-01 -2.57112950e-01 4.94490027e-01
2.25035056e-01 -6.23220466e-02 -4.30330992e-01 -5.50118387e-01
6.35915279e-01 4.72826809e-01 -5.61311364e-01 6.45233750e-01
2.29717549e-02 1.01752961e+00 -1.36322558e+00 5.53270942e-03
1.25397891e-01 4.02036637e-01 -7.78760374e-01 3.13210875e-01
-7.89254963e-01 4.77236390e-01 -6.40812576e-01 1.35867745e-01
5.24209440e-01 1.85739279e-01 3.25332195e-01 3.95478457e-01
4.92281094e-03 3.77522826e-01 -1.18122590e+00 1.83305061e+00
-5.24079561e-01 5.09504564e-02 2.87164897e-01 -1.11681187e+00
8.46433759e-01 1.33645954e-02 4.08862770e-01 -6.14119828e-01
1.31238326e-01 2.13943645e-01 2.45663617e-02 -1.78530365e-01
3.83229196e-01 -1.63699612e-01 1.42678926e-02 7.28678048e-01
3.13539445e-01 -2.57441938e-01 3.09301671e-02 1.80110097e-01
1.09243453e+00 9.03895557e-01 -8.92020762e-02 -6.71987057e-01
6.44724444e-02 -4.17534411e-01 5.12676001e-01 9.98659134e-01
-2.46577993e-01 3.91950518e-01 9.00700927e-01 -1.91003621e-01
-1.11684382e+00 -1.47946358e+00 -6.08106293e-02 9.84852254e-01
-2.43434355e-01 -2.82508343e-01 -8.28662515e-01 -7.75478959e-01
5.32335639e-01 1.06495357e+00 -9.09601152e-01 -3.99693787e-01
-3.76140058e-01 -4.20412928e-01 5.11012316e-01 5.51578820e-01
1.71095625e-01 -1.18453825e+00 -7.05019295e-01 3.62828285e-01
1.49160013e-01 -4.65908647e-01 -7.39381373e-01 3.47241133e-01
-6.52694046e-01 -8.66485655e-01 -9.93121982e-01 -8.43966827e-02
4.00909334e-01 -6.71746910e-01 1.24077988e+00 -5.21530449e-01
-5.64255528e-02 6.44382536e-01 -1.55293401e-02 -6.22968674e-01
-4.62356746e-01 8.94841738e-03 3.15160483e-01 -1.77264228e-01
-1.26868308e-01 -8.39773357e-01 -1.02545488e+00 -7.03943223e-02
-8.78101945e-01 -2.56257236e-01 3.20725173e-01 1.05881262e+00
5.55663347e-01 -4.39525455e-01 7.12075353e-01 -5.98287225e-01
9.01439786e-01 -6.73456132e-01 -7.32285321e-01 -5.65058030e-02
-6.32615149e-01 6.04410112e-01 8.52707148e-01 -8.50685894e-01
-1.17122173e+00 -3.23777534e-02 -1.19877622e-01 -6.25380695e-01
2.93362439e-02 2.82118469e-01 4.64995056e-02 5.52597046e-01
8.83313477e-01 3.20597410e-01 4.77081627e-01 -4.95656043e-01
9.55543697e-01 4.18244094e-01 5.72832406e-01 -1.29483247e+00
5.27561963e-01 3.08577716e-01 -1.20636128e-01 -4.49011922e-01
-8.57089758e-01 1.10357299e-01 -1.88242197e-01 6.78034686e-03
7.27787077e-01 -9.71200228e-01 -9.90277708e-01 5.07998407e-01
-7.40269065e-01 -1.19675410e+00 -1.17045200e+00 6.32549644e-01
-1.19643605e+00 -7.96633139e-02 -8.04984212e-01 -1.20991242e+00
-2.80583292e-01 -9.68011439e-01 9.72047329e-01 2.11083800e-01
-1.88457742e-01 -1.26267862e+00 5.09257495e-01 -4.25019145e-01
6.21168196e-01 3.60042095e-01 7.47857451e-01 -5.86891115e-01
-6.24371693e-02 2.58069992e-01 3.01449358e-01 6.40465677e-01
-3.45865786e-02 -2.17212483e-01 -8.40061963e-01 -5.76366782e-01
-1.86881851e-02 -7.08231449e-01 8.47778797e-01 7.94196606e-01
1.27235115e+00 -5.75190365e-01 3.07444986e-02 7.55377889e-01
9.74888146e-01 -8.88339430e-02 4.97561663e-01 2.58710384e-01
4.62219954e-01 1.79631889e-01 3.44598711e-01 8.23673785e-01
4.77161884e-01 4.28188801e-01 5.99016309e-01 2.33525634e-01
9.23390687e-02 -9.43079829e-01 7.64220178e-01 2.20699534e-01
5.37896976e-02 8.52826089e-02 -5.23186028e-01 3.53765011e-01
-2.14602637e+00 -1.04721975e+00 4.41630989e-01 2.38767028e+00
1.36775756e+00 1.06926173e-01 6.66416407e-01 -4.21703935e-01
2.81013846e-01 1.93427563e-01 -1.19088745e+00 -5.31911850e-01
1.85681030e-01 4.32094604e-01 5.46213806e-01 4.04756635e-01
-7.95206785e-01 6.51856422e-01 6.33634377e+00 1.04834700e+00
-1.01471090e+00 -9.05133039e-02 6.16568208e-01 -3.90912443e-01
-6.04100347e-01 -1.05988011e-01 -5.50118208e-01 8.93791258e-01
1.08522522e+00 -3.99908870e-01 8.17011774e-01 1.12692308e+00
3.08709145e-01 -6.25918582e-02 -1.03196025e+00 6.33053422e-01
-4.92733389e-01 -9.50465858e-01 -2.10900575e-01 2.22026572e-01
7.43378460e-01 2.13465914e-01 3.90737832e-01 8.86133790e-01
1.05790842e+00 -1.00167012e+00 1.00274014e+00 7.26420641e-01
7.79164433e-01 -8.18456173e-01 3.12950701e-01 7.33575523e-01
-6.88247323e-01 -3.62603039e-01 -5.65360427e-01 -1.20893784e-01
-6.46468848e-02 4.36610579e-01 -3.39830041e-01 1.75226033e-01
5.12627363e-01 6.41476333e-01 4.40946519e-02 8.17861438e-01
-4.99562681e-01 8.33598137e-01 -4.97184187e-01 -6.21692836e-02
3.44212353e-01 -7.22212434e-01 5.61303139e-01 8.87762606e-01
3.31500590e-01 -1.28510118e-01 4.02713388e-01 1.28382242e+00
1.48573201e-02 -6.11883178e-02 -5.48203826e-01 -1.47816777e-01
4.56789881e-01 1.18870759e+00 -2.36614257e-01 -2.67960459e-01
6.76534325e-02 7.09477067e-01 6.56785727e-01 5.96965253e-01
-8.99867237e-01 -1.46186635e-01 1.07609284e+00 1.00070611e-01
4.17279750e-01 -2.24363774e-01 9.34515372e-02 -1.16612816e+00
-1.12554841e-01 -9.28913176e-01 2.83881903e-01 -5.46094120e-01
-1.53115511e+00 1.78231865e-01 1.37722418e-01 -1.02809584e+00
-6.54822826e-01 -4.50775117e-01 -8.07313085e-01 1.13785481e+00
-1.34407151e+00 -8.27186704e-01 4.39780802e-01 6.83768630e-01
4.06812765e-02 -1.80780366e-02 7.69085169e-01 -7.95728788e-02
-4.20233518e-01 5.95851243e-01 4.89387661e-01 1.51546567e-03
6.07361615e-01 -1.81025302e+00 2.45471761e-01 2.10996911e-01
-2.13265106e-01 4.71990347e-01 8.08392584e-01 -6.56904221e-01
-1.49073005e+00 -1.06969392e+00 -1.79702833e-01 -4.77415532e-01
8.98038387e-01 -5.30730903e-01 -8.25555503e-01 6.98864102e-01
2.07773775e-01 2.96010762e-01 2.61027187e-01 1.25808222e-02
-1.34888977e-01 3.23368572e-02 -1.13701856e+00 6.93991661e-01
9.76721227e-01 -4.52285200e-01 -3.91016871e-01 2.35380024e-01
6.18877590e-01 -7.12817550e-01 -7.41013587e-01 -1.27620623e-01
4.25978750e-01 -8.06053877e-01 8.27700555e-01 -8.43927026e-01
4.20932263e-01 -1.91836394e-02 1.66478485e-01 -2.21191669e+00
-2.19591662e-01 -9.11609948e-01 -5.40127039e-01 9.90789413e-01
5.85993491e-02 -8.90977859e-01 5.07442236e-01 7.53552556e-01
-1.67484194e-01 -1.03373718e+00 -9.82154131e-01 -1.09807658e+00
7.27942824e-01 -2.61662900e-01 7.20527053e-01 5.04857719e-01
2.56837681e-02 8.70856494e-02 -4.35918212e-01 -3.26704085e-01
7.41180420e-01 -1.08838022e-01 6.54651284e-01 -8.41446817e-01
-6.48397565e-01 -5.69190085e-01 1.83660805e-01 -1.20806623e+00
5.97639561e-01 -8.84335756e-01 1.52148202e-01 -1.16897237e+00
-1.05626613e-01 -5.60793579e-01 -4.08675879e-01 5.26978612e-01
-1.82494223e-01 -4.78284955e-01 2.43325561e-01 -1.65115878e-01
-4.20712054e-01 1.48560882e+00 1.53007662e+00 -8.54947641e-02
-4.66483533e-01 1.12893820e-01 -5.77526450e-01 7.51935303e-01
7.92828083e-01 -4.67735261e-01 -8.78177106e-01 -1.08460493e-01
5.57651445e-02 2.07869560e-01 3.10423464e-01 -7.39757001e-01
-1.64431021e-01 -4.37274218e-01 3.38214815e-01 6.79016951e-03
1.87443405e-01 -3.70336503e-01 -2.83346266e-01 5.91944456e-01
-8.41991842e-01 6.72664959e-03 1.27775490e-01 7.87531853e-01
2.98540831e-01 -5.53464331e-03 8.12178850e-01 -1.06648356e-01
1.28651187e-02 5.24605155e-01 -5.27633131e-01 6.48462951e-01
8.42114747e-01 4.52378541e-01 -3.11871290e-01 -6.43143475e-01
-6.51729167e-01 4.16970402e-01 4.56140339e-01 6.70131519e-02
2.71988034e-01 -1.55280614e+00 -6.78397298e-01 8.48787948e-02
-2.99238354e-01 1.60152689e-01 1.38077006e-01 5.85779727e-01
-7.57473931e-02 -1.52871490e-01 -1.37229592e-01 -3.01525950e-01
-3.03268701e-01 5.09001374e-01 5.51132679e-01 -5.14437139e-01
-8.17413330e-01 4.29816693e-01 5.70180640e-02 -7.79482305e-01
2.14381337e-01 -5.82996428e-01 4.64187898e-02 -7.58827403e-02
5.60370505e-01 2.48135164e-01 -4.88745272e-01 -3.32512483e-02
1.87554471e-02 -9.94525105e-02 2.88433969e-01 -3.71822059e-01
1.35315573e+00 1.75883830e-01 3.26380819e-01 5.11703968e-01
7.44945765e-01 -1.45325631e-01 -2.11114526e+00 -2.22425878e-01
-3.21269900e-01 -2.46015057e-01 1.64004080e-02 -1.12544703e+00
-8.96777868e-01 7.20194578e-01 5.02632201e-01 1.40893728e-01
6.56121373e-01 -3.98749202e-01 5.06212890e-01 1.42169759e-01
3.32570463e-01 -1.49670029e+00 3.91259082e-02 6.42138422e-01
9.41421747e-01 -9.12244022e-01 -8.06514993e-02 5.71999073e-01
-9.30502117e-01 7.97998428e-01 5.65219998e-01 -6.07212543e-01
7.54356205e-01 3.92460048e-01 -8.21017399e-02 2.79630452e-01
-9.61053789e-01 -1.91580370e-01 1.65450275e-01 8.19372535e-01
1.19561493e-01 3.79047900e-01 -3.55946571e-01 7.30216682e-01
-2.05031797e-01 1.08877487e-01 3.47481668e-01 8.10355365e-01
-1.17315836e-01 -1.27747190e+00 6.09212108e-02 4.36999559e-01
-2.13439047e-01 8.16717893e-02 1.92472011e-01 7.38653481e-01
-3.00223947e-01 3.63882244e-01 1.82923540e-01 3.59357372e-02
2.34057307e-01 6.48954883e-02 4.50486869e-01 -4.36339825e-01
-4.37288046e-01 -9.29058045e-02 -2.35450894e-01 -8.05926323e-01
2.22300708e-01 -7.34126091e-01 -1.31311655e+00 -2.57235318e-01
2.14519769e-01 2.31243581e-01 7.26726234e-01 8.94463062e-01
4.46600914e-01 5.98938882e-01 5.06672502e-01 -1.05114675e+00
-1.80043936e+00 -9.96868074e-01 -1.02385056e+00 5.51344752e-01
4.39997643e-01 -8.35932910e-01 -3.99932534e-01 -5.56095362e-01] | [4.062713623046875, 2.4839680194854736] |
2acc3816-c59a-4ac8-bb10-d04f802ca026 | partial-label-learning-with-self-guided | 1902.03045 | null | http://arxiv.org/abs/1902.03045v1 | http://arxiv.org/pdf/1902.03045v1.pdf | Partial Label Learning with Self-Guided Retraining | Partial label learning deals with the problem where each training instance is
assigned a set of candidate labels, only one of which is correct. This paper
provides the first attempt to leverage the idea of self-training for dealing
with partially labeled examples. Specifically, we propose a unified formulation
with proper constraints to train the desired model and perform pseudo-labeling
jointly. For pseudo-labeling, unlike traditional self-training that manually
differentiates the ground-truth label with enough high confidence, we introduce
the maximum infinity norm regularization on the modeling outputs to
automatically achieve this consideratum, which results in a convex-concave
optimization problem. We show that optimizing this convex-concave problem is
equivalent to solving a set of quadratic programming (QP) problems. By
proposing an upper-bound surrogate objective function, we turn to solving only
one QP problem for improving the optimization efficiency. Extensive experiments
on synthesized and real-world datasets demonstrate that the proposed approach
significantly outperforms the state-of-the-art partial label learning
approaches. | ['Lei Feng', 'Bo An'] | 2019-02-08 | null | null | null | null | ['partial-label-learning'] | ['methodology'] | [ 4.77940351e-01 7.32872188e-01 -8.21834624e-01 -7.76679039e-01
-1.19778025e+00 -5.03543735e-01 2.67020762e-01 1.74819767e-01
-3.37514073e-01 7.80084968e-01 -3.60162079e-01 -1.92361116e-01
1.46779492e-01 -3.78509551e-01 -7.87763298e-01 -6.84581518e-01
3.95339668e-01 6.66686952e-01 -1.81665599e-01 5.15839398e-01
8.89028162e-02 8.70347321e-02 -1.40900958e+00 1.16328284e-01
1.08286643e+00 1.11638796e+00 -7.65737100e-03 8.25644583e-02
-5.95791340e-02 9.25914526e-01 -3.45209062e-01 -4.44955438e-01
3.28971714e-01 -4.72380519e-01 -9.43897247e-01 8.35385144e-01
5.35338104e-01 -1.05578668e-01 2.73318738e-01 1.33093631e+00
1.72310770e-01 2.74044164e-02 7.63020933e-01 -1.45259869e+00
-4.85910267e-01 3.44981402e-01 -6.78167284e-01 -3.30381423e-01
7.13575929e-02 -1.77338257e-01 1.31570208e+00 -9.43706393e-01
3.57083797e-01 1.01744854e+00 5.98277807e-01 5.72432399e-01
-1.41553867e+00 -5.66845596e-01 2.97244012e-01 -9.83905047e-02
-1.40659809e+00 -2.76924282e-01 8.93111050e-01 -4.97574925e-01
3.31275225e-01 1.44781023e-01 2.92292327e-01 7.11633742e-01
-5.74595094e-01 1.02650619e+00 1.43434989e+00 -5.52921057e-01
4.02500927e-01 6.79911792e-01 5.89895546e-01 9.57792819e-01
2.35385850e-01 -8.67696851e-02 -1.77823097e-01 -2.02160358e-01
3.74277860e-01 -7.89135620e-02 -2.94395536e-01 -7.16040730e-01
-1.10030580e+00 7.85260081e-01 2.83407092e-01 -1.26159228e-02
-2.00362489e-01 -1.73767228e-02 1.19326510e-01 1.43011644e-01
7.73935080e-01 4.21372563e-01 -6.99335277e-01 3.36064249e-01
-1.00215054e+00 -6.28381148e-02 8.99953723e-01 1.08407891e+00
1.15849602e+00 4.30679284e-02 -1.39190361e-01 1.03655982e+00
5.62860131e-01 2.33399048e-01 2.29444951e-01 -1.18734634e+00
4.67595845e-01 7.63083935e-01 4.27916497e-01 -5.38367629e-01
-3.50035578e-01 -8.90611649e-01 -5.15962899e-01 2.46016532e-02
5.76927781e-01 -3.56837720e-01 -7.60075033e-01 1.88314426e+00
6.75582051e-01 5.94457865e-01 1.20381147e-01 8.70095193e-01
4.19482946e-01 6.54311597e-01 1.50483295e-01 -7.71659493e-01
1.13960326e+00 -1.45334303e+00 -7.61725187e-01 -4.68987465e-01
9.17573094e-01 -4.68528330e-01 1.04625571e+00 2.75815099e-01
-7.92545021e-01 -4.82156157e-01 -1.12264216e+00 2.11291239e-01
5.57547435e-02 5.56383848e-01 5.23470104e-01 7.09949791e-01
-6.85452282e-01 6.45962834e-01 -5.38847506e-01 -9.99289099e-03
2.73755670e-01 4.89436865e-01 -3.13003689e-01 -2.99228262e-03
-8.45326662e-01 7.43438184e-01 6.28050387e-01 6.32452518e-02
-8.55708957e-01 -6.50212288e-01 -9.08077657e-01 -5.92467114e-02
8.67172003e-01 -4.46191311e-01 1.40585470e+00 -1.08588088e+00
-1.52756751e+00 1.37764144e+00 -3.86275023e-01 -1.99407294e-01
5.01421809e-01 -1.50069088e-01 -1.28126472e-01 -8.33484530e-02
4.24802244e-01 5.83674610e-01 8.21496606e-01 -1.91024077e+00
-7.94707954e-01 -2.07939193e-01 2.36377090e-01 2.14113951e-01
-5.34302711e-01 -2.38572657e-01 -5.43360472e-01 -4.46771145e-01
2.74870157e-01 -1.07093263e+00 -3.25641572e-01 -1.19518176e-01
-6.83168590e-01 -3.11713398e-01 5.05090892e-01 -2.48316854e-01
1.14367068e+00 -2.05223751e+00 -2.63071489e-02 1.13197960e-01
2.23930433e-01 1.81310371e-01 3.94867994e-02 2.74329968e-02
-1.23240508e-01 9.14913882e-03 -4.82654601e-01 -8.81163836e-01
1.58573687e-02 4.36617970e-01 -2.72395253e-01 5.82059145e-01
1.31725103e-01 7.26365745e-01 -1.12060618e+00 -7.08067596e-01
-5.73421642e-02 -1.92131177e-02 -4.90969509e-01 3.64642501e-01
-3.80140513e-01 6.21158600e-01 -6.70286417e-01 6.16453826e-01
6.47668183e-01 -7.83837259e-01 3.37056309e-01 -2.31394380e-01
1.35135695e-01 -9.94775537e-03 -1.46768630e+00 1.47365272e+00
-4.92715269e-01 3.98151875e-02 2.01865822e-01 -1.37825513e+00
8.21639121e-01 3.31287593e-01 6.12023532e-01 -1.64106384e-01
3.40146780e-01 3.57319057e-01 -4.66930330e-01 -3.43810350e-01
1.63539108e-02 -3.68468016e-01 4.87883501e-02 6.59920096e-01
2.66681969e-01 -6.76502660e-02 2.47441500e-01 -1.43264771e-01
5.90734601e-01 2.68787235e-01 4.11641717e-01 -2.87755966e-01
7.64290333e-01 4.02351618e-02 9.74377096e-01 6.35671020e-01
-1.43293723e-01 4.55424398e-01 4.16566104e-01 -2.25284666e-01
-8.28121960e-01 -7.03484297e-01 -2.97176719e-01 9.77026224e-01
2.82711893e-01 -2.03467578e-01 -9.29567277e-01 -1.16278398e+00
-8.49232301e-02 8.84549379e-01 -6.28587127e-01 -4.51107733e-02
-4.97846901e-01 -8.48943710e-01 3.90420035e-02 2.26880372e-01
4.74591285e-01 -6.94948196e-01 -2.44922042e-01 2.19327956e-01
-1.64126351e-01 -1.15206409e+00 -6.43089592e-01 6.11660123e-01
-9.87011790e-01 -1.20035553e+00 -7.15416551e-01 -1.15148628e+00
1.19692492e+00 2.32713059e-01 1.05969763e+00 8.01001340e-02
1.10755444e-01 1.71720982e-01 -1.95285439e-01 -8.49082395e-02
-5.26143432e-01 8.74312669e-02 -1.82863381e-02 4.71025616e-01
1.74879789e-01 -3.05364460e-01 -1.66353077e-01 4.48249280e-01
-7.27330029e-01 2.46040612e-01 4.84231293e-01 1.10723329e+00
1.14214838e+00 1.69612691e-01 8.18303883e-01 -1.53465652e+00
3.56907785e-01 -4.46482509e-01 -7.37958372e-01 6.18159711e-01
-1.11956000e+00 4.46938932e-01 7.48158216e-01 -4.90344852e-01
-1.03522325e+00 5.97073138e-01 2.34906256e-01 -5.32894790e-01
-8.23057070e-02 4.77089107e-01 -3.38107973e-01 -1.64521158e-01
4.91513819e-01 1.76526710e-01 -2.08154336e-01 -4.94467705e-01
5.58792114e-01 6.39706135e-01 4.78449672e-01 -7.19741285e-01
8.50674033e-01 3.74274313e-01 1.23694949e-01 -2.44977027e-01
-1.94277418e+00 -7.32203782e-01 -7.98152447e-01 -1.81330159e-01
6.40205562e-01 -8.56831253e-01 -6.41184449e-01 1.68898508e-01
-8.94680858e-01 -4.21827525e-01 -4.97414857e-01 3.61625582e-01
-7.37372994e-01 4.70287085e-01 -4.62479055e-01 -8.57909322e-01
-3.47601920e-02 -1.12317348e+00 1.08978605e+00 1.22768402e-01
2.76240520e-02 -1.08326268e+00 1.33175313e-01 6.99364722e-01
-2.72293150e-01 1.76561996e-01 7.28322983e-01 -7.78960586e-01
-3.46222311e-01 -2.57400542e-01 -3.02026272e-01 6.10576868e-01
1.79791927e-01 -4.51716661e-01 -1.02299607e+00 -3.73213202e-01
4.05822664e-01 -7.09264576e-01 7.58283734e-01 1.99728861e-01
1.03237867e+00 -4.51011926e-01 -4.69322562e-01 6.43367469e-01
1.39527345e+00 4.95168678e-02 -1.05446808e-01 2.47306690e-01
7.64977455e-01 6.75657094e-01 1.15733361e+00 4.39156085e-01
5.36590278e-01 5.22985339e-01 4.69576746e-01 -2.50045925e-01
1.54221207e-01 -3.74859005e-01 -2.77616885e-02 6.86713815e-01
4.22854960e-01 -1.46708414e-01 -6.24170840e-01 3.93426687e-01
-2.12591362e+00 -5.39526105e-01 -1.69448391e-01 2.24643993e+00
1.10189331e+00 4.26821299e-02 -9.34403613e-02 1.42622441e-01
8.64944160e-01 1.18164346e-02 -8.77084732e-01 7.04375729e-02
2.25592982e-02 -8.08896199e-02 5.46517670e-01 6.38330162e-01
-1.53247130e+00 1.06962335e+00 6.32476711e+00 8.26123238e-01
-9.78951991e-01 3.36479574e-01 9.04441357e-01 2.44607538e-01
-2.05899775e-01 2.64343947e-01 -8.76070857e-01 2.87332207e-01
7.53555536e-01 -5.47815748e-02 4.29247767e-01 1.14726889e+00
-4.78270650e-02 -5.97580485e-02 -1.34217155e+00 8.86341274e-01
1.41764775e-01 -9.27654028e-01 -3.32695693e-01 2.51097288e-02
1.25172114e+00 -4.18848306e-01 2.83717620e-03 4.29604620e-01
3.89426261e-01 -6.39080286e-01 9.22437668e-01 2.18560547e-01
8.95825565e-01 -5.37916183e-01 5.87281108e-01 7.18899846e-01
-1.00511336e+00 -2.36572355e-01 -3.23650509e-01 9.80464965e-02
3.17560881e-01 8.56402397e-01 -8.02833259e-01 4.78598654e-01
6.66517019e-02 7.74752498e-01 -3.28593165e-01 9.34594870e-01
-5.81557393e-01 8.46606076e-01 -1.60501868e-01 2.53198713e-01
1.51338667e-01 -3.92897487e-01 2.94205725e-01 1.04064560e+00
5.34189753e-02 1.61103591e-01 6.95176065e-01 8.84592175e-01
-3.21383744e-01 3.14448744e-01 -1.87837973e-01 8.99220631e-02
4.67098147e-01 1.41133678e+00 -7.58080423e-01 -4.96013612e-01
-5.24470210e-01 1.10881424e+00 7.21121907e-01 3.64874005e-01
-7.01939404e-01 1.24244750e-01 9.81638953e-02 -1.59387946e-01
-4.81238663e-02 1.79458886e-01 -4.55891818e-01 -1.27540648e+00
6.95094317e-02 -6.54855788e-01 4.07826424e-01 -5.04195809e-01
-1.34225845e+00 4.82942402e-01 -1.31062001e-01 -1.31904566e+00
-2.94633210e-01 -4.57002312e-01 -2.91687578e-01 7.40279019e-01
-1.78289199e+00 -1.19906592e+00 -2.34871522e-01 3.35524321e-01
4.17302191e-01 3.27696912e-02 8.40047956e-01 4.27674919e-01
-7.43400872e-01 6.16717756e-01 1.64650381e-01 -1.55621558e-01
8.30538869e-01 -1.52255297e+00 -1.61693349e-01 6.31853282e-01
2.12343872e-01 3.73207986e-01 7.55208313e-01 -5.34661829e-01
-9.58286762e-01 -1.37461686e+00 9.56083655e-01 -1.06511392e-01
6.24292910e-01 -2.33082503e-01 -8.53400886e-01 7.99715817e-01
-1.42821595e-01 3.36053550e-01 9.00851250e-01 2.75284145e-02
-3.37524325e-01 -3.96012142e-02 -1.21866202e+00 2.45036751e-01
9.33066487e-01 -3.43478471e-01 -3.95395070e-01 8.08526278e-01
7.97303915e-01 -3.56458247e-01 -7.14172065e-01 5.99396467e-01
4.31741700e-02 -5.41064918e-01 8.15092444e-01 -5.45447290e-01
1.61145881e-01 -4.72692817e-01 -1.28946155e-01 -1.27986050e+00
-2.45260492e-01 -4.40344721e-01 -3.09520513e-01 1.32175314e+00
5.56227505e-01 -5.17050326e-01 1.17975569e+00 7.42745280e-01
-2.55698860e-01 -1.12778771e+00 -6.72614872e-01 -8.30743074e-01
-2.27694467e-01 -3.25416744e-01 2.47311756e-01 1.28762615e+00
1.18335243e-02 5.29913902e-01 -5.92645228e-01 3.16760927e-01
1.03397512e+00 3.66883785e-01 5.51301837e-01 -1.36484516e+00
-5.92105269e-01 6.57309836e-04 3.71844508e-02 -1.46636999e+00
8.25399578e-01 -1.09562731e+00 3.06115955e-01 -1.50411355e+00
4.35776979e-01 -8.17200780e-01 -3.82947803e-01 7.47612596e-01
-4.74359453e-01 2.22239017e-01 6.95229471e-02 4.19932753e-01
-9.10939217e-01 5.51389694e-01 1.35419440e+00 -2.59497732e-01
-1.90325901e-01 3.26475680e-01 -8.76986086e-01 6.75397992e-01
6.19860232e-01 -8.05032551e-01 -6.40207946e-01 -2.02280208e-01
8.54420811e-02 2.61394739e-01 1.61362216e-02 -7.82545924e-01
2.10731044e-01 -2.56971210e-01 -2.12736815e-01 -4.52125788e-01
3.44117194e-01 -9.47912455e-01 -5.00218049e-02 3.15252841e-01
-6.61617577e-01 -6.13682747e-01 -2.84454226e-01 6.58939838e-01
-3.06407094e-01 -7.91398406e-01 9.84675288e-01 -7.37341270e-02
-4.38563317e-01 2.78703839e-01 7.31562600e-02 2.51649469e-01
1.27124274e+00 -2.57320292e-02 -1.93605777e-02 -2.74995238e-01
-9.86709952e-01 6.73499942e-01 3.43262017e-01 9.53654945e-03
1.50075823e-01 -1.32446992e+00 -5.71942747e-01 -1.45173445e-01
1.25783101e-01 1.73321709e-01 -2.17178166e-02 6.22371852e-01
4.06978615e-02 5.23140311e-01 2.89024293e-01 -5.79681456e-01
-1.01455808e+00 9.43680286e-01 3.09379965e-01 -6.68169737e-01
-1.42637327e-01 8.59186172e-01 3.57953697e-01 -8.14465702e-01
3.78242761e-01 -2.76470575e-02 -3.07319164e-01 3.51907127e-02
1.04988612e-01 2.77292848e-01 -7.28225634e-02 -7.29928792e-01
-3.11022960e-02 5.90568960e-01 4.40918608e-03 -4.62767109e-02
1.08913970e+00 -2.04892531e-01 -9.33904126e-02 6.48259759e-01
1.30629194e+00 -2.17470959e-01 -1.47489643e+00 -6.04142547e-01
2.29831144e-01 -2.68257558e-01 5.53870387e-02 -7.20538437e-01
-1.09707129e+00 6.20647967e-01 3.34713191e-01 5.81829213e-02
9.81375813e-01 -5.37099876e-02 5.39117336e-01 4.97452557e-01
6.18008256e-01 -1.15518415e+00 1.65764078e-01 3.77636254e-01
4.67261702e-01 -1.60221636e+00 -8.47420618e-02 -8.45899701e-01
-7.50696599e-01 9.03047323e-01 8.33374381e-01 -4.89834175e-02
7.07394242e-01 -2.48674862e-02 6.67281896e-02 -1.46440743e-02
-6.22321606e-01 -1.92261562e-01 3.82237107e-01 1.70672491e-01
3.96943361e-01 1.64709255e-01 -4.50602621e-01 6.90825343e-01
5.48900366e-02 8.24346021e-03 2.14472935e-01 8.40502858e-01
-5.87068975e-01 -1.44297051e+00 -3.37260991e-01 4.44950610e-01
-4.67034638e-01 1.37832686e-01 -2.02919751e-01 4.74300444e-01
1.78474635e-01 1.18580401e+00 -2.06824809e-01 -1.96984068e-01
1.85146734e-01 3.08983594e-01 3.68101031e-01 -1.16722715e+00
-2.63803303e-01 3.09676439e-01 8.05420205e-02 -2.11119413e-01
-6.16618037e-01 -5.48640668e-01 -1.39413452e+00 3.90972614e-01
-7.82706261e-01 4.32694584e-01 4.65203136e-01 1.06218684e+00
-1.57845917e-03 4.22913909e-01 9.90022123e-01 -6.48229063e-01
-9.29980934e-01 -8.69855464e-01 -7.59036243e-01 4.87939507e-01
2.25876182e-01 -8.01342905e-01 -4.59334224e-01 1.87156871e-01] | [9.414383888244629, 4.036522388458252] |
4266c56d-93f0-489c-a066-a8da4f17b04c | pruning-meets-low-rank-parameter-efficient | 2305.18403 | null | https://arxiv.org/abs/2305.18403v2 | https://arxiv.org/pdf/2305.18403v2.pdf | Pruning Meets Low-Rank Parameter-Efficient Fine-Tuning | Large pre-trained models (LPMs), such as LLaMA and ViT-G, have shown exceptional performance across various tasks. Although parameter-efficient fine-tuning (PEFT) has emerged to cheaply fine-tune these large models on downstream tasks, their deployment is still hindered by the vast model scale and computational costs. Neural network pruning offers a solution for model compression by removing redundant parameters, but most existing methods rely on computing parameter gradients. However, obtaining the gradients is computationally prohibitive for LPMs, which necessitates the exploration of alternative approaches. To this end, we propose a unified framework for efficient fine-tuning and deployment of LPMs, termed LoRAPrune. We first design a PEFT-aware pruning criterion, which utilizes the values and gradients of Low-Rank Adaption (LoRA), rather than the gradients of pre-trained parameters for importance estimation. We then propose an iterative pruning procedure to remove redundant parameters while maximizing the advantages of PEFT. Thus, our LoRAPrune delivers an accurate, compact model for efficient inference in a highly cost-effective manner. Experimental results on various tasks demonstrate that our method achieves state-of-the-art results. For instance, in the VTAB-1k benchmark, LoRAPrune utilizes only 0.76% of the trainable parameters and outperforms magnitude and movement pruning methods by a significant margin, achieving a mean Top-1 accuracy that is 5.7% and 4.3% higher, respectively. Moreover, our approach achieves comparable performance to PEFT methods, highlighting its efficacy in delivering high-quality results while benefiting from the advantages of pruning. | ['Hao Chen', 'Bohan Zhuang', 'Xinyi Yu', 'Linlin Ou', 'Zhen Yang', 'Chunhua Shen', 'Mingyang Zhang'] | 2023-05-28 | null | null | null | null | ['model-compression'] | ['methodology'] | [ 1.58352822e-01 -1.99555367e-01 -3.67242515e-01 -3.80088449e-01
-1.05047262e+00 -3.60534459e-01 5.40029228e-01 -5.62904701e-02
-6.63764238e-01 7.14781821e-01 1.31770834e-01 -3.53966326e-01
-3.31261933e-01 -4.22988534e-01 -7.33082116e-01 -6.53905392e-01
3.94684374e-02 3.69397640e-01 2.48826474e-01 -3.21584120e-02
2.91838348e-01 3.03409129e-01 -1.45808077e+00 2.41161123e-01
1.01944041e+00 1.19321156e+00 5.52580595e-01 5.24435878e-01
1.72471739e-02 4.40178365e-01 -4.74621922e-01 -6.44203305e-01
2.31497362e-01 -1.21939875e-01 -7.73010612e-01 -4.33222532e-01
5.60923934e-01 -4.32462662e-01 -2.41566047e-01 7.95311868e-01
5.87925971e-01 2.15797395e-01 5.95861018e-01 -7.01703846e-01
-1.58705056e-01 8.24621618e-01 -6.79211020e-01 4.18084800e-01
-4.08739895e-01 8.25178623e-02 1.37736332e+00 -9.82083261e-01
2.58227557e-01 1.17476118e+00 8.53850305e-01 3.38681310e-01
-1.22759295e+00 -7.65307963e-01 3.86086434e-01 2.57183999e-01
-1.61329877e+00 -6.97099268e-01 3.81372899e-01 -1.69345975e-01
1.41579378e+00 2.51336724e-01 2.95862824e-01 9.30926323e-01
1.56322062e-01 8.63004565e-01 8.75978887e-01 -3.86981189e-01
1.22451678e-01 -3.08068004e-02 1.00225613e-01 7.00834811e-01
3.04901570e-01 -2.89508887e-02 -7.68052638e-01 -2.46243447e-01
6.22058630e-01 -1.49966598e-01 -2.07901478e-01 -1.11612834e-01
-9.02938843e-01 7.73602247e-01 4.02347505e-01 1.79122761e-01
-3.16658825e-01 4.49822813e-01 4.61445451e-01 1.34437755e-02
5.95334291e-01 6.12306595e-01 -7.90014803e-01 -4.71195281e-01
-1.28184330e+00 1.29911199e-01 5.93236446e-01 6.91986203e-01
5.25547743e-01 4.62268852e-02 -5.27818918e-01 1.29709637e+00
1.54165417e-01 3.13870847e-01 4.52317476e-01 -9.50267017e-01
7.71322787e-01 3.50058913e-01 -1.71276748e-01 -8.14375699e-01
-2.90259242e-01 -1.03803921e+00 -9.26787794e-01 -2.82850146e-01
2.23562673e-01 9.13544297e-02 -9.50364888e-01 1.94025302e+00
2.63637275e-01 1.92253396e-01 -1.79778278e-01 6.08991861e-01
4.37670499e-01 5.77952147e-01 2.58022636e-01 -1.66144446e-01
1.27672303e+00 -1.11820877e+00 -4.04059619e-01 -6.00138843e-01
6.70817196e-01 -6.55174434e-01 1.31026518e+00 5.59040964e-01
-1.10984945e+00 -4.59473699e-01 -1.13475132e+00 -1.34859756e-01
5.20659238e-02 4.08611298e-01 8.58688354e-01 5.60781658e-01
-8.19245875e-01 9.12675798e-01 -1.04311001e+00 5.45909368e-02
7.75388598e-01 4.91928965e-01 3.51829715e-02 -1.68843299e-01
-1.10399449e+00 8.80577624e-01 5.43060124e-01 1.33507147e-01
-9.24797416e-01 -1.06435859e+00 -5.40993750e-01 4.77967858e-01
5.92160583e-01 -8.26490641e-01 1.39289331e+00 -2.58005351e-01
-1.49859357e+00 3.77521515e-01 -2.63646603e-01 -7.92713940e-01
4.75407004e-01 -6.03818476e-01 -1.48423865e-01 9.56897959e-02
-9.42384973e-02 7.23181784e-01 9.88195598e-01 -7.91985750e-01
-9.00934398e-01 -9.90028381e-02 8.01298544e-02 2.13689640e-01
-8.01618576e-01 -9.90185812e-02 -9.74609911e-01 -6.70666873e-01
-5.90727106e-03 -7.92005599e-01 -3.47504497e-01 -3.60652566e-01
-3.24028105e-01 -2.37415835e-01 4.61702317e-01 -6.51785553e-01
1.92427099e+00 -2.04154325e+00 -4.51496467e-02 1.40698284e-01
5.06328642e-01 6.29805624e-01 -2.29714081e-01 1.40656561e-01
3.93021196e-01 2.25996092e-01 -3.81314158e-01 -5.64517677e-01
-4.72099930e-02 3.00608993e-01 -3.50131482e-01 7.67402798e-02
2.58394361e-01 9.66890097e-01 -7.75592327e-01 -4.76969242e-01
3.15008909e-02 6.20930731e-01 -8.66415203e-01 -1.81865364e-01
-2.38335133e-01 1.37169864e-02 -5.45857191e-01 5.07336795e-01
4.11322296e-01 -4.17745948e-01 1.90233573e-01 -2.62480795e-01
1.03133962e-01 7.09111571e-01 -1.05294633e+00 1.46364343e+00
-7.23229349e-01 4.58568722e-01 -1.01391748e-02 -8.90981853e-01
6.12214208e-01 -5.63413724e-02 1.78955749e-01 -7.63707042e-01
7.06091076e-02 4.42467153e-01 -1.28611028e-01 -1.00678444e-01
4.99553412e-01 1.56313062e-01 1.28151253e-02 3.04636151e-01
2.63867229e-02 5.72678894e-02 5.61569571e-01 2.74863750e-01
1.15767241e+00 1.76074132e-01 3.21321785e-01 -2.76558191e-01
3.08712244e-01 -2.60601729e-01 6.38108134e-01 9.12032247e-01
3.79604548e-02 2.79811710e-01 3.01911265e-01 -1.49183989e-01
-8.04515064e-01 -7.74222195e-01 -3.04609925e-01 1.39081538e+00
-2.73176938e-01 -8.40868950e-01 -8.48967791e-01 -5.62038720e-01
4.84945104e-02 6.58355653e-01 -4.11504447e-01 -3.08200359e-01
-8.47312808e-01 -9.88078833e-01 6.94128394e-01 6.26105785e-01
6.13854766e-01 -7.10497141e-01 -5.75371325e-01 3.50408942e-01
-3.84547114e-01 -1.11423981e+00 -5.30260384e-01 3.39700580e-01
-1.26893032e+00 -7.05044925e-01 -5.97221673e-01 -3.62877190e-01
5.04165113e-01 3.37914288e-01 1.26962042e+00 1.31791428e-01
-1.79913238e-01 -2.78461665e-01 -1.42066702e-01 -2.51551330e-01
-1.08482301e-01 6.62475944e-01 2.40194649e-02 -2.26652324e-01
1.41136751e-01 -7.43089497e-01 -7.20459282e-01 3.06739599e-01
-6.73036695e-01 1.38115272e-01 1.11993110e+00 9.51871753e-01
9.81878102e-01 2.47525014e-02 7.01742291e-01 -1.06372213e+00
6.79810941e-01 -2.28258297e-01 -6.02154195e-01 3.09833676e-01
-1.12136769e+00 3.69541764e-01 6.88874245e-01 -5.05378306e-01
-1.17517531e+00 -1.63279593e-01 -8.87930021e-02 -5.10012448e-01
3.98662060e-01 7.74884522e-01 8.33072513e-02 -3.62576544e-02
7.84473181e-01 1.50533348e-01 -4.24399585e-01 -7.80881882e-01
4.03513074e-01 4.15326864e-01 5.43957353e-01 -6.89753890e-01
7.11991668e-01 1.23007879e-01 -7.51549553e-04 -6.19054675e-01
-1.22784054e+00 -4.05621260e-01 -3.22681695e-01 2.93645144e-01
2.63051152e-01 -1.01563251e+00 -4.94128317e-01 1.91451862e-01
-8.02396476e-01 -4.27967638e-01 -1.23553529e-01 4.37330693e-01
-2.64286220e-01 4.35261369e-01 -7.30385482e-01 -6.37989819e-01
-8.14658642e-01 -1.09201670e+00 9.91028845e-01 -6.46670237e-02
-2.88875163e-01 -7.79015064e-01 -1.79813191e-01 5.45277357e-01
5.56401968e-01 -2.80742049e-01 1.09662008e+00 -4.17480886e-01
-6.27405643e-01 -8.04333836e-02 -3.33139479e-01 4.57082719e-01
-2.46768236e-01 -2.12110892e-01 -1.12002122e+00 -3.63853514e-01
-1.83157921e-01 -2.96532065e-01 1.34641290e+00 6.59424365e-01
1.52768958e+00 -4.04225945e-01 -5.72411954e-01 9.14009750e-01
1.32741892e+00 -1.00555182e-01 3.99960816e-01 3.22695345e-01
7.09458947e-01 1.47264153e-01 6.77044868e-01 5.38435876e-01
2.16602504e-01 7.47604072e-01 1.91287324e-01 1.07353047e-01
-2.18671769e-01 -3.22900116e-01 2.29709119e-01 9.80413139e-01
-2.72688299e-01 -1.10034503e-01 -7.57939696e-01 4.38729376e-01
-1.81612372e+00 -6.05371237e-01 4.71113026e-02 2.29711032e+00
1.19600725e+00 4.60569143e-01 -1.61459282e-01 1.51487142e-01
3.00189674e-01 2.01349348e-01 -7.37897456e-01 -3.99156183e-01
6.97387606e-02 4.51065898e-01 5.98766565e-01 3.78891975e-01
-1.04349124e+00 1.16212833e+00 6.04163790e+00 1.50664163e+00
-7.99301505e-01 2.71759182e-01 7.77201295e-01 -6.02376163e-01
-1.38931602e-01 -1.10635884e-01 -1.46618223e+00 3.41556430e-01
1.18360722e+00 8.66278261e-03 6.44275725e-01 9.85711873e-01
2.17344612e-01 -2.03396231e-02 -9.68213916e-01 9.22290325e-01
-2.47708142e-01 -1.33501697e+00 1.08837150e-01 4.80709001e-02
8.32878947e-01 3.56039137e-01 1.03511408e-01 7.78294921e-01
1.29809424e-01 -9.01280224e-01 7.20255911e-01 1.80638656e-01
9.34949219e-01 -9.10994112e-01 5.88320613e-01 4.85453695e-01
-1.22604680e+00 -2.08220050e-01 -6.68886781e-01 4.51832265e-02
6.05295785e-02 9.76073682e-01 -8.32089603e-01 3.78768384e-01
8.39571595e-01 4.91388768e-01 -6.20520830e-01 9.70496595e-01
-3.27396661e-01 9.80155289e-01 -5.68031609e-01 1.32362638e-02
3.30059975e-01 -4.44314107e-02 3.41138870e-01 1.40182638e+00
1.91075936e-01 -6.98504746e-02 -9.39153805e-02 6.57568097e-01
-4.33188140e-01 6.54824302e-02 5.57162054e-02 -1.47300765e-01
8.01645458e-01 1.31805682e+00 -4.72755432e-01 -3.66111517e-01
-1.85269877e-01 6.39017940e-01 7.89499104e-01 2.47926325e-01
-1.00563037e+00 -3.57393116e-01 6.75312281e-01 -3.92375886e-02
6.75644457e-01 -1.41024247e-01 -4.21123058e-01 -9.61699307e-01
1.93625808e-01 -9.55516100e-01 4.76401627e-01 -2.41889775e-01
-9.19122338e-01 6.72379076e-01 1.07540511e-01 -8.78665030e-01
-3.92915845e-01 -3.68713647e-01 -2.99713492e-01 7.89408743e-01
-1.76345849e+00 -1.01768625e+00 -1.13422602e-01 2.29546532e-01
7.47070074e-01 -3.65279689e-02 6.18553221e-01 4.82386947e-01
-8.61310959e-01 1.14484751e+00 2.97662526e-01 -3.86145592e-01
5.94704390e-01 -9.61824000e-01 5.56084037e-01 8.86924565e-01
2.65861213e-01 8.79168391e-01 4.04833883e-01 -5.03376901e-01
-1.18373203e+00 -1.16921651e+00 1.05499172e+00 -2.78401077e-01
5.03350973e-01 -3.34267855e-01 -9.43489194e-01 4.98885363e-01
-3.72788757e-01 -1.36761606e-01 5.49632132e-01 6.11082852e-01
-4.91699964e-01 -3.27189535e-01 -7.52993584e-01 7.65082300e-01
1.29761863e+00 -3.24948102e-01 -3.73561352e-01 3.19213867e-01
7.60162652e-01 -4.25306469e-01 -8.87867451e-01 5.83123803e-01
6.38884842e-01 -8.08900774e-01 1.18113875e+00 -5.09793878e-01
4.50616628e-01 -4.77355206e-03 -1.54048800e-01 -1.13941646e+00
-5.56457281e-01 -7.54683256e-01 -6.96341455e-01 1.27712202e+00
7.31158793e-01 -6.47392154e-01 7.51942098e-01 6.19050384e-01
-3.15933943e-01 -1.37834573e+00 -8.17088485e-01 -7.64091015e-01
-8.70567337e-02 -7.29040861e-01 5.44933319e-01 6.24934494e-01
-4.72984761e-01 5.90841472e-01 -4.30242002e-01 -8.10220242e-02
6.12068594e-01 -1.10109830e-02 6.07528985e-01 -1.17348230e+00
-7.39350557e-01 -8.10086191e-01 1.93502590e-01 -1.41607463e+00
4.65237796e-02 -9.18922901e-01 -2.78680753e-02 -1.41720653e+00
4.08569008e-01 -6.90533519e-01 -5.63962102e-01 7.14956701e-01
-5.47892272e-01 2.69877076e-01 1.53228834e-01 4.47275907e-01
-6.08286142e-01 6.22227550e-01 1.09351695e+00 2.38310173e-02
-2.87407070e-01 1.05376393e-01 -8.81044686e-01 7.84527123e-01
8.98895562e-01 -6.48812652e-01 -5.13494432e-01 -6.99971259e-01
3.11778367e-01 -3.58555466e-01 5.64375147e-02 -9.01736736e-01
1.54437304e-01 2.40242872e-02 3.69873643e-01 -5.69631636e-01
4.37700480e-01 -3.46099406e-01 -5.34512550e-02 4.21200961e-01
-2.76642859e-01 -2.53279716e-01 3.56675714e-01 6.62853301e-01
-1.39840603e-01 -2.70007074e-01 8.22550654e-01 5.88364378e-02
-7.22100317e-01 3.87611300e-01 3.44729088e-02 2.12060601e-01
4.21118766e-01 -2.09772453e-01 -2.01480284e-01 3.52917574e-02
-2.87653208e-01 2.68064976e-01 -8.66718665e-02 2.34051630e-01
4.26631093e-01 -1.04439807e+00 -6.98353767e-01 1.37413442e-01
-4.04449590e-02 1.15139425e-01 3.46950382e-01 1.00668871e+00
-2.36818224e-01 7.83644736e-01 2.73413301e-01 -5.15765250e-01
-1.26366448e+00 1.60999820e-01 2.14934777e-02 -9.48631167e-01
-5.59087098e-01 1.24375737e+00 3.62519473e-01 -1.97617039e-01
3.05535555e-01 -4.81951058e-01 -8.12191516e-03 -4.11688201e-02
5.72293818e-01 5.15205681e-01 3.77932042e-01 -1.33828864e-01
-2.53761709e-01 3.94532681e-01 -4.73268956e-01 1.24708578e-01
1.41235852e+00 1.29983742e-02 1.58915043e-01 -4.31992859e-02
1.12951839e+00 -3.56520228e-02 -1.48727977e+00 -5.49389601e-01
7.66825601e-02 -4.64350551e-01 5.39666355e-01 -9.59921956e-01
-1.11167181e+00 8.53619933e-01 1.86117485e-01 -2.70986408e-01
1.37395132e+00 -1.25267357e-01 1.11979735e+00 6.55482173e-01
4.70043778e-01 -1.21782625e+00 -7.55736828e-02 7.07887173e-01
6.79076493e-01 -9.80218291e-01 2.90325254e-01 -3.33459496e-01
-5.08403659e-01 7.03106403e-01 4.81452852e-01 2.63037175e-01
3.76990408e-01 2.14854643e-01 -3.93592060e-01 3.36466283e-02
-1.04007816e+00 -4.70845625e-02 6.09865189e-01 1.03762202e-01
3.41047943e-01 5.09165116e-02 -3.32751215e-01 7.69232512e-01
-3.87705505e-01 -1.06783099e-01 -2.41192698e-01 6.77100539e-01
-6.56886637e-01 -1.05474341e+00 -5.87719604e-02 8.90103817e-01
-7.63703585e-01 -5.65642595e-01 1.31741334e-02 7.57879138e-01
-1.64687663e-01 8.41001034e-01 -2.24786639e-01 -3.85164618e-01
2.98817277e-01 7.49124680e-03 4.58703369e-01 -4.95079309e-01
-6.45891726e-01 2.17231199e-01 3.75973165e-01 -6.92312837e-01
2.95470357e-02 -2.92078167e-01 -1.08738089e+00 -4.02874261e-01
-5.07134140e-01 -1.41641358e-02 6.41276598e-01 9.43681121e-01
7.01046765e-01 5.85993946e-01 3.11095178e-01 -8.51836264e-01
-1.05258238e+00 -1.06573522e+00 -3.09266746e-01 4.43066955e-02
-1.35654509e-02 -8.34398031e-01 -2.91427970e-01 -2.40133584e-01] | [8.779716491699219, 3.5550107955932617] |
d17a8d9b-097d-463a-bd62-8238c961dcbf | population-wise-labeling-of-sulcal-graphs | 2301.13532 | null | https://arxiv.org/abs/2301.13532v1 | https://arxiv.org/pdf/2301.13532v1.pdf | Population-wise Labeling of Sulcal Graphs using Multi-graph Matching | Population-wise matching of the cortical fold is necessary to identify biomarkers of neurological or psychiatric disorders. The difficulty comes from the massive interindividual variations in the morphology and spatial organization of the folds. This task is challenging at both methodological and conceptual levels. In the widely used registration-based techniques, these variations are considered as noise and the matching of folds is only implicit. Alternative approaches are based on the extraction and explicit identification of the cortical folds. In particular, representing cortical folding patterns as graphs of sulcal basins-termed sulcal graphs-enables to formalize the task as a graph-matching problem. In this paper, we propose to address the problem of sulcal graph matching directly at the population level using multi-graph matching techniques. First, we motivate the relevance of multi-graph matching framework in this context. We then introduce a procedure to generate populations of artificial sulcal graphs, which allows us benchmarking several state of the art multi-graph matching methods. Our results on both artificial and real data demonstrate the effectiveness of multi-graph matching techniques to obtain a population-wise consistent labeling of cortical folds at the sulcal basins level. | ['Guillaume Auzias', 'S. Takerkart', 'François-Xavier Dupé', 'Rohit Yadav'] | 2023-01-31 | null | null | null | null | ['graph-matching'] | ['graphs'] | [ 2.18218207e-01 1.59053907e-01 2.74962187e-01 -2.14015618e-01
-6.23307645e-01 -5.45652807e-01 5.83543539e-01 5.46916306e-01
-2.66242743e-01 5.18622696e-01 -1.33864969e-01 1.78766057e-01
-3.12191546e-01 -8.06482553e-01 -6.07306719e-01 -5.49317658e-01
-3.79097164e-02 6.35234058e-01 4.09537703e-01 -2.62423694e-01
3.94098252e-01 7.07341135e-01 -1.51835454e+00 1.54192418e-01
9.24195051e-01 5.03153145e-01 6.94650710e-02 3.65913898e-01
-2.43304372e-01 -3.28216627e-02 -2.93783784e-01 -5.08337855e-01
2.18247190e-01 -6.60933137e-01 -8.40803802e-01 1.25939026e-01
9.68587101e-01 4.80782956e-01 3.09447765e-01 1.36451960e+00
4.52952832e-01 -3.22564125e-01 8.52463663e-01 -9.53114986e-01
-3.49011540e-01 2.73827672e-01 -6.85975611e-01 1.45382434e-01
3.57081711e-01 -2.63448477e-01 9.33095634e-01 -6.30924284e-01
9.38063860e-01 1.18036473e+00 6.93062186e-01 4.11468178e-01
-1.53736758e+00 -4.37739998e-01 -4.60995547e-02 2.52690434e-01
-1.40198815e+00 -3.01796913e-01 8.46061587e-01 -9.08177137e-01
5.37996709e-01 2.95133471e-01 8.54554832e-01 5.90439439e-01
3.66846532e-01 1.50019333e-01 1.62323105e+00 -5.61968923e-01
2.81351089e-01 -3.93933296e-01 4.83022094e-01 8.50418150e-01
4.52375263e-01 -2.71953106e-01 -4.49218750e-01 -2.48474091e-01
9.07362282e-01 -2.29259804e-01 -1.42652631e-01 -6.10507488e-01
-1.15558207e+00 6.29280984e-01 4.31915253e-01 9.59279716e-01
-3.24935555e-01 4.98431325e-02 2.49096200e-01 1.36552289e-01
4.88799304e-01 3.20463479e-01 3.56942639e-02 4.89698380e-01
-1.13001871e+00 4.20431823e-01 4.29454625e-01 6.01665676e-01
1.10729957e+00 -3.57351005e-01 7.72166699e-02 7.83520043e-01
4.59362179e-01 1.37276396e-01 5.63600779e-01 -6.44651115e-01
3.37766528e-01 1.19641411e+00 -4.39729989e-01 -1.24480855e+00
-5.36233366e-01 -1.41447991e-01 -9.17346299e-01 4.27516222e-01
8.30436230e-01 2.00559571e-01 -8.31686854e-01 1.80309677e+00
4.34880316e-01 2.50966221e-01 -3.10067713e-01 6.81671143e-01
4.54933137e-01 -2.75837660e-01 8.27252418e-02 -1.96787834e-01
1.60622978e+00 -3.75194937e-01 -5.39779067e-01 -2.29286447e-01
4.69483435e-01 -5.93795955e-01 7.57659137e-01 -1.28519714e-01
-1.09791589e+00 -2.72521794e-01 -1.04507256e+00 2.71280110e-01
-5.23128569e-01 1.48831289e-02 2.10718811e-01 6.27902210e-01
-1.25557339e+00 6.80110335e-01 -8.96911144e-01 -7.53281057e-01
4.09005076e-01 4.92958277e-01 -8.49849641e-01 4.19451326e-01
-7.10208654e-01 1.10477614e+00 1.86169490e-01 1.26386613e-01
-7.33552948e-02 -4.52298671e-01 -8.54303300e-01 -1.59323081e-01
-1.66453138e-01 -6.41969323e-01 5.65425694e-01 -1.07145655e+00
-1.00396347e+00 1.55926251e+00 -2.07005113e-01 -1.82445124e-01
6.25060081e-01 4.34574306e-01 -1.80553526e-01 1.70767769e-01
1.98062032e-01 1.65246859e-01 7.97603548e-01 -1.18910003e+00
-6.84316531e-02 -9.46880579e-01 -3.27209681e-01 -9.15789157e-02
2.50499342e-02 4.66900691e-02 -3.97023447e-02 -6.72016501e-01
7.00040638e-01 -8.40525389e-01 -2.20623434e-01 2.61494163e-02
-1.76903084e-01 -1.95733443e-01 4.15416807e-01 -6.83521748e-01
9.06482339e-01 -2.01206994e+00 3.91862899e-01 5.23608387e-01
6.95375264e-01 7.08302995e-03 -5.08501567e-02 2.64544100e-01
-3.27058077e-01 7.19719455e-02 -5.75183809e-01 -2.06940502e-01
-2.22274348e-01 -8.38453323e-02 3.50839674e-01 9.71213102e-01
1.11794814e-01 9.50036347e-01 -7.20111370e-01 -7.76670873e-01
1.68636039e-01 3.38469505e-01 -4.83560324e-01 3.62678580e-02
1.22070938e-01 4.95962679e-01 -4.92876083e-01 4.54115927e-01
7.74090350e-01 -2.20892727e-01 4.73088950e-01 -3.31088066e-01
-3.24603096e-02 -3.17692012e-01 -1.14816141e+00 1.77278507e+00
-1.37651652e-01 3.88474733e-01 2.08566740e-01 -1.33697736e+00
1.05231476e+00 2.51309752e-01 5.78288913e-01 -7.27891803e-01
1.66598842e-01 6.33550465e-01 2.41997391e-01 -2.01832637e-01
4.24551666e-02 -4.40749466e-01 6.72115758e-02 3.57494503e-01
4.30843830e-01 -1.75927743e-01 2.65576392e-01 -2.19586998e-01
9.81817067e-01 4.50453237e-02 5.52613080e-01 -8.56126666e-01
8.49827766e-01 -3.26060265e-01 4.09884930e-01 1.37640879e-01
-1.77286476e-01 9.02370155e-01 5.30040026e-01 -4.64739770e-01
-1.05231702e+00 -1.14470196e+00 -3.66222024e-01 2.96163529e-01
1.56146422e-01 -3.41137946e-01 -1.20789063e+00 -5.48105955e-01
7.33897090e-02 -6.16894849e-02 -9.42237139e-01 6.29187450e-02
-7.64286280e-01 -1.09285378e+00 4.61375803e-01 2.76887976e-02
2.65664458e-01 -9.55639780e-01 -7.32188106e-01 1.64903998e-01
5.55763729e-02 -9.30472732e-01 -2.23017111e-01 -1.30982846e-01
-9.98615921e-01 -1.56313562e+00 -9.52717483e-01 -8.64471197e-01
1.02299583e+00 -3.16673458e-01 9.64360237e-01 4.39187407e-01
-5.63555241e-01 2.38362178e-01 -2.00361565e-01 2.49472726e-02
-4.44809020e-01 1.52775524e-02 -8.38800520e-02 3.46197397e-01
1.31161928e-01 -1.09917855e+00 -5.69404185e-01 3.04967970e-01
-7.60000885e-01 -3.13947946e-02 4.31896418e-01 5.13816059e-01
7.69758582e-01 -6.47853136e-01 5.68680406e-01 -1.07614315e+00
8.71872842e-01 -2.98518896e-01 -6.15216792e-01 6.71226025e-01
-6.09623313e-01 3.53822976e-01 4.20379847e-01 -1.81057990e-01
-4.06852543e-01 3.80246043e-02 -1.08466908e-01 -8.65875278e-03
-2.39422157e-01 3.07396978e-01 -7.42225349e-02 -5.98231554e-01
7.92172611e-01 1.48807541e-01 2.37767980e-01 -4.85995740e-01
2.75395453e-01 4.44476664e-01 3.92253548e-01 -6.34022713e-01
5.26316166e-01 3.97198141e-01 4.98713344e-01 -8.15079749e-01
-2.10523143e-01 -3.90587866e-01 -1.06557560e+00 -5.21419466e-01
9.40712214e-01 -2.76118934e-01 -6.11891687e-01 5.37372291e-01
-1.21692884e+00 -2.32982412e-01 -1.58322394e-01 9.48222652e-02
-8.37706447e-01 6.30329251e-01 -3.16883683e-01 -4.71689612e-01
-4.36038375e-01 -1.29993927e+00 1.19822943e+00 8.04993734e-02
-3.41480583e-01 -1.21040976e+00 6.48812234e-01 6.91016391e-02
1.88521072e-01 6.89852595e-01 1.19200206e+00 -5.56720495e-01
-3.37618619e-01 -1.18936278e-01 -2.22779632e-01 -1.76655337e-01
3.64944696e-01 -1.68560311e-01 -8.09755206e-01 -8.76121521e-02
-1.03781214e-02 9.67097506e-02 4.76801366e-01 2.38119334e-01
7.60286450e-01 5.24191707e-02 -3.78227353e-01 5.08566201e-01
1.79004574e+00 -1.15520127e-01 5.62286198e-01 2.14409634e-01
6.16707087e-01 1.05495965e+00 -9.88165662e-02 1.81118364e-03
2.59936690e-01 8.99137497e-01 1.35358170e-01 1.33564845e-02
-4.16977555e-01 9.76438746e-02 -2.32428834e-01 9.40841079e-01
-4.00281012e-01 2.12543771e-01 -1.17834008e+00 5.53716481e-01
-1.89795375e+00 -8.64921212e-01 -5.36800802e-01 2.29126072e+00
6.25916719e-01 -2.14513972e-01 2.23232388e-01 8.57432187e-02
1.06391573e+00 -1.62578136e-01 -2.49716952e-01 -3.02989781e-01
-2.00059861e-01 4.88402933e-01 2.99842805e-01 6.43774867e-01
-8.07156265e-01 6.58650219e-01 6.79537249e+00 6.02013588e-01
-1.02388465e+00 2.49643415e-01 5.68938732e-01 4.98705328e-01
-3.29996288e-01 8.44360217e-02 -4.79555488e-01 4.08270091e-01
7.26814747e-01 -3.14328730e-01 5.21813452e-01 1.94751561e-01
-5.27220555e-02 -1.94029585e-02 -1.06609333e+00 1.08902478e+00
1.74612090e-01 -1.31201518e+00 8.73581395e-02 1.58115968e-01
4.66027141e-01 -1.50698662e-01 -3.99793625e-01 -4.63340193e-01
-1.15694091e-01 -9.65633750e-01 6.41413271e-01 7.87902057e-01
8.48301589e-01 -3.10383618e-01 4.96357113e-01 1.29663264e-02
-1.51931775e+00 4.41670060e-01 -1.15963332e-01 1.94990218e-01
1.52170911e-01 4.56916898e-01 -4.03051376e-01 8.40593755e-01
4.08556074e-01 5.63211262e-01 -8.84590268e-01 1.19285595e+00
-3.11910789e-02 -1.39567032e-02 -1.73353374e-01 8.17641392e-02
-1.04765363e-01 -7.94597507e-01 4.60816234e-01 1.10956335e+00
3.25130552e-01 -1.95173845e-01 -1.43016338e-01 1.27875054e+00
1.39029235e-01 6.25257552e-01 -6.56867623e-01 1.66810647e-01
7.68325701e-02 1.52289355e+00 -1.19208348e+00 -5.61527833e-02
-3.79542381e-01 7.91434348e-01 8.39966416e-01 5.97519763e-02
-2.69493997e-01 -1.90554693e-01 4.56896693e-01 3.95775795e-01
-2.63711691e-01 -2.66705483e-01 -4.46632296e-01 -1.14582872e+00
2.15767205e-01 -6.75315201e-01 2.26033732e-01 -4.21757549e-01
-1.32801628e+00 8.31427276e-01 1.35431215e-01 -9.56405461e-01
-2.49937400e-01 -6.03991866e-01 -7.36676633e-01 1.04815829e+00
-1.05422962e+00 -1.38862395e+00 -3.95300180e-01 5.59025943e-01
2.11486258e-02 1.48799494e-02 9.26945567e-01 2.89667457e-01
-2.35761121e-01 3.29839587e-01 -9.11552310e-02 3.73994671e-02
4.67689157e-01 -1.46026790e+00 4.26168263e-01 8.25687766e-01
1.61410898e-01 7.17134178e-01 6.34787917e-01 -7.13337839e-01
-1.02076423e+00 -6.95421517e-01 1.11997402e+00 -3.80930394e-01
8.28010619e-01 -5.87917805e-01 -1.01274157e+00 5.18018305e-01
4.02650200e-02 1.98682427e-01 4.56617177e-01 -6.77707884e-03
-1.64404705e-01 -5.20474240e-02 -1.26325929e+00 5.60953975e-01
1.06113994e+00 -5.81581712e-01 -7.46792078e-01 2.46317714e-01
-4.53287698e-02 6.05779253e-02 -1.15994596e+00 3.16557676e-01
6.04193270e-01 -1.02293718e+00 8.78867865e-01 -3.03192526e-01
-8.71448405e-03 -2.44760394e-01 1.15558773e-01 -1.26823604e+00
-1.24353878e-01 -3.98324966e-01 3.54133636e-01 1.13946438e+00
2.52097785e-01 -7.72076309e-01 8.38033855e-01 5.32962799e-01
6.74462542e-02 -6.09038770e-01 -1.23311210e+00 -6.56230927e-01
3.71875077e-01 1.33247226e-01 4.83657032e-01 9.60860193e-01
3.47813755e-01 1.65477231e-01 2.03367263e-01 -2.20848247e-01
8.34281445e-01 2.93891996e-01 4.19515610e-01 -1.60929847e+00
6.47046193e-02 -7.81966805e-01 -1.04038048e+00 -8.42863545e-02
3.39397401e-01 -1.25617385e+00 -1.50997341e-01 -1.51082051e+00
2.26138636e-01 -3.65087777e-01 -2.12134402e-02 1.75550133e-01
-4.93115932e-02 3.09411258e-01 1.47570804e-01 1.18123375e-01
-2.09519431e-01 2.79168338e-01 1.34887135e+00 7.53735900e-02
2.87167588e-03 -3.34956080e-01 -2.55371034e-01 6.98841035e-01
8.33921850e-01 -3.13325316e-01 -1.42706499e-01 -1.27059817e-01
1.96609944e-01 -7.89554715e-02 4.51844573e-01 -1.13375592e+00
2.64070511e-01 6.89086616e-02 -5.60266487e-02 -8.96297917e-02
5.31310327e-02 -7.45624065e-01 5.04138708e-01 5.68906307e-01
-4.60037552e-02 5.19240975e-01 -9.96889360e-03 3.25875342e-01
-2.23618850e-01 -2.48579174e-01 1.05152118e+00 -3.44227463e-01
-2.55675673e-01 3.08426112e-01 -3.40607375e-01 8.50142762e-02
1.06676292e+00 -4.49626863e-01 -1.01183929e-01 2.53268123e-01
-9.68340516e-01 -2.80436516e-01 7.01306045e-01 1.63284823e-01
5.25818348e-01 -1.30900455e+00 -6.54095650e-01 2.87093192e-01
2.53067493e-01 -3.59551549e-01 1.25303373e-01 1.12655628e+00
-7.49462426e-01 1.17033295e-01 -7.64871657e-01 -5.81329226e-01
-1.45519376e+00 3.36291075e-01 8.01057577e-01 -2.41814256e-01
-4.10324097e-01 3.95534009e-01 2.68626153e-01 -4.85178530e-01
-3.50428998e-01 -3.89511943e-01 -3.95448983e-01 2.77594805e-01
1.07208349e-01 2.42494568e-01 3.29015017e-01 -1.06507683e+00
-5.49834609e-01 1.28319192e+00 2.71363944e-01 -5.17267510e-02
1.29319668e+00 -3.00660469e-02 -5.91998160e-01 5.09946227e-01
1.05626988e+00 3.30528654e-02 -6.15778625e-01 1.24590673e-01
4.18358505e-01 -2.19573647e-01 -3.80306870e-01 -4.29343075e-01
-9.79230464e-01 7.52757728e-01 7.92255521e-01 3.42011839e-01
8.75429273e-01 1.68947324e-01 2.41053358e-01 -1.94109559e-01
7.22059071e-01 -7.47606754e-01 -2.46904686e-01 4.85690497e-02
9.96314526e-01 -8.37784529e-01 -1.34800509e-01 -5.44489861e-01
9.51849893e-02 1.21478152e+00 4.72460806e-01 -6.82801783e-01
7.41230369e-01 1.09026663e-01 -1.17615148e-01 -5.99598050e-01
-7.87857249e-02 -3.22958022e-01 6.69592917e-01 4.08734858e-01
7.27888227e-01 7.59590715e-02 -9.75633383e-01 3.21046591e-01
-7.65631720e-02 -2.39425927e-01 1.63234234e-01 7.28313088e-01
-3.59075546e-01 -1.68384814e+00 -3.73604596e-01 5.53556144e-01
-4.46464032e-01 9.09574777e-02 -7.18059838e-01 8.04359555e-01
2.21302286e-01 5.60100555e-01 -2.32534185e-02 -2.37221792e-01
4.34554428e-01 1.15249380e-01 9.65752065e-01 -5.99563181e-01
-8.59438539e-01 -2.65244190e-02 -2.14129999e-01 -5.38444698e-01
-7.45973885e-01 -9.51313376e-01 -1.16385722e+00 -8.69816467e-02
-2.57583827e-01 4.98899668e-02 5.53813696e-01 1.08833361e+00
2.85388857e-01 3.17249894e-01 1.41395003e-01 -7.72912979e-01
-7.06892386e-02 -8.90689373e-01 -8.19299221e-01 9.79603589e-01
2.40364179e-01 -8.29477251e-01 -1.65988937e-01 7.08078295e-02] | [14.046255111694336, -2.4460673332214355] |
39b952e1-9cca-4cfb-92c2-be8213b00985 | an-accurate-iris-segmentation-framework-under | null | null | http://openaccess.thecvf.com/content_iccv_2015/html/Zhao_An_Accurate_Iris_ICCV_2015_paper.html | http://openaccess.thecvf.com/content_iccv_2015/papers/Zhao_An_Accurate_Iris_ICCV_2015_paper.pdf | An Accurate Iris Segmentation Framework Under Relaxed Imaging Constraints Using Total Variation Model | This paper proposes a novel and more accurate iris segmentation framework to automatically segment iris region from the face images acquired with relaxed imaging under visible or near-infrared illumination, which provides strong feasibility for applications in surveillance, forensics and the search for missing children, etc. The proposed framework is built on a novel total-variation based formulation which uses l1 norm regularization to robustly suppress noisy texture pixels for the accurate iris localization. A series of novel and robust post processing operations are introduced to more accurately localize the limbic boundaries. Our experimental results on three publicly available databases, i.e., FRGC, UBIRIS.v2 and CASIA.v4-distance, achieve significant performance improvement in terms of iris segmentation accuracy over the state-of-the-art approaches in the literature. Besides, we have shown that using iris masks generated from the proposed approach helps to improve iris recognition performance as well. Unlike prior work, all the implementations in this paper are made publicly available to further advance research and applications in biometrics at-d-distance. | ['Kumar Ajay', 'Zijing Zhao'] | 2015-12-01 | null | null | null | iccv-2015-12 | ['iris-segmentation'] | ['medical'] | [ 3.89448553e-01 -4.25791919e-01 -2.95999497e-01 -4.30298001e-01
-6.31231844e-01 -3.64974976e-01 1.21246822e-01 -1.48585305e-01
-1.70907438e-01 4.47542340e-01 5.56396358e-02 -2.20669121e-01
-3.92904222e-01 -3.19783241e-01 -3.95713836e-01 -1.05773020e+00
2.37584785e-01 -3.86785269e-02 -2.63168871e-01 2.28953928e-01
6.06605411e-01 6.22632325e-01 -1.87398207e+00 -1.12644181e-01
1.30151355e+00 7.45725274e-01 -5.56282103e-01 4.82283771e-01
1.57401115e-01 2.54797578e-01 -4.19202209e-01 -3.32935572e-01
5.96515596e-01 -3.84719014e-01 -6.52268887e-01 5.09521067e-01
1.04930365e+00 -3.60619664e-01 4.61433977e-02 1.39136255e+00
6.95285559e-01 1.37160778e-01 4.08712506e-01 -3.80513638e-01
-7.40884662e-01 -1.03477249e-02 -1.50131321e+00 2.89837837e-01
3.43070418e-01 3.49141330e-01 2.25396633e-01 -4.83624458e-01
3.11907142e-01 1.10153139e+00 5.10647416e-01 5.07183671e-01
-9.61169779e-01 -6.91768765e-01 -3.30050200e-01 -5.63705489e-02
-1.56752145e+00 -5.91296434e-01 5.52596390e-01 -3.96005213e-01
4.77808088e-01 6.76203430e-01 4.00306642e-01 3.75156939e-01
-2.10354999e-01 7.15684414e-01 1.77937651e+00 -8.10603619e-01
-3.60276043e-01 5.59156314e-02 2.64280081e-01 8.62855852e-01
3.95244360e-01 5.45304537e-01 -3.00838500e-01 -1.11753784e-01
9.71897423e-01 -7.95419961e-02 -2.23096907e-01 9.07828752e-03
-8.10052514e-01 3.37742537e-01 1.80344000e-01 3.46017510e-01
-3.04745823e-01 -6.60596907e-01 8.04400742e-02 -1.46885007e-03
6.27789319e-01 1.70535386e-01 -4.60081622e-02 2.85231625e-03
-1.07007205e+00 -1.88913822e-01 1.53450862e-01 5.74049115e-01
5.07451952e-01 -2.15625763e-01 -4.63039398e-01 1.08848548e+00
5.16434908e-01 6.19362354e-01 2.35410780e-01 -5.10257244e-01
1.75142094e-01 8.54137301e-01 1.70547724e-01 -9.56517816e-01
-2.01272875e-01 -5.11261404e-01 -7.34927654e-01 2.05491111e-01
5.34294546e-01 -2.12512076e-01 -1.39041042e+00 9.63602066e-01
8.75344455e-01 9.02294815e-01 3.80695611e-02 1.19695592e+00
1.05150306e+00 6.07010908e-02 -2.38103926e-01 -2.46406361e-01
1.39741850e+00 -8.17187130e-01 -6.55955374e-01 3.31895977e-01
1.53343871e-01 -1.46869862e+00 7.46953964e-01 5.60693324e-01
-9.67737138e-01 -6.64224148e-01 -8.17668021e-01 1.03145406e-01
9.72507298e-02 7.45043099e-01 5.39278209e-01 1.35371184e+00
-1.04535556e+00 9.51767713e-02 -8.75509799e-01 -4.68583226e-01
6.06339633e-01 7.19168961e-01 -3.20008218e-01 -7.81335533e-02
-4.21019554e-01 4.24034536e-01 4.76684645e-02 3.83897722e-01
-9.12322998e-02 -3.54412317e-01 -8.76124799e-01 -5.07542431e-01
1.85503840e-01 -2.43350521e-01 6.43125534e-01 -8.90678823e-01
-1.76053655e+00 1.41670561e+00 -7.35849738e-01 -1.72471628e-01
8.20993856e-02 9.52126645e-03 -5.35094798e-01 1.83527753e-01
-2.55494207e-01 7.41039813e-02 8.62626791e-01 -8.83122861e-01
-6.05554163e-01 -1.04036438e+00 -2.68448740e-01 1.95057571e-01
-1.64231643e-01 6.41442716e-01 -7.21904576e-01 -7.49830008e-01
2.19414771e-01 -9.50350940e-01 -4.04575206e-02 -2.29040444e-01
-5.59195280e-01 -2.12296113e-01 5.50612211e-01 -9.41747367e-01
1.13900805e+00 -2.23320270e+00 -2.16352135e-01 5.60060859e-01
-1.05454355e-01 1.09099054e+00 -8.15258846e-02 -2.41933763e-01
-1.07238993e-01 6.31905068e-03 -2.47972757e-01 -2.91280448e-01
-4.42684054e-01 -1.52632177e-01 2.75642991e-01 1.06913519e+00
3.78577621e-03 5.30919015e-01 -4.50156093e-01 -5.23096144e-01
5.84984422e-01 8.25935423e-01 -1.53318524e-01 1.65775754e-02
3.51725817e-01 9.50284302e-01 -4.20257717e-01 1.18403280e+00
1.28318191e+00 1.04749948e-01 -3.48396003e-01 -8.67439434e-03
-2.77612031e-01 -3.27547878e-01 -1.39057362e+00 1.70457840e+00
-6.79891109e-02 1.91572040e-01 2.87334025e-01 -8.30838442e-01
1.05226076e+00 3.88557732e-01 3.51915061e-01 -4.66016740e-01
3.10532361e-01 8.43391865e-02 -5.71458600e-03 -6.80578351e-01
2.29408890e-01 4.63140942e-02 7.90905118e-01 3.00460964e-01
-3.65983099e-01 3.69681060e-01 1.35606185e-01 -4.73789275e-01
2.22044975e-01 2.28809640e-01 2.41712660e-01 -2.43923381e-01
1.29096711e+00 -1.48143739e-01 6.60281181e-01 4.46024209e-01
-5.57847261e-01 4.68514472e-01 2.08016863e-04 -4.57896799e-01
-4.25478071e-01 -6.50718570e-01 -9.01676595e-01 3.66221279e-01
2.58344948e-01 9.63513553e-02 -9.59558308e-01 -5.53443968e-01
8.30001850e-03 7.14877546e-02 -5.64960182e-01 6.10907733e-01
-2.09143743e-01 -1.27938318e+00 5.11072695e-01 -3.67739610e-02
7.78842509e-01 -5.66451669e-01 -1.38178080e-01 -3.65302473e-01
1.66111574e-01 -8.25675309e-01 -5.87453246e-01 -1.03871858e+00
-8.87940407e-01 -1.42690551e+00 -9.64297593e-01 -8.68588328e-01
1.22433245e+00 2.27977261e-01 4.68656600e-01 3.83297861e-01
-1.00579274e+00 2.65056968e-01 -2.29829550e-01 -4.27087724e-01
8.03442970e-02 -4.18496519e-01 2.61000246e-02 7.21303165e-01
1.00520980e+00 8.02312642e-02 -9.25611854e-01 3.99510711e-01
-5.23902774e-01 -4.38791424e-01 5.15634477e-01 8.96305382e-01
8.93677235e-01 2.82566935e-01 5.98337129e-02 -9.75404263e-01
3.42879534e-01 1.05694778e-01 -1.06879950e+00 4.00628418e-01
-7.81555891e-01 -1.99951962e-01 4.38294895e-02 -6.44243881e-02
-1.44490063e+00 1.08988523e-01 -6.98587159e-03 -1.29975900e-01
-6.47637844e-01 1.53803498e-01 3.65710258e-01 -9.01430488e-01
5.80044806e-01 2.40387842e-01 3.35564762e-01 -8.37912500e-01
4.11826111e-02 1.22371662e+00 6.47694528e-01 -5.85588634e-01
7.04100728e-01 6.28997207e-01 2.09889308e-01 -8.97125959e-01
-6.61428869e-01 -1.00284052e+00 -6.97851300e-01 4.55039814e-02
7.68677890e-01 -8.35863829e-01 -1.02786565e+00 9.56902027e-01
-6.09515607e-01 3.38845164e-01 2.19966173e-01 8.27008069e-01
-8.84289574e-03 8.38472605e-01 -5.86271405e-01 -1.07770002e+00
-5.94514608e-01 -1.46881700e+00 1.03734159e+00 1.09410155e+00
4.45988387e-01 -9.27081525e-01 1.61909640e-01 1.03588808e+00
2.57201672e-01 4.10573274e-01 4.58284646e-01 -1.26958206e-01
-5.42938709e-01 -1.71671078e-01 -4.61300135e-01 2.97571927e-01
5.08480847e-01 4.97518599e-01 -1.13360548e+00 -6.33561194e-01
-9.84467268e-02 7.60436133e-02 6.90413117e-01 7.74444580e-01
1.22689474e+00 -2.15279013e-01 -2.77157515e-01 1.21030700e+00
1.62687814e+00 3.12455207e-01 8.53796780e-01 2.20443383e-01
4.54245538e-01 7.54665434e-01 7.50865459e-01 3.09990108e-01
-1.33802965e-02 6.50425494e-01 3.02251074e-02 -7.27944851e-01
-2.68756449e-01 2.24561110e-01 -3.03594291e-01 3.35983336e-01
-5.44149697e-01 2.77557135e-01 -9.47424650e-01 7.74948001e-01
-1.61513507e+00 -8.61540675e-01 -3.53415221e-01 2.66583776e+00
1.02239561e+00 -6.66055620e-01 1.64771557e-01 -4.97837476e-02
9.89738226e-01 -9.48928371e-02 -5.19691885e-01 -4.66502309e-01
-1.18547134e-01 8.43462050e-01 6.12639189e-01 6.30799770e-01
-1.38459277e+00 8.71306360e-01 5.95007658e+00 7.83661008e-01
-1.28323710e+00 -1.24019571e-01 8.96653056e-01 -9.83038023e-02
3.59733164e-01 -1.07512295e-01 -8.15547884e-01 2.93111145e-01
5.17259359e-01 2.90703684e-01 5.46866477e-01 1.76800922e-01
3.05594444e-01 -3.84300888e-01 -2.43760183e-01 1.35383725e+00
3.41618896e-01 -9.79636788e-01 -4.73380983e-01 1.97692722e-01
1.11776447e+00 -2.23450959e-01 6.74026012e-01 -5.70508182e-01
-1.04433224e-01 -1.30870390e+00 -5.48688650e-01 7.59730339e-01
1.18866360e+00 -9.65285778e-01 9.07765031e-01 -6.34443834e-02
-8.80150914e-01 2.60234535e-01 -3.20878774e-01 1.77153185e-01
-4.42794949e-01 5.78654587e-01 -7.83493340e-01 7.80829012e-01
6.48268521e-01 8.67835879e-01 -7.37733424e-01 1.50121570e+00
-1.05075322e-01 7.54849732e-01 -1.80672362e-01 4.07464862e-01
-7.46460333e-02 -6.24782443e-01 5.81699789e-01 8.09715390e-01
1.01694338e-01 3.80909562e-01 -1.25966981e-01 6.89607441e-01
2.38951206e-01 5.87935746e-01 -3.08842450e-01 1.38740510e-01
4.79831547e-02 1.25357246e+00 -5.26385307e-01 7.78343389e-03
-7.18103349e-01 6.06262684e-01 -3.32977802e-01 4.50822413e-01
-3.20957929e-01 -5.30427277e-01 7.28176951e-01 1.09800979e-01
1.17073879e-01 4.18626033e-02 -4.41487283e-01 -1.18072641e+00
8.13725516e-02 -1.13274992e+00 3.65260959e-01 -2.47969300e-01
-1.13741457e+00 4.14675832e-01 -2.48875052e-01 -1.15420842e+00
7.78848678e-02 -5.70497215e-01 -5.26498377e-01 1.58744621e+00
-1.74708128e+00 -1.41032934e+00 -2.28842899e-01 7.46983051e-01
9.82488990e-02 -5.32836914e-01 6.53275192e-01 3.74956071e-01
-1.17005777e+00 1.05130219e+00 2.59246618e-01 3.84833992e-01
1.06075406e+00 -9.91017282e-01 1.37143716e-01 1.33226883e+00
1.35852039e-01 1.09594917e+00 3.29797000e-01 -6.56249046e-01
-1.32275581e+00 -6.96459889e-01 5.53253055e-01 -2.33003438e-01
-1.78057775e-02 3.39767069e-01 -7.08853602e-01 4.15579498e-01
3.29892248e-01 1.57841802e-01 9.25848961e-01 3.31473380e-01
-4.24056686e-02 -2.18102962e-01 -1.61194527e+00 3.45800668e-01
6.19780362e-01 -3.83897722e-01 -2.90864557e-01 3.71450126e-01
-2.06045508e-01 -9.52741325e-01 -1.03000879e+00 7.21580565e-01
5.59796154e-01 -1.20173085e+00 9.08302248e-01 -4.19042140e-01
-1.96957916e-01 -7.19292104e-01 3.41840178e-01 -6.54809594e-01
8.71806294e-02 -9.87879932e-01 3.33737999e-01 1.40940285e+00
1.95317194e-01 -9.31181312e-01 9.72838521e-01 6.46693528e-01
3.96570235e-01 -7.13872015e-01 -8.13067138e-01 -3.99524689e-01
-2.76756227e-01 1.91383466e-01 7.45525539e-01 8.31902683e-01
-3.03419590e-01 -4.16907370e-01 -2.84791112e-01 5.79241157e-01
1.20898700e+00 4.39197659e-01 8.65504622e-01 -1.21131754e+00
-1.66614987e-02 -2.05842331e-01 -7.18431532e-01 -8.14368725e-01
-6.78793117e-02 -5.16881764e-01 -3.56482744e-01 -1.01756132e+00
3.07987183e-01 -5.81540227e-01 -3.30430388e-01 5.16237617e-01
-4.88525778e-01 7.33807147e-01 -4.04250026e-01 2.49944195e-01
2.47575510e-02 -1.01338290e-01 1.54279196e+00 -1.03486657e-01
-4.58223790e-01 4.54320073e-01 -7.89860964e-01 6.20562136e-01
6.55238688e-01 4.25660564e-03 -2.42539734e-01 -3.57402593e-01
-5.38900554e-01 -5.10658324e-02 1.40902326e-01 -7.78335989e-01
3.18878740e-01 -1.49230734e-01 4.45256740e-01 -4.45194006e-01
7.84368664e-02 -5.39416611e-01 -8.67905095e-02 1.67853594e-01
-4.72069606e-02 -3.94779503e-01 4.06337798e-01 3.00353408e-01
-4.07358766e-01 -5.87840155e-02 1.16007793e+00 1.35724455e-01
-4.95670140e-01 4.18235540e-01 4.62771118e-01 -2.07813025e-01
1.14557278e+00 -4.61922407e-01 -5.77787757e-01 3.11180670e-02
-6.29853666e-01 5.18068410e-02 7.67207563e-01 2.30840504e-01
6.52773142e-01 -8.47947478e-01 -9.45316017e-01 1.00612795e+00
1.42143339e-01 -2.13261023e-01 4.48356450e-01 1.26586890e+00
-7.94451535e-01 6.12717986e-01 -2.70559430e-01 -9.48820591e-01
-2.01070571e+00 3.64000469e-01 4.44595695e-01 2.06177503e-01
-4.63313580e-01 1.05222428e+00 -1.21167675e-01 -3.46231848e-01
3.24315012e-01 -2.44867817e-01 -4.49480921e-01 -5.11686504e-01
9.12361860e-01 3.75352472e-01 1.51187748e-01 -1.14075124e+00
-3.03333610e-01 1.34307945e+00 -3.17048758e-01 1.79024845e-01
9.29684639e-01 -2.34179705e-01 -5.78972459e-01 -2.90387034e-01
7.08804905e-01 4.51811343e-01 -9.12674665e-01 -3.72510523e-01
-2.14480519e-01 -1.31191993e+00 3.68417621e-01 -1.08401549e+00
-1.26870382e+00 7.33366489e-01 1.30864227e+00 -4.65093762e-01
1.52576506e+00 -4.87373233e-01 6.04690850e-01 -3.33794981e-01
2.01824978e-01 -8.53618205e-01 -7.44584680e-01 -1.45494401e-01
3.69086117e-01 -1.56483221e+00 2.92335838e-01 -5.89919746e-01
-4.05271500e-01 9.78718102e-01 5.02076983e-01 1.30777106e-01
5.95556855e-01 -3.24720778e-02 6.69399381e-01 -9.48832855e-02
4.86136302e-02 -5.35000324e-01 9.41534758e-01 7.35384524e-01
8.28849852e-01 3.22514139e-02 -6.00356340e-01 -1.51522130e-01
1.53130069e-01 7.09207803e-02 2.84408152e-01 6.06762052e-01
-1.71083570e-01 -1.47364306e+00 -9.67088759e-01 5.63987136e-01
-1.00653875e+00 5.81013598e-02 -2.63414264e-01 4.73984063e-01
2.93855965e-01 1.36251855e+00 -1.69623911e-01 -6.03506416e-02
-4.32627928e-03 -2.66452432e-01 6.78212345e-01 -5.45814991e-01
-6.59061313e-01 4.65012282e-01 -3.24595869e-01 -4.27410275e-01
-1.12017024e+00 -7.00213909e-01 -7.99966812e-01 -3.58702630e-01
-6.21449828e-01 -4.29904833e-02 7.75171161e-01 7.08076119e-01
4.13551211e-01 -5.86545505e-02 6.52693868e-01 -1.68527156e-01
-2.08861098e-01 -8.57029438e-01 -8.93711507e-01 2.67668158e-01
6.69259608e-01 -5.62536716e-01 -2.06580475e-01 -4.01552096e-02] | [3.7464797496795654, -3.629788637161255] |
81b2b474-751a-440b-b326-d3878610d88a | spatiotemporal-implicit-neural-representation | 2301.00127 | null | https://arxiv.org/abs/2301.00127v2 | https://arxiv.org/pdf/2301.00127v2.pdf | Spatiotemporal implicit neural representation for unsupervised dynamic MRI reconstruction | Supervised Deep-Learning (DL)-based reconstruction algorithms have shown state-of-the-art results for highly-undersampled dynamic Magnetic Resonance Imaging (MRI) reconstruction. However, the requirement of excessive high-quality ground-truth data hinders their applications due to the generalization problem. Recently, Implicit Neural Representation (INR) has appeared as a powerful DL-based tool for solving the inverse problem by characterizing the attributes of a signal as a continuous function of corresponding coordinates in an unsupervised manner. In this work, we proposed an INR-based method to improve dynamic MRI reconstruction from highly undersampled k-space data, which only takes spatiotemporal coordinates as inputs. Specifically, the proposed INR represents the dynamic MRI images as an implicit function and encodes them into neural networks. The weights of the network are learned from sparsely-acquired (k, t)-space data itself only, without external training datasets or prior images. Benefiting from the strong implicit continuity regularization of INR together with explicit regularization for low-rankness and sparsity, our proposed method outperforms the compared scan-specific methods at various acceleration factors. E.g., experiments on retrospective cardiac cine datasets show an improvement of 5.5 ~ 7.1 dB in PSNR for extremely high accelerations (up to 41.6-fold). The high-quality and inner continuity of the images provided by INR has great potential to further improve the spatiotemporal resolution of dynamic MRI, without the need of any training data. | ['Hongjiang Wei', 'Yuyao Zhang', 'Zhiyong Zhang', 'Qing Wu', 'Ruimin Feng', 'Jie Feng'] | 2022-12-31 | null | null | null | null | ['mri-reconstruction'] | ['computer-vision'] | [ 3.75803560e-01 -1.62255272e-01 -8.03219303e-02 -4.79966193e-01
-7.64347136e-01 -6.87177386e-03 1.59663320e-01 -2.11511627e-02
-5.03175020e-01 5.93124926e-01 3.56675714e-01 -1.03640109e-01
-5.12425661e-01 -5.12364626e-01 -6.92189395e-01 -1.03906024e+00
-3.41964453e-01 2.00172886e-01 6.89974949e-02 -1.03326768e-01
-1.31408786e-02 6.05137944e-01 -1.01085889e+00 3.22308272e-01
8.42430532e-01 1.03382897e+00 5.57671189e-01 2.85805792e-01
9.76623967e-02 9.23445404e-01 -3.83574702e-02 2.40281343e-01
1.86712444e-01 -3.67228419e-01 -6.28542185e-01 1.16981812e-01
1.40884906e-01 -4.74244326e-01 -7.98737824e-01 1.03708851e+00
6.65852487e-01 1.82732910e-01 3.63274902e-01 -4.69377995e-01
-7.14566469e-01 3.64232957e-01 -7.46656716e-01 5.48080087e-01
-2.20352679e-01 1.19443916e-01 5.32402098e-01 -1.14503562e+00
6.14572108e-01 5.45242250e-01 6.60996675e-01 2.98538089e-01
-1.40723908e+00 -6.08581126e-01 -2.08592445e-01 3.51104528e-01
-1.49047089e+00 -3.31258446e-01 1.10061681e+00 -4.74645913e-01
6.19007945e-01 1.98436141e-01 6.02675736e-01 7.85787642e-01
2.24663362e-01 6.19328856e-01 1.25569320e+00 -1.91147760e-01
1.72703657e-02 -2.87499219e-01 1.61988914e-01 5.49492896e-01
1.84214413e-01 7.43919313e-02 -3.34796280e-01 -2.73757502e-02
1.38062072e+00 1.93664372e-01 -6.98742986e-01 -5.01406133e-01
-1.52516198e+00 7.92232156e-01 8.97399306e-01 5.81278741e-01
-8.02649021e-01 5.27463108e-02 5.78776419e-01 -1.66676212e-02
3.86950314e-01 4.32617992e-01 -1.22333795e-01 1.20710135e-01
-9.36324537e-01 -1.50094271e-01 4.57089692e-02 5.01675487e-01
4.42390561e-01 5.13124287e-01 -7.29601597e-03 9.69721854e-01
3.57051753e-02 4.68764395e-01 7.87624836e-01 -9.98048723e-01
4.11263883e-01 1.16804115e-01 -6.81126341e-02 -1.43352246e+00
-6.08931124e-01 -1.06266618e+00 -1.50750554e+00 -9.84092355e-02
3.32525939e-01 4.20166403e-02 -7.71126926e-01 1.73083699e+00
2.65115470e-01 3.49153340e-01 -1.71240300e-01 1.45509756e+00
7.26014435e-01 7.25531459e-01 -1.65687814e-01 -6.32371008e-01
1.06202257e+00 -4.98574346e-01 -1.02846539e+00 1.27788678e-01
5.43347299e-01 -4.16303307e-01 1.04728138e+00 4.63456571e-01
-1.12117958e+00 -4.63503718e-01 -1.00481820e+00 7.51640201e-02
2.61176676e-01 3.06660116e-01 5.85568130e-01 1.61598966e-01
-8.86382163e-01 5.99999547e-01 -1.12044442e+00 3.77426803e-01
3.44256490e-01 5.02616584e-01 -5.30844212e-01 -3.39454204e-01
-1.19154835e+00 5.83780229e-01 8.93717259e-02 5.03585517e-01
-7.87862539e-01 -9.74188745e-01 -6.83148503e-01 -2.30338201e-01
3.39736819e-01 -4.18499261e-01 6.53375745e-01 -7.87777424e-01
-1.39097142e+00 6.98016942e-01 1.74549386e-01 -3.41174573e-01
3.81064981e-01 -2.00319782e-01 -5.10805488e-01 6.68252110e-01
6.10052273e-02 2.07550451e-01 8.88233423e-01 -1.23058307e+00
1.02560021e-01 -5.39779186e-01 -1.26441255e-01 1.84920728e-01
-3.62830579e-01 -2.77222484e-01 -3.55757326e-01 -1.04857993e+00
7.43934989e-01 -8.98885131e-01 -5.43376088e-01 1.74972191e-01
-1.00860223e-01 4.91299838e-01 4.86156553e-01 -1.18574476e+00
1.15948224e+00 -2.25789285e+00 2.62216061e-01 2.13375360e-01
5.28709769e-01 1.83559731e-01 -2.52207760e-02 -1.71254396e-01
-4.62788045e-01 -3.90006512e-01 -5.19312739e-01 1.53915465e-01
-6.20818198e-01 1.98031381e-01 -7.25689903e-02 8.78463089e-01
1.43426685e-02 7.22369552e-01 -1.09087276e+00 -3.86156887e-01
2.25048020e-01 8.63582671e-01 -4.93483216e-01 2.75246650e-01
3.40255827e-01 1.09437287e+00 -4.97438967e-01 2.37730652e-01
6.75693333e-01 -5.00382245e-01 2.77027488e-01 -8.33980381e-01
-5.28104343e-02 4.85545211e-02 -9.72225130e-01 2.14669251e+00
-5.01726329e-01 4.91630524e-01 3.00878227e-01 -1.46724415e+00
7.67820120e-01 4.26023006e-01 1.04502344e+00 -1.00507486e+00
1.75909325e-02 3.98080260e-01 2.40144297e-01 -6.80323362e-01
6.53453171e-02 -4.28051889e-01 1.71980351e-01 3.70256722e-01
-1.71300918e-01 2.14536279e-01 -1.72171339e-01 1.34481102e-01
9.31963682e-01 -1.51459128e-01 -4.40052617e-03 -4.15106565e-01
7.15954602e-01 -2.13967830e-01 6.91335678e-01 5.79985559e-01
5.19332103e-02 7.49424458e-01 6.99729696e-02 -6.46716774e-01
-1.13636017e+00 -1.03937721e+00 -5.61368763e-01 4.72779959e-01
2.37106215e-02 2.59932578e-02 -4.74066734e-01 -2.70242602e-01
-3.37997019e-01 3.65869403e-01 -5.88664711e-01 -1.37101769e-01
-1.19322050e+00 -1.03515828e+00 3.16543132e-01 5.09530544e-01
4.72614825e-01 -7.44589269e-01 -4.82567281e-01 4.39655304e-01
-5.20557642e-01 -1.31063175e+00 -3.95226032e-01 1.09451450e-01
-1.41772282e+00 -7.50928998e-01 -1.10327578e+00 -6.96046531e-01
1.02864981e+00 3.94671440e-01 7.13321030e-01 9.42008477e-03
-3.93647552e-01 6.32314533e-02 -2.39223614e-01 4.38271672e-01
-1.19503736e-01 -3.52280021e-01 2.13841870e-01 3.21871042e-01
-3.41258794e-01 -9.18080688e-01 -9.39648092e-01 2.79671550e-01
-1.07676709e+00 3.32276672e-01 8.39552224e-01 1.23063970e+00
9.90862906e-01 -5.02575375e-02 7.15986192e-01 -8.39225113e-01
3.21805835e-01 -5.09252310e-01 -3.12781096e-01 -3.71830916e-04
-3.64455462e-01 2.35205457e-01 8.29440653e-01 -6.01548195e-01
-8.67211163e-01 9.32487622e-02 -1.82758093e-01 -6.80111706e-01
2.55339622e-01 7.91761518e-01 6.93389252e-02 -2.50560552e-01
7.54595280e-01 6.60315394e-01 2.58204401e-01 -5.74579418e-01
2.77276516e-01 2.84110278e-01 7.46561289e-01 -6.21492028e-01
5.78636885e-01 6.55122459e-01 2.15401813e-01 -8.63059044e-01
-7.79992640e-01 -4.22310919e-01 -5.66358268e-01 -2.11103976e-01
7.46425986e-01 -9.92103398e-01 -4.61097181e-01 3.61001879e-01
-7.27815449e-01 -2.42793992e-01 -2.99088538e-01 1.01728737e+00
-5.46589017e-01 7.59744585e-01 -9.89531457e-01 -3.05599153e-01
-5.35964787e-01 -1.38294172e+00 5.19917369e-01 -2.95631260e-01
2.03169838e-01 -6.73716724e-01 -2.82600999e-01 4.21340495e-01
6.38342738e-01 4.72235680e-01 9.75543261e-01 -1.13576144e-01
-6.41638994e-01 -8.31498355e-02 -3.21700245e-01 4.99329239e-01
2.54428476e-01 -9.45801795e-01 -5.28150260e-01 -5.20277143e-01
6.34445488e-01 -1.92237765e-01 4.52395380e-01 7.27803230e-01
1.70830202e+00 -3.91761094e-01 1.45823300e-01 1.00629520e+00
1.52914953e+00 2.63782479e-02 5.36285102e-01 1.74983367e-01
8.89311194e-01 3.92162472e-01 3.22951019e-01 5.29302359e-01
2.13591345e-02 7.69569993e-01 3.27395856e-01 -3.30246031e-01
-2.68129021e-01 3.06117106e-02 2.76615154e-02 1.53202832e+00
-3.08875591e-01 4.30901676e-01 -8.12956035e-01 5.89823723e-01
-1.63097489e+00 -6.58183455e-01 -3.02040011e-01 2.26038122e+00
1.06524885e+00 -1.10677078e-01 -1.25964358e-01 2.68504739e-01
6.11438334e-01 2.23397300e-01 -8.53381693e-01 1.55834824e-01
-2.76269633e-02 1.75291136e-01 4.19100195e-01 3.09952050e-01
-7.99542248e-01 3.50814372e-01 5.44315290e+00 8.42569709e-01
-1.55262470e+00 4.71829057e-01 7.53795743e-01 -1.60510138e-01
-1.46577507e-01 -3.50049973e-01 -8.30314755e-02 3.72304142e-01
7.45247126e-01 1.36470413e-02 6.02144837e-01 4.61813003e-01
4.30139095e-01 1.60454988e-01 -6.83816373e-01 1.31360960e+00
-1.86657310e-02 -1.55911005e+00 9.50033292e-02 -1.09888916e-03
7.00543165e-01 5.31131960e-02 2.39556402e-01 -1.61336914e-01
-4.50902224e-01 -9.60329592e-01 4.68149096e-01 6.47976875e-01
1.24445498e+00 -6.81111097e-01 7.16546237e-01 3.41427356e-01
-8.12469304e-01 8.89426395e-02 -3.30067098e-01 1.97971150e-01
3.28162193e-01 9.36153769e-01 -4.80374843e-01 6.66473448e-01
7.08727598e-01 8.58920395e-01 -2.48052925e-02 8.53547990e-01
-3.52309309e-02 7.38023460e-01 -2.25037873e-01 5.23114443e-01
3.20860952e-01 -3.30569416e-01 5.29364347e-01 1.03454626e+00
2.20256746e-01 7.28322387e-01 1.66924208e-01 7.37736166e-01
8.57079700e-02 1.58621281e-01 -3.46054912e-01 1.47521600e-01
-1.50119215e-01 1.25359356e+00 -6.11188769e-01 -3.04173738e-01
-3.01116854e-01 7.44450271e-01 1.35674894e-01 4.74961817e-01
-6.79656208e-01 -9.19553339e-02 2.55620033e-01 6.09158397e-01
2.77734578e-01 -5.33386409e-01 -3.70454431e-01 -1.18155527e+00
1.30874738e-01 -9.17663455e-01 1.78655982e-01 -7.26062179e-01
-1.24853122e+00 7.91214705e-01 -9.32404920e-02 -1.40226614e+00
-1.26834124e-01 -3.49528939e-01 -4.62017991e-02 8.85192037e-01
-1.41944504e+00 -7.15799928e-01 -3.09922278e-01 7.93625474e-01
2.69948453e-01 -5.15253581e-02 6.25565588e-01 7.44100928e-01
-3.34774375e-01 3.52904409e-01 3.33645046e-01 3.75134319e-01
3.60815227e-01 -8.49502742e-01 -2.68231630e-01 6.27926946e-01
-7.98164010e-02 8.32591772e-01 5.28889298e-01 -5.07289827e-01
-1.63006914e+00 -9.42065299e-01 2.70599455e-01 1.18748136e-01
7.26511717e-01 -6.50371462e-02 -1.12885070e+00 3.61623257e-01
-2.71755248e-01 7.49496579e-01 6.35286033e-01 -3.37692857e-01
-5.53815849e-02 -4.46834922e-01 -1.08915913e+00 2.84037113e-01
9.03530180e-01 -6.66864097e-01 -4.50344533e-01 4.35614735e-01
6.40701234e-01 -5.81396580e-01 -1.28708768e+00 6.45740330e-01
4.55840051e-01 -7.49894083e-01 1.32506931e+00 -3.80563349e-01
5.77791572e-01 -4.68833387e-01 -1.69587895e-01 -1.09178567e+00
-6.16694450e-01 -2.67480850e-01 -2.21368358e-01 4.72881824e-01
3.39937210e-03 -5.39141119e-01 6.41087413e-01 3.71904045e-01
-3.79833102e-01 -1.14066160e+00 -1.06493914e+00 -6.01270974e-01
-9.94326621e-02 -3.73447657e-01 1.24851517e-01 1.27990258e+00
-2.66081780e-01 1.03843138e-01 -6.62103951e-01 3.37917626e-01
9.06023204e-01 6.67977333e-02 1.11512966e-01 -6.92620218e-01
-4.92647976e-01 -1.60423070e-02 -5.38000584e-01 -1.24975228e+00
-6.88012168e-02 -1.08452761e+00 1.24106249e-02 -1.27796984e+00
1.59640417e-01 -8.36689591e-01 -7.34268129e-01 2.25957602e-01
3.53680961e-02 4.46484864e-01 2.37140264e-02 4.86487150e-01
-2.70820558e-01 7.01018393e-01 1.69543231e+00 -1.55424371e-01
-8.64560343e-03 -2.40105107e-01 -4.36112106e-01 5.77217698e-01
6.95199728e-01 -4.77580488e-01 -4.82457697e-01 -6.93229496e-01
1.47318728e-02 6.12976670e-01 2.42722198e-01 -1.07447863e+00
2.75178462e-01 8.38965327e-02 5.23412406e-01 -4.35015827e-01
4.60967004e-01 -8.66330326e-01 3.75489801e-01 5.22603750e-01
-4.11811084e-01 -2.17371117e-02 -2.00835485e-02 5.17788172e-01
-4.06540692e-01 -7.04122484e-02 9.14751828e-01 -2.83660114e-01
-4.59064305e-01 6.02500081e-01 -1.41251892e-01 5.18215634e-02
5.57669759e-01 -1.67861015e-01 3.17855120e-01 -3.06786180e-01
-1.05968237e+00 -2.72632778e-01 8.60031024e-02 -3.98720428e-02
1.01174796e+00 -1.37861371e+00 -7.67082512e-01 3.56586635e-01
-1.82186753e-01 8.45392868e-02 8.24907303e-01 1.42771626e+00
-5.63928962e-01 3.12766224e-01 -4.37053353e-01 -8.94455731e-01
-6.30978763e-01 5.19175887e-01 3.54815185e-01 -4.93454903e-01
-1.27902544e+00 5.06556451e-01 3.95964652e-01 -2.37299025e-01
3.74498102e-03 -3.60768288e-01 -1.61361322e-01 -4.06312287e-01
6.35469317e-01 1.12612061e-01 1.58751220e-01 -9.04265523e-01
-3.16245049e-01 6.08431280e-01 -1.06762700e-01 -7.71061629e-02
1.76458347e+00 1.05000697e-01 -5.81605658e-02 4.05006140e-01
1.50748515e+00 -1.38682768e-01 -1.28111148e+00 -5.96189201e-01
-2.24249944e-01 -5.66678166e-01 5.74059010e-01 -6.44280076e-01
-1.57381988e+00 9.37996447e-01 7.78252304e-01 -2.89184272e-01
1.26826131e+00 -2.99396336e-01 1.01204014e+00 2.45349541e-01
5.53301990e-01 -7.92930245e-01 2.71191895e-01 1.70207664e-01
1.16471791e+00 -1.08516526e+00 3.48210633e-01 -3.76012594e-01
-5.71474612e-01 1.08831775e+00 1.64613888e-01 -3.67048830e-01
6.59647226e-01 8.76474082e-02 -5.48294075e-02 -3.59577805e-01
-1.70063242e-01 3.33699346e-01 3.62405241e-01 4.10346597e-01
4.91702318e-01 1.47494495e-01 -5.18254519e-01 5.33448398e-01
2.38255218e-01 1.54940680e-01 3.16337496e-01 6.29203439e-01
-1.52366394e-02 -7.77348936e-01 -4.18650627e-01 6.00914598e-01
-5.16590536e-01 -7.16498122e-02 5.36566019e-01 5.34787178e-01
-1.06244311e-01 4.40858275e-01 -1.97365090e-01 -1.37199178e-01
2.36031070e-01 -4.87499595e-01 5.47413886e-01 -3.49293768e-01
-3.75188768e-01 2.14707792e-01 -1.87251404e-01 -7.00282395e-01
-5.05264878e-01 -6.06994212e-01 -1.59938288e+00 1.19615890e-01
-8.71521160e-02 1.34251580e-01 6.59538984e-01 6.74165070e-01
3.33591580e-01 5.81660330e-01 8.10563743e-01 -8.10502589e-01
-5.56784630e-01 -7.15746701e-01 -6.84194028e-01 6.76297307e-01
4.34922904e-01 -6.61940098e-01 -2.18608260e-01 -5.56879193e-02] | [13.48510456085205, -2.4202301502227783] |
7f797c9f-81ca-4694-b633-89bb28c3fbb3 | describing-language-variation-in-the | null | null | https://aclanthology.org/2022.digitam-1.4 | https://aclanthology.org/2022.digitam-1.4.pdf | Describing Language Variation in the Colophons of Armenian Manuscripts | The colophons of Armenian manuscripts constitute a large textual corpus spanning a millennium of written culture. These texts are highly diverse and rich in terms of linguistic variation. This poses a challenge to NLP tools, especially considering the fact that linguistic resources designed or suited for Armenian are still scarce. In this paper, we deal with a sub-corpus of colophons written to commemorate the rescue of a manuscript and dating from 1286 to ca. 1450, a thematic group distinguished by a particularly high concentration of words exhibiting linguistic variation. The text is processed (lemmatization, POS-tagging, and inflectional tagging) using the tools of the GREgORI Project and evaluated. Through a selection of examples, we show how variation is dealt with at each linguistic level (phonology, orthography, flexion, vocabulary, syntax). Complex variation, at the level of tokens or lemmata, is considered as well. The results of this work are used to enrich and refine the linguistic resources of the GREgORI project, which in turn benefits the processing of other texts. | ['Emmanuel Van Elverdinghe', 'Bastien Kindt'] | null | null | null | null | digitam-lrec-2022-6 | ['lemmatization', 'culture'] | ['natural-language-processing', 'speech'] | [-1.51880115e-01 -3.21473092e-01 6.40408322e-02 -1.25283882e-01
-4.10374582e-01 -1.09125757e+00 1.02313137e+00 5.79107881e-01
-6.45969331e-01 1.01224458e+00 5.63098073e-01 -3.49225044e-01
-2.58130550e-01 -7.50873327e-01 -1.76405966e-01 -3.87644202e-01
1.76794812e-01 6.82051837e-01 -2.81655993e-02 -6.17515266e-01
6.81276381e-01 7.36208439e-01 -1.48984766e+00 4.19425577e-01
7.29633987e-01 2.49679983e-01 3.77383858e-01 3.83199960e-01
-7.30678201e-01 1.19575724e-01 -1.04702401e+00 -6.74732149e-01
-1.55850366e-01 -3.67785186e-01 -8.64800274e-01 1.01044625e-01
4.44734901e-01 2.86586910e-01 1.30506560e-01 1.03363061e+00
2.34858185e-01 -2.08577793e-02 7.68991649e-01 -3.65536302e-01
-3.81902575e-01 1.24964499e+00 -3.02863605e-02 2.94295788e-01
3.72922897e-01 -4.09806430e-01 1.07642400e+00 -1.11077821e+00
1.17072904e+00 1.33386338e+00 4.60518628e-01 2.28347883e-01
-8.72587800e-01 -2.29679227e-01 -1.11899212e-01 -7.26375654e-02
-1.28479564e+00 -4.61863369e-01 7.15437651e-01 -7.65040874e-01
8.92955601e-01 4.71429862e-02 7.35898852e-01 8.99553359e-01
1.62425861e-01 3.82277548e-01 9.67219591e-01 -9.02156532e-01
1.34320498e-01 3.60088557e-01 2.03947812e-01 4.44977820e-01
4.51557934e-01 -5.27241528e-01 -5.54512382e-01 3.48944999e-02
3.30185562e-01 -4.57323968e-01 -9.35769454e-02 1.70446500e-01
-1.26075256e+00 7.40233004e-01 -5.17918944e-01 1.19935596e+00
-1.97897732e-01 -3.08012068e-01 8.86624992e-01 3.66268247e-01
3.48961115e-01 5.43590367e-01 -4.98500526e-01 -5.21416783e-01
-1.19311225e+00 2.24512205e-01 9.56046343e-01 9.11156595e-01
4.90803272e-01 2.65741795e-02 2.79349297e-01 1.14891088e+00
2.86557615e-01 6.52970433e-01 5.05399644e-01 -6.28061354e-01
7.10908234e-01 6.40984774e-01 -1.34240896e-01 -8.52092922e-01
-3.71669620e-01 -5.53539768e-02 -4.04071957e-01 -1.80855855e-01
6.44862771e-01 -1.15714088e-01 -7.29223669e-01 1.46945226e+00
2.02658087e-01 -1.05770171e+00 2.54991919e-01 4.18000430e-01
9.48508799e-01 8.01109850e-01 2.57189751e-01 -5.12957394e-01
1.63488138e+00 -4.63208050e-01 -8.59952092e-01 -2.21299548e-02
4.61343080e-01 -1.28519046e+00 1.11791074e+00 5.01952529e-01
-1.48891807e+00 -3.55294734e-01 -8.76396954e-01 -1.27281457e-01
-8.55226576e-01 1.64259911e-01 3.79277050e-01 9.35421765e-01
-7.49285281e-01 5.53479135e-01 -5.39455771e-01 -6.49042249e-01
1.21943608e-01 -3.95098366e-02 -2.94254035e-01 1.89908043e-01
-1.26782811e+00 9.61181164e-01 6.36364996e-01 4.18310519e-03
-1.02010943e-01 -6.23267770e-01 -9.08193290e-01 -2.92873502e-01
1.89282849e-01 -1.73681732e-02 8.26167762e-01 -8.85594368e-01
-1.39149022e+00 1.32641113e+00 2.36941054e-01 -8.99283402e-03
4.93531466e-01 3.24489661e-02 -5.96663952e-01 3.92784148e-01
1.16870075e-01 6.17666021e-02 3.76093835e-01 -8.59363735e-01
-6.56579494e-01 -4.23634619e-01 -3.04362535e-01 -1.60059050e-01
-3.24211776e-01 6.95822835e-01 -3.21580201e-01 -1.04887640e+00
6.08573966e-02 -6.48354173e-01 1.55278325e-01 -6.17059708e-01
-8.93031340e-03 -3.53766322e-01 3.90053958e-01 -1.21529293e+00
1.48201895e+00 -1.96975744e+00 2.31817335e-01 4.18144137e-01
-1.32257253e-01 2.37456575e-01 2.93493629e-01 9.76242006e-01
2.36660093e-01 4.64470863e-01 -6.21020377e-01 -1.81082442e-01
3.19947302e-01 5.55031300e-01 -4.70047668e-02 6.10180199e-01
9.87320021e-02 5.97417772e-01 -8.94200742e-01 -6.52733088e-01
6.01227880e-02 5.16027808e-01 -1.06508084e-01 -4.75968421e-01
-2.43487522e-01 1.32101238e-01 -3.23344976e-01 7.44793594e-01
3.95129085e-01 5.51402271e-01 4.12331402e-01 2.23559216e-01
-9.14435148e-01 6.03552878e-01 -9.51093197e-01 1.69331968e+00
-6.51747882e-01 6.54303730e-01 9.31822062e-02 -6.58807039e-01
1.22324431e+00 1.90052286e-01 1.41544431e-01 -4.38228428e-01
4.85842675e-01 5.98438263e-01 1.78094089e-01 -6.50929272e-01
1.18217540e+00 -3.66078496e-01 -6.71086550e-01 2.87389994e-01
6.15411699e-02 -4.08844620e-01 9.91256177e-01 1.31482720e-01
7.58093953e-01 2.18298450e-01 7.50075638e-01 -7.96305001e-01
9.65156436e-01 2.18734518e-01 5.87002277e-01 1.66250318e-01
-1.65850800e-02 5.16206861e-01 7.28627026e-01 -1.64296716e-01
-1.41912580e+00 -8.96256685e-01 -6.40344739e-01 8.33040476e-01
-5.00137269e-01 -5.93973398e-01 -6.50384307e-01 -3.43231618e-01
-1.13818869e-01 6.99405551e-01 -5.00792205e-01 3.39122385e-01
-8.84054899e-01 -5.61407506e-01 8.04569483e-01 -2.38279719e-02
2.43025199e-01 -1.65046287e+00 -5.42201817e-01 3.98568124e-01
2.94100121e-02 -1.11792028e+00 4.30140980e-02 -6.00970536e-02
-5.45697689e-01 -7.34283984e-01 -7.06390262e-01 -1.04494929e+00
2.58517325e-01 -5.76192677e-01 1.22403109e+00 -2.60113422e-02
-2.80094296e-01 3.32894087e-01 -6.68960512e-01 -4.60197508e-01
-1.03536868e+00 3.08045357e-01 -1.71062455e-01 -4.53513235e-01
3.31681728e-01 -2.41434932e-01 2.98519671e-01 -2.82247096e-01
-1.06014848e+00 -6.47173941e-01 2.94952363e-01 3.83575827e-01
3.13178480e-01 -2.49391079e-01 5.20457506e-01 -1.24621046e+00
4.45953071e-01 -5.28314769e-01 -4.43062276e-01 2.53150553e-01
-1.30575791e-01 -2.04170242e-01 8.12470376e-01 -2.37066224e-01
-1.29323542e+00 -3.89091939e-01 -3.06902289e-01 5.61976910e-01
-2.42200673e-01 8.16856205e-01 -5.29592216e-01 2.31697351e-01
5.06977499e-01 7.40503892e-02 -6.94350675e-02 -6.81771576e-01
2.88867682e-01 9.37817991e-01 6.96958005e-01 -8.67174029e-01
6.74964726e-01 2.65269130e-01 5.36062121e-02 -1.52864254e+00
-3.08222055e-01 -4.16813612e-01 -9.78929758e-01 -3.25880110e-01
7.09908187e-01 -5.85450113e-01 -2.01670289e-01 3.83130252e-01
-1.08886588e+00 -2.05390975e-01 -5.12799740e-01 4.05162096e-01
-2.88787901e-01 6.50372267e-01 -6.95453823e-01 -5.41933656e-01
-2.79426545e-01 -6.58840656e-01 7.06472158e-01 1.06405623e-01
-5.70199251e-01 -1.37955976e+00 3.50704044e-01 1.22460313e-01
1.95315424e-02 4.66238946e-01 1.41701519e+00 -5.87683797e-01
2.35662371e-01 9.59789529e-02 2.18120262e-01 2.15576172e-01
5.14746346e-02 4.85696495e-01 -4.73181248e-01 -6.78359866e-02
-1.01141669e-01 -7.91554898e-03 5.98453701e-01 -8.84508789e-02
2.66270667e-01 3.36602591e-02 2.15575457e-01 1.11807585e-01
1.53668189e+00 1.36516452e-01 7.78583109e-01 7.92986274e-01
3.45707595e-01 8.85721266e-01 5.56973994e-01 6.95969939e-01
1.67051643e-01 2.73703784e-01 -1.66467577e-01 6.22975469e-01
-3.97206187e-01 2.57995218e-01 6.23317957e-01 1.39193654e+00
-1.20821156e-01 -2.12599128e-01 -1.40965760e+00 1.13577080e+00
-1.29646587e+00 -8.38846684e-01 -5.26762366e-01 2.13802862e+00
9.50378120e-01 5.09233177e-02 1.62715137e-01 3.65508765e-01
8.05248439e-01 3.06547642e-01 4.37378198e-01 -1.05520928e+00
-5.89075863e-01 5.88812947e-01 1.85733199e-01 6.68771029e-01
-6.54145956e-01 1.37372315e+00 6.15007544e+00 7.62583613e-01
-8.37206423e-01 -1.95787102e-01 -1.93328992e-01 -4.19658609e-03
-6.08187020e-01 -1.35758352e-02 -8.87850106e-01 6.50251269e-01
1.12253928e+00 -2.86411401e-02 1.41784251e-01 3.14737767e-01
2.45787725e-01 -2.69878954e-01 -6.15485549e-01 6.73099518e-01
5.39269090e-01 -1.20205784e+00 1.34473145e-01 -8.90108943e-02
6.18088961e-01 -1.64594159e-01 -2.02700287e-01 2.71347433e-01
-1.75069198e-01 -6.52654827e-01 1.21524501e+00 3.52401316e-01
6.09421849e-01 -9.80287433e-01 6.54110610e-01 -1.37862802e-01
-8.40621293e-01 3.19933385e-01 -6.53357923e-01 -7.31293261e-02
3.74401480e-01 7.39633143e-01 -5.32680511e-01 6.42242849e-01
3.37613165e-01 6.58100486e-01 -7.09351242e-01 7.43907690e-01
-5.59545040e-01 8.37685347e-01 -2.86927730e-01 -5.03536999e-01
5.45386910e-01 -7.04049289e-01 9.95819688e-01 1.85967243e+00
2.99677610e-01 -1.20311789e-01 -3.04246433e-02 5.62218010e-01
-2.64413259e-03 8.81162107e-01 -2.91206568e-01 -4.21674937e-01
5.93261898e-01 1.21251786e+00 -1.09117305e+00 -3.44044030e-01
-2.57915378e-01 6.68194115e-01 2.46410176e-01 9.43785831e-02
-3.74361813e-01 -7.69035280e-01 4.32247132e-01 1.48131818e-01
4.89507109e-01 -6.37435138e-01 -4.35148925e-01 -9.20403659e-01
2.36737788e-01 -8.41646194e-01 4.05122042e-01 -2.25365579e-01
-1.25242758e+00 5.83284795e-01 -4.79386970e-02 -6.77555799e-01
-1.55250564e-01 -6.86085701e-01 -2.79465348e-01 8.55672777e-01
-1.04776466e+00 -1.18990171e+00 2.04320312e-01 2.78855920e-01
4.27409917e-01 -5.92155576e-01 8.03059697e-01 4.25485462e-01
-4.59440142e-01 2.79957950e-01 2.99271584e-01 2.14167431e-01
7.08509207e-01 -1.29514551e+00 2.80094087e-01 7.28903949e-01
6.78277090e-02 6.06130779e-01 6.42853796e-01 -8.33892822e-01
-1.10851872e+00 -8.02337348e-01 1.84841442e+00 -2.51460016e-01
1.28848767e+00 -4.26011592e-01 -7.89726019e-01 4.08471614e-01
3.65535527e-01 -6.43705726e-01 8.29043329e-01 1.54643819e-01
-1.53085202e-01 2.94418812e-01 -1.10567617e+00 7.66057730e-01
8.41990113e-01 -5.52018762e-01 -1.18203664e+00 2.47180521e-01
2.13392198e-01 -1.84818581e-02 -1.22387350e+00 -1.98987231e-01
5.56764126e-01 -6.94588780e-01 4.08723593e-01 -4.52086687e-01
5.44721961e-01 -1.39575541e-01 -1.42789796e-01 -1.06160426e+00
-7.87465572e-02 -1.03231430e+00 6.01700187e-01 1.97552824e+00
6.70517087e-01 -3.42581838e-01 5.37437081e-01 1.69209182e-01
-4.67729300e-01 1.21475728e-02 -8.52926075e-01 -7.21823812e-01
6.37620449e-01 -4.13366765e-01 3.79843950e-01 1.02274334e+00
3.97884935e-01 3.82641673e-01 1.71453834e-01 -5.00917196e-01
4.34025705e-01 -4.88035530e-02 3.27157259e-01 -1.36202276e+00
1.23580746e-01 -7.29151964e-01 -5.53651273e-01 -1.66049108e-01
3.50685716e-01 -1.18624377e+00 4.44220230e-02 -1.43513000e+00
-3.12455863e-01 -4.07946944e-01 3.09733033e-01 1.58307344e-01
2.03381523e-01 3.08554024e-01 4.11498368e-01 1.72152042e-01
-1.31101221e-01 2.11739615e-01 7.94251442e-01 2.04581276e-01
-4.77140576e-01 -4.91178632e-01 -3.42645854e-01 8.73136759e-01
9.72126901e-01 -5.22614717e-01 2.89650798e-01 -3.08541685e-01
7.86677897e-01 -3.81476760e-01 -2.00913399e-01 -7.78193295e-01
-8.63393098e-02 1.50144575e-02 1.74011201e-01 -7.96144903e-01
6.31221607e-02 -6.22160375e-01 1.57521795e-02 4.19312656e-01
5.39867207e-02 2.02772990e-01 3.63654584e-01 -1.74809784e-01
-2.40762547e-01 -8.46215606e-01 8.25096905e-01 -2.97500163e-01
-7.30241179e-01 -5.54791167e-02 -1.14181161e+00 5.29786646e-01
7.81473160e-01 -2.41536438e-01 -1.66282788e-01 2.75738358e-01
-6.98672473e-01 -2.00766280e-01 7.87983179e-01 9.84501094e-02
1.80749491e-01 -1.04414380e+00 -9.79694903e-01 2.85422262e-02
5.35829812e-02 -2.84191161e-01 2.08792705e-02 4.59252834e-01
-1.19860280e+00 2.31908381e-01 -5.50412834e-01 -7.87922144e-02
-1.16834533e+00 3.85625690e-01 -2.36519918e-01 -2.76056379e-01
-6.38268590e-01 4.03447092e-01 -5.49898446e-01 -4.64137465e-01
-1.78338997e-02 -1.67431951e-01 -6.95894599e-01 8.49029839e-01
4.15897697e-01 5.68315625e-01 1.78759649e-01 -1.07523787e+00
-4.43422407e-01 7.49684453e-01 1.57570824e-01 -5.57741165e-01
1.46226633e+00 -3.02215368e-01 -7.36403048e-01 8.44701052e-01
1.02548766e+00 8.71567369e-01 -3.68244290e-01 9.25238878e-02
4.57237691e-01 -2.16293931e-01 -2.58583695e-01 -7.32707620e-01
-5.25866508e-01 7.82693446e-01 -3.49083006e-01 2.51814663e-01
5.31664371e-01 -1.78286200e-03 5.86263597e-01 3.18385512e-01
6.02640063e-02 -1.59121776e+00 -5.52389085e-01 1.16825998e+00
9.18853223e-01 -5.61247706e-01 1.88884765e-01 -3.46054822e-01
-6.09945297e-01 1.44751191e+00 -1.76184237e-01 -2.04486728e-01
5.94170511e-01 3.98392439e-01 5.69695607e-02 -2.53360927e-01
-1.48305357e-01 -2.40431949e-01 1.22593917e-01 4.34958994e-01
7.07718194e-01 2.16725077e-02 -1.39888763e+00 6.39876962e-01
-8.67597759e-01 -5.29961705e-01 9.43191350e-01 9.99888897e-01
-7.09331572e-01 -1.41462278e+00 -6.81272447e-01 -1.92585979e-02
-9.17956591e-01 -3.50251317e-01 -8.01225960e-01 1.37059402e+00
3.54346573e-01 7.50025630e-01 2.95776099e-01 2.17180565e-01
1.29052818e-01 2.79331326e-01 8.38741481e-01 -8.13187242e-01
-1.28309226e+00 3.04615855e-01 6.07275426e-01 2.02399790e-01
-5.26882470e-01 -1.21973586e+00 -1.31371629e+00 -5.29868722e-01
2.26619706e-01 5.57198644e-01 9.91398692e-01 1.05693722e+00
-5.24785519e-01 3.96484911e-01 9.36491638e-02 -6.42693758e-01
-1.39555752e-01 -1.03444207e+00 -1.07274675e+00 1.32019430e-01
-2.89541125e-01 -3.03372175e-01 -4.43856865e-01 2.70153373e-01] | [10.319713592529297, 10.268806457519531] |
4f3e50bc-1de9-467d-9826-0ffd9d80144f | improving-model-understanding-and-trust-with | 2206.02790 | null | https://arxiv.org/abs/2206.02790v1 | https://arxiv.org/pdf/2206.02790v1.pdf | Improving Model Understanding and Trust with Counterfactual Explanations of Model Confidence | In this paper, we show that counterfactual explanations of confidence scores help users better understand and better trust an AI model's prediction in human-subject studies. Showing confidence scores in human-agent interaction systems can help build trust between humans and AI systems. However, most existing research only used the confidence score as a form of communication, and we still lack ways to explain why the algorithm is confident. This paper also presents two methods for understanding model confidence using counterfactual explanation: (1) based on counterfactual examples; and (2) based on visualisation of the counterfactual space. | ['Liz Sonenberg', 'Ronal Singh', 'Tim Miller', 'Thao Le'] | 2022-06-06 | null | null | null | null | ['counterfactual-explanation'] | ['miscellaneous'] | [-2.87079275e-01 1.01243770e+00 -2.54413158e-01 -5.45683861e-01
1.11533865e-01 -4.24142361e-01 1.00992656e+00 5.20984173e-01
-1.88624650e-01 1.13676596e+00 3.09704304e-01 -1.10044503e+00
-4.04351167e-02 -6.94373250e-01 -6.19021475e-01 2.10003108e-02
-2.94441968e-01 4.30646926e-01 -9.72498134e-02 -1.39582813e-01
8.09635043e-01 8.16843137e-02 -1.44267976e+00 4.18731630e-01
1.24383664e+00 5.40020108e-01 -2.49673843e-01 6.83659613e-01
1.52381405e-01 1.11779952e+00 -7.70788014e-01 -7.86872327e-01
1.94666892e-01 -6.87405407e-01 -8.58121276e-01 -4.34172422e-01
-1.70425326e-01 -4.24403369e-01 4.47788388e-01 9.87043858e-01
-1.09800756e-01 -1.83749706e-01 6.48520708e-01 -2.10264182e+00
-6.19100630e-01 8.62142563e-01 -6.22208379e-02 -2.47076049e-01
9.53149438e-01 3.98532271e-01 7.49974549e-01 -3.26716602e-01
4.43395913e-01 1.58830917e+00 6.84596181e-01 6.15726769e-01
-1.27704155e+00 -8.90086830e-01 2.17799380e-01 4.90696520e-01
-9.01959717e-01 -9.63097885e-02 7.18002558e-01 -6.22547746e-01
1.06308246e+00 6.42788887e-01 1.38640964e+00 6.25688195e-01
3.86292487e-01 5.11460483e-01 1.76829720e+00 -7.15217888e-01
6.32859468e-01 8.30156922e-01 2.06559792e-01 5.56193054e-01
9.00916874e-01 9.21491563e-01 -3.81205022e-01 -7.02483237e-01
6.69543147e-01 -2.17341751e-01 -3.74896288e-01 -6.69754088e-01
-1.17924678e+00 1.25098610e+00 5.06864607e-01 4.04087007e-01
-6.03149056e-01 1.44316629e-01 1.28498167e-01 5.48671961e-01
5.50234318e-01 1.22421098e+00 -7.51023233e-01 -1.47388339e-01
-6.73672199e-01 4.94978428e-01 1.35218024e+00 3.44152898e-01
3.83953273e-01 -1.20844342e-01 2.74260044e-01 1.50312632e-01
1.06581473e+00 2.01335996e-01 5.80942810e-01 -1.03253973e+00
-1.55435815e-01 6.63623393e-01 7.75515258e-01 -8.56052160e-01
-2.48804197e-01 -1.27375826e-01 -1.15401760e-01 1.10991514e+00
3.41651589e-01 -2.85780489e-01 -5.08804440e-01 1.47509110e+00
1.38465166e-01 2.62808651e-01 5.18512189e-01 8.01945746e-01
4.41590399e-01 3.91403407e-01 3.56980920e-01 -6.11713827e-01
9.67130244e-01 -7.73537517e-01 -9.17575955e-01 -3.71673524e-01
1.07908010e+00 -4.35330212e-01 1.02722359e+00 4.76794988e-01
-1.00783682e+00 -2.48493165e-01 -1.62651277e+00 5.46318114e-01
-4.46657807e-01 -6.39530838e-01 1.26674938e+00 1.06919396e+00
-9.99537051e-01 8.50200593e-01 -5.48707366e-01 -2.15588763e-01
1.75672352e-01 7.46216625e-02 -1.76379420e-02 1.96158215e-01
-1.41432714e+00 1.59706712e+00 5.17388761e-01 -1.71015114e-01
-4.22638237e-01 -6.44820929e-01 -1.08904517e+00 -4.95629990e-03
9.16073024e-02 -9.77789104e-01 1.56218994e+00 -1.24262965e+00
-1.57102084e+00 5.05694747e-01 8.60776454e-02 -1.01157951e+00
8.12928259e-01 -1.98347926e-01 -3.42371672e-01 -3.84967208e-01
4.60765548e-02 4.78459567e-01 2.80526519e-01 -1.65633869e+00
-5.79739332e-01 -3.91071796e-01 4.62647080e-01 1.86812416e-01
5.24988592e-01 3.33165266e-02 8.43241751e-01 -2.05281898e-01
-1.25792325e-01 -8.07413161e-01 -4.53285038e-01 1.57483980e-01
-4.05382998e-02 -3.85198146e-01 6.10317588e-01 -2.74335057e-01
1.30908000e+00 -1.46187818e+00 -8.72328997e-01 3.20384830e-01
1.85415402e-01 2.64943033e-01 4.14518148e-01 3.49974483e-01
-4.25788313e-01 6.70027852e-01 6.52946979e-02 4.72223639e-01
3.64128977e-01 -1.39803007e-01 -2.76115239e-01 -5.38184568e-02
-1.95909038e-01 6.61129415e-01 -1.07028544e+00 -4.35541958e-01
5.22876263e-01 4.68858443e-02 -5.42886376e-01 3.72155666e-01
-1.49310932e-01 -1.92220584e-01 -2.29123935e-01 1.48546128e-02
6.10651255e-01 2.91820080e-03 4.68454987e-01 -1.78178232e-02
-2.85387248e-01 6.56721771e-01 -1.10266757e+00 1.01440060e+00
-4.52955067e-01 6.04392469e-01 -5.02884090e-01 -3.90207916e-01
7.78175533e-01 6.87319756e-01 -2.41090938e-01 -2.71667600e-01
2.34380066e-02 2.15884775e-01 4.90472764e-01 -4.37325180e-01
1.02332346e-01 -7.20163643e-01 1.13386281e-01 1.11738551e+00
-5.62558174e-01 -8.66335034e-01 -6.53584525e-02 2.73089796e-01
4.16270733e-01 1.97238494e-02 1.24697876e+00 -3.27162355e-01
2.44982153e-01 2.82712698e-01 2.72798717e-01 8.76717925e-01
-3.58215928e-01 -2.98668612e-02 6.02639616e-01 -9.13040042e-01
-7.27341294e-01 -7.94275701e-01 -9.49355029e-03 3.60983819e-01
4.62252162e-02 -3.55682939e-01 -7.58469522e-01 -1.04458916e+00
1.31764725e-01 2.12130475e+00 -8.83442819e-01 -2.76156217e-01
1.35370180e-01 -3.16663355e-01 9.49907452e-02 5.26044011e-01
5.31716049e-01 -9.16410208e-01 -1.13291943e+00 -2.13664677e-02
-2.10881960e-02 -1.34872317e-01 3.68656293e-02 -2.98636347e-01
-1.07379031e+00 -1.31458831e+00 2.17483100e-02 9.34766755e-02
4.45501626e-01 3.74000669e-01 1.28317583e+00 8.83645892e-01
5.07003546e-01 2.43879288e-01 -2.23031670e-01 -1.43347204e+00
-8.73606503e-01 -1.09413421e+00 1.81909889e-01 -9.29451227e-01
6.43023133e-01 -3.54812384e-01 -6.13446891e-01 3.85971367e-01
-4.42798823e-01 4.31199431e-01 3.13479543e-01 7.99952924e-01
-2.96551406e-01 -8.62583742e-02 6.16011977e-01 -1.14287388e+00
1.39725065e+00 -4.21800554e-01 -4.50922847e-01 5.29496849e-01
-1.66498017e+00 3.20688970e-02 1.84765667e-01 -2.55460292e-01
-1.26666605e+00 -4.45859015e-01 3.27681661e-01 4.59082395e-01
-2.44628280e-01 6.16176069e-01 4.21647727e-03 2.70587713e-01
1.00316691e+00 -4.89335716e-01 1.78524509e-01 2.28900433e-01
4.98805016e-01 5.99095225e-01 2.24940795e-02 -3.14457864e-01
4.01904553e-01 7.98589736e-02 -5.44074118e-01 -3.87689024e-02
-6.18772328e-01 3.20700049e-01 -3.98187369e-01 -4.60912466e-01
4.17141169e-01 -4.03263658e-01 -1.07790673e+00 -1.79954529e-01
-1.26145971e+00 -3.87381226e-01 -3.96910936e-01 9.27979767e-01
-7.91814148e-01 6.86201230e-02 1.61927845e-02 -1.48282754e+00
-8.67809355e-02 -8.00373316e-01 1.87078729e-01 3.18921804e-01
-1.05210090e+00 -1.24674809e+00 2.61524975e-01 3.42962176e-01
3.15097511e-01 2.80653864e-01 8.54292929e-01 -9.43801880e-01
-1.37189910e-01 -4.34466660e-01 4.27762233e-02 -1.21695183e-01
-6.24553151e-02 1.90729931e-01 -9.97154057e-01 3.81159224e-02
4.44118559e-01 -2.10816696e-01 -8.53115413e-03 3.75126690e-01
5.42305052e-01 -1.03883243e+00 -5.18759966e-01 -5.85483193e-01
9.07734573e-01 6.72862351e-01 8.36269319e-01 6.00989223e-01
-2.42387116e-01 7.73235202e-01 1.18606102e+00 3.03581327e-01
5.75763583e-01 4.75894779e-01 3.49926263e-01 -2.13269085e-01
5.12646019e-01 -4.51220304e-01 1.61973298e-01 -1.38454381e-02
-2.98826873e-01 1.69300541e-01 -1.02313793e+00 2.89017260e-01
-2.17554307e+00 -1.09232903e+00 -1.29223093e-01 2.27415395e+00
7.13389874e-01 5.41589618e-01 1.32668078e-01 2.49086142e-01
5.88972867e-01 -1.65055320e-01 -4.04716194e-01 -9.74398553e-01
3.83695722e-01 -3.85641485e-01 -6.58771992e-02 8.86875689e-01
-4.53682989e-01 5.41833937e-01 7.58084345e+00 -4.93285321e-02
-4.68583703e-01 -1.54390037e-01 6.95658863e-01 7.20160753e-02
-7.21775353e-01 6.06365263e-01 3.26761425e-01 2.26249382e-01
1.10340452e+00 -1.08615446e+00 -1.25005513e-01 1.02924919e+00
4.54970002e-01 -5.08581281e-01 -1.57756972e+00 4.99839604e-01
-3.50080341e-01 -1.38373399e+00 7.41389394e-02 -8.24228525e-02
5.35650730e-01 -7.78396487e-01 -2.24144980e-01 2.76299715e-01
1.10921764e+00 -1.11316168e+00 8.74436378e-01 6.13462031e-01
1.53846011e-01 -7.50466466e-01 1.38883686e+00 5.42938352e-01
-3.79521638e-01 2.90165301e-02 -1.63021311e-01 -9.70054328e-01
5.75037934e-02 2.92856842e-01 -1.55597746e+00 3.70303959e-01
4.41510797e-01 2.56752074e-01 -3.71680290e-01 1.00628889e+00
-6.56337798e-01 5.06138623e-01 5.02418205e-02 -3.94572765e-01
-4.27037962e-02 2.00999118e-02 2.54417330e-01 9.49268937e-01
8.94525424e-02 4.82791811e-01 -2.93231934e-01 9.78130937e-01
6.56100750e-01 -1.37229756e-01 -1.10698569e+00 1.35026008e-01
7.32840836e-01 5.21792769e-01 -5.15748024e-01 -5.79298258e-01
-2.51416564e-01 5.35668492e-01 -9.24298018e-02 1.69405341e-01
-5.02030730e-01 3.54893953e-02 6.67262852e-01 4.20383438e-02
-4.44084615e-01 5.85241355e-02 -6.77319169e-01 -8.78572941e-01
-4.44839478e-01 -1.00114048e+00 4.76539046e-01 -1.44359267e+00
-9.99464035e-01 3.90772700e-01 3.84109467e-01 -9.80067313e-01
-9.68450904e-01 -3.33987921e-01 -9.78627503e-01 7.30248451e-01
-8.33601117e-01 -9.31529105e-01 -2.09445395e-02 -1.66425090e-02
1.77059457e-01 1.56682298e-01 1.20623386e+00 -7.44070053e-01
-7.45474771e-02 2.56642669e-01 -5.88465810e-01 -5.10725260e-01
5.25145650e-01 -1.42588103e+00 5.66018641e-01 5.73971987e-01
6.98845908e-02 1.09172058e+00 1.45145535e+00 -8.16335022e-01
-5.69423020e-01 -8.94131362e-02 9.20771182e-01 -7.05403507e-01
5.89758873e-01 1.87805340e-01 -9.49276686e-01 7.74739683e-01
6.58063054e-01 -6.39637232e-01 9.11862254e-01 4.93555099e-01
-2.93139786e-01 3.45332175e-01 -1.70416331e+00 9.90924299e-01
6.57861948e-01 -2.68797368e-01 -1.40764153e+00 1.92803055e-01
7.95443892e-01 2.62459535e-02 -5.16310930e-01 2.86462605e-01
1.04153562e+00 -1.49912715e+00 6.67050719e-01 -1.03692710e+00
2.65485346e-01 -3.85878772e-01 1.64690271e-01 -1.84052086e+00
-1.00851186e-01 -6.26834273e-01 -4.95152269e-03 6.71461761e-01
9.34733272e-01 -1.07430458e+00 4.91913080e-01 1.77552092e+00
2.71649688e-01 -5.97824216e-01 -5.86236417e-01 -3.05935174e-01
-8.08796287e-02 -7.48387218e-01 1.11655462e+00 1.35353220e+00
1.03194666e+00 1.80357739e-01 -9.19487029e-02 4.71300036e-02
4.23227072e-01 1.08538255e-01 8.12303483e-01 -1.45595694e+00
-3.25611643e-02 -4.47917789e-01 -4.08416867e-01 -4.69690621e-01
2.90010273e-02 -2.69142330e-01 -9.33283716e-02 -1.64829063e+00
3.49403858e-01 -1.31629661e-01 -7.72315403e-03 3.07210863e-01
-3.00683111e-01 -3.02113891e-01 3.15022379e-01 1.34936899e-01
-3.49296451e-01 2.51401067e-01 9.11998987e-01 1.70658275e-01
-1.12666167e-01 1.33977145e-01 -1.24866581e+00 1.29066825e+00
8.97081494e-01 -6.12435222e-01 -6.03066742e-01 1.86358586e-01
4.06481445e-01 1.81455910e-01 6.76576197e-01 -7.08800554e-01
1.83832824e-01 -7.16264486e-01 2.57319599e-01 -7.67397508e-02
9.53676477e-02 -1.04802620e+00 4.94128168e-01 1.19118559e+00
-4.90888715e-01 1.96800292e-01 3.93186986e-01 4.64477360e-01
-6.64296187e-03 -5.74434578e-01 4.15203273e-01 -2.22606882e-01
-4.63850945e-01 -7.20373392e-01 -6.82808280e-01 -6.44974232e-01
1.12995076e+00 -4.65629339e-01 -4.85546410e-01 -8.82198870e-01
-6.00302100e-01 4.22603071e-01 6.84922874e-01 1.94332555e-01
7.84576237e-01 -1.25803483e+00 -6.46009922e-01 2.24993993e-02
1.16671346e-01 -8.16902101e-01 -6.79289326e-02 3.25338602e-01
-4.11469668e-01 7.11011589e-01 -4.52109098e-01 -1.60825685e-01
-1.30871153e+00 6.23375118e-01 5.32716990e-01 -2.87644863e-01
-3.11504547e-02 6.96997702e-01 2.52277523e-01 -4.72906739e-01
-1.03969581e-01 -3.08691949e-01 -5.13425231e-01 -3.62000138e-01
8.40522587e-01 4.16282296e-01 -3.01567763e-01 -9.39414725e-02
-3.54923487e-01 -5.03310524e-02 -2.52486408e-01 -6.36418164e-01
1.12669933e+00 -2.07833320e-01 6.15382195e-02 7.64215052e-01
2.83844054e-01 -3.55466098e-01 -1.11711788e+00 3.03111553e-01
3.20149332e-01 -1.03292370e+00 -2.49435790e-02 -1.48836613e+00
-1.32391140e-01 5.15939653e-01 6.43851161e-01 8.53095293e-01
6.81586981e-01 -9.43312421e-02 -1.25555679e-01 4.44431514e-01
6.14912927e-01 -8.99272382e-01 -2.34707609e-01 -1.66918635e-01
1.58707511e+00 -1.50155115e+00 4.94349539e-01 -3.51033121e-01
-1.15195072e+00 9.67993855e-01 7.55961657e-01 2.79127747e-01
7.19599962e-01 3.78627400e-03 4.36972469e-01 -4.02932793e-01
-1.20160604e+00 3.54628921e-01 2.09564731e-01 8.52043808e-01
1.05140710e+00 5.65598845e-01 -8.66216898e-01 8.74897301e-01
-8.09276819e-01 3.52257550e-01 9.08744395e-01 8.49767208e-01
-3.93796623e-01 -9.27644491e-01 -5.04122913e-01 5.75243473e-01
-9.20044258e-02 1.00274742e-01 -9.34411049e-01 1.17361152e+00
-2.14628309e-01 1.53769565e+00 -1.75926894e-01 -4.60243911e-01
5.32382488e-01 1.10916436e-01 3.70383769e-01 -6.11688793e-01
-6.62951589e-01 -5.74411273e-01 7.09663510e-01 -6.91600263e-01
-6.27119601e-01 -6.53362751e-01 -1.10167837e+00 -6.53799832e-01
-7.72104263e-01 9.56049383e-01 7.24165976e-01 9.09877539e-01
5.68171553e-02 7.86767974e-02 3.48563850e-01 -5.10818243e-01
-5.43053985e-01 -1.27325463e+00 -3.30388993e-01 2.62333721e-01
8.87392387e-02 -7.97777474e-01 -6.35550201e-01 -2.82545954e-01] | [8.79911994934082, 5.925322532653809] |
b634392f-2a72-4691-a3e0-a221d26017fb | on-consistency-in-graph-neural-network | 2205.13733 | null | https://arxiv.org/abs/2205.13733v2 | https://arxiv.org/pdf/2205.13733v2.pdf | Towards Faithful and Consistent Explanations for Graph Neural Networks | Uncovering rationales behind predictions of graph neural networks (GNNs) has received increasing attention over recent years. Instance-level GNN explanation aims to discover critical input elements, like nodes or edges, that the target GNN relies upon for making predictions. Though various algorithms are proposed, most of them formalize this task by searching the minimal subgraph which can preserve original predictions. However, an inductive bias is deep-rooted in this framework: several subgraphs can result in the same or similar outputs as the original graphs. Consequently, they have the danger of providing spurious explanations and fail to provide consistent explanations. Applying them to explain weakly-performed GNNs would further amplify these issues. To address this problem, we theoretically examine the predictions of GNNs from the causality perspective. Two typical reasons of spurious explanations are identified: confounding effect of latent variables like distribution shift, and causal factors distinct from the original input. Observing that both confounding effects and diverse causal rationales are encoded in internal representations, we propose a simple yet effective countermeasure by aligning embeddings. Concretely, concerning potential shifts in the high-dimensional space, we design a distribution-aware alignment algorithm based on anchors. This new objective is easy to compute and can be incorporated into existing techniques with no or little effort. Theoretical analysis shows that it is in effect optimizing a more faithful explanation objective in design, which further justifies the proposed approach. | ['Suhang Wang', 'Xiang Zhang', 'Dongsheng Luo', 'Tianxiang Zhao'] | 2022-05-27 | null | null | null | null | ['network-interpretation'] | ['computer-vision'] | [ 4.60527062e-01 9.06515062e-01 -5.83355606e-01 -3.84837478e-01
1.52685329e-01 -5.47076464e-01 6.39669538e-01 3.55908513e-01
2.74872303e-01 6.35362625e-01 6.14704549e-01 -6.91152215e-01
-4.46712643e-01 -8.34596157e-01 -9.94636893e-01 -5.72795212e-01
-5.60113303e-02 -1.10426605e-01 2.80398726e-02 -1.16765864e-01
3.62919599e-01 4.27621484e-01 -1.53317881e+00 7.34368414e-02
7.91714311e-01 4.91774231e-01 -3.67809124e-02 2.55093127e-01
-2.84991920e-01 6.69637620e-01 -4.54871505e-01 -5.97476721e-01
1.25705883e-01 -6.74822807e-01 -5.11523247e-01 -1.83374718e-01
3.83583784e-01 -8.60857069e-02 -3.28691512e-01 1.28748739e+00
3.80876027e-02 -3.79648536e-01 6.99814737e-01 -1.62990093e+00
-1.20883751e+00 1.27847195e+00 -4.49339420e-01 1.21564239e-01
2.15708129e-02 3.23574245e-02 1.54317844e+00 -7.70707190e-01
3.84106517e-01 1.23086274e+00 6.75323904e-01 6.33326352e-01
-1.48432457e+00 -6.45625293e-01 5.36969900e-01 6.59411922e-02
-9.37237620e-01 -2.08248824e-01 1.04178989e+00 -2.98777521e-01
6.52593315e-01 6.26455784e-01 5.62683284e-01 1.56334102e+00
2.78165400e-01 6.49951339e-01 6.30323231e-01 -2.85802186e-01
3.75733972e-01 2.60444820e-01 3.02930355e-01 4.54627752e-01
1.02781451e+00 1.71821699e-01 -5.55830479e-01 -2.38674313e-01
6.61054432e-01 1.49412394e-01 -4.94448721e-01 -5.94390273e-01
-9.93144214e-01 1.07660508e+00 6.97651803e-01 3.63477707e-01
-3.44158828e-01 5.19086063e-01 1.00078218e-01 1.61358356e-01
2.98242748e-01 6.03721499e-01 -6.70980275e-01 3.59874278e-01
-5.27688086e-01 1.96097463e-01 5.78176320e-01 8.66996229e-01
8.22280586e-01 2.14946523e-01 3.38495523e-02 1.49940237e-01
6.11370802e-01 2.94887841e-01 3.94256920e-01 -3.52247745e-01
4.18528438e-01 1.03542554e+00 -2.58375466e-01 -1.71674788e+00
-4.71675426e-01 -7.63135731e-01 -1.02006364e+00 -3.12750265e-02
3.68907750e-01 2.26772930e-02 -7.90375829e-01 2.17248058e+00
1.52075859e-02 3.60840887e-01 -1.39727844e-02 9.46334541e-01
6.00760520e-01 2.95761228e-01 5.51212691e-02 -1.46139726e-01
1.19043756e+00 -5.91303766e-01 -7.76096642e-01 -4.90723580e-01
5.97654819e-01 -3.52989346e-01 1.15173030e+00 -9.67489108e-02
-6.46045268e-01 -3.24891746e-01 -1.26038706e+00 2.31620118e-01
-2.33488530e-01 -1.51573107e-01 8.41320395e-01 8.36908042e-01
-9.79180455e-01 8.07845891e-01 -4.92437571e-01 -4.29561168e-01
1.95934236e-01 4.94195879e-01 -2.02220425e-01 1.89917222e-01
-1.33580470e+00 6.75319910e-01 2.70687938e-01 1.75698653e-01
-5.94520092e-01 -8.27524483e-01 -6.37290418e-01 4.44550693e-01
4.44740146e-01 -7.43718207e-01 6.95053816e-01 -1.06612372e+00
-9.39330101e-01 3.25849354e-01 -1.44090950e-01 -4.79955137e-01
2.84397066e-01 1.59490611e-02 -4.70996857e-01 -2.96003401e-01
1.40266329e-01 2.50285476e-01 9.44895506e-01 -1.39425874e+00
-2.31325284e-01 -4.62680012e-01 6.31871596e-02 -3.34142625e-01
-7.43315279e-01 -4.94914144e-01 -7.76132196e-02 -6.74174607e-01
5.90275347e-01 -7.54043341e-01 -3.99517000e-01 -1.84523165e-01
-9.78223383e-01 -9.23820436e-02 4.89864379e-01 -2.51311392e-01
1.45443439e+00 -1.93562794e+00 1.29936352e-01 4.72207755e-01
7.98703253e-01 -6.14416972e-02 -1.44215271e-01 4.75657046e-01
-4.98021960e-01 6.80245340e-01 -1.23851679e-01 5.50008342e-02
1.50096402e-01 2.99557000e-01 -8.76707137e-01 4.50467199e-01
3.79311383e-01 9.93578136e-01 -8.14776659e-01 7.03638270e-02
-5.43447286e-02 2.57331163e-01 -7.90715575e-01 4.95774858e-02
-2.58265406e-01 1.18252270e-01 -4.75550234e-01 2.62249261e-01
6.27609968e-01 -5.80562413e-01 6.23543561e-01 -2.95935690e-01
1.80743895e-02 4.40295696e-01 -1.11565685e+00 1.16444004e+00
-2.25090701e-02 6.01719201e-01 -4.51824993e-01 -1.23842978e+00
1.08820331e+00 7.60203525e-02 8.83963183e-02 -4.37443018e-01
1.40435606e-01 2.32335091e-01 1.84513137e-01 -3.45947266e-01
4.66853976e-01 -1.72013029e-01 -1.81338645e-03 5.38029134e-01
-1.40677616e-01 6.31911993e-01 -4.43690628e-01 2.60637611e-01
1.34937859e+00 -1.56698406e-01 5.65046787e-01 -3.91573310e-01
1.83758363e-01 -1.04772985e-01 7.08155632e-01 8.97032320e-01
-2.02486888e-02 4.34831977e-01 1.10274351e+00 -4.50996011e-01
-1.02084732e+00 -1.06258535e+00 1.73739299e-01 7.08414912e-01
2.56241828e-01 -5.71812809e-01 -4.61730212e-01 -9.34459448e-01
2.02786878e-01 8.81498814e-01 -1.05750275e+00 -5.33697903e-01
-3.34797740e-01 -5.95199466e-01 5.30150235e-01 5.89632750e-01
-5.42178713e-02 -7.13114798e-01 -5.22568464e-01 1.04301892e-01
-1.23820286e-02 -7.10223436e-01 -7.59884715e-02 4.71486837e-01
-1.18272710e+00 -1.16653621e+00 -2.52035230e-01 -3.29828471e-01
9.90872145e-01 4.78030890e-01 1.03702700e+00 3.53202790e-01
8.72549415e-02 -5.61157800e-02 -1.11467674e-01 -3.52902561e-01
-3.90502393e-01 4.83216252e-03 3.25427622e-01 -9.00187269e-02
4.49813128e-01 -9.99407470e-01 -5.45334518e-01 2.74064362e-01
-8.46069574e-01 1.46567896e-01 7.96200514e-01 7.47666240e-01
2.31379911e-01 -2.12754756e-02 8.00343990e-01 -1.00453293e+00
7.19560742e-01 -7.49448419e-01 -4.95670319e-01 2.24409342e-01
-1.06687200e+00 5.95541894e-01 9.82261658e-01 -4.64884937e-01
-5.01254320e-01 -2.10362628e-01 1.80489510e-01 -5.65699100e-01
-1.36892468e-01 6.75746083e-01 -4.37161207e-01 3.76193672e-01
8.66094112e-01 2.38716137e-02 -2.05510929e-01 -3.32362831e-01
6.49776697e-01 6.94707558e-02 2.47495085e-01 -3.59208226e-01
1.10706520e+00 3.31714332e-01 1.19837195e-01 -4.56429213e-01
-7.25109816e-01 7.66290203e-02 -3.55084360e-01 -5.97925745e-02
5.87237775e-01 -5.31691670e-01 -7.20729947e-01 -3.32711160e-01
-1.36639023e+00 1.91575155e-01 -2.75301993e-01 3.49109381e-01
-2.20093146e-01 1.34025082e-01 -2.19645016e-02 -8.11340272e-01
-3.73290619e-04 -1.00241923e+00 4.41459060e-01 1.25537723e-01
-5.80133557e-01 -1.10984910e+00 4.20625620e-02 -1.41999453e-01
3.22103173e-01 2.75656193e-01 1.54887307e+00 -1.05188644e+00
-7.45858848e-01 3.23054641e-02 -5.33119500e-01 -1.42771810e-01
3.59514058e-01 -6.09994344e-02 -1.11862779e+00 -4.32485975e-02
-1.24345206e-01 3.41289461e-01 8.45147133e-01 4.41524506e-01
9.88990188e-01 -7.35565484e-01 -5.31289697e-01 3.66995603e-01
1.54048789e+00 -1.66475847e-01 3.64928722e-01 2.71086693e-01
8.81644726e-01 7.64777541e-01 1.18282922e-01 3.65089774e-01
2.15851456e-01 5.06153107e-01 9.89739060e-01 -1.38646707e-01
2.86064483e-02 -7.70423710e-01 3.33966523e-01 5.97911298e-01
9.21036452e-02 -4.36052680e-01 -8.77119422e-01 6.80761218e-01
-2.12443495e+00 -1.00805640e+00 -5.86767018e-01 2.06309199e+00
3.70952159e-01 2.66659409e-01 -1.22484535e-01 2.45911017e-01
9.00449693e-01 2.68231332e-01 -8.14938307e-01 -4.41870958e-01
-1.36494726e-01 -2.89161146e-01 4.79856551e-01 2.93848813e-01
-4.50318933e-01 5.96377015e-01 6.29914665e+00 4.85780209e-01
-1.10483694e+00 -1.83868706e-01 6.68341696e-01 1.59811035e-01
-1.46213949e+00 4.16024476e-01 -7.23561466e-01 3.34996134e-01
8.41975212e-01 -3.49829584e-01 2.95438021e-01 8.80876303e-01
4.17243421e-01 4.39371318e-01 -1.34292948e+00 6.07488632e-01
-8.22449028e-02 -1.59562314e+00 4.51638550e-01 3.43182206e-01
5.87400913e-01 -2.23926395e-01 2.05422670e-01 3.42155658e-02
3.26078922e-01 -1.09466219e+00 7.43040383e-01 2.54684329e-01
4.39181447e-01 -6.93477690e-01 6.08688056e-01 2.63247788e-01
-9.85957921e-01 -2.62752682e-01 -6.48144960e-01 -3.73474598e-01
-1.87253356e-01 9.40196753e-01 -9.35600877e-01 5.91307163e-01
3.27080399e-01 7.12020993e-01 -5.79576492e-01 4.14931506e-01
-7.37365782e-01 8.86138082e-01 -4.78657335e-02 -2.38686442e-01
6.35879785e-02 -8.79238248e-02 6.67872369e-01 7.22128689e-01
4.52428043e-01 -1.38066471e-01 -5.07211208e-01 1.62041652e+00
-1.63715050e-01 1.32635040e-02 -1.04463506e+00 -1.51867688e-01
4.29458112e-01 9.88600671e-01 -7.47607112e-01 -3.54423630e-03
-3.81066859e-01 6.93479419e-01 4.20585662e-01 4.10926461e-01
-1.14111722e+00 4.95234057e-02 8.82757366e-01 3.18884760e-01
1.87336922e-01 2.09846094e-01 -7.35951841e-01 -1.18345785e+00
9.49108824e-02 -7.32820868e-01 1.82429388e-01 -4.71790433e-01
-1.36863899e+00 5.20406365e-01 -1.90765679e-01 -1.10417175e+00
-9.71352756e-02 -3.76684785e-01 -8.56125355e-01 6.26210034e-01
-1.52194357e+00 -9.23865676e-01 -1.75741851e-01 4.29699421e-01
1.83515012e-01 -5.70847467e-02 7.21662760e-01 2.78649647e-02
-6.97086811e-01 7.42023706e-01 -1.15393430e-01 -1.88571915e-01
4.13189292e-01 -1.24526918e+00 5.59824109e-01 1.18833876e+00
4.82470304e-01 1.11009645e+00 1.22731030e+00 -8.29255402e-01
-1.43689287e+00 -1.12544537e+00 1.15139067e+00 -3.67952198e-01
1.05615115e+00 -4.06824023e-01 -1.01327610e+00 7.00584412e-01
5.51503971e-02 -1.69190302e-01 6.89487815e-01 5.84836185e-01
-5.91011822e-01 7.31737912e-02 -7.29247212e-01 9.86963689e-01
1.39701796e+00 -3.27968448e-01 -6.04314089e-01 1.03559094e-02
1.03834951e+00 1.84322849e-01 -3.75060648e-01 4.03365165e-01
4.86436278e-01 -1.17888105e+00 8.63927245e-01 -9.56684470e-01
8.79534066e-01 -1.79709271e-01 -3.01374674e-01 -1.38729119e+00
-4.87885594e-01 -6.01710975e-01 -1.46641478e-01 1.37632871e+00
6.74757779e-01 -8.24312270e-01 1.03453636e+00 6.99403167e-01
-7.38019240e-04 -7.28628516e-01 -6.59159005e-01 -7.23796070e-01
-9.79102477e-02 -6.53546035e-01 1.00287151e+00 1.13388741e+00
1.94838479e-01 5.37853539e-01 -5.42325675e-01 5.65867305e-01
7.49333441e-01 3.08543026e-01 7.82974839e-01 -1.46051264e+00
-2.60714144e-01 -5.16215503e-01 -5.53064764e-01 -9.00081158e-01
1.88380346e-01 -1.09700525e+00 -2.49901488e-01 -1.40919542e+00
1.81398600e-01 -4.28834587e-01 -4.45064962e-01 6.64973617e-01
-2.38092005e-01 -1.56437680e-01 7.54244253e-02 2.74401695e-01
-4.33198847e-02 6.12297833e-01 8.65028858e-01 -1.15249567e-01
-8.49098638e-02 -8.99899751e-02 -1.39451981e+00 8.24453056e-01
8.86243522e-01 -7.50094950e-01 -7.12750733e-01 -3.32610756e-01
7.62052476e-01 -1.12570561e-01 6.77298129e-01 -5.74761868e-01
1.52524754e-01 -1.74968109e-01 1.49591237e-01 -2.72137463e-01
-3.02449644e-01 -1.07010019e+00 2.88458198e-01 5.92781842e-01
-5.15001297e-01 1.35462701e-01 1.17344372e-01 9.83913898e-01
-4.06257659e-02 -1.81981474e-01 2.47438282e-01 2.76898474e-01
-5.07020116e-01 9.04964581e-02 -3.28198612e-01 -1.73378795e-01
7.36886203e-01 -3.71729553e-01 -6.78039849e-01 -4.59641218e-01
-4.71816748e-01 -1.85076118e-01 3.27227324e-01 6.34086132e-01
7.12783635e-01 -1.51885295e+00 -3.06851298e-01 3.11019361e-01
2.30240002e-01 -4.54605162e-01 2.90821809e-02 8.22700202e-01
1.87935814e-01 4.97233897e-01 -4.56457697e-02 -5.37879229e-01
-9.10795271e-01 6.76597059e-01 5.05289473e-02 -1.78496078e-01
-6.04943395e-01 7.34400570e-01 6.16196692e-01 -3.42102200e-01
5.73631525e-02 -6.31207585e-01 -2.97126204e-01 6.39552474e-02
2.87232369e-01 7.84148648e-02 -2.21787527e-01 -1.23895757e-01
-4.41924661e-01 2.96372801e-01 1.12410933e-01 3.21405977e-01
1.48339891e+00 -1.26763821e-01 4.03173734e-03 3.79753917e-01
9.81536210e-01 2.70974636e-01 -1.01525640e+00 -1.44153368e-02
2.43017614e-01 -4.44940060e-01 -2.07344949e-01 -5.71819186e-01
-1.08254468e+00 9.42392647e-01 2.07793206e-01 7.42877126e-01
9.01756346e-01 8.87988359e-02 3.01810831e-01 5.38813248e-02
1.09942757e-01 -4.49409395e-01 1.30664945e-01 9.67701077e-02
9.38798964e-01 -1.01781321e+00 -3.39580961e-02 -5.17279923e-01
-2.44787887e-01 1.23998964e+00 6.37896299e-01 -5.21447994e-02
4.04180437e-01 4.29537222e-02 -1.23906352e-01 -5.69072008e-01
-8.82833004e-01 -4.82441857e-03 3.14453840e-01 4.32863861e-01
2.82713294e-01 7.84914792e-02 -2.49778345e-01 1.05584741e+00
-2.08444953e-01 -4.49980229e-01 7.34219313e-01 2.57313609e-01
-3.78304034e-01 -9.27252352e-01 -2.57585078e-01 3.93814206e-01
-1.20908596e-01 -2.43425667e-01 -5.84000230e-01 8.88979197e-01
-4.62188199e-02 9.65519667e-01 -1.52317330e-01 -8.02002788e-01
3.63380075e-01 -1.38161197e-01 -1.03070758e-01 -4.56956953e-01
-3.62501502e-01 -1.90037757e-01 -1.28506854e-01 -6.54724181e-01
-2.08492279e-01 -3.96663666e-01 -1.24924362e+00 -5.09522915e-01
-6.08490825e-01 9.10248756e-02 5.85677147e-01 1.02228212e+00
5.18132925e-01 7.70978272e-01 6.04255974e-01 -3.91813964e-01
-6.04295909e-01 -5.11461973e-01 -4.64993209e-01 2.60376185e-01
3.69734019e-01 -6.32403672e-01 -9.08474445e-01 -3.47426355e-01] | [8.324769973754883, 5.94536828994751] |
cf7bd50f-a193-4c2c-902d-1267cde08150 | rst-modnet-real-time-spatio-temporal-moving | 1912.00438 | null | https://arxiv.org/abs/1912.00438v1 | https://arxiv.org/pdf/1912.00438v1.pdf | RST-MODNet: Real-time Spatio-temporal Moving Object Detection for Autonomous Driving | Moving Object Detection (MOD) is a critical task for autonomous vehicles as moving objects represent higher collision risk than static ones. The trajectory of the ego-vehicle is planned based on the future states of detected moving objects. It is quite challenging as the ego-motion has to be modelled and compensated to be able to understand the motion of the surrounding objects. In this work, we propose a real-time end-to-end CNN architecture for MOD utilizing spatio-temporal context to improve robustness. We construct a novel time-aware architecture exploiting temporal motion information embedded within sequential images in addition to explicit motion maps using optical flow images.We demonstrate the impact of our algorithm on KITTI dataset where we obtain an improvement of 8% relative to the baselines. We compare our algorithm with state-of-the-art methods and achieve competitive results on KITTI-Motion dataset in terms of accuracy at three times better run-time. The proposed algorithm runs at 23 fps on a standard desktop GPU targeting deployment on embedded platforms. | ['Senthil Yogamani', 'Hazem Rashed', 'Mohamed Ramzy', 'Ahmad El Sallab'] | 2019-12-01 | null | null | null | null | ['moving-object-detection'] | ['computer-vision'] | [-1.67027131e-01 -2.18873620e-01 6.89996406e-03 -3.40665460e-01
-3.11332822e-01 -4.28607523e-01 6.52600408e-01 -2.22760051e-01
-9.76522088e-01 2.73583323e-01 -2.12681610e-02 -3.02675039e-01
2.89033711e-01 -6.30184054e-01 -8.42347085e-01 -5.39779603e-01
-2.47785062e-01 3.70328963e-01 9.46891248e-01 -2.38804176e-01
8.71717781e-02 6.16285622e-01 -1.64901745e+00 2.00595751e-01
3.67430001e-01 8.88988495e-01 4.55657750e-01 1.17370641e+00
2.41697118e-01 8.24689031e-01 -1.81026831e-02 -2.91731339e-02
4.45204437e-01 9.25981253e-02 -5.82308531e-01 1.15710655e-02
7.03523934e-01 -7.25845575e-01 -7.84292579e-01 8.28753233e-01
1.69546321e-01 2.88988322e-01 3.29992145e-01 -1.34204638e+00
1.60349071e-01 3.16702500e-02 -5.34757555e-01 6.81475699e-01
-1.26098916e-01 4.90576833e-01 5.43112874e-01 -8.31665158e-01
8.82602751e-01 1.24514246e+00 4.77023721e-01 5.19617677e-01
-9.00210798e-01 -5.82456291e-01 2.67730981e-01 8.69807839e-01
-1.16206110e+00 -6.84996068e-01 5.23648322e-01 -3.89681906e-01
1.12910557e+00 -1.72495097e-01 5.22722781e-01 7.22643435e-01
4.74679857e-01 8.33477437e-01 3.28091502e-01 -5.82756288e-03
2.17738003e-01 -1.22523844e-01 1.32205144e-01 8.12858939e-01
1.28162444e-01 4.03661340e-01 -2.89518118e-01 3.91115814e-01
3.01565856e-01 -3.97222564e-02 3.03691439e-02 -4.37208980e-01
-1.25480080e+00 6.74326837e-01 6.30266368e-01 1.47671849e-01
-6.08534813e-01 6.38958752e-01 4.41391647e-01 -4.92722876e-02
3.22889984e-01 -4.36368197e-01 -3.55071396e-01 -2.60428756e-01
-1.09933960e+00 5.15795112e-01 3.92814517e-01 8.47489655e-01
7.88527012e-01 2.30974525e-01 -1.70601070e-01 1.44936740e-01
4.51780885e-01 5.80429733e-01 2.92252332e-01 -1.13711214e+00
5.78286469e-01 3.04457963e-01 4.18545395e-01 -1.07152450e+00
-6.30993366e-01 -4.08263683e-01 -5.23576677e-01 5.69182336e-01
4.47881937e-01 -1.01358429e-01 -1.10588324e+00 1.59550154e+00
6.39459074e-01 8.19492817e-01 1.69470966e-01 1.15834868e+00
6.42817676e-01 8.40039074e-01 1.28555879e-01 -5.54910190e-02
1.25126719e+00 -1.27640307e+00 -7.06885874e-01 -4.86791283e-01
8.92506659e-01 -6.02892399e-01 4.34457511e-01 3.28507298e-03
-9.06567991e-01 -6.02243721e-01 -1.17691338e+00 -1.63202465e-01
-2.10030451e-01 2.26759970e-01 3.96353453e-01 4.50177789e-01
-1.13373137e+00 6.10397041e-01 -1.49916637e+00 -4.39870507e-01
5.54520547e-01 4.20382321e-01 -5.30732453e-01 -4.21154052e-02
-7.20137000e-01 8.64098012e-01 4.23788369e-01 1.97425038e-01
-1.22217572e+00 -6.52775764e-01 -9.05504704e-01 -1.52704373e-01
3.10512125e-01 -7.20703065e-01 1.32365012e+00 -7.00776577e-01
-1.35084844e+00 5.61875045e-01 -6.04289889e-01 -9.93740797e-01
9.39246476e-01 -3.04257661e-01 -2.46178195e-01 2.54876345e-01
2.01584119e-02 9.52596545e-01 7.14147329e-01 -9.24800098e-01
-1.25247204e+00 -2.22615048e-01 -2.25559566e-02 1.08999886e-01
1.61456391e-02 -1.17094949e-01 -7.47705400e-01 -2.52279699e-01
-6.36647642e-02 -1.37792373e+00 -3.11894715e-01 2.15114757e-01
7.58965015e-02 -8.10997784e-02 1.61243176e+00 -5.02638876e-01
8.19778681e-01 -2.06071568e+00 -1.38797835e-01 -4.03647959e-01
2.56967276e-01 7.10011303e-01 -1.42996296e-01 4.13283473e-03
1.54357031e-01 -3.73372346e-01 8.67476016e-02 -6.65427029e-01
-1.66837528e-01 2.20594943e-01 -8.58526081e-02 8.58936906e-01
3.34355772e-01 1.00655091e+00 -8.42613697e-01 -3.26580256e-01
6.48162842e-01 7.34389007e-01 -6.07702434e-01 7.52914175e-02
-1.32736832e-01 4.74351525e-01 -3.37183177e-01 3.11846942e-01
8.90399456e-01 1.65957153e-01 -2.70813107e-01 -1.61734834e-01
-4.75078642e-01 1.71452522e-01 -1.13377070e+00 1.44804418e+00
-4.16527361e-01 1.19078779e+00 3.09094824e-02 -7.56822824e-01
3.71026844e-01 1.54606074e-01 3.68375450e-01 -6.83589876e-01
2.51216859e-01 5.34195676e-02 1.85631111e-01 -5.38047731e-01
8.34815502e-01 1.76704541e-01 2.99450487e-01 4.01169620e-02
-2.36866504e-01 3.33906442e-01 3.18346649e-01 2.17679769e-01
1.14627099e+00 2.44393364e-01 -6.39897659e-02 -3.33949924e-01
5.35404563e-01 3.38787258e-01 6.18408442e-01 4.40387040e-01
-5.70706248e-01 4.05117959e-01 9.52090546e-02 -8.86336386e-01
-1.09014714e+00 -7.06457078e-01 1.08907439e-01 8.09249580e-01
4.76985514e-01 -1.33369431e-01 -6.11584961e-01 -4.45616722e-01
-2.03420430e-01 6.74618781e-01 -6.78512335e-01 -9.03367982e-05
-9.73126948e-01 -5.07933259e-01 2.91804790e-01 8.13188493e-01
7.29093552e-01 -7.74024606e-01 -1.30654705e+00 5.83255470e-01
-2.48087682e-02 -1.73941290e+00 -4.05984372e-01 -3.21116090e-01
-8.81255805e-01 -9.63936687e-01 -4.69381094e-01 -6.18912339e-01
3.27385366e-01 8.12384307e-01 7.96667457e-01 -1.05097294e-01
-2.78795630e-01 1.07595898e-01 -1.81558177e-01 -4.09488201e-01
-3.78014177e-01 -3.87353562e-02 4.17693630e-02 1.77015573e-01
2.18044117e-01 -2.67057329e-01 -1.07812440e+00 3.75747114e-01
-7.40473747e-01 2.27987006e-01 4.18886811e-01 4.25451428e-01
3.30695182e-01 7.83982277e-02 4.72329594e-02 -3.43339682e-01
-3.67718667e-01 -4.06870693e-01 -9.36062157e-01 -2.93146521e-01
-2.05983415e-01 -5.83341680e-02 3.46211195e-01 -3.17591310e-01
-9.67180371e-01 5.26412845e-01 -3.97039130e-02 -6.73946202e-01
-1.40038013e-01 -2.84294859e-02 1.73296660e-01 -1.97464660e-01
2.70778388e-01 1.45354971e-01 -1.69482734e-02 -1.05037890e-01
3.76691043e-01 2.87246644e-01 7.72567868e-01 2.79161818e-02
7.60445952e-01 1.18341398e+00 2.89769948e-01 -8.91897380e-01
-3.31087172e-01 -7.51993775e-01 -7.27015495e-01 -4.91105050e-01
1.02625692e+00 -1.26263165e+00 -8.54554832e-01 4.85986918e-01
-1.32097900e+00 -4.64773834e-01 2.76395708e-01 6.30870461e-01
-5.11608422e-01 3.54586840e-01 -3.96400392e-01 -6.82548463e-01
-3.52764606e-01 -1.41147578e+00 1.14748287e+00 2.04090178e-01
1.62071452e-01 -9.30585861e-01 -4.52681482e-02 2.64508724e-01
5.87448061e-01 3.81048352e-01 2.08167151e-01 -1.27246112e-01
-1.03582323e+00 -3.66512537e-01 -3.30888510e-01 -9.63153318e-02
-2.17666954e-01 1.62608624e-01 -9.29191530e-01 -5.21506190e-01
-1.84385642e-01 3.60040724e-01 1.22959399e+00 4.91873235e-01
5.49412489e-01 -1.36034936e-01 -6.52786314e-01 6.70043826e-01
1.50005305e+00 1.50875717e-01 5.85374653e-01 5.21307170e-01
9.53998685e-01 6.55376911e-01 9.06419694e-01 3.05745155e-01
6.87169194e-01 9.55795705e-01 7.50373781e-01 1.61268026e-01
-2.70410657e-01 1.14917606e-01 4.59991038e-01 2.79484898e-01
-7.55573139e-02 -3.39933753e-01 -9.78526771e-01 1.02990353e+00
-2.16192603e+00 -1.15128303e+00 -6.52563334e-01 2.09552622e+00
-5.68920597e-02 3.06487828e-01 2.24383801e-01 5.34303896e-02
5.57605922e-01 1.98488370e-01 -5.68565130e-01 -1.51947156e-01
2.37241253e-01 -3.03934306e-01 9.18755710e-01 8.34496558e-01
-1.41158676e+00 1.22809386e+00 5.40549517e+00 4.87630188e-01
-1.47230268e+00 2.91395247e-01 4.72600192e-01 -3.97029251e-01
4.16783094e-01 2.57062502e-02 -1.06405270e+00 3.27584684e-01
1.16187596e+00 -2.46657133e-01 -1.15520999e-01 9.05856311e-01
8.19532394e-01 -3.38184565e-01 -1.01196706e+00 8.43649685e-01
-8.32252353e-02 -1.47557652e+00 -3.41790110e-01 1.94826707e-01
7.31480300e-01 6.92532003e-01 -1.04089312e-01 1.47814050e-01
6.94763064e-02 -7.33865440e-01 1.05393398e+00 3.08900774e-01
1.50861874e-01 -9.23490882e-01 8.12487841e-01 4.12795603e-01
-1.56394076e+00 5.90335242e-02 -3.31860125e-01 -2.24504441e-01
5.27746499e-01 2.58776307e-01 -1.00928152e+00 4.14834142e-01
7.42735803e-01 8.10936213e-01 -5.54875553e-01 1.10580826e+00
2.02507123e-01 4.87786263e-01 -5.14176309e-01 1.53095573e-01
7.50821054e-01 1.18513674e-01 8.45042765e-01 1.37503648e+00
3.41411054e-01 -9.05382819e-03 1.83743224e-01 5.41260123e-01
2.83664048e-01 -2.62756228e-01 -5.84494591e-01 2.87160695e-01
-2.22862009e-02 1.25417447e+00 -8.02708983e-01 -5.37464917e-01
-4.27428335e-01 9.39961135e-01 1.42991558e-01 1.71800479e-01
-1.13843894e+00 -9.25730169e-02 1.03646624e+00 2.75184005e-01
8.21372569e-01 -7.46252835e-01 1.24518976e-01 -8.95378172e-01
1.29807130e-01 -8.97401422e-02 1.94760617e-02 -5.28626978e-01
-3.39548767e-01 6.82373583e-01 -3.79450582e-02 -1.36393130e+00
-4.07000184e-01 -6.54058695e-01 -6.17548943e-01 5.58811247e-01
-1.74960160e+00 -1.06832540e+00 -6.70803249e-01 4.08670366e-01
8.55158389e-01 7.50306696e-02 2.46353388e-01 4.31246907e-01
-4.90706325e-01 3.07312727e-01 7.56826103e-02 2.71063298e-02
3.27652127e-01 -8.65216434e-01 8.60413611e-01 1.27816451e+00
-1.05435001e-02 8.83739591e-02 9.63577628e-01 -5.26326120e-01
-1.61839128e+00 -1.60757732e+00 8.20671320e-01 -4.82093930e-01
5.67118049e-01 -1.60201699e-01 -8.64337206e-01 6.84863150e-01
1.96089402e-01 5.44737518e-01 -4.23032977e-02 -7.16534853e-01
-3.72414254e-02 -1.04191944e-01 -9.28613007e-01 6.06001556e-01
1.06438267e+00 -5.91468252e-02 -1.55302405e-01 1.26117095e-01
6.52988851e-01 -6.90218151e-01 -1.90323919e-01 4.75307703e-01
5.98668098e-01 -1.08389008e+00 7.50413179e-01 -4.03017670e-01
1.25932634e-01 -8.43210936e-01 -5.30951507e-02 -8.75267446e-01
-2.81301647e-01 -6.15556836e-01 -1.81354702e-01 6.94935322e-01
2.89139953e-02 -3.98824334e-01 1.05987585e+00 5.86521268e-01
-2.17933744e-01 -5.11418700e-01 -1.20575190e+00 -8.54525208e-01
-2.45739743e-01 -9.25388098e-01 1.64175421e-01 4.82113868e-01
-5.36203623e-01 7.89801627e-02 -4.06698912e-01 6.44483447e-01
7.47910738e-01 -4.58538160e-02 1.07135832e+00 -7.79117703e-01
8.90002176e-02 -3.60822231e-01 -1.12674272e+00 -1.12420416e+00
2.39899158e-01 -5.65549612e-01 2.41514742e-01 -1.40082502e+00
1.89801655e-03 1.65761318e-02 -1.63758807e-02 2.63456047e-01
-9.99928936e-02 6.01372182e-01 3.75163198e-01 5.17071746e-02
-7.51976073e-01 5.71478665e-01 9.97912228e-01 -1.57176465e-01
-1.71351925e-01 -3.40363905e-02 1.69024244e-01 7.30382085e-01
7.14485109e-01 -5.97419083e-01 -3.16560179e-01 -6.92428470e-01
-2.77998954e-01 3.30663882e-02 7.00951457e-01 -1.45326698e+00
5.72397709e-01 5.43559380e-02 1.15036458e-01 -1.08166015e+00
6.35779798e-01 -7.54794002e-01 2.15649694e-01 7.75265753e-01
8.81182030e-02 2.96966225e-01 6.13221049e-01 8.44056666e-01
-5.99433258e-02 1.14915207e-01 9.72485662e-01 1.99352667e-01
-1.33345306e+00 5.02657592e-01 -6.38269663e-01 -2.98513919e-01
1.40090895e+00 -2.24524245e-01 -1.89080060e-01 -3.29102308e-01
-4.48453069e-01 4.12090689e-01 4.41854745e-01 7.12873161e-01
5.98463774e-01 -1.11964917e+00 -7.89451063e-01 1.60312243e-02
1.09950371e-01 2.71888692e-02 5.46275914e-01 9.14022386e-01
-1.11392844e+00 8.65583718e-01 -2.04885066e-01 -9.95139837e-01
-1.60308516e+00 5.59547782e-01 4.76194650e-01 7.95296207e-02
-9.69901085e-01 6.41890228e-01 3.91977102e-01 1.32039160e-01
1.35886475e-01 -4.99187857e-01 -1.89474717e-01 -2.52032220e-01
8.11117172e-01 5.60640812e-01 2.10983947e-01 -1.15151894e+00
-6.19170070e-01 6.70342803e-01 -3.69217068e-01 -1.88074514e-01
1.12649834e+00 -2.38226771e-01 3.59271675e-01 6.79066554e-02
1.43173647e+00 -3.76329273e-01 -1.90078986e+00 -8.77491310e-02
-4.78844419e-02 -6.26701236e-01 5.70986390e-01 -1.99710876e-01
-1.24838150e+00 7.77084947e-01 1.12014544e+00 -3.17145079e-01
7.40434766e-01 -1.28565848e-01 1.01357269e+00 3.59354913e-01
4.70043808e-01 -7.48425245e-01 -1.69273302e-01 6.05236769e-01
5.42248666e-01 -1.59066367e+00 -5.78350350e-02 -4.27041441e-01
-4.87386227e-01 9.98456836e-01 5.33638537e-01 -4.07574624e-01
5.59590340e-01 8.24300274e-02 2.94222027e-01 4.99301068e-02
-9.97139513e-01 -5.26695907e-01 2.69600481e-01 4.43391144e-01
-7.85433128e-02 -1.28157511e-01 -8.62621367e-02 -2.15849996e-01
3.30340192e-02 7.91369900e-02 5.51544428e-01 8.94563556e-01
-3.75016421e-01 -6.96641743e-01 -1.50794372e-01 -1.18280381e-01
-4.41760927e-01 2.45767623e-01 2.30321154e-01 8.24690521e-01
1.63868055e-01 1.14179254e+00 3.41231793e-01 -3.26577753e-01
2.98502743e-01 -2.38340944e-01 2.12875918e-01 -2.28555396e-01
-1.58853889e-01 -4.65905070e-02 7.77715370e-02 -1.03787673e+00
-6.10990703e-01 -7.34787524e-01 -1.43970203e+00 -5.34284413e-01
-7.02589378e-02 -3.75247031e-01 1.05311441e+00 1.00662720e+00
7.21410811e-01 6.43610239e-01 5.28266191e-01 -1.52340293e+00
7.05486909e-02 -7.35948682e-01 2.23493222e-02 2.98238456e-01
6.88102603e-01 -6.25608921e-01 -1.75089732e-01 1.76950485e-01] | [8.149144172668457, -1.4163340330123901] |
9c602d26-89b7-491f-a896-70d74707d942 | 3dsam-adapter-holistic-adaptation-of-sam-from | 2306.13465 | null | https://arxiv.org/abs/2306.13465v1 | https://arxiv.org/pdf/2306.13465v1.pdf | 3DSAM-adapter: Holistic Adaptation of SAM from 2D to 3D for Promptable Medical Image Segmentation | Despite that the segment anything model (SAM) achieved impressive results on general-purpose semantic segmentation with strong generalization ability on daily images, its demonstrated performance on medical image segmentation is less precise and not stable, especially when dealing with tumor segmentation tasks that involve objects of small sizes, irregular shapes, and low contrast. Notably, the original SAM architecture is designed for 2D natural images, therefore would not be able to extract the 3D spatial information from volumetric medical data effectively. In this paper, we propose a novel adaptation method for transferring SAM from 2D to 3D for promptable medical image segmentation. Through a holistically designed scheme for architecture modification, we transfer the SAM to support volumetric inputs while retaining the majority of its pre-trained parameters for reuse. The fine-tuning process is conducted in a parameter-efficient manner, wherein most of the pre-trained parameters remain frozen, and only a few lightweight spatial adapters are introduced and tuned. Regardless of the domain gap between natural and medical data and the disparity in the spatial arrangement between 2D and 3D, the transformer trained on natural images can effectively capture the spatial patterns present in volumetric medical images with only lightweight adaptations. We conduct experiments on four open-source tumor segmentation datasets, and with a single click prompt, our model can outperform domain state-of-the-art medical image segmentation models on 3 out of 4 tasks, specifically by 8.25%, 29.87%, and 10.11% for kidney tumor, pancreas tumor, colon cancer segmentation, and achieve similar performance for liver tumor segmentation. We also compare our adaptation method with existing popular adapters, and observed significant performance improvement on most datasets. | ['Qi Dou', 'Pheng-Ann Heng', 'Jingyang Zhang', 'Zhao Wang', 'Jinpeng Li', 'Wenao Ma', 'Yuan Zhong', 'Shizhan Gong'] | 2023-06-23 | null | null | null | null | ['tumor-segmentation', 'medical-image-segmentation'] | ['computer-vision', 'medical'] | [ 2.82952964e-01 2.62196988e-01 -3.31878811e-01 -3.69495064e-01
-9.06446755e-01 -6.09111965e-01 1.85464248e-01 1.48385271e-01
-4.69257176e-01 3.35487038e-01 -7.62404501e-02 -6.64862752e-01
2.14768678e-01 -7.01736271e-01 -6.09372497e-01 -6.17171943e-01
3.74689773e-02 6.21826351e-01 5.85466623e-01 -1.46436557e-01
-1.20464712e-01 6.01497829e-01 -8.26984406e-01 1.67022347e-01
1.03294945e+00 1.05417955e+00 4.33085114e-01 6.59987867e-01
-5.56527734e-01 3.50140810e-01 -4.49271142e-01 -2.25986191e-03
2.07918510e-01 -1.95049152e-01 -1.06497884e+00 2.73591280e-01
4.00384635e-01 -3.96243721e-01 -1.70551166e-01 7.77942300e-01
6.28425479e-01 -2.68157065e-01 6.78887248e-01 -8.19145560e-01
-5.40619373e-01 5.27422845e-01 -5.77747345e-01 2.33974025e-01
-1.87129416e-02 4.65509355e-01 3.28409225e-01 -6.34838939e-01
6.42850399e-01 7.26460695e-01 9.29198921e-01 7.24273205e-01
-1.09255767e+00 -5.79118788e-01 1.66004196e-01 -3.45965207e-01
-1.17014849e+00 -1.15591392e-01 5.17088175e-01 -4.46780711e-01
9.05484855e-01 3.23266804e-01 7.86891937e-01 8.63051653e-01
2.54000992e-01 8.96218419e-01 8.89578581e-01 -5.15246466e-02
1.18967332e-01 1.58261254e-01 8.12187791e-02 7.94930935e-01
4.40146811e-02 -2.43438587e-01 2.10280493e-02 2.20651459e-02
1.03657258e+00 9.13331658e-02 -2.54521072e-01 -4.81311947e-01
-1.36563516e+00 5.89873791e-01 7.19215751e-01 4.33303386e-01
-2.05270678e-01 5.63130379e-02 4.37402785e-01 3.23087983e-02
4.80876684e-01 4.06579077e-01 -5.41508794e-01 8.34631454e-03
-1.03554261e+00 -1.30872458e-01 5.85386515e-01 1.19597304e+00
5.14148176e-01 -1.48057073e-01 -4.70243126e-01 7.82732129e-01
-1.82989277e-02 5.90232015e-01 6.61047757e-01 -7.16313183e-01
2.59935111e-01 7.67924368e-01 -1.96340084e-01 -5.14124572e-01
-1.00116849e+00 -7.14283109e-01 -9.16278303e-01 -1.02470674e-01
5.95734715e-01 1.57144275e-02 -1.57873821e+00 1.50331604e+00
5.28248489e-01 4.14105207e-02 -1.02214441e-01 9.00529742e-01
1.29790592e+00 2.26086736e-01 3.84405375e-01 1.32040277e-01
1.54617238e+00 -1.07176411e+00 -3.37169677e-01 -4.13408786e-01
8.48595381e-01 -7.15787888e-01 1.43186224e+00 -9.14001837e-02
-1.27305055e+00 -3.70401055e-01 -8.31987858e-01 -2.40375444e-01
-3.19927156e-01 1.93386842e-02 8.59941006e-01 6.59537017e-01
-1.02467477e+00 3.90144199e-01 -1.10446358e+00 -4.74480540e-01
8.28020811e-01 4.79446948e-01 -1.59283936e-01 -6.16131015e-02
-7.47885466e-01 5.84289968e-01 3.58354568e-01 -1.91431254e-01
-8.03175688e-01 -1.47575808e+00 -8.10972095e-01 -2.12414563e-01
2.42912203e-01 -1.01852441e+00 1.36941206e+00 -9.63418067e-01
-1.51087487e+00 1.29269958e+00 1.42052257e-02 -2.84227610e-01
8.08648586e-01 2.90340662e-01 -1.75217822e-01 4.17004645e-01
1.87380806e-01 1.14685380e+00 6.45967662e-01 -1.04436243e+00
-5.27342856e-01 -3.52245182e-01 -9.80711654e-02 2.35911548e-01
-3.30435574e-01 -3.67515564e-01 -9.01879430e-01 -7.65861630e-01
2.79279888e-01 -1.02515054e+00 -4.97203737e-01 4.26901489e-01
-3.73984009e-01 1.83850542e-01 9.24564540e-01 -5.73310256e-01
8.69889855e-01 -2.18584967e+00 -1.45655900e-01 1.87553093e-01
2.46075809e-01 2.21373215e-01 -1.12261593e-01 -3.55993956e-01
1.22218356e-01 1.42475307e-01 -6.53144538e-01 -2.33072162e-01
-2.16504306e-01 2.33579233e-01 1.73314475e-02 3.34930778e-01
3.25608961e-02 1.34931982e+00 -9.34778929e-01 -8.22156608e-01
3.36662292e-01 3.54378164e-01 -7.24447906e-01 2.44349182e-01
-3.06461722e-01 7.90534437e-01 -5.99366486e-01 9.57502246e-01
9.21950400e-01 -7.32337832e-01 2.18752939e-02 -4.76595879e-01
1.67167261e-01 8.08901936e-02 -6.81738675e-01 2.18332314e+00
-6.01122141e-01 2.09570572e-01 2.87035048e-01 -8.56395900e-01
5.21803916e-01 2.86398321e-01 9.93294835e-01 -9.29115117e-01
6.37855679e-02 2.32793450e-01 -1.32883817e-01 -5.30863881e-01
1.72824502e-01 -4.01554964e-02 -1.59923434e-01 2.98902661e-01
-8.40898529e-02 -6.37272358e-01 1.22773901e-01 1.68165803e-01
1.09936690e+00 -1.58125564e-01 1.04533933e-01 -4.29565519e-01
3.68748605e-01 5.59334934e-01 3.45062226e-01 7.30850697e-01
-3.77264082e-01 8.57323349e-01 3.37482452e-01 -4.86751139e-01
-8.43299687e-01 -1.38275003e+00 -4.50638443e-01 9.58140254e-01
4.39048052e-01 -1.01866849e-01 -9.06190574e-01 -9.50541198e-01
-5.82358651e-02 4.19211984e-01 -5.94884217e-01 -1.49400430e-02
-6.47412598e-01 -9.50153530e-01 6.43408120e-01 7.69569516e-01
7.90314436e-01 -8.93525422e-01 -8.24006200e-01 3.24026644e-01
-2.81623930e-01 -1.22620285e+00 -7.34306455e-01 1.64916962e-01
-1.22052860e+00 -9.98736441e-01 -1.05093849e+00 -9.76526976e-01
9.32608306e-01 3.47207755e-01 1.33774424e+00 2.19039038e-01
-5.59353888e-01 2.51801908e-01 -1.36334300e-01 -3.65972370e-01
-4.30664062e-01 5.06119788e-01 -6.37670994e-01 -6.43342435e-01
1.12471171e-01 -3.65138203e-01 -1.03612435e+00 5.03863394e-01
-1.07248998e+00 5.93027532e-01 7.37181425e-01 9.21643019e-01
8.23750615e-01 -3.55120748e-01 2.32098505e-01 -1.09821224e+00
2.17891052e-01 -3.29997748e-01 -2.98822165e-01 1.17321648e-01
-3.70894015e-01 -1.34633169e-01 5.09462535e-01 -5.49486518e-01
-9.00811136e-01 2.67989397e-01 -3.71165574e-01 -1.73629120e-01
-3.50240558e-01 3.12549949e-01 1.46913692e-01 -2.55036682e-01
6.66459024e-01 2.09890097e-01 2.72266805e-01 -3.37188870e-01
1.72193751e-01 4.93389308e-01 5.00468731e-01 -5.85298061e-01
4.65499341e-01 7.66896307e-01 -1.18107140e-01 -5.54682493e-01
-7.14981079e-01 -4.51283604e-01 -5.29601872e-01 8.12735558e-02
9.87811685e-01 -8.53333235e-01 -4.00111198e-01 6.04939282e-01
-6.90973401e-01 -8.84957254e-01 -4.52048987e-01 2.02574283e-01
-4.51197028e-01 2.00585276e-01 -9.71346438e-01 1.14905678e-01
-6.96779728e-01 -1.58374429e+00 1.29271388e+00 2.18752697e-01
-2.25106373e-01 -1.08177841e+00 -3.57331634e-01 4.62867558e-01
9.06643927e-01 2.90043563e-01 1.21072710e+00 -4.91510212e-01
-5.37403226e-01 4.09754738e-02 -5.20035028e-01 -2.86082663e-02
4.26511437e-01 -3.28740716e-01 -8.08019280e-01 -3.93956274e-01
-2.19127864e-01 -1.52812526e-01 7.31300414e-01 7.88128734e-01
1.72629619e+00 2.23312099e-02 -7.02887118e-01 1.15967512e+00
1.13401246e+00 2.64899842e-02 3.85972261e-01 3.34889501e-01
7.54114985e-01 1.88810229e-01 4.19358522e-01 1.49133071e-01
4.15327698e-01 6.11509264e-01 5.00219285e-01 -8.63499761e-01
-4.36569661e-01 5.82074281e-03 -1.70549780e-01 6.57430947e-01
1.90367877e-01 6.25222400e-02 -1.13642979e+00 6.92546308e-01
-1.46873116e+00 -3.89067501e-01 1.02858759e-01 1.90605867e+00
1.14053273e+00 1.11401990e-01 4.11624536e-02 -3.84454459e-01
2.79118031e-01 -5.10681309e-02 -9.22243953e-01 -8.53786692e-02
3.20324823e-02 3.64608407e-01 9.51996028e-01 2.61674881e-01
-1.24630952e+00 9.73225296e-01 6.84103823e+00 9.54846323e-01
-1.55339527e+00 2.35430986e-01 9.55267429e-01 -1.63547948e-01
-3.98858011e-01 -2.96938211e-01 -4.14218277e-01 4.39101309e-01
5.52796662e-01 1.81233361e-01 4.88266051e-02 7.57113039e-01
5.51790418e-03 2.63857450e-02 -9.72242653e-01 9.41260397e-01
-3.21833432e-01 -1.52717483e+00 8.99246633e-02 -1.49906182e-03
7.44081318e-01 1.64913520e-01 2.98369735e-01 2.09852248e-01
1.06160119e-01 -1.22885275e+00 5.31791985e-01 7.08095878e-02
1.12626374e+00 -4.06925559e-01 6.44282579e-01 2.12545618e-01
-1.08023357e+00 2.22806856e-01 -1.80666953e-01 5.52069008e-01
1.07608207e-01 7.15258360e-01 -1.12952065e+00 3.81655931e-01
7.56461799e-01 6.27140224e-01 -6.60622537e-01 9.18041348e-01
9.20187011e-02 5.34785271e-01 -5.26290417e-01 2.85159200e-01
4.23982531e-01 1.62671357e-01 2.93829441e-01 1.42878282e+00
3.34986329e-01 5.89045174e-02 2.55529135e-01 7.81226397e-01
-9.98403877e-02 6.03325479e-02 -1.68086097e-01 3.86651278e-01
2.60485649e-01 1.36488795e+00 -1.05289805e+00 -4.44657683e-01
-3.62297684e-01 9.19666588e-01 4.14826460e-02 2.95880824e-01
-1.24442828e+00 1.49945179e-02 4.80816513e-01 3.25463444e-01
3.37335855e-01 -7.05582201e-02 -6.74191058e-01 -8.55490744e-01
-6.75296783e-02 -8.02924395e-01 5.51069975e-01 -4.80737418e-01
-1.16946578e+00 6.38682008e-01 -5.00586703e-02 -1.10394990e+00
-1.25820152e-02 -6.56571329e-01 -5.10141551e-01 5.91176331e-01
-1.64348471e+00 -1.40095937e+00 -7.43800819e-01 7.73744702e-01
4.80766386e-01 1.75582573e-01 7.80418098e-01 3.93613279e-01
-4.43825096e-01 8.57213140e-01 -1.34518981e-01 2.55103171e-01
7.78605700e-01 -1.37338376e+00 4.09730464e-01 3.95461947e-01
-4.05077785e-01 2.54152805e-01 3.76472801e-01 -4.57890183e-01
-1.36330020e+00 -1.33231413e+00 1.94976121e-01 -2.59369910e-01
2.90662825e-01 -2.24619180e-01 -8.73456120e-01 6.18492186e-01
3.17264311e-02 4.82092410e-01 6.70608044e-01 -2.74048120e-01
-1.34450346e-01 -8.71953070e-02 -1.56374109e+00 7.03071713e-01
1.28096211e+00 -1.42684892e-01 -1.84969336e-01 5.39437234e-01
8.18111956e-01 -1.22163498e+00 -1.23027074e+00 6.22202396e-01
3.40526909e-01 -7.29447603e-01 1.33842301e+00 -3.54047239e-01
3.74415845e-01 -2.30357617e-01 1.46438330e-01 -1.14559114e+00
-2.92714179e-01 -2.80781597e-01 1.45649865e-01 7.48419464e-01
5.42573988e-01 -7.22270131e-01 1.11090434e+00 7.29183435e-01
-5.88468969e-01 -9.22702491e-01 -8.47686768e-01 -4.85783637e-01
3.89486909e-01 -2.64886588e-01 7.22657442e-01 8.39480519e-01
-2.12826520e-01 -1.43756703e-01 3.81836027e-01 5.36321513e-02
5.35415113e-01 3.68261099e-01 6.16612673e-01 -6.30012095e-01
-2.16620669e-01 -7.03974247e-01 -2.10259467e-01 -1.31706548e+00
-1.86692163e-01 -1.12274277e+00 -8.34921300e-02 -1.57274008e+00
2.77918220e-01 -9.01090920e-01 -2.82101452e-01 6.63549721e-01
-1.87212020e-01 3.90583217e-01 3.71959209e-02 2.64071405e-01
-4.15526807e-01 1.72619253e-01 1.82930350e+00 -6.12955749e-01
-2.97032535e-01 6.50442485e-03 -7.52471507e-01 7.00922132e-01
8.06194127e-01 -2.77416199e-01 -4.41334128e-01 -8.03357244e-01
-4.30596620e-01 1.98181961e-02 4.40002173e-01 -9.72817361e-01
2.57849663e-01 -2.24053606e-01 6.50911212e-01 -5.40473044e-01
1.05019249e-01 -8.96399558e-01 1.10980257e-01 6.16809964e-01
-1.73461989e-01 -1.58217907e-01 6.56546831e-01 1.20688066e-01
-5.44986539e-02 9.85988006e-02 1.03086221e+00 -3.55828255e-01
-7.65021265e-01 6.63531482e-01 -1.45337805e-01 3.21392566e-01
1.19404709e+00 -4.82299089e-01 -2.90094912e-01 1.37167022e-01
-8.80265296e-01 3.20663661e-01 6.67786956e-01 2.23620579e-01
4.63663459e-01 -1.04615474e+00 -5.90983987e-01 2.74663329e-01
1.91901773e-01 6.91173911e-01 6.54285073e-01 1.06473649e+00
-9.21176612e-01 4.33587968e-01 -1.47476032e-01 -1.13333058e+00
-8.84581089e-01 4.65754002e-01 7.03042686e-01 -6.32598221e-01
-9.78502929e-01 7.86179304e-01 5.19759297e-01 -5.43690205e-01
1.11114517e-01 -8.77420545e-01 2.54867017e-01 -4.51744139e-01
3.30848336e-01 -1.31329849e-01 4.45711076e-01 -2.68363297e-01
-5.67045093e-01 7.68439353e-01 -2.82831520e-01 3.21167946e-01
1.17765880e+00 -3.53171933e-03 1.89353526e-01 -8.47871974e-02
1.24768293e+00 -2.27739453e-01 -1.52188885e+00 -3.15715730e-01
-3.52550656e-01 -1.73011228e-01 1.67460412e-01 -1.04829550e+00
-1.57267284e+00 6.56976044e-01 7.31300056e-01 -1.19572924e-02
1.35171318e+00 2.07632795e-01 1.26380098e+00 -1.39651135e-01
2.76759177e-01 -9.74920273e-01 -6.64903671e-02 3.35552573e-01
4.89847988e-01 -1.37857747e+00 -2.24733409e-02 -7.54527748e-01
-5.06164789e-01 9.20390904e-01 6.73179626e-01 6.35856688e-02
6.84983552e-01 6.62104130e-01 2.39721701e-01 -2.76131958e-01
-3.66399914e-01 -5.17836288e-02 3.49726319e-01 7.63263166e-01
4.62109745e-01 2.53174275e-01 3.27984318e-02 4.17589694e-01
-1.85270518e-01 1.91992354e-02 1.46794394e-01 9.51844335e-01
-2.13383466e-01 -8.08939099e-01 -1.90524235e-01 5.87201238e-01
-4.64454830e-01 -6.43333271e-02 6.70775250e-02 9.77206528e-01
-1.06931515e-02 5.12148559e-01 3.25865686e-01 1.42455101e-01
5.20823956e-01 -3.63041818e-01 5.36178172e-01 -5.05842090e-01
-7.95039475e-01 3.18319686e-02 -3.45806390e-01 -7.70331264e-01
-2.19031215e-01 -3.57950717e-01 -1.60593009e+00 -2.33684316e-01
1.08016185e-01 -2.65064180e-01 6.54793799e-01 8.45172524e-01
5.17819524e-01 8.62333596e-01 2.89860129e-01 -7.30567455e-01
-4.13595438e-01 -6.49593055e-01 -2.38667235e-01 6.96053922e-01
1.95583746e-01 -4.02721852e-01 -4.89085075e-03 8.15377906e-02] | [14.678135871887207, -2.5143747329711914] |
3a1a302f-4534-4f9c-bf7f-7e53cf432be4 | losh-long-short-text-joint-prediction-network | 2306.08736 | null | https://arxiv.org/abs/2306.08736v1 | https://arxiv.org/pdf/2306.08736v1.pdf | LoSh: Long-Short Text Joint Prediction Network for Referring Video Object Segmentation | Referring video object segmentation (RVOS) aims to segment the target instance referred by a given text expression in a video clip. The text expression normally contains sophisticated descriptions of the instance's appearance, actions, and relations with others. It is therefore rather difficult for an RVOS model to capture all these attributes correspondingly in the video; in fact, the model often favours more on the action- and relation-related visual attribute of the instance. This can end up with incomplete or even incorrect mask prediction of the target instance. In this paper, we tackle this problem by taking a subject-centric short text expression from the original long text expression. The short one retains only the appearance-related information of the target instance so that we can use it to focus the model's attention on the instance's appearance. We let the model make joint predictions using both long and short text expressions and introduce a long-short predictions intersection loss to align the joint predictions. Besides the improvement on the linguistic part, we also introduce a forward-backward visual consistency loss, which utilizes optical flows to warp visual features between the annotated frames and their temporal neighbors for consistency. We build our method on top of two state of the art transformer-based pipelines for end-to-end training. Extensive experiments on A2D-Sentences and JHMDB-Sentences datasets show impressive improvements of our method. | ['Zijie Yue', 'Miaojing Shi', 'Linfeng Yuan'] | 2023-06-14 | null | null | null | null | ['referring-expression-segmentation', 'referring-video-object-segmentation', 'video-object-segmentation', 'video-semantic-segmentation'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [ 2.20213220e-01 2.87015438e-01 -2.82645851e-01 -5.39761066e-01
-4.90116745e-01 -3.66198212e-01 5.49550414e-01 -1.47530315e-02
-3.05878997e-01 5.18213451e-01 5.30804880e-02 -3.32800001e-02
3.73270214e-01 -4.03584749e-01 -8.64877701e-01 -5.18130898e-01
2.45136350e-01 5.63869596e-01 7.52416074e-01 -6.33960441e-02
-3.47504430e-02 3.76112521e-01 -1.27327693e+00 6.20101452e-01
4.08363640e-01 1.35211039e+00 4.50920939e-01 4.65594321e-01
-2.00198591e-01 1.39407969e+00 -5.42620540e-01 -6.82572007e-01
2.36539245e-01 -6.03960991e-01 -1.13741994e+00 5.27567983e-01
8.45602036e-01 -4.66022789e-01 -5.22093654e-01 1.05502939e+00
1.88115254e-01 2.96938688e-01 4.40326601e-01 -1.39365268e+00
-3.50124210e-01 4.43083525e-01 -7.21094549e-01 4.25559461e-01
4.75692809e-01 2.26411074e-01 1.20469189e+00 -6.91435337e-01
1.23389459e+00 1.27087343e+00 4.03814971e-01 5.99260569e-01
-1.28026080e+00 -3.65948558e-01 7.29567468e-01 4.95671540e-01
-1.07626581e+00 -5.55635691e-01 9.88735199e-01 -4.92101490e-01
8.53954673e-01 1.37915835e-01 9.03501213e-01 1.22321963e+00
-5.52083813e-02 1.19343174e+00 8.34871888e-01 -1.43765092e-01
6.92752674e-02 8.09730887e-02 -1.32291302e-01 8.02112520e-01
-3.96593958e-01 -6.27120286e-02 -3.97634357e-01 4.26280618e-01
6.06815934e-01 9.09803435e-03 -3.46344948e-01 -5.69589555e-01
-1.12457287e+00 4.37878460e-01 5.03946006e-01 3.22175920e-01
-5.44605494e-01 2.52187818e-01 4.06596214e-01 1.44464504e-02
5.66215992e-01 -8.05568025e-02 -5.77563107e-01 -3.39895599e-02
-1.25707686e+00 2.28249878e-01 5.48566520e-01 1.10302675e+00
7.36156702e-01 -1.69007465e-01 -4.87124085e-01 7.77319849e-01
1.58697858e-01 8.98621082e-02 1.23501092e-01 -1.20119548e+00
4.94907677e-01 5.38440049e-01 -3.38365883e-02 -1.02932990e+00
-2.18465433e-01 -2.27924526e-01 -4.70356345e-01 9.72954258e-02
6.04913652e-01 1.63034439e-01 -1.04313719e+00 1.76114118e+00
5.42032182e-01 4.34797198e-01 -1.45761371e-01 1.11575532e+00
8.60032022e-01 7.57863343e-01 2.53275484e-01 -2.59427100e-01
1.47300816e+00 -1.29836881e+00 -8.46488714e-01 -4.82858896e-01
5.81182539e-01 -5.99710405e-01 8.31649423e-01 9.33636278e-02
-1.34629965e+00 -6.56004965e-01 -6.13893867e-01 -3.97448272e-01
-1.04885943e-01 3.03128976e-02 1.89161047e-01 -3.37464325e-02
-8.80722046e-01 5.58656633e-01 -8.29393923e-01 -3.34019899e-01
6.10541105e-01 6.53500706e-02 -5.07475019e-01 -1.87846959e-01
-1.04305077e+00 9.35028195e-01 3.28074872e-01 2.39192545e-01
-7.66457021e-01 -5.76350868e-01 -9.88335431e-01 -2.28323210e-02
6.89610898e-01 -7.03527510e-01 1.22136879e+00 -1.61756170e+00
-1.16167903e+00 1.12140584e+00 -4.53312337e-01 -5.39833963e-01
8.48966300e-01 -3.81162055e-02 -1.07476994e-01 5.18305898e-01
2.98312962e-01 1.05037534e+00 9.70857799e-01 -1.20787287e+00
-8.92294049e-01 -2.82315522e-01 3.78208280e-01 2.53308117e-01
2.11962819e-01 1.49497047e-01 -1.06717193e+00 -6.81991100e-01
2.22250614e-02 -9.79421258e-01 4.46851328e-02 4.85334396e-01
-5.16068339e-01 -2.82022506e-01 9.91395116e-01 -7.83455670e-01
1.02704573e+00 -2.24527812e+00 4.08126354e-01 -1.61786377e-01
1.50204644e-01 2.81547040e-01 -1.77219346e-01 1.60678718e-02
-1.45768300e-01 -1.85472861e-01 -2.85787821e-01 -7.44225562e-01
-2.44603083e-01 4.95717049e-01 -2.73480982e-01 5.79236984e-01
4.08361673e-01 9.20738578e-01 -9.81830180e-01 -8.01457345e-01
4.07608479e-01 4.98905063e-01 -5.77014208e-01 3.54665756e-01
-6.19388521e-01 5.65670013e-01 -4.21812445e-01 5.67486227e-01
4.22722310e-01 -1.62737504e-01 -1.01436481e-01 -6.50684416e-01
8.51353630e-02 2.38395825e-01 -9.62292790e-01 1.82889259e+00
-3.29828858e-01 8.21350515e-01 3.73153426e-02 -1.21298528e+00
5.42858839e-01 2.73499131e-01 6.47840083e-01 -7.16613889e-01
4.05866578e-02 -2.47319438e-03 -1.25546277e-01 -9.12771225e-01
1.46585211e-01 -7.91493356e-02 2.19743997e-01 8.26884881e-02
1.84966877e-01 2.57289186e-02 4.52270180e-01 4.46730971e-01
7.31548905e-01 7.40562379e-01 8.24902058e-02 2.44135544e-01
7.68141687e-01 -9.47994087e-03 6.94122553e-01 3.51761013e-01
-4.12962943e-01 8.48481178e-01 7.02696919e-01 -4.99376297e-01
-9.13576186e-01 -7.39395499e-01 2.76651736e-02 1.06406248e+00
3.76050889e-01 -4.20477808e-01 -6.69191062e-01 -1.12744379e+00
-3.36922914e-01 8.48164380e-01 -8.56871068e-01 3.41036879e-02
-6.19687200e-01 -8.52691010e-02 2.12623402e-01 6.14128053e-01
5.64725220e-01 -1.16959643e+00 -6.67835414e-01 2.83256382e-01
-5.87903738e-01 -1.78495693e+00 -7.54942119e-01 1.09424554e-01
-5.58544457e-01 -1.03351545e+00 -6.59419000e-01 -8.34483624e-01
7.41411746e-01 2.98791360e-02 1.09922552e+00 1.80905297e-01
-9.38241631e-02 3.53898466e-01 -5.29854774e-01 -4.20011617e-02
-5.07608771e-01 -4.14421648e-01 -3.31063122e-01 4.22702461e-01
9.96799842e-02 -3.01604360e-01 -5.91194868e-01 3.35763901e-01
-8.32498372e-01 3.97062212e-01 3.19190025e-01 7.76151121e-01
8.01305354e-01 -2.66180426e-01 3.90352011e-02 -7.56136596e-01
-2.10229293e-01 -1.69101283e-01 -4.11985338e-01 3.45001072e-01
-1.54547945e-01 7.19117150e-02 4.55316663e-01 -4.57153440e-01
-9.66667652e-01 4.74890769e-01 -1.39225572e-01 -9.85855460e-01
-1.33895308e-01 1.13304555e-01 -1.41285941e-01 2.59606510e-01
5.44312298e-02 2.05219954e-01 -5.63621037e-02 -3.05618405e-01
4.07865256e-01 3.53755862e-01 7.19574869e-01 -5.10530114e-01
4.39477712e-01 5.76737702e-01 2.97393519e-02 -7.17839241e-01
-1.20113099e+00 -5.89506328e-01 -8.27412426e-01 -3.91464502e-01
1.31058991e+00 -8.13847840e-01 -4.41686332e-01 1.44096330e-01
-1.55323553e+00 -4.10456091e-01 -4.00686949e-01 1.58393204e-01
-8.12305808e-01 4.13755208e-01 -3.90648514e-01 -5.24576485e-01
3.86623554e-02 -1.45810115e+00 1.43311858e+00 -8.28666314e-02
-1.75336778e-01 -9.48973000e-01 -2.18734607e-01 5.34315467e-01
-8.42690542e-02 1.85893461e-01 6.77230239e-01 -7.64803410e-01
-6.09908700e-01 -1.11958541e-01 -3.80509257e-01 4.71445858e-01
-5.43977395e-02 2.33572632e-01 -9.59875166e-01 5.14748320e-03
-6.06254488e-03 -1.24874920e-01 9.27416921e-01 4.24667209e-01
1.14312136e+00 -3.32672447e-01 -3.41448694e-01 6.86104655e-01
1.15289891e+00 1.04194738e-01 6.54810965e-01 1.84201330e-01
1.00495672e+00 9.53294218e-01 9.19045031e-01 2.15825755e-02
3.93351644e-01 1.15056586e+00 6.88511372e-01 -1.72043771e-01
-4.59709078e-01 -2.32449159e-01 5.72013974e-01 2.38505900e-01
-1.63124204e-01 -3.26019734e-01 -5.91585815e-01 5.17504036e-01
-1.95595407e+00 -1.22031951e+00 -1.83685601e-01 1.82903945e+00
7.73600519e-01 4.68935370e-02 2.16360852e-01 -6.21852800e-02
7.25473523e-01 3.26240957e-01 -4.22964990e-01 -2.70067185e-01
-7.69142061e-02 -2.84442782e-01 7.09007308e-02 4.35623825e-01
-1.30279505e+00 1.06298745e+00 5.01305628e+00 8.53449762e-01
-1.15754867e+00 1.74496740e-01 8.50265801e-01 -2.28570685e-01
7.99204409e-02 8.25752616e-02 -7.35627234e-01 4.55041438e-01
5.78049421e-01 1.86613545e-01 2.11866662e-01 6.11941397e-01
3.52862388e-01 -3.01629663e-01 -1.54903519e+00 9.18159246e-01
1.61953345e-01 -1.35648966e+00 1.15004681e-01 -2.04082906e-01
5.31099319e-01 -1.51890084e-01 -1.95750937e-01 9.65201780e-02
-2.62127072e-01 -7.00837910e-01 1.34451604e+00 3.89896750e-01
7.51578927e-01 -4.60542172e-01 6.35676444e-01 1.91960126e-01
-1.26384866e+00 3.83492261e-02 -2.63826270e-02 2.22108632e-01
4.49383080e-01 1.63803563e-01 -6.39676094e-01 4.39995259e-01
7.29348242e-01 1.17432702e+00 -5.87614655e-01 7.59422779e-01
-4.28361803e-01 3.08987975e-01 -1.70304030e-01 3.10221136e-01
4.74847883e-01 -2.48521447e-01 7.83976912e-01 1.22538149e+00
-4.71222587e-02 1.61779389e-01 3.05346847e-01 9.36629891e-01
-1.79484077e-02 -3.63688022e-02 -5.17832458e-01 1.78608626e-01
5.14239296e-02 1.14589405e+00 -7.51525700e-01 -6.08836472e-01
-6.18378162e-01 1.44423318e+00 2.38656744e-01 4.86912191e-01
-1.09051776e+00 1.01682246e-01 4.70344961e-01 2.99839616e-01
7.29671717e-01 1.61093742e-01 1.07423283e-01 -1.16565311e+00
2.80660570e-01 -6.66928232e-01 3.99306267e-01 -1.11863887e+00
-1.03553963e+00 6.83589637e-01 -8.52141250e-03 -1.32037783e+00
-3.55478019e-01 -3.83949459e-01 -4.40746337e-01 5.89912772e-01
-1.47550297e+00 -1.30765474e+00 -1.92468047e-01 6.22060359e-01
9.95041192e-01 3.27134699e-01 3.51635277e-01 3.44248712e-01
-6.49446845e-01 3.38742226e-01 -5.07389426e-01 3.72158766e-01
6.67425096e-01 -1.12248659e+00 8.68891552e-02 9.34986711e-01
4.57288325e-01 1.03725769e-01 9.58534420e-01 -6.24317408e-01
-9.02114749e-01 -1.25287056e+00 9.90892053e-01 -5.38009167e-01
7.23584831e-01 -2.09851608e-01 -9.79017496e-01 9.91175294e-01
2.15777919e-01 5.40026486e-01 1.36657488e-02 -4.14705664e-01
-2.99290568e-01 -1.50247589e-01 -1.00939357e+00 5.73847353e-01
1.13823164e+00 -5.37194312e-01 -6.75593197e-01 3.23552787e-01
7.17916250e-01 -5.02801120e-01 -5.50715983e-01 3.08059275e-01
4.32895094e-01 -1.08595181e+00 1.02918184e+00 -7.89743364e-01
7.73754597e-01 -3.59332323e-01 -1.30378932e-01 -8.89546692e-01
1.49141764e-03 -5.00887096e-01 -1.45612836e-01 1.37557709e+00
1.62270129e-01 6.49338663e-02 5.90496361e-01 7.99434125e-01
-2.14598805e-01 -9.11942899e-01 -1.02271664e+00 -6.07421219e-01
-2.52000242e-01 -5.82963169e-01 1.31278798e-01 8.18779111e-01
-2.11480141e-01 3.89130503e-01 -4.25568014e-01 3.25738154e-02
3.91678065e-01 1.84532270e-01 6.55418992e-01 -8.75057101e-01
-2.39799663e-01 -3.85370553e-01 -6.95531487e-01 -1.29294550e+00
4.75028485e-01 -7.30051816e-01 2.00098351e-01 -1.57890868e+00
1.69754446e-01 -1.78671271e-01 -3.89278233e-02 6.19543731e-01
-1.97234377e-01 3.58908236e-01 4.45737422e-01 7.19406158e-02
-8.84811342e-01 4.44568515e-01 1.46917558e+00 -4.11929697e-01
-1.14290990e-01 2.18368955e-02 -4.09641638e-02 8.65632296e-01
2.65319854e-01 -5.48344851e-01 -3.78514469e-01 -3.98910642e-01
-1.09274842e-01 3.44675303e-01 7.18611121e-01 -6.54082358e-01
2.27796406e-01 -1.78216055e-01 3.40901852e-01 -4.87944186e-01
5.55678844e-01 -1.06952357e+00 6.90229163e-02 2.45048285e-01
-5.64515233e-01 -6.37709200e-02 5.29818386e-02 6.40474737e-01
-2.90765852e-01 -1.40136376e-01 9.18403149e-01 -7.27120414e-02
-9.74087834e-01 5.02645791e-01 -1.12703182e-01 1.98858127e-01
1.21690893e+00 -4.87604976e-01 -4.66565527e-02 -4.03444499e-01
-1.03900385e+00 3.67743552e-01 5.63621521e-01 4.41346198e-01
7.19234288e-01 -1.04022908e+00 -6.08640134e-01 1.64238969e-03
1.53605178e-01 2.39976332e-01 3.66855621e-01 1.38847852e+00
-3.48569214e-01 1.89175367e-01 -1.97116047e-01 -8.18729579e-01
-1.58392382e+00 7.28585243e-01 6.21626675e-01 -4.79765944e-02
-9.60163713e-01 9.14195657e-01 6.60423875e-01 2.78435975e-01
3.91270816e-01 -3.57122511e-01 -3.12382072e-01 2.68615484e-01
4.40268040e-01 7.20089450e-02 -2.34636739e-01 -1.25766981e+00
-4.64506865e-01 6.86428726e-01 -1.22989275e-01 -8.71950164e-02
1.25562239e+00 -4.11119670e-01 -8.90981704e-02 5.53269804e-01
1.43134975e+00 -1.51847795e-01 -1.68923879e+00 -3.00879538e-01
-2.43356839e-01 -5.28304458e-01 1.11030839e-01 -6.41784906e-01
-1.49866605e+00 9.77969587e-01 2.80615598e-01 -8.46722797e-02
1.19082832e+00 4.68397915e-01 7.58019269e-01 5.91540486e-02
8.88130367e-02 -9.92916584e-01 1.52234390e-01 2.73144066e-01
9.37162340e-01 -1.13801455e+00 -5.42954654e-02 -4.68209803e-01
-1.02784848e+00 1.19150424e+00 6.28901064e-01 2.93500032e-02
2.03562558e-01 -6.40479103e-02 3.77407223e-02 -2.30662003e-01
-7.97170162e-01 -4.14955199e-01 5.82834065e-01 5.10671377e-01
1.80255309e-01 -3.16050977e-01 -3.52113694e-02 2.14516044e-01
2.80101269e-01 -7.99443722e-02 4.17198330e-01 5.72963715e-01
-1.64374888e-01 -9.76372004e-01 -1.39300302e-01 2.22843274e-01
-6.12299144e-01 -4.48573306e-02 -3.68800521e-01 7.32479513e-01
3.75174403e-01 7.18584597e-01 3.79740417e-01 -1.41491964e-01
3.13333750e-01 1.29407585e-01 5.58526218e-01 -5.13512850e-01
-3.48319501e-01 3.12534690e-01 2.20489800e-01 -1.02727151e+00
-9.38984632e-01 -9.24604118e-01 -1.55237246e+00 8.55250359e-02
-1.00421615e-01 -1.55840605e-01 4.70209837e-01 1.18776953e+00
1.04369469e-01 6.37355268e-01 4.45225716e-01 -9.58927035e-01
-1.32275879e-01 -6.17065549e-01 -3.85556042e-01 8.86920393e-01
6.02530003e-01 -7.90924609e-01 -3.45091730e-01 4.86016542e-01] | [9.633481979370117, 0.5527945756912231] |
6394fe1e-83d2-4cd6-9607-392875b27a05 | uofl-at-semeval-2016-task-4-multi-domain | null | null | https://aclanthology.org/S16-1024 | https://aclanthology.org/S16-1024.pdf | UofL at SemEval-2016 Task 4: Multi Domain word2vec for Twitter Sentiment Classification | null | ['Adel Elmaghraby', 'Omar Abdelwahab'] | 2016-06-01 | null | null | null | semeval-2016-6 | ['twitter-sentiment-analysis'] | ['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] | [-7.290785312652588, 3.691213607788086] |
aea05bf2-bb68-460d-8c20-9623bd0fd5e6 | neural-software-analysis | 2011.07986 | null | https://arxiv.org/abs/2011.07986v2 | https://arxiv.org/pdf/2011.07986v2.pdf | Neural Software Analysis | Many software development problems can be addressed by program analysis tools, which traditionally are based on precise, logical reasoning and heuristics to ensure that the tools are practical. Recent work has shown tremendous success through an alternative way of creating developer tools, which we call neural software analysis. The key idea is to train a neural machine learning model on numerous code examples, which, once trained, makes predictions about previously unseen code. In contrast to traditional program analysis, neural software analysis naturally handles fuzzy information, such as coding conventions and natural language embedded in code, without relying on manually encoded heuristics. This article gives an overview of neural software analysis, discusses when to (not) use it, and presents three example analyses. The analyses address challenging software development problems: bug detection, type prediction, and code completion. The resulting tools complement and outperform traditional program analyses, and are used in industrial practice. | ['Satish Chandra', 'Michael Pradel'] | 2020-11-16 | null | null | null | null | ['type-prediction'] | ['computer-code'] | [ 1.75224945e-01 3.60826135e-01 -2.45970517e-01 -3.65469992e-01
-2.40136579e-01 -4.94418651e-01 -4.60182838e-02 5.34286559e-01
1.99038535e-01 3.09843779e-01 -3.16416383e-01 -7.14119196e-01
-3.96313779e-02 -7.78731883e-01 -7.01994538e-01 -6.71038777e-02
-1.00269832e-01 7.42773265e-02 1.52539417e-01 -1.70729637e-01
6.77601457e-01 2.24826988e-02 -1.91734135e+00 5.49498200e-01
1.09918761e+00 5.34044862e-01 -4.31626011e-03 8.48725200e-01
-5.58347821e-01 1.44936872e+00 -5.90852797e-01 -4.58198160e-01
-8.57918710e-02 -2.15647846e-01 -8.37076068e-01 -4.14838016e-01
5.86185902e-02 -1.28866717e-01 4.85039920e-01 1.58068275e+00
-3.10211629e-01 -1.51314408e-01 2.91628391e-01 -1.40375602e+00
-8.61028969e-01 9.30312514e-01 -4.22231048e-01 -1.92786846e-02
7.37718284e-01 -7.50665814e-02 9.09101605e-01 -7.21943676e-01
4.43109632e-01 8.84261370e-01 1.26414049e+00 6.92325115e-01
-1.32350540e+00 -3.62990916e-01 -2.04307273e-01 1.27759963e-01
-1.18565619e+00 -2.76469558e-01 8.97137225e-01 -1.16897142e+00
1.50553989e+00 4.07984816e-02 4.89435315e-01 6.73690736e-01
4.71939951e-01 6.54728770e-01 5.25997102e-01 -9.77290869e-01
4.45830882e-01 2.87328422e-01 6.55474901e-01 1.17380023e+00
2.25001797e-01 1.65021390e-01 -4.35220171e-03 -4.56921816e-01
3.11557859e-01 -7.35846069e-03 -2.48059079e-01 -3.45076710e-01
-8.91328990e-01 8.64747584e-01 1.54665068e-01 5.39055586e-01
-2.79988378e-01 3.17708760e-01 7.56125212e-01 5.90130568e-01
7.68630430e-02 9.67958152e-01 -6.24730885e-01 -6.03528738e-01
-1.12606359e+00 2.33139917e-01 1.17531705e+00 9.54204738e-01
9.59263861e-01 3.75174731e-01 3.66131812e-01 7.06222355e-01
4.45418954e-01 8.96683857e-02 6.48100019e-01 -1.04243767e+00
1.97912619e-01 1.24003506e+00 -1.79009482e-01 -1.19654703e+00
-2.18192846e-01 1.02782007e-02 -4.15167719e-01 8.38021338e-01
1.59926847e-01 -1.67817309e-01 -6.10810339e-01 1.33752310e+00
-1.24015711e-01 -2.27508366e-01 2.61205006e-02 2.27749780e-01
4.84273583e-01 5.03296614e-01 -1.99188143e-01 -1.24968647e-03
7.47541726e-01 -8.45119476e-01 -6.32046223e-01 -3.81238431e-01
1.02517509e+00 -4.12970424e-01 1.15631223e+00 9.70764697e-01
-1.17368603e+00 -4.64354724e-01 -1.22204828e+00 1.07313178e-01
-7.16494739e-01 6.38164729e-02 8.52924407e-01 9.39257443e-01
-1.38247645e+00 8.78218412e-01 -9.14555788e-01 -5.14186285e-02
2.45278314e-01 5.51232994e-01 -2.04459235e-01 5.79195507e-02
-7.63255119e-01 9.12240386e-01 7.06308842e-01 7.39341751e-02
-6.36626363e-01 -5.54903686e-01 -1.20193326e+00 2.50256807e-01
3.93243819e-01 -2.87569135e-01 1.62577033e+00 -1.64585328e+00
-1.23959529e+00 7.49004841e-01 4.75099646e-02 -4.53510433e-01
-1.71415657e-01 -1.29756391e-01 -2.54057825e-01 -3.86149049e-01
-1.08049318e-01 2.27597088e-01 6.87712848e-01 -1.06989062e+00
-5.40620089e-01 -5.60294203e-02 2.56005943e-01 -7.92961001e-01
-4.17919517e-01 4.53866333e-01 -4.04332876e-02 -5.17849922e-01
-2.70152569e-01 -7.27879226e-01 -2.98943251e-01 -2.02806979e-01
-3.55791301e-01 -2.64780283e-01 4.61961567e-01 -9.98920143e-01
1.61888385e+00 -2.01940536e+00 3.08116496e-01 5.02777576e-01
4.01617348e-01 1.39625415e-01 6.65062964e-02 1.77575573e-01
-3.45729709e-01 3.74314815e-01 -4.73295301e-01 5.03659844e-01
3.10872942e-01 6.31974917e-03 -1.65248245e-01 5.31907827e-02
3.96482050e-01 7.71207571e-01 -9.83964324e-01 -3.48685890e-01
1.35000214e-01 1.02354795e-01 -9.34349120e-01 1.63895234e-01
-7.29652524e-01 -3.63649994e-01 -3.31795901e-01 1.02029741e+00
2.65226454e-01 -1.96510568e-01 5.70129827e-02 3.98898154e-01
-1.55584008e-01 1.25678450e-01 -8.37931216e-01 1.40948856e+00
-7.39480078e-01 9.93828714e-01 -7.58031234e-02 -1.03486276e+00
1.17609870e+00 2.54263401e-01 -6.01134636e-02 -1.81150764e-01
1.10270873e-01 2.35129014e-01 1.52997375e-01 -1.05162954e+00
3.28365117e-01 1.39876500e-01 -3.25194746e-01 5.55190802e-01
6.58393875e-02 -2.03199685e-01 3.44360828e-01 -1.43191949e-01
1.49208486e+00 3.82922143e-01 7.47987032e-01 -2.58727193e-01
5.83360076e-01 3.13936055e-01 6.72782779e-01 7.53817439e-01
-6.08300790e-02 1.94360211e-01 1.11358750e+00 -9.02300060e-01
-1.04254103e+00 -6.44620001e-01 3.02656949e-01 1.13324082e+00
-4.34652507e-01 -7.85208821e-01 -1.17053795e+00 -9.81773794e-01
-1.52050883e-01 1.11249626e+00 -8.10732126e-01 -4.21346277e-01
-7.20433116e-01 -1.87080771e-01 6.74484134e-01 8.32428217e-01
-9.60948244e-02 -1.46769166e+00 -9.67706025e-01 3.74176443e-01
2.45281354e-01 -3.61733556e-01 9.31366980e-02 5.75681329e-01
-8.12199593e-01 -1.23463237e+00 -3.36692855e-02 -9.85856056e-01
6.76319003e-01 -3.53697002e-01 1.55496097e+00 9.23737943e-01
-3.19428533e-01 1.39373064e-01 -4.32497323e-01 -4.14110869e-01
-1.00770712e+00 1.32902674e-02 -1.78005800e-01 -6.73939347e-01
9.45477664e-01 -5.82667649e-01 4.06978548e-01 -1.34275839e-01
-7.97417402e-01 -1.42127275e-01 6.00667119e-01 1.10736942e+00
6.88839257e-02 5.06059527e-01 4.93197173e-01 -1.10096395e+00
8.11199188e-01 -6.02163315e-01 -9.59036589e-01 6.69253647e-01
-6.98475420e-01 4.12631005e-01 9.60135102e-01 -2.00125441e-01
-1.11696434e+00 3.59141797e-01 -2.44529784e-01 -2.01958477e-01
-3.39536250e-01 1.12439787e+00 -1.29900677e-02 -3.18592429e-01
1.13999701e+00 -2.48013921e-02 -4.26584706e-02 -2.31768534e-01
5.86165395e-03 8.03933024e-01 4.71544772e-01 -8.44121754e-01
7.47061670e-01 -3.50240082e-01 -4.76614296e-01 -3.94853324e-01
-2.64771074e-01 1.09356672e-01 -6.22410834e-01 -2.52977103e-01
6.49270892e-01 -2.30752423e-01 -5.65495253e-01 1.41024664e-01
-1.36694026e+00 -5.02811134e-01 -9.52919796e-02 -4.60906923e-02
-6.18542671e-01 2.81721026e-01 -5.41687846e-01 -1.01133585e+00
-1.09483473e-01 -1.38717353e+00 5.61564922e-01 2.14731514e-01
-8.66571963e-01 -1.02614224e+00 1.98531285e-01 -2.05629338e-02
4.92159754e-01 5.39609551e-01 1.61892962e+00 -7.10415959e-01
-4.23292875e-01 -5.46252072e-01 8.39741826e-02 6.07487500e-01
1.80544138e-01 7.95103848e-01 -7.74257779e-01 2.00836048e-01
1.87337190e-01 -3.62149954e-01 2.45746747e-01 2.09448606e-01
1.36714935e+00 -4.07920629e-01 -2.56757647e-01 3.58368367e-01
1.30170786e+00 5.88341236e-01 5.37559390e-01 5.08180797e-01
5.00939369e-01 9.22643721e-01 4.53346133e-01 2.57731974e-01
1.40176609e-01 3.49447817e-01 6.23415291e-01 3.40808988e-01
4.96517211e-01 7.43119717e-02 5.41072071e-01 7.88071632e-01
-1.00658759e-01 3.03650558e-01 -1.73553431e+00 5.91395497e-01
-1.96519411e+00 -8.06300342e-01 -1.49611486e-02 1.91806149e+00
9.73135471e-01 4.18482035e-01 -4.25196625e-02 4.82525766e-01
6.78527117e-01 -4.98766035e-01 -3.11625987e-01 -1.05814028e+00
4.49421197e-01 3.11616510e-01 -6.92719743e-02 3.62829357e-01
-9.65224087e-01 6.79898262e-01 6.73678970e+00 4.16151047e-01
-9.02734160e-01 -2.58230031e-01 1.67870715e-01 2.65988052e-01
-4.35104489e-01 5.03928401e-02 -2.04643592e-01 2.70304739e-01
1.16192031e+00 -3.58645797e-01 8.28865349e-01 1.74002993e+00
-2.30882898e-01 4.56961617e-02 -1.62538230e+00 4.91622537e-01
1.26845002e-01 -1.53153861e+00 -4.00274247e-01 -4.69193697e-01
4.74596441e-01 -2.89082527e-01 -2.12605923e-01 7.00827718e-01
5.06242752e-01 -1.20584548e+00 6.66442871e-01 5.68438947e-01
3.54872525e-01 -8.73178065e-01 1.07369518e+00 4.44056273e-01
-6.81623280e-01 -5.36447585e-01 -2.18015119e-01 -3.48970056e-01
-4.24673796e-01 5.36479354e-01 -8.36959779e-01 8.67858380e-02
1.00516891e+00 8.00459683e-01 -9.09836531e-01 1.02707720e+00
-4.14919913e-01 4.42366511e-01 1.28457118e-02 -4.09258753e-01
-1.19319528e-01 2.06213102e-01 1.87933818e-01 1.38110352e+00
4.01149869e-01 -4.56305653e-01 -7.55888969e-02 1.51207769e+00
3.27506512e-01 -2.68925697e-01 -9.32519436e-01 -3.36861372e-01
3.75893712e-01 8.78275454e-01 -6.65823042e-01 -2.15150520e-01
-7.27206647e-01 3.20714861e-01 4.42226171e-01 1.53676540e-01
-6.82730556e-01 -1.05033469e+00 3.64718944e-01 -6.84553236e-02
1.84206560e-01 1.54669762e-01 -6.23254061e-01 -9.37207401e-01
4.54781055e-01 -1.41329718e+00 2.26081550e-01 -8.35324407e-01
-8.75494719e-01 7.49969304e-01 -1.85882717e-01 -9.63936150e-01
-7.41289496e-01 -8.79572451e-01 -9.03052807e-01 5.84735990e-01
-1.02807033e+00 -7.61719286e-01 -1.78239375e-01 -4.72389199e-02
5.04949272e-01 -5.96021473e-01 9.85938489e-01 1.54255629e-01
-5.54146230e-01 5.71462929e-01 -2.22311899e-01 1.24754451e-01
5.00894226e-02 -1.54063368e+00 5.46021461e-01 1.00014102e+00
-1.61981374e-01 1.15372884e+00 8.14022005e-01 -7.59387076e-01
-1.52476645e+00 -8.96790504e-01 8.90353799e-01 -5.53443611e-01
9.16801512e-01 -2.95552105e-01 -1.36314225e+00 8.37629259e-01
1.95794314e-01 -1.65655345e-01 8.02840590e-01 3.12853009e-01
-6.53156817e-01 2.69632459e-01 -1.26099133e+00 4.09963787e-01
4.73894924e-01 -6.71417058e-01 -1.09450996e+00 2.18925864e-01
6.42295122e-01 -2.72072822e-01 -9.13043618e-01 2.48327121e-01
4.90421802e-01 -1.29346931e+00 6.46285832e-01 -7.82759130e-01
8.73045266e-01 -4.39382851e-01 4.16159146e-02 -1.21910334e+00
-3.60279351e-01 -6.31013095e-01 -4.51054990e-01 1.16427684e+00
5.47800362e-01 -2.39890710e-01 8.68838012e-01 9.87744331e-01
-3.28316748e-01 -7.01588333e-01 -4.54487771e-01 -5.55354595e-01
5.56103662e-02 -5.97748399e-01 7.77907908e-01 1.19779634e+00
7.27162063e-01 -1.39210522e-01 6.88012764e-02 -3.41237783e-02
2.80847490e-01 9.88535136e-02 5.31597614e-01 -1.59104776e+00
-6.89389884e-01 -9.39636946e-01 -6.08586788e-01 -2.08060846e-01
6.83305204e-01 -7.95978606e-01 5.78252494e-01 -1.04316330e+00
-4.08859253e-02 -1.44622266e-01 -2.35090867e-01 9.35497820e-01
-1.67741813e-03 -2.79539943e-01 -3.79141331e-01 -2.01041698e-01
-4.64958221e-01 -7.43992478e-02 3.13911051e-01 -3.99826229e-01
-2.76906997e-01 2.47520003e-02 -7.99747467e-01 1.21740675e+00
7.16990054e-01 -6.12232625e-01 -1.96961656e-01 -4.75357711e-01
1.12329578e+00 1.47849798e-01 2.34011114e-01 -1.07402956e+00
3.35660636e-01 -2.44841412e-01 1.47644784e-02 -7.60475397e-02
-4.69213188e-01 -9.45562840e-01 2.03045071e-04 6.34309590e-01
-5.45183659e-01 3.77069890e-01 3.50120068e-01 1.71979144e-01
-4.30372208e-01 -1.06049669e+00 5.75040936e-01 -2.35487297e-01
-1.02471244e+00 -3.71285230e-01 -7.30968237e-01 -1.61803558e-01
1.07280064e+00 -4.20750856e-01 -1.35197282e-01 9.08377990e-02
-7.31348395e-01 4.90180850e-02 8.50862503e-01 5.44237971e-01
6.75974131e-01 -1.12087595e+00 -2.34664679e-01 5.38616478e-01
4.34573770e-01 -7.31045380e-02 -1.22056425e-01 5.18059194e-01
-9.25010204e-01 7.34862834e-02 -4.43069816e-01 -4.48834956e-01
-1.17789054e+00 1.01157355e+00 4.72441345e-01 3.17783989e-02
-5.35756052e-01 7.67963409e-01 -2.34066591e-01 -6.91583574e-01
3.26443911e-01 -7.31862545e-01 -2.52161145e-01 -4.29443657e-01
7.39282787e-01 1.58838004e-01 2.76296079e-01 6.63238540e-02
-5.01061201e-01 4.17615473e-01 -1.23826563e-01 4.33755308e-01
1.59436929e+00 5.92616558e-01 -6.94399059e-01 6.08797133e-01
7.47547746e-01 -2.14702502e-01 -8.27773452e-01 1.48965955e-01
8.83628726e-01 -2.86698878e-01 4.79821898e-02 -8.94520044e-01
-8.85845900e-01 1.02675629e+00 1.19819932e-01 8.48273695e-01
1.07783496e+00 -3.05170447e-01 3.14802945e-01 1.06102931e+00
3.38891029e-01 -1.16232038e+00 -1.50162369e-01 6.61985278e-01
7.50521898e-01 -1.20314729e+00 -3.18938017e-01 -2.02408701e-01
-2.65705734e-01 1.75724888e+00 9.89845335e-01 -3.29554081e-01
4.66763586e-01 1.06293046e+00 -9.95045155e-02 -3.17180127e-01
-7.39645064e-01 5.52156866e-01 8.71960372e-02 9.51701939e-01
7.75631666e-01 -2.67365694e-01 4.09756511e-01 1.03709793e+00
-3.01409364e-01 4.34613645e-01 9.01487172e-01 1.37216473e+00
-5.54405272e-01 -1.08347499e+00 -4.99358773e-01 6.94899201e-01
-5.34093380e-01 -2.20502332e-01 -6.61052227e-01 7.87829220e-01
1.84798241e-01 6.81559682e-01 -2.68685937e-01 -7.38181651e-01
1.15990192e-01 3.23684990e-01 4.46636736e-01 -1.00784135e+00
-1.09855545e+00 -5.66409707e-01 1.38283581e-01 -5.95668614e-01
-2.00654324e-02 -4.73197907e-01 -1.18012631e+00 -1.25225857e-01
-3.85528654e-01 3.45030069e-01 5.89560211e-01 8.68153870e-01
1.31540671e-01 7.59264588e-01 1.03617504e-01 -8.07012498e-01
-4.50040698e-01 -4.76608872e-01 -1.05124325e-01 6.62781373e-02
6.19461060e-01 -5.96428871e-01 -3.59358311e-01 4.67784762e-01] | [7.672738552093506, 7.706411838531494] |
0ef54f9f-9229-4358-8a54-930935af16d3 | verifai-verified-generative-ai | 2307.02796 | null | https://arxiv.org/abs/2307.02796v1 | https://arxiv.org/pdf/2307.02796v1.pdf | VerifAI: Verified Generative AI | Generative AI has made significant strides, yet concerns about the accuracy and reliability of its outputs continue to grow. Such inaccuracies can have serious consequences such as inaccurate decision-making, the spread of false information, privacy violations, legal liabilities, and more. Although efforts to address these risks are underway, including explainable AI and responsible AI practices such as transparency, privacy protection, bias mitigation, and social and environmental responsibility, misinformation caused by generative AI will remain a significant challenge. We propose that verifying the outputs of generative AI from a data management perspective is an emerging issue for generative AI. This involves analyzing the underlying data from multi-modal data lakes, including text files, tables, and knowledge graphs, and assessing its quality and consistency. By doing so, we can establish a stronger foundation for evaluating the outputs of generative AI models. Such an approach can ensure the correctness of generative AI, promote transparency, and enable decision-making with greater confidence. Our vision is to promote the development of verifiable generative AI and contribute to a more trustworthy and responsible use of AI. | ['Lei Cao', 'Ju Fan', 'Chenyu Yang', 'Nan Tang'] | 2023-07-06 | null | null | null | null | ['knowledge-graphs', 'misinformation', 'management', 'decision-making'] | ['knowledge-base', 'miscellaneous', 'miscellaneous', 'reasoning'] | [ 3.49537134e-01 6.50842369e-01 3.37389484e-02 -3.83731723e-01
-7.75208056e-01 -9.21601176e-01 6.15441442e-01 4.66833860e-01
-8.73525888e-02 8.38501811e-01 4.62026983e-01 -4.82731551e-01
-1.02833256e-01 -1.11795723e+00 -8.29611957e-01 -4.69419241e-01
3.64489287e-01 3.33550662e-01 -2.42383808e-01 2.14420930e-01
3.47213745e-01 3.68614852e-01 -1.25768435e+00 2.13788942e-01
1.13494575e+00 8.74360681e-01 -7.39872992e-01 3.52840811e-01
-4.20658439e-02 1.23056018e+00 -7.95632958e-01 -1.01989794e+00
2.30150923e-01 -3.59854370e-01 -6.23128831e-01 -1.72735244e-01
2.42832929e-01 -6.87819660e-01 3.27995270e-01 1.45821071e+00
8.49888399e-02 -6.64518893e-01 4.24135327e-01 -1.69856954e+00
-1.03390372e+00 1.08330691e+00 -2.14208588e-01 -2.61218697e-01
-1.48431659e-02 6.91435695e-01 9.21188474e-01 -3.25259209e-01
6.76978946e-01 1.24274278e+00 5.96787393e-01 1.90312982e-01
-1.23752260e+00 -1.14377213e+00 1.02833435e-02 -1.18470177e-01
-1.21008158e+00 -6.43595219e-01 3.70016366e-01 -7.43945599e-01
3.66074860e-01 5.55887878e-01 9.70199347e-01 1.06446493e+00
3.95847827e-01 4.86745149e-01 9.36582029e-01 -1.45395234e-01
4.47514564e-01 4.29262489e-01 -1.02251843e-02 6.07228160e-01
1.41242266e+00 1.47260755e-01 -6.85063064e-01 -3.66090566e-01
3.76561671e-01 1.04403943e-01 -2.03857183e-01 -2.13055298e-01
-1.20439029e+00 6.74874067e-01 3.39495450e-01 1.28416210e-01
-5.57732224e-01 4.13317472e-01 -1.31861782e-02 3.48455757e-02
5.73148549e-01 7.97359645e-01 -6.73142821e-02 -2.58479387e-01
-7.23557293e-01 4.15700078e-01 7.95789421e-01 8.15051079e-01
5.72396934e-01 1.61839038e-01 1.32933155e-01 -5.89466095e-02
7.07000852e-01 1.05576909e+00 -1.55717924e-01 -1.22436440e+00
3.59344870e-01 1.11219108e+00 3.02493870e-01 -1.44774377e+00
4.82557192e-02 -3.59690428e-01 -5.67022324e-01 5.23517251e-01
3.60615760e-01 -2.25774541e-01 -8.95054519e-01 1.63193905e+00
2.13439494e-01 -4.82670397e-01 2.24325761e-01 9.31603432e-01
4.09514934e-01 5.93584776e-01 2.21825197e-01 1.98745489e-01
8.82014632e-01 3.88854258e-02 -9.06102598e-01 -2.98663974e-01
6.01723850e-01 -3.53589177e-01 7.12317050e-01 2.83031791e-01
-1.10475707e+00 1.74981937e-01 -1.01191318e+00 -7.43768588e-02
-2.32457891e-01 -3.50811273e-01 6.69657528e-01 7.69796252e-01
-5.92856646e-01 3.20945114e-01 -8.97224665e-01 -8.93779024e-02
9.72644806e-01 -4.84572016e-02 -3.74145269e-01 -1.11073099e-01
-9.13560033e-01 1.01463246e+00 1.11978322e-01 5.00513434e-01
-6.74303889e-01 -8.38359237e-01 -7.64201701e-01 2.22418576e-01
4.69236761e-01 -5.54457009e-01 1.02755499e+00 -9.68556941e-01
-5.44001043e-01 6.35382116e-01 2.26672500e-01 -5.46048522e-01
8.05250347e-01 -2.67998785e-01 -2.20376119e-01 -4.77183342e-01
1.52931914e-01 2.98559248e-01 3.12906206e-01 -1.43453181e+00
-6.44962132e-01 -7.62527704e-01 -4.75627571e-01 -9.83151272e-02
-9.17839184e-02 -1.95425972e-01 4.74235773e-01 -2.27236047e-01
1.24974944e-01 -9.58495796e-01 -5.51160537e-02 3.07867557e-01
-6.22792542e-01 3.18775356e-01 4.88016129e-01 -7.40191340e-01
1.08199084e+00 -2.24294639e+00 -3.23509037e-01 2.98126638e-01
6.16675317e-01 2.85881013e-01 3.85632575e-01 3.65483016e-01
6.57688975e-01 8.98218095e-01 -3.36389542e-01 1.97333753e-01
1.07241817e-01 3.70734483e-02 -5.29190600e-01 4.47531223e-01
5.87828040e-01 9.67082381e-01 -7.84448504e-01 -1.20464630e-01
-1.76311657e-02 4.22837615e-01 -2.71353871e-01 -1.51035683e-02
-2.02592179e-01 4.52450775e-02 -4.93563861e-01 8.01035404e-01
4.42623228e-01 -4.40994889e-01 3.03440541e-01 9.51043665e-02
-3.04872066e-01 4.26523775e-01 -9.83685613e-01 6.26632631e-01
1.15467990e-02 9.50539112e-01 3.15161571e-02 -2.37421259e-01
1.06450737e+00 1.62512332e-01 -2.42857845e-03 -7.98818231e-01
1.81117848e-01 3.07863683e-01 2.93218821e-01 -3.30370933e-01
7.12639511e-01 5.03601357e-02 -4.66230251e-02 9.62307751e-01
-6.23911917e-01 -3.88138264e-01 -1.97869703e-01 3.92080903e-01
8.83002162e-01 6.93273451e-03 2.00669557e-01 -9.49511304e-02
-1.92682073e-01 6.63871050e-01 7.29353666e-01 4.79626864e-01
-1.62625432e-01 3.85453999e-01 5.25147438e-01 -4.97440994e-01
-1.12094700e+00 -9.13039267e-01 8.96042064e-02 1.62745714e-01
-1.03640936e-01 -2.08210468e-01 -5.41720569e-01 -4.82417196e-01
5.72249293e-01 1.39628804e+00 -6.49413586e-01 -5.46389341e-01
-4.28846776e-02 -4.67149764e-01 5.89601219e-01 5.68676233e-01
4.54412639e-01 -8.18442822e-01 -1.06020904e+00 1.71246678e-01
-1.88525617e-01 -8.13472807e-01 3.34830396e-02 -2.18062416e-01
-5.09686887e-01 -1.24877405e+00 -9.04477984e-02 9.86725911e-02
9.71980393e-01 7.08049983e-02 9.75338519e-01 2.07832351e-01
-6.94413185e-02 1.09901987e-01 -1.01310663e-01 -1.03378570e+00
-1.17832410e+00 -3.74119490e-01 -3.05457532e-01 -3.18582535e-01
4.22202647e-01 -3.46599549e-01 -3.71893466e-01 1.21491596e-01
-9.93278325e-01 1.94546536e-01 6.56725645e-01 2.80593991e-01
3.50728691e-01 -1.98210433e-01 5.12840807e-01 -1.32695162e+00
7.96529830e-01 -5.19012153e-01 -8.36580575e-01 5.24704397e-01
-1.43072438e+00 1.52734984e-02 3.62773329e-01 -1.07801631e-01
-9.62378681e-01 -3.04286182e-01 5.67668915e-01 1.29290879e-01
-3.01838480e-02 6.82266712e-01 -1.40422806e-01 1.65064916e-01
8.79244447e-01 -1.47531107e-01 1.91175207e-01 -1.40011311e-01
4.57717657e-01 7.84821391e-01 4.73193228e-01 -1.42348498e-01
8.54682624e-01 5.29572189e-01 -1.92932263e-01 -4.78809178e-01
-5.27023852e-01 2.88760394e-01 4.94291931e-02 -3.49929810e-01
4.42573667e-01 -8.29784930e-01 -6.08591318e-01 2.93347120e-01
-8.78666401e-01 -1.63711578e-01 -4.26874489e-01 3.11509699e-01
5.09453565e-02 3.14640589e-02 -1.54915482e-01 -1.09543908e+00
-4.95262176e-01 -1.01355243e+00 4.51734334e-01 1.26437277e-01
-7.78791189e-01 -4.32085603e-01 1.73226595e-02 5.86212516e-01
6.58041894e-01 5.92438579e-01 9.65825319e-01 -6.66490793e-01
-1.02955556e+00 -5.30960202e-01 -3.07444304e-01 3.74716073e-01
2.81868309e-01 7.68390298e-01 -8.54804039e-01 2.34500661e-01
-1.84467018e-01 -4.05726641e-01 1.98254451e-01 4.14167978e-02
4.64971423e-01 -9.71255362e-01 -1.46167353e-01 1.37649000e-01
1.11562014e+00 4.89744574e-01 5.27176023e-01 1.94163322e-01
5.49460948e-01 8.61196876e-01 4.73801553e-01 3.53792578e-01
7.69514084e-01 -8.97429138e-02 4.65502232e-01 8.57516304e-02
1.19651340e-01 -4.07415807e-01 6.89083114e-02 2.62814969e-01
1.11624189e-01 -3.91555995e-01 -1.40688872e+00 6.69070721e-01
-1.85782278e+00 -1.05558515e+00 -2.06415370e-01 2.20349836e+00
7.33911991e-01 1.47895485e-01 -9.84541103e-02 3.13676596e-01
4.69238430e-01 -2.50711650e-01 -7.73667693e-01 -4.72267032e-01
-5.44526428e-02 -5.35385370e-01 5.43294370e-01 1.69089675e-01
-4.32334125e-01 6.81110442e-01 6.21392870e+00 -2.32856616e-01
-1.06519866e+00 -3.28479439e-01 9.05954242e-01 2.73734964e-02
-1.15127575e+00 1.44740030e-01 -2.84989357e-01 4.24913794e-01
8.77587557e-01 -7.19087780e-01 4.40298080e-01 8.00435185e-01
2.22981095e-01 -3.55428457e-01 -9.77869034e-01 4.37447459e-01
-3.33173573e-02 -1.61998618e+00 3.13029587e-02 3.83704185e-01
6.51613533e-01 -2.18618944e-01 6.10724166e-02 -3.34013671e-01
1.09909391e+00 -1.23548853e+00 1.23283696e+00 6.30250156e-01
4.56338614e-01 -8.61788690e-01 7.85629570e-01 3.17447841e-01
-3.87947619e-01 1.40785594e-02 -5.12603857e-02 -3.43508303e-01
-3.72230746e-02 8.17257822e-01 -9.86180007e-01 1.54818699e-01
5.17970324e-01 4.73617077e-01 -5.35330951e-01 7.98120081e-01
-3.35990936e-01 7.76774526e-01 -1.39363334e-01 -2.15838462e-01
-1.72739685e-01 -7.08074793e-02 3.01411361e-01 6.88377619e-01
1.49147153e-01 5.92258722e-02 -4.61913466e-01 1.49050677e+00
-1.82144970e-01 -3.51082534e-01 -8.91436160e-01 -1.07389522e+00
7.48834729e-01 7.79897749e-01 -5.69957674e-01 -1.45184994e-01
-1.08131997e-01 2.15068340e-01 9.59238559e-02 2.05949545e-01
-3.66644025e-01 1.13422409e-01 7.33687460e-01 5.78968562e-02
-2.90542364e-01 -1.09308310e-01 -1.01424623e+00 -8.34544539e-01
1.28266290e-01 -1.24976242e+00 1.67676851e-01 -7.13893235e-01
-9.99330580e-01 4.25584376e-01 -2.13991836e-01 -7.48164117e-01
-5.20273089e-01 -1.22037232e-02 -1.38922751e-01 7.64698327e-01
-1.05960584e+00 -1.10587931e+00 -3.83665472e-01 -1.78094491e-01
-3.19852948e-01 1.50928706e-01 7.47755826e-01 -1.83508784e-01
-3.51850122e-01 2.25904197e-01 -4.80430499e-02 7.13435337e-02
4.97455150e-01 -8.22775781e-01 7.51349449e-01 1.09566498e+00
2.67391890e-01 7.42343128e-01 6.08084857e-01 -1.07947695e+00
-1.63682830e+00 -1.03818333e+00 9.12541687e-01 -8.69896412e-01
7.18702912e-01 -2.28418007e-01 -9.45397377e-01 9.56929326e-01
-2.82782823e-01 -4.51055110e-01 7.66853929e-01 4.61015478e-02
-5.07013738e-01 -1.64575338e-01 -1.33399558e+00 5.66962719e-01
8.65160823e-01 -5.12633801e-01 -3.95681560e-01 4.51265350e-02
6.67294741e-01 -7.97326714e-02 -7.22615421e-01 5.59168011e-02
7.80600190e-01 -9.99888301e-01 3.86469781e-01 -5.61483264e-01
5.66588461e-01 -4.71318483e-01 -1.16874419e-01 -1.08211541e+00
-2.60615706e-01 -4.15274829e-01 2.62420416e-01 1.41961503e+00
8.83884311e-01 -8.56774330e-01 5.51518381e-01 1.73589563e+00
1.38242140e-01 -2.07834736e-01 -4.53803003e-01 -4.03002948e-01
-4.18130420e-02 -6.25753760e-01 1.13119352e+00 1.08183682e+00
-3.96731384e-02 -1.52316257e-01 -2.87142098e-01 4.85634089e-01
8.44962180e-01 2.17778936e-01 1.04245126e+00 -1.31784463e+00
4.12979335e-01 -2.71641880e-01 -6.04418337e-01 1.00118332e-01
-4.29013461e-01 -4.33443457e-01 3.80724929e-02 -1.87666285e+00
2.73256600e-01 -5.94279945e-01 1.93117216e-01 7.72333860e-01
-2.32934758e-01 1.66900799e-01 4.43854928e-01 3.47792119e-01
-2.27490261e-01 2.47785598e-01 9.83934045e-01 -1.78299502e-01
5.95904738e-02 -3.64482790e-01 -1.37482750e+00 6.18072391e-01
5.88767052e-01 -8.54234695e-01 -2.98611164e-01 -6.52847767e-01
7.55480766e-01 -2.47246921e-01 6.92788363e-01 -7.17553973e-01
3.27116817e-01 -5.47266006e-01 2.91838735e-01 -3.43602985e-01
-1.15796976e-01 -1.11556160e+00 8.62246275e-01 6.28078103e-01
-5.69598496e-01 -1.48297530e-02 1.14311136e-01 3.23075801e-01
3.61493006e-02 2.56787717e-01 2.51294583e-01 -6.31391304e-03
-1.15336098e-01 -9.98868048e-02 -2.72590488e-01 1.31548136e-01
1.04396737e+00 -8.76360685e-02 -8.99419487e-01 -5.32270849e-01
-8.18279013e-02 4.51787412e-01 1.07328176e+00 3.16288143e-01
5.89857697e-01 -8.52938533e-01 -9.28168297e-01 3.27452511e-01
2.49818981e-01 2.92263359e-01 -1.30326897e-01 3.77847880e-01
-4.39812720e-01 1.14035182e-01 -1.07531019e-01 -2.33192533e-01
-9.57822740e-01 9.04607102e-02 3.71220112e-02 3.60293835e-01
-6.36583507e-01 5.20544946e-01 -1.89795539e-01 -4.18728404e-03
2.92669296e-01 -4.58593220e-01 3.59503955e-01 -1.72313049e-01
6.88305914e-01 4.94844586e-01 -1.79166466e-01 -4.18935835e-01
-4.95195359e-01 -1.69140369e-01 6.22604713e-02 -2.01921821e-01
1.44998944e+00 -5.38005307e-02 -2.39292845e-01 5.59006870e-01
1.97549060e-01 3.01334918e-01 -1.31250322e+00 1.79557636e-01
1.06452875e-01 -8.47553670e-01 1.65440872e-01 -1.52311897e+00
-1.01682365e+00 7.54704952e-01 2.13583604e-01 5.79000473e-01
5.78584015e-01 -1.70940518e-01 3.64582390e-01 2.65369356e-01
4.51151401e-01 -5.54798484e-01 -4.57883030e-01 -4.33976918e-01
9.07479346e-01 -1.32006812e+00 3.02923888e-01 -1.60735264e-01
-9.17835116e-01 8.38799179e-01 4.69351262e-01 4.87554759e-01
1.49799511e-01 3.96481276e-01 3.66161942e-01 -2.80639321e-01
-8.08867276e-01 1.91387579e-01 5.74823059e-02 5.96141219e-01
3.85258317e-01 5.52447736e-01 -1.93347242e-02 5.15620291e-01
-2.58741379e-01 4.11627024e-01 7.23383248e-01 9.18737531e-01
-2.33380437e-01 -5.99675655e-01 -5.17914116e-01 5.37616789e-01
-3.07086676e-01 -1.28383815e-01 -1.07901144e+00 8.12419891e-01
-6.66205734e-02 1.10988808e+00 -2.58128028e-02 -4.01662946e-01
8.91468599e-02 2.72972658e-02 -6.46296814e-02 -3.06157231e-01
-5.49247384e-01 -4.58973885e-01 5.29968381e-01 -5.85352063e-01
-1.79890007e-01 -8.33145201e-01 -1.12129939e+00 -9.58256841e-01
-2.52235174e-01 9.25846100e-02 1.06504345e+00 8.24907303e-01
9.27807808e-01 1.92655474e-01 1.48506030e-01 1.49820358e-01
-7.35346198e-01 -4.96516049e-01 -2.17914850e-01 3.36464554e-01
3.15901697e-01 -2.29792342e-01 -4.56311941e-01 -2.04647426e-02] | [8.850357055664062, 6.112376689910889] |
b8bef4a4-4c86-40e9-9d6d-64c8603a8877 | lung-nodules-detection-and-segmentation-using | 1907.07676 | null | https://arxiv.org/abs/1907.07676v1 | https://arxiv.org/pdf/1907.07676v1.pdf | Lung Nodules Detection and Segmentation Using 3D Mask-RCNN | Accurate assessment of Lung nodules is a time consuming and error prone ingredient of the radiologist interpretation work. Automating 3D volume detection and segmentation can improve workflow as well as patient care. Previous works have focused either on detecting lung nodules from a full CT scan or on segmenting them from a small ROI. We adapt the state of the art architecture for 2D object detection and segmentation, MaskRCNN, to handle 3D images and employ it to detect and segment lung nodules from CT scans. We report on competitive results for the lung nodule detection on LUNA16 data set. The added value of our method is that in addition to lung nodule detection, our framework produces 3D segmentations of the detected nodules. | ['Guy Engelhard', 'Evi Kopelowitz'] | 2019-07-17 | null | null | null | null | ['lung-nodule-detection'] | ['medical'] | [ 1.93835035e-01 4.49099123e-01 -9.94535610e-02 -1.67536363e-01
-7.65583038e-01 -6.89651370e-01 2.64072955e-01 6.30985349e-02
-2.96891004e-01 -4.20860127e-02 -3.24195586e-02 -8.35412741e-01
1.21904649e-01 -6.59664750e-01 -3.37062001e-01 -2.90348321e-01
-2.51982007e-02 1.19972360e+00 1.12341177e+00 1.93646222e-01
-1.59712106e-01 1.04228652e+00 -7.94555366e-01 4.32740420e-01
4.18048024e-01 7.76935935e-01 2.68849730e-01 1.15210760e+00
-2.63273269e-01 6.35168016e-01 -1.79697052e-01 2.32286919e-02
6.80138588e-01 -6.32796288e-01 -9.92859960e-01 4.25283372e-01
4.74235535e-01 -5.95719576e-01 -9.89178047e-02 7.03275561e-01
4.38769817e-01 -6.07115209e-01 9.10175800e-01 -5.89018524e-01
1.83910862e-01 7.55064607e-01 -4.55305129e-01 5.37882924e-01
-2.17390209e-01 2.09821984e-01 5.82000434e-01 -9.69735980e-01
3.13051671e-01 8.00779283e-01 8.75132799e-01 7.22276330e-01
-8.31142902e-01 -3.44875485e-01 -4.45742995e-01 -3.69461745e-01
-1.27182007e+00 1.78544354e-02 5.33207990e-02 -6.00115538e-01
7.69891918e-01 7.09214330e-01 1.00332093e+00 2.52562374e-01
3.15184027e-01 7.37610996e-01 8.64592671e-01 -4.25645798e-01
-2.22819168e-02 1.16224371e-01 4.55938578e-02 1.12381303e+00
8.40551794e-01 2.54461262e-02 4.92417544e-01 -2.15308174e-01
1.34830558e+00 2.12634176e-01 -1.29414335e-01 -6.50058985e-01
-1.52910316e+00 7.52441287e-01 6.52484000e-01 4.19052303e-01
-6.06684268e-01 4.61934954e-01 3.63978356e-01 -3.15896302e-01
1.88186169e-01 3.13362062e-01 -1.09222688e-01 2.77021378e-01
-1.11558115e+00 -9.22482535e-02 9.25058007e-01 7.15233088e-01
8.73655975e-02 -3.35985929e-01 -8.68690252e-01 4.10957664e-01
4.92452949e-01 4.87841994e-01 5.18617809e-01 -6.07263446e-01
1.12700790e-01 9.15871620e-01 -1.53613567e-01 -7.94011280e-02
-7.48689950e-01 -3.18887472e-01 -5.68372250e-01 3.28488708e-01
4.63129073e-01 8.26936290e-02 -1.68844044e+00 5.84918201e-01
4.85571653e-01 -7.29501918e-02 -4.31986362e-01 1.02906668e+00
1.33011961e+00 -6.97973594e-02 9.70006287e-02 9.39801782e-02
1.72317839e+00 -1.10885727e+00 -1.58388689e-01 -2.62371391e-01
7.27282643e-01 -8.52445900e-01 7.06085503e-01 -1.98444933e-01
-1.34253705e+00 -3.49380523e-01 -8.10575426e-01 6.95665702e-02
1.82346612e-01 4.50816154e-01 3.40993375e-01 1.01092303e+00
-8.98694277e-01 2.65932888e-01 -1.36470282e+00 -4.66862917e-01
8.93322051e-01 7.83200443e-01 -4.20517810e-02 1.06370628e-01
-5.19447565e-01 1.14799392e+00 6.71542048e-01 -1.18495598e-01
-1.00854182e+00 -8.93285453e-01 -3.63020360e-01 5.06831557e-02
8.31422567e-01 -9.61759388e-01 1.83114159e+00 -2.74888813e-01
-9.89450216e-01 1.29832923e+00 2.44520470e-01 -8.01090240e-01
9.26809669e-01 1.63044006e-01 1.86949566e-01 3.55150521e-01
-1.56493694e-01 6.83089793e-01 6.98588014e-01 -8.93147528e-01
-7.69427299e-01 -1.28376409e-01 -4.98326749e-01 2.01895699e-01
3.69138658e-01 -1.99045569e-01 -6.21718168e-01 -2.33630121e-01
3.99754345e-01 -1.07103467e+00 -8.31857324e-01 2.45911226e-01
-6.16359055e-01 -3.27515975e-02 1.03927183e+00 -5.10797918e-01
1.00956178e+00 -1.64692211e+00 -4.56602424e-01 4.86857891e-01
7.48178601e-01 4.80852932e-01 3.91073555e-01 -3.73400807e-01
7.58776888e-02 3.81898791e-01 -3.20026934e-01 2.05217868e-01
-2.06820264e-01 2.96194166e-01 2.32527569e-01 3.19633096e-01
1.10021606e-01 1.40922165e+00 -6.03715777e-01 -1.16026175e+00
5.79829156e-01 2.13877320e-01 -3.94102544e-01 1.29088625e-01
-1.24558121e-01 3.54265779e-01 -7.34862864e-01 7.27204144e-01
6.81973994e-01 -6.19335532e-01 1.50163963e-01 -2.32840240e-01
-1.52412072e-01 2.38644168e-01 -9.80777144e-01 1.37946248e+00
-2.38108650e-01 1.72889650e-01 2.34444767e-01 -3.48757744e-01
5.62784374e-01 7.27956116e-01 9.17448878e-01 -7.68181384e-02
4.01634544e-01 6.38962448e-01 5.83051801e-01 -5.11970162e-01
-1.36275545e-01 -2.86511004e-01 1.90797433e-01 7.85178900e-01
-2.39875406e-01 -8.11412334e-01 3.13778102e-01 9.74599123e-02
1.48492897e+00 -1.85375690e-01 1.06158352e+00 -3.63872081e-01
5.89686930e-01 7.28922725e-01 -2.10287690e-01 9.13598061e-01
-4.79530185e-01 8.77268255e-01 3.64883780e-01 -3.56410980e-01
-1.28537214e+00 -1.11828589e+00 -4.36009169e-01 5.80045283e-01
-1.91639960e-01 1.37187615e-01 -8.22867513e-01 -1.49060118e+00
4.26148921e-02 2.12269530e-01 -6.29666090e-01 3.05479258e-01
-9.37612891e-01 -8.29248846e-01 4.86704797e-01 6.49326384e-01
3.92149627e-01 -1.02805603e+00 -1.30132174e+00 2.49402761e-01
1.58297662e-02 -8.91523600e-01 -6.81926429e-01 5.89466870e-01
-1.23660064e+00 -1.40084481e+00 -9.30275083e-01 -9.21745837e-01
8.59157503e-01 4.75556582e-01 1.62590849e+00 4.08090323e-01
-1.09470761e+00 4.32900876e-01 2.00128350e-02 -6.58438981e-01
-9.78362739e-01 1.56564981e-01 -5.72094262e-01 -7.37627447e-01
1.30451441e-01 -1.43223787e-02 -7.89830327e-01 5.32618880e-01
-9.09615934e-01 1.25464350e-01 1.30671024e+00 4.25077051e-01
1.04639530e+00 -6.39024526e-02 1.62035674e-02 -1.43824983e+00
2.78040051e-01 -2.96081066e-01 -5.81399679e-01 1.58599585e-01
-3.44461352e-01 4.91275452e-02 1.21948048e-01 -2.11554646e-01
-8.86511922e-01 8.74560535e-01 -2.09559843e-01 -3.62696022e-01
-4.27650511e-01 -1.18435755e-01 4.79435742e-01 -4.55841273e-01
9.10557985e-01 -7.74759799e-02 1.13798626e-01 -3.24776560e-01
2.33961135e-01 4.34162825e-01 3.96144331e-01 1.89335421e-01
9.91377950e-01 7.36807644e-01 5.50010383e-01 -5.20280838e-01
-9.08358097e-01 -1.18723035e+00 -1.28023660e+00 -1.73391283e-01
1.01713872e+00 -5.70842147e-01 -2.43339330e-01 -4.10765082e-01
-8.23251784e-01 -2.69366860e-01 -9.48263109e-01 6.10930264e-01
-3.75022471e-01 3.23138595e-01 -4.41583395e-01 -3.70369703e-01
-6.70255482e-01 -1.03989172e+00 1.21604228e+00 1.01548977e-01
-2.12953135e-01 -9.47629511e-01 1.58684880e-01 1.56020775e-01
6.25587642e-01 2.43119612e-01 8.85053337e-01 -1.10655499e+00
-7.82169104e-01 -6.44279242e-01 -4.09533620e-01 2.07935758e-02
2.82715648e-01 -2.55486727e-01 -7.39085734e-01 7.60465488e-02
1.49714157e-01 -1.32644065e-02 1.03908026e+00 8.68705869e-01
1.18225539e+00 2.34699458e-01 -7.81571686e-01 3.41886729e-01
1.39473820e+00 -7.38006318e-03 1.29714668e-01 -1.03582464e-01
7.50165641e-01 8.98094326e-02 3.66268486e-01 1.16633639e-01
-3.12444687e-01 1.32939667e-01 8.26139569e-01 -5.56257129e-01
-7.27885067e-01 1.95455417e-01 -3.10324937e-01 3.87532234e-01
-3.66290897e-01 -1.16299748e-01 -1.40397942e+00 5.49756050e-01
-1.42756772e+00 -5.81154525e-01 -8.77245843e-01 1.95788491e+00
5.42823136e-01 2.94990331e-01 3.62689614e-01 5.90384519e-03
6.43918097e-01 -3.25819492e-01 -4.58313733e-01 1.09514847e-01
6.69184446e-01 4.74985003e-01 9.07381237e-01 2.10015655e-01
-1.48756337e+00 4.50435311e-01 7.38062620e+00 6.51272893e-01
-9.95672047e-01 3.71485680e-01 4.88356560e-01 -2.21787602e-01
6.79546967e-02 -3.44584584e-01 -7.91262150e-01 -2.45282531e-01
4.75095570e-01 2.25814637e-02 -2.21254319e-01 1.03548253e+00
8.12954828e-02 -3.05910766e-01 -1.35551810e+00 4.92397457e-01
-1.51645660e-01 -1.25377512e+00 -7.40116835e-02 2.80465901e-01
6.70240521e-01 3.58009815e-01 -2.35964000e-01 2.17946947e-01
3.15390080e-01 -1.33467817e+00 2.54532039e-01 1.59451962e-01
7.46165156e-01 -3.36160600e-01 1.02123976e+00 5.96579075e-01
-1.33865130e+00 3.47833246e-01 -2.96729773e-01 7.23328948e-01
3.67455594e-02 5.00132203e-01 -2.37334943e+00 1.62823424e-01
3.98583621e-01 4.06378731e-02 -1.11855233e+00 1.70001245e+00
-4.50344160e-02 8.74188721e-01 -5.42243898e-01 -7.71794766e-02
3.35845798e-01 2.20712975e-01 5.94896257e-01 1.48727465e+00
4.03142065e-01 2.91646719e-01 5.19810200e-01 9.56241965e-01
-1.70510978e-01 2.35206991e-01 -4.10023540e-01 1.50241911e-01
1.89704038e-02 1.72529733e+00 -1.60411489e+00 -5.75470030e-01
-3.25094134e-01 7.01842606e-01 -2.36390412e-01 -4.63638932e-01
-9.41617966e-01 1.78549916e-01 -4.10307854e-01 6.79124415e-01
5.35058022e-01 7.11846650e-02 -4.57968086e-01 -5.10060430e-01
-1.28100008e-01 -2.33660847e-01 6.23346806e-01 -3.09037536e-01
-1.06609321e+00 6.88180923e-01 3.37575823e-02 -1.28107882e+00
-3.50552350e-01 -7.63544917e-01 -7.68270433e-01 5.70362508e-01
-1.25190973e+00 -1.25333786e+00 -7.54814208e-01 2.96183974e-01
6.23365760e-01 2.50320435e-01 5.94929814e-01 1.10323608e-01
1.00026749e-01 -3.56132947e-02 -2.20379189e-01 3.48785967e-01
3.77552032e-01 -1.82138360e+00 5.02935350e-01 4.73375410e-01
1.87642828e-01 -2.00882375e-01 1.30816087e-01 -6.99255288e-01
-1.21502113e+00 -1.52487540e+00 4.10268754e-01 -1.06350458e+00
2.45921254e-01 1.38057079e-02 -7.51290917e-01 7.32961655e-01
4.82462719e-02 6.42077804e-01 3.89261544e-01 -5.45597970e-01
2.45584875e-01 5.86946785e-01 -1.40164733e+00 3.16855699e-01
8.05261791e-01 6.65797368e-02 -4.66920823e-01 7.12578058e-01
3.50572258e-01 -1.01369774e+00 -7.70631135e-01 6.02736473e-01
3.98740590e-01 -1.02230537e+00 1.34627748e+00 -3.55863810e-01
2.76863370e-02 -2.46627182e-01 5.63933909e-01 -7.53012359e-01
-4.86793399e-01 7.68145695e-02 -3.69213931e-02 2.40729287e-01
5.54631352e-01 -5.62584912e-03 1.38259315e+00 3.73230666e-01
-4.17020142e-01 -6.60716474e-01 -8.21151197e-01 -5.68602026e-01
1.13046043e-01 -2.32489884e-01 1.69827610e-01 2.91360438e-01
-7.31384873e-01 8.18714499e-02 4.30146664e-01 6.83687702e-02
6.30080462e-01 1.89102829e-01 4.64066386e-01 -1.44383025e+00
-2.62825012e-01 -8.01118851e-01 -1.99696541e-01 -7.17315257e-01
-7.06306994e-01 -1.34609234e+00 2.84003437e-01 -2.10177779e+00
7.24657178e-01 -2.12216988e-01 -4.00217474e-02 3.83315355e-01
-1.37078300e-01 7.53023207e-01 8.16722587e-02 4.24711794e-01
-5.96824586e-01 -4.35489535e-01 1.99763215e+00 1.12502933e-01
-1.60362199e-01 8.39655936e-01 -1.74199998e-01 9.66643572e-01
7.01118052e-01 -7.71785736e-01 -9.72298235e-02 6.43542930e-02
-3.08873057e-01 1.35869429e-01 6.41000092e-01 -1.04334605e+00
1.23076819e-01 2.21331939e-02 7.51042485e-01 -1.33332539e+00
1.56828910e-01 -1.16415668e+00 -1.30425811e-01 1.24039328e+00
-9.12123397e-02 -3.54959011e-01 1.79224327e-01 4.14928287e-01
1.54588223e-01 -7.87545741e-01 1.11023176e+00 -9.73729193e-01
-3.55319947e-01 4.41118121e-01 -6.46621466e-01 2.56120637e-02
1.42539799e+00 -3.15642893e-01 3.04212958e-01 2.07875609e-01
-1.11729038e+00 1.16275005e-01 1.85619459e-01 -2.43410811e-01
4.46642309e-01 -1.26539695e+00 -1.09217429e+00 1.30039021e-01
-1.40954033e-01 6.02869868e-01 -1.36995867e-01 1.19941306e+00
-1.17061877e+00 9.41710770e-01 3.42050456e-02 -1.05988610e+00
-1.65109968e+00 3.13286483e-01 8.88595343e-01 -7.96035111e-01
-7.70605505e-01 8.90343010e-01 2.40953624e-01 -3.65661949e-01
1.21390499e-01 -1.02820921e+00 -2.10977837e-01 -3.56203109e-01
8.08307007e-02 4.17298228e-01 4.69276071e-01 -1.63030833e-01
-6.68170869e-01 4.43695426e-01 -1.57120064e-01 5.89712076e-02
7.81536162e-01 3.39097261e-01 1.34927034e-01 1.42891690e-01
6.24300718e-01 6.76075891e-02 -7.94828475e-01 -1.01799935e-01
2.03261033e-01 -2.10596427e-01 2.74229020e-01 -9.64609206e-01
-9.04731870e-01 6.70375645e-01 8.96103561e-01 5.62960327e-01
7.21861660e-01 6.39723659e-01 6.09679282e-01 4.60089475e-01
-2.07403392e-01 -5.61672151e-01 1.88277885e-01 3.07234019e-01
5.82046568e-01 -1.60155571e+00 4.62107778e-01 -9.25595105e-01
-5.15882909e-01 1.45305789e+00 6.88506663e-01 -2.55515993e-01
7.29448736e-01 4.47085500e-01 1.76474422e-01 -6.57480240e-01
-5.25069416e-01 -5.68919539e-01 9.77427840e-01 3.49297523e-01
8.65827560e-01 2.68316567e-01 -1.72412425e-01 2.85637885e-01
2.13963941e-01 -1.72031790e-01 5.38148999e-01 1.14169729e+00
-1.05351782e+00 -8.30119014e-01 -8.80703330e-01 9.72015381e-01
-6.73798978e-01 1.85151845e-02 -7.89647877e-01 1.36008632e+00
1.00345217e-01 2.38401547e-01 6.21707141e-02 5.48479617e-01
4.76663888e-01 9.77352336e-02 6.10121965e-01 -1.30639923e+00
-1.06832731e+00 4.56345230e-01 -3.43871932e-03 -2.86516786e-01
-2.19059825e-01 -5.53992987e-01 -1.70639360e+00 2.57099301e-01
-4.80442613e-01 -4.29589115e-02 7.47277081e-01 6.91716373e-01
-1.58755988e-01 1.00787067e+00 3.75994831e-01 -7.51435518e-01
-8.52971792e-01 -9.63416338e-01 -3.99875283e-01 1.48242787e-01
7.39133134e-02 -1.34085909e-01 -5.02311885e-02 3.77790853e-02] | [15.381422996520996, -2.151587963104248] |
687936fa-6a91-44bf-b693-f9c7966f8984 | learning-semantic-aligned-feature | 2112.06714 | null | https://arxiv.org/abs/2112.06714v1 | https://arxiv.org/pdf/2112.06714v1.pdf | Learning Semantic-Aligned Feature Representation for Text-based Person Search | Text-based person search aims to retrieve images of a certain pedestrian by a textual description. The key challenge of this task is to eliminate the inter-modality gap and achieve the feature alignment across modalities. In this paper, we propose a semantic-aligned embedding method for text-based person search, in which the feature alignment across modalities is achieved by automatically learning the semantic-aligned visual features and textual features. First, we introduce two Transformer-based backbones to encode robust feature representations of the images and texts. Second, we design a semantic-aligned feature aggregation network to adaptively select and aggregate features with the same semantics into part-aware features, which is achieved by a multi-head attention module constrained by a cross-modality part alignment loss and a diversity loss. Experimental results on the CUHK-PEDES and Flickr30K datasets show that our method achieves state-of-the-art performances. | ['Min Zhang', 'Min Cao', 'Shiping Li'] | 2021-12-13 | null | null | null | null | ['person-search'] | ['computer-vision'] | [ 1.60054520e-01 -4.21890289e-01 -1.14709459e-01 -6.49215877e-01
-1.05019403e+00 -2.77230740e-01 9.41032887e-01 -1.90530092e-01
-6.54951036e-01 3.49680126e-01 6.65642679e-01 3.57352883e-01
-3.53052139e-01 -4.68974531e-01 -5.77526033e-01 -6.45292699e-01
3.42881769e-01 2.86971748e-01 1.48371309e-01 -7.29651004e-02
-5.34485020e-02 2.90763378e-03 -1.84902656e+00 5.73102057e-01
7.72179723e-01 1.41185749e+00 1.59442410e-01 2.84671545e-01
-7.91776925e-02 5.29430211e-01 -7.08404556e-03 -8.91305864e-01
3.31686646e-01 -4.88029152e-01 -9.71198380e-01 5.28711915e-01
8.53413463e-01 -1.58960223e-01 -9.24009740e-01 9.91310179e-01
9.11281765e-01 2.57264972e-01 6.26230001e-01 -1.46149647e+00
-8.79194081e-01 1.19397089e-01 -4.57575679e-01 -6.69534039e-03
5.92822194e-01 3.11285466e-01 1.22627497e+00 -1.13724530e+00
4.27228004e-01 1.61813343e+00 3.90794426e-01 5.32256782e-01
-1.03990364e+00 -4.66574252e-01 2.97074527e-01 8.66427064e-01
-1.75027466e+00 -4.91287410e-01 8.52460861e-01 -3.61115873e-01
7.79599369e-01 5.28624237e-01 8.54653716e-01 1.16776252e+00
-2.02266023e-01 1.16015720e+00 9.56016123e-01 -3.52386713e-01
-2.94877619e-01 2.50040472e-01 -2.16880403e-02 9.86644506e-01
2.86811404e-02 -2.65328656e-03 -1.03789699e+00 -2.18878597e-01
4.54748243e-01 3.78897786e-01 -2.96772748e-01 -6.91547751e-01
-1.35884488e+00 8.37900400e-01 6.82105601e-01 1.28586337e-01
-2.42366165e-01 1.45068154e-01 6.99952483e-01 2.10859463e-01
3.19063604e-01 1.19686447e-01 -2.95469314e-01 1.94891676e-01
-7.42477179e-01 2.92266876e-01 2.65463173e-01 1.19254315e+00
6.40818715e-01 -4.61119533e-01 -7.99249053e-01 1.03253698e+00
5.44050038e-01 8.46957982e-01 5.67826450e-01 -7.16859043e-01
6.46526694e-01 7.82144487e-01 9.26916823e-02 -1.13762784e+00
-1.13705739e-01 -2.98172742e-01 -8.47289205e-01 -2.92835653e-01
1.44274667e-01 2.77543366e-01 -9.58022296e-01 1.77300763e+00
3.78701985e-01 -4.48014885e-02 -8.47048163e-02 1.30668032e+00
1.15501690e+00 2.88509399e-01 1.70878127e-01 1.34882540e-01
1.93444443e+00 -1.34574986e+00 -5.86829364e-01 -2.73419082e-01
1.74475729e-01 -5.82163334e-01 1.14605403e+00 -2.72257715e-01
-8.78324091e-01 -6.49979055e-01 -8.17023873e-01 -4.68581080e-01
-5.60702085e-01 2.35219464e-01 2.72146553e-01 5.24599671e-01
-8.25102627e-01 4.06848639e-02 -3.77593338e-01 -6.82740569e-01
5.26905119e-01 2.18431741e-01 -6.32019222e-01 -2.99494535e-01
-1.35180902e+00 6.99786484e-01 4.49213475e-01 -5.34962639e-02
-5.04126608e-01 -4.42135364e-01 -1.13302958e+00 1.69285864e-01
3.34333956e-01 -1.35602415e+00 7.23448217e-01 -9.90625441e-01
-1.13149571e+00 1.26314962e+00 -3.97559077e-01 -2.13118821e-01
7.02193737e-01 -2.50000894e-01 -5.20360351e-01 5.35526872e-01
4.82756853e-01 7.78329015e-01 1.04393303e+00 -1.19287741e+00
-9.05117691e-01 -5.64350128e-01 -1.28967300e-01 5.78676581e-01
-7.76022613e-01 1.98280379e-01 -1.22211325e+00 -6.77155316e-01
-1.18456498e-01 -7.90710688e-01 -1.05702735e-01 3.98957282e-01
-5.79378307e-01 -4.50855821e-01 4.50869679e-01 -8.60881984e-01
9.70284998e-01 -1.98840201e+00 3.75175476e-01 2.81855196e-01
1.58742920e-01 -8.39494020e-02 -3.41949135e-01 3.51601511e-01
3.11665416e-01 -2.53328770e-01 7.70960525e-02 -6.40450120e-01
3.70330930e-01 -9.86435115e-02 -6.37104958e-02 4.61097747e-01
2.64667690e-01 1.19741511e+00 -8.77573252e-01 -9.98147368e-01
3.16376835e-01 6.27282739e-01 -3.67260545e-01 1.22998297e-01
1.13981105e-01 3.27640206e-01 -7.44130433e-01 8.43214810e-01
4.51879263e-01 -4.65391248e-01 -7.92437941e-02 -6.73198521e-01
1.62393287e-01 -1.68181673e-01 -9.35056210e-01 2.08655190e+00
-1.69302791e-01 3.37107420e-01 -2.53936678e-01 -1.02399600e+00
5.88688195e-01 1.09385245e-01 5.27620375e-01 -1.17164993e+00
1.17479011e-01 8.33899528e-02 -8.37407053e-01 -5.99525034e-01
5.61169922e-01 3.39041412e-01 -4.41436440e-01 1.76920101e-01
1.92159563e-01 4.73466367e-01 3.42505611e-02 1.43106461e-01
6.98507905e-01 -9.67690870e-02 4.97963373e-03 -1.47035420e-01
7.89559543e-01 -1.74758866e-01 3.25459927e-01 8.55045378e-01
-1.43506497e-01 7.72320628e-01 -1.28566727e-01 -5.89599550e-01
-1.05949903e+00 -1.00287330e+00 -7.87250772e-02 1.23860526e+00
6.37470603e-01 -5.88082671e-01 -4.75787044e-01 -9.85890746e-01
1.49150997e-01 2.42285818e-01 -6.98383808e-01 -3.26590955e-01
-3.00281316e-01 -5.05292952e-01 2.57147133e-01 4.99110192e-01
9.66224492e-01 -7.03935146e-01 -5.39060175e-01 -1.19189709e-01
-7.06011891e-01 -1.27053165e+00 -1.11345875e+00 -2.32973963e-01
-2.38768250e-01 -1.04457271e+00 -1.06236756e+00 -1.09174049e+00
7.81114399e-01 5.13763249e-01 9.57906663e-01 1.31322965e-01
-4.39877838e-01 9.05985236e-01 -4.84609395e-01 6.28321767e-02
5.11392355e-01 5.02782501e-02 -1.43387213e-01 4.93182719e-01
5.22386193e-01 -9.18645039e-02 -9.55941141e-01 3.97337526e-01
-7.07091033e-01 9.41630676e-02 5.47944009e-01 1.21339178e+00
8.55180502e-01 -1.31164789e-01 1.58351302e-01 7.48754432e-03
4.27642435e-01 -1.97638229e-01 -9.95257348e-02 7.35656738e-01
-4.45390165e-01 1.25313297e-01 4.02796000e-01 -4.26183134e-01
-8.75155270e-01 3.86690587e-01 1.75037920e-01 -5.20490646e-01
4.41951528e-02 2.72828519e-01 -5.41817367e-01 -8.20337385e-02
1.53752536e-01 9.21635151e-01 -1.16596706e-01 -4.79787260e-01
6.55126750e-01 7.07163453e-01 6.45709515e-01 -5.42795122e-01
8.85998249e-01 6.59382284e-01 -1.58911109e-01 -7.86982656e-01
-9.57442462e-01 -7.25486398e-01 -6.01746142e-01 -2.89929479e-01
1.04396045e+00 -1.24062145e+00 -6.62298560e-01 5.03563821e-01
-8.75607133e-01 2.20868021e-01 -2.56979108e-01 4.10744548e-01
-6.14864767e-01 6.64946079e-01 -1.65664732e-01 -5.33669531e-01
-5.88643551e-01 -1.07623255e+00 1.47398341e+00 4.53200549e-01
2.71803528e-01 -6.27109468e-01 -2.97196269e-01 7.14606822e-01
1.74279466e-01 -6.36814088e-02 5.40135443e-01 -6.93880498e-01
-6.25249863e-01 -2.85191208e-01 -6.94382489e-01 2.76336726e-03
1.05994336e-01 -7.39300847e-01 -6.82528019e-01 -5.80573142e-01
-5.44942915e-01 -5.89358687e-01 1.27548611e+00 1.22163102e-01
1.28963220e+00 -4.35445249e-01 -6.22587502e-01 7.26180673e-01
1.26807523e+00 -4.50373143e-01 4.54051971e-01 4.67602521e-01
7.93735027e-01 4.43754762e-01 5.28174996e-01 4.15157735e-01
8.98583889e-01 1.13921773e+00 2.18446851e-01 3.26509848e-02
-3.37754875e-01 -5.73840976e-01 8.64092857e-02 3.99998844e-01
1.35158241e-01 -1.78812012e-01 -5.31734765e-01 8.27961862e-01
-2.18556356e+00 -1.26368153e+00 2.92211324e-01 2.05558348e+00
8.69267106e-01 -2.91258574e-01 4.30127800e-01 -2.37996191e-01
8.62240195e-01 1.65686563e-01 -4.73631740e-01 3.84043127e-01
-3.33234221e-01 -2.88936496e-01 2.65545309e-01 2.34142393e-01
-1.49429703e+00 8.97352815e-01 5.49667072e+00 1.09874392e+00
-7.32937992e-01 3.34421396e-01 4.03902799e-01 -2.52142936e-01
-3.28072637e-01 -3.19427788e-01 -9.38343942e-01 7.20487356e-01
2.57499307e-01 -2.24952117e-01 2.27377057e-01 6.04972184e-01
-1.04105890e-01 3.73586893e-01 -1.02979076e+00 1.44478309e+00
5.93307793e-01 -1.16547310e+00 4.65965658e-01 -3.52708921e-02
5.75692534e-01 -2.81310648e-01 3.04919004e-01 1.95392981e-01
6.46057501e-02 -9.40558910e-01 1.02558410e+00 9.63547766e-01
9.07935262e-01 -7.16810048e-01 5.12769520e-01 -1.60173699e-01
-1.68005359e+00 -2.80259997e-01 -3.36357534e-01 5.04981935e-01
3.34135890e-01 1.80688187e-01 -1.72325253e-01 8.27379465e-01
1.03180480e+00 8.89311194e-01 -9.17255461e-01 1.34961355e+00
1.13771945e-01 -3.01823728e-02 -2.45379448e-01 -1.35006025e-01
8.30427855e-02 8.60682577e-02 5.88577926e-01 1.30878913e+00
2.66091168e-01 -1.82939753e-01 4.93345588e-01 8.35690618e-01
3.84387560e-03 2.69124538e-01 -2.66313940e-01 1.51813552e-01
5.79633653e-01 1.05698228e+00 -2.21726328e-01 -2.68980682e-01
-6.60595298e-01 1.57565606e+00 3.08875382e-01 4.19299215e-01
-8.08561563e-01 -4.11293954e-01 6.72251642e-01 -8.99382234e-02
6.10452473e-01 1.72550619e-01 7.36582850e-04 -1.44072115e+00
4.72518295e-01 -7.77040720e-01 7.97226429e-01 -7.85773396e-01
-1.76421142e+00 6.07968211e-01 -7.28808045e-02 -1.34064651e+00
3.99100333e-02 -4.08496201e-01 -1.85366914e-01 8.75283122e-01
-1.71981466e+00 -2.05689168e+00 -5.91021359e-01 1.09832489e+00
5.93863845e-01 -4.70759183e-01 6.21198893e-01 4.80440378e-01
-2.40809977e-01 1.11782575e+00 3.42681035e-02 3.75017911e-01
7.70723581e-01 -9.64603066e-01 8.32646787e-02 6.17413759e-01
1.64163232e-01 4.77817029e-01 4.58427757e-01 -4.89814669e-01
-1.43759620e+00 -1.24713445e+00 1.16530049e+00 -3.80386531e-01
4.36027288e-01 -2.20645070e-01 -4.93324101e-01 4.55507845e-01
3.20929945e-01 2.36001298e-01 6.86514795e-01 -5.14101349e-02
-6.61915481e-01 -2.96548933e-01 -9.28715527e-01 6.96081161e-01
1.45543790e+00 -7.87675083e-01 -5.95434844e-01 4.62259859e-01
6.85885489e-01 -1.55159935e-01 -7.41763890e-01 2.65690327e-01
7.97653854e-01 -7.51393676e-01 1.48719394e+00 -8.16785574e-01
1.91686839e-01 -3.31728160e-01 -4.70398933e-01 -1.08964705e+00
-5.11918008e-01 -2.96108752e-01 8.09682235e-02 1.21485150e+00
9.97779071e-02 -4.94157583e-01 4.41271245e-01 5.92804432e-01
1.05655089e-01 -5.24482369e-01 -1.22486246e+00 -7.86824942e-01
-1.87559545e-01 1.63474698e-02 6.70500696e-01 7.80959070e-01
-1.73006520e-01 3.69970411e-01 -7.51798391e-01 1.03167877e-01
1.00328600e+00 3.61262202e-01 6.18338764e-01 -1.14854980e+00
1.62354857e-02 -5.09274662e-01 -6.94847226e-01 -1.27314579e+00
3.67373675e-01 -1.06404650e+00 -8.58795941e-02 -1.60069454e+00
1.02751994e+00 -4.27892320e-02 -5.34565866e-01 4.84815419e-01
-4.05936807e-01 2.72966087e-01 2.94538051e-01 2.40749687e-01
-1.29062819e+00 1.14064276e+00 1.06187606e+00 -5.18233359e-01
1.88672796e-01 -2.41303504e-01 -7.78643072e-01 3.46214801e-01
3.45987618e-01 -1.41153917e-01 -2.98403054e-01 -6.82832241e-01
1.36145607e-01 -2.97876477e-01 1.06485295e+00 -8.43115270e-01
4.66476202e-01 -1.28537223e-01 7.49641001e-01 -7.99962938e-01
7.78065741e-01 -1.01446176e+00 -1.59487888e-01 1.97386682e-01
-6.67257607e-01 -7.31926486e-02 -2.50931829e-01 9.76016998e-01
-2.44457051e-01 2.59572059e-01 5.64826965e-01 -8.39249715e-02
-9.22443211e-01 6.56531692e-01 1.59321830e-01 1.82511955e-01
8.31576705e-01 -2.49666199e-01 -2.99466044e-01 -3.72521013e-01
-6.09544516e-01 7.64951766e-01 4.39107388e-01 7.23763525e-01
9.51582849e-01 -2.06290793e+00 -8.66520584e-01 1.96760222e-01
6.81568742e-01 -4.90601093e-01 4.85682786e-01 8.12106669e-01
1.52796060e-01 5.62347591e-01 -1.74684048e-01 -6.23869836e-01
-1.56012690e+00 5.64608097e-01 4.58793610e-01 -9.19532776e-02
-6.86753452e-01 9.95594919e-01 1.18562862e-01 -3.78292590e-01
3.43736947e-01 4.05384898e-01 -2.74266541e-01 1.15223534e-01
7.54660904e-01 6.28799424e-02 -2.02615038e-01 -1.13065982e+00
-6.63342774e-01 8.30210149e-01 -2.42664199e-02 -8.14519152e-02
1.18184757e+00 -6.30201638e-01 2.25378796e-02 -2.92200204e-02
1.41144025e+00 -3.76216799e-01 -1.18630302e+00 -9.04591739e-01
-2.62184530e-01 -8.63274753e-01 -1.24131009e-01 -7.28915989e-01
-1.18519688e+00 5.33581376e-01 9.33592737e-01 -1.46136031e-01
1.16065753e+00 4.28407848e-01 1.03024864e+00 4.07008559e-01
1.65212855e-01 -1.21178794e+00 4.54581559e-01 2.89068937e-01
9.47796285e-01 -1.41287065e+00 -1.32592782e-01 -2.41913483e-01
-7.89765537e-01 8.39879632e-01 6.01789951e-01 2.91361839e-01
5.08320212e-01 -6.47019029e-01 -3.20761174e-01 -2.38396183e-01
-5.56705296e-01 -8.89894247e-01 9.81356263e-01 6.40974462e-01
-1.47871599e-02 -9.09435749e-02 -2.37432852e-01 8.92519593e-01
1.62786134e-02 -2.45562345e-01 -4.21726763e-01 7.63956666e-01
-5.24133325e-01 -8.74829471e-01 -2.78941482e-01 4.06526804e-01
-2.36558229e-01 -2.39005238e-01 -5.30556738e-01 3.80910873e-01
7.06530064e-02 1.06192219e+00 -1.23772718e-01 -5.81223369e-01
3.48525792e-01 2.87302256e-01 5.25614858e-01 -1.12104259e-01
-3.05051833e-01 -9.25149098e-02 1.10928990e-01 -6.21591806e-01
-5.72735190e-01 -7.50458419e-01 -6.88050210e-01 -6.41115615e-03
-2.38996595e-01 4.78450358e-02 2.72331923e-01 9.49629307e-01
7.00410485e-01 1.60305530e-01 6.79129124e-01 -6.55950904e-01
-6.21375978e-01 -7.01729834e-01 -4.59294200e-01 9.44385886e-01
2.94230670e-01 -7.92982459e-01 -1.66942164e-01 1.96889564e-01] | [14.629586219787598, 0.8663081526756287] |
daf4c657-44ae-4f03-8f5d-c482ea9a3a39 | seeing-glass-joint-point-cloud-and-depth | 2110.00087 | null | https://arxiv.org/abs/2110.00087v1 | https://arxiv.org/pdf/2110.00087v1.pdf | Seeing Glass: Joint Point Cloud and Depth Completion for Transparent Objects | The basis of many object manipulation algorithms is RGB-D input. Yet, commodity RGB-D sensors can only provide distorted depth maps for a wide range of transparent objects due light refraction and absorption. To tackle the perception challenges posed by transparent objects, we propose TranspareNet, a joint point cloud and depth completion method, with the ability to complete the depth of transparent objects in cluttered and complex scenes, even with partially filled fluid contents within the vessels. To address the shortcomings of existing transparent object data collection schemes in literature, we also propose an automated dataset creation workflow that consists of robot-controlled image collection and vision-based automatic annotation. Through this automated workflow, we created Toronto Transparent Objects Depth Dataset (TODD), which consists of nearly 15000 RGB-D images. Our experimental evaluation demonstrates that TranspareNet outperforms existing state-of-the-art depth completion methods on multiple datasets, including ClearGrasp, and that it also handles cluttered scenes when trained on TODD. Code and dataset will be released at https://www.pair.toronto.edu/TranspareNet/ | ['Animesh Garg', 'Florian Shkurti', 'Alàn Aspuru-Guzik', 'Sagi Eppel', 'Yi Ru Wang', 'Haoping Xu'] | 2021-09-30 | null | null | null | null | ['transparent-objects'] | ['computer-vision'] | [ 1.39268920e-01 -7.13556781e-02 5.79860151e-01 -3.94632638e-01
-4.16215599e-01 -8.14236224e-01 3.06577682e-01 -2.52804458e-01
-1.95897236e-01 3.00575286e-01 5.67897176e-03 -4.21801619e-02
1.31487370e-01 -7.74430275e-01 -4.20124024e-01 -3.86830062e-01
2.19544813e-01 7.00675428e-01 6.16454422e-01 -2.03237496e-02
4.03074473e-01 6.28412247e-01 -1.74559319e+00 2.85446823e-01
8.21681619e-01 1.21862876e+00 5.96149087e-01 8.03883195e-01
-3.62982064e-01 7.32262909e-01 -3.80794585e-01 -2.90037006e-01
8.96699309e-01 3.67588550e-01 -6.49467766e-01 3.27688724e-01
7.16772616e-01 -1.08309603e+00 -4.22260225e-01 6.02109730e-01
4.38147753e-01 -1.21762648e-01 5.56668937e-01 -1.24167037e+00
-7.38108337e-01 -1.79468900e-01 -6.41876876e-01 -2.59230167e-01
7.55016744e-01 5.43924987e-01 5.70377707e-01 -1.33508325e+00
7.55696952e-01 1.30059528e+00 6.14097297e-01 5.76290429e-01
-9.38263953e-01 -3.52549136e-01 5.87105229e-02 -2.95992106e-01
-1.00383353e+00 -4.13927466e-01 6.54372633e-01 -5.93534350e-01
1.00073266e+00 1.07332751e-01 1.00494993e+00 8.29545677e-01
-9.27584022e-02 5.90116084e-01 1.11131644e+00 -1.80117354e-01
3.06738555e-01 -1.01290286e-01 -1.57157019e-01 7.79457211e-01
4.41381723e-01 2.81234216e-02 -7.69161224e-01 2.48311292e-02
1.13986146e+00 1.15260400e-01 -3.72404844e-01 -8.64787638e-01
-1.51359534e+00 3.32053423e-01 5.38819015e-01 -5.06774127e-01
-2.11366862e-01 3.11331213e-01 -3.56561430e-02 1.70556784e-01
4.75250274e-01 2.69803971e-01 -5.58679342e-01 -3.01670820e-01
-2.44697705e-01 3.02538663e-01 9.20338213e-01 1.50910985e+00
1.09366822e+00 -3.68804842e-01 3.09863359e-01 5.09920955e-01
6.65360630e-01 7.28113711e-01 -2.37226412e-01 -1.62259233e+00
7.64596164e-01 8.67053866e-01 4.43953454e-01 -4.98928338e-01
-4.34862405e-01 4.93294775e-01 -2.26110905e-01 8.33481014e-01
6.42189145e-01 2.45998763e-02 -1.07338512e+00 7.67806888e-01
9.21747983e-01 -4.28276062e-01 2.99258024e-01 1.20841718e+00
1.44478095e+00 4.13960218e-01 -6.06823623e-01 2.51682103e-01
1.00416291e+00 -9.81372535e-01 -4.70011979e-01 -3.05538923e-01
3.32131863e-01 -8.76898944e-01 1.51260781e+00 6.96272969e-01
-1.20893157e+00 -1.36103839e-01 -8.50758016e-01 -8.38043809e-01
-2.29466096e-01 -1.99432671e-03 1.14928162e+00 4.72893178e-01
-9.71968114e-01 1.14946954e-01 -8.81260276e-01 -4.03238565e-01
5.86276472e-01 2.44925261e-01 -4.92994696e-01 -5.67244053e-01
-2.93534011e-01 6.24789774e-01 -3.50364447e-02 2.17309043e-01
-1.06779504e+00 -1.05774391e+00 -7.89128244e-01 -6.68323398e-01
3.51511031e-01 -8.23985636e-01 1.49210739e+00 -2.41047189e-01
-1.82171726e+00 1.15081251e+00 4.00132127e-02 2.29118943e-01
8.41305077e-01 -6.77068830e-01 3.55053246e-01 6.32813454e-01
2.23284252e-02 8.81991923e-01 5.54905236e-01 -1.73804319e+00
-3.81965965e-01 -5.33806980e-01 3.61894101e-01 4.91150111e-01
-5.41768894e-02 -1.72274977e-01 -7.86909103e-01 -5.84477000e-02
6.92635477e-01 -8.75784874e-01 -1.72411546e-01 1.28330863e+00
-5.19247711e-01 1.40259027e-01 9.53031003e-01 -1.65182784e-01
1.68767452e-01 -1.96113372e+00 -4.80041318e-02 -7.44209886e-02
6.28095627e-01 1.73513182e-02 4.16890765e-03 4.63607341e-01
6.15589142e-01 -2.44578943e-01 -3.46097529e-01 -8.52017820e-01
8.14533159e-02 3.43223959e-01 -2.34561995e-01 6.13784492e-01
-9.63695720e-02 5.81342697e-01 -8.27125549e-01 -5.02793729e-01
5.06624579e-01 6.02233231e-01 -4.61196005e-01 3.91096652e-01
-5.17147720e-01 7.33330607e-01 -4.53347147e-01 1.33553815e+00
1.16841638e+00 3.49677987e-02 -2.78104514e-01 -1.10278629e-01
-6.09700203e-01 2.11359069e-01 -1.02845633e+00 2.39219022e+00
-2.74612874e-01 6.13687634e-01 4.14508969e-01 1.00916341e-01
1.01616871e+00 1.12739615e-01 6.57252729e-01 -4.77335393e-01
4.47490588e-02 4.70770419e-01 -4.67503071e-01 -8.74206483e-01
6.89422488e-01 2.27244422e-01 2.50883102e-01 4.30728495e-01
-3.87037426e-01 -1.32896614e+00 -6.36579245e-02 2.56387711e-01
1.25442100e+00 7.25143015e-01 -3.35153908e-01 2.36004576e-01
-1.93354398e-01 3.65077496e-01 5.36448777e-01 4.64886665e-01
-2.54775226e-01 1.27083075e+00 -2.89436672e-02 -5.54995835e-01
-9.83213305e-01 -1.45119226e+00 -3.14092934e-01 4.55166548e-01
7.58699834e-01 -1.33404449e-01 -3.99634421e-01 -2.07279995e-01
4.04291213e-01 2.27066968e-02 -4.74813461e-01 3.63802373e-01
-1.69221595e-01 -3.81065637e-01 9.13968310e-02 3.72308075e-01
9.04207647e-01 -8.95541131e-01 -1.13340759e+00 -5.97788617e-02
-1.08295575e-01 -1.52321219e+00 -2.64291968e-02 5.54210283e-02
-9.92631435e-01 -1.30293024e+00 -7.15173662e-01 -4.76595670e-01
5.87506294e-01 1.03317916e+00 1.16395950e+00 -5.11789732e-02
-6.10676885e-01 9.18851435e-01 -5.27674317e-01 -7.62927592e-01
3.51021811e-02 -4.16369587e-01 6.93118796e-02 -4.70848113e-01
1.63490057e-01 -6.64120734e-01 -1.02186155e+00 3.62731785e-01
-8.85392129e-01 2.74650246e-01 3.81863862e-01 -1.27772406e-01
6.92405820e-01 -4.83096361e-01 -2.70816267e-01 -5.73060930e-01
6.87655881e-02 -2.20244050e-01 -9.61302102e-01 -4.70235124e-02
9.69524123e-03 -3.76355886e-01 -2.22590789e-01 -2.74352729e-01
-1.17957246e+00 4.46241170e-01 2.82784909e-01 -6.33558333e-01
-1.58013418e-01 -5.82495257e-02 -2.01828733e-01 -4.99408782e-01
7.69701421e-01 -1.03182934e-01 -1.49809748e-01 -5.48361659e-01
3.93697560e-01 7.17730701e-01 6.15701318e-01 -5.90505481e-01
9.56509471e-01 1.23942661e+00 -6.36400795e-03 -9.46899951e-01
-7.57847130e-01 -6.09589696e-01 -1.11720109e+00 -4.57122952e-01
8.18226635e-01 -1.07567430e+00 -8.43553126e-01 7.57238746e-01
-1.44433093e+00 -8.86411726e-01 -2.60528624e-01 6.59269989e-01
-7.48002827e-01 4.16265309e-01 -5.96641481e-01 -8.51443172e-01
-1.74306870e-01 -1.02898800e+00 1.56971443e+00 1.86789885e-01
8.25564042e-02 -6.24735832e-01 1.56659827e-01 6.42592549e-01
8.12758133e-02 5.83476067e-01 3.48768324e-01 5.68643570e-01
-1.43481529e+00 -1.22319593e-03 -3.70398968e-01 -5.81102399e-03
3.03502709e-01 3.04595530e-01 -1.37649977e+00 6.43646494e-02
-1.59489974e-01 -6.89888775e-01 7.63442874e-01 8.52874890e-02
9.87028658e-01 6.86184168e-02 -1.53982386e-01 9.84354317e-01
1.49305177e+00 -2.08005100e-01 8.03844094e-01 5.80750585e-01
9.11861122e-01 7.90678859e-01 1.00039005e+00 7.59818196e-01
7.56999671e-01 5.25445640e-01 1.17473412e+00 -2.00116247e-01
-3.32116991e-01 9.83671620e-02 1.62151888e-01 4.70365494e-01
-4.68795449e-01 -1.59240484e-01 -1.00608361e+00 3.25858116e-01
-1.48531365e+00 -4.05675381e-01 -8.06593060e-01 2.09329414e+00
7.78058052e-01 -5.33670783e-02 -7.16295242e-02 4.49479483e-02
1.41630322e-02 -3.38997930e-01 -7.82261729e-01 -1.81495935e-01
-2.09753647e-01 -1.23119310e-01 5.16698837e-01 4.14345682e-01
-8.20554733e-01 9.69305336e-01 6.07767057e+00 -2.69723445e-01
-7.58221090e-01 -7.12262839e-02 -2.21922502e-01 -3.06413591e-01
-4.93277520e-01 -1.32984752e-02 -4.76591617e-01 -2.03248896e-02
6.77384809e-02 3.65992337e-01 2.57359356e-01 8.12458456e-01
3.38500768e-01 -8.38222206e-01 -1.13374615e+00 1.28189111e+00
-2.13525109e-02 -9.24000621e-01 -1.60744920e-01 7.84619227e-02
9.00843740e-01 4.64286536e-01 6.05056696e-02 -4.58014965e-01
5.25889218e-01 -5.61777472e-01 1.12594295e+00 6.51574433e-01
8.04130018e-01 -3.85995694e-02 2.26157531e-01 1.37322590e-01
-1.15358949e+00 3.89049985e-02 -6.86029911e-01 -2.11955354e-01
7.79423863e-02 9.80326235e-01 -9.37721670e-01 2.77670622e-01
1.25867844e+00 7.62076795e-01 -3.89743596e-01 1.50677228e+00
-3.20638657e-01 -1.11855470e-01 -5.24371266e-01 2.02969074e-01
-1.05863623e-01 -4.28608745e-01 4.42066520e-01 7.56859899e-01
3.21888059e-01 2.47064561e-01 1.72432736e-01 1.04573190e+00
-8.20965786e-03 -3.87153566e-01 -7.94117749e-01 3.44276398e-01
5.23983538e-01 1.28600395e+00 -6.44984126e-01 -3.64081934e-02
-5.26324093e-01 1.23406148e+00 4.36062545e-01 4.59639370e-01
-6.23429000e-01 -4.09351856e-01 9.66536105e-01 8.11249167e-02
5.30993715e-02 -8.50064754e-01 -5.29726505e-01 -1.17923272e+00
5.30404747e-01 -1.80269927e-01 -1.08578816e-01 -1.57939446e+00
-1.17413735e+00 2.34618500e-01 -8.51132870e-02 -1.60059869e+00
5.63992202e-01 -8.92650247e-01 -1.79838344e-01 6.78729236e-01
-1.97825754e+00 -1.31368840e+00 -1.31458724e+00 8.22757304e-01
5.68861485e-01 4.27868336e-01 7.39918292e-01 4.33193631e-02
-1.73102379e-01 -2.83270329e-01 4.91838977e-02 -1.27977520e-01
7.26810277e-01 -1.35198283e+00 2.80306518e-01 5.77344835e-01
-3.35480899e-01 3.51649016e-01 4.95506346e-01 -6.47318721e-01
-2.03984571e+00 -9.87363636e-01 1.08280160e-01 -9.28228378e-01
3.04381937e-01 -6.37031674e-01 -6.80196762e-01 7.31397629e-01
-2.04143837e-01 3.87051523e-01 2.92837948e-01 -5.30426145e-01
-4.75876540e-01 -9.69919339e-02 -1.44909728e+00 5.31348825e-01
1.47977853e+00 -4.74598140e-01 -5.20442486e-01 6.82296753e-01
9.41570759e-01 -1.05367625e+00 -9.07996058e-01 3.18926603e-01
6.76464140e-01 -1.40594482e+00 1.20541155e+00 2.53680736e-01
6.53819084e-01 -5.99819481e-01 -3.13257664e-01 -9.38043892e-01
2.06094816e-01 -7.18951523e-01 -2.49417812e-01 9.74378884e-01
2.19692782e-01 -5.88160813e-01 8.77838612e-01 1.14990616e+00
-6.62294269e-01 -2.54122466e-01 -8.64180565e-01 -5.50381005e-01
-3.18339556e-01 -6.25142038e-01 4.42504972e-01 6.32201195e-01
-1.94320366e-01 -2.08374992e-01 8.24678615e-02 5.10961115e-01
1.02381516e+00 4.19447958e-01 1.36655247e+00 -1.41317153e+00
1.46247625e-01 5.13852574e-02 -3.39959115e-01 -1.16003478e+00
-3.65718424e-01 -5.47871888e-01 4.66418803e-01 -2.22923088e+00
-1.92745272e-02 -9.62255478e-01 8.05314600e-01 5.29997349e-01
1.90732583e-01 4.90443289e-01 -3.49317975e-02 5.64461708e-01
-6.06732786e-01 7.20728517e-01 1.83399284e+00 1.16466032e-02
-4.61137921e-01 -1.49534702e-01 -2.99814463e-01 7.89758205e-01
7.83826172e-01 -2.07054153e-01 -3.35229337e-01 -1.24952197e+00
2.21840277e-01 -2.36311570e-01 5.65232337e-01 -1.30294418e+00
1.09355465e-01 -3.63737345e-01 4.04264927e-01 -7.99963832e-01
1.06934595e+00 -1.07794976e+00 -1.35989457e-01 2.61878937e-01
2.04121128e-01 -1.89157918e-01 9.79822204e-02 5.88147461e-01
3.61124098e-01 3.56253944e-02 5.36612034e-01 -5.95552266e-01
-8.44982803e-01 6.39459312e-01 -2.93160647e-01 -1.58137590e-01
1.15086925e+00 -7.88638711e-01 -7.00925410e-01 -7.00502247e-02
-2.82053590e-01 2.97953546e-01 1.03255737e+00 3.25832307e-01
1.24716175e+00 -9.79123414e-01 -5.34027636e-01 1.81409910e-01
5.56825519e-01 1.05819321e+00 -1.41327158e-02 6.06185734e-01
-1.40192568e+00 4.83594760e-02 -1.23339809e-01 -9.01844919e-01
-1.21568143e+00 1.58857986e-01 3.08263928e-01 7.27119267e-01
-1.09169781e+00 1.02196145e+00 3.16107899e-01 -8.30142677e-01
3.45576316e-01 -1.06727564e+00 4.02747214e-01 -5.40022492e-01
4.92036402e-01 4.80975777e-01 2.51920640e-01 -2.39861891e-01
-1.52286857e-01 9.24018323e-01 4.01046425e-01 -2.66716689e-01
1.54408193e+00 -4.70800817e-01 -3.02126735e-01 5.85847735e-01
8.11228633e-01 7.05807433e-02 -1.96534753e+00 -1.96801633e-01
-4.71504688e-01 -1.05887532e+00 -5.37058525e-02 -6.70274734e-01
-9.91243899e-01 1.02147126e+00 4.60598528e-01 -1.89621717e-01
8.55187893e-01 2.03812018e-01 7.16358066e-01 7.24561512e-01
8.18687081e-01 -9.08826828e-01 5.35217702e-01 6.43043339e-01
1.09813046e+00 -1.37709296e+00 2.00774953e-01 -1.11506712e+00
-4.81791764e-01 1.30025542e+00 8.97720754e-01 2.48200297e-01
4.17631805e-01 4.76103842e-01 5.05900383e-01 -5.76098323e-01
-4.69832927e-01 -2.66736895e-01 -2.40475684e-01 1.24598444e+00
-1.35531006e-02 -2.51596361e-01 5.85543990e-01 -1.81817651e-01
-3.12737942e-01 9.75651294e-02 8.61601532e-01 1.51318729e+00
-6.68847144e-01 -7.45806813e-01 -7.56963849e-01 1.89366579e-01
2.05639020e-01 -5.33142034e-03 -6.37065351e-01 6.40581667e-01
1.63618270e-02 1.10508406e+00 7.59008303e-02 -3.73353437e-02
3.75700921e-01 -5.47096789e-01 8.35841835e-01 -8.82237136e-01
-2.47217372e-01 -2.01955825e-01 5.95797636e-02 -9.57924306e-01
-7.41951942e-01 -6.69693530e-01 -1.59111464e+00 -1.26138166e-01
-1.80051446e-01 -4.82734144e-01 1.07623851e+00 4.26538318e-01
2.13051587e-01 1.05381846e-01 6.04040682e-01 -1.61075246e+00
4.18628678e-02 -8.66742671e-01 -6.85587585e-01 2.37130269e-01
5.50704062e-01 -8.48741889e-01 -3.79268855e-01 8.63167197e-02] | [7.021670341491699, -1.9960604906082153] |
13a28cbc-c03a-45a4-990a-e15e7ba690d7 | hexatagging-projective-dependency-parsing-as | 2306.05477 | null | https://arxiv.org/abs/2306.05477v1 | https://arxiv.org/pdf/2306.05477v1.pdf | Hexatagging: Projective Dependency Parsing as Tagging | We introduce a novel dependency parser, the hexatagger, that constructs dependency trees by tagging the words in a sentence with elements from a finite set of possible tags. In contrast to many approaches to dependency parsing, our approach is fully parallelizable at training time, i.e., the structure-building actions needed to build a dependency parse can be predicted in parallel to each other. Additionally, exact decoding is linear in time and space complexity. Furthermore, we derive a probabilistic dependency parser that predicts hexatags using no more than a linear model with features from a pretrained language model, i.e., we forsake a bespoke architecture explicitly designed for the task. Despite the generality and simplicity of our approach, we achieve state-of-the-art performance of 96.4 LAS and 97.4 UAS on the Penn Treebank test set. Additionally, our parser's linear time complexity and parallelism significantly improve computational efficiency, with a roughly 10-times speed-up over previous state-of-the-art models during decoding. | ['Ryan Cotterell', 'Tianyu Liu', 'Afra Amini'] | 2023-06-08 | null | null | null | null | ['dependency-parsing'] | ['natural-language-processing'] | [-1.80592418e-01 6.58484578e-01 -2.53153116e-01 -7.95697927e-01
-1.41693056e+00 -8.70324135e-01 1.87268451e-01 3.26452136e-01
-4.82169122e-01 7.43194997e-01 2.81277776e-01 -7.74253070e-01
3.39669824e-01 -7.17271388e-01 -8.02421987e-01 -5.23292780e-01
-2.51964420e-01 7.73156881e-01 4.58426088e-01 -9.99291167e-02
-2.07956452e-02 1.89825937e-01 -9.60602105e-01 8.49277750e-02
5.54442525e-01 5.36819339e-01 3.25737953e-01 1.06023180e+00
-3.27422321e-01 5.32780588e-01 -5.41298687e-01 -6.97891593e-01
-1.08281478e-01 -3.33202213e-01 -1.13709438e+00 -2.97261536e-01
3.32492515e-02 -1.92801461e-01 -3.07291746e-01 8.74075055e-01
1.44282296e-01 -1.67294875e-01 1.83850989e-01 -6.00411892e-01
-3.96183699e-01 1.19048631e+00 -2.19812080e-01 4.07309592e-01
4.06159550e-01 -4.32112038e-01 1.48424423e+00 -6.44276679e-01
3.66790354e-01 1.17993438e+00 5.07319152e-01 6.07326925e-01
-1.24634826e+00 -5.35955369e-01 5.67520678e-01 -2.81188458e-01
-9.55985129e-01 -4.06292647e-01 2.52114445e-01 -1.24447010e-01
1.43747413e+00 -5.03178276e-02 4.86846209e-01 8.37203920e-01
5.88742018e-01 7.60540903e-01 9.39494491e-01 -7.31521368e-01
1.47733390e-01 -3.15151334e-01 6.96169555e-01 1.02962911e+00
2.09045291e-01 -2.49575302e-01 -5.66859126e-01 -1.39075741e-01
5.01333237e-01 -4.37349886e-01 1.98829368e-01 -5.89125268e-02
-9.08048630e-01 9.53887284e-01 -1.44008726e-01 2.53755569e-01
3.54960077e-02 1.43858805e-01 4.47187990e-01 1.03768840e-01
2.64546305e-01 2.03913778e-01 -1.08593595e+00 -3.70999068e-01
-7.34437764e-01 -6.74767643e-02 1.34356177e+00 1.14661920e+00
6.79330111e-01 -2.02525824e-01 1.37552559e-01 5.04575551e-01
3.26869488e-01 4.40813631e-01 2.97994912e-01 -8.34989190e-01
8.17145467e-01 1.41012892e-01 -4.90241572e-02 -4.43773344e-02
-6.10785484e-01 -1.55616879e-01 -3.11482668e-01 -1.55410424e-01
6.21171117e-01 -4.97394115e-01 -9.26974893e-01 1.80361807e+00
2.77543902e-01 -3.64897549e-02 3.92936587e-01 4.94495451e-01
4.70572352e-01 8.23274314e-01 4.43635315e-01 -1.99280605e-01
1.81117356e+00 -8.95876884e-01 -5.51316202e-01 -7.52269804e-01
1.11373174e+00 -7.22443461e-01 7.45349526e-01 3.90712947e-01
-1.38762105e+00 -2.62195975e-01 -1.01333070e+00 -9.46712345e-02
8.34332258e-02 -1.05418168e-01 1.14653337e+00 8.75757992e-01
-1.05755901e+00 6.21428728e-01 -1.37389684e+00 -1.00618124e-01
1.06069379e-01 5.80273390e-01 -5.98267853e-01 -1.28665611e-01
-1.02142954e+00 8.91795337e-01 7.24229395e-01 -1.01392806e-01
-4.55215901e-01 -3.31006974e-01 -1.16671491e+00 2.75275052e-01
2.22698376e-01 -3.09559375e-01 1.81744623e+00 -6.01180732e-01
-1.73035324e+00 7.45206594e-01 -6.21896565e-01 -4.94214863e-01
-2.23621532e-01 -5.32478869e-01 -3.63391966e-01 -1.16553325e-02
2.37607062e-01 3.57220948e-01 2.26454467e-01 -7.56941378e-01
-9.32495773e-01 -1.99237645e-01 2.04633966e-01 9.26127434e-02
-3.91250029e-02 6.54881418e-01 -8.01555037e-01 -4.00083333e-01
4.48307455e-01 -1.02024639e+00 -4.59086150e-01 -6.78977668e-01
-4.03711796e-01 -4.23453331e-01 2.58745760e-01 -8.72554362e-01
1.34400463e+00 -2.02486491e+00 6.65755048e-02 -1.04733126e-03
1.29310992e-02 2.96787143e-01 -1.69473767e-01 5.02879322e-01
-1.14335932e-01 2.46239632e-01 -5.01011968e-01 -5.77158868e-01
5.12516908e-02 8.77027810e-01 -1.80231228e-01 3.05707276e-01
3.88842076e-01 8.96064222e-01 -9.81264353e-01 -5.60687482e-01
-1.56904384e-01 1.64553657e-01 -6.05770111e-01 4.46493149e-01
-2.47223914e-01 3.19315225e-01 -5.99036217e-01 4.77135301e-01
5.34023941e-01 -1.52277112e-01 9.16822553e-01 3.87784183e-01
-2.29108408e-01 1.20032203e+00 -7.63754904e-01 1.91623950e+00
-6.92604184e-01 2.31613234e-01 1.10663235e-01 -9.77573276e-01
7.96841562e-01 4.69836295e-01 2.75362339e-02 -3.76875043e-01
2.06743926e-02 3.63048643e-01 1.08166523e-01 -5.36209978e-02
3.41369063e-01 -2.24753216e-01 -7.39831567e-01 6.24518812e-01
4.04144138e-01 7.16673136e-02 4.12868708e-01 2.67979801e-01
1.46489048e+00 3.29502821e-01 5.24094224e-01 -2.33381212e-01
2.45181009e-01 1.01217779e-03 9.60995078e-01 6.43708527e-01
4.94075194e-02 2.28084758e-01 8.95610154e-01 -4.92782474e-01
-8.43840122e-01 -1.16687131e+00 -2.30076969e-01 1.48468387e+00
-2.89310217e-01 -7.30651021e-01 -8.75206947e-01 -1.10529590e+00
-4.36387390e-01 7.90770650e-01 -3.29101533e-01 3.50851625e-01
-1.36773598e+00 -8.40959191e-01 6.68983638e-01 9.03161049e-01
1.25214115e-01 -1.10443163e+00 -4.51289982e-01 7.23464847e-01
-1.37510393e-02 -1.56603169e+00 -2.24787757e-01 1.00606441e+00
-9.75793540e-01 -9.21937883e-01 5.61519489e-02 -1.31915975e+00
6.76957309e-01 -1.75096944e-01 1.59427476e+00 -1.22976815e-02
1.68505803e-01 -3.01617652e-01 -6.19571507e-01 -1.24350421e-01
-7.17882633e-01 2.57960826e-01 -2.26374820e-01 -7.87787139e-01
5.79003930e-01 -5.80988169e-01 -1.29065603e-01 -1.42983228e-01
-4.70843017e-01 3.58711071e-02 7.76233613e-01 1.05879045e+00
5.33600152e-01 -7.52732158e-02 1.50409326e-01 -1.29629743e+00
9.78934914e-02 -5.65564811e-01 -7.97811270e-01 2.28846654e-01
-2.92794317e-01 5.17750084e-01 8.38576078e-01 2.16379296e-02
-1.20214403e+00 5.51940918e-01 -7.39423156e-01 4.56273288e-01
-2.51798719e-01 5.55493355e-01 -3.51076990e-01 2.78067738e-01
1.13036059e-01 2.42895056e-02 -4.60071623e-01 -7.13606179e-01
3.81963968e-01 4.20571357e-01 6.37822568e-01 -8.54269266e-01
6.69565558e-01 1.58375304e-03 2.21750350e-04 -4.12686229e-01
-1.30248141e+00 -3.19353223e-01 -1.03861165e+00 5.31967103e-01
9.75728929e-01 -9.80713010e-01 -5.05643368e-01 2.27161705e-01
-1.50657594e+00 -3.77033472e-01 3.35159115e-02 4.58084494e-01
-2.93227941e-01 5.46181440e-01 -1.14145696e+00 -6.51167631e-01
-4.38890189e-01 -8.93361926e-01 9.73355055e-01 1.02203213e-01
-2.15893701e-01 -1.02001822e+00 2.07054317e-01 1.46679550e-01
-3.14072281e-01 -1.63340583e-01 1.11560631e+00 -1.00040436e+00
-3.71751726e-01 -1.79173097e-01 -4.17454727e-02 3.45618308e-01
-1.84082583e-01 -2.08841011e-01 -7.27671206e-01 -1.30479142e-01
-3.50331776e-02 -2.64006972e-01 6.06595635e-01 1.55440241e-01
8.76148999e-01 -2.08740622e-01 -4.60901916e-01 4.67891753e-01
1.13570404e+00 2.10921183e-01 3.96392405e-01 2.64384985e-01
5.00940561e-01 5.22250593e-01 6.96729481e-01 3.15479964e-01
6.42920434e-01 3.52870077e-01 1.38248712e-01 1.86471000e-01
7.78022036e-02 -4.69816566e-01 5.86018026e-01 1.30287063e+00
1.81866914e-01 -3.23261142e-01 -1.09114468e+00 5.40897012e-01
-1.72092593e+00 -3.77022117e-01 -2.78419644e-01 2.02192640e+00
1.03347564e+00 5.15713155e-01 -1.98567063e-01 1.02465607e-01
6.07634723e-01 9.44065750e-02 -5.15437163e-02 -1.02223682e+00
6.90409392e-02 9.93977129e-01 9.14573073e-01 6.99634373e-01
-1.21593618e+00 1.51754630e+00 7.04169750e+00 4.59752351e-01
-6.78726971e-01 2.58538336e-01 5.11383355e-01 2.11445719e-01
-2.21178129e-01 3.64867181e-01 -1.40803587e+00 2.77792007e-01
1.66995633e+00 1.85673326e-01 2.55060554e-01 9.35733259e-01
-3.19979787e-01 -7.07560703e-02 -1.18030453e+00 3.80463898e-01
-1.60973445e-01 -1.08582747e+00 -3.68827075e-01 -1.95449777e-02
2.12201312e-01 3.34414899e-01 -4.92482096e-01 4.61352646e-01
1.21187246e+00 -8.83146942e-01 6.96673155e-01 -2.59180158e-01
6.81439698e-01 -8.67369831e-01 8.18422973e-01 5.36332726e-01
-1.27908826e+00 -1.78642236e-02 -4.78285372e-01 -3.05011362e-01
7.03935325e-01 6.84793353e-01 -7.14457929e-01 4.76266503e-01
5.48580945e-01 2.19557658e-01 -1.55643240e-01 5.01591682e-01
-1.15889776e+00 1.20675647e+00 -4.90554184e-01 -5.59946299e-02
4.04161543e-01 5.52894063e-02 1.29112169e-01 1.72565866e+00
2.87480444e-01 4.48467880e-01 4.16657627e-01 -7.29642361e-02
-1.20918088e-01 -2.99969967e-03 -1.21152006e-01 -3.93844731e-02
8.14342320e-01 1.16699350e+00 -6.95044696e-01 -3.66392255e-01
-6.76174700e-01 1.00847661e+00 8.57523739e-01 -1.57561049e-01
-8.24665844e-01 -4.03926581e-01 6.40052378e-01 -3.66283625e-01
8.75920475e-01 -6.91480279e-01 -4.03723598e-01 -1.07084107e+00
1.60234034e-01 -5.15612304e-01 5.25301576e-01 -2.97750175e-01
-1.07593083e+00 8.51239204e-01 -7.96325356e-02 -5.62315762e-01
-5.53066850e-01 -8.69409740e-01 -5.75416982e-01 1.08090961e+00
-1.65583301e+00 -9.98820305e-01 4.27703470e-01 1.71420038e-01
5.19527078e-01 1.92838177e-01 1.62341619e+00 -8.47762898e-02
-6.09182894e-01 6.61244094e-01 -1.24251612e-01 5.55402040e-01
4.80547905e-01 -1.50678480e+00 1.47564447e+00 1.20025277e+00
4.29858625e-01 6.47742391e-01 4.47965443e-01 -4.98483300e-01
-1.54750049e+00 -8.71388376e-01 1.75354290e+00 -5.10454297e-01
1.02128661e+00 -7.74803460e-01 -9.25125003e-01 1.12319469e+00
1.66736826e-01 1.71976730e-01 1.00848484e+00 7.15511024e-01
-7.81077385e-01 1.72799155e-01 -7.51185060e-01 3.13278317e-01
1.14837408e+00 -3.38563204e-01 -1.05855882e+00 1.79010719e-01
8.25650156e-01 -7.59853482e-01 -1.03504777e+00 8.61215293e-02
5.87454379e-01 -7.33496070e-01 5.55586576e-01 -7.74824321e-01
3.07456613e-01 2.80875657e-02 -2.18909293e-01 -1.20168722e+00
-6.86198533e-01 -7.99093068e-01 -1.84594542e-01 1.26275611e+00
7.81970024e-01 -6.39450669e-01 9.06233370e-01 7.47567654e-01
-5.69075108e-01 -7.04655588e-01 -1.09525430e+00 -6.57244444e-01
3.92702192e-01 -6.72571599e-01 5.63462675e-01 6.12714708e-01
3.46814990e-01 6.94951653e-01 -1.38481125e-01 5.18429399e-01
4.74514633e-01 1.69188485e-01 4.44447964e-01 -1.22104406e+00
-7.07638264e-01 1.47745699e-01 -3.32945287e-02 -1.53881836e+00
7.15308845e-01 -7.94243813e-01 5.46830297e-01 -1.53760898e+00
7.59261698e-02 -8.39659870e-01 -3.75072241e-01 9.35602367e-01
-3.71107101e-01 -9.61866900e-02 1.99670941e-01 1.31877273e-01
-5.26543438e-01 4.86026220e-02 8.12535822e-01 3.60157460e-01
3.24380472e-02 -7.44358152e-02 -8.89374077e-01 9.03795540e-01
1.04366446e+00 -1.00948286e+00 -9.14400294e-02 -8.11326683e-01
7.82972351e-02 6.34831190e-01 -3.39911222e-01 -6.87917233e-01
9.70910639e-02 -1.79077834e-02 1.13127649e-01 -4.72622722e-01
3.71541977e-01 -3.00616711e-01 -3.38057190e-01 3.56857598e-01
7.18497962e-04 3.61960322e-01 2.93160379e-01 4.13600206e-01
-1.70062900e-01 -5.92693508e-01 5.44816136e-01 -2.49018565e-01
-7.29717255e-01 1.20976575e-01 -7.26971447e-01 2.38251656e-01
7.05016792e-01 1.78379580e-01 -1.59170076e-01 2.19197690e-01
-6.37495458e-01 8.16490501e-02 8.15021396e-02 1.70050681e-01
1.37291804e-01 -6.59843981e-01 -6.98408544e-01 2.12434560e-01
-1.30130440e-01 2.71355808e-01 -1.84750512e-01 1.87814042e-01
-5.22189915e-01 6.26618266e-01 4.91113812e-02 -2.06118986e-01
-1.32459092e+00 2.51327544e-01 -2.90651023e-01 -9.16236758e-01
-7.90718019e-01 1.09637105e+00 1.41086102e-01 -4.39865261e-01
-1.26600832e-01 -4.31788802e-01 -4.96953353e-03 -3.86164039e-01
5.29500902e-01 -3.07572842e-01 1.90282509e-01 -4.73025292e-01
-5.85834324e-01 2.32493907e-01 -3.32714707e-01 -2.77576298e-01
1.39643693e+00 2.25518614e-01 -3.53144765e-01 2.06356600e-01
9.27726686e-01 3.65741402e-01 -1.25845289e+00 -1.45671770e-01
4.56061006e-01 -4.76805773e-03 -1.41376570e-01 -8.85049820e-01
-6.18261993e-01 6.82303369e-01 -2.19019949e-01 2.12347303e-02
8.88097405e-01 3.66565287e-01 1.16084385e+00 6.95370138e-01
5.64486504e-01 -9.08850789e-01 -4.58204061e-01 1.12216067e+00
1.26436904e-01 -1.05455506e+00 -1.90257207e-01 -8.71869206e-01
-6.12093270e-01 1.14503610e+00 3.05004239e-01 -3.49414915e-01
5.95706403e-01 1.04335272e+00 1.16092697e-01 1.63079366e-01
-9.75450099e-01 -1.37672544e-01 -2.61265069e-01 5.22858500e-01
8.87391746e-01 4.39493835e-01 -5.06735861e-01 1.05963421e+00
-5.34631193e-01 -3.58543754e-01 4.31806594e-01 1.27505326e+00
-7.25558162e-01 -1.94897711e+00 -1.80233449e-01 1.64657280e-01
-9.45032001e-01 -5.54541230e-01 6.36199117e-02 7.89580047e-01
-2.24827826e-01 1.10006487e+00 1.70918137e-01 -2.52530366e-01
2.20890239e-01 4.83673155e-01 5.63050032e-01 -1.17144334e+00
-6.43169999e-01 -1.23666046e-04 7.96802163e-01 -5.50780773e-01
-1.34797007e-01 -7.39833236e-01 -1.84251857e+00 -1.96061075e-01
-2.19756261e-01 6.64210141e-01 7.92029381e-01 1.26914716e+00
1.33223012e-01 3.08304429e-01 4.64085430e-01 -5.78460991e-01
-5.41144788e-01 -8.67997110e-01 -4.91286516e-01 -2.23168686e-01
-1.05875127e-01 -2.73258865e-01 -4.46295813e-02 5.47873564e-02] | [10.338915824890137, 9.66922378540039] |
83ac2a42-a015-4b6c-9125-18b98fcda352 | sparse-recovery-via-bootstrapping | null | null | https://openreview.net/forum?id=BUCHknhWq8D | https://openreview.net/pdf?id=BUCHknhWq8D | Sparse Recovery via Bootstrapping: Collaborative or Independent? | Sparse regression problems have traditionally been solved using all available measurements simultaneously. However, this approach fails in challenging scenarios such as when the noise level is high or there are missing data / adversarial samples.
We propose JOBS (Joint-Sparse Optimization via Bootstrap Samples) -- a \emph{collaborative} sparse-regression framework on bootstrapped samples from the pool of available measurements via a joint-sparse constraint to ensure support consistency. In comparison to traditional bagging which solves sub-problems in an \emph{independent} fashion across bootstrapped samples, JOBS achieves state-of-the-art performance with the added advantage of having a sparser solution while requiring a lower number of observation samples.
Analysis of theoretical performance limits is employed to determine critical optimal parameters: the number of bootstrap samples $K$ and the number of elements $L$ in each bootstrap sample. Theoretical results indicate a better bound than Bagging (i.e. higher probability of achieving the same or better performance). Simulation results are used to validate this parameter selection. JOBS is robust to adversarial samples that fool the baseline method, as shown by better generalization in an image reconstruction task where the adversary has similar occlusions or alignment as the test sample. Furthermore, JOBS also improves discriminative performance in a facial recognition task in a sparse-representation-based classification setting. | ['Anonymous'] | 2021-01-01 | null | null | null | null | ['sparse-representation-based-classification'] | ['computer-vision'] | [ 2.85411656e-01 -1.03677614e-02 -2.58675069e-01 -5.16069651e-01
-1.53197336e+00 -2.30469123e-01 1.17521271e-01 -3.71603698e-01
-2.88533330e-01 9.51146841e-01 1.01980507e-01 1.06548681e-03
-1.90805316e-01 -4.07714784e-01 -9.62077677e-01 -1.14464235e+00
-3.10609527e-02 3.83702964e-01 -3.17267865e-01 5.91134243e-02
-6.82245940e-02 4.79034543e-01 -1.47486496e+00 4.23303172e-02
7.22414732e-01 1.01736832e+00 -1.49756372e-02 3.57254922e-01
3.10553133e-01 5.88254809e-01 -6.89663112e-01 -4.18250471e-01
7.92401433e-01 -5.87178886e-01 -9.49663371e-02 4.73040819e-01
8.54473829e-01 -3.85032564e-01 -2.53093064e-01 1.14016807e+00
6.32779300e-01 1.24408521e-01 4.38224792e-01 -1.35094094e+00
-2.24011883e-01 3.48067850e-01 -1.00463009e+00 9.19457525e-02
2.54360765e-01 -1.32626966e-01 7.13416100e-01 -1.05395353e+00
4.21423107e-01 1.08072674e+00 8.74545217e-01 4.20833945e-01
-1.61459696e+00 -1.11847270e+00 -3.80754173e-02 -1.32636949e-01
-1.52883172e+00 -9.52883959e-01 8.42762291e-01 -1.91288278e-01
4.41775769e-01 3.92833829e-01 1.38344005e-01 1.20117342e+00
-1.02475338e-01 6.15094543e-01 1.30227780e+00 -3.29760551e-01
4.61467534e-01 3.84112626e-01 -7.35313212e-03 6.44309521e-01
5.78959703e-01 1.15219034e-01 -7.64693856e-01 -6.36171401e-01
5.76648474e-01 -7.69817382e-02 -3.36767524e-01 -5.62634587e-01
-4.86605853e-01 1.15912795e+00 1.53645262e-01 -3.41597535e-02
-4.53423142e-01 2.02263277e-02 1.80669829e-01 3.99342775e-01
4.84625816e-01 1.27783924e-01 -2.21215755e-01 1.90284178e-01
-1.55990684e+00 3.78498644e-01 8.17198634e-01 9.68424082e-01
6.10011637e-01 7.61242032e-01 4.16734256e-02 9.25131738e-01
1.22741424e-02 9.51903403e-01 3.19092929e-01 -1.11562395e+00
8.28635454e-01 -2.15396807e-01 3.18192333e-01 -1.17551470e+00
-2.86270957e-02 -7.31146157e-01 -9.74957466e-01 3.65072936e-01
5.23238122e-01 -2.86512524e-01 -7.98346102e-01 1.92200506e+00
3.17337632e-01 5.26371539e-01 -5.29624522e-02 9.36551213e-01
4.37622517e-01 4.98645604e-01 -3.96812469e-01 -6.64183915e-01
9.74577248e-01 -4.44249690e-01 -5.93487382e-01 -3.94313514e-01
1.84021652e-01 -6.74820423e-01 7.52993941e-01 6.49648786e-01
-1.03637373e+00 -4.29074019e-01 -9.42995071e-01 4.30094868e-01
3.12075585e-01 2.00467587e-01 4.21252251e-01 1.11541331e+00
-7.97682106e-01 3.72278571e-01 -7.47684836e-01 1.01647139e-01
6.86481416e-01 5.51708877e-01 -7.51854956e-01 -5.29201210e-01
-5.18521965e-01 5.01658916e-01 -1.32914990e-01 1.28514007e-01
-8.50349069e-01 -6.53220236e-01 -1.00469208e+00 -7.25004002e-02
3.03513944e-01 -4.15822744e-01 7.97762394e-01 -1.05264258e+00
-1.10473382e+00 6.09446645e-01 -6.54854596e-01 -6.69110835e-01
4.92191136e-01 -3.33349884e-01 -2.09186271e-01 2.80781597e-01
3.09431404e-01 3.04007560e-01 1.55413139e+00 -1.37281644e+00
-2.36769840e-01 -5.33505857e-01 -4.15696442e-01 -1.13524608e-01
-2.06837609e-01 -1.83408745e-02 -3.12543839e-01 -8.33039939e-01
3.52754533e-01 -1.19211686e+00 -6.06965601e-01 -2.55036443e-01
-4.27287877e-01 5.59903979e-01 8.00935924e-01 -7.25796580e-01
7.92406738e-01 -2.37122226e+00 8.73794258e-02 6.59087539e-01
1.17512628e-01 6.34909198e-02 -2.06339702e-01 2.56196052e-01
-3.39426309e-01 -1.94398716e-01 -3.05032253e-01 -7.56284714e-01
-3.74281883e-01 2.32139617e-01 -5.20835757e-01 9.39459503e-01
1.52625395e-02 3.26467186e-01 -5.14605165e-01 -3.16915929e-01
2.01818552e-02 3.53460222e-01 -7.43309259e-01 2.32220888e-01
3.11950773e-01 6.57409787e-01 -3.69691879e-01 7.13042855e-01
9.11604524e-01 -2.39073455e-01 1.56439081e-01 -3.73610221e-02
6.34078026e-01 -4.49370712e-01 -1.57162046e+00 1.32991564e+00
-5.38442612e-01 5.09531677e-01 7.56779253e-01 -1.33705235e+00
1.11221063e+00 3.46419185e-01 5.58391392e-01 -3.79496902e-01
1.32460982e-01 2.43007258e-01 -2.23572627e-01 -3.35536897e-01
1.98921263e-01 -3.76260519e-01 -5.80200292e-02 8.67908821e-02
2.37522591e-02 -2.06172183e-01 -1.52727157e-01 2.09013462e-01
1.10697794e+00 -2.33800098e-01 3.47927898e-01 -1.16270937e-01
1.36193573e-01 -3.23450118e-01 9.43267822e-01 1.05184829e+00
4.21666913e-02 7.85464525e-01 3.05789441e-01 3.10165063e-02
-1.01146460e+00 -8.81642878e-01 -3.27284396e-01 7.82101035e-01
-7.20549673e-02 -2.32051224e-01 -6.79786623e-01 -5.11623323e-01
3.46186310e-01 7.79165387e-01 -6.69263661e-01 1.46027878e-01
-5.22319376e-01 -8.25751424e-01 3.30711663e-01 5.10463238e-01
2.74003446e-01 -6.02058113e-01 -3.42211366e-01 7.75360465e-02
-7.94487968e-02 -1.46849108e+00 -2.88261026e-01 2.37387970e-01
-8.92914414e-01 -9.15802240e-01 -5.09437442e-01 -4.89818811e-01
1.09033692e+00 5.40439188e-01 7.83956647e-01 -3.61511618e-01
-3.18599969e-01 5.13589740e-01 -2.75195062e-01 -3.46128136e-01
-1.69777945e-01 -4.27590549e-01 2.29925901e-01 3.23897183e-01
6.73158932e-03 -7.57460594e-01 -2.61757135e-01 3.70313913e-01
-7.58303940e-01 -4.35295790e-01 4.43980038e-01 1.34390080e+00
6.87365294e-01 1.07062154e-01 7.23267853e-01 -1.03609669e+00
3.41440797e-01 -7.64197469e-01 -6.85777903e-01 -1.93618029e-01
-4.56705242e-01 -2.03533724e-01 7.64804006e-01 -5.46479225e-01
-7.31516600e-01 1.88037395e-01 -2.53389310e-02 -1.00152409e+00
1.58473663e-02 4.16431248e-01 -2.34958246e-01 -3.25092345e-01
6.63759232e-01 3.29542488e-01 4.15729076e-01 -4.39374268e-01
1.51442766e-01 4.61310387e-01 4.45103168e-01 -6.51924789e-01
1.02776194e+00 6.20327532e-01 3.36908013e-01 -1.06625843e+00
-8.64517868e-01 -4.64998573e-01 -1.49559081e-01 1.21540442e-01
1.71219930e-01 -1.30436623e+00 -4.28734154e-01 2.09547356e-01
-5.21710157e-01 -3.05833146e-02 -4.82345402e-01 6.32431149e-01
-7.22252011e-01 4.58190978e-01 -2.33544707e-01 -1.08521914e+00
-1.86691433e-01 -1.23994172e+00 1.16098833e+00 4.80035096e-02
-1.68172061e-01 -4.92836863e-01 -3.22717607e-01 7.04921305e-01
2.48573020e-01 5.16325533e-01 3.99719477e-01 -7.79949486e-01
-4.09983903e-01 -7.31931567e-01 1.88139931e-03 7.02885747e-01
-3.78500558e-02 -4.13847923e-01 -9.46912766e-01 -8.83016825e-01
3.80888760e-01 -5.47618151e-01 7.57142544e-01 7.21987605e-01
1.39247847e+00 -6.47127926e-01 -1.92999110e-01 8.41554105e-01
1.44041908e+00 -5.97557388e-02 5.46131968e-01 -1.57431990e-01
4.58853245e-01 4.82874870e-01 7.14659393e-01 7.46965349e-01
-2.51879215e-01 7.67702281e-01 4.64133203e-01 -4.76532392e-02
1.25718668e-01 -1.84595972e-01 3.82646054e-01 5.14273465e-01
2.63506025e-01 3.70788835e-02 -3.50555092e-01 5.57084620e-01
-1.67373335e+00 -1.23769248e+00 1.27587795e-01 2.54784846e+00
7.09655643e-01 -1.86414947e-03 3.51479679e-01 2.73925453e-01
5.98180890e-01 2.20534176e-01 -5.65306842e-01 -1.03630416e-01
-2.49779209e-01 6.52209520e-01 7.56266594e-01 5.48852503e-01
-8.98946404e-01 5.77242374e-01 6.55549479e+00 1.11638999e+00
-9.78467822e-01 3.33755553e-01 7.64390349e-01 -6.44329309e-01
-9.46967080e-02 -4.54458036e-02 -1.06933796e+00 6.76015019e-01
7.83044934e-01 7.75951296e-02 4.78112906e-01 9.90572572e-01
1.96038455e-01 -2.30938122e-01 -7.94656694e-01 1.25745094e+00
5.02442539e-01 -1.21911299e+00 -3.64675194e-01 2.01691896e-01
7.89498031e-01 1.43114645e-02 2.58951515e-01 1.04135670e-01
3.41793478e-01 -1.11648512e+00 5.40445566e-01 1.89981461e-01
9.49109435e-01 -6.57841921e-01 6.89192712e-01 5.27658105e-01
-7.79349267e-01 -4.23249483e-01 -4.39795405e-01 2.15620428e-01
4.34155725e-02 1.05162168e+00 -8.06149781e-01 4.55229223e-01
5.96482098e-01 3.31754237e-01 -2.48168886e-01 9.31961179e-01
-9.09830481e-02 1.14135051e+00 -6.17216110e-01 3.16246390e-01
-8.24047625e-02 -3.99966806e-01 7.75562465e-01 1.08966923e+00
5.02146840e-01 9.47694853e-02 3.94641817e-01 3.86206061e-01
5.37395887e-02 1.12436496e-01 -7.75846958e-01 3.78125817e-01
7.82418549e-01 9.74413991e-01 -3.16018850e-01 -3.37384604e-02
-3.00794721e-01 7.28945136e-01 2.27331385e-01 5.41141689e-01
-7.76731670e-01 -7.49642849e-02 5.94634354e-01 2.43622050e-01
6.86915696e-01 -1.74967632e-01 -3.43596339e-01 -1.05810368e+00
1.67262241e-01 -1.39874899e+00 3.13717037e-01 -5.04180551e-01
-1.30554152e+00 4.12829250e-01 5.13905697e-02 -1.25059211e+00
-4.56851661e-01 -2.06063166e-01 -4.05243874e-01 9.03349459e-01
-1.00949883e+00 -1.12320375e+00 -2.39329979e-01 6.66503251e-01
4.47898000e-01 -5.32923758e-01 8.99540782e-01 1.19441271e-01
-7.45514691e-01 9.86156166e-01 4.71301287e-01 8.05734545e-02
5.23675382e-01 -7.97004104e-01 -1.11810021e-01 1.04608881e+00
3.32055509e-01 5.79086065e-01 1.00876749e+00 -5.84635079e-01
-1.50258672e+00 -9.03092086e-01 2.96621114e-01 -2.98382938e-01
4.41653937e-01 -4.60270673e-01 -6.91447556e-01 7.20928371e-01
-3.30762476e-01 5.49961388e-01 1.03327572e+00 2.42674693e-01
-6.05400801e-01 -4.21854109e-01 -1.62703192e+00 2.19491020e-01
8.29455912e-01 -3.55917603e-01 -2.33022839e-01 3.98651123e-01
2.08290771e-01 -4.59119856e-01 -8.12217772e-01 4.98462975e-01
6.34768486e-01 -1.11347866e+00 1.09663260e+00 -5.96318901e-01
9.63225439e-02 -1.46680534e-01 -7.22992539e-01 -1.15697086e+00
-1.84814364e-01 -8.70401382e-01 -2.33046174e-01 1.06975019e+00
1.61735415e-01 -7.49265194e-01 1.17911386e+00 4.39141184e-01
3.43811721e-01 -8.64211261e-01 -1.40291786e+00 -1.12209260e+00
-2.12250471e-01 -3.26448172e-01 2.49063969e-01 8.05278838e-01
-4.04614419e-01 7.02575669e-02 -9.03260410e-01 4.15523559e-01
1.04572308e+00 1.54795125e-01 1.25847888e+00 -8.68190050e-01
-7.13847816e-01 1.53831646e-01 -6.15821958e-01 -7.88046300e-01
4.38366592e-01 -5.92243135e-01 -6.64160922e-02 -8.17728102e-01
1.77853018e-01 -7.26644933e-01 -8.33530426e-02 4.79726017e-01
-2.22850963e-01 3.58623177e-01 3.04702073e-01 2.15553656e-01
-1.66528672e-01 4.71463680e-01 7.14650452e-01 6.09766841e-02
-1.94632390e-03 3.66192222e-01 -8.56672287e-01 5.29411435e-01
6.01527870e-01 -7.16139913e-01 -4.78814751e-01 -5.80552556e-02
-2.63398141e-01 5.71971714e-01 3.92279446e-01 -1.08417094e+00
-4.22898456e-02 -1.40922099e-01 5.56933880e-01 -2.72576123e-01
8.22386205e-01 -1.04148710e+00 4.08346981e-01 2.35748544e-01
-1.84150755e-01 -1.82645321e-01 2.01030210e-01 9.18226480e-01
-2.31746897e-01 -4.30025756e-01 1.05532706e+00 2.18096286e-01
4.60907705e-02 2.13745832e-01 6.63528889e-02 7.82275572e-02
1.09634781e+00 -4.47445452e-01 -6.85204044e-02 -7.85129130e-01
-7.51286447e-01 -1.13668144e-01 3.20672661e-01 -2.80009229e-02
6.12605393e-01 -1.13552630e+00 -1.05687451e+00 5.85880458e-01
-1.13881797e-01 -9.49920863e-02 1.93698719e-01 8.98716986e-01
-6.32404611e-02 -9.63243917e-02 4.79111671e-02 -5.45012653e-01
-1.56116295e+00 3.90325963e-01 3.86984311e-02 -9.45396498e-02
-5.27952611e-01 1.11894608e+00 -2.68385224e-02 -1.22427911e-01
3.12303066e-01 2.28038415e-01 3.68935496e-01 5.83419278e-02
5.12290061e-01 3.64376724e-01 1.68152079e-01 -7.76013017e-01
-1.53865337e-01 3.95569742e-01 -6.88495785e-02 -1.74945042e-01
1.43747735e+00 1.30722642e-01 1.00631997e-01 1.61250681e-01
1.31099761e+00 7.38149464e-01 -1.36288905e+00 -3.24600637e-01
-4.66970801e-01 -1.09217262e+00 3.36869620e-02 -4.45220828e-01
-1.32786107e+00 4.37668145e-01 5.27405441e-01 3.92129235e-02
1.12450469e+00 -2.66607612e-01 5.67236543e-01 2.24476084e-01
6.50198996e-01 -7.52892196e-01 2.11800896e-02 -1.10888913e-01
1.01559317e+00 -1.25877213e+00 4.26325500e-01 -6.08591795e-01
-6.27558708e-01 6.05046988e-01 3.67345870e-01 -5.18977702e-01
7.48970568e-01 5.22069693e-01 -1.64017290e-01 1.15781866e-01
-4.23654467e-01 1.52963892e-01 1.26523927e-01 6.94140673e-01
-9.73316580e-02 1.02489531e-01 3.98225375e-02 6.20688140e-01
-3.78106713e-01 -1.11259192e-01 3.79685521e-01 8.40498626e-01
-3.34350914e-01 -8.85375500e-01 -8.79193425e-01 9.35213804e-01
-6.14299655e-01 1.24413393e-01 3.37932855e-02 6.61659539e-01
-1.10934794e-01 1.17361653e+00 -1.77776590e-01 -2.59019822e-01
1.37303129e-01 -5.80133311e-02 5.23332059e-01 -5.48590481e-01
-3.53810161e-01 2.47591794e-01 1.63998812e-01 -7.95933485e-01
-2.84640610e-01 -1.05394411e+00 -6.73318207e-01 -3.70084971e-01
-6.22911632e-01 3.47546905e-01 6.63504303e-01 7.89455295e-01
3.39853019e-01 -3.30966413e-02 1.12079132e+00 -8.29248309e-01
-1.16495752e+00 -7.83002853e-01 -9.71287310e-01 3.78581911e-01
3.67341042e-01 -6.71245039e-01 -7.04164505e-01 -8.28676969e-02] | [7.076717853546143, 4.481830596923828] |
99e5a717-9008-44f3-820f-e82ff53dfa4e | revisiting-facial-key-point-detection-an | 2205.07121 | null | https://arxiv.org/abs/2205.07121v1 | https://arxiv.org/pdf/2205.07121v1.pdf | Revisiting Facial Key Point Detection: An Efficient Approach Using Deep Neural Networks | Facial landmark detection is a widely researched field of deep learning as this has a wide range of applications in many fields. These key points are distinguishing characteristic points on the face, such as the eyes center, the eye's inner and outer corners, the mouth center, and the nose tip from which human emotions and intent can be explained. The focus of our work has been evaluating transfer learning models such as MobileNetV2 and NasNetMobile, including custom CNN architectures. The objective of the research has been to develop efficient deep learning models in terms of model size, parameters, and inference time and to study the effect of augmentation imputation and fine-tuning on these models. It was found that while augmentation techniques produced lower RMSE scores than imputation techniques, they did not affect the inference time. MobileNetV2 architecture produced the lowest RMSE and inference time. Moreover, our results indicate that manually optimized CNN architectures performed similarly to Auto Keras tuned architecture. However, manually optimized architectures yielded better inference time and training curves. | ['Sabeesh Ethiraj', 'Bharath Kumar Bolla', 'Prathima Dileep'] | 2022-05-14 | null | null | null | null | ['facial-landmark-detection'] | ['computer-vision'] | [-5.68937004e-01 3.18940103e-01 -3.10241520e-01 -6.07855082e-01
-8.14859495e-02 -1.87276810e-01 4.58990812e-01 -1.86895877e-01
-4.57773477e-01 5.11776507e-01 4.90236543e-02 -8.89996439e-02
-1.09655559e-01 -5.33200145e-01 -4.51969147e-01 -5.10576427e-01
3.62900607e-02 1.43675908e-01 -5.04221201e-01 -9.63351205e-02
3.20720255e-01 6.02058172e-01 -1.52073312e+00 -8.69282037e-02
3.20231944e-01 1.17664981e+00 -5.65279603e-01 3.11851799e-01
1.51434720e-01 4.33713377e-01 -5.27884007e-01 -7.77093887e-01
8.45397562e-02 -1.12502864e-02 -4.10592198e-01 -3.11693549e-01
7.02023208e-01 -5.17816782e-01 -8.23910907e-02 6.93031371e-01
8.12532604e-01 2.34647736e-01 4.22327280e-01 -1.51339257e+00
-4.90647465e-01 2.47886643e-01 -5.41621506e-01 4.16651275e-03
1.74781695e-01 3.58413979e-02 7.74193764e-01 -8.44683886e-01
3.53234380e-01 1.30988324e+00 1.03734946e+00 7.59682119e-01
-1.16322672e+00 -1.13339877e+00 -2.46914163e-01 -1.72748417e-01
-1.57531917e+00 -9.87436712e-01 5.65652490e-01 -2.25019291e-01
1.02617359e+00 -6.80333152e-02 2.71846801e-01 1.17345631e+00
4.23707157e-01 2.67787308e-01 1.08063543e+00 -2.58369952e-01
1.89903811e-01 6.56093121e-01 1.17185026e-01 8.60065818e-01
1.40220165e-01 3.33585404e-02 -5.21314919e-01 -3.38739723e-01
7.88161337e-01 -1.08222000e-01 1.28558725e-01 1.15766443e-01
-5.20266414e-01 1.02568901e+00 2.64816016e-01 1.52266160e-01
-5.20846963e-01 3.04392338e-01 4.25644666e-01 1.09220318e-01
5.16669869e-01 2.66328782e-01 -7.95136690e-01 -2.09192336e-01
-8.72279763e-01 -1.59116238e-02 8.38796020e-01 5.80322862e-01
7.51372397e-01 1.35954410e-01 1.39197782e-02 8.65908146e-01
5.81645548e-01 3.29682291e-01 3.91805410e-01 -1.17063606e+00
1.86424479e-01 7.21791625e-01 -1.35992557e-01 -1.25782263e+00
-5.65763175e-01 -5.30943334e-01 -6.90447330e-01 1.77691817e-01
4.00665343e-01 -7.96553969e-01 -9.12424624e-01 1.88945639e+00
1.92291617e-01 2.44959936e-01 -8.14264193e-02 6.96909070e-01
1.07489622e+00 2.69425601e-01 3.42179716e-01 2.44985580e-01
1.21681595e+00 -4.88214135e-01 -5.87025583e-01 -3.73814493e-01
6.36749566e-01 -8.95100713e-01 8.80688488e-01 5.37237860e-02
-1.13335598e+00 -6.37635112e-01 -7.61769772e-01 -2.08832011e-01
-3.38727266e-01 4.72787291e-01 8.84577930e-01 1.00588906e+00
-1.17110586e+00 6.17536306e-01 -8.18583667e-01 -4.56480980e-01
5.64469337e-01 6.95867658e-01 -4.74967539e-01 3.15900117e-01
-8.76983643e-01 1.11276913e+00 1.44903641e-02 3.33227217e-01
-5.43051362e-01 -7.01065481e-01 -8.16655636e-01 1.81509525e-01
-4.89357561e-02 -4.03992623e-01 1.05886996e+00 -1.26257682e+00
-1.45514905e+00 9.58451748e-01 -3.89401555e-01 -3.61418605e-01
1.99310288e-01 -1.16584703e-01 -3.76280546e-01 -3.22415590e-01
-1.97159529e-01 1.15463889e+00 7.71712422e-01 -9.03607070e-01
-2.78895080e-01 -6.46948278e-01 -1.21500574e-01 -1.83415130e-01
-3.47942680e-01 -2.80347466e-02 -4.00095731e-01 -2.78048217e-01
2.06854083e-02 -1.16413319e+00 1.85011432e-01 7.89047629e-02
-1.87070936e-01 -4.10881758e-01 7.05863118e-01 -7.12156713e-01
7.80236840e-01 -2.27644706e+00 -5.50906181e-01 5.54282427e-01
1.42030776e-01 3.41534585e-01 -1.68653592e-01 1.87743738e-01
-1.68451741e-01 2.52785653e-01 4.62840855e-01 -4.31927145e-01
-1.62827801e-02 2.23196119e-01 1.32765472e-01 4.79182631e-01
2.04043522e-01 9.00220394e-01 -2.32687384e-01 -4.14702028e-01
1.14886232e-01 9.50596511e-01 -6.83850646e-01 -1.56794176e-01
8.93708542e-02 1.32976010e-01 -6.14344776e-02 7.86941528e-01
7.35923707e-01 -3.06383483e-02 -4.93562110e-02 -2.14746311e-01
8.02149698e-02 2.19508275e-01 -8.67212415e-01 1.26344132e+00
-5.35005093e-01 1.00458992e+00 1.64708272e-01 -3.99690717e-01
1.25189388e+00 3.38773996e-01 3.12580556e-01 -7.57999480e-01
5.46920478e-01 -1.30408481e-01 2.22632810e-01 -4.97924298e-01
3.86679798e-01 -2.48884838e-02 4.89389002e-01 3.56314868e-01
-3.53413448e-02 4.68659788e-01 -3.68210047e-01 -2.77719468e-01
5.16882658e-01 1.04698224e-03 1.98633038e-02 -2.47517064e-01
3.36141646e-01 -4.11183178e-01 6.12688839e-01 1.40423268e-01
-3.15427750e-01 4.33093518e-01 6.68363392e-01 -3.56415957e-01
-6.43956006e-01 -5.79407632e-01 -1.71466500e-01 1.26037943e+00
-4.73896831e-01 -3.59224856e-01 -9.61928427e-01 -3.18215907e-01
1.09002832e-02 7.55289018e-01 -9.29761291e-01 -1.76030576e-01
-6.46370798e-02 -6.18556857e-01 8.27266932e-01 5.98639369e-01
6.15597188e-01 -1.12277937e+00 -5.02679765e-01 -3.16597134e-01
-1.71214148e-01 -1.01850474e+00 -1.32890731e-01 -1.74995586e-01
-1.07607460e+00 -1.15825951e+00 -4.22379583e-01 -6.09973133e-01
6.33790791e-01 -2.05953971e-01 1.03733623e+00 1.49037093e-01
-6.46519586e-02 3.20456654e-01 1.33804306e-01 -8.40342522e-01
-1.45302564e-02 1.92279503e-01 5.27751036e-02 6.58215582e-02
9.44359124e-01 -4.40283626e-01 -5.70002377e-01 4.41405684e-01
-6.37868047e-01 -4.28093016e-01 5.52601099e-01 5.31481504e-01
2.42771819e-01 -3.80933806e-02 4.58618671e-01 -6.37036800e-01
8.24608564e-01 -5.65074742e-01 -4.33382303e-01 4.33225818e-02
-8.41260612e-01 -1.50395378e-01 2.68922988e-02 -3.69991630e-01
-8.82778585e-01 -2.26028077e-02 -2.86769688e-01 -4.58017588e-01
-2.91211128e-01 3.84214044e-01 -1.46200638e-02 -2.11592302e-01
6.14924967e-01 -3.16199452e-01 5.50870657e-01 -2.22394690e-01
-1.13091476e-01 6.40878975e-01 1.72404572e-01 -4.45380807e-01
3.36529553e-01 4.06729907e-01 -1.92439426e-02 -1.17660534e+00
-5.13099611e-01 -1.32007405e-01 -6.49340570e-01 -1.38477325e-01
7.34180033e-01 -8.93730998e-01 -1.16765404e+00 3.72103423e-01
-9.07864094e-01 -2.58517921e-01 1.33469954e-01 4.97312635e-01
-4.66179214e-02 -1.56101748e-01 -4.05418068e-01 -7.56167114e-01
-6.01398528e-01 -1.12496698e+00 7.91648507e-01 5.70536137e-01
-6.37872636e-01 -1.23295784e+00 -7.52392635e-02 4.78599131e-01
5.49630821e-01 2.73228079e-01 9.03410971e-01 -8.03464413e-01
-1.81611612e-01 -4.16712850e-01 -2.99418628e-01 5.30531108e-01
1.33653060e-01 4.41328555e-01 -1.35619521e+00 -1.85970753e-01
-2.85062581e-01 -3.73253226e-01 4.28306133e-01 5.95045447e-01
1.00839150e+00 -2.42988214e-01 -8.54667202e-02 8.14225852e-01
1.14551139e+00 6.43399954e-02 8.70995224e-01 3.96152020e-01
3.14258575e-01 6.64028227e-01 1.40172854e-01 3.41577768e-01
2.96134800e-01 4.06501412e-01 4.78375226e-01 -1.52840793e-01
8.90741274e-02 -1.48810446e-01 3.58190119e-01 2.29545370e-01
-2.58673191e-01 1.34721532e-01 -8.62180531e-01 2.98479736e-01
-1.35400832e+00 -7.13500559e-01 5.51503636e-02 2.06672955e+00
3.43184918e-01 -4.24785763e-02 1.23517938e-01 2.37656571e-02
3.69666547e-01 -4.63279746e-02 -5.41896880e-01 -8.11469197e-01
2.00221404e-01 4.34610546e-01 3.37633222e-01 2.46886089e-01
-7.48697460e-01 9.32810664e-01 6.87347174e+00 3.30712169e-01
-1.49206126e+00 -2.21419297e-02 9.01211619e-01 -2.45457545e-01
8.98380354e-02 -2.48922408e-01 -1.08253169e+00 3.15230072e-01
1.26096380e+00 3.26910883e-01 3.40611160e-01 9.52876985e-01
3.33108544e-01 -1.17744461e-01 -1.00518644e+00 9.94119942e-01
1.94259420e-01 -1.04520988e+00 -2.92661935e-01 3.42496842e-01
4.91446406e-01 1.32923484e-01 3.15981001e-01 4.80806589e-01
-1.64289087e-01 -1.34466577e+00 2.05889046e-01 4.65820462e-01
6.41099811e-01 -9.41919267e-01 9.24372673e-01 -2.38686055e-02
-4.78983909e-01 -6.10085912e-02 -3.68341923e-01 -2.42799684e-01
-3.98729652e-01 1.43478110e-01 -1.12054241e+00 -2.18296379e-01
7.84726620e-01 2.30190203e-01 -5.93321145e-01 7.91334152e-01
-1.32421732e-01 7.01549470e-01 -2.78379977e-01 -1.10824518e-01
1.69482067e-01 -1.73996150e-01 -5.29372655e-02 9.59711611e-01
1.58651069e-01 -1.05219394e-01 -3.61225128e-01 7.89576054e-01
-2.82863528e-01 8.18748400e-02 -5.54479420e-01 -8.87483582e-02
4.61423844e-01 1.26752067e+00 -5.43349862e-01 1.93593040e-01
-4.80498016e-01 4.77384984e-01 5.84246814e-02 4.50099230e-01
-1.06072199e+00 -2.62839407e-01 1.20222175e+00 4.19427335e-01
-1.09213004e-02 -8.08834955e-02 -5.32010853e-01 -6.27724290e-01
-2.74213105e-01 -8.90519142e-01 2.51233041e-01 -8.30893517e-01
-7.84311593e-01 3.82761389e-01 -1.62981048e-01 -4.39474136e-01
-1.98042825e-01 -6.11132622e-01 -7.14043796e-01 9.99973714e-01
-1.20297122e+00 -1.19557834e+00 -5.27304649e-01 7.12739170e-01
1.24853931e-01 -2.83623666e-01 1.00351250e+00 3.69357765e-01
-8.95097435e-01 1.25868535e+00 -1.07888132e-01 5.43136239e-01
7.09576488e-01 -5.94124496e-01 2.04505920e-01 4.42460477e-01
-2.36562602e-02 1.06199479e+00 4.41686988e-01 -5.08839488e-01
-1.22399795e+00 -1.00067925e+00 8.66346240e-01 -3.42940778e-01
2.42794290e-01 -6.78924099e-02 -6.96340144e-01 9.72351968e-01
1.81119472e-01 -3.66074562e-01 9.09823716e-01 6.35887027e-01
-3.73905152e-01 -2.94163197e-01 -1.37815440e+00 5.88249862e-01
3.99886519e-01 -5.24298728e-01 -1.38294235e-01 1.57115255e-02
1.01486836e-02 -3.81279141e-01 -8.85792434e-01 4.18475360e-01
9.60520625e-01 -1.10589802e+00 9.78303432e-01 -5.67697346e-01
4.68922377e-01 3.64371330e-01 6.38258606e-02 -1.02674854e+00
-2.39817739e-01 -4.21213061e-01 -8.36853832e-02 1.39370179e+00
6.54762328e-01 -9.94859517e-01 1.17972863e+00 1.18580782e+00
2.03577474e-01 -8.18933070e-01 -6.85006499e-01 -2.40660176e-01
-5.09345196e-02 -5.18973053e-01 5.49317002e-01 9.04686868e-01
-6.66078866e-01 3.32735389e-01 -1.82060629e-01 3.86495516e-02
4.57933336e-01 -4.04575169e-01 1.02700782e+00 -1.34014511e+00
2.48347268e-01 -4.35833544e-01 -4.34364915e-01 -3.28829557e-01
3.80673230e-01 -6.76391602e-01 -3.53259325e-01 -1.31276774e+00
-1.25655010e-01 -4.11495954e-01 -2.92217880e-01 9.91052747e-01
1.87912434e-01 3.66275847e-01 8.31989348e-02 -1.31476164e-01
1.44976109e-01 1.73719332e-01 9.35094297e-01 2.30504230e-01
-3.87021929e-01 2.59172916e-01 -8.81964624e-01 8.00591469e-01
1.07094038e+00 -2.54915982e-01 -5.49090922e-01 -5.63033402e-01
1.92820206e-01 -2.24081725e-01 4.69354451e-01 -7.75368392e-01
2.50387609e-01 8.50709900e-02 9.92261946e-01 -4.33314264e-01
6.98667347e-01 -7.89741874e-01 1.62203759e-01 1.90856799e-01
-9.52999890e-02 3.13455462e-01 6.67286992e-01 4.11335789e-02
1.15544960e-01 -1.52701870e-01 8.11766267e-01 5.36416993e-02
-4.84570175e-01 1.97616905e-01 -2.77294636e-01 -1.54543594e-01
1.00252867e+00 -4.77701664e-01 -1.26200661e-01 -6.33741260e-01
-6.36306643e-01 5.12419008e-02 3.73755902e-01 6.27729833e-01
6.55134678e-01 -9.60544646e-01 -5.81704676e-01 5.25746167e-01
-3.02951753e-01 -2.31597573e-01 2.73460969e-02 1.02158928e+00
-5.04802287e-01 5.73639929e-01 -3.97227496e-01 -5.30459225e-01
-1.49896836e+00 -1.19833253e-01 6.65673792e-01 5.66752255e-01
-2.13028774e-01 9.74945366e-01 -2.80409485e-01 -5.38777888e-01
7.23519385e-01 -2.94149637e-01 -1.93514004e-01 1.95681408e-01
4.43298131e-01 7.06180751e-01 1.91676915e-01 -6.69669509e-01
-4.43624705e-01 4.16722059e-01 -2.86912359e-02 2.73392685e-02
1.25974572e+00 1.17514752e-01 -2.14594200e-01 1.15866885e-01
1.33850586e+00 -1.30094392e-02 -1.00645459e+00 1.57697842e-01
-8.40360299e-02 -4.61672306e-01 2.97846437e-01 -8.94030452e-01
-1.34497809e+00 8.99947703e-01 9.17101681e-01 -2.34412000e-01
9.52408195e-01 -3.49174201e-01 6.24702692e-01 3.79919767e-01
4.04576063e-02 -9.76577640e-01 -1.38103098e-01 4.91918325e-01
5.95057487e-01 -1.41566813e+00 -5.85001782e-02 5.31590022e-02
-6.08775437e-01 9.23393488e-01 9.06405091e-01 9.01216865e-02
9.63093579e-01 1.05883643e-01 3.28672796e-01 -4.23029482e-01
-6.06684506e-01 1.24210320e-01 3.95421654e-01 5.69882214e-01
8.52322280e-01 -1.86878234e-01 7.82837644e-02 4.65143144e-01
-5.11003673e-01 2.34075859e-01 1.32188097e-01 6.06905162e-01
-5.15856706e-02 -7.69672215e-01 -2.45401517e-01 5.79830527e-01
-8.11457634e-01 -3.82673703e-02 -6.73284292e-01 1.12095082e+00
2.26244360e-01 1.11275613e+00 3.63068104e-01 -4.85838592e-01
1.39181778e-01 4.96871173e-01 2.59614527e-01 -4.34201628e-01
-7.42092073e-01 -2.77249247e-01 9.42050591e-02 -6.25011086e-01
-2.28894666e-01 -4.52611089e-01 -1.19826555e+00 -8.79997194e-01
-4.49369699e-01 9.82763991e-02 1.23070407e+00 9.25549567e-01
6.57027304e-01 1.74116746e-01 4.22234833e-01 -6.02608442e-01
-3.49910796e-01 -1.26653421e+00 -2.75446713e-01 2.69305632e-02
1.71608672e-01 -5.51273167e-01 -2.73760051e-01 -2.31627420e-01] | [13.444416999816895, 1.302703857421875] |
d467b19e-df2a-4e99-9488-6ae765170586 | entity-projection-via-machine-translation-for | 1909.05356 | null | https://arxiv.org/abs/1909.05356v2 | https://arxiv.org/pdf/1909.05356v2.pdf | Entity Projection via Machine Translation for Cross-Lingual NER | Although over 100 languages are supported by strong off-the-shelf machine translation systems, only a subset of them possess large annotated corpora for named entity recognition. Motivated by this fact, we leverage machine translation to improve annotation-projection approaches to cross-lingual named entity recognition. We propose a system that improves over prior entity-projection methods by: (a) leveraging machine translation systems twice: first for translating sentences and subsequently for translating entities; (b) matching entities based on orthographic and phonetic similarity; and (c) identifying matches based on distributional statistics derived from the dataset. Our approach improves upon current state-of-the-art methods for cross-lingual named entity recognition on 5 diverse languages by an average of 4.1 points. Further, our method achieves state-of-the-art F_1 scores for Armenian, outperforming even a monolingual model trained on Armenian source data. | ['Bhargavi Paranjape', 'Alankar Jain', 'Zachary C. Lipton'] | 2019-08-31 | entity-projection-via-machine-translation-for-1 | https://aclanthology.org/D19-1100 | https://aclanthology.org/D19-1100.pdf | ijcnlp-2019-11 | ['cross-lingual-ner'] | ['natural-language-processing'] | [-1.65203869e-01 -1.79656103e-01 -5.58047831e-01 -4.70173657e-01
-1.76044679e+00 -1.16782463e+00 8.45756114e-01 8.46798122e-02
-5.46288729e-01 9.51816738e-01 5.80804288e-01 -6.04641616e-01
4.70632941e-01 -5.82056463e-01 -6.76605701e-01 -1.60108775e-01
4.12903637e-01 1.01057100e+00 -2.21570387e-01 -1.21022522e-01
2.29772270e-01 4.13896412e-01 -8.60227168e-01 4.64525551e-01
1.25423372e+00 3.16430598e-01 -1.32814571e-01 1.46592736e-01
-6.06134892e-01 2.58891642e-01 -3.00761849e-01 -9.62922752e-01
3.37855756e-01 -5.41688442e-01 -9.58446145e-01 -2.90264219e-01
5.29644370e-01 1.83308989e-01 5.59257045e-02 1.05991471e+00
4.62221533e-01 -1.59859985e-01 6.20087385e-01 -7.00064838e-01
-1.28830755e+00 8.30197155e-01 -2.95200437e-01 -7.42504746e-02
4.58509713e-01 4.87017967e-02 1.31872582e+00 -1.46020567e+00
1.09633458e+00 7.87981272e-01 8.61727476e-01 3.86697441e-01
-1.40249693e+00 -7.34471560e-01 -1.17874548e-01 -1.03718489e-01
-1.58575523e+00 -8.92507911e-01 3.10308963e-01 -4.73847479e-01
1.47427654e+00 -1.42364904e-01 1.69516936e-01 1.05769682e+00
-9.76291522e-02 6.63159609e-01 1.34707975e+00 -7.20605552e-01
4.06833738e-02 4.19677883e-01 -2.08731323e-01 5.09618402e-01
2.88011461e-01 -1.03956908e-01 -7.17543423e-01 -3.86442244e-01
3.27054441e-01 -4.82023686e-01 1.21819012e-01 -3.73135209e-02
-1.48622918e+00 7.88310885e-01 -1.57257706e-01 5.03577352e-01
-5.58127940e-01 -2.85950780e-01 4.45259809e-01 4.42506731e-01
6.67485058e-01 7.98684835e-01 -9.55994427e-01 -5.07959068e-01
-1.22233832e+00 -2.87496716e-01 1.10197091e+00 1.15942717e+00
8.92873824e-01 -9.40800738e-03 4.74133827e-02 1.12643409e+00
1.80255696e-01 9.59152520e-01 5.38635552e-01 -3.55942965e-01
8.05042207e-01 7.12707818e-01 2.18622103e-01 -3.21184129e-01
-1.40475824e-01 -2.37160757e-01 -2.47353584e-01 -3.72158408e-01
5.46478689e-01 -2.61294127e-01 -6.99073255e-01 1.80973494e+00
3.28662455e-01 -2.58320361e-01 3.28940392e-01 5.22137761e-01
4.52857882e-01 5.35946190e-01 2.21563116e-01 -5.35003170e-02
1.41750038e+00 -9.35679793e-01 -3.44265997e-01 -2.83681095e-01
9.56241548e-01 -1.23499799e+00 1.00731277e+00 -8.36703926e-02
-9.47836220e-01 -3.25251341e-01 -5.80173135e-01 -1.65381879e-01
-5.90007365e-01 6.04306698e-01 4.61739242e-01 1.02742314e+00
-9.78966475e-01 3.16579551e-01 -9.57738698e-01 -7.24357545e-01
1.17701754e-01 2.65612811e-01 -7.58532643e-01 -4.50121472e-03
-1.03395772e+00 1.24195874e+00 2.87999630e-01 -2.22637296e-01
-5.81435740e-01 -8.24383676e-01 -8.29282165e-01 -1.70210928e-01
-1.16264723e-01 -5.61703086e-01 1.06626141e+00 -7.39425361e-01
-1.65386689e+00 1.47261786e+00 -3.94096524e-01 -1.33461878e-01
3.94817024e-01 -4.48329896e-01 -8.41192603e-01 -1.25411022e-02
6.54897690e-01 5.32839894e-01 1.12030007e-01 -7.58887172e-01
-6.15573227e-01 -4.40518618e-01 -3.76803100e-01 2.78552771e-01
-3.51515710e-01 6.97670758e-01 -5.44421554e-01 -6.05411232e-01
1.77689828e-02 -1.10896337e+00 -1.47399426e-01 -4.18440253e-01
-4.93569881e-01 -1.69366702e-01 3.31540793e-01 -1.07129896e+00
1.05785251e+00 -1.79495502e+00 -6.99426755e-02 3.42041552e-02
-3.31553906e-01 3.22111934e-01 -3.76583457e-01 7.33049870e-01
-1.30818347e-02 2.89805651e-01 -1.34125382e-01 -4.74099696e-01
2.33882055e-01 -1.89617779e-02 -5.26332974e-01 5.36597311e-01
5.30848384e-01 1.08029974e+00 -8.51107657e-01 -4.52695817e-01
-2.92676166e-02 4.15544838e-01 -4.56822813e-01 -5.92447538e-03
7.24564940e-02 5.24104893e-01 -2.55221367e-01 7.93885887e-01
4.56754982e-01 5.31931818e-02 6.29913151e-01 2.18915924e-01
-5.34631073e-01 1.18267417e+00 -8.88020813e-01 2.00926566e+00
-6.91923738e-01 4.09912765e-01 -2.66500026e-01 -6.77218616e-01
1.22252274e+00 4.86219794e-01 4.20088708e-01 -4.53712612e-01
-1.19108371e-01 9.40062940e-01 -2.40744054e-01 -1.94135621e-01
6.50733054e-01 -2.55651355e-01 -4.81264144e-01 6.99324429e-01
3.62244606e-01 -9.63684916e-03 1.34436280e-01 -1.15701392e-01
9.50789273e-01 4.98115391e-01 6.12835050e-01 -4.66194004e-01
4.14877772e-01 4.45215493e-01 7.50544906e-01 4.73603547e-01
-2.38485157e-01 5.23693502e-01 9.57314968e-02 -3.69217724e-01
-1.32017577e+00 -1.15082014e+00 -1.76265389e-01 1.35937440e+00
-3.82150680e-01 -2.90619969e-01 -7.21313894e-01 -8.19237292e-01
-6.21547513e-02 1.05848217e+00 -3.49994063e-01 2.79918194e-01
-9.12503600e-01 -6.05984509e-01 1.21611357e+00 6.22087777e-01
2.73073912e-01 -1.05271208e+00 2.14419588e-01 4.20726895e-01
-4.21593904e-01 -1.40329492e+00 -6.90248609e-01 1.82872593e-01
-7.50912964e-01 -5.94088912e-01 -7.44279861e-01 -9.70937014e-01
4.06854361e-01 -1.16730109e-02 1.36303937e+00 -5.86212516e-01
2.77698845e-01 8.26067105e-02 -7.96818212e-02 -2.28586063e-01
-6.51725352e-01 6.93603992e-01 4.40229386e-01 -1.75550804e-01
9.07879949e-01 -4.88163084e-01 -1.30775437e-01 3.64735305e-01
-3.07991505e-01 -2.71773368e-01 8.52926016e-01 6.45077944e-01
7.23338902e-01 -8.32356989e-01 7.08133161e-01 -1.15650845e+00
2.64685094e-01 -5.87330222e-01 -6.23911619e-01 6.48305058e-01
-6.43443406e-01 2.50252173e-03 7.62792349e-01 -3.81612182e-01
-1.21885693e+00 2.24276692e-01 -1.61265969e-01 -1.42669201e-01
-4.00761575e-01 4.05180961e-01 -1.17814377e-01 9.56987813e-02
6.93570077e-01 2.92682737e-01 -6.55830622e-01 -4.93625432e-01
7.81799853e-01 9.84056652e-01 6.56221926e-01 -8.31333339e-01
8.38540733e-01 2.69302487e-01 -5.28145373e-01 -6.44537270e-01
-7.86903381e-01 -8.07429910e-01 -1.19145191e+00 2.82878608e-01
1.08602512e+00 -1.14550471e+00 -1.19161122e-01 2.31032744e-01
-1.35711169e+00 5.02740294e-02 -1.52039379e-01 9.07297790e-01
-4.15868342e-01 1.43049918e-02 -8.48993242e-01 -6.29190207e-01
-4.87443060e-01 -9.41118598e-01 1.34023285e+00 -2.37316592e-03
-4.67583746e-01 -1.10715914e+00 8.09075117e-01 5.29093564e-01
3.92570764e-01 -1.99871063e-01 7.90913939e-01 -1.33183277e+00
-3.09309453e-01 -1.61912650e-01 -1.52989358e-01 2.97347922e-02
1.54198691e-01 -2.85348952e-01 -8.26058805e-01 -1.18809290e-01
-2.92467296e-01 -2.36080751e-01 4.00261730e-01 -5.25277145e-02
-1.46411592e-03 -1.54315129e-01 -4.04584706e-01 5.64648449e-01
1.32473576e+00 -6.89398721e-02 4.69973028e-01 4.27490145e-01
6.03104711e-01 4.53951627e-01 2.93847978e-01 -4.54446226e-02
6.50684714e-01 6.96817994e-01 -4.68898565e-01 -1.09442480e-01
-2.05337197e-01 -6.68206573e-01 7.08369315e-01 1.40568328e+00
7.91976005e-02 1.96106359e-01 -1.32690203e+00 1.11366451e+00
-1.51013696e+00 -8.97977114e-01 -9.69408527e-02 2.28947616e+00
1.09416676e+00 -1.87121615e-01 6.13118112e-02 -8.42333436e-01
1.04191852e+00 -2.22479925e-01 -4.45305258e-01 -4.30868804e-01
-3.86515081e-01 5.75195253e-01 6.14180326e-01 2.47703969e-01
-1.01871550e+00 1.54706371e+00 6.81572771e+00 7.24287033e-01
-1.19039762e+00 3.64973605e-01 2.94170827e-01 2.18778834e-01
-4.03487504e-01 4.46500033e-01 -1.21985745e+00 1.71617553e-01
1.19501042e+00 -2.45387241e-01 4.09269243e-01 9.73391056e-01
-2.17215613e-01 4.28603679e-01 -1.08735108e+00 7.93359041e-01
2.43970916e-01 -1.16961718e+00 -7.37576857e-02 2.53998935e-01
1.18813431e+00 8.01792085e-01 -1.96926877e-01 4.51703370e-01
8.07795882e-01 -7.75877476e-01 6.58019185e-01 1.99979916e-01
1.12184179e+00 -5.19820631e-01 5.56126058e-01 2.10059837e-01
-1.18072319e+00 4.27563727e-01 -3.99448603e-01 2.98841089e-01
4.34968323e-01 4.56217200e-01 -9.29951310e-01 7.70038724e-01
2.83724427e-01 4.34627920e-01 -4.30398911e-01 8.59009564e-01
-5.40905595e-01 1.01634848e+00 -5.00226736e-01 4.77389731e-02
3.76547754e-01 -4.93779361e-01 4.25318480e-01 1.60857797e+00
5.38204551e-01 -7.41595104e-02 3.12997431e-01 7.82595038e-01
-4.66599047e-01 8.43395829e-01 -7.89514780e-01 -4.44333166e-01
8.30904365e-01 1.14528227e+00 -5.73039830e-01 -4.50338542e-01
-8.46729398e-01 1.24025047e+00 8.00116360e-01 1.99811772e-01
-6.17396891e-01 -4.89868522e-01 7.34186769e-01 -3.60474825e-01
5.60207725e-01 -4.84915167e-01 -5.63449800e-01 -1.65034080e+00
9.80858654e-02 -9.26955760e-01 4.07599837e-01 -4.61124897e-01
-1.59756625e+00 8.23926806e-01 -5.55403233e-01 -1.20651650e+00
-3.80812794e-01 -6.69459939e-01 -3.65934193e-01 1.14441597e+00
-1.34152901e+00 -1.58886671e+00 5.15812457e-01 2.94824660e-01
3.49628925e-01 -4.61475521e-01 1.32572508e+00 5.67605615e-01
-4.71301436e-01 7.94942200e-01 3.77534896e-01 5.66231012e-01
1.24632812e+00 -1.04283226e+00 1.07379436e+00 1.01167083e+00
8.57561111e-01 1.10096729e+00 1.14512093e-01 -8.49950671e-01
-1.45649564e+00 -1.11250734e+00 1.96340311e+00 -9.24738586e-01
1.10243511e+00 -4.93079185e-01 -7.06684291e-01 1.02818120e+00
3.13065112e-01 -1.63188368e-01 1.19482851e+00 6.85792565e-01
-9.45584595e-01 3.05829942e-01 -1.12462771e+00 5.05075395e-01
1.17802131e+00 -1.19859922e+00 -9.85073209e-01 5.35876930e-01
4.30407077e-01 -1.93661034e-01 -1.02491713e+00 -8.71615261e-02
6.94690406e-01 -4.62877989e-01 6.59095943e-01 -1.17048728e+00
3.39875519e-01 -3.03520948e-01 -5.89166105e-01 -1.28399670e+00
-2.91046023e-01 -8.05219769e-01 4.67747808e-01 1.58160079e+00
1.00829053e+00 -8.22991490e-01 5.45707524e-01 5.54217100e-01
-3.14494789e-01 -3.31412971e-01 -1.02206385e+00 -1.04968774e+00
5.47229707e-01 -4.28271800e-01 6.18315935e-01 1.58521450e+00
2.28747204e-01 6.09592140e-01 -2.40174249e-01 3.04482728e-01
3.13021928e-01 3.81908625e-01 8.01972091e-01 -9.54678178e-01
-3.24926317e-01 -4.37934965e-01 -2.86870897e-01 -9.66217816e-01
6.00807488e-01 -1.44148982e+00 7.19896331e-02 -1.30335176e+00
4.23749298e-01 -5.69845319e-01 -1.62065774e-01 8.13543558e-01
-1.68554902e-01 6.95239127e-01 3.90389189e-02 6.55050159e-01
-5.38041711e-01 2.98262388e-01 3.75560284e-01 2.25070715e-01
-2.47089833e-01 -3.45342964e-01 -7.56821454e-01 5.68142951e-01
5.30658305e-01 -7.48997748e-01 3.79977167e-01 -8.22699785e-01
2.26231471e-01 -7.68807605e-02 -4.39239115e-01 -6.49015665e-01
1.61035165e-01 -2.36477688e-01 2.10123450e-01 -4.71474022e-01
7.62086408e-03 -3.47851396e-01 2.59295642e-01 1.98238656e-01
-2.41344780e-01 4.08422083e-01 2.09402546e-01 3.24480273e-02
-2.42533430e-01 -9.02190953e-02 6.50953889e-01 -4.18741889e-02
-6.17054164e-01 4.27586772e-02 -3.81523222e-01 4.37544078e-01
5.97606242e-01 1.60083726e-01 -5.37271261e-01 -5.66830742e-04
-3.45520020e-01 -2.37347052e-01 8.11171532e-01 3.75251442e-01
-2.15606570e-01 -1.53344679e+00 -1.06542015e+00 1.32071972e-01
4.83300209e-01 -9.67869461e-01 -2.83915401e-01 8.76312435e-01
-3.63880694e-01 7.36568630e-01 -1.62516609e-01 -3.55845362e-01
-8.75447512e-01 2.44567990e-01 -9.48193967e-02 -5.46121299e-01
-2.38473207e-01 7.06221879e-01 -5.17980084e-02 -1.44894767e+00
-5.51681578e-01 1.50693074e-01 3.63649160e-01 -1.09383557e-02
2.42228284e-01 2.55222112e-01 4.18871343e-01 -1.12014687e+00
-7.27474988e-01 6.68199539e-01 -6.99828789e-02 -6.66749358e-01
1.32858670e+00 -3.84137370e-02 -1.36981130e-01 3.96080405e-01
1.19157493e+00 7.64133036e-01 -3.33528996e-01 -4.21830833e-01
4.21870172e-01 -2.64945447e-01 -3.92345876e-01 -1.02519321e+00
-4.25668806e-01 6.14295542e-01 6.65948465e-02 -3.85499001e-01
7.20851243e-01 6.70059845e-02 1.00977385e+00 6.48440063e-01
7.28903413e-01 -1.11110747e+00 -5.97240150e-01 8.65765870e-01
3.24232697e-01 -1.25490856e+00 -1.48324087e-01 -2.58062184e-01
-7.33037472e-01 8.87544036e-01 4.34532017e-02 -3.57872322e-02
3.47788692e-01 2.05583051e-01 5.43541312e-01 -1.71756539e-02
-5.55825651e-01 -2.04956815e-01 5.47224402e-01 4.91909027e-01
8.79620850e-01 3.58501881e-01 -4.96944755e-01 4.55997109e-01
-4.12662864e-01 -2.02827215e-01 1.55636862e-01 6.27722502e-01
-1.01873949e-01 -1.59107506e+00 -3.05448383e-01 2.12807029e-01
-7.93025196e-01 -7.57478833e-01 -6.20515108e-01 7.90913165e-01
-1.36035487e-01 9.44132566e-01 -7.52045587e-02 -1.86692089e-01
2.73321241e-01 6.33770645e-01 3.52991194e-01 -8.58708084e-01
-9.06225264e-01 2.97860354e-02 5.23827910e-01 -3.27777058e-01
-2.41035238e-01 -1.09936655e+00 -1.00352931e+00 -2.35712871e-01
-5.32021046e-01 4.20121998e-01 1.01360559e+00 9.69251633e-01
8.13178480e-01 -4.07500088e-01 4.21981663e-01 -4.30782646e-01
-6.14588022e-01 -1.13246310e+00 -2.17992842e-01 4.25866872e-01
-4.64240104e-01 -2.67549664e-01 -6.95208088e-02 2.38873810e-01] | [10.662985801696777, 9.974372863769531] |
effebeaa-0216-4221-b04d-e68ac9b62364 | contour-and-centreline-tracking-of-vessels | 1707.03710 | null | http://arxiv.org/abs/1707.03710v1 | http://arxiv.org/pdf/1707.03710v1.pdf | Contour and Centreline Tracking of Vessels from Angiograms using the Classical Image Processing Techniques | This article deals with the problem of vessel edge and centerline detection
using classical image processing techniques due to their simpleness and
easiness to be implemented. The method is divided into four steps: the vessel
enhancement which implies a non-linear filtering proposed by Frangi, the
thresholding using Otsu method and the contour detection using the Canny edge
detector due to its good performances for the small vessels and the
morphological skeletonisation. The algorithms are tested on real data collected
from a cardiac catheterism laboratory and it is accurate for images with good
spatial resolution (512*512). The output image can be used for further
processing in order to find the vessel length or its radius. | ['Tache Irina Andra'] | 2017-06-13 | null | null | null | null | ['contour-detection'] | ['computer-vision'] | [ 1.47033572e-01 -9.71110910e-02 1.61497295e-01 5.48377559e-02
1.31402433e-01 -4.36245888e-01 2.55795598e-01 6.20149672e-01
-7.92416334e-01 7.06683636e-01 -4.12185900e-02 -5.60291946e-01
-1.05495468e-01 -6.56667233e-01 2.46927530e-01 -6.17893696e-01
-2.90136397e-01 3.14519584e-01 7.06418335e-01 1.20682292e-01
4.91817623e-01 1.12109506e+00 -1.06495512e+00 -1.24217376e-01
5.04750609e-01 8.48067462e-01 -3.10219750e-02 1.11953032e+00
-9.08138230e-02 8.19795609e-01 -2.72464156e-01 -2.23968178e-02
4.60310519e-01 -5.96938550e-01 -6.55189216e-01 3.03259254e-01
-8.94031003e-02 -3.55069071e-01 -1.38839092e-02 7.50224233e-01
5.95889986e-01 -4.64075133e-02 9.57736611e-01 -5.26110709e-01
7.54263923e-02 7.96053838e-03 -6.74383998e-01 8.87887359e-01
-1.69066899e-02 -4.30151708e-02 3.12617481e-01 -5.13738096e-01
5.88413119e-01 1.04303324e+00 6.52851760e-01 4.35044020e-02
-1.15427816e+00 -2.03239605e-01 -8.26517999e-01 3.23603064e-01
-1.32867849e+00 -3.27517577e-02 4.36883390e-01 -6.91671073e-01
6.44431233e-01 2.48867899e-01 7.52375782e-01 -1.28851846e-01
4.91350204e-01 9.91292670e-02 1.34351945e+00 -7.29346097e-01
9.00169536e-02 2.49149218e-01 3.01204920e-01 7.97287285e-01
5.10805249e-01 3.29666197e-01 3.66710007e-01 5.96842263e-03
1.34293437e+00 -1.84475910e-02 -1.37377545e-01 -2.12548912e-01
-9.83598351e-01 8.33327234e-01 2.98777848e-01 9.92093921e-01
-6.03844404e-01 -1.58848226e-01 7.08878160e-01 2.62760073e-01
6.77969493e-03 2.59487271e-01 -5.40568270e-02 -3.38956267e-02
-9.67861772e-01 -3.31905991e-01 7.06880212e-01 4.41817462e-01
7.72003680e-02 -9.03320536e-02 1.73225090e-01 4.16114748e-01
3.74796331e-01 3.36053260e-02 6.74889803e-01 -7.81563401e-01
-1.03932641e-01 7.12650836e-01 -9.48858708e-02 -9.83940184e-01
-6.13689423e-01 -1.45982191e-01 -7.20487356e-01 1.07971323e+00
1.07775307e+00 -3.80903155e-01 -8.97793651e-01 7.58848548e-01
4.41287905e-01 1.11347415e-01 2.36680489e-02 8.33378613e-01
9.21712041e-01 6.14099979e-01 2.21526593e-01 -3.68337691e-01
1.75030923e+00 -6.42117083e-01 -7.93425262e-01 3.04031819e-01
6.21913135e-01 -1.20888925e+00 3.84597838e-01 4.89818931e-01
-1.00506759e+00 -6.09902382e-01 -9.56745565e-01 1.23089507e-01
-3.54351908e-01 5.16107261e-01 4.82340336e-01 7.83821046e-01
-7.45426059e-01 5.43369591e-01 -8.44322741e-01 -5.90191424e-01
2.93832958e-01 3.03245634e-01 -4.71941322e-01 3.46915573e-01
-3.49207610e-01 1.13173354e+00 6.49854541e-01 2.04189181e-01
-2.49634031e-02 -2.00581059e-01 -7.27641821e-01 -5.76319024e-02
-3.40414271e-02 -5.15145957e-01 5.81975579e-01 -8.95981371e-01
-1.58443296e+00 1.12230992e+00 3.42018642e-02 -5.44834912e-01
7.99506843e-01 3.52202356e-01 -1.78180382e-01 8.17818582e-01
-1.63824186e-01 9.90394279e-02 5.82258046e-01 -5.92059255e-01
-7.50981212e-01 -4.43977982e-01 -4.83214945e-01 8.67024139e-02
2.08997160e-01 2.65419573e-01 -4.06590216e-02 -7.23179221e-01
4.45009887e-01 -7.33271003e-01 -4.55496907e-01 1.41500935e-01
-7.33757839e-02 2.30162200e-02 6.50328875e-01 -9.24980819e-01
1.22650301e+00 -2.20541286e+00 -3.40140730e-01 5.84869802e-01
2.07129762e-01 4.47921306e-01 4.47405607e-01 3.38471413e-01
-3.52882564e-01 -2.35108510e-01 -1.09911636e-01 3.64315748e-01
-6.34153724e-01 1.16474785e-01 3.73121619e-01 7.35659182e-01
-2.16135547e-01 2.74715215e-01 -5.14025509e-01 -1.24122512e+00
6.47491693e-01 5.61225414e-01 1.06006384e-01 2.33886801e-02
8.18893731e-01 4.96544600e-01 -4.36186522e-01 2.91937590e-01
5.60875654e-01 1.55147865e-01 -2.14523524e-02 -2.81369269e-01
-4.25961226e-01 -4.70352501e-01 -1.58871102e+00 9.82592404e-01
-1.07500024e-01 6.53533220e-01 3.36029321e-01 -8.91611278e-01
1.29639840e+00 6.62282348e-01 6.39847457e-01 -3.70431900e-01
7.04436839e-01 4.60339069e-01 2.94542968e-01 -7.47521579e-01
-2.21027695e-02 -5.87182581e-01 6.95930958e-01 3.01090389e-01
-2.39486367e-01 -1.07087381e-01 7.12255001e-01 -1.60289049e-01
6.68894291e-01 -1.34303108e-01 7.61517406e-01 -4.89256978e-01
1.01161265e+00 2.82456666e-01 2.29891747e-01 1.83495224e-01
-4.75766718e-01 3.55102986e-01 5.95118165e-01 -5.96605599e-01
-1.01313043e+00 -7.27149308e-01 -8.08720112e-01 1.94853783e-01
1.27923191e-01 2.48978212e-01 -7.65238047e-01 -3.62497211e-01
-1.19504496e-01 1.76434293e-01 -6.29832923e-01 3.35512489e-01
-6.20070338e-01 -6.71202660e-01 4.29848991e-02 2.78143942e-01
5.19240677e-01 -1.13263810e+00 -1.43267787e+00 4.70398128e-01
3.44895691e-01 -6.27988160e-01 -2.60559954e-02 5.32373972e-02
-1.53683329e+00 -1.21861458e+00 -9.27137017e-01 -9.39213574e-01
7.90470183e-01 -7.15255141e-02 6.25749469e-01 4.31148320e-01
-9.27390337e-01 2.37800982e-02 -3.63312155e-01 -4.29413915e-01
-7.41861045e-01 -3.04462403e-01 -5.65585971e-01 -1.18435450e-01
3.99371535e-01 -5.09208739e-01 -7.38764882e-01 2.12118998e-01
-6.65848196e-01 -3.58308673e-01 8.14704835e-01 5.36834061e-01
4.39781547e-01 2.77824968e-01 1.74129575e-01 -1.01142836e+00
6.05373263e-01 1.62658215e-01 -7.10553765e-01 -3.81255224e-02
-6.44641519e-01 -2.87354648e-01 4.26159114e-01 4.87606786e-02
-7.91596293e-01 1.73093885e-01 -1.69700310e-01 2.77918905e-01
-5.19883454e-01 4.15702313e-02 5.64678431e-01 -4.45335388e-01
8.42987478e-01 -4.67154458e-02 5.87536216e-01 -4.62557256e-01
-3.14040966e-02 5.65303147e-01 6.35005355e-01 1.34196714e-01
5.03691614e-01 5.99127769e-01 6.57543242e-01 -1.02423513e+00
-2.42040977e-02 -8.61104429e-01 -1.03541255e+00 -4.59653050e-01
1.12047350e+00 -1.48209974e-01 -5.50907195e-01 2.59284377e-01
-9.71744418e-01 1.98134452e-01 -3.09809655e-01 8.67053747e-01
-3.52469802e-01 8.79620790e-01 -9.51930344e-01 -9.88182247e-01
-7.66896963e-01 -9.54176426e-01 1.31312907e-01 5.35586059e-01
-4.48648855e-02 -1.36248851e+00 -3.23421806e-02 -6.02309778e-02
4.82425392e-01 7.30094433e-01 9.80380416e-01 -3.47509652e-01
-6.24778718e-02 -6.63607299e-01 -3.15103322e-01 3.02379310e-01
2.72928197e-02 3.58098269e-01 -3.94081771e-01 -4.45363894e-02
2.41641328e-01 3.70686889e-01 6.19837523e-01 9.98916626e-01
4.29414123e-01 1.52393624e-01 -2.90632606e-01 3.96238923e-01
1.83483601e+00 5.81286252e-01 7.67773628e-01 5.30979812e-01
-1.28050074e-02 6.24479890e-01 6.25085235e-01 4.34413999e-01
-5.07833421e-01 3.11173975e-01 3.67789328e-01 -8.00201952e-01
-2.40602195e-01 4.57312346e-01 -3.12910646e-01 2.85878837e-01
-4.69062418e-01 3.40437114e-01 -6.92769885e-01 5.07255971e-01
-1.47198415e+00 -9.13174033e-01 -9.86000001e-01 2.31491232e+00
3.65827143e-01 4.30802941e-01 4.01273906e-01 4.16829288e-01
8.15907598e-01 -4.76012409e-01 1.20497055e-01 -8.22088897e-01
2.46647745e-01 4.45606411e-01 8.43443811e-01 6.77802145e-01
-1.14111352e+00 2.52107978e-01 6.75492191e+00 3.73914361e-01
-1.19929588e+00 -1.39378216e-02 6.75254762e-01 4.29311901e-01
7.25820720e-01 3.55047323e-02 -4.42025870e-01 4.58507061e-01
5.00311553e-01 -1.25990972e-01 -1.56597003e-01 5.90845346e-01
4.80938554e-01 -9.63393390e-01 -3.87925357e-01 8.99389565e-01
-2.11212710e-01 -1.11503923e+00 -3.24139953e-01 1.57389119e-01
2.51006246e-01 -4.47959542e-01 -5.45915902e-01 -3.49460512e-01
-4.32086140e-01 -7.66367793e-01 1.25495359e-01 5.58407068e-01
3.72355282e-01 -6.04105234e-01 1.06118190e+00 2.34031435e-02
-1.09404230e+00 -4.47179278e-04 -4.15744275e-01 -2.26087332e-01
3.92367005e-01 4.51083452e-01 -1.06371844e+00 2.20361233e-01
2.85310507e-01 3.43508750e-01 -5.03276110e-01 1.77656031e+00
1.12239227e-01 4.50585425e-01 -6.33500159e-01 -1.35795940e-02
4.07853335e-01 -8.43378007e-01 6.93464458e-01 1.38527381e+00
2.65525609e-01 6.72756284e-02 -1.89163372e-01 4.81253415e-01
7.11229980e-01 9.15524185e-01 -3.36513609e-01 5.05314052e-01
3.60307917e-02 1.42514527e+00 -1.59990144e+00 -5.82200408e-01
-2.80943573e-01 7.05226600e-01 -5.44488847e-01 -2.49774512e-02
-3.25473040e-01 -7.26687253e-01 -2.56878197e-01 6.02183759e-01
2.96472132e-01 -5.40204197e-02 -4.79850382e-01 -5.16325057e-01
-3.35837215e-01 -2.11503163e-01 8.65594923e-01 -3.66566956e-01
-7.01098800e-01 7.41213858e-01 8.22573230e-02 -1.13545978e+00
-1.12155855e-01 -1.05883610e+00 -9.19348598e-01 1.19932520e+00
-1.21164036e+00 -8.81121099e-01 -3.00455809e-01 4.65593338e-01
3.30755353e-01 -1.45372152e-01 6.36880338e-01 3.48507077e-01
-4.53883260e-01 8.74758419e-03 2.19890401e-01 2.70773321e-01
4.82720882e-01 -1.56828523e+00 -4.36662197e-01 1.02870965e+00
-4.26241189e-01 3.66153270e-01 8.03612947e-01 -4.90514606e-01
-5.04684091e-01 -2.91986048e-01 1.08614337e+00 1.82931155e-01
4.91893470e-01 4.47188437e-01 -5.99278569e-01 2.44498700e-01
3.45469177e-01 1.86055258e-01 5.84589958e-01 -5.79067945e-01
4.88492548e-01 -2.05121990e-02 -1.38576472e+00 2.59750724e-01
-5.70273288e-02 4.01979446e-01 -6.70147598e-01 2.95896828e-01
-5.30010998e-01 -3.27824563e-01 -1.24123192e+00 7.79090375e-02
6.03826940e-01 -1.34451556e+00 6.60831511e-01 -2.30211377e-01
9.50514674e-02 -4.28702533e-01 7.23222494e-01 -6.29353344e-01
-4.28749532e-01 -5.69970787e-01 8.05112660e-01 9.66065586e-01
4.65529382e-01 -7.07469881e-01 7.44156241e-01 3.02672207e-01
2.28451505e-01 -5.29297888e-01 -1.08097637e+00 -3.38654459e-01
-3.09111476e-01 3.98961425e-01 -2.26699412e-01 6.23773158e-01
1.79666027e-01 2.85379440e-01 4.54077348e-02 -2.72263914e-01
7.66464293e-01 1.51701629e-01 5.19201994e-01 -1.50741506e+00
-3.32402289e-01 -7.64736831e-01 -9.18682635e-01 -3.68922234e-01
-6.89169466e-01 -4.63863850e-01 -4.33503628e-01 -1.62057900e+00
-2.16122866e-01 -3.84859204e-01 2.48002671e-02 1.11635417e-01
3.79206315e-02 4.94375288e-01 -1.10257305e-01 2.06385374e-01
1.24999732e-01 -5.39949059e-01 1.38357592e+00 4.86268878e-01
-4.61641431e-01 3.97725344e-01 -1.77733570e-01 7.25748420e-01
6.21591508e-01 -4.36118066e-01 -1.77913859e-01 3.57098252e-01
-2.83954471e-01 3.09375793e-01 1.36966243e-01 -1.18564236e+00
6.89189658e-02 3.53585333e-01 4.79279906e-01 -5.98549306e-01
-1.55059010e-01 -1.15920544e+00 2.73815572e-01 1.14325666e+00
-3.36043909e-02 2.01866627e-01 2.87393946e-02 2.26917237e-01
-2.47363091e-01 -1.12116611e+00 1.31235790e+00 -4.89163756e-01
-6.73200190e-01 -2.24888548e-01 -7.62008309e-01 -4.62380707e-01
1.56315017e+00 -8.31627905e-01 1.22501649e-01 -2.36149162e-01
-1.20262623e+00 -2.86409020e-01 4.32642400e-01 -6.09498858e-01
3.30185175e-01 -8.06353152e-01 -1.00294137e+00 1.41197532e-01
-8.80413726e-02 -3.55220914e-01 2.20802873e-01 1.62045944e+00
-1.78597558e+00 1.42477915e-01 -7.31769562e-01 -4.73511517e-01
-1.81216097e+00 6.91504240e-01 6.95088267e-01 -3.63034874e-01
-1.01243341e+00 4.23774034e-01 -3.00958991e-01 6.74448013e-01
2.29811150e-04 -3.68778437e-01 -8.76145840e-01 -1.14081472e-01
6.56268179e-01 7.93695927e-01 -3.77631001e-02 -8.62954855e-01
-1.37718707e-01 9.65130448e-01 2.64796168e-01 3.14342268e-02
1.12356007e+00 -1.77081898e-01 -4.78061318e-01 1.77217364e-01
1.09806418e+00 7.48151317e-02 -8.95847857e-01 6.71924725e-02
-6.36078510e-03 -5.08182526e-01 4.95017499e-01 -6.28783941e-01
-9.55087900e-01 9.42947268e-01 1.03699863e+00 6.07033849e-01
1.20802546e+00 -4.26875204e-01 4.70017135e-01 -1.15526140e-01
-1.63310617e-01 -1.15156305e+00 -5.91241419e-01 2.25778334e-02
4.92830753e-01 -9.48631585e-01 3.73005629e-01 -6.84905231e-01
-5.02493858e-01 2.00118446e+00 -2.77781710e-02 -4.91486698e-01
9.43197966e-01 3.92773092e-01 4.03839618e-01 -1.50748327e-01
-6.03281632e-02 -4.30234253e-01 1.45186767e-01 5.49962521e-01
7.09579945e-01 -1.09483888e-02 -1.43555045e+00 -2.73754388e-01
2.15543017e-01 1.82312697e-01 7.10280597e-01 8.76468003e-01
-8.35230708e-01 -9.75119293e-01 -7.57658422e-01 6.53443813e-01
-9.66178954e-01 2.38619417e-01 -2.28326712e-02 9.80037928e-01
9.73168314e-02 8.23972642e-01 -2.13390380e-01 3.94582540e-01
5.42300582e-01 -1.21419363e-01 5.33168852e-01 -1.62267670e-01
-8.85641456e-01 7.79184520e-01 3.91009077e-03 -1.65915564e-01
-3.21052551e-01 -8.76481950e-01 -1.43420362e+00 -3.98513861e-02
-2.98911065e-01 4.40340906e-01 8.91975582e-01 7.86966026e-01
-2.86804229e-01 3.39469314e-01 4.43749279e-01 -4.13865417e-01
-2.64606059e-01 -1.10827279e+00 -1.10695696e+00 6.12314105e-01
1.51187763e-01 -5.31097412e-01 -3.60883325e-01 4.34798449e-01] | [14.727148056030273, -2.9061691761016846] |
fe573c28-5cbe-430a-8b95-251bfb2973de | multi-agent-reinforcement-learning-2 | 2305.06446 | null | https://arxiv.org/abs/2305.06446v3 | https://arxiv.org/pdf/2305.06446v3.pdf | Cooperative Multi-Agent Reinforcement Learning: Asynchronous Communication and Linear Function Approximation | We study multi-agent reinforcement learning in the setting of episodic Markov decision processes, where multiple agents cooperate via communication through a central server. We propose a provably efficient algorithm based on value iteration that enable asynchronous communication while ensuring the advantage of cooperation with low communication overhead. With linear function approximation, we prove that our algorithm enjoys an $\tilde{\mathcal{O}}(d^{3/2}H^2\sqrt{K})$ regret with $\tilde{\mathcal{O}}(dHM^2)$ communication complexity, where $d$ is the feature dimension, $H$ is the horizon length, $M$ is the total number of agents, and $K$ is the total number of episodes. We also provide a lower bound showing that a minimal $\Omega(dM)$ communication complexity is required to improve the performance through collaboration. | ['Quanquan Gu', 'Tianhao Wang', 'Jiafan He', 'Yifei Min'] | 2023-05-10 | null | null | null | null | ['multi-agent-reinforcement-learning'] | ['methodology'] | [-1.74726784e-01 7.21637249e-01 2.28655055e-01 -1.14024915e-01
-1.02989233e+00 -5.30727029e-01 8.51015653e-03 6.40421689e-01
-1.12595069e+00 9.73252952e-01 -3.15834045e-01 -4.10896957e-01
-6.24769330e-01 -1.13362360e+00 -6.56455457e-01 -1.05705631e+00
-7.54521787e-01 7.20377028e-01 4.84933816e-02 -1.51379794e-01
2.49872699e-01 1.43187055e-02 -1.11047256e+00 -7.95479119e-01
6.54868007e-01 1.43566310e+00 1.76964756e-02 8.37941587e-01
2.09111810e-01 9.86351192e-01 -8.49664330e-01 -3.81558448e-01
5.54647267e-01 -8.19967389e-01 -7.45817125e-01 4.25873369e-01
-7.37184882e-01 -6.00761235e-01 -6.01657987e-01 9.11710560e-01
7.04453588e-01 4.22280014e-01 2.44978666e-01 -1.15722656e+00
2.30536088e-02 8.50294650e-01 -7.51642942e-01 1.82681933e-01
1.65829822e-01 3.43906254e-01 8.74958396e-01 2.39777640e-01
6.40015364e-01 9.01674151e-01 4.36866330e-03 2.67676085e-01
-7.92310953e-01 -5.90094447e-01 3.70151490e-01 3.60620171e-02
-1.05971396e+00 1.19946413e-01 2.87210494e-01 -9.14524123e-02
8.60564053e-01 2.32618228e-01 8.18783939e-01 3.15173715e-02
1.98760912e-01 6.85652316e-01 1.17629969e+00 -5.21506011e-01
8.84848595e-01 -2.42804959e-01 -2.84225553e-01 8.35089862e-01
1.61934242e-01 5.27473018e-02 -3.97401601e-01 -1.34473428e-01
8.42830598e-01 8.61678421e-02 5.37860813e-03 2.30665542e-02
-9.21155334e-01 1.09295022e+00 -1.75986052e-01 -1.55248433e-01
-6.25458837e-01 4.45124269e-01 1.05141744e-01 8.60028446e-01
7.69752190e-02 3.79380286e-01 -4.21633124e-01 -6.68769538e-01
-4.74873662e-01 4.23972040e-01 8.76853585e-01 1.01989686e+00
7.53030956e-01 1.65945604e-01 1.32229760e-01 2.43349001e-01
-2.40144119e-01 1.12237012e+00 -1.02338620e-01 -1.82249522e+00
6.54941678e-01 3.79075676e-01 7.08441675e-01 -4.33798581e-01
-6.82885528e-01 -3.67210746e-01 -8.08361530e-01 2.06439301e-01
4.67820883e-01 -9.05019462e-01 6.08100835e-03 1.91815317e+00
2.89165527e-01 -6.20352387e-01 2.06666216e-01 5.26418686e-01
-4.26622182e-01 5.67161083e-01 -4.45496589e-01 -9.85530615e-01
7.94901073e-01 -6.52150869e-01 -2.99992770e-01 -9.58783999e-02
6.35203958e-01 -2.73306161e-01 6.17655277e-01 3.23365897e-01
-1.79667962e+00 3.11244160e-01 -7.98077464e-01 7.96161473e-01
1.79022998e-01 -6.34112656e-01 5.96053541e-01 5.42025685e-01
-7.09333956e-01 5.34580231e-01 -8.35200906e-01 3.47968280e-01
-1.38173075e-02 6.03717446e-01 6.52862638e-02 1.52362019e-01
-7.98374236e-01 4.68175203e-01 9.21261404e-03 -2.07075208e-01
-1.20271385e+00 -7.61987790e-02 -4.82730359e-01 1.55983269e-01
8.42217326e-01 -3.50543141e-01 1.41133595e+00 -2.67183810e-01
-1.53989375e+00 2.06246004e-01 3.46190035e-01 -8.27108145e-01
7.18626142e-01 3.08109671e-01 1.68065578e-01 6.02983773e-01
4.01681215e-02 2.12408796e-01 4.50464725e-01 -6.00840032e-01
-1.20811546e+00 -6.76746607e-01 4.98918891e-01 5.50082982e-01
-3.17405105e-01 -1.81243658e-01 1.26960397e-01 -1.46858707e-01
3.15609947e-02 -1.14129245e+00 -6.87570095e-01 -4.16900218e-01
7.96606094e-02 -1.22033738e-01 1.08255222e-01 -2.53026575e-01
9.93835628e-01 -1.85608125e+00 3.72027516e-01 4.60365087e-01
3.87369543e-01 -3.36342335e-01 1.25884920e-01 8.06527257e-01
9.19763565e-01 -1.19217917e-01 -2.14567274e-01 -2.22247571e-01
5.21483719e-01 2.75824249e-01 4.04347898e-03 5.31880796e-01
-7.97523618e-01 3.27393830e-01 -6.80745304e-01 -9.91468653e-02
-1.37581695e-02 -2.35991762e-03 -2.97082454e-01 1.06160983e-01
-4.11066204e-01 1.66791841e-01 -6.58311605e-01 3.17017138e-01
3.74567628e-01 -5.35906255e-01 6.58795476e-01 9.84919131e-01
-1.08846419e-01 4.19717319e-02 -1.76100469e+00 1.20382774e+00
-3.49621713e-01 4.06304821e-02 6.07780516e-01 -9.20895100e-01
5.35664737e-01 3.12075227e-01 8.43443394e-01 -1.00786090e+00
4.91806358e-01 1.03012912e-01 -3.95889044e-01 -1.12278327e-01
1.65069297e-01 -1.86484560e-01 -4.68554825e-01 1.05631697e+00
-2.79751331e-01 8.12367722e-02 2.67880112e-01 1.47007808e-01
1.68634582e+00 -3.97681266e-01 -9.07425582e-02 1.56147042e-02
-1.55156285e-01 -1.21757686e-01 1.00456500e+00 1.06218266e+00
-4.40512449e-01 -4.23567593e-01 1.04769945e+00 -3.80962431e-01
-1.01248503e+00 -8.16354573e-01 5.04590213e-01 9.18903828e-01
5.92836976e-01 -2.49813214e-01 -8.37128758e-01 -3.75487626e-01
2.82584000e-02 6.02092028e-01 -8.41347039e-01 8.17881674e-02
-2.66603559e-01 -8.51330459e-01 3.60825568e-01 4.71578598e-01
8.17298174e-01 -1.01235628e+00 -1.53753078e+00 4.58766311e-01
2.92177256e-02 -8.31735015e-01 -2.90808767e-01 4.91616637e-01
-5.75730205e-01 -7.75096416e-01 -6.68909371e-01 -2.57620513e-01
7.46339023e-01 2.17229769e-01 4.45947140e-01 -2.95663327e-01
-1.02360062e-01 7.40220189e-01 -3.32119286e-01 -6.34806335e-01
-4.69381595e-03 -2.06980873e-02 9.46014151e-02 -4.87831801e-01
7.88735747e-02 -5.70186079e-01 -1.10796475e+00 -4.41683307e-02
-8.00990820e-01 -1.11225732e-01 4.79370952e-01 6.18157327e-01
7.03149140e-01 4.23682570e-01 4.70065176e-01 -3.61588240e-01
4.09240693e-01 -3.33366245e-02 -1.26214910e+00 9.39931720e-02
-6.44554675e-01 1.72595516e-01 6.95961714e-01 -4.15793508e-02
-5.96243739e-01 3.52294445e-02 4.05774444e-01 1.35637045e-01
2.97835469e-01 3.05089116e-01 3.66355211e-01 1.21562839e-01
2.38831922e-01 5.69150388e-01 2.81831175e-01 -1.59753248e-01
8.46566930e-02 3.24674219e-01 1.33321271e-03 -4.50628459e-01
3.93107831e-01 4.62674737e-01 4.59649473e-01 -5.61947525e-01
-5.55346012e-01 2.27274984e-01 1.98482975e-01 -1.37250751e-01
4.29077804e-01 -8.01092625e-01 -2.09991622e+00 1.38539895e-01
-3.26369196e-01 -7.91835070e-01 -4.63378936e-01 5.79448700e-01
-1.02484524e+00 2.33542353e-01 -6.69210672e-01 -1.52995956e+00
-3.02343577e-01 -8.02735031e-01 3.63278419e-01 3.48871142e-01
3.86031836e-01 -5.04768729e-01 -1.05056874e-01 5.80824971e-01
4.57015336e-01 1.80096298e-01 8.07997525e-01 -2.92215049e-01
-8.24334264e-01 -1.74807668e-01 1.51667371e-01 1.05666965e-01
-2.42520288e-01 -6.23655677e-01 2.27272455e-02 -6.65751100e-01
8.29690769e-02 -3.96953404e-01 1.79155603e-01 2.91369885e-01
8.26165080e-01 -9.73596215e-01 3.46960351e-02 -9.15304199e-02
1.35343373e+00 6.78462923e-01 3.38212103e-01 3.65095258e-01
-1.30184829e-01 2.85048187e-01 7.18253791e-01 1.60794747e+00
6.88413858e-01 2.36659020e-01 6.73274219e-01 3.07166278e-01
8.34653437e-01 2.39373788e-01 3.83477986e-01 3.94678414e-01
-6.19692914e-02 -1.59022257e-01 -7.01921999e-01 6.62245810e-01
-1.90073395e+00 -8.73123884e-01 6.78519249e-01 2.55825520e+00
8.67300272e-01 4.64293033e-01 6.67492688e-01 2.62262285e-01
3.60850096e-01 -1.80348024e-01 -6.15801871e-01 -6.62257791e-01
-1.14181437e-01 3.66180420e-01 9.41909730e-01 6.86399937e-01
-5.20851970e-01 5.56680560e-01 4.53799438e+00 5.87318242e-01
-7.57949114e-01 1.94792673e-02 4.35248554e-01 -8.46941888e-01
3.98560241e-02 5.72400279e-02 -4.99460578e-01 8.05450022e-01
1.18659008e+00 -4.05557990e-01 8.99414062e-01 7.60879815e-01
2.26222634e-01 -7.86385000e-01 -6.60458267e-01 7.31453717e-01
-4.51019615e-01 -1.28338885e+00 -6.43249214e-01 5.53016722e-01
8.13664079e-01 -3.45431089e-01 3.64645496e-02 8.33103210e-02
8.29667032e-01 -6.09557867e-01 5.94937980e-01 8.50721356e-03
2.70137966e-01 -1.48641860e+00 6.42098367e-01 7.12151766e-01
-1.14079893e+00 -7.21659899e-01 -2.39063427e-01 -3.68557483e-01
2.05542818e-01 3.55230898e-01 -5.49531460e-01 5.42613089e-01
7.32020557e-01 -5.83828092e-01 2.60832518e-01 6.73791111e-01
9.57947075e-02 2.97781527e-02 -6.91687226e-01 -3.13726544e-01
5.30101836e-01 -3.22037756e-01 3.13327283e-01 5.14737070e-01
2.78511465e-01 8.20477009e-01 6.59457803e-01 1.21920608e-01
-8.33731517e-02 -3.99899259e-02 -1.75715849e-01 -1.44170135e-01
9.49771285e-01 8.23028564e-01 -9.06012297e-01 -2.46879622e-01
1.99898630e-02 1.01490629e+00 1.92213923e-01 3.53368744e-02
-8.12112331e-01 -7.85477161e-01 4.67575967e-01 -7.81548843e-02
6.17397785e-01 -5.81416488e-01 -1.60730183e-01 -4.89193499e-01
2.70655274e-01 -3.65357667e-01 5.32119632e-01 -1.08383723e-01
-7.43013024e-01 2.22328424e-01 -2.89142609e-01 -8.41710329e-01
-5.21462739e-01 1.40874535e-01 3.17297056e-02 1.93316370e-01
-1.01587832e+00 -4.21796232e-01 7.37722218e-02 6.29250407e-01
2.59185415e-02 -2.53596216e-01 9.33353961e-01 -1.30119920e-01
-5.35697579e-01 5.75686693e-01 5.23264945e-01 -1.04238078e-01
-5.25378510e-02 -1.20900798e+00 -3.72927994e-01 4.42882568e-01
-6.17280424e-01 -2.88388114e-02 8.87656450e-01 -1.49987251e-01
-1.71328521e+00 -6.66973531e-01 6.65456057e-01 2.24919260e-01
3.99223596e-01 -7.95996860e-02 -5.99112478e-04 5.47372878e-01
2.90029317e-01 -3.70940268e-01 8.31220090e-01 -2.42585525e-01
3.99607234e-02 -4.01672691e-01 -1.33221400e+00 6.26302958e-01
7.06129909e-01 -1.55495867e-01 -3.33742797e-02 2.45526060e-01
4.68490273e-01 -3.63186538e-01 -1.26951325e+00 8.92780721e-02
2.83363700e-01 -7.86214054e-01 1.90016940e-01 -7.34824762e-02
-2.36961365e-01 1.56212002e-02 -3.72739077e-01 -9.62710559e-01
1.47611961e-01 -1.37722373e+00 -2.95035332e-01 3.64643246e-01
5.58877766e-01 -6.79510057e-01 9.21923816e-01 9.62209642e-01
3.47401530e-01 -7.75405824e-01 -1.47335255e+00 -8.36932659e-01
3.64158660e-01 2.56538600e-01 3.24201107e-01 5.60412884e-01
4.55678672e-01 -1.19387740e-02 -3.27918202e-01 3.62806991e-02
9.29222941e-01 2.96059668e-01 4.61135864e-01 -8.57669473e-01
-1.01230097e+00 -1.14654541e-01 1.30143464e-01 -7.58972347e-01
-3.03951263e-01 6.31648162e-03 -1.84682935e-01 -1.32261574e+00
3.68941009e-01 -6.96707904e-01 -2.88128257e-01 5.05204499e-01
3.63905162e-01 -5.09121954e-01 5.74067473e-01 -1.33876771e-01
-1.35853684e+00 7.66495645e-01 1.03841591e+00 1.41275272e-01
-3.85996491e-01 6.96572810e-02 -6.12038970e-01 4.62422848e-01
8.99871349e-01 -3.64210188e-01 -3.44276190e-01 -6.15683615e-01
5.88601828e-01 1.30940831e+00 -5.03639579e-02 -7.37623453e-01
2.66495615e-01 -5.41638374e-01 -1.07269891e-01 -2.60728419e-01
4.88736808e-01 -5.68131447e-01 -1.61966030e-02 9.08384025e-01
-6.28247201e-01 4.95023727e-01 -2.66485631e-01 8.19049418e-01
2.56052107e-01 3.27588804e-02 7.70171285e-01 -4.15222198e-01
3.05567440e-02 1.80895954e-01 -7.90853858e-01 -3.53180505e-02
1.46997058e+00 1.45539209e-01 -2.91814625e-01 -9.60304081e-01
-6.80471241e-01 8.94797027e-01 2.88062572e-01 -2.53891557e-01
2.33467892e-01 -6.11604989e-01 -3.82968932e-01 -3.10058028e-01
-4.57736701e-01 2.82000229e-02 7.28758931e-01 7.90099382e-01
-2.01025248e-01 2.56611973e-01 -1.65146634e-01 -7.00557008e-02
-9.09069836e-01 4.02404457e-01 3.40512663e-01 -4.81183171e-01
-5.39067030e-01 7.72217453e-01 -5.43820560e-01 2.25528359e-01
7.12336004e-01 2.24338874e-01 7.57460713e-01 -8.87304917e-02
4.96092051e-01 9.65045810e-01 -2.11911947e-01 3.07113498e-01
-9.25799161e-02 -1.11476451e-01 -2.23549128e-01 -1.03786695e+00
1.37915921e+00 -5.13525605e-01 5.94605990e-02 6.71148077e-02
7.31962740e-01 -4.56101358e-01 -1.76901650e+00 -3.19480926e-01
-2.05359846e-01 -4.91279662e-02 -1.15256302e-01 -7.48774886e-01
-9.27873135e-01 5.86750925e-01 6.73591197e-01 5.56705832e-01
1.10856509e+00 -1.46667555e-01 7.81488955e-01 6.35392845e-01
1.03888249e+00 -1.94930387e+00 1.64412305e-01 6.00953460e-01
3.00651103e-01 -7.42618322e-01 -2.13413224e-01 2.39048570e-01
-8.96552324e-01 6.84457541e-01 4.22207773e-01 -1.11631870e-01
3.51818502e-01 2.39806086e-01 -3.66356939e-01 -1.09416164e-01
-9.89672780e-01 -3.75974625e-01 -1.17854655e+00 4.00204770e-02
-3.20925117e-01 4.79859293e-01 -5.88534176e-01 6.30301476e-01
-2.16679975e-01 -1.34145916e-01 8.56902778e-01 1.72357166e+00
-9.68570352e-01 -1.14814913e+00 1.64936781e-02 3.72082919e-01
-6.57326519e-01 3.91305059e-01 -1.07866205e-01 5.38078249e-01
-2.45527193e-01 1.25795650e+00 1.93483338e-01 1.75333321e-02
1.10797904e-01 -3.74864899e-02 7.78241992e-01 -4.26494256e-02
-5.17535031e-01 2.89749235e-01 -6.58020824e-02 -4.65052605e-01
-8.73495340e-02 -4.91120368e-01 -1.74635458e+00 -8.26732635e-01
2.53706485e-01 5.67292869e-01 5.33402383e-01 9.78213847e-01
5.45880973e-01 1.77462175e-01 1.27567422e+00 -7.42681623e-02
-9.62794542e-01 -7.66773283e-01 -9.29557025e-01 -1.63001388e-01
1.31491229e-01 -4.74124968e-01 -3.51406813e-01 -6.20400667e-01] | [4.356756687164307, 2.8810746669769287] |
4a92ca55-31a7-4448-8b9b-326730b22049 | weakly-supervised-image-segmentation-beyond | 2301.12053 | null | https://arxiv.org/abs/2301.12053v1 | https://arxiv.org/pdf/2301.12053v1.pdf | Weakly Supervised Image Segmentation Beyond Tight Bounding Box Annotations | Weakly supervised image segmentation approaches in the literature usually achieve high segmentation performance using tight bounding box supervision and decrease the performance greatly when supervised by loose bounding boxes. However, compared with loose bounding box, it is much more difficult to acquire tight bounding box due to its strict requirements on the precise locations of the four sides of the box. To resolve this issue, this study investigates whether it is possible to maintain good segmentation performance when loose bounding boxes are used as supervision. For this purpose, this work extends our previous parallel transformation based multiple instance learning (MIL) for tight bounding box supervision by integrating an MIL strategy based on polar transformation to assist image segmentation. The proposed polar transformation based MIL formulation works for both tight and loose bounding boxes, in which a positive bag is defined as pixels in a polar line of a bounding box with one endpoint located inside the object enclosed by the box and the other endpoint located at one of the four sides of the box. Moreover, a weighted smooth maximum approximation is introduced to incorporate the observation that pixels closer to the origin of the polar transformation are more likely to belong to the object in the box. The proposed approach was evaluated on two public datasets using dice coefficient when bounding boxes at different precision levels were considered in the experiments. The results demonstrate that the proposed approach achieves state-of-the-art performance for bounding boxes at all precision levels and is robust to mild and moderate errors in the loose bounding box annotations. The codes are available at \url{https://github.com/wangjuan313/wsis-beyond-tightBB}. | ['Bin Xia', 'Juan Wang'] | 2023-01-28 | null | null | null | null | ['multiple-instance-learning'] | ['methodology'] | [ 2.37688720e-01 2.49949083e-01 -4.76259232e-01 -4.03432459e-01
-1.01473224e+00 -4.78766084e-01 1.45960137e-01 4.30115134e-01
-5.37282944e-01 8.38725626e-01 -2.32246920e-01 -1.06372833e-01
-2.12921754e-01 -6.97528899e-01 -1.07070017e+00 -1.08734739e+00
1.66727439e-01 7.08658934e-01 5.00382185e-01 1.36759087e-01
2.96006620e-01 3.33248347e-01 -1.03827000e+00 1.30994275e-01
1.11343884e+00 9.94483948e-01 7.20992163e-02 2.65711606e-01
-2.93057039e-02 7.45115653e-02 -6.04399502e-01 -3.74442816e-01
5.58895588e-01 -6.19408973e-02 -6.55143023e-01 1.13030642e-01
4.41204995e-01 -1.65306672e-01 3.79484773e-01 1.12293804e+00
2.08891109e-01 1.39918983e-01 7.55767703e-01 -1.17021585e+00
-3.42638552e-01 3.90391052e-01 -1.18085110e+00 1.33369207e-01
2.62584630e-02 -1.42970413e-01 1.04167461e+00 -6.47250175e-01
4.58764285e-01 7.52486467e-01 7.20821381e-01 2.49612704e-01
-1.36262655e+00 -7.28261054e-01 2.38436162e-01 -1.87357053e-01
-1.46314871e+00 5.96268550e-02 6.09930933e-01 -5.08615077e-01
4.41337049e-01 3.34827393e-01 3.26539099e-01 4.85853344e-01
-1.69311594e-02 7.52999306e-01 1.24690747e+00 -3.63519222e-01
1.15208305e-01 4.55838531e-01 4.60445762e-01 5.40104747e-01
4.00796771e-01 -1.92602918e-01 -9.30085108e-02 -9.60229412e-02
8.01073372e-01 -1.61208451e-01 -2.22877979e-01 -5.58733284e-01
-1.05216599e+00 8.29313397e-01 7.00779319e-01 3.11723262e-01
-1.49003223e-01 -1.41178608e-01 4.00187284e-01 -4.86107945e-01
6.21710837e-01 1.15632236e-01 -3.95323664e-01 9.31954160e-02
-1.31842387e+00 2.52689660e-01 5.91164470e-01 9.71805155e-01
6.08372211e-01 -4.70018953e-01 -1.35734871e-01 8.11185956e-01
2.17879415e-01 1.57592177e-01 7.86459744e-02 -8.44668210e-01
6.70603454e-01 7.10087359e-01 1.39937118e-01 -8.62792790e-01
-2.14702755e-01 -3.43290687e-01 -7.80731559e-01 1.50491923e-01
9.84874666e-01 -2.87899487e-02 -9.50904131e-01 1.36448979e+00
7.35371351e-01 2.30270237e-01 -2.44712666e-01 1.03016174e+00
6.55291378e-01 7.42443740e-01 6.16730489e-02 -1.67072803e-01
1.62421131e+00 -1.02267253e+00 -5.66994965e-01 -8.36308300e-02
6.00616693e-01 -6.44806206e-01 1.16876948e+00 3.75216633e-01
-8.67631972e-01 -4.91629004e-01 -1.03882682e+00 -1.10734895e-01
-3.04805458e-01 3.36229354e-01 3.55933458e-01 7.14457929e-01
-5.59769809e-01 3.57079357e-01 -8.97564352e-01 -2.43024584e-02
6.91360116e-01 4.24561769e-01 -3.73123795e-01 5.95298260e-02
-8.55963707e-01 6.50331736e-01 4.87598598e-01 1.41393021e-01
-4.39619541e-01 -9.21248674e-01 -7.91336417e-01 -5.90477474e-02
5.49903870e-01 -2.70918190e-01 7.55996823e-01 -9.94929135e-01
-9.90717232e-01 9.50321198e-01 -1.37168854e-01 -6.39188230e-01
8.49988163e-01 -1.95475265e-01 1.94806397e-01 7.98620060e-02
2.86754042e-01 7.38996267e-01 6.63125515e-01 -1.38083220e+00
-5.98653376e-01 -7.00146616e-01 -7.06261843e-02 2.25607708e-01
-1.97261021e-01 2.76747886e-02 -6.50782526e-01 -6.65340900e-01
2.43841916e-01 -9.22346056e-01 -1.37766778e-01 5.87233379e-02
-6.43086433e-01 -2.43777648e-01 9.41245794e-01 -8.31438184e-01
1.28404224e+00 -2.10425925e+00 -1.12924188e-01 3.45216662e-01
1.80893149e-02 3.55761439e-01 4.18727696e-01 -9.21123028e-02
-3.66086029e-02 2.88688153e-01 -6.14035249e-01 -3.23585868e-01
-5.01165539e-02 3.00213963e-01 -1.20573536e-01 7.18955994e-01
3.19922306e-02 4.82445449e-01 -6.63088620e-01 -7.37177789e-01
2.61482209e-01 4.83562827e-01 -3.05885911e-01 -4.04314734e-02
-1.00518294e-01 5.45524895e-01 -3.84853601e-01 6.03862822e-01
9.24466908e-01 -1.76928490e-01 -1.82893127e-01 -2.14230031e-01
-1.30440608e-01 -8.37144628e-02 -1.56979632e+00 1.25480127e+00
-3.70735414e-02 2.56501853e-01 1.66607037e-01 -1.10789454e+00
1.01411855e+00 2.45605469e-01 5.86132407e-01 -3.20368469e-01
2.90975384e-02 3.13465714e-01 3.34501714e-02 -1.19216800e-01
2.90382147e-01 -3.14735144e-01 1.18324563e-01 -7.78567567e-02
-3.24170649e-01 -2.51822770e-01 4.18114215e-01 -4.11857106e-02
5.43094337e-01 2.45534882e-01 5.50942004e-01 -3.55049819e-01
8.33820105e-01 5.34084849e-02 9.26685095e-01 4.51119572e-01
-5.27427375e-01 7.03019917e-01 7.83680916e-01 -2.25036532e-01
-1.01077986e+00 -1.00997794e+00 -7.64139414e-01 8.04400384e-01
3.66682917e-01 -1.30871266e-01 -9.80582476e-01 -7.81498969e-01
-2.61150431e-02 5.16976118e-01 -6.62835896e-01 3.47695976e-01
-5.35288572e-01 -8.65711331e-01 4.48548973e-01 6.40047073e-01
8.34356308e-01 -6.93328857e-01 -4.97510970e-01 -1.81600839e-01
-2.22237438e-01 -1.19561088e+00 -5.43946147e-01 2.12152213e-01
-1.03000474e+00 -1.05168700e+00 -9.76390123e-01 -6.17106259e-01
1.03066587e+00 -1.20828480e-01 6.62202477e-01 -1.36203319e-01
-2.09094081e-02 -1.55213609e-01 -3.19636971e-01 -2.86478728e-01
2.06709243e-02 8.33341330e-02 -2.86965489e-01 1.41261565e-02
2.14843139e-01 -2.61787847e-02 -5.75935304e-01 7.88208365e-01
-9.14962411e-01 4.22544479e-02 2.47227848e-01 8.81674111e-01
1.15313017e+00 -4.59892936e-02 2.79781282e-01 -9.85598922e-01
1.55615196e-01 -4.24826920e-01 -9.16376770e-01 1.39689624e-01
-4.36100483e-01 -1.50721803e-01 4.12452251e-01 -2.88125575e-01
-9.04656827e-01 1.60987586e-01 -1.49970958e-02 -9.89151970e-02
-3.83803546e-01 1.93536326e-01 -2.50342011e-01 7.07924366e-02
3.78501952e-01 -1.39198199e-01 -2.01726794e-01 -2.96662301e-01
1.38440952e-01 5.87714136e-01 3.03807944e-01 -8.10574174e-01
6.01224124e-01 7.73717105e-01 1.17791936e-01 -5.95971882e-01
-9.72875476e-01 -6.96082830e-01 -8.78634334e-01 -1.73598513e-01
1.07718837e+00 -5.56109488e-01 -5.17117202e-01 1.86686948e-01
-7.74915695e-01 -2.31605679e-01 -1.11138545e-01 4.83716220e-01
-4.27427590e-01 4.79291648e-01 -4.38884020e-01 -7.20414519e-01
-3.99183482e-01 -1.45749390e+00 1.21914852e+00 4.57279712e-01
-2.72962630e-01 -9.37407970e-01 -3.18078250e-02 9.88500118e-01
-1.65193617e-01 4.78412151e-01 7.99009681e-01 -8.45174789e-01
-4.89788353e-01 -4.17584717e-01 -3.87505800e-01 6.30441248e-01
8.17551389e-02 1.79969624e-01 -8.09074223e-01 -2.27851227e-01
-1.03450112e-01 6.30443096e-02 6.63958371e-01 8.11979949e-01
1.15926671e+00 -1.38791516e-01 -4.49085891e-01 6.26617014e-01
1.47860527e+00 2.64626354e-01 5.44273019e-01 6.10128641e-01
6.01408839e-01 6.90225840e-01 1.05393684e+00 2.45255843e-01
1.89568833e-01 7.96544373e-01 3.55413944e-01 -2.26307705e-01
1.66003153e-01 9.46676135e-02 1.52424365e-01 1.65899441e-01
-9.33192223e-02 -1.21125244e-01 -1.07905579e+00 6.59985304e-01
-1.68959594e+00 -5.80642641e-01 -4.49949652e-01 2.53080821e+00
8.89499366e-01 3.67404938e-01 2.17586339e-01 2.72393852e-01
9.06216145e-01 1.39282525e-01 -4.61530268e-01 -4.96936202e-01
9.15951356e-02 -9.46272090e-02 7.31701553e-01 5.93864083e-01
-1.35567808e+00 7.27923989e-01 4.63607168e+00 1.05762386e+00
-8.45322967e-01 1.87250555e-01 1.04820848e+00 -6.59367368e-02
2.35543415e-01 2.89167427e-02 -1.16048765e+00 6.15723252e-01
3.80269796e-01 3.67400557e-01 -6.99746385e-02 7.88135290e-01
3.48493665e-01 -6.37930572e-01 -9.56146836e-01 5.65663040e-01
-4.61775996e-02 -9.96452153e-01 -3.02165866e-01 7.62791261e-02
9.10946786e-01 -1.80159435e-01 1.97527543e-01 -7.11063221e-02
-2.17721820e-01 -9.51550543e-01 5.44590652e-01 7.39880949e-02
6.48427904e-01 -6.42926931e-01 9.07700777e-01 5.61658323e-01
-1.24779272e+00 1.81791440e-01 -1.97408825e-01 8.90316963e-02
1.03483990e-01 7.77269125e-01 -1.07159758e+00 4.41094130e-01
8.18756938e-01 3.76404136e-01 -2.96626478e-01 1.12827778e+00
-3.61082643e-01 5.72261691e-01 -4.65526819e-01 2.35779136e-01
4.28694338e-01 -6.06855989e-01 6.19707406e-01 1.04333460e+00
-2.14038976e-02 -2.15019286e-02 2.99258232e-01 1.06414080e+00
1.41158504e-02 3.62102896e-01 -3.81418526e-01 4.69461024e-01
3.47242743e-01 1.20754945e+00 -1.15640628e+00 -3.23812574e-01
-4.83746022e-01 5.10515273e-01 1.79356813e-01 3.16653162e-01
-1.23811293e+00 -3.12830925e-01 3.23018551e-01 3.20453733e-01
3.18167716e-01 -5.67502528e-02 -9.39726353e-01 -6.55416071e-01
4.03978348e-01 -6.11440957e-01 6.06649756e-01 -5.70590734e-01
-1.02991831e+00 4.02278483e-01 3.14106286e-01 -1.03269112e+00
3.19408804e-01 -5.78012705e-01 -4.37492579e-01 9.12088454e-01
-1.28857660e+00 -1.20391929e+00 -2.84472644e-01 3.74559760e-01
3.24403822e-01 4.41787630e-01 3.90335470e-01 3.76141936e-01
-7.70365894e-01 5.40501416e-01 1.77248180e-01 3.60093683e-01
6.54414117e-01 -1.29929006e+00 -1.91720143e-01 9.28089738e-01
8.25552642e-02 6.04226112e-01 7.41050065e-01 -8.60220194e-01
-4.17669415e-01 -9.08291340e-01 6.45904899e-01 -3.45226288e-01
4.29977506e-01 -3.07568610e-01 -1.05689335e+00 7.64155269e-01
-1.95000291e-01 3.73695314e-01 6.49760902e-01 -4.36362140e-02
-1.66455433e-01 -2.07059622e-01 -1.46683121e+00 3.68808389e-01
3.99341702e-01 -3.66285108e-02 -5.35468757e-01 3.55203599e-01
4.04256225e-01 -6.94592714e-01 -1.10996699e+00 6.80403054e-01
3.46712142e-01 -7.80440569e-01 9.25392807e-01 -1.89875439e-01
3.89374226e-01 -6.29168868e-01 -6.70916736e-02 -9.39518929e-01
2.93479264e-01 -4.02075760e-02 9.17139202e-02 1.39647257e+00
5.07539511e-01 -6.06031060e-01 1.03431618e+00 6.39853299e-01
-8.50638524e-02 -1.21741939e+00 -1.18065488e+00 -7.31706858e-01
4.05733436e-01 -2.64116079e-01 3.64762068e-01 7.50550270e-01
-6.98107854e-02 -2.09021136e-01 -1.59473997e-02 4.22104210e-01
6.78189278e-01 1.53902948e-01 6.82412803e-01 -1.03442454e+00
-6.26634881e-02 -3.32018763e-01 -2.81460851e-01 -9.94722843e-01
-6.02962216e-03 -8.42559993e-01 1.00856677e-01 -1.54329932e+00
3.48212034e-01 -7.04364896e-01 -1.17366277e-01 4.33973759e-01
-2.14385599e-01 3.76157522e-01 1.55942082e-01 1.32225960e-01
-3.85681689e-01 2.68251359e-01 1.36901879e+00 2.96078622e-02
-2.60007054e-01 1.85527802e-01 -3.96886200e-01 1.07501197e+00
8.84364069e-01 -4.94057775e-01 -1.43027008e-01 -2.89120615e-01
-3.34966071e-02 -6.31058961e-02 3.87920856e-01 -7.02067614e-01
2.08366945e-01 5.33850007e-02 3.11755031e-01 -7.58073211e-01
1.90986574e-01 -1.07961965e+00 -1.07750101e-02 2.23869398e-01
-2.36071482e-01 -2.51680553e-01 2.05983952e-01 4.23890203e-01
-2.11297870e-01 -5.01201153e-01 1.11059725e+00 3.27684171e-02
-3.12410176e-01 1.44853249e-01 1.01552196e-01 1.12060450e-01
1.53158271e+00 -6.31528199e-01 -2.61903644e-01 1.33916333e-01
-8.16809058e-01 3.74264091e-01 4.31558400e-01 1.87169313e-02
3.97097170e-01 -9.65893924e-01 -4.36385006e-01 8.36615488e-02
4.46915701e-02 5.12253225e-01 9.96376649e-02 1.34951067e+00
-6.86293423e-01 4.68215138e-01 -8.00893322e-05 -8.66988599e-01
-1.55408478e+00 3.24552208e-01 5.06455302e-01 -5.14005482e-01
-4.97305512e-01 9.66273248e-01 3.47411782e-01 -5.51825523e-01
3.82388115e-01 -6.95496917e-01 -5.54323941e-02 1.39436215e-01
7.59585276e-02 6.44395411e-01 -4.10223268e-02 -8.20244610e-01
-3.99769872e-01 7.76739120e-01 -1.11520931e-01 1.35142997e-01
1.07656789e+00 1.10152401e-01 -1.86733454e-01 3.69187921e-01
1.00440764e+00 3.36709544e-02 -1.37955534e+00 -1.07151888e-01
1.05568729e-01 -6.61804974e-01 8.82534236e-02 -8.39029312e-01
-1.05498385e+00 7.84953475e-01 5.89000225e-01 -4.31478024e-02
9.28303540e-01 4.79971990e-02 8.00740957e-01 -2.07625851e-01
2.03342527e-01 -1.19723451e+00 -2.68848658e-01 1.95681289e-01
5.90245783e-01 -1.44874930e+00 2.15644717e-01 -7.72164226e-01
-8.29014838e-01 8.91326606e-01 7.95252323e-01 -1.61733791e-01
5.17187953e-01 3.56831074e-01 -1.80731162e-01 -5.72415292e-02
5.72300032e-02 7.66148642e-02 5.58161676e-01 3.25578779e-01
3.92197877e-01 1.58786923e-01 -5.79688072e-01 6.92823768e-01
4.98801246e-02 -2.07664937e-01 1.12931356e-01 7.18510628e-01
-3.26841652e-01 -8.82116795e-01 -8.33034992e-01 4.99139845e-01
-6.81712866e-01 -6.24290779e-02 -8.83478671e-02 1.00472331e+00
5.63430429e-01 7.10340619e-01 1.57141402e-01 3.34656656e-01
2.73409486e-01 -8.00953060e-02 3.29232812e-01 -5.61172187e-01
-6.03752375e-01 2.71061361e-01 -1.36812344e-01 -3.39185953e-01
-4.96218950e-01 -8.92755628e-01 -1.82865667e+00 3.09376400e-02
-6.17470026e-01 2.04554051e-01 5.39923429e-01 9.18645084e-01
-5.22791371e-02 2.52603412e-01 1.77424416e-01 -5.44752359e-01
-4.33102101e-01 -8.70132029e-01 -6.99114501e-01 2.91366726e-01
2.46325746e-01 -7.51200855e-01 -3.66051853e-01 8.05820897e-02] | [9.580742835998535, 0.3040638566017151] |
3e1b3731-cf26-4679-a114-6efb54245d47 | symmetry-and-group-in-attribute-object | 2004.00587 | null | https://arxiv.org/abs/2004.00587v1 | https://arxiv.org/pdf/2004.00587v1.pdf | Symmetry and Group in Attribute-Object Compositions | Attributes and objects can compose diverse compositions. To model the compositional nature of these general concepts, it is a good choice to learn them through transformations, such as coupling and decoupling. However, complex transformations need to satisfy specific principles to guarantee the rationality. In this paper, we first propose a previously ignored principle of attribute-object transformation: Symmetry. For example, coupling peeled-apple with attribute peeled should result in peeled-apple, and decoupling peeled from apple should still output apple. Incorporating the symmetry principle, a transformation framework inspired by group theory is built, i.e. SymNet. SymNet consists of two modules, Coupling Network and Decoupling Network. With the group axioms and symmetry property as objectives, we adopt Deep Neural Networks to implement SymNet and train it in an end-to-end paradigm. Moreover, we propose a Relative Moving Distance (RMD) based recognition method to utilize the attribute change instead of the attribute pattern itself to classify attributes. Our symmetry learning can be utilized for the Compositional Zero-Shot Learning task and outperforms the state-of-the-art on widely-used benchmarks. Code is available at https://github.com/DirtyHarryLYL/SymNet. | ['Yong-Lu Li', 'Yue Xu', 'Xiaohan Mao', 'Cewu Lu'] | 2020-04-01 | symmetry-and-group-in-attribute-object-1 | http://openaccess.thecvf.com/content_CVPR_2020/html/Li_Symmetry_and_Group_in_Attribute-Object_Compositions_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Li_Symmetry_and_Group_in_Attribute-Object_Compositions_CVPR_2020_paper.pdf | cvpr-2020-6 | ['compositional-zero-shot-learning'] | ['computer-vision'] | [ 2.56169736e-01 -7.38727525e-02 -8.41849819e-02 -6.53205514e-01
-1.20625488e-01 -4.51292634e-01 6.65101290e-01 -3.09925467e-01
-6.12531267e-02 2.08932504e-01 2.04512537e-01 1.13315634e-01
-1.90594286e-01 -1.04804564e+00 -6.62132144e-01 -8.94905210e-01
4.44579512e-01 4.09708649e-01 1.80951998e-01 -2.60430396e-01
1.60487629e-02 1.95877150e-01 -1.67326021e+00 3.91943187e-01
8.15567374e-01 1.16820955e+00 1.34121373e-01 1.58416897e-01
-2.15075284e-01 8.31103146e-01 -1.65711701e-01 -6.29416764e-01
4.70876455e-01 -6.62488341e-01 -6.02770329e-01 8.34937766e-02
2.97060132e-01 -3.05125952e-01 -2.52651870e-01 1.12459719e+00
4.67631549e-01 2.84576863e-01 8.15073133e-01 -1.65297592e+00
-1.36342800e+00 9.74809825e-01 -1.73529908e-01 -3.43797922e-01
1.61980301e-01 1.88437894e-01 1.54318750e+00 -1.04745901e+00
4.75490242e-01 1.21789789e+00 3.20636034e-01 6.46037936e-01
-1.25241399e+00 -8.37791264e-01 2.96287119e-01 4.57625210e-01
-1.45972168e+00 -3.77355546e-01 1.03362679e+00 -3.74509633e-01
5.70784330e-01 3.97785753e-01 7.16591179e-01 1.10026181e+00
-1.17973328e-01 9.86862779e-01 7.40039051e-01 -1.83135584e-01
1.51237056e-01 -3.93345952e-02 -6.68356419e-02 4.25365001e-01
4.71744388e-02 4.32479009e-02 -3.70833248e-01 1.49485052e-01
4.58331168e-01 4.28578794e-01 -1.55708492e-01 -8.69062245e-01
-1.53701782e+00 6.85138941e-01 6.01946414e-01 2.28148639e-01
-1.50900140e-01 2.00473085e-01 4.07247961e-01 4.75957990e-01
1.04375541e-01 2.95808762e-01 -2.95176744e-01 1.88365117e-01
-3.58381599e-01 9.27449465e-02 6.17234647e-01 1.31842017e+00
7.53482163e-01 3.39208096e-02 -9.36851576e-02 9.51815426e-01
3.83634388e-01 3.34439009e-01 4.35016185e-01 -9.08914804e-01
8.52688402e-02 8.09102297e-01 -2.87652671e-01 -9.30631459e-01
-1.57033816e-01 -3.85824531e-01 -1.15434647e+00 5.71165569e-02
4.38828394e-02 2.92652458e-01 -7.22942770e-01 1.96919417e+00
4.17908072e-01 2.93616205e-01 -6.93653896e-03 1.02722168e+00
8.69197726e-01 7.09147513e-01 -5.46615832e-02 6.85618892e-02
1.28775883e+00 -1.02084720e+00 -4.99648094e-01 1.91080749e-01
5.82778335e-01 -6.73458874e-01 1.43629611e+00 2.99801230e-01
-9.33420837e-01 -5.86697042e-01 -1.17910039e+00 -1.58942774e-01
-4.14178431e-01 -2.41128374e-02 6.96530521e-01 2.85327256e-01
-5.71362495e-01 6.76788449e-01 -6.67069912e-01 -4.18481886e-01
3.88645381e-01 2.38819852e-01 -3.17101121e-01 3.59311998e-02
-1.20781302e+00 6.39255643e-01 5.41173935e-01 4.72512208e-02
-7.91708350e-01 -7.86598265e-01 -9.02293742e-01 2.28245080e-01
5.33446252e-01 -8.44803691e-01 1.28979886e+00 -1.08458352e+00
-1.65016282e+00 8.09974551e-01 2.90584356e-01 -4.57849540e-02
5.47240913e-01 -9.83321741e-02 -5.62859893e-01 -2.65826851e-01
8.91399682e-02 5.99316359e-01 6.30272865e-01 -1.08452642e+00
-7.61534989e-01 -2.04078928e-01 1.14790723e-02 1.31535485e-01
-5.81892431e-01 -1.46756312e-02 -2.30186224e-01 -6.36355698e-01
2.57146060e-01 -8.89257193e-01 2.73451686e-01 1.52104646e-01
-6.09489858e-01 -4.60332215e-01 7.11674571e-01 -2.34644219e-01
1.01252961e+00 -2.37575507e+00 3.27027678e-01 1.78007945e-01
3.03137481e-01 -1.26131093e-02 -2.23348007e-01 4.53404516e-01
-2.81201988e-01 -1.14249855e-01 -4.42848265e-01 -7.11073121e-03
4.69735950e-01 1.11642979e-01 -2.44576693e-01 4.23469812e-01
2.65382051e-01 1.01642287e+00 -9.68803704e-01 -3.24068874e-01
2.36902222e-01 3.91479105e-01 -6.00861311e-01 1.62768096e-01
-2.42778018e-01 1.74501404e-01 -3.36258382e-01 9.48816299e-01
6.95198476e-01 -1.34887785e-01 6.64362013e-02 -5.07055223e-01
-8.07221383e-02 7.32588172e-02 -1.37195611e+00 1.67057252e+00
-3.35586518e-01 9.59906280e-02 -2.48098314e-01 -1.07809162e+00
1.24975824e+00 1.31057858e-01 5.87166488e-01 -6.38256729e-01
2.84718633e-01 3.42795640e-01 3.14002395e-01 -3.01937371e-01
3.15075852e-02 -1.79203928e-01 -1.39338881e-01 4.44123060e-01
9.32134837e-02 1.29224718e-01 1.02000587e-01 1.18442439e-02
8.27290416e-01 4.35587525e-01 4.44266528e-01 -3.57472956e-01
7.93062329e-01 -2.65247613e-01 9.69308496e-01 2.23896831e-01
8.58851988e-03 5.15909672e-01 4.81231809e-01 -6.85325384e-01
-1.21832609e+00 -1.28079832e+00 4.70261760e-02 1.41112447e+00
4.25565332e-01 -4.08292323e-01 -4.29849416e-01 -5.11724055e-01
5.69565035e-02 8.95220995e-01 -6.25829101e-01 -6.48492515e-01
-5.25883079e-01 -4.80722159e-01 3.69203359e-01 5.91461420e-01
7.11351216e-01 -1.16627705e+00 -4.91881520e-01 2.23658174e-01
-6.13298267e-02 -7.54815578e-01 -8.06445420e-01 1.36717483e-01
-4.07729983e-01 -9.78897929e-01 -4.77300346e-01 -1.02387750e+00
5.78904688e-01 2.31677637e-01 1.07758486e+00 4.84319553e-02
-1.37186885e-01 1.09688625e-01 -4.54962939e-01 -2.65436500e-01
-1.83161587e-01 1.99628413e-01 6.71268255e-02 4.88154799e-01
6.23291731e-01 -9.61322427e-01 -6.35902822e-01 6.04417086e-01
-1.08137691e+00 2.63834685e-01 8.46375227e-01 9.13262546e-01
6.74301863e-01 -2.24585906e-01 4.68245596e-01 -7.72934973e-01
5.41426659e-01 -3.31743866e-01 -4.41771418e-01 4.23100531e-01
-5.95416844e-01 2.04335511e-01 9.74751949e-01 -8.18547428e-01
-8.25474620e-01 2.89366335e-01 1.00682892e-01 -5.80022395e-01
-8.69641230e-02 1.72421023e-01 -9.48407710e-01 1.91881150e-01
3.60760003e-01 4.42486405e-01 1.33439768e-02 -4.58243996e-01
6.56703889e-01 6.02636456e-01 6.67767823e-01 -7.98490703e-01
8.48304808e-01 5.05411804e-01 5.93620166e-03 -3.18529785e-01
-6.69188976e-01 -2.06618115e-01 -6.67930126e-01 -1.20467164e-01
8.35955083e-01 -6.62354112e-01 -9.71577942e-01 5.32159746e-01
-9.84421134e-01 -1.02635538e-02 -3.85576010e-01 3.31103384e-01
-6.92826033e-01 1.41030669e-01 -3.15024942e-01 -1.76703066e-01
-3.82309824e-01 -9.63252783e-01 7.97328472e-01 6.81052133e-02
-2.09172398e-01 -5.91200173e-01 -1.31329179e-01 -9.84671861e-02
4.10960078e-01 2.30613098e-01 1.13337672e+00 -9.50093746e-01
-6.67734206e-01 1.24877924e-03 -1.64298490e-01 5.03010511e-01
3.60368013e-01 3.13172847e-01 -7.69879103e-01 -3.21824476e-02
-5.27386516e-02 -1.81550771e-01 8.08317006e-01 -1.59536660e-01
1.19274521e+00 -3.71364295e-01 -1.04458861e-01 9.21934962e-01
1.27770293e+00 3.88423830e-01 6.09232664e-01 4.37911123e-01
9.69126701e-01 6.37915909e-01 4.73630965e-01 5.73410571e-01
5.46731353e-01 6.65488183e-01 3.97984326e-01 2.81101614e-01
-1.47981226e-01 -3.94185454e-01 3.23093832e-01 1.15017891e+00
-4.46181260e-02 5.08892201e-02 -7.71162570e-01 2.60903686e-01
-2.01201272e+00 -1.10208714e+00 5.81968762e-02 2.20142150e+00
8.60183954e-01 3.10809724e-02 1.11951195e-02 2.66055483e-02
9.09349024e-01 5.61720952e-02 -6.81744456e-01 -2.61885852e-01
-1.06184416e-01 -1.41201690e-01 1.14840731e-01 2.10266963e-01
-1.10523903e+00 8.78844380e-01 4.55130100e+00 6.85172737e-01
-1.07026541e+00 -1.16172805e-02 3.40769678e-01 -4.75352332e-02
-6.03464067e-01 1.28385335e-01 -6.49510741e-01 6.70421660e-01
2.40670025e-01 -4.33566242e-01 6.59250379e-01 1.00328088e+00
-2.12541923e-01 6.60731196e-01 -1.59133887e+00 9.85560060e-01
1.56208441e-01 -9.93840098e-01 3.22029203e-01 -1.98017150e-01
4.93317455e-01 -5.02779722e-01 6.31726235e-02 5.35681784e-01
4.16517168e-01 -8.78854692e-01 8.38967979e-01 5.73874712e-01
6.74061477e-01 -7.72770703e-01 4.62701172e-01 2.15215147e-01
-1.52492476e+00 -1.01865999e-01 -6.30080044e-01 1.07165188e-01
-1.29329264e-01 4.38985914e-01 -4.20484573e-01 7.36031234e-01
5.59824824e-01 1.00849771e+00 -3.53906304e-01 8.34955931e-01
-4.11529720e-01 1.32515311e-01 -1.72836944e-01 -1.55426204e-01
1.27467841e-01 -4.75422412e-01 2.83994317e-01 8.72066557e-01
5.51223040e-01 1.31814718e-01 9.69689414e-02 1.18359077e+00
-2.77027637e-01 2.17073172e-01 -5.45538545e-01 1.35933444e-01
8.21499884e-01 1.39407063e+00 -4.13520247e-01 -4.37227905e-01
-5.36224484e-01 9.88216281e-01 1.90272599e-01 1.17795117e-01
-9.51116204e-01 -6.37548923e-01 8.13629389e-01 -1.37319177e-01
4.45281059e-01 1.15972795e-01 -1.95555031e-01 -1.29700565e+00
1.93474278e-01 -1.07003641e+00 4.41501528e-01 -6.50006652e-01
-1.57160330e+00 4.19870287e-01 9.39343497e-03 -1.64838278e+00
1.37285277e-01 -4.62167293e-01 -8.88635278e-01 6.19533181e-01
-1.25457919e+00 -1.40606391e+00 -6.92966282e-01 6.94528103e-01
4.73459154e-01 -3.29753280e-01 6.97795093e-01 3.82948488e-01
-5.19742489e-01 6.32457197e-01 -3.66464220e-02 1.51040226e-01
9.10320044e-01 -1.17806196e+00 3.78259689e-01 5.72080195e-01
8.58391747e-02 6.78754091e-01 4.99083191e-01 -3.81599069e-01
-1.47737515e+00 -1.18635273e+00 8.02444100e-01 -2.56603807e-01
7.84817576e-01 -5.66540897e-01 -1.04446507e+00 7.25311995e-01
1.51419461e-01 6.22935705e-02 7.50874043e-01 -1.58232853e-01
-8.92028093e-01 -6.39254630e-01 -7.74185419e-01 8.10847878e-01
1.41767704e+00 -3.47200155e-01 -7.74992466e-01 2.43555903e-01
9.59063768e-01 4.87762615e-02 -8.79879475e-01 5.54484010e-01
6.84675276e-01 -8.04764926e-01 9.48018789e-01 -5.91506779e-01
5.79143405e-01 -4.46661800e-01 -4.65997010e-01 -1.43665445e+00
-8.33898902e-01 -4.66809154e-01 -5.77387176e-02 1.55346036e+00
4.02258843e-01 -7.98486292e-01 5.30838966e-01 5.46810210e-01
-2.70695984e-01 -7.04943240e-01 -6.69439554e-01 -1.17460608e+00
6.48178756e-02 -2.32122708e-02 1.21022367e+00 1.09370065e+00
-1.64184973e-01 5.30842781e-01 -4.66676444e-01 -9.71852392e-02
5.90899885e-01 3.80273193e-01 6.92985237e-01 -1.45708156e+00
-2.84226477e-01 -7.41613686e-01 -5.48767865e-01 -8.46100211e-01
2.66081661e-01 -1.31383276e+00 -9.86022502e-02 -1.29660726e+00
2.85389274e-01 -3.32018018e-01 -6.47458434e-01 5.46420932e-01
-7.63723487e-03 -1.39011532e-01 3.77959698e-01 4.67671096e-01
-4.39317256e-01 1.04307604e+00 1.34411573e+00 -3.78125757e-01
-6.70846552e-02 -1.57219425e-01 -9.51727152e-01 5.21791160e-01
1.06332016e+00 -4.59464490e-01 -5.94127893e-01 -4.05831635e-01
2.92159468e-01 -4.88544017e-01 2.76484936e-01 -9.84817743e-01
3.37285757e-01 -3.24180126e-01 2.45914429e-01 -3.24531019e-01
2.93321222e-01 -1.08749652e+00 3.42515647e-01 5.07516742e-01
-5.41936874e-01 2.04268917e-01 -5.23499370e-01 3.40456873e-01
-1.55140027e-01 -6.33281991e-02 7.26497948e-01 -1.08695313e-01
-9.24934447e-01 5.67370713e-01 1.53981537e-01 3.86052765e-02
1.31064737e+00 -9.30881128e-02 -4.43692833e-01 -6.11746609e-02
-5.74357033e-01 3.29217523e-01 5.74017704e-01 6.56865180e-01
6.84739113e-01 -2.08991599e+00 -7.96393216e-01 3.50857258e-01
5.08332908e-01 -5.85225821e-02 5.18414527e-02 8.41427922e-01
-4.11403477e-01 3.43594663e-02 -5.60887575e-01 -4.95345563e-01
-1.07102251e+00 9.26680207e-01 1.99266478e-01 2.58559585e-01
-7.03936756e-01 7.08258033e-01 4.44688529e-01 -9.37084258e-01
2.94954747e-01 -2.81416565e-01 -8.27738922e-03 1.09913357e-01
4.20053124e-01 2.74046868e-01 -1.70333594e-01 -3.55644941e-01
-4.36472178e-01 6.34922147e-01 -7.42684156e-02 1.63710415e-01
1.30341601e+00 1.26371518e-01 -3.62908632e-01 6.50068820e-01
1.19240904e+00 -3.59591246e-01 -1.06733644e+00 -5.71423173e-01
8.47654268e-02 -5.17144918e-01 -3.85208547e-01 -4.86264557e-01
-1.03986549e+00 8.78394306e-01 3.78023505e-01 1.76108077e-01
1.14549923e+00 3.50931995e-02 6.94116354e-01 6.82836890e-01
2.16406539e-01 -1.08657944e+00 1.83746427e-01 3.39142710e-01
8.79314303e-01 -1.02177751e+00 -1.97179690e-01 -4.84062105e-01
-8.93696249e-01 1.06603515e+00 9.13560510e-01 -1.55319065e-01
6.68998539e-01 -3.23227085e-02 -6.91312701e-02 4.03497592e-02
-9.41140234e-01 -1.14780746e-01 3.50957483e-01 3.99491221e-01
3.74219626e-01 3.73196483e-01 -2.43892774e-01 8.09822440e-01
-2.87405699e-01 -1.40740260e-01 2.07136989e-01 7.89090335e-01
-5.04529715e-01 -1.16765726e+00 -1.00338116e-01 3.44532192e-01
4.40645367e-02 -7.04190135e-02 -5.88856101e-01 7.22964227e-01
3.76210898e-01 5.89258194e-01 1.72828272e-01 -6.66533232e-01
5.12165606e-01 1.82480529e-01 2.42630213e-01 -6.15569413e-01
-3.56095403e-01 -1.62361428e-01 -2.12115437e-01 -5.88021517e-01
-3.64752680e-01 -6.22170150e-01 -1.25173092e+00 -5.85275471e-01
5.34104835e-03 -1.15443207e-01 3.48479569e-01 6.08197808e-01
3.43509883e-01 6.35224760e-01 8.35669875e-01 -3.27680379e-01
-9.42540765e-01 -8.44206810e-01 -5.87167919e-01 7.00311840e-01
-3.83394919e-02 -6.87662721e-01 -3.19115520e-01 1.30835116e-01] | [10.114606857299805, 2.350795269012451] |
b02db73a-fc1c-466e-bd05-e0e546942c12 | privacy-inference-empowered-stealthy-backdoor | 2306.08011 | null | https://arxiv.org/abs/2306.08011v1 | https://arxiv.org/pdf/2306.08011v1.pdf | Privacy Inference-Empowered Stealthy Backdoor Attack on Federated Learning under Non-IID Scenarios | Federated learning (FL) naturally faces the problem of data heterogeneity in real-world scenarios, but this is often overlooked by studies on FL security and privacy. On the one hand, the effectiveness of backdoor attacks on FL may drop significantly under non-IID scenarios. On the other hand, malicious clients may steal private data through privacy inference attacks. Therefore, it is necessary to have a comprehensive perspective of data heterogeneity, backdoor, and privacy inference. In this paper, we propose a novel privacy inference-empowered stealthy backdoor attack (PI-SBA) scheme for FL under non-IID scenarios. Firstly, a diverse data reconstruction mechanism based on generative adversarial networks (GANs) is proposed to produce a supplementary dataset, which can improve the attacker's local data distribution and support more sophisticated strategies for backdoor attacks. Based on this, we design a source-specified backdoor learning (SSBL) strategy as a demonstration, allowing the adversary to arbitrarily specify which classes are susceptible to the backdoor trigger. Since the PI-SBA has an independent poisoned data synthesis process, it can be integrated into existing backdoor attacks to improve their effectiveness and stealthiness in non-IID scenarios. Extensive experiments based on MNIST, CIFAR10 and Youtube Aligned Face datasets demonstrate that the proposed PI-SBA scheme is effective in non-IID FL and stealthy against state-of-the-art defense methods. | ['Longfei Zheng', 'Jun Wu', 'Gaolei Li', 'Haochen Mei'] | 2023-06-13 | null | null | null | null | ['backdoor-attack'] | ['adversarial'] | [-9.11098197e-02 -2.60810554e-01 -2.38787889e-01 -1.98224932e-01
-4.27533865e-01 -1.03528106e+00 5.30404031e-01 -5.05433261e-01
-2.46116724e-02 5.90026617e-01 4.09686305e-02 -5.17466187e-01
-3.25476490e-02 -1.02395916e+00 -6.97804630e-01 -1.03574347e+00
3.36695351e-02 -8.24829265e-02 7.54254907e-02 -1.96113110e-01
-1.41491294e-01 6.14919126e-01 -1.01188517e+00 3.18181545e-01
6.23461545e-01 9.13046896e-01 -4.63773698e-01 1.18775725e-01
-1.88878685e-01 7.61913657e-01 -7.37840533e-01 -9.26131666e-01
7.62159050e-01 -4.89051372e-01 -3.44498992e-01 -3.92753243e-01
1.95954129e-01 -6.66742325e-01 -6.61005914e-01 1.26443481e+00
5.95160544e-01 -3.74416083e-01 3.08247685e-01 -1.83527267e+00
-6.20174229e-01 7.05574811e-01 -6.50235534e-01 2.14295518e-02
2.25876182e-01 6.27167284e-01 4.67074573e-01 -5.09329736e-01
4.33951676e-01 1.46090829e+00 5.07522941e-01 9.27428365e-01
-1.08562422e+00 -1.53233767e+00 1.62374794e-01 -1.85714543e-01
-1.25866103e+00 -3.99820060e-01 8.63355935e-01 -1.94824964e-01
5.85823432e-02 5.73467314e-01 3.53526294e-01 1.59486997e+00
1.29012272e-01 8.12411726e-01 1.31003988e+00 8.19260404e-02
3.58131409e-01 3.93087655e-01 -4.36533242e-02 5.19951165e-01
5.79948425e-01 5.64851165e-01 -4.70313013e-01 -8.31362188e-01
7.43015826e-01 4.56596941e-01 -3.75434816e-01 -3.92642558e-01
-5.06345630e-01 1.03396714e+00 3.41657281e-01 -1.51764229e-01
-2.17289105e-02 -4.59299497e-02 4.98907477e-01 4.08912063e-01
2.16991425e-01 -6.75000399e-02 -4.06168431e-01 2.91160792e-01
-6.32024109e-01 4.03588295e-01 8.66138101e-01 7.11645246e-01
5.89556575e-01 1.44920379e-01 -3.51555020e-01 2.69784600e-01
4.40288812e-01 6.27706945e-01 4.10152942e-01 -7.64102757e-01
5.55520236e-01 3.87385517e-01 -1.78747207e-01 -1.25479066e+00
2.14865774e-01 -5.11578858e-01 -9.11897779e-01 2.47163698e-01
3.39078873e-01 -4.40419376e-01 -6.20258152e-01 2.18062186e+00
6.23474240e-01 4.19475943e-01 2.31143385e-01 7.81466901e-01
5.35659611e-01 4.42144156e-01 4.25101221e-02 -3.93604070e-01
1.32126307e+00 -5.45111895e-01 -7.78302014e-01 2.33993456e-01
2.70077527e-01 -2.06286952e-01 1.08328235e+00 3.39792341e-01
-6.14562035e-01 -4.72977720e-02 -8.50537121e-01 3.12136739e-01
-3.63590300e-01 -4.11836177e-01 8.97662342e-01 1.28928387e+00
-5.69857001e-01 8.77471939e-02 -6.51350796e-01 3.05105653e-02
1.06356287e+00 4.66410279e-01 -5.11312246e-01 -7.83061758e-02
-1.46085095e+00 -9.00629535e-03 2.52231777e-01 -1.69013858e-01
-1.50617945e+00 -8.26497376e-01 -5.67448735e-01 8.21030438e-02
6.83399439e-01 -6.10482574e-01 8.16514015e-01 -7.03069627e-01
-1.53512633e+00 5.51173210e-01 2.99348116e-01 -6.59860432e-01
9.45321023e-01 8.53645355e-02 -6.09225690e-01 8.50503445e-02
-1.53210595e-01 6.37300462e-02 1.16516161e+00 -1.36937344e+00
-2.51307219e-01 -7.49394536e-01 2.29109064e-01 -1.61839738e-01
-7.41062462e-01 1.36026338e-01 -1.48630664e-01 -8.41660440e-01
-3.67781132e-01 -8.27393591e-01 -6.18160442e-02 1.30102411e-01
-7.66704738e-01 8.94497931e-02 1.60728085e+00 -3.11769456e-01
1.23151612e+00 -2.33938336e+00 -3.27173144e-01 3.33821714e-01
3.42384696e-01 6.57288432e-01 -9.23483148e-02 3.84516805e-01
9.26090255e-02 4.70922142e-01 -2.02344194e-01 -2.95409530e-01
-6.63364306e-02 2.05402672e-01 -8.30416441e-01 5.43846965e-01
-1.87672243e-01 6.52823985e-01 -6.87501669e-01 -3.26427311e-01
-1.59188993e-02 5.60175657e-01 -6.81996763e-01 3.74528795e-01
-2.01731518e-01 6.57361746e-01 -7.71024287e-01 8.81618023e-01
1.27239537e+00 -3.73941436e-02 1.19272485e-01 -5.80856577e-02
2.44994387e-01 -2.47568205e-01 -1.15704167e+00 1.13246870e+00
-1.70471653e-01 -5.45696393e-02 3.09325099e-01 -3.90508622e-01
8.29299688e-01 2.98401445e-01 3.00410062e-01 -3.03992271e-01
3.53295833e-01 4.35582213e-02 -1.92557752e-01 -3.16576660e-01
-1.00815102e-01 -1.74829766e-01 -1.79997027e-01 6.38838172e-01
-1.08274333e-01 6.78417385e-01 -5.21969020e-01 2.19865769e-01
1.16833234e+00 -2.59128273e-01 2.61599664e-02 -1.56367421e-01
7.81001091e-01 -5.22232771e-01 9.49361145e-01 8.23965073e-01
-3.78589302e-01 2.81527787e-01 6.96237981e-01 -4.13823336e-01
-4.61802214e-01 -1.19568765e+00 9.59459227e-03 8.66382658e-01
2.77729243e-01 -4.49301660e-01 -9.24419940e-01 -1.34639919e+00
3.12193155e-01 5.33859253e-01 -5.69620967e-01 -4.20885861e-01
-3.70318174e-01 -6.99432313e-01 1.19293547e+00 6.65050000e-02
1.08120632e+00 -7.56298602e-01 -1.85966030e-01 -1.72962949e-01
3.76220495e-02 -9.49109674e-01 -6.24494910e-01 -2.44792119e-01
-4.97618675e-01 -1.18241024e+00 -3.22535098e-01 -1.55517697e-01
6.07946157e-01 3.28741521e-01 4.67478067e-01 6.12280704e-03
-1.00233443e-01 2.03692451e-01 -1.57455549e-01 -4.44999188e-01
-5.10742366e-01 -6.02763295e-02 2.40510002e-01 6.03361905e-01
3.79256397e-01 -7.25199759e-01 -7.49964416e-01 3.86171609e-01
-1.25115097e+00 -5.60332775e-01 3.31588805e-01 7.66898155e-01
3.97237748e-01 3.17318499e-01 5.83489656e-01 -1.37539041e+00
8.58234882e-01 -8.22876036e-01 -7.09033430e-01 2.99995512e-01
-5.64714551e-01 -6.87680766e-02 1.12798977e+00 -7.94467747e-01
-1.11175776e+00 1.78775303e-02 -1.40904918e-01 -1.06910884e+00
-4.18675598e-03 -1.82448234e-02 -9.93342698e-01 -4.48992103e-01
5.17298460e-01 4.83152568e-01 1.93324521e-01 -5.20126283e-01
4.29587901e-01 8.40528548e-01 3.93914998e-01 -6.81300581e-01
1.20530128e+00 7.60795355e-01 1.89434588e-01 -5.34159064e-01
-5.00896394e-01 2.37127230e-01 2.29797408e-01 1.14629798e-01
5.71170449e-01 -9.00749505e-01 -1.16458189e+00 8.22982192e-01
-9.18790877e-01 1.05007634e-01 -5.50193787e-02 1.29468650e-01
-2.58038223e-01 4.64936674e-01 -5.31666934e-01 -9.27177966e-01
-6.61633551e-01 -1.25603127e+00 6.42988861e-01 3.71610940e-01
3.00994396e-01 -8.25083017e-01 -1.10345967e-01 5.22310853e-01
4.07471865e-01 6.41687393e-01 7.12761104e-01 -1.05149567e+00
-7.64700830e-01 -2.29754746e-01 8.38688836e-02 4.83578652e-01
2.70056039e-01 -1.30932212e-01 -1.02029109e+00 -6.26874506e-01
4.52729642e-01 -3.81161690e-01 5.68349779e-01 -8.48021433e-02
1.65058064e+00 -1.09979212e+00 -2.78530180e-01 1.14851558e+00
1.47696650e+00 1.78520769e-01 6.44856572e-01 1.02536149e-01
6.93116188e-01 3.66990268e-01 3.62874895e-01 7.57237256e-01
2.01164976e-01 3.44124317e-01 8.04660738e-01 7.99781829e-02
2.14574024e-01 -9.00088072e-01 4.84802037e-01 1.60758980e-02
5.13369024e-01 -3.87220502e-01 -3.96237165e-01 3.69454809e-02
-1.52609980e+00 -1.13157809e+00 1.64606482e-01 2.39662766e+00
8.09234142e-01 -4.64231372e-02 2.54342347e-01 -2.29729153e-02
6.47377551e-01 3.23669791e-01 -8.16169024e-01 -3.24172527e-01
-2.04885304e-01 7.98742697e-02 7.03576744e-01 1.35150120e-01
-9.52056050e-01 9.00301337e-01 5.26801825e+00 1.12120104e+00
-1.23021686e+00 4.67170417e-01 8.43465447e-01 -1.18289448e-01
-6.05584383e-01 5.21847457e-02 -1.02965236e+00 9.74485397e-01
6.40510261e-01 -2.48611227e-01 6.73632622e-01 9.39954400e-01
1.43586081e-02 5.36713660e-01 -8.45789969e-01 9.97853994e-01
-1.02064207e-01 -1.34030998e+00 3.15458268e-01 4.98769015e-01
4.59198803e-01 -2.75354207e-01 4.54881996e-01 2.41946608e-01
6.67056143e-01 -9.44856763e-01 2.43386686e-01 2.21406043e-01
1.02761960e+00 -1.20822597e+00 4.14012939e-01 5.99851251e-01
-7.63522446e-01 -5.23793519e-01 -2.98366904e-01 4.23256487e-01
-1.43487796e-01 3.88426781e-01 -2.26594388e-01 6.19530439e-01
6.84507012e-01 1.03993237e-01 -3.71109515e-01 4.39155221e-01
-3.30381781e-01 8.18219781e-01 -5.30264974e-01 1.34508476e-01
1.68570802e-01 -8.45478252e-02 7.26919830e-01 7.82960117e-01
8.64015371e-02 2.02474996e-01 7.42982626e-02 1.09717071e+00
-5.46456099e-01 -8.11535418e-02 -1.04563642e+00 4.96951342e-02
9.68458056e-01 1.18657529e+00 -4.14735861e-02 8.65143612e-02
-5.65382652e-02 8.98126841e-01 7.57701844e-02 2.77010232e-01
-9.72184837e-01 -3.76626521e-01 1.09151876e+00 2.10103720e-01
9.10198987e-02 1.14405341e-01 -2.46810187e-02 -1.34210169e+00
-1.55229524e-01 -1.34874773e+00 7.13007510e-01 -1.71644509e-01
-1.69749212e+00 5.59821367e-01 -1.77496433e-01 -1.03358519e+00
8.76719877e-02 -6.82544932e-02 -8.36244166e-01 7.75390804e-01
-1.41331255e+00 -1.44816864e+00 -6.84547201e-02 1.31802452e+00
-8.70118737e-02 -5.00530779e-01 6.91922784e-01 2.55987048e-01
-8.90008152e-01 1.38854527e+00 6.77393228e-02 3.64717454e-01
4.51566309e-01 -5.28234899e-01 1.15494713e-01 9.97205198e-01
4.51122299e-02 9.93681312e-01 4.88076419e-01 -8.19373727e-01
-1.79757285e+00 -1.39601791e+00 8.10493827e-02 -3.63506496e-01
4.33368176e-01 -6.64166391e-01 -8.05134296e-01 7.37495303e-01
-1.27646968e-01 5.10841548e-01 1.04495895e+00 -4.32211161e-01
-8.28070641e-01 -5.16231716e-01 -2.02000928e+00 6.53642178e-01
9.63602066e-01 -6.80079877e-01 -4.80596498e-02 4.01876599e-01
1.04559803e+00 -1.76035285e-01 -5.59657276e-01 3.81080717e-01
4.23606366e-01 -1.12897980e+00 1.00644982e+00 -8.21435928e-01
-2.92987879e-02 -2.63811082e-01 -2.09970191e-01 -6.74314678e-01
1.47034079e-01 -1.27373183e+00 -5.19572258e-01 1.69077599e+00
-1.53282270e-01 -1.07304764e+00 1.11726344e+00 5.62305212e-01
6.15225315e-01 -7.05631673e-01 -1.07628334e+00 -1.00463724e+00
1.85340181e-01 -1.73731282e-01 1.29693568e+00 1.05220687e+00
-4.76337999e-01 -1.87288612e-01 -7.23828971e-01 5.02134025e-01
1.08775711e+00 -1.68737546e-01 1.14285743e+00 -8.51481438e-01
-4.22421694e-01 3.44694816e-02 -1.43471241e-01 -6.64491653e-01
3.34171832e-01 -7.51152694e-01 -6.38433158e-01 -5.44620574e-01
7.61591941e-02 -5.53099036e-01 -4.14296508e-01 6.77843213e-01
-1.55611895e-02 7.73135126e-02 2.44831368e-01 3.23892504e-01
-1.58553645e-01 6.73973918e-01 1.00172246e+00 1.67879183e-03
-1.07357100e-01 2.94291109e-01 -9.83873188e-01 4.61828828e-01
7.83564687e-01 -7.00544178e-01 -8.54920745e-01 -5.42510971e-02
-2.08993122e-01 1.93887845e-01 5.15808225e-01 -7.37642646e-01
1.67227030e-01 -3.93856287e-01 1.05582386e-01 -3.69894415e-01
8.25267136e-02 -1.14822817e+00 4.36440021e-01 7.04491496e-01
-2.05466658e-01 -1.00801542e-01 -1.84766133e-03 9.03526425e-01
2.85440125e-02 2.56390572e-01 7.83501804e-01 -2.50728011e-01
-1.43491045e-01 9.15430486e-01 1.09316237e-01 2.17900649e-01
1.34516239e+00 -9.77875739e-02 -6.78117394e-01 -2.85345703e-01
-2.34210283e-01 2.30303034e-01 6.03287578e-01 2.89108723e-01
6.69606030e-01 -1.28059602e+00 -5.55612266e-01 8.24553668e-01
-7.83132389e-02 -2.87462711e-01 3.94031107e-01 4.06482965e-01
-2.37869903e-01 7.41508305e-02 -1.71611711e-01 -2.23425344e-01
-1.37745667e+00 9.02115703e-01 3.79722089e-01 -2.80782551e-01
-4.78925377e-01 8.37200940e-01 4.16371405e-01 -5.17283022e-01
3.13814700e-01 4.47195947e-01 1.76554590e-01 -2.45891064e-01
7.15752840e-01 3.28744352e-01 -2.81650990e-01 -5.15448093e-01
-4.07751679e-01 7.34480396e-02 -1.71047151e-01 9.58940983e-02
1.04213953e+00 -3.69736888e-02 -2.03068212e-01 -2.09645778e-01
1.26169479e+00 4.46144819e-01 -1.35049844e+00 -1.35463879e-01
-5.60308099e-01 -1.02401376e+00 -1.32728517e-01 -8.64170313e-01
-1.53082299e+00 6.93850815e-01 7.31018245e-01 2.40992382e-01
1.26305985e+00 -3.82163703e-01 1.09728420e+00 6.41154647e-02
6.37803733e-01 -4.15284008e-01 -1.28865801e-02 2.21552365e-02
4.75820094e-01 -9.14439678e-01 -2.22411647e-01 -5.30821562e-01
-5.90580881e-01 7.46342242e-01 8.09046268e-01 -9.65647120e-03
8.57135296e-01 3.91382992e-01 -3.35172452e-02 -5.93899786e-02
-6.67780817e-01 4.54600722e-01 -3.95527065e-01 6.75653696e-01
-4.75492239e-01 -2.73740962e-02 -2.58174986e-01 1.13507712e+00
2.18292978e-02 9.05060768e-02 3.83518010e-01 8.48234475e-01
6.99060187e-02 -1.40312386e+00 -4.69021797e-01 1.71129882e-01
-8.76847327e-01 1.27205372e-01 -4.66099799e-01 6.55023694e-01
3.68382812e-01 8.40806961e-01 -4.06392276e-01 -8.31523359e-01
-9.57046896e-02 -1.20155439e-01 3.70530188e-02 -2.87623227e-01
-9.53444004e-01 -1.38775766e-01 -3.94804120e-01 -6.93738222e-01
1.42676905e-01 -4.24734384e-01 -1.02830219e+00 -7.49586284e-01
-3.27607930e-01 2.37026781e-01 5.12284875e-01 5.74287415e-01
7.13594735e-01 8.97819772e-02 1.36072242e+00 2.20286883e-02
-1.12426233e+00 -3.11979711e-01 -7.23291159e-01 4.51880902e-01
2.28026733e-01 -4.67667997e-01 -7.28446960e-01 -5.04671097e-01] | [5.8331403732299805, 7.117053985595703] |
6f73e663-17f0-4f39-830d-038f3fed5c10 | improved-dual-correlation-reduction-network | 2202.12533 | null | https://arxiv.org/abs/2202.12533v1 | https://arxiv.org/pdf/2202.12533v1.pdf | Improved Dual Correlation Reduction Network | Deep graph clustering, which aims to reveal the underlying graph structure and divide the nodes into different clusters without human annotations, is a fundamental yet challenging task. However, we observed that the existing methods suffer from the representation collapse problem and easily tend to encode samples with different classes into the same latent embedding. Consequently, the discriminative capability of nodes is limited, resulting in sub-optimal clustering performance. To address this problem, we propose a novel deep graph clustering algorithm termed Improved Dual Correlation Reduction Network (IDCRN) through improving the discriminative capability of samples. Specifically, by approximating the cross-view feature correlation matrix to an identity matrix, we reduce the redundancy between different dimensions of features, thus improving the discriminative capability of the latent space explicitly. Meanwhile, the cross-view sample correlation matrix is forced to approximate the designed clustering-refined adjacency matrix to guide the learned latent representation to recover the affinity matrix even across views, thus enhancing the discriminative capability of features implicitly. Moreover, we avoid the collapsed representation caused by the over-smoothing issue in Graph Convolutional Networks (GCNs) through an introduced propagation regularization term, enabling IDCRN to capture the long-range information with the shallow network structure. Extensive experimental results on six benchmarks have demonstrated the effectiveness and the efficiency of IDCRN compared to the existing state-of-the-art deep graph clustering algorithms. | ['Xihong Yang', 'Wenxuan Tu', 'Xinwang Liu', 'Sihang Zhou', 'Yue Liu'] | 2022-02-25 | null | null | null | null | ['graph-clustering'] | ['graphs'] | [-3.21721107e-01 -7.55683929e-02 -4.95199300e-02 -2.18623862e-01
-2.02879369e-01 -4.93643910e-01 3.33667547e-01 2.58536004e-02
7.51867518e-03 1.22069776e-01 2.09390074e-01 2.02612206e-01
-5.87915480e-01 -8.12837481e-01 -3.04280847e-01 -1.23814189e+00
-8.84543434e-02 2.95569181e-01 -4.26885039e-02 2.51957744e-01
-5.14486842e-02 3.27428102e-01 -1.15076089e+00 -7.98103772e-03
9.52659667e-01 7.85203576e-01 1.67350352e-01 -9.14980769e-02
-1.35770485e-01 5.27812719e-01 -2.96553701e-01 -2.13391036e-01
1.87674031e-01 -2.89879382e-01 -2.67666787e-01 4.51669782e-01
3.04067075e-01 -1.35055259e-01 -9.00442302e-01 1.49966145e+00
3.59906256e-01 1.76866874e-01 4.60236669e-01 -1.36530018e+00
-8.89678419e-01 6.84358537e-01 -9.53074753e-01 -2.29168326e-01
-9.53318477e-02 6.90970756e-03 1.12474871e+00 -8.27942729e-01
5.01918316e-01 1.39281714e+00 4.85045046e-01 2.70314038e-01
-1.48033822e+00 -8.40559959e-01 4.66407478e-01 1.62696138e-01
-1.93431532e+00 -2.30042711e-02 1.11732042e+00 -4.71110880e-01
3.39197040e-01 1.64498433e-01 7.05137551e-01 7.42957890e-01
-3.49252611e-01 6.39754057e-01 5.50924718e-01 2.29040369e-01
-1.24550425e-02 -9.51380506e-02 2.13025153e-01 1.02347040e+00
4.18636262e-01 -3.51016670e-01 -2.40153909e-01 -3.94501537e-02
8.70948613e-01 5.94655514e-01 -5.43590903e-01 -9.56967294e-01
-1.07983935e+00 7.99437046e-01 9.89811182e-01 4.39262986e-01
-9.70613658e-02 4.00272794e-02 6.01097047e-01 -3.81202921e-02
3.98301780e-01 -2.78662406e-02 -7.88352638e-02 4.19606209e-01
-6.74984813e-01 -2.10841462e-01 3.27474147e-01 1.01557481e+00
1.11054742e+00 -6.68442249e-02 -1.52641341e-01 7.61557043e-01
5.66139221e-01 8.56909230e-02 4.57374036e-01 -7.88514555e-01
7.40828812e-01 1.39801323e+00 -4.47969317e-01 -1.77456677e+00
-3.36977184e-01 -9.51134145e-01 -1.69270611e+00 -3.33632529e-01
2.15366736e-01 9.06270966e-02 -8.19897473e-01 1.76358950e+00
2.74119705e-01 1.76813692e-01 -2.09362924e-01 1.04857481e+00
7.45710433e-01 6.77259803e-01 -2.38525555e-01 -3.10399115e-01
1.22738385e+00 -9.27980065e-01 -6.96175516e-01 -1.08534314e-01
5.78306735e-01 -4.20941561e-01 9.59044874e-01 1.10542774e-01
-5.49142420e-01 -6.30501628e-01 -1.11336017e+00 -2.87962426e-02
-1.41910374e-01 1.62550420e-01 8.66471171e-01 2.59075522e-01
-7.47548997e-01 5.88352084e-01 -9.11695063e-01 3.40375155e-02
5.55833280e-01 3.59621674e-01 -5.40936410e-01 -4.96928692e-01
-9.12018180e-01 -1.95178881e-01 5.60291111e-01 4.40875292e-01
-5.48575401e-01 -5.43672442e-01 -7.71833718e-01 4.07732755e-01
4.35555011e-01 -3.96580398e-01 4.00020778e-02 -6.84443295e-01
-1.00981653e+00 4.72782254e-01 -8.84652957e-02 1.81848451e-01
3.41427833e-01 7.40307570e-02 -4.33802485e-01 1.64980620e-01
2.36747175e-01 3.34175825e-01 6.22792959e-01 -1.38413429e+00
-1.32960290e-01 -7.36573815e-01 -1.33885399e-01 2.57631540e-01
-6.25325739e-01 -6.24466479e-01 -1.04226995e+00 -7.43759692e-01
7.24286973e-01 -9.11772132e-01 -2.68378496e-01 -5.75144440e-02
-5.74617803e-01 -2.72500783e-01 9.48656797e-01 -5.07981181e-01
1.54817092e+00 -2.51736951e+00 5.58563828e-01 5.23136735e-01
9.06481326e-01 1.81098972e-02 -2.06734329e-01 4.29154575e-01
-1.90008298e-01 8.96891728e-02 -3.13635409e-01 -3.04322720e-01
-9.94367003e-02 3.64399433e-01 7.38772750e-02 6.89839244e-01
-2.79600695e-02 7.30150700e-01 -1.06476235e+00 -5.66817403e-01
1.24823555e-01 6.73695803e-01 -5.15112638e-01 1.58734307e-01
2.21757978e-01 5.42428076e-01 -6.21901214e-01 3.27585995e-01
1.04575861e+00 -6.76609814e-01 5.02262533e-01 -7.06774712e-01
2.25120410e-01 -7.17687160e-02 -1.35800886e+00 1.96671712e+00
-1.61037654e-01 2.68429339e-01 3.00943911e-01 -1.31663215e+00
9.68169808e-01 -2.35886518e-02 6.62669957e-01 -3.68024498e-01
-3.25697362e-02 2.00110264e-02 2.15482846e-01 -2.56878346e-01
-1.12978998e-03 2.39227414e-01 5.54576963e-02 2.50012040e-01
-5.02313487e-02 7.04661787e-01 9.46202576e-02 6.39715254e-01
9.50894833e-01 -2.29030520e-01 -8.71585235e-02 -4.33324605e-01
6.83116972e-01 -4.90838885e-01 9.26670492e-01 2.39559725e-01
3.29462253e-02 4.90087539e-01 6.47049725e-01 -4.02150333e-01
-7.58690298e-01 -1.06476188e+00 -3.29211690e-02 6.70812845e-01
4.76237178e-01 -7.40143061e-01 -6.68669760e-01 -8.82973552e-01
-5.06222919e-02 1.08292781e-01 -6.76147640e-01 -4.44669843e-01
-5.51495969e-01 -8.38564754e-01 2.05711752e-01 4.99695331e-01
5.18544078e-01 -4.72312152e-01 2.57801801e-01 1.74685523e-01
-3.45857561e-01 -9.97541726e-01 -7.38882244e-01 2.35023573e-02
-8.55746984e-01 -1.07032084e+00 -5.06681621e-01 -8.95089090e-01
1.20563090e+00 7.67674863e-01 7.92761266e-01 4.86460447e-01
-1.07876420e-01 -2.98991036e-02 -3.54180664e-01 6.65294826e-01
1.11205518e-01 1.40523687e-01 -6.25382140e-02 3.39082092e-01
5.59184313e-01 -1.01972353e+00 -9.36886907e-01 2.91073382e-01
-9.64414835e-01 1.53583035e-01 6.48280621e-01 1.11194479e+00
6.04724765e-01 5.32297492e-01 3.13505679e-01 -7.84739137e-01
3.92892003e-01 -4.86329466e-01 -5.27307034e-01 2.25185588e-01
-6.39159620e-01 1.27072528e-01 8.65898430e-01 -3.15183967e-01
-6.57144547e-01 1.04109906e-01 2.44466245e-01 -8.22000623e-01
1.30144909e-01 6.46783531e-01 -7.59386063e-01 1.59764230e-01
1.78189687e-02 4.74073738e-01 1.15987577e-01 -5.45106351e-01
4.83251601e-01 3.63595665e-01 3.36357892e-01 -3.72278631e-01
9.74533498e-01 6.24877572e-01 2.98291117e-01 -5.17575502e-01
-8.19404006e-01 -6.83280945e-01 -8.22087944e-01 2.52200644e-02
8.77994716e-01 -1.13112175e+00 -7.27972627e-01 3.39488864e-01
-9.03189242e-01 9.91292447e-02 1.06802374e-01 4.13907826e-01
8.79356414e-02 9.13983762e-01 -7.69722104e-01 -4.42258656e-01
-1.41334459e-01 -1.16911614e+00 1.12175083e+00 1.43353626e-01
3.04297358e-01 -9.98094380e-01 -4.16218489e-02 2.78165042e-01
-8.41905400e-02 3.16197187e-01 1.08688116e+00 -3.62573832e-01
-8.28694880e-01 -2.05007270e-01 -6.31724775e-01 1.92977205e-01
4.44062293e-01 2.14263089e-02 -6.32928908e-01 -6.86707795e-01
-1.50974005e-01 6.93653850e-03 1.05721080e+00 1.08149610e-01
1.50736761e+00 -4.21945721e-01 -5.66159666e-01 8.99744570e-01
1.47016168e+00 -1.17927015e-01 4.37232375e-01 -1.53745279e-01
1.58838904e+00 7.19942093e-01 1.61367401e-01 3.56885105e-01
4.38492715e-01 4.76329952e-01 4.48628545e-01 -2.18803398e-02
1.11340865e-01 -5.17546177e-01 5.19374833e-02 1.51154876e+00
-1.29415737e-02 -6.89849108e-02 -6.17985785e-01 5.03194153e-01
-2.16674256e+00 -7.41559863e-01 -3.92123789e-01 2.06091905e+00
4.39508766e-01 -3.52577977e-02 -1.18520647e-01 6.22539036e-02
1.03271425e+00 4.00525659e-01 -6.44079447e-01 4.29138571e-01
3.26577425e-02 -3.90536487e-01 3.58068109e-01 2.04390168e-01
-9.21624422e-01 7.15556204e-01 4.34208345e+00 1.24820232e+00
-8.65401864e-01 8.56933999e-04 5.57185769e-01 1.37609765e-01
-5.02418756e-01 2.13750675e-01 -5.33996701e-01 6.42294705e-01
3.06814432e-01 -3.78299691e-02 6.23874366e-01 7.94671714e-01
-5.92430308e-02 5.20331264e-01 -8.69126439e-01 1.10504746e+00
-3.74129675e-02 -1.09826553e+00 2.85552233e-01 4.50450718e-01
7.22354650e-01 -1.35746524e-01 3.20716240e-02 2.85829097e-01
2.85433441e-01 -8.86007845e-01 8.29914734e-02 3.54950577e-01
7.31458187e-01 -1.02614105e+00 5.72961986e-01 3.27859640e-01
-1.77442968e+00 -8.03103447e-02 -9.00662005e-01 1.42126322e-01
-1.00127317e-01 8.63760948e-01 -3.31618935e-01 9.58820641e-01
7.07612157e-01 1.27647603e+00 -7.82473385e-01 5.59893131e-01
5.43952398e-02 3.23692977e-01 -1.16886251e-01 2.77334690e-01
5.08177459e-01 -9.41580892e-01 3.89745772e-01 7.59305537e-01
2.74205446e-01 -9.16093811e-02 5.27712464e-01 9.89345431e-01
-2.35226259e-01 1.36957660e-01 -5.56429863e-01 -1.67273968e-01
5.11823714e-01 1.67138755e+00 -9.55121577e-01 -1.72013685e-01
-4.93558645e-01 1.04080439e+00 6.87711418e-01 5.43318093e-01
-6.81574762e-01 -3.86705965e-01 6.14870727e-01 8.10304508e-02
4.48003829e-01 -3.73541474e-01 2.70708855e-02 -1.42332375e+00
1.87847435e-01 -6.83108747e-01 4.00706321e-01 -3.37613612e-01
-1.60212708e+00 5.36158264e-01 -2.18238324e-01 -1.42783010e+00
3.52905393e-01 -3.04474771e-01 -5.91690361e-01 7.28679478e-01
-1.00449586e+00 -1.30690813e+00 -6.90531433e-01 6.18304372e-01
1.57434940e-01 -1.88824177e-01 5.38285077e-01 6.56196535e-01
-1.01098311e+00 6.72526538e-01 4.47759479e-01 4.98011351e-01
6.03172481e-01 -1.20494270e+00 9.78678986e-02 7.48014867e-01
4.21257988e-02 1.07484555e+00 1.13820076e-01 -6.88119352e-01
-1.41939986e+00 -1.48971248e+00 1.20949723e-01 -2.06870865e-02
7.34938383e-01 -7.41727352e-01 -1.40260684e+00 4.95595098e-01
-2.78670099e-02 3.68626118e-01 9.00996149e-01 2.37799466e-01
-6.69937313e-01 -4.46285516e-01 -6.92534626e-01 6.17258728e-01
1.36310804e+00 -7.80423164e-01 -2.72571091e-02 2.69330770e-01
1.03510642e+00 6.90220520e-02 -1.14982474e+00 3.96316975e-01
4.13670391e-01 -1.07620311e+00 8.95908833e-01 -3.32035065e-01
2.98534334e-01 -7.20613897e-01 -5.95133305e-02 -1.27637041e+00
-9.26563382e-01 -3.14292133e-01 -2.76292711e-01 1.61205626e+00
-5.45610003e-02 -5.89043021e-01 8.47173035e-01 3.46942544e-01
-5.26604541e-02 -7.88570940e-01 -8.82947564e-01 -5.74438691e-01
-1.25009149e-01 1.33183271e-01 7.46093214e-01 1.40234327e+00
-2.02006355e-01 7.26423681e-01 -3.86365652e-01 4.20748502e-01
1.01770914e+00 4.38122332e-01 7.43996263e-01 -1.30700231e+00
-3.13166767e-01 -4.44153786e-01 -7.07438469e-01 -1.27383208e+00
2.58414626e-01 -1.10256052e+00 -2.98591971e-01 -1.49963319e+00
5.43300688e-01 -5.48149765e-01 -4.79395002e-01 1.48302317e-01
-5.24168015e-01 3.86930406e-02 -5.05181327e-02 5.46778023e-01
-8.78635764e-01 9.25306380e-01 1.46798086e+00 -2.47539788e-01
-6.79048523e-02 -3.77806664e-01 -7.86481798e-01 6.05785549e-01
4.86435682e-01 -4.34663653e-01 -7.52731025e-01 -4.63651657e-01
2.96632081e-01 -7.28973672e-02 3.73136342e-01 -8.91375661e-01
4.38716769e-01 5.82516603e-02 5.73458195e-01 -7.18052447e-01
5.30014858e-02 -1.06621397e+00 3.48089308e-01 4.17591125e-01
-7.54233673e-02 -1.51446193e-01 -2.67356455e-01 1.14187884e+00
-3.38740617e-01 1.75774023e-01 6.53833389e-01 -4.55627218e-02
-3.82044971e-01 8.08005154e-01 1.10370703e-01 -1.49583235e-01
8.25007677e-01 -1.18900888e-01 -3.08194816e-01 -8.49738941e-02
-5.75885355e-01 5.60377479e-01 6.54331088e-01 3.71139795e-01
6.18121445e-01 -1.77951872e+00 -4.48445380e-01 3.50995064e-01
2.57222861e-01 5.87162375e-01 8.17635477e-01 9.52384710e-01
-2.91133702e-01 8.28407481e-02 1.57812051e-02 -8.40798199e-01
-9.46187913e-01 1.01268435e+00 2.14207128e-01 -5.08521020e-01
-8.96331906e-01 5.64527869e-01 8.30398619e-01 -4.94922131e-01
1.79117098e-01 1.31675944e-01 -3.77494156e-01 1.61260486e-01
2.45712027e-01 3.26323479e-01 -3.18941176e-01 -7.87603021e-01
-4.03555512e-01 7.07598805e-01 -2.92378306e-01 5.53286135e-01
1.33560753e+00 -3.93452138e-01 -5.15139997e-01 3.27166736e-01
1.67141962e+00 1.20315596e-01 -1.38714635e+00 -3.17954540e-01
-1.88061163e-01 -5.74688554e-01 5.07833473e-02 5.40544204e-02
-1.61248124e+00 1.03142655e+00 5.20884275e-01 3.32924247e-01
9.79469001e-01 -6.54251128e-02 6.69233561e-01 2.86888719e-01
8.20843503e-02 -8.63649607e-01 2.75611103e-01 4.66672555e-02
5.69302678e-01 -1.17082882e+00 2.81038672e-01 -6.96317911e-01
-5.01533508e-01 1.00286961e+00 8.47169757e-01 -4.24803346e-01
9.06226456e-01 -2.25622177e-01 -2.34475881e-01 -6.51390433e-01
-3.68122220e-01 -4.43191864e-02 3.72422963e-01 4.88741994e-01
2.96187222e-01 1.32623628e-01 -6.55468479e-02 6.45048380e-01
2.52653450e-01 -6.94005847e-01 6.87759519e-02 2.51783520e-01
-1.55054247e-02 -9.73460853e-01 1.00789271e-01 5.09302318e-01
-2.33104482e-01 4.05523507e-03 -4.58327204e-01 6.46423876e-01
1.12343147e-01 7.19326973e-01 1.55220225e-01 -6.00544393e-01
6.07035309e-02 -4.42878336e-01 5.05436510e-02 -5.79788804e-01
-1.14505775e-01 5.37108123e-01 -4.03391063e-01 -5.22232711e-01
-5.23047924e-01 -4.25936282e-01 -1.25079906e+00 -1.44049987e-01
-4.49845403e-01 3.93357605e-01 1.39586385e-02 7.20953107e-01
6.59357846e-01 7.26290882e-01 9.51097667e-01 -6.21172011e-01
-3.21852386e-01 -8.73444796e-01 -9.49971199e-01 6.71532273e-01
1.29778922e-01 -7.22051620e-01 -5.81977069e-01 -3.18925470e-01] | [7.452425956726074, 5.963438510894775] |
f4c52355-079a-4e33-aecc-7b26eba1fd47 | a-sentinel-2-multi-year-multi-country | 2204.00951 | null | https://arxiv.org/abs/2204.00951v2 | https://arxiv.org/pdf/2204.00951v2.pdf | A Sentinel-2 multi-year, multi-country benchmark dataset for crop classification and segmentation with deep learning | In this work we introduce Sen4AgriNet, a Sentinel-2 based time series multi country benchmark dataset, tailored for agricultural monitoring applications with Machine and Deep Learning. Sen4AgriNet dataset is annotated from farmer declarations collected via the Land Parcel Identification System (LPIS) for harmonizing country wide labels. These declarations have only recently been made available as open data, allowing for the first time the labeling of satellite imagery from ground truth data. We proceed to propose and standardise a new crop type taxonomy across Europe that address Common Agriculture Policy (CAP) needs, based on the Food and Agriculture Organization (FAO) Indicative Crop Classification scheme. Sen4AgriNet is the only multi-country, multi-year dataset that includes all spectral information. It is constructed to cover the period 2016-2020 for Catalonia and France, while it can be extended to include additional countries. Currently, it contains 42.5 million parcels, which makes it significantly larger than other available archives. We extract two sub-datasets to highlight its value for diverse Deep Learning applications; the Object Aggregated Dataset (OAD) and the Patches Assembled Dataset (PAD). OAD capitalizes zonal statistics of each parcel, thus creating a powerful label-to-features instance for classification algorithms. On the other hand, PAD structure generalizes the classification problem to parcel extraction and semantic segmentation and labeling. The PAD and OAD are examined under three different scenarios to showcase and model the effects of spatial and temporal variability across different years and different countries. | ['Ioannis Papoutsis', 'Dimitrios Zografakis', 'Maria Sdraka', 'Dimitrios Sykas'] | 2022-04-02 | null | null | null | null | ['crop-classification'] | ['miscellaneous'] | [ 7.22299963e-02 -1.46485537e-01 -4.03865486e-01 -2.08977208e-01
-4.55600947e-01 -1.14683688e+00 6.93329453e-01 9.33113813e-01
-3.03206146e-01 8.37611675e-01 -2.70063188e-02 -5.40047407e-01
-2.55874872e-01 -1.50528371e+00 -7.97551334e-01 -8.70224476e-01
-5.34741640e-01 1.90959826e-01 -3.72034907e-01 -4.91826892e-01
-4.56486017e-01 8.29987168e-01 -1.73092675e+00 3.61210525e-01
9.33521807e-01 1.11719680e+00 5.74537754e-01 3.41881275e-01
5.24024963e-02 1.65720209e-01 -4.21645343e-01 3.56285833e-02
5.77652335e-01 -1.02147512e-01 -6.20578647e-01 -1.58683121e-01
3.88173789e-01 -2.25061938e-01 3.00383568e-01 1.04427540e+00
3.77932608e-01 -2.85703421e-01 6.41109467e-01 -1.03152704e+00
-5.25092304e-01 6.08271003e-01 -7.65779257e-01 -1.97344825e-01
-1.67952225e-01 -2.33005404e-01 9.25080538e-01 -3.39213997e-01
6.20913982e-01 1.00587285e+00 9.08250749e-01 -2.76316881e-01
-1.17790020e+00 -5.95377028e-01 2.66777664e-01 -1.12551577e-01
-1.37226546e+00 -1.68648008e-02 2.08696574e-01 -8.29477966e-01
8.98846090e-01 5.30669451e-01 8.43294561e-01 6.12780690e-01
2.39470437e-01 5.48979282e-01 1.43091059e+00 -3.65272909e-01
1.95157796e-01 -2.48442948e-01 8.34619403e-02 2.49695584e-01
5.14530778e-01 3.18689436e-01 9.33138430e-02 -1.24709278e-01
7.72996962e-01 -1.20889302e-03 -1.18166029e-01 -5.43578982e-01
-1.18560135e+00 9.12713408e-01 8.53979230e-01 3.57336700e-01
-9.85046387e-01 -3.55198234e-01 6.29445195e-01 5.43639243e-01
9.01706576e-01 2.91784167e-01 -1.14015758e+00 5.70166051e-01
-1.08491278e+00 5.93935847e-01 5.94970286e-01 7.52963245e-01
8.95883620e-01 7.85597414e-02 1.91831544e-01 9.82601702e-01
8.48293826e-02 1.08163905e+00 -1.67046085e-01 -7.36169398e-01
2.73535192e-01 6.69541061e-01 4.16900009e-01 -1.02485800e+00
-8.64416063e-01 -5.41528881e-01 -1.12438560e+00 2.64613509e-01
5.39392352e-01 -4.20117319e-01 -1.08526945e+00 1.64380932e+00
2.37190485e-01 -3.88547897e-01 4.04007435e-01 8.52451503e-01
9.45178747e-01 8.97029340e-01 3.01200390e-01 -2.15572000e-01
1.57865119e+00 -3.34533453e-01 -6.18703008e-01 2.72547305e-02
7.49706805e-01 -3.81305665e-01 5.61745107e-01 3.82337838e-01
-3.00460190e-01 -4.84113306e-01 -9.61805701e-01 6.26717627e-01
-1.26218152e+00 3.73841882e-01 1.01294231e+00 4.38943952e-01
-8.89303803e-01 4.43684340e-01 -6.26039267e-01 -6.79754794e-01
5.22389770e-01 -6.15097322e-02 -7.55816519e-01 1.51387483e-01
-1.43509829e+00 8.39932024e-01 8.32021117e-01 4.10893738e-01
-8.01654220e-01 -8.14698219e-01 -1.09310699e+00 7.97339454e-02
1.27737507e-01 4.81537879e-02 6.03760719e-01 -1.11261857e+00
-8.37202251e-01 1.24324274e+00 3.36564422e-01 -7.51060903e-01
2.10778832e-01 -1.90794289e-01 -6.46002114e-01 -1.92253649e-01
5.84353089e-01 7.61691034e-01 2.68126309e-01 -1.21455646e+00
-1.00731766e+00 -5.86067796e-01 7.47985020e-02 -1.74694136e-01
-6.86743706e-02 1.18759155e-01 2.34040618e-01 -1.07017589e+00
9.04591531e-02 -9.01647866e-01 -3.30493718e-01 -4.25385386e-01
-1.64932445e-01 1.10439725e-01 5.55516601e-01 -1.04408038e+00
9.06758606e-01 -1.96525347e+00 -8.92800912e-02 4.91585396e-02
-2.48935401e-01 3.03143680e-01 -5.41901290e-01 6.43060684e-01
-4.20025140e-01 1.73305273e-01 -7.69450963e-01 5.56899011e-01
-1.04478337e-01 3.41465503e-01 -4.67870921e-01 5.45309424e-01
5.57568669e-01 7.15032399e-01 -8.65753949e-01 4.16711019e-03
2.97156215e-01 1.57302275e-01 -8.04873630e-02 -8.75854269e-02
-5.13319075e-01 4.99502093e-01 -3.15735400e-01 1.10805428e+00
1.49261558e+00 3.75665963e-01 3.65954190e-01 -3.73441994e-01
-7.16900647e-01 -3.37634951e-01 -1.00794995e+00 1.43115211e+00
-4.14839149e-01 5.74921846e-01 3.65314156e-01 -1.26409829e+00
1.20023167e+00 2.68915862e-01 8.29718649e-01 -7.85571456e-01
-1.00591607e-01 5.56472242e-01 -2.34536603e-01 -3.67345899e-01
5.55726707e-01 2.16095552e-01 -4.44848955e-01 -9.11011323e-02
1.96204320e-01 -1.23788290e-01 3.68715912e-01 -4.52644795e-01
3.61276865e-01 6.25165522e-01 5.45760393e-01 -8.95867527e-01
3.02705884e-01 6.24717593e-01 8.22112620e-01 5.70878148e-01
-1.39607772e-01 4.77662891e-01 6.90685570e-01 -9.03700113e-01
-1.03470707e+00 -9.22187746e-01 -8.73254359e-01 1.13216710e+00
-2.59673208e-01 1.69554785e-01 -5.80141068e-01 -5.49897194e-01
5.29948413e-01 5.67118526e-01 -7.69586325e-01 7.71731555e-01
-2.70985007e-01 -1.65789604e+00 7.10657001e-01 2.59029716e-01
8.13762784e-01 -1.27753603e+00 -7.82446802e-01 3.45599413e-01
-1.56888753e-01 -1.08809936e+00 4.69887912e-01 6.53776407e-01
-5.02756715e-01 -1.22538614e+00 -9.09736216e-01 -5.71984053e-01
1.73331171e-01 2.36705616e-02 1.21544945e+00 -7.67273724e-01
-1.49314597e-01 5.56663089e-02 -6.76824212e-01 -8.53980958e-01
-4.11600143e-01 2.95472771e-01 -1.71582073e-01 -8.96912664e-02
5.75122952e-01 -4.98674780e-01 -2.55640268e-01 2.11640015e-01
-8.86899889e-01 1.37483373e-01 5.25654018e-01 6.24304593e-01
8.09446990e-01 1.16276659e-01 9.50086951e-01 -6.55796885e-01
-1.59399897e-01 -9.18180645e-01 -1.14620161e+00 3.36956918e-01
-2.58599699e-01 -4.87373590e-01 3.16455573e-01 1.71479464e-01
-9.29662049e-01 4.60420400e-01 7.76341185e-02 2.78444082e-01
-7.79950976e-01 1.26760364e+00 -3.57003361e-01 1.48952797e-01
5.01403034e-01 9.38414335e-02 -2.70467192e-01 -7.19546735e-01
3.81559700e-01 8.62098634e-01 8.98462296e-01 -3.73469591e-01
4.84590650e-01 3.76205057e-01 1.45105114e-02 -1.19368517e+00
-7.68409669e-01 -5.21163821e-01 -1.17420363e+00 -6.43989444e-02
9.13542747e-01 -1.49167466e+00 -2.33666852e-01 9.74767804e-01
-1.14551294e+00 -5.76593220e-01 -1.90064088e-01 4.48900700e-01
-4.52586174e-01 -1.49411902e-01 -1.10884175e-01 -6.26636922e-01
-3.98325980e-01 -6.74006343e-01 1.03471768e+00 -6.36630580e-02
1.21831514e-01 -9.08568442e-01 1.24182887e-01 -2.32168093e-01
4.29095030e-01 1.42902851e+00 7.82188535e-01 -3.56471211e-01
5.35834916e-02 -1.57925606e-01 -5.98381102e-01 4.83181030e-01
5.69297373e-01 1.06553882e-02 -1.08190596e+00 -4.85903412e-01
-3.19126457e-01 -3.73052470e-02 1.30326867e+00 8.03074062e-01
7.80683458e-01 -1.36117950e-01 -1.51479498e-01 8.20964277e-01
1.84232175e+00 2.56104410e-01 4.19879049e-01 7.24247754e-01
4.79346931e-01 1.04131341e+00 1.01416039e+00 4.00698185e-01
2.66593099e-01 5.00731528e-01 1.01053464e+00 -6.52453780e-01
2.35138595e-01 2.15736434e-01 2.50526041e-01 1.95973992e-01
-2.27298230e-01 -2.54460156e-01 -1.32996762e+00 1.04806888e+00
-1.79970670e+00 -8.62296999e-01 -4.67639267e-01 2.22954702e+00
6.94127262e-01 -4.38120127e-01 3.36262509e-02 4.56150956e-02
5.57946563e-01 5.66306233e-01 -3.59086543e-01 -3.04315299e-01
-9.91795719e-01 4.57897753e-01 1.43242109e+00 2.49848381e-01
-1.89146841e+00 1.20244384e+00 5.49157429e+00 6.20474875e-01
-1.22150183e+00 1.72396585e-01 4.53828990e-01 3.54394525e-01
6.76419064e-02 -1.57131597e-01 -6.97109461e-01 3.25628705e-02
8.99367511e-01 1.19357472e-02 6.65260702e-02 5.32657325e-01
3.75713676e-01 -2.97766894e-01 -4.91817147e-01 3.87937576e-01
-5.78965366e-01 -1.35871708e+00 3.39012109e-02 1.40445441e-01
9.17392790e-01 7.14992583e-01 -2.21415043e-01 1.57503083e-01
4.45580453e-01 -9.35007870e-01 7.69513905e-01 3.01934600e-01
1.09223223e+00 -7.11084306e-01 1.08047211e+00 1.09582424e-01
-1.56359136e+00 -2.99560487e-01 -4.51345354e-01 -4.12145890e-02
-1.45977408e-01 6.70031428e-01 2.48264819e-02 1.28850079e+00
1.18462682e+00 1.08476770e+00 -4.41665888e-01 8.20104420e-01
1.20061912e-01 6.10346675e-01 -4.92205203e-01 6.06738627e-01
6.99679255e-01 -4.46360201e-01 5.04893005e-01 1.39896858e+00
6.50014281e-01 -6.61339834e-02 2.42200911e-01 5.97590864e-01
2.49194011e-01 3.64430726e-01 -7.87722170e-01 -1.26774520e-01
5.36597550e-01 1.31960034e+00 -7.01691270e-01 6.84063360e-02
-3.49518806e-01 6.56640708e-01 -2.18601003e-01 4.64127600e-01
-4.99535620e-01 -4.06847775e-01 9.66521561e-01 -6.00587167e-02
2.67594248e-01 -2.45077819e-01 3.92764099e-02 -8.56254041e-01
-1.84637278e-01 -7.94990122e-01 4.27143753e-01 -6.49972796e-01
-1.06584096e+00 3.90338063e-01 3.69572550e-01 -1.27855420e+00
-3.93315926e-02 -8.61741543e-01 -1.59848690e-01 1.27263916e+00
-2.04203606e+00 -1.78419673e+00 -6.53993666e-01 9.16933715e-02
2.17928365e-01 -4.54409212e-01 1.42168057e+00 1.68808430e-01
-4.96270746e-01 -1.79014052e-03 6.30753756e-01 2.44069993e-01
4.58507270e-01 -1.32356405e+00 6.54100776e-01 6.13487184e-01
-3.07508290e-01 3.61845153e-03 5.16664207e-01 -5.60920835e-01
-8.93621564e-01 -1.93261874e+00 7.51295030e-01 -5.04458919e-02
7.69007206e-01 -2.77918637e-01 -8.16699088e-01 6.75006270e-01
1.65354311e-01 6.99754804e-02 4.34849381e-01 -1.04999214e-01
-3.22421461e-01 -6.09045923e-01 -1.20796156e+00 -6.32008538e-02
4.66321170e-01 -2.26820439e-01 3.58298258e-03 5.92955291e-01
6.17658615e-01 -2.65945971e-01 -1.37616837e+00 7.91106701e-01
7.51675069e-01 -8.97297144e-01 7.08017409e-01 -3.19815993e-01
3.95165354e-01 -3.08121353e-01 -7.25309610e-01 -1.68364489e+00
-4.32017922e-01 -2.05592990e-01 4.91173327e-01 1.32753110e+00
2.62974024e-01 -9.26167607e-01 3.86728108e-01 -4.96134520e-01
-1.56732380e-01 -2.22366184e-01 -7.75833011e-01 -1.03573895e+00
7.02075362e-01 -3.47751319e-01 1.31605053e+00 1.38190901e+00
-6.85029984e-01 -4.40233409e-01 -8.71975441e-03 6.42604589e-01
4.02259916e-01 4.96245176e-01 6.23496592e-01 -1.74536383e+00
4.35846031e-01 -5.09639800e-01 -1.64607376e-01 -2.78453708e-01
1.51292488e-01 -8.45714569e-01 -2.86145836e-01 -1.35681808e+00
-1.89296797e-01 -8.86806250e-01 -3.35446894e-01 1.09228730e+00
8.69713798e-02 1.91679165e-01 6.78276643e-02 1.10048413e-01
3.75711530e-01 2.18215257e-01 5.99932671e-01 -4.17481452e-01
-2.77742565e-01 -5.43504097e-02 -4.18365687e-01 5.89366078e-01
1.00844729e+00 -3.65558952e-01 2.49214873e-01 -5.16405046e-01
3.42662483e-01 -1.79155365e-01 6.45649433e-01 -9.46906686e-01
-6.75307751e-01 -4.73481596e-01 2.91336060e-01 -1.21512389e+00
-3.48005056e-01 -9.51557934e-01 5.45927405e-01 4.52038229e-01
4.69211899e-02 -2.00106069e-01 8.94883871e-01 2.39607647e-01
-5.05562067e-01 1.91365361e-01 7.46650517e-01 -1.39063209e-01
-1.03646243e+00 4.56974000e-01 -5.61974466e-01 -4.80590582e-01
9.06699479e-01 -4.38146479e-02 -3.18230599e-01 1.16797917e-01
-6.81513667e-01 6.00072801e-01 5.20315647e-01 6.52394056e-01
-2.26652786e-01 -1.35091269e+00 -1.44152594e+00 4.24847931e-01
5.94258010e-01 4.32945751e-02 2.12125525e-01 4.57062155e-01
-1.07058275e+00 8.13062251e-01 -6.90486789e-01 -6.47241771e-01
-9.48035479e-01 4.37340498e-01 5.37453413e-01 -3.71915549e-01
-3.08513999e-01 2.81736195e-01 2.28128374e-01 -1.05942905e+00
-2.52909124e-01 -5.40779352e-01 -8.50789070e-01 9.62424099e-01
2.65698075e-01 1.18895490e-02 2.18232438e-01 -9.53652322e-01
-2.46342853e-01 4.88907337e-01 5.63393176e-01 3.26958686e-01
1.85858095e+00 3.09091136e-02 -6.02016628e-01 4.60233361e-01
8.63013327e-01 -1.83370218e-01 -1.09944880e+00 -1.96226135e-01
3.02449882e-01 4.51000640e-04 2.80181915e-01 -1.19390559e+00
-1.25454426e+00 6.26442492e-01 1.09928167e+00 6.14179254e-01
1.41970038e+00 -4.75104034e-01 3.79283309e-01 2.75238663e-01
5.19561946e-01 -7.68630981e-01 -1.07420909e+00 5.84714174e-01
1.25970316e+00 -1.41922832e+00 -3.31880264e-02 -3.24618429e-01
-2.14704618e-01 1.10211313e+00 1.63692892e-01 1.20497659e-01
7.75061250e-01 2.44346276e-01 1.91995397e-01 -1.59667581e-01
-1.71414897e-01 -7.28590429e-01 -1.59234460e-02 1.09824872e+00
5.08545041e-01 8.65241766e-01 -2.49100894e-01 7.48058081e-01
-1.17879868e-01 4.28033173e-02 2.31773615e-01 7.04787374e-01
-4.63649571e-01 -9.54774201e-01 -6.84632659e-01 5.67551315e-01
-3.70707780e-01 -1.92815289e-01 -1.98043242e-01 1.02364826e+00
6.01835012e-01 7.56289780e-01 3.45991910e-01 1.11802749e-01
3.81727606e-01 -3.65254402e-01 2.07422584e-01 -4.70460773e-01
-6.20715499e-01 -7.20096612e-03 1.09144077e-01 -3.49736065e-01
-8.04517388e-01 -9.35204685e-01 -7.49046683e-01 -4.01787519e-01
-2.45051771e-01 3.37070003e-02 8.03760350e-01 5.90736628e-01
1.56941101e-01 3.86931002e-01 7.07841814e-01 -1.26360989e+00
-7.54564106e-02 -1.15879714e+00 -1.10893357e+00 -2.46585254e-02
5.14651954e-01 -5.96089959e-01 5.99487796e-02 -3.57375890e-02] | [9.417177200317383, -1.6168004274368286] |
75cd0bf1-b088-444e-b6d3-577c08020e72 | ergodic-limits-relaxations-and-geometric | 2109.04526 | null | https://arxiv.org/abs/2109.04526v1 | https://arxiv.org/pdf/2109.04526v1.pdf | Ergodic Limits, Relaxations, and Geometric Properties of Random Walk Node Embeddings | Random walk based node embedding algorithms learn vector representations of nodes by optimizing an objective function of node embedding vectors and skip-bigram statistics computed from random walks on the network. They have been applied to many supervised learning problems such as link prediction and node classification and have demonstrated state-of-the-art performance. Yet, their properties remain poorly understood. This paper studies properties of random walk based node embeddings in the unsupervised setting of discovering hidden block structure in the network, i.e., learning node representations whose cluster structure in Euclidean space reflects their adjacency structure within the network. We characterize the ergodic limits of the embedding objective, its generalization, and related convex relaxations to derive corresponding non-randomized versions of the node embedding objectives. We also characterize the optimal node embedding Grammians of the non-randomized objectives for the expected graph of a two-community Stochastic Block Model (SBM). We prove that the solution Grammian has rank $1$ for a suitable nuclear norm relaxation of the non-randomized objective. Comprehensive experimental results on SBM random networks reveal that our non-randomized ergodic objectives yield node embeddings whose distribution is Gaussian-like, centered at the node embeddings of the expected network within each community, and concentrate in the linear degree-scaling regime as the number of nodes increases. | ['Prakash Ishwar', 'Daniel Sussman', 'Christy Lin'] | 2021-09-09 | null | null | null | null | ['stochastic-block-model'] | ['graphs'] | [ 2.36502420e-02 7.07225025e-01 -5.03396332e-01 -1.10940330e-01
-2.96036184e-01 -4.95426238e-01 5.24591863e-01 1.62119791e-01
-2.59893328e-01 2.38092020e-01 2.27962762e-01 -4.69836712e-01
-8.05027664e-01 -9.56589699e-01 -6.06202543e-01 -1.17741597e+00
-8.76076698e-01 8.79509687e-01 2.16829926e-02 -5.21409772e-02
-1.87189460e-01 3.40672016e-01 -8.82921159e-01 -4.74027008e-01
6.46001935e-01 3.46331716e-01 2.08573714e-02 9.57153320e-01
3.61132734e-02 4.32355076e-01 -5.98563477e-02 -3.84118438e-01
1.63161024e-01 -1.04328774e-01 -8.42868209e-01 5.73062189e-02
-6.71062842e-02 1.22078545e-01 -1.19217384e+00 1.23596573e+00
2.22367242e-01 7.04046711e-02 1.06615341e+00 -1.90663612e+00
-8.51224661e-01 1.04114056e+00 -5.85304976e-01 2.15022191e-01
-8.81060660e-02 -3.45648974e-01 1.70033371e+00 -6.19007170e-01
7.89411008e-01 1.19193101e+00 7.60510564e-01 4.04858559e-01
-1.74503243e+00 -3.15054446e-01 -2.06404716e-01 7.88098574e-02
-1.69493151e+00 -1.00172702e-02 8.32933843e-01 -6.04142785e-01
6.92799628e-01 3.06267202e-01 5.83501816e-01 9.08142805e-01
1.65760890e-01 4.16949332e-01 5.82454503e-01 -2.42741957e-01
9.17797238e-02 1.69213116e-01 3.26493233e-01 1.07526195e+00
8.01095724e-01 -1.42555967e-01 -2.47827590e-01 -4.37823206e-01
4.27878231e-01 2.58397877e-01 -3.61872256e-01 -1.05373204e+00
-1.29003429e+00 1.24261987e+00 7.55728304e-01 5.42636573e-01
-4.10831809e-01 5.97538412e-01 2.54418790e-01 3.89524668e-01
4.42347586e-01 -1.55189680e-02 -2.11297855e-01 1.33886635e-01
-8.01952720e-01 -9.76932496e-02 9.98735845e-01 9.75659192e-01
9.11096931e-01 -3.34652290e-02 1.81776926e-01 3.37374926e-01
6.76500082e-01 7.57015288e-01 1.83965247e-02 -7.08589315e-01
4.71603632e-01 5.51294982e-01 -1.87877864e-01 -1.47465694e+00
-5.27234018e-01 -6.31601930e-01 -1.33624291e+00 -4.38935488e-01
2.75182158e-01 1.73282444e-01 -3.32393825e-01 1.89603066e+00
2.10573435e-01 1.92579195e-01 -1.12451158e-01 5.11646330e-01
5.22964358e-01 8.60656381e-01 -1.40781194e-01 -2.87088543e-01
8.35342526e-01 -6.30417705e-01 -6.20084465e-01 1.72472090e-01
9.47058201e-01 -1.07068270e-01 6.45929396e-01 -4.69838142e-01
-7.25017488e-01 1.44036515e-02 -9.71336544e-01 2.74465889e-01
-7.92167187e-02 -2.46275395e-01 3.05599183e-01 5.45518100e-01
-1.31992328e+00 7.50340641e-01 -8.72712016e-01 -4.79156792e-01
2.34344780e-01 4.17154282e-01 -5.99221110e-01 -3.42325509e-01
-1.05029881e+00 3.64260644e-01 2.51361340e-01 2.79939383e-01
-8.12588394e-01 -6.15391612e-01 -1.03297973e+00 3.43460202e-01
2.59047747e-01 -4.35797751e-01 3.39933723e-01 -2.82982349e-01
-6.12415135e-01 7.89329469e-01 -1.52591184e-01 -2.73571968e-01
1.15595713e-01 4.77760613e-01 -3.68900180e-01 3.30399930e-01
3.01276445e-01 2.63021290e-01 7.75812447e-01 -9.16067302e-01
1.53151289e-01 -2.60695219e-01 4.69010547e-02 -2.50893205e-01
-8.88765097e-01 -4.62293595e-01 -3.32277752e-02 -3.62408370e-01
4.28334147e-01 -1.23723638e+00 -4.02146339e-01 2.21271992e-01
-5.31340480e-01 -4.07244116e-01 6.80611134e-01 -3.23359549e-01
1.48145628e+00 -2.29142785e+00 5.96073568e-01 7.62529373e-01
8.04484844e-01 -3.34141076e-01 -5.37455976e-01 9.35575962e-01
-2.53836393e-01 5.74605346e-01 -2.03456298e-01 -3.18762995e-02
2.36020193e-01 3.08228582e-01 5.85681945e-02 1.07416487e+00
-3.66876796e-02 9.65411007e-01 -1.06898522e+00 -5.60106874e-01
-1.16039246e-01 3.96999091e-01 -6.65544569e-01 -1.43492483e-02
1.76576465e-01 -2.75393009e-01 -3.93492252e-01 2.02005982e-01
5.09169579e-01 -8.41859162e-01 6.05392814e-01 -1.97801888e-01
5.24731815e-01 -9.64598879e-02 -1.22137010e+00 1.10865641e+00
-2.42817253e-01 9.80382085e-01 4.39927369e-01 -1.38818574e+00
6.82984293e-01 3.50548476e-01 8.23554218e-01 4.02747452e-01
-1.62534654e-01 5.19900694e-02 2.55759388e-01 -4.21163231e-01
1.57408565e-01 -8.56334418e-02 -2.28926893e-02 7.23465800e-01
2.59334911e-02 4.78826582e-01 2.33110607e-01 1.01934505e+00
1.59337366e+00 -7.94151008e-01 2.36550242e-01 -7.75707781e-01
2.98439473e-01 -3.97425026e-01 2.83399314e-01 5.91670871e-01
-1.33429691e-01 1.89483270e-01 9.86487269e-01 -8.67917463e-02
-1.08523822e+00 -1.25990796e+00 -1.68543551e-02 9.44594026e-01
1.66630551e-01 -7.27194309e-01 -5.71556687e-01 -5.22267997e-01
3.03280741e-01 1.58941448e-01 -8.54303241e-01 -4.40301627e-01
-3.89830500e-01 -9.27721024e-01 3.21190685e-01 2.59044945e-01
4.60642874e-02 -5.72635114e-01 1.11227170e-01 2.36716181e-01
-1.77141920e-01 -1.15722990e+00 -8.53216469e-01 7.98797905e-02
-9.03135777e-01 -1.32181764e+00 -5.04035234e-01 -1.03719521e+00
9.55668867e-01 5.47047496e-01 9.20890570e-01 1.83046952e-01
-4.22249883e-01 7.08338201e-01 -1.50924534e-01 5.90331793e-01
-3.15147936e-01 4.55339104e-01 4.25642103e-01 2.04282671e-01
1.45600885e-01 -8.18723619e-01 -6.19341910e-01 4.88652349e-01
-8.96382153e-01 -3.67525905e-01 5.18199325e-01 9.14867938e-01
3.26625943e-01 3.51774871e-01 3.22358221e-01 -6.64208829e-01
6.11399293e-01 -9.72980261e-01 -4.63403195e-01 2.31606260e-01
-7.70104349e-01 6.04673505e-01 2.96716779e-01 -5.52490532e-01
2.41370425e-02 -2.52850860e-01 4.30063248e-01 -2.87978888e-01
7.13518441e-01 7.47216463e-01 -1.77934483e-01 1.97364273e-03
3.67263079e-01 1.80528179e-01 1.01987399e-01 -1.58048898e-01
5.39531648e-01 6.10111177e-01 3.83448005e-02 -3.98908526e-01
1.47716248e+00 5.61314166e-01 5.31428397e-01 -1.29695857e+00
-3.91071767e-01 -7.04405844e-01 -6.28851414e-01 -1.26972750e-01
7.24629879e-01 -5.83258748e-01 -9.09071803e-01 -1.80954471e-01
-1.00556266e+00 -1.28054187e-01 -2.87469476e-01 4.75544244e-01
-4.20712173e-01 5.47504365e-01 -7.42032409e-01 -8.93825412e-01
-1.74367905e-01 -8.13847244e-01 6.01798117e-01 -2.30667338e-01
-1.31750807e-01 -1.52958584e+00 3.92681152e-01 -6.19287938e-02
3.15094322e-01 1.39571652e-01 1.38500690e+00 -7.22747326e-01
-6.20018423e-01 -5.23273766e-01 -3.36509585e-01 1.47681602e-03
-2.29268312e-01 8.16215500e-02 -2.53197193e-01 -5.70664287e-01
-5.15123725e-01 3.20958823e-01 8.13620687e-01 4.77086753e-01
6.83741629e-01 -6.70288742e-01 -7.01242626e-01 5.73809266e-01
1.44406176e+00 -6.97338164e-01 1.87720060e-01 -5.00233173e-02
7.25350082e-01 5.80825508e-01 8.71906877e-02 3.66969764e-01
2.54111141e-01 4.50938582e-01 7.37900674e-01 5.04134119e-01
3.18715364e-01 -4.92665172e-01 4.17377353e-01 1.26092315e+00
8.90031978e-02 -2.99814314e-01 -8.75256002e-01 9.20472264e-01
-1.82052720e+00 -1.12152338e+00 -3.65244538e-01 1.99352229e+00
4.89671856e-01 1.40489385e-01 2.77913719e-01 2.27799729e-01
9.99356568e-01 6.13491476e-01 -3.62336904e-01 -1.53892964e-01
-1.01429455e-01 -4.94628139e-02 8.22965622e-01 7.80777097e-01
-9.66329753e-01 3.99043202e-01 6.03817892e+00 6.86460435e-01
-4.53352779e-01 3.93452883e-01 1.47161663e-01 5.71122169e-02
-8.42371523e-01 4.43528779e-02 -6.04715765e-01 3.30786616e-01
1.16646373e+00 -4.28930730e-01 4.83916610e-01 8.93693924e-01
1.95325986e-01 4.98373240e-01 -1.12589371e+00 8.66153061e-01
-1.84908777e-01 -1.42184687e+00 -1.90026149e-01 8.92107546e-01
9.29050863e-01 2.97038794e-01 6.19883388e-02 1.84281878e-02
4.49294776e-01 -1.16772413e+00 7.99168199e-02 2.59380758e-01
8.74834597e-01 -5.99565506e-01 7.33169019e-01 2.71611780e-01
-1.59450495e+00 -1.25752449e-01 -4.69148368e-01 2.08802208e-01
1.17337376e-01 7.56319046e-01 -6.86968148e-01 3.00655127e-01
3.30641717e-01 9.86443162e-01 -3.70081127e-01 7.50331402e-01
1.49906605e-01 7.69470036e-01 -4.84665394e-01 -4.21808302e-01
3.65706563e-01 -5.95366657e-01 8.78051698e-01 7.90799797e-01
1.83071405e-01 -6.05376996e-02 6.12303950e-02 7.41278172e-01
-4.51604694e-01 7.49761313e-02 -9.53401744e-01 -6.50022984e-01
7.66392529e-01 1.33266985e+00 -7.85256922e-01 2.13763610e-01
-1.87031090e-01 7.35056400e-01 4.87254143e-01 5.35802782e-01
-5.72464108e-01 -5.31866074e-01 9.37919378e-01 5.52863419e-01
5.80703020e-01 -5.21107733e-01 2.01737866e-01 -1.02553499e+00
1.85821936e-01 -2.14217469e-01 2.47691959e-01 -2.25461662e-01
-1.35288835e+00 4.26934451e-01 -2.60910630e-04 -9.11312401e-01
-1.50598809e-01 -5.90192795e-01 -7.12759018e-01 4.34464097e-01
-9.15736735e-01 -7.87211359e-01 7.05899894e-02 4.04868782e-01
-1.45723283e-01 -7.08364621e-02 8.02116632e-01 1.73560947e-01
-6.26999855e-01 6.97391272e-01 9.79221761e-01 4.71784443e-01
-4.40579876e-02 -1.32831013e+00 1.53842986e-01 6.01674139e-01
4.82476115e-01 7.13653326e-01 7.75652468e-01 -5.91931939e-01
-1.85308516e+00 -1.26314700e+00 1.00969100e+00 -1.53011009e-01
1.36923969e+00 -5.15783608e-01 -6.37061954e-01 8.01155984e-01
-3.18142891e-01 5.94238758e-01 7.46618211e-01 2.18964249e-01
-4.08306479e-01 -8.62268880e-02 -1.04201233e+00 4.96040285e-01
1.20332325e+00 -7.86246896e-01 8.54518414e-02 6.61142886e-01
8.24819505e-01 4.87265438e-01 -1.18996227e+00 6.54109120e-02
5.38824916e-01 -3.09849858e-01 1.09678316e+00 -9.39958274e-01
3.49749476e-01 4.90137143e-03 -4.60504264e-01 -1.15648150e+00
-5.51157057e-01 -8.92008245e-01 -4.36834484e-01 9.93138015e-01
4.15552914e-01 -6.70051098e-01 1.08393633e+00 8.10146779e-02
6.20090246e-01 -1.06909728e+00 -1.22662187e+00 -9.15096581e-01
-6.60825893e-02 -3.78492028e-02 1.19222522e-01 8.43398452e-01
1.28645450e-01 4.89875883e-01 -2.91385710e-01 2.86921501e-01
1.19562328e+00 -3.21796164e-02 4.51630801e-01 -1.54266679e+00
-3.06515217e-01 -4.32295650e-01 -1.00278783e+00 -1.07912874e+00
4.88407880e-01 -1.46679282e+00 -1.83648080e-01 -1.54850769e+00
6.26461387e-01 -5.91545522e-01 -1.12790398e-01 -1.57615975e-01
9.73932743e-02 9.51926857e-02 4.25873175e-02 4.02843267e-01
-7.52447248e-01 7.14344680e-01 6.85869992e-01 -3.99027377e-01
2.51751598e-02 1.60989881e-01 -5.19305408e-01 5.40412724e-01
5.85764706e-01 -8.82326365e-01 -3.58670384e-01 -6.88094944e-02
5.06610513e-01 4.30839695e-03 3.18517178e-01 -5.41425288e-01
1.80036396e-01 5.42468876e-02 -3.55645895e-01 -6.11250281e-01
2.63399392e-01 -1.03661871e+00 2.02844143e-01 8.22351098e-01
-6.10196412e-01 1.45522922e-01 -7.36724854e-01 1.37345111e+00
2.22689331e-01 -4.49252039e-01 5.16714096e-01 4.02876407e-01
-4.80581820e-02 7.42034316e-01 -3.75869334e-01 3.74315947e-01
1.10795152e+00 5.19717636e-04 -1.28185481e-01 -7.28016913e-01
-9.31307435e-01 3.23750645e-01 3.28523666e-01 -1.51691558e-02
6.17899299e-01 -1.65535712e+00 -9.18466628e-01 -1.37410685e-01
1.59194723e-01 -3.84308249e-01 -3.70055437e-02 1.04512942e+00
-4.41041946e-01 4.84965026e-01 4.89581436e-01 -7.06086934e-01
-1.34115160e+00 5.34628391e-01 1.64274216e-01 -5.97899914e-01
-5.26164114e-01 6.51540577e-01 -6.24558516e-02 -5.93347549e-01
2.17391878e-01 1.00112505e-01 8.08550045e-02 6.94959909e-02
9.03576836e-02 6.36913955e-01 -4.73169923e-01 -7.52512813e-01
-4.23905998e-01 4.33083534e-01 6.99958950e-02 -2.67634634e-02
1.55341792e+00 -8.73435661e-02 -5.46151102e-01 4.72334474e-01
2.07624221e+00 -1.53874278e-01 -6.33212507e-01 -6.63205802e-01
1.85320228e-01 -1.85440570e-01 -1.45249352e-01 3.66305262e-01
-1.13788939e+00 8.29147518e-01 2.74135441e-01 6.50896907e-01
3.64940971e-01 6.76253676e-01 4.71398443e-01 6.33486569e-01
1.89534962e-01 -6.15649879e-01 6.40108436e-02 3.05104524e-01
4.48829740e-01 -9.58529294e-01 -1.05856573e-02 -4.98983771e-01
-8.80033523e-03 1.07260954e+00 -6.57348428e-03 -3.05322558e-01
1.41982186e+00 -1.17902152e-01 -7.58624256e-01 -3.63887548e-01
-6.72079802e-01 -1.07417472e-01 1.48370832e-01 6.56466067e-01
2.10371375e-01 3.93317342e-01 -1.21411644e-01 2.20024273e-01
-7.98397213e-02 -7.89009094e-01 5.89860678e-01 4.48346257e-01
-5.35352588e-01 -8.41018498e-01 6.51999637e-02 8.03532064e-01
-1.24095991e-01 -2.78394908e-01 -1.52486274e-02 7.43391633e-01
-5.99195063e-01 7.79478788e-01 1.76730251e-03 -5.62859535e-01
-2.61082023e-01 -9.47768614e-02 2.31808364e-01 -5.88120162e-01
2.38644153e-01 -3.76059085e-01 -1.66789338e-01 -1.20832212e-01
-3.40570658e-01 -7.41891623e-01 -8.84652734e-01 -7.98663318e-01
-4.95923072e-01 6.02594137e-01 5.43785930e-01 7.05464303e-01
1.48947537e-01 2.50276834e-01 1.03647399e+00 -5.53562462e-01
-9.51182067e-01 -9.68886614e-01 -9.26587522e-01 2.56474018e-01
3.32581341e-01 -4.46862876e-01 -8.98648381e-01 -4.84792441e-01] | [7.080700397491455, 5.864130020141602] |
997f4e71-010b-43f2-989f-1d82baf16b6e | which-spurious-correlations-impact-reasoning | 2306.12146 | null | https://arxiv.org/abs/2306.12146v1 | https://arxiv.org/pdf/2306.12146v1.pdf | Which Spurious Correlations Impact Reasoning in NLI Models? A Visual Interactive Diagnosis through Data-Constrained Counterfactuals | We present a human-in-the-loop dashboard tailored to diagnosing potential spurious features that NLI models rely on for predictions. The dashboard enables users to generate diverse and challenging examples by drawing inspiration from GPT-3 suggestions. Additionally, users can receive feedback from a trained NLI model on how challenging the newly created example is and make refinements based on the feedback. Through our investigation, we discover several categories of spurious correlations that impact the reasoning of NLI models, which we group into three categories: Semantic Relevance, Logical Fallacies, and Bias. Based on our findings, we identify and describe various research opportunities, including diversifying training data and assessing NLI models' robustness by creating adversarial test suites. | ['Mennatallah El-Assady', 'Afra Amini', 'Robin Chan'] | 2023-06-21 | null | null | null | null | ['logical-fallacies'] | ['miscellaneous'] | [ 3.14471662e-01 7.10958183e-01 -2.98272908e-01 -6.99762523e-01
-6.44717097e-01 -7.46369123e-01 5.82493365e-01 7.53835365e-02
2.12592199e-01 1.00081193e+00 2.96944737e-01 -4.64413732e-01
-2.34680712e-01 -6.71366215e-01 -8.40635836e-01 2.40535349e-01
3.43135029e-01 6.71254277e-01 -1.33974031e-01 1.44890314e-02
4.80724096e-01 3.65105540e-01 -1.35051215e+00 7.05353498e-01
1.31055987e+00 7.22453296e-01 -5.64339519e-01 3.85314286e-01
-5.95442690e-02 1.15243709e+00 -1.09355474e+00 -5.98146915e-01
2.11242750e-01 -2.99973100e-01 -6.31419718e-01 -6.92864537e-01
5.76065600e-01 -1.70373470e-01 -1.77887261e-01 7.80374408e-01
3.25972587e-01 -3.47085372e-02 6.49688005e-01 -1.75284529e+00
-9.21231270e-01 8.62364590e-01 4.30702507e-01 1.83467016e-01
8.11898649e-01 7.18066752e-01 8.87573659e-01 -9.55945611e-01
8.07230711e-01 1.38602293e+00 1.00561035e+00 7.57484198e-01
-1.33609581e+00 -1.28326190e+00 2.49059588e-01 3.31764668e-01
-1.11705542e+00 -4.56828892e-01 5.66275716e-01 -4.21455741e-01
1.31892812e+00 5.47498465e-01 3.92541349e-01 1.62519920e+00
2.02282041e-01 7.33430743e-01 1.13959980e+00 -1.43196568e-01
3.66232604e-01 4.69516516e-01 2.90518910e-01 4.57790494e-01
6.13603175e-01 5.66097021e-01 -7.82601535e-01 -4.97845590e-01
3.44376057e-01 -4.55214083e-01 1.36056263e-02 2.83223659e-01
-1.06057084e+00 6.12576246e-01 2.06859082e-01 -8.01166967e-02
-2.05785796e-01 -2.36972012e-02 5.05865775e-02 3.36165756e-01
2.39908442e-01 1.46196318e+00 -9.57498014e-01 -1.39825836e-01
-8.02918375e-01 8.01429212e-01 1.01943564e+00 1.26394689e+00
6.45754218e-01 -1.09016985e-01 -4.98098999e-01 2.74082005e-01
3.11114579e-01 4.52360064e-01 4.95528162e-01 -9.95082378e-01
5.12751639e-01 1.01079106e+00 3.85260403e-01 -9.36648786e-01
-3.79017830e-01 -5.54530203e-01 -3.31929505e-01 2.92201728e-01
3.76229554e-01 -4.07456070e-01 -7.43115425e-01 1.51709652e+00
1.42892882e-01 1.61503002e-01 1.15063928e-01 6.06222153e-01
1.00064540e+00 1.61868006e-01 4.50475305e-01 1.62104711e-01
6.47601843e-01 -6.53379261e-01 -3.94789368e-01 -4.36811805e-01
8.52582633e-01 -4.13005322e-01 1.47346115e+00 5.37531078e-01
-1.06654060e+00 -6.70297623e-01 -1.04831278e+00 1.12427399e-01
-3.95964384e-01 -1.65328234e-01 6.00453734e-01 3.55525583e-01
-7.11756051e-01 1.08442700e+00 -2.67591357e-01 -1.38592273e-01
5.35174191e-01 1.45476297e-01 2.65932754e-02 -6.62526414e-02
-1.43091559e+00 1.13833547e+00 3.56002182e-01 -2.02353731e-01
-7.55515814e-01 -1.45842648e+00 -7.06166029e-01 2.30772033e-01
4.15302902e-01 -9.29224193e-01 1.30291569e+00 -8.42345059e-01
-1.18844903e+00 1.94241226e-01 -3.48001748e-01 -4.18986768e-01
8.61272812e-01 -2.03160197e-01 -6.69023871e-01 -3.01167071e-01
2.92499810e-01 7.10122168e-01 5.49206018e-01 -1.18226182e+00
-5.28934121e-01 -9.58688110e-02 5.82472596e-04 -2.61112470e-02
6.85504377e-02 4.67270836e-02 1.97996676e-01 -6.03911817e-01
-1.12394527e-01 -7.96929777e-01 -1.83820635e-01 -1.30755931e-01
-7.90707529e-01 -2.50155032e-01 6.63099170e-01 -5.48639417e-01
1.48758221e+00 -1.80931449e+00 -4.79234368e-01 5.21269917e-01
1.38232395e-01 3.68603915e-01 -2.77629375e-01 2.99354166e-01
-1.81058124e-01 1.03094280e+00 2.40215048e-01 3.57961357e-01
1.70060083e-01 -1.70908377e-01 -5.58826864e-01 -4.85327303e-01
6.27978563e-01 1.25296533e+00 -1.14845741e+00 -4.84373778e-01
-5.46106545e-04 -1.07722282e-02 -7.36797571e-01 2.28102669e-01
-6.60478413e-01 6.97093189e-01 -4.48271066e-01 5.87243140e-01
3.87558639e-01 -4.42245334e-01 -5.76697886e-02 -1.68625280e-01
3.74988735e-01 9.06789899e-01 -7.22650051e-01 9.22954559e-01
-1.78686053e-01 4.31375891e-01 -7.93181241e-01 -4.72512484e-01
1.16396236e+00 1.60918236e-01 -2.21856937e-01 -5.53996325e-01
-4.61969465e-01 1.13506138e-01 1.80804893e-01 -6.28690600e-01
2.73606569e-01 -3.85256521e-02 -2.06832998e-02 5.56039095e-01
-1.33907631e-01 -3.22639257e-01 1.48382425e-01 1.62109077e-01
1.37154472e+00 2.83962667e-01 3.74314934e-01 5.15848771e-02
2.99905300e-01 4.12694722e-01 8.08653235e-01 1.18693507e+00
-2.12446153e-01 2.01974452e-01 5.19717395e-01 -5.96542776e-01
-9.72583950e-01 -9.91930485e-01 5.57467714e-02 6.66424572e-01
-3.46380562e-01 -4.98602271e-01 -3.46838266e-01 -1.28087020e+00
3.68638515e-01 1.73809266e+00 -7.28270888e-01 -4.56717789e-01
-1.14484310e-01 -1.77189365e-01 7.43103385e-01 7.67310739e-01
1.81170553e-01 -1.41831672e+00 -2.55360246e-01 7.82932565e-02
-1.59956992e-01 -6.43196464e-01 -3.65812294e-02 -5.66972308e-02
-9.24395442e-01 -1.25801134e+00 3.22647452e-01 -3.78858387e-01
6.55102849e-01 -1.11654758e-01 1.44711483e+00 2.19355032e-01
-1.95675761e-01 1.12263776e-01 -3.61147314e-01 -6.45493269e-01
-9.41922367e-01 -1.15581537e-02 -1.35040715e-01 -1.00341558e+00
7.66719639e-01 -5.45840561e-01 -3.43076319e-01 7.44306624e-01
-4.80843782e-01 2.11953565e-01 4.58539069e-01 5.86357057e-01
4.50671345e-01 6.09068759e-02 7.71864235e-01 -1.38583052e+00
8.80285263e-01 -6.66357875e-01 -3.52308154e-01 7.31974542e-01
-1.17937839e+00 2.20215917e-01 9.34516251e-01 -5.11426568e-01
-1.23168957e+00 -4.22744185e-01 1.65151685e-01 -1.81880593e-01
-3.95694613e-01 6.73275352e-01 -3.48821133e-01 -7.74164870e-02
1.09383488e+00 -1.68466702e-01 -4.65267628e-01 -1.75030977e-01
4.45775628e-01 4.34393048e-01 2.76453525e-01 -6.60624146e-01
9.28778350e-01 -3.29615176e-01 -4.46894228e-01 -2.16532871e-01
-1.07425618e+00 3.16378415e-01 -3.22881520e-01 -1.01792529e-01
8.25813338e-02 -5.70061266e-01 -7.35013485e-01 6.47508278e-02
-9.13506448e-01 -7.69781768e-01 -4.86504972e-01 2.21038461e-01
-4.02683884e-01 -4.08387296e-02 -3.46828133e-01 -4.48068649e-01
-4.51510757e-01 -8.31644893e-01 3.14239979e-01 3.41125190e-01
-1.29790807e+00 -9.02807653e-01 -1.68257490e-01 6.00666046e-01
5.45646727e-01 2.44365409e-01 1.49571526e+00 -1.25594044e+00
-5.66891253e-01 -2.58710682e-01 -2.54485279e-01 2.74525523e-01
-1.82429895e-01 2.15106368e-01 -9.37996089e-01 1.64302647e-01
-1.86130047e-01 -5.46425521e-01 1.40159532e-01 -1.64016206e-02
1.28522158e+00 -7.26419330e-01 -4.67797697e-01 4.42742258e-01
8.78104627e-01 2.16132402e-01 5.80156982e-01 3.67787600e-01
4.87490207e-01 5.15183628e-01 9.53145504e-01 2.93793201e-01
7.22217038e-02 1.00666106e-01 6.76068142e-02 3.65640283e-01
3.37931439e-02 -9.77680206e-01 1.91298261e-01 -4.33008447e-02
4.26719278e-01 -3.24288577e-01 -1.04098618e+00 1.15625486e-01
-1.70629847e+00 -8.22890162e-01 9.90972295e-02 2.00655246e+00
1.02632141e+00 6.18700325e-01 -3.92527044e-01 -1.16419427e-01
3.86950821e-01 -2.91044086e-01 -1.21224976e+00 -4.93856847e-01
-1.21946996e-02 3.77085209e-01 2.62515008e-01 6.57408774e-01
-5.03729165e-01 1.24899709e+00 7.63479042e+00 4.89072055e-01
-9.13454175e-01 -3.71440440e-01 7.48734355e-01 -2.57997513e-01
-9.74384069e-01 1.97676077e-01 -9.45238054e-01 4.80127990e-01
1.07550275e+00 -6.53631926e-01 5.23903131e-01 9.27222013e-01
3.99613261e-01 1.94229007e-01 -1.57810867e+00 1.90850794e-01
-2.33283043e-02 -1.47001398e+00 5.53982556e-01 -2.46255651e-01
6.37093842e-01 -1.87955275e-01 7.99372271e-02 6.84116304e-01
8.25079203e-01 -1.35333419e+00 6.57825470e-01 7.88257778e-01
7.51917541e-01 -6.68202639e-01 4.75205809e-01 4.06172067e-01
-2.93693066e-01 -8.78438205e-02 -2.22445309e-01 -4.69984025e-01
-2.57485598e-01 4.90439385e-01 -1.58008134e+00 8.36878195e-02
3.81559849e-01 5.96611321e-01 -1.01009440e+00 8.72342944e-01
-9.42657471e-01 1.05697453e+00 -1.29426926e-01 -2.43746892e-01
-3.14668477e-01 2.13096976e-01 4.33100849e-01 1.12440443e+00
2.76628673e-01 1.05342507e-01 -8.65558460e-02 1.56239283e+00
-9.30071622e-03 -2.70702064e-01 -6.99709713e-01 -2.70572662e-01
1.01131201e+00 9.50412214e-01 -1.49208918e-01 -5.90339005e-01
-1.39006540e-01 6.78892255e-01 2.66331911e-01 5.78691423e-01
-7.30235934e-01 -3.60457689e-01 8.90968800e-01 1.20982803e-01
-5.73988408e-02 1.69431373e-01 -9.05791640e-01 -1.02518046e+00
-9.53253731e-03 -1.34486139e+00 1.63871512e-01 -1.43275177e+00
-1.56631792e+00 3.35074008e-01 4.18873727e-02 -8.82897496e-01
-5.34443855e-01 -4.43034232e-01 -9.08217549e-01 1.00338566e+00
-1.22278547e+00 -7.68195033e-01 -4.15744036e-01 1.69484928e-01
2.06862688e-01 -1.50472239e-01 1.07252288e+00 -2.99330533e-01
-4.76959497e-01 1.11615872e+00 -4.29571837e-01 -1.46549642e-02
9.19564545e-01 -1.01681757e+00 1.17212915e+00 5.61503291e-01
-1.46675423e-01 1.06558108e+00 9.06229496e-01 -1.32412636e+00
-8.23772371e-01 -1.23283613e+00 1.33808851e+00 -8.43611360e-01
7.42555678e-01 -2.33213827e-02 -1.11243773e+00 9.96002197e-01
-3.11361998e-01 -2.66436517e-01 1.11789370e+00 3.62765074e-01
-5.78047872e-01 3.35257500e-01 -1.61861002e+00 7.28146017e-01
1.39785087e+00 -3.33637953e-01 -8.26146603e-01 3.52019072e-01
9.80147600e-01 -3.80577385e-01 -8.96908462e-01 4.06048089e-01
4.47190404e-01 -9.02863801e-01 8.77062380e-01 -1.30588877e+00
7.62772858e-01 -1.50602862e-01 3.27170968e-01 -1.42450118e+00
-6.32736504e-01 -6.09662771e-01 -7.04945996e-02 1.16474295e+00
1.26828492e+00 -7.35450685e-01 1.01906300e+00 1.58259141e+00
2.99062859e-02 -7.10557640e-01 -2.17754111e-01 -6.21329188e-01
1.00105181e-02 -8.01452100e-01 1.01583779e+00 9.99929547e-01
4.56894219e-01 4.51772690e-01 1.60515234e-02 8.62832144e-02
3.04725111e-01 -8.14147890e-02 8.97985339e-01 -1.36814582e+00
-3.07542294e-01 -2.16879129e-01 5.49435206e-02 -6.35521293e-01
1.64828226e-01 -1.01416123e+00 -1.90018579e-01 -1.34141672e+00
9.22103971e-02 -6.84241354e-01 -1.52568549e-01 8.74641359e-01
-5.72505236e-01 -1.80996984e-01 3.10394019e-01 2.88789839e-01
-4.98418778e-01 1.20120876e-01 1.02506745e+00 -1.07570946e-01
-2.49259993e-01 6.51887879e-02 -1.30972409e+00 8.40881169e-01
8.98785174e-01 -4.78184223e-01 -5.74428141e-01 -2.95872420e-01
7.77332067e-01 -1.40825674e-01 3.38817149e-01 -1.12236714e+00
1.26918763e-01 -5.33965647e-01 8.54269505e-01 -3.78522336e-01
-1.08876735e-01 -4.44566965e-01 1.12771578e-01 3.49290371e-01
-1.10339558e+00 -1.51909798e-01 4.85284984e-01 2.97438443e-01
7.04613402e-02 -2.99975812e-01 2.81205535e-01 -1.65999874e-01
-4.89547372e-01 4.28848900e-02 -1.98846340e-01 4.59115922e-01
6.88452423e-01 -2.03409523e-01 -6.13068581e-01 -4.39053118e-01
-7.23157942e-01 5.26257515e-01 5.49785674e-01 4.51069772e-01
7.03332603e-01 -1.06235278e+00 -6.70974851e-01 5.61385214e-01
3.25836390e-01 -1.54596031e-01 -4.48396280e-02 2.18837559e-01
-3.25710736e-02 4.62456614e-01 -9.19797942e-02 1.86804105e-02
-8.46415460e-01 5.93874931e-01 2.17611760e-01 -3.43385339e-01
-3.97316307e-01 9.00378942e-01 -2.09808685e-02 -7.30797768e-01
2.93443054e-01 -5.74474514e-01 2.12692954e-02 -4.39922512e-01
5.79061925e-01 4.62767005e-01 -9.01153758e-02 1.58998296e-01
-3.79525006e-01 -1.63679048e-01 -1.44851044e-01 -1.26469880e-01
1.05685759e+00 2.57855564e-01 2.93234438e-01 2.83261091e-01
7.46522605e-01 -1.74068063e-01 -1.23077190e+00 -1.75763562e-01
1.09267242e-01 -5.90550482e-01 -4.15249169e-01 -2.02262807e+00
-2.99376518e-01 5.83813310e-01 8.21912382e-03 -7.16612414e-02
5.64417005e-01 -3.07032436e-01 6.09763563e-01 7.09808230e-01
1.58446759e-01 -1.03986144e+00 1.75363868e-02 5.42461812e-01
1.21837032e+00 -1.13502610e+00 -8.30994993e-02 -3.74506861e-01
-9.26823139e-01 1.03988242e+00 1.03234529e+00 -8.97670016e-02
4.05801594e-01 2.73054808e-01 1.22960031e-01 -2.70231739e-02
-1.18315804e+00 5.76029360e-01 2.38735735e-01 8.62782121e-01
5.37576795e-01 7.55921230e-02 -1.25512138e-01 9.97215450e-01
-6.49812639e-01 5.89727521e-01 3.64876300e-01 5.61961889e-01
-2.35407814e-01 -9.68379498e-01 -3.64828885e-01 9.02095437e-01
8.78702328e-02 -3.76523614e-01 -1.00155675e+00 7.74119616e-01
1.04728222e-01 1.08478892e+00 -2.37439975e-01 -8.45219374e-01
5.88826239e-01 5.19562006e-01 1.89029902e-01 -8.20812643e-01
-8.84165823e-01 -7.98289895e-01 5.92844129e-01 -9.15477157e-01
3.74914765e-01 -6.29999757e-01 -1.20164895e+00 -3.42916250e-01
-6.83426205e-03 1.90936282e-01 4.88492727e-01 1.05773222e+00
8.44968915e-01 4.27701741e-01 1.90633550e-01 -4.74605471e-01
-8.42367530e-01 -1.25830328e+00 2.19440863e-01 6.34526014e-01
-1.84622079e-01 -3.83670896e-01 -5.81353009e-01 -2.65948892e-01] | [10.613188743591309, 8.12588882446289] |
0c5e7683-9ac5-4533-bfe1-166d09b39012 | constraining-linear-chain-crfs-to-regular | 2106.07306 | null | https://arxiv.org/abs/2106.07306v5 | https://arxiv.org/pdf/2106.07306v5.pdf | Constraining Linear-chain CRFs to Regular Languages | A major challenge in structured prediction is to represent the interdependencies within output structures. When outputs are structured as sequences, linear-chain conditional random fields (CRFs) are a widely used model class which can learn \textit{local} dependencies in the output. However, the CRF's Markov assumption makes it impossible for CRFs to represent distributions with \textit{nonlocal} dependencies, and standard CRFs are unable to respect nonlocal constraints of the data (such as global arity constraints on output labels). We present a generalization of CRFs that can enforce a broad class of constraints, including nonlocal ones, by specifying the space of possible output structures as a regular language $\mathcal{L}$. The resulting regular-constrained CRF (RegCCRF) has the same formal properties as a standard CRF, but assigns zero probability to all label sequences not in $\mathcal{L}$. Notably, RegCCRFs can incorporate their constraints during training, while related models only enforce constraints during decoding. We prove that constrained training is never worse than constrained decoding, and show empirically that it can be substantially better in practice. Additionally, we demonstrate a practical benefit on downstream tasks by incorporating a RegCCRF into a deep neural model for semantic role labeling, exceeding state-of-the-art results on a standard dataset. | ['Sebastian Padó', 'Roman Klinger', 'Sean Papay'] | 2021-06-14 | constraining-linear-chain-crfs-to-regular-1 | https://openreview.net/forum?id=jbrgwbv8nD | https://openreview.net/pdf?id=jbrgwbv8nD | iclr-2022-4 | ['semantic-role-labeling'] | ['natural-language-processing'] | [ 5.49723387e-01 5.47805727e-01 -5.48697472e-01 -9.22475994e-01
-5.99759102e-01 -1.02210855e+00 5.07642388e-01 7.49182925e-02
-3.69713932e-01 9.81672943e-01 3.92173856e-01 -7.16174364e-01
1.25636876e-01 -7.35630751e-01 -1.03062952e+00 -5.93703449e-01
1.45389978e-02 7.62460113e-01 1.18147261e-01 3.21685191e-04
-1.59114257e-01 3.11276168e-01 -1.49500549e+00 5.02045691e-01
4.86761332e-01 7.62863755e-01 5.16228616e-01 4.25769240e-01
-3.23709488e-01 1.11399961e+00 -6.25148058e-01 -2.64817268e-01
-1.36027977e-01 -3.41915637e-01 -1.04962564e+00 -1.52359650e-01
2.31090039e-01 4.33260249e-03 -6.61471188e-02 8.79445314e-01
-8.76959264e-02 2.19884634e-01 8.56541395e-01 -1.08446014e+00
-6.69889271e-01 1.14282715e+00 4.15562093e-02 -4.23804931e-02
1.96575075e-01 -1.62212774e-01 1.55722606e+00 -5.86130321e-01
5.74645877e-01 1.52258682e+00 4.13212746e-01 7.07938254e-01
-1.61500895e+00 -6.14227831e-01 6.82727695e-01 -2.03416541e-01
-1.19452941e+00 -2.69825578e-01 1.17373124e-01 -4.55189615e-01
1.39609563e+00 1.54623210e-01 3.35563608e-02 1.03795707e+00
1.40136480e-01 8.05781543e-01 1.05589032e+00 -7.03830123e-01
1.67128101e-01 -1.47319481e-01 4.99079078e-01 6.94650054e-01
-2.06702188e-01 3.68908426e-04 -5.86977839e-01 -4.98463959e-02
7.23757088e-01 -1.00595891e-01 9.96367726e-03 1.95573062e-01
-1.12215817e+00 8.26797128e-01 1.13451026e-01 2.84007311e-01
4.31576073e-01 5.17951906e-01 1.55389994e-01 2.77128577e-01
2.99489141e-01 9.21786055e-02 -1.03069222e+00 5.71676195e-02
-5.97437799e-01 9.46241990e-02 8.51813376e-01 1.44142234e+00
1.01855993e+00 -1.46792859e-01 -4.10159498e-01 8.05431187e-01
5.53004622e-01 4.87602472e-01 1.31226495e-01 -1.17026222e+00
5.75438499e-01 2.52459466e-01 6.65172637e-02 -2.56161422e-01
-3.63335043e-01 -2.95475215e-01 -6.68673575e-01 -1.95085302e-01
5.97148120e-01 -1.74437463e-01 -1.26227391e+00 2.17207599e+00
-2.38917708e-01 6.96770847e-02 -8.51948261e-02 5.32274067e-01
3.91106904e-01 8.80434334e-01 6.64783537e-01 -3.26787502e-01
1.14000630e+00 -7.84264088e-01 -6.89413488e-01 -4.91938621e-01
1.16153145e+00 -5.01158774e-01 1.02609527e+00 2.23088637e-01
-7.77159631e-01 -5.03136098e-01 -6.61502600e-01 -2.74382383e-01
-2.24196821e-01 8.88831392e-02 9.31123137e-01 6.89125121e-01
-1.29116499e+00 5.23982823e-01 -9.30931985e-01 1.85732514e-01
1.18702143e-01 7.09463775e-01 -3.10034186e-01 -1.22233063e-01
-1.27370512e+00 9.00266767e-01 5.50158918e-01 2.44792365e-02
-9.94146585e-01 -4.43823367e-01 -1.18997145e+00 2.24388182e-01
4.26708639e-01 -2.82354623e-01 1.62076795e+00 -8.26431870e-01
-1.37032676e+00 9.29763615e-01 -7.51747072e-01 -3.58390838e-01
-1.01032898e-01 -8.13884810e-02 -3.74760807e-01 -2.43501902e-01
9.46425870e-02 8.05785537e-01 5.11249959e-01 -1.20096588e+00
-5.63246846e-01 -8.70246440e-02 2.27697760e-01 1.46204352e-01
8.45513195e-02 4.20118690e-01 -5.18711030e-01 -5.74201941e-01
2.01142073e-01 -1.01134920e+00 -4.61413175e-01 -4.40148294e-01
-8.34548593e-01 -7.22666323e-01 2.38474965e-01 -2.29875013e-01
1.29656839e+00 -2.02542281e+00 3.58821712e-02 3.11064303e-01
-1.71902329e-01 2.46305130e-02 8.09236690e-02 2.67457932e-01
-8.89797062e-02 5.25455952e-01 -6.22684181e-01 -7.51862645e-01
3.89462918e-01 9.89089906e-01 -5.36717594e-01 3.19636554e-01
3.11813861e-01 9.24826264e-01 -7.21955061e-01 -4.72189009e-01
-3.83460820e-02 1.24971785e-01 -6.22892678e-01 1.84840053e-01
-1.03380406e+00 6.08576894e-01 -3.01090598e-01 4.73682493e-01
4.15126264e-01 -5.21835864e-01 8.34755540e-01 4.15496320e-01
-4.04226445e-02 9.02548134e-01 -9.29293394e-01 1.53445339e+00
-4.90215659e-01 3.37749213e-01 2.85893437e-02 -1.15780115e+00
9.04691041e-01 3.43130380e-01 1.37429059e-01 -4.85959202e-01
-1.08305059e-01 6.09952249e-02 -6.80396855e-02 1.83129180e-02
3.61252755e-01 -3.82663369e-01 -4.92132187e-01 5.66018879e-01
2.90615797e-01 6.17147423e-02 1.50627553e-01 3.44787478e-01
1.06778610e+00 5.30615091e-01 -4.43400815e-02 -4.22748268e-01
2.28298739e-01 -2.17498004e-01 8.34544718e-01 1.03378093e+00
2.55736977e-01 4.30736542e-01 6.90294385e-01 -1.13188840e-01
-8.22510600e-01 -1.11264110e+00 -3.74581337e-01 1.69708300e+00
3.24253663e-02 -6.45626724e-01 -5.08033514e-01 -9.77927744e-01
-7.64647275e-02 8.75948548e-01 -3.99508804e-01 1.81914836e-01
-8.79535079e-01 -6.39028251e-01 8.04861903e-01 8.00815523e-01
-4.23434414e-02 -1.42658854e+00 -2.28015885e-01 3.05496097e-01
-8.41648728e-02 -1.18747139e+00 -3.00768971e-01 1.15907371e+00
-8.49679649e-01 -7.59620130e-01 -8.70058089e-02 -1.18014264e+00
8.74011457e-01 -5.75312972e-01 1.48847687e+00 2.82966167e-01
2.98206151e-01 -2.00126860e-02 -2.23874941e-01 -9.62843075e-02
-5.27235389e-01 1.20388374e-01 -1.63933232e-01 -3.76773864e-01
3.28779042e-01 -5.36058962e-01 -8.74357391e-03 3.28129619e-01
-9.55328643e-01 7.60104805e-02 3.31272870e-01 7.95510054e-01
8.47195089e-01 -1.17177004e-02 4.36519623e-01 -1.55504310e+00
4.66760881e-02 -5.89438558e-01 -5.27949452e-01 3.14340085e-01
-4.69485670e-01 6.08159363e-01 9.83188212e-01 -1.87036291e-01
-1.28051913e+00 3.70320469e-01 -4.81289089e-01 -4.27314267e-02
-6.69957817e-01 6.48324370e-01 -3.84452492e-01 6.81875229e-01
4.74211276e-01 1.92037731e-01 -4.15992558e-01 -8.48710597e-01
4.60246205e-01 4.27421987e-01 4.94695663e-01 -1.23453033e+00
4.14120942e-01 1.31022096e-01 1.04669854e-01 -4.33753192e-01
-1.32601762e+00 -2.44885311e-01 -7.75112689e-01 2.94610143e-01
8.44150603e-01 -1.09661508e+00 -8.32250774e-01 2.35922039e-01
-1.23468685e+00 -9.80784595e-01 -1.49833620e-01 1.72619149e-01
-7.05846846e-01 3.06133747e-01 -1.04669321e+00 -8.38724136e-01
1.95568010e-01 -1.01402497e+00 9.88353550e-01 -8.88224989e-02
-2.49204069e-01 -1.19734454e+00 -3.61389339e-01 3.37055922e-02
1.89668480e-02 -1.90621689e-01 1.52597916e+00 -7.91812003e-01
-5.55185616e-01 4.42052573e-01 -2.99919043e-02 4.50452358e-01
-1.49429753e-01 -2.12744042e-01 -7.61802614e-01 9.44835544e-02
-3.73776197e-01 -5.87121069e-01 1.09803462e+00 2.59280264e-01
1.45982838e+00 -3.63419414e-01 -6.02298915e-01 3.71781945e-01
1.29164672e+00 1.83776274e-01 5.74550509e-01 -1.36715338e-01
6.75672889e-01 5.27389467e-01 4.61740166e-01 2.54189551e-01
5.22357047e-01 5.71472585e-01 4.56342667e-01 1.98234662e-01
7.18596950e-02 -6.80915594e-01 6.02995455e-01 5.99600136e-01
1.91720068e-01 -5.86331666e-01 -8.12719047e-01 4.92335230e-01
-1.85441864e+00 -6.42579496e-01 -5.50506830e-01 2.01422024e+00
1.40698004e+00 2.56285518e-01 -3.38116378e-01 -8.67589787e-02
8.17648709e-01 5.87009266e-02 -3.05519998e-01 -5.69485545e-01
-2.96879739e-01 8.15923989e-01 6.03358984e-01 8.28379631e-01
-1.11944222e+00 1.37639523e+00 7.08051014e+00 8.21673751e-01
-5.51051259e-01 1.76875547e-01 7.40935504e-01 3.07167560e-01
-8.35018694e-01 3.71454567e-01 -1.56798840e+00 5.43022275e-01
1.22141325e+00 5.19719124e-01 4.04230893e-01 7.02967227e-01
-1.42394468e-01 -1.24455363e-01 -1.51268029e+00 4.74792302e-01
-3.11445326e-01 -1.21120584e+00 -2.00154290e-01 3.76298316e-02
6.98936403e-01 1.29693791e-01 -5.34494258e-02 6.45796657e-01
1.18577552e+00 -1.58051002e+00 9.81525540e-01 2.87961781e-01
1.20663309e+00 -6.43273532e-01 3.47034335e-01 8.08936894e-01
-1.06602573e+00 -1.74503252e-01 -4.58106995e-01 -4.51031327e-01
3.11600447e-01 6.97463632e-01 -6.92191005e-01 2.67351717e-01
4.97461319e-01 8.69378746e-01 -1.64873123e-01 3.24206680e-01
-8.92306924e-01 1.10131216e+00 -4.82860029e-01 6.58445209e-02
1.29120708e-01 -5.64504080e-02 2.88012233e-02 1.61189151e+00
4.15180586e-02 2.74084985e-01 4.94456708e-01 8.63073826e-01
-2.44532555e-01 -2.46565118e-01 -4.78668720e-01 -2.97693331e-02
5.61756492e-01 7.01645374e-01 -7.19014645e-01 -3.77736390e-01
-3.45607162e-01 7.56691098e-01 7.27404654e-01 4.87379819e-01
-7.43705630e-01 1.76488146e-01 6.20340705e-01 -9.76011530e-02
4.79757696e-01 -4.40808803e-01 -4.60565954e-01 -1.17624986e+00
-2.30942070e-01 -5.98155379e-01 6.45656407e-01 -5.81036031e-01
-1.35584915e+00 3.58425826e-01 2.05150694e-01 -6.69611275e-01
-4.68313605e-01 -8.53050232e-01 -9.86873358e-02 1.21045172e+00
-1.59873080e+00 -1.09111357e+00 5.84358931e-01 8.27493191e-01
4.22180057e-01 2.00960904e-01 1.35586476e+00 -4.16664407e-02
-2.77739435e-01 5.70666254e-01 -1.09151952e-01 2.73361921e-01
3.93413991e-01 -1.50572085e+00 2.78054237e-01 8.25296104e-01
3.37524295e-01 9.13942337e-01 4.89490002e-01 -6.77781641e-01
-1.12334275e+00 -1.23226583e+00 1.45859301e+00 -6.11439288e-01
4.80635166e-01 -7.63601720e-01 -7.52584100e-01 1.58865821e+00
3.72788943e-02 2.71925956e-01 7.92808771e-01 5.05217969e-01
-6.81921601e-01 4.41323280e-01 -7.32025504e-01 3.50121170e-01
1.43410075e+00 -7.16058791e-01 -5.44742286e-01 4.29933578e-01
1.14834368e+00 -4.88496423e-01 -7.42962182e-01 4.64116603e-01
2.97205895e-01 -7.54427373e-01 6.39114916e-01 -7.51089513e-01
5.31634748e-01 -3.46071482e-01 -5.07126808e-01 -1.06049228e+00
-3.15816075e-01 -5.54734111e-01 -2.60904312e-01 1.20087886e+00
9.00101781e-01 -5.31997085e-01 8.67378473e-01 7.49496460e-01
-5.43566406e-01 -5.40246904e-01 -8.13535571e-01 -7.13613331e-01
4.93784666e-01 -9.62790370e-01 3.57970655e-01 8.60655844e-01
6.98619708e-02 4.33217853e-01 -3.44398528e-01 2.90454626e-01
3.15628767e-01 1.95618749e-01 1.63496718e-01 -1.28520894e+00
-7.41030335e-01 -1.41635146e-02 5.40282167e-02 -1.77989697e+00
9.43692327e-01 -1.37873173e+00 5.23495138e-01 -1.57024372e+00
7.61821344e-02 -1.29196942e+00 -3.13248217e-01 1.28397977e+00
-2.66334787e-02 1.39677852e-01 1.69210866e-01 2.24922925e-01
-7.06721425e-01 3.08342576e-01 1.14672327e+00 3.62598933e-02
-3.17614414e-02 9.33497548e-02 -7.35576093e-01 6.73703730e-01
6.57335520e-01 -7.75721252e-01 -4.65380728e-01 -5.97289622e-01
5.45506239e-01 3.02796394e-01 4.85689379e-02 -4.56672907e-01
9.82809067e-02 -3.22410733e-01 2.65174568e-01 -4.66296256e-01
3.49109977e-01 -6.73398137e-01 8.99611861e-02 -1.01248607e-01
-8.58557761e-01 -2.82587856e-01 4.08009663e-02 5.36614239e-01
-1.55189961e-01 -4.33705032e-01 4.25572753e-01 -3.32773298e-01
-6.07800424e-01 1.38452008e-01 -7.07061887e-01 2.32778743e-01
5.85214913e-01 1.74274817e-01 -3.38179380e-01 -1.99149832e-01
-1.13919783e+00 2.47011945e-01 2.48999193e-01 1.41885996e-01
2.37092271e-01 -8.65204394e-01 -3.23957115e-01 3.02331865e-01
-1.36525750e-01 5.94107747e-01 -1.27156988e-01 3.95151943e-01
-3.01758554e-02 5.97716928e-01 2.85551906e-01 -5.65980613e-01
-8.93439829e-01 3.05230916e-01 1.89501211e-01 -7.04747915e-01
-4.08644885e-01 1.08457136e+00 3.53515446e-01 -9.33358550e-01
3.75560135e-01 -5.82855582e-01 -1.42564520e-01 -3.07463616e-01
1.24977857e-01 -4.71350163e-01 1.63716339e-02 -6.55822337e-01
-3.44376206e-01 9.41969007e-02 -3.57753225e-03 -2.32818663e-01
1.15822697e+00 -2.61580609e-02 -3.02089334e-01 5.71586132e-01
1.05217803e+00 1.02526974e-02 -1.47897983e+00 -4.05197322e-01
3.77430886e-01 -1.33381654e-02 -4.28893656e-01 -1.14569449e+00
-6.02919042e-01 9.10086095e-01 -1.96914345e-01 1.69225544e-01
8.82919610e-01 3.85868788e-01 4.63724673e-01 4.94265437e-01
6.43229544e-01 -8.12360525e-01 -1.80385739e-01 1.33242285e+00
2.30578616e-01 -8.24934840e-01 -4.75895852e-01 -7.41458833e-01
-8.04251969e-01 8.89614820e-01 4.64760780e-01 3.41588855e-02
7.90312707e-01 7.51802802e-01 -7.63116851e-02 1.59339339e-01
-1.10108411e+00 -3.54323834e-01 -6.90768659e-02 6.18012547e-01
8.17421138e-01 3.37910473e-01 -1.80043697e-01 8.44917715e-01
-3.44998300e-01 -3.37900370e-01 2.97335088e-01 9.91920233e-01
-4.55647022e-01 -1.74867928e+00 -9.02918801e-02 4.70529824e-01
-8.39127123e-01 -4.21204150e-01 3.10118161e-02 4.34067279e-01
4.74639624e-01 1.24412823e+00 2.27083683e-01 -1.40795961e-01
-1.04077600e-01 5.31324446e-01 4.93114591e-01 -1.35703981e+00
-5.87804437e-01 8.46989080e-02 3.80401462e-01 -3.51356477e-01
-4.67323273e-01 -5.62605977e-01 -2.07575798e+00 8.63304809e-02
-1.60283864e-01 1.85583785e-01 5.14305472e-01 1.45883346e+00
1.58048168e-01 2.37894937e-01 4.00706202e-01 -3.47824275e-01
-4.58767682e-01 -9.59272802e-01 -7.87930250e-01 3.76582891e-01
2.04336807e-01 -4.85256970e-01 -2.38001779e-01 4.81025010e-01] | [10.4258451461792, 9.431923866271973] |
ca00e7ba-258a-4490-b440-25f030294780 | predict-then-propagate-graph-neural-networks | 1810.05997 | null | https://arxiv.org/abs/1810.05997v6 | https://arxiv.org/pdf/1810.05997v6.pdf | Predict then Propagate: Graph Neural Networks meet Personalized PageRank | Neural message passing algorithms for semi-supervised classification on graphs have recently achieved great success. However, for classifying a node these methods only consider nodes that are a few propagation steps away and the size of this utilized neighborhood is hard to extend. In this paper, we use the relationship between graph convolutional networks (GCN) and PageRank to derive an improved propagation scheme based on personalized PageRank. We utilize this propagation procedure to construct a simple model, personalized propagation of neural predictions (PPNP), and its fast approximation, APPNP. Our model's training time is on par or faster and its number of parameters on par or lower than previous models. It leverages a large, adjustable neighborhood for classification and can be easily combined with any neural network. We show that this model outperforms several recently proposed methods for semi-supervised classification in the most thorough study done so far for GCN-like models. Our implementation is available online. | ['Johannes Gasteiger', 'Stephan Günnemann', 'Aleksandar Bojchevski'] | 2018-10-14 | predict-then-propagate-graph-neural-networks-1 | https://openreview.net/forum?id=H1gL-2A9Ym | https://openreview.net/pdf?id=H1gL-2A9Ym | iclr-2019-5 | ['node-classification-on-non-homophilic'] | ['graphs'] | [ 3.50710094e-01 6.26007318e-01 -6.78378224e-01 -5.83790123e-01
-5.75781524e-01 -3.06809574e-01 6.47416353e-01 5.79759121e-01
-3.37271482e-01 7.48297453e-01 2.53736526e-02 -7.43145466e-01
-2.63724923e-01 -1.02537143e+00 -7.38724470e-01 -4.88866001e-01
-4.99713808e-01 6.92780375e-01 8.08879852e-01 -1.44286349e-01
5.93901426e-02 5.66957891e-01 -1.09533370e+00 2.40957469e-01
6.02147102e-01 8.52645338e-01 -1.77059218e-01 1.17478836e+00
-1.27318010e-01 1.07641792e+00 -1.78661793e-01 -5.48313618e-01
5.10809012e-02 -1.71845376e-01 -1.26391876e+00 -2.45264143e-01
5.46145976e-01 -4.24191713e-01 -5.11071980e-01 7.79016256e-01
1.31338894e-01 1.47965983e-01 6.20280504e-01 -1.32402492e+00
-4.62590367e-01 1.34584582e+00 -2.01718643e-01 1.30672872e-01
-2.94005703e-02 -3.79207522e-01 1.31239605e+00 -4.43661332e-01
7.13208735e-01 8.27361226e-01 1.22201002e+00 3.98846060e-01
-1.40545189e+00 -3.82634073e-01 2.56402493e-01 1.64547786e-01
-1.20637155e+00 -1.24071188e-01 6.53509140e-01 -1.30195662e-01
1.02962124e+00 1.10695548e-01 7.66269028e-01 8.31800997e-01
1.88243106e-01 8.55454922e-01 7.76728332e-01 -3.99077386e-01
2.04921335e-01 2.61863202e-01 7.23664939e-01 1.06785047e+00
2.75473714e-01 -1.90405145e-01 -4.61231202e-01 -4.36082274e-01
4.86556590e-01 3.69959958e-02 -1.16880327e-01 -3.49275291e-01
-7.15593278e-01 9.97488976e-01 8.52653682e-01 1.24828115e-01
-1.60611778e-01 4.94008303e-01 2.64202952e-01 4.09932286e-01
9.22041297e-01 6.17285073e-02 -5.87653697e-01 5.41187674e-02
-1.07776558e+00 1.71643496e-01 1.35922229e+00 6.60582185e-01
1.01834691e+00 -2.89351642e-01 4.15244289e-02 7.86221027e-01
4.57359850e-01 2.80543175e-02 1.60495177e-01 -7.38955796e-01
1.52399957e-01 6.84694946e-01 -2.00497806e-01 -1.04946280e+00
-8.17052662e-01 -8.04606616e-01 -9.06071246e-01 2.07575619e-01
5.27268708e-01 -2.11547524e-01 -8.52630794e-01 1.48250854e+00
4.11333561e-01 2.57326126e-01 -1.09480880e-01 4.17778820e-01
7.80608833e-01 5.98333061e-01 8.87866840e-02 1.46579459e-01
8.11098635e-01 -1.41720724e+00 -5.59247732e-02 -1.46089822e-01
1.20913970e+00 -1.85198545e-01 4.89392221e-01 2.70719379e-01
-9.82427537e-01 -3.47085334e-02 -1.14909220e+00 -5.43398075e-02
-6.47497654e-01 -1.60016522e-01 1.27594638e+00 7.45059192e-01
-1.79854846e+00 1.37378645e+00 -9.69464839e-01 -6.33149207e-01
5.81851602e-01 5.84720731e-01 -3.33597541e-01 1.50937229e-01
-1.14432299e+00 6.69405639e-01 4.24320698e-01 4.17905822e-02
-6.84443414e-01 -5.47327995e-01 -6.75823569e-01 1.55313596e-01
1.49725199e-01 -5.95414221e-01 1.37949431e+00 -9.16503668e-01
-1.46746314e+00 5.70274413e-01 -7.12987855e-02 -1.03858638e+00
4.91312295e-01 1.83690920e-01 -1.45514444e-01 2.38657266e-01
-3.73656005e-01 6.10509813e-01 6.65880263e-01 -9.23050106e-01
-5.40940285e-01 -1.13917843e-01 3.32889885e-01 9.82315540e-02
-7.72329330e-01 -5.83724678e-02 -4.98817474e-01 -2.95992613e-01
3.63553874e-02 -8.80338609e-01 -5.88521183e-01 1.39128372e-01
-5.81450939e-01 -3.20103943e-01 3.69789004e-01 -3.89848500e-01
1.19156754e+00 -1.63808644e+00 -1.66892648e-01 7.30593741e-01
7.02020288e-01 3.00358027e-01 -2.95844615e-01 6.27136946e-01
1.92294613e-01 2.53483355e-01 -2.68257588e-01 -4.44079846e-01
-1.54325748e-02 2.28859797e-01 -4.28783670e-02 4.23249871e-01
-6.56208536e-03 9.03562486e-01 -1.05883729e+00 -6.53998911e-01
-3.00659716e-01 6.01760685e-01 -6.31564498e-01 -2.64168054e-01
-2.74910569e-01 -2.69646883e-01 -4.01095927e-01 1.84599906e-01
6.70196950e-01 -9.01037335e-01 3.79745811e-01 1.34046793e-01
2.58786231e-01 5.06681681e-01 -9.90764976e-01 1.14710009e+00
-2.33278006e-01 7.57131934e-01 1.33705333e-01 -9.52814102e-01
7.27938533e-01 -8.06033146e-03 3.48759234e-01 3.84175777e-02
5.28053306e-02 7.95916542e-02 -1.81831926e-01 1.50547296e-01
6.11433983e-01 3.34433287e-01 5.60020983e-01 8.47666919e-01
2.36405998e-01 3.10136557e-01 4.88520801e-01 7.74094701e-01
1.68428969e+00 -2.02991497e-02 9.70219225e-02 -1.74502149e-01
3.68032783e-01 9.16173831e-02 1.61821783e-01 1.36347890e+00
-1.69312418e-01 3.88056189e-01 7.88520277e-01 -4.21498060e-01
-7.76361048e-01 -8.68721426e-01 5.78140188e-03 1.44852626e+00
-6.37637153e-02 -8.83545876e-01 -7.80029297e-01 -1.09933805e+00
-2.02091057e-02 3.30461413e-01 -6.86571240e-01 -1.19247136e-03
-4.03085202e-01 -9.45981085e-01 7.18289971e-01 7.02776372e-01
2.25986555e-01 -8.86472166e-01 2.68644035e-01 2.29660377e-01
4.48469132e-01 -9.11488593e-01 -6.19722158e-02 4.74211991e-01
-1.15362096e+00 -1.05002320e+00 -5.63333154e-01 -9.02236462e-01
9.04564738e-01 1.74698845e-01 1.13534153e+00 5.08589983e-01
1.99707076e-01 3.94789279e-01 -4.29259479e-01 -1.51820391e-01
-7.06958532e-01 8.00783217e-01 -1.44737899e-01 -5.28578088e-02
1.15698375e-01 -8.08158100e-01 -4.69044775e-01 -4.15723724e-03
-7.43036509e-01 2.93017536e-01 7.52330005e-01 8.13360631e-01
2.09344774e-01 6.60659373e-02 3.66031289e-01 -1.75109732e+00
5.59377670e-01 -5.20963013e-01 -5.03462017e-01 5.31659797e-02
-1.10623264e+00 2.13172808e-01 7.94512153e-01 -2.89184958e-01
-7.32156992e-01 1.13437474e-01 -3.65267754e-01 1.83058798e-01
-4.19452712e-02 8.31234217e-01 6.04023516e-01 -6.77288294e-01
7.39515603e-01 2.49814820e-02 -6.88382238e-02 -3.95524114e-01
4.92674410e-01 5.27313769e-01 1.14504464e-01 -1.06189847e-01
1.04689312e+00 3.93969297e-01 2.41821080e-01 -7.44397044e-01
-8.36970508e-01 -2.61458874e-01 -6.41380906e-01 -2.74483323e-01
3.46967578e-01 -5.99189878e-01 -5.28014421e-01 5.22739530e-01
-1.11543250e+00 -8.44010472e-01 6.43307418e-02 4.35673803e-01
-2.36932859e-01 4.45268631e-01 -1.29714870e+00 -6.26199901e-01
-5.21625280e-01 -4.96576488e-01 5.07355332e-01 -1.45037696e-02
1.69140659e-02 -1.40094864e+00 9.88446400e-02 1.28353655e-01
8.33736837e-01 -1.07314385e-01 8.52328002e-01 -1.20181131e+00
-5.85187554e-01 -5.70530832e-01 -5.44708192e-01 2.75496781e-01
-3.77801389e-01 1.92330196e-01 -8.34390461e-01 -3.51740122e-01
-7.90905178e-01 -3.08168322e-01 1.35469854e+00 1.81718960e-01
1.14224100e+00 -4.84634161e-01 -6.69137776e-01 5.99070251e-01
1.31293321e+00 -6.70160949e-01 3.78667593e-01 2.25923210e-01
7.82708526e-01 3.66236597e-01 4.57753241e-02 1.55424729e-01
6.04600847e-01 2.43734166e-01 3.28600854e-01 -1.81896850e-01
-1.72973856e-01 -2.86045045e-01 3.42356682e-01 9.41044629e-01
-1.61407024e-01 -3.58134747e-01 -9.22948599e-01 3.82083684e-01
-1.92427862e+00 -6.71437860e-01 -2.44488344e-01 1.93727911e+00
6.85986459e-01 6.32202864e-01 -4.16606516e-02 9.52224657e-02
6.07169211e-01 3.31862330e-01 -3.21115673e-01 -5.82059026e-01
2.16434807e-01 1.62656292e-01 9.59856868e-01 8.06253672e-01
-1.24977553e+00 9.42392647e-01 7.10136986e+00 8.13155890e-01
-8.10593367e-01 1.90733328e-01 6.65343761e-01 3.17786455e-01
-2.28055775e-01 3.06711495e-01 -1.04156351e+00 2.18808964e-05
1.38548243e+00 3.16139683e-02 5.04908562e-01 1.02864182e+00
-1.72351569e-01 1.75826639e-01 -1.22199965e+00 3.52854908e-01
-4.63511609e-02 -1.55731940e+00 1.01744704e-01 -8.55272487e-02
7.38618553e-01 7.15261281e-01 -2.90518671e-01 2.80007511e-01
9.72671688e-01 -7.70236850e-01 2.19599575e-01 3.71268392e-01
3.08305860e-01 -5.12757540e-01 7.02123642e-01 3.39872479e-01
-1.09401441e+00 -2.18189284e-02 -3.95758778e-01 -1.22384794e-01
-1.83840230e-01 7.94515073e-01 -1.06430972e+00 3.95300478e-01
4.75956261e-01 9.35337603e-01 -8.44184458e-01 1.10702062e+00
-2.92994112e-01 1.14254451e+00 -6.50929511e-01 -6.00326002e-01
3.84970009e-01 1.71146497e-01 4.51617986e-01 1.46644747e+00
-2.60032676e-02 -2.94661552e-01 1.88039303e-01 4.69822407e-01
-4.29295689e-01 1.92800939e-01 -3.92280340e-01 -1.14187635e-01
2.60165632e-01 1.44517899e+00 -9.16508675e-01 -5.05013525e-01
-3.95196706e-01 9.04091656e-01 1.10997975e+00 4.40957785e-01
-6.68385327e-01 -7.03441024e-01 2.52661437e-01 5.99296615e-02
4.10668075e-01 -1.62108302e-01 6.32796884e-02 -8.88046503e-01
-1.82748497e-01 -2.14388564e-01 5.46172976e-01 -4.45414901e-01
-1.33712351e+00 7.79366374e-01 -9.01512504e-02 -7.75211215e-01
-2.54866540e-01 -8.03918421e-01 -6.74191535e-01 3.96617532e-01
-1.69235766e+00 -1.23455954e+00 -1.59350753e-01 2.62682557e-01
-3.40739265e-03 1.45156279e-01 8.48869860e-01 1.42257929e-01
-5.72585762e-01 8.55647504e-01 3.40101451e-01 4.17512506e-01
4.66933966e-01 -1.41567779e+00 4.60669070e-01 7.04795480e-01
1.86658025e-01 4.61573005e-01 5.43179512e-01 -5.52669466e-01
-1.23921835e+00 -1.38194084e+00 1.23237252e+00 -3.22351933e-01
1.02934754e+00 -4.86195236e-01 -8.74142468e-01 9.15091395e-01
-1.50042623e-02 1.19321600e-01 5.89341402e-01 5.03006935e-01
-5.08690119e-01 -2.95558393e-01 -8.55535626e-01 6.05629504e-01
1.10257840e+00 -3.76819819e-01 3.28781232e-02 6.90507174e-01
7.89546967e-01 -2.07766518e-01 -9.46065485e-01 3.44439857e-02
5.86441040e-01 -9.00191069e-01 6.67894781e-01 -6.62557900e-01
2.47955278e-01 1.54439155e-02 2.78506458e-01 -1.29848731e+00
-5.37308574e-01 -6.50238693e-01 -3.56530488e-01 9.70081687e-01
1.07874250e+00 -1.08276176e+00 1.29869282e+00 4.83636320e-01
-1.02984263e-02 -1.02947736e+00 -4.56415683e-01 -5.38489521e-01
2.15197615e-02 -5.41857064e-01 3.46986115e-01 6.70897901e-01
2.74757594e-01 2.85331815e-01 -1.67894319e-01 2.86597699e-01
8.13511908e-01 -1.00163566e-02 6.25535548e-01 -1.54249156e+00
-5.55025041e-01 -5.04995823e-01 -6.02948010e-01 -1.29130197e+00
2.97353119e-01 -1.45298564e+00 -5.69264442e-02 -1.80504489e+00
3.62812757e-01 -7.33986497e-01 -6.16285264e-01 8.31755519e-01
1.67060271e-01 5.79437315e-01 -1.24435261e-01 1.51142851e-01
-9.98606861e-01 6.23397417e-02 7.65806794e-01 -2.03506708e-01
-1.89942494e-01 1.79568693e-01 -5.62137902e-01 8.70878994e-01
1.06011951e+00 -6.84218168e-01 -1.96231157e-01 -2.75860220e-01
7.39403725e-01 -1.93177283e-01 2.72892118e-01 -1.09801328e+00
6.49941027e-01 3.25024098e-01 3.57500076e-01 -5.14931202e-01
1.46036685e-01 -4.77409273e-01 -1.47735804e-01 4.08968449e-01
-8.95415187e-01 -2.95183420e-01 -2.46268839e-01 9.95628238e-01
1.88654751e-01 -3.09612960e-01 4.68281180e-01 7.97072519e-03
-4.49252307e-01 6.13390863e-01 -5.53031206e-01 -3.00199747e-01
4.91481066e-01 2.87709013e-02 -4.51404542e-01 -5.98240316e-01
-8.51961970e-01 3.27025086e-01 3.40980053e-01 -4.28216998e-03
5.07656932e-01 -1.07556558e+00 -7.36872911e-01 2.51723966e-03
-1.33259711e-03 -3.21038663e-01 -6.89879656e-02 1.02625895e+00
-8.28499019e-01 3.16742837e-01 2.19884485e-01 -2.97388196e-01
-1.19739902e+00 2.57507980e-01 3.03621292e-01 -7.07735896e-01
-7.83458233e-01 8.34177077e-01 -4.65462416e-01 -7.26796627e-01
5.04130006e-01 -2.46272624e-01 -3.66237670e-01 -3.15842688e-01
5.26624918e-01 4.44822490e-01 2.37376206e-02 -2.96713859e-01
-1.67789280e-01 8.20080116e-02 -4.33590204e-01 1.27067119e-01
1.45649469e+00 -3.77601348e-02 -4.58571225e-01 4.46974099e-01
1.33302200e+00 -1.89663425e-01 -1.13032186e+00 -4.27977502e-01
2.35931817e-02 -3.98355303e-03 2.45394424e-01 -5.51402271e-01
-1.00796854e+00 5.67248464e-01 1.87536433e-01 6.32710755e-01
7.01297343e-01 1.81084722e-01 7.54722476e-01 9.82939363e-01
2.46630982e-01 -9.06574130e-01 -4.27449793e-01 7.12101221e-01
2.35771313e-01 -1.02929652e+00 2.22485900e-01 -6.23591006e-01
-5.28374780e-03 1.32904148e+00 3.28529894e-01 -3.37956578e-01
1.15180528e+00 1.96605340e-01 -1.38871223e-01 -1.85045108e-01
-1.10230315e+00 -6.24907464e-02 1.02697477e-01 5.95763266e-01
4.28834707e-01 7.78793022e-02 -3.02584559e-01 3.32180768e-01
-1.53071180e-01 -6.53252378e-02 6.64101720e-01 7.43476510e-01
-7.10824311e-01 -1.05132771e+00 2.98850894e-01 9.82525706e-01
-4.23437089e-01 -4.06462133e-01 -5.17428279e-01 5.15573442e-01
-4.01198775e-01 8.38207901e-01 -1.23607099e-01 -5.95944226e-01
-4.51040924e-01 5.09630749e-03 2.96426117e-01 -3.80345106e-01
-5.39853096e-01 -6.34758532e-01 5.79327822e-01 -6.99263871e-01
-4.23741788e-01 -2.94254214e-01 -1.18863988e+00 -6.27204776e-01
-5.97854733e-01 2.62853622e-01 7.25812495e-01 7.88693547e-01
3.95548344e-01 2.85639495e-01 5.30887604e-01 -1.00776148e+00
-6.18579984e-01 -8.24319184e-01 -7.87142098e-01 3.49658616e-02
3.15457284e-01 -1.56189010e-01 -5.71936429e-01 -4.16898549e-01] | [6.98975133895874, 6.189652919769287] |
1be5b5f2-06af-49fb-88d8-c9c4893f089f | a-call-for-standardization-and-validation-of | 2306.00539 | null | https://arxiv.org/abs/2306.00539v1 | https://arxiv.org/pdf/2306.00539v1.pdf | A Call for Standardization and Validation of Text Style Transfer Evaluation | Text Style Transfer (TST) evaluation is, in practice, inconsistent. Therefore, we conduct a meta-analysis on human and automated TST evaluation and experimentation that thoroughly examines existing literature in the field. The meta-analysis reveals a substantial standardization gap in human and automated evaluation. In addition, we also find a validation gap: only few automated metrics have been validated using human experiments. To this end, we thoroughly scrutinize both the standardization and validation gap and reveal the resulting pitfalls. This work also paves the way to close the standardization and validation gap in TST evaluation by calling out requirements to be met by future research. | ['Sophie Fellenz', 'Marius Kloft', 'Mayank Nagda', 'Phil Ostheimer'] | 2023-06-01 | null | null | null | null | ['style-transfer', 'text-style-transfoer'] | ['computer-vision', 'natural-language-processing'] | [ 3.28656465e-01 8.36883187e-02 -2.42573872e-01 -2.99865305e-01
-8.27830017e-01 -8.83351743e-01 5.87733924e-01 2.60968208e-01
-3.67290258e-01 5.31243861e-01 2.32467324e-01 -5.61479568e-01
-1.41338110e-01 -1.92912877e-01 -3.48710716e-01 7.52886459e-02
4.51337487e-01 3.96364152e-01 2.46001840e-01 -2.70832717e-01
8.05499136e-01 1.94639862e-02 -1.53343165e+00 4.49470073e-01
1.07607043e+00 4.73499358e-01 -7.39440024e-02 4.61825401e-01
-3.41412038e-01 6.34618402e-01 -1.01990080e+00 -8.52477789e-01
-1.24016106e-02 -4.48819131e-01 -1.18598354e+00 -3.29715345e-04
6.43995404e-01 -2.62384683e-01 5.85324429e-02 8.41998518e-01
8.28989148e-01 -1.31651357e-01 5.64497650e-01 -1.47856152e+00
-9.24363732e-01 5.53895235e-01 -2.09359691e-01 1.50718659e-01
6.06858075e-01 2.76177227e-01 1.08711731e+00 -9.47528899e-01
7.86359906e-01 1.21049976e+00 7.63309121e-01 4.93252635e-01
-8.23000968e-01 -6.86290383e-01 -1.64946109e-01 9.38770249e-02
-1.10346043e+00 -3.46923590e-01 3.45135510e-01 -6.77351356e-01
1.10325921e+00 3.84639680e-01 7.74727643e-01 1.36542106e+00
9.38053504e-02 9.44381475e-01 1.48289692e+00 -8.90673816e-01
1.31125867e-01 6.10003233e-01 3.53341967e-01 2.63323218e-01
6.82673872e-01 -7.08899498e-02 -5.44350863e-01 1.79912612e-01
3.81722838e-01 -6.61647499e-01 -4.29774821e-02 6.69932216e-02
-1.18793523e+00 5.92711806e-01 -3.10114294e-01 7.36916661e-01
3.79307335e-03 -2.19928756e-01 8.89459312e-01 6.23192012e-01
5.72380304e-01 7.81154454e-01 -3.04277986e-01 -6.36172175e-01
-1.30558217e+00 7.34786019e-02 8.72467935e-01 1.17612302e+00
4.21999007e-01 -1.15069143e-01 -7.28714585e-01 7.30979443e-01
3.45917165e-01 4.82140601e-01 5.06972313e-01 -8.36285472e-01
6.45193398e-01 7.36711442e-01 2.54034311e-01 -8.70548129e-01
-8.73747244e-02 -3.38290304e-01 -2.03860492e-01 2.12119102e-01
2.89608240e-01 -1.65453479e-01 -5.32827079e-01 1.37552428e+00
-6.52110204e-02 -4.91510212e-01 -2.98718750e-01 7.18484700e-01
8.24145794e-01 2.89390922e-01 2.41790175e-01 -9.76656750e-02
1.27981114e+00 -9.52821016e-01 -1.07225919e+00 -1.45162776e-01
1.11551642e+00 -1.27764750e+00 1.77862418e+00 3.32635939e-01
-1.20569754e+00 -5.57680488e-01 -1.08258533e+00 -4.61571254e-02
-5.05003452e-01 3.32162738e-01 3.89297962e-01 1.25488317e+00
-1.06672001e+00 6.22548759e-01 -5.39095342e-01 -9.29800332e-01
2.20129386e-01 4.47001122e-02 -1.36628583e-01 2.08729148e-01
-1.27692175e+00 1.38975692e+00 2.45634183e-01 -1.52740851e-02
-4.58069384e-01 -6.26099586e-01 -5.74862003e-01 -3.36187094e-01
2.73099333e-01 -6.51035607e-01 1.71797121e+00 -1.10075009e+00
-1.55310524e+00 9.52406883e-01 -9.27655473e-02 -7.35792518e-02
6.34175777e-01 -4.18337971e-01 -4.16450560e-01 -1.78728595e-01
2.53359765e-01 3.28346729e-01 3.34805101e-01 -1.07774508e+00
-7.14121819e-01 -3.44808280e-01 -1.64428204e-01 3.41071904e-01
-6.88068688e-01 4.90537792e-01 -3.23822230e-01 -1.06265783e+00
-3.89472425e-01 -9.01810884e-01 3.94206136e-01 -4.21829760e-01
-2.37921402e-01 -4.10747141e-01 8.02513719e-01 -6.53052509e-01
1.92582953e+00 -2.03067708e+00 -1.26761660e-01 1.03243448e-01
3.95533919e-01 4.05721486e-01 -1.01768270e-01 8.58903885e-01
2.60945171e-01 7.21386850e-01 3.25748995e-02 -5.31624496e-01
3.88410121e-01 -2.73673479e-02 -2.10293844e-01 1.63984999e-01
-2.35038072e-01 1.08263409e+00 -8.32731247e-01 -6.85100913e-01
1.09137587e-01 3.01786005e-01 -1.35033354e-01 5.97450845e-02
-1.60689354e-02 1.72122359e-01 -4.30015683e-01 7.01969028e-01
3.17213416e-01 -1.63475558e-01 1.27768114e-01 -9.10884421e-03
-5.90816200e-01 5.03749192e-01 -8.13027680e-01 1.54416561e+00
-3.37932706e-01 1.27308309e+00 -3.97893161e-01 -3.28227401e-01
7.93992162e-01 5.82139432e-01 2.69035339e-01 -7.24095404e-01
3.38757128e-01 5.84521770e-01 -7.09564537e-02 -6.79841161e-01
8.52391362e-01 -1.36601552e-01 5.46850264e-02 8.06245089e-01
8.67490191e-03 -2.12282047e-01 3.26756090e-01 6.77515194e-02
7.71226883e-01 1.36087641e-01 1.29732445e-01 -1.72291830e-01
3.72920603e-01 2.12889999e-01 1.55138776e-01 7.64320076e-01
-4.95165169e-01 4.25692618e-01 3.31453353e-01 -1.57539487e-01
-1.26860785e+00 -6.14181340e-01 7.85774142e-02 1.09249425e+00
-4.75118123e-02 -7.86356986e-01 -1.36019528e+00 -8.77754986e-01
-2.45969877e-01 9.59093809e-01 -8.84877920e-01 1.61733516e-02
-2.74102837e-01 -4.26952779e-01 9.46047544e-01 5.83123922e-01
5.05775392e-01 -1.16421771e+00 -8.81039977e-01 -9.33745690e-03
-4.20023710e-01 -1.21157622e+00 -4.27263260e-01 -4.14251953e-01
-9.68830526e-01 -8.61148298e-01 -5.99349380e-01 -6.35198355e-01
1.68804854e-01 2.28436068e-01 1.18085158e+00 2.07598686e-01
8.16751346e-02 6.70061171e-01 -7.09004998e-01 -7.05003202e-01
-5.43727696e-01 2.95943856e-01 -1.44278333e-01 -5.84102452e-01
7.94748843e-01 -5.51367179e-02 -4.03740466e-01 7.56779909e-01
-9.90707695e-01 -7.72878900e-02 7.42290497e-01 6.06407702e-01
1.24234147e-01 2.10914873e-02 5.50769866e-01 -7.25739419e-01
1.20031095e+00 -1.72021493e-01 -4.04497772e-01 6.89679444e-01
-1.21887875e+00 -1.34390280e-01 -1.48183759e-03 -3.24342221e-01
-1.10688424e+00 -7.98702836e-01 1.37471780e-01 -5.88743538e-02
-1.60920009e-01 6.54201806e-01 1.36495680e-01 -2.09701270e-01
7.89798021e-01 -1.49577126e-01 2.24552513e-03 -4.32886571e-01
-2.46041398e-02 8.76404524e-01 -8.65897685e-02 -6.44915402e-01
7.00339615e-01 -3.48613784e-02 -5.89724422e-01 -7.02794313e-01
-7.53958762e-01 -2.89844960e-01 -5.98475456e-01 -6.28760397e-01
7.08148122e-01 -4.44576323e-01 -3.57276022e-01 2.29205132e-01
-9.28584397e-01 -6.12912059e-01 -2.74612159e-01 5.76972663e-01
-3.90921712e-01 5.29920340e-01 -6.01575971e-01 -7.10052252e-01
-6.05477035e-01 -1.29858744e+00 9.64999974e-01 2.31504720e-02
-1.09141695e+00 -1.17990184e+00 3.69411260e-01 6.85790539e-01
7.17834473e-01 -4.71029803e-03 5.91787457e-01 -7.06633449e-01
-1.69418622e-02 -2.31047139e-01 -1.68583155e-01 3.03530127e-01
6.31211773e-02 4.03558314e-01 -8.77726614e-01 -3.06317180e-01
1.81547869e-02 -3.35811049e-01 7.60948062e-02 2.06494302e-01
7.00137258e-01 -1.42109290e-01 -5.67444637e-02 -6.78807264e-03
1.00653791e+00 1.57070607e-01 7.21063673e-01 9.25997853e-01
4.82374400e-01 1.06320965e+00 8.58994901e-01 2.21102700e-01
3.63001376e-01 6.87284291e-01 -3.92170757e-01 1.31540492e-01
-3.45174968e-01 -3.94250751e-01 4.35459405e-01 1.10556960e+00
-1.38182908e-01 -5.50604284e-01 -1.15656674e+00 3.08338135e-01
-1.62614894e+00 -7.40009844e-01 -1.96770281e-01 2.35594010e+00
6.43864572e-01 3.76688510e-01 3.50549340e-01 4.66605008e-01
6.85176551e-01 -2.86985368e-01 7.42432550e-02 -5.24215698e-01
-1.55792162e-01 1.44967318e-01 2.56057322e-01 2.98273832e-01
-5.83186030e-01 1.05846334e+00 7.71870852e+00 8.11610281e-01
-1.01881766e+00 1.30401716e-01 2.81172186e-01 1.38372764e-01
-4.94785011e-01 1.58037126e-01 -4.52049583e-01 3.18441063e-01
9.14742887e-01 -4.93082255e-01 1.69008806e-01 3.40394497e-01
3.48178983e-01 1.82764336e-01 -1.11385787e+00 7.33038425e-01
1.23644553e-01 -7.54312098e-01 8.52884352e-02 -1.39740622e-02
5.03654897e-01 -1.44135654e-01 3.85165125e-01 4.93890047e-01
-8.04458112e-02 -8.81241262e-01 1.06209099e+00 4.42568846e-02
7.94175684e-01 -2.23056301e-01 8.43261242e-01 -8.31628144e-02
-7.20805228e-01 9.09297690e-02 1.11850575e-01 -2.17541829e-01
4.75651436e-02 8.37325901e-02 -9.41572666e-01 5.82378924e-01
7.12565899e-01 6.62054598e-01 -9.13972735e-01 9.64196384e-01
-3.80383879e-01 9.89467978e-01 1.58367142e-01 -3.71530354e-01
2.17868149e-01 -2.02457458e-01 5.07785499e-01 1.63254941e+00
3.29103410e-01 -2.08734885e-01 -2.45932087e-01 6.21334195e-01
1.07474126e-01 6.36502743e-01 -7.30313063e-01 -3.51156801e-01
6.41481280e-01 8.85287166e-01 -9.25270259e-01 -3.80125225e-01
-4.79050606e-01 9.64905918e-01 1.46187178e-03 4.81286883e-01
-6.53429985e-01 -4.48470950e-01 2.36628920e-01 1.07325576e-01
1.35304211e-02 -1.38828769e-01 -9.61757362e-01 -1.05750573e+00
1.84815884e-01 -1.14052045e+00 4.12433654e-01 -8.23155820e-01
-1.21034932e+00 5.09587348e-01 3.21848750e-01 -1.47873962e+00
-3.86750810e-02 -5.02083123e-01 -3.54145110e-01 6.25993848e-01
-1.00799572e+00 -9.28229630e-01 -3.43671888e-01 9.94733199e-02
6.72147512e-01 -3.96947265e-02 5.66790164e-01 4.04545158e-01
-7.19160795e-01 1.02401853e+00 -1.66307181e-01 -1.47057146e-01
1.02354574e+00 -1.11506784e+00 5.38212538e-01 6.04016960e-01
-3.19504052e-01 9.12970960e-01 7.65682936e-01 -9.53799307e-01
-1.23657167e+00 -7.08427608e-01 1.19873559e+00 -9.13149595e-01
6.57620966e-01 -1.09376423e-01 -9.43029821e-01 7.29153693e-01
7.47823954e-01 -8.72998059e-01 1.07079339e+00 1.71527922e-01
-4.40988123e-01 2.57144302e-01 -1.13085699e+00 8.94151151e-01
8.00016999e-01 -7.05341637e-01 -8.63574207e-01 -5.44837005e-02
5.09039760e-01 -2.06118330e-01 -1.03726935e+00 4.40475285e-01
9.32990134e-01 -7.76368916e-01 5.51104248e-01 -2.89550185e-01
3.46491724e-01 -5.45755178e-02 -4.99477684e-02 -1.22649300e+00
-1.13163345e-01 -8.39150429e-01 3.96284729e-01 1.46011019e+00
6.58494532e-01 -6.17304206e-01 5.95980525e-01 8.41550350e-01
-3.37742537e-01 -5.01625061e-01 -5.77343702e-01 -9.19741154e-01
3.73876423e-01 -6.46683455e-01 5.19186676e-01 1.23899984e+00
5.00210345e-01 5.72521210e-01 -1.83229879e-01 -5.00043988e-01
4.95766908e-01 -4.37980205e-01 8.58438432e-01 -1.10083222e+00
2.72812843e-01 -9.04578269e-01 -1.64772406e-01 -5.60769081e-01
-7.60787874e-02 -6.98486567e-01 -1.95373297e-01 -1.60788178e+00
2.77742654e-01 -7.56740198e-02 -5.17815202e-02 3.73003691e-01
-5.05509675e-02 1.52023584e-01 2.29599714e-01 2.92973191e-01
-7.32163072e-01 4.53984439e-01 1.31822908e+00 1.21090874e-01
-2.95427144e-01 -3.90941054e-01 -7.79205084e-01 4.01888520e-01
1.28823674e+00 -4.12281662e-01 -5.05497515e-01 -8.04101884e-01
4.46302474e-01 -3.89142871e-01 4.15627053e-03 -9.27183628e-01
2.17545360e-01 -1.08419627e-01 -1.08113550e-01 -5.18770218e-01
-2.43705362e-01 -7.16112256e-01 -1.12755381e-01 1.67412996e-01
-5.64721406e-01 5.67574203e-01 4.49765652e-01 -9.66744572e-02
-2.13284612e-01 -3.45336109e-01 2.90762305e-01 1.21580012e-01
-3.54866952e-01 -1.62957996e-01 -7.38934875e-01 3.30687910e-01
8.31368983e-01 -7.29059637e-01 -3.43896478e-01 -3.81798893e-01
-6.07768744e-02 2.35702887e-01 6.61989987e-01 6.09186709e-01
4.88062233e-01 -1.26847076e+00 -6.48134470e-01 -1.02621213e-01
4.78233784e-01 -5.58139265e-01 -1.23953074e-03 1.03909588e+00
-4.59293038e-01 6.70035660e-01 -2.85775602e-01 -3.06846410e-01
-1.16000104e+00 3.39701533e-01 3.84442583e-02 -6.85653836e-02
-3.49709451e-01 1.69996694e-01 -3.53327841e-01 -2.86613733e-01
4.14962411e-01 -4.46347505e-01 -3.32876742e-01 6.90578222e-02
3.91998887e-01 9.40818727e-01 2.85366267e-01 -5.42751908e-01
-7.51090348e-02 4.64453578e-01 1.72222517e-02 -7.68639326e-01
7.34563768e-01 -4.48753297e-01 -1.27300218e-01 8.07011247e-01
1.05392027e+00 1.13799445e-01 -3.57400835e-01 6.83552818e-03
5.13918340e-01 -5.48576176e-01 -7.77755380e-02 -9.62909877e-01
-4.82288212e-01 6.97372913e-01 5.67921340e-01 4.38687086e-01
1.00204408e+00 -2.47586370e-01 6.38877690e-01 3.30464333e-01
9.71903950e-02 -1.59599090e+00 1.63137108e-01 6.07139766e-01
8.79049838e-01 -1.07307148e+00 -2.13354781e-01 -3.80569011e-01
-9.15285707e-01 1.11503053e+00 7.58998692e-01 5.17375588e-01
3.48515451e-01 1.22030638e-01 1.71949759e-01 -4.69722003e-01
-6.79690003e-01 1.87493816e-01 4.49369788e-01 3.79202962e-01
1.26039386e+00 3.32284756e-02 -9.76937950e-01 4.32767242e-01
-5.09150684e-01 4.59006727e-01 2.81307548e-01 1.26260734e+00
-2.16004893e-01 -1.35600781e+00 -5.30600369e-01 2.30671853e-01
-5.22985160e-01 -4.78968807e-02 -9.95101333e-01 1.12244117e+00
-2.65847743e-01 1.34325397e+00 -3.71679097e-01 -7.78135359e-01
7.11521506e-01 1.32600099e-01 5.23475409e-01 -5.12360573e-01
-1.07599199e+00 -5.31120636e-02 3.13416779e-01 -2.98940718e-01
-3.84018362e-01 -6.78001940e-01 -6.42586410e-01 -3.53549808e-01
-4.02898192e-01 3.08274329e-01 7.47638762e-01 9.60718691e-01
2.89362729e-01 3.76098275e-01 2.93996871e-01 -1.92838922e-01
-3.89775723e-01 -1.40525019e+00 -1.31252706e-01 4.54681277e-01
-1.27460420e-01 -6.11372352e-01 -3.87639731e-01 -4.78706323e-03] | [11.447616577148438, 9.58362865447998] |
b1403c41-2721-4208-9431-3ba405409ec4 | semi-supervised-multimodal-representation | 2306.15711 | null | https://arxiv.org/abs/2306.15711v1 | https://arxiv.org/pdf/2306.15711v1.pdf | Semi-supervised Multimodal Representation Learning through a Global Workspace | Recent deep learning models can efficiently combine inputs from different modalities (e.g., images and text) and learn to align their latent representations, or to translate signals from one domain to another (as in image captioning, or text-to-image generation). However, current approaches mainly rely on brute-force supervised training over large multimodal datasets. In contrast, humans (and other animals) can learn useful multimodal representations from only sparse experience with matched cross-modal data. Here we evaluate the capabilities of a neural network architecture inspired by the cognitive notion of a "Global Workspace": a shared representation for two (or more) input modalities. Each modality is processed by a specialized system (pretrained on unimodal data, and subsequently frozen). The corresponding latent representations are then encoded to and decoded from a single shared workspace. Importantly, this architecture is amenable to self-supervised training via cycle-consistency: encoding-decoding sequences should approximate the identity function. For various pairings of vision-language modalities and across two datasets of varying complexity, we show that such an architecture can be trained to align and translate between two modalities with very little need for matched data (from 4 to 7 times less than a fully supervised approach). The global workspace representation can be used advantageously for downstream classification tasks and for robust transfer learning. Ablation studies reveal that both the shared workspace and the self-supervised cycle-consistency training are critical to the system's performance. | ['Rufin VanRullen', 'Léopold Maytié', 'Benjamin Devillers'] | 2023-06-27 | null | null | null | null | ['image-generation', 'image-captioning', 'transfer-learning'] | ['computer-vision', 'computer-vision', 'miscellaneous'] | [ 8.88910651e-01 2.79961467e-01 -9.87534150e-02 -4.22702223e-01
-9.01577532e-01 -9.26948667e-01 1.17305183e+00 -1.03062704e-01
-3.95338714e-01 5.98525941e-01 2.36163273e-01 -1.48336872e-01
2.75474042e-02 -3.67884696e-01 -1.06551874e+00 -6.68023467e-01
7.75239095e-02 3.50922614e-01 -1.46235749e-01 -9.62139741e-02
3.55533287e-02 2.67460912e-01 -1.62464619e+00 7.69631863e-01
5.61822236e-01 8.78811061e-01 6.29278541e-01 6.54103875e-01
-7.50352889e-02 4.46501762e-01 -9.97568071e-02 -5.19214198e-02
2.71220505e-01 -7.53102899e-01 -8.91480029e-01 -8.30666944e-02
6.26911283e-01 -2.33799294e-01 -4.25101757e-01 7.20728517e-01
4.23770249e-01 -4.11016569e-02 8.96341085e-01 -1.20185995e+00
-8.29856217e-01 3.69971573e-01 -2.30798513e-01 -1.54259965e-01
3.32926720e-01 3.52167279e-01 8.33599269e-01 -7.35877872e-01
9.22908604e-01 9.63692665e-01 3.85637760e-01 9.17789340e-01
-1.79856133e+00 -2.82235831e-01 4.59223613e-02 -3.14409345e-01
-1.01304626e+00 -9.00547326e-01 3.62011433e-01 -6.39106750e-01
9.93661940e-01 -1.10263273e-03 5.32722294e-01 1.41004074e+00
2.73070157e-01 5.18642306e-01 1.14867401e+00 -3.73525470e-01
1.07630141e-01 3.75184137e-03 -3.98027986e-01 7.12583780e-01
-2.23526545e-02 1.82550266e-01 -1.13221860e+00 6.43605217e-02
9.24516022e-01 -1.08930200e-01 -3.11447710e-01 -5.19601047e-01
-1.75040770e+00 4.86674339e-01 5.58591306e-01 3.73359442e-01
-2.06090376e-01 4.97289002e-01 3.11632693e-01 5.93750358e-01
-1.26158386e-01 6.98224545e-01 -3.52188826e-01 9.97046232e-02
-9.52039540e-01 -8.21774155e-02 5.61299384e-01 1.02674973e+00
8.40369344e-01 1.11290719e-02 -6.67311177e-02 6.63225710e-01
5.12724638e-01 5.71335912e-01 8.24851036e-01 -1.06075978e+00
5.15267432e-01 3.94412637e-01 -2.53844112e-02 -6.35750532e-01
-3.47375214e-01 -1.81554839e-01 -7.86289513e-01 2.39175484e-01
6.21056437e-01 -7.07915658e-03 -9.78562415e-01 2.45164824e+00
-2.98697501e-01 -2.39707425e-01 3.07864100e-01 8.64951253e-01
5.49204350e-01 6.47505462e-01 2.34163597e-01 -6.42005727e-02
1.20090008e+00 -6.04642272e-01 -2.57920325e-01 -7.57012248e-01
6.13814771e-01 -6.16257071e-01 9.96816278e-01 1.71205550e-01
-1.34378588e+00 -4.89989936e-01 -1.15020251e+00 -3.25113982e-01
-4.01550591e-01 1.79046348e-01 7.24228919e-01 2.04214066e-01
-1.32655978e+00 5.60908735e-01 -7.79603064e-01 -7.16416001e-01
4.09859180e-01 5.14422655e-01 -9.80133772e-01 -2.23407410e-02
-9.72900808e-01 1.10780954e+00 3.78214002e-01 8.28691795e-02
-1.28504014e+00 -4.69192654e-01 -1.02485907e+00 -3.59334052e-02
-4.21714574e-01 -1.25158429e+00 9.71385658e-01 -1.54971147e+00
-1.37573457e+00 1.27174902e+00 -2.51180559e-01 -2.89698482e-01
1.46746859e-01 3.47696617e-02 3.16252224e-02 4.22752261e-01
1.15169369e-01 1.27126896e+00 1.14953601e+00 -1.42236531e+00
-6.76499605e-02 -3.14993382e-01 -1.59598812e-01 3.33538920e-01
-3.48437130e-01 -9.92077738e-02 -3.53284806e-01 -4.54949915e-01
4.74813551e-01 -8.97803128e-01 1.16000816e-01 4.06416416e-01
-5.10766879e-02 4.20539945e-01 4.32167977e-01 -6.02227271e-01
5.04381478e-01 -2.18294597e+00 8.14349651e-01 1.52395964e-01
1.06951870e-01 -1.49187326e-01 -6.43662155e-01 6.25360131e-01
-3.13556403e-01 -3.41141294e-03 -4.36946362e-01 -4.51912910e-01
-8.27098116e-02 2.36534119e-01 -4.96151179e-01 5.56978583e-01
4.62431937e-01 1.20552850e+00 -8.97661805e-01 -2.14308307e-01
-9.08996463e-02 3.90840828e-01 -3.87243897e-01 4.71900672e-01
-2.15721563e-01 7.23546326e-01 9.64527503e-02 4.76467192e-01
1.67590052e-01 -3.48506898e-01 5.33054233e-01 -5.18437564e-01
1.26488775e-01 2.83377498e-01 -6.34470642e-01 2.42264128e+00
-7.87656724e-01 8.95165920e-01 1.71915859e-01 -1.10522008e+00
7.30455637e-01 4.84836876e-01 2.16291219e-01 -8.59423399e-01
2.11514860e-01 3.50839466e-01 -2.16524862e-02 -3.91066641e-01
9.61555541e-02 -4.18043673e-01 -1.38487041e-01 8.90567899e-01
7.79408813e-01 -1.67329282e-01 -5.21002337e-02 3.08530301e-01
9.84167337e-01 3.22884798e-01 -5.22684166e-03 -1.60724133e-01
7.39450902e-02 -3.06133577e-03 5.98380379e-02 7.60828078e-01
4.45059352e-02 9.85095680e-01 5.04403889e-01 -5.35234883e-02
-1.30397606e+00 -1.45004535e+00 -1.98961142e-02 1.35855520e+00
1.11151464e-01 -1.35030359e-01 -5.81355751e-01 -2.57794708e-01
-1.44267038e-01 5.24606943e-01 -6.81445003e-01 -6.09868050e-01
-2.90623963e-01 -3.19766283e-01 8.95785153e-01 5.43898523e-01
1.99261621e-01 -1.04466760e+00 -6.53741300e-01 5.26419990e-02
-9.73631516e-02 -1.00099695e+00 -2.88318187e-01 4.83079046e-01
-9.85304952e-01 -7.18705177e-01 -8.72271359e-01 -1.06365860e+00
1.10891247e+00 2.17426717e-01 1.01973009e+00 5.09401923e-03
-1.39319062e-01 7.91209459e-01 1.53502561e-02 1.66780367e-01
-4.99056727e-01 5.87308519e-02 2.16029078e-01 6.88737258e-02
-2.65246719e-01 -9.05877590e-01 -5.13169289e-01 2.25856468e-01
-1.17648637e+00 6.22237027e-01 8.59457910e-01 1.22717083e+00
3.23760450e-01 -8.37695777e-01 6.46916986e-01 -5.55028975e-01
6.09280229e-01 -5.18270135e-01 -3.10033143e-01 3.72361273e-01
-1.06431916e-01 5.35749674e-01 3.79937261e-01 -6.42339110e-01
-1.04479182e+00 4.51317906e-01 2.99017996e-01 -3.75388235e-01
-2.42663458e-01 6.34607494e-01 -1.32338136e-01 -8.41313973e-02
9.15314674e-01 4.83705670e-01 4.13933158e-01 -9.71741453e-02
8.83520186e-01 3.43854904e-01 8.43255281e-01 -8.21971178e-01
8.37199211e-01 5.13009250e-01 -2.20259819e-02 -5.06587684e-01
-4.26815331e-01 3.04346699e-02 -9.34543073e-01 -7.99497291e-02
1.02262425e+00 -9.52172041e-01 -6.02971971e-01 3.79087448e-01
-1.31718111e+00 -5.87290525e-01 -1.84022263e-01 4.45442617e-01
-1.02913415e+00 -3.76694277e-02 -6.22007728e-01 -4.36948836e-01
-5.60046136e-02 -1.00814390e+00 1.21473277e+00 1.41884819e-01
-4.25798774e-01 -9.39329803e-01 9.30564925e-02 2.49069214e-01
4.85693455e-01 1.11309074e-01 1.04196095e+00 -4.75633979e-01
-6.34157121e-01 -8.18914473e-02 -3.40024650e-01 1.57201633e-01
9.82577447e-03 -2.35691562e-01 -1.15077293e+00 -3.28253895e-01
-3.06152821e-01 -9.92537141e-01 1.03412127e+00 -2.94647901e-03
7.38089323e-01 -1.76051468e-01 -3.76828104e-01 6.87986195e-01
1.20970917e+00 -1.96548283e-01 5.96211374e-01 1.43025935e-01
4.00734991e-01 8.40121329e-01 -7.92634264e-02 -1.78448245e-01
2.61187702e-01 3.74983788e-01 2.59441286e-01 -4.90046889e-02
-1.52122423e-01 -5.41805923e-01 6.78099155e-01 6.75602376e-01
1.44087106e-01 8.20212066e-02 -9.96539354e-01 4.96326506e-01
-1.86819994e+00 -1.02367687e+00 4.70364958e-01 2.34011054e+00
1.10159183e+00 9.17989463e-02 -2.58781493e-01 -3.76381963e-01
3.83375674e-01 -1.05732642e-01 -5.46748281e-01 -3.60281020e-01
-3.25732976e-01 8.99128169e-02 2.17623189e-01 3.96637261e-01
-8.64105821e-01 9.17527080e-01 7.18226767e+00 3.10721457e-01
-1.27467310e+00 2.63181597e-01 4.53870356e-01 -2.83271044e-01
-5.63730836e-01 1.53073385e-01 -3.53917181e-01 1.46954834e-01
1.16248310e+00 -2.79194415e-02 8.48287761e-01 3.62463653e-01
-1.65193632e-01 -2.17136979e-01 -1.74132144e+00 1.07138205e+00
2.40965888e-01 -1.58461595e+00 1.34575129e-01 -6.88745677e-02
6.71037197e-01 8.39361250e-02 3.30364227e-01 1.72207981e-01
1.84748873e-01 -1.42035270e+00 9.28393364e-01 8.47256303e-01
1.10557365e+00 -9.10317004e-02 1.97329447e-01 3.75597060e-01
-9.04096663e-01 -8.21575243e-03 -1.75008610e-01 6.22534938e-02
2.28634015e-01 2.53723040e-02 -4.64108139e-01 1.88070893e-01
3.60895872e-01 6.08744502e-01 -5.62211812e-01 6.09491169e-01
-9.22825411e-02 2.11596504e-01 -2.58763760e-01 3.25331897e-01
3.11191343e-02 6.73710406e-02 3.22286248e-01 1.08246982e+00
3.80501002e-01 -1.67300820e-01 -2.76364774e-01 1.14540660e+00
-1.98427692e-01 -2.56315738e-01 -1.06992233e+00 -4.28457022e-01
2.35785663e-01 1.18120551e+00 -4.49786991e-01 -2.84410864e-01
-3.83260190e-01 1.31126142e+00 5.34687996e-01 6.76134348e-01
-5.01360297e-01 -5.47569133e-02 5.58293164e-01 -1.07119620e-01
7.40375742e-02 -5.53100407e-01 -5.29854774e-01 -1.31947911e+00
-1.20805189e-01 -7.51213193e-01 7.96442106e-03 -1.30078018e+00
-1.39825845e+00 6.50204539e-01 -3.74190733e-02 -1.30747879e+00
-5.72204411e-01 -7.51151204e-01 -4.82736021e-01 1.21448064e+00
-1.05973649e+00 -1.59423065e+00 -1.43346578e-01 7.25354970e-01
6.08321540e-02 -4.17357832e-01 1.07227528e+00 5.63064069e-02
-1.66224554e-01 5.64771414e-01 -5.39185256e-02 2.16279123e-02
9.11554098e-01 -9.55499172e-01 1.20385207e-01 6.25088811e-01
4.15696800e-01 1.08302331e+00 5.00487566e-01 -3.28430384e-01
-1.78419459e+00 -8.63626301e-01 6.26670122e-01 -6.21561408e-01
6.09857857e-01 -6.33567214e-01 -8.03456783e-01 8.63627374e-01
5.61129510e-01 4.17538598e-04 8.73984754e-01 1.05576113e-01
-9.40132141e-01 -4.10879627e-02 -8.41126978e-01 7.06848085e-01
1.09658170e+00 -1.21755946e+00 -7.09007263e-01 2.53874332e-01
5.19998550e-01 -2.72413552e-01 -7.20519066e-01 1.17925340e-02
8.02387118e-01 -7.43822873e-01 8.38580608e-01 -9.91348088e-01
6.57387137e-01 -4.46160406e-01 -4.05807197e-01 -1.31720006e+00
-1.17317431e-01 -6.31475806e-01 -1.64588597e-02 1.09573472e+00
6.55588388e-01 -4.53607082e-01 2.72366971e-01 6.94148302e-01
-1.31578848e-01 -4.24262524e-01 -1.08695340e+00 -6.47344649e-01
1.73275858e-01 -3.20999295e-01 6.02517650e-02 8.89938772e-01
3.15852195e-01 6.60402656e-01 -3.16118449e-01 2.91291643e-02
3.31342787e-01 2.67569989e-01 6.01927459e-01 -8.50962043e-01
-4.37848836e-01 -5.86230040e-01 -4.02714431e-01 -1.07853663e+00
2.89112240e-01 -1.45217621e+00 2.27552563e-01 -1.48243678e+00
3.21231812e-01 -2.19296634e-01 -1.11619495e-01 8.68544579e-01
3.35792899e-01 3.51005644e-01 3.22643280e-01 4.82680142e-01
-4.71599668e-01 5.97165585e-01 1.05148804e+00 -2.04273343e-01
-1.00965165e-01 -5.49017191e-01 -8.66053522e-01 4.36906397e-01
5.54348230e-01 -3.29594523e-01 -4.44838822e-01 -8.14541280e-01
5.38972974e-01 5.73335327e-02 6.57702267e-01 -8.74694526e-01
3.90700787e-01 -7.63440728e-02 6.71879828e-01 1.46408603e-01
4.55100298e-01 -7.81794250e-01 1.64559469e-01 2.33808771e-01
-8.06878805e-01 -7.70821422e-02 4.00510699e-01 5.73391914e-01
-1.20528862e-01 -1.51600346e-01 7.69282699e-01 -1.98459998e-01
-5.82968891e-01 -5.61555140e-02 -5.54783285e-01 -9.79040787e-02
8.40147674e-01 -2.13943899e-01 -5.66328943e-01 -4.56940174e-01
-9.96025622e-01 -3.29038972e-04 5.72168946e-01 3.75265896e-01
6.92474484e-01 -1.59858990e+00 -4.27134126e-01 4.74266678e-01
2.82264143e-01 -3.81412715e-01 5.01832888e-02 8.92548025e-01
-3.30792457e-01 3.91428947e-01 -7.30040550e-01 -8.88257205e-01
-6.47267580e-01 3.83733273e-01 5.24933279e-01 1.13917969e-01
-1.55196935e-01 6.40132248e-01 3.14406186e-01 -5.62616527e-01
-5.68089820e-02 -8.61105546e-02 2.07949728e-01 1.15843721e-01
4.19651538e-01 -3.81617874e-01 -1.13031581e-01 -6.87703669e-01
-2.17465535e-01 4.47336495e-01 1.08400643e-01 -7.55722046e-01
1.18937588e+00 -1.17659211e-01 -3.19231510e-01 6.74153566e-01
1.26980305e+00 -4.33713347e-01 -1.37484467e+00 -1.68558776e-01
-2.55356222e-01 -2.01924473e-01 -2.30329558e-01 -7.86707103e-01
-8.27694535e-01 1.19032252e+00 5.19248366e-01 -1.56439826e-01
1.16025913e+00 3.01271558e-01 4.26577330e-01 5.56080282e-01
3.98315907e-01 -8.24898005e-01 4.79202896e-01 4.87925768e-01
1.21700633e+00 -1.11997259e+00 -3.36431146e-01 2.13961750e-01
-8.09158921e-01 1.20555615e+00 5.84999740e-01 6.72615841e-02
3.88228118e-01 1.63716644e-01 1.61479749e-02 -1.60535388e-02
-1.06915009e+00 -1.36977762e-01 4.84331340e-01 8.85273933e-01
5.72006643e-01 -2.86014438e-01 2.05124855e-01 5.54882109e-01
1.97616499e-02 -1.61115155e-01 1.66132405e-01 9.33667541e-01
-2.84537017e-01 -1.11509669e+00 -2.07557634e-01 4.73588169e-01
2.40347996e-01 -1.85891569e-01 -3.68365765e-01 3.94380867e-01
4.85946536e-02 5.87270916e-01 1.94077954e-01 -3.79352719e-01
-1.22989632e-01 4.68242228e-01 9.09464061e-01 -7.54114211e-01
-4.42758411e-01 9.98430029e-02 -1.32115930e-01 -6.44578516e-01
-6.05402589e-01 -7.23407865e-01 -1.16457331e+00 5.78760467e-02
1.39137119e-01 -4.65241522e-01 6.18907630e-01 1.08367574e+00
5.70101976e-01 3.21382374e-01 1.95677161e-01 -1.22530973e+00
-3.46522480e-01 -7.38369584e-01 -2.68978745e-01 5.34694672e-01
3.20252180e-01 -5.72233260e-01 -2.29108498e-01 6.59203470e-01] | [11.065253257751465, 1.4168344736099243] |
6121fa7a-a76c-4f41-808c-3644b1d02ece | t-gap-learning-to-walk-across-time-for | 2012.10595 | null | https://arxiv.org/abs/2012.10595v1 | https://arxiv.org/pdf/2012.10595v1.pdf | T-GAP: Learning to Walk across Time for Temporal Knowledge Graph Completion | Temporal knowledge graphs (TKGs) inherently reflect the transient nature of real-world knowledge, as opposed to static knowledge graphs. Naturally, automatic TKG completion has drawn much research interests for a more realistic modeling of relational reasoning. However, most of the existing mod-els for TKG completion extend static KG embeddings that donot fully exploit TKG structure, thus lacking in 1) account-ing for temporally relevant events already residing in the lo-cal neighborhood of a query, and 2) path-based inference that facilitates multi-hop reasoning and better interpretability. In this paper, we propose T-GAP, a novel model for TKG completion that maximally utilizes both temporal information and graph structure in its encoder and decoder. T-GAP encodes query-specific substructure of TKG by focusing on the temporal displacement between each event and the query times-tamp, and performs path-based inference by propagating attention through the graph. Our empirical experiments demonstrate that T-GAP not only achieves superior performance against state-of-the-art baselines, but also competently generalizes to queries with unseen timestamps. Through extensive qualitative analyses, we also show that T-GAP enjoys from transparent interpretability, and follows human intuition in its reasoning process. | ['U Kang', 'Jinhong Jung', 'JaeHun Jung'] | 2020-12-19 | null | null | null | null | ['temporal-knowledge-graph-completion'] | ['knowledge-base'] | [-3.33483577e-01 4.85355079e-01 -6.79234505e-01 -3.65017176e-01
-4.95400757e-01 -6.78644598e-01 6.44630671e-01 7.19625533e-01
-2.89974332e-01 4.67648208e-01 7.23693669e-01 -5.09414077e-01
-4.10718322e-01 -1.25909078e+00 -9.92455661e-01 -9.34222788e-02
-6.47078812e-01 7.51906693e-01 5.30220687e-01 -1.74416572e-01
-2.20674202e-01 2.77371645e-01 -1.03157842e+00 3.29590976e-01
6.69810832e-01 8.93123627e-01 -3.67066383e-01 5.04435956e-01
1.11271152e-02 1.54532802e+00 -3.45572561e-01 -8.56742501e-01
-9.36138704e-02 5.82114188e-03 -1.12746096e+00 -4.65307534e-01
2.49900967e-01 -5.39707065e-01 -1.34004486e+00 6.23583496e-01
-4.66771908e-02 4.65509027e-01 3.30895662e-01 -1.32708621e+00
-1.23009562e+00 1.29682779e+00 -6.12047613e-02 4.84418362e-01
4.42422003e-01 2.55202830e-01 1.54878199e+00 -5.38938761e-01
8.43004167e-01 1.39707303e+00 5.56235492e-01 8.25486109e-02
-9.65787828e-01 -2.31131807e-01 5.81710160e-01 7.60962486e-01
-1.53546476e+00 -1.10052951e-01 5.41120708e-01 -3.64587940e-02
1.44451714e+00 2.45367527e-01 7.69805133e-01 9.57337022e-01
1.83953643e-01 8.86702061e-01 4.21727180e-01 -7.14628920e-02
2.05319881e-01 -4.81207997e-01 4.72831696e-01 7.74848342e-01
2.03052834e-01 8.99182335e-02 -9.45368171e-01 -1.19350396e-01
6.44180000e-01 1.32550716e-01 -2.56218791e-01 -1.64682120e-01
-1.42242479e+00 5.54201722e-01 8.11367154e-01 4.11250032e-02
-4.70951319e-01 8.92136395e-01 4.61972713e-01 2.29206994e-01
2.97113985e-01 2.70073742e-01 -4.28824961e-01 -3.39406371e-01
-5.97980976e-01 3.33961308e-01 9.47509050e-01 1.21257281e+00
6.12822056e-01 -2.33187303e-01 -6.36415660e-01 6.97676167e-02
3.19628716e-01 2.98916698e-01 1.50651604e-01 -9.73100603e-01
5.74170530e-01 8.78547192e-01 1.09859295e-01 -1.10241699e+00
-2.09145069e-01 -2.22113490e-01 -3.96156996e-01 -5.73416650e-01
2.40238562e-01 2.35453695e-01 -8.58290374e-01 2.05398989e+00
3.67599696e-01 5.34156561e-01 1.39772549e-01 7.00120091e-01
7.46444881e-01 7.23373830e-01 8.21279511e-02 3.54097299e-02
1.48165381e+00 -8.87402117e-01 -7.47558117e-01 -3.01139832e-01
7.01023221e-01 6.23551309e-02 1.02443862e+00 3.95971164e-02
-1.04868960e+00 -2.14582622e-01 -1.02246332e+00 -6.84434295e-01
-5.05408525e-01 -5.14705896e-01 1.13405788e+00 1.32458463e-01
-1.07861686e+00 5.14550447e-01 -1.21070194e+00 -3.98549497e-01
3.32682937e-01 3.35854478e-02 -2.62366146e-01 -3.91124725e-01
-1.84792733e+00 7.55157948e-01 8.74321222e-01 2.61329681e-01
-9.73627448e-01 -7.02457428e-01 -1.08204699e+00 3.17880571e-01
1.00010395e+00 -8.96093965e-01 1.38915515e+00 5.53897407e-04
-1.12466156e+00 4.55411851e-01 -3.90828252e-01 -7.73775458e-01
3.94776225e-01 -3.98101091e-01 -7.20928907e-01 2.73183495e-01
4.31835242e-02 3.29841942e-01 4.11827356e-01 -8.44517231e-01
-3.95105004e-01 -3.76796424e-01 7.77461231e-01 1.98817000e-01
-1.46126479e-01 -4.50106800e-01 -1.22606063e+00 -6.05727434e-01
8.60245079e-02 -6.78765416e-01 -7.92866126e-02 3.66016030e-02
-5.11821568e-01 -4.90575999e-01 7.13549435e-01 -6.26903117e-01
1.90123236e+00 -2.09805179e+00 6.10403754e-02 2.02208698e-01
4.87023920e-01 3.93851623e-02 9.57695320e-02 1.06681287e+00
2.61452675e-01 1.82928726e-01 1.86146237e-02 -2.61121273e-01
3.98692638e-01 8.26235175e-01 -8.44534814e-01 1.66050985e-01
-1.38226319e-02 1.66795099e+00 -1.47281682e+00 -5.79968512e-01
4.80488464e-02 1.26097918e-01 -5.96396625e-01 5.15331440e-02
-9.09573555e-01 -1.93179086e-01 -6.97460592e-01 6.69174850e-01
9.46249440e-02 -7.15936840e-01 5.32020271e-01 -3.98718327e-01
3.10228407e-01 5.93357563e-01 -9.12616134e-01 1.90589523e+00
-2.43207917e-01 4.10270005e-01 -5.25771439e-01 -7.06021547e-01
3.23787242e-01 3.81293803e-01 3.18561167e-01 -8.40243340e-01
-3.02479208e-01 -1.78977519e-01 -3.49517435e-01 -5.19412816e-01
9.01333570e-01 3.09938431e-01 -3.61086756e-01 6.41989410e-01
-5.87080121e-02 1.28293723e-01 3.85655463e-01 1.01794100e+00
1.36533999e+00 1.49514198e-01 2.42560849e-01 1.42821193e-01
1.28981665e-01 -5.49327657e-02 6.72117352e-01 1.02119935e+00
-8.63233358e-02 1.11570992e-01 7.47175038e-01 -4.39107686e-01
-5.97178280e-01 -1.30431557e+00 3.76423925e-01 9.76606011e-01
4.79502589e-01 -9.55279231e-01 -2.86107808e-01 -7.65072346e-01
5.75604856e-01 1.04862201e+00 -8.72354209e-01 -6.27487481e-01
-6.65977418e-01 -2.60427445e-01 9.35997725e-01 1.06363547e+00
4.37448353e-01 -9.29160655e-01 -2.79408962e-01 4.90378410e-01
-4.98720169e-01 -1.34801555e+00 -6.57780886e-01 -1.17958151e-01
-8.26977193e-01 -1.26914537e+00 3.12409312e-01 -2.26353481e-01
3.59118372e-01 2.48007417e-01 1.39257383e+00 1.96387380e-01
-9.98098701e-02 7.48486400e-01 -4.51097935e-01 -1.88037232e-01
-5.98912798e-02 -7.84914643e-02 -1.30676299e-01 -1.90775782e-01
5.33110976e-01 -7.21421242e-01 -5.27410150e-01 1.94524214e-01
-1.10988653e+00 -8.88168812e-02 4.04091477e-01 5.06578624e-01
7.09449947e-01 3.34264845e-01 4.57410067e-01 -1.09799874e+00
5.86151004e-01 -6.26214266e-01 -5.07160723e-01 7.28890002e-01
-8.69513988e-01 4.80593264e-01 5.72185218e-01 -2.91782260e-01
-1.17198634e+00 -7.56973445e-01 2.49207139e-01 -6.72458053e-01
4.26901937e-01 1.12207055e+00 -1.02411695e-01 4.60219324e-01
6.00719333e-01 2.60296762e-01 -4.62611824e-01 -1.48829401e-01
1.08986795e+00 1.59982014e-02 1.02132642e+00 -1.13803494e+00
8.30824792e-01 7.97209322e-01 -1.37985051e-01 -2.37398699e-01
-1.16298699e+00 -3.51618856e-01 -2.94591457e-01 4.35957052e-02
6.13680005e-01 -9.50009823e-01 -1.12696743e+00 2.73773491e-01
-1.13740659e+00 -6.26026273e-01 -6.30853176e-01 4.48548406e-01
-4.45441842e-01 5.08323491e-01 -1.05210268e+00 -6.39593303e-01
-1.57615051e-01 -6.92321599e-01 9.22888219e-01 -2.28499584e-02
-2.41457596e-01 -1.29376233e+00 -3.74911204e-02 3.38627070e-01
2.05552980e-01 2.30411410e-01 1.14072239e+00 -5.48318923e-01
-1.32016242e+00 -1.48972511e-01 -3.23321849e-01 -1.23559430e-01
7.03171268e-02 -1.24192491e-01 -8.61082315e-01 -1.41876742e-01
-3.65874767e-01 -2.77220547e-01 8.74715328e-01 8.60212520e-02
1.19336939e+00 -7.57386386e-01 -5.63137054e-01 6.88663483e-01
1.32584202e+00 7.30149373e-02 6.67219579e-01 -3.18616144e-02
8.28285873e-01 1.90921798e-01 7.13840306e-01 4.01583195e-01
1.12546647e+00 4.77527618e-01 6.36439443e-01 3.83712322e-01
-1.18447594e-01 -9.66848612e-01 3.13219547e-01 7.69311309e-01
-1.73967451e-01 -7.20736563e-01 -8.99885058e-01 9.52248693e-01
-2.36176419e+00 -1.08250761e+00 8.03524442e-03 1.97514391e+00
1.02912426e+00 3.19025129e-01 -2.34817177e-01 -2.13167980e-01
3.53106380e-01 6.04237258e-01 -8.40377986e-01 -1.35466635e-01
-5.48418351e-02 7.27149174e-02 7.49572039e-01 6.20373547e-01
-7.87493527e-01 1.20107186e+00 6.23598480e+00 5.86430013e-01
-5.96488953e-01 -2.23020818e-02 -1.95295755e-02 -9.91471857e-02
-7.94118106e-01 3.98424923e-01 -5.81695735e-01 3.11264694e-01
9.22576249e-01 -6.65914834e-01 7.51966536e-01 5.68228304e-01
-6.74169958e-02 -4.33427515e-03 -1.59311497e+00 6.71827137e-01
-1.67082787e-01 -1.53355432e+00 2.58423388e-01 -1.07624188e-01
5.58499873e-01 -1.97792873e-02 -1.87229797e-01 6.22447670e-01
9.14704561e-01 -9.46936786e-01 1.00242805e+00 7.53849030e-01
6.65051103e-01 -5.19374847e-01 4.85855401e-01 1.27329513e-01
-1.74928117e+00 -5.31968363e-02 -1.10449925e-01 -6.74127340e-02
3.51814538e-01 5.96749902e-01 -1.02537763e+00 1.31701326e+00
6.04471982e-01 1.01511395e+00 -5.32944977e-01 5.53499579e-01
-7.56188989e-01 7.35474229e-01 -3.27540070e-01 1.10825770e-01
3.36547732e-01 7.45664984e-02 4.34232652e-01 1.13232541e+00
2.75092889e-02 3.13663632e-01 2.05933422e-01 9.50030506e-01
-4.40403014e-01 -5.48717916e-01 -5.30640125e-01 -6.63085818e-01
1.05871642e+00 6.47342920e-01 -3.92986327e-01 -6.24160051e-01
-4.71610785e-01 9.64832366e-01 6.63450599e-01 8.60057592e-01
-1.10013413e+00 -2.48567343e-01 8.10087979e-01 1.60658099e-02
4.56025124e-01 -4.50176209e-01 -1.37296114e-02 -1.21172953e+00
2.86950618e-01 -5.38607240e-01 1.18360245e+00 -9.60630059e-01
-1.30048180e+00 2.44415417e-01 3.39637548e-01 -5.65564156e-01
-3.12725067e-01 -2.03964591e-01 -4.78469670e-01 6.68147147e-01
-1.40509343e+00 -1.43668330e+00 -2.61579901e-01 8.14222395e-01
-1.67859271e-02 4.74346608e-01 6.66672707e-01 2.07087517e-01
-3.74980718e-01 5.92017472e-01 -2.68005818e-01 3.34707886e-01
5.60381651e-01 -1.43420660e+00 8.15548897e-01 1.11992860e+00
3.29136968e-01 1.11632001e+00 5.22601902e-01 -8.70492160e-01
-1.90852666e+00 -1.34011269e+00 1.08293295e+00 -7.68868506e-01
1.05146766e+00 -1.60095960e-01 -1.10787725e+00 1.56291509e+00
-1.12593405e-01 1.24903046e-01 3.21285605e-01 6.15606487e-01
-9.95116234e-01 -3.10098380e-01 -6.75461233e-01 7.42895067e-01
1.47407687e+00 -1.12434065e+00 -1.01078868e+00 3.81096750e-01
1.45005691e+00 -6.57152295e-01 -1.21807826e+00 2.91197598e-01
3.53755087e-01 -7.08688915e-01 1.19017553e+00 -9.42343175e-01
2.33945504e-01 -5.75071573e-01 -9.50195864e-02 -8.68690908e-01
-2.93701768e-01 -7.26172984e-01 -1.06295097e+00 9.99421954e-01
2.08520412e-01 -6.80014789e-01 7.02506304e-01 7.93024063e-01
-3.41598988e-01 -7.92818069e-01 -7.06070125e-01 -9.54339027e-01
-4.17067230e-01 -8.94855082e-01 9.06555057e-01 1.01579928e+00
3.45852792e-01 1.85083240e-01 -2.36378551e-01 6.46855891e-01
5.25526285e-01 4.27425981e-01 5.17022312e-01 -1.01122785e+00
-3.95586699e-01 -1.63527697e-01 -3.28126371e-01 -1.31609786e+00
1.13122620e-01 -9.72586691e-01 -1.44360855e-01 -2.09166002e+00
2.74073239e-02 -2.20345721e-01 -3.63064408e-01 8.94072473e-01
-2.89395273e-01 -3.52368861e-01 -2.16071248e-01 1.09931447e-01
-1.01098061e+00 6.07606053e-01 1.29942167e+00 -3.21328461e-01
2.50805058e-02 -4.99266624e-01 -7.60475338e-01 2.68505216e-01
2.34220162e-01 -3.38785440e-01 -1.12976539e+00 -6.72381461e-01
7.98554301e-01 3.37942690e-01 5.67568898e-01 -5.30977249e-01
8.07844877e-01 -4.03253466e-01 -2.70677328e-01 -6.04540944e-01
4.26892996e-01 -5.61355531e-01 4.07149136e-01 1.40731975e-01
-2.98533618e-01 6.51654825e-02 1.27880946e-01 1.26870406e+00
-5.01561761e-01 3.47434253e-01 -5.43039814e-02 -1.38723731e-01
-1.25236845e+00 8.40704322e-01 9.29788649e-02 4.24568027e-01
7.48347163e-01 2.33648773e-02 -7.08294511e-01 -6.61656320e-01
-6.44991815e-01 7.90817499e-01 4.96189445e-01 4.07492548e-01
5.38220227e-01 -1.36124742e+00 -3.20999444e-01 -2.81442285e-01
5.30662835e-01 4.04553950e-01 4.86893296e-01 7.47052848e-01
-4.01707768e-01 5.75346589e-01 4.43926692e-01 -1.98231593e-01
-5.66249371e-01 1.00469375e+00 1.01689115e-01 -5.71055114e-01
-8.76068652e-01 8.51494551e-01 1.38453528e-01 -1.63790151e-01
3.89606804e-01 -9.25491869e-01 2.37708539e-01 -2.23437920e-01
3.27302456e-01 2.00100332e-01 4.42540161e-02 -8.42770860e-02
-4.89927351e-01 -4.17300127e-02 -1.59794256e-01 -1.42888561e-01
1.01756108e+00 -6.92772195e-02 -2.66500741e-01 4.90393728e-01
8.61022949e-01 -9.80243012e-02 -1.08314645e+00 -8.09156597e-01
2.01211929e-01 -4.81231213e-01 -1.15513891e-01 -8.58869672e-01
-8.38098168e-01 5.24977505e-01 -4.81735557e-01 2.26691574e-01
9.22314703e-01 2.28927404e-01 1.05449617e+00 8.64197493e-01
6.90233290e-01 -8.85193884e-01 2.06055388e-01 6.96648240e-01
6.25704050e-01 -7.68659353e-01 3.46861295e-02 -4.09701467e-01
-6.29052877e-01 9.01815057e-01 6.02811158e-01 3.32394302e-01
5.46486914e-01 -5.86283505e-02 -2.64558911e-01 -6.41970038e-01
-1.23541093e+00 -1.48008615e-01 2.44906232e-01 4.26145017e-01
7.27572804e-03 1.78482920e-01 -2.08602962e-03 7.88863361e-01
-1.09951454e-03 2.68121898e-01 3.18962425e-01 1.10060871e+00
-9.30486321e-02 -9.38745320e-01 1.34849668e-01 3.07342231e-01
-8.37427303e-02 -1.97846666e-01 -2.82364607e-01 1.00599098e+00
-2.66383320e-01 8.44357193e-01 -9.44053754e-03 -4.50468808e-01
3.81757647e-01 9.36348736e-02 4.95519072e-01 -5.56938112e-01
-2.68602848e-01 -6.07459426e-01 4.59817678e-01 -1.06729424e+00
-1.29794002e-01 -4.12343353e-01 -1.65114748e+00 -7.09980905e-01
6.97918311e-02 3.19718063e-01 -2.56257206e-02 8.39808464e-01
6.08805835e-01 7.84367085e-01 8.59917775e-02 6.20266423e-02
-4.39476222e-01 -5.90935409e-01 -6.61381423e-01 5.55944204e-01
1.69011384e-01 -6.13594055e-01 -1.01497546e-01 5.98094538e-02] | [8.572199821472168, 7.914571285247803] |
c810d120-5654-42b4-8087-8a26bbcf830b | semi-supervised-image-to-image-translation | 1901.08212 | null | http://arxiv.org/abs/1901.08212v1 | http://arxiv.org/pdf/1901.08212v1.pdf | Semi-Supervised Image-to-Image Translation | Image-to-image translation is a long-established and a difficult problem in
computer vision. In this paper we propose an adversarial based model for
image-to-image translation. The regular deep neural-network based methods
perform the task of image-to-image translation by comparing gram matrices and
using image segmentation which requires human intervention. Our generative
adversarial network based model works on a conditional probability approach.
This approach makes the image translation independent of any local, global and
content or style features. In our approach we use a bidirectional
reconstruction model appended with the affine transform factor that helps in
conserving the content and photorealism as compared to other models. The
advantage of using such an approach is that the image-to-image translation is
semi-supervised, independant of image segmentation and inherits the properties
of generative adversarial networks tending to produce realistic. This method
has proven to produce better results than Multimodal Unsupervised
Image-to-image translation. | ['Manan Oza', 'Sudhir Bagul', 'Himanshu Vaghela'] | 2019-01-24 | null | null | null | null | ['multimodal-unsupervised-image-to-image'] | ['computer-vision'] | [ 5.12243509e-01 4.32746142e-01 3.24227244e-01 -3.62880617e-01
-5.19207060e-01 -6.69403613e-01 1.05036402e+00 -3.35291237e-01
-4.82754976e-01 7.82468200e-01 4.37357975e-03 -2.00734302e-01
2.13833973e-01 -1.00292480e+00 -1.11498177e+00 -7.52538621e-01
4.32321012e-01 6.35784447e-01 1.90727666e-01 -3.30519468e-01
1.07216105e-01 6.38920963e-01 -1.12011313e+00 1.32347912e-01
5.36233425e-01 5.72495282e-01 -3.18718776e-02 1.02813089e+00
-1.19574465e-01 9.44088280e-01 -4.23863113e-01 -5.64219475e-01
6.07753098e-01 -8.35470319e-01 -8.78026426e-01 2.82037646e-01
5.56219220e-01 -2.90170580e-01 -1.99315935e-01 1.24578071e+00
4.06657904e-01 -1.49601817e-01 1.03267717e+00 -1.27408993e+00
-8.78928602e-01 3.12195480e-01 -3.84190559e-01 -3.21700633e-01
3.05464923e-01 1.44205332e-01 4.09702480e-01 -5.39041042e-01
1.00166702e+00 1.30903494e+00 5.59994876e-01 4.65102166e-01
-1.67757010e+00 -2.87821382e-01 -5.60305953e-01 -2.32543588e-01
-1.22092867e+00 -2.47365788e-01 8.22856426e-01 -5.30812919e-01
4.79625434e-01 3.04616690e-01 4.73969668e-01 1.01095057e+00
4.11126077e-01 3.43069524e-01 1.60725296e+00 -8.97598982e-01
5.02964891e-02 5.34423888e-01 -6.25064611e-01 6.54687643e-01
-6.68444633e-02 3.14289868e-01 -1.47875458e-01 2.39528045e-01
1.14503109e+00 -3.43315043e-02 -5.24918586e-02 -4.68627691e-01
-1.11869001e+00 9.39563930e-01 6.02584183e-01 5.59009016e-01
-3.62576067e-01 4.23699796e-01 2.87449121e-01 5.61867952e-01
2.89059132e-01 3.28550994e-01 -5.31568713e-02 2.67691672e-01
-1.25322473e+00 2.87442863e-01 6.78522587e-01 6.96940660e-01
7.82858372e-01 4.43433285e-01 6.53630355e-03 5.09910405e-01
2.89936423e-01 8.74850869e-01 6.11263216e-01 -9.65177238e-01
8.55284259e-02 3.88423324e-01 -6.45860657e-02 -1.21553731e+00
4.88584042e-02 -9.59402025e-02 -9.05903220e-01 8.06178153e-01
4.02217895e-01 -8.97652879e-02 -1.25306690e+00 1.76371133e+00
2.53133953e-01 -4.07317847e-01 1.46004364e-01 5.24780035e-01
5.34514964e-01 6.37832463e-01 1.45705029e-01 -8.68650561e-04
9.83626008e-01 -7.64947116e-01 -7.91941643e-01 1.59716487e-01
8.46758559e-02 -1.30313754e+00 8.74510229e-01 2.40567833e-01
-1.17323864e+00 -6.67573035e-01 -9.16320860e-01 -1.01945885e-01
-5.51950455e-01 2.51508737e-03 2.85219885e-02 8.66334558e-01
-1.31777871e+00 6.40759945e-01 -6.41448557e-01 -5.82525492e-01
9.21621844e-02 5.23528874e-01 -7.37237692e-01 1.62866771e-01
-9.82916236e-01 1.16719937e+00 6.79372311e-01 -1.73465669e-01
-9.25657153e-01 -1.93737835e-01 -8.94653320e-01 -2.75599599e-01
-2.98464429e-02 -9.00633276e-01 9.11399305e-01 -1.92127669e+00
-1.87992322e+00 1.25497878e+00 5.19567691e-02 -5.96520841e-01
1.00280166e+00 4.25762869e-02 1.88213829e-02 2.45584339e-01
-1.92985460e-01 1.19161284e+00 1.39245689e+00 -1.52705491e+00
-1.30003691e-01 -1.02734514e-01 -9.03187767e-02 -5.38560562e-03
-1.80791557e-01 -5.69615625e-02 -2.80290127e-01 -8.78964782e-01
-4.19345275e-02 -1.21443558e+00 -1.19150065e-01 2.16575526e-02
-4.52851295e-01 3.66158903e-01 8.34353030e-01 -8.84621203e-01
3.83876145e-01 -1.79717386e+00 3.96095693e-01 2.90383101e-01
-1.03783160e-01 3.60257089e-01 -2.33625904e-01 6.24456286e-01
-3.63207579e-01 6.47613555e-02 -4.32236761e-01 -5.16245902e-01
-1.13117948e-01 4.69339430e-01 -8.42064470e-02 5.82551718e-01
2.20418140e-01 1.17803252e+00 -5.18326938e-01 -7.50884712e-01
5.35387814e-01 7.91815162e-01 -5.12209535e-01 4.45651650e-01
-1.75388783e-01 7.75834143e-01 -2.47482713e-02 1.42204583e-01
7.15085566e-01 3.49047095e-01 -9.21111554e-02 -1.13995932e-01
1.47897646e-01 -4.96193260e-01 -8.27803850e-01 1.70168018e+00
-5.49102724e-01 7.90167689e-01 -6.67240173e-02 -1.06022406e+00
9.81306791e-01 4.23021674e-01 2.91692466e-01 -4.42120373e-01
4.60712016e-01 2.90028244e-01 -3.45398672e-02 -3.09406638e-01
3.10811132e-01 -5.15755594e-01 2.72468805e-01 4.34015989e-01
1.96461871e-01 -7.73383737e-01 -6.31422922e-02 1.92677602e-01
5.49936593e-01 6.17390394e-01 1.36613771e-01 -3.16201389e-01
6.69694185e-01 -6.39676303e-02 1.01774082e-01 4.64638352e-01
1.46569476e-01 9.06286299e-01 4.23522413e-01 -2.67269880e-01
-1.68478310e+00 -1.01599729e+00 1.11036859e-01 4.19262201e-01
-2.38289744e-01 3.04693043e-01 -1.32222819e+00 -4.99493718e-01
-4.58997369e-01 7.02430665e-01 -8.91305327e-01 -1.04144521e-01
-5.24381101e-01 -3.78434271e-01 5.60352325e-01 4.38176095e-02
5.63765407e-01 -1.35071516e+00 -2.14377999e-01 1.17925420e-01
-9.15052369e-02 -1.09868276e+00 -4.16943759e-01 4.89083715e-02
-9.61238384e-01 -7.24435687e-01 -1.24352217e+00 -8.97529721e-01
1.01965499e+00 -2.86772192e-01 1.18874931e+00 -1.58642009e-01
-2.52688050e-01 3.75597596e-01 -4.10645872e-01 -5.26090562e-01
-1.30549264e+00 -2.03435924e-02 -2.39017963e-01 2.54199445e-01
-1.10415690e-01 -7.62783349e-01 -4.98725057e-01 1.33186549e-01
-1.66482830e+00 1.06866807e-01 6.30557477e-01 7.68461168e-01
4.33446199e-01 -2.89593153e-02 1.01721972e-01 -1.03383195e+00
6.14154458e-01 8.59768130e-03 -5.41441023e-01 -5.57219330e-03
-6.16203904e-01 1.92674801e-01 6.45845413e-01 -4.84285563e-01
-8.65959108e-01 4.71304715e-01 -2.06139669e-01 -5.03703058e-01
-3.60243887e-01 2.81843767e-02 -2.10217521e-01 -5.35593212e-01
8.01745117e-01 4.02162731e-01 3.89581770e-01 -1.00057662e-01
5.79283059e-01 2.99830675e-01 6.08931065e-01 -2.96867102e-01
1.40103066e+00 5.68408668e-01 3.73995125e-01 -6.87374711e-01
-1.82589412e-01 4.31566872e-02 -8.82326126e-01 -2.98265457e-01
1.39548588e+00 -5.83719015e-01 -3.72243017e-01 7.12230802e-01
-1.35645223e+00 -3.80923152e-01 -5.02908766e-01 2.70888776e-01
-8.41045916e-01 4.51502234e-01 -6.56969607e-01 -6.57341421e-01
-3.30775231e-01 -1.23442709e+00 9.10065234e-01 1.58877745e-01
-8.81242007e-03 -1.17204285e+00 1.64796546e-01 2.89250046e-01
6.20115995e-01 6.87695920e-01 7.28070080e-01 -4.02673274e-01
-6.05467737e-01 -4.00276572e-01 -1.71130806e-01 9.63642478e-01
7.09742084e-02 1.29874006e-01 -8.73928428e-01 -6.14742637e-02
2.68218458e-01 -3.52331489e-01 5.51569939e-01 2.98199594e-01
5.33833861e-01 -2.73456931e-01 2.41004691e-01 4.96417224e-01
1.80442142e+00 6.30663633e-02 1.20168591e+00 3.88628989e-01
9.18292880e-01 6.16997182e-01 2.34023318e-01 -2.72645175e-01
-6.31443337e-02 7.06669807e-01 5.59520960e-01 -6.59448266e-01
-3.00144106e-01 -2.63124347e-01 4.72806096e-01 4.75732177e-01
-1.58287257e-01 -3.48637044e-01 -6.17710590e-01 5.64716578e-01
-1.63499558e+00 -9.00355875e-01 -2.96299875e-01 2.20024610e+00
8.09337020e-01 1.67575449e-01 -3.96710336e-02 9.40113962e-02
7.51960158e-01 -1.83375075e-01 1.79404486e-02 -9.18289900e-01
-2.36201912e-01 4.07126874e-01 8.69048893e-01 7.03436255e-01
-9.70428944e-01 1.09974742e+00 6.12420368e+00 8.78347099e-01
-1.25098014e+00 3.79645944e-01 6.86659575e-01 3.85229409e-01
-3.23342174e-01 -2.99340785e-02 -2.05532342e-01 2.65123188e-01
8.56550097e-01 4.41020161e-01 4.58945572e-01 6.10784590e-01
2.23259225e-01 -2.40546286e-01 -9.66749907e-01 7.75104284e-01
4.03286308e-01 -1.00314724e+00 4.27573174e-01 2.44560376e-01
1.04195976e+00 -1.98745459e-01 1.45800352e-01 -2.21216038e-01
3.23673666e-01 -1.21458018e+00 9.90587234e-01 7.47043073e-01
7.41667747e-01 -7.76925087e-01 8.74113202e-01 1.79622456e-01
-5.71146905e-01 5.36799014e-01 -1.38317421e-01 3.64436567e-01
3.15190375e-01 2.69745201e-01 -6.87830031e-01 5.81279993e-01
3.72568220e-01 8.00274163e-02 -4.57746446e-01 5.76415360e-01
-2.72858560e-01 4.09449875e-01 -2.18760833e-01 1.59491777e-01
4.84287113e-01 -5.69738865e-01 6.94978237e-01 1.21582985e+00
2.62063861e-01 -5.39951086e-01 -9.23113525e-02 1.05369425e+00
1.32798525e-02 3.84510845e-01 -1.04234886e+00 9.87553969e-03
-3.35921466e-01 1.06917834e+00 -9.33879375e-01 -3.85325730e-01
-1.05405664e-02 1.67752767e+00 -2.49238789e-01 2.60332733e-01
-8.69657099e-01 -3.07629853e-01 -7.94261023e-02 3.09609532e-01
4.23885673e-01 -2.15692177e-01 -1.03605494e-01 -8.12531173e-01
-1.10185839e-01 -1.06733131e+00 -3.52817327e-01 -9.42586184e-01
-9.54728365e-01 1.00645280e+00 2.34523299e-03 -1.12380791e+00
-4.13613200e-01 -6.17835522e-01 -5.70948660e-01 1.05700028e+00
-1.16788864e+00 -1.90447593e+00 -1.38897523e-01 7.50177443e-01
4.20577377e-01 -3.85005623e-01 8.72037470e-01 1.73759416e-01
-1.03205340e-02 5.08399010e-01 1.78274810e-01 2.76328772e-01
8.86172891e-01 -1.34794486e+00 1.48945272e-01 1.12961042e+00
2.85057545e-01 5.21680951e-01 1.20306981e+00 -5.16813219e-01
-1.01211166e+00 -1.04154027e+00 8.92352581e-01 -5.38278461e-01
5.11483192e-01 -1.98410243e-01 -5.78510702e-01 6.19744658e-01
1.00016654e+00 -1.21366724e-01 2.79649854e-01 -6.07049048e-01
-2.10470811e-01 -1.42560035e-01 -1.42110097e+00 5.86596131e-01
3.51440310e-01 -6.97356164e-01 -5.07083237e-01 5.14124453e-01
4.95946646e-01 -2.75798976e-01 -7.61816263e-01 -1.17037129e-02
3.16949308e-01 -1.19082785e+00 1.02897668e+00 -2.65814394e-01
7.36168623e-01 -3.84924948e-01 -6.70918599e-02 -1.18388844e+00
-2.09838920e-03 -8.44191670e-01 7.15205967e-01 1.43114424e+00
5.03315687e-01 -5.95000744e-01 6.44941211e-01 5.04235029e-01
2.75821030e-01 -1.27207264e-01 -7.02811003e-01 -7.12484658e-01
3.46354783e-01 -3.70311253e-02 1.20161898e-01 8.46770942e-01
-8.00943673e-01 2.97608465e-01 -7.96858549e-01 -1.55447528e-01
6.79241776e-01 -1.60113737e-01 1.07342482e+00 -8.39027584e-01
-5.58383584e-01 -2.78309762e-01 -6.74577594e-01 -3.29304457e-01
1.57975078e-01 -8.82045746e-01 -6.48549646e-02 -1.21431196e+00
1.40458494e-01 -8.71738568e-02 9.88830701e-02 2.55360574e-01
2.25544408e-01 9.62042689e-01 3.85665625e-01 2.27149919e-01
4.43156920e-02 3.20270687e-01 1.50098896e+00 -9.22677368e-02
1.07442707e-01 -2.38216639e-01 -1.88300461e-01 6.62882149e-01
7.87493527e-01 -7.81621218e-01 -3.83092254e-01 -1.77699983e-01
8.40643570e-02 -1.34491175e-01 5.06721795e-01 -8.63112211e-01
-1.31385073e-01 6.82089403e-02 4.21687305e-01 -1.39931917e-01
3.21994752e-01 -1.12054789e+00 5.60679197e-01 7.25863039e-01
-2.84340262e-01 1.92507029e-01 8.07739720e-02 4.69577193e-01
-4.43935364e-01 -5.21651149e-01 1.19388926e+00 -4.74039555e-01
-2.20891088e-01 5.32364100e-02 -3.54970455e-01 -3.65391463e-01
1.08418846e+00 -3.69409502e-01 2.87241071e-01 -8.15267384e-01
-8.54697227e-01 -6.47475064e-01 8.54358852e-01 2.54917324e-01
2.50086576e-01 -1.31142664e+00 -8.02671790e-01 6.07958026e-02
-1.97400838e-01 -3.40701520e-01 -1.51440818e-02 8.61931920e-01
-1.26255715e+00 3.18225414e-01 -6.89089000e-01 -6.50208354e-01
-1.50838876e+00 6.82274759e-01 4.26003486e-01 -2.67885149e-01
-4.21580970e-01 6.13501847e-01 1.52336851e-01 -5.51183283e-01
-1.33861601e-01 -3.56261283e-02 3.39892460e-03 -3.06087852e-01
1.87676996e-02 -1.04996257e-01 -1.77267604e-02 -1.01838803e+00
8.09707642e-02 8.29708099e-01 2.03567237e-01 -6.23876572e-01
1.20082390e+00 -5.52731939e-02 -4.39466238e-01 2.18775466e-01
1.37879252e+00 1.73453912e-01 -9.09883916e-01 2.36280332e-03
-4.67725635e-01 -3.33164930e-01 2.19578117e-01 -7.23396361e-01
-1.18156421e+00 8.38591874e-01 9.83293414e-01 2.75670797e-01
1.20154488e+00 -4.95609999e-01 6.83151245e-01 4.03088657e-03
1.09096609e-01 -9.96280432e-01 6.73276111e-02 1.90842286e-01
1.17747307e+00 -1.40731466e+00 -9.87396464e-02 -1.48533523e-01
-6.14982665e-01 1.11189783e+00 1.39973596e-01 -5.33404410e-01
4.47200239e-01 2.68560410e-01 4.47398305e-01 1.14047252e-01
1.03541054e-01 -1.42725140e-01 4.20808017e-01 7.44686365e-01
4.30596739e-01 8.09501931e-02 -4.53070074e-01 -4.47402239e-01
-1.14398114e-01 8.08461681e-02 4.55897957e-01 7.45593429e-01
-1.44579252e-02 -1.76653051e+00 -8.06226313e-01 -2.23125383e-01
-7.46560693e-01 -2.81989574e-01 -6.59130216e-01 9.07057881e-01
3.29912573e-01 6.58083856e-01 -1.76568151e-01 -1.99801728e-01
1.41954292e-02 1.15725793e-01 8.17414403e-01 -2.25583389e-01
-8.89793038e-01 2.87778527e-01 -2.59389251e-01 -4.24537450e-01
-7.56386936e-01 -5.92742205e-01 -8.84980023e-01 -4.90705758e-01
-1.12475328e-01 -1.51148587e-01 1.16752732e+00 8.23375762e-01
4.78434451e-02 5.90457678e-01 5.11329591e-01 -1.08178902e+00
-1.48513541e-01 -1.03400087e+00 -4.19281006e-01 7.99450159e-01
1.04057372e-01 -1.70919716e-01 -1.69733092e-01 7.34482646e-01] | [11.687090873718262, -0.3814775347709656] |
7633bc7c-cc89-4841-abd7-9883deaae6cd | versatile-multi-modal-pre-training-for-human | 2203.13815 | null | https://arxiv.org/abs/2203.13815v1 | https://arxiv.org/pdf/2203.13815v1.pdf | Versatile Multi-Modal Pre-Training for Human-Centric Perception | Human-centric perception plays a vital role in vision and graphics. But their data annotations are prohibitively expensive. Therefore, it is desirable to have a versatile pre-train model that serves as a foundation for data-efficient downstream tasks transfer. To this end, we propose the Human-Centric Multi-Modal Contrastive Learning framework HCMoCo that leverages the multi-modal nature of human data (e.g. RGB, depth, 2D keypoints) for effective representation learning. The objective comes with two main challenges: dense pre-train for multi-modality data, efficient usage of sparse human priors. To tackle the challenges, we design the novel Dense Intra-sample Contrastive Learning and Sparse Structure-aware Contrastive Learning targets by hierarchically learning a modal-invariant latent space featured with continuous and ordinal feature distribution and structure-aware semantic consistency. HCMoCo provides pre-train for different modalities by combining heterogeneous datasets, which allows efficient usage of existing task-specific human data. Extensive experiments on four downstream tasks of different modalities demonstrate the effectiveness of HCMoCo, especially under data-efficient settings (7.16% and 12% improvement on DensePose Estimation and Human Parsing). Moreover, we demonstrate the versatility of HCMoCo by exploring cross-modality supervision and missing-modality inference, validating its strong ability in cross-modal association and reasoning. | ['Ziwei Liu', 'Zhongang Cai', 'Liang Pan', 'Fangzhou Hong'] | 2022-03-25 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Hong_Versatile_Multi-Modal_Pre-Training_for_Human-Centric_Perception_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Hong_Versatile_Multi-Modal_Pre-Training_for_Human-Centric_Perception_CVPR_2022_paper.pdf | cvpr-2022-1 | ['human-parsing'] | ['computer-vision'] | [ 8.26950371e-02 5.01662195e-02 -2.71233618e-01 -4.73820597e-01
-1.16399431e+00 -3.86507154e-01 5.99031210e-01 -4.71269004e-02
-5.59863746e-01 4.42336887e-01 5.00360131e-01 1.60027713e-01
-2.56648138e-02 -5.53973258e-01 -1.01242709e+00 -6.13621712e-01
2.86823422e-01 6.69613183e-01 2.68078953e-01 -1.82184070e-01
-1.31202340e-01 2.51449376e-01 -1.92929995e+00 7.38863409e-01
6.52886093e-01 1.17412436e+00 4.26499248e-01 5.45524836e-01
6.61312640e-02 7.61018455e-01 3.85737512e-03 -5.31792879e-01
2.06854373e-01 -1.35224104e-01 -8.45759690e-01 1.75441429e-01
7.28379250e-01 -3.64029586e-01 -9.59232375e-02 6.96356297e-01
6.48544908e-01 1.41506180e-01 5.26780486e-01 -1.46327007e+00
-3.27475399e-01 2.76950032e-01 -6.57089353e-01 -1.99760586e-01
4.60083991e-01 4.22943294e-01 1.10069954e+00 -9.38748837e-01
6.63984716e-01 1.35997498e+00 6.30415797e-01 6.72175646e-01
-1.17606843e+00 -4.86739665e-01 1.87886402e-01 2.82445967e-01
-1.06251073e+00 -3.78513694e-01 7.15251565e-01 -5.07033646e-01
8.07112038e-01 2.06404448e-01 4.80724543e-01 1.60877407e+00
-1.78006276e-01 1.05069017e+00 1.36844981e+00 -3.25606436e-01
4.68841121e-02 3.48113775e-02 5.44253364e-02 1.03934181e+00
-7.89279640e-02 -6.59792591e-03 -1.05391192e+00 1.21804453e-01
8.07534099e-01 8.04696083e-02 -5.93637675e-02 -6.08859360e-01
-1.45319998e+00 6.29201353e-01 5.35987020e-01 -4.23360318e-02
-4.09292042e-01 2.47713223e-01 4.92087662e-01 -4.82658856e-02
1.15229987e-01 2.39483938e-02 -7.66938031e-01 -6.42738044e-02
-6.40340805e-01 1.72230795e-01 3.74634892e-01 1.09405458e+00
8.19939077e-01 -4.46395308e-01 -3.20394963e-01 7.97379851e-01
4.85595047e-01 6.98646963e-01 2.37068310e-01 -1.15758038e+00
8.02830040e-01 6.74754083e-01 -1.80088609e-01 -6.59185410e-01
-5.61580479e-01 -2.29774803e-01 -9.45043623e-01 2.98908502e-02
5.40832937e-01 1.14709258e-01 -1.05160594e+00 1.99365735e+00
6.65626228e-01 1.70502454e-01 -5.40273190e-02 9.50802088e-01
8.21526408e-01 3.46581936e-01 5.84165871e-01 1.43516019e-01
1.88575029e+00 -1.10454690e+00 -3.61026376e-01 -3.12093467e-01
4.95921552e-01 -5.63313186e-01 1.57676256e+00 3.73180509e-01
-1.04950690e+00 -8.47369790e-01 -5.76771975e-01 -6.94065392e-01
-3.57619792e-01 1.44660413e-01 9.68585670e-01 2.05336332e-01
-6.46447003e-01 7.37530366e-02 -1.01803958e+00 -2.59398997e-01
5.36667824e-01 2.15581506e-01 -8.31656456e-01 -5.42579770e-01
-9.31851506e-01 6.57748282e-01 6.41365826e-01 2.08570212e-02
-9.22530830e-01 -9.12868381e-01 -1.11126685e+00 -1.30681664e-01
4.38743830e-01 -1.37171674e+00 8.83238316e-01 -5.88942051e-01
-1.37346518e+00 1.18533170e+00 -2.14952111e-01 -1.67664826e-01
6.72912538e-01 -5.16478539e-01 4.68591712e-02 3.69345814e-01
2.55696803e-01 1.05168283e+00 8.98490667e-01 -1.36990643e+00
-7.96886623e-01 -6.01619899e-01 2.11681798e-01 3.79320294e-01
-2.83858329e-01 -2.65442729e-01 -8.95918310e-01 -4.07255173e-01
1.20009109e-01 -9.91117775e-01 6.85043558e-02 1.08167186e-01
-4.58898753e-01 -3.42441082e-01 3.83941203e-01 -7.99628258e-01
5.94840407e-01 -2.23136759e+00 6.24613941e-01 3.74454521e-02
1.90715954e-01 -1.83432698e-01 -1.50818095e-01 1.26629367e-01
2.48144448e-01 -4.82426345e-01 -3.50009590e-01 -9.81471479e-01
2.70292342e-01 4.38220650e-01 -1.09060884e-01 3.11352581e-01
2.80613780e-01 1.17030334e+00 -8.52599978e-01 -8.20173442e-01
4.28364187e-01 7.36676097e-01 -7.06184089e-01 5.42980850e-01
-3.42589855e-01 8.41152608e-01 -3.14891368e-01 1.01647663e+00
5.18051386e-01 -6.02150679e-01 -3.32828239e-02 -8.41109157e-01
1.14577487e-01 -6.23564459e-02 -1.01068425e+00 2.45023108e+00
-7.71317065e-01 5.74635975e-02 6.48406744e-02 -7.74702907e-01
3.38478565e-01 1.58612177e-01 3.87787879e-01 -9.92401361e-01
8.02105740e-02 1.41741008e-01 -5.34636021e-01 -6.83734119e-01
3.12460989e-01 -2.21276835e-01 -3.09852481e-01 1.15524821e-01
4.32752967e-01 -1.24605708e-02 -6.33305535e-02 3.22848797e-01
9.13722992e-01 5.42669356e-01 8.21966752e-02 1.27683178e-01
4.42761242e-01 -1.70945436e-01 4.45109516e-01 5.70667803e-01
-7.29704052e-02 6.90960467e-01 2.92352021e-01 -9.70284268e-02
-8.25052202e-01 -1.23906815e+00 -7.34158754e-02 1.69523799e+00
1.36085272e-01 -4.13672656e-01 -5.05986810e-01 -7.83567965e-01
2.60841865e-02 4.13858265e-01 -7.60613680e-01 9.67667550e-02
-5.16276896e-01 -5.44257462e-01 3.34157407e-01 8.73152137e-01
6.24603868e-01 -1.01596832e+00 -7.35502064e-01 -1.83000267e-01
-5.30454159e-01 -1.53860259e+00 -1.00515433e-01 2.38380656e-01
-5.93197405e-01 -1.09705758e+00 -7.62902617e-01 -5.23803651e-01
5.92110336e-01 7.99413919e-02 1.29334760e+00 -2.40255952e-01
-3.18373710e-01 9.40353155e-01 -3.14620048e-01 3.30695137e-02
1.01006061e-01 1.31778359e-01 -1.60171360e-01 4.11954969e-02
1.11710668e-01 -5.46072721e-01 -6.74376786e-01 1.86102286e-01
-8.01065087e-01 4.66808707e-01 7.90574968e-01 9.31206644e-01
9.47513819e-01 -3.10338050e-01 1.52487263e-01 -9.96125340e-01
-1.63241148e-01 -5.44101179e-01 -4.49570984e-01 4.17199850e-01
-1.21890016e-01 1.74492612e-01 3.34152400e-01 -2.15880722e-01
-1.29518831e+00 3.82460415e-01 -1.21120729e-01 -4.96565908e-01
-3.76683205e-01 3.57593685e-01 -6.10495746e-01 2.29722410e-01
4.19434369e-01 9.73854735e-02 -1.14888258e-01 -5.00507057e-01
8.93790007e-01 2.69246578e-01 8.94123614e-01 -9.78251994e-01
6.37726068e-01 7.85976350e-01 2.16985151e-01 -6.08437240e-01
-1.36201108e+00 -6.53399408e-01 -7.37677693e-01 6.74149916e-02
1.31769621e+00 -1.45093358e+00 -8.66620541e-01 4.58606064e-01
-1.08297682e+00 -4.59891826e-01 -1.78975582e-01 3.58374596e-01
-7.60476291e-01 3.65364254e-01 -7.10301816e-01 -5.28352678e-01
-2.36959890e-01 -1.08426952e+00 1.70011199e+00 -6.52788533e-03
-1.16093587e-02 -1.01311493e+00 -1.23502471e-01 1.12976432e+00
4.53554559e-03 4.64668751e-01 7.71712244e-01 -2.02457979e-01
-7.69242406e-01 1.75740167e-01 -5.49720168e-01 2.73982197e-01
-2.04610571e-01 -5.08572936e-01 -1.22015071e+00 -2.10807905e-01
-3.70730311e-01 -9.20413554e-01 1.08857155e+00 2.71099210e-01
1.30730557e+00 1.90750971e-01 -2.77282238e-01 9.26044703e-01
1.32492864e+00 -6.83216393e-01 4.28326726e-01 2.74545252e-01
1.27925372e+00 8.27832758e-01 7.64466345e-01 4.33582455e-01
1.00728738e+00 5.90653360e-01 5.57979047e-01 -1.75961778e-01
-3.65521669e-01 -4.12677646e-01 1.96903467e-01 7.56085634e-01
-2.11981371e-01 -4.61508967e-02 -8.66337359e-01 4.85943049e-01
-2.09546733e+00 -7.36978233e-01 -1.03985429e-01 1.90849626e+00
8.86458695e-01 9.56370961e-03 2.93418825e-01 -7.16579407e-02
2.65083641e-01 -1.01372622e-01 -4.94810641e-01 3.72329891e-01
-1.43509567e-01 1.85264543e-01 2.64477998e-01 3.13025117e-01
-1.22879934e+00 1.00810695e+00 4.83340454e+00 6.76698387e-01
-7.91212201e-01 4.62204099e-01 4.05308634e-01 -4.04428057e-02
-3.76941800e-01 -2.12866321e-01 -9.58766937e-01 4.07710254e-01
6.16667092e-01 6.82919860e-01 2.34037071e-01 7.71067858e-01
-2.25649700e-01 -1.84610933e-01 -1.30659807e+00 1.50365114e+00
9.35101882e-02 -1.14070916e+00 -6.22885302e-02 -1.48791922e-02
5.36972344e-01 1.20237567e-01 8.28827918e-02 5.50550640e-01
1.84637681e-01 -8.74931455e-01 8.26332211e-01 7.04908669e-01
8.79874229e-01 -6.53115094e-01 6.30854309e-01 2.80398011e-01
-1.23086119e+00 -2.78527942e-02 -1.40064225e-01 1.74897984e-01
3.67437840e-01 6.28123105e-01 -3.71590108e-01 7.74193227e-01
8.97602737e-01 8.02387834e-01 -6.29987359e-01 4.73203659e-01
-3.98248553e-01 3.12886000e-01 -4.22208309e-01 6.32222772e-01
-7.32813915e-03 2.14649856e-01 7.89819956e-02 1.27026534e+00
6.71330988e-02 -1.07795000e-02 3.78735721e-01 6.43134356e-01
-4.24317457e-02 -1.27981514e-01 -2.37712860e-01 2.94876784e-01
3.77132893e-01 1.33780670e+00 -4.93874788e-01 -2.60372937e-01
-5.74919224e-01 1.32949746e+00 6.12387538e-01 2.47283444e-01
-1.09664452e+00 2.40908861e-01 4.50940937e-01 9.26593915e-02
3.55753601e-01 -3.14947605e-01 -4.02400792e-01 -1.41930079e+00
7.23964274e-02 -8.56821895e-01 8.00328553e-01 -7.52526045e-01
-1.64957774e+00 4.13737714e-01 1.99008927e-01 -9.88042116e-01
-2.49092206e-01 -7.25097895e-01 7.15437680e-02 7.69923389e-01
-1.70590425e+00 -2.18532801e+00 -7.78442979e-01 1.13846469e+00
3.65088761e-01 -3.66568118e-02 7.50616908e-01 4.15455014e-01
-4.31430042e-01 7.07471013e-01 -6.18339241e-01 -3.41861956e-02
8.27765882e-01 -1.40541649e+00 3.41457799e-02 5.93694091e-01
1.40763298e-01 6.22910440e-01 3.83762270e-01 -4.53959107e-01
-1.73139977e+00 -1.03894627e+00 4.90589589e-01 -8.96673679e-01
5.73329806e-01 -5.22862434e-01 -7.74497867e-01 7.31820464e-01
-1.18253268e-01 4.84744906e-01 7.57212281e-01 4.92379874e-01
-7.61592507e-01 -3.26628350e-02 -1.07602143e+00 4.49661195e-01
1.39190221e+00 -8.42430472e-01 -5.04647195e-01 3.68412584e-01
9.03675437e-01 -6.52493477e-01 -1.04468226e+00 5.94166994e-01
5.92601001e-01 -1.11828518e+00 1.31623471e+00 -5.04286945e-01
5.52560747e-01 -2.15259388e-01 -6.50709987e-01 -7.49701023e-01
-1.83151230e-01 -1.81512192e-01 -5.45221150e-01 1.37521803e+00
-1.73579715e-03 -3.29573005e-01 7.04166710e-01 6.70795441e-01
-1.29078075e-01 -6.27241433e-01 -8.70111644e-01 -4.52993572e-01
-1.80615723e-01 -7.22635031e-01 1.73394278e-01 9.85380888e-01
-5.06300986e-01 5.81201077e-01 -5.50183654e-01 4.19932663e-01
9.73324895e-01 8.22173953e-02 1.00309241e+00 -1.06376433e+00
-7.43360162e-01 -3.25788441e-03 -2.65626103e-01 -1.07811677e+00
3.51148754e-01 -8.61796141e-01 -2.67516896e-02 -1.46215415e+00
3.68712664e-01 -6.04406178e-01 -4.00785893e-01 8.18775594e-01
-2.85650313e-01 3.88757676e-01 2.63334304e-01 1.56414688e-01
-1.05445445e+00 6.10642016e-01 1.20650852e+00 9.22575518e-02
9.18781012e-02 -3.39725614e-01 -5.20983994e-01 6.64065242e-01
3.58048588e-01 -2.12738693e-01 -5.29560685e-01 -6.24071956e-01
3.66671711e-01 -6.51652589e-02 8.80032361e-01 -9.52567101e-01
1.27888575e-01 -1.66750565e-01 3.82738441e-01 -6.86071575e-01
6.99482620e-01 -9.14354146e-01 -2.19056103e-02 3.89305465e-02
-1.67177916e-01 -1.85598046e-01 9.62289870e-02 7.25414574e-01
-2.20745817e-01 2.63851643e-01 7.23566055e-01 -1.75979197e-01
-1.17047858e+00 3.97359729e-01 5.03985167e-01 3.16978246e-01
8.01803291e-01 5.95498420e-02 -1.61716685e-01 7.46422783e-02
-8.35204959e-01 2.72757679e-01 4.31492120e-01 3.56284052e-01
5.27764559e-01 -1.25110567e+00 -5.27296841e-01 2.22113878e-01
5.49947023e-01 4.71199572e-01 6.66328311e-01 1.00193655e+00
-1.55062675e-01 1.90767452e-01 -3.71208787e-01 -1.06891954e+00
-1.14841831e+00 4.66013998e-01 -4.15692164e-04 -2.49873057e-01
-6.75624073e-01 1.19424915e+00 3.78515065e-01 -5.50163686e-01
2.75311530e-01 -2.54677534e-01 1.92384645e-01 6.12858832e-02
4.53740150e-01 2.82546192e-01 -2.03531105e-02 -6.12633288e-01
-4.43759441e-01 6.53560340e-01 1.23195432e-01 -2.31385201e-01
1.40436161e+00 -2.05629751e-01 -1.77679777e-01 6.61585689e-01
1.23050976e+00 -5.95127530e-02 -1.66997600e+00 -3.38769734e-01
-1.44283190e-01 -4.22325790e-01 -3.47631052e-02 -9.09147859e-01
-9.17793393e-01 1.03767967e+00 6.40459418e-01 -4.18666422e-01
1.22129309e+00 4.66275245e-01 9.69416916e-01 2.14044347e-01
6.50853455e-01 -1.00768065e+00 3.77697468e-01 3.72622967e-01
8.06254685e-01 -1.68067610e+00 -1.68649018e-01 -6.15434945e-01
-8.02896678e-01 7.12714195e-01 7.95775950e-01 2.13869303e-01
4.63223368e-01 3.89504693e-02 -4.52497378e-02 -3.49925995e-01
-6.62119508e-01 -5.78534663e-01 6.59554422e-01 6.19498312e-01
4.16281193e-01 1.56598017e-01 2.05290437e-01 6.75176978e-01
-6.16574474e-02 -1.26654610e-01 -3.21400940e-01 8.95646811e-01
-6.27820268e-02 -1.05992544e+00 -2.40785658e-01 5.96788414e-02
2.67455988e-02 -6.50520623e-02 1.18851960e-01 7.93647826e-01
5.49870491e-01 5.54895520e-01 -1.22192577e-01 -2.12548658e-01
2.83265799e-01 1.29326850e-01 9.22050476e-01 -5.41067064e-01
-3.92853349e-01 -2.01878175e-02 8.34647045e-02 -9.80558574e-01
-6.73627853e-01 -6.18952215e-01 -1.33459711e+00 1.24674721e-03
1.11193329e-01 -4.83615011e-01 5.98755181e-01 1.16427219e+00
4.54298854e-01 5.59955299e-01 5.50944060e-02 -9.08459842e-01
-2.84123361e-01 -7.52621412e-01 -2.52935261e-01 9.60753322e-01
2.30120853e-01 -1.05376112e+00 -1.64868578e-01 2.29806796e-01] | [8.012614250183105, -0.43323829770088196] |
166093b9-7be1-41fd-94b5-69fb1bb17765 | dynamic-object-removal-for-effective-slam | 2303.10923 | null | https://arxiv.org/abs/2303.10923v1 | https://arxiv.org/pdf/2303.10923v1.pdf | Dynamic Object Removal for Effective Slam | This research paper focuses on the problem of dynamic objects and their impact on effective motion planning and localization. The paper proposes a two-step process to address this challenge, which involves finding the dynamic objects in the scene using a Flow-based method and then using a deep Video inpainting algorithm to remove them. The study aims to test the validity of this approach by comparing it with baseline results using two state-of-the-art SLAM algorithms, ORB-SLAM2 and LSD, and understanding the impact of dynamic objects and the corresponding trade-offs. The proposed approach does not require any significant modifications to the baseline SLAM algorithms, and therefore, the computational effort required remains unchanged. The paper presents a detailed analysis of the results obtained and concludes that the proposed method is effective in removing dynamic objects from the scene, leading to improved SLAM performance. | ['Raj Kolamuri', 'Abhishek Bamotra', 'Phani Krishna Uppala'] | 2023-03-20 | null | null | null | null | ['video-inpainting', 'motion-planning'] | ['computer-vision', 'robots'] | [ 2.07476653e-02 -2.52257049e-01 1.04733273e-01 -1.50430098e-01
-2.48926222e-01 -5.05995929e-01 7.15034366e-01 -1.00492397e-02
-7.84955859e-01 8.42609227e-01 -1.33702753e-03 -6.77210018e-02
-3.16248268e-01 -6.08277142e-01 -5.27526319e-01 -5.90896368e-01
-2.88873821e-01 7.40106225e-01 5.63327014e-01 -2.65948474e-01
7.89290071e-01 1.06220865e+00 -1.76009071e+00 -4.65774894e-01
7.28716075e-01 5.48043907e-01 5.83795547e-01 7.42353320e-01
-1.47486672e-01 6.88929200e-01 -5.65548778e-01 1.83869913e-01
5.45418084e-01 -4.43966448e-01 -6.70888841e-01 8.12219530e-02
7.91432679e-01 -2.87674546e-01 -3.70684713e-01 9.49828923e-01
4.23778087e-01 5.32941461e-01 3.38605613e-01 -1.26327968e+00
3.61099333e-01 -2.40086734e-01 -4.31048244e-01 1.75934181e-01
5.29325485e-01 9.01807249e-02 5.07919133e-01 -7.70084202e-01
1.03798282e+00 1.31531703e+00 7.49858558e-01 7.07830861e-02
-1.07580054e+00 -5.34783006e-01 -1.96910575e-01 4.22038585e-01
-1.45530736e+00 -5.67890346e-01 6.33123100e-01 -5.22072196e-01
1.08554614e+00 1.41980380e-01 7.34438062e-01 4.75409776e-01
6.83763087e-01 2.67220408e-01 1.05241072e+00 -6.23402953e-01
3.20127338e-01 1.06800362e-01 -2.72109034e-03 7.22635388e-01
7.33634353e-01 2.32767269e-01 -6.46958232e-01 -5.10146059e-02
6.80496156e-01 -2.37628430e-01 -2.51446605e-01 -9.08115208e-01
-1.23005283e+00 7.05688894e-01 2.70395249e-01 4.42174822e-01
-5.33361733e-01 4.13360864e-01 3.25622857e-01 2.26333216e-02
2.51455903e-01 4.25987720e-01 -9.17775631e-02 -2.26656422e-01
-1.40515077e+00 5.91235816e-01 5.66167712e-01 9.79747772e-01
1.08774769e+00 -8.27746689e-02 1.03424616e-01 2.03898191e-01
5.22456646e-01 4.98047978e-01 1.58360362e-01 -1.12536025e+00
2.71065980e-01 3.91046882e-01 5.06291568e-01 -1.43692076e+00
-4.58463490e-01 -4.01860029e-02 -2.30751261e-02 6.14036977e-01
8.40986669e-02 8.42262283e-02 -9.97135997e-01 1.21576428e+00
6.05111063e-01 5.13333827e-02 -2.39222776e-02 9.93561864e-01
4.06146228e-01 6.93525553e-01 -2.24625438e-01 -9.98422652e-02
9.45200741e-01 -1.19027758e+00 -1.10284591e+00 -3.81713092e-01
6.00694120e-01 -1.10429955e+00 6.46096885e-01 2.20766008e-01
-7.95767009e-01 -5.75436771e-01 -1.30308771e+00 -1.93903074e-01
-1.68708697e-01 -9.22980681e-02 7.55725086e-01 3.80498350e-01
-1.03853250e+00 7.31416762e-01 -9.56230819e-01 -1.02216804e+00
-9.17304978e-02 3.55978340e-01 -3.92972857e-01 -3.01153481e-01
-6.98283851e-01 1.37687314e+00 5.89240313e-01 9.27393436e-02
-9.55166101e-01 -4.32279527e-01 -1.03487682e+00 -2.51546770e-01
3.62582266e-01 -6.34489000e-01 1.09432805e+00 -7.91853786e-01
-1.52657151e+00 7.57232308e-01 -6.10870779e-01 -5.00878394e-01
9.33769703e-01 -5.86594939e-01 -1.36368312e-02 2.14414060e-01
3.73843282e-01 6.25323296e-01 5.05175233e-01 -1.40252090e+00
-9.36299026e-01 -1.97295800e-01 5.62079577e-03 4.67974752e-01
4.87833261e-01 -1.81838199e-01 -7.44960070e-01 -1.29330233e-01
5.74024022e-01 -1.27158999e+00 -4.37984437e-01 -3.46465334e-02
7.83583596e-02 2.57731378e-01 1.03608263e+00 -3.65924776e-01
8.12349319e-01 -2.04153109e+00 1.27928406e-01 1.93560272e-02
-3.21999133e-01 2.58415081e-02 1.63926445e-02 8.41644049e-01
4.32927519e-01 -2.95944899e-01 -1.32798433e-01 -5.78159869e-01
-3.56754661e-01 4.91834790e-01 -2.56977290e-01 9.78067458e-01
-3.52296323e-01 4.63356674e-01 -9.61163878e-01 -5.05759180e-01
8.69724631e-01 3.85023266e-01 -2.31616229e-01 6.98632076e-02
-1.45692974e-01 7.15630591e-01 -2.64116257e-01 5.74454546e-01
9.99311864e-01 5.80922484e-01 3.99415614e-03 3.25208530e-02
-7.45135128e-01 2.74378151e-01 -1.38618684e+00 2.16152334e+00
-3.77071410e-01 1.01955605e+00 2.60814309e-01 -4.69989449e-01
9.49434459e-01 -1.56051293e-01 5.68198264e-01 -6.78119063e-01
1.21391349e-01 4.65938628e-01 -1.12786867e-01 -4.69761670e-01
1.12164998e+00 -3.11514530e-02 1.11741386e-01 -5.09221340e-03
-1.52410731e-01 -4.91928935e-01 2.73478121e-01 2.46448711e-01
1.06302047e+00 5.52271247e-01 3.39888364e-01 -6.51007056e-01
7.71472871e-01 7.49964058e-01 5.65872610e-01 8.82494390e-01
-4.09518689e-01 2.85089582e-01 -9.17221755e-02 -5.39618015e-01
-8.62412691e-01 -7.44342923e-01 9.87251922e-02 3.93474132e-01
1.03089464e+00 -3.85665596e-01 -4.24567878e-01 -2.30334684e-01
2.49745935e-01 8.18724096e-01 -4.88571465e-01 1.50239795e-01
-6.77807331e-01 -4.03146714e-01 2.05610529e-01 -1.46223918e-01
5.68057835e-01 -9.93020356e-01 -1.39519358e+00 2.88365304e-01
-3.49018753e-01 -1.17480707e+00 7.83621818e-02 1.27978669e-02
-1.20100129e+00 -1.06342864e+00 -2.12512389e-01 -3.77293080e-01
7.16018081e-01 8.77727389e-01 9.09293056e-01 1.52767196e-01
-4.23858285e-01 4.97654676e-01 -3.95141631e-01 -3.89010787e-01
-3.42590690e-01 -1.08241990e-01 8.11399966e-02 -1.71706438e-01
1.29864320e-01 -3.09844434e-01 -5.29799223e-01 3.26617837e-01
-6.64164007e-01 1.53442606e-01 6.76431596e-01 2.36618876e-01
6.40916705e-01 1.27534568e-01 -2.41467237e-01 -6.53969944e-01
1.10523805e-01 -2.04116687e-01 -1.05854261e+00 -2.86354452e-01
-7.01879799e-01 7.05205426e-02 1.75022513e-01 -4.56499830e-02
-1.00093830e+00 5.39956510e-01 1.52099162e-01 -2.50938863e-01
-1.41909897e-01 2.35833094e-01 6.78946450e-02 -7.96434164e-01
2.57146776e-01 1.48197010e-01 -1.25816222e-02 -3.35035145e-01
1.68531358e-01 1.73481569e-01 5.33072770e-01 -1.89951122e-01
1.03182232e+00 1.15504313e+00 4.71632510e-01 -1.07399189e+00
-4.31389153e-01 -9.45477307e-01 -8.86505842e-01 -4.99642015e-01
7.20765948e-01 -7.31240869e-01 -3.19790214e-01 5.02016008e-01
-1.27649498e+00 -1.95958346e-01 -1.42413750e-01 7.52164423e-01
-8.29464614e-01 5.84823549e-01 -3.20550799e-01 -9.20973659e-01
6.11697184e-03 -1.23193407e+00 1.05851889e+00 1.45005867e-01
-1.07286029e-01 -7.99726725e-01 4.36294198e-01 3.00283283e-01
5.26499748e-01 4.99894589e-01 3.75889719e-01 -7.13175163e-02
-1.26225376e+00 -2.01431617e-01 7.46502716e-04 -4.07436967e-01
-5.72303217e-03 -8.81440416e-02 -6.82787061e-01 -5.54631233e-01
4.44962196e-02 1.42309502e-01 6.81899488e-01 4.67918843e-01
1.36983857e-01 2.21047014e-01 -5.32503426e-01 8.14689398e-01
1.86826766e+00 2.98257262e-01 8.61881077e-01 1.10076320e+00
5.64498663e-01 6.68144524e-01 1.44835210e+00 2.96671957e-01
2.16587022e-01 9.79503274e-01 9.53038573e-01 5.24340495e-02
-1.05647542e-01 -2.13407636e-01 2.90262640e-01 3.75234544e-01
-3.75998840e-02 -1.78088874e-01 -8.77103090e-01 7.30517387e-01
-2.05358195e+00 -7.10715175e-01 -5.74856222e-01 2.19602084e+00
-1.25448033e-01 -9.31721926e-02 -4.93648410e-01 -4.94042411e-02
5.65023005e-01 4.38228369e-01 -2.17122763e-01 -3.02964270e-01
1.51855946e-01 -1.71348482e-01 1.09476459e+00 1.02079725e+00
-1.00174320e+00 1.35560179e+00 6.74196625e+00 3.47276270e-01
-1.30599213e+00 1.74150586e-01 -6.07982874e-01 -2.87833288e-02
1.20304421e-01 6.01193845e-01 -8.90452445e-01 1.15294449e-01
8.82645965e-01 -1.82490632e-01 3.54865253e-01 7.97622859e-01
6.26017511e-01 -1.11190975e+00 -6.69217646e-01 8.86940420e-01
4.46621478e-01 -1.22894931e+00 -7.82537684e-02 2.14637011e-01
6.02110624e-01 2.16747805e-01 -5.86329520e-01 1.35020008e-02
-7.56734461e-02 -4.85333204e-01 1.12198591e+00 5.21157384e-01
1.09303392e-01 -7.89934516e-01 9.72732604e-01 3.06285083e-01
-1.14468908e+00 1.43839300e-01 -3.51414949e-01 -4.15633410e-01
5.70261598e-01 4.07682925e-01 -1.13159359e+00 8.47051561e-01
6.71565354e-01 4.03925955e-01 -5.58431029e-01 1.51303327e+00
-1.85849965e-01 1.39705554e-01 -3.68115395e-01 8.78307968e-02
4.18147802e-01 -2.69286007e-01 9.94179368e-01 1.21567178e+00
4.56420213e-01 -2.49576628e-01 2.95057058e-01 6.29439116e-01
6.28666878e-01 5.17507754e-02 -8.21804523e-01 3.89325261e-01
3.15808475e-01 1.01540244e+00 -1.05372620e+00 -3.26105505e-01
-2.04395562e-01 1.18285310e+00 -4.31055017e-02 1.83299646e-01
-7.97738016e-01 -3.10261160e-01 6.39615953e-01 3.20662141e-01
2.16445357e-01 -8.49103391e-01 -1.94104090e-01 -9.58682716e-01
-6.19762912e-02 -3.24927181e-01 -3.75602324e-03 -8.76686811e-01
-2.66354769e-01 3.07589501e-01 2.88163185e-01 -1.29385841e+00
-1.58919901e-01 -2.01443136e-01 -1.58717588e-01 8.49870086e-01
-1.74647915e+00 -1.12998235e+00 -7.13626623e-01 3.92031252e-01
8.28126907e-01 2.46583760e-01 4.12436694e-01 2.94400185e-01
1.01168146e-02 -1.95977449e-01 1.68583155e-01 -5.59773743e-01
7.39904702e-01 -7.38692999e-01 3.07992071e-01 1.52163732e+00
-2.82783806e-02 5.36232233e-01 1.33909750e+00 -1.06770396e+00
-1.74458671e+00 -8.67599249e-01 1.02407098e+00 -4.02476728e-01
2.69994289e-01 -3.52489829e-01 -5.93856990e-01 7.09166527e-01
7.64480233e-02 -1.90693811e-01 -6.65027201e-02 -3.19158971e-01
3.56367469e-01 8.73428136e-02 -1.22750771e+00 3.84139836e-01
9.88379776e-01 -8.94603208e-02 -7.53488064e-01 2.62081683e-01
5.02843022e-01 -6.81744874e-01 -1.40157342e-01 6.12324595e-01
5.81502974e-01 -1.44374478e+00 7.76535809e-01 8.81686434e-02
-1.56092510e-01 -7.64202535e-01 -8.72384235e-02 -1.03598464e+00
-2.76620716e-01 -4.68123645e-01 3.64806578e-02 8.43414307e-01
-1.89706266e-01 -5.91190457e-01 9.45635974e-01 2.35689476e-01
-3.85849595e-01 -1.78942923e-02 -1.08684635e+00 -9.72925007e-01
-7.40595877e-01 -3.08447868e-01 1.62864954e-03 7.66573489e-01
-7.77958095e-01 -1.17652498e-01 -4.83445317e-01 6.23704672e-01
6.81692183e-01 2.00856566e-01 1.52028978e+00 -1.18267584e+00
8.48833919e-02 3.69607098e-02 -8.35518777e-01 -8.04148972e-01
5.34136482e-02 -4.28364158e-01 4.36185122e-01 -1.83988750e+00
-1.55538648e-01 -3.96129310e-01 1.48930684e-01 -1.05190817e-02
1.44357368e-01 7.36392885e-02 4.78298903e-01 5.70961237e-01
-5.64362347e-01 4.88075078e-01 8.94153714e-01 2.25025281e-01
-4.78830665e-01 -1.85531706e-01 1.82919487e-01 8.35360527e-01
3.38826746e-01 -8.24632049e-01 -3.59273911e-01 -5.20750761e-01
-7.07784146e-02 -8.17390811e-03 2.96068311e-01 -1.52905416e+00
3.99307251e-01 -3.32571894e-01 1.41638909e-02 -1.19959974e+00
4.97917175e-01 -1.26696467e+00 5.06798208e-01 1.08659577e+00
3.96690935e-01 3.02140534e-01 5.19960344e-01 5.57222426e-01
-1.86495885e-01 -6.69728398e-01 8.70316923e-01 -2.34827161e-01
-1.43562162e+00 -9.84463096e-02 -5.38965404e-01 -4.57903951e-01
1.45288038e+00 -5.49678147e-01 -4.56415489e-02 -4.79397058e-01
-4.61389512e-01 8.66801888e-02 1.06017768e+00 4.68377769e-01
3.82601202e-01 -9.65391338e-01 -3.43561590e-01 2.42635369e-01
3.56313549e-02 6.63449988e-02 2.60861605e-01 8.78839135e-01
-1.48017967e+00 5.82418144e-01 -4.57606018e-01 -8.13229799e-01
-1.43220496e+00 4.00560617e-01 3.15264761e-01 -1.32648364e-01
-7.21323609e-01 4.93103802e-01 -1.22877434e-02 -2.95593530e-01
1.66423753e-01 -1.04205668e-01 1.27427280e-01 -8.89219418e-02
3.55506450e-01 6.66821301e-01 1.38779163e-01 -9.31853652e-01
-6.27566338e-01 9.24513042e-01 2.89727122e-01 -4.74828452e-01
1.08607697e+00 -6.24731779e-01 -2.75265694e-01 4.49904144e-01
7.50187278e-01 5.19269347e-01 -1.19618142e+00 1.83865830e-01
2.64238685e-01 -1.02501154e+00 1.45463228e-01 -4.83300030e-01
-6.17388844e-01 5.84038377e-01 7.33920395e-01 -3.93281430e-01
8.35266471e-01 -2.81210840e-01 6.08560741e-01 3.14324260e-01
1.04258835e+00 -9.13668215e-01 -2.16861665e-01 8.95771682e-01
7.38910615e-01 -1.09646368e+00 5.42553723e-01 -6.72615170e-01
-2.37598270e-01 1.11433792e+00 4.63666022e-01 -4.04731840e-01
1.16379932e-02 1.77227527e-01 1.75320059e-01 -1.64541215e-01
-2.29876295e-01 -2.95876771e-01 -1.30257159e-01 4.30300295e-01
9.40780621e-03 -3.34013313e-01 -8.22587788e-01 -5.63675046e-01
-1.33000717e-01 1.07829414e-01 7.83956230e-01 1.59721851e+00
-7.03171492e-01 -9.61655796e-01 -7.10800767e-01 -2.12997377e-01
-1.28030106e-01 1.40628070e-01 -3.59228849e-01 1.31267357e+00
2.48925418e-01 7.70272017e-01 -8.72572437e-02 -2.48946488e-01
4.75652546e-01 -1.43813387e-01 5.83987355e-01 -4.37062770e-01
-3.33988637e-01 2.44395714e-02 1.04651883e-01 -1.00580299e+00
-7.30341971e-01 -8.17437470e-01 -1.43950260e+00 -2.74714082e-01
-3.24725509e-01 2.02516347e-01 1.22039354e+00 8.14399064e-01
5.79641283e-01 3.31849307e-01 3.13384354e-01 -1.53901815e+00
-1.15013659e-01 -6.69325829e-01 -3.70094895e-01 3.11511129e-01
5.21627545e-01 -1.00768864e+00 -4.16190416e-01 -2.12162212e-01] | [7.293731212615967, -2.106858015060425] |
1529f223-346d-4091-ba7e-0137ed4a3085 | video-text-pre-training-with-learned-regions | 2112.01194 | null | https://arxiv.org/abs/2112.01194v2 | https://arxiv.org/pdf/2112.01194v2.pdf | Video-Text Pre-training with Learned Regions | Video-Text pre-training aims at learning transferable representations from large-scale video-text pairs via aligning the semantics between visual and textual information. State-of-the-art approaches extract visual features from raw pixels in an end-to-end fashion. However, these methods operate at frame-level directly and thus overlook the spatio-temporal structure of objects in video, which yet has a strong synergy with nouns in textual descriptions. In this work, we propose a simple yet effective module for video-text representation learning, namely RegionLearner, which can take into account the structure of objects during pre-training on large-scale video-text pairs. Given a video, our module (1) first quantizes visual features into semantic clusters, then (2) generates learnable masks and uses them to aggregate the features belonging to the same semantic region, and finally (3) models the interactions between different aggregated regions. In contrast to using off-the-shelf object detectors, our proposed module does not require explicit supervision and is much more computationally efficient. We pre-train the proposed approach on the public WebVid2M and CC3M datasets. Extensive evaluations on four downstream video-text retrieval benchmarks clearly demonstrate the effectiveness of our RegionLearner. The code will be available at https://github.com/ruiyan1995/Region_Learner. | ['Jinhui Tang', 'Guanyu Cai', 'Xudong Lin', 'Alex Jinpeng Wang', 'Yixiao Ge', 'Mike Zheng Shou', 'Rui Yan'] | 2021-12-02 | null | null | null | null | ['video-text-retrieval'] | ['computer-vision'] | [ 2.94274539e-02 -2.83585340e-01 -3.01737458e-01 -3.54735792e-01
-9.48476255e-01 -5.89654803e-01 7.80313194e-01 2.84319162e-01
-6.29562438e-01 3.19924444e-01 2.85565972e-01 -1.44752949e-01
2.11995512e-01 -5.87828755e-01 -9.47608173e-01 -6.04453325e-01
6.99536800e-02 2.97739118e-01 4.42208290e-01 1.09540656e-01
7.99075514e-02 1.38931170e-01 -1.52198493e+00 8.36674631e-01
4.01295930e-01 1.11928391e+00 5.09460628e-01 7.63104916e-01
-1.79157734e-01 8.71298552e-01 -2.38725364e-01 -1.08234577e-01
2.18592376e-01 -3.54795843e-01 -6.89744771e-01 2.69423246e-01
8.46657753e-01 -5.65549314e-01 -8.38434756e-01 9.19464767e-01
2.88842201e-01 2.02759281e-01 6.31809056e-01 -1.13856828e+00
-5.63715458e-01 4.34521735e-01 -6.38105929e-01 4.09929574e-01
2.94439673e-01 2.57547379e-01 1.18986368e+00 -1.29072869e+00
7.57911682e-01 1.28909063e+00 3.20486754e-01 4.32686180e-01
-9.98524070e-01 -6.54876173e-01 4.50018167e-01 3.22458088e-01
-1.49194741e+00 -4.48142678e-01 5.70886195e-01 -5.10263681e-01
9.59100664e-01 1.29699916e-01 6.58026636e-01 1.13281059e+00
-6.91330507e-02 1.24131298e+00 6.82317615e-01 -2.28940830e-01
-5.54756001e-02 1.13131091e-01 -1.02041308e-02 7.63395607e-01
-2.23279931e-02 -1.62229732e-01 -6.88880086e-01 1.05151758e-01
7.54121482e-01 4.42221612e-01 -3.52792799e-01 -6.24548912e-01
-1.43668425e+00 6.38216496e-01 5.66371620e-01 3.88714105e-01
-2.99719900e-01 4.47282165e-01 4.75051880e-01 1.16186626e-01
5.76438367e-01 -2.37784266e-01 -2.71669537e-01 -1.86502598e-02
-1.22917855e+00 1.04071736e-01 2.84173071e-01 9.92977738e-01
9.81432319e-01 -3.23768198e-01 -3.75684738e-01 6.92656159e-01
3.85184377e-01 4.85227227e-01 4.32238519e-01 -5.21883368e-01
8.55097353e-01 6.40872598e-01 -7.08406717e-02 -8.31726849e-01
-1.20312586e-01 -1.08026914e-01 -6.21892035e-01 -1.45689756e-01
3.65230978e-01 9.76589248e-02 -1.07215142e+00 1.43703651e+00
3.70222539e-01 4.71056700e-01 1.91933140e-02 1.12067986e+00
1.03669047e+00 9.39177692e-01 2.49727219e-01 1.07736625e-01
1.43056762e+00 -1.16994750e+00 -5.66274881e-01 -3.07107747e-01
6.83734179e-01 -6.55805290e-01 1.00738645e+00 -3.06136673e-03
-1.12014019e+00 -4.87559080e-01 -8.46536994e-01 -3.61876875e-01
-5.27758300e-01 2.94418871e-01 3.28001529e-01 8.81352499e-02
-1.03807557e+00 2.88862407e-01 -8.59739602e-01 -5.02232194e-01
6.18577182e-01 2.28154302e-01 -4.85519737e-01 -2.99180120e-01
-9.88772750e-01 4.68148023e-01 5.67170620e-01 -7.41383899e-03
-1.15977156e+00 -5.96749544e-01 -1.03262925e+00 1.60878733e-01
6.69274867e-01 -6.57373428e-01 1.07390368e+00 -1.28873909e+00
-1.04084003e+00 8.77239764e-01 -3.69753987e-01 -3.76579076e-01
4.05112475e-01 -3.64553750e-01 -1.25159547e-01 7.26036072e-01
7.76416883e-02 1.07766068e+00 1.07247221e+00 -1.26737654e+00
-8.37636709e-01 -2.02349663e-01 2.27647781e-01 3.62641603e-01
-5.31693697e-01 2.23122805e-01 -1.00103128e+00 -8.30377340e-01
-2.37812638e-01 -8.46888363e-01 5.35963401e-02 3.41385394e-01
-1.62638009e-01 -2.74636507e-01 1.04023123e+00 -5.82220316e-01
1.10709763e+00 -2.32009244e+00 1.61587790e-01 5.28846681e-02
2.84455478e-01 2.56468982e-01 -3.56762677e-01 4.79226083e-01
-3.57452594e-02 -1.30137950e-01 -1.06924579e-01 -4.23106223e-01
-6.58070892e-02 -9.12790596e-02 -2.16511607e-01 5.88847220e-01
1.63940236e-01 1.17442560e+00 -9.98470902e-01 -7.93739796e-01
6.09221637e-01 5.71010232e-01 -6.72212839e-01 2.36211106e-01
-4.70890254e-01 1.60915077e-01 -5.14844656e-01 6.68988824e-01
3.89068872e-01 -4.19568300e-01 1.59952044e-02 -3.11020911e-01
-2.00605989e-02 2.06109926e-01 -1.08336544e+00 2.06495738e+00
-3.69860113e-01 9.07308102e-01 6.39701216e-03 -1.17723382e+00
4.42316443e-01 3.56227458e-01 5.97355187e-01 -7.06311166e-01
1.29791901e-01 -5.39684528e-03 -4.66302693e-01 -6.33803964e-01
4.96547163e-01 1.44147620e-01 -1.11483447e-01 3.39076668e-01
3.67806345e-01 2.57928342e-01 2.48272642e-01 4.79905993e-01
9.66917217e-01 3.34545821e-01 3.29857022e-01 -2.29998175e-02
5.45330226e-01 -1.43882409e-01 1.26189023e-01 5.85674345e-01
-1.07583560e-01 7.89368868e-01 3.24176401e-01 -3.41827452e-01
-9.09736931e-01 -1.12670255e+00 1.66165903e-01 1.32045722e+00
3.77471268e-01 -7.13021934e-01 -5.71751475e-01 -8.64377856e-01
4.19969894e-02 3.77258241e-01 -6.94449782e-01 2.84811649e-02
-4.52130288e-01 -2.01423049e-01 3.99156421e-01 6.37289643e-01
3.87154728e-01 -9.75139976e-01 -4.81577873e-01 1.57628302e-02
-3.51294607e-01 -1.35089803e+00 -8.64158213e-01 -1.71628390e-02
-6.28337264e-01 -1.06870484e+00 -7.30238497e-01 -9.96568799e-01
6.83654487e-01 7.13102520e-01 9.38026190e-01 2.01021120e-01
-4.43464905e-01 7.43000329e-01 -4.82779771e-01 -1.22932620e-01
-6.63688034e-02 -3.94675601e-03 -2.91840136e-01 2.91819781e-01
4.63815719e-01 -1.93620563e-01 -8.13477397e-01 2.66032904e-01
-1.10166478e+00 2.21163929e-01 6.34874761e-01 6.53367341e-01
7.82302558e-01 -1.50536492e-01 1.75639734e-01 -6.52223825e-01
5.86817376e-02 -5.54313362e-01 -4.78586614e-01 3.68580043e-01
-3.08810081e-02 -1.14865184e-01 4.20875639e-01 -3.73366117e-01
-9.28408206e-01 3.71554375e-01 2.65424073e-01 -9.74823236e-01
-2.49597088e-01 4.60647732e-01 -1.49945676e-01 3.87581706e-01
2.85849780e-01 3.90161097e-01 -1.93526134e-01 -3.35909933e-01
5.54835916e-01 6.92603230e-01 5.39976358e-01 -5.21494865e-01
9.83598113e-01 7.66419530e-01 -3.21480274e-01 -1.06278169e+00
-8.01699460e-01 -1.00923872e+00 -7.38662183e-01 -3.50574017e-01
1.16115355e+00 -1.33957613e+00 -3.61636788e-01 2.65203595e-01
-1.05059755e+00 -4.67194289e-01 -7.68382549e-02 6.01761162e-01
-5.10908186e-01 4.17696089e-01 -5.83626866e-01 -4.35351104e-01
-2.91158766e-01 -8.75383735e-01 1.48600280e+00 4.21314836e-02
7.93285519e-02 -9.47600126e-01 -1.47497788e-01 5.37386239e-01
-2.61394065e-02 -7.89566785e-02 5.31986415e-01 -6.29266858e-01
-9.59699392e-01 -1.79039374e-01 -4.64127600e-01 2.04472333e-01
-3.32440808e-02 -3.89136896e-02 -9.15962994e-01 -4.57433015e-01
-3.81013274e-01 -3.85013252e-01 1.20633817e+00 2.74678409e-01
1.33527625e+00 -3.35750967e-01 -3.81169677e-01 5.80022871e-01
1.49287009e+00 -1.69202164e-01 6.42837107e-01 3.13832372e-01
1.05743635e+00 5.96036494e-01 8.39707792e-01 3.90990466e-01
3.95507395e-01 7.79841840e-01 4.95474428e-01 -2.19014212e-01
-2.92765886e-01 -3.99621010e-01 6.34871781e-01 5.64897060e-01
9.85761061e-02 -4.77830201e-01 -7.82845557e-01 6.98232293e-01
-2.18413138e+00 -1.13877296e+00 1.08933501e-01 1.93955302e+00
6.83987498e-01 -9.95709095e-03 1.76330954e-01 -2.53925025e-01
7.85735905e-01 3.45020145e-01 -3.58849257e-01 1.57046884e-01
-1.72545835e-02 -2.63139233e-02 4.84110653e-01 3.16779822e-01
-1.35179412e+00 1.17384279e+00 4.52808046e+00 9.51985240e-01
-1.13672364e+00 1.33481592e-01 4.74798888e-01 -6.12325430e-01
-1.45878404e-01 -8.46084654e-02 -8.07974160e-01 3.37356716e-01
7.60528743e-01 -8.26424062e-02 3.26047659e-01 7.43097663e-01
2.94670910e-01 5.77131249e-02 -1.35354590e+00 1.01006782e+00
3.17451686e-01 -1.47601902e+00 3.62103671e-01 -7.06985071e-02
7.07339585e-01 2.70309925e-01 -6.41209707e-02 2.57346034e-01
-6.39329702e-02 -9.38286662e-01 9.95464265e-01 3.71231139e-01
8.31582725e-01 -5.76662183e-01 5.79252303e-01 1.84435695e-01
-1.60631835e+00 -4.81579676e-02 -3.48523438e-01 2.36538365e-01
2.70293970e-02 3.86799395e-01 -7.29900658e-01 4.79767919e-01
9.12530720e-01 1.27363753e+00 -7.31398284e-01 9.79263604e-01
-2.42454097e-01 5.60306668e-01 -2.31757924e-01 8.39222521e-02
6.05367184e-01 3.73001434e-02 4.53358889e-01 1.56168342e+00
1.29674047e-01 -6.42423052e-03 3.30042958e-01 6.76445603e-01
-2.92594403e-01 2.42876351e-01 -6.45583272e-01 -2.00102970e-01
3.27769160e-01 1.30544436e+00 -7.86595047e-01 -6.72294140e-01
-9.59368229e-01 9.95085597e-01 3.64743233e-01 5.28722286e-01
-9.42681193e-01 -1.02286480e-01 5.52553535e-01 2.83329368e-01
8.24305117e-01 -9.86571461e-02 1.54478416e-01 -1.41805792e+00
2.73466140e-01 -8.04041684e-01 5.92742801e-01 -9.26520765e-01
-1.13419986e+00 2.64935076e-01 1.75455153e-01 -1.43423009e+00
-1.15116283e-01 -6.45445168e-01 -3.74817491e-01 4.57875103e-01
-1.51697385e+00 -1.36879158e+00 -4.79818106e-01 1.01198351e+00
1.06955147e+00 6.67516813e-02 4.26350981e-01 4.05496657e-01
-4.21942443e-01 5.11385143e-01 2.13713571e-01 4.21961069e-01
7.87859499e-01 -1.01761055e+00 2.41341710e-01 7.64955282e-01
5.89931548e-01 3.96098822e-01 3.28196943e-01 -5.32022059e-01
-1.38568318e+00 -1.48847163e+00 7.52223432e-01 -4.26773548e-01
7.40181208e-01 -6.65677190e-01 -9.95244205e-01 7.97519088e-01
3.30375791e-01 3.51089865e-01 4.17275399e-01 -2.41844907e-01
-5.31840622e-01 -6.18850067e-02 -5.55232644e-01 6.16568446e-01
1.09054506e+00 -8.33424926e-01 -6.46501243e-01 4.34147567e-01
6.25791192e-01 -2.92022109e-01 -5.81043184e-01 1.63061425e-01
4.81136113e-01 -7.49111652e-01 1.16755116e+00 -5.96913218e-01
5.65046132e-01 -3.45505714e-01 -1.70118198e-01 -9.93214130e-01
-1.76305562e-01 -3.40222269e-01 -1.27824441e-01 1.24614179e+00
2.71802545e-01 -2.08452791e-01 6.44678354e-01 1.70727521e-01
-6.61087707e-02 -6.26301050e-01 -8.04202676e-01 -7.09267974e-01
-8.97182375e-02 -4.04798806e-01 6.34791180e-02 9.01717067e-01
6.74174801e-02 2.94738144e-01 -3.66782904e-01 2.00689584e-01
5.00155985e-01 2.40132242e-01 7.69071579e-01 -8.50093782e-01
-2.44165495e-01 -4.53360379e-01 -5.01133621e-01 -1.26473904e+00
3.33393127e-01 -1.08022535e+00 1.83655351e-01 -1.60658574e+00
6.00662529e-01 -8.63688961e-02 -4.76066262e-01 6.11137271e-01
-2.93100536e-01 3.67843479e-01 3.46280485e-01 2.91531950e-01
-1.23250079e+00 5.77634454e-01 1.05015910e+00 -3.51580650e-01
1.82897877e-02 -3.84127676e-01 -2.27224588e-01 7.86566794e-01
7.56501317e-01 -5.56135774e-01 -4.61204320e-01 -5.00798166e-01
2.13476294e-03 5.90683855e-02 7.09376216e-01 -8.34735155e-01
1.96032584e-01 -1.09264456e-01 4.62596416e-01 -8.32624495e-01
3.51345509e-01 -1.01026428e+00 -2.31637016e-01 2.39727929e-01
-4.82689857e-01 -4.81817611e-02 2.35409379e-01 8.39412272e-01
-3.04326326e-01 -1.08610235e-01 5.57167053e-01 -1.49156600e-01
-9.65776026e-01 5.03891289e-01 -4.65402424e-01 1.40689522e-01
1.14636528e+00 -6.53624982e-02 -2.81613737e-01 -3.02105963e-01
-5.00688851e-01 3.52628171e-01 6.48814857e-01 7.78717339e-01
8.06661308e-01 -1.25350463e+00 -7.06791401e-01 5.92001788e-02
5.11184275e-01 4.68737669e-02 3.03467065e-01 7.00409651e-01
-4.20699060e-01 5.91461897e-01 2.31989808e-02 -9.23386037e-01
-1.58340621e+00 8.18316936e-01 1.99772745e-01 -1.55359820e-01
-8.41310084e-01 6.93923056e-01 8.34540248e-01 -9.57687385e-03
4.30990368e-01 -3.34204525e-01 -1.13443792e-01 1.52808771e-01
6.19552314e-01 4.20857109e-02 -1.23506196e-01 -8.39466929e-01
-5.05503297e-01 6.68244302e-01 -2.18646660e-01 -5.38549125e-02
1.18176961e+00 -2.82082558e-01 1.66467607e-01 4.61872369e-01
1.56258893e+00 -2.67095655e-01 -1.49683213e+00 -5.54461181e-01
-1.12503804e-01 -5.45656323e-01 1.62066787e-01 -3.38266253e-01
-1.28310680e+00 1.03437960e+00 3.78446877e-01 -2.79251963e-01
1.01394522e+00 3.77162188e-01 8.43623221e-01 5.28986931e-01
6.16908930e-02 -1.06284428e+00 4.85778809e-01 2.96869963e-01
6.97679341e-01 -1.30388844e+00 5.77603541e-02 -3.02591980e-01
-7.06243932e-01 1.08754587e+00 6.01606488e-01 -1.19859576e-01
4.67210919e-01 -1.19355749e-02 -8.01378563e-02 -2.14517072e-01
-1.05288911e+00 -4.26606506e-01 5.52060425e-01 3.25030893e-01
3.27134579e-01 -1.93675861e-01 3.64003666e-02 2.00993314e-01
3.85516196e-01 4.92823981e-02 2.06280142e-01 9.97021079e-01
-5.02755404e-01 -7.55490422e-01 -2.91457474e-01 3.68881464e-01
-3.68720412e-01 -3.60603273e-01 -2.31979430e-01 8.37450802e-01
5.96176609e-02 7.24344254e-01 2.27119789e-01 -1.71504766e-01
1.47704259e-01 5.76183908e-02 4.52755660e-01 -6.71582937e-01
-4.67617720e-01 3.67210060e-01 -5.96537590e-02 -8.52297008e-01
-5.91941595e-01 -8.52467060e-01 -1.40272033e+00 3.70677151e-02
-1.04689129e-01 -3.18478867e-02 5.28214872e-01 8.32241416e-01
3.55282336e-01 4.83557463e-01 3.78519744e-01 -1.04967761e+00
-1.53985143e-01 -8.05608392e-01 -2.99358696e-01 6.97026670e-01
4.52406377e-01 -6.36108935e-01 -3.29986423e-01 5.01841843e-01] | [10.067136764526367, 0.913196325302124] |
ae11741d-70f6-4480-be8c-5162adcdce53 | learning-modulated-loss-for-rotated-object | 1911.08299 | null | https://arxiv.org/abs/1911.08299v3 | https://arxiv.org/pdf/1911.08299v3.pdf | Learning Modulated Loss for Rotated Object Detection | Popular rotated detection methods usually use five parameters (coordinates of the central point, width, height, and rotation angle) to describe the rotated bounding box and l1-loss as the loss function. In this paper, we argue that the aforementioned integration can cause training instability and performance degeneration, due to the loss discontinuity resulted from the inherent periodicity of angles and the associated sudden exchange of width and height. This problem is further pronounced given the regression inconsistency among five parameters with different measurement units. We refer to the above issues as rotation sensitivity error (RSE) and propose a modulated rotation loss to dismiss the loss discontinuity. Our new loss is combined with the eight-parameter regression to further solve the problem of inconsistent parameter regression. Experiments show the state-of-art performances of our method on the public aerial image benchmark DOTA and UCAS-AOD. Its generalization abilities are also verified on ICDAR2015, HRSC2016, and FDDB. Qualitative improvements can be seen in Fig 1, and the source code will be released with the publication of the paper. | ['Yue Guo', 'Xue Yang', 'Silong Peng', 'Junchi Yan', 'Wen Qian'] | 2019-11-19 | null | null | null | null | ['object-detection-in-aerial-images'] | ['computer-vision'] | [ 8.94939601e-02 -1.54919758e-01 -2.62102693e-01 -2.18703225e-01
-6.21627867e-01 -3.56539607e-01 3.41317117e-01 -2.79416800e-01
-4.34439272e-01 5.37636697e-01 -1.61160544e-01 -2.46302158e-01
-4.73578155e-01 -3.34035695e-01 -6.23246610e-01 -8.44211757e-01
-2.26393938e-01 -2.22124428e-01 2.37379789e-01 -4.09775287e-01
8.26931819e-02 5.87743044e-01 -1.44547093e+00 -4.22248542e-02
1.11672330e+00 1.20367944e+00 4.93417978e-02 2.14644611e-01
5.19313753e-01 4.16732162e-01 -8.23603809e-01 -4.66084182e-01
5.84838569e-01 -5.52881919e-02 -5.72435737e-01 7.28458017e-02
4.82648700e-01 -5.13842642e-01 -2.25189760e-01 1.11091304e+00
5.09490192e-01 3.04991705e-03 5.75327098e-01 -1.05148828e+00
-4.60453004e-01 4.10122991e-01 -1.07436740e+00 4.61262427e-02
5.82897626e-02 -1.42965436e-01 9.46643651e-01 -9.19461131e-01
5.84880829e-01 9.83608305e-01 1.00645959e+00 1.52218789e-01
-8.95853519e-01 -6.58013225e-01 3.34427685e-01 1.55935585e-01
-1.69926763e+00 -5.30949607e-02 6.42803550e-01 -4.34982896e-01
6.45345569e-01 3.44262123e-01 3.40885460e-01 1.11322284e+00
1.24559686e-01 4.39025730e-01 7.98811316e-01 -4.20640379e-01
-1.76887900e-01 6.97694272e-02 -2.31624246e-02 7.87202001e-01
4.60732073e-01 2.61779577e-01 6.29941896e-02 1.05662890e-01
8.76826823e-01 -1.84072748e-01 -4.41947639e-01 -3.85922670e-01
-9.53655243e-01 7.17819095e-01 1.05092466e+00 -1.29838549e-02
-4.26252596e-02 -1.84469521e-01 2.85110146e-01 3.30063432e-01
5.69631696e-01 8.64699543e-01 -3.84174138e-01 3.57890636e-01
-6.71805382e-01 1.74565330e-01 1.67049780e-01 1.05590892e+00
4.17462230e-01 1.64962281e-02 -3.06145772e-02 1.16972733e+00
2.22279608e-01 5.74460447e-01 4.12835062e-01 -7.19053626e-01
5.76994479e-01 5.05590498e-01 1.67897180e-01 -1.26883805e+00
-6.85436666e-01 -7.22300410e-01 -1.03966463e+00 -2.05636583e-02
3.44120085e-01 -2.96553105e-01 -7.76823819e-01 1.65792191e+00
2.22163066e-01 -5.37346257e-03 -8.47634226e-02 1.22926879e+00
7.95964181e-01 2.95193762e-01 -2.19012663e-01 -6.46099970e-02
1.31695175e+00 -8.66258085e-01 -6.32003367e-01 -3.49209368e-01
7.63766885e-01 -7.54249156e-01 9.06458676e-01 4.35392946e-01
-6.47810102e-01 -6.75556242e-01 -1.51251078e+00 6.69868588e-02
-3.20924848e-01 9.73290324e-01 6.53666794e-01 2.54309356e-01
-4.22584772e-01 6.86929345e-01 -5.78639865e-01 -2.81019598e-01
3.30949992e-01 1.35874957e-01 -1.88363925e-01 8.53052735e-02
-1.14768445e+00 9.29267108e-01 2.09050365e-02 5.17530024e-01
-1.00405306e-01 -7.04672992e-01 -7.04696000e-01 -3.84993434e-01
4.89487141e-01 -4.76858348e-01 8.94092441e-01 -6.89691603e-01
-1.30773795e+00 8.14077675e-01 4.79063958e-01 -5.14017165e-01
8.77061546e-01 -6.49718821e-01 -4.81165439e-01 -9.60576087e-02
-4.42526191e-02 6.31124794e-01 9.74702656e-01 -1.20210302e+00
-5.93473911e-01 -5.94549179e-01 -6.42630551e-03 9.00362805e-02
-1.09551787e-01 -1.92453966e-01 -4.05982345e-01 -8.38439286e-01
5.51264763e-01 -1.11391592e+00 -9.19038728e-02 -1.06750354e-01
-5.17283440e-01 1.05788037e-01 8.44839215e-01 -5.96409440e-01
1.55471635e+00 -2.52232575e+00 9.82837379e-02 2.95722429e-02
-1.44762397e-01 2.64746815e-01 -1.70445636e-01 -4.32738177e-02
-3.04882824e-01 1.31922156e-01 -2.81321228e-01 -6.56663552e-02
-3.51231635e-01 -1.26253977e-01 -6.85103893e-01 7.71986723e-01
3.09954971e-01 3.81929040e-01 -7.04241931e-01 -1.43159524e-01
1.70147166e-01 6.65058315e-01 -3.72585326e-01 -1.95931911e-01
3.39314729e-01 1.91703901e-01 -3.96752089e-01 6.33278787e-01
9.51853931e-01 -4.15761545e-02 -3.64282094e-02 -8.43981385e-01
-2.48024479e-01 3.86289544e-02 -1.43941081e+00 1.34210742e+00
-2.45893434e-01 7.06104100e-01 -9.12471935e-02 -7.88094401e-01
1.14402437e+00 -8.73248354e-02 3.19698960e-01 -6.07825160e-01
1.63556442e-01 2.15804055e-01 6.00970350e-03 -4.39266801e-01
7.40027130e-01 3.00294012e-01 1.01063676e-01 -3.67424965e-01
-2.22133934e-01 -4.09609377e-01 4.25036112e-03 -2.21303791e-01
5.78525484e-01 3.80692452e-01 3.09406221e-01 -1.72591403e-01
4.41729218e-01 -1.48948416e-01 6.57957137e-01 4.99600589e-01
-1.32629842e-01 8.70996654e-01 5.65032125e-01 -4.85459507e-01
-1.06521618e+00 -6.53359890e-01 -8.39899540e-01 6.75922632e-01
3.53996813e-01 -4.15659338e-01 -3.90436560e-01 -5.90183020e-01
5.53873507e-03 5.16610086e-01 -6.17652714e-01 -2.41098374e-01
-6.03450477e-01 -1.23987842e+00 6.62291348e-01 5.94640493e-01
7.32867718e-01 -4.35467243e-01 -6.97120130e-01 -8.67011398e-02
-3.11964750e-01 -1.42277706e+00 -1.42067879e-01 2.16124177e-01
-9.25142229e-01 -1.16049111e+00 -6.31786346e-01 -4.63730276e-01
6.79059625e-01 3.90571237e-01 7.52060175e-01 9.97699946e-02
-6.10427499e-01 -2.04430804e-01 -4.83333886e-01 -3.99079919e-01
6.07809462e-02 2.11823270e-01 4.29096855e-02 -3.92038047e-01
-2.32535481e-01 -1.17676936e-01 -6.56957030e-01 8.08503687e-01
-7.29318857e-01 -8.41189083e-03 4.46163654e-01 9.15558279e-01
5.03084183e-01 -1.45619825e-01 4.35889781e-01 -5.83889663e-01
2.81899840e-01 -2.39050582e-01 -9.60979640e-01 1.65709436e-01
-8.46449733e-01 -8.86347331e-03 4.81347859e-01 -3.10445666e-01
-8.43454838e-01 -4.58152592e-02 1.23363130e-01 -4.16816562e-01
1.46771669e-01 3.01904112e-01 1.13552414e-01 -7.97898769e-02
7.37620711e-01 -2.89887965e-01 -3.59162718e-01 -2.76134461e-01
1.29860893e-01 5.28913438e-01 6.17980242e-01 -3.06644529e-01
8.05632174e-01 4.84994233e-01 2.58571237e-01 -8.89716625e-01
-1.08543634e+00 -3.91050726e-01 -4.61073488e-01 -1.46482438e-01
4.76604581e-01 -1.19815600e+00 -2.74589419e-01 6.56560004e-01
-1.16403520e+00 -5.47020398e-02 -1.35574400e-01 7.04902530e-01
-1.76989615e-01 3.77394706e-01 -3.81759882e-01 -6.74991071e-01
-1.73939899e-01 -1.29487896e+00 1.12998676e+00 2.39339441e-01
1.72641397e-01 -5.59348524e-01 -1.72255725e-01 1.49155408e-01
1.99579746e-01 4.56048667e-01 7.28433251e-01 -1.09373137e-01
-2.62878597e-01 -4.07437354e-01 -5.29779673e-01 5.92117250e-01
3.65820788e-02 5.49720824e-01 -1.00876939e+00 -5.12299597e-01
-1.40571043e-01 -1.87117070e-01 1.08048844e+00 5.50521731e-01
1.31751204e+00 -1.18664771e-01 -3.72544467e-01 1.05052495e+00
1.52972138e+00 3.12501043e-02 7.74280608e-01 5.90060711e-01
6.44955933e-01 5.18546283e-01 1.01858950e+00 4.66333717e-01
1.51944444e-01 9.74668145e-01 6.97862804e-01 -3.44494075e-01
-8.35928023e-02 2.01946478e-02 2.24254981e-01 5.42593241e-01
-5.83539724e-01 -2.21023366e-01 -6.48157179e-01 1.09288350e-01
-1.56513333e+00 -8.00750434e-01 -3.22207570e-01 2.39079070e+00
4.48907107e-01 8.06673840e-02 -7.46883824e-02 2.08394960e-01
6.11366332e-01 3.74522388e-01 -4.08299655e-01 -1.08113386e-01
-5.06839931e-01 -3.26301038e-01 1.14245903e+00 4.67400342e-01
-1.34250200e+00 7.85277188e-01 6.09207726e+00 7.58877277e-01
-1.36249542e+00 -3.26154262e-01 6.14726663e-01 1.41456261e-01
2.98987806e-01 -1.42096415e-01 -9.82427895e-01 3.49137276e-01
2.93161303e-01 3.92280966e-01 3.18085015e-01 7.68835366e-01
3.69575694e-02 -4.17253515e-03 -7.65638173e-01 9.68407750e-01
1.18723363e-02 -7.61967719e-01 -5.73812276e-02 -6.52350262e-02
6.63457096e-01 1.65237784e-01 3.60277921e-01 3.75947684e-01
-3.04996014e-01 -1.00664115e+00 5.57161152e-01 3.02412748e-01
1.05649936e+00 -6.95080340e-01 8.87555063e-01 -7.23445714e-02
-1.21537006e+00 -4.52299833e-01 -5.95228374e-01 2.47838855e-01
-3.22572142e-01 4.43603337e-01 -7.15322793e-01 1.05003071e+00
9.37324107e-01 6.91056490e-01 -7.85852432e-01 1.10542357e+00
-3.87916118e-01 2.94556260e-01 -3.15384895e-01 3.15303504e-01
7.16528594e-02 -4.30484653e-01 7.96839416e-01 1.04442835e+00
4.30647314e-01 -2.65302688e-01 -2.07747772e-01 6.33422852e-01
1.64354388e-02 -6.13508001e-02 -4.40169215e-01 4.15134668e-01
5.93140662e-01 1.25049663e+00 -5.57109416e-01 1.60071954e-01
-3.41753721e-01 7.12386727e-01 5.70537671e-02 3.72660309e-01
-1.26089776e+00 -6.31297350e-01 4.91950125e-01 1.51072338e-01
4.75448608e-01 -1.11135542e-02 -2.22986415e-01 -1.03478038e+00
2.74314016e-01 -8.77299011e-01 2.18839958e-01 -6.84765339e-01
-1.13561094e+00 7.95952678e-01 3.39245260e-01 -1.71895409e+00
9.80479941e-02 -8.19869518e-01 -3.86480957e-01 4.11366284e-01
-1.48452818e+00 -8.40505958e-01 -6.63693309e-01 1.85654480e-02
3.65090132e-01 -2.25280911e-01 5.76131463e-01 5.25476515e-01
-9.49995577e-01 8.68863344e-01 1.43903553e-01 1.88437700e-01
1.09621084e+00 -9.87520933e-01 2.21274123e-01 7.67061591e-01
-7.38227740e-02 3.55200022e-01 8.74159396e-01 -1.30454436e-01
-1.02531230e+00 -1.11792064e+00 5.23413002e-01 -2.49110833e-01
4.79089946e-01 -1.14564262e-01 -9.14051473e-01 5.24850786e-01
-2.18545854e-01 2.09515557e-01 2.64147907e-01 3.37317139e-02
-4.62767422e-01 -3.82036090e-01 -1.05072033e+00 4.67778414e-01
1.06257522e+00 -9.20943469e-02 -2.99659610e-01 2.62551785e-01
7.41614521e-01 -7.51078188e-01 -7.78902531e-01 1.10361063e+00
4.87980008e-01 -8.86722505e-01 1.23732460e+00 -3.35143179e-01
3.72724056e-01 -4.34322268e-01 -2.24488735e-01 -1.21960521e+00
-2.24081278e-01 -4.07729894e-01 2.38208577e-01 1.08243322e+00
5.42020023e-01 -6.72005594e-01 2.13533372e-01 -1.30458638e-01
-1.48429200e-01 -1.09315860e+00 -9.13887858e-01 -9.46893394e-01
-2.54715588e-02 -1.99857280e-01 6.17759228e-01 9.94102597e-01
-2.74364769e-01 3.36743534e-01 -4.21534777e-01 3.96127671e-01
4.62028056e-01 -1.54222921e-01 7.48452127e-01 -1.22710466e+00
-1.42458260e-01 -3.42352659e-01 -3.15807313e-01 -1.16655171e+00
-3.10984403e-01 -4.50572252e-01 3.41532519e-03 -9.74625766e-01
-2.57024080e-01 -5.62271774e-01 -1.68786511e-01 2.67339915e-01
-1.14816159e-01 9.91691574e-02 2.31808648e-01 3.64261240e-01
-3.82650733e-01 6.73738837e-01 1.48311996e+00 8.90383963e-03
-3.52318794e-01 -4.67042252e-02 -4.96283740e-01 8.94459188e-01
7.23207951e-01 -3.63082916e-01 -2.82067060e-01 -5.22728264e-01
3.85477901e-01 -1.69691890e-01 6.66621745e-01 -1.14352107e+00
-3.83984186e-02 8.13780054e-02 6.15042686e-01 -6.46072865e-01
3.46284240e-01 -7.38522530e-01 -2.59417921e-01 4.40107554e-01
-1.90240070e-01 1.35689288e-01 4.00076717e-01 4.02882069e-01
-2.13096827e-01 -5.25509156e-02 1.02276027e+00 4.41797227e-01
-3.95133168e-01 4.77071777e-02 2.25011930e-01 -8.52720290e-02
9.01705444e-01 -2.39082932e-01 -7.47767270e-01 3.80554870e-02
-3.39041501e-01 2.35961914e-01 3.30890507e-01 5.19363105e-01
5.05966067e-01 -1.15718913e+00 -7.28502691e-01 3.82332057e-01
1.69893757e-01 2.10022718e-01 3.02248359e-01 1.09158313e+00
-5.33369303e-01 2.81300932e-01 -1.45489082e-01 -6.35909259e-01
-1.17132807e+00 2.37924382e-01 5.98154247e-01 -2.59829015e-01
-6.35312796e-01 9.36583221e-01 4.25089374e-02 -2.70100832e-01
4.41069663e-01 -7.22066879e-01 -2.05394000e-01 2.28831127e-01
2.36962393e-01 5.43806195e-01 8.72618333e-02 -3.61302108e-01
-5.48686206e-01 9.29072201e-01 9.92193935e-04 3.66704404e-01
1.12728620e+00 -2.10873205e-02 2.63015598e-01 9.89632830e-02
9.89714444e-01 2.83782240e-02 -1.31499767e+00 -9.99880210e-02
-2.93734699e-01 -4.41174716e-01 8.53053182e-02 -7.73016095e-01
-1.13793194e+00 5.61153471e-01 9.93160486e-01 1.20115452e-01
1.17952287e+00 -2.92956233e-01 4.56365377e-01 3.48454177e-01
1.62457507e-02 -1.23932874e+00 -1.67018414e-01 5.08747160e-01
1.25312698e+00 -1.35271990e+00 5.81590354e-01 -6.22075975e-01
-7.06030190e-01 1.16159141e+00 7.68737674e-01 -2.77937680e-01
4.06632572e-01 2.26490274e-01 5.00045717e-03 8.30354765e-02
-3.74730319e-01 7.30645284e-02 4.74294782e-01 3.97257626e-01
4.88151878e-01 -9.33459122e-03 -4.91373628e-01 6.27931535e-01
-3.33373040e-01 -2.63604164e-01 2.16448903e-01 5.58694661e-01
-1.48772866e-01 -7.13266373e-01 -3.80782098e-01 3.80476043e-02
-4.20195490e-01 8.01295638e-02 -2.97784060e-01 1.29370368e+00
3.16530585e-01 6.61202669e-01 2.21209198e-01 -5.10637105e-01
6.59928381e-01 -5.04503965e-01 6.37864053e-01 -1.10285558e-01
-2.31764466e-01 8.33969638e-02 -1.10345647e-01 -5.85531771e-01
-3.28586727e-01 -4.15561914e-01 -1.10656524e+00 -5.57357781e-02
-6.57428980e-01 -1.20961867e-01 7.95575321e-01 5.93648016e-01
2.91150719e-01 6.38226628e-01 8.51996303e-01 -3.74683410e-01
-1.10023844e+00 -1.10310459e+00 -4.48474228e-01 2.67185092e-01
5.77356517e-01 -9.02198374e-01 -7.46199071e-01 -1.75756052e-01] | [8.710329055786133, -0.8182957172393799] |
a40665c0-f866-4789-a4ef-41c2d4896886 | beyond-farthest-point-sampling-in-point-wise | 2107.04291 | null | https://arxiv.org/abs/2107.04291v3 | https://arxiv.org/pdf/2107.04291v3.pdf | Task-Aware Sampling Layer for Point-Wise Analysis | Sampling, grouping, and aggregation are three important components in the multi-scale analysis of point clouds. In this paper, we present a novel data-driven sampler learning strategy for point-wise analysis tasks. Unlike the widely used sampling technique, Farthest Point Sampling (FPS), we propose to learn sampling and downstream applications jointly. Our key insight is that uniform sampling methods like FPS are not always optimal for different tasks: sampling more points around boundary areas can make the point-wise classification easier for segmentation. Towards this end, we propose a novel sampler learning strategy that learns sampling point displacement supervised by task-related ground truth information and can be trained jointly with the underlying tasks. We further demonstrate our methods in various point-wise analysis tasks, including semantic part segmentation, point cloud completion, and keypoint detection. Our experiments show that jointly learning of the sampler and task brings better performance than using FPS in various point-based networks. | ['Shuguang Cui', 'Xiaoguang Han', 'Chongyang Ma', 'Haibin Huang', 'Lichang Chen', 'Yiqun Lin'] | 2021-07-09 | null | null | null | null | ['point-cloud-completion'] | ['computer-vision'] | [ 7.64549002e-02 -6.32120669e-02 -2.44668230e-01 -4.81926054e-01
-1.18466473e+00 -4.17025924e-01 4.67550784e-01 3.70558411e-01
-2.06771359e-01 4.27749306e-01 -3.14294875e-01 -2.12816492e-01
-1.90196365e-01 -9.16762650e-01 -1.19178224e+00 -4.49056178e-01
-9.46641862e-02 8.65101814e-01 5.79570949e-01 2.07150891e-01
2.78019011e-01 1.17384052e+00 -1.43019032e+00 2.70559520e-01
1.04109275e+00 1.02466309e+00 4.51542109e-01 5.71346164e-01
-3.97627562e-01 3.71744275e-01 -4.06885594e-01 9.82623398e-02
3.57047200e-01 1.55064389e-01 -8.37585986e-01 2.25637823e-01
6.46304429e-01 -3.43819022e-01 3.36249381e-01 8.38373661e-01
2.28288412e-01 2.45616361e-01 8.11334491e-01 -1.32228613e+00
-7.62792379e-02 4.11794364e-01 -9.30635571e-01 -1.59864604e-01
-5.36258183e-02 2.37163417e-02 1.16163206e+00 -1.13812518e+00
3.69587421e-01 1.51931596e+00 1.00001287e+00 1.64163768e-01
-1.18865681e+00 -6.20617390e-01 5.01671970e-01 -1.77541092e-01
-1.19283342e+00 -4.83414046e-02 1.14312410e+00 -4.69491959e-01
5.85130394e-01 1.52362183e-01 6.88914001e-01 5.60335398e-01
-2.36911356e-01 1.23862982e+00 7.14027345e-01 -1.74022287e-01
4.51737225e-01 -1.05054162e-01 -4.94734459e-02 5.49397886e-01
2.11276710e-01 -3.51130724e-01 -3.64599317e-01 -2.79150635e-01
1.22193992e+00 4.04200882e-01 -2.94874441e-02 -9.26550984e-01
-1.17603469e+00 8.53975952e-01 7.27480829e-01 -1.83251843e-01
-5.40809095e-01 5.66074431e-01 2.69200474e-01 2.81025879e-02
1.06796515e+00 4.31215733e-01 -6.46081507e-01 1.40698567e-01
-1.26003683e+00 4.73765492e-01 4.59958106e-01 1.09583628e+00
1.29877615e+00 -2.32546166e-01 -1.11402024e-03 8.49283874e-01
3.59938741e-01 5.62562704e-01 -7.31068552e-02 -1.26876473e+00
5.06327152e-01 7.73682058e-01 2.46110603e-01 -6.49825454e-01
-4.09736276e-01 -3.65640461e-01 -6.24935448e-01 4.08575296e-01
4.02042776e-01 -1.72860920e-01 -9.76452112e-01 1.18271863e+00
7.39153802e-01 3.56451929e-01 -3.90859306e-01 5.59086263e-01
6.63034856e-01 5.73068142e-01 1.58401225e-02 2.54017860e-01
1.12925923e+00 -9.17196214e-01 -1.77854553e-01 -3.32253844e-01
6.34049892e-01 -5.67081094e-01 1.22447920e+00 4.06236470e-01
-1.04660320e+00 -6.27204597e-01 -7.81389534e-01 -2.26018026e-01
-2.38914490e-01 2.69895941e-01 7.45834112e-01 3.94892618e-02
-9.62795496e-01 8.82682502e-01 -1.04051542e+00 -1.71046034e-01
1.12937450e+00 1.81438997e-01 9.99509543e-02 1.24549806e-01
-4.68820274e-01 3.20791066e-01 6.81708846e-03 -1.91515520e-01
-8.46467614e-01 -1.16438580e+00 -7.23436117e-01 2.22091675e-02
4.37719196e-01 -7.37318277e-01 1.38576734e+00 -4.99773175e-01
-1.18519056e+00 8.42327118e-01 -4.40733522e-01 -5.84046483e-01
7.50593007e-01 -6.37779117e-01 5.84877431e-01 2.19848841e-01
3.94410789e-01 9.75909054e-01 9.73743737e-01 -1.51329732e+00
-9.60180283e-01 -4.73039687e-01 7.95452371e-02 2.45988354e-01
7.64110014e-02 -3.27453464e-01 -4.29830760e-01 -3.07334989e-01
4.09713477e-01 -5.85941494e-01 -5.02267361e-01 5.72659373e-01
-3.74854267e-01 -7.01303244e-01 1.09964800e+00 -3.34510386e-01
5.91519117e-01 -2.14240479e+00 -7.09906518e-02 2.13476196e-01
4.51026231e-01 -6.85191751e-02 7.05239698e-02 2.71305561e-01
9.51479822e-02 4.80815545e-02 -2.52457052e-01 -8.28854382e-01
-1.38717175e-01 1.17347881e-01 -3.58194977e-01 4.33907449e-01
4.61211681e-01 9.97266710e-01 -1.06531453e+00 -5.82743883e-01
5.74189782e-01 4.44808811e-01 -5.43755114e-01 1.11835925e-02
-5.85464656e-01 3.02408367e-01 -6.91357315e-01 8.96056473e-01
8.99226844e-01 -4.96746808e-01 -4.54035819e-01 -2.85109490e-01
-5.70720173e-02 1.97923169e-01 -1.19833469e+00 1.82561803e+00
-6.21296167e-01 6.12399578e-01 2.45183304e-01 -7.43172646e-01
8.88544977e-01 -4.29472029e-02 6.17445171e-01 -1.24881297e-01
-2.31100291e-01 1.77243829e-01 -5.25271416e-01 -2.10796054e-02
4.79365587e-01 3.14444751e-02 2.87926286e-01 2.27721617e-01
-9.89309251e-02 -7.66370177e-01 -5.16860411e-02 1.24272697e-01
8.03255260e-01 3.52492034e-01 2.77835280e-01 -3.30742359e-01
1.94868386e-01 1.59419373e-01 4.60299671e-01 7.85593092e-01
2.14998215e-03 8.56164336e-01 5.70655227e-01 -4.67232198e-01
-9.70521688e-01 -1.03095496e+00 -1.51699066e-01 1.05960441e+00
3.12302798e-01 -2.65047789e-01 -7.40298152e-01 -9.26454008e-01
3.89673710e-01 4.70503390e-01 -4.01524574e-01 4.36082065e-01
-7.07006693e-01 -3.02043319e-01 6.51058406e-02 8.82465184e-01
4.29795921e-01 -9.71837580e-01 -6.00402474e-01 4.85653058e-02
9.82952788e-02 -9.82485652e-01 -3.64112943e-01 3.04272532e-01
-1.36821699e+00 -1.21766686e+00 -8.19977939e-01 -5.85415483e-01
8.50264072e-01 7.57412374e-01 1.22628224e+00 -7.39655346e-02
4.14712019e-02 3.54948729e-01 -1.31490633e-01 -8.19841027e-01
-3.68879475e-02 5.13457537e-01 -4.19141203e-01 -2.28412464e-01
2.87881404e-01 -6.67239368e-01 -5.95676780e-01 2.28681237e-01
-7.24320769e-01 4.36960459e-02 6.00820661e-01 4.74971950e-01
9.88219321e-01 -3.93080339e-02 3.45957428e-01 -1.07702804e+00
4.37085271e-01 -2.67469972e-01 -7.07949698e-01 3.61546949e-02
-1.16656974e-01 -1.04383804e-01 3.71263951e-01 -1.55592382e-01
-6.59584165e-01 3.13875586e-01 -2.58347780e-01 -9.22980905e-01
-3.74200255e-01 6.63770884e-02 -1.78795353e-01 -2.48915210e-01
5.93145072e-01 3.17166075e-02 3.95768471e-02 -6.51520669e-01
4.54775125e-01 3.98118049e-01 1.41683385e-01 -7.57452607e-01
8.53113592e-01 1.08191860e+00 1.38417438e-01 -9.86977875e-01
-8.74069929e-01 -8.54484975e-01 -9.27815855e-01 -2.31572434e-01
7.89909065e-01 -1.14736485e+00 -5.64772606e-01 4.43683773e-01
-1.51480830e+00 -5.87267578e-01 -7.12628961e-01 8.42477083e-02
-6.96245492e-01 1.89207554e-01 -3.57615888e-01 -9.57165420e-01
-4.33963686e-01 -1.21801805e+00 1.89068091e+00 7.03660995e-02
-2.18465365e-02 -1.09062386e+00 -1.66352928e-01 4.37326729e-02
-4.40009087e-02 2.86689430e-01 5.92291772e-01 -5.04888058e-01
-1.07897866e+00 -8.73656869e-02 -4.75226253e-01 2.83742160e-01
2.81324565e-01 7.37418085e-02 -1.02920425e+00 -1.27778575e-01
-4.79675680e-02 -1.93336368e-01 9.88910556e-01 7.67144799e-01
1.84070361e+00 1.14877671e-01 -7.04165041e-01 9.16277945e-01
1.36561441e+00 -1.60786822e-01 2.82806814e-01 2.30047747e-01
1.12841797e+00 6.02110028e-01 8.61853361e-01 3.48628700e-01
4.40923125e-01 3.28451216e-01 6.96165383e-01 -3.69001836e-01
-7.46017173e-02 -3.92936826e-01 -1.85604945e-01 2.19983578e-01
-8.70310441e-02 -1.09927738e-02 -1.13445246e+00 7.16270864e-01
-1.96080637e+00 -6.79116726e-01 -3.30962151e-01 2.04393959e+00
6.02646589e-01 1.76491186e-01 2.87820071e-01 -6.73560053e-02
6.00063801e-01 3.73452157e-01 -7.21662223e-01 1.56907558e-01
1.95128188e-01 1.85393795e-01 7.36720324e-01 5.86508274e-01
-1.30428958e+00 1.02466536e+00 6.07104301e+00 1.10697544e+00
-8.76922429e-01 1.19262181e-01 7.10080087e-01 -6.27715066e-02
-3.41863960e-01 -1.12528667e-01 -9.86808002e-01 3.42614710e-01
2.55147874e-01 2.90227145e-01 9.54792053e-02 1.31704903e+00
3.74555081e-01 -4.67502102e-02 -1.31646669e+00 9.96432662e-01
-2.37930879e-01 -1.63057482e+00 1.54776797e-01 1.06959350e-01
6.81004584e-01 4.33022231e-01 -2.44382590e-01 8.63568410e-02
5.79378784e-01 -6.32206321e-01 7.18688488e-01 3.48122299e-01
6.71223044e-01 -7.63740957e-01 3.11576009e-01 5.01615405e-01
-1.36213422e+00 1.11701623e-01 -4.50541168e-01 9.25379023e-02
2.89446324e-01 1.01334500e+00 -8.49040687e-01 5.02151072e-01
6.72116160e-01 1.08215511e+00 -4.52179402e-01 1.12380552e+00
-2.17917830e-01 4.02270764e-01 -5.90543330e-01 8.88281763e-02
2.93647170e-01 -2.69332916e-01 6.09025478e-01 1.04842687e+00
1.30104870e-01 -3.82029593e-01 3.20464373e-01 1.11343515e+00
-9.13383067e-02 -1.70786574e-01 -6.98718548e-01 4.20117706e-01
7.41315246e-01 1.30769110e+00 -1.10388708e+00 -3.95267308e-01
-2.34943166e-01 6.62620246e-01 5.24329841e-01 3.44736725e-01
-6.57544017e-01 -3.00455809e-01 9.50364530e-01 4.52325076e-01
4.84073430e-01 -4.12041634e-01 -6.71828568e-01 -8.47015083e-01
1.00536399e-01 -3.30839604e-01 4.92676012e-02 -7.80114293e-01
-1.43741798e+00 7.62742087e-02 2.57515222e-01 -1.30435789e+00
3.93424835e-03 -6.06849849e-01 -8.68364453e-01 8.23905230e-01
-1.89242244e+00 -1.25111425e+00 -4.45954591e-01 2.86851048e-01
1.06714964e+00 2.10892141e-01 2.06657767e-01 1.53439134e-01
-2.17825234e-01 8.69071186e-02 4.38386314e-02 6.16517179e-02
4.42498952e-01 -1.57334435e+00 8.36553156e-01 6.84379458e-01
9.49951559e-02 5.92620492e-01 2.73060948e-01 -8.59679759e-01
-1.05886900e+00 -1.47067571e+00 4.19631422e-01 -5.22808135e-01
4.99776363e-01 -5.52650690e-01 -9.90878820e-01 6.39197707e-01
-3.33375633e-01 1.53141499e-01 1.69847533e-01 2.68495828e-01
-7.29061216e-02 -2.42787719e-01 -1.14385629e+00 3.82584423e-01
1.14082670e+00 -4.89790350e-01 -2.08607346e-01 6.96334302e-01
1.01696908e+00 -6.29308045e-01 -6.11304820e-01 4.90385592e-01
1.83461547e-01 -7.09164560e-01 1.37461662e+00 -3.08370024e-01
4.40662563e-01 -3.42390388e-01 3.52530442e-02 -1.38039851e+00
-2.54476666e-01 -4.21473682e-01 -2.00215295e-01 1.15350223e+00
2.18547806e-01 -6.19965851e-01 1.33367217e+00 2.91008711e-01
-3.43379021e-01 -9.43004727e-01 -8.98314536e-01 -7.33732522e-01
2.35343188e-01 -6.93945050e-01 1.03613138e+00 7.37416208e-01
-7.52973199e-01 1.32681385e-01 7.59243369e-02 4.62523520e-01
9.08387065e-01 3.20601881e-01 1.17966199e+00 -1.69627190e+00
1.09137539e-02 -5.21527827e-01 -1.79122522e-01 -1.62272847e+00
2.31350195e-02 -6.86160266e-01 1.41770944e-01 -1.95509851e+00
-1.14924967e-01 -8.00666273e-01 2.22098619e-01 2.40561977e-01
-2.90411681e-01 -8.00736174e-02 1.82941109e-01 5.23461103e-01
-6.68942988e-01 4.69213098e-01 1.40806818e+00 -1.80317611e-02
-3.94526243e-01 4.19251829e-01 -4.28997457e-01 9.67843354e-01
7.36091077e-01 -3.09577405e-01 -4.88318592e-01 -6.61876678e-01
1.55267164e-01 -1.69194445e-01 5.92476547e-01 -1.07175410e+00
1.72263712e-01 -2.21420810e-01 4.41347986e-01 -1.32108223e+00
5.75912416e-01 -8.57840419e-01 -2.47824371e-01 6.31376877e-02
-1.52698293e-01 -2.20305428e-01 9.14911479e-02 7.16395974e-01
-2.20794734e-02 -1.43795505e-01 6.26686275e-01 -3.97798985e-01
-5.00511110e-01 5.65981448e-01 1.96363494e-01 -7.02956989e-02
1.03532588e+00 -3.60468537e-01 8.23213682e-02 -1.22437045e-01
-5.61231077e-01 6.23894751e-01 5.55557549e-01 1.00373119e-01
6.25540614e-01 -1.14647055e+00 -5.22737563e-01 1.67072028e-01
-1.69857647e-02 1.19795012e+00 2.46297829e-02 6.42651439e-01
-6.50195241e-01 1.69223189e-01 3.74619871e-01 -1.26012218e+00
-1.11930001e+00 2.12786645e-01 3.24810445e-01 -5.39013185e-02
-7.10910559e-01 1.13685167e+00 4.68839735e-01 -7.90542662e-01
2.76862085e-01 -9.41305399e-01 1.12118199e-01 1.43171921e-01
1.33269057e-01 5.50638914e-01 1.34612843e-01 -1.30959138e-01
-1.91214785e-01 7.00564444e-01 -1.11066654e-01 1.44468114e-01
1.47018898e+00 5.05984984e-02 -5.03667966e-02 7.26624668e-01
9.60147977e-01 -1.76360622e-01 -1.72288227e+00 -3.83253425e-01
-6.99163899e-02 -6.18714929e-01 1.17399976e-01 -2.12893412e-01
-1.05679619e+00 1.11429262e+00 2.26307333e-01 3.17434937e-01
7.49512434e-01 3.39000076e-01 7.25257814e-01 3.10700059e-01
5.70578694e-01 -1.07599688e+00 -1.15842909e-01 3.71626884e-01
7.89898276e-01 -1.52666950e+00 1.93497568e-01 -8.36956799e-01
-1.93669379e-01 1.02763784e+00 6.89630747e-01 -4.09860283e-01
7.50507176e-01 1.57243386e-01 -2.11216331e-01 -5.29805422e-01
-4.53740120e-01 -2.78307736e-01 2.39074543e-01 6.37747705e-01
9.49062333e-02 -5.40758204e-03 2.32751831e-01 8.18703547e-02
-8.66835043e-02 1.46415448e-02 3.85157168e-02 1.01161432e+00
-6.63274944e-01 -9.13465619e-01 -5.28504610e-01 7.47934103e-01
-4.19887826e-02 1.33904368e-01 -3.07522446e-01 9.19442773e-01
8.34050477e-02 3.56940120e-01 4.04888004e-01 9.99172777e-02
3.73971820e-01 -7.58217946e-02 1.90265730e-01 -8.97598684e-01
-2.80034721e-01 1.17843732e-01 -2.97176957e-01 -7.76589692e-01
-4.60249275e-01 -8.83645177e-01 -1.33031714e+00 -3.58771868e-02
-3.24171990e-01 1.04968481e-01 7.86414027e-01 8.95497322e-01
4.86139923e-01 6.74817622e-01 6.20704710e-01 -1.57672131e+00
-3.91348451e-01 -7.90173888e-01 -4.62412477e-01 1.78407118e-01
5.67246616e-01 -6.42599523e-01 -4.16707277e-01 -1.25683263e-01] | [8.038382530212402, -3.4402241706848145] |
53e86f43-d43a-4069-9cd2-81aa3efa72e9 | real-time-optical-flow-for-vehicular | 2112.10591 | null | https://arxiv.org/abs/2112.10591v1 | https://arxiv.org/pdf/2112.10591v1.pdf | Real-Time Optical Flow for Vehicular Perception with Low- and High-Resolution Event Cameras | Event cameras capture changes of illumination in the observed scene rather than accumulating light to create images. Thus, they allow for applications under high-speed motion and complex lighting conditions, where traditional framebased sensors show their limits with blur and over- or underexposed pixels. Thanks to these unique properties, they represent nowadays an highly attractive sensor for ITS-related applications. Event-based optical flow (EBOF) has been studied following the rise in popularity of these neuromorphic cameras. The recent arrival of high-definition neuromorphic sensors, however, challenges the existing approaches, because of the increased resolution of the events pixel array and a much higher throughput. As an answer to these points, we propose an optimized framework for computing optical flow in real-time with both low- and high-resolution event cameras. We formulate a novel dense representation for the sparse events flow, in the form of the "inverse exponential distance surface". It serves as an interim frame, designed for the use of proven, state-of-the-art frame-based optical flow computation methods. We evaluate our approach on both low- and high-resolution driving sequences, and show that it often achieves better results than the current state of the art, while also reaching higher frame rates, 250Hz at 346 x 260 pixels and 77Hz at 1280 x 720 pixels. | ['Franck Davoine', 'Julien Moreau', 'Vincent Brebion'] | 2021-12-20 | null | null | null | null | ['event-based-optical-flow'] | ['computer-vision'] | [ 4.16673094e-01 -6.16136611e-01 3.07002902e-01 4.99013737e-02
-1.27122357e-01 -3.38497430e-01 6.36772931e-01 -3.36721204e-02
-9.66427743e-01 9.29669917e-01 1.03335828e-01 3.30174923e-01
-3.98473293e-02 -6.66719079e-01 -7.00539649e-01 -7.13568151e-01
-5.23343608e-02 -1.43955514e-01 6.56539857e-01 1.45809308e-01
4.18546110e-01 5.63272953e-01 -2.31843114e+00 2.29700908e-01
4.93797034e-01 1.15514255e+00 3.74210715e-01 6.24234796e-01
6.81623593e-02 9.04608428e-01 -4.20866668e-01 -3.39909732e-01
2.24404052e-01 -5.08964062e-01 -2.40398735e-01 -1.13333091e-01
7.98623562e-01 -5.77929199e-01 -6.51999414e-01 1.06632650e+00
4.21938896e-01 1.22067764e-01 7.98332617e-02 -9.38667893e-01
-3.26384664e-01 8.62059221e-02 -3.42189103e-01 8.16936612e-01
6.17133915e-01 4.34220701e-01 6.58630908e-01 -8.76994252e-01
1.03405964e+00 9.10433650e-01 3.75592232e-01 7.79037833e-01
-1.38689232e+00 -3.38089436e-01 -2.57194906e-01 6.44893765e-01
-1.00769341e+00 -5.88355482e-01 6.44024014e-01 -3.12902033e-01
1.12222147e+00 1.46407224e-02 1.12971592e+00 1.15300834e+00
6.01416051e-01 4.35119957e-01 1.20052993e+00 -4.98154536e-02
4.77907926e-01 -3.73943090e-01 2.81501067e-04 6.18035257e-01
4.63471144e-01 2.59001821e-01 -1.11188507e+00 1.31824851e-01
1.13954997e+00 1.42736539e-01 -7.62670755e-01 -2.97046825e-02
-1.38686967e+00 3.30584973e-01 2.66490549e-01 3.26593280e-01
-4.96386766e-01 5.76366127e-01 2.24774435e-01 2.99659632e-02
1.19798467e-01 2.52228796e-01 1.84349671e-01 -7.28852570e-01
-9.64343667e-01 2.64393210e-01 6.25228524e-01 5.53535819e-01
8.36591899e-01 1.63812116e-01 -1.56760048e-02 2.43533030e-01
1.41852289e-01 4.09105390e-01 6.13936424e-01 -1.13614643e+00
6.22034036e-02 3.21377307e-01 1.57052889e-01 -8.94467294e-01
-2.96345770e-01 -2.41981223e-01 -8.79829645e-01 4.62044567e-01
5.85558116e-01 1.31113753e-01 -5.21370769e-01 1.64057136e+00
1.22215904e-01 6.62063301e-01 -1.45830661e-01 1.19747865e+00
2.68153876e-01 6.96865439e-01 -2.60738581e-01 -6.04828179e-01
1.38978541e+00 -4.01773781e-01 -1.01437318e+00 -1.04224585e-01
-9.53833088e-02 -4.58941281e-01 9.26871717e-01 6.74967527e-01
-1.32114840e+00 -4.98405874e-01 -1.02771449e+00 -2.18458220e-01
-1.50721103e-01 -1.74929529e-01 6.17657542e-01 6.81647599e-01
-1.23330057e+00 7.22832143e-01 -1.10007000e+00 -3.46315593e-01
5.71967185e-01 3.69080663e-01 -2.53716379e-01 7.30304793e-03
-7.56262839e-01 6.98624253e-01 -9.55941454e-02 4.91494052e-02
-8.11496794e-01 -9.50561285e-01 -5.31896412e-01 7.07977861e-02
-6.83874041e-02 -6.96795166e-01 8.53992462e-01 -8.32679808e-01
-1.80116498e+00 8.43081772e-01 -2.74150997e-01 -7.95131922e-01
4.87091959e-01 -3.10244173e-01 -2.82438576e-01 7.05380380e-01
-3.47977459e-01 5.87816715e-01 9.54898834e-01 -6.30693257e-01
-5.06148458e-01 -5.75947702e-01 7.28099719e-02 -1.20453738e-01
-5.24040580e-01 7.59412050e-02 -1.44610927e-01 -2.86622316e-01
-2.74482876e-01 -6.57841444e-01 1.74831375e-02 5.12666404e-01
3.24171871e-01 1.70134932e-01 9.73118842e-01 -1.39937341e-01
1.01445210e+00 -2.10609341e+00 5.38396798e-02 -4.13450360e-01
2.71009445e-01 5.64162433e-01 1.73391670e-01 2.40816429e-01
1.18159167e-01 -5.24772286e-01 -2.75057703e-01 -3.56815070e-01
-4.32411462e-01 2.47988030e-01 -2.73994416e-01 7.11416364e-01
4.08393711e-01 7.63259888e-01 -9.92274761e-01 -2.95854121e-01
7.19141781e-01 9.85416532e-01 -5.47946513e-01 1.01408839e-01
-2.69568283e-02 6.49525523e-01 -8.43277574e-02 1.43259168e-01
5.17293215e-01 -1.21030338e-01 -1.77185342e-01 -3.38712692e-01
-7.41085827e-01 1.24738924e-01 -1.33035696e+00 2.12194777e+00
-1.36714280e-01 1.17733228e+00 3.16186659e-02 -6.90233052e-01
1.00276613e+00 2.15482786e-01 8.23346376e-01 -1.01757944e+00
3.30894768e-01 3.06220084e-01 -1.98896408e-01 -5.38717806e-01
5.54536521e-01 -7.84286186e-02 4.94241118e-01 1.72494754e-01
2.39355385e-01 6.72489929e-04 5.39900482e-01 -8.10431018e-02
1.43851018e+00 2.52910823e-01 6.42352551e-02 -3.36263508e-01
5.79712689e-01 -5.17403781e-01 4.22365874e-01 4.51523393e-01
-5.01782954e-01 7.25517929e-01 2.09656715e-01 -6.95412934e-01
-9.11255181e-01 -1.06807268e+00 -4.61252928e-01 2.78011650e-01
3.70646238e-01 -4.08763975e-01 -8.51771712e-01 2.33545974e-01
-3.58353913e-01 9.58712101e-02 -3.16300452e-01 1.73061378e-02
-6.80968463e-01 -8.04304063e-01 3.62571359e-01 1.63718298e-01
7.87531435e-01 -1.04495633e+00 -1.73649573e+00 5.29599428e-01
-2.13162741e-03 -1.61583841e+00 2.61404607e-02 -3.97382528e-02
-9.62054372e-01 -9.98736382e-01 -8.29910874e-01 -4.53134418e-01
3.69052619e-01 1.37735337e-01 9.11051631e-01 -4.08095986e-01
-6.51912093e-01 4.62637931e-01 -1.14689834e-01 -1.10728391e-01
1.34653389e-01 -5.16901970e-01 5.60564324e-02 5.80846786e-01
1.22927167e-01 -1.00149691e+00 -1.10773218e+00 3.77809070e-02
-1.25199544e+00 -3.31156477e-02 2.72951484e-01 5.54203331e-01
5.87895513e-01 -3.86695206e-01 2.55516648e-01 -4.71026391e-01
2.08313718e-01 -6.54541254e-02 -9.31678891e-01 -2.51043022e-01
-4.19050723e-01 3.87179218e-02 8.42921972e-01 -4.63937879e-01
-1.02820015e+00 2.07424209e-01 -6.43610060e-02 -5.50046384e-01
-1.21294506e-01 1.48632415e-02 3.17078352e-01 -3.56790185e-01
7.88051248e-01 2.86385387e-01 -8.51148888e-02 -1.26829773e-01
-5.41812070e-02 3.29075307e-01 9.79663134e-01 -3.13652813e-01
4.48618323e-01 1.15749669e+00 2.86796361e-01 -1.01141977e+00
-3.20694715e-01 -4.50135410e-01 -4.02524978e-01 -6.52106464e-01
9.05031323e-01 -8.07270467e-01 -9.88099277e-01 8.94097030e-01
-1.42698038e+00 -2.14224100e-01 -5.79483628e-01 6.74694300e-01
-6.68292582e-01 3.35112989e-01 -7.74128914e-01 -9.57090735e-01
-2.23642960e-01 -1.18283796e+00 1.03677356e+00 6.02128804e-01
1.61743104e-01 -7.75759816e-01 2.40509480e-01 -8.52167085e-02
6.48502350e-01 4.62106556e-01 3.48215997e-01 3.96200866e-01
-1.03156030e+00 4.09517735e-02 -2.94119209e-01 2.12833077e-01
-1.62793070e-01 1.38125747e-01 -1.26978242e+00 -8.24643373e-02
3.34982246e-01 -4.39738482e-02 8.21427941e-01 4.17888910e-01
1.14702106e+00 2.45605588e-01 4.92861345e-02 7.62947261e-01
1.92871726e+00 4.02159318e-02 1.07474983e+00 1.30132645e-01
3.99731129e-01 3.66373718e-01 9.17805731e-02 8.84155929e-01
1.64889172e-01 7.71584749e-01 7.32688904e-01 2.90443063e-01
-4.52729076e-01 6.85904473e-02 6.10569358e-01 6.68291032e-01
-4.35975403e-01 -1.61263838e-01 -6.41080201e-01 5.66675425e-01
-1.75244343e+00 -1.19598556e+00 -5.06032825e-01 2.58438778e+00
6.57064378e-01 1.19694263e-01 9.12851021e-02 2.86687702e-01
6.93086803e-01 2.65288293e-01 -4.78178918e-01 -2.29415074e-01
-4.06878918e-01 6.42162383e-01 3.42629850e-01 1.69346258e-01
-6.96499407e-01 4.81687307e-01 5.74578142e+00 2.97187895e-01
-1.42388892e+00 1.01082385e-01 2.15583265e-01 -3.85990292e-01
-8.59560519e-02 -9.55728069e-03 -8.39223087e-01 8.45205545e-01
1.30979955e+00 -1.12630047e-01 6.34423852e-01 2.81709045e-01
4.41938311e-01 -4.82888997e-01 -1.07121682e+00 1.58553648e+00
2.63344347e-01 -1.64852154e+00 -5.31539507e-02 2.07716212e-01
6.00184262e-01 1.37909472e-01 2.47747898e-02 -4.55344230e-01
-4.30983365e-01 -5.36731184e-01 8.52138638e-01 6.88300192e-01
9.10262704e-01 -3.73452008e-01 3.18521589e-01 1.30158499e-01
-1.19002736e+00 -1.51780948e-01 -4.60703671e-01 -4.65980381e-01
4.67819214e-01 1.02030623e+00 4.08726074e-02 1.94625884e-01
8.33053172e-01 1.10455632e+00 -2.49855801e-01 1.23712492e+00
9.21364129e-02 2.84590006e-01 -5.55536985e-01 -1.36125967e-01
-2.41257120e-02 -2.12767959e-01 6.15701914e-01 1.04695380e+00
4.92187828e-01 2.78793126e-02 -3.47062171e-01 1.08907521e+00
4.44491208e-02 -3.89396340e-01 -5.90485990e-01 2.35022709e-01
1.47709340e-01 1.25804448e+00 -7.04762280e-01 -1.19431034e-01
-5.71850181e-01 1.10630703e+00 1.80524677e-01 2.15856671e-01
-8.11850309e-01 -3.83125782e-01 8.36813569e-01 1.69946581e-01
1.81493834e-01 -6.52298689e-01 -4.39309925e-02 -1.47819316e+00
3.02513987e-01 -3.58366132e-01 -8.54256526e-02 -7.78992057e-01
-7.87510931e-01 4.67511743e-01 -3.40022475e-01 -1.24045074e+00
-8.06916356e-02 -8.66391778e-01 -3.24736834e-01 4.04094011e-01
-1.91587710e+00 -4.34103519e-01 -6.79376245e-01 9.15810168e-01
3.98494750e-01 2.20875815e-01 7.28412747e-01 6.07389569e-01
-4.83080864e-01 9.48614255e-03 1.96883485e-01 -1.65448993e-01
5.87269485e-01 -9.22933877e-01 1.96889758e-01 1.26476240e+00
4.03038114e-01 2.20567405e-01 6.15722895e-01 -1.88934118e-01
-2.05143714e+00 -7.68720925e-01 6.60074055e-01 -2.22585425e-01
6.68486893e-01 -5.02390385e-01 -7.79366136e-01 1.61252066e-01
1.31007090e-01 5.51904976e-01 2.85507202e-01 -6.53458953e-01
-8.89556110e-02 -5.06272554e-01 -1.11532176e+00 4.27298695e-01
1.23008013e+00 -7.10922837e-01 -3.17067623e-01 -6.94159046e-02
1.96325168e-01 -3.50901693e-01 -7.72035658e-01 1.38621509e-01
6.47904873e-01 -1.56789219e+00 7.50954449e-01 2.72150457e-01
5.52223980e-01 -4.22998816e-01 7.60523677e-02 -8.91200781e-01
-1.36626154e-01 -1.12792838e+00 -3.76786739e-01 8.47986758e-01
-2.47716919e-01 -8.20954740e-01 8.74725580e-01 3.58315319e-01
-1.72514275e-01 -4.76018697e-01 -1.39887071e+00 -6.21938646e-01
-5.57975531e-01 -4.64779437e-01 6.35966510e-02 4.40497637e-01
5.26017398e-02 1.02834508e-01 -1.15411572e-01 -1.20364524e-01
7.88077831e-01 -1.37351803e-03 3.54166687e-01 -1.14272559e+00
-3.16119701e-01 -4.31339771e-01 -1.15295565e+00 -1.10885096e+00
-5.08884527e-02 -3.48236412e-01 1.31127136e-02 -1.01901555e+00
-4.27930802e-02 1.47958234e-01 -1.79789335e-01 -9.08560082e-02
1.31583825e-01 6.56846046e-01 2.84586877e-01 1.82990700e-01
-5.63340962e-01 5.21799386e-01 1.09197104e+00 3.08371216e-01
1.65558923e-02 -5.64422786e-01 7.33428523e-02 5.30622303e-01
2.72020221e-01 -1.29158005e-01 -2.51154035e-01 -6.92326486e-01
3.16935480e-01 -5.80521710e-02 7.50005364e-01 -1.80768979e+00
6.03528023e-01 2.51361489e-01 3.33558142e-01 -1.33522928e-01
6.28856242e-01 -8.33624899e-01 3.20904791e-01 6.35711253e-01
-1.92393318e-01 2.38828659e-01 5.25957569e-02 5.58050990e-01
-3.48607302e-01 -1.63258389e-01 9.14670587e-01 -1.40121639e-01
-8.34251940e-01 4.04808134e-01 -5.62147379e-01 8.75854492e-02
9.28080082e-01 -5.95388889e-01 -7.04693913e-01 -9.80152711e-02
-2.39075884e-01 -4.97739911e-01 6.20582402e-01 1.60994813e-01
8.99810374e-01 -9.79884386e-01 -4.47918355e-01 5.55677176e-01
-1.27720818e-01 -1.26013845e-01 2.24623293e-01 9.16446388e-01
-7.48158157e-01 1.36919305e-01 -8.85120451e-01 -9.13047254e-01
-8.25648785e-01 2.92931885e-01 2.27728754e-01 9.03119370e-02
-9.04149234e-01 6.21134400e-01 4.82459776e-02 7.35459089e-01
1.21149406e-01 -4.01462972e-01 -7.96985254e-02 4.55284026e-03
8.96021843e-01 6.23778582e-01 1.84118763e-01 -4.49893206e-01
-2.35053852e-01 8.42376649e-01 4.28381085e-01 -2.85092354e-01
1.32999825e+00 -2.03564525e-01 -6.86402246e-02 7.35560596e-01
1.05424762e+00 -2.55385071e-01 -1.81511319e+00 1.29408780e-02
-1.79300219e-01 -7.98074543e-01 1.64932817e-01 -1.22135051e-01
-1.06228316e+00 1.23985219e+00 8.33463073e-01 2.93816894e-01
1.38371003e+00 -4.22580123e-01 1.03168917e+00 5.73419072e-02
7.49623477e-01 -8.58983576e-01 1.29572839e-01 4.34037030e-01
4.12338704e-01 -7.96044409e-01 -1.52254209e-01 -3.42031389e-01
4.82614376e-02 1.30760229e+00 3.20125431e-01 -5.28968394e-01
3.65418643e-01 6.41169250e-01 -3.77524853e-01 -9.16169398e-03
-8.86593878e-01 -2.60827214e-01 -2.25008950e-01 6.47183061e-01
2.84998834e-01 -3.34589303e-01 -3.68150592e-01 -1.26293823e-01
5.09057604e-02 4.78342533e-01 9.33262229e-01 7.65024066e-01
-4.12816793e-01 -8.63553345e-01 -1.92154765e-01 3.28196496e-01
-6.15204394e-01 4.08071317e-02 1.31706417e-01 3.11990619e-01
1.67372555e-01 6.65020764e-01 3.71519715e-01 -8.78078118e-02
3.50509435e-01 -1.03163593e-01 9.43218648e-01 -1.98543057e-01
-6.86688960e-01 -1.09679945e-01 -2.86440015e-01 -1.11565042e+00
-9.20980096e-01 -8.87412906e-01 -1.23305225e+00 -2.27648914e-01
5.27995899e-02 -3.91029656e-01 8.73837411e-01 7.60013640e-01
5.93564630e-01 3.48933399e-01 4.76579398e-01 -1.10079265e+00
-8.39740559e-02 -5.00216424e-01 -5.23620069e-01 6.81162477e-01
5.30619979e-01 -5.30640364e-01 -3.92800987e-01 4.30968046e-01] | [8.6663179397583, -1.2876839637756348] |
be108769-f303-480c-8134-80b345b1bbc9 | multi-scale-self-calibrated-network-for-image | 2104.08838 | null | https://arxiv.org/abs/2104.08838v1 | https://arxiv.org/pdf/2104.08838v1.pdf | Multi-scale Self-calibrated Network for Image Light Source Transfer | Image light source transfer (LLST), as the most challenging task in the domain of image relighting, has attracted extensive attention in recent years. In the latest research, LLST is decomposed three sub-tasks: scene reconversion, shadow estimation, and image re-rendering, which provides a new paradigm for image relighting. However, many problems for scene reconversion and shadow estimation tasks, including uncalibrated feature information and poor semantic information, are still unresolved, thereby resulting in insufficient feature representation. In this paper, we propose novel down-sampling feature self-calibrated block (DFSB) and up-sampling feature self-calibrated block (UFSB) as the basic blocks of feature encoder and decoder to calibrate feature representation iteratively because the LLST is similar to the recalibration of image light source. In addition, we fuse the multi-scale features of the decoder in scene reconversion task to further explore and exploit more semantic information, thereby providing more accurate primary scene structure for image re-rendering. Experimental results in the VIDIT dataset show that the proposed approach significantly improves the performance for LLST. | ['Yuntao Wu', 'Yanduo Zhang', 'Tao Lu', 'Yuanzhi Wang'] | 2021-04-18 | null | null | null | null | ['image-relighting'] | ['computer-vision'] | [ 6.93204701e-01 -4.79792982e-01 1.59294903e-01 -3.92107219e-01
-2.70362586e-01 -2.97506899e-01 5.00368059e-01 -2.19050050e-01
-1.74040094e-01 6.79572940e-01 3.57531786e-01 7.62965307e-02
2.79180318e-01 -7.13289142e-01 -7.96408892e-01 -9.16179717e-01
8.64864290e-01 -1.48534670e-01 5.31744719e-01 -3.10948133e-01
3.92098218e-01 4.79648978e-01 -1.50777626e+00 1.69193074e-01
1.15313709e+00 9.33510602e-01 7.03840673e-01 4.06080097e-01
-2.72231996e-01 9.28892970e-01 -3.72510016e-01 -1.82139337e-01
1.15957022e-01 -5.98036051e-01 -5.87060332e-01 3.50228906e-01
3.70305508e-01 -6.74689710e-01 -5.79707980e-01 1.18389630e+00
4.57890719e-01 3.01885635e-01 3.73102248e-01 -1.27323377e+00
-4.56034273e-01 7.16471672e-02 -9.87272084e-01 1.95874915e-01
4.67955828e-01 6.12604432e-02 2.04206884e-01 -9.90902841e-01
5.40397406e-01 1.51860881e+00 3.16447020e-01 1.16564766e-01
-1.09799242e+00 -1.03385711e+00 -4.48067598e-02 5.24479330e-01
-1.27964067e+00 -5.97145021e-01 1.09296834e+00 -2.00145856e-01
2.85354614e-01 4.40136045e-01 7.45023072e-01 6.38406754e-01
2.50072867e-01 8.02727401e-01 1.51222754e+00 -2.88121074e-01
-1.60333738e-02 2.96899229e-01 -1.66592255e-01 6.28251493e-01
1.60777569e-01 1.76873475e-01 -6.62487209e-01 9.97505188e-02
1.04378998e+00 1.31118283e-01 -7.69423366e-01 -4.22261477e-01
-1.16089737e+00 4.85501021e-01 7.45557666e-01 4.96309735e-02
-3.15569304e-02 7.00499266e-02 3.00649762e-01 5.67229800e-02
4.06540185e-01 -9.97876748e-02 -2.37535790e-01 8.33206698e-02
-6.61911786e-01 -7.43376017e-02 3.99496555e-01 1.02135491e+00
1.18693268e+00 2.01236278e-01 -2.86820352e-01 9.41343904e-01
1.44977197e-01 7.14938581e-01 4.30614382e-01 -8.55189979e-01
5.60636818e-01 4.38824713e-01 2.07559690e-01 -1.34148216e+00
-1.36968121e-01 -4.51747060e-01 -1.10874856e+00 2.17944030e-02
-6.81220368e-02 1.45307124e-01 -7.33868897e-01 1.35934734e+00
7.19848394e-01 7.19929039e-01 1.03293195e-01 1.03456962e+00
9.33064520e-01 9.45905089e-01 -1.63564876e-01 -4.48724985e-01
1.33650243e+00 -1.08348334e+00 -6.99892342e-01 -2.23652631e-01
-2.55796616e-03 -1.09019625e+00 1.11593914e+00 3.05146903e-01
-7.02099204e-01 -8.74570549e-01 -9.56380665e-01 -5.65060616e-01
8.17718059e-02 4.72652875e-02 6.96376085e-01 3.51684451e-01
-7.07909107e-01 2.02837795e-01 -3.31672758e-01 -2.10556611e-01
4.33471411e-01 -5.25271259e-02 -3.01646054e-01 -4.34091747e-01
-1.09182608e+00 5.92744291e-01 5.63480794e-01 2.70744324e-01
-8.01740944e-01 -7.54250824e-01 -7.37792432e-01 -8.48467946e-02
4.89137471e-01 -6.14110291e-01 8.69001627e-01 -9.74006355e-01
-1.59685111e+00 5.44406056e-01 -3.39355707e-01 -2.19959076e-02
5.00047028e-01 -5.34958579e-02 -4.93831277e-01 2.36421898e-02
1.52155325e-01 4.21061456e-01 1.24480999e+00 -1.64466500e+00
-6.63529575e-01 -3.98209661e-01 -2.29334652e-01 7.34167993e-01
-1.50851488e-01 -1.74228951e-01 -7.13274360e-01 -8.51331711e-01
2.28977188e-01 -7.10748434e-01 9.21972934e-03 1.24389447e-01
-4.54834491e-01 2.73493439e-01 1.02360916e+00 -7.68579900e-01
9.61644053e-01 -2.22941756e+00 1.55715734e-01 -4.40905280e-02
9.27902982e-02 2.13255376e-01 -1.71454296e-01 2.80825198e-01
-2.55578101e-01 -5.23749590e-01 -3.07530940e-01 -1.55812904e-01
-5.43823659e-01 2.90293515e-01 -7.11554468e-01 4.62408483e-01
-2.16360316e-01 7.26368845e-01 -1.00147331e+00 -6.40715837e-01
7.34703124e-01 6.35812879e-01 -3.42131972e-01 3.27225089e-01
7.24740699e-02 9.56884027e-01 -5.63805640e-01 3.93433124e-01
1.30920935e+00 1.37979537e-01 -3.14599574e-01 -8.08755398e-01
-2.49124169e-01 -2.22286656e-01 -1.29871833e+00 1.95447922e+00
-6.59990191e-01 6.86693609e-01 -3.06805849e-01 -5.36714911e-01
8.44802797e-01 -3.98584932e-01 4.37748969e-01 -9.13357437e-01
2.31994748e-01 2.21868958e-02 -5.22038698e-01 -5.89403331e-01
5.71234345e-01 1.10638559e-01 2.61503071e-01 2.31574550e-01
-4.60487962e-01 -5.17225802e-01 -2.06286147e-01 2.23283827e-01
4.99340028e-01 3.59435648e-01 2.82758236e-01 3.25975828e-02
8.70880842e-01 -1.51330382e-01 7.34019220e-01 1.76576853e-01
8.96984860e-02 7.73020446e-01 1.35055169e-01 -2.08316803e-01
-8.88756454e-01 -9.07594621e-01 -2.41791755e-01 7.54700065e-01
1.00882483e+00 -3.19858581e-01 -8.29857230e-01 -3.80560935e-01
-8.36893469e-02 1.00603890e+00 -3.90479028e-01 -3.65966469e-01
-5.82738876e-01 -7.89384782e-01 -4.85192165e-02 2.33934879e-01
1.25512230e+00 -9.81898189e-01 -6.47754610e-01 1.61873266e-01
-5.96902013e-01 -1.15050197e+00 -8.28663349e-01 -3.15636337e-01
-5.90448201e-01 -1.10258663e+00 -6.63644910e-01 -6.40027702e-01
7.26461411e-01 1.16366160e+00 7.28892744e-01 2.35362858e-01
-5.13159037e-01 2.58922335e-02 -3.21387440e-01 -7.82925636e-02
-3.87930214e-01 -4.18682009e-01 -2.58113444e-01 5.02335668e-01
-2.09959716e-01 -5.02809823e-01 -9.67269421e-01 5.36008716e-01
-1.20549607e+00 8.15528512e-01 6.11036837e-01 7.33379781e-01
6.20309889e-01 2.90117323e-01 7.24882334e-02 -1.08229566e+00
2.47641698e-01 -1.16155371e-01 -5.88905632e-01 2.99032599e-01
-4.93635774e-01 2.75605936e-02 6.49935782e-01 -2.10501835e-01
-1.81244397e+00 -2.78594568e-02 -4.13323268e-02 -4.34316516e-01
6.90718293e-02 -1.00374594e-01 -4.45426852e-01 -5.53188860e-01
3.55784357e-01 8.10139298e-01 -2.03529358e-01 -5.34858048e-01
4.79790956e-01 7.01439619e-01 6.66399121e-01 -2.36258298e-01
1.19704103e+00 7.66270936e-01 1.64498314e-01 -7.24929094e-01
-9.09987986e-01 -2.76081175e-01 -5.32989919e-01 -2.84567773e-01
6.52198017e-01 -1.06324553e+00 -7.01859772e-01 8.05616200e-01
-1.08235526e+00 -1.09533966e-01 -3.14706355e-01 1.95132494e-01
-4.86873806e-01 7.94556737e-01 -1.54463559e-01 -2.74260163e-01
-3.36256891e-01 -1.33193088e+00 1.28917348e+00 6.11363113e-01
5.58213294e-01 -5.63769698e-01 -8.44055340e-02 5.15153944e-01
3.21368665e-01 3.29801559e-01 8.66434336e-01 4.31961179e-01
-8.82290721e-01 2.85274267e-01 -7.66849220e-01 4.01297480e-01
4.18009967e-01 -2.96244740e-01 -9.51268971e-01 -4.04035658e-01
1.10901043e-01 -9.66435373e-02 9.29770291e-01 4.81697731e-02
1.39629531e+00 1.93493329e-02 -4.06674147e-01 1.13553989e+00
1.53561258e+00 1.77387044e-01 8.21585357e-01 3.24799716e-01
1.06448495e+00 3.15302610e-01 1.12321901e+00 5.48927546e-01
5.02214432e-01 8.25229347e-01 4.62006003e-01 -3.82664800e-01
-7.81865656e-01 -5.70389807e-01 2.58033186e-01 7.20720291e-01
8.37336406e-02 -1.07154481e-01 -4.89496678e-01 -4.64061461e-02
-1.72651160e+00 -6.72597766e-01 -3.77973288e-01 2.14933181e+00
8.51218104e-01 -3.13726127e-01 -4.84525204e-01 -7.83296302e-02
7.89060533e-01 4.44866031e-01 -9.31961238e-01 1.21631332e-01
-2.78118193e-01 -2.46554583e-01 5.27009368e-01 5.39968431e-01
-7.01943517e-01 1.14851940e+00 4.93046761e+00 1.29898787e+00
-9.70450640e-01 1.89025685e-01 5.88130534e-01 3.43944162e-01
-4.96149033e-01 3.92457038e-01 -6.51202321e-01 7.79648423e-01
-2.88032424e-02 -1.82576403e-01 9.18858886e-01 5.20413816e-01
3.26970220e-01 -6.44755363e-01 -6.54427409e-01 1.42708588e+00
4.71906006e-01 -8.91740739e-01 3.11841220e-01 -3.25596660e-01
8.66305351e-01 -2.81816334e-01 8.55419859e-02 -5.57098538e-04
8.67133066e-02 -4.46380228e-01 6.95638776e-01 5.20615935e-01
1.24957860e+00 -7.23255217e-01 2.87271976e-01 1.97758391e-01
-1.48716438e+00 -1.43301338e-01 -7.29982734e-01 4.83304054e-01
2.52411515e-01 6.94812238e-01 -6.21785283e-01 8.31083536e-01
7.15969801e-01 1.07701504e+00 -7.61589289e-01 1.01929903e+00
-2.93334246e-01 1.89270079e-01 -7.31417313e-02 4.75186169e-01
-2.92871416e-01 -4.50059682e-01 5.48626065e-01 7.38080263e-01
1.91604704e-01 2.86381245e-01 9.52580571e-02 6.32227004e-01
3.45694497e-02 1.48742735e-01 -1.52660415e-01 5.08414328e-01
3.13767135e-01 1.26686919e+00 -6.84748769e-01 -3.20221186e-01
-2.45405883e-01 1.46565044e+00 1.34054884e-01 5.89389563e-01
-1.06467462e+00 -3.99532527e-01 3.75298887e-01 1.29132017e-01
-1.50609910e-01 5.94660603e-02 1.41191229e-01 -1.53313816e+00
-1.95840925e-01 -5.82668185e-01 7.07591549e-02 -1.51048350e+00
-7.94152081e-01 5.64250588e-01 1.19396657e-01 -1.43882501e+00
1.44001380e-01 -5.15778847e-02 -3.64878327e-01 8.32222164e-01
-2.13620305e+00 -1.42575049e+00 -1.21125638e+00 1.00211072e+00
9.64778781e-01 1.77096054e-01 1.85064122e-01 2.68516302e-01
-6.19540572e-01 2.61969715e-01 5.00206053e-01 -2.71444976e-01
8.54256511e-01 -7.37478375e-01 1.61992595e-01 9.21510160e-01
-1.80794895e-01 8.33064988e-02 6.53705537e-01 -7.76127338e-01
-1.53831875e+00 -1.43843794e+00 3.06062847e-02 1.10017255e-01
1.38625458e-01 -2.95353293e-01 -8.44342649e-01 4.44628358e-01
6.41415641e-02 1.37568545e-02 3.92039642e-02 -6.26929879e-01
-1.52960777e-01 -4.97972012e-01 -1.03408611e+00 6.34548068e-01
1.21260631e+00 -3.16841036e-01 -2.19364151e-01 5.02161205e-01
7.54544854e-01 -6.83795094e-01 -4.29450184e-01 2.80165315e-01
4.24235255e-01 -1.23436511e+00 1.24604642e+00 1.80202723e-01
2.40459695e-01 -6.97809398e-01 -1.34953335e-01 -1.30801642e+00
-2.58916587e-01 -4.83142346e-01 2.11115316e-01 1.44719851e+00
-5.38722992e-01 -8.65933895e-01 3.95336509e-01 1.42814741e-01
-6.20581917e-02 -4.71074343e-01 -7.42283583e-01 -3.15906137e-01
-5.82336187e-01 -5.78941815e-02 9.49011207e-01 8.13241363e-01
-6.86739028e-01 3.32774490e-01 -6.16517544e-01 5.65584525e-02
8.85453999e-01 7.32687831e-01 1.16342044e+00 -9.75384593e-01
-2.08386034e-01 7.56960437e-02 -2.17457950e-01 -1.29576325e+00
9.89578292e-02 -7.14840174e-01 9.33779329e-02 -1.54210186e+00
6.37198091e-01 -6.17518783e-01 -4.91317734e-03 2.31011167e-01
-4.66305763e-01 4.40814704e-01 2.14396700e-01 4.34960127e-01
-2.85965651e-01 1.00867486e+00 1.67114866e+00 -8.46524760e-02
-1.00678794e-01 -2.05534950e-01 -6.16653919e-01 7.51811922e-01
4.91093397e-01 -3.61443847e-01 -7.52494514e-01 -6.28591895e-01
-5.98449484e-02 2.71184742e-01 5.14448345e-01 -9.04230773e-01
1.77349299e-01 -4.92205828e-01 6.89054012e-01 -8.25658202e-01
5.44867992e-01 -9.82411742e-01 5.09976864e-01 3.17834973e-01
2.42329184e-02 -3.00768495e-01 4.78115231e-02 9.00047004e-01
-2.06428900e-01 9.05015040e-03 9.89060462e-01 4.03017476e-02
-9.95512545e-01 4.60360706e-01 2.06635118e-01 -1.06793493e-01
1.18438244e+00 -5.25944889e-01 -3.42486441e-01 -2.72152662e-01
-1.13853395e-01 1.03350915e-01 6.65432870e-01 6.14761829e-01
1.02850842e+00 -1.27343452e+00 -6.67068303e-01 6.09110773e-01
1.10987946e-01 2.88408011e-01 8.34780455e-01 6.52971148e-01
-7.60563850e-01 -7.43308961e-02 -2.63537109e-01 -6.02870941e-01
-1.31734216e+00 6.52069449e-01 1.73481971e-01 1.63220182e-01
-8.79509747e-01 6.88888550e-01 8.93247783e-01 1.22239381e-01
-2.21861690e-01 -1.84105158e-01 -1.75664961e-01 -1.41582891e-01
4.99336660e-01 4.16864663e-01 -8.72756019e-02 -1.01787627e+00
-9.27086100e-02 1.04104209e+00 -1.36279926e-01 2.19636425e-01
1.13365722e+00 -7.88357437e-01 -3.75685066e-01 1.73335582e-01
1.30312145e+00 4.79417431e-05 -1.36398041e+00 -5.55086493e-01
-8.21846664e-01 -1.27028608e+00 5.04370868e-01 -5.87159157e-01
-1.26302016e+00 9.02831614e-01 7.33996630e-01 -4.05999839e-01
1.62837303e+00 -3.57766956e-01 1.16735494e+00 7.51479939e-02
6.43325746e-01 -8.43646288e-01 2.60367036e-01 6.00513965e-02
1.08852470e+00 -1.09781158e+00 2.39379600e-01 -9.70241547e-01
-6.98969483e-01 1.02347374e+00 7.35147536e-01 2.20110007e-02
4.92906272e-01 6.59133419e-02 -3.06355208e-01 1.21109225e-01
-1.83327854e-01 1.14143848e-01 1.35173962e-01 4.51023966e-01
-2.08638042e-01 -1.48938701e-01 -4.97052586e-03 2.33061686e-01
-7.38574192e-02 6.76233545e-02 5.45159638e-01 6.45764887e-01
-5.55939555e-01 -8.29966664e-01 -5.85985303e-01 3.57494026e-01
2.30500177e-01 -2.82481015e-01 9.25823301e-02 4.31910485e-01
2.44113803e-01 9.41689968e-01 -1.77825481e-01 -3.32412720e-01
3.23970050e-01 -7.77663350e-01 6.28711939e-01 -3.75027567e-01
-8.51558000e-02 1.98707938e-01 -3.83536279e-01 -7.93703735e-01
-3.60919207e-01 -4.80636477e-01 -1.16548467e+00 -3.11295241e-01
-6.42183185e-01 -7.32715800e-02 6.53414547e-01 6.42065048e-01
2.71844089e-01 6.48341179e-01 1.08180618e+00 -9.86226797e-01
-1.66668802e-01 -7.06389725e-01 -7.13605464e-01 4.88655269e-01
3.79678309e-01 -9.38987672e-01 -2.36599818e-01 3.13185006e-01] | [10.530145645141602, -2.35139799118042] |
312c5b8f-7da1-499c-9520-5f992ba4d29a | in-or-out-fixing-imagenet-out-of-distribution | 2306.00826 | null | https://arxiv.org/abs/2306.00826v1 | https://arxiv.org/pdf/2306.00826v1.pdf | In or Out? Fixing ImageNet Out-of-Distribution Detection Evaluation | Out-of-distribution (OOD) detection is the problem of identifying inputs which are unrelated to the in-distribution task. The OOD detection performance when the in-distribution (ID) is ImageNet-1K is commonly being tested on a small range of test OOD datasets. We find that most of the currently used test OOD datasets, including datasets from the open set recognition (OSR) literature, have severe issues: In some cases more than 50$\%$ of the dataset contains objects belonging to one of the ID classes. These erroneous samples heavily distort the evaluation of OOD detectors. As a solution, we introduce with NINCO a novel test OOD dataset, each sample checked to be ID free, which with its fine-grained range of OOD classes allows for a detailed analysis of an OOD detector's strengths and failure modes, particularly when paired with a number of synthetic "OOD unit-tests". We provide detailed evaluations across a large set of architectures and OOD detection methods on NINCO and the unit-tests, revealing new insights about model weaknesses and the effects of pretraining on OOD detection performance. We provide code and data at https://github.com/j-cb/NINCO. | ['Matthias Hein', 'Maximilian Müller', 'Julian Bitterwolf'] | 2023-06-01 | null | null | null | null | ['open-set-learning'] | ['miscellaneous'] | [ 8.44161306e-03 -1.04377598e-01 -6.59242272e-02 -4.41912115e-01
-6.78974211e-01 -8.71292233e-01 6.88657999e-01 -2.30687037e-02
-1.56544209e-01 4.07812327e-01 -2.07476914e-01 -6.04809165e-01
3.76341343e-02 -5.86624324e-01 -8.47645581e-01 -3.89861047e-01
-2.28579223e-01 6.14633024e-01 3.02079350e-01 2.20957994e-01
5.24766110e-02 3.66982639e-01 -2.01956749e+00 4.58569169e-01
6.87898695e-01 1.27267146e+00 -1.03623331e-01 7.30313659e-01
-5.19120507e-02 4.75477099e-01 -1.32081044e+00 -1.77804157e-01
7.54264832e-01 -2.97316134e-01 -4.65907663e-01 -1.49476537e-02
1.28673124e+00 -4.48212296e-01 -4.42233533e-01 1.39277542e+00
8.51932704e-01 -1.68534324e-01 1.00831687e+00 -1.69810164e+00
-8.81029129e-01 2.25345448e-01 -2.89119273e-01 7.52869964e-01
1.49244338e-01 6.44078851e-01 9.72780883e-01 -1.11763263e+00
7.95719922e-01 1.39299452e+00 5.59792876e-01 5.66122234e-01
-1.56554663e+00 -9.89948630e-01 -2.25419551e-01 -1.44257203e-01
-1.41008949e+00 -4.20527518e-01 1.39431655e-01 -5.63201249e-01
1.27435303e+00 2.48054594e-01 3.25549543e-01 1.40978539e+00
2.84380555e-01 8.10014307e-01 1.06492567e+00 -3.39978755e-01
3.24690551e-01 3.67239058e-01 4.46425706e-01 4.03997093e-01
7.01298892e-01 5.87710559e-01 -4.24255490e-01 -1.45467147e-01
5.08021414e-01 -3.45624447e-01 -2.41648164e-02 -3.64740252e-01
-9.88111913e-01 6.06604576e-01 5.29468179e-01 6.61350042e-02
1.34151474e-01 1.25761688e-01 4.65285093e-01 8.48977625e-01
5.07003009e-01 7.06664443e-01 -3.93116444e-01 4.59410660e-02
-6.85918689e-01 3.36844385e-01 9.84970629e-01 1.25903130e+00
7.97978520e-01 1.70116037e-01 -4.62959707e-01 9.64040875e-01
3.11959684e-01 5.18082082e-01 6.67868972e-01 -6.68624341e-01
4.67788279e-01 5.82176507e-01 2.21618675e-02 -8.20105076e-01
-1.60604969e-01 -3.80971700e-01 -3.74811858e-01 5.19940257e-01
7.51345932e-01 -1.50019646e-01 -1.51734972e+00 1.52299845e+00
1.42681167e-01 -1.01287812e-01 -6.09089844e-02 8.23513448e-01
1.07679713e+00 3.71666640e-01 -6.46733642e-02 5.87514639e-01
1.43555975e+00 -5.50899208e-01 -4.71520543e-01 -6.84805095e-01
8.81510675e-01 -7.52764225e-01 1.28115273e+00 3.55841577e-01
-4.20248270e-01 -6.09399080e-01 -1.27706194e+00 1.24608077e-01
-8.81696522e-01 -1.82281986e-01 5.46361089e-01 7.88888872e-01
-9.90607381e-01 2.64999628e-01 -2.68499792e-01 -4.34376985e-01
6.80144548e-01 2.72439778e-01 -2.65319705e-01 -4.66094971e-01
-1.18415916e+00 8.69274735e-01 4.07924116e-01 -2.71869808e-01
-1.44072902e+00 -6.83010161e-01 -8.87461126e-01 1.11663593e-02
2.19413504e-01 -1.60156816e-01 1.22550118e+00 -8.48914921e-01
-4.15945023e-01 1.22373152e+00 2.36598074e-01 -3.59175265e-01
7.78790593e-01 -2.53524244e-01 -5.82020164e-01 -1.62199244e-01
5.54109633e-01 9.05656815e-01 9.46425200e-01 -1.18818879e+00
-6.03666067e-01 -2.03785107e-01 -2.57438123e-01 -6.17354438e-02
-1.66639805e-01 -8.03970546e-02 -3.55235904e-01 -8.08317542e-01
6.66342378e-02 -7.79599667e-01 2.97245622e-01 -4.75804992e-02
-9.47473884e-01 -3.97705019e-01 7.18040168e-01 -2.23674960e-02
1.22771955e+00 -2.39635253e+00 -6.37307823e-01 3.19166392e-01
5.59335053e-01 1.74106807e-01 -2.78710663e-01 5.94241247e-02
-5.23465216e-01 5.01798451e-01 -1.96077339e-02 -4.48338650e-02
4.51867551e-01 1.18533984e-01 -3.54144394e-01 5.46207368e-01
5.84616125e-01 6.22736573e-01 -5.88996410e-01 -2.95531183e-01
1.48161814e-01 7.11874366e-02 -4.75471616e-01 3.81892532e-01
-1.58394858e-01 -2.40342125e-01 -1.14284128e-01 1.21017742e+00
5.00925958e-01 -1.78704441e-01 -3.64351362e-01 9.48104709e-02
-6.14210078e-03 4.16563362e-01 -1.45521140e+00 9.05267119e-01
9.37872380e-02 1.08553827e+00 -4.19551820e-01 -4.83612835e-01
9.57085609e-01 1.73699901e-01 -1.21277027e-01 -9.52381909e-01
1.73014089e-01 5.86385548e-01 5.59723139e-01 -4.69264001e-01
3.60732377e-01 1.80676773e-01 -3.17081180e-03 2.77249485e-01
4.57479239e-01 -1.14624761e-01 4.92285728e-01 9.31508541e-02
1.56777072e+00 -3.21793020e-01 7.33797699e-02 -6.34046912e-01
-2.42956370e-01 2.14858890e-01 4.88786101e-01 1.45876968e+00
-5.91288626e-01 8.78407240e-01 8.88303936e-01 -3.75383288e-01
-1.13272738e+00 -1.34964132e+00 -9.33943808e-01 8.68861794e-01
5.07651232e-02 -1.37463540e-01 -6.48953199e-01 -7.11608887e-01
6.17720187e-01 7.78325140e-01 -7.64226556e-01 -3.84072363e-01
7.64640868e-02 -9.85915601e-01 1.02681744e+00 4.31000113e-01
3.08255047e-01 -1.27281618e+00 -3.85366648e-01 3.60217318e-02
4.39758509e-01 -9.17055190e-01 -2.92323381e-01 1.07385433e+00
-6.11031771e-01 -1.32033169e+00 -8.20370674e-01 -9.25884843e-01
6.53690159e-01 1.90410510e-01 1.50752795e+00 -7.94256851e-03
-6.73968136e-01 1.54787317e-01 -1.65816233e-01 -8.13877106e-01
-7.05425382e-01 -9.38848704e-02 1.86809778e-01 -3.38982642e-01
1.03045332e+00 -1.58194244e-01 -2.36908004e-01 5.86551726e-01
-8.06904852e-01 -5.93888283e-01 7.37069309e-01 8.18066955e-01
4.78390664e-01 3.75067219e-02 5.10559678e-01 -1.06741142e+00
8.57721686e-01 -7.87461102e-01 -6.15426123e-01 -2.07980856e-01
-7.49582589e-01 8.98471624e-02 2.87240148e-01 -7.41567969e-01
-5.09229898e-01 -3.88481468e-01 1.17065839e-01 -8.39912474e-01
-6.80666089e-01 8.55510011e-02 -2.19313323e-01 2.42930591e-01
1.23595667e+00 -1.62792206e-01 -1.93070620e-01 -5.99625826e-01
-1.68156147e-01 7.85434067e-01 3.67772996e-01 -5.07853448e-01
7.38803267e-01 1.30902857e-01 -6.41149879e-01 -9.09262002e-01
-6.84708416e-01 -4.56911653e-01 -1.62716106e-01 -7.14233965e-02
7.49770522e-01 -9.97172058e-01 -3.36274743e-01 6.27444088e-01
-8.29289436e-01 -7.36166716e-01 -3.74119610e-01 5.07642746e-01
-9.91753787e-02 -1.26544297e-01 -4.86919284e-01 -6.04771137e-01
2.22586334e-01 -1.47945333e+00 8.94870937e-01 2.64307827e-01
-4.87175971e-01 -6.89459980e-01 -1.75939091e-02 -2.54755318e-01
1.66276693e-01 1.95401996e-01 8.91151428e-01 -1.25992346e+00
-3.04356158e-01 -5.18571436e-01 -5.72560787e-01 6.06660545e-01
-7.97344744e-03 -4.99858037e-02 -1.33616078e+00 -2.79899389e-01
-3.88333321e-01 -6.87598586e-01 1.06256878e+00 2.84660548e-01
8.83495867e-01 1.04419172e-01 -2.65098542e-01 5.57980835e-01
1.34334409e+00 2.16632232e-01 5.67106009e-01 5.24145603e-01
5.66464782e-01 4.20398653e-01 6.08931482e-01 2.08147824e-01
-6.64762408e-02 2.69378722e-01 6.00025952e-01 -1.01675496e-01
-3.73812467e-01 -1.42292678e-01 4.90950018e-01 2.81002522e-02
6.87398493e-01 -7.47731328e-01 -1.20313811e+00 7.76123941e-01
-1.23935044e+00 -7.11565137e-01 -1.04386181e-01 2.21800447e+00
6.76805854e-01 8.26646984e-01 9.00456607e-02 1.21190377e-01
9.26181257e-01 8.37169811e-02 -8.52229297e-01 -4.59289134e-01
-2.26620331e-01 1.56952590e-01 7.12743878e-01 1.20668814e-01
-1.30808473e+00 6.23044789e-01 6.96599960e+00 6.96221948e-01
-7.67830670e-01 -1.14548400e-01 7.28336513e-01 -2.44186334e-02
-1.05674595e-01 -3.06265742e-01 -1.28722763e+00 7.00282335e-01
8.71911049e-01 2.46169657e-01 1.14155076e-01 1.13508093e+00
-1.68087423e-01 -3.11780393e-01 -1.55388701e+00 9.58681107e-01
1.57242790e-01 -9.40126717e-01 -4.76011448e-02 3.53721857e-01
6.70209527e-01 7.21851826e-01 2.56485641e-02 6.93050563e-01
6.63374364e-01 -1.11975920e+00 1.00202274e+00 -4.17451598e-02
1.06591415e+00 -3.97904366e-01 7.16512680e-01 1.88263938e-01
-7.73983002e-01 -1.01380698e-01 -6.16565108e-01 9.32311043e-02
-5.92894614e-01 6.17064774e-01 -9.40294862e-01 -1.67048946e-01
1.27615058e+00 4.65914905e-01 -1.05984592e+00 1.10518420e+00
-9.91418585e-02 6.71530545e-01 -6.41663551e-01 -9.72668156e-02
2.04596862e-01 3.77058238e-01 6.26018524e-01 1.24915516e+00
1.41068846e-01 -5.27480841e-01 2.62473710e-02 1.27085268e+00
-3.25085402e-01 -5.13689399e-01 -1.02837384e+00 -2.62144506e-01
7.34885156e-01 7.68493235e-01 -7.09949970e-01 -4.79246706e-01
-4.81873393e-01 5.08126438e-01 2.32107788e-01 2.58757770e-01
-8.19780290e-01 -8.95264804e-01 1.09220970e+00 1.30515844e-02
2.36168340e-01 3.02249312e-01 -2.12051690e-01 -1.17837918e+00
5.41585051e-02 -1.23427844e+00 6.27808094e-01 -6.34457231e-01
-1.77335608e+00 4.84145582e-01 -6.62987977e-02 -1.42906570e+00
-8.65143687e-02 -1.09335124e+00 -5.99884868e-01 7.70309806e-01
-1.45131910e+00 -2.73659199e-01 -3.50533724e-01 2.50534773e-01
6.00097060e-01 -2.29967058e-01 6.05549634e-01 6.19615912e-01
-7.44961023e-01 9.31549191e-01 9.51107442e-02 6.17895603e-01
1.14126146e+00 -1.57697213e+00 6.31056309e-01 8.27156782e-01
1.21993251e-01 3.48965228e-01 9.47450936e-01 -6.23133421e-01
-1.03921425e+00 -1.05190778e+00 6.56722009e-01 -8.54286611e-01
5.15820980e-01 -6.01612628e-01 -8.72948647e-01 9.38948452e-01
-1.39507771e-01 4.92463201e-01 4.44848835e-01 7.54903862e-03
-4.96436000e-01 3.44915129e-02 -1.06332493e+00 3.22731912e-01
1.06128728e+00 -5.65478206e-01 -8.16620111e-01 2.25101992e-01
4.27040368e-01 -5.38322568e-01 -7.89239824e-01 2.79803902e-01
4.49341595e-01 -1.23675525e+00 9.33648646e-01 -6.41722202e-01
3.42811823e-01 -2.54436910e-01 -4.06798780e-01 -1.24614239e+00
-5.84710874e-02 -1.56855673e-01 -4.02067378e-02 1.39122176e+00
5.34070849e-01 -8.35900426e-01 5.08452594e-01 4.42104161e-01
-3.92462462e-01 -4.34937239e-01 -8.19707513e-01 -1.08090222e+00
-1.14512444e-01 -8.54135931e-01 3.25535685e-01 1.01710510e+00
-3.62736076e-01 1.06176168e-01 1.07594825e-01 3.97288024e-01
6.86835229e-01 -2.13723704e-01 8.79104793e-01 -1.19332683e+00
-2.65666664e-01 -3.83471310e-01 -6.73859596e-01 -8.77358139e-01
-2.54624605e-01 -1.07551777e+00 2.53993988e-01 -1.06844473e+00
-7.57358968e-02 -6.04546666e-01 -3.60641509e-01 5.51939368e-01
-5.18006040e-03 5.53497910e-01 2.03833833e-01 4.67134982e-01
-5.24658501e-01 1.13518424e-01 8.88246596e-01 -2.40910321e-01
-1.94461629e-01 -3.07245940e-01 -5.43760538e-01 7.31694400e-01
5.37883520e-01 -1.02987075e+00 -1.36248037e-01 -4.65207100e-01
-4.54952791e-02 -4.33196038e-01 6.17073417e-01 -1.24133015e+00
-2.05252364e-01 2.71606088e-01 8.67759049e-01 -5.35565019e-01
-8.20160061e-02 -5.61046779e-01 -2.60215789e-01 6.59707546e-01
-5.36399484e-01 1.47798970e-01 5.22282958e-01 5.36533415e-01
-1.27121434e-01 -2.30725318e-01 8.29551637e-01 -2.03804836e-01
-1.04139650e+00 1.26955882e-01 -5.05015433e-01 8.10981512e-01
9.23243642e-01 -3.38100046e-01 -5.32000244e-01 1.05035953e-01
-4.84935850e-01 3.82715464e-01 3.63176376e-01 7.68391192e-01
3.25931549e-01 -1.23993576e+00 -4.64059174e-01 7.63582349e-01
7.41989076e-01 1.55713335e-01 -2.75943935e-01 4.94153887e-01
-6.31025910e-01 2.54427046e-01 -2.87623703e-01 -8.53634596e-01
-8.67429614e-01 2.79920369e-01 6.59528136e-01 6.17078282e-02
-3.94767851e-01 9.23066974e-01 5.98315716e-01 -6.13384008e-01
5.61440706e-01 -3.99808049e-01 2.07097918e-01 2.36578390e-01
4.53156561e-01 3.08457673e-01 1.89932778e-01 -1.58628076e-01
-4.31054056e-01 -2.05593422e-01 -1.05709389e-01 1.61076739e-01
1.09847939e+00 1.46195278e-01 4.11869854e-01 6.57557428e-01
1.23213589e+00 -2.98499882e-01 -1.29008877e+00 -2.45194603e-02
1.60308704e-01 -5.45144141e-01 -6.67944551e-02 -9.09079731e-01
-7.79603362e-01 8.69993031e-01 9.80164766e-01 5.70533276e-01
4.83373344e-01 2.73169160e-01 1.94638893e-01 4.91661340e-01
-6.81815948e-03 -1.08341730e+00 1.63596496e-01 4.98933524e-01
6.85663164e-01 -1.64377820e+00 -3.10027868e-01 2.65568141e-02
-3.45611125e-01 1.02289164e+00 1.21267617e+00 -5.48671126e-01
5.89903533e-01 5.16331375e-01 2.41001204e-01 -4.56450135e-01
-7.96768188e-01 -2.62010962e-01 1.16359763e-01 6.76913619e-01
3.45679075e-01 -1.22478634e-01 2.36825734e-01 3.11081409e-01
-1.96696311e-01 -3.79534155e-01 3.81913722e-01 8.97156596e-01
-3.80009681e-01 -5.15034199e-01 -5.09368539e-01 1.12050045e+00
-3.51777047e-01 -2.67084628e-01 -5.89342952e-01 1.25620329e+00
2.28396729e-01 5.09511411e-01 6.55637443e-01 -4.32215750e-01
5.86694121e-01 3.30941617e-01 6.34675995e-02 -8.92391562e-01
-5.13113856e-01 -5.07149398e-02 1.76254943e-01 -6.02675080e-01
2.06648663e-01 -5.72108448e-01 -9.98526394e-01 -1.81386977e-01
-5.67529440e-01 -3.63551766e-01 3.84400278e-01 6.05746329e-01
4.50486243e-01 6.48173571e-01 2.62348473e-01 -6.99648678e-01
-8.78602862e-01 -1.16192627e+00 -8.61234426e-01 7.17001498e-01
5.25755048e-01 -9.61179495e-01 -1.00856781e+00 -2.39584729e-01] | [9.360740661621094, 2.7460591793060303] |
479474ae-661f-48cd-912e-7a20ed5418ba | your-day-in-your-pocket-complex-activity | 2301.06993 | null | https://arxiv.org/abs/2301.06993v1 | https://arxiv.org/pdf/2301.06993v1.pdf | Your Day in Your Pocket: Complex Activity Recognition from Smartphone Accelerometers | Human Activity Recognition (HAR) enables context-aware user experiences where mobile apps can alter content and interactions depending on user activities. Hence, smartphones have become valuable for HAR as they allow large, and diversified data collection. Although previous work in HAR managed to detect simple activities (i.e., sitting, walking, running) with good accuracy using inertial sensors (i.e., accelerometer), the recognition of complex daily activities remains an open problem, specially in remote work/study settings when people are more sedentary. Moreover, understanding the everyday activities of a person can support the creation of applications that aim to support their well-being. This paper investigates the recognition of complex activities exclusively using smartphone accelerometer data. We used a large smartphone sensing dataset collected from over 600 users in five countries during the pandemic and showed that deep learning-based, binary classification of eight complex activities (sleeping, eating, watching videos, online communication, attending a lecture, sports, shopping, studying) can be achieved with AUROC scores up to 0.76 with partially personalized models. This shows encouraging signs toward assessing complex activities only using phone accelerometer data in the post-pandemic world. | ['Daniel Gatica-Perez', 'Lakmal Meegahapola', 'Emma Bouton--Bessac'] | 2023-01-17 | null | null | null | null | ['human-activity-recognition', 'human-activity-recognition'] | ['computer-vision', 'time-series'] | [ 3.80009472e-01 -1.94792897e-01 -5.70442915e-01 -1.39530122e-01
-3.45538020e-01 -3.17946583e-01 5.60681999e-01 2.99035639e-01
-3.92323583e-01 7.04072475e-01 7.81623662e-01 -2.47834176e-01
-1.43341199e-01 -7.71262884e-01 -3.01262408e-01 -6.31221473e-01
-2.01714337e-01 2.10391685e-01 -4.97747138e-02 -1.57007858e-01
-8.21401775e-02 4.33120608e-01 -1.80954421e+00 2.55905867e-01
4.74678427e-01 9.03647065e-01 8.51409137e-03 1.00598955e+00
1.90462872e-01 6.06007218e-01 -8.29763591e-01 6.67382926e-02
-3.00854325e-01 -5.64964950e-01 -4.30725574e-01 9.95819122e-02
2.30171112e-03 -4.52233732e-01 1.43991604e-01 2.10752323e-01
7.58455575e-01 1.96922809e-01 1.26013130e-01 -1.06437838e+00
-9.77833718e-02 1.13525189e-01 -1.87545959e-02 5.13716340e-01
1.08367431e+00 9.55421031e-02 4.06775653e-01 -3.11077505e-01
1.58677146e-01 4.71710443e-01 1.18464398e+00 3.29561293e-01
-7.99625814e-01 -4.22005028e-01 -3.64582449e-01 2.65041322e-01
-1.13612628e+00 -5.13969839e-01 4.60981607e-01 -4.66113269e-01
1.41442192e+00 6.49547577e-01 1.33991337e+00 1.79788601e+00
4.20106977e-01 3.55762482e-01 9.48150158e-01 -2.00458542e-01
3.90035599e-01 1.25072494e-01 -2.75920220e-02 2.98376054e-01
5.23942113e-01 -4.04415339e-01 -6.45736277e-01 -1.14264667e-01
3.23058099e-01 7.54194498e-01 5.12733348e-02 -3.05143818e-02
-1.57880127e+00 3.49949688e-01 -1.73259228e-02 8.30769241e-01
-7.24278331e-01 -1.03504188e-01 3.99050623e-01 -1.19256578e-01
4.23422098e-01 2.69323766e-01 -6.86498940e-01 -1.14958477e+00
-9.69143093e-01 8.60490128e-02 8.43346417e-01 2.24708289e-01
3.31213295e-01 -1.34992033e-01 -1.72753930e-01 6.45163953e-01
1.19079582e-01 6.07964456e-01 9.66935754e-01 -9.79048133e-01
2.48757631e-01 7.50597060e-01 3.50905180e-01 -1.19513226e+00
-1.21219516e+00 -4.80817080e-01 -1.04872990e+00 -5.24683595e-01
3.69734526e-01 -5.90840876e-01 -2.39647433e-01 1.39997375e+00
6.46967173e-01 3.39962333e-01 -3.28577965e-01 4.28416908e-01
5.68913162e-01 3.53665471e-01 2.67316878e-01 -3.74588251e-01
1.68429840e+00 -6.31237924e-01 -8.68655264e-01 -3.21251422e-01
8.49034905e-01 -1.90104857e-01 1.26490831e+00 6.28142834e-01
-7.11002767e-01 -6.93588793e-01 -1.07383394e+00 1.54459268e-01
-9.30393696e-01 -9.74885300e-02 7.23987103e-01 1.46187103e+00
-8.92652094e-01 6.59649491e-01 -1.39841008e+00 -8.72857332e-01
4.86470550e-01 4.81439501e-01 -3.30229074e-01 3.16204160e-01
-9.79600191e-01 5.67931354e-01 1.77586958e-01 -3.03037733e-01
-3.26862544e-01 -3.34717810e-01 -7.82023966e-01 1.94841232e-02
1.51172448e-02 -7.18118966e-01 9.15142000e-01 -7.06804037e-01
-1.57793713e+00 6.99241579e-01 -1.68178767e-01 -5.59515715e-01
3.54099929e-01 -4.91639465e-01 -8.80530119e-01 -5.11754118e-02
1.01665080e-01 3.00902188e-01 4.47282046e-01 -2.86408871e-01
-7.58787513e-01 -7.12989032e-01 -2.89219711e-02 3.10398340e-01
-7.00243294e-01 -5.40802591e-02 -7.20080212e-02 -2.85832107e-01
-2.36525685e-01 -1.08434284e+00 1.63437262e-01 -6.58548832e-01
-9.81560722e-02 -6.66841213e-03 8.40153575e-01 -1.13374794e+00
1.60741210e+00 -1.95806420e+00 -3.11966926e-01 -2.28294861e-02
1.99172750e-01 2.37446323e-01 7.26138055e-01 3.47415149e-01
2.25662783e-01 7.59424269e-02 -1.25021011e-01 -2.79647112e-01
6.84816688e-02 4.30481255e-01 4.65297192e-01 4.70969826e-01
-4.09242272e-01 8.90335500e-01 -9.57499743e-01 -3.24197084e-01
5.51983714e-01 8.53696465e-01 -5.03867030e-01 -3.47653478e-02
4.90048110e-01 7.43497670e-01 -1.07464537e-01 7.44878709e-01
6.30898550e-02 -4.37147796e-01 3.85472894e-01 2.84714937e-01
-6.13095574e-02 3.95947248e-01 -1.06363165e+00 1.60450637e+00
-7.71778047e-01 7.09445000e-01 -2.56166369e-01 -1.05020475e+00
3.98996055e-01 3.68667871e-01 9.77914214e-01 -8.22829425e-01
5.70263900e-02 -3.49319428e-02 -2.26573870e-01 -8.65515649e-01
2.95650095e-01 1.72424003e-01 -2.68130392e-01 3.66341531e-01
-3.30022454e-01 5.76385260e-01 1.71875715e-01 -3.72444391e-01
1.41007733e+00 1.47733331e-01 9.02627230e-01 1.01288587e-01
3.81170690e-01 -3.50025624e-01 2.40108415e-01 7.07022130e-01
-7.19882429e-01 3.67275774e-01 -1.31305316e-02 -5.33042669e-01
-3.68528903e-01 -9.97016072e-01 9.99648869e-02 1.63872933e+00
-2.62247980e-01 -4.48950946e-01 -8.67090762e-01 -4.48463827e-01
-3.82993162e-01 4.35902625e-01 -6.57639563e-01 -1.59349397e-01
-6.17800355e-01 -1.09419644e+00 5.59096694e-01 5.24948597e-01
1.01462758e+00 -1.09903157e+00 -1.20203221e+00 4.11245346e-01
-5.05812764e-01 -1.10012543e+00 -2.58662909e-01 4.92221154e-02
-9.70716238e-01 -9.90671575e-01 -6.86271131e-01 -4.53888446e-01
-1.08841784e-01 2.87094116e-01 9.98078108e-01 -2.04231039e-01
-2.99130287e-02 8.08141470e-01 -2.86234140e-01 -4.60892797e-01
9.41295251e-02 3.73881072e-01 4.29397911e-01 1.53430670e-01
6.82411492e-01 -7.92308807e-01 -9.89417374e-01 4.34153616e-01
-4.65772927e-01 -1.52522445e-01 3.85387719e-01 1.60019025e-01
4.72719729e-01 -5.50181977e-03 4.81829494e-01 -5.32312810e-01
6.13803804e-01 -9.29987073e-01 3.70230466e-01 -9.21072289e-02
-6.64624095e-01 -7.73471594e-01 3.88632923e-01 -4.97857183e-01
-7.40409851e-01 -4.04796116e-02 -5.18314004e-01 5.77956378e-01
-7.59685218e-01 9.84891951e-02 -3.32050681e-01 3.25158536e-01
8.24147701e-01 2.09376514e-01 -4.02688950e-01 -3.75180721e-01
-2.71337956e-01 1.00815094e+00 3.51133794e-01 2.05214452e-02
-3.19868065e-02 6.10233009e-01 -4.50485013e-02 -1.57209873e+00
-6.41646326e-01 -8.31370354e-01 -5.47174156e-01 -4.01850462e-01
1.14549613e+00 -8.71403217e-01 -9.58953738e-01 5.61731160e-01
-4.94157881e-01 -4.89168197e-01 -2.94009149e-01 5.99067748e-01
-4.25056577e-01 3.75946641e-01 -2.32865259e-01 -9.75215614e-01
-3.61939371e-01 -6.18702829e-01 1.35070539e+00 5.28783739e-01
-1.02996778e+00 -1.12698722e+00 3.54175776e-01 9.98507440e-01
6.38929248e-01 8.26057673e-01 9.44926888e-02 -6.14485919e-01
1.53697327e-01 -4.80519831e-01 3.49871784e-01 1.68498054e-01
6.21237218e-01 -5.18082857e-01 -9.32576776e-01 -8.99895802e-02
9.60899666e-02 1.44231141e-01 8.80079751e-04 4.61106777e-01
1.17379677e+00 -6.45682752e-01 -5.26617467e-01 5.73202014e-01
7.48248458e-01 5.42351305e-01 9.63245571e-01 4.31696594e-01
7.64475822e-01 1.61259025e-01 1.58936113e-01 6.83094740e-01
6.76991224e-01 8.02018881e-01 2.05576912e-01 2.32700091e-02
9.04987305e-02 1.71903763e-02 6.61810279e-01 6.76341534e-01
-5.63631475e-01 -1.73236862e-01 -1.00713706e+00 3.34202707e-01
-1.42638600e+00 -1.08692086e+00 -2.00740740e-01 2.32770252e+00
4.32908475e-01 2.90258110e-01 7.02240229e-01 6.82400346e-01
3.14568251e-01 1.87729940e-01 -5.23227036e-01 -5.96613765e-01
-1.03429323e-02 2.93344766e-01 3.42607141e-01 2.72956584e-02
-1.20261812e+00 1.55492380e-01 6.03969622e+00 2.68040389e-01
-1.06157911e+00 5.94962835e-01 6.41004145e-01 -4.45265561e-01
4.40777063e-01 -7.10999072e-01 -5.80576658e-01 9.54823494e-01
1.70731080e+00 6.10891283e-01 5.06093740e-01 8.09166193e-01
7.87996173e-01 -4.84887570e-01 -6.30438805e-01 1.29608250e+00
5.33435307e-02 -1.03496134e+00 -5.23154497e-01 4.68905360e-01
6.62332237e-01 2.28623077e-01 -1.54740572e-01 4.39932287e-01
-4.84402478e-01 -7.10188329e-01 4.66070771e-02 7.90215552e-01
6.34432852e-01 -6.04564846e-01 7.65901625e-01 5.21825075e-01
-1.01060152e+00 -3.14356267e-01 5.66116214e-01 -6.49426460e-01
1.03582934e-01 6.79326832e-01 -9.30027604e-01 -1.05699999e-02
1.13720345e+00 7.94450819e-01 -8.13341677e-01 7.09920228e-01
1.91657692e-01 8.68925214e-01 -5.00024676e-01 -3.24778736e-01
-1.07420117e-01 -1.31810173e-01 1.18589483e-01 1.53064287e+00
7.02793300e-01 -5.56930453e-02 -1.29383102e-01 -1.71336994e-01
1.68172941e-01 8.67268164e-03 -8.74011278e-01 -2.78661102e-01
-5.83873726e-02 1.10558939e+00 -9.13434565e-01 -3.30561548e-01
-3.62182200e-01 1.13118804e+00 -3.41616482e-01 2.02550441e-02
-8.29958677e-01 -3.71832758e-01 8.70093882e-01 6.64684415e-01
-1.63707107e-01 -5.70649147e-01 -4.05733228e-01 -9.63235140e-01
-1.42788002e-02 -8.99737239e-01 4.13236707e-01 -6.52093291e-01
-4.41988140e-01 -1.54577777e-01 -1.76315069e-01 -1.08129847e+00
-5.59785128e-01 -4.83484894e-01 -5.77897131e-01 1.73732817e-01
-6.17314398e-01 -8.64528358e-01 -7.45994806e-01 6.87416553e-01
6.13272071e-01 2.60369897e-01 8.75846803e-01 5.94920218e-01
-6.61452770e-01 3.26181889e-01 2.33151942e-01 -1.02463260e-01
2.99897403e-01 -1.18988144e+00 2.10421711e-01 6.05897009e-01
1.87116072e-01 7.18040526e-01 4.55595434e-01 -5.90152681e-01
-1.32964337e+00 -8.53166521e-01 1.32222962e+00 -9.31550026e-01
2.57010520e-01 -3.71642590e-01 -3.84988457e-01 6.41349792e-01
-8.65102187e-03 -3.43855381e-01 1.48134601e+00 2.27447316e-01
4.31852311e-01 -2.10404575e-01 -1.35831153e+00 6.53648138e-01
1.40906894e+00 -5.62937617e-01 -4.52392191e-01 4.45008695e-01
2.18567669e-01 -2.92611748e-01 -9.90258992e-01 9.81917679e-02
1.16036141e+00 -1.41115844e+00 1.21813834e+00 -3.61967683e-01
-3.70118171e-02 5.80334999e-02 -4.36704010e-02 -1.03365040e+00
-1.37505397e-01 -6.06860816e-01 -8.24661493e-01 8.48489046e-01
-9.30502489e-02 -8.66239190e-01 8.95008743e-01 4.11439568e-01
-4.15514782e-02 -5.34814715e-01 -9.16655064e-01 -4.99928176e-01
-8.39506030e-01 -7.51920521e-01 9.02584076e-01 1.03749228e+00
1.68949515e-01 3.05249155e-01 -5.06484210e-01 -5.71996495e-02
2.27025282e-02 -6.36540830e-01 8.94808114e-01 -1.34275615e+00
-3.90100718e-01 -3.69981468e-01 -5.71729004e-01 -6.81567430e-01
-5.51068187e-01 -3.69861811e-01 -5.98506033e-01 -1.75994480e+00
-1.58420771e-01 1.70575678e-01 -2.27002189e-01 3.59193057e-01
1.57637194e-01 5.19335747e-01 -3.59499872e-01 -1.34907946e-01
-9.27092850e-01 -7.32654929e-02 8.30719292e-01 -8.16600993e-02
-5.68870008e-01 5.02922475e-01 -7.77708292e-01 7.50560045e-01
1.22707474e+00 1.18070859e-02 -2.90884286e-01 3.52382138e-02
5.81946552e-01 -8.12419504e-02 2.78359443e-01 -1.69435871e+00
-7.57634044e-02 -1.29635781e-01 8.21905792e-01 -3.11891884e-01
5.52700102e-01 -8.85322571e-01 5.18372416e-01 8.69938672e-01
1.10245541e-01 7.22972080e-02 4.69832346e-02 5.47522604e-01
2.66080290e-01 3.25365692e-01 2.37755507e-01 -2.05621660e-01
-5.06608307e-01 -1.23777494e-01 -8.29157174e-01 -1.30203038e-01
7.71661937e-01 -8.21625710e-01 -1.09312192e-01 -4.35412228e-01
-1.29884708e+00 -2.25080192e-01 1.57399967e-01 5.87451160e-01
1.33413017e-01 -1.22929513e+00 3.67970541e-02 2.55466849e-01
2.22066924e-01 -3.78128260e-01 4.48526084e-01 1.20834720e+00
-5.82301855e-01 6.61587059e-01 -3.51093560e-01 -6.24029756e-01
-1.26790333e+00 1.94468766e-01 3.13151151e-01 -4.07503664e-01
-3.05601627e-01 1.92607298e-01 -5.46537161e-01 -4.06103075e-01
2.20219016e-01 -6.51900828e-01 -5.56912959e-01 5.11306882e-01
8.38724613e-01 1.25735366e+00 4.82547879e-01 -7.61388361e-01
-6.31209016e-01 3.12561750e-01 6.35837555e-01 1.90637290e-01
1.13319135e+00 -4.59373981e-01 3.88976783e-01 6.99988246e-01
1.24965191e+00 1.43585084e-02 -9.08493638e-01 3.70904386e-01
1.39481097e-01 -2.04450801e-01 3.78269851e-02 -7.71263361e-01
-5.35747051e-01 8.67926776e-01 1.44264710e+00 7.75221348e-01
1.25732934e+00 -2.75195569e-01 1.06662798e+00 5.45443892e-01
4.76382852e-01 -1.39767087e+00 6.53440580e-02 2.72169501e-01
2.41342321e-01 -1.22723353e+00 -1.50748402e-01 4.40584153e-01
-4.63686198e-01 6.28582597e-01 1.22791499e-01 5.43577135e-01
7.51096070e-01 -6.11649901e-02 -3.51753831e-01 -8.65731314e-02
-1.24871626e-01 -3.11138898e-01 1.14113465e-01 1.12401092e+00
6.24829829e-01 5.92177927e-01 -2.13076025e-01 5.67386389e-01
-4.32562798e-01 3.74892801e-01 4.64904398e-01 1.12475693e+00
-5.05665302e-01 -6.29330397e-01 -5.46573162e-01 9.42429423e-01
-8.34044218e-01 4.50630784e-01 -3.43578011e-01 6.40359402e-01
7.84257531e-01 1.26243329e+00 1.77553356e-01 -3.82572979e-01
4.31447625e-01 4.45040554e-01 2.84577161e-01 -4.93613541e-01
-8.11577380e-01 -3.89300466e-01 3.46838415e-01 -8.21761549e-01
-6.85890317e-01 -1.03391945e+00 -9.08336818e-01 -3.84813666e-01
5.20959914e-01 -2.44125932e-01 9.42257643e-01 1.16388381e+00
5.54395854e-01 6.67484820e-01 3.99206489e-01 -1.12175536e+00
2.51394689e-01 -1.01032972e+00 -5.74753582e-01 1.93532944e-01
5.19753218e-01 -5.43524146e-01 -2.55574137e-01 1.63033545e-01] | [7.330082416534424, 0.7140392661094666] |
d0a70a11-6b84-4871-8841-8f7e0c902993 | robust-regression-for-safe-exploration-in | 1906.05819 | null | https://arxiv.org/abs/1906.05819v2 | https://arxiv.org/pdf/1906.05819v2.pdf | Robust Regression for Safe Exploration in Control | We study the problem of safe learning and exploration in sequential control problems. The goal is to safely collect data samples from operating in an environment, in order to learn to achieve a challenging control goal (e.g., an agile maneuver close to a boundary). A central challenge in this setting is how to quantify uncertainty in order to choose provably-safe actions that allow us to collect informative data and reduce uncertainty, thereby achieving both improved controller safety and optimality. To address this challenge, we present a deep robust regression model that is trained to directly predict the uncertainty bounds for safe exploration. We derive generalization bounds for learning, and connect them with safety and stability bounds in control. We demonstrate empirically that our robust regression approach can outperform the conventional Gaussian process (GP) based safe exploration in settings where it is difficult to specify a good GP prior. | ['Yisong Yue', 'Soon-Jo Chung', 'Anima Anandkumar', 'Anqi Liu', 'Guanya Shi'] | 2019-06-13 | null | https://openreview.net/forum?id=PdxyQilsUrG | https://openreview.net/pdf?id=PdxyQilsUrG | l4dc-2020-6 | ['safe-exploration'] | ['robots'] | [ 3.70870590e-01 5.06792963e-01 -3.51210207e-01 1.83611467e-01
-1.25464082e+00 -6.41469777e-01 3.73978734e-01 2.82205582e-01
-2.12756604e-01 1.04708719e+00 -1.62669316e-01 -5.36852717e-01
-6.67397022e-01 -6.06886148e-01 -1.03443062e+00 -1.03408921e+00
-4.67629731e-01 4.73964006e-01 -1.15523972e-01 7.30709732e-02
3.41738135e-01 3.60577285e-01 -1.19287717e+00 -5.83543599e-01
1.11234784e+00 1.01398540e+00 1.16100065e-01 6.39773309e-01
7.45721817e-01 5.49380839e-01 -2.20687672e-01 3.98611605e-01
4.88483310e-01 -1.29630625e-01 -5.73721647e-01 -6.27644062e-02
-1.29395753e-01 -2.73578644e-01 1.92387812e-02 1.24748147e+00
3.16669554e-01 7.63695598e-01 6.65560067e-01 -1.31976926e+00
4.51561026e-02 6.28587127e-01 -5.21299303e-01 -1.62214443e-01
-1.21179350e-01 6.52896345e-01 7.92259872e-01 -2.60820724e-02
2.10368574e-01 1.26957047e+00 3.79324317e-01 5.66365361e-01
-1.56977761e+00 -5.33138275e-01 3.93841565e-01 -5.28290570e-01
-1.23576760e+00 -1.93218186e-01 2.53594846e-01 -6.93873227e-01
5.04089057e-01 1.05435759e-01 4.42691565e-01 1.27629066e+00
7.84387589e-01 4.73111749e-01 7.92964220e-01 -5.98713495e-02
8.63407791e-01 -1.88258976e-01 -1.21914558e-01 4.47840005e-01
5.30045033e-01 6.35813892e-01 -2.86568403e-01 -2.47510985e-01
7.86438048e-01 -7.25968704e-02 -4.57280487e-01 -7.37835646e-01
-1.04442453e+00 1.04004204e+00 3.40540767e-01 -4.42651480e-01
-4.47926670e-01 6.25794590e-01 1.97594315e-01 2.41844535e-01
1.14517557e-02 1.09786654e+00 -5.24993300e-01 -4.89640832e-01
-3.66409957e-01 6.97521567e-01 9.50795650e-01 1.00203204e+00
2.93730050e-01 4.22612071e-01 -4.18443471e-01 1.23398162e-01
3.54340762e-01 6.28144205e-01 -3.25631469e-01 -1.45984495e+00
7.25413918e-01 6.96221963e-02 9.53431785e-01 -5.04313409e-01
-2.77064651e-01 -5.13146877e-01 -4.77417439e-01 8.28261554e-01
4.74124044e-01 -8.29481900e-01 -8.54334414e-01 1.96435082e+00
2.65674800e-01 2.35407785e-01 1.38744116e-01 7.97841847e-01
-3.96745116e-01 9.05594289e-01 -1.85188130e-02 -5.04808605e-01
7.95919657e-01 -5.41919112e-01 -5.40841162e-01 -4.69399393e-01
4.14776087e-01 -1.12755768e-01 1.15558660e+00 8.00980747e-01
-1.09514511e+00 -9.65865254e-02 -1.11456323e+00 5.05868316e-01
2.67725617e-01 -3.82732719e-01 1.80787370e-01 4.20285851e-01
-5.98070621e-01 1.06043267e+00 -1.30420339e+00 -4.69339900e-02
3.63705337e-01 5.34695446e-01 4.77822358e-03 2.58584470e-01
-7.70433486e-01 1.07211518e+00 6.97900951e-01 1.61641419e-01
-1.79602373e+00 -9.91365850e-01 -8.59530449e-01 7.20064342e-02
1.36023295e+00 -5.89666069e-01 1.50278437e+00 -1.28582507e-01
-1.67975247e+00 -8.13859403e-02 2.66589254e-01 -9.09421504e-01
6.85164571e-01 -7.09403694e-01 4.05531317e-01 -1.84592485e-01
-2.05395430e-01 4.84318525e-01 1.09828663e+00 -1.35408080e+00
-7.10359097e-01 -2.48764083e-01 2.03514457e-01 3.68698239e-01
2.94018313e-02 -3.93303990e-01 -6.25878870e-02 -1.95865989e-01
-2.80069739e-01 -1.39221275e+00 -9.78212237e-01 -1.98421761e-01
-6.55452311e-01 7.53394142e-02 6.83118463e-01 -5.83672166e-01
9.05742645e-01 -1.70007110e+00 6.48099422e-01 5.03456652e-01
1.38340279e-01 -1.10296771e-01 1.39416948e-01 3.47603887e-01
4.37946379e-01 2.39835903e-01 -2.94338465e-01 -2.40974963e-01
2.15939879e-01 4.18744326e-01 -9.47823942e-01 6.63334608e-01
3.95942658e-01 6.35421395e-01 -9.64987934e-01 9.47239250e-03
7.19234049e-02 2.02695727e-01 -5.38115323e-01 3.73407245e-01
-7.92551339e-01 1.07711756e+00 -9.16415572e-01 3.51987541e-01
1.28002524e-01 1.23092331e-01 6.06710128e-02 4.46337789e-01
-1.55901909e-01 -3.91412713e-02 -1.36053944e+00 1.08455837e+00
-6.54881895e-01 2.84692824e-01 5.42774856e-01 -8.00250649e-01
9.08513486e-01 -1.70133095e-02 3.84314477e-01 -5.89301363e-02
5.32418251e-01 -2.51823161e-02 -2.11160511e-01 -1.60136446e-01
4.03458536e-01 -2.85544097e-01 -3.78093481e-01 2.73418993e-01
-4.42311615e-01 -7.69302607e-01 -2.61156142e-01 -3.66554052e-01
1.09804821e+00 2.21804172e-01 2.73786366e-01 -6.56617880e-01
2.10275978e-01 -1.34343402e-02 8.94078851e-01 7.94421732e-01
-6.47459924e-02 6.69025704e-02 1.13400793e+00 -7.27162370e-03
-9.39518988e-01 -1.18032944e+00 8.07010010e-02 5.89856803e-01
1.46911174e-01 -5.92585504e-02 -5.65796971e-01 -4.41884816e-01
2.17450619e-01 1.17756414e+00 -6.51373148e-01 -5.54445565e-01
-3.94380748e-01 -3.09619576e-01 -5.38620763e-02 5.97595632e-01
-3.63326771e-03 -4.21391279e-01 -1.01538658e+00 2.00614825e-01
3.98439109e-01 -6.40284479e-01 -4.54346538e-01 7.13706553e-01
-8.22274923e-01 -1.01709926e+00 -2.15153858e-01 -3.08073699e-01
6.06682837e-01 -4.11691368e-01 6.19051635e-01 -6.61886454e-01
4.14936729e-02 3.89381528e-01 2.18143657e-01 -6.57459140e-01
-4.50919747e-01 2.87130754e-02 3.71048599e-01 -3.43778461e-01
-3.65438610e-01 -3.76856536e-01 -2.56538510e-01 2.89656311e-01
-5.53486526e-01 -2.92043626e-01 2.63152450e-01 8.80444467e-01
9.51518893e-01 6.67192161e-01 2.87109345e-01 -4.26735818e-01
9.83202100e-01 -5.18033504e-01 -1.56398952e+00 1.11061677e-01
-6.33318484e-01 5.06821454e-01 6.95474803e-01 -6.49362385e-01
-7.18599916e-01 1.43137991e-01 4.18461770e-01 -7.92414427e-01
1.91320181e-01 4.53895122e-01 -2.51344651e-01 -2.07163006e-01
5.14154196e-01 -8.08547363e-02 3.85283709e-01 -9.76565629e-02
3.76847833e-01 1.61485568e-01 4.61158633e-01 -1.35787416e+00
9.13299561e-01 1.24829076e-01 5.99035084e-01 -5.12244761e-01
-8.90686929e-01 7.53971264e-02 -4.88789439e-01 -1.12750940e-01
7.51334190e-01 -8.13350320e-01 -1.27950335e+00 -2.33428225e-01
-4.84676391e-01 -9.98909295e-01 -5.49400270e-01 3.99273932e-01
-1.34414685e+00 -8.74520093e-02 -2.62484133e-01 -1.47673059e+00
-8.92189741e-02 -1.28996074e+00 1.04191506e+00 4.53106225e-01
-2.10875645e-01 -1.00064194e+00 1.35253876e-01 -2.09080786e-01
2.26560846e-01 9.08840597e-01 6.09897852e-01 -4.83146220e-01
-8.22733939e-01 1.10307783e-02 4.34076458e-01 3.00229043e-01
-7.90485442e-02 3.48795913e-02 -4.06336904e-01 -7.42498398e-01
4.04411197e-01 -5.66860497e-01 6.80114210e-01 7.45293856e-01
1.15800548e+00 -7.17889011e-01 -4.86794233e-01 6.00306749e-01
1.15040588e+00 5.72807372e-01 2.48823822e-01 3.33625346e-01
4.72724438e-01 7.10241556e-01 1.25251687e+00 7.49229193e-01
6.33102879e-02 4.70218986e-01 5.66808343e-01 6.51943922e-01
9.74742532e-01 -5.50772786e-01 4.03950036e-01 -1.11583546e-01
1.23904109e-01 -1.42127067e-01 -1.02410734e+00 3.11251789e-01
-2.19585967e+00 -6.59482121e-01 3.40279371e-01 2.81258178e+00
9.60245132e-01 4.90756214e-01 1.92914251e-02 -2.87764937e-01
4.33122963e-01 -2.01248989e-01 -1.04229820e+00 -4.00381625e-01
5.14301062e-01 -3.69615927e-02 9.30110395e-01 8.01123083e-01
-1.34613025e+00 5.74495733e-01 6.49748564e+00 7.45148242e-01
-9.16283965e-01 -3.42712760e-01 8.55371654e-01 -2.99391985e-01
-1.34009868e-01 5.26849180e-02 -1.13839638e+00 3.35012525e-01
1.20600188e+00 -5.31055272e-01 8.61173511e-01 1.13237727e+00
6.69454217e-01 -1.36150822e-01 -1.34120977e+00 5.44475973e-01
-5.52091956e-01 -9.81298268e-01 -5.66145539e-01 2.58053869e-01
1.02111876e+00 -2.22036377e-01 4.14677620e-01 4.74590719e-01
1.04891253e+00 -1.36569095e+00 5.92095256e-01 6.82433546e-01
4.76422191e-01 -1.22614098e+00 3.75563830e-01 6.10163331e-01
-9.09150422e-01 -6.60756171e-01 -1.99200049e-01 3.75900008e-02
3.49238038e-01 3.11429560e-01 -8.74170780e-01 1.45351827e-01
3.32113415e-01 4.97770160e-01 1.57468870e-01 1.19051898e+00
-5.01949847e-01 4.67456937e-01 -6.03360832e-01 -1.21305980e-01
4.53966558e-01 -4.18761909e-01 9.44369197e-01 5.11263132e-01
5.28491974e-01 -1.34963915e-01 6.57401919e-01 1.41410244e+00
3.73707443e-01 -5.16050637e-01 -8.01090419e-01 -3.13221902e-01
5.65345764e-01 9.09417391e-01 -4.29504752e-01 2.23951817e-01
3.41641098e-01 3.59972745e-01 1.50991023e-01 4.40694720e-01
-9.67379630e-01 -3.40980053e-01 1.00730014e+00 -2.60488629e-01
2.20179334e-01 -7.26895750e-01 -5.67062438e-01 -5.64049959e-01
-2.03337312e-01 -8.13294888e-01 3.20378751e-01 -2.94602692e-01
-1.10012269e+00 9.80938375e-02 2.96809316e-01 -1.13221407e+00
-6.96315229e-01 -5.43219209e-01 -7.55611241e-01 1.02536535e+00
-1.14662611e+00 -6.82119131e-01 1.67649046e-01 2.73323238e-01
3.31370562e-01 -9.15539172e-03 2.75369316e-01 -6.18972421e-01
-8.73868644e-01 1.39966860e-01 3.30271482e-01 -6.36563122e-01
4.42224592e-01 -1.63253784e+00 1.68683290e-01 9.23562825e-01
-4.93118674e-01 7.39702344e-01 1.20122492e+00 -1.02137983e+00
-1.87794912e+00 -1.46503973e+00 -2.94881493e-01 -7.89771557e-01
1.05564499e+00 -2.33477548e-01 -7.62092292e-01 7.75834084e-01
-3.54564101e-01 -1.53275043e-01 -3.95662487e-02 6.73785284e-02
5.94230331e-02 -6.23350926e-02 -1.10003614e+00 9.37942088e-01
6.63159490e-01 8.47269874e-03 -4.68766809e-01 2.78458357e-01
1.19661987e+00 -6.77782059e-01 -9.95411098e-01 5.04228711e-01
5.03696650e-02 -4.05161344e-02 8.52160156e-01 -8.31149459e-01
3.83342691e-02 -3.72337878e-01 -1.45306185e-01 -1.63302326e+00
-7.86060654e-03 -1.59538543e+00 -6.45815909e-01 7.81889617e-01
3.86489391e-01 -6.67591035e-01 7.00091898e-01 9.56255198e-01
-2.41900131e-01 -9.77278769e-01 -1.04955232e+00 -1.19315088e+00
5.38334608e-01 -3.90267432e-01 2.77275860e-01 2.03046918e-01
1.24288194e-01 6.49406686e-02 -5.95175743e-01 7.51993299e-01
9.32242334e-01 -1.99415423e-02 7.20668316e-01 -1.02114058e+00
-4.12583500e-01 -5.20752549e-01 1.22103766e-01 -8.59685361e-01
4.07218575e-01 -1.07627086e-01 7.82052755e-01 -1.34095311e+00
-3.43658477e-02 -6.49434865e-01 -1.12796992e-01 2.89805055e-01
-2.48741522e-01 -6.67138994e-01 1.54216215e-01 -1.68269023e-01
-6.18263483e-01 8.60006750e-01 1.12761092e+00 -4.85449545e-02
-7.16276050e-01 4.92623359e-01 -9.26860452e-01 4.41665828e-01
8.50443006e-01 -2.23602116e-01 -9.06822681e-01 5.69127053e-02
2.40101323e-01 4.35174316e-01 2.27147594e-01 -9.28706467e-01
1.77379757e-01 -9.36350167e-01 -5.29227629e-02 -5.13651013e-01
3.04023951e-01 -8.98150444e-01 1.03289418e-01 8.12587440e-01
-5.96107185e-01 -2.88193613e-01 3.95886511e-01 1.19611073e+00
2.05775574e-01 -1.77392170e-01 8.86321664e-01 2.55095512e-01
-3.72159511e-01 4.95267630e-01 -2.91350782e-01 3.37667376e-01
1.49185920e+00 3.21842849e-01 -2.75319964e-01 -5.45030534e-01
-7.29193807e-01 1.15041482e+00 4.27207112e-01 4.53280389e-01
4.75902945e-01 -9.56073225e-01 -3.32940340e-01 3.32800834e-03
-1.51298031e-01 4.13443536e-01 -3.45084220e-02 7.47948468e-01
-1.03557773e-01 5.04448414e-01 -4.85298410e-02 -6.42639339e-01
-6.53087080e-01 7.80413508e-01 5.28179467e-01 -2.19778925e-01
-6.30373120e-01 6.94840789e-01 1.53699994e-01 -2.63124377e-01
6.10959768e-01 -8.28507781e-01 1.70916960e-01 -3.33904028e-01
6.56789601e-01 4.60321575e-01 -2.08777085e-01 2.76288688e-01
9.30201858e-02 2.71325469e-01 2.20569059e-01 -2.75808811e-01
1.15859640e+00 -7.10359067e-02 2.70820081e-01 5.67267179e-01
7.08600342e-01 -2.72879571e-01 -2.36136174e+00 2.09347084e-01
2.23532870e-01 -5.14672935e-01 2.60490149e-01 -5.95086277e-01
-5.20377874e-01 6.83898509e-01 3.61231327e-01 3.21693271e-02
6.84452415e-01 -3.35377753e-01 5.35399079e-01 7.43048131e-01
5.55364788e-01 -1.21315515e+00 5.04309796e-02 6.43889487e-01
9.84005928e-01 -1.05367589e+00 -2.02642158e-01 -6.24275543e-02
-1.00729525e+00 1.11043584e+00 7.29784846e-01 -1.78826526e-01
5.50211966e-01 7.14184761e-01 -6.13226593e-01 1.28996328e-01
-1.16195357e+00 4.59717698e-02 2.57150173e-01 7.57505894e-01
-3.80434453e-01 2.04123229e-01 3.24569315e-01 5.97720921e-01
-1.08140677e-01 -2.20667616e-01 5.20310581e-01 1.04886663e+00
-9.12922263e-01 -7.54102111e-01 -6.06247485e-01 5.61449170e-01
-1.79811642e-01 4.31769401e-01 1.19528651e-01 7.03118682e-01
-3.91469598e-01 9.48898256e-01 -5.54307178e-02 -2.57056922e-01
3.00362110e-01 -1.81151897e-01 3.91924232e-01 -7.67524600e-01
4.41541187e-02 1.49219379e-01 2.70810485e-01 -9.06623363e-01
4.65319216e-01 -7.95176625e-01 -1.23988283e+00 -1.69505298e-01
-1.37745619e-01 2.15332314e-01 4.45358247e-01 8.54378700e-01
1.20089225e-01 6.24332547e-01 6.05194688e-01 -8.01187813e-01
-1.65945804e+00 -5.02488494e-01 -6.93647206e-01 -4.15750682e-01
6.94064498e-01 -9.98159945e-01 -4.83463228e-01 -2.90159851e-01] | [4.683619976043701, 2.217881917953491] |
d4ed177a-816b-4057-b90e-daa5b2dbd47c | joint-feature-learning-and-relation-modeling | 2203.11991 | null | https://arxiv.org/abs/2203.11991v4 | https://arxiv.org/pdf/2203.11991v4.pdf | Joint Feature Learning and Relation Modeling for Tracking: A One-Stream Framework | The current popular two-stream, two-stage tracking framework extracts the template and the search region features separately and then performs relation modeling, thus the extracted features lack the awareness of the target and have limited target-background discriminability. To tackle the above issue, we propose a novel one-stream tracking (OSTrack) framework that unifies feature learning and relation modeling by bridging the template-search image pairs with bidirectional information flows. In this way, discriminative target-oriented features can be dynamically extracted by mutual guidance. Since no extra heavy relation modeling module is needed and the implementation is highly parallelized, the proposed tracker runs at a fast speed. To further improve the inference efficiency, an in-network candidate early elimination module is proposed based on the strong similarity prior calculated in the one-stream framework. As a unified framework, OSTrack achieves state-of-the-art performance on multiple benchmarks, in particular, it shows impressive results on the one-shot tracking benchmark GOT-10k, i.e., achieving 73.7% AO, improving the existing best result (SwinTrack) by 4.3\%. Besides, our method maintains a good performance-speed trade-off and shows faster convergence. The code and models are available at https://github.com/botaoye/OSTrack. | ['Xilin Chen', 'Shiguang Shan', 'Bingpeng Ma', 'Hong Chang', 'Botao Ye'] | 2022-03-22 | null | null | null | null | ['visual-object-tracking'] | ['computer-vision'] | [-1.97867632e-01 -4.83943224e-01 -6.36312842e-01 -1.89267308e-01
-7.88785934e-01 -4.06578243e-01 6.86340690e-01 -9.16675478e-02
-4.15674865e-01 2.91286469e-01 -1.07545994e-01 -6.02127239e-02
-2.68646255e-02 -4.36802387e-01 -4.42088306e-01 -8.48834455e-01
-2.24206373e-02 1.34836793e-01 9.79291260e-01 7.71390349e-02
4.95493300e-02 3.05814773e-01 -1.43450105e+00 -3.19768414e-02
6.99490964e-01 1.38933158e+00 2.22130761e-01 4.83231246e-01
1.12312406e-01 5.65840602e-01 -3.93133968e-01 -4.10401344e-01
3.85366619e-01 -8.91076550e-02 -1.14831366e-01 -1.26622409e-01
6.64571881e-01 -2.53134310e-01 -7.95639575e-01 1.12280285e+00
5.90130508e-01 -2.74662077e-02 1.63500860e-01 -1.35043311e+00
-1.34921178e-01 4.02890801e-01 -9.04644489e-01 5.67414701e-01
-1.70253813e-01 4.99782503e-01 9.48544800e-01 -1.04798067e+00
3.41470957e-01 1.11706018e+00 6.99627876e-01 3.35922956e-01
-1.15408170e+00 -1.19703388e+00 4.81020600e-01 3.26517373e-01
-1.49925220e+00 -5.43346047e-01 5.61664641e-01 -4.36963826e-01
4.20925975e-01 2.92356253e-01 7.92147458e-01 9.87246335e-01
1.85552567e-01 9.85911369e-01 8.19185317e-01 3.92672718e-02
-1.47168130e-01 -5.08577563e-02 3.73705089e-01 6.61985338e-01
3.50596100e-01 5.27216434e-01 -7.87925065e-01 -1.51999801e-01
5.86949527e-01 1.88758895e-01 -2.97214329e-01 -4.61297303e-01
-1.36291480e+00 5.53683102e-01 6.04564607e-01 1.92836270e-01
-1.33023327e-02 3.57868701e-01 5.27765155e-01 -1.53886164e-02
2.69677699e-01 -2.01285109e-01 -2.42007434e-01 -1.49370477e-01
-1.11762869e+00 6.14328943e-02 3.54858786e-01 1.10355079e+00
5.88492930e-01 9.64059457e-02 -5.80420375e-01 5.30390084e-01
5.89940488e-01 8.61089766e-01 2.37836167e-01 -5.99546671e-01
4.04917896e-01 5.57873905e-01 7.27428198e-02 -1.17788363e+00
-3.01504791e-01 -9.57133651e-01 -6.61135852e-01 -1.50594190e-02
4.29573953e-01 -5.10044470e-02 -8.23554277e-01 1.58155811e+00
7.56126642e-01 6.77821279e-01 -1.19440839e-01 1.01801205e+00
8.25314522e-01 5.18325031e-01 1.70887813e-01 -3.09607208e-01
1.66355193e+00 -1.21710563e+00 -6.83888733e-01 -3.58926505e-01
4.25308347e-01 -8.71360421e-01 4.95541990e-01 2.08288297e-01
-5.86525023e-01 -7.67395258e-01 -1.03516781e+00 3.33626509e-01
-2.04521734e-02 5.45348763e-01 6.18420839e-01 4.69251156e-01
-6.09346926e-01 2.96374053e-01 -1.05368900e+00 -2.28406683e-01
5.62329412e-01 3.62030864e-01 -1.01065807e-01 -2.59862840e-02
-9.79860067e-01 5.19987583e-01 4.43147421e-01 2.99810588e-01
-1.05319965e+00 -8.11544418e-01 -6.48615956e-01 -1.04872182e-01
9.46995616e-01 -4.03877914e-01 1.23603916e+00 -5.18704891e-01
-1.30656075e+00 3.68648857e-01 -1.53904513e-01 -3.74622732e-01
5.46559274e-01 -4.67329592e-01 -6.19231462e-01 -6.80343136e-02
6.69098496e-02 3.62296820e-01 9.00184929e-01 -9.86535311e-01
-9.92477179e-01 -4.45165843e-01 -2.26254091e-01 -2.25629769e-02
-4.19598639e-01 2.26356998e-01 -1.21470869e+00 -7.86999106e-01
-2.63070464e-02 -1.03886807e+00 -1.97453350e-01 3.12106818e-01
-2.99889684e-01 -1.42162353e-01 9.84186232e-01 -4.33598965e-01
1.49155235e+00 -2.37816525e+00 -1.06307432e-01 1.00108661e-01
4.60284770e-01 6.61479712e-01 3.15356404e-02 1.31674305e-01
2.11457491e-01 -4.83041584e-01 1.04264520e-01 -2.00594619e-01
-2.58096345e-02 -1.38354480e-01 -3.92253399e-02 7.31752038e-01
4.69712093e-02 8.60347807e-01 -9.60202396e-01 -6.75781786e-01
1.93481475e-01 3.53603631e-01 -3.43321234e-01 1.21773459e-01
-3.94496098e-02 3.12508076e-01 -6.88094199e-01 6.47638679e-01
8.41081500e-01 -3.82215679e-01 1.87396258e-01 -5.09446084e-01
-4.67545450e-01 -6.95262775e-02 -1.30272210e+00 1.76617956e+00
-1.54457554e-01 6.59515262e-01 6.88223913e-02 -6.09951913e-01
1.01227403e+00 -2.27699932e-02 5.51420450e-01 -6.96511567e-01
2.13201851e-01 2.98778880e-02 1.96578503e-01 -4.01095934e-02
3.46329331e-01 4.30143178e-01 9.32093784e-02 -2.31469776e-02
-3.57102826e-02 7.10516512e-01 2.08927840e-01 3.70086670e-01
1.14156270e+00 3.13745737e-01 1.88596651e-01 -2.85656393e-01
7.09353805e-01 9.37983990e-02 1.08835673e+00 6.83669209e-01
-4.22176570e-01 1.62476361e-01 1.80746242e-01 -2.02507690e-01
-3.86981249e-01 -9.70363677e-01 -9.56852436e-02 9.61500764e-01
7.72800446e-01 -8.30232799e-01 -3.44805568e-01 -7.33375549e-01
1.62797600e-01 3.18047523e-01 -5.14486909e-01 -2.70462304e-01
-5.87914407e-01 -6.74149871e-01 5.82677543e-01 6.30406857e-01
4.92899746e-01 -5.94858050e-01 -6.15181863e-01 2.77558208e-01
-6.40454749e-03 -1.25987530e+00 -7.29754984e-01 -1.15571074e-01
-7.18094289e-01 -1.21274853e+00 -5.29715598e-01 -4.88073915e-01
3.63502771e-01 6.96032524e-01 8.60009730e-01 1.55542329e-01
-4.52038854e-01 2.46644914e-01 -3.05518806e-01 -5.43646552e-02
1.65423378e-01 -1.06321037e-01 2.24549435e-02 2.46496499e-01
3.78596991e-01 -2.80053139e-01 -7.60452628e-01 7.14791358e-01
-4.91783321e-01 3.24771879e-03 7.64088452e-01 8.64898920e-01
8.07559371e-01 -1.18007883e-01 2.49998391e-01 -2.72893339e-01
-1.41037211e-01 -4.45584834e-01 -1.15700459e+00 2.63581365e-01
-5.55525839e-01 -1.82277992e-01 4.70344424e-01 -6.58014059e-01
-1.15114629e+00 3.45258296e-01 2.47689068e-01 -7.64449596e-01
1.74518660e-01 1.36155579e-02 -2.88670957e-01 -3.63130718e-01
3.78477514e-01 4.50517029e-01 -2.51892656e-02 -6.22254193e-01
1.91418931e-01 4.01365846e-01 5.97088873e-01 -3.70274514e-01
1.30662918e+00 5.96272171e-01 -7.17176124e-02 -6.01104856e-01
-8.88296902e-01 -8.23575974e-01 -3.61789882e-01 -5.22450030e-01
5.30114353e-01 -1.18652225e+00 -9.34711635e-01 5.96576810e-01
-8.09718728e-01 -1.16470419e-01 -4.94001172e-02 8.39099705e-01
-8.61439332e-02 4.79135484e-01 -5.02927840e-01 -8.58937442e-01
-4.73040700e-01 -1.15935826e+00 1.02858949e+00 5.28769255e-01
3.76275390e-01 -5.06488204e-01 -1.19988518e-02 1.12769790e-01
4.37597305e-01 1.26688750e-02 7.96664581e-02 -6.43307984e-01
-9.83231246e-01 -1.73170060e-01 -5.53990543e-01 -6.33537173e-02
-3.23039219e-02 5.78210317e-03 -8.01889062e-01 -5.11355460e-01
-2.29450643e-01 -2.27394216e-02 9.78525937e-01 2.83056170e-01
8.73831928e-01 1.23673454e-01 -7.66291857e-01 7.35615194e-01
1.36743295e+00 1.70457229e-01 3.10631216e-01 3.21437478e-01
7.77260959e-01 2.05705002e-01 1.31519639e+00 6.36191308e-01
3.19807351e-01 1.22484541e+00 5.02615690e-01 6.38466850e-02
-4.24834996e-01 -2.00164393e-01 6.63509309e-01 7.30763137e-01
1.74062774e-01 -1.15684059e-03 -7.48284757e-01 3.69672149e-01
-2.13332057e+00 -9.90757704e-01 -3.87395471e-01 2.29316044e+00
5.64662457e-01 4.38231707e-01 3.56722832e-01 -2.93294609e-01
1.00222671e+00 3.60341638e-01 -7.09114432e-01 5.67657351e-01
1.16953559e-01 -1.43616006e-01 7.91610718e-01 3.48930150e-01
-1.30904460e+00 1.06588674e+00 5.03041697e+00 1.31782246e+00
-1.10947824e+00 2.28073746e-01 1.82926312e-01 -3.50822240e-01
2.90567547e-01 1.74240023e-01 -1.55956376e+00 7.21765220e-01
7.13688314e-01 -2.87037849e-01 -2.96515319e-03 9.04139221e-01
1.94333568e-01 -1.22631721e-01 -8.25566828e-01 1.07835412e+00
-3.60274129e-02 -1.35944772e+00 -3.67580384e-01 1.61047474e-01
1.34863198e-01 2.07994372e-01 -4.74081412e-02 3.98388803e-01
1.14292443e-01 -3.63372803e-01 6.54764771e-01 4.06020164e-01
5.88397622e-01 -6.08106196e-01 6.10497475e-01 3.49682122e-01
-1.93418348e+00 -4.37177196e-02 -3.04177374e-01 3.19527119e-01
1.66465476e-01 7.35732496e-01 -3.17233622e-01 9.91264880e-01
8.61893058e-01 9.43887770e-01 -7.47254848e-01 1.64261186e+00
-9.77934375e-02 6.63847983e-01 -4.90606993e-01 1.94088444e-01
1.88716426e-01 1.01730607e-01 8.37296069e-01 1.31056607e+00
2.28487685e-01 -8.13150629e-02 7.33096659e-01 5.04378200e-01
2.21101150e-01 3.88487950e-02 -1.43335015e-01 3.15953642e-01
6.48494422e-01 1.52872217e+00 -7.65526593e-01 -3.09618026e-01
-5.52170336e-01 4.88919079e-01 1.04092218e-01 6.49665594e-02
-1.47436690e+00 -2.51482338e-01 7.09568024e-01 -6.54316619e-02
7.60813832e-01 -1.54754817e-01 2.57458508e-01 -1.26444626e+00
2.02547774e-01 -7.87324727e-01 4.93133426e-01 -1.66701183e-01
-1.15683365e+00 6.39392555e-01 1.32936677e-02 -1.63901377e+00
1.85937434e-01 -3.41962814e-01 -5.28140426e-01 5.30490935e-01
-1.62698996e+00 -1.24607539e+00 -7.14675844e-01 5.84233344e-01
4.72805858e-01 -7.87818283e-02 1.59672916e-01 6.66677713e-01
-9.48061645e-01 9.25073028e-01 -6.86237663e-02 2.42179260e-01
9.02912617e-01 -7.76882231e-01 4.03672487e-01 1.09271014e+00
4.42371666e-02 4.87835884e-01 4.09017771e-01 -8.16884756e-01
-1.75207007e+00 -1.26310420e+00 2.81794280e-01 -1.16032474e-01
9.53325152e-01 -3.92127931e-01 -8.33320260e-01 4.23624128e-01
-2.45298855e-02 5.23664474e-01 5.85147500e-01 5.84530756e-02
-5.62600195e-01 -4.79190499e-01 -5.78094602e-01 5.02079725e-01
1.29826617e+00 3.51116434e-02 -3.49975795e-01 2.45072380e-01
6.94944978e-01 -5.46956122e-01 -1.00941694e+00 4.43729013e-01
5.76513946e-01 -8.70391726e-01 9.38591778e-01 -1.52289152e-01
-3.31516981e-01 -9.06146049e-01 -2.73310114e-02 -7.16200113e-01
-6.36463106e-01 -8.07150245e-01 -6.73447311e-01 1.41245043e+00
1.51455671e-01 -6.36593878e-01 7.99659848e-01 1.84529394e-01
-1.28386199e-01 -9.82003808e-01 -9.18258011e-01 -1.25860322e+00
-6.52694166e-01 -3.56946826e-01 3.74817282e-01 5.79406619e-01
-4.41071391e-01 3.00464690e-01 -6.85361028e-01 3.83796722e-01
1.08245766e+00 4.19345468e-01 1.00859201e+00 -1.25705945e+00
-5.19105196e-01 -2.85563856e-01 -5.00352144e-01 -1.49302542e+00
-5.88631555e-02 -8.27444434e-01 6.73223510e-02 -1.03837013e+00
3.77049774e-01 -6.36241078e-01 -5.71564794e-01 4.96707320e-01
-3.64041805e-01 1.76123351e-01 5.71446836e-01 4.93590087e-01
-1.12969816e+00 8.14068615e-01 1.09112787e+00 -1.05510041e-01
-4.19577323e-02 1.35541692e-01 -5.46400785e-01 6.22211099e-01
6.50346816e-01 -7.56422877e-01 -1.30128533e-01 -3.07755888e-01
-4.43678051e-01 -1.00295037e-01 4.91757005e-01 -1.27679384e+00
6.93845093e-01 -5.89965060e-02 5.06776273e-01 -8.81365120e-01
4.36681062e-01 -8.59211802e-01 2.57086605e-01 7.87371397e-01
8.56969692e-03 -1.52235940e-01 3.56140912e-01 1.03391171e+00
-2.13417962e-01 4.65534441e-02 8.09126139e-01 3.90777946e-01
-1.03318763e+00 7.33373761e-01 1.18469350e-01 8.79274681e-02
1.30457211e+00 -2.00905219e-01 -4.33368027e-01 6.91304654e-02
-4.19913739e-01 7.04832792e-01 3.30678374e-01 5.42475522e-01
4.10719097e-01 -1.50657141e+00 -5.70765793e-01 -1.05013177e-01
3.05519223e-01 -2.52130896e-01 4.08292592e-01 1.25379658e+00
4.34716456e-02 2.82111317e-01 4.54202555e-02 -9.86390352e-01
-1.42461932e+00 4.63910103e-01 1.40996128e-01 -4.66959417e-01
-9.05341983e-01 6.96622789e-01 3.95286143e-01 1.20867386e-01
3.61857980e-01 1.44006342e-01 -1.12736858e-01 9.25829858e-02
9.00900483e-01 3.53489637e-01 -2.64818192e-01 -7.48586595e-01
-7.62357533e-01 6.83226466e-01 -3.27895075e-01 2.86186308e-01
9.89981055e-01 2.64067575e-02 2.22192019e-01 2.96309646e-02
8.74261856e-01 1.94946185e-01 -1.63563764e+00 -5.64365685e-01
1.17405355e-01 -9.26760495e-01 1.80910215e-01 -4.48526353e-01
-1.48629224e+00 5.13227582e-01 8.29764009e-01 -2.05928847e-01
1.15402973e+00 -8.09105337e-02 8.75658214e-01 1.69881135e-01
2.49661908e-01 -7.58565903e-01 1.27305046e-01 4.42503393e-01
4.69106257e-01 -1.20516682e+00 2.18452111e-01 -6.71127975e-01
-5.01754522e-01 9.25764441e-01 9.76122081e-01 -1.35054141e-01
6.03986561e-01 3.64505678e-01 -1.42516673e-01 -1.84163868e-01
-7.73702860e-01 -4.39428151e-01 4.64783043e-01 4.17314231e-01
1.46366730e-01 -1.08614787e-01 -2.90472776e-01 5.87569773e-01
3.76679301e-01 -1.74686834e-02 -1.35424346e-01 7.54603267e-01
-4.95484740e-01 -1.03804922e+00 -4.31780726e-01 3.32659572e-01
-3.48086149e-01 -1.90105923e-02 -1.36560844e-02 9.57437932e-01
9.28986594e-02 8.68289113e-01 -2.20255658e-01 -6.02785408e-01
4.01801467e-01 -4.25911903e-01 2.50245094e-01 -3.08432221e-01
-7.09908128e-01 4.67694312e-01 3.31844166e-02 -1.11602497e+00
-4.48282331e-01 -6.58035517e-01 -1.07362521e+00 -1.12795688e-01
-7.89370120e-01 2.29806721e-01 4.32676196e-01 6.80998862e-01
6.82145655e-01 6.38209760e-01 5.74884951e-01 -7.99729705e-01
-6.30879283e-01 -6.71189487e-01 -2.02610955e-01 -4.28414010e-02
1.22855090e-01 -1.13590801e+00 -1.54622316e-01 -2.95853645e-01] | [6.306602478027344, -2.134387969970703] |
9cf6582d-c026-4dbd-8af1-d4af5d481d76 | a-simple-language-model-for-task-oriented | 2005.00796 | null | https://arxiv.org/abs/2005.00796v4 | https://arxiv.org/pdf/2005.00796v4.pdf | A Simple Language Model for Task-Oriented Dialogue | Task-oriented dialogue is often decomposed into three tasks: understanding user input, deciding actions, and generating a response. While such decomposition might suggest a dedicated model for each sub-task, we find a simple, unified approach leads to state-of-the-art performance on the MultiWOZ dataset. SimpleTOD is a simple approach to task-oriented dialogue that uses a single, causal language model trained on all sub-tasks recast as a single sequence prediction problem. This allows SimpleTOD to fully leverage transfer learning from pre-trained, open domain, causal language models such as GPT-2. SimpleTOD improves over the prior state-of-the-art in joint goal accuracy for dialogue state tracking, and our analysis reveals robustness to noisy annotations in this setting. SimpleTOD also improves the main metrics used to evaluate action decisions and response generation in an end-to-end setting: inform rate by 8.1 points, success rate by 9.7 points, and combined score by 7.2 points. | ['Chien-Sheng Wu', 'Ehsan Hosseini-Asl', 'Richard Socher', 'Semih Yavuz', 'Bryan McCann'] | 2020-05-02 | null | http://proceedings.neurips.cc/paper/2020/hash/e946209592563be0f01c844ab2170f0c-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/e946209592563be0f01c844ab2170f0c-Paper.pdf | neurips-2020-12 | ['end-to-end-dialogue-modelling'] | ['natural-language-processing'] | [ 2.49448583e-01 7.82426715e-01 -3.72434527e-01 -4.41693008e-01
-1.29635978e+00 -7.57999718e-01 1.07836115e+00 1.31905779e-01
-3.00053775e-01 1.01590610e+00 8.33168805e-01 -2.43805960e-01
1.32677957e-01 -4.08018768e-01 -2.69502938e-01 -2.26893276e-01
1.56013861e-01 8.04024935e-01 2.67906696e-01 -5.83917737e-01
2.71921396e-01 -4.95725781e-01 -7.81627774e-01 7.65447795e-01
6.54578269e-01 7.57519543e-01 -1.71862617e-01 1.24772072e+00
-1.17627889e-01 1.36604965e+00 -7.59759426e-01 -4.95951265e-01
-1.31863520e-01 -7.05243230e-01 -1.53529155e+00 3.60931791e-02
1.33236110e-01 -5.20780563e-01 -2.95147240e-01 4.33928818e-01
7.04543412e-01 2.14009374e-01 5.07100046e-01 -1.09145141e+00
-3.63789201e-01 7.83006132e-01 -2.30308250e-01 -2.07352638e-01
9.81355846e-01 6.28421903e-01 1.28663814e+00 -3.25707257e-01
6.39620960e-01 1.62865007e+00 7.98601389e-01 1.02828491e+00
-1.41338861e+00 -2.12760061e-01 1.31950125e-01 -2.01504171e-01
-3.52332503e-01 -7.50641108e-01 2.96396822e-01 -5.43869257e-01
1.31013966e+00 2.62763560e-01 2.30851680e-01 1.52007663e+00
4.44699498e-03 1.06338251e+00 1.27736640e+00 -4.22861099e-01
-2.65925061e-02 -1.53694838e-01 4.14032131e-01 6.64194763e-01
-5.92376053e-01 -2.50549972e-01 -9.10113037e-01 -5.04791975e-01
3.92910540e-01 -4.66469079e-01 -1.62101194e-01 2.08824024e-01
-1.19905424e+00 9.67324138e-01 -1.84223339e-01 4.01348323e-02
-3.56051326e-01 1.83693856e-01 7.16261804e-01 4.43006575e-01
7.30016947e-01 5.99852204e-01 -8.07007194e-01 -9.27414954e-01
-4.59656179e-01 6.01011276e-01 1.39825332e+00 7.22290754e-01
1.91365302e-01 -2.05097333e-01 -7.50081480e-01 8.90590787e-01
1.31322414e-01 2.15733692e-01 3.07813287e-01 -1.20390499e+00
8.21311772e-01 5.54239988e-01 3.12171251e-01 -3.33110213e-01
-7.41020381e-01 7.14658275e-02 -3.32445532e-01 -2.65824825e-01
1.11140108e+00 -7.73462534e-01 -3.57215703e-01 1.96560466e+00
2.76262373e-01 -3.54309589e-01 3.45173329e-01 7.64519393e-01
7.65534580e-01 7.05459833e-01 2.33655140e-01 -1.80166379e-01
1.40651524e+00 -1.11181486e+00 -7.90500760e-01 -5.00081658e-01
1.02344596e+00 -6.51350915e-01 1.26963711e+00 4.69391644e-01
-1.27325141e+00 -1.57814831e-01 -5.98715186e-01 -3.21815819e-01
1.41804934e-01 6.39291853e-02 6.85992897e-01 4.23854947e-01
-9.99951363e-01 5.96758425e-01 -4.50434506e-01 -4.64340627e-01
-9.22180153e-03 8.67368504e-02 -2.23916098e-01 2.64750063e-01
-1.46769357e+00 1.18736434e+00 2.43328869e-01 -3.73330086e-01
-1.08354056e+00 -8.06100905e-01 -6.49144948e-01 -8.68038759e-02
6.75169945e-01 -7.20238566e-01 2.27517438e+00 -4.17056382e-01
-2.09241128e+00 8.34833801e-01 -2.01043546e-01 -6.37847960e-01
7.05464602e-01 -6.01364851e-01 1.24399737e-01 1.48992259e-02
9.38387364e-02 4.25656319e-01 4.19184148e-01 -7.02264607e-01
-6.27206802e-01 -3.32803786e-01 3.56304586e-01 3.36195052e-01
-1.53281450e-01 3.53561610e-01 -2.50986546e-01 -1.17459156e-01
-4.35626566e-01 -9.95776772e-01 -6.86045215e-02 -3.77960145e-01
-5.87407947e-01 -8.66238952e-01 4.79072303e-01 -9.70507681e-01
1.29762471e+00 -1.59026396e+00 2.82789946e-01 -7.37460673e-01
2.60915130e-01 1.78577498e-01 -6.03563897e-02 7.70660996e-01
3.97690795e-02 2.15142667e-01 -1.86924666e-01 -5.63542783e-01
2.29604453e-01 -7.40639791e-02 -2.11503476e-01 1.44500062e-01
4.47248638e-01 9.87094462e-01 -1.28637457e+00 -2.22623333e-01
9.46939439e-02 3.32894772e-02 -5.88731945e-01 6.47835016e-01
-8.02054465e-01 4.57382143e-01 -5.24029493e-01 2.89834552e-02
-1.31168008e-01 -3.04343551e-01 4.29314345e-01 3.40670526e-01
1.29997088e-02 1.29133761e+00 -5.93587101e-01 2.03670549e+00
-6.70798182e-01 6.61010027e-01 2.81608701e-02 -6.57512963e-01
7.73792803e-01 8.05136263e-01 3.06113750e-01 -4.99374539e-01
1.99689884e-02 -7.11782724e-02 9.38880742e-02 -6.98962688e-01
4.87748325e-01 -6.89027086e-02 -5.90972602e-01 1.08622682e+00
4.54193860e-01 -1.20190002e-01 1.97335169e-01 5.54964066e-01
1.31798637e+00 3.28942418e-01 3.19600224e-01 -8.17631409e-02
2.21095175e-01 2.80513108e-01 2.98499167e-01 6.99609160e-01
-1.53867051e-01 3.12492579e-01 1.07910907e+00 -3.60557884e-01
-9.03069317e-01 -4.31352466e-01 3.27849388e-01 1.70154130e+00
-4.96380746e-01 -6.96759343e-01 -9.58234727e-01 -1.01953888e+00
-1.68586820e-01 1.13512480e+00 -6.64440334e-01 -6.92761391e-02
-4.57370788e-01 -7.49890745e-01 9.64805543e-01 1.98032856e-01
5.50881207e-01 -8.88483882e-01 -5.02307713e-01 4.36129630e-01
-7.69224524e-01 -1.14177084e+00 -5.13395011e-01 9.26088542e-02
-5.33092856e-01 -1.22641885e+00 -6.15967035e-01 -7.79813156e-02
-1.01051286e-01 -4.90833598e-04 1.41223145e+00 -6.86241761e-02
1.33043498e-01 4.51683939e-01 -3.85232240e-01 -2.29878321e-01
-9.58279312e-01 1.46416008e-01 -1.09388620e-01 -7.60924369e-02
2.32311666e-01 -2.54078652e-03 -3.77141058e-01 2.56411076e-01
-1.56127378e-01 2.72448659e-01 9.65004414e-02 1.03362799e+00
-3.27059746e-01 -7.66604602e-01 6.31588161e-01 -1.09239352e+00
1.28963923e+00 -5.28198183e-01 -1.04877232e-02 1.95691243e-01
-6.13236964e-01 2.74916381e-01 3.19242150e-01 -2.89127022e-01
-1.39677918e+00 -2.94110805e-01 -1.69845998e-01 1.08688347e-01
-3.30728084e-01 3.72397453e-01 1.67536363e-01 6.44472599e-01
1.01369798e+00 1.64260149e-01 2.71392554e-01 -4.23729300e-01
5.97735941e-01 8.77346396e-01 4.70143199e-01 -7.11154878e-01
1.90503478e-01 -7.90199265e-02 -3.96127373e-01 -5.79483390e-01
-1.19431424e+00 -3.64057541e-01 -4.86733675e-01 -1.97782889e-01
1.03787935e+00 -9.23778176e-01 -1.14240134e+00 4.68114555e-01
-1.58457661e+00 -1.21209574e+00 6.07881993e-02 -1.04710691e-01
-8.22335064e-01 3.36907059e-01 -8.98016989e-01 -1.22860706e+00
-6.41807675e-01 -8.57466817e-01 1.15944266e+00 -5.18934354e-02
-9.44068015e-01 -1.22091103e+00 1.73891887e-01 7.83737838e-01
2.16243446e-01 1.02081019e-02 6.64101362e-01 -1.14624465e+00
-1.29114022e-03 -8.03603232e-02 -3.58199999e-02 1.93571895e-01
3.05494033e-02 -1.86228782e-01 -1.03211343e+00 4.35368828e-02
-1.27109140e-01 -1.02483988e+00 7.15601027e-01 -4.65091579e-02
4.90845919e-01 -5.48492312e-01 -1.72204614e-01 -2.58994579e-01
4.98745531e-01 1.51949346e-01 2.71888942e-01 1.72780585e-02
4.02979463e-01 1.09278154e+00 8.10464740e-01 4.16667521e-01
9.00082529e-01 9.27603483e-01 1.86745420e-01 1.04668990e-01
-1.44213304e-01 -3.64363372e-01 6.71310008e-01 3.18067402e-01
4.03434858e-02 -4.50857103e-01 -1.09602201e+00 5.30498981e-01
-2.34461164e+00 -1.09920394e+00 -3.76753628e-01 1.88692856e+00
1.44244719e+00 2.23438278e-01 5.85718513e-01 -1.32699758e-01
5.03152847e-01 3.67834926e-01 -2.57176518e-01 -6.33081138e-01
2.55571038e-01 -2.01807600e-02 8.66709724e-02 7.39937305e-01
-1.00315654e+00 1.21184778e+00 6.71619225e+00 7.38130212e-01
-9.42723811e-01 4.28163201e-01 7.97094524e-01 -8.23777989e-02
-8.43176793e-04 6.63406923e-02 -9.13218975e-01 3.98201644e-01
1.26523399e+00 -1.55709296e-01 2.49197811e-01 6.60924911e-01
5.11214674e-01 -1.81965113e-01 -1.28543019e+00 4.50221002e-01
-1.07329972e-01 -1.25715411e+00 -3.12024802e-01 -8.16008747e-02
4.68095422e-01 -6.85693324e-02 -2.56424099e-01 6.56594098e-01
1.03558493e+00 -8.50914538e-01 4.57464784e-01 3.05163592e-01
7.20832109e-01 -2.61586964e-01 5.08344293e-01 7.98809111e-01
-3.75931978e-01 -2.17713485e-03 7.10657015e-02 -4.89502460e-01
2.94533283e-01 1.60592705e-01 -1.35057366e+00 3.72908652e-01
2.61479765e-01 5.31505466e-01 -2.36171022e-01 3.53738695e-01
-6.01850986e-01 9.27542567e-01 -7.01129586e-02 -2.88371742e-01
4.67840672e-01 1.17838964e-01 5.37806630e-01 1.45699954e+00
-3.72778028e-01 2.91332662e-01 3.25172156e-01 6.22559965e-01
-5.99976704e-02 -6.97891563e-02 -4.63301569e-01 -2.63301730e-01
3.19414586e-01 1.17045903e+00 1.74690783e-02 -3.92323971e-01
-2.46556967e-01 9.90113497e-01 4.38678682e-01 4.51102220e-02
-6.87891841e-01 -1.41666695e-01 6.17458522e-01 -1.21303648e-01
-1.93698779e-01 -1.20772839e-01 -3.15115184e-01 -9.85196292e-01
-2.67071575e-01 -1.18915677e+00 5.15123844e-01 -7.70086467e-01
-1.10959148e+00 3.55550200e-01 4.70323339e-02 -7.13064313e-01
-9.91656661e-01 -4.17306125e-01 -8.34912360e-01 1.14160085e+00
-1.12209570e+00 -1.15331149e+00 -4.56979647e-02 4.45533067e-01
1.16617107e+00 -1.03283804e-02 1.23473585e+00 -2.14322537e-01
-6.70155346e-01 5.34480274e-01 -4.66526866e-01 2.97852933e-01
1.20555723e+00 -1.56446671e+00 6.27260208e-01 7.25415826e-01
-3.28945369e-01 5.13166428e-01 8.61086667e-01 -5.99762380e-01
-1.33131385e+00 -8.49238753e-01 1.32296002e+00 -1.06124985e+00
9.51837420e-01 -5.56624651e-01 -7.22532332e-01 7.22255409e-01
6.18010581e-01 -5.83526075e-01 6.85754299e-01 7.86238730e-01
-4.90801811e-01 5.25652230e-01 -9.07230735e-01 7.08664894e-01
9.70999777e-01 -6.12238050e-01 -7.95728326e-01 7.25272238e-01
9.45834816e-01 -7.52219975e-01 -1.12398493e+00 -1.56857803e-01
5.18645644e-01 -8.44286263e-01 6.14490509e-01 -1.18211985e+00
9.44519341e-01 4.01878238e-01 1.00607201e-01 -1.48818660e+00
-2.15519935e-01 -1.48696780e+00 -3.58419985e-01 1.13636792e+00
8.23830843e-01 -4.58763689e-01 5.76993644e-01 1.03264666e+00
-2.12197721e-01 -7.08476782e-01 -5.96337795e-01 -2.98052132e-01
2.23789901e-01 -4.37159956e-01 1.54168293e-01 9.08519387e-01
7.26015568e-01 1.24270523e+00 -6.67028308e-01 -3.49183679e-01
3.26490283e-01 -6.54275194e-02 1.12552643e+00 -1.18314421e+00
-5.22047579e-01 -4.31134790e-01 3.82872969e-01 -1.54066396e+00
2.84173906e-01 -6.41950071e-01 1.86463520e-01 -1.70360529e+00
6.21751323e-02 -3.48542452e-01 2.52220899e-01 8.78554940e-01
-5.44812322e-01 -1.86244741e-01 2.86358237e-01 2.44258530e-02
-9.37467873e-01 5.25356054e-01 1.16217899e+00 -8.90956447e-02
-3.02148998e-01 9.06226858e-02 -9.62050676e-01 5.88428319e-01
8.91839564e-01 -3.35065782e-01 -3.92687589e-01 -4.18622911e-01
1.13285340e-01 7.67534912e-01 2.72637069e-01 -4.08651203e-01
2.93957740e-01 -3.06402951e-01 -2.30749577e-01 -1.26248956e-01
5.83868623e-01 7.48883337e-02 -4.45939183e-01 3.60650510e-01
-9.41273391e-01 -3.13362449e-01 2.54357308e-01 4.72904444e-01
4.47893515e-02 -2.22870126e-01 4.76840019e-01 -3.70841563e-01
-3.94570678e-01 -1.47069335e-01 -6.00517988e-01 5.52213252e-01
7.52026498e-01 -9.42825433e-03 -7.23761439e-01 -9.18161869e-01
-6.92118168e-01 5.96898317e-01 -1.35907754e-01 6.05924010e-01
1.36827186e-01 -6.83989882e-01 -1.02718472e+00 -5.03364086e-01
8.47460330e-02 -1.09122343e-01 3.85482833e-02 9.94315267e-01
7.87616372e-02 8.27016473e-01 1.34364158e-01 -5.79334557e-01
-1.40504050e+00 2.16952544e-02 3.02751392e-01 -8.29904318e-01
-3.04643929e-01 1.03877127e+00 2.74597015e-02 -7.02863097e-01
1.81632906e-01 -6.34390041e-02 -1.44906551e-01 1.21266425e-01
4.65827167e-01 1.87527239e-01 -5.85982129e-02 -2.39581645e-01
-2.29682401e-02 -1.36589780e-01 -1.42422557e-01 -4.70145911e-01
1.15718102e+00 -1.23588987e-01 6.73189759e-02 7.75859535e-01
6.85612500e-01 -2.20196113e-01 -1.44718289e+00 -3.56578171e-01
7.76372552e-02 -6.24557631e-03 -8.96330699e-02 -1.38334203e+00
-1.77912280e-01 1.01364160e+00 -1.28949925e-01 5.88469148e-01
6.21845186e-01 -1.41389862e-01 7.32072473e-01 4.89343941e-01
3.82240266e-01 -9.32343721e-01 4.95529801e-01 1.06780863e+00
1.00221384e+00 -1.34414411e+00 -2.26383850e-01 -2.69497424e-01
-1.17675209e+00 9.44415629e-01 7.00085819e-01 2.34535038e-01
-5.79318916e-03 1.01497069e-01 4.17406261e-02 -1.60259008e-01
-1.59670234e+00 -7.22749606e-02 -5.06931692e-02 3.73316884e-01
9.10313785e-01 4.86705303e-02 -2.34338850e-01 7.73764789e-01
-1.71798989e-01 -1.19162381e-01 5.63536882e-01 7.30830073e-01
-2.60892749e-01 -1.20648611e+00 -7.13963956e-02 3.26220781e-01
-5.85029900e-01 -1.86334208e-01 -9.74456787e-01 7.14031994e-01
-7.09296942e-01 1.50356686e+00 -1.04170948e-01 -5.20214915e-01
4.97135609e-01 7.06023276e-01 5.05201638e-01 -9.01514053e-01
-1.21387172e+00 -1.17268696e-01 1.06030226e+00 -6.75302267e-01
-2.53634691e-01 -6.73391759e-01 -1.11615539e+00 -4.65691715e-01
-2.35200927e-01 3.46833378e-01 3.36171836e-01 1.02732778e+00
5.02055585e-01 5.34325361e-01 4.66992557e-01 -5.68518460e-01
-9.87695277e-01 -1.49680674e+00 1.18609279e-01 3.99507523e-01
2.60524333e-01 -4.89954472e-01 -8.79188552e-02 5.51234856e-02] | [12.768430709838867, 8.057605743408203] |
d4b62751-87e1-4c8e-b6cd-0321e05f2356 | layoutreader-pre-training-of-text-and-layout | 2108.11591 | null | https://arxiv.org/abs/2108.11591v2 | https://arxiv.org/pdf/2108.11591v2.pdf | LayoutReader: Pre-training of Text and Layout for Reading Order Detection | Reading order detection is the cornerstone to understanding visually-rich documents (e.g., receipts and forms). Unfortunately, no existing work took advantage of advanced deep learning models because it is too laborious to annotate a large enough dataset. We observe that the reading order of WORD documents is embedded in their XML metadata; meanwhile, it is easy to convert WORD documents to PDFs or images. Therefore, in an automated manner, we construct ReadingBank, a benchmark dataset that contains reading order, text, and layout information for 500,000 document images covering a wide spectrum of document types. This first-ever large-scale dataset unleashes the power of deep neural networks for reading order detection. Specifically, our proposed LayoutReader captures the text and layout information for reading order prediction using the seq2seq model. It performs almost perfectly in reading order detection and significantly improves both open-source and commercial OCR engines in ordering text lines in their results in our experiments. We will release the dataset and model at \url{https://aka.ms/layoutreader}. | ['Furu Wei', 'Jingbo Shang', 'Lei Cui', 'Yiheng Xu', 'Zilong Wang'] | 2021-08-26 | null | https://aclanthology.org/2021.emnlp-main.389 | https://aclanthology.org/2021.emnlp-main.389.pdf | emnlp-2021-11 | ['document-layout-analysis'] | ['computer-vision'] | [ 3.59322548e-01 -2.96602428e-01 -5.14022633e-02 -3.22077215e-01
-7.15715587e-01 -1.10837662e+00 7.36363769e-01 3.40729982e-01
-2.30690375e-01 4.61350009e-02 4.46917146e-01 -8.46775651e-01
6.66609854e-02 -7.70594954e-01 -1.13983810e+00 -1.55938655e-01
3.61129582e-01 3.34444314e-01 7.71388486e-02 -6.24380484e-02
8.32144856e-01 4.99807537e-01 -1.30613148e+00 6.79849505e-01
1.03164899e+00 1.00899565e+00 6.77622497e-01 1.35397243e+00
-5.44975460e-01 7.62564003e-01 -8.50587189e-01 -3.33663046e-01
7.70459548e-02 -3.49202126e-01 -1.00380063e+00 5.18944813e-03
1.01367962e+00 -9.39654648e-01 -5.79640388e-01 8.83495331e-01
5.74925959e-01 -2.59689093e-01 6.87981308e-01 -8.46928596e-01
-1.53647590e+00 7.59372115e-01 -5.95055640e-01 8.35836977e-02
5.51252961e-01 1.61753967e-01 1.43732941e+00 -7.57293582e-01
6.36394680e-01 1.03643394e+00 2.74726540e-01 2.76806474e-01
-5.28217614e-01 -2.33980030e-01 -1.05799943e-01 3.15079212e-01
-9.23964202e-01 -5.09880900e-01 5.43478012e-01 -5.48623025e-01
1.08318269e+00 5.06374776e-01 5.88935256e-01 1.12153590e+00
2.37096220e-01 1.59227872e+00 7.81486452e-01 -8.23581100e-01
-4.83384579e-02 -3.95527840e-01 6.16407573e-01 9.06837404e-01
1.36168256e-01 -3.50960910e-01 -4.89058375e-01 5.62274396e-01
5.52729964e-01 -5.41226380e-02 -2.65144378e-01 3.03388517e-02
-1.13851464e+00 2.86471605e-01 1.92536414e-01 2.65956931e-02
2.44214967e-01 -2.54355986e-02 3.37355763e-01 2.30960950e-01
1.88037783e-01 5.59176445e-01 -3.65945637e-01 -3.37435275e-01
-9.25381839e-01 7.51447901e-02 7.20277011e-01 1.15373659e+00
4.72248644e-01 -3.00214738e-01 -2.52378374e-01 9.08801198e-01
1.83853179e-01 6.18671656e-01 2.86921024e-01 -7.42143512e-01
8.43444705e-01 7.22775519e-01 -5.44831380e-02 -7.48264134e-01
-4.48136181e-01 -1.80186659e-01 -7.97073185e-01 -1.18537687e-01
6.20576560e-01 1.48689240e-01 -8.45400393e-01 9.22813177e-01
-1.32731512e-01 -5.60671806e-01 -4.09021050e-01 6.85847998e-01
8.57266068e-01 8.78041685e-01 -5.03514528e-01 4.41563398e-01
1.55206382e+00 -1.16115379e+00 -6.98848724e-01 -4.10401762e-01
5.47144711e-01 -1.07697368e+00 1.69276023e+00 6.31905019e-01
-1.09890831e+00 -6.48054540e-01 -1.30475688e+00 -7.91637480e-01
-6.41800404e-01 6.02445960e-01 5.21452546e-01 5.05466402e-01
-8.53402674e-01 4.47831988e-01 -6.93492711e-01 -4.98724461e-01
6.02126181e-01 -1.58738375e-01 -8.43137801e-02 -2.50746638e-01
-8.75272393e-01 2.88175553e-01 1.45118058e-01 2.49305427e-01
-4.20557380e-01 -5.95623672e-01 -7.14307368e-01 1.46612599e-01
2.51969963e-01 -3.31202567e-01 1.49278033e+00 -6.18765414e-01
-1.28922355e+00 9.03046310e-01 -2.92139739e-01 -9.74771604e-02
6.05641484e-01 -6.66254878e-01 -4.11723673e-01 8.92640352e-02
-1.56428874e-01 5.83333313e-01 7.93042839e-01 -9.94953871e-01
-6.58863246e-01 -5.03508747e-01 -7.70950411e-03 -8.73789415e-02
-5.71619928e-01 5.80391362e-02 -8.57167602e-01 -5.61463177e-01
-3.63016665e-01 -5.50806999e-01 6.33080423e-01 1.03315532e-01
-1.09329009e+00 -3.18522453e-01 6.97514713e-01 -1.11343408e+00
1.34313023e+00 -2.15846920e+00 -2.56384164e-01 -9.31671783e-02
3.58725429e-01 3.03150535e-01 -4.55480099e-01 6.35996521e-01
2.17924312e-01 4.13293302e-01 2.53796309e-01 -1.25267178e-01
5.69286883e-01 -2.18096510e-01 -4.50641185e-01 3.24923605e-01
-1.61772370e-02 1.32453763e+00 -6.61415458e-01 -4.25424933e-01
8.18107054e-02 1.08587503e-01 -2.93703437e-01 3.14965576e-01
-5.61726868e-01 -2.18352839e-01 -7.82402158e-02 8.99861157e-01
7.26915359e-01 -6.71154976e-01 1.36000767e-01 -7.89459348e-02
-2.56485879e-01 5.70539117e-01 -8.64504755e-01 1.64261472e+00
-3.85769069e-01 1.40638089e+00 -1.82783887e-01 -3.92656982e-01
6.36898518e-01 -3.84630680e-01 3.44226621e-02 -1.29602504e+00
-2.86458228e-02 -2.22403109e-02 -1.54068982e-02 -5.06097794e-01
1.10000861e+00 1.13309789e+00 -7.94631019e-02 6.99509263e-01
-4.55117226e-01 1.47248968e-01 5.77069283e-01 5.82191229e-01
1.21975315e+00 1.20323129e-01 6.61587119e-02 4.52721771e-03
9.54365954e-02 -2.03498259e-01 -4.15768214e-02 1.00298691e+00
2.17331439e-01 6.79795742e-01 6.48750067e-01 -5.06231427e-01
-1.47010946e+00 -1.13532114e+00 -2.14516316e-02 1.52229440e+00
-4.86401618e-02 -5.52347124e-01 -8.65991056e-01 -4.81732816e-01
-4.16523777e-02 5.75041890e-01 -4.05757993e-01 1.43112302e-01
-7.73018301e-01 -3.73589188e-01 7.96402693e-01 9.67642367e-01
3.92445475e-01 -1.06166971e+00 -3.16548467e-01 -1.63144886e-01
-5.75260036e-02 -1.08540988e+00 -8.99585664e-01 2.23875523e-01
-2.77657837e-01 -1.23043597e+00 -7.43567944e-01 -1.02527082e+00
7.09500015e-01 1.62743226e-01 1.33439660e+00 2.04202145e-01
-5.64690411e-01 2.06102133e-01 -4.90446717e-01 -5.94208598e-01
-5.25166988e-01 4.21728313e-01 -6.17708445e-01 -4.57267314e-01
5.25732815e-01 2.17136785e-01 -7.05369592e-01 -5.17284237e-02
-1.04263806e+00 1.52368367e-01 7.83787370e-01 6.30411148e-01
3.59580666e-01 1.88177135e-02 -3.39337289e-02 -9.11964476e-01
9.44046199e-01 5.99277616e-02 -9.61938143e-01 7.11295187e-01
-5.64840436e-01 4.17523794e-02 8.06794882e-01 -4.75822985e-02
-9.01045859e-01 -1.31581083e-01 -2.21812010e-01 2.96156555e-01
-3.47235948e-01 2.59173691e-01 -4.86311108e-01 5.05876660e-01
2.40502611e-01 5.69750249e-01 -4.08069491e-01 -7.78345883e-01
4.98798728e-01 1.12436068e+00 8.24926496e-01 -4.48034853e-01
7.10571945e-01 1.66574717e-01 -3.35167110e-01 -9.81619954e-01
-8.62788916e-01 -4.43466693e-01 -9.16793823e-01 -1.22759007e-02
7.40107119e-01 -5.47393441e-01 -9.22887444e-01 1.02469075e+00
-1.43263686e+00 -6.28441095e-01 1.21919684e-01 -2.27421954e-01
-1.58310384e-01 7.92799771e-01 -9.36712146e-01 -5.28689206e-01
-4.68488723e-01 -1.08837318e+00 1.42090356e+00 2.26246268e-01
-2.02639520e-01 -6.90050304e-01 -1.70360152e-02 5.45473635e-01
-2.35130750e-02 -4.24724430e-01 1.49702263e+00 -5.98757148e-01
-8.91461134e-01 -1.73567787e-01 -7.23434806e-01 3.23975712e-01
6.84191436e-02 5.83756924e-01 -8.39634240e-01 1.19855050e-02
-9.51695025e-01 -2.94811308e-01 1.07891953e+00 2.62531132e-01
1.82814300e+00 -2.73134112e-01 -1.22538924e-01 7.18643785e-01
1.28129077e+00 3.32294971e-01 9.40061212e-01 6.74176276e-01
1.15791512e+00 3.51024061e-01 3.10560614e-01 6.29620910e-01
6.10910237e-01 2.93356001e-01 3.75052780e-01 -1.46653056e-01
-3.16662997e-01 -5.46697438e-01 1.71497032e-01 9.03382957e-01
3.93163145e-01 -1.14765346e+00 -1.19968259e+00 4.80182797e-01
-1.38333952e+00 -8.71391535e-01 -3.82265538e-01 1.77166128e+00
8.61365616e-01 1.55399173e-01 -8.11107010e-02 2.10154355e-01
5.68590164e-01 3.31072122e-01 -5.82104087e-01 -5.90413749e-01
-2.79866457e-01 -4.82787937e-03 5.52878201e-01 3.08617711e-01
-1.21912432e+00 8.53803456e-01 6.13483429e+00 6.27169192e-01
-8.74654710e-01 -5.26721239e-01 7.59758353e-01 -3.55822369e-02
-4.70176935e-01 -4.13009495e-01 -1.21849620e+00 7.88715065e-01
7.35723674e-01 2.98163801e-01 5.82046509e-01 5.86969912e-01
3.00787300e-01 -1.41276419e-01 -1.38971066e+00 8.03353190e-01
1.94698423e-01 -1.56780612e+00 9.52042118e-02 1.64151296e-01
4.60797757e-01 -6.77423403e-02 3.25812966e-01 5.70154004e-02
3.83949250e-01 -1.21653736e+00 9.78735983e-01 5.73515236e-01
9.97902393e-01 -4.56702203e-01 4.61340308e-01 1.89766005e-01
-9.89489436e-01 -3.70159820e-02 -4.96181279e-01 6.10189736e-02
9.85903852e-03 6.69160426e-01 -7.43305206e-01 1.39294416e-01
6.78987741e-01 7.80891597e-01 -1.08412397e+00 1.00468051e+00
-3.29553336e-01 4.99850243e-01 -1.69860553e-02 -3.82740438e-01
2.56444961e-01 -1.95637383e-02 3.59762646e-02 1.39806938e+00
2.68231153e-01 -4.47201908e-01 -2.40470488e-02 6.36458278e-01
-7.19061613e-01 1.58410832e-01 -3.29169691e-01 -7.29524910e-01
6.14442825e-01 1.14569807e+00 -7.10659266e-01 -3.03454340e-01
-4.82143372e-01 1.18063688e+00 3.85696888e-01 2.27230772e-01
-6.55877590e-01 -9.60284710e-01 4.01257366e-01 1.08283348e-01
4.27095324e-01 -3.80430102e-01 -3.85453850e-01 -1.06223285e+00
4.11313534e-01 -1.07022727e+00 1.95548370e-01 -1.14635217e+00
-1.08305085e+00 2.90748894e-01 -6.52133822e-01 -9.70519543e-01
-1.16784737e-01 -1.16195047e+00 -5.14832675e-01 5.48482001e-01
-1.51138067e+00 -1.04180551e+00 -6.20639443e-01 1.75295129e-01
9.92834687e-01 -7.92671517e-02 3.66924465e-01 2.62829125e-01
-8.77743840e-01 8.38359475e-01 8.39349270e-01 9.32128906e-01
9.18008029e-01 -1.86198044e+00 9.92131770e-01 9.06855941e-01
4.39967692e-01 7.79862940e-01 2.15133533e-01 -5.21758080e-01
-1.91038167e+00 -8.68861377e-01 8.64490807e-01 -5.39788187e-01
7.58802295e-01 -9.24557447e-01 -7.94861674e-01 7.51561344e-01
7.32131481e-01 -5.15923858e-01 6.81404591e-01 1.16211444e-01
-5.96637189e-01 -7.86990076e-02 -3.21686685e-01 7.67663360e-01
1.12786198e+00 -7.05541015e-01 -4.30656493e-01 5.88124812e-01
7.55233586e-01 -5.57841778e-01 -7.28306174e-01 -2.49390453e-01
8.98159683e-01 -9.28990722e-01 7.36920953e-01 -4.17656571e-01
1.00923944e+00 6.55343235e-02 4.32486758e-02 -1.17316699e+00
-3.12505245e-01 -5.65893650e-01 -3.13919485e-01 1.46010172e+00
5.00858665e-01 -1.45633981e-01 8.13765585e-01 2.43411839e-01
-2.91830599e-01 -6.15161240e-01 -8.68908018e-02 -7.17230976e-01
2.33101502e-01 -4.91848230e-01 1.01076448e+00 4.52635080e-01
-1.18632235e-01 4.16931599e-01 -1.77163586e-01 5.91915920e-02
3.79442185e-01 4.24464852e-01 8.12093616e-01 -9.91328776e-01
-3.82588059e-01 -7.88757741e-01 4.47468199e-02 -1.85429859e+00
1.08728603e-01 -8.26794446e-01 3.45249698e-02 -2.02223349e+00
2.90206581e-01 -5.43367825e-02 3.63676026e-02 4.25538957e-01
-2.03703478e-01 -7.43647590e-02 1.88052267e-01 2.18847707e-01
-9.38805938e-01 1.38972461e-01 1.38131928e+00 -7.98432350e-01
1.41428024e-01 -3.14765900e-01 -7.11905837e-01 5.69160044e-01
7.73280501e-01 2.02170312e-01 -2.55882829e-01 -1.28891301e+00
7.70851016e-01 -3.90240133e-01 3.28517169e-01 -6.72303259e-01
2.57755488e-01 6.50672764e-02 1.03232682e+00 -1.39070821e+00
-3.16848993e-01 -3.17372978e-01 -7.81545579e-01 7.75136352e-02
-8.19292665e-01 3.24996442e-01 -3.78567725e-02 4.19961780e-01
-1.25086643e-02 -3.94347072e-01 4.13822591e-01 -2.03411028e-01
-1.13948178e+00 8.11271667e-02 -4.82763797e-01 2.08627850e-01
3.99001241e-01 -2.42116913e-01 -1.04351556e+00 -3.69426131e-01
3.64445220e-03 1.14581056e-01 7.12992370e-01 6.78141057e-01
5.95803201e-01 -7.87211239e-01 -4.59128082e-01 2.88215399e-01
3.11462075e-01 4.34745327e-02 3.36470571e-03 2.92960137e-01
-1.09595430e+00 8.31587791e-01 -5.01247533e-02 -5.23387492e-01
-1.19542217e+00 5.96621871e-01 -8.02547932e-02 -1.32411569e-01
-5.76477766e-01 8.77160609e-01 1.01062067e-01 -4.05110657e-01
4.88851100e-01 -6.69304132e-01 -7.17190802e-02 -4.60648797e-02
9.54580307e-01 5.09254873e-01 3.37526172e-01 2.88522355e-02
-1.23319089e-01 4.96492296e-01 -5.27447760e-01 3.05994123e-01
1.25610292e+00 -2.05092028e-01 -2.95778722e-01 2.40548044e-01
1.45301962e+00 3.23470742e-01 -1.25312424e+00 1.51706710e-01
2.30253175e-01 -2.98516631e-01 -2.66995039e-02 -1.06442928e+00
-7.39648759e-01 1.19314218e+00 3.73097718e-01 3.63432854e-01
1.06226349e+00 -1.25408201e-02 1.17175436e+00 7.71426320e-01
-2.00752288e-01 -1.45239317e+00 2.44315311e-01 8.01660240e-01
6.19141459e-01 -1.41043389e+00 -4.75033559e-03 -1.11285888e-01
-2.13239208e-01 1.34218621e+00 7.57490933e-01 3.47396433e-01
8.41717944e-02 6.81398571e-01 2.18158904e-02 -7.74102435e-02
-6.42724037e-01 -2.92140227e-02 5.77271581e-01 5.83206654e-01
7.50624895e-01 -4.38922495e-02 1.21292315e-01 3.93391818e-01
-4.46860969e-01 -2.92949587e-01 6.40111029e-01 9.39211130e-01
-5.22141933e-01 -1.08371210e+00 -3.35518628e-01 7.40749955e-01
-3.82356793e-01 -4.80135590e-01 -8.53029072e-01 6.97280169e-01
-2.54008323e-01 8.00187409e-01 4.61544275e-01 -1.89391583e-01
5.08655794e-02 -2.31558476e-02 5.02350867e-01 -3.78361970e-01
-1.24605894e-01 1.55578917e-02 -6.73558190e-02 -5.28280854e-01
3.18383366e-01 -5.57831526e-01 -1.09380019e+00 -5.79709411e-01
-3.46975285e-03 -5.23512065e-01 7.94528484e-01 8.41717303e-01
4.65893924e-01 7.40591705e-01 3.34936768e-01 -3.17133904e-01
-4.57238823e-01 -8.14935625e-01 -5.32083809e-01 2.22058713e-01
3.74736756e-01 2.51726568e-01 -9.38728545e-03 2.98436821e-01] | [11.57256031036377, 2.4432485103607178] |
e7e93717-d2df-4c2a-bc26-0393a79e7160 | mammut-a-simple-architecture-for-joint | 2303.16839 | null | https://arxiv.org/abs/2303.16839v2 | https://arxiv.org/pdf/2303.16839v2.pdf | MaMMUT: A Simple Architecture for Joint Learning for MultiModal Tasks | The development of language models have moved from encoder-decoder to decoder-only designs. In addition, the common knowledge has it that the two most popular multimodal tasks, the generative and contrastive tasks, tend to conflict with one another, are hard to accommodate in one architecture, and further need complex adaptations for downstream tasks. We propose a novel paradigm of training with a decoder-only model for multimodal tasks, which is surprisingly effective in jointly learning of these disparate vision-language tasks. This is done with a simple model, called MaMMUT. It consists of a single vision encoder and a text decoder, and is able to accommodate contrastive and generative learning by a novel two-pass approach on the text decoder. We demonstrate that joint learning of these diverse objectives is simple, effective, and maximizes the weight-sharing of the model across these tasks. Furthermore, the same architecture enables straightforward extensions to open-vocabulary object detection and video-language tasks. The model tackles a diverse range of tasks, while being modest in capacity. Our model achieves the state of the art on image-text and text-image retrieval, video question answering and open-vocabulary detection tasks, outperforming much larger and more extensively trained foundational models. It shows very competitive results on VQA and Video Captioning, especially considering its capacity. Ablations confirm the flexibility and advantages of our approach. | ['Anelia Angelova', 'Claire Cui', 'Zhifeng Chen', 'Andrew Dai', 'Luowei Zhou', 'Abhijit Ogale', 'Wei Li', 'Ben Caine', 'Xiyang Luo', 'Dahun Kim', 'AJ Piergiovanni', 'Weicheng Kuo'] | 2023-03-29 | null | null | null | null | ['open-vocabulary-object-detection', 'video-captioning', 'video-question-answering'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 3.30794394e-01 1.71429873e-01 6.94136471e-02 -2.39232853e-01
-1.17272091e+00 -5.60034633e-01 1.04698431e+00 -3.46708506e-01
-7.03686297e-01 4.91363525e-01 2.84988314e-01 -3.92443568e-01
2.10341886e-01 -3.06029290e-01 -9.76396441e-01 -6.26299679e-01
2.81575590e-01 7.15899885e-01 3.01383197e-01 -3.05726409e-01
-1.31317422e-01 -7.69229159e-02 -1.64225674e+00 5.19732773e-01
5.20008981e-01 1.01675200e+00 6.78052247e-01 9.86327231e-01
-5.29664420e-02 1.00195348e+00 -2.72647232e-01 -8.22805822e-01
5.97229637e-02 -3.22218567e-01 -7.22405195e-01 2.66468138e-01
8.51366460e-01 -6.12331450e-01 -6.25957608e-01 6.43318772e-01
6.93745375e-01 -2.92279780e-01 8.00467968e-01 -1.11845982e+00
-1.02816212e+00 3.90939116e-01 -4.39218491e-01 7.00527802e-02
3.15846950e-01 3.75209093e-01 1.30562150e+00 -1.14688039e+00
5.62208652e-01 1.45021081e+00 5.63927472e-01 7.16600955e-01
-1.17811835e+00 -2.21329451e-01 1.88599154e-01 5.42864762e-02
-1.32960486e+00 -7.46349871e-01 1.09941602e-01 -5.45694292e-01
1.24369264e+00 6.87789768e-02 3.92413437e-01 1.52655208e+00
2.11641714e-01 1.07682562e+00 6.01003528e-01 -3.15572292e-01
-1.12008892e-01 1.90056458e-01 -3.24748039e-01 8.68347526e-01
1.18686110e-01 -1.21627308e-01 -4.97962385e-01 4.30332646e-02
4.99515146e-01 -5.63396141e-02 -3.26895326e-01 -5.11494994e-01
-1.28155875e+00 1.01726031e+00 8.82547572e-02 1.00884184e-01
-7.22695589e-02 4.51997995e-01 4.16962713e-01 4.66735303e-01
2.77272642e-01 2.25908130e-01 -2.23515704e-01 2.42724493e-02
-1.07010913e+00 1.67139322e-01 9.50303257e-01 1.24151278e+00
5.97980917e-01 8.00222978e-02 -4.31760579e-01 7.94386744e-01
6.11703932e-01 8.79582822e-01 5.79759955e-01 -6.97634220e-01
5.00273824e-01 2.56745547e-01 -1.63762867e-01 -4.99997258e-01
-3.19849133e-01 -3.57639998e-01 -5.92198551e-01 2.44484954e-02
3.55822831e-01 -1.30058885e-01 -1.14136314e+00 1.78818929e+00
-5.61512746e-02 -1.48700535e-01 1.10154234e-01 7.87206650e-01
9.66480434e-01 7.36597538e-01 2.13212907e-01 7.76577322e-03
1.52754092e+00 -1.09258723e+00 -5.13519526e-01 -7.37546623e-01
5.27564645e-01 -8.38210285e-01 1.07262325e+00 1.68668836e-01
-1.45224178e+00 -4.99115109e-01 -9.63012040e-01 -5.63064098e-01
-3.41031432e-01 2.39114761e-01 5.13669312e-01 5.44507325e-01
-1.50684500e+00 -7.38012940e-02 -5.99786997e-01 -6.05488896e-01
3.94628584e-01 3.12731832e-01 -2.23078877e-01 -2.73566961e-01
-9.60120916e-01 1.07861495e+00 3.35284084e-01 -1.58769473e-01
-1.38243568e+00 -3.29005957e-01 -1.01127851e+00 1.49250865e-01
6.33382618e-01 -1.40806007e+00 1.45937622e+00 -1.13459265e+00
-1.26712728e+00 1.09708273e+00 -8.89331996e-02 -5.73178530e-01
5.26371181e-01 -2.22011372e-01 -1.65830493e-01 3.35273504e-01
-1.09012518e-02 1.33801234e+00 1.46254551e+00 -1.06461048e+00
-5.00269413e-01 -2.18062818e-01 1.38614669e-01 3.98618191e-01
-4.24204022e-01 4.32921164e-02 -9.09656167e-01 -4.38876420e-01
-3.92996192e-01 -7.64658451e-01 2.18274184e-02 1.71287239e-01
-9.65587571e-02 -1.85103148e-01 7.23376751e-01 -6.46336436e-01
1.14409339e+00 -2.19490337e+00 3.18122566e-01 -3.69854689e-01
3.28821689e-01 3.16166222e-01 -5.49649954e-01 5.98931074e-01
1.91497058e-01 -3.27399522e-02 -2.89202094e-01 -5.77304900e-01
2.54013777e-01 3.32069218e-01 -4.84902620e-01 2.74423331e-01
5.55789173e-01 1.48627949e+00 -6.94077611e-01 -6.14070892e-01
9.96203646e-02 5.39421976e-01 -6.33863091e-01 2.83557385e-01
-6.08780086e-01 6.30619079e-02 -2.35460445e-01 6.22201979e-01
1.77887544e-01 -7.08598435e-01 5.37370741e-02 -1.26130283e-01
1.47865983e-02 1.57964468e-01 -8.49679589e-01 1.93655562e+00
-3.82309854e-01 8.70914638e-01 2.89032966e-01 -9.69215572e-01
4.14280117e-01 4.94276255e-01 7.85646401e-03 -7.43360937e-01
2.61894345e-01 2.88862497e-01 -3.13336812e-02 -8.33114207e-01
5.75002313e-01 -1.58402231e-02 -1.46734819e-01 3.75528157e-01
7.11459517e-01 -3.27346861e-01 3.19099993e-01 4.57285434e-01
1.00738573e+00 2.00042889e-01 2.95116991e-01 5.75479306e-02
4.20315236e-01 -1.14920236e-01 -1.91176757e-01 1.11696494e+00
-1.93372020e-03 7.63277888e-01 4.01246309e-01 -3.10385972e-02
-1.33157909e+00 -1.17083919e+00 -2.63664126e-02 1.52296531e+00
-6.66532815e-02 -4.41147566e-01 -6.45292759e-01 -5.28758824e-01
-6.86956942e-02 4.04684395e-01 -4.83247608e-01 -1.57438695e-01
-3.91278416e-01 -6.16202354e-01 7.32200444e-01 4.50322926e-01
2.50400037e-01 -7.84174681e-01 -4.45129722e-01 2.92065963e-02
-2.12620452e-01 -1.45672655e+00 -5.75810075e-01 4.51601483e-02
-4.82029945e-01 -7.52245069e-01 -1.12507200e+00 -9.95956063e-01
3.06741148e-01 3.73650074e-01 1.43178630e+00 -1.58741146e-01
-3.66391063e-01 1.09335577e+00 -2.42300764e-01 -4.22136188e-01
-6.32122397e-01 1.38148710e-01 -3.40496570e-01 1.38234541e-01
1.51083156e-01 -2.31121957e-01 -5.43802977e-01 1.56178311e-01
-1.19956338e+00 4.12988700e-02 9.97651041e-01 1.03080881e+00
2.98804849e-01 -6.89114869e-01 5.36626697e-01 -3.92091513e-01
6.62097037e-01 -5.10931909e-01 -3.88649732e-01 4.29316223e-01
-3.69669169e-01 1.40350997e-01 2.64043182e-01 -4.16807294e-01
-8.36065948e-01 6.27791062e-02 -3.58283460e-01 -5.47635615e-01
-9.45804864e-02 3.98312420e-01 -1.37587979e-01 -5.77495322e-02
5.41967452e-01 5.79026341e-01 1.55614257e-01 -2.70739645e-01
7.87279546e-01 7.27884650e-01 6.89755976e-01 -1.90637410e-01
7.39943802e-01 4.88829583e-01 -2.52966881e-01 -1.07275844e+00
-7.67185450e-01 -4.62989450e-01 -4.19819564e-01 -4.70190346e-02
1.17306876e+00 -1.48361480e+00 -4.92760777e-01 4.73064661e-01
-1.34775150e+00 -2.57979333e-01 -2.58257776e-01 3.19696635e-01
-6.87986076e-01 5.57477653e-01 -4.78416353e-01 -6.62951231e-01
-2.46373892e-01 -1.23166943e+00 1.55486226e+00 -6.39351308e-02
2.70224158e-02 -1.07716846e+00 -1.15726180e-01 5.20127714e-01
5.33248723e-01 -4.30703133e-01 7.24203765e-01 -7.51524091e-01
-8.69772851e-01 -3.71026481e-03 -4.06233847e-01 4.84749407e-01
-4.00178760e-01 -3.51277977e-01 -1.14743876e+00 -4.83166099e-01
-4.86227684e-02 -9.36002791e-01 1.40602434e+00 2.50766903e-01
8.49838734e-01 -2.13559583e-01 -2.47656092e-01 5.75286567e-01
1.27773261e+00 -1.50589928e-01 6.08549476e-01 4.90179770e-02
7.15112686e-01 4.16955411e-01 1.37905851e-01 2.14853257e-01
7.57092953e-01 6.39081478e-01 6.39922678e-01 -1.39010936e-01
-3.94860804e-01 -1.01620130e-01 7.44231820e-01 7.86069453e-01
2.24981472e-01 -7.31904864e-01 -9.40378547e-01 6.41403377e-01
-1.92264271e+00 -1.01200092e+00 5.48898429e-02 1.93158686e+00
7.97006786e-01 -3.69825661e-02 9.05950442e-02 -4.48783696e-01
2.82554626e-01 3.53809536e-01 -4.94632721e-01 -2.87757099e-01
-3.18363905e-01 1.23520546e-01 3.93588305e-01 3.42222661e-01
-1.25598645e+00 9.40754473e-01 7.33494616e+00 8.64471853e-01
-9.71738935e-01 4.38909292e-01 3.66620362e-01 -3.09361428e-01
-3.60902518e-01 -2.21367955e-01 -9.75528300e-01 2.40039200e-01
1.09371340e+00 1.50153145e-01 3.15896362e-01 6.87576234e-01
-1.02039389e-01 -1.99347753e-02 -1.42671156e+00 1.19917011e+00
7.21442699e-01 -1.33388233e+00 5.14597654e-01 -1.06855646e-01
4.97661710e-01 4.65458721e-01 2.66107917e-01 5.60440004e-01
1.54750004e-01 -1.19881010e+00 8.89349103e-01 4.77031052e-01
9.64348197e-01 -2.00585842e-01 4.83599067e-01 3.64422828e-01
-9.31741238e-01 -3.16026390e-01 -2.81927675e-01 1.12516575e-01
2.68122941e-01 2.01398842e-02 -6.74887717e-01 3.57146084e-01
4.70340759e-01 5.61143637e-01 -7.30101883e-01 9.18359280e-01
-7.02375174e-03 3.32616717e-01 -8.74055773e-02 -1.48499068e-02
5.43981016e-01 1.72339097e-01 5.87018251e-01 1.49692166e+00
2.59203374e-01 -3.68371248e-01 2.01377913e-01 8.69074285e-01
-1.86649233e-01 4.04768251e-03 -8.51707995e-01 -2.54180372e-01
2.24140417e-02 1.09098494e+00 -1.47477865e-01 -3.97490621e-01
-9.44527388e-01 1.09711790e+00 3.78675312e-01 3.88788760e-01
-1.00575435e+00 -1.56990327e-02 3.72949451e-01 -2.39160676e-02
7.65503287e-01 -3.63409311e-01 1.35254562e-01 -1.40562296e+00
3.92325036e-02 -1.03042817e+00 3.71469229e-01 -1.00028419e+00
-1.29714561e+00 5.95502198e-01 -5.35154939e-02 -9.86089587e-01
-5.72131753e-01 -9.28229034e-01 -1.85778752e-01 7.16997445e-01
-1.82342362e+00 -1.51055837e+00 -2.10818559e-01 9.45530653e-01
8.42229187e-01 -2.18575925e-01 6.27511322e-01 5.12617886e-01
-3.37912053e-01 6.08792424e-01 1.74294859e-01 3.02582346e-02
8.19381654e-01 -1.10529244e+00 3.56725961e-01 7.79924273e-01
4.18421268e-01 3.24695528e-01 4.90087748e-01 -2.92750120e-01
-1.82233131e+00 -1.15499473e+00 8.46741676e-01 -7.31915176e-01
8.82998824e-01 -7.77728438e-01 -6.71226740e-01 8.91604424e-01
6.59995914e-01 -2.45690465e-01 5.07740021e-01 -1.58630461e-01
-5.74092805e-01 1.62993565e-01 -4.29211289e-01 6.66788280e-01
9.38282907e-01 -8.40437651e-01 -8.22424054e-01 4.74476337e-01
8.21419477e-01 -3.36020052e-01 -4.97539699e-01 1.40886515e-01
5.83400667e-01 -7.79881775e-01 1.18551731e+00 -6.00090384e-01
6.25063717e-01 9.72113293e-03 -2.61316508e-01 -9.19638813e-01
-2.02490687e-01 -6.81090593e-01 -3.51930529e-01 9.61878121e-01
5.71848869e-01 -3.65848184e-01 2.76846677e-01 1.95865303e-01
-2.85544723e-01 -6.30559146e-01 -9.48530197e-01 -7.27732062e-01
5.45342565e-02 -4.55778927e-01 -2.46294253e-02 5.44694781e-01
-2.64040709e-01 1.01396072e+00 -6.41284406e-01 -2.87514739e-02
3.38223517e-01 -1.63988084e-01 7.85812080e-01 -9.38416600e-01
-7.33464301e-01 -5.68751931e-01 -2.32048303e-01 -1.60592771e+00
-2.08689962e-02 -1.04899013e+00 1.06801406e-01 -1.69398355e+00
4.78388637e-01 1.95793062e-01 1.11408211e-01 4.14480925e-01
-1.57723770e-01 3.83195192e-01 3.55220318e-01 3.14723253e-01
-1.03021896e+00 6.49741292e-01 1.10301626e+00 -2.97109455e-01
1.93034917e-01 -3.42971414e-01 -8.02201152e-01 6.42105579e-01
2.96917140e-01 -1.33296356e-01 -5.71943462e-01 -9.98905122e-01
4.34793532e-01 1.87125430e-02 7.46407747e-01 -7.69437075e-01
3.30590546e-01 3.31952810e-01 1.84558749e-01 -4.54275697e-01
5.92617452e-01 -6.81633711e-01 -2.82731742e-01 1.70502782e-01
-4.38721120e-01 7.83411041e-02 2.43204549e-01 7.84848630e-01
-2.29698479e-01 -1.79883912e-01 6.22647166e-01 -3.53437215e-01
-8.98760140e-01 3.27118218e-01 -5.18409431e-01 3.63265723e-01
9.00874913e-01 -2.43269086e-01 -4.70546842e-01 -6.25937521e-01
-7.37740278e-01 4.46314454e-01 2.24789530e-01 7.06268728e-01
6.33240819e-01 -9.13713455e-01 -1.01984036e+00 8.01407769e-02
3.40188891e-01 -2.05216050e-01 1.29272461e-01 1.03797448e+00
-1.23919338e-01 7.71537364e-01 6.01174794e-02 -1.00194430e+00
-1.16533482e+00 7.34269917e-01 3.85416210e-01 -2.90252328e-01
-4.36286956e-01 8.47431183e-01 7.75400221e-01 -1.28388613e-01
3.84368509e-01 -1.06280245e-01 -1.01809710e-01 2.06254095e-01
5.29552817e-01 -2.91694850e-02 2.47376300e-02 -5.40554106e-01
-1.37482673e-01 4.51772243e-01 -1.98610991e-01 -1.48244828e-01
1.04822779e+00 -5.26102126e-01 3.99811305e-02 4.24064100e-01
1.14184439e+00 -3.37659419e-01 -1.30501676e+00 -2.75965840e-01
-2.18784943e-01 1.41681628e-02 -3.62841748e-02 -8.14152420e-01
-6.07868373e-01 1.19332087e+00 3.95707160e-01 1.51456118e-01
1.03385854e+00 3.97400767e-01 6.77292228e-01 6.01353049e-01
4.62009758e-03 -8.48680198e-01 3.49396020e-01 8.00852001e-01
8.81740570e-01 -1.48509800e+00 -2.37958357e-01 -8.93786326e-02
-7.14955270e-01 9.54472840e-01 4.50720996e-01 1.57113999e-01
1.89268544e-01 3.09124708e-01 -1.55535474e-01 -3.14674497e-01
-1.21143830e+00 -5.50450802e-01 5.22478759e-01 5.22153318e-01
4.14385259e-01 -3.26859891e-01 6.55294396e-03 2.95479745e-01
1.36037871e-01 -1.34114906e-01 3.38188261e-01 8.36330473e-01
-5.64547122e-01 -8.60172093e-01 -1.81707650e-01 3.06952357e-01
-4.64992702e-01 -5.08881927e-01 -4.10743892e-01 9.37038839e-01
-4.27910611e-02 9.36579168e-01 1.33412883e-01 -9.37282294e-02
9.59411077e-03 3.95016611e-01 6.97121263e-01 -6.18279397e-01
-5.19427061e-01 2.64353335e-01 1.76511914e-01 -5.76822460e-01
-4.19770658e-01 -5.37153602e-01 -5.69853365e-01 8.77719596e-02
-3.83462995e-01 -2.70783901e-01 6.45209372e-01 1.08997643e+00
5.17867267e-01 5.10901630e-01 2.10052758e-01 -7.15836704e-01
-9.24497306e-01 -7.51474202e-01 -2.78881282e-01 1.86325073e-01
5.91645360e-01 -4.14216578e-01 -1.54623806e-01 3.33641231e-01] | [10.924966812133789, 1.5475414991378784] |
0ab4a0a2-af2c-4835-b2ac-97ec28b8f4de | benchmarking-shadow-removal-for-facial | 2111.13790 | null | https://arxiv.org/abs/2111.13790v1 | https://arxiv.org/pdf/2111.13790v1.pdf | Benchmarking Shadow Removal for Facial Landmark Detection and Beyond | Facial landmark detection is a very fundamental and significant vision task with many important applications. In practice, facial landmark detection can be affected by a lot of natural degradations. One of the most common and important degradations is the shadow caused by light source blocking. While many advanced shadow removal methods have been proposed to recover the image quality in recent years, their effects to facial landmark detection are not well studied. For example, it remains unclear whether shadow removal could enhance the robustness of facial landmark detection to diverse shadow patterns or not. In this work, for the first attempt, we construct a novel benchmark to link two independent but related tasks (i.e., shadow removal and facial landmark detection). In particular, the proposed benchmark covers diverse face shadows with different intensities, sizes, shapes, and locations. Moreover, to mine hard shadow patterns against facial landmark detection, we propose a novel method (i.e., adversarial shadow attack), which allows us to construct a challenging subset of the benchmark for a comprehensive analysis. With the constructed benchmark, we conduct extensive analysis on three state-of-the-art shadow removal methods and three landmark detectors. The observation of this work motivates us to design a novel detection-aware shadow removal framework, which empowers shadow removal to achieve higher restoration quality and enhance the shadow robustness of deployed facial landmark detectors. | ['Song Wang', 'Yang Liu', 'Wei Feng', 'Hongkai Yu', 'Felix Juefei-Xu', 'Qing Guo', 'Lan Fu'] | 2021-11-27 | null | null | null | null | ['shadow-removal', 'facial-landmark-detection'] | ['computer-vision', 'computer-vision'] | [ 5.35033047e-01 -2.36674041e-01 2.37222716e-01 -1.78171977e-01
-1.90083653e-01 -5.20073235e-01 5.14652014e-01 -3.89668822e-01
-9.11781862e-02 7.49826372e-01 -1.82185415e-02 -1.53384164e-01
2.04747796e-01 -5.03935993e-01 -5.26013255e-01 -1.22179663e+00
2.35864848e-01 -2.39031658e-01 6.46387041e-01 -2.70862162e-01
1.82725012e-01 8.59561563e-01 -1.62495756e+00 -2.41120130e-01
1.02259541e+00 8.03959906e-01 1.65703997e-01 1.46555632e-01
1.68461874e-01 2.34570444e-01 -9.59272861e-01 -3.18786144e-01
2.81272203e-01 -3.59527797e-01 7.45782554e-02 2.91660100e-01
3.14206570e-01 -4.48625773e-01 -2.07812905e-01 9.17918324e-01
6.11898601e-01 3.56850103e-02 5.50682485e-01 -1.56263006e+00
-3.31239283e-01 -1.35307863e-01 -9.87632453e-01 2.15274498e-01
2.96917945e-01 3.61106157e-01 5.36192954e-01 -1.13561678e+00
3.21566910e-01 1.37266576e+00 5.01076102e-01 4.04902786e-01
-8.95028234e-01 -1.19215512e+00 8.82815868e-02 3.62292290e-01
-1.41866922e+00 -7.77325273e-01 1.02497566e+00 1.05784997e-01
-1.32404581e-01 3.53345007e-01 4.33704108e-01 9.84908700e-01
3.17840576e-01 8.00936282e-01 1.69156003e+00 -3.22792262e-01
1.62766591e-01 2.47556984e-01 -1.90176398e-01 9.37436819e-01
5.02208173e-01 2.06355691e-01 -5.87802052e-01 -2.11830914e-01
3.91401649e-01 6.09376878e-02 -7.54159212e-01 -1.29946604e-01
-7.00170159e-01 4.05889094e-01 3.24209839e-01 -8.35230723e-02
-1.88230798e-02 -6.62907362e-02 1.05083734e-01 -1.20007671e-01
3.36107790e-01 -3.00546229e-01 -1.57423198e-01 4.12872732e-01
-8.24644804e-01 -1.72725450e-02 6.50868595e-01 6.36256218e-01
8.26128840e-01 1.37362793e-01 -3.63253236e-01 6.15777433e-01
4.98399764e-01 1.15469539e+00 4.90654632e-02 -3.37862611e-01
1.17129587e-01 3.93148065e-01 7.27976188e-02 -1.41063881e+00
-3.91146928e-01 -4.95225102e-01 -7.85097182e-01 6.17767692e-01
3.03534687e-01 -1.09238409e-01 -9.02101338e-01 1.66844058e+00
7.07671106e-01 7.05847800e-01 8.96319821e-02 1.03143156e+00
1.01571763e+00 4.10695761e-01 -7.20600039e-02 -5.98020256e-01
1.46246910e+00 -7.46389866e-01 -8.38119626e-01 -1.78177625e-01
-5.86147048e-02 -1.16385984e+00 9.98283684e-01 3.02396536e-01
-5.96852243e-01 -4.00540203e-01 -1.22033191e+00 2.40745768e-01
-1.13337673e-01 2.61856735e-01 5.91100991e-01 1.03024876e+00
-7.27728307e-01 5.66829182e-02 -4.90496457e-01 -3.04813296e-01
5.03926396e-01 9.51945484e-02 -1.47536948e-01 -3.78333986e-01
-1.08226466e+00 6.77793622e-01 -2.28908852e-01 3.73305976e-01
-1.03912175e+00 -5.85203409e-01 -4.82410729e-01 -1.09218143e-01
7.35913754e-01 -3.11050683e-01 8.55839908e-01 -9.22227085e-01
-1.33077645e+00 6.36661708e-01 -4.68459964e-01 -1.44020081e-01
4.79334354e-01 -1.54988870e-01 -5.75844407e-01 6.53075725e-02
-6.97092190e-02 3.07051204e-02 1.39553607e+00 -1.84591830e+00
-4.33052778e-01 -4.99528915e-01 1.39173819e-02 2.26432279e-01
-5.35331488e-01 1.90011889e-01 -6.21343851e-01 -7.51346469e-01
-3.60700302e-02 -1.02910626e+00 6.84109554e-02 1.34394616e-01
-6.90266490e-01 2.32222732e-02 1.26566803e+00 -5.28990686e-01
1.06298327e+00 -2.25397587e+00 -2.59926289e-01 2.62861937e-01
8.41669217e-02 4.61429000e-01 -9.70620885e-02 2.66722053e-01
2.41016477e-01 -1.40674695e-01 -4.81228113e-01 -6.11978471e-01
-2.04583667e-02 2.52632171e-01 -5.84588230e-01 8.73793006e-01
-6.09118417e-02 5.07246375e-01 -5.73448718e-01 -6.09296918e-01
1.51507586e-01 7.48503983e-01 9.48521048e-02 2.21141264e-01
5.53020984e-02 4.51306134e-01 -5.14029622e-01 1.18222845e+00
1.23559177e+00 3.34655315e-01 -6.24078549e-02 -3.57436627e-01
1.00467855e-03 -5.31788826e-01 -1.25804806e+00 9.56886888e-01
-3.56240600e-01 6.45084679e-01 3.21900785e-01 -4.67185348e-01
1.00735247e+00 1.51472062e-01 2.77564794e-01 -5.69035888e-01
1.49630129e-01 2.32523873e-01 -1.42660871e-01 -3.30092937e-01
3.18576127e-01 -1.96199790e-01 1.39653444e-01 3.91565233e-01
-6.39907718e-01 5.19322157e-02 -2.27736071e-01 2.27065254e-02
1.14105535e+00 1.16889496e-04 9.66798514e-02 -9.17190462e-02
8.42639089e-01 -4.91706461e-01 8.67130816e-01 2.54278988e-01
-3.75696987e-01 5.21559775e-01 2.81744003e-01 -6.63512424e-02
-3.05197209e-01 -1.12977850e+00 -2.29384020e-01 9.27942514e-01
7.90804029e-01 -9.65557992e-02 -7.04370558e-01 -6.94211304e-01
1.11807607e-01 4.65846032e-01 -5.38710415e-01 -2.77209312e-01
-5.29467940e-01 -9.74199295e-01 8.05992901e-01 1.52398661e-01
8.41762483e-01 -1.06246626e+00 -7.26125121e-01 -3.06163490e-01
-3.36145912e-03 -1.22665119e+00 -4.55914766e-01 -2.34877348e-01
-3.80543828e-01 -1.44544196e+00 -6.86196744e-01 -6.51452541e-01
7.74871111e-01 1.02073824e+00 8.93352866e-01 4.61065143e-01
-5.20078897e-01 3.05463523e-01 -4.17064875e-01 -6.77504420e-01
-2.22374320e-01 -4.07979757e-01 2.19918534e-01 6.98414683e-01
-6.47181794e-02 -6.37168646e-01 -8.93330276e-01 6.72454059e-01
-1.20018303e+00 -1.82267487e-01 9.70308065e-01 6.03383243e-01
4.19202149e-01 4.07707363e-01 4.40451086e-01 -8.47128332e-01
6.11447275e-01 -3.02686274e-01 -6.92973793e-01 3.10739070e-01
-5.83934903e-01 -2.50461519e-01 7.53304958e-01 -2.14589551e-01
-1.39726985e+00 8.38010907e-02 1.57249644e-01 -3.54870141e-01
-2.51090139e-01 6.83936151e-03 -7.52297163e-01 -5.09264767e-01
3.68623286e-01 3.51126194e-01 -1.99572399e-01 -3.46436143e-01
2.98386276e-01 4.56693560e-01 4.16619241e-01 -3.49149972e-01
1.55130541e+00 9.95187581e-01 3.98474246e-01 -1.09472883e+00
-7.13755071e-01 -3.70586097e-01 -2.90156931e-01 -4.05462563e-01
4.18464035e-01 -6.19529665e-01 -6.72907054e-01 6.87246501e-01
-1.09297526e+00 -2.61874437e-01 3.65127385e-01 -8.07628781e-02
-6.45009205e-02 7.76817918e-01 1.21280923e-02 -1.02502203e+00
-3.47708941e-01 -1.09447515e+00 1.10254776e+00 6.00310385e-01
4.73676920e-01 -6.92384958e-01 -1.66569516e-01 2.74993122e-01
3.25797021e-01 5.09836137e-01 5.61903358e-01 -7.24581210e-03
-8.79588664e-01 2.08705645e-02 -4.42800701e-01 3.13004196e-01
3.91520500e-01 7.17446432e-02 -1.14609826e+00 -3.55556399e-01
4.13683280e-02 -1.82643086e-02 9.75347459e-01 1.66702211e-01
8.81953537e-01 -1.02995962e-01 -5.11485755e-01 7.33387649e-01
1.33057427e+00 9.89127532e-02 7.90614903e-01 1.70219406e-01
6.05621517e-01 5.11974990e-01 1.06687582e+00 5.19529462e-01
-7.50711933e-02 8.75746667e-01 5.90999067e-01 -5.53660035e-01
-4.05935019e-01 8.59611705e-02 4.37577456e-01 1.43843576e-01
-4.26803343e-02 -3.50476772e-01 -6.41214490e-01 1.46673620e-01
-1.51527739e+00 -6.52564287e-01 -1.29963428e-01 2.13220024e+00
5.87309778e-01 -2.34231260e-02 -3.48058313e-01 2.77313828e-01
7.46701956e-01 5.38738191e-01 -6.47160769e-01 2.22979590e-01
-4.14143473e-01 2.23947480e-01 5.13507903e-01 3.43649149e-01
-8.81431580e-01 9.99555945e-01 4.81082058e+00 9.71769571e-01
-1.07185864e+00 1.10334732e-01 4.16616619e-01 2.09596872e-01
-2.86801606e-01 6.52110502e-02 -8.53672564e-01 5.72249234e-01
1.08867243e-01 -5.69489924e-03 2.69511968e-01 5.19379675e-01
4.71805036e-01 -3.97546917e-01 -5.57686806e-01 9.82688308e-01
5.95526397e-01 -6.67469859e-01 -1.52765647e-01 7.54482970e-02
5.98415732e-01 -5.02152979e-01 3.06373030e-01 -5.92083260e-02
-4.77299560e-03 -1.00785685e+00 3.22992355e-01 4.09257144e-01
8.24207485e-01 -8.24159503e-01 7.17898846e-01 8.67476091e-02
-1.31715786e+00 -1.33497372e-01 -2.62181938e-01 3.34758371e-01
2.28044659e-01 8.56348991e-01 -7.25765586e-01 5.07391691e-01
5.74262857e-01 3.15528214e-01 -6.80513322e-01 1.07868946e+00
-7.09847033e-01 6.41061902e-01 -1.89916477e-01 1.18183054e-01
-2.46261567e-01 -1.35230660e-01 7.99660265e-01 1.08298063e+00
1.38837799e-01 2.62923688e-01 1.91034615e-01 5.18456876e-01
-1.53413177e-01 1.46148831e-01 -5.89188635e-01 5.71320057e-01
6.48962736e-01 1.52137506e+00 -9.19853866e-01 9.00124609e-02
-3.84777337e-01 1.13849306e+00 -2.36263409e-01 6.31737053e-01
-1.19922006e+00 -2.61330158e-01 8.61389935e-01 2.06646577e-01
-6.34520426e-02 -2.70078391e-01 -1.31920889e-01 -8.13475311e-01
1.30560428e-01 -8.81476283e-01 -1.73867699e-02 -5.58535635e-01
-8.81360233e-01 4.34455693e-01 -2.46959850e-01 -1.04512143e+00
4.87131119e-01 -3.23798269e-01 -9.19122338e-01 6.59154177e-01
-2.10735726e+00 -1.08860373e+00 -1.05397999e+00 8.88576567e-01
4.08676475e-01 -1.93675980e-01 3.96227360e-01 3.14465135e-01
-7.33857989e-01 7.89887846e-01 2.88382415e-02 6.14191592e-02
1.00794506e+00 -6.71586275e-01 6.55048713e-02 1.13047826e+00
1.34512251e-02 3.95146728e-01 8.92433465e-01 -5.88101983e-01
-1.66033375e+00 -1.10218847e+00 2.17900857e-01 -1.36228189e-01
3.16003114e-01 -3.17511737e-01 -8.12258661e-01 1.71809807e-01
-1.10826239e-01 1.21631093e-01 3.97219658e-01 -3.39096606e-01
-1.79192543e-01 -5.26525795e-01 -1.15730226e+00 8.17894995e-01
1.03196287e+00 -4.10728961e-01 -2.64432341e-01 3.90202641e-01
4.10513848e-01 -2.38579243e-01 -1.39707804e-01 7.54453540e-01
6.26648307e-01 -1.27758634e+00 1.02823782e+00 1.98702633e-01
-3.16941068e-02 -5.81339955e-01 -1.85445294e-01 -1.12710238e+00
2.02257350e-01 -6.95842326e-01 -9.30157006e-02 1.56663942e+00
-9.80438888e-02 -9.91722584e-01 8.22450221e-01 1.15404814e-01
-1.18634857e-01 -8.22051108e-01 -9.03627753e-01 -7.20933139e-01
-6.44262135e-01 -7.33793229e-02 5.57270944e-01 5.33069432e-01
-7.12726653e-01 4.52706553e-02 -4.68156964e-01 7.04399407e-01
9.19126093e-01 3.47716779e-01 1.14944375e+00 -1.22603774e+00
-4.78721671e-02 -1.44680560e-01 -2.66285062e-01 -8.44134808e-01
3.31459254e-01 -2.92764604e-01 3.56422067e-01 -1.33377326e+00
2.36621693e-01 -6.35227382e-01 -3.07127297e-01 3.89177412e-01
-4.08705145e-01 6.51751399e-01 2.02932581e-01 3.40377092e-01
-3.53621602e-01 7.65994966e-01 1.26417077e+00 -5.26904427e-02
-2.56215215e-01 2.71087557e-01 -6.08822286e-01 8.58548045e-01
8.16234529e-01 -3.54644954e-01 -4.83502507e-01 -1.25014961e-01
-1.71629697e-01 -1.67167649e-01 6.52979970e-01 -1.08156216e+00
7.36300647e-02 -4.33658153e-01 2.96616197e-01 -4.24651504e-01
6.36344194e-01 -9.36394930e-01 -8.94926637e-02 4.95398849e-01
4.05927658e-01 -3.02869052e-01 1.58643305e-01 9.33379233e-01
-6.96097612e-02 1.00690678e-01 9.30657685e-01 1.47769481e-01
-6.93435788e-01 3.65664810e-01 -5.08788973e-02 -1.92423537e-01
1.26186156e+00 -2.71097511e-01 -5.75705111e-01 -3.65013599e-01
-1.39024854e-01 -1.77736394e-02 6.43728912e-01 3.39580268e-01
8.70155275e-01 -9.63539541e-01 -7.96968699e-01 3.14518243e-01
-7.41654858e-02 -2.58522660e-01 2.56575972e-01 9.96764898e-01
-2.68660665e-01 1.62436649e-01 -1.83628634e-01 -3.57773542e-01
-1.85723281e+00 4.77669477e-01 9.13590640e-02 1.84232399e-01
-4.83177841e-01 6.79098248e-01 5.87223887e-01 2.04009071e-01
3.84227008e-01 8.31903890e-02 -1.40763924e-01 -5.96718192e-02
5.17926753e-01 3.94397140e-01 5.44227697e-02 -7.63861537e-01
-4.58538681e-01 7.57198393e-01 2.70613968e-01 1.73738688e-01
9.07137990e-01 -2.85763443e-01 -1.36016101e-01 4.36974466e-02
8.47363949e-01 6.72533274e-01 -1.28105772e+00 -1.99717224e-01
-3.40656906e-01 -8.37230921e-01 -4.17950749e-02 -5.67806721e-01
-1.33504236e+00 7.46696413e-01 7.72289395e-01 3.21560055e-02
1.57820249e+00 -1.55574381e-01 1.08686709e+00 1.43301278e-01
4.66708094e-01 -8.59758794e-01 3.99130285e-01 5.37983961e-02
8.84283841e-01 -1.19430673e+00 3.16080272e-01 -9.19728875e-01
-3.18288714e-01 9.24923956e-01 5.84095836e-01 2.05241397e-01
5.94482422e-01 4.36662674e-01 1.65568560e-01 -1.19055100e-01
-1.37586415e-01 -4.27999079e-01 2.03981981e-01 6.01080835e-01
-7.39241838e-02 -3.85078527e-02 -4.14191365e-01 4.10922796e-01
1.61263481e-01 -2.46490315e-01 4.72244769e-01 7.25306153e-01
-6.49432778e-01 -9.74798739e-01 -9.61299241e-01 6.45318255e-02
-3.18272948e-01 -2.84082536e-03 -5.00367403e-01 7.20498204e-01
3.30524862e-01 1.23202038e+00 -5.99440992e-01 -3.22257638e-01
2.05347344e-01 -2.62577832e-01 2.15044633e-01 -3.28631312e-01
1.16710772e-03 -4.09942418e-02 -1.98609993e-01 -4.93188947e-01
-4.91470069e-01 -6.00819409e-01 -1.22198594e+00 -3.22232276e-01
-4.70512956e-01 -1.57663032e-01 6.11760199e-01 8.90030324e-01
3.38598154e-02 5.21499634e-01 9.99150634e-01 -8.31786156e-01
-2.93416649e-01 -6.42213345e-01 -7.70172119e-01 3.49763870e-01
3.79628211e-01 -1.20274103e+00 -6.03228033e-01 -9.81736109e-02] | [10.870951652526855, -4.09283971786499] |
12f429a0-fed4-49a9-a3a1-85c01d490963 | application-of-machine-learning-in-1 | 2112.01998 | null | https://arxiv.org/abs/2112.01998v1 | https://arxiv.org/pdf/2112.01998v1.pdf | Application of Machine Learning in understanding plant virus pathogenesis: Trends and perspectives on emergence, diagnosis, host-virus interplay and management | Inclusion of high throughput technologies in the field of biology has generated massive amounts of biological data in the recent years. Now, transforming these huge volumes of data into knowledge is the primary challenge in computational biology. The traditional methods of data analysis have failed to carry out the task. Hence, researchers are turning to machine learning based approaches for the analysis of high-dimensional big data. In machine learning, once a model is trained with a training dataset, it can be applied on a testing dataset which is independent. In current times, deep learning algorithms further promote the application of machine learning in several field of biology including plant virology. Considering a significant progress in the application of machine learning in understanding plant virology, this review highlights an introductory note on machine learning and comprehensively discusses the trends and prospects of machine learning in diagnosis of viral diseases, understanding host-virus interplay and emergence of plant viruses. | ['Supriya Chakraborty', 'Hariprasad Kodamana', 'Srija Chakraborty', 'Dibyendu Ghosh'] | 2021-12-03 | null | null | null | null | ['virology'] | ['miscellaneous'] | [ 3.20351452e-01 -3.38952720e-01 -1.70790985e-01 -6.57263473e-02
1.40961662e-01 -5.09768009e-01 3.69129509e-01 6.37209415e-01
-5.91066480e-02 6.90866709e-01 -4.67049122e-01 -6.00744545e-01
1.21101309e-02 -9.94792700e-01 -5.67814112e-01 -1.07712173e+00
-1.04915528e-02 7.38131762e-01 -2.16806028e-02 -3.32439929e-01
3.91847268e-02 8.13053787e-01 -1.28960013e+00 4.56797719e-01
6.64809346e-01 8.62218976e-01 5.90750813e-01 8.88967574e-01
-4.65784878e-01 2.23516837e-01 -6.16316617e-01 3.37191299e-03
-1.62975863e-01 -2.55029798e-01 -8.22915792e-01 -2.34773601e-04
-5.20494461e-01 -1.55233666e-02 1.53415620e-01 5.00568211e-01
4.67089206e-01 -3.25218230e-01 5.38082302e-01 -1.06395411e+00
-8.22288930e-01 -6.70471713e-02 -5.89107752e-01 1.53096318e-01
-1.12488233e-01 1.02792166e-01 7.06727684e-01 -9.02435541e-01
6.79304242e-01 1.13531888e+00 4.66364145e-01 3.08827817e-01
-1.25996435e+00 -6.36678636e-02 -1.97582692e-01 3.81319761e-01
-8.76227617e-01 1.02133200e-01 3.08099955e-01 -6.96793437e-01
9.40397799e-01 1.72764793e-01 7.17751443e-01 8.21220636e-01
4.85808969e-01 5.38685441e-01 7.79215217e-01 -2.45811507e-01
3.58366966e-01 1.90673191e-02 -4.86973748e-02 2.84927577e-01
2.57619023e-01 1.23155251e-01 -1.24501407e-01 -1.12668723e-01
4.13083792e-01 3.10754240e-01 9.61283445e-02 -1.18545465e-01
-9.38378394e-01 1.03364766e+00 3.89862567e-01 6.14947736e-01
-5.72653830e-01 -6.43171847e-01 5.77852309e-01 1.35905802e-01
5.19287825e-01 4.49468672e-01 -1.03389585e+00 2.22399086e-01
-3.99326861e-01 1.54608386e-02 8.38055134e-01 2.52418309e-01
4.15982097e-01 1.46581680e-01 3.08723658e-01 8.55575621e-01
2.22061714e-03 3.50374281e-01 3.76587510e-01 -4.32901233e-01
-4.63999510e-01 9.02147710e-01 -1.28314972e-01 -9.27024364e-01
-5.59204280e-01 -2.35026926e-01 -1.32121909e+00 1.22734845e-01
4.84341800e-01 -3.87990922e-02 -8.43187928e-01 1.23764396e+00
8.20041120e-01 -7.84154236e-03 -1.65113974e-02 6.13976777e-01
8.35123181e-01 9.20845270e-01 6.30859435e-02 -3.54980767e-01
1.33830070e+00 -2.87337512e-01 -6.73489630e-01 -1.09889671e-01
6.63889229e-01 -8.13777745e-01 7.06483066e-01 4.90874887e-01
-3.83855492e-01 -4.92533416e-01 -7.31904447e-01 4.79987524e-02
-1.01048517e+00 -1.44578323e-01 8.45703304e-01 3.75066072e-01
-7.43642867e-01 6.23832226e-01 -6.31750464e-01 -7.89111555e-01
7.35035360e-01 4.80565429e-01 -5.22622168e-01 -1.93734214e-01
-8.77761483e-01 1.02036631e+00 6.90846920e-01 5.15773058e-01
-1.04944170e+00 -7.59862185e-01 -2.83955127e-01 1.39772177e-01
2.58026600e-01 -3.24334174e-01 7.30668604e-01 -4.35848385e-01
-1.33449161e+00 1.25663781e+00 4.03488949e-02 -1.17340252e-01
-1.93797201e-01 1.03397099e-02 -2.39122689e-01 -3.01508337e-01
-2.62334287e-01 3.29017609e-01 5.22014380e-01 -1.09223366e+00
-6.23582423e-01 -7.73456275e-01 -2.49799550e-01 -7.33628929e-01
-3.12664807e-01 1.46573320e-01 2.33930513e-01 -1.66862890e-01
-6.95219859e-02 -9.91979539e-01 -3.62815052e-01 9.68588293e-02
-1.39967948e-01 -2.37254098e-01 1.30235767e+00 -4.88480747e-01
4.18969840e-01 -1.99071681e+00 4.41687077e-01 -9.29821581e-02
4.25720900e-01 1.12349057e+00 -3.86070937e-01 6.94754720e-01
-9.84869078e-02 1.57188177e-01 -5.04267104e-02 5.32455683e-01
-6.82005167e-01 3.17107946e-01 -2.15977579e-01 2.82157242e-01
4.10971940e-01 1.08369553e+00 -7.88775265e-01 -9.72496197e-02
5.15343845e-01 7.67661333e-01 -2.23082721e-01 3.90910685e-01
-3.91805351e-01 7.85365403e-01 -5.24125934e-01 7.87194371e-01
8.69032800e-01 -7.08567798e-01 4.29389060e-01 -8.18695724e-02
-1.41552106e-01 -1.57132998e-01 -5.16537666e-01 1.12037432e+00
-2.74171352e-01 5.74115157e-01 2.41682351e-01 -1.62369585e+00
9.24728990e-01 4.38671738e-01 6.87965333e-01 -2.89554358e-01
2.26179823e-01 1.54329240e-01 3.77435029e-01 -5.41230559e-01
-4.03195739e-01 -4.27011997e-02 5.07265627e-01 1.58794299e-01
2.06273705e-01 -1.15751410e-02 5.42498082e-02 -1.45139664e-01
8.90129387e-01 -2.47596111e-02 5.14978230e-01 -8.57595168e-03
6.85306907e-01 1.56587392e-01 6.81168914e-01 1.14276335e-01
-3.76689762e-01 3.43249366e-02 5.86858273e-01 -1.12557387e+00
-1.27950633e+00 -5.53259969e-01 -4.21817899e-01 1.17490470e+00
-4.14566219e-01 -2.61846423e-01 -4.23616111e-01 -3.42972785e-01
1.32213861e-01 4.59841564e-02 -6.81376040e-01 -2.24529654e-01
-2.86659300e-01 -1.48385048e+00 4.09706235e-01 1.63828731e-01
1.40292540e-01 -1.19344437e+00 -3.98891658e-01 2.55468130e-01
2.00617746e-01 -1.15633357e+00 7.29584932e-01 4.42036957e-01
-8.25132191e-01 -1.22324157e+00 -4.80855644e-01 -9.82959569e-01
3.76524150e-01 4.63946372e-01 8.04111958e-01 3.35856020e-01
-1.03189099e+00 -5.99983692e-01 -4.73845065e-01 -1.04088783e+00
-6.19338393e-01 1.59797460e-01 -5.35351075e-02 -1.90252811e-01
6.59690797e-01 -4.54620361e-01 -3.87981772e-01 4.10301797e-02
-9.10160899e-01 -4.24513407e-02 5.62371075e-01 1.20072007e+00
4.53803062e-01 1.03880428e-01 8.75611901e-01 -8.43564808e-01
3.26205492e-01 -6.65460289e-01 -7.37971187e-01 4.72919911e-01
-3.34266871e-01 -2.81579912e-01 8.58954310e-01 -3.20319802e-01
-6.21211469e-01 -1.07554816e-01 -2.83358037e-01 1.33125752e-01
-4.53407168e-01 7.70780563e-01 -2.02590808e-01 -2.02753782e-01
4.89982128e-01 7.88559020e-02 3.57507855e-01 -3.99301827e-01
1.42538980e-01 8.16049695e-01 -1.34418592e-01 4.54410501e-02
4.86309320e-01 4.63121653e-01 5.28153002e-01 -1.31694520e+00
-8.54952455e-01 -4.26921606e-01 -9.78730559e-01 -1.21801838e-01
9.39874589e-01 -3.23742330e-01 -1.19180334e+00 7.48541713e-01
-1.06049001e+00 -7.15990812e-02 -4.88395803e-02 3.35955530e-01
-6.73994645e-02 -9.56957862e-02 -6.07417762e-01 -5.72511435e-01
-3.42114866e-01 -1.00298727e+00 1.01491547e+00 3.79607797e-01
-3.59158665e-02 -1.03724408e+00 5.76945186e-01 3.60507756e-01
3.92117739e-01 4.29342628e-01 1.48110139e+00 -8.19673598e-01
-3.33618015e-01 -7.02747166e-01 -2.71719545e-01 2.49273703e-01
5.48670471e-01 4.44982678e-01 -1.32926571e+00 -1.25295341e-01
1.03090554e-01 -6.18528426e-01 5.62760353e-01 7.10659564e-01
1.18258893e+00 1.72494918e-01 -6.80797994e-01 5.55206537e-01
1.42300713e+00 4.12154585e-01 4.66888845e-01 3.16852987e-01
7.90953875e-01 8.95251513e-01 5.02969205e-01 3.54123265e-01
-1.35473430e-01 2.42596492e-01 5.48173726e-01 -3.52738023e-01
5.03624022e-01 3.77488852e-01 -3.54324251e-01 6.17388368e-01
-3.12157273e-01 -3.47746611e-01 -1.05131745e+00 1.92268044e-01
-1.65473855e+00 -1.15204406e+00 -1.81190923e-01 1.99338043e+00
6.98459029e-01 -1.02603115e-01 -1.45135298e-02 2.65771538e-01
6.00462317e-01 -9.51527655e-02 -7.95168936e-01 -6.94180608e-01
-4.12972957e-01 1.19924836e-01 -1.98505875e-02 1.17047243e-01
-1.11288452e+00 1.10257602e+00 6.91172171e+00 4.84859109e-01
-1.63375735e+00 -1.75601155e-01 8.55112731e-01 2.56040156e-01
1.12038910e-01 -1.07967444e-01 -7.60491133e-01 2.64460057e-01
1.17425275e+00 5.32805026e-02 3.45428973e-01 6.67439580e-01
4.01134044e-01 -1.37140587e-01 -9.59052205e-01 4.37642157e-01
-3.22779238e-01 -1.57210338e+00 2.83807013e-02 4.11339283e-01
5.34763753e-01 1.70608893e-01 -4.47444096e-02 1.98018610e-01
2.03084394e-01 -1.30534029e+00 -6.03470862e-01 -4.77692224e-02
3.65945160e-01 -5.22837341e-01 8.96733165e-01 6.53287828e-01
-8.03305268e-01 -8.04428756e-02 -8.65227818e-01 -2.32932016e-01
2.53980421e-02 9.36324358e-01 -1.25760245e+00 3.85249108e-01
6.50854528e-01 5.36335766e-01 -3.15301120e-01 8.43186736e-01
2.30174482e-01 4.06728119e-01 -1.59340903e-01 3.40763450e-04
9.62007567e-02 -2.88634896e-01 1.28075048e-01 7.33379960e-01
-2.73300428e-02 -1.13405950e-01 5.69503367e-01 5.46427786e-01
3.21676403e-01 2.12089688e-01 -7.80874074e-01 -7.58039832e-01
1.18918195e-01 1.64704883e+00 -1.06426358e+00 -2.18015119e-01
-4.90792304e-01 5.13064444e-01 2.44522229e-01 -2.79976372e-02
-2.84762442e-01 -8.16650987e-02 6.45541966e-01 -1.33758172e-01
2.37243086e-01 -2.66759008e-01 -3.46312016e-01 -7.51320422e-01
-5.31534374e-01 -1.07869637e+00 3.14265788e-01 -5.75538158e-01
-1.15547514e+00 3.87148768e-01 -5.23243606e-01 -5.75412750e-01
-1.65840924e-01 -1.11807287e+00 -4.89551783e-01 9.07060206e-01
-1.14429748e+00 -1.34971583e+00 -2.43710324e-01 1.48085073e-01
6.32398427e-01 -4.60126787e-01 1.60462010e+00 1.31090492e-01
-7.54173517e-01 -3.12575340e-01 8.03819716e-01 -3.16775478e-02
5.82788467e-01 -7.03031123e-01 4.00496751e-01 3.40619951e-01
-1.41532496e-01 3.40267062e-01 6.58321261e-01 -7.79622495e-01
-1.58674443e+00 -7.32017398e-01 9.22301888e-01 -3.12657714e-01
6.49597883e-01 -4.83751625e-01 -1.20873308e+00 3.97666544e-01
-2.93958634e-02 1.50539866e-02 1.21793580e+00 2.61984110e-01
-1.66540250e-01 -2.18025029e-01 -1.23842835e+00 3.03807497e-01
2.24739447e-01 -3.58805686e-01 1.25491187e-01 8.09936702e-01
5.61687410e-01 3.03461961e-02 -9.08218324e-01 5.24316072e-01
7.47199476e-01 -5.46432614e-01 1.10487437e+00 -1.41225886e+00
3.21477771e-01 -1.29716113e-01 3.50358114e-02 -1.28920364e+00
-5.78829467e-01 -2.33597010e-01 -5.87788373e-02 9.46590900e-01
3.57813481e-03 -3.39416385e-01 6.28982008e-01 3.02998483e-01
2.48864219e-01 -9.70713019e-01 -4.94075119e-01 -4.18670744e-01
1.83681384e-01 2.66434699e-01 4.58385736e-01 9.59918261e-01
4.66229813e-03 3.99650872e-01 -2.24754423e-01 -3.67357880e-02
3.99138361e-01 3.70241404e-01 6.98387623e-01 -1.81076479e+00
8.87565240e-02 -1.73303410e-01 -7.68774867e-01 9.67554525e-02
1.33299977e-01 -7.89610088e-01 -2.45132193e-01 -1.53375089e+00
2.20554203e-01 -2.10536897e-01 -2.10190058e-01 3.83964747e-01
-2.71315783e-01 2.06762068e-02 -6.74216300e-02 1.14427373e-01
1.68121740e-01 1.76169753e-01 1.37409329e+00 -1.34499699e-01
-2.09822655e-01 1.41350880e-01 -5.67427814e-01 6.67818129e-01
1.15167737e+00 -2.87766367e-01 -2.03950480e-01 -1.89718083e-01
2.70887256e-01 -3.42566550e-01 2.32227042e-01 -5.60118020e-01
-3.13318938e-01 -6.01363778e-01 9.32057202e-01 -4.08885211e-01
2.59483933e-01 -7.33274639e-01 3.74195687e-02 8.84687543e-01
-1.04513071e-01 5.93093447e-02 4.63592499e-01 3.35604966e-01
-7.45102167e-02 -3.41199599e-02 1.03182173e+00 -2.79888034e-01
-7.40747511e-01 4.63288486e-01 -7.84538329e-01 -1.38393730e-01
1.39542735e+00 -4.34154347e-02 -4.88654703e-01 1.61805496e-01
-7.35111475e-01 5.40649891e-02 9.58293527e-02 5.47233164e-01
4.82993335e-01 -7.83421755e-01 -7.76230693e-01 4.07428533e-01
-8.08394030e-02 -6.62834868e-02 3.08265567e-01 8.59705806e-01
-8.31099868e-01 9.07091200e-01 -8.54049146e-01 -9.21038270e-01
-1.42106700e+00 1.16001260e+00 3.38618755e-02 -1.35435089e-01
-3.45079035e-01 6.60563767e-01 5.02818465e-01 -4.61423934e-01
-1.96632389e-02 1.35436729e-01 -6.02521539e-01 9.13589075e-02
7.12068439e-01 1.57292768e-01 2.66005278e-01 -4.80806679e-01
-4.13288265e-01 2.47724518e-01 -4.44410622e-01 4.49096560e-01
2.00490856e+00 3.07363898e-01 -5.55955410e-01 7.38145053e-01
1.04800212e+00 -6.52704597e-01 -4.74584073e-01 2.05607876e-01
1.95158422e-01 -4.61899161e-01 -3.57370861e-02 -7.62807906e-01
-8.84955943e-01 1.14596140e+00 6.89628124e-01 5.08154988e-01
1.03201759e+00 -3.90806831e-02 5.40176570e-01 6.05199397e-01
9.71815586e-02 -8.92372429e-01 -3.55903385e-03 6.29108131e-01
6.42427325e-01 -1.68419063e+00 -2.95404911e-01 -3.69561046e-01
-2.01478601e-01 9.44074035e-01 5.40273726e-01 8.12234282e-02
1.02737546e+00 6.02278948e-01 1.02947555e-01 -2.54203171e-01
-1.06354642e+00 -9.51086357e-02 -2.18613833e-01 1.10318661e+00
1.01879406e+00 -6.04481846e-02 -2.45605335e-01 -2.49226727e-02
2.33957708e-01 5.67259252e-01 3.35015863e-01 1.00598550e+00
-7.18034923e-01 -1.44866908e+00 -3.65496546e-01 4.81679112e-01
-6.20247424e-01 2.70071439e-02 -6.60949886e-01 3.57184619e-01
1.31951779e-01 7.76572883e-01 -8.71208012e-02 -1.37045756e-01
-1.83785558e-01 2.28483260e-01 3.71574372e-01 -6.37083411e-01
-4.44533020e-01 3.52967046e-02 -3.50240380e-01 -1.80029020e-01
-2.71335036e-01 -3.28012466e-01 -1.09979606e+00 -7.52058804e-01
-4.43539500e-01 4.78654057e-02 1.18292010e+00 1.11690438e+00
6.32660270e-01 6.40840709e-01 4.96600300e-01 -7.20934093e-01
-2.64337838e-01 -8.18143606e-01 -5.20915389e-01 5.86252064e-02
3.71327728e-01 -4.10957903e-01 9.41669121e-02 2.95997798e-01] | [5.398075103759766, 5.479242324829102] |
57b26eba-2450-439b-a0e4-25b314ae0d97 | recurrent-neural-network-language-model | 1611.00196 | null | http://arxiv.org/abs/1611.00196v1 | http://arxiv.org/pdf/1611.00196v1.pdf | Recurrent Neural Network Language Model Adaptation Derived Document Vector | In many natural language processing (NLP) tasks, a document is commonly
modeled as a bag of words using the term frequency-inverse document frequency
(TF-IDF) vector. One major shortcoming of the frequency-based TF-IDF feature
vector is that it ignores word orders that carry syntactic and semantic
relationships among the words in a document, and they can be important in some
NLP tasks such as genre classification. This paper proposes a novel distributed
vector representation of a document: a simple recurrent-neural-network language
model (RNN-LM) or a long short-term memory RNN language model (LSTM-LM) is
first created from all documents in a task; some of the LM parameters are then
adapted by each document, and the adapted parameters are vectorized to
represent the document. The new document vectors are labeled as DV-RNN and
DV-LSTM respectively. We believe that our new document vectors can capture some
high-level sequential information in the documents, which other current
document representations fail to capture. The new document vectors were
evaluated in the genre classification of documents in three corpora: the Brown
Corpus, the BNC Baby Corpus and an artificially created Penn Treebank dataset.
Their classification performances are compared with the performance of TF-IDF
vector and the state-of-the-art distributed memory model of paragraph vector
(PV-DM). The results show that DV-LSTM significantly outperforms TF-IDF and
PV-DM in most cases, and combinations of the proposed document vectors with
TF-IDF or PV-DM may further improve performance. | ['Brian Kan Wing Mak', 'Wei Li'] | 2016-11-01 | null | null | null | null | ['genre-classification'] | ['computer-vision'] | [ 5.28612174e-02 -3.88333142e-01 -5.38216531e-01 -5.42589664e-01
-3.67857933e-01 -5.17656922e-01 8.40256929e-01 1.70291960e-01
-7.47822046e-01 7.35585749e-01 6.90276325e-01 -4.50877160e-01
-2.82960415e-01 -6.73641562e-01 -4.44360346e-01 -6.97244048e-01
-6.62173480e-02 4.66617703e-01 1.75351590e-01 -1.64090604e-01
5.89029074e-01 2.85388649e-01 -1.47452354e+00 6.32040799e-01
3.52635086e-01 1.01164341e+00 7.13808417e-01 5.86812317e-01
-1.07246208e+00 8.42289746e-01 -9.23065305e-01 -9.03590992e-02
-3.76264274e-01 -3.28145623e-01 -9.42171156e-01 -4.34245467e-01
5.73939979e-02 -7.80815035e-02 -5.07755220e-01 7.49059141e-01
3.77877474e-01 7.67028570e-01 1.02316332e+00 -6.86169386e-01
-8.92579436e-01 1.01927924e+00 -4.49595153e-01 5.00020921e-01
9.97487381e-02 -6.19168580e-01 9.62465644e-01 -9.25811470e-01
5.71846068e-01 1.49875724e+00 3.90824914e-01 5.17928064e-01
-7.02049315e-01 -3.52952421e-01 3.72663379e-01 4.12338734e-01
-1.32130671e+00 -1.47631466e-01 7.35807002e-01 -3.44208658e-01
1.93694997e+00 -5.59673295e-04 3.92699271e-01 1.50081182e+00
1.02738631e+00 7.56503999e-01 3.57246011e-01 -8.42067599e-01
9.76077691e-02 3.82318571e-02 7.54802227e-01 3.79590929e-01
-2.08595190e-02 -7.72211999e-02 -6.19518518e-01 -3.50420922e-01
6.71597362e-01 3.48635972e-01 -2.16108039e-01 2.55238146e-01
-9.68364060e-01 1.11846650e+00 9.63883400e-02 1.20241690e+00
-3.78719956e-01 1.43144101e-01 9.70796943e-01 2.71231085e-01
6.91797018e-01 3.37180085e-02 -7.32838690e-01 -2.62051612e-01
-8.94399226e-01 1.86939240e-01 6.90265834e-01 6.50472283e-01
3.19900304e-01 3.88991773e-01 -4.30956572e-01 1.45341909e+00
4.48578864e-01 4.72497076e-01 1.59596074e+00 -3.81763816e-01
4.42306250e-01 2.95479834e-01 -2.68836200e-01 -1.28982759e+00
-3.67896259e-01 -3.97330135e-01 -8.60872149e-01 -5.62870979e-01
-2.61949599e-01 1.41099006e-01 -9.09835279e-01 1.56689513e+00
-1.62680686e-01 -1.12837017e-01 2.64385104e-01 4.57887441e-01
1.04686952e+00 1.56557369e+00 -2.92105637e-02 -5.24435997e-01
1.23712325e+00 -1.12406361e+00 -1.08332753e+00 -3.85355741e-01
9.22251999e-01 -5.86233318e-01 8.61769557e-01 3.39110821e-01
-6.96197391e-01 -8.49659920e-01 -8.33179176e-01 -3.91615415e-03
-7.71744549e-01 -5.78746684e-02 3.96578223e-01 3.94809186e-01
-8.49224269e-01 5.80212593e-01 -5.46825528e-01 -3.68624806e-01
-1.56699754e-02 2.36100703e-01 -1.89860225e-01 3.25804092e-02
-1.53969657e+00 9.70186651e-01 8.11468244e-01 3.30241807e-02
-6.95998311e-01 -2.35506684e-01 -9.41064835e-01 2.88415521e-01
-1.51869198e-02 -3.79043758e-01 1.26123047e+00 -8.24134529e-01
-1.47811580e+00 3.36847305e-01 -5.05688369e-01 -5.73085010e-01
-4.15244132e-01 -1.82720944e-01 -7.01835573e-01 -2.29002666e-02
-1.14368200e-01 3.29415113e-01 9.16897714e-01 -8.46963108e-01
-7.09442079e-01 -3.77818584e-01 -4.93051559e-01 1.43609837e-01
-9.62985158e-01 1.13022968e-01 -8.74352083e-02 -9.28167999e-01
-2.03357607e-01 -5.47052324e-01 2.82422230e-02 -7.90714204e-01
-9.72150266e-02 -8.66744280e-01 9.92833912e-01 -6.07933462e-01
1.81385541e+00 -2.20298123e+00 1.63116962e-01 1.12238221e-01
-1.82020098e-01 5.70021510e-01 -4.46443260e-01 7.24236846e-01
6.32136911e-02 2.71379292e-01 9.84017849e-02 -3.55237037e-01
-1.33043423e-01 7.56651998e-01 -6.76066339e-01 2.69041419e-01
-4.27403480e-01 8.43852818e-01 -6.25366569e-01 -5.12039900e-01
9.35502797e-02 7.34597564e-01 5.19739650e-02 4.54333834e-02
-2.93299466e-01 -4.67063002e-02 -3.98842633e-01 1.35834247e-01
2.24996880e-01 1.58161018e-02 1.55258268e-01 -8.84345472e-02
-6.69260845e-02 5.31559110e-01 -8.55216801e-01 1.78358924e+00
-6.11341715e-01 5.57958543e-01 -5.72678864e-01 -1.27103031e+00
1.13615155e+00 6.21085703e-01 1.50157064e-01 -7.37755239e-01
2.81475842e-01 1.17027164e-01 2.07187012e-02 -5.08961916e-01
7.55145669e-01 -1.64393634e-01 -2.26496026e-01 5.03630400e-01
5.80615401e-01 4.97728556e-01 3.38811189e-01 1.53513134e-01
8.24798107e-01 -1.81135029e-01 3.18054914e-01 -2.55179763e-01
6.63243055e-01 -3.68992954e-01 4.31995600e-01 8.22116911e-01
3.64097774e-01 3.60525787e-01 2.18898267e-01 -4.72402483e-01
-7.11845517e-01 -6.53823197e-01 -2.84231484e-01 1.63233137e+00
-4.67846483e-01 -6.46473408e-01 -3.01885992e-01 -6.08269811e-01
-3.47035634e-03 1.23338270e+00 -4.57412332e-01 -3.67416799e-01
-6.28215373e-01 -4.41801012e-01 7.24506259e-01 6.98357940e-01
2.02384830e-01 -1.48516369e+00 -1.26117885e-01 5.60357749e-01
-1.94397613e-01 -7.54396200e-01 -6.02391422e-01 4.13144022e-01
-9.07565475e-01 -3.73166621e-01 -8.93562198e-01 -1.12900424e+00
2.94495255e-01 1.33863598e-01 9.02188599e-01 -2.61503190e-01
1.80767730e-01 2.17153609e-01 -7.68624067e-01 -3.80562156e-01
-4.53059733e-01 1.83370650e-01 1.40384570e-01 6.49979934e-02
6.00109696e-01 -3.06715816e-01 -2.06649363e-01 -1.83618218e-02
-8.70848298e-01 -4.83424395e-01 3.37906837e-01 1.14858365e+00
4.62049365e-01 2.41648003e-01 6.72218978e-01 -9.43160474e-01
1.06853914e+00 -4.40568388e-01 -9.72035527e-02 2.68170208e-01
-5.81520915e-01 2.18011990e-01 8.21459234e-01 -9.28751290e-01
-1.21566176e+00 -4.23475891e-01 -2.50620544e-01 -5.36772788e-01
6.47820681e-02 1.17405427e+00 5.68502061e-02 3.11207294e-01
5.25512934e-01 7.09809303e-01 -2.94471920e-01 -7.02462554e-01
2.61724800e-01 1.10040092e+00 2.34714404e-01 -4.78410542e-01
-2.13157423e-02 -2.40252420e-01 -3.98556650e-01 -1.03041160e+00
-7.78818190e-01 -6.14712119e-01 -5.34201920e-01 9.77106765e-02
6.17006481e-01 -5.09999931e-01 -3.63225132e-01 4.38381821e-01
-1.65660870e+00 9.67390537e-02 -1.63228929e-01 7.60553658e-01
-3.16097170e-01 2.79570043e-01 -9.29514050e-01 -8.30204904e-01
-7.30146229e-01 -7.73997903e-01 8.27605188e-01 7.59204179e-02
-3.77570093e-01 -1.56900263e+00 2.81588048e-01 -2.03248993e-01
6.71962678e-01 -3.06147784e-01 1.48178113e+00 -1.14480436e+00
5.96644580e-01 -4.47456479e-01 2.65414000e-01 7.69863486e-01
1.58872724e-01 -1.31674424e-01 -8.15117776e-01 -3.67654830e-01
3.33037883e-01 -1.14665762e-01 1.20775950e+00 6.85330153e-01
1.09126294e+00 -5.65884352e-01 -5.35059750e-01 3.02727818e-01
1.39747000e+00 7.56100953e-01 4.20284837e-01 2.35534310e-01
7.42249370e-01 3.56190622e-01 3.05415541e-01 3.42256248e-01
6.10472411e-02 3.16766500e-01 -9.74892303e-02 3.57807010e-01
2.12391585e-01 -2.51365125e-01 5.38594127e-01 1.53923965e+00
-1.45523891e-01 -9.26020861e-01 -9.44494605e-01 5.48992932e-01
-1.96898890e+00 -9.54151690e-01 1.10050671e-01 2.04599762e+00
6.67695582e-01 2.23154351e-01 -1.25597417e-01 2.77351201e-01
7.93555319e-01 5.62782526e-01 -2.14011744e-01 -1.19889081e+00
-1.81539237e-01 2.04813048e-01 2.26139009e-01 4.25149739e-01
-8.18127155e-01 1.02219367e+00 5.95566654e+00 1.18341482e+00
-1.23743057e+00 3.02628368e-01 3.22013497e-01 -1.35584116e-01
-1.51904136e-01 -2.97714472e-01 -1.19634295e+00 5.22257626e-01
1.68892443e+00 -4.88669902e-01 1.38884023e-01 8.78652036e-01
-5.43631576e-02 2.79802024e-01 -1.05993938e+00 1.04413199e+00
5.21182954e-01 -1.29936481e+00 7.50926495e-01 1.13039784e-01
2.62296617e-01 5.51826879e-02 4.34329696e-02 7.98617244e-01
-1.33566797e-01 -1.07991421e+00 3.78971875e-01 6.16531014e-01
7.32610106e-01 -1.03825200e+00 1.04130733e+00 6.23580515e-01
-1.21449673e+00 -2.15736061e-01 -1.01949143e+00 -3.57476175e-02
1.53457671e-01 5.74317813e-01 -6.03102982e-01 5.11304975e-01
5.56518555e-01 9.09523368e-01 -2.87756115e-01 5.08780420e-01
1.57325059e-01 8.56298804e-01 -7.19742998e-02 -5.34657240e-01
5.65916359e-01 1.68273091e-01 4.87588972e-01 1.63607872e+00
4.98404741e-01 -2.38953665e-01 8.08794126e-02 4.82888848e-01
-3.25070210e-02 5.27389288e-01 -7.91021705e-01 -4.97198105e-01
4.67777371e-01 1.11777198e+00 -5.76683283e-01 -5.24851739e-01
-5.09801209e-01 9.05923247e-01 2.56899416e-01 4.40685570e-01
-5.28320909e-01 -6.64069235e-01 2.82612354e-01 -1.56614214e-01
6.41039968e-01 -1.87804058e-01 3.23709637e-01 -9.91509736e-01
-1.17417485e-01 -6.36137903e-01 3.37495506e-01 -5.97472787e-01
-1.58386564e+00 1.09372556e+00 1.95366040e-01 -1.00155008e+00
-8.18795383e-01 -6.53507650e-01 -5.21084249e-01 9.74248886e-01
-1.11511040e+00 -1.08368444e+00 5.14352679e-01 6.43765688e-01
9.18190718e-01 -7.85562575e-01 1.26729286e+00 1.40558988e-01
-3.60758126e-01 5.64670384e-01 6.71099782e-01 1.87212422e-01
5.60906589e-01 -8.59026432e-01 2.68536180e-01 2.61529356e-01
4.20449317e-01 9.41591799e-01 3.30171615e-01 -6.45673990e-01
-1.13964021e+00 -1.15277994e+00 1.46466875e+00 1.28090875e-02
5.17944872e-01 -4.76810575e-01 -1.18298113e+00 9.07945752e-01
4.26561177e-01 1.80702936e-02 9.30158615e-01 1.05818130e-01
-2.34687641e-01 -7.21955672e-02 -8.26223433e-01 1.33235037e-01
7.22445309e-01 -4.98987436e-01 -1.19620609e+00 4.35698152e-01
1.09441495e+00 8.11049640e-02 -8.17341149e-01 1.98810428e-01
5.08112788e-01 -4.33597982e-01 8.74713421e-01 -7.90838718e-01
2.62792885e-01 2.26653725e-01 -5.75963140e-01 -1.64880300e+00
-7.10928977e-01 -2.24951897e-02 -3.43227118e-01 1.66153812e+00
3.21777523e-01 -7.48548985e-01 2.33877897e-01 -2.35565007e-02
-2.05017611e-01 -7.21297681e-01 -1.08563066e+00 -1.00564373e+00
3.42915267e-01 -5.30297101e-01 3.82924139e-01 8.69839251e-01
2.08759248e-01 8.96253586e-01 -3.95457298e-01 -4.65734482e-01
1.16650179e-01 -7.47669861e-02 3.31659168e-02 -1.18160057e+00
-3.06590587e-01 -5.31426132e-01 -5.00796735e-01 -1.07602549e+00
6.23683453e-01 -1.18571174e+00 -2.44332954e-01 -1.69183123e+00
2.02616844e-02 2.21800610e-01 -7.26086915e-01 5.58257461e-01
2.16568276e-01 -4.90095317e-01 2.79077403e-02 2.83909172e-01
-3.77216339e-01 7.71653593e-01 9.16889846e-01 -4.57124203e-01
-3.80849481e-01 -1.94637164e-01 -2.77628094e-01 7.78984547e-01
5.72476864e-01 -8.30729604e-01 -4.72983122e-01 -5.59380233e-01
-2.64566690e-02 2.37497091e-01 -2.25362599e-01 -6.20866895e-01
3.75786990e-01 -3.69124971e-02 6.35249376e-01 -9.03442919e-01
3.49731207e-01 -5.55791318e-01 -3.49627644e-01 4.13925678e-01
-5.69912434e-01 2.78031439e-01 2.13293076e-01 4.73794073e-01
-5.68245649e-01 -6.56296194e-01 3.36328834e-01 -3.32940787e-01
-7.17950702e-01 -3.42930369e-02 -1.02400529e+00 -2.32537135e-01
5.35778999e-01 -1.68491378e-01 -3.00260663e-01 -2.65884489e-01
-4.23325002e-01 -1.05182894e-01 -3.39338601e-01 8.14516783e-01
1.01574111e+00 -1.51886725e+00 -6.29374683e-01 4.16077614e-01
-6.73981085e-02 -2.70657301e-01 3.02561581e-01 2.19421357e-01
-1.28303275e-01 1.01336694e+00 -7.89096504e-02 -4.50614870e-01
-1.16607738e+00 7.44895041e-01 1.04243331e-01 -5.11317790e-01
-6.94434345e-01 9.06691134e-01 2.89815813e-01 -4.02441740e-01
4.97877717e-01 -4.40428704e-01 -8.91256571e-01 2.41442308e-01
7.66363144e-01 1.51186258e-01 1.85230419e-01 -8.76845121e-01
-4.89587814e-01 6.32173359e-01 -6.03251874e-01 -2.32671127e-01
1.40705597e+00 -9.55066904e-02 -3.03869903e-01 1.22113991e+00
1.60795355e+00 -3.19638699e-01 -9.80813578e-02 -4.93931383e-01
1.12676106e-01 -1.34417742e-01 3.77078801e-01 -5.57934523e-01
-9.11760986e-01 9.23234701e-01 4.70030427e-01 1.06477439e-01
7.56537735e-01 -1.15736015e-01 1.16067970e+00 7.10457861e-01
3.08606535e-01 -1.15487933e+00 -2.59833108e-03 1.19391441e+00
1.02409971e+00 -5.24141490e-01 -2.42335111e-01 3.19453180e-01
-5.70789516e-01 1.43386960e+00 3.12223077e-01 -1.95761561e-01
8.82017851e-01 2.34691918e-01 -2.98974086e-02 8.82576108e-02
-1.09852147e+00 3.61757189e-01 5.05640328e-01 3.15210044e-01
8.18147063e-01 -2.87649244e-01 -6.95659339e-01 1.11100411e+00
-2.89948136e-01 -5.36653250e-02 2.74302989e-01 1.06131828e+00
-6.09690011e-01 -1.15061867e+00 -3.30290169e-01 7.35544980e-01
-6.27165735e-01 -1.59931406e-01 -1.12048343e-01 4.61076289e-01
4.92372476e-02 9.31697130e-01 3.55858952e-01 -6.27452135e-01
4.07895669e-02 5.79377711e-01 2.88350701e-01 -8.31745028e-01
-7.62513518e-01 3.66516113e-01 1.93180721e-02 -2.29287654e-01
-3.87292355e-01 -3.13249648e-01 -1.46045077e+00 -1.12002730e-01
-5.53985119e-01 5.33626258e-01 8.20437372e-01 1.15850985e+00
2.23724276e-01 7.65510917e-01 4.17985946e-01 -8.90212119e-01
-4.80002642e-01 -1.48012865e+00 -9.17797685e-01 1.63505584e-01
2.08526328e-01 -6.45112395e-01 -5.33760786e-01 9.12808627e-03] | [10.854142189025879, 8.334634780883789] |
a3a6ae62-b330-45a0-8410-8a2b195157d1 | crnns-for-urban-sound-tagging-with | 2008.10413 | null | https://arxiv.org/abs/2008.10413v2 | https://arxiv.org/pdf/2008.10413v2.pdf | CRNNs for Urban Sound Tagging with spatiotemporal context | This paper describes CRNNs we used to participate in Task 5 of the DCASE 2020 challenge. This task focuses on hierarchical multilabel urban sound tagging with spatiotemporal context. The code is available on our GitHub repository at https://github.com/multitel-ai/urban-sound-tagging. | ['Nicolas Riche', 'Augustin Arnault'] | 2020-08-24 | null | null | null | null | ['audio-tagging', 'environmental-sound-classification'] | ['audio', 'audio'] | [-5.83848834e-01 4.23307978e-02 -2.02433676e-01 -3.83704543e-01
-1.42372096e+00 -7.26863265e-01 6.87623799e-01 1.34528831e-01
-6.81586862e-01 6.36201382e-01 7.79748499e-01 -3.18503976e-01
4.47632819e-01 -5.78431726e-01 -4.04953003e-01 -1.77793652e-01
-3.99328947e-01 3.07522327e-01 3.79508436e-01 -1.37586683e-01
-2.67414242e-01 -1.31173879e-02 -1.25476491e+00 5.14939725e-01
2.66828895e-01 6.69700563e-01 3.96316320e-01 1.17980528e+00
2.31356025e-01 1.43215990e+00 -3.13025177e-01 -6.12337217e-02
1.91374704e-01 -3.00186664e-01 -1.23363638e+00 -1.13550496e+00
5.94986558e-01 1.69631708e-02 -7.95342624e-01 1.03739357e+00
7.03736484e-01 5.32302082e-01 1.67008668e-01 -1.19057894e+00
-6.16524160e-01 1.35398066e+00 3.03773403e-01 7.23886967e-01
4.16648209e-01 -2.67969668e-01 1.46851516e+00 -1.04628015e+00
6.97103322e-01 1.09863484e+00 7.64606297e-01 6.80222154e-01
-5.65961659e-01 -1.12602508e+00 -4.68019629e-03 1.91085964e-01
-1.92881751e+00 -8.66377831e-01 6.15798593e-01 -3.64287555e-01
9.84856665e-01 4.31331486e-01 5.03134489e-01 1.22073853e+00
-3.94939154e-01 1.03751874e+00 7.01448619e-01 -2.72133857e-01
-5.34620881e-02 -3.97778988e-01 2.10139975e-01 8.28527808e-01
-2.34864548e-01 2.82123119e-01 -5.39222300e-01 7.59228393e-02
4.41571891e-01 -3.85561109e-01 2.02945825e-02 4.20898944e-01
-1.38332868e+00 8.75003874e-01 8.03162396e-01 1.00373948e+00
-1.12526923e-01 8.73252511e-01 5.74649930e-01 -8.02673101e-02
1.14128828e+00 3.03464085e-01 -2.75527388e-01 -5.53734004e-01
-1.02765739e+00 2.59625465e-01 5.98276794e-01 9.20599759e-01
4.14515018e-01 1.24393322e-01 2.94082128e-02 1.06218791e+00
3.44180465e-01 6.86682642e-01 1.56947345e-01 -1.41737235e+00
5.43822765e-01 -2.86423951e-01 1.58268064e-01 -9.85336721e-01
-6.37484074e-01 -2.56775707e-01 -5.99599540e-01 -4.65948343e-01
1.13151215e-01 -4.13665533e-01 -8.03798735e-01 1.72039223e+00
1.48677856e-01 6.57327652e-01 -9.46791023e-02 8.71322274e-01
1.63066387e+00 9.24451530e-01 6.59303427e-01 3.37854683e-01
1.18264878e+00 -1.12867558e+00 -7.41855264e-01 -8.37433711e-02
9.68520701e-01 -8.08978379e-01 8.20312381e-01 -1.64389491e-01
-8.23816478e-01 -5.48517287e-01 -5.03914654e-01 -2.57301331e-01
-9.10668135e-01 1.20175749e-01 5.56648850e-01 1.66891232e-01
-1.55712962e+00 -2.00657919e-03 -8.29124331e-01 -7.76780605e-01
3.17049980e-01 -2.12699041e-01 -1.16093896e-01 -7.96036422e-03
-1.95990944e+00 5.63238382e-01 5.22628665e-01 1.34426981e-01
-1.10350013e+00 -6.19336903e-01 -8.05645466e-01 -5.22185624e-01
1.02284074e-01 2.66459081e-02 2.03043079e+00 -6.01261735e-01
-1.05754972e+00 8.96276176e-01 -2.15621382e-01 -4.24938560e-01
2.52251625e-01 -3.84517491e-01 -9.05417323e-01 1.15631334e-01
7.13075519e-01 1.13664424e+00 -1.53838828e-01 -1.20030940e+00
-9.57153082e-01 1.76496178e-01 2.33057216e-01 1.64920136e-01
2.48752266e-01 5.73758781e-01 -4.22534764e-01 -8.89795959e-01
-2.76931196e-01 -1.12301564e+00 -2.95593232e-01 -7.62852669e-01
-5.15252173e-01 -4.33590621e-01 5.75589836e-01 -9.27869916e-01
1.38107538e+00 -2.19574118e+00 -4.37792957e-01 -1.84053317e-01
9.45290327e-02 7.84674436e-02 -3.60548794e-01 5.94983697e-01
-1.33825839e-02 5.76082289e-01 -1.54702634e-01 -4.84292120e-01
1.72215462e-01 -2.32171729e-01 -1.20910339e-01 2.78598547e-01
-4.64369386e-01 1.06877315e+00 -1.50336599e+00 -6.38898671e-01
1.96158484e-01 5.11064410e-01 -1.64512113e-01 6.09933883e-02
-2.49852419e-01 4.86910492e-01 -5.11597991e-01 6.36242926e-01
2.70473152e-01 8.70584846e-02 3.69573571e-02 3.40243690e-02
-7.24931955e-01 6.03573203e-01 -7.05551088e-01 1.95124352e+00
-7.08880007e-01 1.01327896e+00 1.38662249e-01 -6.99288905e-01
2.48410404e-01 6.42464876e-01 6.91681921e-01 -8.97337317e-01
1.57370642e-02 1.88146815e-01 -5.66881180e-01 -3.17203015e-01
8.57050776e-01 1.58491999e-01 -8.37908268e-01 3.72161537e-01
6.93910271e-02 -2.50776619e-01 3.11633080e-01 4.39766049e-01
1.16040576e+00 6.99748918e-02 3.37988228e-01 -4.73496586e-01
2.66844273e-01 2.73664325e-01 5.91761947e-01 8.33115816e-01
-5.67026258e-01 5.78634381e-01 -1.34121969e-01 -6.90678239e-01
-5.80320537e-01 -9.40448284e-01 -9.32935029e-02 1.47631085e+00
-2.75749773e-01 -6.15151584e-01 -4.62943852e-01 -6.16338670e-01
-4.99822646e-01 9.50842381e-01 -5.75598717e-01 5.46009898e-01
-6.73610091e-01 -1.12579785e-01 1.57159626e+00 4.72634792e-01
5.10006547e-01 -1.38838148e+00 -4.42773134e-01 1.40309110e-01
-8.69431674e-01 -1.28387427e+00 -6.24007821e-01 2.55515695e-01
-6.86559528e-02 -7.96522260e-01 -5.16471267e-01 -1.00687277e+00
1.98479835e-02 2.09679246e-01 1.45599520e+00 6.86183199e-02
-1.45876303e-01 6.14656508e-01 -1.09745538e+00 -4.25033182e-01
-2.73175508e-01 5.02961636e-01 -1.41190603e-01 -6.30471706e-01
2.96517968e-01 -4.02609825e-01 -4.18440610e-01 3.12878489e-02
-5.58799505e-01 4.04017754e-02 -8.45797881e-02 2.01709330e-01
5.95441937e-01 -1.96611181e-01 4.25316304e-01 -6.46945477e-01
4.15219545e-01 -9.56817269e-01 -7.17229724e-01 -2.72643641e-02
2.41608858e-01 -6.88862681e-01 3.44059139e-01 2.74410844e-02
-8.42341244e-01 2.90405214e-01 -6.55922294e-01 -6.17893450e-02
-7.67107546e-01 6.39895558e-01 1.33345753e-01 2.61415571e-01
5.33468068e-01 -3.71031798e-02 -1.08652329e+00 -3.63440752e-01
9.28743362e-01 9.67679560e-01 6.28446460e-01 -5.19612849e-01
5.86638510e-01 2.78172046e-01 -5.09158909e-01 -9.70696270e-01
-1.03230798e+00 -7.68205881e-01 -8.37324798e-01 -6.52710319e-01
1.35881746e+00 -1.48072588e+00 -1.82127222e-01 3.76233518e-01
-1.17847323e+00 -1.29262018e+00 -3.25761825e-01 4.50562954e-01
-3.20036501e-01 -1.42024130e-01 -6.68016732e-01 -7.32967317e-01
-3.61497819e-01 -4.35957909e-01 9.88952398e-01 4.37255241e-02
-3.95376921e-01 -1.20512378e+00 6.93499923e-01 3.94903988e-01
4.82820600e-01 2.05076769e-01 1.51531354e-01 -6.82403445e-01
-3.43035251e-01 -1.63315654e-01 -8.59977026e-03 1.85192227e-02
-9.98047590e-02 3.11354753e-02 -1.14423978e+00 1.39427662e-01
-9.51843798e-01 -5.84996402e-01 1.03164768e+00 4.81331378e-01
8.84920299e-01 -4.52828228e-01 -3.36559623e-01 2.67424077e-01
1.14949965e+00 3.26749146e-01 1.68047518e-01 1.40968099e-01
9.81095493e-01 3.53733182e-01 7.29558885e-01 2.88874477e-01
1.02215064e+00 8.97869647e-01 2.44489253e-01 -5.12960330e-02
-4.51299608e-01 -4.86260891e-01 2.33567670e-01 1.30038607e+00
-2.52185520e-02 -6.42589986e-01 -1.74993634e+00 1.17317545e+00
-1.99524105e+00 -1.30742335e+00 -2.16474921e-01 1.59406602e+00
7.43575037e-01 -4.41675276e-01 3.74197483e-01 -4.59500134e-01
6.07277274e-01 8.40783775e-01 1.64493144e-01 -1.75265640e-01
1.94408447e-02 7.20040277e-02 4.61945862e-01 1.16211319e+00
-1.64976120e+00 1.88417876e+00 5.71259689e+00 1.02592885e+00
-8.73547792e-01 1.07443619e+00 4.16885257e-01 -2.51136512e-01
-1.85656235e-01 -6.93738237e-02 -6.45561755e-01 3.14429551e-01
1.60107684e+00 7.30763301e-02 5.49852788e-01 5.14040291e-01
5.65120876e-01 -9.63034332e-02 -3.53019536e-01 4.92048830e-01
-1.20708615e-01 -1.28583813e+00 -4.26218778e-01 -3.50240529e-01
7.51937747e-01 1.22145808e+00 -1.34095252e-02 4.33374912e-01
1.02162373e+00 -7.20457375e-01 1.12259114e+00 5.57077825e-01
1.06095338e+00 -6.29818618e-01 7.01675773e-01 -1.18223000e-02
-1.84156561e+00 1.08005732e-01 9.79779661e-02 4.00787592e-02
4.87698436e-01 5.96909463e-01 -6.38135612e-01 5.89611650e-01
1.03644478e+00 1.11642969e+00 -5.73048711e-01 9.47209656e-01
-5.49241960e-01 1.07262099e+00 -4.03122783e-01 -1.11863270e-01
4.53918785e-01 2.80756295e-01 7.26085067e-01 1.90306687e+00
4.48497117e-01 4.13483173e-01 3.31691176e-01 3.76479745e-01
-3.45884144e-01 3.92946541e-01 -9.91360188e-01 -2.44823515e-01
1.05767429e+00 1.20474470e+00 -8.39412391e-01 -3.32071722e-01
-2.86258429e-01 5.20693183e-01 2.64995724e-01 3.65857601e-01
-9.72138345e-01 -3.41283768e-01 4.79688019e-01 -8.28802958e-02
3.08874343e-02 -4.83440846e-01 8.26231614e-02 -9.72418725e-01
-5.72921455e-01 -5.09150386e-01 5.85823774e-01 -1.08846974e+00
-6.61453009e-01 9.66955185e-01 2.39402637e-01 -1.26485133e+00
-2.92315900e-01 1.39681280e-01 -5.89021921e-01 3.40313971e-01
-1.10299444e+00 -1.60861003e+00 -2.12738425e-01 6.00303352e-01
7.08581448e-01 -9.57622230e-02 9.32642519e-01 8.14345717e-01
-2.37160534e-01 3.36799711e-01 -2.39105716e-01 5.52655578e-01
6.43482268e-01 -1.15995550e+00 8.40792120e-01 1.05516231e+00
6.39966190e-01 -1.88923569e-03 5.40429175e-01 -7.34405398e-01
-4.02821124e-01 -1.60442030e+00 1.78404808e+00 -7.04899192e-01
1.07463861e+00 -6.11141622e-01 -1.89894229e-01 1.04301226e+00
6.52998030e-01 2.45568037e-01 5.72962880e-01 1.74949214e-01
-2.63411671e-01 1.09194472e-01 -6.65720046e-01 4.29528564e-01
1.21358919e+00 -9.31704640e-01 -8.25310200e-02 6.17157876e-01
1.02254295e+00 -2.37857148e-01 -1.00641084e+00 2.97564089e-01
5.16921997e-01 -3.44775677e-01 6.92566872e-01 -1.54200837e-01
2.38227099e-01 -4.97187138e-01 -7.16804624e-01 -1.33818555e+00
-5.65718055e-01 -2.85105944e-01 2.66694337e-01 1.40591586e+00
7.10122526e-01 -4.16049570e-01 3.30416411e-01 -1.77673981e-01
-3.01676124e-01 -5.39240167e-02 -1.48469901e+00 -7.22290576e-01
2.79558361e-01 -1.05873752e+00 3.80936265e-01 1.23405075e+00
4.00485955e-02 1.88299671e-01 -5.06419122e-01 4.33667600e-01
4.42553937e-01 -4.35499102e-01 4.31343734e-01 -1.09012258e+00
2.39082336e-01 -4.55242634e-01 -1.28473230e-02 -6.71240628e-01
5.50982833e-01 -1.20475721e+00 5.46338201e-01 -1.69274402e+00
-8.36597830e-02 -4.70122397e-01 -4.49099898e-01 1.06783521e+00
2.62029290e-01 7.40684450e-01 3.65644991e-01 2.59088665e-01
-1.51484108e+00 3.53225052e-01 4.37882185e-01 -3.02600294e-01
-9.98559296e-02 -2.22270843e-02 -3.58236879e-01 3.56687784e-01
1.61543310e+00 -8.90371263e-01 9.64954197e-02 -7.25216985e-01
2.52755553e-01 1.20858513e-01 3.59240741e-01 -1.29060745e+00
3.41997325e-01 8.18937942e-02 -3.29343230e-01 -6.99574351e-01
2.80831546e-01 -5.89297950e-01 4.61501598e-01 3.53576660e-01
-6.95475161e-01 2.49601752e-01 4.48978245e-01 4.40617912e-02
-4.94065225e-01 2.02879123e-02 5.20058274e-01 -3.90442491e-01
-1.09559941e+00 2.80066550e-01 -9.07968044e-01 3.97493005e-01
4.33116734e-01 6.64578617e-01 -5.43148875e-01 -7.05268383e-01
-1.00173736e+00 2.32040927e-01 1.10865451e-01 7.82737851e-01
2.46053606e-01 -1.67613161e+00 -9.44987774e-01 -4.36219811e-01
2.66670734e-01 -3.06768388e-01 3.07426333e-01 7.42104232e-01
-6.20335698e-01 8.55679929e-01 1.02036176e-02 -8.42137560e-02
-1.15030861e+00 -3.80729847e-02 5.49160242e-01 -2.74451166e-01
-3.77741337e-01 1.11809695e+00 -6.41062185e-02 -1.03551662e+00
1.63038418e-01 -2.67899007e-01 -2.97843903e-01 1.52715012e-01
3.18948030e-01 3.63745153e-01 -2.80469030e-01 -1.26457167e+00
-6.93359137e-01 4.02501762e-01 7.39368618e-01 -5.11906922e-01
1.31214511e+00 -4.19651508e-01 -2.62294374e-02 9.23950851e-01
1.26510715e+00 1.11407287e-01 -5.36666751e-01 -1.52714446e-01
1.31271914e-01 1.73065498e-01 4.16042894e-01 -1.08600497e+00
-1.16904664e+00 5.69449544e-01 6.68356597e-01 4.99997169e-01
7.23120213e-01 3.79238576e-01 8.44186783e-01 3.91884625e-01
5.03296912e-01 -1.19909954e+00 -5.35776019e-01 1.01407170e+00
1.08328509e+00 -1.24284935e+00 -2.90300369e-01 2.39093066e-03
-6.67319894e-01 5.50637066e-01 3.25870842e-01 4.20888402e-02
1.20816553e+00 3.60344201e-01 5.37568748e-01 -4.29524571e-01
-7.67925978e-01 -8.15509439e-01 1.73744664e-01 5.76810420e-01
8.22789252e-01 7.74709821e-01 1.07771605e-02 3.70896339e-01
-3.88849109e-01 -3.51762585e-02 2.73215681e-01 8.05279493e-01
-4.16322470e-01 -8.08349371e-01 1.46204472e-01 -8.78469869e-02
-6.30408108e-01 -6.77685082e-01 -5.95468521e-01 8.92830789e-01
3.50426316e-01 1.19099557e+00 -2.08334997e-03 -7.03567684e-01
3.96495640e-01 7.97397792e-02 -1.75890803e-01 -7.76464701e-01
-4.88465190e-01 -8.41076951e-03 7.67298579e-01 -8.37734044e-01
-7.97960639e-01 -9.87045527e-01 -1.27887106e+00 -1.56045243e-01
4.21732545e-01 6.34650946e-01 3.80847007e-01 3.92366022e-01
2.74225265e-01 3.12904328e-01 2.78605938e-01 -1.13469851e+00
8.02768230e-01 -1.22039104e+00 -3.95073384e-01 -1.52084798e-01
3.49912107e-01 -2.55676210e-01 -4.10500437e-01 2.11128607e-01] | [15.209370613098145, 5.099123001098633] |
76c1df46-1d92-4e4d-ace5-75fba43db19c | language-resources-to-support-language | null | null | https://aclanthology.org/2022.lrec-1.58 | https://aclanthology.org/2022.lrec-1.58.pdf | Language Resources to Support Language Diversity – the ELRA Achievements | This article highlights ELRA’s latest achievements in the field of Language Resources (LRs) identification, sharing and production. It also reports on ELRA’s involvement in several national and international projects, as well as in the organization of events for the support of LRs and related Language Technologies, including for under-resourced languages. Over the past few years, ELRA, together with its operational agency ELDA, has continued to increase its catalogue offer of LRs, establishing worldwide partnerships for the production of various types of LRs (SMS, tweets, crawled data, MT aligned data, speech LRs, sentiment-based data, etc.). Through their consistent involvement in EU-funded projects, ELRA and ELDA have contributed to improve the access to multilingual information in the context of the pandemic, develop tools for the de-identification of texts in the legal and medical domains, support the EU eTranslation Machine Translation system, and set up a European platform providing access to both resources and services. In December 2019, ELRA co-organized the LT4All conference, whose main topics were Language Technologies for enabling linguistic diversity and multilingualism worldwide. Moreover, although LREC was cancelled in 2020, ELRA published the LREC 2020 proceedings for the Main conference and Workshops papers, and carried on its dissemination activities while targeting the new LREC edition for 2022. | ['Hélène Mazo', 'Khalid Choukri', 'Victoria Arranz', 'Valérie Mapelli'] | null | null | null | null | lrec-2022-6 | ['de-identification'] | ['natural-language-processing'] | [-2.71152020e-01 1.90770835e-01 -4.50935155e-01 -4.21939716e-02
-1.16688716e+00 -7.00940490e-01 9.82451200e-01 5.86076617e-01
-8.35441053e-01 6.93305612e-01 7.90787578e-01 -6.87332690e-01
1.25699684e-01 -6.15327835e-01 -3.43356840e-02 -3.77825461e-02
3.26132238e-01 8.66781890e-01 -9.74375457e-02 -5.75631142e-01
1.51770189e-01 6.18236661e-01 -9.43610370e-01 5.20357788e-01
9.99432325e-01 4.00038451e-01 4.42846805e-01 3.87472719e-01
-3.58790159e-01 7.79182613e-01 -2.82638401e-01 -4.06484187e-01
1.07675733e-03 -2.64477491e-01 -1.26067603e+00 -5.22515476e-01
-2.06139624e-01 -6.26616273e-03 3.03180143e-02 8.01410675e-01
5.99708200e-01 -1.44874007e-01 2.54853070e-01 -6.45511329e-01
-2.79353708e-01 5.50401390e-01 -2.74782628e-01 2.10283697e-01
6.81517184e-01 -1.44668773e-01 1.05768168e+00 -9.83984053e-01
1.14903963e+00 1.16743648e+00 7.38960862e-01 1.92768767e-01
-5.38448870e-01 -5.47501147e-01 -3.27646524e-01 -4.97253649e-02
-1.24627197e+00 -8.61803293e-01 -3.84155326e-02 -4.95584339e-01
1.36331534e+00 5.18978953e-01 1.79466486e-01 1.01747084e+00
-1.04607068e-01 6.61594927e-01 1.36140108e+00 -8.80526125e-01
-5.67736179e-02 4.38134611e-01 -2.58614600e-01 4.03221428e-01
1.50218770e-01 -4.46636587e-01 -4.07923073e-01 -4.77107644e-01
1.64144203e-01 -4.36925769e-01 1.71185713e-02 5.29386163e-01
-1.42990506e+00 9.68074262e-01 -4.27758545e-01 1.33125699e+00
-4.81567234e-01 -6.80025339e-01 8.00989091e-01 5.31926155e-01
7.84161925e-01 3.35145593e-01 -5.78883111e-01 -5.47239065e-01
-7.29574323e-01 6.05408847e-02 9.29010570e-01 4.68212575e-01
4.56683069e-01 -3.30864966e-01 1.28247231e-01 1.32038438e+00
3.32758009e-01 8.43595743e-01 6.04503751e-01 -7.66037226e-01
6.69241846e-01 6.82961106e-01 2.85377055e-01 -1.03119826e+00
-7.82412291e-01 -1.80121467e-01 -7.15942383e-01 -7.28149414e-01
1.54036820e-01 -4.85175282e-01 -2.36479089e-01 1.33710039e+00
1.26647756e-01 -6.57880664e-01 4.04607147e-01 6.00935102e-01
7.66807973e-01 6.05596185e-01 -5.37333488e-02 -4.26812291e-01
1.18932247e+00 -5.36223173e-01 -8.38229418e-01 1.18280254e-01
1.03047216e+00 -1.22917223e+00 5.83749473e-01 3.39168042e-01
-1.06996608e+00 -3.09217900e-01 -4.27109510e-01 1.58671975e-01
-7.43612647e-01 7.07936585e-02 2.82660067e-01 6.28607690e-01
-1.37760365e+00 1.18031017e-02 -5.83566546e-01 -1.13981879e+00
3.50690223e-02 2.00224146e-01 -6.12336040e-01 -1.74699366e-01
-1.77686596e+00 1.24831545e+00 2.36029670e-01 1.80394202e-02
-1.57402247e-01 -2.88813114e-01 -7.92599857e-01 -6.50516629e-01
8.30467865e-02 -3.03621083e-01 6.84781730e-01 -4.91667897e-01
-1.20257211e+00 1.44652772e+00 4.65777218e-02 -4.63772118e-01
5.03554642e-01 5.12402132e-02 -1.30920625e+00 -1.94970131e-01
6.37455285e-01 3.87331814e-01 -1.18238263e-01 -2.50726700e-01
-1.05947530e+00 -3.02711338e-01 -1.33163810e-01 1.80617884e-01
-3.42290550e-01 1.00416160e+00 -2.08270550e-01 -5.04155695e-01
-3.09407055e-01 -7.09677219e-01 -1.89905718e-01 -1.03882182e+00
-2.45348677e-01 -1.28424317e-01 6.02534473e-01 -1.27153933e+00
1.19255722e+00 -1.98264194e+00 6.75908998e-02 2.27156609e-01
-1.01490326e-01 4.39383984e-01 5.02624623e-02 1.35956132e+00
3.21224093e-01 3.43289673e-01 -2.68552992e-02 -7.99791962e-02
1.71586081e-01 2.32548326e-01 -4.41318750e-01 4.55555141e-01
-2.54468650e-01 6.51119828e-01 -1.27636468e+00 -2.33781353e-01
2.78427988e-01 4.76334572e-01 7.42600337e-02 -3.83546799e-01
1.49306819e-01 7.57163405e-01 -4.15732741e-01 4.27258849e-01
3.55569392e-01 3.52141224e-02 7.68988431e-01 1.03596933e-01
-1.01669264e+00 4.61461365e-01 -8.53099346e-01 1.27581739e+00
-9.04788136e-01 7.07930446e-01 2.46656194e-01 -6.14890099e-01
9.23689604e-01 6.54535651e-01 7.83585846e-01 -1.02678180e+00
-4.88715991e-02 1.06804383e+00 -2.72042453e-01 -4.69335496e-01
9.46066797e-01 1.23870820e-01 -2.36082271e-01 5.99353075e-01
-2.14920059e-01 1.67245165e-01 1.26547828e-01 2.67467946e-01
7.31741488e-01 2.28106622e-02 4.55958664e-01 -3.73024195e-01
1.13867784e+00 9.55889076e-02 1.29735634e-01 2.02443719e-01
-1.16166204e-01 -1.55542046e-01 -9.05624181e-02 -3.27125937e-01
-1.01478219e+00 -3.34608167e-01 -5.97516775e-01 1.05544937e+00
-5.82076609e-01 -4.43559766e-01 -6.84047878e-01 -4.70133364e-01
-3.68821025e-01 7.32942104e-01 -1.36478424e-01 5.52337587e-01
-7.52620161e-01 -7.49108255e-01 8.90855014e-01 -4.30958033e-01
5.93209028e-01 -1.23406732e+00 -4.71561998e-01 3.78240466e-01
-8.27897906e-01 -1.31365955e+00 -2.65895784e-01 -1.26200259e-01
-1.77418515e-01 -9.17135596e-01 -9.84013796e-01 -6.54265642e-01
9.68577117e-02 -1.06234744e-01 6.21411264e-01 -2.78259754e-01
-2.27307513e-01 5.23877859e-01 -5.09550631e-01 -4.06173319e-01
-1.01049340e+00 4.35477287e-01 2.57548004e-01 -1.85270265e-01
3.37519407e-01 -2.51499354e-03 -2.01462716e-01 1.29888415e-01
-5.48401475e-01 -1.32280067e-01 1.32486239e-01 5.58024228e-01
3.25613141e-01 -3.25712055e-01 1.17762649e+00 -9.57645535e-01
8.05799782e-01 -5.56755364e-01 -5.34449756e-01 4.46480989e-01
-6.34632349e-01 -4.32878554e-01 3.81431580e-01 3.82923156e-01
-9.48831499e-01 -6.30248308e-01 -5.24482965e-01 5.65710664e-01
-5.21890558e-02 7.65584648e-01 1.11330830e-01 -1.58918053e-01
3.48574817e-01 1.41302899e-01 -1.79894432e-01 -6.31545961e-01
4.47394013e-01 1.48662877e+00 3.71260315e-01 -2.08589882e-01
3.32133681e-01 3.36038977e-01 -4.30511534e-01 -1.14131665e+00
-5.00969768e-01 -1.02087653e+00 -4.54941422e-01 -3.60176146e-01
8.92133176e-01 -9.79949951e-01 -3.75817478e-01 5.62007546e-01
-1.05859888e+00 -1.67025506e-01 -6.69716671e-02 5.20691216e-01
-2.06924438e-01 3.32960010e-01 -7.07432687e-01 -8.21777582e-01
-5.06566942e-01 -8.71580482e-01 7.57963181e-01 -3.66409212e-01
-6.09284401e-01 -1.38586557e+00 4.84349281e-01 6.62898242e-01
6.89809382e-01 3.40742737e-01 7.29678273e-01 -8.83541644e-01
2.97010630e-01 -1.67432725e-01 -1.52723506e-01 9.51194465e-02
2.56817162e-01 -1.02346525e-01 -5.32559633e-01 -3.26287985e-01
-4.03663069e-02 -2.52419651e-01 4.21884179e-01 -7.72776827e-02
1.71553597e-01 -3.97292286e-01 -3.93605411e-01 -1.45000651e-01
1.14659417e+00 2.99410164e-01 3.79128397e-01 6.99802339e-01
2.72018284e-01 9.82375562e-01 7.98951983e-01 4.84087437e-01
7.58280993e-01 1.00319755e+00 -3.17526013e-01 -3.08900535e-01
1.47756450e-02 -2.89714988e-02 7.34081328e-01 1.62389457e+00
-1.15684219e-01 -2.68632006e-02 -1.21758926e+00 4.48856801e-01
-1.68077481e+00 -6.67933643e-01 -4.60697800e-01 2.27879953e+00
8.25954199e-01 -2.96758682e-01 3.80482852e-01 -2.53812551e-01
6.53633118e-01 5.40877357e-02 1.52175456e-01 -7.99583316e-01
-4.48173136e-01 2.27587238e-01 8.03529024e-01 8.73907447e-01
-6.81349456e-01 1.20054626e+00 6.43279123e+00 9.67333376e-01
-1.20781088e+00 6.06930614e-01 6.21238947e-01 4.24480081e-01
-5.03827929e-01 -2.39559621e-01 -9.73293066e-01 2.20335633e-01
1.32272911e+00 -5.35200119e-01 4.96242404e-01 1.21477365e-01
5.55175364e-01 -6.07667957e-03 -2.45560095e-01 4.36548620e-01
4.14713509e-02 -1.68045604e+00 -2.61158705e-01 5.29788673e-01
8.99590552e-01 9.72100735e-01 2.22064573e-02 1.74874052e-01
3.48304123e-01 -8.77194285e-01 6.64672077e-01 4.83179152e-01
1.29007983e+00 -8.77309799e-01 9.57376659e-01 4.86169338e-01
-8.89094293e-01 -2.08983235e-02 -1.18304595e-01 1.83421746e-01
4.18371141e-01 6.22953057e-01 -7.87697136e-01 1.05935621e+00
4.39452231e-01 5.73568165e-01 -2.65127152e-01 3.25555325e-01
2.34982684e-01 4.30198610e-01 -5.31515572e-03 4.68060970e-02
6.11650050e-01 -2.84911156e-01 7.45682359e-01 1.50742078e+00
7.31161386e-02 -2.20475078e-01 2.63717175e-01 1.05366297e-01
3.56720462e-02 9.04920220e-01 -4.80590701e-01 -3.84714842e-01
6.80942476e-01 1.19402802e+00 -7.05343723e-01 -2.30706707e-01
-4.50369984e-01 8.25674951e-01 4.02165949e-02 1.16148941e-01
-2.46610388e-01 -4.38293755e-01 4.11640946e-03 2.01474011e-01
-6.90331608e-02 -1.93851918e-01 2.49782935e-01 -6.57280564e-01
-4.19007301e-01 -1.41810787e+00 4.86011803e-01 -3.66383106e-01
-7.42445529e-01 9.79430914e-01 -1.29062533e-02 -7.03958988e-01
-6.14375532e-01 -3.90929848e-01 1.01696871e-01 1.20861244e+00
-1.25833178e+00 -1.35551715e+00 7.95762300e-01 3.74886245e-01
2.76041508e-01 -7.02043235e-01 1.20261645e+00 7.42981195e-01
-3.24185431e-01 2.99657315e-01 4.33597058e-01 5.13049774e-02
8.29763114e-01 -5.90197444e-01 4.64030445e-01 2.33787969e-01
1.41036779e-01 6.70920551e-01 3.85355264e-01 -7.16863930e-01
-1.06666172e+00 -1.11437535e+00 2.07573032e+00 -3.03691417e-01
1.26382327e+00 -3.28501493e-01 -1.15939617e-01 4.87822086e-01
4.22502190e-01 -8.94509614e-01 7.14915276e-01 2.50546392e-02
6.67319587e-03 8.46450478e-02 -1.35731578e+00 4.01103854e-01
5.49795389e-01 -9.88558769e-01 -2.98005909e-01 1.02641582e+00
7.46658385e-01 -2.09619626e-01 -1.09251726e+00 1.94461316e-01
4.48600084e-01 -7.09749758e-01 6.14789009e-01 -2.94232488e-01
-1.23657785e-01 -6.67006895e-02 -3.52851331e-01 -8.14940631e-01
2.33760718e-02 -9.73032296e-01 5.99471450e-01 1.35246718e+00
4.38414276e-01 -1.20047653e+00 -5.96671738e-02 -8.93137231e-02
-1.61338687e-01 -5.06425738e-01 -1.15805614e+00 -7.61951566e-01
-4.16283645e-02 -5.96899390e-01 6.02945626e-01 1.32611215e+00
2.01728612e-01 2.02212587e-01 -3.35976899e-01 -2.29235783e-01
-8.88527483e-02 -3.18677723e-01 5.30792117e-01 -7.59139776e-01
-9.52142030e-02 -4.58405703e-01 -1.16175033e-01 -3.27477992e-01
1.61356941e-01 -1.20841277e+00 -3.80887896e-01 -1.53883696e+00
4.44926620e-02 -4.56000060e-01 -4.80941758e-02 4.52019364e-01
3.50920618e-01 3.78952503e-01 5.22683002e-02 4.74423379e-01
-4.52800393e-01 -2.36873865e-01 9.54033077e-01 1.12881921e-01
-1.55388579e-01 -1.48243248e-01 -6.18618011e-01 4.55518931e-01
6.97977722e-01 -2.10172474e-01 2.22295344e-01 -1.27886921e-01
6.99715614e-01 -3.50179672e-02 -3.92137803e-02 -6.11202776e-01
9.05211142e-04 -2.05876946e-01 -1.82500273e-01 -7.53798187e-01
-1.36107579e-01 -6.04949772e-01 5.97022414e-01 7.79739261e-01
-1.64653778e-01 3.49028170e-01 2.21074134e-01 -2.43082866e-01
-3.58824044e-01 -3.67755651e-01 3.65717649e-01 -1.97452158e-01
-4.62619156e-01 1.92923278e-01 -7.97586799e-01 3.57727826e-01
5.52079678e-01 -2.44482588e-02 -3.13014209e-01 -4.60065454e-01
-5.18975675e-01 2.35392064e-01 4.74808067e-01 4.47699010e-01
2.00294685e-02 -9.73374605e-01 -1.16718721e+00 1.23636210e-02
-3.29708941e-02 -7.59028912e-01 1.64753094e-01 1.16577494e+00
-5.78879952e-01 9.97296393e-01 4.80942242e-02 -1.36584103e-01
-1.06004083e+00 1.77711278e-01 1.58297345e-01 -9.19222951e-01
-4.85798657e-01 2.91943729e-01 -3.16412985e-01 -8.86416674e-01
-1.99446857e-01 4.00693625e-01 -6.11837506e-01 3.47724378e-01
5.97016275e-01 6.46669924e-01 3.74026597e-02 -1.35086036e+00
-5.63922703e-01 2.89807349e-01 7.09800571e-02 -6.51816428e-01
1.38614595e+00 -4.25163686e-01 -6.38570070e-01 7.12507248e-01
1.08016598e+00 7.50118792e-01 2.29520313e-02 4.44908217e-02
5.53132176e-01 1.76665038e-01 -2.37416271e-02 -1.16482699e+00
-6.28204823e-01 4.43631738e-01 3.08025658e-01 1.19114041e-01
8.44409704e-01 -5.87331988e-02 8.80419016e-01 3.10582314e-02
7.10159421e-01 -1.44335592e+00 -5.91741920e-01 1.01515746e+00
9.66283739e-01 -8.03511918e-01 -1.48876056e-01 -2.08335847e-01
-5.24707615e-01 1.13462877e+00 -4.78821874e-01 7.63040423e-01
4.64689732e-01 2.07294151e-01 2.99549878e-01 -2.35886574e-01
-4.03500795e-01 8.09355918e-03 1.51681751e-01 8.19877446e-01
8.47727478e-01 3.24177891e-01 -9.91073370e-01 2.93747693e-01
-2.25414887e-01 1.34414896e-01 1.19836859e-01 3.74789804e-01
-2.85760194e-01 -1.89719844e+00 -3.34810138e-01 3.30778956e-01
-9.07213569e-01 -5.58965623e-01 -5.50350726e-01 9.21248019e-01
3.97843301e-01 9.76353705e-01 -1.56204358e-01 -2.24581733e-01
3.15702856e-01 1.68003961e-02 2.46734828e-01 -4.88586217e-01
-1.20183384e+00 5.15758842e-02 9.72885430e-01 -2.86564589e-01
-4.68374193e-01 -1.12487209e+00 -1.13038814e+00 -6.14796638e-01
-1.00114129e-01 6.94493234e-01 8.77220511e-01 9.59162354e-01
2.12028801e-01 2.46194452e-02 8.68904352e-01 -2.16006443e-01
-1.08259432e-01 -8.85579705e-01 -2.98267990e-01 -2.00884119e-01
-1.27674937e-01 1.15814745e-01 -7.15747699e-02 -4.43133205e-01] | [10.550886154174805, 10.219650268554688] |
9749629a-9f47-42ce-979f-1e936a501981 | how-to-train-pointgoal-navigation-agents-on-a | 2012.06117 | null | https://arxiv.org/abs/2012.06117v1 | https://arxiv.org/pdf/2012.06117v1.pdf | How to Train PointGoal Navigation Agents on a (Sample and Compute) Budget | PointGoal navigation has seen significant recent interest and progress, spurred on by the Habitat platform and associated challenge. In this paper, we study PointGoal navigation under both a sample budget (75 million frames) and a compute budget (1 GPU for 1 day). We conduct an extensive set of experiments, cumulatively totaling over 50,000 GPU-hours, that let us identify and discuss a number of ostensibly minor but significant design choices -- the advantage estimation procedure (a key component in training), visual encoder architecture, and a seemingly minor hyper-parameter change. Overall, these design choices to lead considerable and consistent improvements over the baselines present in Savva et al. Under a sample budget, performance for RGB-D agents improves 8 SPL on Gibson (14% relative improvement) and 20 SPL on Matterport3D (38% relative improvement). Under a compute budget, performance for RGB-D agents improves by 19 SPL on Gibson (32% relative improvement) and 35 SPL on Matterport3D (220% relative improvement). We hope our findings and recommendations will make serve to make the community's experiments more efficient. | ['Dhruv Batra', 'Irfan Essa', 'Erik Wijmans'] | 2020-12-11 | null | null | null | null | ['pointgoal-navigation'] | ['robots'] | [-1.83745205e-01 4.05171402e-02 2.59848405e-03 -2.78405905e-01
-8.47347677e-01 -7.09679723e-01 7.12171078e-01 2.36360326e-01
-9.46677864e-01 5.81014156e-01 3.75626534e-01 -4.14917350e-01
2.64471948e-01 -8.31064045e-01 -8.89405072e-01 -3.57617229e-01
-7.21286297e-01 2.37502933e-01 4.53107655e-01 -6.20342433e-01
3.40345591e-01 1.41341493e-01 -1.69752479e+00 -1.42308950e-01
5.42752087e-01 8.03683102e-01 2.29780734e-01 1.09962678e+00
4.27066356e-01 8.15071344e-01 -4.23667401e-01 -2.05991209e-01
5.98455846e-01 -2.30184868e-01 -6.51383698e-01 -3.05015862e-01
5.11425436e-01 -7.61443913e-01 -3.88055086e-01 8.99988711e-01
8.33224595e-01 4.94714022e-01 2.99621254e-01 -1.17021096e+00
-2.47850195e-01 3.12115729e-01 -5.87659478e-01 4.08301115e-01
4.78993833e-01 8.27503204e-01 9.46704865e-01 -4.05366510e-01
5.89196026e-01 1.22161245e+00 1.04242086e+00 2.93502212e-01
-9.67166185e-01 -3.05085659e-01 4.15617734e-01 -3.07013672e-02
-1.09089351e+00 -5.01806974e-01 -1.03179663e-01 -2.20963404e-01
1.60221338e+00 7.58111998e-02 1.04856634e+00 8.46395373e-01
2.98932403e-01 7.14863956e-01 9.08391893e-01 -1.64925635e-01
5.15222371e-01 -3.92960429e-01 4.64315787e-02 9.47696447e-01
3.04026514e-01 6.63172662e-01 -5.74502826e-01 -1.81798581e-02
1.11329782e+00 -5.55896759e-01 -2.99645096e-01 -3.48412395e-01
-1.37094963e+00 9.17319059e-01 7.21757829e-01 -3.27062339e-01
-3.68934393e-01 7.94503033e-01 4.62123394e-01 2.46822134e-01
2.12895781e-01 4.96874750e-01 -2.77971506e-01 -9.92650867e-01
-4.74587679e-01 8.20006311e-01 9.47803497e-01 1.47415125e+00
6.13428175e-01 1.11395910e-01 1.75724387e-01 3.38987678e-01
5.08798778e-01 6.65607035e-01 2.27376014e-01 -1.35323489e+00
4.37983662e-01 5.86510114e-02 4.34294105e-01 -8.92446458e-01
-8.03145051e-01 -4.37376052e-01 -2.74778783e-01 6.61486685e-01
5.27398467e-01 -4.88941014e-01 -9.51283455e-01 1.79497457e+00
3.53305906e-01 2.71094050e-02 6.39497787e-02 1.24270439e+00
5.63398242e-01 6.47720337e-01 1.63607791e-01 4.09132987e-01
1.40937579e+00 -1.27262211e+00 -1.08176097e-01 -6.09279990e-01
8.57595086e-01 -6.27098382e-01 1.42472863e+00 2.00989112e-01
-1.19281054e+00 -4.92846191e-01 -1.42784774e+00 -3.31872702e-01
-1.09445065e-01 -4.09602910e-01 1.07154226e+00 7.00709820e-01
-1.42489612e+00 5.31487882e-01 -1.21852636e+00 -8.54187667e-01
5.20737469e-02 4.18355703e-01 -1.06932312e-01 -7.29377121e-02
-6.78550899e-01 9.49396670e-01 6.07797392e-02 -1.77580938e-01
-1.13317204e+00 -5.40715218e-01 -7.98991740e-01 -7.89604485e-02
3.20156157e-01 -1.05051756e+00 1.68380928e+00 -2.60760248e-01
-1.56709254e+00 6.96712971e-01 1.79803789e-01 -8.34469855e-01
8.39170516e-01 -5.50953269e-01 9.65964049e-02 1.12161465e-01
2.24229619e-01 1.18193007e+00 6.73525557e-02 -1.01371658e+00
-1.07708728e+00 -3.21190625e-01 6.58403575e-01 9.23310876e-01
1.92723110e-01 -1.93031639e-01 -8.40537608e-01 -3.44200373e-01
2.56240610e-02 -1.18015361e+00 -4.31414604e-01 1.46466538e-01
9.93391722e-02 5.81942797e-02 2.19233871e-01 -5.35708189e-01
5.05520105e-01 -1.97095060e+00 9.40407589e-02 -2.17736423e-01
2.44865835e-01 -2.82839656e-01 -1.15042135e-01 2.80572563e-01
4.34683502e-01 -2.79007051e-02 -1.07924454e-01 -4.27407742e-01
2.27560103e-01 1.70076355e-01 -6.32134303e-02 5.45590281e-01
-2.32968763e-01 9.27579522e-01 -1.07418704e+00 -1.16169058e-01
3.60814512e-01 4.82785076e-01 -1.00016189e+00 -1.07240016e-02
-1.54128745e-01 2.52351612e-01 -4.09182161e-01 4.62323636e-01
4.27660733e-01 -1.72088519e-01 -1.16743967e-01 2.48227287e-02
-5.58599591e-01 5.29223621e-01 -8.99125814e-01 2.28722620e+00
-3.34705383e-01 6.35380268e-01 9.08562243e-02 -1.36992320e-01
5.00766337e-01 -1.03101641e-01 2.08643109e-01 -1.08967555e+00
1.43080756e-01 1.61078915e-01 9.72152054e-02 -3.56969297e-01
1.10037506e+00 8.93419906e-02 -8.42412412e-02 3.72499496e-01
-2.32242659e-01 -4.22571123e-01 2.92270929e-01 2.65298754e-01
1.39468336e+00 6.66290581e-01 5.04580885e-02 -5.81273377e-01
-3.29812020e-01 6.10572040e-01 1.48739040e-01 9.98724043e-01
-6.59740567e-01 5.52857697e-01 4.61192399e-01 -7.04693258e-01
-1.39914286e+00 -9.37674403e-01 1.82079434e-01 1.32986677e+00
5.87285578e-01 -5.66914141e-01 -6.87900782e-01 -4.01437342e-01
1.52043596e-01 6.04659200e-01 -6.59390867e-01 -8.88874605e-02
-7.52755404e-01 -1.00760293e+00 7.92687535e-01 7.68730104e-01
8.36244345e-01 -7.58711934e-01 -1.41156054e+00 1.65644929e-01
2.28451807e-02 -8.07444751e-01 -2.70223945e-01 3.82543057e-01
-9.81029987e-01 -7.44910717e-01 -7.25537717e-01 -5.80300033e-01
2.14501739e-01 6.92478955e-01 1.32805359e+00 2.41890885e-02
-4.65280823e-02 4.04667765e-01 -4.36332405e-01 -2.14877576e-01
1.56522259e-01 1.44631237e-01 6.95276111e-02 -1.14864480e+00
1.31445244e-01 -3.55994552e-01 -8.56834888e-01 2.99276024e-01
-3.74745607e-01 3.28869313e-01 4.15711761e-01 6.06655061e-01
3.94901663e-01 -5.47877550e-01 8.41437429e-02 -4.44082320e-01
5.04046261e-01 -4.46141303e-01 -8.92826915e-01 -3.53108495e-01
-5.85649014e-01 7.29108751e-02 1.30382180e-01 -2.43699760e-03
-8.27757239e-01 -3.43129873e-01 -3.97865534e-01 -1.22406958e-02
2.71639116e-02 3.92772436e-01 3.14501882e-01 -2.46995479e-01
9.69659209e-01 -4.19935882e-02 3.22624780e-02 -2.81666368e-01
5.52401125e-01 2.68590897e-01 4.72957492e-01 -7.16972232e-01
3.37487459e-01 4.63055313e-01 -2.79686391e-01 -6.10285699e-01
-7.29101896e-02 -2.09705725e-01 -1.24542363e-01 -1.11963779e-01
9.18622434e-01 -1.33060694e+00 -1.17805874e+00 4.00107294e-01
-6.97030306e-01 -1.19321024e+00 -2.33057559e-01 5.91273189e-01
-8.46308649e-01 1.57006189e-01 -7.09213972e-01 -6.00872040e-01
-2.24218830e-01 -1.37169445e+00 1.21084213e+00 3.73598725e-01
-3.02901804e-01 -7.50961900e-01 2.21900031e-01 1.15190923e-01
6.86293781e-01 2.29627565e-01 5.39659023e-01 3.69926007e-03
-9.27784562e-01 4.37201411e-02 -4.99979556e-01 -3.76675606e-01
-2.74541259e-01 -4.29021955e-01 -6.77200258e-01 -6.86773539e-01
-2.61749774e-01 -3.67033005e-01 7.06655443e-01 4.28867042e-01
4.82063651e-01 5.26102223e-02 -3.47803712e-01 9.68584538e-01
1.48760021e+00 4.19476330e-01 4.46299016e-01 9.35610235e-01
5.63340366e-01 1.81565523e-01 7.12450445e-01 3.72148216e-01
8.29909801e-01 7.90314555e-01 8.23141277e-01 8.06657523e-02
-2.39358515e-01 -1.98852196e-01 1.60995841e-01 4.17082489e-01
-5.07818937e-01 -5.16852796e-01 -1.08967161e+00 4.91233587e-01
-1.87323058e+00 -6.22013688e-01 1.89419612e-02 2.25844526e+00
3.17677796e-01 2.73620039e-01 4.39521939e-01 -5.44924080e-01
3.04535747e-01 3.70498329e-01 -8.61187041e-01 -2.93108284e-01
8.26702043e-02 -1.71709299e-01 9.04407918e-01 7.75766432e-01
-9.23979700e-01 1.09520578e+00 7.30938387e+00 2.17338189e-01
-9.47616637e-01 -8.23437497e-02 5.06551027e-01 -5.40517449e-01
-9.06136706e-02 -3.42062078e-02 -7.37664998e-01 3.43275666e-01
1.03065217e+00 -2.16844290e-01 6.32807612e-01 1.15999699e+00
6.23284020e-02 -5.40472865e-01 -1.08702862e+00 1.01210415e+00
1.24378670e-02 -1.10761094e+00 -3.95118862e-01 4.16399360e-01
5.56349754e-01 8.56425166e-01 2.92360727e-02 5.66286385e-01
9.32972610e-01 -1.01740289e+00 1.04999411e+00 -2.26983847e-03
5.57980418e-01 -8.73955905e-01 7.78988302e-01 1.18900508e-01
-1.12455261e+00 1.66536242e-01 -5.67274630e-01 -5.68026900e-01
3.82650703e-01 -1.89755782e-01 -6.15667641e-01 3.76160771e-01
1.19028628e+00 7.24527121e-01 -3.21543783e-01 1.19487786e+00
-1.87493116e-01 3.74259681e-01 -7.18877554e-01 -3.89878809e-01
7.62038469e-01 -3.90156768e-02 5.51511824e-01 1.11376047e+00
2.67744124e-01 2.47292042e-01 1.07737601e-01 3.12181443e-01
7.27330372e-02 -1.47729129e-01 -4.46069390e-01 2.41911411e-01
4.68003362e-01 8.19668651e-01 -9.06636596e-01 -3.46870720e-01
-3.21341693e-01 1.14211106e+00 4.66149509e-01 2.64531016e-01
-1.20751584e+00 -3.66429687e-01 1.14501309e+00 -1.36845782e-01
3.27628136e-01 -7.31766522e-01 -4.76596564e-01 -9.91964400e-01
-1.27923459e-01 -8.24841440e-01 1.70152098e-01 -8.86032820e-01
-7.13618159e-01 7.05923855e-01 1.09021813e-02 -1.07462513e+00
-2.46302247e-01 -5.67534506e-01 -1.92529917e-01 8.00941885e-01
-1.37206101e+00 -9.30998981e-01 -6.23459578e-01 1.00623392e-01
5.40075600e-01 1.35350525e-01 7.93779433e-01 2.26929486e-01
-2.18951717e-01 4.90324438e-01 3.08225583e-02 -1.86016023e-01
4.68047380e-01 -1.34605634e+00 1.24112034e+00 8.39571714e-01
-1.37526557e-01 7.09155023e-01 9.45487440e-01 -6.77095890e-01
-1.72124350e+00 -6.89263940e-01 3.87596428e-01 -6.19896531e-01
4.21557724e-01 -3.11175346e-01 -3.98394316e-01 1.10953915e+00
4.33400750e-01 -4.21022356e-01 3.84757310e-01 3.37932169e-01
-1.58018097e-01 3.04008067e-01 -9.91910994e-01 1.12451982e+00
1.59329915e+00 -6.75075501e-02 -4.39792454e-01 2.41915375e-01
8.91408563e-01 -1.14389539e+00 -6.86882496e-01 9.69818383e-02
7.86153078e-01 -1.31404340e+00 1.01979232e+00 -3.67524624e-01
4.20254648e-01 -3.31745803e-01 -4.47907329e-01 -1.32014620e+00
-4.80055392e-01 -5.90893805e-01 5.89477047e-02 4.13812488e-01
1.93272039e-01 -6.31804645e-01 1.07508063e+00 6.40804708e-01
-6.15753293e-01 -7.25155115e-01 -6.96370661e-01 -6.78571999e-01
8.30090791e-03 -5.44260204e-01 6.02849543e-01 4.34573501e-01
5.93156517e-02 2.69882172e-01 -3.29254895e-01 1.09084852e-01
6.02804005e-01 -1.26991108e-01 1.30713391e+00 -5.41882157e-01
-4.15149868e-01 -5.60709715e-01 -5.04647136e-01 -1.97366333e+00
-6.87637985e-01 -4.27476436e-01 3.35475534e-01 -1.77350247e+00
-1.45398214e-01 -6.19840860e-01 1.17153056e-01 4.52075332e-01
-1.53268337e-01 3.11723948e-01 4.26527530e-01 1.53784990e-01
-7.51521230e-01 7.29937494e-01 1.01735175e+00 2.22076252e-01
-2.79738247e-01 -4.35975432e-01 -9.58692074e-01 6.92612350e-01
5.49247384e-01 -1.73744246e-01 -4.41489547e-01 -1.23081839e+00
3.57504547e-01 -1.26584128e-01 3.51715177e-01 -1.40499651e+00
3.28701258e-01 7.17393681e-02 2.29043216e-01 -3.51177067e-01
6.94072604e-01 -4.66881245e-01 -1.14373621e-02 8.10469329e-01
-2.21342854e-02 6.43270552e-01 7.56718159e-01 6.26163542e-01
2.17485458e-01 2.13007763e-01 4.78944182e-01 -2.71817833e-01
-1.22391081e+00 3.74960378e-02 -5.05096257e-01 1.51424587e-01
9.30131674e-01 -4.82363313e-01 -6.01356030e-01 -6.00146532e-01
-3.55071902e-01 5.46719432e-01 9.01234090e-01 4.67737526e-01
3.60893667e-01 -1.02685928e+00 -5.61675131e-01 5.33609949e-02
3.45916897e-02 1.73454955e-01 2.11564869e-01 6.18324041e-01
-1.25904822e+00 3.66304725e-01 -4.78067249e-01 -6.63005888e-01
-8.90948474e-01 1.01784512e-01 3.86338443e-01 6.82911798e-02
-9.54146862e-01 1.47206354e+00 9.51367170e-02 -3.23114067e-01
2.76820064e-01 -6.26993597e-01 2.39478737e-01 -3.57516795e-01
4.80566621e-01 5.51369250e-01 -1.03669770e-01 -3.76409769e-01
-3.90185684e-01 5.70680678e-01 -1.19511038e-01 -5.04875839e-01
1.45512700e+00 -2.76826322e-01 4.91913885e-01 2.10043699e-01
8.79774630e-01 -2.18918428e-01 -2.01338053e+00 2.23670736e-01
-3.67721826e-01 -5.24265349e-01 1.78428814e-01 -9.04665053e-01
-6.06012881e-01 6.34997427e-01 8.93236279e-01 7.55294114e-02
8.07280481e-01 -1.54330805e-01 7.82947481e-01 4.34109300e-01
1.09030497e+00 -6.71053112e-01 -9.95239243e-02 7.77650952e-01
7.12448180e-01 -1.21033263e+00 3.23040481e-03 -1.67190507e-01
-5.80054402e-01 6.27981126e-01 7.32716322e-01 -4.34574813e-01
1.50407076e-01 3.87042135e-01 5.20624705e-02 -2.70738631e-01
-7.28006959e-01 -2.73821175e-01 -3.99295658e-01 7.67675042e-01
3.49048525e-01 -9.24811885e-02 -3.93412150e-02 1.78132340e-01
-6.48747325e-01 -1.77434489e-01 3.16949755e-01 1.28749061e+00
-7.25020111e-01 -6.84371352e-01 -1.25849590e-01 1.87929109e-01
6.81123137e-02 -2.69567341e-01 6.57429621e-02 1.24857128e+00
-1.14490427e-01 9.31862533e-01 3.40628535e-01 -3.23444247e-01
6.44973159e-01 -4.06073689e-01 6.87831819e-01 -3.92123252e-01
-7.87156165e-01 7.50018656e-02 4.13654476e-01 -1.02303779e+00
-1.68768078e-01 -6.24401391e-01 -1.38150275e+00 -9.69685197e-01
-2.40174625e-02 1.00337401e-01 7.55888999e-01 5.25156140e-01
5.79337776e-01 6.66838408e-01 -2.97896892e-01 -1.22598088e+00
-3.65579754e-01 -8.34049523e-01 -1.94330916e-01 -8.50743577e-02
1.11534752e-01 -7.45472431e-01 -2.07300380e-01 -8.95616636e-02] | [4.528714179992676, 0.8366382122039795] |
35e4bd91-2c65-4db4-bf06-0916faecbace | span-detection-for-aspect-based-sentiment | 2110.07833 | null | https://arxiv.org/abs/2110.07833v1 | https://arxiv.org/pdf/2110.07833v1.pdf | Span Detection for Aspect-Based Sentiment Analysis in Vietnamese | Aspect-based sentiment analysis plays an essential role in natural language processing and artificial intelligence. Recently, researchers only focused on aspect detection and sentiment classification but ignoring the sub-task of detecting user opinion span, which has enormous potential in practical applications. In this paper, we present a new Vietnamese dataset (UIT-ViSD4SA) consisting of 35,396 human-annotated spans on 11,122 feedback comments for evaluating the span detection in aspect-based sentiment analysis. Besides, we also propose a novel system using Bidirectional Long Short-Term Memory (BiLSTM) with a Conditional Random Field (CRF) layer (BiLSTM-CRF) for the span detection task in Vietnamese aspect-based sentiment analysis. The best result is a 62.76% F1 score (macro) for span detection using BiLSTM-CRF with embedding fusion of syllable embedding, character embedding, and contextual embedding from XLM-RoBERTa. In future work, span detection will be extended in many NLP tasks such as constructive detection, emotion recognition, complaint analysis, and opinion mining. Our dataset is freely available at https://github.com/kimkim00/UIT-ViSD4SA for research purposes. | ['Kiet Van Nguyen', 'Duc-Vu Nguyen', 'Phuc Huynh Pham', 'Luong Luc Phan', 'Sieu Khai Huynh', 'Kim Thi-Thanh Nguyen'] | 2021-10-15 | null | null | null | null | ['vietnamese-aspect-based-sentiment-analysis'] | ['natural-language-processing'] | [-7.90031627e-02 -3.46838415e-01 -7.15603307e-02 -3.59521389e-01
-9.25595522e-01 -4.73219723e-01 8.62795562e-02 4.69954461e-01
-6.34674013e-01 4.67570812e-01 4.27820355e-01 -3.15985292e-01
6.99762940e-01 -7.13689327e-01 -1.84639812e-01 -4.13469166e-01
2.61758029e-01 7.81223476e-02 -1.73271939e-01 -4.39425558e-01
6.22454405e-01 -2.12358236e-02 -1.18528187e+00 5.75258911e-01
8.52095127e-01 1.11713660e+00 8.30379725e-02 9.11753833e-01
-2.36842260e-01 5.34117937e-01 -9.10583019e-01 -7.73244441e-01
-4.57314610e-01 -4.71050590e-01 -7.16655552e-01 7.35646263e-02
-2.67401282e-02 7.10319653e-02 2.84944236e-01 8.31968129e-01
7.74575472e-01 3.51642966e-01 4.65297282e-01 -1.05735028e+00
-8.84276628e-01 4.99734461e-01 -8.96904588e-01 2.01386109e-01
5.14332414e-01 -1.36249647e-01 1.31917536e+00 -1.03584087e+00
5.20471394e-01 1.05998886e+00 5.75881124e-01 3.26012105e-01
-5.16046345e-01 -4.61792856e-01 2.08510637e-01 4.76015240e-01
-1.04160786e+00 -9.20045525e-02 8.48308861e-01 -1.31685108e-01
1.57922828e+00 1.80823579e-01 8.86078358e-01 9.90681708e-01
5.59702516e-01 1.19804907e+00 1.04228413e+00 -5.31746745e-01
3.09927538e-02 3.43683749e-01 4.97566134e-01 6.47877097e-01
-1.90046579e-01 -4.26394671e-01 -8.81453216e-01 -1.49853900e-01
-1.10584516e-02 -2.41801664e-01 1.24067508e-01 8.25828195e-01
-1.13599765e+00 1.06041455e+00 2.85935011e-02 3.24000597e-01
-4.00815606e-01 -3.17502260e-01 1.02610373e+00 4.61811960e-01
9.10236597e-01 1.55741647e-01 -8.78879666e-01 -4.45258081e-01
-6.69071257e-01 3.25051807e-02 7.14005351e-01 9.49141264e-01
4.21477705e-01 3.01254898e-01 -4.02359575e-01 9.16819632e-01
2.78139651e-01 9.26338911e-01 8.88183653e-01 -3.37231666e-01
4.96971577e-01 6.59414232e-01 -2.36509547e-01 -9.97941196e-01
-4.51180011e-01 -2.92507112e-01 -6.59232259e-01 -4.34296459e-01
-2.40574509e-01 -4.94584113e-01 -7.49533057e-01 1.38520885e+00
3.40791315e-01 -5.41845322e-01 9.19018760e-02 6.45317435e-01
1.12502432e+00 9.08571422e-01 2.31392533e-01 -3.21910083e-01
1.95775831e+00 -1.12786639e+00 -7.58292794e-01 -2.57718742e-01
7.55200148e-01 -1.35406172e+00 1.21609318e+00 5.40927649e-01
-8.52030993e-01 -6.34781599e-01 -8.12918425e-01 -2.53762513e-01
-6.44006848e-01 4.45256442e-01 5.54813087e-01 6.37645245e-01
-7.67305315e-01 -2.06179574e-01 -5.25339603e-01 -4.30175841e-01
-1.60173140e-03 7.68475235e-02 -2.27445096e-01 2.05189630e-01
-1.47665524e+00 6.70917630e-01 1.75354436e-01 2.67771721e-01
-3.07952434e-01 -1.69700503e-01 -1.22191310e+00 -2.78888196e-01
1.10708967e-01 -3.93307954e-01 1.34447610e+00 -7.28968024e-01
-1.11471498e+00 7.84979880e-01 -7.26055086e-01 -1.71414271e-01
-1.36524543e-01 -4.61167574e-01 -8.87964785e-01 7.00701550e-02
2.80209631e-01 5.09696662e-01 8.80103230e-01 -6.61791503e-01
-5.28823078e-01 -4.86171752e-01 -2.62471318e-01 2.91569769e-01
-6.85654581e-01 6.84639871e-01 -1.76255286e-01 -7.55659580e-01
-3.57550353e-01 -8.40154588e-01 -4.07120660e-02 -6.11649811e-01
-5.78735352e-01 -6.54757798e-01 7.80569315e-01 -1.09478378e+00
1.45154595e+00 -2.02530742e+00 -4.24520642e-01 -2.25664899e-01
-1.65291965e-01 2.87019402e-01 -2.63356239e-01 6.62186623e-01
1.61749303e-01 2.11752042e-01 -1.88313127e-01 -5.04483283e-01
5.67669719e-02 -8.76387656e-02 -3.73314828e-01 1.71428189e-01
3.37647438e-01 1.10316765e+00 -6.46703899e-01 -9.20521975e-01
-1.07340865e-01 4.81645495e-01 -1.49593517e-01 9.86851975e-02
-1.26864523e-01 2.38707468e-01 -3.60748649e-01 9.48129535e-01
5.75015664e-01 1.73242882e-01 -1.83302522e-01 -8.45228434e-02
-2.60195196e-01 4.53522027e-01 -7.74386585e-01 1.54983032e+00
-6.81677759e-01 6.97637737e-01 -1.33168563e-01 -6.96645319e-01
1.12689304e+00 5.59891999e-01 1.37385696e-01 -8.12906325e-01
3.65475088e-01 -7.90010858e-03 -1.35444313e-01 -7.09008992e-01
1.21933210e+00 -3.66649956e-01 -6.30587161e-01 5.83253562e-01
1.16105214e-01 -8.28914717e-02 5.91518164e-01 2.34078094e-01
6.97825789e-01 -1.00491397e-01 4.84115243e-01 1.22084156e-01
7.17386127e-01 1.06770642e-01 7.99933493e-01 1.69712573e-01
-4.24717724e-01 5.26129782e-01 5.62322021e-01 -7.60489926e-02
-7.92728901e-01 -6.28692508e-01 -1.78197891e-01 1.26466990e+00
-2.68664122e-01 -5.70393205e-01 -5.05364835e-01 -8.07378352e-01
-2.98871428e-01 9.02347803e-01 -5.38013041e-01 5.14633246e-02
-5.04629135e-01 -8.12822342e-01 7.14119613e-01 6.46976173e-01
5.39658785e-01 -1.70314634e+00 -4.68667269e-01 2.78166950e-01
-5.98929942e-01 -1.13429248e+00 -8.63720119e-01 1.47560582e-01
-7.36332953e-01 -7.29114234e-01 -6.36063993e-01 -1.09748101e+00
3.58293355e-01 7.00322762e-02 1.14488494e+00 -2.37885565e-01
-3.35191220e-01 1.87239766e-01 -8.10617328e-01 -7.07110763e-01
2.55231895e-02 1.92016527e-01 -1.07869484e-01 -2.39177182e-01
9.84962046e-01 -1.73161089e-01 -5.45959413e-01 1.30133420e-01
-6.69494271e-01 -2.48100862e-01 6.80480361e-01 7.34061360e-01
5.28463244e-01 -8.97098035e-02 7.93559730e-01 -9.25351560e-01
9.58031476e-01 -3.88212949e-01 -4.53869663e-02 6.77490979e-02
-6.10694826e-01 -4.06786978e-01 7.90223658e-01 -3.83319288e-01
-1.28838444e+00 -3.97213429e-01 -7.80655265e-01 1.78791851e-01
-2.38594055e-01 8.80079865e-01 -1.30283982e-01 5.72463810e-01
1.88789979e-01 4.77113128e-01 -2.95880049e-01 -3.47444028e-01
1.71217069e-01 1.11373222e+00 8.39172229e-02 -1.20122567e-01
2.97375530e-01 9.77279246e-02 -5.46260059e-01 -1.29630244e+00
-1.11028004e+00 -7.85031557e-01 -4.05864179e-01 -1.51417390e-01
1.02340066e+00 -1.00249732e+00 -7.93176472e-01 6.84501410e-01
-1.25160944e+00 6.21779040e-02 -1.09063409e-01 5.19332349e-01
-1.11073844e-01 5.87701440e-01 -1.09672165e+00 -1.02838433e+00
-1.19370735e+00 -8.44198048e-01 1.26835120e+00 3.52276325e-01
-5.56403160e-01 -1.01428747e+00 2.48844683e-01 7.65953481e-01
2.79102117e-01 -5.02290726e-02 8.57141674e-01 -6.57446384e-01
3.99789721e-01 -3.90341789e-01 -7.21371360e-03 6.56004250e-01
-1.84623018e-01 -1.36462096e-02 -9.65204835e-01 8.54307413e-02
2.15444148e-01 -5.30530572e-01 1.01617694e+00 2.99644828e-01
6.87298656e-01 -2.71683753e-01 2.97263652e-01 -3.99987437e-02
1.33007061e+00 2.82937258e-01 6.11697257e-01 2.34379813e-01
6.81896508e-01 5.93531728e-01 1.11598837e+00 6.06617272e-01
5.89956403e-01 7.56046399e-02 1.03449344e-03 -1.07701086e-02
9.90506709e-02 -8.27517211e-02 9.74674821e-01 1.67015803e+00
1.68903649e-01 -4.73069608e-01 -5.80997109e-01 8.94779205e-01
-1.62581956e+00 -8.22197735e-01 -5.54421425e-01 1.70113838e+00
7.04701841e-01 2.80326277e-01 1.87692821e-01 3.43003631e-01
6.76255167e-01 5.29370666e-01 -3.91185373e-01 -1.30773962e+00
-1.58462286e-01 1.51154846e-01 -4.01047245e-02 2.69814491e-01
-1.20501161e+00 9.21590626e-01 4.76593828e+00 9.55115020e-01
-1.09468091e+00 4.07517463e-01 6.19821727e-01 1.63495377e-01
-3.28767210e-01 -2.63972580e-01 -1.00771141e+00 4.70674485e-01
1.07727873e+00 3.88076305e-02 -2.25224316e-01 9.70265687e-01
4.05114889e-01 -4.81139421e-01 -3.03588241e-01 7.04894483e-01
5.84272325e-01 -6.22934580e-01 -1.33079469e-01 -1.78436369e-01
5.48861980e-01 -2.59131957e-02 2.00961441e-01 6.43150389e-01
-3.47582728e-01 -6.81713164e-01 5.09625494e-01 3.00083518e-01
7.90261507e-01 -1.13699377e+00 1.21132946e+00 2.63909400e-01
-1.50192189e+00 3.12000751e-01 -4.69078064e-01 -1.32549271e-01
4.69315588e-01 9.21373248e-01 -6.30621076e-01 4.93392885e-01
6.15070164e-01 8.75845671e-01 -4.91831899e-01 6.94786251e-01
-5.61630905e-01 1.01410007e+00 1.11033425e-01 -5.89091480e-01
1.34620488e-01 -1.44946367e-01 5.81656098e-01 1.55024731e+00
1.95419952e-01 -2.65163571e-01 1.29755080e-01 3.03959101e-01
-8.48765448e-02 5.61880469e-01 -3.74348640e-01 -2.98212081e-01
-6.00769632e-02 1.68489850e+00 -1.00564206e+00 -3.59453678e-01
-5.42195618e-01 9.75406766e-01 2.15887487e-01 2.17354685e-01
-7.99422324e-01 -7.65820444e-01 3.80550027e-01 -2.84503102e-01
5.11441708e-01 -2.36454979e-01 -4.88859057e-01 -1.26669931e+00
2.28089601e-01 -8.20801973e-01 3.25838327e-01 -6.76753461e-01
-1.39051425e+00 9.25063968e-01 -4.87047732e-01 -1.23257899e+00
-1.26007184e-01 -5.69525778e-01 -1.01007891e+00 8.04208398e-01
-1.47658694e+00 -1.40819633e+00 5.98198809e-02 3.96173388e-01
1.02426624e+00 5.63175306e-02 1.02185726e+00 2.95495450e-01
-6.17440462e-01 6.32184267e-01 -2.58805901e-01 2.43448049e-01
9.98221397e-01 -1.28126204e+00 4.18295920e-01 8.54155898e-01
2.04728004e-02 5.42270601e-01 6.04838133e-01 -7.04934597e-01
-1.28030634e+00 -1.24664652e+00 1.69092155e+00 -2.95767933e-01
7.21907437e-01 -3.32057625e-01 -5.54180145e-01 5.65699339e-01
6.87677443e-01 -5.98676443e-01 1.14152563e+00 2.67445207e-01
-2.37346396e-01 -7.29752705e-02 -9.85244811e-01 3.19645822e-01
5.59598684e-01 -6.82827890e-01 -5.37274420e-01 2.15335950e-01
9.73017156e-01 1.19358068e-02 -1.03388584e+00 2.96914995e-01
5.42190135e-01 -6.89215600e-01 4.33025390e-01 -3.00872058e-01
6.08579576e-01 -1.00745812e-01 -1.15321718e-01 -1.24156153e+00
1.40625581e-01 -2.32988983e-01 -5.05316742e-02 1.57676065e+00
7.14780271e-01 -5.77681065e-01 4.93340313e-01 1.84735879e-01
-2.32232258e-01 -1.05896389e+00 -5.78268886e-01 -4.18409348e-01
-8.17178488e-02 -9.24047232e-01 2.55824476e-01 5.29341102e-01
2.63307452e-01 1.06064939e+00 -4.18723792e-01 -3.15720171e-01
2.61083245e-01 5.94109476e-01 2.71913350e-01 -7.30218470e-01
-1.45868361e-01 -2.18119860e-01 -1.33593544e-01 -8.69689345e-01
2.26676539e-01 -6.18790209e-01 6.70738667e-02 -1.70287395e+00
4.34354424e-01 7.24916607e-02 -2.15507299e-01 5.14951110e-01
-3.76693726e-01 4.51550096e-01 4.73450497e-02 -1.41069680e-01
-9.41280067e-01 8.01662087e-01 1.35413945e+00 -1.04965709e-01
-1.83678567e-01 1.85660765e-01 -8.52435350e-01 6.25164092e-01
1.30963826e+00 -4.62438941e-01 -7.87241086e-02 -1.57731265e-01
5.84491789e-01 3.49663645e-02 -2.40297094e-01 -5.24323106e-01
4.78273407e-02 8.00435543e-02 4.81902212e-01 -1.27623236e+00
3.04651231e-01 -1.71886459e-01 -7.60170877e-01 5.30433714e-01
-1.58499837e-01 6.52639568e-01 2.28590965e-01 2.41361499e-01
-5.76931655e-01 -4.52449262e-01 1.69855505e-01 -1.93109348e-01
-8.47198009e-01 1.07129805e-01 -1.01297927e+00 2.86660314e-01
8.33948731e-01 2.22950280e-01 -4.10192162e-01 -4.20091987e-01
-4.65537190e-01 2.84064740e-01 3.29912342e-02 4.29893792e-01
1.00997293e+00 -8.79861653e-01 -8.69572163e-01 1.73804715e-01
3.86267275e-01 -3.10379535e-01 5.13044238e-01 9.72398937e-01
-2.16153488e-01 4.45445806e-01 1.09247208e-01 -2.11853653e-01
-1.58323157e+00 4.91697460e-01 -5.76385140e-01 -6.13233328e-01
-2.04468787e-01 8.84808898e-01 -2.85927474e-01 -7.10703075e-01
-3.93470302e-02 -1.59485906e-01 -7.40495861e-01 4.77749735e-01
4.23980027e-01 4.07877296e-01 2.56123871e-01 -9.90912616e-01
-3.90774041e-01 4.84101385e-01 -1.58396706e-01 -2.65459001e-01
1.03395069e+00 -3.07497740e-01 -4.11616802e-01 8.05418134e-01
1.34745979e+00 3.90828788e-01 -1.60285711e-01 8.37991685e-02
-3.41687049e-03 3.95309664e-02 -1.38026714e-01 -7.80062497e-01
-7.86178410e-01 1.03563547e+00 2.54076511e-01 2.90243328e-01
1.24648154e+00 -1.52158439e-01 1.57640243e+00 3.22038710e-01
1.02318395e-02 -1.47404194e+00 1.03539936e-01 9.65908408e-01
7.47532129e-01 -1.42040455e+00 -1.81627110e-01 -3.08572575e-02
-1.30554795e+00 9.58207965e-01 5.98046541e-01 -6.88754320e-02
7.44932890e-01 1.99977636e-01 4.11142409e-01 -2.50924468e-01
-9.24273014e-01 -3.86661887e-01 2.51763880e-01 3.86886150e-01
8.60839963e-01 2.36075804e-01 -8.26588452e-01 8.19534600e-01
-5.52188873e-01 -4.77874577e-01 4.19667304e-01 1.08155954e+00
-4.83097494e-01 -1.09165525e+00 -2.17305467e-01 7.30327785e-01
-9.13895667e-01 -5.59318960e-01 -2.76807308e-01 3.35658252e-01
1.21904552e-01 1.50206292e+00 -4.33838330e-02 -4.66834128e-01
2.28780806e-01 2.67632306e-01 1.67472195e-02 -6.51268303e-01
-9.42966700e-01 4.31664616e-01 5.05364954e-01 -1.58951208e-01
-5.70292413e-01 -6.97078109e-01 -1.28268349e+00 -1.63644731e-01
-4.80076224e-01 5.52126586e-01 8.09088051e-01 9.28238213e-01
2.53311336e-01 5.05896509e-01 7.44886041e-01 -2.06656247e-01
-4.92591672e-02 -1.48994195e+00 -7.52863824e-01 4.63498831e-02
7.23825991e-02 -8.72685686e-02 -3.32121342e-01 8.07323158e-02] | [11.373903274536133, 6.741175651550293] |
a875ad95-998f-4c44-89cc-3e9a12204b88 | bifsmnv2-pushing-binary-neural-networks-for | 2211.06987 | null | https://arxiv.org/abs/2211.06987v2 | https://arxiv.org/pdf/2211.06987v2.pdf | BiFSMNv2: Pushing Binary Neural Networks for Keyword Spotting to Real-Network Performance | Deep neural networks, such as the Deep-FSMN, have been widely studied for keyword spotting (KWS) applications while suffering expensive computation and storage. Therefore, network compression technologies like binarization are studied to deploy KWS models on edge. In this paper, we present a strong yet efficient binary neural network for KWS, namely BiFSMNv2, pushing it to the real-network accuracy performance. First, we present a Dual-scale Thinnable 1-bit-Architecture to recover the representation capability of the binarized computation units by dual-scale activation binarization and liberate the speedup potential from an overall architecture perspective. Second, we also construct a Frequency Independent Distillation scheme for KWS binarization-aware training, which distills the high and low-frequency components independently to mitigate the information mismatch between full-precision and binarized representations. Moreover, we propose the Learning Propagation Binarizer, a general and efficient binarizer that enables the forward and backward propagation of binary KWS networks to be continuously improved through learning. We implement and deploy the BiFSMNv2 on ARMv8 real-world hardware with a novel Fast Bitwise Computation Kernel, which is proposed to fully utilize registers and increase instruction throughput. Comprehensive experiments show our BiFSMNv2 outperforms existing binary networks for KWS by convincing margins across different datasets and achieves comparable accuracy with the full-precision networks (only a tiny 1.51% drop on Speech Commands V1-12). We highlight that benefiting from the compact architecture and optimized hardware kernel, BiFSMNv2 can achieve an impressive 25.1x speedup and 20.2x storage-saving on edge hardware. | ['Xianglong Liu', 'Jie Luo', 'Jiakai Wang', 'Zejun Ma', 'Yang Zhang', 'Xiaoyang Li', 'Yifu Ding', 'Xudong Ma', 'Haotong Qin'] | 2022-11-13 | null | null | null | null | ['keyword-spotting'] | ['speech'] | [ 1.70927867e-01 -2.51611114e-01 -8.91143441e-01 -3.88613492e-01
-5.45450687e-01 3.50578129e-02 8.40780735e-02 -4.85323481e-02
-6.80593550e-01 4.14190054e-01 2.27025673e-02 -1.09837377e+00
-8.10880661e-02 -9.80342805e-01 -1.11080873e+00 -5.94826758e-01
1.27527595e-01 -1.28599986e-01 3.71728867e-01 -3.80422056e-01
-1.20190360e-01 4.56697583e-01 -1.40251589e+00 4.59450454e-01
4.94161755e-01 1.51958418e+00 2.80538291e-01 8.49693239e-01
-8.43956247e-02 9.40271795e-01 -6.90161288e-01 -5.70721269e-01
4.32354420e-01 2.14507088e-01 -5.29220164e-01 -9.30335879e-01
6.37312710e-01 -9.42191243e-01 -1.14968097e+00 1.06484032e+00
6.84893727e-01 -1.57808319e-01 2.72422343e-01 -1.05210817e+00
-6.80946767e-01 1.41384017e+00 -4.76429015e-01 4.59762961e-01
-4.82531130e-01 2.58636773e-01 9.61777687e-01 -6.13609076e-01
-4.48580459e-02 1.10719025e+00 8.87673914e-01 3.80854249e-01
-8.23066413e-01 -1.23047709e+00 -1.31629586e-01 8.54771793e-01
-1.72111464e+00 -6.97425842e-01 5.58126330e-01 3.86161327e-01
1.57161176e+00 3.99303049e-01 7.68017590e-01 9.79905486e-01
6.77715167e-02 1.09457517e+00 6.21462464e-01 -4.35311943e-01
1.34729847e-01 -2.87119597e-01 6.95607126e-01 7.91999876e-01
5.92508137e-01 5.63778430e-02 -8.40618610e-01 7.41149038e-02
5.57159364e-01 3.00564528e-01 -4.91430789e-01 3.73292744e-01
-7.06605017e-01 6.10399842e-01 7.53090084e-01 1.55272365e-01
-2.50912726e-01 9.71450210e-01 7.64491141e-01 3.24780613e-01
8.42384696e-02 -1.15210094e-01 -8.76565516e-01 -4.75649983e-01
-1.43797576e+00 -7.95147195e-02 6.58637226e-01 1.11015308e+00
6.77770793e-01 7.35034347e-01 -4.52538311e-01 6.40412509e-01
1.21490039e-01 7.04056799e-01 1.01108789e+00 -4.05937999e-01
6.50188208e-01 5.21761417e-01 -8.86589885e-01 -9.12201524e-01
-3.09728026e-01 -5.02502620e-01 -1.39849114e+00 -1.72408730e-01
-9.75539088e-02 1.00431539e-01 -1.20855105e+00 1.59251058e+00
9.13382974e-03 3.57636839e-01 1.13846198e-01 7.15859175e-01
1.08982730e+00 9.14656937e-01 -9.36373249e-02 5.22295088e-02
1.49305189e+00 -1.23605430e+00 -7.76225984e-01 -1.31882310e-01
7.79059291e-01 -4.09890264e-01 1.08536363e+00 3.86881590e-01
-1.22005808e+00 -6.56845689e-01 -1.78013122e+00 -3.68411779e-01
-5.20996511e-01 2.89506137e-01 9.72808778e-01 9.34607387e-01
-1.23031378e+00 9.60449398e-01 -1.18316603e+00 5.43856502e-01
5.02111256e-01 7.96123207e-01 2.36741845e-02 -6.04973966e-03
-1.50885749e+00 5.00220597e-01 9.58239675e-01 8.30700472e-02
-7.05385447e-01 -9.23812449e-01 -9.26529884e-01 6.85689211e-01
3.42128366e-01 -3.84570479e-01 1.35891438e+00 -6.06733799e-01
-1.70606554e+00 5.62603511e-02 1.05059613e-02 -1.14459956e+00
-1.18999317e-01 -8.82450789e-02 -5.54951370e-01 1.94946051e-01
-6.52882099e-01 6.15709484e-01 8.42734754e-01 -2.99947530e-01
-5.57408273e-01 -1.02360450e-01 -1.34783462e-01 2.54345983e-02
-1.25747287e+00 -2.89968818e-01 -7.04312265e-01 -1.10663748e+00
-1.05304219e-01 -5.02649546e-01 1.47781193e-01 -1.98443919e-01
-4.50178891e-01 -2.78645694e-01 9.51067388e-01 -7.43486762e-01
1.88879359e+00 -2.07463813e+00 -3.01103234e-01 1.70788109e-01
4.76002693e-01 9.37162340e-01 5.65769970e-02 -1.70977294e-01
-2.58577764e-02 3.99732664e-02 7.14368373e-02 -4.96758312e-01
1.77219927e-01 2.64879465e-01 -4.77827609e-01 3.29211056e-01
5.42943217e-02 1.27292526e+00 -4.72379118e-01 -3.60471815e-01
4.99023451e-03 6.00601554e-01 -5.81992090e-01 -1.82660036e-02
2.50442941e-02 -8.27675819e-01 1.15854293e-01 9.02764440e-01
8.99798691e-01 -4.04568255e-01 1.21415049e-01 -7.82055199e-01
1.43244892e-01 7.35424876e-01 -8.61545622e-01 1.59525990e+00
-5.23292184e-01 7.23675072e-01 -1.09948451e-03 -1.17732465e+00
6.84119999e-01 -9.95483156e-03 1.69611145e-02 -1.08621609e+00
3.51646096e-01 4.19708878e-01 -7.05060288e-02 1.89242601e-01
1.03727663e+00 3.15460742e-01 -6.54473454e-02 2.84195065e-01
4.13635969e-01 2.47490972e-01 1.98203903e-02 3.14739734e-01
1.05448472e+00 -4.46960628e-01 -4.75958176e-02 -1.37129173e-01
2.52676070e-01 -3.45031023e-01 3.89418364e-01 8.84474933e-01
-7.10748211e-02 5.19493259e-02 1.19895048e-01 -4.45748866e-01
-9.33506191e-01 -6.97289705e-01 -3.23987693e-01 1.32979345e+00
1.52098238e-01 -8.72925580e-01 -7.58712113e-01 -3.50101382e-01
-2.34564953e-02 5.37406921e-01 -1.53251603e-01 -8.71224999e-01
-7.79904425e-01 -9.57729280e-01 1.33078051e+00 8.48928988e-01
9.33324814e-01 -6.33951247e-01 -7.80321181e-01 -7.12605342e-02
2.53029078e-01 -9.58736956e-01 -5.83814204e-01 9.31621075e-01
-8.05701196e-01 -5.45474827e-01 -5.88226318e-01 -7.28420258e-01
-5.51569387e-02 3.69478703e-01 9.15664792e-01 2.61601090e-01
-2.70139396e-01 -3.38395566e-01 -3.03542525e-01 -1.07026875e-01
-2.19573110e-01 4.88482386e-01 3.47730190e-01 -5.37446082e-01
5.37648678e-01 -7.24966228e-01 -7.63074994e-01 -7.14810938e-02
-9.94619250e-01 3.59507322e-01 9.69267309e-01 1.14202094e+00
6.01176918e-01 3.20251435e-01 1.69469446e-01 -4.28187191e-01
4.25770551e-01 -3.55728686e-01 -5.62451065e-01 2.25483507e-01
-1.17267168e+00 2.83654213e-01 8.16093087e-01 -8.32291663e-01
-4.40236419e-01 -3.23583364e-01 -4.00415003e-01 -7.92426169e-01
4.79627848e-01 4.30261195e-01 -3.44535429e-03 -4.84450102e-01
5.47483563e-01 5.41064620e-01 -1.31060705e-01 -3.62755448e-01
4.26763296e-01 9.42317367e-01 8.42340469e-01 -4.92683470e-01
5.72505534e-01 2.83665899e-02 -8.76903608e-02 -5.65271735e-01
-3.39659184e-01 -1.49353385e-01 1.02354430e-01 3.28021765e-01
2.58971483e-01 -1.22597766e+00 -1.19933927e+00 5.80922306e-01
-1.12839317e+00 -7.20300555e-01 -1.73334673e-01 2.13237062e-01
1.19559325e-01 4.93386209e-01 -1.18880749e+00 -5.28762341e-01
-1.25805402e+00 -1.51682270e+00 9.54149783e-01 4.56372768e-01
1.08818211e-01 -5.43571830e-01 -7.14397669e-01 -2.38295440e-02
8.35485458e-01 -5.89908242e-01 9.75150108e-01 -6.91333592e-01
-6.45248652e-01 -8.53895023e-03 -7.86654174e-01 5.72465897e-01
-2.57542282e-01 -3.30083996e-01 -9.38468397e-01 -3.93732339e-01
-2.94856858e-02 -6.02010250e-01 1.38943744e+00 2.58736670e-01
1.87417841e+00 -6.08721614e-01 -2.46214047e-01 1.43456256e+00
1.31589532e+00 1.91056475e-01 5.24641812e-01 2.81839997e-01
8.70234311e-01 -4.20278877e-01 3.15223006e-03 3.55485022e-01
3.15329731e-01 5.37124395e-01 5.57689965e-01 -7.50284642e-02
-4.93533313e-01 -2.69931167e-01 3.28067541e-01 1.31728542e+00
3.88590753e-01 -2.65273064e-01 -6.49819016e-01 3.33563834e-01
-1.46077955e+00 -4.77649271e-01 1.29886419e-01 1.95367134e+00
1.44000602e+00 6.43083990e-01 -3.29998732e-01 6.38482690e-01
4.28858072e-01 3.20900351e-01 -7.97666252e-01 -5.78748465e-01
-1.65939689e-01 9.35383856e-01 1.39086306e+00 2.82516450e-01
-9.81267154e-01 9.79444802e-01 5.37376356e+00 2.01040864e+00
-1.44561625e+00 1.97780445e-01 1.02522755e+00 -2.07027882e-01
-1.28695726e-01 -4.06843543e-01 -1.55684555e+00 7.14445889e-01
1.64041817e+00 1.53513327e-01 5.62779725e-01 1.19885588e+00
-4.89786655e-01 2.87558109e-01 -8.76577914e-01 1.23848879e+00
-6.01589158e-02 -1.72809243e+00 1.31595910e-01 -1.72676623e-01
4.00756657e-01 9.15358886e-02 3.46565366e-01 7.47749925e-01
2.20202968e-01 -1.18267429e+00 9.07313406e-01 -6.94887862e-02
1.52157915e+00 -1.06825078e+00 7.38207221e-01 2.22530946e-01
-1.38346040e+00 -1.96209013e-01 -5.59860170e-01 1.17989024e-02
-2.62305681e-02 7.06426680e-01 -8.59921932e-01 3.17811638e-01
1.06454635e+00 2.88181126e-01 -4.51049894e-01 5.57968915e-01
1.34113906e-02 9.17267263e-01 -6.28709018e-01 -3.81214261e-01
3.20869148e-01 3.70861381e-01 -1.46400824e-01 1.63901913e+00
2.73698062e-01 -4.22284659e-03 -2.48192847e-01 6.34199798e-01
-4.73147094e-01 -2.07799047e-01 -6.21359572e-02 -1.83541566e-01
8.11450541e-01 1.24675548e+00 -4.97557044e-01 -6.40237391e-01
-1.75440997e-01 1.02614474e+00 4.68887627e-01 -1.05568795e-02
-1.04460895e+00 -9.04392242e-01 8.40938389e-01 -2.59857088e-01
5.46879232e-01 -1.40659600e-01 -4.55004692e-01 -9.92295861e-01
-5.52126840e-02 -9.43234801e-01 1.77908614e-01 -4.96581227e-01
-5.70243061e-01 7.41267204e-01 2.45080870e-02 -4.72058237e-01
4.38615531e-02 -1.00234199e+00 -3.39487791e-01 8.39425147e-01
-1.89839792e+00 -1.09940195e+00 -2.84417391e-01 6.32979512e-01
5.16151071e-01 -2.83047915e-01 5.58681071e-01 7.23877847e-01
-8.08277845e-01 1.65091324e+00 2.00633883e-01 2.10880935e-01
4.37436253e-01 -8.77602398e-01 8.85602057e-01 8.66303027e-01
7.61114061e-02 9.06994343e-01 1.79517582e-01 -5.89385867e-01
-2.02704859e+00 -1.19162190e+00 3.86125565e-01 5.43807924e-01
7.66912818e-01 -4.49444920e-01 -8.82098675e-01 3.96961480e-01
3.92338634e-02 1.89150155e-01 4.99178112e-01 -2.31984764e-01
-7.36117244e-01 -4.46151644e-01 -8.34192574e-01 6.94845378e-01
1.13091016e+00 -5.79744101e-01 -3.85171361e-02 3.36454540e-01
1.59488702e+00 -8.37242067e-01 -6.44572079e-01 8.04583192e-01
4.79874760e-01 -7.06102192e-01 1.27098942e+00 -4.26882714e-01
3.79266262e-01 1.92618236e-01 -5.56926787e-01 -7.26020217e-01
-2.00559050e-01 -6.51878536e-01 -1.24792647e+00 9.25315082e-01
1.46600470e-01 -5.24069130e-01 1.25354517e+00 9.42212120e-02
-5.06565094e-01 -1.23929060e+00 -1.10009038e+00 -7.32599080e-01
-7.27371201e-02 -8.08454812e-01 8.91484082e-01 6.42790973e-01
-5.55701107e-02 1.74611270e-01 -4.87377942e-01 -4.33248729e-02
2.90812492e-01 -1.51384085e-01 4.08770919e-01 -5.77308118e-01
-8.18297267e-01 -8.40828896e-01 -3.98401856e-01 -1.72133040e+00
1.12487458e-01 -9.90098655e-01 -3.29428315e-02 -6.82542145e-01
5.31715006e-02 -6.05373919e-01 -4.85464394e-01 8.95972729e-01
-9.69525799e-02 4.74135071e-01 -2.23699724e-03 3.59845161e-02
-4.54582781e-01 5.47675550e-01 7.07942784e-01 -4.62185860e-01
3.89368348e-02 -3.03145885e-01 -7.42504716e-01 4.69944358e-01
6.67541206e-01 -2.73636341e-01 -3.81500781e-01 -6.51041567e-01
1.95699081e-01 -1.95952863e-01 9.63445902e-02 -1.20653117e+00
7.47628272e-01 1.33420855e-01 3.39483410e-01 -7.07348466e-01
5.80635130e-01 -6.82728350e-01 -1.06116958e-01 8.46910477e-01
-7.21500367e-02 2.14901298e-01 5.84957778e-01 4.14320886e-01
5.14029451e-02 -2.62321264e-01 6.96760893e-01 2.65772045e-01
-8.80548477e-01 5.25523484e-01 -3.84026133e-02 -2.47036874e-01
5.61398447e-01 -1.08619623e-01 -7.13201523e-01 -1.44405067e-01
-2.49267276e-02 3.39261219e-02 -8.56337249e-02 -1.84597902e-03
8.67815554e-01 -1.45981717e+00 -2.49288857e-01 5.88805556e-01
-3.81897599e-01 3.44440371e-01 4.37531054e-01 6.13320529e-01
-9.49634135e-01 8.44874084e-01 8.73947367e-02 -3.70059371e-01
-1.22093654e+00 5.86112440e-01 1.75321713e-01 -5.20431399e-01
-6.13264561e-01 1.35500240e+00 -1.80979565e-01 -4.13250586e-04
6.79669559e-01 -7.44667828e-01 2.64603347e-01 -4.89692986e-01
8.61280620e-01 5.26798546e-01 5.58712423e-01 -2.64939278e-01
-1.70189798e-01 8.49681273e-02 -3.92454505e-01 4.65733677e-01
1.10237169e+00 3.13292414e-01 -1.41597539e-01 -1.66115150e-01
1.62624931e+00 -4.11256254e-01 -8.87819707e-01 -4.40522403e-01
-3.13216567e-01 1.62211508e-02 6.73692584e-01 -4.71624225e-01
-1.53710210e+00 9.64026332e-01 6.88335538e-01 -6.83630109e-02
1.49858379e+00 -3.66328955e-01 1.59082329e+00 6.72197998e-01
9.14215669e-02 -1.00336599e+00 -2.84473971e-02 6.88425839e-01
1.32347330e-01 -9.35153961e-01 8.52021053e-02 -1.59241647e-01
7.55145103e-02 1.33229351e+00 7.39366472e-01 1.68717280e-02
7.60814130e-01 9.49164271e-01 -5.85484505e-01 4.17094007e-02
-6.95392847e-01 2.85679281e-01 3.64300489e-01 2.44883835e-01
-4.15593386e-02 1.25633433e-01 -1.11039288e-01 1.05801487e+00
-5.46101630e-01 -1.02920286e-01 6.91799223e-02 8.99916828e-01
-3.54340404e-01 -7.90611386e-01 -1.77065834e-01 7.04006016e-01
-6.15984857e-01 -9.31621015e-01 2.01445207e-01 4.39260334e-01
2.72010714e-02 3.53647411e-01 2.16779977e-01 -1.15324056e+00
1.86884142e-02 -1.45526361e-02 1.37612700e-01 -1.10260144e-01
-6.45403266e-01 -1.63574129e-01 -1.69878289e-01 -5.59310794e-01
3.31270486e-01 3.36929649e-01 -1.19766331e+00 -1.00441086e+00
-6.57728374e-01 -1.54149190e-01 1.01942444e+00 6.46993577e-01
5.75697839e-01 1.00488281e+00 1.94743037e-01 -1.02436447e+00
-1.01736689e+00 -8.22673261e-01 -4.04376179e-01 -2.61174709e-01
4.21734184e-01 -3.40652853e-01 -3.68177950e-01 -4.82283980e-01] | [8.577640533447266, 3.043286085128784] |
b301923b-0ef4-4e86-83c7-74f109169898 | pt2pc-learning-to-generate-3d-point-cloud | 2003.08624 | null | https://arxiv.org/abs/2003.08624v2 | https://arxiv.org/pdf/2003.08624v2.pdf | PT2PC: Learning to Generate 3D Point Cloud Shapes from Part Tree Conditions | 3D generative shape modeling is a fundamental research area in computer vision and interactive computer graphics, with many real-world applications. This paper investigates the novel problem of generating 3D shape point cloud geometry from a symbolic part tree representation. In order to learn such a conditional shape generation procedure in an end-to-end fashion, we propose a conditional GAN "part tree"-to-"point cloud" model (PT2PC) that disentangles the structural and geometric factors. The proposed model incorporates the part tree condition into the architecture design by passing messages top-down and bottom-up along the part tree hierarchy. Experimental results and user study demonstrate the strengths of our method in generating perceptually plausible and diverse 3D point clouds, given the part tree condition. We also propose a novel structural measure for evaluating if the generated shape point clouds satisfy the part tree conditions. | ['Xinchen Yan', 'Leonidas J. Guibas', 'Kaichun Mo', 'He Wang'] | 2020-03-19 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/32_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123510681.pdf | eccv-2020-8 | ['3d-shape-generation'] | ['computer-vision'] | [ 3.58784229e-01 2.80728430e-01 2.26375446e-01 -5.04994452e-01
-6.22490466e-01 -5.83523810e-01 8.79568398e-01 -1.57659069e-01
6.07641578e-01 2.93999940e-01 -5.63870370e-03 -3.37552577e-01
1.81267131e-02 -1.19322491e+00 -9.29170907e-01 -5.15241027e-01
2.02937737e-01 8.56290340e-01 2.10151598e-02 -5.62878177e-02
2.44330779e-01 9.38315749e-01 -1.81239748e+00 3.22941184e-01
9.79145944e-01 7.25018263e-01 3.03811759e-01 7.56047428e-01
-4.35654581e-01 1.15499772e-01 -3.98369819e-01 -4.64425504e-01
4.48940843e-01 -3.32036793e-01 -4.01126385e-01 5.91111958e-01
5.35800278e-01 -8.18052962e-02 4.77167547e-01 8.82050991e-01
5.54040849e-01 -1.59755737e-01 9.21938121e-01 -1.43304598e+00
-6.07709646e-01 3.30661774e-01 -7.43089855e-01 -5.22720993e-01
2.83057749e-01 1.44875154e-01 7.93087244e-01 -1.13815153e+00
5.25955737e-01 1.63628769e+00 4.68933195e-01 3.64457756e-01
-1.41964102e+00 -6.66687250e-01 3.35652888e-01 -3.77253026e-01
-1.55328512e+00 -1.91927829e-03 1.22162092e+00 -7.79567778e-01
5.95386624e-01 4.78664368e-01 7.41278470e-01 7.39383161e-01
1.48835987e-01 6.75957859e-01 9.63735759e-01 -3.29665720e-01
4.06255096e-01 1.23879155e-02 -3.36043894e-01 7.39787340e-01
2.76364833e-01 2.78909802e-01 -3.41397285e-01 -5.38894236e-01
1.30537212e+00 -1.16667993e-01 9.20745954e-02 -8.63604486e-01
-8.07927608e-01 8.43938589e-01 5.10779321e-01 1.59373041e-02
-4.19306606e-01 4.08418119e-01 -2.59955734e-01 1.93432067e-02
7.24436879e-01 1.56648047e-02 -3.34301174e-01 1.35889947e-01
-8.24921846e-01 5.02521873e-01 6.25650048e-01 1.71349156e+00
8.14829648e-01 3.38476986e-01 -2.69452929e-01 5.50909996e-01
6.76903009e-01 6.13583982e-01 -1.69515178e-01 -7.16448009e-01
1.51471063e-01 7.82004893e-01 3.65427174e-02 -9.15352106e-01
6.60540238e-02 -5.88678181e-01 -9.43621874e-01 6.28213346e-01
-3.20197850e-01 1.20416984e-01 -1.13833904e+00 1.53154802e+00
6.79612815e-01 4.54631090e-01 -2.62260318e-01 6.74226105e-01
7.69298911e-01 6.04259849e-01 6.37571886e-02 6.57266676e-02
1.23109710e+00 -5.62654436e-01 -1.58467278e-01 1.37882635e-01
2.35218436e-01 -8.95558059e-01 1.03274453e+00 3.00106019e-01
-1.24261761e+00 -8.41701269e-01 -8.50955427e-01 -2.97233686e-02
6.28645420e-02 1.84178427e-01 5.43423474e-01 8.88973594e-01
-1.12006319e+00 1.82618469e-01 -7.98801363e-01 1.28738638e-02
4.62648571e-01 2.03577787e-01 1.20053798e-01 -1.58362091e-02
-3.89896870e-01 4.51520592e-01 2.09010780e-01 -6.05047680e-03
-8.80028784e-01 -8.66473079e-01 -8.81817937e-01 7.76078701e-02
9.93100181e-02 -1.52821493e+00 1.27364123e+00 -5.33300340e-01
-1.41552782e+00 7.97282636e-01 -3.85084867e-01 -3.15390080e-02
6.45695567e-01 -1.74250901e-01 4.75440398e-02 -1.89368710e-01
-4.46728542e-02 7.45983243e-01 1.12876976e+00 -2.07878137e+00
-4.55361664e-01 -5.24653912e-01 -1.94671452e-01 3.25192600e-01
4.95526433e-01 -4.26175147e-01 -5.54240823e-01 -7.66846240e-01
4.66361821e-01 -1.00230992e+00 -5.09658277e-01 1.70084476e-01
-6.25669777e-01 -2.70429492e-01 9.73788977e-01 -2.77965575e-01
6.28588617e-01 -1.98786116e+00 5.42181954e-02 7.64973998e-01
1.03551082e-01 -1.28181607e-01 5.32265343e-02 4.58279908e-01
-1.76404174e-02 3.00516814e-01 -4.22105640e-01 -5.01371384e-01
9.31846425e-02 2.42377713e-01 -5.10083258e-01 1.47086047e-02
4.11459923e-01 9.49684739e-01 -7.67928660e-01 -2.93796122e-01
4.13316280e-01 7.37254977e-01 -7.57481933e-01 2.90189177e-01
-6.48181856e-01 6.11893773e-01 -7.13792026e-01 5.94272971e-01
1.01508021e+00 -1.67079195e-01 -1.37910992e-01 -1.08121872e-01
-1.31451666e-01 -1.90643549e-01 -1.09064138e+00 1.71621466e+00
-4.90489960e-01 6.56028837e-02 -1.12868942e-01 -2.56700158e-01
1.39197040e+00 2.14922577e-01 2.55076468e-01 -2.46910349e-01
1.05338730e-01 8.93573239e-02 -2.95598805e-01 -6.17266893e-02
5.75700402e-01 -2.97538787e-01 -7.95006081e-02 1.37084991e-01
-3.34957600e-01 -8.47824275e-01 -3.39077592e-01 2.50229597e-01
5.38464248e-01 6.70333266e-01 1.26320124e-01 -1.61056280e-01
4.12555486e-01 -1.27107561e-01 2.47782633e-01 5.13452053e-01
5.63360035e-01 1.06831336e+00 3.00489277e-01 -2.43753716e-01
-1.19537675e+00 -1.30894780e+00 8.52525383e-02 6.45945013e-01
2.06962302e-01 -4.09768343e-01 -7.25535512e-01 -4.03780013e-01
1.74145445e-01 1.18936038e+00 -6.09082580e-01 -6.47304282e-02
-5.18099904e-01 -1.85633942e-01 2.75991142e-01 5.67419648e-01
2.61568904e-01 -8.33206534e-01 -6.51492298e-01 3.11732292e-04
1.68020830e-01 -9.37878788e-01 -3.81529659e-01 -3.27056974e-01
-1.12079275e+00 -6.91449881e-01 -5.69745779e-01 -7.78305411e-01
1.10249555e+00 3.31706345e-01 1.33429289e+00 2.43210196e-01
-1.97699711e-01 5.88417113e-01 -3.17533135e-01 -6.21725619e-01
-7.52885818e-01 -2.02534020e-01 -3.47405195e-01 1.79143921e-01
-1.26903772e-01 -9.04620588e-01 -5.49649179e-01 2.60416895e-01
-9.96627271e-01 7.34710932e-01 6.58251107e-01 3.85406852e-01
1.26946831e+00 6.00347184e-02 2.71110535e-01 -9.17476356e-01
5.17991245e-01 -2.73736894e-01 -7.61018395e-01 8.78617391e-02
-5.56985974e-01 6.08103313e-02 2.94993281e-01 -3.89638953e-02
-1.09827137e+00 3.56039792e-01 -2.67608136e-01 -8.97272348e-01
-3.41375023e-01 3.45175266e-01 -6.38533473e-01 6.19713478e-02
3.46759707e-01 4.72838372e-01 -2.57055730e-01 -5.96088707e-01
7.25557148e-01 1.34810925e-01 5.76405704e-01 -9.96618092e-01
1.28941500e+00 5.25475323e-01 3.57304811e-01 -8.56623471e-01
-2.56896853e-01 -2.44090006e-01 -7.34360099e-01 -2.65612751e-01
9.02714789e-01 -8.56790781e-01 -4.81439918e-01 2.80153424e-01
-1.50873280e+00 1.08404569e-01 -3.10978353e-01 -1.45245373e-01
-8.68252039e-01 2.03930452e-01 5.87301478e-02 -1.13811159e+00
-3.59221607e-01 -1.00931501e+00 1.70595658e+00 -8.52238480e-03
-4.98236157e-02 -7.49326706e-01 -6.75432160e-02 1.30147845e-01
1.35627270e-01 8.10253084e-01 1.36050427e+00 -1.83713719e-01
-1.11116779e+00 -9.59549248e-02 -2.39991218e-01 8.11984017e-02
1.85093269e-01 2.20198259e-01 -9.53642249e-01 -1.71136603e-01
-3.06809731e-02 2.73114771e-01 3.22071791e-01 3.78247201e-01
1.29892135e+00 -2.67396182e-01 -3.37162197e-01 7.15553045e-01
1.60584438e+00 3.00752819e-01 6.43248200e-01 -4.22281414e-01
1.10326874e+00 4.87248749e-01 2.98482388e-01 5.96550643e-01
4.78145301e-01 6.42839313e-01 6.49533212e-01 -6.65458217e-02
-3.66217673e-01 -9.44565773e-01 -1.21140502e-01 8.36302876e-01
-1.02762789e-01 -2.68796861e-01 -8.05726111e-01 3.69174182e-01
-1.62804246e+00 -8.05771649e-01 -4.56104100e-01 2.20377588e+00
4.85671818e-01 9.03608054e-02 -1.31203339e-01 3.97620350e-02
7.53074646e-01 -1.82732835e-01 -4.20830727e-01 -4.00788635e-01
9.32162553e-02 3.49999875e-01 4.77846116e-02 4.90315050e-01
-5.85395575e-01 8.99878144e-01 5.69041300e+00 9.03961003e-01
-1.00740910e+00 -2.62865692e-01 6.22664809e-01 3.95247459e-01
-1.02272511e+00 2.87341833e-01 -7.70195246e-01 2.44624913e-01
2.56804168e-01 -1.89817622e-01 2.35328190e-02 9.84654725e-01
2.53616452e-01 1.62840366e-01 -1.24783552e+00 1.06434035e+00
-1.04555693e-02 -1.44867086e+00 7.07010567e-01 2.96355695e-01
8.88966918e-01 -4.48351413e-01 1.56659395e-01 3.15037854e-02
5.27050316e-01 -1.04362822e+00 1.15618622e+00 5.77263117e-01
9.68017220e-01 -9.43396747e-01 3.63441147e-02 7.37481296e-01
-1.40618837e+00 6.23977482e-01 -1.52576983e-01 2.56701559e-01
4.44545180e-01 5.33139169e-01 -1.29795980e+00 8.99283051e-01
3.19429487e-01 2.21859664e-01 -5.20455360e-01 1.16847932e+00
-3.95672500e-01 4.73336220e-01 -2.28496000e-01 5.82746752e-02
9.07607451e-02 -4.14122760e-01 9.14825261e-01 1.00967264e+00
6.79843843e-01 1.99402243e-01 2.50246763e-01 1.48977268e+00
-4.46518278e-03 -6.68948293e-02 -8.60934913e-01 2.69070774e-01
6.16795123e-01 9.37696755e-01 -9.28806961e-01 -1.66395023e-01
1.02395050e-01 9.02361393e-01 -4.76271398e-02 2.25043163e-01
-7.05762506e-01 -3.64190564e-02 3.19563240e-01 4.03170496e-01
6.51724815e-01 -6.46678865e-01 -8.68208826e-01 -6.65517449e-01
8.73880740e-03 -4.20200229e-01 -1.47937760e-01 -1.10571480e+00
-1.24070728e+00 6.35264218e-01 3.19560230e-01 -1.49995339e+00
-4.60355580e-01 -2.09923834e-01 -7.08835721e-01 1.02511489e+00
-1.25723398e+00 -1.81797230e+00 -4.58383024e-01 6.00966990e-01
6.63474321e-01 6.60320297e-02 8.97129595e-01 -1.08200930e-01
6.27684668e-02 3.57996345e-01 -3.39741319e-01 -3.05901915e-01
-1.27972335e-01 -1.21720469e+00 9.96765971e-01 8.36916566e-01
4.08493996e-01 7.07911015e-01 7.45717227e-01 -1.00035095e+00
-1.34648883e+00 -1.37356365e+00 5.39267659e-01 -6.01696074e-01
-2.24276513e-01 -6.36708438e-01 -7.56555617e-01 4.70868140e-01
4.60373051e-02 -4.07805115e-01 6.07375562e-01 -9.04756859e-02
-4.79562789e-01 2.71272510e-01 -1.41900158e+00 6.65054858e-01
1.31536973e+00 -1.60742402e-01 -3.88763040e-01 -7.09486678e-02
1.00446236e+00 -5.23246765e-01 -5.78212380e-01 6.62797749e-01
3.57300520e-01 -9.81243014e-01 1.08861887e+00 -3.35712671e-01
4.94231582e-01 -6.58163965e-01 -2.72221178e-01 -1.32501578e+00
-4.96052355e-01 -7.36152112e-01 -1.97088689e-01 1.15591896e+00
2.86035419e-01 -1.72221869e-01 1.12623036e+00 6.20582819e-01
-5.36231041e-01 -8.43718350e-01 -8.69789183e-01 -7.09238648e-01
8.17354321e-02 -7.14138269e-01 1.15114701e+00 6.89331055e-01
-1.00703537e+00 2.80657142e-01 -1.46676809e-01 4.15342748e-01
7.74968207e-01 6.30265892e-01 1.12051046e+00 -1.52018785e+00
-3.71072441e-01 -5.65572083e-01 -5.00478625e-01 -1.24443090e+00
-2.28270069e-01 -9.39304650e-01 7.59896240e-04 -1.81267524e+00
-9.81455520e-02 -8.54469359e-01 2.74540365e-01 7.26006478e-02
-3.90741788e-02 4.86257002e-02 3.17644924e-01 4.50280234e-02
1.79836020e-01 8.74197781e-01 1.56412375e+00 -1.19808435e-01
-2.54951358e-01 3.30483049e-01 -8.34680259e-01 8.17336023e-01
5.25163949e-01 -3.83812219e-01 -9.65346277e-01 -6.69595420e-01
8.23010206e-02 1.83902010e-01 6.80414140e-01 -8.50872934e-01
7.12553710e-02 -3.83723319e-01 4.13758785e-01 -1.34384429e+00
6.65289223e-01 -1.09715044e+00 6.34975970e-01 2.44657233e-01
7.18290859e-04 2.46955276e-01 1.41250953e-01 6.95952654e-01
1.43124163e-01 1.31363247e-03 6.63104296e-01 -6.01158366e-02
-3.58100027e-01 5.78726113e-01 1.55795559e-01 -2.17689574e-01
1.09838879e+00 -6.58532321e-01 1.08736902e-01 -3.31975788e-01
-5.69522321e-01 4.86294702e-02 5.71980596e-01 5.97009718e-01
1.13578749e+00 -1.61964679e+00 -9.60700750e-01 6.74087942e-01
2.19324008e-01 6.08298123e-01 2.06782848e-01 2.99163405e-02
-6.26779079e-01 3.02997082e-02 -1.35576050e-03 -1.02445090e+00
-1.30370712e+00 3.50483239e-01 -3.31195034e-02 8.83907601e-02
-6.09273314e-01 9.99792218e-01 6.64463580e-01 -5.95364630e-01
-5.08328862e-02 -5.20086288e-01 2.03216374e-01 -4.91994202e-01
-9.40677896e-03 8.92294571e-02 1.06894843e-01 -7.13660538e-01
3.41573507e-02 8.18412423e-01 2.96647072e-01 -3.67349267e-01
1.16788411e+00 1.02890424e-01 6.23959303e-02 2.99226791e-01
8.12463641e-01 2.23116130e-01 -1.09516585e+00 -6.14823066e-02
-2.17138737e-01 -8.35003912e-01 -2.28220865e-01 -7.26761043e-01
-8.68126273e-01 7.92321622e-01 3.98359150e-01 1.69324532e-01
9.00259793e-01 1.43911079e-01 6.19784057e-01 -1.73076354e-02
8.61962914e-01 -4.47237641e-01 -1.14960641e-01 4.28232014e-01
1.32433641e+00 -6.12547815e-01 -3.26093882e-02 -1.04331660e+00
-6.10717535e-01 7.56086111e-01 6.49438322e-01 -2.83220857e-01
7.67449737e-01 2.36159340e-01 -2.52574563e-01 -4.01045650e-01
-8.23383808e-01 -9.76881012e-03 6.34613156e-01 9.21436787e-01
2.72352189e-01 3.96886319e-01 -5.63779697e-02 4.88395959e-01
-7.36651123e-01 -1.15133412e-01 1.73644572e-01 9.17897701e-01
-3.87563437e-01 -1.32261372e+00 -3.46211553e-01 1.52658820e-01
2.27095231e-01 2.62804013e-02 -5.74411094e-01 5.67625523e-01
3.42446148e-01 5.61276555e-01 1.43980309e-01 -5.05220115e-01
4.40150350e-01 -1.08768570e-03 6.13674641e-01 -7.57307947e-01
-4.81168419e-01 3.11184645e-01 -1.25913128e-01 -1.24873064e-01
-1.62322104e-01 -6.04940116e-01 -1.25433779e+00 -1.83028102e-01
-2.52043724e-01 3.81062850e-02 9.24797773e-01 5.06207407e-01
7.94042528e-01 4.46835428e-01 9.17845488e-01 -1.18176889e+00
-3.31660807e-01 -6.09353721e-01 -2.57205904e-01 4.20768917e-01
3.20765488e-02 -6.19643450e-01 1.48889467e-01 2.92529732e-01] | [8.813520431518555, -3.6378421783447266] |
947f25de-3dbf-4266-8489-f5dd346538dc | weather2k-a-multivariate-spatio-temporal | 2302.10493 | null | https://arxiv.org/abs/2302.10493v1 | https://arxiv.org/pdf/2302.10493v1.pdf | Weather2K: A Multivariate Spatio-Temporal Benchmark Dataset for Meteorological Forecasting Based on Real-Time Observation Data from Ground Weather Stations | Weather forecasting is one of the cornerstones of meteorological work. In this paper, we present a new benchmark dataset named Weather2K, which aims to make up for the deficiencies of existing weather forecasting datasets in terms of real-time, reliability, and diversity, as well as the key bottleneck of data quality. To be specific, our Weather2K is featured from the following aspects: 1) Reliable and real-time data. The data is hourly collected from 2,130 ground weather stations covering an area of 6 million square kilo- meters. 2) Multivariate meteorological variables. 20 meteorological factors and 3 constants for position information are provided with a length of 40,896 time steps. 3) Applicable to diverse tasks. We conduct a set of baseline tests on time series forecasting and spatio-temporal forecasting. To the best of our knowledge, our Weather2K is the first attempt to tackle weather forecasting task by taking full advantage of the strengths of observation data from ground weather stations. Based on Weather2K, we further propose Meteorological Factors based Multi-Graph Convolution Network (MFMGCN), which can effectively construct the intrinsic correlation among geographic locations based on meteorological factors. Sufficient experiments show that MFMGCN improves both the forecasting performance and temporal robustness. We hope our Weather2K can significantly motivate researchers to develop efficient and accurate algorithms to advance the task of weather forecasting. The dataset can be available at https://github.com/bycnfz/weather2k/. | ['Ziheng Yang', 'Bin Zhang', 'Gaozhen Nie', 'Ming Wu', 'Yutong Xiong', 'Xun Zhu'] | 2023-02-21 | null | null | null | null | ['weather-forecasting', 'spatio-temporal-forecasting'] | ['miscellaneous', 'time-series'] | [-6.57933891e-01 -4.28214163e-01 6.29725084e-02 -5.56491613e-01
6.73401728e-02 -6.13094270e-01 6.40425622e-01 1.07311174e-01
-1.14767395e-01 7.64815450e-01 3.92525107e-01 -6.89736664e-01
-3.97868663e-01 -1.36403918e+00 -4.54968959e-01 -8.20401251e-01
-7.08264709e-01 -3.28917392e-02 -1.11928936e-02 -7.54856586e-01
-2.97150165e-02 5.77218056e-01 -1.39855707e+00 -2.56948054e-01
1.02224779e+00 1.03492486e+00 1.12355843e-01 8.29796374e-01
-6.40087351e-02 4.25203532e-01 -4.96114463e-01 -2.38904864e-01
2.74638921e-01 -2.55355477e-01 -4.51984733e-01 -4.00105268e-01
2.59816885e-01 -2.33689576e-01 -6.13591135e-01 7.54845262e-01
6.03175402e-01 5.99969268e-01 3.91207963e-01 -1.50051880e+00
-7.25223839e-01 5.47074437e-01 -2.87844032e-01 6.06518984e-01
-1.14986256e-01 1.73152596e-01 9.31743085e-01 -7.28826523e-01
1.21459186e-01 9.11506653e-01 8.42694521e-01 3.43994647e-02
-6.68104768e-01 -7.52785981e-01 5.64364195e-01 8.44406784e-02
-1.41019154e+00 -2.70874858e-01 4.67762917e-01 -4.75066274e-01
9.62131202e-01 5.31970680e-01 4.94254380e-01 7.95591116e-01
4.19375062e-01 5.93098365e-02 1.02495563e+00 -1.60136987e-02
9.03410167e-02 -2.67830938e-01 5.02352476e-01 4.95596051e-01
2.95207351e-01 4.11469996e-01 -4.10451531e-01 -2.44037569e-01
4.93881285e-01 3.19573373e-01 -6.15981519e-01 6.10031784e-01
-1.26278937e+00 8.17817986e-01 7.19604313e-01 4.48684394e-01
-3.16637069e-01 1.62248343e-01 1.66156545e-01 4.72012877e-01
1.04420364e+00 3.86434376e-01 -6.20660901e-01 -2.33924435e-03
-8.20306838e-01 2.89510220e-01 8.12356114e-01 7.24281669e-01
7.44194865e-01 5.07570148e-01 6.99813291e-02 4.89934951e-01
1.92438543e-01 1.19413817e+00 3.72644395e-01 -4.27968353e-01
6.70051634e-01 4.64042842e-01 1.00345761e-01 -1.66878903e+00
-1.01501679e+00 -4.04756337e-01 -1.47614515e+00 -1.19487435e-01
2.18003720e-01 -8.17895532e-01 -6.85060918e-01 1.45980287e+00
2.87402391e-01 7.25638330e-01 -3.73837017e-02 1.12871313e+00
1.21439946e+00 1.28943896e+00 -2.47615382e-01 -7.53270015e-02
1.13282681e+00 -8.49181056e-01 -7.93398499e-01 2.20335081e-01
6.66504323e-01 -6.14596546e-01 8.34891319e-01 6.19585365e-02
-4.22594965e-01 -6.33030176e-01 -6.65958166e-01 4.65000391e-01
-1.02175248e+00 1.19208425e-01 9.13333356e-01 3.72635752e-01
-1.28920782e+00 4.99928445e-01 -6.53417587e-01 -2.73030698e-01
-2.34362543e-01 -1.65871337e-01 -3.17681998e-01 6.93511637e-03
-1.93555033e+00 7.97548115e-01 2.89876968e-01 5.22538364e-01
-5.13482630e-01 -8.62109184e-01 -8.82938981e-01 1.44654021e-01
1.77448109e-01 -5.37846923e-01 7.89821208e-01 -2.80516386e-01
-9.92614567e-01 -7.86407739e-02 -1.10265628e-01 -2.44218871e-01
2.89119091e-02 -2.04466954e-02 -1.21434700e+00 -2.50283718e-01
-1.56173512e-01 -1.49553344e-01 4.71611857e-01 -8.89224291e-01
-6.51105821e-01 -1.72033772e-01 2.65979040e-02 -3.26786228e-02
-5.92224836e-01 -1.82994604e-01 -3.27447385e-01 -9.77369368e-01
-4.28348482e-01 -8.49185526e-01 -4.73076105e-01 -4.70040858e-01
-4.29345816e-01 -1.40680611e-01 5.87710857e-01 -9.16243732e-01
1.68148613e+00 -1.98087573e+00 -1.90686762e-01 4.42381799e-01
3.73193145e-01 3.99067372e-01 -3.61037493e-01 9.28232849e-01
-1.04099408e-01 2.64802933e-01 -2.59629399e-01 -2.82447100e-01
-5.29285744e-02 3.34930658e-01 -8.54641080e-01 5.45946419e-01
6.65009767e-02 1.09435177e+00 -7.55617917e-01 1.25375614e-01
2.25687474e-01 5.85647047e-01 6.34066686e-02 2.34555855e-01
-8.73371139e-02 4.16204900e-01 -5.84258020e-01 4.23987478e-01
8.99849832e-01 -3.04435194e-01 -2.93092310e-01 2.45285332e-02
-3.92651081e-01 2.26606369e-01 -1.32836998e+00 1.27210176e+00
-3.96470457e-01 4.92260456e-01 -1.95926785e-01 -7.42011786e-01
1.11241210e+00 4.32642490e-01 2.99582064e-01 -7.24725842e-01
-1.58492923e-01 -6.99574202e-02 -2.77301937e-01 -4.51642334e-01
8.00661862e-01 3.82161915e-01 -4.08059098e-02 4.52777296e-01
-3.44259739e-01 -1.45941034e-01 2.75007546e-01 3.05672497e-01
6.27230704e-01 -3.06379467e-01 1.34812459e-01 -3.60017031e-01
9.03645679e-02 7.97369555e-02 5.58248639e-01 6.06165349e-01
2.38526203e-02 4.14853454e-01 2.98029393e-01 -1.08793044e+00
-7.99920142e-01 -4.90894496e-01 -8.89901891e-02 9.50663209e-01
-9.11441669e-02 -4.66218948e-01 -1.15611941e-01 -3.38546962e-01
3.35509181e-01 6.85957730e-01 -8.55380416e-01 2.81196743e-01
-3.35809439e-01 -1.30741382e+00 8.25782597e-01 3.32626492e-01
4.35198665e-01 -7.09295988e-01 -9.74457115e-02 2.13765085e-01
-2.83138305e-01 -9.64701474e-01 -3.89905572e-01 -1.07681528e-01
-7.40776300e-01 -8.91332924e-01 -6.84708059e-01 -5.48651889e-02
4.57327127e-01 7.62943029e-01 1.35918343e+00 4.13533151e-01
-1.17444722e-02 1.31065175e-01 -5.49395740e-01 -3.83211195e-01
1.40749052e-01 3.54467005e-01 2.93143541e-01 4.07078415e-02
-1.26237432e-02 -8.78709733e-01 -6.11476779e-01 4.94991928e-01
-1.00696051e+00 -5.03876917e-02 1.53579712e-01 4.69516098e-01
1.47179976e-01 4.55647290e-01 6.45866215e-01 -5.93603253e-01
6.91252470e-01 -1.05802155e+00 -9.98582304e-01 2.91523129e-01
-8.26147676e-01 -3.10449779e-01 8.77413273e-01 -5.53732887e-02
-7.20380068e-01 -2.17791378e-01 -1.00225406e-02 -6.51306137e-02
-3.76486450e-01 1.24241364e+00 4.21276659e-01 -8.70213360e-02
5.83144188e-01 3.83052289e-01 -4.11202908e-01 -5.85857034e-01
5.63575447e-01 5.27550817e-01 4.65496033e-01 -3.75642508e-01
1.08889222e+00 3.16856295e-01 1.83139011e-01 -8.09038699e-01
-7.89332449e-01 -6.30595744e-01 -2.70286649e-01 -4.22356963e-01
5.25286078e-01 -1.15247548e+00 -7.71034658e-01 7.45952189e-01
-8.37275028e-01 -6.00737333e-01 3.81832749e-01 6.22938097e-01
3.92088801e-01 3.57461274e-01 -5.33142209e-01 -8.51064205e-01
-4.75200355e-01 -6.85426831e-01 6.34117723e-01 1.70594558e-01
3.16675276e-01 -1.43684423e+00 4.03366655e-01 -1.81454912e-01
1.00166678e+00 5.65970123e-01 4.40072119e-01 -2.99997360e-01
-2.34904706e-01 -1.86347872e-01 -3.06522220e-01 7.86864012e-02
2.76113510e-01 3.37140620e-01 -9.68450129e-01 -3.88064474e-01
-4.77639228e-01 -5.76815754e-02 1.15555167e+00 3.68644446e-01
1.04730642e+00 -4.84561920e-01 -1.97877541e-01 8.79893601e-01
1.37521780e+00 -2.03458816e-02 4.61128116e-01 1.85823292e-01
9.59099650e-01 6.40463412e-01 3.77633989e-01 6.53097689e-01
9.45000291e-01 3.62085372e-01 6.13053024e-01 -4.47422624e-01
2.57564157e-01 1.45803303e-01 1.28262773e-01 1.33084357e+00
-5.82039773e-01 -7.88870335e-01 -1.40088737e+00 5.86090863e-01
-1.94322956e+00 -9.70969558e-01 -5.89896023e-01 1.96487939e+00
4.24781471e-01 -2.43989244e-01 2.60425340e-02 -1.02628740e-02
5.22147775e-01 6.80116177e-01 -3.31438273e-01 -1.17309824e-01
-3.60146701e-01 -1.57893568e-01 9.38417315e-01 7.16837704e-01
-1.41433704e+00 6.59828782e-01 5.70082426e+00 6.94192410e-01
-1.37201262e+00 -1.15999140e-01 6.73392177e-01 -4.19945605e-02
-4.80950415e-01 -2.27311835e-01 -1.04422092e+00 7.58794665e-01
1.28709197e+00 -1.42497435e-01 6.80132747e-01 6.53177559e-01
5.72320640e-01 1.87102690e-01 -3.22010815e-01 5.75237572e-01
-2.24423572e-01 -1.54408073e+00 -1.30979419e-01 -5.21650305e-03
9.46240664e-01 6.02404892e-01 2.03033611e-02 1.27003685e-01
5.17885327e-01 -1.23110557e+00 1.31744385e-01 9.53907967e-01
7.31971323e-01 -7.40898728e-01 8.87211740e-01 3.46743017e-01
-1.67864311e+00 1.44193456e-01 -4.36663419e-01 -5.00353396e-01
2.47442558e-01 1.37415564e+00 -4.28049505e-01 1.30701733e+00
1.09972358e+00 1.07712340e+00 -3.11176628e-01 1.00881445e+00
-1.38221383e-01 1.13322353e+00 -4.71539795e-01 -5.57503514e-02
5.07055104e-01 -4.08669680e-01 3.46745163e-01 1.33175099e+00
6.69202566e-01 4.40883160e-01 3.20972294e-01 2.16717809e-01
1.88391700e-01 1.98530868e-01 -7.52576888e-01 -7.64171258e-02
5.81897974e-01 1.47722316e+00 -5.24070740e-01 -4.47990745e-01
-4.85299319e-01 4.26384062e-01 2.52214581e-01 6.74890578e-01
-7.76576281e-01 -5.79145372e-01 9.98142719e-01 -3.35792303e-01
2.36767426e-01 -6.90868616e-01 -1.99028909e-01 -1.39560258e+00
5.24889454e-02 -5.59630871e-01 5.08819461e-01 -8.19798887e-01
-1.49517190e+00 9.28032875e-01 -1.11785263e-01 -1.19110453e+00
-3.15775089e-02 -4.15054768e-01 -9.91538107e-01 1.38007057e+00
-2.20655704e+00 -1.12005210e+00 -8.04969072e-01 6.72117710e-01
-4.14802926e-03 -9.73860398e-02 1.05381954e+00 4.35585797e-01
-7.54433393e-01 2.00135335e-01 4.61427569e-01 2.61250585e-01
6.62675977e-01 -1.14397490e+00 1.34568572e+00 1.11936700e+00
-7.39734769e-02 5.59853077e-01 4.52420741e-01 -6.35913908e-01
-1.34241128e+00 -1.62968647e+00 1.38971794e+00 -3.06901425e-01
9.05224919e-01 -2.43441626e-01 -1.07428777e+00 4.67380106e-01
8.91351700e-02 2.12399036e-01 5.78776479e-01 2.59883851e-01
-4.10875350e-01 -3.14700931e-01 -5.99695921e-01 4.37399745e-01
7.24926353e-01 -1.09884322e-01 -2.05437988e-01 6.89727604e-01
1.04480577e+00 -5.12429237e-01 -1.19398093e+00 2.81249374e-01
1.85050920e-01 -7.00260222e-01 6.92851782e-01 -5.89908302e-01
3.55925798e-01 -5.33044815e-01 -5.33145107e-02 -1.97258818e+00
-6.53162003e-01 -4.89688903e-01 -2.58836132e-02 1.15776300e+00
5.87185442e-01 -1.26358998e+00 2.50552781e-02 4.94469404e-01
-2.96982676e-01 -6.37858212e-01 -7.75174499e-01 -7.72661746e-01
4.67471257e-02 -7.02691376e-01 1.31592119e+00 1.50325263e+00
-2.90618092e-01 -1.08423874e-01 -9.68639612e-01 8.47666979e-01
1.48219898e-01 6.05251431e-01 7.75413632e-01 -1.30385303e+00
-1.90611929e-01 -4.04197812e-01 -9.79274809e-02 -8.64924967e-01
-1.08975373e-01 -6.59321606e-01 -2.93303281e-01 -1.57539499e+00
-4.64103162e-01 -5.92264056e-01 -4.21695888e-01 9.43988144e-01
-3.48625779e-01 2.70284325e-01 -4.25280184e-02 2.56454855e-01
-2.53659755e-01 5.85017085e-01 1.24160278e+00 1.00420401e-01
-1.13239653e-01 9.11631733e-02 -4.42059040e-01 1.83981404e-01
1.18536568e+00 -3.25233608e-01 -3.58381540e-01 -6.32880330e-01
6.71987772e-01 1.73733652e-01 2.73284942e-01 -7.01651275e-01
2.93838888e-01 -7.12273419e-01 2.23838732e-01 -6.38727725e-01
1.32380322e-01 -6.31270647e-01 3.68708372e-01 3.28448147e-01
7.63846040e-02 6.56522274e-01 3.39583844e-01 4.35369760e-01
-2.92174667e-01 4.49686527e-01 -2.18648016e-02 5.07938042e-02
-1.05479062e+00 8.09687376e-01 -3.44229847e-01 -5.14097512e-03
7.64655292e-01 4.36377764e-01 -9.37482595e-01 -6.81864440e-01
-3.57020944e-01 7.65751064e-01 7.03844205e-02 5.14681399e-01
3.67426127e-01 -1.27418458e+00 -1.12255704e+00 1.62021413e-01
2.34221205e-01 -1.36863440e-01 5.49694419e-01 6.96138263e-01
-4.65806007e-01 6.40992403e-01 1.18478246e-01 -1.60540715e-01
-7.97029376e-01 5.65250933e-01 2.76388496e-01 -1.12583764e-01
-7.20073581e-01 6.04282439e-01 -2.26989493e-01 -8.44965994e-01
2.55073328e-02 -7.75155067e-01 -4.51191992e-01 2.00535506e-01
8.13126147e-01 4.17920083e-01 2.42140725e-01 -6.62538052e-01
-3.76794130e-01 5.21166384e-01 6.07259274e-01 1.20148629e-01
1.63097095e+00 -2.88509518e-01 -2.60314107e-01 5.52659750e-01
8.61794770e-01 -4.92108464e-02 -1.00311708e+00 -4.10520047e-01
-3.32454562e-01 -5.07015049e-01 2.64373273e-01 -8.44255686e-01
-1.58535433e+00 7.17295647e-01 2.81819999e-01 7.29210615e-01
1.22735703e+00 -5.04192173e-01 1.07202649e+00 4.93259817e-01
2.25488003e-02 -6.28835738e-01 -6.32969499e-01 8.71149659e-01
9.34246004e-01 -1.31663418e+00 -6.21156394e-02 -1.46763742e-01
-5.71606755e-01 1.19535875e+00 3.14100236e-01 -8.86615217e-02
1.13221133e+00 2.15241209e-01 5.60878694e-01 -3.62403542e-01
-7.86950290e-01 -3.78151923e-01 6.91834867e-01 3.34491760e-01
3.91554028e-01 5.82993746e-01 -1.09988116e-02 4.97619689e-01
-3.67652208e-01 -1.97191507e-01 3.64278466e-01 5.33630490e-01
-3.12490642e-01 -6.33601367e-01 -4.80982453e-01 4.91748512e-01
-1.50421575e-01 -5.37385762e-01 7.63090253e-02 4.91259277e-01
-1.32888287e-01 1.19663560e+00 -1.38976678e-01 -6.04153812e-01
2.89245963e-01 -1.12763204e-01 -6.47169590e-01 -1.94917675e-02
-5.75275838e-01 -4.50194001e-01 4.19392213e-02 -6.26810312e-01
-4.74471301e-01 -2.66238272e-01 -8.37796152e-01 -9.81861472e-01
-1.07661091e-01 4.53070670e-01 8.32570434e-01 8.64091635e-01
7.02561498e-01 6.72816992e-01 9.91420567e-01 -9.30057287e-01
-2.62889862e-01 -9.61365640e-01 -7.92699575e-01 1.48879528e-01
7.03951299e-01 -4.77938086e-01 -5.25486708e-01 -2.61374861e-01] | [6.652734279632568, 2.763509511947632] |
496107a5-212f-4cda-a0b5-a1878e9b27fb | graph-mining-for-cybersecurity-a-survey | 2304.00485 | null | https://arxiv.org/abs/2304.00485v1 | https://arxiv.org/pdf/2304.00485v1.pdf | Graph Mining for Cybersecurity: A Survey | The explosive growth of cyber attacks nowadays, such as malware, spam, and intrusions, caused severe consequences on society. Securing cyberspace has become an utmost concern for organizations and governments. Traditional Machine Learning (ML) based methods are extensively used in detecting cyber threats, but they hardly model the correlations between real-world cyber entities. In recent years, with the proliferation of graph mining techniques, many researchers investigated these techniques for capturing correlations between cyber entities and achieving high performance. It is imperative to summarize existing graph-based cybersecurity solutions to provide a guide for future studies. Therefore, as a key contribution of this paper, we provide a comprehensive review of graph mining for cybersecurity, including an overview of cybersecurity tasks, the typical graph mining techniques, and the general process of applying them to cybersecurity, as well as various solutions for different cybersecurity tasks. For each task, we probe into relevant methods and highlight the graph types, graph approaches, and task levels in their modeling. Furthermore, we collect open datasets and toolkits for graph-based cybersecurity. Finally, we outlook the potential directions of this field for future research. | ['Junping Du', 'Yanfang Ye', 'Qi Li', 'Yong Fang', 'Chuan Shi', 'Cheng Yang', 'Bo Yan'] | 2023-04-02 | null | null | null | null | ['graph-mining'] | ['graphs'] | [ 7.30424747e-02 -3.69006582e-02 -3.79997939e-01 2.60378987e-01
1.60160735e-01 -1.07416415e+00 6.72817409e-01 9.00786400e-01
5.80848530e-02 3.10393393e-01 -3.33701819e-01 -1.04999948e+00
-4.13615167e-01 -1.20439851e+00 -8.91493782e-02 -1.59972414e-01
-6.84360385e-01 1.68347135e-01 3.67500305e-01 -4.32942122e-01
5.00782430e-01 8.19811165e-01 -8.25558186e-01 8.95616487e-02
7.45824158e-01 8.80491793e-01 -5.97390115e-01 7.44944394e-01
1.38135955e-01 7.67360091e-01 -8.60925496e-01 -1.04297686e+00
3.51987630e-01 -1.19078770e-01 -7.80144036e-01 4.70314585e-02
3.68248075e-01 1.85964778e-01 -7.28639483e-01 1.43098438e+00
-1.21608503e-01 1.13783121e-01 4.91691113e-01 -2.01711249e+00
-4.63738412e-01 6.04524553e-01 -7.31097162e-01 5.21775365e-01
4.00273234e-01 7.59151205e-02 1.12524688e+00 -2.58876413e-01
7.60223031e-01 1.02166593e+00 5.46934247e-01 4.47827429e-02
-6.62339687e-01 -7.54663944e-01 5.09047806e-01 5.11461854e-01
-1.04800546e+00 2.01705843e-01 8.52019489e-01 -2.59836674e-01
1.48849797e+00 4.72994000e-01 7.11068153e-01 8.86136174e-01
9.43967760e-01 2.07613677e-01 8.72311771e-01 -3.56365889e-01
7.38339648e-02 -7.12949261e-02 7.12661624e-01 8.66421938e-01
1.09514141e+00 1.21788241e-01 -2.55239336e-03 -6.95432663e-01
3.45865935e-01 1.83599710e-01 -7.83314481e-02 -3.96267086e-01
-9.78006423e-01 9.59621489e-01 5.91705561e-01 5.89858472e-01
8.70050862e-02 -2.32109904e-01 7.34109581e-01 6.12410903e-01
6.11580610e-01 7.45362520e-01 -4.39409286e-01 2.10659251e-01
-1.86383605e-01 8.39891806e-02 1.18122423e+00 7.67390430e-01
4.32971984e-01 2.56419241e-01 7.03817368e-01 1.55036360e-01
3.39782536e-01 7.69560859e-02 -2.19702438e-01 -8.26223344e-02
6.78808689e-01 9.91487026e-01 -6.14705503e-01 -2.02485180e+00
-6.26180053e-01 -3.83221775e-01 -9.93895054e-01 1.11289443e-02
1.21425465e-01 -7.69230649e-02 -6.68267906e-01 1.01852798e+00
4.61143792e-01 5.41761756e-01 -2.45363116e-01 4.80155170e-01
4.66430694e-01 4.16972637e-01 2.68710345e-01 -3.46093237e-01
1.45874250e+00 -6.07676327e-01 -6.59245729e-01 -1.91346407e-01
8.74159336e-01 -5.79298615e-01 4.43730742e-01 6.35945320e-01
-3.11162472e-01 -5.36467284e-02 -1.01944995e+00 4.93259460e-01
-1.29795444e+00 -7.80540228e-01 1.04121935e+00 1.27110684e+00
-8.86013687e-01 6.99255586e-01 -6.81120932e-01 -4.43734229e-01
1.70486733e-01 1.88862875e-01 -3.64156634e-01 -3.10528260e-02
-1.49892473e+00 1.01019728e+00 6.28087044e-01 -4.21663612e-01
-6.47183716e-01 -3.52709115e-01 -1.01839590e+00 4.20334656e-03
1.01457679e+00 -3.20012838e-01 6.25356793e-01 -1.09222069e-01
-5.41186810e-01 7.15257287e-01 7.44925380e-01 -4.87175226e-01
-1.12340100e-01 2.06239834e-01 -1.12931705e+00 2.21801605e-02
-2.96848685e-01 -5.71356654e-01 8.64283085e-01 -1.16009855e+00
-4.09393966e-01 -6.28247917e-01 4.50510085e-01 -2.39160836e-01
-7.31143653e-01 4.17224735e-01 -3.27555016e-02 -6.82045102e-01
-3.35755348e-01 -8.85569274e-01 -5.92083752e-01 -7.79313147e-01
-8.97552133e-01 -1.91565767e-01 1.16666448e+00 -5.47363579e-01
1.77875531e+00 -1.92921865e+00 -1.27125457e-01 7.21335530e-01
9.59055424e-01 5.79333782e-01 -2.37857014e-01 1.17464030e+00
-4.17201966e-01 7.26858914e-01 -1.31857276e-01 2.29113884e-02
-3.34859379e-02 -1.28357440e-01 -4.83503252e-01 5.71424723e-01
-1.85312144e-03 8.35639238e-01 -1.08869457e+00 -2.37768576e-01
2.79310524e-01 2.04831511e-01 -4.62128446e-02 -2.35689506e-02
-1.51238158e-01 9.96652991e-02 -5.61308801e-01 9.73025799e-01
7.59618461e-01 -4.56618160e-01 7.93193161e-01 1.53816734e-02
1.89102292e-01 2.79973447e-01 -9.67424810e-01 7.20508218e-01
-6.01877831e-02 2.60427624e-01 2.46816397e-01 -9.95762169e-01
8.02649081e-01 2.13619217e-01 6.59699261e-01 -3.40993166e-01
4.32934582e-01 -7.16674998e-02 1.88807130e-01 -7.39647076e-02
5.36882043e-01 2.60577023e-01 -3.13260257e-01 8.94957006e-01
-8.45378414e-02 -2.32268661e-01 3.58097970e-01 8.06157708e-01
1.61810756e+00 -7.91866779e-01 9.35961306e-01 1.22650221e-01
6.66887939e-01 2.61144817e-01 2.13859349e-01 4.30809885e-01
-3.25970262e-01 -4.42146748e-01 7.85492301e-01 -7.46709764e-01
-6.22695267e-01 -6.96087241e-01 2.27435842e-01 7.99755037e-01
2.14667857e-01 -1.22386694e+00 -7.41841912e-01 -1.51691413e+00
2.75598228e-01 5.98101497e-01 -3.17888796e-01 -3.56915265e-01
-4.67007935e-01 -8.03365111e-01 5.93692958e-01 -5.11828586e-02
2.47146025e-01 -9.21461582e-01 1.35342255e-01 2.05781236e-02
3.69212925e-02 -1.58745027e+00 -2.10549071e-01 -3.09140354e-01
-9.86077189e-01 -2.00709772e+00 3.43304276e-01 -3.15573543e-01
6.86700404e-01 9.78027105e-01 1.16324055e+00 7.60143161e-01
-5.60842037e-01 8.21529984e-01 -6.63777590e-01 -6.25778317e-01
-5.81266105e-01 3.49698663e-02 3.86055380e-01 -7.80043006e-02
3.60204369e-01 -3.22397053e-01 -8.66068006e-02 2.06438884e-01
-1.12320697e+00 -7.03069627e-01 3.30255866e-01 6.15526065e-02
1.31019577e-01 1.08298254e+00 3.06872189e-01 -1.14569354e+00
1.33618104e+00 -8.93708289e-01 -8.27192724e-01 4.31840539e-01
-1.16654027e+00 -6.25978529e-01 7.73003221e-01 -2.77663440e-01
-4.32814389e-01 -4.33486938e-01 3.14934671e-01 -3.38725567e-01
-1.84702322e-01 8.46117437e-01 -1.15675531e-01 -7.38362372e-01
7.21080244e-01 -1.01496145e-01 -4.79104035e-02 -2.41879597e-01
3.78176242e-01 3.45608592e-01 -2.34671518e-01 -2.23343864e-01
1.24661636e+00 2.00917572e-01 4.40631449e-01 -1.03936589e+00
-5.43338656e-01 -6.20152295e-01 -6.55280411e-01 -5.55104613e-01
5.86661160e-01 -1.78006902e-01 -9.18345034e-01 3.71031582e-01
-1.14355624e+00 5.17411046e-02 3.57826680e-01 1.70202821e-01
1.11014605e-01 1.15746355e+00 -8.35612833e-01 -8.57703984e-01
-3.81621420e-01 -8.18062901e-01 5.38459003e-01 -1.14117883e-01
-5.90725280e-02 -1.70545816e+00 4.28020447e-01 5.35830081e-01
1.97236806e-01 6.79251134e-01 1.14305282e+00 -1.19153678e+00
-6.45727873e-01 -8.88915777e-01 -3.37959409e-01 1.53639555e-01
4.80556130e-01 7.87341669e-02 -3.72161806e-01 -6.41878963e-01
-1.20343409e-01 -2.03484707e-02 5.23849487e-01 -9.07563493e-02
9.87966359e-01 -5.73928773e-01 -7.29669392e-01 3.04282784e-01
1.36732352e+00 2.09933519e-01 3.96486044e-01 3.54734600e-01
1.01227999e+00 7.98812270e-01 6.97886944e-01 3.46240014e-01
3.16132247e-01 2.47439265e-01 9.01169360e-01 1.06501251e-01
3.71395200e-01 -1.08356982e-01 2.19737306e-01 9.67583239e-01
-5.30593395e-01 -5.85176468e-01 -1.24325037e+00 -6.53246939e-02
-1.67093122e+00 -9.16197598e-01 -5.60370028e-01 2.01440930e+00
8.95047262e-02 2.91428387e-01 3.30952287e-01 4.06275451e-01
8.44292760e-01 4.77036059e-01 -2.45641246e-01 -6.50788963e-01
1.70606464e-01 8.82048756e-02 6.83647096e-01 3.55322182e-01
-1.44982958e+00 1.23710644e+00 6.66263819e+00 6.20807946e-01
-8.57298374e-01 -1.15367763e-01 2.35105082e-01 5.69614589e-01
-1.01997457e-01 3.46903503e-01 -5.93751788e-01 5.13477102e-02
1.08696127e+00 -5.67758918e-01 6.51574373e-01 9.26491261e-01
-1.39647350e-01 2.90076196e-01 -6.16485834e-01 6.63645446e-01
2.93038845e-01 -1.25038958e+00 2.80581623e-01 5.59744000e-01
5.14796734e-01 -1.05217397e-01 -2.76606549e-02 2.21026450e-01
5.51333010e-01 -7.37346768e-01 -1.40260771e-01 -5.77445626e-02
2.63419956e-01 -9.23047364e-01 6.80789351e-01 3.58299643e-01
-1.55733919e+00 -2.91372925e-01 -2.57144451e-01 -2.32327402e-01
9.20295566e-02 8.16854894e-01 -9.37381566e-01 1.36155975e+00
4.70061868e-01 9.97113168e-01 -6.28878951e-01 8.60911131e-01
-4.70540434e-01 4.88660336e-01 1.59213752e-01 -1.93919301e-01
1.87059999e-01 -4.46556658e-01 7.14134812e-01 1.23414588e+00
-6.53110743e-02 2.24509180e-01 5.06777227e-01 2.88400978e-01
8.60877801e-03 -4.79187677e-03 -1.40278530e+00 -9.23999190e-01
4.59859341e-01 1.64341319e+00 -1.13391340e+00 -1.21061780e-01
-5.92286110e-01 3.55108351e-01 1.66018128e-01 1.84630349e-01
-7.27557540e-01 -7.40193903e-01 8.20431888e-01 2.37042040e-01
-4.26364362e-01 -9.46122348e-01 -3.00664693e-01 -1.34460962e+00
-4.25727516e-01 -1.45282817e+00 1.20535743e+00 -4.20820117e-02
-1.63331163e+00 7.74812698e-01 1.08569451e-01 -1.39427304e+00
-3.20375785e-02 -8.74976516e-01 -8.05845678e-01 3.27637225e-01
-1.28083432e+00 -1.15933180e+00 -2.53765374e-01 8.13124597e-01
-1.22643281e-02 -3.03164631e-01 8.05178225e-01 7.89008066e-02
-7.54671752e-01 1.58565968e-01 -7.14536250e-01 4.27582741e-01
4.71202374e-01 -9.61912334e-01 1.17475593e+00 1.15356553e+00
4.41479027e-01 8.71651232e-01 3.99305463e-01 -1.35278904e+00
-1.74020588e+00 -1.13282120e+00 8.25336874e-01 -6.78776860e-01
1.59299791e+00 -6.51644170e-01 -9.15076613e-01 8.21580827e-01
2.31837302e-01 -3.54297400e-01 8.88634861e-01 2.97326744e-01
-7.43057370e-01 1.89575553e-01 -1.11985326e+00 7.18016565e-01
1.24800611e+00 -5.29883802e-01 -3.85015368e-01 7.69236565e-01
8.82211149e-01 1.10433456e-02 -9.76021886e-01 3.10548991e-01
-1.65411517e-01 -8.76904130e-01 1.01369846e+00 -1.15067375e+00
-2.55785704e-01 -1.78844035e-02 3.24879676e-01 -1.20632648e+00
-6.59156203e-01 -7.82297492e-01 -5.81404746e-01 6.90284431e-01
2.45252490e-01 -1.25309145e+00 7.49067545e-01 4.92605478e-01
8.90102312e-02 -5.77863276e-01 -5.93108892e-01 -1.21946597e+00
-2.31136188e-01 -5.83787143e-01 5.60533226e-01 1.75650072e+00
7.27454782e-01 3.22233170e-01 -2.81179398e-01 5.39391577e-01
1.15461147e+00 6.47906363e-02 9.53001380e-01 -1.50183451e+00
1.90237988e-04 -6.04832888e-01 -8.47035646e-01 -1.61081046e-01
3.07088375e-01 -9.86746311e-01 -9.48986471e-01 -1.57579184e+00
-5.49082346e-02 -2.97830403e-01 -4.47277933e-01 5.93498826e-01
1.17367432e-02 9.46790259e-03 3.11971843e-01 6.91088736e-02
-6.32095098e-01 -2.14786828e-01 1.21459854e+00 -4.10292715e-01
-1.33120909e-01 2.32376128e-01 -7.31037915e-01 6.20956421e-01
9.74728227e-01 -4.55188245e-01 -5.95788240e-01 1.69151902e-01
6.20824754e-01 1.14077911e-01 2.24376753e-01 -6.09965563e-01
5.02834797e-01 -6.54795945e-01 -1.51610300e-01 -3.54860604e-01
8.98094848e-02 -9.75285470e-01 -5.68092056e-02 7.64396310e-01
3.70950818e-01 4.95221764e-01 1.70465037e-01 8.56715083e-01
-3.63197327e-01 1.07480995e-01 5.31222224e-01 -9.11288410e-02
-7.34588861e-01 9.69898999e-01 -4.69662935e-01 -1.59603551e-01
1.59021854e+00 -3.00436262e-02 -6.67626321e-01 -6.22110546e-01
-6.20931447e-01 2.71012127e-01 3.31173033e-01 8.55650723e-01
9.22072351e-01 -8.58442307e-01 -4.24182832e-01 4.95145507e-02
8.28297213e-02 -7.61000156e-01 1.48300052e-01 6.60833657e-01
-3.90160412e-01 5.97026408e-01 -3.08817446e-01 9.20946002e-02
-1.62268054e+00 1.36885095e+00 -1.03257177e-02 -7.70452976e-01
-3.72670352e-01 2.79472351e-01 -2.17795908e-01 -4.34772402e-01
1.75915048e-01 1.69775747e-02 -5.38870096e-01 -1.81994617e-01
5.46649992e-01 7.24460065e-01 2.41819352e-01 -4.29299951e-01
-7.32461631e-01 3.87094826e-01 -2.54336864e-01 5.14326394e-01
1.26132023e+00 2.57927656e-01 -6.54384196e-01 -1.64975658e-01
8.38144064e-01 2.33798757e-01 -6.87620416e-02 7.64862671e-02
5.50671875e-01 -6.42739475e-01 -1.71311289e-01 -6.67366624e-01
-1.00286293e+00 6.20724261e-01 -2.97699153e-01 1.42914474e+00
1.13633668e+00 -9.76762176e-02 7.87976027e-01 4.05456603e-01
1.04080904e+00 -7.92479336e-01 3.42081994e-01 7.88746119e-01
6.84622884e-01 -9.26918745e-01 4.59645182e-01 -1.25431943e+00
-6.08202994e-01 1.26611793e+00 6.13387704e-01 -3.44591409e-01
1.22012818e+00 2.56002575e-01 -2.42322460e-01 -6.66876912e-01
-5.30170381e-01 7.03794509e-03 3.49392742e-01 8.73157203e-01
6.71446994e-02 3.41995567e-01 -1.44198596e-01 2.21171439e-01
-4.84771132e-02 -8.18431318e-01 5.95310748e-01 1.10867095e+00
-3.05796891e-01 -1.64062965e+00 -4.72150207e-01 4.99691427e-01
-4.53748584e-01 -1.42253712e-01 -1.45616961e+00 1.04604459e+00
-3.48440200e-01 1.43097484e+00 -2.96695352e-01 -1.09015667e+00
5.08018017e-01 -3.94936323e-01 1.58184662e-01 -7.59891391e-01
-7.95906007e-01 -5.15729070e-01 4.21230555e-01 -7.14877069e-01
1.20171219e-01 -2.48658955e-01 -1.03544688e+00 -1.02350557e+00
-6.41892254e-01 1.78528950e-01 6.83385193e-01 5.00427365e-01
5.50952613e-01 3.85890365e-01 8.76476526e-01 -2.48294160e-01
-3.81800383e-01 -6.54677689e-01 -9.19154823e-01 2.17423528e-01
-1.64226070e-01 -6.58937752e-01 -8.26824903e-01 -4.66868341e-01] | [6.24870491027832, 7.236874580383301] |
62edfa7c-dce7-4096-b90c-16ce9fdcfc9a | aesthetic-image-captioning-from-weakly | 1908.11310 | null | https://arxiv.org/abs/1908.11310v1 | https://arxiv.org/pdf/1908.11310v1.pdf | Aesthetic Image Captioning From Weakly-Labelled Photographs | Aesthetic image captioning (AIC) refers to the multi-modal task of generating critical textual feedbacks for photographs. While in natural image captioning (NIC), deep models are trained in an end-to-end manner using large curated datasets such as MS-COCO, no such large-scale, clean dataset exists for AIC. Towards this goal, we propose an automatic cleaning strategy to create a benchmarking AIC dataset, by exploiting the images and noisy comments easily available from photography websites. We propose a probabilistic caption-filtering method for cleaning the noisy web-data, and compile a large-scale, clean dataset "AVA-Captions", (230, 000 images with 5 captions per image). Additionally, by exploiting the latent associations between aesthetic attributes, we propose a strategy for training the convolutional neural network (CNN) based visual feature extractor, the first component of the AIC framework. The strategy is weakly supervised and can be effectively used to learn rich aesthetic representations, without requiring expensive ground-truth annotations. We finally show-case a thorough analysis of the proposed contributions using automatic metrics and subjective evaluations. | ['Koustav Ghosal', 'Aljosa Smolic', 'Aakanksha Rana'] | 2019-08-29 | null | null | null | null | ['aesthetic-image-captioning'] | ['computer-vision'] | [ 4.11733955e-01 5.19187152e-01 2.01637536e-01 -4.08021867e-01
-1.48254001e+00 -6.49258137e-01 5.75845122e-01 1.67675242e-01
-2.05153778e-01 5.79608917e-01 5.76358795e-01 6.88357651e-02
3.00702602e-01 -3.27848941e-01 -1.12716961e+00 -5.09191513e-01
2.55003572e-01 2.65630484e-01 -4.26737487e-01 -5.33265173e-02
1.16191946e-01 -2.86472794e-02 -1.62721777e+00 6.18305087e-01
7.76233017e-01 1.24401867e+00 1.10800192e-01 7.78461874e-01
5.74975424e-02 8.27709436e-01 -6.16579115e-01 -1.04488349e+00
2.13720709e-01 -3.84201646e-01 -1.04156899e+00 5.45998871e-01
6.03158474e-01 -2.72470713e-01 -1.01917893e-01 1.00255513e+00
3.79849613e-01 -9.95406210e-02 6.03070617e-01 -1.49826634e+00
-1.30504632e+00 6.32780313e-01 -3.33504498e-01 -4.86267507e-01
2.57363498e-01 4.95026618e-01 1.30787635e+00 -9.53970790e-01
9.43602324e-01 1.18057287e+00 6.42860234e-01 6.55985594e-01
-1.20676732e+00 -1.58005878e-01 -7.69303041e-03 1.48248240e-01
-1.03681290e+00 -4.27472204e-01 8.92398477e-01 -4.37503427e-01
3.83527458e-01 4.92704839e-01 4.42772180e-01 1.72302723e+00
-5.31368852e-01 1.06696928e+00 1.13039505e+00 -4.88978595e-01
1.94961086e-01 5.17393708e-01 -1.11418836e-01 4.34876263e-01
3.77738737e-02 -3.20675135e-01 -4.89323795e-01 2.47942079e-02
2.75156051e-01 -2.68863767e-01 -2.37463132e-01 -3.49616349e-01
-1.13076532e+00 8.29325557e-01 6.95888281e-01 -3.12320665e-02
-3.93673599e-01 3.83051902e-01 6.55828714e-01 8.28724876e-02
7.44785190e-01 7.31938064e-01 -2.48102605e-01 -1.98639110e-01
-8.05846453e-01 7.90233091e-02 6.00696921e-01 1.12935412e+00
7.60681808e-01 -1.80683419e-01 -6.47427022e-01 9.10645783e-01
1.22073218e-01 6.19738340e-01 1.15238488e-01 -1.02332544e+00
5.40274858e-01 5.85361302e-01 3.08979809e-01 -1.00442958e+00
-3.03262100e-02 -2.86453247e-01 -8.66107225e-01 1.13305897e-01
5.57626747e-02 -4.96237576e-02 -1.03208077e+00 1.68365705e+00
1.06012262e-01 -3.66352797e-01 1.64475828e-01 1.06735039e+00
1.01494622e+00 6.44310355e-01 4.22423154e-01 4.03958466e-03
1.40746856e+00 -1.37603116e+00 -7.51053333e-01 -4.05035615e-01
4.34164971e-01 -9.34893906e-01 1.70143032e+00 4.56712276e-01
-1.11948287e+00 -3.04974318e-01 -9.43616450e-01 -4.59023654e-01
-5.93370974e-01 4.64891315e-01 4.36253220e-01 3.38906258e-01
-1.11628354e+00 4.07252699e-01 -1.87409922e-01 -3.20145339e-01
8.43924403e-01 -2.33600572e-01 -6.20903730e-01 -2.62012988e-01
-9.11457419e-01 8.36545765e-01 1.65096775e-01 1.83211505e-01
-1.21380568e+00 -4.95844126e-01 -9.64747310e-01 1.93561822e-01
2.47081578e-01 -6.62901103e-01 1.24408591e+00 -1.80335367e+00
-1.24250758e+00 1.13329458e+00 -3.51343816e-03 -4.99182969e-01
5.95587790e-01 -4.76633012e-01 -1.32179558e-01 5.06311655e-01
1.93336383e-01 1.14659595e+00 1.05017531e+00 -1.98633182e+00
-2.00382192e-02 1.65456608e-01 2.09190980e-01 1.10194921e-01
-6.98159456e-01 8.05010572e-02 -6.72262013e-01 -7.21150219e-01
-6.67119265e-01 -8.65983307e-01 -2.75750905e-01 5.84623814e-02
-7.32724428e-01 1.29123062e-01 6.65336370e-01 -9.21211243e-01
9.83254671e-01 -2.23324060e+00 8.61824304e-02 9.50274542e-02
1.98644489e-01 3.10815513e-01 -5.15436947e-01 4.05312300e-01
-2.33487979e-01 4.41588134e-01 -4.83956695e-01 -1.08992100e+00
2.59354919e-01 1.06190093e-01 -2.27564245e-01 1.26943752e-01
6.28363431e-01 1.30734754e+00 -1.07104039e+00 -6.34218931e-01
2.93251306e-01 5.17160892e-01 -4.29121137e-01 5.53651750e-01
-5.27476668e-01 1.39779672e-01 -1.20323755e-01 6.35854900e-01
6.72343552e-01 -5.29107571e-01 -1.03848673e-01 -4.38047588e-01
1.99770629e-01 -9.80165526e-02 -7.34510481e-01 1.84248745e+00
-7.49940276e-01 9.69956279e-01 1.35020345e-01 -5.54597676e-01
8.58503282e-01 3.09615880e-01 3.32625687e-01 -6.78319335e-01
3.23769152e-01 -7.68251121e-02 -9.73926961e-01 -7.51888931e-01
8.13494921e-01 2.35387534e-01 -4.47986275e-01 3.49023819e-01
2.62506485e-01 -2.79147238e-01 3.33237886e-01 4.83067721e-01
1.04140139e+00 6.10185899e-02 -4.15137894e-02 1.51962236e-01
4.08107400e-01 2.75780350e-01 6.18082918e-02 5.28188467e-01
-2.36144066e-01 1.41486347e+00 6.05059564e-01 -4.83719260e-01
-1.60093904e+00 -9.06968296e-01 2.98684031e-01 9.06335115e-01
-5.95214106e-02 -7.36330748e-01 -1.18272007e+00 -7.80663371e-01
-3.30752730e-01 5.65076172e-01 -1.02559471e+00 -9.56488773e-02
-1.63201168e-01 -4.88493890e-01 2.64252931e-01 1.79103196e-01
3.62476647e-01 -1.46383655e+00 -1.73171431e-01 -1.20225973e-01
-6.86541677e-01 -1.27928638e+00 -4.76293772e-01 6.00011684e-02
-1.74454182e-01 -1.15047836e+00 -7.08782732e-01 -7.90885150e-01
8.73149753e-01 3.69674027e-01 1.57719183e+00 1.28886178e-01
-3.28482658e-01 6.60149157e-01 -7.03073978e-01 -4.21165913e-01
-5.61676741e-01 8.80781747e-03 -4.85986471e-01 3.38081688e-01
1.32579967e-01 -3.92454267e-01 -7.14885831e-01 -2.85091121e-02
-1.38935435e+00 4.18030322e-01 9.94204104e-01 8.46766710e-01
7.15284944e-01 -5.91427028e-01 3.86801630e-01 -8.63121569e-01
7.00806975e-01 -5.71845591e-01 -2.03907073e-01 4.06176120e-01
-4.15460527e-01 -2.56036390e-02 6.67062283e-01 -3.06558728e-01
-9.20226514e-01 4.35687691e-01 -1.56737044e-01 -7.61581182e-01
-3.22911233e-01 3.72780174e-01 -3.28915030e-01 1.20682448e-01
7.37080038e-01 3.10029183e-02 -1.70413435e-01 -4.52615082e-01
9.71644640e-01 7.16523230e-01 9.23916638e-01 -4.35235858e-01
1.03550172e+00 5.03332078e-01 -4.15135682e-01 -5.37325501e-01
-1.44784153e+00 -4.58165735e-01 -4.49123949e-01 -5.10981023e-01
1.18678355e+00 -1.12062836e+00 -4.21129346e-01 1.91378251e-01
-1.32139373e+00 -2.24616244e-01 -5.16910911e-01 -1.36413366e-01
-7.84525275e-01 4.16573018e-01 -3.01559776e-01 -7.14540362e-01
-8.76539767e-01 -1.08310962e+00 1.47205341e+00 1.62032947e-01
-3.02889645e-01 -5.30001462e-01 5.37903979e-02 8.39878738e-01
3.60636145e-01 7.66539454e-01 5.07757723e-01 -3.95948946e-01
-5.52100241e-01 -3.23062450e-01 -6.96569622e-01 9.59889829e-01
-3.62498105e-01 1.90645486e-01 -1.33123922e+00 8.72872621e-02
-3.82766753e-01 -9.58830416e-01 7.65614986e-01 1.58591956e-01
1.57480967e+00 -7.55180418e-01 1.76679417e-01 5.25606990e-01
1.62070203e+00 -6.37541473e-01 9.48384464e-01 5.08159041e-01
9.59682345e-01 6.13110244e-01 6.27887487e-01 4.75448400e-01
5.76261997e-01 3.24599594e-01 9.65441883e-01 -5.33183634e-01
-3.85807931e-01 -4.65234548e-01 3.43703449e-01 7.26284564e-01
2.51521558e-01 -4.28449959e-01 -6.31314874e-01 8.97325873e-01
-1.97252440e+00 -7.94860661e-01 -2.71332681e-01 1.81862378e+00
9.46528494e-01 -8.95846710e-02 -1.50040597e-01 -8.10498372e-02
7.04536438e-01 1.55204013e-01 -2.56946027e-01 -5.85512757e-01
-4.19192940e-01 -1.02589196e-02 5.00158787e-01 3.05651039e-01
-1.32733941e+00 8.88091564e-01 5.95386076e+00 8.53688300e-01
-6.74497128e-01 2.70881951e-01 1.14010572e+00 -8.71512294e-02
-6.57318234e-01 -2.95336246e-02 -1.28767967e-01 4.55628842e-01
1.04565489e+00 2.83078492e-01 3.39342624e-01 1.30169570e+00
3.58353376e-01 -7.26001859e-02 -9.53496993e-01 1.26618290e+00
4.21059519e-01 -1.48602271e+00 2.56486595e-01 -2.03622401e-01
1.02331710e+00 -8.64221603e-02 2.20910236e-01 1.39914632e-01
3.87651384e-01 -1.04234588e+00 1.05277002e+00 6.36071444e-01
9.56849873e-01 -6.72068238e-01 7.93534994e-01 -2.13708907e-01
-7.18322277e-01 -1.06429616e-02 -1.65082112e-01 3.06190223e-01
3.46177459e-01 8.65794420e-01 -6.18190885e-01 4.26781476e-01
1.01634228e+00 7.17271149e-01 -1.13624120e+00 1.08920002e+00
-4.12003100e-01 5.21150649e-01 5.17193452e-02 2.06331946e-02
4.18791324e-01 -2.26103142e-02 3.81557345e-01 1.36644840e+00
8.53450000e-02 -3.78376454e-01 -2.85440445e-01 9.64866519e-01
-5.86771369e-01 2.21698865e-01 -4.99995857e-01 -2.42872879e-01
5.30183241e-02 1.83728564e+00 -5.00151932e-01 -4.17286903e-01
-1.72305539e-01 1.25165558e+00 5.06570280e-01 3.28526407e-01
-1.02136457e+00 -2.61046171e-01 4.87506658e-01 -1.56424329e-01
2.39354730e-01 1.30284429e-01 -5.55154741e-01 -1.18004632e+00
3.94496560e-01 -1.01074815e+00 5.54066040e-02 -1.48604453e+00
-1.57049823e+00 7.81086564e-01 -2.07020983e-01 -1.37921274e+00
4.49215285e-02 -4.58927035e-01 -6.62029386e-01 5.21760345e-01
-1.70468056e+00 -1.62560129e+00 -8.98798525e-01 5.43069184e-01
6.38625562e-01 3.51941466e-01 8.65384042e-01 2.54129857e-01
-4.79761064e-01 6.08177066e-01 1.33721292e-01 1.20008163e-01
1.07802951e+00 -1.45656109e+00 3.79136652e-01 9.13097262e-01
-8.33281130e-03 6.90897331e-02 9.55848157e-01 -3.13292056e-01
-1.13132715e+00 -1.50045609e+00 9.18494165e-01 -6.19876325e-01
6.52204692e-01 -7.65004277e-01 -6.54876709e-01 3.77361238e-01
7.80731082e-01 -8.38922784e-02 6.47796392e-01 -1.36800915e-01
-5.64361572e-01 -6.81685880e-02 -1.00404084e+00 7.07266271e-01
9.36090648e-01 -4.86507684e-01 -3.65601361e-01 8.81321490e-01
1.15164328e+00 1.56148318e-02 -7.25713015e-01 -2.47000381e-02
2.46483281e-01 -9.15789664e-01 9.84862387e-01 -4.18853939e-01
1.14685297e+00 -1.91186652e-01 5.93522638e-02 -1.38338470e+00
-8.82198140e-02 -1.00188959e+00 7.81953037e-02 1.70803630e+00
6.56527936e-01 2.03836471e-01 6.61100864e-01 7.67247975e-01
-4.52376753e-01 -4.14506316e-01 -6.65755510e-01 -3.56840253e-01
-3.07615221e-01 -5.70371449e-01 4.34866458e-01 6.94290996e-01
-2.22797424e-01 3.94177020e-01 -8.61250758e-01 -9.04304460e-02
5.70165277e-01 -8.76302570e-02 1.06236279e+00 -8.55303347e-01
1.93583984e-02 -2.70911962e-01 -1.08345360e-01 -3.50094587e-01
-2.29062550e-02 -6.42525733e-01 4.32346940e-01 -1.72003913e+00
4.40327466e-01 -1.76281124e-01 -8.67756642e-03 5.50847948e-01
-3.14998865e-01 6.46950483e-01 2.60271877e-01 1.70358986e-01
-1.18897676e+00 7.82648683e-01 1.21867788e+00 -2.93269336e-01
2.58078456e-01 -6.34744048e-01 -9.90848660e-01 5.39112091e-01
7.08793581e-01 -4.52259779e-01 -2.07861111e-01 -5.89670897e-01
6.07134521e-01 -5.68973422e-01 6.88799739e-01 -9.27445114e-01
-2.62108535e-01 -5.99854775e-02 2.87551701e-01 -4.63812530e-01
4.39339668e-01 -8.68442118e-01 -8.03367875e-05 -1.54717833e-01
-6.48334086e-01 6.67102113e-02 6.83670118e-02 6.57105923e-01
-3.93295884e-01 -2.95856029e-01 7.37076640e-01 -1.98283195e-01
-5.18024206e-01 2.13199928e-01 -1.18680887e-01 1.61989078e-01
1.06505466e+00 1.35027081e-01 -5.60454965e-01 -6.25780046e-01
-6.29552782e-01 2.85335839e-01 7.18322039e-01 6.29443049e-01
6.34495497e-01 -1.50144720e+00 -7.68306077e-01 -2.94624567e-01
6.90534472e-01 1.20761164e-01 4.63726729e-01 4.89094436e-01
-6.29810929e-01 6.32825121e-02 -2.69484341e-01 -4.17429328e-01
-1.10639560e+00 9.25725222e-01 -1.70163840e-01 -2.83298880e-01
-3.95750284e-01 7.82161951e-01 1.28296733e-01 -1.42304808e-01
2.84270793e-01 -1.89917430e-01 -2.48744011e-01 1.65948555e-01
6.10749602e-01 -1.55589059e-01 1.77859411e-01 -6.99635088e-01
6.84937984e-02 1.42073676e-01 1.50769502e-01 -1.42433003e-01
1.71344459e+00 -5.66134632e-01 -6.21160045e-02 6.56301901e-02
1.58404875e+00 -3.13337326e-01 -1.40928137e+00 2.53890343e-02
-7.58007094e-02 -4.12227631e-01 -1.61302894e-01 -9.26728427e-01
-1.09540546e+00 8.32362890e-01 4.47061598e-01 2.82285810e-01
1.35137415e+00 7.08482265e-02 9.68112290e-01 2.57039249e-01
-2.62135327e-01 -1.39672995e+00 5.33704817e-01 1.70110106e-01
1.33095372e+00 -1.71437526e+00 -2.77261347e-01 -2.77591318e-01
-1.28162169e+00 8.87850165e-01 5.21105945e-01 -2.58201659e-01
1.05287060e-01 -1.35735184e-01 2.37950221e-01 -2.54744142e-01
-7.98652470e-01 -3.31452698e-01 4.66006786e-01 7.61038542e-01
1.20966071e-02 4.66316231e-02 -4.94504087e-02 7.89448380e-01
-2.29480281e-01 -2.10729703e-01 8.22321117e-01 5.84405422e-01
-2.95411348e-01 -8.90348494e-01 -3.12961966e-01 2.26230681e-01
-2.89306372e-01 -4.08332258e-01 -7.91443944e-01 5.09097159e-01
5.58100902e-02 1.07290888e+00 -1.39692724e-01 -3.42064649e-01
2.63053209e-01 5.45582771e-02 -7.49333128e-02 -3.78824413e-01
-7.00962126e-01 -2.09726125e-01 4.02211279e-01 -7.51414597e-01
-5.46431899e-01 -5.17738521e-01 -4.58737135e-01 -1.27785623e-01
-1.27112865e-02 7.25744441e-02 1.00155091e+00 6.81262434e-01
4.76466537e-01 3.66207421e-01 7.72715151e-01 -1.12944615e+00
-6.37484491e-02 -9.69648302e-01 -6.39921939e-03 1.19294322e+00
2.19550774e-01 -3.16584915e-01 -5.97048819e-01 6.70625627e-01] | [11.016257286071777, 1.0503814220428467] |
10d2ba19-460c-4581-9d7b-4119863df9cd | structure-aware-face-clustering-on-a-large | 2103.13225 | null | https://arxiv.org/abs/2103.13225v2 | https://arxiv.org/pdf/2103.13225v2.pdf | Structure-Aware Face Clustering on a Large-Scale Graph with $\bf{10^{7}}$ Nodes | Face clustering is a promising method for annotating unlabeled face images. Recent supervised approaches have boosted the face clustering accuracy greatly, however their performance is still far from satisfactory. These methods can be roughly divided into global-based and local-based ones. Global-based methods suffer from the limitation of training data scale, while local-based ones are difficult to grasp the whole graph structure information and usually take a long time for inference. Previous approaches fail to tackle these two challenges simultaneously. To address the dilemma of large-scale training and efficient inference, we propose the STructure-AwaRe Face Clustering (STAR-FC) method. Specifically, we design a structure-preserved subgraph sampling strategy to explore the power of large-scale training data, which can increase the training data scale from ${10^{5}}$ to ${10^{7}}$. During inference, the STAR-FC performs efficient full-graph clustering with two steps: graph parsing and graph refinement. And the concept of node intimacy is introduced in the second step to mine the local structural information. The STAR-FC gets 91.97 pairwise F-score on partial MS1M within 310s which surpasses the state-of-the-arts. Furthermore, we are the first to train on very large-scale graph with 20M nodes, and achieve superior inference results on 12M testing data. Overall, as a simple and effective method, the proposed STAR-FC provides a strong baseline for large-scale face clustering. Code is available at \url{https://sstzal.github.io/STAR-FC/}. | ['Jie zhou', 'Jiwen Lu', 'Dalong Du', 'Guan Huang', 'Zheng Zhu', 'Wanhua Li', 'Shuai Shen'] | 2021-03-24 | null | null | null | null | ['face-clustering'] | ['computer-vision'] | [-8.16053823e-02 2.46150807e-01 -3.45690936e-01 -6.29603505e-01
-8.61872315e-01 -3.26573551e-01 3.20515752e-01 -1.88671723e-01
-8.59927014e-02 4.42296743e-01 -2.35609468e-02 -1.59673989e-01
-1.81884497e-01 -7.87057042e-01 -6.19743586e-01 -7.72770762e-01
-9.10222083e-02 6.49381816e-01 1.71572492e-01 1.64561570e-01
-4.65724356e-02 3.80213737e-01 -1.52347827e+00 7.27634877e-02
7.58755803e-01 8.34510326e-01 1.23551659e-01 2.25519195e-01
-2.24704176e-01 4.99381065e-01 -3.18312615e-01 -7.75739849e-01
2.43097264e-02 -5.25098562e-01 -9.58779693e-01 2.93866038e-01
5.74483514e-01 -9.48034897e-02 -2.01411307e-01 1.24196768e+00
5.13072789e-01 -8.44938606e-02 4.16363239e-01 -1.51767957e+00
-2.89992243e-01 6.87561393e-01 -1.02398503e+00 -1.13082133e-01
2.49880120e-01 -3.03774402e-02 9.39870596e-01 -1.00500536e+00
6.41758382e-01 1.40298128e+00 6.84295356e-01 7.82820821e-01
-9.72327590e-01 -1.10840702e+00 2.10547760e-01 1.47084132e-01
-1.74773490e+00 -6.89354062e-01 8.84439528e-01 -1.75300911e-01
5.58187783e-01 2.48570681e-01 4.16126490e-01 8.39607537e-01
-5.59312582e-01 6.63652003e-01 1.20504189e+00 -2.76295036e-01
1.57142371e-01 -2.01055646e-01 2.19893176e-02 1.16917360e+00
2.21908256e-01 -2.28036568e-01 -5.47195196e-01 -5.66051379e-02
7.39001691e-01 -8.36136863e-02 -1.47028983e-01 -6.52774354e-04
-6.90793455e-01 9.35381770e-01 5.51777303e-01 2.89300144e-01
-9.43980087e-03 8.31322595e-02 1.82984129e-01 1.10244796e-01
5.01069367e-01 2.70373542e-02 -3.02444696e-01 5.58865033e-02
-1.18828380e+00 -9.51309204e-02 7.51413107e-01 1.01638782e+00
9.13339555e-01 -1.98291823e-01 3.82403024e-02 9.93700206e-01
4.80210811e-01 3.95762056e-01 -1.69179291e-02 -1.14817786e+00
3.61407191e-01 7.27138400e-01 -5.45626402e-01 -1.24056709e+00
-3.74447495e-01 -3.48081023e-01 -1.07545459e+00 -1.23820595e-01
5.61343193e-01 -1.72716975e-01 -9.51542675e-01 1.87955558e+00
8.04189444e-01 6.51521027e-01 -4.10189956e-01 7.07834482e-01
9.84284639e-01 3.23126614e-01 1.53348232e-02 -5.21352828e-01
1.49522686e+00 -1.01197529e+00 -5.46839774e-01 -9.11770463e-02
6.55680835e-01 -6.42733634e-01 1.01132178e+00 2.74344623e-01
-9.96226251e-01 -4.74238783e-01 -6.36852682e-01 1.80889636e-01
-7.84335136e-02 9.35305431e-02 8.08794379e-01 8.30444694e-01
-1.31369817e+00 5.04937112e-01 -8.34300339e-01 -4.00626212e-01
9.23389852e-01 6.07326090e-01 -5.41781187e-01 -5.84523797e-01
-7.66601264e-01 1.42549157e-01 2.60257810e-01 1.78838432e-01
-7.81041861e-01 -3.83203089e-01 -7.75699914e-01 -5.79751329e-03
7.92411745e-01 -3.69899005e-01 7.72503674e-01 -6.28651142e-01
-1.24378467e+00 9.68128502e-01 -3.92554700e-01 8.15395936e-02
8.96418765e-02 2.06082687e-01 -4.10470098e-01 4.25424308e-01
1.43456519e-01 8.90904248e-01 8.33584011e-01 -1.30042231e+00
-3.24019790e-01 -7.61278629e-01 -3.08054276e-02 -2.06129849e-02
-7.11156487e-01 2.15610892e-01 -1.31208670e+00 -5.41511297e-01
2.35851407e-01 -1.06190395e+00 -9.55105722e-02 -1.06648840e-01
-4.39118624e-01 -5.28830886e-01 7.10227013e-01 -4.34253901e-01
1.34576714e+00 -2.02314687e+00 3.84916104e-02 3.82440776e-01
4.48696703e-01 4.00842190e-01 -3.00151885e-01 1.84520900e-01
8.34552199e-02 3.81050259e-01 -2.39567921e-01 -7.49114513e-01
-7.13105202e-02 2.43243143e-01 3.28065693e-01 3.28071028e-01
1.12201750e-01 1.03034878e+00 -7.77495623e-01 -9.59593654e-01
-5.88220842e-02 5.96159399e-01 -7.24452376e-01 1.58082098e-01
-7.55605474e-02 5.22464335e-01 -4.44638729e-01 8.43393445e-01
1.03991044e+00 -7.26350307e-01 8.19152713e-01 -1.97363138e-01
4.39598501e-01 -7.77553990e-02 -1.43468440e+00 1.74381757e+00
-6.05478808e-02 1.43974766e-01 5.45495927e-01 -1.04456663e+00
7.66322672e-01 1.82215407e-01 5.44839621e-01 -4.66208577e-01
1.84359401e-01 -3.44188348e-03 -1.66579381e-01 -3.35228086e-01
-6.30294681e-02 -3.88747863e-02 8.83304924e-02 4.07802284e-01
1.88760832e-01 1.95962921e-01 3.05496544e-01 4.79872882e-01
1.19331121e+00 5.93742542e-02 -1.48788858e-02 -2.84574717e-01
6.05304480e-01 -4.28693771e-01 7.55036235e-01 3.03647608e-01
-1.85933024e-01 4.91538137e-01 6.78906262e-01 -1.58361748e-01
-4.22361135e-01 -7.94922709e-01 -1.95540991e-02 1.13180041e+00
1.59808844e-01 -1.04187059e+00 -1.24170268e+00 -1.13477802e+00
-1.13997914e-01 1.28773600e-01 -5.17410636e-01 -1.49362357e-02
-5.24881899e-01 -8.72997880e-01 6.34577453e-01 4.77714688e-01
5.72728336e-01 -1.04348111e+00 1.86540648e-01 -2.09139109e-01
-4.48179245e-01 -1.39611900e+00 -5.99969149e-01 -3.31818581e-01
-7.76751697e-01 -1.23594987e+00 -2.70880699e-01 -9.69359517e-01
1.07725966e+00 2.97819197e-01 1.22922802e+00 7.44703352e-01
-3.23973835e-01 1.85798198e-01 -3.04218113e-01 4.03057560e-02
-1.66666843e-02 1.11359388e-01 4.65780832e-02 6.64924830e-02
5.35289407e-01 -8.00099254e-01 -7.01148331e-01 5.97433925e-01
-5.70686638e-01 -3.58818695e-02 6.99176788e-01 5.87013066e-01
6.92305326e-01 3.84390742e-01 5.47454596e-01 -1.13707376e+00
1.62107006e-01 -4.09115374e-01 -4.07999635e-01 2.82195687e-01
-7.45960712e-01 -9.17516276e-02 3.71276438e-01 -2.45681927e-01
-9.18022454e-01 2.45644957e-01 -3.65164876e-01 -4.59084481e-01
-2.92792141e-01 4.14976716e-01 -6.14036620e-01 -1.73771769e-01
2.33823702e-01 -2.40330212e-02 4.78624627e-02 -5.63539147e-01
2.14345008e-01 5.00745952e-01 4.19089854e-01 -7.28791773e-01
9.47541654e-01 5.15348136e-01 1.06701396e-01 -8.30375791e-01
-8.06637526e-01 -4.44006920e-01 -7.25166440e-01 -3.58001649e-01
8.44865739e-01 -9.39799070e-01 -1.02263403e+00 4.83366281e-01
-7.43134975e-01 -4.22787696e-01 2.08496600e-01 1.28534645e-01
-3.86396870e-02 6.90397978e-01 -7.26741433e-01 -7.23343790e-01
-2.65122831e-01 -1.07331693e+00 1.28489971e+00 2.03111231e-01
-5.26436232e-03 -9.00018215e-01 -3.77398044e-01 8.98612380e-01
4.86484244e-02 2.17270672e-01 6.61875308e-01 -4.26330388e-01
-5.95734417e-01 -1.03189321e-02 -4.97227937e-01 1.10819161e-01
1.12631127e-01 -1.27480716e-01 -9.03607607e-01 -5.79115152e-01
-2.68579304e-01 -5.21949947e-01 7.34183848e-01 1.82607710e-01
1.53579962e+00 -2.42490262e-01 -6.21451437e-01 6.36662304e-01
1.29401851e+00 -3.11261863e-01 6.23203158e-01 -3.97693068e-01
1.02039719e+00 5.91928065e-01 6.33365631e-01 3.67280364e-01
6.00195527e-01 6.66869462e-01 3.69473636e-01 -1.35887831e-01
-3.37287843e-01 -3.80206674e-01 2.75802016e-01 9.67975736e-01
-2.58319139e-01 -2.18972072e-01 -8.54028225e-01 3.02607954e-01
-1.62487030e+00 -7.66840935e-01 -1.77721992e-01 2.07670784e+00
8.79050910e-01 1.26510277e-01 2.96122760e-01 1.25341520e-01
9.37556624e-01 1.13500901e-01 -3.46812189e-01 2.49722257e-01
1.59636304e-01 3.27950120e-01 5.04807346e-02 5.92081666e-01
-1.02487946e+00 1.20860624e+00 5.04870462e+00 1.21393383e+00
-8.29231501e-01 2.81239808e-01 7.61022627e-01 -2.60890834e-02
-1.41887486e-01 1.05837151e-01 -1.08761430e+00 5.68750799e-01
9.27163839e-01 3.67489845e-01 5.73059142e-01 7.62403131e-01
-1.09107196e-01 -1.53848054e-02 -8.75519335e-01 1.19501352e+00
2.39623219e-01 -1.04256082e+00 -2.41821688e-02 3.68474811e-01
5.98919451e-01 -1.06548309e-01 -1.49134874e-01 3.22042674e-01
2.80131161e-01 -1.23347139e+00 6.57810196e-02 7.79198036e-02
1.13806450e+00 -7.93275595e-01 5.64426780e-01 4.01083082e-01
-1.72308874e+00 5.12702204e-02 -2.92299092e-01 1.45997941e-01
-4.18797098e-02 7.36134827e-01 -4.58688140e-01 6.63264453e-01
9.51294780e-01 5.97632885e-01 -7.82973051e-01 5.31290948e-01
-3.09577107e-01 9.24612164e-01 -4.91345584e-01 2.20349163e-01
-7.77678788e-02 -3.48459572e-01 1.51046455e-01 9.66273308e-01
1.03788592e-01 4.26418185e-01 5.31667352e-01 4.33670312e-01
-5.95696747e-01 8.74857381e-02 -3.42239201e-01 -1.29975304e-01
7.18815148e-01 1.71383214e+00 -1.25820839e+00 -1.46332547e-01
-5.08103490e-01 1.02926147e+00 6.65784359e-01 1.58676654e-01
-1.01361251e+00 -1.84766069e-01 4.11236435e-01 2.97311693e-01
4.96056885e-01 -1.61536515e-01 8.70533362e-02 -1.07493365e+00
6.01209551e-02 -1.09849429e+00 7.19594538e-01 -4.47668672e-01
-1.22252154e+00 6.62622154e-01 -9.94802918e-03 -6.80866539e-01
5.02409004e-02 -5.06689727e-01 -4.30700481e-01 2.42293164e-01
-1.25515008e+00 -1.57399476e+00 -6.24431193e-01 9.10396934e-01
2.81052738e-01 -3.31069790e-02 8.47798347e-01 6.14200473e-01
-7.85435855e-01 1.01904428e+00 -2.38785654e-01 4.68798161e-01
6.97051823e-01 -1.03346491e+00 1.93934873e-01 8.15103889e-01
4.13682580e-01 6.81113899e-01 1.23253278e-01 -7.29588330e-01
-1.46274245e+00 -1.14190495e+00 7.55324721e-01 -5.43271005e-01
2.72707194e-01 -7.54458964e-01 -8.83714855e-01 5.66033840e-01
-1.13212354e-01 4.31856245e-01 7.07395673e-01 4.68412757e-01
-5.83203793e-01 -3.51978242e-01 -1.25548279e+00 4.69414353e-01
1.62667084e+00 -5.73707938e-01 -1.15729859e-02 4.29523587e-01
6.07879758e-01 -1.73133075e-01 -1.18262446e+00 4.55977172e-01
3.92887950e-01 -1.05219221e+00 9.29653406e-01 -2.62610137e-01
2.05114305e-01 -2.05150306e-01 -5.36800772e-02 -7.17945099e-01
-3.57548952e-01 -7.59099007e-01 -2.48579249e-01 1.75454199e+00
3.21110517e-01 -6.20449960e-01 1.27479649e+00 4.59268570e-01
1.15123592e-01 -1.02907324e+00 -8.43472064e-01 -6.29104137e-01
4.92055248e-03 -3.19366604e-01 7.94245183e-01 1.08897793e+00
-1.91029742e-01 5.58021784e-01 -3.08871567e-01 3.16270798e-01
9.60136354e-01 2.26955548e-01 8.23153734e-01 -1.44116080e+00
-2.51619011e-01 -3.06284279e-01 -2.94571072e-01 -1.01823640e+00
4.09083694e-01 -1.07612121e+00 -1.86499640e-01 -1.39713478e+00
5.22752345e-01 -5.30972183e-01 -1.89203173e-01 8.43096495e-01
-3.80583882e-01 7.24000752e-01 8.19573998e-02 1.40267700e-01
-8.91388118e-01 3.89509201e-01 1.21609426e+00 9.79626626e-02
1.56606644e-01 -1.43206939e-01 -8.52708995e-01 6.81525409e-01
8.00248921e-01 -5.21918654e-01 -6.47473693e-01 -2.07320482e-01
-1.83855053e-02 -1.10455845e-02 2.39469901e-01 -9.12415206e-01
3.79851758e-01 -4.97541726e-02 2.96489239e-01 -4.31393594e-01
3.72950792e-01 -6.80234253e-01 1.83497414e-01 3.29901338e-01
1.18991271e-01 -1.01198621e-01 -6.50036186e-02 6.24246180e-01
-1.81242019e-01 1.45265266e-01 8.65924597e-01 -1.53073803e-01
-6.63512826e-01 8.72575521e-01 3.62500757e-01 3.30213666e-01
9.09237981e-01 -5.44339158e-02 -1.90490007e-01 -3.75417113e-01
-6.18038297e-01 3.70920002e-01 4.65673447e-01 3.04194182e-01
5.61713278e-01 -1.25668871e+00 -6.78501606e-01 1.83924764e-01
-7.70605654e-02 1.82840511e-01 3.03055316e-01 1.01494682e+00
-1.88974231e-01 9.72235110e-03 2.33811587e-01 -7.12713718e-01
-1.67889678e+00 6.43225014e-01 7.19623491e-02 -3.32148254e-01
-4.16364759e-01 1.20319331e+00 6.19429164e-02 -5.34553945e-01
2.38342330e-01 2.94930071e-01 -1.38759360e-01 7.61918202e-02
4.47144657e-01 2.75728971e-01 -7.45064393e-02 -8.39476526e-01
-5.80365181e-01 9.60050941e-01 5.09798713e-02 1.91823646e-01
1.29224586e+00 -2.08392784e-01 -4.99123037e-01 -9.01980251e-02
1.35321736e+00 4.73261215e-02 -1.07471609e+00 -2.26492435e-01
7.48063996e-02 -5.54714203e-01 -1.27737418e-01 -5.83093584e-01
-1.58706915e+00 8.53619397e-01 4.04190689e-01 -3.63410451e-02
1.36725879e+00 3.41821581e-01 8.08580399e-01 1.11715764e-01
4.41112310e-01 -1.01103568e+00 3.34850460e-01 1.38383418e-01
5.96167803e-01 -1.44456625e+00 1.93656206e-01 -1.03938389e+00
-4.53579932e-01 7.29232967e-01 8.40012074e-01 2.48838797e-01
8.82150054e-01 2.88228482e-01 -1.24521084e-01 -6.29036009e-01
-4.87836748e-01 -2.38285482e-01 3.49679947e-01 4.10744160e-01
3.86618018e-01 3.56822833e-02 7.63001293e-03 5.69725275e-01
-8.28876495e-02 -1.27768695e-01 -1.85554162e-01 5.68595409e-01
-2.17936903e-01 -1.35970354e+00 -1.30819723e-01 4.57357943e-01
-5.04381597e-01 3.18181477e-02 -4.52876836e-01 8.15885723e-01
2.16158420e-01 1.32597971e+00 -1.29171252e-01 -5.59416175e-01
-8.14117864e-02 5.87580055e-02 6.24158919e-01 -5.72686613e-01
-2.35546082e-01 2.10994646e-01 1.61049828e-01 -8.36146176e-01
-6.09878480e-01 -6.15893245e-01 -1.45381963e+00 -7.22409904e-01
-5.01786292e-01 3.48226577e-01 3.87282312e-01 8.80246758e-01
5.64279854e-01 1.46587327e-01 7.95007408e-01 -6.21636987e-01
-1.34170741e-01 -8.97992373e-01 -5.68176329e-01 5.41640520e-01
-2.40565643e-01 -5.98113120e-01 -3.08528960e-01 -2.05248557e-02] | [13.458288192749023, 1.003105640411377] |
e6502583-cf5d-4ddf-be3f-631766a15e53 | irb-nlp-at-semeval-2022-task-1-exploring-the | 2205.06840 | null | https://arxiv.org/abs/2205.06840v1 | https://arxiv.org/pdf/2205.06840v1.pdf | IRB-NLP at SemEval-2022 Task 1: Exploring the Relationship Between Words and Their Semantic Representations | What is the relation between a word and its description, or a word and its embedding? Both descriptions and embeddings are semantic representations of words. But, what information from the original word remains in these representations? Or more importantly, which information about a word do these two representations share? Definition Modeling and Reverse Dictionary are two opposite learning tasks that address these questions. The goal of the Definition Modeling task is to investigate the power of information laying inside a word embedding to express the meaning of the word in a humanly understandable way -- as a dictionary definition. Conversely, the Reverse Dictionary task explores the ability to predict word embeddings directly from its definition. In this paper, by tackling these two tasks, we are exploring the relationship between words and their semantic representations. We present our findings based on the descriptive, exploratory, and predictive data analysis conducted on the CODWOE dataset. We give a detailed overview of the systems that we designed for Definition Modeling and Reverse Dictionary tasks, and that achieved top scores on SemEval-2022 CODWOE challenge in several subtasks. We hope that our experimental results concerning the predictive models and the data analyses we provide will prove useful in future explorations of word representations and their relationships. | ['Ivan Grubišić', 'Damir Korenčić'] | 2022-05-13 | null | https://aclanthology.org/2022.semeval-1.5 | https://aclanthology.org/2022.semeval-1.5.pdf | semeval-naacl-2022-7 | ['reverse-dictionary'] | ['natural-language-processing'] | [ 3.88632715e-02 9.04209092e-02 -4.52660143e-01 -4.09014970e-01
-1.35365874e-01 -5.63702524e-01 9.23007071e-01 5.66174805e-01
-6.46979690e-01 2.51432061e-01 9.73728240e-01 -4.32659000e-01
-1.65211305e-01 -8.82521212e-01 -2.07569063e-01 -3.42180401e-01
8.00827984e-03 4.55309272e-01 -3.02943766e-01 -6.09294593e-01
2.73712903e-01 1.24267153e-01 -1.67966163e+00 4.91112143e-01
2.72074670e-01 8.63564789e-01 3.62143904e-01 2.34484181e-01
-5.44326901e-01 6.04235232e-01 -3.54011744e-01 -1.71013281e-01
1.37592368e-02 -1.66177601e-01 -8.57994258e-01 -3.55377495e-02
3.75403851e-01 5.83531857e-02 -3.98645014e-01 8.06744993e-01
2.38827348e-01 2.88884610e-01 9.09098566e-01 -1.17525947e+00
-1.31965220e+00 7.17639804e-01 2.97693051e-02 3.43281150e-01
5.33113062e-01 -2.63986550e-02 1.60001171e+00 -1.21552360e+00
6.95199490e-01 1.32084048e+00 8.55542719e-01 6.50458932e-01
-1.28336644e+00 -5.75213015e-01 2.90480077e-01 1.35293633e-01
-1.46672368e+00 -4.72645164e-01 5.08258164e-01 -7.69344866e-01
1.48832381e+00 1.18266605e-01 7.79357314e-01 1.23594713e+00
1.22733623e-01 6.00761533e-01 9.80250895e-01 -5.86876929e-01
-7.25456402e-02 2.91666389e-01 7.25963116e-01 6.39338851e-01
3.91635537e-01 3.54920417e-01 -6.26978397e-01 -9.72092152e-02
4.70973402e-01 -1.59930903e-02 -3.68807465e-01 -1.36245161e-01
-1.16107202e+00 1.17350543e+00 3.91647995e-01 5.44682443e-01
-4.13895160e-01 4.26085502e-01 5.17410457e-01 2.69647181e-01
5.93844235e-01 1.02822256e+00 -6.14491880e-01 -2.06051156e-01
-5.54751337e-01 2.32668877e-01 8.85809839e-01 8.77155066e-01
7.40842164e-01 -1.12440296e-01 -5.26068266e-03 9.21831012e-01
5.60176849e-01 3.01897734e-01 8.64392042e-01 -3.19029033e-01
1.19099185e-01 4.83818233e-01 -5.35371639e-02 -1.21583498e+00
-3.92553329e-01 -2.82631487e-01 -2.59023279e-01 -2.95817077e-01
-1.02096021e-01 -1.92150183e-03 -1.09823442e+00 2.01808858e+00
-2.18533367e-01 1.20906353e-01 1.02648444e-01 6.08415782e-01
1.23643136e+00 7.20061123e-01 5.08434653e-01 1.99908242e-01
1.75321865e+00 -7.05293775e-01 -9.98117149e-01 -7.50990689e-01
1.13530529e+00 -6.17158532e-01 1.19001329e+00 -6.49448708e-02
-5.31735182e-01 -5.84622085e-01 -1.17560208e+00 -4.40517455e-01
-1.04259050e+00 -1.06954686e-01 5.80820501e-01 5.94263196e-01
-8.57756495e-01 4.29452151e-01 -5.30583262e-01 -7.01759100e-01
2.13092625e-01 -3.40580544e-03 -5.60250700e-01 -2.10553348e-01
-1.63744843e+00 1.54966915e+00 6.93365872e-01 -4.25628483e-01
-4.99916434e-01 -8.90096009e-01 -1.28426909e+00 8.72923657e-02
-6.61599115e-02 -6.27601862e-01 9.47816193e-01 -7.87008703e-01
-4.85666335e-01 9.81500208e-01 -2.99792737e-01 -4.52745318e-01
-5.36714017e-01 -1.08463183e-01 -6.49578929e-01 -3.37563902e-01
1.87412664e-01 5.99139035e-01 4.62783247e-01 -1.34406936e+00
-5.14969528e-01 -2.98686057e-01 1.89279243e-01 7.97904283e-02
-5.74271142e-01 -1.40345380e-01 -1.31981730e-01 -9.61583912e-01
2.54837163e-02 -8.00989628e-01 -3.17846909e-02 3.49552855e-02
5.65388985e-02 -5.35594642e-01 5.34465909e-01 -6.68992102e-01
1.78011668e+00 -2.40893865e+00 9.21938568e-02 1.47455171e-01
4.03076231e-01 2.96123445e-01 -4.42054778e-01 9.55700278e-01
-3.80044281e-01 5.68103313e-01 -6.28768653e-02 -4.05417323e-01
3.72721225e-01 6.21204257e-01 -6.72326744e-01 3.19350690e-01
1.32453039e-01 1.01164591e+00 -1.03891742e+00 -1.15186960e-01
2.17467621e-01 6.38977170e-01 -5.06889343e-01 1.30922675e-01
-2.62471914e-01 -4.30767447e-01 -2.75474817e-01 2.39054725e-01
1.20417885e-01 -2.05032434e-02 2.56232649e-01 -5.29587269e-01
1.28287330e-01 8.41693401e-01 -8.09360743e-01 1.57159710e+00
-8.57532382e-01 1.06556535e+00 -4.13976699e-01 -1.00928605e+00
1.05761886e+00 2.79596180e-01 7.28768781e-02 -7.97994435e-01
-1.68391109e-01 2.83522874e-01 5.02941757e-02 -5.48012793e-01
8.25940967e-01 -7.65593290e-01 -2.52517462e-01 5.95728815e-01
2.21300423e-01 -9.36879590e-02 1.37343064e-01 1.48180068e-01
1.06166208e+00 -1.24574371e-01 7.30751455e-01 -4.93243933e-01
1.04781687e-01 6.14874288e-02 2.80120969e-01 4.84465510e-01
6.96029887e-02 3.55479777e-01 3.26947212e-01 -7.33272791e-01
-9.37110782e-01 -1.10925996e+00 -4.69580680e-01 1.09564388e+00
4.69824933e-02 -1.25939250e+00 8.04192573e-02 -4.07357126e-01
3.20478082e-01 1.41134810e+00 -1.10136461e+00 -4.84491110e-01
-2.69776136e-01 -5.55267870e-01 5.73070049e-01 7.62472451e-01
-8.06865096e-02 -1.02652323e+00 -4.63390708e-01 2.83434302e-01
-8.31120238e-02 -8.62880349e-01 -4.98319864e-01 4.36804622e-01
-5.40148914e-01 -9.51870203e-01 7.57049350e-03 -9.51605976e-01
4.46076423e-01 2.22960845e-01 1.69323623e+00 1.73278585e-01
-2.53604263e-01 6.99251533e-01 -7.23692358e-01 -5.51330328e-01
-4.34454203e-01 -3.70850265e-02 8.03900063e-02 -3.62911850e-01
1.21580637e+00 -4.77801859e-01 -3.12943846e-01 3.59224598e-03
-1.12746036e+00 -1.06442593e-01 3.02829504e-01 7.91287601e-01
4.53290969e-01 -1.58666641e-01 5.54945886e-01 -9.37479854e-01
1.06816590e+00 -9.14862156e-01 2.15522289e-01 2.67832607e-01
-8.40787351e-01 4.67255414e-01 2.22320274e-01 -3.81096989e-01
-3.92836899e-01 -2.42606804e-01 -3.12882990e-01 -2.17963964e-01
2.15383276e-01 9.86710608e-01 6.46106154e-02 5.71544170e-01
7.33573675e-01 3.17854643e-01 4.82686497e-02 -6.32635474e-01
7.71668315e-01 7.45568335e-01 1.37161463e-01 -7.24826515e-01
5.95204890e-01 2.00344011e-01 -4.38799620e-01 -1.04844868e+00
-1.17710412e+00 -7.52834439e-01 -5.51659465e-01 2.77913928e-01
1.10739315e+00 -9.87693429e-01 8.35696533e-02 -2.01421991e-01
-1.30061197e+00 1.40120536e-01 -7.20424056e-01 2.26633236e-01
-3.46028060e-01 9.81942937e-02 -1.31247625e-01 -4.97755170e-01
-1.96492687e-01 -8.69411945e-01 8.90559137e-01 -2.45981157e-01
-9.48159695e-01 -1.66671777e+00 3.27992350e-01 4.59803082e-02
6.36137068e-01 1.33070320e-01 1.48303068e+00 -1.14286757e+00
1.14044614e-01 -2.94151932e-01 -1.38808593e-01 4.64942545e-01
4.58759993e-01 -4.31566864e-01 -7.05818892e-01 -3.71453851e-01
-1.09890200e-01 -3.28839839e-01 1.11961389e+00 -3.23979855e-02
9.20591235e-01 -3.12554628e-01 -3.31165820e-01 5.03618896e-01
1.56878626e+00 7.71868154e-02 3.77614081e-01 3.44321936e-01
5.74392378e-01 7.32355654e-01 4.32507008e-01 3.57704699e-01
4.33399260e-01 6.49823308e-01 1.90620869e-01 5.23853190e-02
-4.36059296e-01 -6.45595968e-01 4.16326821e-01 1.07295477e+00
3.79127592e-01 -2.22695202e-01 -1.16701412e+00 9.70315218e-01
-1.57164180e+00 -8.24786901e-01 2.45399363e-02 1.81659162e+00
8.98441374e-01 -1.02376170e-01 -2.65863061e-01 -9.03498828e-02
4.85170186e-01 6.87735498e-01 -2.02988878e-01 -8.61733615e-01
-2.52942201e-02 6.00967646e-01 1.27929911e-01 5.57275236e-01
-9.08794940e-01 9.44270074e-01 6.96244192e+00 8.56984556e-01
-7.76805997e-01 2.01694697e-01 4.67620082e-02 7.13529885e-02
-8.73079956e-01 1.12714335e-01 -7.73904741e-01 2.52529949e-01
1.16459870e+00 -5.56907594e-01 2.25715056e-01 6.42495632e-01
-7.39340335e-02 2.60370016e-01 -1.66491687e+00 9.80506182e-01
4.28606123e-01 -1.43742168e+00 6.79692984e-01 1.26021370e-01
2.31544957e-01 3.15197557e-02 8.48118588e-02 5.72424471e-01
3.03021520e-01 -1.61054349e+00 6.41568959e-01 4.71196830e-01
8.19537222e-01 -4.60690439e-01 6.71804070e-01 1.32111117e-01
-1.27821517e+00 -5.55248698e-03 -4.52643484e-01 -3.13209891e-01
7.09481388e-02 3.50704879e-01 -6.78537786e-01 1.62206665e-01
7.37820417e-02 1.18203437e+00 -6.02263629e-01 4.09306824e-01
-5.27323961e-01 4.84652728e-01 2.41170488e-02 -1.39915168e-01
4.25494343e-01 1.43409492e-02 4.36418444e-01 1.63648319e+00
2.08960950e-01 9.55296606e-02 1.73857093e-01 1.21383739e+00
-1.42479375e-01 8.68539419e-03 -8.85339737e-01 -6.17313564e-01
7.23695457e-01 8.78528774e-01 -1.58978641e-01 -1.78698033e-01
-5.69131434e-01 5.24631441e-01 5.10530770e-01 1.80266678e-01
-5.48143864e-01 -3.23736846e-01 1.39948225e+00 3.66019845e-01
1.47142559e-01 -5.21636128e-01 -2.15920016e-01 -1.14197397e+00
-2.56761104e-01 -6.17686868e-01 2.84663230e-01 -8.76972258e-01
-1.64819288e+00 5.61629593e-01 1.43365264e-01 -8.18977594e-01
-3.14762175e-01 -9.36435580e-01 -5.05119801e-01 1.07351255e+00
-1.37145853e+00 -1.02425992e+00 -1.55298514e-02 3.22648585e-01
5.81042826e-01 -3.04452598e-01 1.35497427e+00 1.48254111e-01
-1.09380677e-01 4.10820991e-01 5.08663431e-02 3.84131461e-01
4.38529402e-01 -1.18782687e+00 5.72092474e-01 4.22490776e-01
7.50497818e-01 1.08176827e+00 7.76471376e-01 -5.12376189e-01
-1.32790422e+00 -9.88914847e-01 1.32293034e+00 -8.06363940e-01
9.99296188e-01 -4.46976453e-01 -9.54637825e-01 9.90384221e-01
2.45289415e-01 -1.05508873e-02 1.25924742e+00 6.06765866e-01
-8.97585034e-01 4.63126719e-01 -6.41908824e-01 3.96858364e-01
9.76344228e-01 -1.07524109e+00 -1.56716180e+00 1.60842434e-01
1.10522926e+00 2.63008654e-01 -9.78308797e-01 5.88558652e-02
6.12860143e-01 -5.53437293e-01 1.09891558e+00 -1.16933107e+00
6.72222137e-01 -2.84619182e-02 -7.46634126e-01 -1.71943331e+00
-4.92757082e-01 8.88399035e-02 -4.92098629e-02 1.16527677e+00
7.02295363e-01 -5.31584859e-01 3.20925295e-01 5.05831540e-01
-6.64489642e-02 -9.81220603e-01 -8.68063509e-01 -8.83006036e-01
6.14068627e-01 -8.42093945e-01 5.92827320e-01 1.33105254e+00
3.14020999e-02 7.42893100e-01 -1.10958489e-02 -1.79205582e-01
2.01456964e-01 -3.66262607e-02 3.09814513e-01 -1.10107148e+00
-8.67825970e-02 -2.67180055e-01 -7.33071446e-01 -1.11368489e+00
5.23763239e-01 -1.53044927e+00 -2.99080938e-01 -1.84108710e+00
1.36897653e-01 -4.47409719e-01 -3.66352051e-01 5.78756630e-01
-1.39672056e-01 -1.22923747e-01 3.59996408e-01 1.65570721e-01
-2.48999998e-01 7.22806752e-01 7.68308103e-01 -2.09949523e-01
1.73835814e-01 -6.91032052e-01 -9.83973742e-01 6.05845928e-01
5.85125685e-01 -4.93665427e-01 -6.24343991e-01 -8.59574735e-01
4.71761703e-01 -4.89331514e-01 2.10568219e-01 -4.67478096e-01
7.58433938e-02 -6.53621033e-02 2.58818474e-02 -2.03268185e-01
4.47588891e-01 -8.75347078e-01 -3.61438319e-02 3.97096813e-01
-6.83216035e-01 3.88101220e-01 2.78030246e-01 4.97941464e-01
-4.12631214e-01 -4.55916017e-01 6.04334652e-01 -1.37680143e-01
-1.23207605e+00 1.42063260e-01 -4.88925606e-01 6.02817714e-01
7.13806510e-01 -2.48333469e-01 -1.75072268e-01 -3.51236135e-01
-9.42284942e-01 1.27546474e-01 3.55741590e-01 9.55256939e-01
8.18536699e-01 -1.74178219e+00 -8.11064005e-01 2.85586745e-01
7.63174713e-01 -5.44463098e-01 -4.42086488e-01 2.25738510e-01
-1.09257527e-01 4.28676546e-01 -1.41542047e-01 -6.75226673e-02
-1.04146743e+00 6.44452035e-01 4.28155780e-01 -1.90853760e-01
-6.00009561e-01 7.49531627e-01 3.78696740e-01 -4.94074613e-01
-7.52721876e-02 -3.89035493e-01 -3.77466530e-01 4.29862738e-01
6.26026869e-01 1.22972326e-02 -2.92940944e-01 -9.54815626e-01
-5.08755744e-01 4.75073159e-01 -9.34051871e-02 4.51514032e-03
1.48705292e+00 -1.36473075e-01 -2.59353489e-01 7.67366529e-01
1.76520240e+00 -2.34701857e-01 -3.75713944e-01 -5.66394031e-01
3.87488514e-01 -5.13913453e-01 3.30021590e-01 -8.80319655e-01
-7.34767258e-01 9.65856493e-01 4.15364534e-01 2.56079346e-01
7.20039964e-01 3.05717647e-01 8.09072852e-01 2.33805358e-01
1.81231469e-01 -8.50920379e-01 -4.27353643e-02 8.55811954e-01
1.26118600e+00 -1.05525732e+00 2.13036314e-02 -2.33963162e-01
-4.88898546e-01 1.33802414e+00 3.68709385e-01 -1.95771664e-01
9.27435100e-01 1.29387677e-01 -1.88021556e-01 -7.02316165e-01
-1.05069709e+00 -4.68642265e-01 5.58155835e-01 7.15756714e-01
8.64342749e-01 2.16424942e-01 -4.13693875e-01 9.88424599e-01
-4.31784600e-01 -2.63629943e-01 3.55332881e-01 7.70972610e-01
-5.30328453e-01 -1.13805366e+00 8.27343687e-02 6.50305688e-01
-4.83981259e-02 -6.51366591e-01 -6.73844755e-01 1.04537594e+00
3.50085437e-01 9.43615019e-01 4.17655528e-01 -7.83355832e-01
4.60361600e-01 3.02035123e-01 2.30044991e-01 -1.25539577e+00
-4.93332922e-01 -6.59268320e-01 5.73895991e-01 -4.69278514e-01
-2.95596778e-01 -5.54924369e-01 -1.35027146e+00 -2.86370009e-01
-9.77713764e-02 3.36194873e-01 6.90535724e-01 1.11095738e+00
2.23666504e-01 4.73636746e-01 1.79134727e-01 -3.03961545e-01
-5.16065300e-01 -9.77569461e-01 -7.56583154e-01 6.50019348e-01
2.64821619e-01 -7.23602116e-01 -6.15891218e-01 -1.43088892e-01] | [10.400944709777832, 8.846542358398438] |
4ef70061-9b9a-4a80-828f-c7b3ded19081 | no-one-left-behind-real-world-federated-class | 2302.00903 | null | https://arxiv.org/abs/2302.00903v1 | https://arxiv.org/pdf/2302.00903v1.pdf | No One Left Behind: Real-World Federated Class-Incremental Learning | Federated learning (FL) is a hot collaborative training framework via aggregating model parameters of decentralized local clients. However, most existing models unreasonably assume that data categories of FL framework are known and fxed in advance. It renders the global model to signifcantly degrade recognition performance on old categories (i.e., catastrophic forgetting), when local clients receive new categories consecutively under limited memory of storing old categories. Moreover, some new local clients that collect novel categories unseen by other clients may be introduced to the FL training irregularly, which further exacerbates the catastrophic forgetting on old categories. To tackle the above issues, we propose a novel Local-Global Anti-forgetting (LGA) model to address local and global catastrophic forgetting on old categories, which is a pioneering work to explore a global class-incremental model in the FL feld. Specifcally, considering tackling class imbalance of local client to surmount local forgetting, we develop a category-balanced gradient-adaptive compensation loss and a category gradient-induced semantic distillation loss. They can balance heterogeneous forgetting speeds of hard-to-forget and easy-to-forget old categories, while ensure intrinsic class relations consistency within different incremental tasks. Moreover, a proxy server is designed to tackle global forgetting caused by Non-IID class imbalance between different clients. It collects perturbed prototype images of new categories from local clients via prototype gradient communication under privacy preservation, and augments them via self-supervised prototype augmentation to choose the best old global model and improve local distillation gain. Experiments on representative datasets verify superior performance of our model against other comparison methods. | ['Dengxin Dai', 'Bernt Schiele', 'Yulun Zhang', 'Gan Sun', 'Yang Cong', 'Jiahua Dong'] | 2023-02-02 | null | null | null | null | ['class-incremental-learning'] | ['computer-vision'] | [-2.79937357e-01 -5.77672720e-02 -2.84519196e-01 -5.90172231e-01
-5.16471863e-01 -6.17586732e-01 3.23495418e-01 -2.48730853e-01
-5.02658665e-01 9.87249553e-01 5.93278818e-02 -1.12967044e-01
-4.78209108e-02 -8.32593679e-01 -7.93728709e-01 -1.03910470e+00
2.22278282e-01 5.73903978e-01 3.15773100e-01 1.81713089e-01
-9.81026590e-02 3.26101780e-01 -1.69049203e+00 5.35698116e-01
1.05280530e+00 1.20054030e+00 5.90764955e-02 3.01268578e-01
-2.82250583e-01 8.65733981e-01 -6.18461370e-01 -5.67095816e-01
2.87742287e-01 -1.08617902e-01 -5.48019469e-01 -7.20817745e-02
5.56977749e-01 -6.85816646e-01 -3.65728825e-01 9.28437829e-01
5.06283164e-01 4.87924069e-02 2.82587171e-01 -1.51899230e+00
-9.92989302e-01 7.16225386e-01 -3.20300102e-01 2.50667363e-01
-2.43252054e-01 4.80256945e-01 6.26080275e-01 -1.21904373e+00
5.29977381e-01 1.20181477e+00 9.59947288e-01 8.41309249e-01
-9.67156947e-01 -1.11817408e+00 7.69117594e-01 4.60998535e-01
-1.42992771e+00 -4.41899240e-01 6.24239564e-01 -7.74681941e-02
5.28965890e-01 3.80410343e-01 3.05797011e-01 1.14687133e+00
8.79324302e-02 8.35077822e-01 1.11468089e+00 1.89039186e-01
5.88656843e-01 5.43863714e-01 1.44586697e-01 4.41449523e-01
4.73713577e-01 -1.46570290e-02 -6.43705547e-01 -5.32187045e-01
4.15585130e-01 7.70124376e-01 -3.55811059e-01 -7.34691978e-01
-1.05125809e+00 5.65888345e-01 6.08286500e-01 1.25042766e-01
-2.09366977e-01 -3.98404561e-02 5.34288168e-01 8.57453227e-01
5.08896708e-01 -2.97326654e-01 -9.92947519e-01 3.23705643e-01
-9.26851273e-01 8.01283419e-02 7.98565149e-01 1.02440941e+00
1.20756912e+00 -7.30884597e-02 -2.93139666e-01 6.79727733e-01
1.10541932e-01 6.66282237e-01 1.09250319e+00 -6.29677951e-01
4.91813242e-01 7.20180988e-01 -1.16931871e-01 -8.43022287e-01
-3.83417718e-02 -7.80257165e-01 -1.01918554e+00 8.39976296e-02
7.62305111e-02 1.44368643e-02 -8.33464384e-01 1.86817479e+00
6.73983514e-01 4.23050195e-01 4.06063758e-02 9.25258219e-01
3.53437215e-01 1.67793706e-01 1.26264289e-01 -5.26773453e-01
8.78022254e-01 -1.09182167e+00 -5.42955220e-01 -1.37122367e-02
2.96520650e-01 -4.24411982e-01 1.35962212e+00 3.19515884e-01
-7.15784550e-01 -4.93444055e-01 -8.44379008e-01 9.60382447e-02
-4.30347294e-01 -9.31123495e-02 5.36027670e-01 5.28892994e-01
-9.78733838e-01 5.27833998e-01 -6.88578427e-01 -3.23885530e-01
9.39310789e-01 4.40925479e-01 -2.38678381e-01 -3.45114112e-01
-9.77064073e-01 3.57168376e-01 2.54245877e-01 -4.39420342e-02
-1.31334364e+00 -1.01808202e+00 -1.27934784e-01 1.58606544e-02
5.11752963e-01 -9.96215284e-01 1.32202280e+00 -1.12435019e+00
-1.19385564e+00 5.65974414e-01 5.71832620e-02 -4.74716336e-01
1.15685332e+00 -2.12439403e-01 -6.37902498e-01 -1.78573936e-01
1.22485138e-01 1.53075591e-01 1.29118943e+00 -1.32295120e+00
-8.89847934e-01 -6.20632887e-01 -2.03809604e-01 2.85126269e-01
-8.45923901e-01 -7.68620968e-01 -1.83148861e-01 -7.93343842e-01
1.97417080e-01 -5.08544087e-01 1.05121788e-02 4.13344562e-01
-8.12432617e-02 -1.44693986e-01 1.49031556e+00 -2.86451936e-01
1.14040542e+00 -2.26504016e+00 -2.14405611e-01 4.98796441e-02
4.46955919e-01 2.46502131e-01 -1.86729744e-01 -1.22347078e-03
3.12785715e-01 -5.50906993e-02 -2.64562726e-01 -6.17183447e-01
-2.11446658e-01 6.98620200e-01 -8.20108652e-01 4.03394163e-01
-2.60889173e-01 7.68853903e-01 -9.17111695e-01 -3.42950732e-01
-1.57179862e-01 4.19899195e-01 -8.28965008e-01 4.19307172e-01
-2.59251535e-01 3.53778362e-01 -3.25680405e-01 9.93214786e-01
1.18368614e+00 -2.00058892e-01 1.02291219e-01 -9.78837609e-02
2.45316505e-01 -5.76880239e-02 -1.24103868e+00 1.53604317e+00
-5.51920295e-01 -3.66025925e-01 3.27844441e-01 -8.56000960e-01
7.05935597e-01 9.11757722e-02 6.57631308e-02 -3.29949737e-01
-2.45060734e-02 4.56649095e-01 -7.55493999e-01 -2.28448883e-01
2.49715745e-01 -2.12672159e-01 1.49952909e-02 7.03082502e-01
2.19415188e-01 5.10124326e-01 -5.02416849e-01 2.21072733e-01
1.10381162e+00 -1.85929403e-01 -2.08547398e-01 -1.75709248e-01
6.00212514e-01 -3.30301195e-01 9.32269812e-01 1.13578701e+00
-4.35746133e-01 6.38609409e-01 -1.22711577e-01 -8.39831769e-01
-7.02982843e-01 -1.42758656e+00 1.54844284e-01 1.48132765e+00
2.91172713e-01 2.89131943e-02 -4.39958155e-01 -1.34541607e+00
5.08373439e-01 4.00391161e-01 -5.01934588e-01 -6.06166542e-01
-4.46533531e-01 -8.09365809e-01 4.04819012e-01 2.84017950e-01
8.96438479e-01 -8.70253742e-01 -2.77426124e-01 4.21553522e-01
-2.09509328e-01 -4.31558222e-01 -8.42250943e-01 1.39321819e-01
-1.11728299e+00 -1.21826875e+00 -6.58377469e-01 -5.63143730e-01
7.75254786e-01 5.90108097e-01 8.48825634e-01 1.35323912e-01
-2.01350406e-01 4.79533494e-01 -1.65303394e-01 -2.99363062e-02
-9.85079482e-02 3.18642527e-01 5.63039064e-01 4.85328287e-01
3.98415089e-01 -1.02041090e+00 -8.89837086e-01 5.90896666e-01
-1.00630689e+00 -3.53515476e-01 6.58253133e-01 1.03356349e+00
6.23902738e-01 -1.58546492e-02 8.81134331e-01 -1.20776570e+00
5.55322170e-01 -9.34950531e-01 -1.60123631e-01 5.44717193e-01
-1.23951077e+00 -1.49513900e-01 1.01221466e+00 -1.00176835e+00
-1.32496202e+00 -2.34925270e-01 3.06552708e-01 -9.95030642e-01
1.01047754e-01 -5.68512604e-02 -4.43121761e-01 -1.61307514e-01
5.09806514e-01 5.33339322e-01 4.09610569e-02 -8.32014024e-01
7.17362463e-01 8.84529710e-01 8.45898926e-01 -6.43777966e-01
9.31731284e-01 7.66395092e-01 -6.89637959e-01 1.69694331e-02
-7.21531868e-01 -3.28658640e-01 -3.24052304e-01 -6.88535720e-03
3.16888541e-02 -1.19688857e+00 -7.46511400e-01 9.13986087e-01
-7.78437018e-01 -3.18976343e-01 -7.90761530e-01 5.59331514e-02
-3.67350280e-01 3.40759695e-01 -6.71832502e-01 -6.49392426e-01
-7.32535064e-01 -6.01872087e-01 7.12742627e-01 3.49001378e-01
3.78902525e-01 -6.43387675e-01 -3.10056098e-02 2.39865109e-01
1.00412512e+00 -2.08232686e-01 7.22225606e-01 -6.93294466e-01
-7.58001328e-01 -5.55619299e-02 -2.64277935e-01 6.42001331e-01
1.77151933e-01 -5.80654979e-01 -1.20468867e+00 -8.92991960e-01
3.93971235e-01 -3.08644325e-01 1.02315998e+00 -4.58397478e-01
1.26981366e+00 -1.19249582e+00 -4.06685948e-01 8.33643854e-01
1.48589671e+00 -4.54403982e-02 1.29272416e-01 2.34632879e-01
7.43119478e-01 3.90107632e-02 5.09652734e-01 7.82097399e-01
5.30129015e-01 1.90899611e-01 4.34767187e-01 2.97804415e-01
-4.20355111e-01 -5.89644670e-01 2.93382436e-01 1.00925708e+00
4.42652494e-01 1.27678096e-01 -4.49985683e-01 5.92338324e-01
-2.01319408e+00 -7.82769561e-01 4.79114473e-01 2.33438969e+00
1.06602871e+00 -1.25670567e-01 -1.19220495e-01 -1.07172936e-01
8.48694205e-01 1.68007128e-02 -1.20854568e+00 7.61812180e-02
-3.18327427e-01 -4.63931151e-02 6.25786066e-01 1.60492152e-01
-8.84367883e-01 7.79673219e-01 5.08630896e+00 9.35871243e-01
-1.17142475e+00 9.15463030e-01 8.97802949e-01 -4.09030437e-01
-5.07462800e-01 5.45571893e-02 -9.81593728e-01 8.50685894e-01
6.23869300e-01 -3.60650450e-01 6.65198982e-01 1.43305099e+00
-3.34067106e-01 2.69642323e-01 -8.39675784e-01 9.47285831e-01
7.93412328e-02 -1.16919804e+00 2.89971203e-01 -2.14035466e-01
8.97916794e-01 2.17743888e-01 3.37526530e-01 7.19598949e-01
5.42896688e-01 -3.67104679e-01 8.94426048e-01 7.96833515e-01
9.49348688e-01 -7.09133625e-01 7.23787785e-01 4.63535756e-01
-1.10093844e+00 -7.67997742e-01 -6.49672687e-01 1.12353608e-01
-3.13663006e-01 9.20923948e-01 -5.86968422e-01 4.77240801e-01
1.16485274e+00 3.23302597e-01 -7.16001034e-01 1.01086247e+00
6.98233694e-02 4.96343881e-01 -5.32970011e-01 5.04019022e-01
-2.34603316e-01 1.36191532e-01 3.91964704e-01 7.65373707e-01
3.23970556e-01 5.26872277e-02 2.50293702e-01 6.36721551e-01
-5.02127290e-01 1.20446151e-02 -3.75976503e-01 5.24910748e-01
9.77468967e-01 1.11729813e+00 -2.51313210e-01 -5.13307154e-01
-2.44330972e-01 1.28421462e+00 6.91450179e-01 4.92451072e-01
-6.84688509e-01 -3.21097940e-01 8.98941815e-01 1.95050672e-01
2.76272148e-01 1.46264985e-01 -2.55752355e-01 -1.64749396e+00
3.54903102e-01 -6.66326165e-01 1.02698004e+00 -3.18003774e-01
-1.97441447e+00 5.64811170e-01 -4.65464234e-01 -1.15178025e+00
3.12137663e-01 1.49640977e-01 -8.09259355e-01 4.42056179e-01
-1.66544843e+00 -1.27906907e+00 -4.82693464e-01 1.12041891e+00
3.46238375e-01 -2.84600794e-01 5.04792690e-01 6.50309861e-01
-4.66311574e-01 1.31639063e+00 5.22098958e-01 -4.62059736e-01
1.10577488e+00 -8.17049742e-01 -7.45488182e-02 7.95541644e-01
-2.82632232e-01 7.11004436e-01 2.98466951e-01 -8.00927937e-01
-1.41919136e+00 -1.79709721e+00 8.66052985e-01 -4.05753106e-01
4.93181735e-01 -3.92355144e-01 -1.25331080e+00 8.19614649e-01
-3.68050963e-01 4.24331129e-01 3.73701990e-01 -2.08679974e-01
-9.78565276e-01 -9.07787681e-01 -1.74798894e+00 2.55417466e-01
1.30262458e+00 -7.12332308e-01 -4.67785805e-01 4.54358071e-01
1.07389057e+00 -5.50653040e-02 -4.98984873e-01 4.15900797e-01
3.23040873e-01 -1.01742625e+00 8.85335863e-01 -4.24312234e-01
-3.88007283e-01 -4.61554766e-01 -7.96689317e-02 -9.28276598e-01
-3.08961779e-01 -8.14026415e-01 -5.94363868e-01 1.54193604e+00
-1.71407133e-01 -1.19588029e+00 1.08509195e+00 6.63842916e-01
-7.81406090e-02 -5.31937540e-01 -1.41941059e+00 -1.06384528e+00
8.72113034e-02 -1.95569228e-02 1.10648453e+00 1.09390068e+00
-3.64813864e-01 -2.49904782e-01 -4.11045790e-01 1.29689902e-01
8.63107026e-01 2.02472940e-01 8.22742879e-01 -1.12010133e+00
-2.19688147e-01 -2.75525600e-01 -3.73596027e-02 -8.35155427e-01
-9.56822410e-02 -9.86855328e-01 -6.31789565e-02 -8.78317118e-01
3.17388147e-01 -9.41313684e-01 -8.05605173e-01 1.06003582e+00
-2.29137480e-01 2.91809499e-01 1.30102068e-01 9.14347947e-01
-9.43549335e-01 8.25708210e-01 8.89284372e-01 -2.35755444e-01
-2.00712904e-01 1.71339065e-01 -8.52518797e-01 2.74229407e-01
5.82198620e-01 -6.36637688e-01 -5.45722246e-01 -3.84947270e-01
-1.37557819e-01 -3.59307885e-01 4.05060887e-01 -1.20210481e+00
6.79939628e-01 -2.28655353e-01 3.40471119e-01 -3.95476222e-01
-2.02788264e-01 -1.03608716e+00 3.87125134e-01 4.55890745e-01
-7.26144463e-02 -1.47995353e-01 -2.51135588e-01 1.06528032e+00
2.40059160e-02 2.80673474e-01 8.36543500e-01 -2.12949440e-01
-6.10373497e-01 8.33522379e-01 1.09119095e-01 -9.43157375e-02
1.18834186e+00 -3.55094522e-02 -5.51718712e-01 -6.49790391e-02
-8.17749858e-01 4.12652612e-01 7.16917872e-01 4.83771265e-01
5.04640579e-01 -1.42712355e+00 -2.92682528e-01 7.03330100e-01
2.89427079e-02 2.06029966e-01 6.92806542e-01 7.27172017e-01
-5.49398251e-02 -1.97527893e-02 -1.07437391e-02 -3.08168709e-01
-9.76187587e-01 1.09226358e+00 3.22349846e-01 -1.74173176e-01
-5.83379030e-01 1.17286420e+00 4.07663226e-01 -7.88495243e-01
5.62509656e-01 -5.27935438e-02 4.29138899e-01 3.10479626e-02
7.40335882e-01 5.31736910e-01 3.08615535e-01 -2.00956821e-01
-4.15385902e-01 1.69340372e-01 -5.86693645e-01 4.72313136e-01
1.19156039e+00 -6.19608998e-01 -1.92086220e-01 3.56350154e-01
1.21457684e+00 -3.17115754e-01 -1.68743312e+00 -7.41356909e-01
-3.14179808e-01 -6.25071883e-01 -3.53716522e-01 -9.93150234e-01
-1.47180879e+00 5.99139094e-01 1.08337104e+00 -3.07272136e-01
1.40341914e+00 -1.36069328e-01 1.11643779e+00 3.75842214e-01
8.85571599e-01 -1.17668629e+00 1.65833861e-01 2.41010338e-01
5.53524852e-01 -9.76151168e-01 -1.47182912e-01 -1.70830414e-02
-4.98943806e-01 8.81372750e-01 9.19125676e-01 1.18823191e-02
7.85692871e-01 -3.72422598e-02 8.64351094e-02 1.92039520e-01
-1.01114488e+00 3.34051430e-01 -3.81493241e-01 4.78541076e-01
-6.03704870e-01 1.02866881e-01 -1.68766603e-01 1.08030915e+00
1.85775727e-01 3.17472905e-01 1.62612095e-01 1.08879840e+00
-3.77196610e-01 -1.10241270e+00 -2.33024687e-01 5.83583832e-01
-2.43431479e-01 7.59951696e-02 4.06735130e-02 2.53812850e-01
6.26373470e-01 7.15227246e-01 3.62423509e-02 -5.62198162e-01
2.85085946e-01 3.44976783e-01 -1.04280718e-01 -2.94170439e-01
-7.69226432e-01 -3.21025848e-01 -6.89327896e-01 -5.55094063e-01
1.75623745e-01 -4.04745579e-01 -9.78236079e-01 -6.70085669e-01
-2.73644120e-01 2.66763508e-01 2.49093056e-01 6.48071587e-01
8.43633354e-01 1.27803478e-02 1.14374542e+00 -4.85773563e-01
-1.16074908e+00 -7.16810942e-01 -7.46822715e-01 5.22904754e-01
3.72385263e-01 -5.82702935e-01 -9.64078009e-01 -1.16897739e-01] | [5.855233192443848, 6.330670356750488] |
dcff982f-1b50-4873-addb-1d00a62f134f | restore-anything-pipeline-segment-anything | 2305.13093 | null | https://arxiv.org/abs/2305.13093v2 | https://arxiv.org/pdf/2305.13093v2.pdf | Restore Anything Pipeline: Segment Anything Meets Image Restoration | Recent image restoration methods have produced significant advancements using deep learning. However, existing methods tend to treat the whole image as a single entity, failing to account for the distinct objects in the image that exhibit individual texture properties. Existing methods also typically generate a single result, which may not suit the preferences of different users. In this paper, we introduce the Restore Anything Pipeline (RAP), a novel interactive and per-object level image restoration approach that incorporates a controllable model to generate different results that users may choose from. RAP incorporates image segmentation through the recent Segment Anything Model (SAM) into a controllable image restoration model to create a user-friendly pipeline for several image restoration tasks. We demonstrate the versatility of RAP by applying it to three common image restoration tasks: image deblurring, image denoising, and JPEG artifact removal. Our experiments show that RAP produces superior visual results compared to state-of-the-art methods. RAP represents a promising direction for image restoration, providing users with greater control, and enabling image restoration at an object level. | ['Christian Holz', 'Jiaxi Jiang'] | 2023-05-22 | null | null | null | null | ['deblurring', 'jpeg-artifact-removal'] | ['computer-vision', 'computer-vision'] | [ 4.66883272e-01 -5.05647838e-01 2.51654863e-01 -2.97389746e-01
-9.35533047e-01 -4.86211032e-01 3.53175312e-01 -1.18274957e-01
-3.59528661e-02 1.94855452e-01 3.07046682e-01 -3.63757908e-01
1.11308441e-01 -7.16170609e-01 -7.62914419e-01 -7.78644800e-01
2.31002524e-01 8.65480956e-03 1.85174540e-01 -3.00270438e-01
4.89950657e-01 5.79342246e-01 -1.57123089e+00 7.04442859e-01
1.20343912e+00 6.09777212e-01 4.86309856e-01 6.79253340e-01
8.46760571e-02 6.87709689e-01 -7.34056592e-01 -7.90936779e-03
3.22209567e-01 -4.31182384e-01 -6.88094318e-01 5.91827810e-01
9.77471411e-01 -7.00129747e-01 -1.20003730e-01 1.22413516e+00
6.33981109e-01 1.49336368e-01 3.76784652e-01 -9.86368001e-01
-1.18375719e+00 3.76921177e-01 -5.87462902e-01 1.02527007e-01
3.95377964e-01 4.99453574e-01 5.59882879e-01 -7.32214034e-01
5.66806734e-01 1.43101931e+00 8.04827571e-01 3.39073598e-01
-1.78897381e+00 -3.94194692e-01 1.95028423e-03 3.12574625e-01
-1.00635123e+00 -5.03082097e-01 8.06246758e-01 -4.74829346e-01
8.32015514e-01 4.18903351e-01 5.48326790e-01 7.68373013e-01
3.18795264e-01 8.50001454e-01 1.42590272e+00 -3.39727640e-01
2.93880582e-01 -2.75258183e-01 8.52486715e-02 2.13012904e-01
-3.61546017e-02 -1.80883333e-02 -4.39621776e-01 1.65788978e-01
9.39795613e-01 1.67290159e-02 -6.74291909e-01 -2.77815610e-01
-1.25100851e+00 4.30746824e-01 5.30781090e-01 6.34288043e-02
-4.35844630e-01 8.09348375e-02 1.49914771e-01 4.14404720e-01
5.74341059e-01 4.18147475e-01 -3.34122419e-01 3.96975398e-01
-1.14731312e+00 4.16247517e-01 5.97367346e-01 6.35661721e-01
8.10176194e-01 1.55079499e-01 -4.17504787e-01 1.04285681e+00
2.19352126e-01 2.30590105e-01 3.02509099e-01 -1.34706306e+00
-1.20503582e-01 3.77759814e-01 2.76135504e-01 -9.59382594e-01
-8.75591412e-02 -3.54004592e-01 -8.85202289e-01 1.05183637e+00
2.57326543e-01 2.68938720e-01 -1.31003284e+00 1.25373971e+00
1.97491378e-01 2.40699165e-02 -8.99584070e-02 1.06665337e+00
9.86059129e-01 8.47273350e-01 1.08232751e-01 9.86932144e-02
1.19702244e+00 -1.14304543e+00 -7.75885224e-01 -2.37741262e-01
7.71165937e-02 -1.08407259e+00 1.43498456e+00 8.43657017e-01
-1.25788331e+00 -8.01345885e-01 -1.07172537e+00 -3.79085124e-01
-1.58877879e-01 5.96679822e-02 3.79580528e-01 6.43144488e-01
-1.56961143e+00 9.60630298e-01 -5.62379837e-01 -2.14103326e-01
6.44395053e-01 3.33564103e-01 -3.63144338e-01 -4.65932667e-01
-6.50653362e-01 7.00160146e-01 1.12691030e-01 7.81198889e-02
-1.00842273e+00 -9.16135132e-01 -8.70533764e-01 5.95707372e-02
9.07681957e-02 -8.06728005e-01 1.17621768e+00 -1.17456639e+00
-1.56258905e+00 8.60495389e-01 -1.04765199e-01 -1.59937650e-01
7.21848369e-01 -4.20558780e-01 -1.29616678e-01 2.19114468e-01
4.60948087e-02 7.45531499e-01 1.22481728e+00 -1.86709869e+00
-3.10702622e-01 -1.28007114e-01 1.01732938e-02 2.35829949e-01
1.84620265e-02 1.63905770e-01 -6.30402625e-01 -9.38082397e-01
2.64077991e-01 -5.48742771e-01 -2.93608755e-01 2.82687277e-01
-5.92803895e-01 1.32287309e-01 1.03038037e+00 -1.02444208e+00
7.62047529e-01 -2.25412703e+00 2.22950742e-01 -2.13804588e-01
2.25834817e-01 3.08648556e-01 -3.55179071e-01 3.78338933e-01
-3.14874262e-01 3.67708921e-01 -4.57258105e-01 -7.06743062e-01
-2.06373617e-01 9.42435563e-02 -1.74307227e-01 5.82680464e-01
3.06059904e-02 6.21778667e-01 -7.77164876e-01 -3.64716917e-01
7.24048138e-01 7.97333598e-01 -6.43136680e-01 3.49699050e-01
-3.21324259e-01 7.13674128e-01 7.11003169e-02 7.54644156e-01
1.22878945e+00 -1.58095792e-01 2.76564937e-02 -6.03508651e-01
-2.14840516e-01 -1.05908513e-01 -1.16821384e+00 1.66991127e+00
-3.73869270e-01 7.45961845e-01 3.99445057e-01 -7.78551936e-01
6.89439178e-01 1.92498982e-01 4.91481572e-01 -8.04315746e-01
-2.17315570e-01 1.19116165e-01 -2.92398065e-01 -7.34060109e-01
6.34290159e-01 7.96516761e-02 4.32464451e-01 4.59484935e-01
-6.93050772e-02 -4.83663440e-01 2.61530250e-01 8.65408555e-02
9.53781247e-01 4.28939462e-01 -1.45969674e-01 -4.25361276e-01
3.10439378e-01 9.30115953e-02 3.95814419e-01 1.06336164e+00
-1.52558699e-01 1.47822368e+00 2.14070916e-01 -4.53958482e-01
-1.09623325e+00 -1.30436718e+00 -1.72620058e-01 8.73925567e-01
5.01572728e-01 -5.96665442e-02 -9.62242365e-01 -2.63809413e-01
-2.55590200e-01 6.47102475e-01 -2.82821268e-01 2.22419024e-01
-6.18682742e-01 -1.00335336e+00 -1.46305308e-01 4.82971929e-02
6.79113567e-01 -1.52050495e+00 -3.74431789e-01 1.95289969e-01
-5.04589140e-01 -9.45755303e-01 -6.35731697e-01 -1.20973743e-01
-9.14707661e-01 -9.79272068e-01 -8.25631201e-01 -1.01807487e+00
8.92048240e-01 7.48789132e-01 1.34258914e+00 4.08109963e-01
-4.54146624e-01 3.09028268e-01 -4.65375721e-01 3.76922905e-01
-6.21279597e-01 -3.47644180e-01 -2.86755353e-01 6.96452186e-02
-2.85454839e-01 -7.80631065e-01 -1.08256090e+00 2.73744822e-01
-1.51472414e+00 3.97748917e-01 3.83566290e-01 7.86321104e-01
6.69119477e-01 4.08651173e-01 3.94099861e-01 -8.04749370e-01
9.18589532e-01 -2.51050770e-01 -3.48590046e-01 2.00355887e-01
-5.39753556e-01 -2.07540154e-01 6.41212225e-01 -5.92884362e-01
-1.25975871e+00 1.14200562e-01 -1.59511596e-01 -1.47388861e-01
-4.40496922e-01 3.82552743e-01 -3.80580008e-01 -2.13614896e-01
7.17871368e-01 4.36827451e-01 1.74858555e-01 -9.02031243e-01
5.13805091e-01 5.50859690e-01 9.13308918e-01 -4.17075068e-01
7.28979886e-01 4.87885863e-01 -4.76200581e-01 -7.62921333e-01
-3.59975696e-01 -3.08048904e-01 -5.93858778e-01 -3.50919724e-01
7.25794971e-01 -9.01090503e-01 -5.91177821e-01 1.05122173e+00
-1.21321714e+00 -8.42547536e-01 -1.88609093e-01 -1.07837625e-01
-5.12711406e-01 5.98375380e-01 -8.73049021e-01 -3.72441411e-01
-4.90471989e-01 -1.68144524e+00 1.16389418e+00 4.48254049e-01
-2.10389365e-02 -7.83173382e-01 -3.64531189e-01 7.44832814e-01
7.51962781e-01 2.93362170e-01 9.46312785e-01 -8.41320585e-03
-9.34089065e-01 -1.15463398e-02 -5.43733656e-01 5.90532184e-01
3.10623735e-01 -1.10087596e-01 -8.70402038e-01 -4.26361740e-01
8.90763302e-04 -1.02380693e-01 1.02543807e+00 5.95506251e-01
1.41442239e+00 -2.89582103e-01 -2.25387607e-02 8.41575325e-01
1.64472640e+00 1.05124958e-01 1.30863667e+00 6.98076367e-01
8.22663128e-01 4.08787996e-01 1.44455940e-01 -9.36639495e-03
2.20465943e-01 6.40978396e-01 5.80637038e-01 -8.35189521e-01
-6.96501076e-01 1.29285917e-01 3.89064938e-01 3.81488681e-01
1.44512624e-01 -4.00920719e-01 -5.70694387e-01 6.21297896e-01
-1.64610863e+00 -8.42130184e-01 -3.20138514e-01 2.07936549e+00
1.07480204e+00 -2.68940777e-01 -3.60648483e-01 7.91399628e-02
6.06513679e-01 1.75959766e-01 -5.73218822e-01 -2.04416424e-01
-2.94594616e-01 2.07417175e-01 2.95533597e-01 7.02717781e-01
-1.13651752e+00 1.01965797e+00 6.86849928e+00 8.88044834e-01
-1.08089840e+00 -5.02479784e-02 1.02873516e+00 1.64727673e-01
-3.31026047e-01 2.87914574e-02 -2.23438591e-01 4.29380298e-01
2.76576608e-01 1.67431056e-01 6.96963131e-01 4.92676854e-01
7.71859229e-01 -4.50322241e-01 -9.13042665e-01 9.32463884e-01
7.17759654e-02 -1.40532553e+00 3.32138687e-01 -6.18057251e-02
9.44074869e-01 -1.14327967e-01 3.07621568e-01 -2.60994464e-01
4.39993203e-01 -1.28037417e+00 7.85034180e-01 6.27710462e-01
6.45316243e-01 -4.66253668e-01 4.02951270e-01 -4.01121937e-03
-6.78864121e-01 1.10717908e-01 -1.60427973e-01 2.83296466e-01
3.66456628e-01 7.24935055e-01 -3.96061957e-01 4.01551694e-01
1.11218393e+00 7.51184344e-01 -5.96840322e-01 1.33603251e+00
-1.53191373e-01 4.91274685e-01 3.75430994e-02 7.77886152e-01
-2.76627153e-01 -4.39590842e-01 5.72014451e-01 1.31007445e+00
2.46429622e-01 -1.33859187e-01 3.12607616e-01 1.05537140e+00
-1.34372890e-01 4.66553727e-03 -1.06870182e-01 3.03953558e-01
1.64866462e-01 1.20692027e+00 -9.37900186e-01 -3.02530080e-01
-2.31019795e-01 1.52896810e+00 -1.50432855e-01 6.66685879e-01
-4.07217503e-01 -2.22177252e-01 8.33816111e-01 2.75365502e-01
3.09614897e-01 -3.70368242e-01 -5.88179111e-01 -1.23071969e+00
4.24994901e-02 -1.45980859e+00 -8.35328475e-02 -1.25645077e+00
-1.35131574e+00 6.59195900e-01 -1.89859033e-01 -1.20887136e+00
2.48501703e-01 -4.62674916e-01 -6.39204860e-01 1.10957217e+00
-1.60936582e+00 -1.31496835e+00 -5.57755888e-01 5.51407158e-01
9.45583344e-01 2.44210750e-01 5.71240067e-01 3.85680795e-01
-5.04924953e-01 1.08408310e-01 2.45417789e-01 -1.87912881e-01
8.93512368e-01 -1.36159730e+00 5.98378778e-01 1.19691563e+00
-1.99383140e-01 6.65868580e-01 1.00187206e+00 -7.29693353e-01
-1.50408936e+00 -8.28334033e-01 4.59604979e-01 -1.95792556e-01
2.05322981e-01 4.45623547e-02 -1.11382926e+00 3.43418837e-01
6.39899790e-01 -5.85216843e-02 3.10542524e-01 -2.68475324e-01
-2.83571273e-01 -1.49702609e-01 -1.36651671e+00 9.35627818e-01
7.29637563e-01 -3.54953825e-01 -1.84213981e-01 4.13276434e-01
6.52385950e-01 -5.30070603e-01 -6.04028881e-01 2.72096187e-01
4.12932873e-01 -1.39563131e+00 1.51967669e+00 -2.21250039e-02
7.80653179e-01 -7.39916265e-01 -4.80404273e-02 -1.42813575e+00
-4.25881594e-01 -6.20053172e-01 1.98224813e-01 1.21059692e+00
1.29844114e-01 -3.85679007e-01 3.54188055e-01 6.31206453e-01
-2.15626270e-01 -3.20792794e-01 -4.78463709e-01 -4.42828983e-01
-1.33876875e-02 -3.52769673e-01 5.84741175e-01 9.41868901e-01
-4.89996642e-01 -1.37301445e-01 -5.48733056e-01 2.89039940e-01
9.96380448e-01 1.70251742e-01 7.72281706e-01 -9.43658948e-01
-3.61940682e-01 -5.20692527e-01 -4.51168641e-02 -1.19182241e+00
-3.11266184e-01 -6.04355991e-01 3.67719591e-01 -2.11142445e+00
2.39468724e-01 -3.45547944e-01 -8.38605016e-02 5.81536233e-01
-3.71838272e-01 8.41539204e-01 1.97846532e-01 4.91139829e-01
-3.07691544e-01 2.64169067e-01 1.52171493e+00 -3.69567811e-01
-4.62582529e-01 -1.91547945e-01 -1.02733350e+00 5.63960075e-01
7.97616482e-01 -2.46885374e-01 -2.57370770e-01 -8.75221491e-01
-1.61238298e-01 -1.53945416e-01 7.18251050e-01 -9.76023257e-01
1.31027147e-01 -1.30366653e-01 6.22968733e-01 -4.33679640e-01
1.70778766e-01 -6.07985198e-01 4.42523748e-01 1.88341290e-01
-2.09453583e-01 -2.37829313e-01 3.25086921e-01 3.45445305e-01
-1.82970792e-01 -4.61785644e-02 1.11130667e+00 -3.36011350e-01
-8.35833848e-01 1.45700634e-01 -5.60191631e-01 -4.93275046e-01
4.86430228e-01 -4.08510059e-01 -4.62326914e-01 -5.31193316e-01
-8.19260061e-01 -3.61353680e-02 8.49311411e-01 4.75136906e-01
7.76380777e-01 -8.28391373e-01 -9.09953415e-01 2.91121155e-01
-2.76888251e-01 -2.53687445e-02 6.86629832e-01 5.00397921e-01
-9.55118239e-01 -4.92345303e-01 -3.37619632e-01 -7.20211625e-01
-1.35625589e+00 4.70400691e-01 5.94081402e-01 -4.88734394e-02
-1.20471609e+00 5.60367286e-01 1.93035021e-01 -4.15583462e-01
5.26822656e-02 -1.37039334e-01 -1.72954500e-01 -2.18155444e-01
7.61554718e-01 2.34030738e-01 1.18645422e-01 -4.78161097e-01
1.35209396e-01 5.34601033e-01 -1.45904154e-01 -1.41514614e-01
1.57033062e+00 -4.46657807e-01 -5.39036810e-01 -1.80952758e-01
9.29283500e-01 -6.75633699e-02 -1.75060976e+00 -7.38132074e-02
-4.23402429e-01 -8.01819980e-01 4.04852569e-01 -1.21392560e+00
-1.41591632e+00 6.73913360e-01 9.64618325e-01 3.27288687e-01
1.58897889e+00 -2.69877493e-01 7.18341887e-01 -2.69364983e-01
2.18452513e-01 -7.16280520e-01 1.61106214e-01 1.39451891e-01
1.23266029e+00 -1.31927907e+00 2.04297394e-01 -4.66275126e-01
-3.71333867e-01 1.22892499e+00 4.59347278e-01 -1.48209810e-01
4.75954801e-01 2.64026433e-01 5.86080670e-01 4.84500639e-03
-7.84198195e-02 -1.28206536e-02 2.58821368e-01 9.38533008e-01
4.62202400e-01 -1.73339605e-01 -1.68636993e-01 -2.99965218e-02
5.87005094e-02 1.01185955e-01 7.42137551e-01 7.42717266e-01
-4.07879621e-01 -1.51883507e+00 -7.83038020e-01 1.95089668e-01
-5.06222069e-01 -4.43094879e-01 -9.88492463e-03 3.29343289e-01
7.24704787e-02 1.34689200e+00 -1.93263814e-01 -1.76145375e-01
1.42721191e-01 -4.45607454e-01 4.21141893e-01 -4.96465981e-01
-7.06069231e-01 2.61907816e-01 -2.70998538e-01 -5.96346378e-01
-3.05649310e-01 -4.51438367e-01 -8.31054270e-01 -3.72519016e-01
6.80831671e-02 -3.30078006e-01 6.82277799e-01 8.65352392e-01
4.55009997e-01 7.60930359e-01 3.71535540e-01 -1.58657134e+00
-1.19490318e-01 -8.46150100e-01 -4.79089856e-01 6.96311533e-01
6.56157017e-01 -1.40607730e-01 -3.90707195e-01 8.50385010e-01] | [11.202908515930176, -2.1902220249176025] |
c682d72f-f765-4a00-b117-d9e1adcdc041 | tarvis-a-unified-approach-for-target-based | 2301.02657 | null | https://arxiv.org/abs/2301.02657v2 | https://arxiv.org/pdf/2301.02657v2.pdf | TarViS: A Unified Approach for Target-based Video Segmentation | The general domain of video segmentation is currently fragmented into different tasks spanning multiple benchmarks. Despite rapid progress in the state-of-the-art, current methods are overwhelmingly task-specific and cannot conceptually generalize to other tasks. Inspired by recent approaches with multi-task capability, we propose TarViS: a novel, unified network architecture that can be applied to any task that requires segmenting a set of arbitrarily defined 'targets' in video. Our approach is flexible with respect to how tasks define these targets, since it models the latter as abstract 'queries' which are then used to predict pixel-precise target masks. A single TarViS model can be trained jointly on a collection of datasets spanning different tasks, and can hot-swap between tasks during inference without any task-specific retraining. To demonstrate its effectiveness, we apply TarViS to four different tasks, namely Video Instance Segmentation (VIS), Video Panoptic Segmentation (VPS), Video Object Segmentation (VOS) and Point Exemplar-guided Tracking (PET). Our unified, jointly trained model achieves state-of-the-art performance on 5/7 benchmarks spanning these four tasks, and competitive performance on the remaining two. Code and model weights are available at: https://github.com/Ali2500/TarViS | ['Bastian Leibe', 'Deva Ramanan', 'Jonathon Luiten', 'Alexander Hermans', 'Ali Athar'] | 2023-01-06 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Athar_TarViS_A_Unified_Approach_for_Target-Based_Video_Segmentation_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Athar_TarViS_A_Unified_Approach_for_Target-Based_Video_Segmentation_CVPR_2023_paper.pdf | cvpr-2023-1 | ['panoptic-segmentation', 'video-instance-segmentation', 'video-object-segmentation', 'video-semantic-segmentation'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [ 4.42311913e-01 -2.47047648e-01 -5.07892549e-01 -3.86833489e-01
-1.20803607e+00 -7.34663367e-01 5.18346965e-01 -5.24225891e-01
-4.81719285e-01 4.40049946e-01 -1.16025992e-01 -3.71458411e-01
1.95315957e-01 -3.29772323e-01 -1.16550326e+00 -6.10057294e-01
5.48583195e-02 5.85702419e-01 9.53744829e-01 -4.13406268e-02
4.06716727e-02 3.10641497e-01 -1.44919491e+00 4.59108233e-01
6.00330293e-01 1.20376158e+00 3.27299058e-01 7.43412912e-01
-3.24920081e-02 5.67652404e-01 -4.59219307e-01 -4.02224213e-01
4.43607599e-01 -5.97414076e-02 -1.07044971e+00 2.40486428e-01
7.70834625e-01 -3.33359331e-01 -1.86556727e-01 8.58259797e-01
2.30846956e-01 2.10946977e-01 5.62870085e-01 -1.46230650e+00
-4.97449607e-01 4.52033430e-01 -8.82132292e-01 2.99056023e-01
8.79783183e-02 3.72171283e-01 1.02145064e+00 -7.34257638e-01
7.26000190e-01 1.24045026e+00 9.05178666e-01 6.60202920e-01
-1.42642033e+00 -5.51857650e-01 6.35285378e-01 2.31572203e-02
-1.18853581e+00 -3.89326096e-01 5.09001911e-01 -6.72440946e-01
8.14303100e-01 3.20089310e-01 5.37627876e-01 1.31935871e+00
1.07268160e-02 1.14862907e+00 7.97814906e-01 4.38505560e-02
-2.40975972e-02 -1.88229784e-01 2.25578800e-01 6.84592783e-01
3.80129851e-02 -6.39069602e-02 -3.04507047e-01 1.10129237e-01
8.35010052e-01 -3.55468541e-02 -3.64194095e-01 -5.57538390e-01
-1.30333006e+00 7.64575064e-01 3.27536732e-01 1.72847956e-01
-1.71916440e-01 5.26473463e-01 6.20501578e-01 9.96996686e-02
5.47885716e-01 1.60901576e-01 -8.93835545e-01 -3.34976800e-02
-1.28932643e+00 3.77011627e-01 7.10129678e-01 1.06646252e+00
7.75151670e-01 4.86552715e-02 -6.08990312e-01 7.84273505e-01
2.53653377e-01 3.40375960e-01 2.93253332e-01 -1.27079189e+00
4.67696130e-01 2.08222106e-01 1.18179128e-01 -4.32725698e-01
-3.89095157e-01 -3.48775536e-01 -4.41201299e-01 2.04915836e-01
6.41131997e-01 -3.54203641e-01 -1.55806422e+00 1.72767913e+00
4.21495795e-01 5.66679776e-01 -2.17474654e-01 1.07778704e+00
9.97492492e-01 7.58821189e-01 2.45927617e-01 2.07268894e-01
1.50384200e+00 -1.47046268e+00 -1.69106692e-01 -5.66184223e-01
4.74634826e-01 -6.96098685e-01 9.56493199e-01 3.71492594e-01
-1.13717222e+00 -6.99295759e-01 -7.70483196e-01 -2.66387880e-01
-4.71932501e-01 1.31092863e-02 5.85322618e-01 3.91869277e-01
-1.29231417e+00 4.19118315e-01 -8.56843531e-01 -3.39937896e-01
7.51907110e-01 4.90661681e-01 -1.00712359e-01 -2.44937316e-02
-9.71703410e-01 6.16724074e-01 4.91064072e-01 1.18464567e-01
-1.29296243e+00 -9.55301762e-01 -7.72409439e-01 -7.95509145e-02
8.57597113e-01 -9.11784887e-01 1.46008587e+00 -1.23962379e+00
-1.25400078e+00 1.13173211e+00 -1.44772246e-01 -5.18877029e-01
6.88186467e-01 -3.62313360e-01 -1.30356923e-01 2.61551410e-01
3.20084244e-01 1.29590964e+00 1.07789183e+00 -1.42576861e+00
-9.81327593e-01 2.86665149e-02 3.14668864e-01 1.69991285e-01
2.72044122e-01 2.59874463e-01 -1.25899231e+00 -6.95728958e-01
-2.64858216e-01 -1.06807339e+00 -2.62366146e-01 1.17814772e-01
-6.03704572e-01 -3.23431253e-01 1.11449003e+00 -5.61527312e-01
8.79082918e-01 -2.12975430e+00 5.08095682e-01 -3.28286052e-01
2.09974676e-01 5.01029551e-01 -4.53489035e-01 -1.48264626e-02
6.14085793e-02 5.73525466e-02 -5.43478012e-01 -6.29342079e-01
3.04719210e-02 2.42016554e-01 -1.90053776e-01 3.21943223e-01
1.89389735e-01 1.15550077e+00 -7.28840530e-01 -5.38061500e-01
2.92113215e-01 3.83242339e-01 -4.23852116e-01 1.66570455e-01
-8.33931088e-01 4.32094544e-01 -5.22202432e-01 8.64737689e-01
5.79156160e-01 -5.06095946e-01 -8.92561004e-02 -2.90404886e-01
-2.59164441e-02 3.31390253e-03 -9.57543552e-01 1.92739618e+00
-1.11710079e-01 6.88309371e-01 3.44530433e-01 -1.17101598e+00
4.49636251e-01 1.98262587e-01 6.60687864e-01 -4.79820609e-01
7.59800896e-02 8.15512147e-03 -1.62102833e-01 -6.17753208e-01
4.33200479e-01 2.54338741e-01 -1.24366418e-01 2.35685512e-01
2.74781257e-01 -9.00833216e-03 4.80295956e-01 1.30828977e-01
9.76144910e-01 6.75133049e-01 -2.72402558e-02 -2.20496342e-01
3.66632789e-01 2.84796357e-01 7.34111190e-01 8.24638069e-01
-3.64004672e-01 8.17375958e-01 4.98799890e-01 -4.64693636e-01
-8.87379110e-01 -1.18833697e+00 -1.61353573e-01 1.67478108e+00
3.73248518e-01 -1.26120627e-01 -7.64988005e-01 -9.42527354e-01
2.43770629e-01 4.98990744e-01 -6.79141045e-01 3.44780713e-01
-7.49710083e-01 -6.81116462e-01 5.84290326e-01 6.11793935e-01
5.02071381e-01 -1.25836551e+00 -6.63479269e-01 1.44214958e-01
-2.55998284e-01 -1.48779988e+00 -7.05178857e-01 1.35553896e-01
-6.76967204e-01 -1.29619646e+00 -9.62768793e-01 -7.50642598e-01
2.43260637e-01 4.23625827e-01 1.43570828e+00 4.21386957e-02
-1.66819707e-01 6.12819791e-01 -2.48569429e-01 -2.25346938e-01
-1.99872762e-01 3.48919839e-01 -3.81999373e-01 5.24692386e-02
3.22587371e-01 -2.17508703e-01 -5.92922449e-01 4.56157029e-01
-1.04929042e+00 2.49502420e-01 4.66411382e-01 5.24411023e-01
8.09347510e-01 -6.45372808e-01 5.10857880e-01 -9.79106247e-01
1.71932325e-01 -5.91239870e-01 -7.20957696e-01 3.15416336e-01
-1.78491529e-02 -3.13187540e-01 4.50865507e-01 -4.60537553e-01
-8.35659206e-01 1.89510003e-01 -1.51847646e-01 -8.14851582e-01
-3.92470807e-01 3.48095000e-01 -5.97266341e-03 -1.88164145e-01
3.68698567e-01 1.83938473e-01 -8.55905190e-02 -4.51795846e-01
3.67665231e-01 2.18242079e-01 6.10134840e-01 -5.56410491e-01
6.10984147e-01 5.25878131e-01 -8.62515420e-02 -6.57655954e-01
-1.17256618e+00 -6.36578679e-01 -6.43432796e-01 -2.68234819e-01
1.37564778e+00 -1.10787416e+00 -5.01325250e-01 5.76528013e-01
-1.06706452e+00 -9.37884152e-01 -6.01287074e-02 9.78614837e-02
-6.89060450e-01 3.34174573e-01 -6.00871384e-01 -2.05318883e-01
-2.91028947e-01 -1.57268620e+00 1.45492780e+00 2.38573715e-01
-3.14624384e-02 -1.13632596e+00 -2.55452067e-01 4.95935887e-01
2.77443171e-01 3.52984488e-01 4.67189223e-01 -7.41107643e-01
-1.07785976e+00 1.40547022e-01 -4.84592646e-01 3.71782929e-01
-1.48912355e-01 1.48820415e-01 -8.96037936e-01 -3.81547391e-01
-2.81746119e-01 -5.04239976e-01 1.32154596e+00 8.39724183e-01
1.40434897e+00 4.46214117e-02 -7.19962418e-01 1.09218287e+00
1.32759249e+00 2.25982666e-01 5.26700616e-01 4.57830608e-01
1.04882216e+00 2.82378763e-01 6.42709076e-01 -2.20860075e-03
5.00315070e-01 9.27080989e-01 5.78892112e-01 -2.42612079e-01
-2.20659956e-01 1.81527957e-01 4.26162928e-01 2.09227636e-01
-9.31953639e-02 -4.42574948e-01 -7.96602905e-01 6.75985456e-01
-2.05223513e+00 -9.12659645e-01 -1.13135710e-01 1.78613734e+00
5.55537224e-01 1.01490028e-01 3.66362691e-01 -5.03062069e-01
7.98806906e-01 5.94201326e-01 -8.61559391e-01 -2.10561737e-01
8.57267007e-02 2.30197869e-02 7.36322403e-01 3.06961507e-01
-1.72373056e+00 1.22952640e+00 6.37916946e+00 9.63251233e-01
-1.13861513e+00 3.04075718e-01 8.22737753e-01 -2.52032697e-01
-1.58727989e-02 -1.43403977e-01 -9.21853840e-01 5.51899314e-01
6.59651399e-01 2.34550104e-01 3.69708747e-01 7.57752240e-01
5.10513894e-02 -6.44345134e-02 -1.17494607e+00 8.42557192e-01
1.63018834e-02 -1.53398466e+00 1.10870348e-02 -2.59563178e-01
6.86533570e-01 5.77659905e-01 1.00330226e-01 4.40976202e-01
3.54881346e-01 -8.62603128e-01 8.59916747e-01 3.24382573e-01
9.04706240e-01 -3.50486755e-01 2.96076834e-01 1.04971938e-01
-1.25933158e+00 -2.72910558e-02 -1.85389236e-01 3.65136802e-01
4.13208336e-01 1.23012401e-01 -3.41515392e-01 4.96156514e-01
9.32667315e-01 8.41703713e-01 -5.33942103e-01 1.18299890e+00
-1.05631556e-02 6.43169105e-01 -3.89968514e-01 3.58460397e-01
7.06420541e-01 -2.45680958e-01 5.44110477e-01 1.48179245e+00
1.45511538e-01 -1.21131808e-01 4.81744409e-01 8.33123088e-01
-2.24072322e-01 -2.91235387e-01 -2.95458615e-01 2.69938707e-01
3.23861152e-01 1.33093691e+00 -1.03614676e+00 -5.28245866e-01
-6.31195188e-01 9.60828424e-01 1.85217217e-01 7.11836278e-01
-1.37885475e+00 -7.04092309e-02 8.50557327e-01 8.92077163e-02
9.87753034e-01 -1.32372722e-01 -1.22606032e-01 -1.11825621e+00
-5.44139668e-02 -8.35354388e-01 4.86910820e-01 -7.36226678e-01
-1.19901526e+00 5.35594940e-01 2.72669107e-01 -1.07508254e+00
-1.40635028e-01 -7.32810199e-01 -6.23519003e-01 6.53959692e-01
-1.51771986e+00 -1.36753559e+00 -2.82328427e-01 6.69721246e-01
9.46483433e-01 -3.67532857e-02 3.21751803e-01 4.72412378e-01
-7.53639579e-01 4.27407354e-01 -4.12306450e-02 3.19388688e-01
6.98407471e-01 -1.16654265e+00 7.23277986e-01 9.54410434e-01
6.80289194e-02 8.92513767e-02 5.61569452e-01 -5.01916409e-01
-1.25702345e+00 -1.56420481e+00 2.83282012e-01 -5.31897008e-01
7.54440904e-01 -4.32536930e-01 -9.16721821e-01 1.21436489e+00
2.16160998e-01 3.29155803e-01 3.16647798e-01 3.41377035e-02
-3.68785471e-01 -2.17936561e-02 -8.54636490e-01 5.69609463e-01
1.24265635e+00 -1.53591722e-01 -3.45672190e-01 5.54430723e-01
9.72926795e-01 -8.47476482e-01 -6.76896453e-01 5.35920620e-01
4.51279819e-01 -1.12323058e+00 1.23216927e+00 -5.89401841e-01
3.83654684e-01 -3.60051721e-01 -1.10144123e-01 -9.53431666e-01
-2.45181978e-01 -5.09914279e-01 -2.29582533e-01 9.95494545e-01
3.91120851e-01 -5.72737157e-01 8.38662863e-01 3.52697283e-01
-5.95956981e-01 -9.94480550e-01 -9.61333632e-01 -7.62979805e-01
3.96233089e-02 -5.66637218e-01 4.70005006e-01 7.94753313e-01
-8.25065196e-01 3.30587059e-01 -3.90126795e-01 1.40333578e-01
6.21226430e-01 3.43052000e-01 8.56397152e-01 -1.10963702e+00
-4.43484992e-01 -6.78866386e-01 -1.73106536e-01 -1.51974964e+00
1.28344029e-01 -8.93666863e-01 2.09331051e-01 -1.74207711e+00
1.29320815e-01 -4.09206480e-01 -1.38964325e-01 6.24454379e-01
-2.56622314e-01 5.57816029e-01 5.25613785e-01 2.46659786e-01
-1.08312213e+00 2.93812037e-01 1.31063890e+00 -9.75754559e-02
-1.13096811e-01 2.32522905e-01 -6.31403744e-01 8.07618618e-01
6.89754188e-01 -3.72092068e-01 -3.85058284e-01 -8.02661896e-01
-2.95293093e-01 1.04928531e-01 7.08639622e-01 -1.03916228e+00
2.74162516e-02 -1.51769489e-01 2.38501772e-01 -7.31524706e-01
5.31195760e-01 -6.77768826e-01 2.22548991e-01 1.68464303e-01
-1.01478145e-01 -5.40344371e-03 4.10693347e-01 4.58266824e-01
-1.51970908e-01 -1.37798995e-01 8.58535409e-01 -1.79333329e-01
-1.29206061e+00 7.46201992e-01 -1.57463163e-01 3.35784435e-01
1.31689346e+00 -3.94889981e-01 -4.32640046e-01 4.93821912e-02
-7.96314359e-01 6.44537330e-01 4.72769260e-01 6.04040742e-01
4.00334984e-01 -9.76889729e-01 -6.38243198e-01 -1.67101011e-01
-1.33738145e-01 4.37156439e-01 3.93588066e-01 9.54350471e-01
-5.90514839e-01 5.54524899e-01 -5.12081832e-02 -1.08491814e+00
-1.23131132e+00 6.56327546e-01 4.76875454e-01 -2.23815978e-01
-7.33325779e-01 1.01282728e+00 5.74586809e-01 -2.56774724e-01
3.58770609e-01 -4.19185013e-01 -1.88423996e-03 1.65495984e-02
1.70381874e-01 1.58657461e-01 -2.63212800e-01 -6.58998787e-01
-2.78563827e-01 6.63020730e-01 -1.75267950e-01 1.19654730e-01
1.14092672e+00 -3.24025266e-02 1.81574449e-01 4.67612535e-01
1.16456234e+00 -5.16099155e-01 -1.95188069e+00 -1.84684590e-01
-3.57627422e-02 -3.57099682e-01 -1.90422460e-01 -7.77108192e-01
-1.42801678e+00 7.14320362e-01 2.86745757e-01 2.42668986e-01
1.18524516e+00 1.93793163e-01 9.08394635e-01 1.32907614e-01
1.88756660e-01 -8.81504774e-01 -8.73103589e-02 5.62846363e-01
6.74772799e-01 -1.38927221e+00 -1.44384921e-01 -5.02805352e-01
-7.52091289e-01 7.80500948e-01 8.09774518e-01 -2.13073701e-01
4.89826292e-01 2.23475710e-01 7.69329518e-02 -1.91451445e-01
-7.57113397e-01 -3.95826787e-01 5.16248107e-01 5.44898868e-01
3.51391971e-01 -8.87702331e-02 1.07744046e-01 2.81097233e-01
3.23640943e-01 1.52786434e-01 2.11436331e-01 7.87416816e-01
-4.15380418e-01 -9.25505996e-01 -2.98325121e-01 6.68707669e-01
-6.35308087e-01 -8.78759548e-02 -3.16962488e-02 1.14557946e+00
2.97587484e-01 5.38230121e-01 2.24730104e-01 -4.86477241e-02
1.54252619e-01 -3.67342494e-02 4.50662166e-01 -5.98704875e-01
-6.22938156e-01 1.82754785e-01 1.48681641e-01 -8.19748700e-01
-7.28261352e-01 -8.41385365e-01 -1.04528141e+00 -1.07446827e-01
9.28731635e-02 -1.85487762e-01 3.55705649e-01 1.03766131e+00
3.50620717e-01 7.34013021e-01 4.22543213e-02 -1.32288122e+00
-1.75246537e-01 -7.45324910e-01 -2.44699433e-01 4.12286282e-01
4.42916274e-01 -8.09919178e-01 -5.89767098e-02 3.18179935e-01] | [9.235295295715332, -0.014846273697912693] |
640fc451-9a44-47af-bd6c-bc45177f4397 | focal-visual-text-attention-for-visual | 1806.01873 | null | https://arxiv.org/abs/1806.01873v2 | https://arxiv.org/pdf/1806.01873v2.pdf | Focal Visual-Text Attention for Visual Question Answering | Recent insights on language and vision with neural networks have been successfully applied to simple single-image visual question answering. However, to tackle real-life question answering problems on multimedia collections such as personal photos, we have to look at whole collections with sequences of photos or videos. When answering questions from a large collection, a natural problem is to identify snippets to support the answer. In this paper, we describe a novel neural network called Focal Visual-Text Attention network (FVTA) for collective reasoning in visual question answering, where both visual and text sequence information such as images and text metadata are presented. FVTA introduces an end-to-end approach that makes use of a hierarchical process to dynamically determine what media and what time to focus on in the sequential data to answer the question. FVTA can not only answer the questions well but also provides the justifications which the system results are based upon to get the answers. FVTA achieves state-of-the-art performance on the MemexQA dataset and competitive results on the MovieQA dataset. | ['Li-Jia Li', 'Junwei Liang', 'Alexander Hauptmann', 'Lu Jiang', 'Liangliang Cao'] | 2018-06-05 | focal-visual-text-attention-for-visual-1 | http://openaccess.thecvf.com/content_cvpr_2018/html/Liang_Focal_Visual-Text_Attention_CVPR_2018_paper.html | http://openaccess.thecvf.com/content_cvpr_2018/papers/Liang_Focal_Visual-Text_Attention_CVPR_2018_paper.pdf | cvpr-2018-6 | ['memex-question-answering'] | ['natural-language-processing'] | [ 1.58332005e-01 -2.06042528e-01 1.14849649e-01 -4.98151094e-01
-7.76775360e-01 -6.29410803e-01 5.38751006e-01 2.11842462e-01
-6.21495783e-01 3.00997823e-01 3.80283117e-01 -4.01871860e-01
-8.33559558e-02 -5.38416207e-01 -7.94965744e-01 -4.28303212e-01
3.34418714e-01 5.52868485e-01 5.44222713e-01 -3.32143635e-01
3.74784738e-01 1.09821241e-02 -1.79011774e+00 1.00563741e+00
3.02378118e-01 1.19693840e+00 5.03905833e-01 1.17791450e+00
-5.88508368e-01 1.59004831e+00 -5.57185113e-01 -5.22527516e-01
-3.02016344e-02 -6.34323418e-01 -1.26572287e+00 3.42489958e-01
1.03023612e+00 -7.61609077e-01 -4.14532930e-01 8.90709102e-01
4.04374450e-01 3.01828980e-01 5.86927652e-01 -1.28934848e+00
-1.02262175e+00 4.29077446e-01 -4.14526790e-01 7.83155560e-01
6.49823964e-01 3.62502813e-01 1.25522554e+00 -1.02583694e+00
6.98385298e-01 1.36636281e+00 -1.43961301e-02 6.93365395e-01
-6.89134419e-01 -1.46175295e-01 4.42998201e-01 9.43755329e-01
-9.94634628e-01 -5.67802668e-01 6.34503007e-01 -4.28254902e-01
7.97967374e-01 4.26336199e-01 5.18620729e-01 8.97876441e-01
-1.06483988e-01 1.31908143e+00 5.09454131e-01 -4.11428392e-01
2.12436154e-01 -1.59867331e-02 2.89062232e-01 8.21096659e-01
-3.44756842e-01 -6.10973656e-01 -6.71705365e-01 4.25170502e-03
4.66902494e-01 1.33383051e-01 -3.19937468e-01 -1.85342044e-01
-1.25011265e+00 8.24541271e-01 6.86273873e-01 3.89182448e-01
-5.37398279e-01 4.20628786e-01 4.64438736e-01 2.64995635e-01
1.25696644e-01 2.53926814e-01 -1.64924145e-01 2.25614205e-01
-7.98224926e-01 3.53433341e-01 7.04552531e-01 7.83503652e-01
6.67607129e-01 -2.00948432e-01 -7.41199672e-01 6.40162349e-01
2.74060339e-01 4.38490093e-01 2.23913327e-01 -1.31956208e+00
5.41485846e-01 7.49292672e-01 1.65577009e-01 -1.08693027e+00
-2.11967900e-01 2.44567588e-01 -6.25198722e-01 -3.20816696e-01
6.31744385e-01 1.39642894e-01 -1.07456267e+00 1.37099636e+00
4.53904063e-01 -1.96094170e-01 -5.72954901e-02 1.37635767e+00
1.50102758e+00 8.95167112e-01 4.69313152e-02 5.68277389e-03
1.64273715e+00 -1.36129570e+00 -8.13000917e-01 -3.64925325e-01
2.42626309e-01 -6.21594489e-01 1.24234521e+00 1.84743479e-01
-1.29470730e+00 -7.53824592e-01 -5.66310823e-01 -6.84614241e-01
-2.52700597e-01 5.76696619e-02 1.28786013e-01 3.11480686e-02
-1.32521260e+00 -1.31481364e-01 -2.97226906e-01 -5.95492423e-01
4.91687328e-01 8.67263526e-02 -4.21031080e-02 -4.86060977e-01
-1.07536221e+00 5.60866356e-01 1.39115900e-01 2.39641398e-01
-1.36886227e+00 -3.72949868e-01 -4.64848697e-01 1.83444455e-01
8.48697722e-01 -9.63526487e-01 1.45916951e+00 -1.38496959e+00
-1.00282037e+00 9.63133037e-01 -4.25922811e-01 -4.06851679e-01
3.60131621e-01 -2.05182523e-01 -1.66534171e-01 9.77723956e-01
2.13554710e-01 9.63098526e-01 1.05424404e+00 -1.24147892e+00
-8.41997027e-01 -4.11163986e-01 6.35306180e-01 3.09858859e-01
-5.12212753e-01 1.61119029e-01 -1.05070448e+00 -9.26443711e-02
-2.65781492e-01 -6.45268917e-01 -1.24758221e-01 1.74570426e-01
-3.10878843e-01 -6.31258011e-01 1.02323425e+00 -8.67651939e-01
9.42125678e-01 -1.83625960e+00 4.19600785e-01 -2.36514807e-01
5.04582822e-01 2.79961735e-01 -3.14655870e-01 5.42446434e-01
3.18725079e-01 -1.28323166e-02 6.04209714e-02 -1.82442129e-01
-1.51106074e-01 2.64480561e-01 -4.63331848e-01 3.14868420e-01
1.49119094e-01 1.20788801e+00 -8.51889908e-01 -8.60732615e-01
2.04406697e-02 2.47709766e-01 -5.71726203e-01 5.26900053e-01
-9.81595695e-01 3.11298996e-01 -6.79229319e-01 8.15212846e-01
2.50141710e-01 -9.08477068e-01 -1.15505971e-01 -2.93358445e-01
2.78876349e-02 -2.94207871e-01 -6.64088428e-01 1.59779549e+00
9.95466765e-03 9.71972764e-01 2.08509505e-01 -9.58863735e-01
5.17356932e-01 1.79833144e-01 2.74705589e-01 -1.06448209e+00
1.76285088e-01 -1.45469755e-01 -2.75186956e-01 -1.30615640e+00
6.89329505e-01 2.61852145e-01 9.64950398e-03 5.57234168e-01
6.48715496e-02 1.66364193e-01 5.94464779e-01 7.35138953e-01
1.01235986e+00 -8.24526027e-02 -2.42718328e-02 1.21498764e-01
9.01051223e-01 3.00656378e-01 -2.04396948e-01 1.03246832e+00
-3.40919346e-01 5.32850683e-01 4.21930522e-01 -7.19639182e-01
-1.12102473e+00 -8.09411585e-01 6.01889729e-01 1.59680390e+00
3.03961873e-01 -1.63235784e-01 -6.82996035e-01 -7.30877936e-01
-1.67470425e-01 5.08986950e-01 -8.22436571e-01 2.22619519e-01
-5.25403798e-01 -1.82524417e-02 1.81746289e-01 3.29814792e-01
7.51141071e-01 -1.44045353e+00 -9.08679783e-01 -2.51741540e-02
-8.02853465e-01 -1.31511438e+00 -7.28398144e-01 -3.84833574e-01
-3.13487232e-01 -1.20586991e+00 -9.66633379e-01 -1.00731361e+00
4.81075048e-01 7.72300124e-01 1.37637949e+00 5.37032485e-01
-3.15389067e-01 1.21915805e+00 -6.33188367e-01 -2.98810482e-01
-6.16286881e-02 -1.30406082e-01 -5.01629233e-01 5.15502155e-01
4.11683053e-01 4.34646476e-03 -9.64964330e-01 3.60128522e-01
-1.23033237e+00 -1.91390038e-01 2.92749941e-01 5.88332057e-01
6.82341516e-01 -4.70963657e-01 5.81515729e-01 -5.28696597e-01
7.21012414e-01 -4.96980906e-01 -3.62294376e-01 8.64747405e-01
-5.64282686e-02 -8.18750635e-02 5.19449770e-01 -4.29191560e-01
-9.16642964e-01 -2.67934706e-02 -3.22591886e-02 -8.62135887e-01
-8.99710208e-02 5.20280302e-01 2.02510893e-01 2.82155067e-01
5.87080181e-01 4.46518242e-01 -1.14095539e-01 -1.71083286e-01
6.33364797e-01 5.98337889e-01 6.30373955e-01 -2.93566495e-01
2.96388537e-01 6.92550600e-01 -3.59240383e-01 -1.00497627e+00
-1.27817357e+00 -8.95725965e-01 -5.31759739e-01 -8.64736974e-01
1.44861126e+00 -7.83911049e-01 -1.29949152e+00 6.26410618e-02
-1.32764018e+00 -1.96515933e-01 -9.50349644e-02 -2.52771437e-01
-6.27171576e-01 5.08865237e-01 -4.31512713e-01 -9.69856203e-01
-5.49459815e-01 -1.08626878e+00 1.11384690e+00 3.10309619e-01
9.37167630e-02 -8.65619361e-01 -2.53273427e-01 1.06342244e+00
2.87918031e-01 -1.06648512e-01 8.54348719e-01 -6.66743219e-01
-1.12345934e+00 3.77545319e-02 -5.39150536e-01 1.51372388e-01
-4.11610276e-01 -1.79352984e-01 -8.52432907e-01 -3.68268758e-01
3.18191275e-02 -7.90638864e-01 1.07979095e+00 3.28164071e-01
1.49054337e+00 -3.40524018e-01 -3.79669070e-02 1.74023092e-01
1.39783549e+00 1.26521572e-01 5.55548847e-01 2.55803704e-01
8.19747865e-01 9.54419494e-01 4.45830166e-01 2.33676136e-01
8.88542771e-01 4.71233249e-01 8.43644679e-01 -3.40983011e-02
-1.85582787e-01 -1.83526590e-01 3.11022609e-01 5.46429217e-01
2.05374449e-01 -6.67026699e-01 -1.02565813e+00 9.59222913e-01
-2.08581638e+00 -1.21451771e+00 -3.80908936e-01 1.66620433e+00
3.10828269e-01 -3.93375516e-01 2.20764354e-01 -1.01413243e-01
6.39435530e-01 3.82424086e-01 -7.81874418e-01 -1.44614935e-01
-5.66420257e-02 -4.50764537e-01 8.57491642e-02 1.75955936e-01
-9.34378326e-01 8.01552057e-01 6.06312561e+00 5.37703872e-01
-8.69385481e-01 2.27702528e-01 6.13262892e-01 -2.46201038e-01
-2.51278907e-01 -1.30375192e-01 -6.90873682e-01 1.73369139e-01
8.31048191e-01 7.19276816e-02 5.41747689e-01 4.70984310e-01
9.65501964e-02 -4.16525692e-01 -1.14718366e+00 1.09120202e+00
6.46239221e-01 -1.67700195e+00 5.29429138e-01 -3.35808754e-01
5.98066032e-01 -8.93714055e-02 1.03245750e-01 1.72536716e-01
1.40300065e-01 -9.14079428e-01 8.73737991e-01 6.66403115e-01
5.38531661e-01 -5.15632331e-01 4.73631918e-01 5.04242182e-01
-1.08667970e+00 -4.52796191e-01 -4.12907571e-01 1.04364768e-01
4.40636516e-01 1.18667044e-01 -9.34314430e-01 2.80832559e-01
1.21218002e+00 6.37071788e-01 -8.51229668e-01 9.13900018e-01
-9.64103118e-02 4.54730034e-01 4.56629917e-02 -4.70990211e-01
5.19243121e-01 1.76517129e-01 3.52516860e-01 8.47590923e-01
3.00902296e-02 3.69037479e-01 1.78421587e-01 7.01701105e-01
-2.52026200e-01 2.47938827e-01 -4.04909879e-01 -3.33414763e-01
4.82119545e-02 1.13222444e+00 -8.06952059e-01 -5.67589760e-01
-5.35532653e-01 9.82452273e-01 5.56134403e-01 6.91642523e-01
-5.71619749e-01 -6.67235851e-02 1.19195037e-01 1.13128379e-01
7.18802869e-01 -3.36088985e-02 3.86984944e-01 -1.15669262e+00
-1.44700734e-02 -9.65917945e-01 9.67595220e-01 -1.53843486e+00
-1.30081296e+00 7.08717287e-01 -1.24705285e-01 -1.00111830e+00
-1.23126470e-01 -5.42571723e-01 -3.88565928e-01 4.13270473e-01
-1.49272394e+00 -1.21934676e+00 -6.32287800e-01 1.17517292e+00
9.48717356e-01 -4.83691245e-02 3.27799469e-01 2.79961050e-01
-2.53634274e-01 1.94317728e-01 -1.24422461e-01 2.63223350e-01
5.49979031e-01 -9.37315583e-01 -5.14484197e-02 7.83015430e-01
4.06143546e-01 2.50745773e-01 8.14862847e-01 -3.38082105e-01
-1.69964409e+00 -8.33636642e-01 8.93644571e-01 -7.31053591e-01
6.90120697e-01 -1.78352103e-01 -9.03759003e-01 6.93151176e-01
7.71203399e-01 1.10271797e-02 4.23186690e-01 -2.10909367e-01
-3.71349424e-01 -6.88862056e-02 -8.08472872e-01 5.75575531e-01
7.47492135e-01 -7.10412085e-01 -7.14595735e-01 6.31150961e-01
9.93954241e-01 -1.80040494e-01 -4.62207764e-01 -3.44637558e-02
3.69039476e-01 -9.79526699e-01 1.09842610e+00 -9.57293987e-01
8.44397187e-01 -4.60518986e-01 -4.51421171e-01 -6.79264724e-01
-3.52337654e-03 -2.19900250e-01 -1.33175537e-01 9.72257435e-01
3.19770217e-01 1.05569087e-01 6.06756508e-01 3.64420831e-01
1.68972075e-01 -4.97273803e-01 -8.34731281e-01 -9.01975557e-02
-3.85361940e-01 -2.40121976e-01 2.53583759e-01 5.82193851e-01
-4.10308540e-01 8.01960707e-01 -7.37747252e-01 3.76875810e-02
5.64618349e-01 3.61853361e-01 8.35570574e-01 -9.32111740e-01
-1.16784595e-01 -2.44220182e-01 -1.04972795e-01 -1.40159285e+00
-5.55704674e-03 -7.10602820e-01 2.05343172e-01 -2.05617237e+00
5.18147886e-01 4.62544441e-01 -7.73735419e-02 1.76577479e-01
-2.09336579e-01 2.67201245e-01 5.86328149e-01 2.91826427e-01
-1.58826554e+00 3.99738520e-01 1.49707770e+00 -5.13149738e-01
-8.25612620e-03 -2.57389545e-01 -6.28510475e-01 4.59518939e-01
3.75254810e-01 -1.86498508e-01 -5.20460010e-01 -8.62207234e-01
6.65064275e-01 4.98023927e-01 6.72042370e-01 -6.14156723e-01
7.53180563e-01 -1.00456052e-01 3.93581867e-01 -1.04480040e+00
4.44913536e-01 -7.20088601e-01 -3.24067593e-01 2.09442735e-01
-8.33669722e-01 2.75292158e-01 -6.20074868e-02 9.46889460e-01
-4.46608722e-01 -3.12440097e-01 3.73413086e-01 -4.80462849e-01
-1.11025071e+00 4.91241932e-01 -4.21647847e-01 2.96643257e-01
1.02057064e+00 3.20333391e-02 -6.81805849e-01 -9.85989809e-01
-8.23186040e-01 9.50136840e-01 1.23831339e-01 6.78204536e-01
1.03420591e+00 -1.15965819e+00 -8.49864185e-01 -4.43289399e-01
5.17810702e-01 -9.86949578e-02 8.57466698e-01 6.35911822e-01
-5.06940067e-01 5.93508303e-01 1.51894463e-04 -9.07772303e-01
-1.43900585e+00 1.06156790e+00 2.82143921e-01 4.12651524e-03
-4.18293059e-01 1.13495517e+00 3.89525712e-01 -4.40872228e-03
3.70708942e-01 -2.08166037e-02 -7.06094742e-01 4.87728179e-01
9.44091201e-01 1.32261310e-02 -2.33217567e-01 -6.79013491e-01
-2.86138713e-01 5.72863817e-01 -1.09063081e-01 -2.16602124e-02
1.17226803e+00 -6.82378352e-01 -1.58194572e-01 4.90091324e-01
1.23932779e+00 -5.05851686e-01 -1.22829890e+00 -4.76155311e-01
-1.43584684e-01 -3.30404043e-01 -4.69360873e-02 -7.25636780e-01
-1.13569975e+00 1.13839197e+00 5.12359381e-01 6.09188914e-01
1.13392365e+00 5.25320590e-01 7.70875573e-01 6.97026789e-01
-1.16675325e-01 -9.65118170e-01 9.38134551e-01 5.59346795e-01
1.15654302e+00 -1.46274567e+00 -9.52177197e-02 2.01120481e-01
-9.50619996e-01 1.07388854e+00 6.89024746e-01 8.18597600e-02
4.29517418e-01 -5.10111153e-01 2.29396448e-01 -6.45942450e-01
-1.21135390e+00 -5.23434460e-01 5.47265828e-01 3.12240005e-01
-7.23262131e-02 -3.15824121e-01 2.06064489e-02 1.76663175e-01
2.62928545e-01 -5.65364547e-02 5.04775107e-01 7.62583256e-01
-8.09030116e-01 -4.64455873e-01 -4.68118012e-01 5.11060059e-01
-4.71054673e-01 6.71323463e-02 -6.07573032e-01 4.88039613e-01
-2.24255785e-01 1.37286329e+00 2.40524217e-01 -2.62533605e-01
2.74887055e-01 1.05956659e-01 3.44422579e-01 -2.91823745e-01
-6.66597009e-01 -1.18700013e-01 4.68083285e-03 -5.52333236e-01
-8.98212254e-01 -4.97732639e-01 -1.13552845e+00 -9.99274030e-02
5.71676381e-02 -1.26077980e-02 4.80449528e-01 1.12624359e+00
2.39503264e-01 5.80037415e-01 3.48203927e-01 -5.84912121e-01
-1.47943377e-01 -6.93681002e-01 1.93862934e-02 4.85554934e-01
8.16119552e-01 -1.67886406e-01 -2.55765349e-01 3.99621308e-01] | [10.597115516662598, 1.2876787185668945] |
72b5cfd5-7d65-4a89-ac96-663052f23915 | depth-based-6dof-object-pose-estimation-using | 2303.02133 | null | https://arxiv.org/abs/2303.02133v2 | https://arxiv.org/pdf/2303.02133v2.pdf | Depth-based 6DoF Object Pose Estimation using Swin Transformer | Accurately estimating the 6D pose of objects is crucial for many applications, such as robotic grasping, autonomous driving, and augmented reality. However, this task becomes more challenging in poor lighting conditions or when dealing with textureless objects. To address this issue, depth images are becoming an increasingly popular choice due to their invariance to a scene's appearance and the implicit incorporation of essential geometric characteristics. However, fully leveraging depth information to improve the performance of pose estimation remains a difficult and under-investigated problem. To tackle this challenge, we propose a novel framework called SwinDePose, that uses only geometric information from depth images to achieve accurate 6D pose estimation. SwinDePose first calculates the angles between each normal vector defined in a depth image and the three coordinate axes in the camera coordinate system. The resulting angles are then formed into an image, which is encoded using Swin Transformer. Additionally, we apply RandLA-Net to learn the representations from point clouds. The resulting image and point clouds embeddings are concatenated and fed into a semantic segmentation module and a 3D keypoints localization module. Finally, we estimate 6D poses using a least-square fitting approach based on the target object's predicted semantic mask and 3D keypoints. In experiments on the LineMod and Occlusion LineMod datasets, SwinDePose outperforms existing state-of-the-art methods for 6D object pose estimation using depth images. This demonstrates the effectiveness of our approach and highlights its potential for improving performance in real-world scenarios. Our code is at https://github.com/zhujunli1993/SwinDePose. | ['Ioannis Stamos', 'Zhujun Li'] | 2023-03-03 | null | null | null | null | ['6d-pose-estimation-1', '6d-pose-estimation', 'robotic-grasping'] | ['computer-vision', 'computer-vision', 'robots'] | [ 1.01073757e-01 -2.45885894e-01 -1.82317406e-01 -4.69784707e-01
-4.92501318e-01 -5.07796228e-01 4.00441319e-01 5.93489967e-02
-4.21213746e-01 1.43498629e-01 -2.27076709e-01 5.39070964e-02
2.48735473e-02 -6.97960854e-01 -1.00408947e+00 -5.30957460e-01
2.10853606e-01 8.44480097e-01 3.47131670e-01 4.55772951e-02
4.84489858e-01 9.84831631e-01 -1.59589386e+00 -3.10598463e-01
8.12065780e-01 1.28191376e+00 5.00737369e-01 2.51043350e-01
-2.30709210e-01 -4.16014120e-02 -3.03495646e-01 -2.03993827e-01
5.11632502e-01 2.70338535e-01 -3.35778177e-01 2.81108052e-01
5.38845360e-01 -6.22106373e-01 -2.52334118e-01 1.04265690e+00
2.41158649e-01 3.33296396e-02 5.02158761e-01 -1.26667643e+00
-1.42577350e-01 -1.28623649e-01 -7.51751482e-01 -3.48314822e-01
4.80199635e-01 3.83270569e-02 6.75092995e-01 -1.11461973e+00
6.19140029e-01 1.27724814e+00 4.93841231e-01 4.21729356e-01
-1.09211349e+00 -7.69448638e-01 2.57754594e-01 1.79884955e-01
-1.33569777e+00 -1.84619948e-01 1.01843190e+00 -4.12761629e-01
7.84170747e-01 2.43835151e-02 6.83587611e-01 8.43408704e-01
-7.72291124e-02 8.68446052e-01 7.64127493e-01 -2.49492317e-01
1.73898026e-01 -5.42446859e-02 -2.39185780e-01 6.09258711e-01
2.82738537e-01 -1.83644548e-01 -4.17381197e-01 -8.34278204e-03
1.16693711e+00 4.25536186e-01 -2.39809752e-01 -1.24141717e+00
-1.35411763e+00 6.45103395e-01 6.34507179e-01 -1.05102710e-01
-4.73760128e-01 1.65956095e-01 9.06341430e-03 -1.73689872e-01
4.93099719e-01 3.23972464e-01 -5.23278415e-01 -2.09440202e-01
-4.12469715e-01 4.49141651e-01 5.98204494e-01 1.09079206e+00
9.06414747e-01 -3.83608460e-01 5.52186847e-01 7.92340159e-01
5.13105273e-01 7.43798673e-01 6.44512847e-02 -1.17136657e+00
5.81363916e-01 9.12146568e-01 2.53337860e-01 -1.13243508e+00
-4.98921245e-01 -3.02592903e-01 -4.04291213e-01 3.27018797e-01
4.16455001e-01 3.68933707e-01 -9.25948739e-01 1.32308555e+00
8.07410538e-01 1.24911182e-01 -2.47857764e-01 1.19261694e+00
6.20511234e-01 4.17504847e-01 -3.59080285e-01 2.95170277e-01
1.23853528e+00 -7.50241995e-01 -2.50048190e-01 -4.22779411e-01
2.95038968e-01 -7.71883547e-01 8.56210291e-01 3.63596529e-01
-8.15054178e-01 -3.36040348e-01 -1.05206215e+00 -3.26906681e-01
-2.04497993e-01 2.10262477e-01 6.18710399e-01 1.90520212e-01
-4.79807705e-01 3.48855078e-01 -1.17202067e+00 -2.68064648e-01
5.06828129e-01 5.13045549e-01 -5.95069766e-01 -3.50669533e-01
-5.55087268e-01 8.92364979e-01 4.34241772e-01 1.89931586e-01
-5.18632352e-01 -6.46950603e-01 -1.17453599e+00 -2.03749031e-01
6.04901731e-01 -5.09670198e-01 1.17149472e+00 -2.68763155e-01
-1.55096328e+00 9.03649092e-01 -1.97964296e-01 -1.04406022e-01
6.75263882e-01 -6.35322213e-01 2.52073646e-01 2.92802960e-01
1.31353080e-01 7.68021107e-01 8.69023561e-01 -1.40415740e+00
-5.05719423e-01 -9.38906133e-01 1.37677819e-01 5.35072327e-01
-4.37116623e-02 -4.03228432e-01 -9.47145760e-01 -3.65767360e-01
8.03907275e-01 -9.76312041e-01 -2.33278990e-01 6.43337250e-01
-3.83139879e-01 -2.97213495e-01 1.11451423e+00 -3.97371441e-01
3.99517804e-01 -2.29703808e+00 3.35589200e-01 2.24925205e-01
9.63168368e-02 1.78608224e-01 -3.16506624e-02 1.42551094e-01
2.30070397e-01 -3.27256680e-01 -2.55150914e-01 -5.75917900e-01
2.49392763e-02 2.18003854e-01 -1.73158397e-03 6.56657338e-01
2.43363380e-01 8.51570845e-01 -7.79164255e-01 -2.33163089e-01
7.79600203e-01 8.01525950e-01 -4.18726951e-01 2.74154425e-01
-4.21455562e-01 4.85068917e-01 -6.19280040e-01 7.19559431e-01
9.70270753e-01 -6.45924509e-02 -2.06580132e-01 -4.35578048e-01
-1.48660988e-01 1.54764548e-01 -1.25873506e+00 2.16013288e+00
-4.25303668e-01 3.09998661e-01 6.49468452e-02 -8.94919038e-01
1.14436960e+00 -5.81710152e-02 6.47506833e-01 -6.40658379e-01
2.26229087e-01 3.57546598e-01 -4.35034364e-01 -4.38479543e-01
2.15126157e-01 1.04726627e-01 2.52888501e-02 1.51568845e-01
-1.49702698e-01 -7.80136943e-01 -1.60474941e-01 -3.31549570e-02
6.91224813e-01 5.17534018e-01 9.40632820e-02 2.28432477e-01
3.46141905e-01 6.84897229e-02 4.42653418e-01 1.95651755e-01
-3.35761048e-02 8.61180961e-01 2.83683598e-01 -3.88697028e-01
-1.08359122e+00 -1.14753163e+00 -2.99102247e-01 2.51106441e-01
6.31773055e-01 -2.36589193e-01 -6.28238142e-01 -6.20070219e-01
5.12097895e-01 4.23154265e-01 -3.21472526e-01 -3.46340537e-02
-6.38187468e-01 -2.12044299e-01 -1.87852412e-01 6.52167261e-01
5.65544784e-01 -6.83019102e-01 -9.47490394e-01 1.01165727e-01
-1.58867300e-01 -1.48231840e+00 -1.90990239e-01 4.13386598e-02
-9.30847883e-01 -1.09729350e+00 -7.11362302e-01 -5.85948408e-01
8.23549807e-01 5.49632311e-01 5.81997871e-01 -2.24616230e-01
-3.84305179e-01 3.39694887e-01 -2.78467089e-01 -4.62077528e-01
1.59083843e-01 4.02768217e-02 3.37545350e-02 -9.28256065e-02
5.17676592e-01 -5.03945053e-01 -9.10237074e-01 5.28434038e-01
-7.54388869e-01 1.86965033e-01 5.05715191e-01 4.23025787e-01
8.34512472e-01 -1.48446426e-01 -1.33808777e-02 -3.99942249e-01
-4.79691736e-02 -2.00006649e-01 -8.65164042e-01 -1.92710221e-01
-1.13607131e-01 -1.11111850e-01 2.30159834e-01 -3.66949558e-01
-7.03636467e-01 5.37283659e-01 -1.97304472e-01 -8.37587953e-01
-1.84669375e-01 2.05125898e-01 -2.90687561e-01 -1.96718588e-01
2.71008551e-01 -5.49099147e-02 3.70408684e-01 -6.14966869e-01
1.87471136e-01 6.64185882e-01 5.14677644e-01 -4.49538440e-01
9.57812786e-01 7.92155385e-01 4.15751189e-02 -8.71930361e-01
-7.88050771e-01 -7.30623841e-01 -8.79244387e-01 -2.34330848e-01
8.87879372e-01 -8.82569015e-01 -7.62912035e-01 6.41286850e-01
-1.29819369e+00 -7.53516704e-02 3.81529331e-02 6.95661008e-01
-6.56533659e-01 3.63875657e-01 -2.39084035e-01 -6.42650425e-01
-1.17177509e-01 -1.41939771e+00 1.53520989e+00 2.18033656e-01
-1.26754835e-01 -5.92690945e-01 -3.72352570e-01 6.14686370e-01
-3.21506076e-02 4.53631252e-01 7.37797737e-01 -1.83069408e-01
-9.20075178e-01 -4.71659720e-01 -3.88012737e-01 3.14391911e-01
2.27024332e-01 -1.11523047e-01 -7.84981489e-01 -1.39934495e-01
4.13496159e-02 -1.85066879e-01 4.39224720e-01 3.19904327e-01
1.15246439e+00 2.01163754e-01 -3.89235407e-01 8.32196295e-01
1.46237612e+00 1.51245564e-01 3.63410354e-01 5.40638983e-01
9.25467134e-01 6.51325941e-01 9.88596857e-01 4.07404006e-01
4.77586657e-01 9.20979738e-01 1.03409541e+00 6.41703159e-02
1.11089148e-01 -3.37065578e-01 -5.27993888e-02 5.92205167e-01
6.60644025e-02 1.11827783e-01 -1.02874303e+00 2.83361584e-01
-1.72814906e+00 -3.19345742e-01 -1.08583882e-01 2.31702638e+00
4.31370795e-01 2.32068971e-01 -1.74662724e-01 4.17611785e-02
5.61411142e-01 -2.92262621e-02 -9.65274334e-01 4.63734418e-02
1.33243680e-01 1.58329874e-01 5.25419772e-01 2.98691630e-01
-1.05711782e+00 1.01998580e+00 4.49415350e+00 5.64059913e-01
-1.29143620e+00 -1.27492040e-01 1.75527334e-01 6.02589101e-02
-1.62166432e-01 -1.00532696e-01 -8.67599308e-01 2.85586655e-01
1.52731985e-01 3.37082058e-01 2.01401412e-01 1.00356615e+00
2.50066929e-02 -3.76061082e-01 -1.11490667e+00 1.21697164e+00
2.40875378e-01 -9.41131711e-01 -1.27548993e-01 3.99691984e-02
5.45889378e-01 1.60100386e-01 2.61996258e-02 -1.10007115e-01
-2.35133946e-01 -7.39591837e-01 8.34721088e-01 1.87782541e-01
7.41266549e-01 -6.99341536e-01 5.51026762e-01 5.27027249e-01
-1.03487539e+00 1.01896204e-01 -4.31840301e-01 -3.15916203e-02
2.32545227e-01 6.43446922e-01 -1.02813828e+00 4.72024322e-01
7.99135268e-01 7.47307062e-01 -2.91542113e-01 1.18460989e+00
-4.18577045e-01 -1.09745778e-01 -7.21939445e-01 7.53706396e-02
9.83318910e-02 -2.52162188e-01 6.24514878e-01 5.50767481e-01
4.39398557e-01 4.64315079e-02 1.24661833e-01 8.47027838e-01
8.35613534e-02 -9.08170640e-02 -5.08133769e-01 1.82287768e-01
5.78047991e-01 1.18315482e+00 -9.46694553e-01 -3.50944400e-02
-3.32527637e-01 1.06759310e+00 1.90627530e-01 2.87602544e-01
-6.25513375e-01 -3.24019730e-01 9.12486017e-01 1.74520910e-01
4.25113320e-01 -7.05599844e-01 -3.90007824e-01 -1.32034302e+00
5.48206806e-01 -5.64027667e-01 -1.36412755e-01 -8.62725794e-01
-8.50242257e-01 3.23685557e-01 2.45794639e-01 -1.31405914e+00
4.12971787e-02 -8.86048198e-01 -1.05771974e-01 7.11048007e-01
-1.55152166e+00 -1.03747666e+00 -7.54223287e-01 3.44977260e-01
7.22128391e-01 4.33263093e-01 6.12518311e-01 1.38372079e-01
-3.19495797e-01 1.27370387e-01 -3.59178968e-02 -3.12372353e-02
4.78791505e-01 -1.00262618e+00 5.22671700e-01 4.34462428e-01
-1.06767034e-02 4.67591673e-01 5.33589244e-01 -5.50805628e-01
-1.74887490e+00 -8.91593814e-01 4.37830478e-01 -4.54647869e-01
2.85100758e-01 -7.21392930e-01 -8.44541013e-01 5.10370791e-01
-5.80514669e-01 2.65667319e-01 1.37861416e-01 -2.75243163e-01
-4.00805026e-01 -1.95778817e-01 -1.15975487e+00 5.01114190e-01
1.23451507e+00 -3.84312123e-01 -4.30999100e-01 2.71571189e-01
6.77476406e-01 -1.00607133e+00 -8.35910082e-01 7.06293225e-01
7.97408700e-01 -8.07130516e-01 1.22927880e+00 5.48903085e-02
3.82322073e-01 -4.41844940e-01 -2.64571071e-01 -1.14357340e+00
9.26220044e-02 -1.72021091e-01 -2.35637892e-02 8.35916936e-01
2.14962717e-02 -6.56138182e-01 1.12748051e+00 6.46722734e-01
-1.31698683e-01 -8.89996588e-01 -1.04940248e+00 -5.86172998e-01
-2.39007488e-01 -5.88435292e-01 7.14253306e-01 6.10550582e-01
-4.42003310e-01 -7.28277955e-03 1.61497772e-01 4.27277714e-01
8.69706035e-01 3.27113420e-01 1.17320740e+00 -1.27051127e+00
1.79750919e-01 -2.79565752e-01 -9.10968900e-01 -1.55089605e+00
1.87082753e-01 -6.38030171e-01 1.91446438e-01 -1.62139022e+00
-7.27395266e-02 -7.61824548e-01 1.84825018e-01 3.56064171e-01
3.97298113e-02 4.55547899e-01 2.36331612e-01 2.05837071e-01
-2.56823480e-01 7.13703275e-01 1.35039854e+00 -1.35921910e-02
-1.37325853e-01 3.28376889e-02 -1.98968381e-01 8.33158910e-01
7.19631672e-01 -2.47316316e-01 -2.32752845e-01 -7.51404762e-01
5.29211434e-03 -1.64602790e-02 5.32868803e-01 -1.02110195e+00
1.25433609e-01 -1.37059987e-01 4.53135014e-01 -9.73809242e-01
8.50718439e-01 -1.20177770e+00 5.65398894e-02 3.60627413e-01
1.58199340e-01 -1.06296770e-01 1.05096377e-01 3.89573246e-01
-9.66579393e-02 -2.17264891e-01 5.50748050e-01 -1.19514093e-01
-7.76654720e-01 5.78226864e-01 3.14672977e-01 -3.86265904e-01
1.34038246e+00 -5.78424811e-01 -1.38463639e-02 -1.06765762e-01
-4.23459947e-01 3.38242799e-01 8.83067131e-01 6.16478086e-01
1.01512849e+00 -1.18013203e+00 -4.48229343e-01 5.28758824e-01
3.12553763e-01 9.14432228e-01 1.60298094e-01 8.51290464e-01
-9.02190030e-01 2.94701308e-01 -3.19047496e-02 -1.32051790e+00
-1.15015900e+00 2.92741060e-01 1.44031212e-01 4.10518378e-01
-7.77481794e-01 8.05889785e-01 3.68368179e-01 -7.30195045e-01
3.44652891e-01 -4.92392927e-01 1.63235981e-02 -2.17702031e-01
2.38988876e-01 2.09199950e-01 1.80713534e-01 -7.08572507e-01
-3.62911880e-01 1.10713792e+00 -1.50019139e-01 5.03295176e-02
1.52079642e+00 -1.24097571e-01 -5.61755449e-02 3.28782052e-01
1.49751854e+00 -2.05872610e-01 -1.66954637e+00 -2.76773661e-01
-8.68227631e-02 -8.94912243e-01 -1.40700519e-01 -3.20667595e-01
-1.02102005e+00 1.02190030e+00 5.76360583e-01 -2.96340704e-01
8.50329518e-01 2.45416686e-01 8.49341214e-01 2.57019043e-01
6.95439994e-01 -7.19079792e-01 2.08458275e-01 4.84611988e-01
9.33168113e-01 -1.37061512e+00 4.44542617e-02 -8.26788366e-01
-2.47836605e-01 1.25133669e+00 6.83615446e-01 -2.22045451e-01
4.81209964e-01 7.07474202e-02 2.05136672e-01 -2.31557935e-01
-5.33599518e-02 1.08991995e-01 2.58410960e-01 5.95999360e-01
-7.87797570e-02 2.96008419e-02 8.79816338e-02 -4.13346700e-02
-2.62667388e-01 -2.39560902e-01 3.46464477e-02 1.12407672e+00
-3.48364979e-01 -1.02421439e+00 -5.14547050e-01 2.57690638e-01
-1.03841081e-01 4.25960630e-01 -1.98534012e-01 7.77723014e-01
1.09674945e-01 5.06913543e-01 2.74037719e-01 -3.12178135e-01
5.20777524e-01 -1.99505091e-01 7.61632621e-01 -7.17231274e-01
6.11775555e-02 1.11501016e-01 -3.46636653e-01 -8.61180186e-01
-4.75817233e-01 -7.50007272e-01 -1.45606828e+00 8.78668278e-02
-4.36014682e-01 -1.50990620e-01 1.41419494e+00 8.07673573e-01
3.67635131e-01 1.08290240e-01 5.92589259e-01 -1.41825414e+00
-5.19963562e-01 -5.92577100e-01 -2.78983414e-01 4.39092219e-01
3.32976073e-01 -1.03279078e+00 -3.33593100e-01 -3.13159525e-01] | [7.511162281036377, -2.6066370010375977] |
233183bd-08b4-406e-ac59-682fa577f5c6 | customics-a-versatile-deep-learning-based | 2209.05485 | null | https://arxiv.org/abs/2209.05485v1 | https://arxiv.org/pdf/2209.05485v1.pdf | CustOmics: A versatile deep-learning based strategy for multi-omics integration | Recent advances in high-throughput sequencing technologies have enabled the extraction of multiple features that depict patient samples at diverse and complementary molecular levels. The generation of such data has led to new challenges in computational biology regarding the integration of high-dimensional and heterogeneous datasets that capture the interrelationships between multiple genes and their functions. Thanks to their versatility and ability to learn synthetic latent representations of complex data, deep learning methods offer promising perspectives for integrating multi-omics data. These methods have led to the conception of many original architectures that are primarily based on autoencoder models. However, due to the difficulty of the task, the integration strategy is fundamental to take full advantage of the sources' particularities without losing the global trends. This paper presents a novel strategy to build a customizable autoencoder model that adapts to the dataset used in the case of high-dimensional multi-source integration. We will assess the impact of integration strategies on the latent representation and combine the best strategies to propose a new method, CustOmics (https://github.com/HakimBenkirane/CustOmics). We focus here on the integration of data from multiple omics sources and demonstrate the performance of the proposed method on test cases for several tasks such as classification and survival analysis. | ['Paul-Henry Cournède', 'Stefan Michiels', 'Yoann Pradat', 'Hakim Benkirane'] | 2022-09-12 | null | null | null | null | ['survival-analysis'] | ['miscellaneous'] | [ 2.20282339e-02 -2.80886501e-01 8.90112966e-02 -2.59117037e-01
-5.19920766e-01 -4.36411709e-01 5.33562899e-01 5.30400634e-01
-1.50198847e-01 8.65936875e-01 2.73019433e-01 1.36321977e-01
-5.09725571e-01 -8.99577320e-01 -4.84552890e-01 -1.04953969e+00
-7.64908716e-02 6.84330046e-01 -4.06648487e-01 -1.39209837e-01
-2.32826740e-01 7.84711897e-01 -1.62478614e+00 5.49057007e-01
6.73944890e-01 7.25998521e-01 2.21162975e-01 6.86630249e-01
-3.01681012e-01 4.57068831e-02 -5.68408906e-01 -2.29039088e-01
1.59951806e-01 -4.86781210e-01 -2.35013142e-01 -2.55128294e-01
-1.31256297e-01 1.15316235e-01 -9.83327180e-02 7.19991446e-01
8.32891643e-01 -2.82776624e-01 6.66469932e-01 -1.05321324e+00
-3.82820487e-01 3.15814018e-01 1.62379831e-01 -2.16526110e-02
1.50255367e-01 2.17798367e-01 7.22149193e-01 -4.96986985e-01
1.00914383e+00 1.06546557e+00 6.44270062e-01 4.76213664e-01
-1.42228365e+00 -2.54859507e-01 -2.81617373e-01 1.88272685e-01
-1.15246737e+00 -3.09298575e-01 5.64496219e-01 -6.27435744e-01
9.13327157e-01 3.40265930e-01 8.66808236e-01 1.56311810e+00
4.79819775e-01 5.54782271e-01 1.27235126e+00 -3.38652939e-01
3.28644454e-01 2.51917511e-01 1.44388020e-01 5.96320748e-01
3.63314062e-01 1.24491449e-03 -4.69062686e-01 -3.04004788e-01
2.30156034e-01 4.56673771e-01 -1.84692308e-01 -3.30414861e-01
-1.39006352e+00 7.45595872e-01 2.53819287e-01 7.38176644e-01
-5.47364295e-01 -8.91666412e-02 4.75665212e-01 1.20933898e-01
3.37314337e-01 6.86468720e-01 -6.09515786e-01 1.35080233e-01
-8.04512501e-01 1.32160574e-01 8.05825472e-01 6.88383102e-01
6.14748001e-01 9.26742256e-02 1.65343042e-02 7.31584489e-01
5.05007729e-02 1.31469145e-01 9.76546168e-01 -5.05284965e-01
-1.21489890e-01 8.89252186e-01 -1.09032996e-01 -8.71309280e-01
-7.65974283e-01 -7.54065216e-01 -1.06626785e+00 -9.93029997e-02
2.64854312e-01 -7.93911740e-02 -8.80978107e-01 1.71691120e+00
5.98388791e-01 -1.70648396e-02 4.58455652e-01 4.99057233e-01
8.32593203e-01 5.91481030e-01 1.18798211e-01 1.00288346e-01
1.26303196e+00 -4.53819156e-01 -7.02354789e-01 3.41063768e-01
5.94955206e-01 -3.65112901e-01 6.21863484e-01 3.26252639e-01
-4.90800112e-01 -4.04710650e-01 -1.16599619e+00 6.73973411e-02
-1.12159657e+00 1.78372785e-01 6.49717271e-01 4.65024859e-01
-7.26978183e-01 7.58940220e-01 -9.70941842e-01 -6.40606821e-01
3.13575417e-01 4.04636949e-01 -5.98775446e-01 3.44597511e-02
-1.13666105e+00 7.13637531e-01 7.90406525e-01 3.01365368e-02
-8.45963001e-01 -7.49846578e-01 -4.34606642e-01 3.10658395e-01
-1.79250371e-02 -1.06747997e+00 3.14388633e-01 -8.51423740e-01
-1.47086680e+00 4.96675104e-01 1.43046767e-01 -3.13869715e-01
4.01071995e-01 -7.57622067e-03 -4.89720672e-01 -5.58627211e-02
-7.06145167e-02 5.57465613e-01 3.63097429e-01 -8.50632727e-01
-1.65955275e-01 -4.57890511e-01 -3.13086301e-01 -2.81308085e-01
-5.65173090e-01 -2.78813124e-01 -1.30896345e-01 -4.84319925e-01
-1.32804781e-01 -9.93836164e-01 -7.75011331e-02 -2.58658946e-01
-4.41859275e-01 1.28733784e-01 4.70448494e-01 -5.31072438e-01
7.25165546e-01 -2.18046403e+00 6.23426437e-01 -2.05717355e-01
1.72554106e-01 3.33841085e-01 -2.46039346e-01 1.03304446e+00
-3.39491189e-01 1.15396544e-01 -1.94812045e-01 -2.45945990e-01
-1.19931355e-01 2.39059567e-01 -5.53495474e-02 3.60883176e-01
2.62847006e-01 9.15142417e-01 -7.15121388e-01 -1.37328610e-01
2.02457711e-01 8.32925260e-01 -3.31801146e-01 4.55238879e-01
-3.96173656e-01 5.07654727e-01 -3.99311721e-01 8.78875375e-01
4.39427137e-01 -3.39888364e-01 3.97798151e-01 -2.92315453e-01
6.91809952e-02 -1.29216954e-01 -8.23041975e-01 1.75259387e+00
-2.95206755e-01 3.85857195e-01 -2.00417712e-01 -1.09802735e+00
9.02176857e-01 3.77253979e-01 7.68336236e-01 -4.17904645e-01
2.11343050e-01 2.54341245e-01 1.82102874e-01 -6.08834922e-01
-9.68572870e-03 -4.96210426e-01 6.25204593e-02 1.89370036e-01
4.92765576e-01 3.02744538e-01 2.03440711e-01 -1.21983849e-01
1.00084722e+00 2.15861604e-01 5.25107741e-01 -2.58960843e-01
4.66733843e-01 5.87685183e-02 6.84064090e-01 3.09872985e-01
-3.72280478e-02 5.04932940e-01 5.37206352e-01 -7.53652215e-01
-1.15898347e+00 -7.82235086e-01 -2.50139952e-01 6.37705386e-01
-5.07166147e-01 -3.93815070e-01 -5.35649240e-01 -3.79133016e-01
1.97557554e-01 3.22164357e-01 -9.29639161e-01 -2.17882395e-01
-6.66079894e-02 -1.56384587e+00 8.49401593e-01 1.88394308e-01
3.03192697e-02 -6.47558689e-01 -5.93721807e-01 4.04147446e-01
4.49715853e-02 -9.61206198e-01 5.03798842e-01 4.81538892e-01
-9.78498697e-01 -1.12823319e+00 -7.42745399e-01 -1.43366933e-01
3.61321747e-01 -3.16011488e-01 6.99971199e-01 -2.43880674e-01
-7.98048079e-01 6.02404289e-02 -2.89806336e-01 -7.05832720e-01
-7.62415886e-01 3.15107644e-01 3.14877331e-01 2.46965200e-01
2.12469190e-01 -6.75284386e-01 -5.20243227e-01 -1.93453431e-01
-1.29072690e+00 1.51501760e-01 7.85774112e-01 1.11493409e+00
5.70063949e-01 -9.20728818e-02 9.50734973e-01 -9.47394371e-01
7.08531916e-01 -9.98239160e-01 -5.26252866e-01 4.87735778e-01
-5.38460970e-01 3.21711302e-01 8.21333885e-01 -3.46777767e-01
-7.40782738e-01 5.17026288e-03 -3.38717639e-01 -4.00556356e-01
-4.45360243e-01 8.49782884e-01 -2.71320015e-01 2.39702404e-01
4.85896111e-01 4.16164756e-01 2.49364525e-01 -6.03436410e-01
1.85378402e-01 5.96932709e-01 8.35041050e-03 -3.39989722e-01
1.70113340e-01 5.70206702e-01 3.23897421e-01 -8.24667931e-01
-2.94064313e-01 -2.64703065e-01 -7.45273352e-01 1.12795115e-01
8.78747702e-01 -9.19053078e-01 -6.32499993e-01 5.70803344e-01
-8.54405582e-01 1.55554041e-02 -1.78404436e-01 5.37045419e-01
-5.80976486e-01 1.09969079e-01 -6.05509222e-01 -4.27083045e-01
-4.53112841e-01 -1.28734362e+00 8.36479425e-01 4.20720242e-02
-1.11306041e-01 -1.02734590e+00 6.88222349e-01 1.36202022e-01
4.37438965e-01 6.55567467e-01 1.43194890e+00 -1.00532722e+00
-5.53212702e-01 -4.84982073e-01 1.11075722e-01 2.74520218e-01
4.64346111e-01 8.41247663e-02 -1.11891735e+00 -1.81752250e-01
-9.21971798e-02 -1.52178422e-01 7.39075184e-01 7.05481470e-02
9.75984454e-01 -4.63868901e-02 -4.68863904e-01 8.28696191e-01
1.77486348e+00 8.03321004e-02 4.43887144e-01 2.14977697e-01
4.92031544e-01 7.88680553e-01 1.57032087e-01 4.11478341e-01
4.81070913e-02 7.33140111e-01 3.71632338e-01 -7.92663097e-02
2.98890527e-02 -1.70187995e-01 1.11509867e-01 9.31486070e-01
6.40182123e-02 -4.08641040e-01 -8.81545782e-01 4.34041023e-01
-1.80542064e+00 -7.32943833e-01 1.56485513e-01 1.97708225e+00
5.51095366e-01 -4.11160350e-01 7.19225854e-02 -5.15201353e-02
3.59475255e-01 -2.25998417e-01 -5.53673863e-01 -4.93653387e-01
-4.01011765e-01 1.33571208e-01 1.23327985e-01 1.30691320e-01
-8.29062700e-01 6.03131831e-01 6.41416216e+00 6.35107398e-01
-1.44380426e+00 8.48415047e-02 5.49522996e-01 -2.22625777e-01
-1.94465578e-01 -3.23013872e-01 -8.60427976e-01 6.15611374e-01
1.50380456e+00 -1.51330844e-01 1.50814056e-01 6.29908562e-01
6.85439706e-02 1.27251029e-01 -1.14926505e+00 6.61019564e-01
-8.43824670e-02 -1.65141296e+00 2.62871146e-01 1.82715848e-01
5.75451434e-01 2.92701155e-01 -1.84263606e-02 2.07132667e-01
3.30417342e-02 -1.00384355e+00 1.00705996e-01 9.11769807e-01
4.08539563e-01 -5.77020109e-01 9.34146404e-01 4.66963053e-01
-5.67559183e-01 -1.44345820e-01 -2.88019806e-01 1.12116896e-01
-9.01609287e-02 8.43867123e-01 -1.43628895e+00 9.90959287e-01
4.74848837e-01 6.56123102e-01 -5.64768076e-01 8.98522735e-01
2.73306102e-01 1.87719196e-01 -3.65177810e-01 -1.33048281e-01
-6.46865144e-02 -2.68868595e-01 3.98295879e-01 1.39440703e+00
6.77633405e-01 -2.39670441e-01 -1.92872882e-01 9.67045665e-01
1.00999355e-01 3.60596716e-01 -7.81933725e-01 -6.64609611e-01
1.45691410e-01 1.38060749e+00 -6.64118946e-01 -4.29135352e-01
-2.79703617e-01 7.34777689e-01 4.04255152e-01 2.40541607e-01
-5.38721144e-01 -8.27013180e-02 8.97067368e-01 4.17735539e-02
1.71125263e-01 -2.34601364e-01 -1.43080261e-02 -1.36039448e+00
-3.15655947e-01 -1.15105271e+00 5.54961681e-01 -6.79155827e-01
-1.30045986e+00 7.94802427e-01 2.00673733e-02 -1.15915024e+00
-4.00204599e-01 -8.34176421e-01 -2.06090659e-01 9.47684407e-01
-1.30545521e+00 -1.32691443e+00 -2.26850122e-01 3.56182545e-01
2.69468695e-01 -5.64345717e-01 1.50904989e+00 3.72564465e-01
-7.81831920e-01 1.98789865e-01 7.62219191e-01 -2.06696525e-01
6.58366919e-01 -1.01874065e+00 -2.21102759e-02 4.90175933e-01
-2.98416018e-02 8.17766190e-01 6.46373332e-01 -6.17137015e-01
-1.55709577e+00 -1.02904785e+00 5.19002497e-01 -3.42770040e-01
5.13452649e-01 -5.87845922e-01 -8.62790525e-01 5.98500073e-01
6.84862211e-02 -1.81064814e-01 1.37134981e+00 1.86329216e-01
-2.74055898e-01 -2.67301708e-01 -1.12128532e+00 4.41284031e-01
4.92789388e-01 -4.93084252e-01 -3.64530891e-01 2.83347845e-01
3.84850442e-01 1.11245304e-01 -1.21680605e+00 5.69252908e-01
7.92849422e-01 -9.29905474e-01 9.34801996e-01 -9.89708066e-01
4.71156299e-01 -2.58806020e-01 -2.61572391e-01 -1.53775680e+00
-2.81882763e-01 2.29834896e-02 -6.57955036e-02 9.02631462e-01
2.39506066e-01 -8.17196786e-01 6.29101217e-01 3.28138888e-01
-3.52655649e-02 -9.24201608e-01 -1.00028181e+00 -4.28619832e-01
-2.48682871e-02 2.43774708e-02 7.92976439e-01 1.02018297e+00
-6.23423792e-02 1.33026689e-01 -3.75722855e-01 3.15279961e-02
3.24052393e-01 3.19339126e-01 7.15725422e-01 -1.37448072e+00
-5.06236494e-01 -2.16307983e-01 -8.24061453e-01 -1.05202682e-01
3.60074788e-02 -1.07614446e+00 -4.45231944e-01 -1.18701458e+00
2.15899840e-01 -2.97247112e-01 -6.48055077e-01 2.77716309e-01
-1.31088123e-02 3.58893685e-02 7.76816085e-02 4.55722697e-02
-7.46501237e-02 6.63222909e-01 8.20876777e-01 -1.85376883e-01
-1.60019219e-01 -3.99236202e-01 -6.97483242e-01 3.29041123e-01
9.08415973e-01 -4.37773436e-01 -1.71882167e-01 -2.02617779e-01
3.51640522e-01 -8.28096271e-03 2.10554034e-01 -1.18178189e+00
1.17063507e-01 -3.82620618e-02 7.15282500e-01 -3.63355011e-01
5.53792536e-01 -7.67000258e-01 9.93405402e-01 6.69004679e-01
-2.18147486e-01 1.59130737e-01 3.49922657e-01 7.02607691e-01
-2.49291107e-01 -9.32510272e-02 6.06977463e-01 -3.88383597e-01
-4.07013297e-01 1.36174217e-01 -4.74113345e-01 -7.11833477e-01
1.03582537e+00 6.17958345e-02 -2.80677766e-01 1.28670976e-01
-1.04207194e+00 -1.79919824e-01 4.16476130e-01 4.74230617e-01
4.62383986e-01 -1.07043409e+00 -7.36021459e-01 4.97033745e-01
3.04006249e-01 -4.54883993e-01 4.71348852e-01 7.28704929e-01
-6.84949338e-01 8.10637951e-01 -8.66172552e-01 -6.10801935e-01
-1.16776657e+00 9.30824518e-01 4.76209313e-01 -4.43118185e-01
-3.99747968e-01 3.83417517e-01 2.95590103e-01 -5.74889600e-01
-3.11774135e-01 -8.18067566e-02 -2.49139965e-01 4.11746711e-01
3.06468695e-01 1.83413506e-01 3.28331381e-01 -3.59405994e-01
-2.35433027e-01 2.79084831e-01 1.39379734e-02 2.70173669e-01
1.81700659e+00 1.45496517e-01 -2.60403633e-01 8.16122174e-01
1.26474750e+00 -1.03911065e-01 -6.02039218e-01 2.77271986e-01
-3.75734568e-02 -1.46309465e-01 -1.31984636e-01 -9.75739598e-01
-8.50912809e-01 9.20964420e-01 8.54603589e-01 5.25173172e-02
1.09493744e+00 -3.49574447e-01 3.80934089e-01 4.42539245e-01
3.33946019e-01 -7.59722948e-01 -1.44791707e-01 1.40891477e-01
7.33545065e-01 -9.98217165e-01 -4.07766104e-02 -1.45611286e-01
-2.21163541e-01 1.33916950e+00 1.55689090e-01 6.39780611e-02
6.29313529e-01 2.12045372e-01 1.75245583e-01 -2.18242034e-01
-1.02682757e+00 -2.01473638e-01 3.96308787e-02 4.66959000e-01
5.51339924e-01 2.55665898e-01 -4.32125330e-01 5.83113432e-01
1.91796839e-01 4.09259081e-01 4.67223167e-01 6.95143342e-01
-2.07813755e-02 -1.61144054e+00 -1.77269578e-01 4.08636302e-01
-4.69991744e-01 9.32721794e-02 -4.23736632e-01 6.96921527e-01
3.11085939e-01 3.33582133e-01 -3.33377779e-01 -4.49056625e-01
3.07771023e-02 7.13104844e-01 2.03527644e-01 -4.53448027e-01
-6.33232951e-01 1.19263984e-01 1.86764300e-02 -3.69904459e-01
-3.93878877e-01 -6.06116891e-01 -8.04189622e-01 -1.03909172e-01
-5.67426942e-02 1.89710230e-01 1.16104603e+00 7.62687683e-01
1.10973036e+00 8.75532150e-01 2.86198109e-01 -6.54282808e-01
-2.88324207e-01 -1.13739467e+00 -5.31813979e-01 4.71864790e-01
3.07530165e-01 -6.51926219e-01 -5.69651127e-02 -2.23971251e-02] | [5.968252658843994, 5.656187534332275] |
34583b2a-9fa9-452b-81a3-ec2ace1dabf6 | fusemodnet-real-time-camera-and-lidar-based | 1910.05395 | null | https://arxiv.org/abs/1910.05395v3 | https://arxiv.org/pdf/1910.05395v3.pdf | FuseMODNet: Real-Time Camera and LiDAR based Moving Object Detection for robust low-light Autonomous Driving | Moving object detection is a critical task for autonomous vehicles. As dynamic objects represent higher collision risk than static ones, our own ego-trajectories have to be planned attending to the future states of the moving elements of the scene. Motion can be perceived using temporal information such as optical flow. Conventional optical flow computation is based on camera sensors only, which makes it prone to failure in conditions with low illumination. On the other hand, LiDAR sensors are independent of illumination, as they measure the time-of-flight of their own emitted lasers. In this work, we propose a robust and real-time CNN architecture for Moving Object Detection (MOD) under low-light conditions by capturing motion information from both camera and LiDAR sensors. We demonstrate the impact of our algorithm on KITTI dataset where we simulate a low-light environment creating a novel dataset "Dark KITTI". We obtain a 10.1% relative improvement on Dark-KITTI, and a 4.25% improvement on standard KITTI relative to our baselines. The proposed algorithm runs at 18 fps on a standard desktop GPU using $256\times1224$ resolution images. | ['Ganesh Sistu', 'Mohamed Ramzy', 'Hazem Rashed', 'Senthil Yogamani', 'Victor Vaquero', 'Ahmad El Sallab'] | 2019-10-11 | null | null | null | null | ['moving-object-detection'] | ['computer-vision'] | [-5.42270280e-02 -5.53790212e-01 8.79695490e-02 -2.65346825e-01
5.09486673e-03 -5.48882604e-01 4.69672918e-01 -3.31069589e-01
-9.04011846e-01 4.78215516e-01 -3.64300728e-01 -2.08933696e-01
3.18172216e-01 -9.71403003e-01 -7.39436269e-01 -6.25521779e-01
-1.38365045e-01 1.68426648e-01 6.70963347e-01 -1.14302449e-01
6.36338964e-02 7.27656484e-01 -1.79506123e+00 -2.71240883e-02
4.42733288e-01 1.02728570e+00 3.33532065e-01 1.08083785e+00
3.81837115e-02 7.14532197e-01 -2.99529016e-01 -4.62182015e-02
5.88153243e-01 -4.53818552e-02 -2.05108806e-01 -3.06644719e-02
9.84568954e-01 -6.83922231e-01 -7.60673821e-01 9.71432626e-01
1.04114503e-01 3.51343244e-01 2.92277008e-01 -1.33707690e+00
-2.50487737e-02 -1.23798288e-01 -6.73571050e-01 6.63640440e-01
1.62542850e-01 6.70387328e-01 4.96440440e-01 -9.97274220e-01
7.02545285e-01 1.20035338e+00 4.59208190e-01 5.07685483e-01
-9.72890615e-01 -7.69410491e-01 1.89321786e-02 5.69619238e-01
-1.22155535e+00 -6.82090402e-01 5.14653146e-01 -4.67102736e-01
1.04302824e+00 -5.90000413e-02 6.71799839e-01 7.28025615e-01
5.28930247e-01 2.61718869e-01 6.59761965e-01 -1.78864822e-02
2.43399039e-01 -9.54482630e-02 2.05824915e-02 8.26330543e-01
5.05657494e-01 5.60743093e-01 -5.88059843e-01 3.36982161e-01
6.20576024e-01 -8.60893633e-03 -1.94280311e-01 -1.93079457e-01
-1.29206908e+00 5.38765728e-01 5.82798898e-01 -1.19417995e-01
-1.13336198e-01 6.47180498e-01 1.02828935e-01 1.28796786e-01
1.72306582e-01 -2.08697826e-01 -1.65371388e-01 -1.90585941e-01
-8.63228440e-01 2.45236471e-01 5.01770854e-01 9.69842076e-01
1.08334398e+00 2.21580058e-01 4.73091044e-02 7.57905394e-02
4.62754309e-01 1.04407024e+00 1.55304819e-01 -1.27682376e+00
6.18771732e-01 2.81156123e-01 4.21530455e-01 -1.20661843e+00
-4.75368440e-01 -2.41270959e-01 -5.99136233e-01 7.92952478e-01
6.04802012e-01 -2.68640071e-01 -8.72030854e-01 1.53818929e+00
5.01440883e-01 6.72624886e-01 1.26410812e-01 1.20167720e+00
6.45811439e-01 7.83407688e-01 -2.03240141e-01 -3.04691762e-01
1.07551444e+00 -8.65783811e-01 -6.83354914e-01 -4.57891136e-01
5.69772005e-01 -7.96846986e-01 6.60781384e-01 2.95173258e-01
-8.78149450e-01 -6.40841484e-01 -1.12289953e+00 -1.64359912e-01
-2.39412814e-01 1.20153666e-01 3.82433742e-01 6.56006277e-01
-1.05834091e+00 4.01909739e-01 -1.25813985e+00 -3.12408566e-01
4.15993363e-01 2.67682642e-01 -3.26791495e-01 -1.97292373e-01
-7.68585742e-01 8.84939551e-01 2.14560285e-01 2.34556437e-01
-9.37758923e-01 -6.92683101e-01 -8.03710818e-01 -5.29965758e-02
3.81978601e-01 -5.62608957e-01 1.08638465e+00 -6.56850636e-01
-1.52787292e+00 5.96289575e-01 -6.49714947e-01 -7.16420412e-01
7.17941523e-01 -3.58134776e-01 -2.86818057e-01 3.84986430e-01
2.12450065e-02 8.99518967e-01 7.56680489e-01 -1.03313351e+00
-1.07856023e+00 -2.25836813e-01 8.53552148e-02 -9.71708633e-03
-8.35266262e-02 -1.69233635e-01 -4.96043533e-01 2.21751891e-02
-5.99357784e-02 -1.19364727e+00 -2.60100275e-01 5.45977116e-01
1.67237837e-02 4.76272628e-02 1.42986798e+00 1.37951270e-01
7.39649117e-01 -2.05989194e+00 -4.65475678e-01 -2.34124690e-01
1.97802961e-01 6.40729904e-01 -7.42381141e-02 2.67948806e-02
2.55058020e-01 -2.97479391e-01 1.03050079e-02 -5.04251003e-01
-4.06505674e-01 3.10012758e-01 -2.92976677e-01 7.85899341e-01
2.52804607e-01 7.95998514e-01 -9.95188594e-01 -4.84939843e-01
6.34176195e-01 6.06820047e-01 -5.16705692e-01 -2.49569118e-02
-1.44131884e-01 4.80690628e-01 -2.13679850e-01 3.95161361e-01
9.94021773e-01 1.97094709e-01 -3.07065994e-01 -1.06375381e-01
-7.44675875e-01 -2.24497765e-02 -1.15401995e+00 1.49100280e+00
-3.87211144e-01 1.40383887e+00 -2.87540094e-03 -4.73228693e-01
7.56380260e-01 -2.66299676e-02 4.13612336e-01 -8.33810210e-01
3.16557109e-01 1.13828368e-01 1.63201943e-01 -5.67208350e-01
7.38313913e-01 3.84817533e-02 3.35084677e-01 2.43385881e-01
-4.49374199e-01 -2.61144675e-02 4.43149477e-01 2.57703036e-01
1.17087233e+00 1.46449581e-01 -1.34077564e-01 -1.32520333e-01
4.73002106e-01 1.74157634e-01 6.89485729e-01 5.45352995e-01
-5.74521005e-01 4.02376026e-01 2.62667108e-02 -7.20631897e-01
-8.76525342e-01 -1.00212240e+00 -1.50568336e-01 5.41375458e-01
6.94592655e-01 -1.55386433e-01 -4.57026064e-01 -2.80979067e-01
-3.28014903e-02 6.51436627e-01 -4.68074560e-01 -2.28783768e-02
-9.58323121e-01 -4.83153462e-01 3.86056751e-01 4.87189651e-01
7.96028376e-01 -7.57270694e-01 -1.64109862e+00 2.66099215e-01
-4.21016552e-02 -1.71635413e+00 -2.75575757e-01 -3.37891817e-01
-7.80191600e-01 -1.12242448e+00 -1.62667751e-01 -3.23086441e-01
4.26044524e-01 1.06369662e+00 8.81611705e-01 -5.73854856e-02
-4.37751889e-01 4.08442348e-01 -9.46049765e-02 -4.70935732e-01
-1.19473122e-01 -3.56342435e-01 2.21255243e-01 2.90505260e-01
3.79813850e-01 -3.08476806e-01 -1.01632047e+00 3.98741007e-01
-7.05600142e-01 8.56878757e-02 1.91747054e-01 3.58996600e-01
3.40216666e-01 4.99354228e-02 -1.60852119e-01 -2.15141401e-01
-3.16643417e-01 -2.31553316e-01 -1.18508232e+00 -2.89558291e-01
-2.72016406e-01 -2.29759201e-01 5.42881548e-01 -2.93398231e-01
-1.19602168e+00 4.17196512e-01 2.97854155e-01 -7.80878186e-01
-1.68572068e-01 -1.23078041e-01 2.07746714e-01 -2.27742881e-01
6.01351619e-01 -1.68841761e-02 -4.63399589e-02 4.69639450e-02
3.35946828e-01 2.86747932e-01 6.92666650e-01 -1.93320066e-01
1.02364874e+00 1.41118169e+00 4.44310963e-01 -1.16713119e+00
-5.06003320e-01 -5.57163835e-01 -7.40566850e-01 -7.70248473e-01
9.88130510e-01 -9.17687356e-01 -1.15154874e+00 5.90528607e-01
-1.35218978e+00 -4.48584467e-01 -2.43947916e-02 7.29309499e-01
-3.92096460e-01 3.40166986e-01 -3.22854459e-01 -8.98125947e-01
-2.11795241e-01 -1.22158504e+00 9.36267078e-01 4.48371410e-01
2.08391637e-01 -9.08180833e-01 1.75268836e-02 1.33372694e-01
4.85226303e-01 2.98786283e-01 1.45708069e-01 3.93752813e-01
-1.18140101e+00 -1.06106222e-01 -5.01859486e-01 6.76749647e-02
-3.11118718e-02 4.61259007e-01 -1.12627161e+00 -3.61248314e-01
-5.80000654e-02 1.83180198e-01 1.09199059e+00 4.45401251e-01
7.38957286e-01 1.45309642e-01 -5.06975949e-01 8.15317750e-01
1.73646200e+00 3.50333661e-01 6.35379136e-01 2.44608581e-01
9.07197058e-01 6.10672534e-01 8.24854314e-01 4.12926883e-01
4.77083445e-01 8.20370913e-01 7.83679962e-01 7.90001005e-02
-3.62126648e-01 1.52328402e-01 5.39710879e-01 3.01734000e-01
-1.93073660e-01 -4.40319926e-01 -1.01953888e+00 5.50226152e-01
-1.76898265e+00 -1.17340660e+00 -8.26002121e-01 2.29285741e+00
1.36365697e-01 2.03817472e-01 -1.31601617e-01 -4.70807739e-02
5.20712912e-01 1.21732885e-02 -5.44888437e-01 -7.49058351e-02
1.23959081e-02 -3.98580693e-02 9.71310139e-01 7.53600955e-01
-1.06331873e+00 1.01048255e+00 5.59278107e+00 2.32168734e-01
-1.54910719e+00 1.67143971e-01 2.38499075e-01 -5.30694008e-01
1.33112013e-01 2.17094854e-01 -1.08884871e+00 5.38540542e-01
1.02648044e+00 -1.77444398e-01 -3.36535871e-02 5.56116045e-01
7.74125278e-01 -6.65016115e-01 -9.30495858e-01 1.10604990e+00
1.11240111e-02 -1.29761326e+00 -2.85500199e-01 2.88192779e-01
5.33336699e-01 4.99695718e-01 -1.65001266e-02 -1.15279384e-01
4.06707488e-02 -7.10548341e-01 9.16966200e-01 4.34625804e-01
5.44716001e-01 -6.69299543e-01 5.45284867e-01 3.88259381e-01
-1.51400769e+00 -4.76347506e-02 -4.95678455e-01 -3.00783217e-01
4.70868230e-01 5.50695956e-01 -8.50256920e-01 2.63709337e-01
7.75231302e-01 9.85142052e-01 -4.54680383e-01 1.07926226e+00
1.58401117e-01 4.58036661e-01 -6.24469221e-01 1.20311007e-01
3.06781471e-01 -2.99512506e-01 7.69467235e-01 1.23563731e+00
5.02317011e-01 1.61708117e-01 2.11105064e-01 7.52979875e-01
1.98124155e-01 -4.46115583e-01 -7.78957427e-01 3.99474055e-01
1.93969131e-01 1.26348722e+00 -8.28331172e-01 -3.87568980e-01
-4.85724360e-01 7.18536615e-01 3.57904434e-02 4.08798665e-01
-1.01738060e+00 -2.76740551e-01 1.17054713e+00 1.77052736e-01
4.63218898e-01 -8.27231646e-01 -1.06541477e-01 -1.17783928e+00
-1.72173008e-02 -1.34145347e-02 -2.93617174e-02 -7.03226328e-01
-5.23419797e-01 4.36362952e-01 -7.81734437e-02 -1.59454203e+00
2.16627475e-02 -8.75112295e-01 -5.87114871e-01 5.80514193e-01
-1.90644336e+00 -7.08779693e-01 -8.23732495e-01 6.36347353e-01
6.68149352e-01 1.27559394e-01 2.80960977e-01 4.85035658e-01
-5.75950027e-01 8.72672200e-02 3.71632539e-02 8.70683268e-02
5.13822377e-01 -7.91508317e-01 5.62320173e-01 1.36119246e+00
1.13320082e-01 3.12559307e-01 8.82407904e-01 -5.21626294e-01
-1.78969204e+00 -1.20449245e+00 7.34014094e-01 -5.39073050e-01
5.35359085e-01 -2.93824464e-01 -8.03831398e-01 5.04749954e-01
1.40262887e-01 4.97235954e-01 6.74760863e-02 -7.43185639e-01
-1.72352955e-01 -4.21834499e-01 -1.00061965e+00 5.44216990e-01
1.20554268e+00 -1.81555465e-01 -1.83956549e-02 3.85174491e-02
5.46103477e-01 -5.64871967e-01 -1.94089517e-01 3.24744910e-01
7.14904904e-01 -1.24620688e+00 9.50063050e-01 -1.17543444e-01
1.60255898e-02 -8.17307055e-01 -6.46609217e-02 -8.57147872e-01
-6.65963367e-02 -6.09282553e-01 -9.02076438e-02 7.07703710e-01
-2.75053661e-02 -7.29815543e-01 8.39853406e-01 4.55948979e-01
-1.74851000e-01 -2.49793440e-01 -1.18271017e+00 -8.50056350e-01
-2.25921601e-01 -9.96214926e-01 1.60103306e-01 5.08481622e-01
-5.47221601e-01 1.52142525e-01 -3.13690126e-01 5.03324091e-01
9.30568635e-01 5.21959402e-02 1.04289699e+00 -1.13455248e+00
1.33527026e-01 -3.23607296e-01 -8.31258476e-01 -1.08711815e+00
7.72209167e-02 -5.07212460e-01 2.77203798e-01 -1.22493124e+00
-1.75805017e-01 -2.60377139e-01 1.19456194e-01 1.04611591e-01
9.10812020e-02 6.19937718e-01 3.41734916e-01 1.21819392e-01
-5.39234281e-01 4.53335285e-01 1.03947306e+00 -6.51025102e-02
-1.36320069e-01 -5.64919263e-02 2.84725398e-01 7.50380516e-01
7.11903155e-01 -4.81441915e-01 -4.06476945e-01 -8.11026633e-01
1.80174872e-01 -4.92942929e-02 6.52266920e-01 -1.51025343e+00
5.93745887e-01 -3.18147808e-01 3.11960340e-01 -8.35483193e-01
6.85603619e-01 -8.21991742e-01 1.56317472e-01 7.55457342e-01
1.08542517e-01 2.63900727e-01 4.65360641e-01 7.49845684e-01
6.03393503e-02 -7.97772482e-02 9.61612105e-01 -4.97365184e-02
-1.03091395e+00 5.12844384e-01 -6.88745201e-01 -9.64421779e-02
1.27120531e+00 -4.55335200e-01 -5.24914980e-01 -2.16829464e-01
-1.16894126e-01 2.79593945e-01 5.55958152e-01 4.02242333e-01
8.74656439e-01 -1.05965149e+00 -5.84451497e-01 2.73379773e-01
5.62262349e-02 1.17761284e-01 2.57788569e-01 8.57574582e-01
-1.05636513e+00 6.42519832e-01 -2.74998128e-01 -1.00018656e+00
-1.45599806e+00 4.27544147e-01 4.35954124e-01 4.48287666e-01
-8.35208416e-01 7.06666112e-01 3.28954458e-01 2.86924213e-01
-1.35931326e-02 -5.46259999e-01 -5.96584529e-02 -3.68245170e-02
6.59330547e-01 7.07984805e-01 4.37328480e-02 -9.80496466e-01
-4.27576602e-01 8.81712615e-01 1.18955672e-01 -2.36714378e-01
8.42402756e-01 -4.25910413e-01 1.37964204e-01 3.51410538e-01
1.20490849e+00 -3.86243686e-02 -1.62305593e+00 -1.31268784e-01
-2.58869708e-01 -9.45075572e-01 3.22602183e-01 -2.82251704e-02
-1.18152416e+00 1.11468756e+00 9.76014853e-01 -6.89199269e-02
9.27789569e-01 -3.13168317e-01 9.33086455e-01 4.83820587e-01
6.18811727e-01 -8.09370220e-01 7.95862824e-02 6.45327747e-01
2.47651219e-01 -1.38112700e+00 -4.09615003e-02 -4.38738108e-01
-1.94484189e-01 1.26947474e+00 7.38316417e-01 -1.55549780e-01
4.80663449e-01 2.53654182e-01 4.29600090e-01 -7.16228485e-02
-8.62948895e-01 -6.16026163e-01 -5.86049538e-03 4.76979285e-01
-1.10425234e-01 -1.22440219e-01 9.20634195e-02 -6.30016863e-01
-7.36463442e-02 1.09147606e-02 8.47669303e-01 8.90914977e-01
-5.33484519e-01 -7.00344265e-01 -3.86672229e-01 -2.25307979e-02
-2.10228592e-01 2.51839221e-01 1.21670455e-01 6.95124328e-01
3.36333543e-01 1.17646813e+00 5.26624620e-01 -1.89528525e-01
2.81848371e-01 -3.06223094e-01 3.30146909e-01 -3.60948265e-01
-2.38511036e-03 -1.84630245e-01 -2.36108735e-01 -9.03932691e-01
-6.60364091e-01 -7.58886099e-01 -1.39025176e+00 -5.11855304e-01
-8.71867687e-02 -3.86850685e-01 9.09848928e-01 8.03931892e-01
4.34291720e-01 3.35238189e-01 5.74315369e-01 -1.19313967e+00
1.25646755e-01 -5.04755795e-01 -1.33769244e-01 2.18745485e-01
7.08298326e-01 -8.03940535e-01 -4.15671438e-01 1.77097097e-01] | [8.12402629852295, -1.5170823335647583] |
9c93ae77-4d7d-4da1-99fe-3baf782b915a | asking-questions-the-human-way-scalable | 2002.00748 | null | https://arxiv.org/abs/2002.00748v2 | https://arxiv.org/pdf/2002.00748v2.pdf | Asking Questions the Human Way: Scalable Question-Answer Generation from Text Corpus | The ability to ask questions is important in both human and machine intelligence. Learning to ask questions helps knowledge acquisition, improves question-answering and machine reading comprehension tasks, and helps a chatbot to keep the conversation flowing with a human. Existing question generation models are ineffective at generating a large amount of high-quality question-answer pairs from unstructured text, since given an answer and an input passage, question generation is inherently a one-to-many mapping. In this paper, we propose Answer-Clue-Style-aware Question Generation (ACS-QG), which aims at automatically generating high-quality and diverse question-answer pairs from unlabeled text corpus at scale by imitating the way a human asks questions. Our system consists of: i) an information extractor, which samples from the text multiple types of assistive information to guide question generation; ii) neural question generators, which generate diverse and controllable questions, leveraging the extracted assistive information; and iii) a neural quality controller, which removes low-quality generated data based on text entailment. We compare our question generation models with existing approaches and resort to voluntary human evaluation to assess the quality of the generated question-answer pairs. The evaluation results suggest that our system dramatically outperforms state-of-the-art neural question generation models in terms of the generation quality, while being scalable in the meantime. With models trained on a relatively smaller amount of data, we can generate 2.8 million quality-assured question-answer pairs from a million sentences found in Wikipedia. | ['Yancheng He', 'Di Niu', 'Haolan Chen', 'Haojie Wei', 'Bang Liu'] | 2020-01-27 | null | null | null | null | ['question-answer-generation'] | ['natural-language-processing'] | [ 1.73688188e-01 7.69509077e-01 3.51879686e-01 -4.11696255e-01
-1.39374697e+00 -7.42969692e-01 7.26348460e-01 -2.21185219e-02
-3.30957651e-01 1.08635283e+00 5.55847943e-01 -5.10030448e-01
1.02206163e-01 -1.10832179e+00 -5.68628848e-01 1.25334486e-01
6.18132830e-01 9.54809070e-01 1.10559076e-01 -7.70981312e-01
3.38800013e-01 -3.49288046e-01 -1.48699522e+00 5.30253589e-01
1.68074846e+00 7.58146882e-01 3.60532850e-01 1.18507934e+00
-7.24556565e-01 1.48404980e+00 -1.06843174e+00 -6.94599271e-01
-3.86243388e-02 -1.25970709e+00 -1.50835693e+00 -2.58668184e-01
4.37510997e-01 -6.08031750e-01 -1.56001553e-01 5.31041205e-01
4.79119509e-01 3.58352453e-01 5.92573822e-01 -1.05865777e+00
-1.27731979e+00 5.82573056e-01 4.70132411e-01 1.48298725e-01
1.07438457e+00 5.65004230e-01 1.34198213e+00 -8.39654684e-01
7.26209581e-01 1.21826720e+00 2.72529721e-01 1.10808516e+00
-8.72368336e-01 -2.61130095e-01 -3.20443064e-01 1.74371421e-01
-4.94099915e-01 -4.95218456e-01 5.54251194e-01 -1.85307145e-01
9.85019863e-01 4.60370719e-01 4.98363227e-01 1.02007687e+00
6.75000399e-02 8.76932204e-01 6.98626459e-01 -4.56642777e-01
1.68256551e-01 1.93583235e-01 1.16497934e-01 6.61762536e-01
-9.71375108e-02 -2.92222500e-01 -5.64532995e-01 -2.37506419e-01
3.30986440e-01 -4.96830940e-01 -4.46976900e-01 2.81505942e-01
-1.19402409e+00 1.14319193e+00 3.38811696e-01 1.74506620e-01
-4.87396091e-01 -8.71412903e-02 5.82587570e-02 8.74873221e-01
3.22542548e-01 1.39568150e+00 -4.71152753e-01 -4.30634230e-01
-6.95980132e-01 9.23469961e-01 1.56814599e+00 1.04986846e+00
7.61927009e-01 -2.31541544e-01 -1.03872943e+00 8.25840771e-01
5.60129546e-02 6.28671348e-01 6.39501870e-01 -1.40562439e+00
8.74261320e-01 9.64278519e-01 5.08589327e-01 -6.88193977e-01
-5.37506118e-02 4.52562608e-03 -5.13109565e-01 -2.88829297e-01
7.02690423e-01 -5.31461120e-01 -4.66337144e-01 1.68980515e+00
3.30224484e-01 -5.97916067e-01 2.43395329e-01 6.69605613e-01
1.39721584e+00 8.03094923e-01 -2.86124855e-01 4.31327894e-02
1.45789683e+00 -1.37879074e+00 -9.56511557e-01 -3.74612719e-01
6.00188494e-01 -6.36823535e-01 1.65341437e+00 -1.99727602e-02
-1.44591117e+00 -6.44343019e-01 -5.74334919e-01 -5.52045345e-01
-7.53089190e-02 -8.99663642e-02 7.94737563e-02 3.97806346e-01
-1.14894807e+00 2.69555688e-01 -1.24501642e-02 -1.15437441e-01
2.58635700e-01 -1.69144303e-01 -2.09047366e-02 -2.09607974e-01
-1.56136954e+00 9.26260114e-01 -8.35857540e-02 -2.42150709e-01
-7.71837234e-01 -7.21988738e-01 -9.98655736e-01 1.83225587e-01
3.44191104e-01 -1.21768641e+00 1.90677404e+00 -7.82177866e-01
-1.79811907e+00 6.33061290e-01 -4.41197842e-01 -3.36069971e-01
3.93000662e-01 -3.23690683e-01 1.95691083e-02 4.82722878e-01
5.88544071e-01 9.48652923e-01 6.18898273e-01 -9.42654014e-01
-5.18044591e-01 4.15348373e-02 4.47696745e-01 4.56493706e-01
-1.25020951e-01 -2.21803010e-01 -3.31106782e-02 -3.59744877e-01
-3.35299879e-01 -4.37711358e-01 -1.52694196e-01 -1.14257812e-01
-3.56866717e-01 -7.78313160e-01 3.49571615e-01 -1.04295969e+00
1.10065281e+00 -1.34740710e+00 -1.02459423e-01 -3.11883628e-01
2.66334057e-01 3.13399792e-01 -7.17529535e-01 7.48376369e-01
4.97547209e-01 2.11949736e-01 -2.02134386e-01 -2.16983214e-01
2.17308387e-01 -4.19523790e-02 -4.35301274e-01 -5.36717057e-01
6.26250327e-01 1.53323150e+00 -1.40157831e+00 -4.91933346e-01
-3.21832806e-01 -1.30212277e-01 -8.72442186e-01 1.23020852e+00
-8.85156751e-01 3.66592556e-01 -5.73032916e-01 2.59173959e-01
7.31381699e-02 -4.75287765e-01 -2.86185652e-01 4.16287303e-01
4.86193210e-01 9.74278867e-01 -6.33634031e-01 1.63583517e+00
-7.17109144e-01 5.50497591e-01 4.35580425e-02 -4.38545018e-01
1.13860774e+00 5.84776163e-01 -2.82926172e-01 -9.68214214e-01
-8.06871504e-02 1.45838127e-01 -3.72240506e-02 -1.02903903e+00
9.39038038e-01 -1.07796207e-01 -2.54019588e-01 1.14631712e+00
2.58331746e-01 -8.91175508e-01 7.23384738e-01 6.89568460e-01
1.52339470e+00 -1.12378284e-01 9.72102880e-02 -3.72895114e-02
6.31645560e-01 4.52447981e-02 -3.06120813e-02 1.03494430e+00
-1.49746537e-01 6.70031071e-01 6.57739937e-01 -2.64353156e-02
-8.65999460e-01 -1.11575210e+00 5.35859168e-01 1.13292468e+00
-1.68893069e-01 -1.85425788e-01 -1.13441670e+00 -8.70488226e-01
-1.50132835e-01 1.18359804e+00 -5.53474605e-01 -1.75110549e-01
-6.18794560e-01 1.20865926e-01 6.75574422e-01 2.47585192e-01
5.61513841e-01 -1.77029860e+00 -4.09912020e-01 4.14551675e-01
-1.01178336e+00 -9.18862581e-01 -8.59434545e-01 -3.96844685e-01
-7.23913610e-01 -1.19043386e+00 -6.51629090e-01 -8.79446447e-01
5.78051090e-01 2.06185982e-01 1.96286035e+00 4.44789529e-01
1.84551641e-01 4.73858595e-01 -6.23689532e-01 -3.20698351e-01
-1.02544367e+00 5.38630664e-01 -5.58478177e-01 -2.61829197e-01
3.01847100e-01 -3.74796212e-01 -7.40563512e-01 1.93916351e-01
-1.00817680e+00 8.38586837e-02 3.48342299e-01 1.00517869e+00
2.16602646e-02 -6.91119552e-01 1.51941741e+00 -1.02687538e+00
1.59065700e+00 -7.38421679e-01 -8.17539021e-02 4.54870194e-01
-4.61906105e-01 3.69918734e-01 9.59804118e-01 -1.00766644e-01
-1.32630229e+00 -6.06174350e-01 -4.59539622e-01 3.70778322e-01
-1.39884740e-01 2.77851999e-01 -3.00288647e-01 5.36666572e-01
1.14022338e+00 3.53311658e-01 2.24735647e-01 -1.18050687e-01
1.02571940e+00 8.49717736e-01 5.42979300e-01 -6.73553884e-01
8.16895127e-01 -1.49182767e-01 -7.53970206e-01 -4.87547398e-01
-1.16879272e+00 -3.89180809e-01 -3.66967991e-02 -3.63911957e-01
8.23145628e-01 -5.81720114e-01 -7.43541062e-01 2.93355018e-01
-1.47840810e+00 -6.23459995e-01 -6.59024715e-01 -2.13430941e-01
-6.04341805e-01 4.26791012e-02 -8.55344534e-01 -7.70633340e-01
-9.61050630e-01 -7.09699810e-01 9.50949132e-01 5.42705297e-01
-6.47837579e-01 -9.55213904e-01 2.86976695e-01 1.31030607e+00
7.27540016e-01 -1.14216767e-01 9.18629229e-01 -8.86881351e-01
-7.78754890e-01 -1.84545308e-01 -1.65931255e-01 5.35422146e-01
2.68482864e-01 -4.65252250e-01 -7.02384651e-01 1.37776017e-01
2.38839060e-01 -1.04863095e+00 4.58140761e-01 -2.10906759e-01
8.22008908e-01 -9.11302745e-01 4.07394260e-01 -1.62262186e-01
7.05931544e-01 -1.30312145e-01 7.52093494e-01 2.29054168e-02
4.35523033e-01 1.16726732e+00 4.73500431e-01 7.30300397e-02
1.00964749e+00 1.19133845e-01 2.84018219e-01 2.90162206e-01
-2.91780293e-01 -6.00421667e-01 2.75641292e-01 1.05946934e+00
4.92118955e-01 -4.81773436e-01 -5.69659591e-01 9.94869411e-01
-1.50110626e+00 -1.10949194e+00 -3.17027897e-01 1.63880825e+00
1.58820283e+00 -1.21943615e-01 4.57932465e-02 -2.30356082e-01
4.13314760e-01 4.94864620e-02 -6.06189132e-01 -4.58137244e-01
1.36500925e-01 5.60834169e-01 -5.33285558e-01 8.48088682e-01
-2.36851126e-01 7.92642653e-01 5.74753094e+00 4.88281846e-01
-4.50118870e-01 2.27431841e-02 6.62009776e-01 -9.40595269e-02
-1.00322723e+00 -9.44694430e-02 -4.68423724e-01 4.42279428e-01
1.03186584e+00 -4.11529243e-01 5.51210403e-01 6.02067590e-01
2.00211018e-01 -1.24413580e-01 -1.19017005e+00 4.71840113e-01
4.30772007e-01 -1.37705517e+00 5.05347669e-01 -3.46507967e-01
8.22348475e-01 -3.71760279e-01 -3.29748064e-01 6.36514843e-01
4.79971379e-01 -1.19565344e+00 4.96403337e-01 9.09045815e-01
4.16518539e-01 -5.61214805e-01 7.79096007e-01 7.89718091e-01
-6.60994232e-01 5.52970096e-02 -8.33944455e-02 -4.26585585e-01
4.81984168e-01 5.70058763e-01 -1.07121050e+00 4.44790572e-01
2.85802484e-01 1.42095741e-02 -7.36699402e-01 5.45827091e-01
-8.23173106e-01 7.74432600e-01 1.01376303e-01 -6.70905828e-01
1.25698492e-01 -1.23478025e-01 2.35420987e-01 8.07374179e-01
2.60402799e-01 2.76106805e-01 -2.55382597e-01 1.35394013e+00
-6.74026847e-01 2.27520242e-01 -4.82139766e-01 -6.50216863e-02
7.14069843e-01 1.32588792e+00 8.08636695e-02 -4.94671196e-01
-1.41090691e-01 1.03000927e+00 6.74901366e-01 2.85945773e-01
-2.65157193e-01 -8.49901974e-01 2.46020332e-01 1.37903064e-01
-2.42463350e-02 2.26190358e-01 -3.40037107e-01 -1.12094867e+00
5.19973218e-01 -1.48551095e+00 1.56931669e-01 -1.07494664e+00
-1.51814330e+00 7.77222574e-01 -5.43876350e-01 -7.21987426e-01
-1.01147771e+00 -2.70645380e-01 -9.43414748e-01 1.16618359e+00
-1.54755068e+00 -8.34704518e-01 -5.00791073e-01 3.25344473e-01
8.29320312e-01 -1.37079880e-01 8.50702703e-01 3.00602373e-02
-7.20290616e-02 6.39037132e-01 -3.86313528e-01 2.98165441e-01
7.71029294e-01 -1.56873655e+00 9.68963921e-01 6.47308350e-01
5.07702120e-02 5.82206428e-01 5.36829591e-01 -5.01736104e-01
-1.19040775e+00 -1.17289495e+00 1.67566037e+00 -1.11094761e+00
4.49026018e-01 -2.70771027e-01 -1.16508114e+00 2.62584716e-01
7.56936967e-01 -5.94303906e-01 8.06463957e-01 -1.48847520e-01
-2.19775140e-01 1.47824749e-01 -1.20666087e+00 6.74431980e-01
7.92947352e-01 -7.85100341e-01 -1.28231299e+00 4.58081156e-01
1.26320994e+00 -2.59330571e-01 -6.78757966e-01 -9.68913510e-02
1.71256915e-01 -9.07284260e-01 5.68499327e-01 -7.76278377e-01
1.05551279e+00 -8.09414461e-02 2.37832263e-01 -1.47864342e+00
1.18043292e-02 -8.84408057e-01 -2.80885220e-01 1.31181204e+00
7.24421799e-01 -5.22724688e-01 7.36046255e-01 9.67743814e-01
-2.12981105e-01 -8.62327874e-01 -7.34257519e-01 -4.57861602e-01
3.82757396e-01 3.82508412e-02 9.37155306e-01 6.05036914e-01
1.70309693e-01 1.12563396e+00 -9.07135978e-02 -5.23150802e-01
2.89858907e-01 1.61229536e-01 1.10979342e+00 -8.16230893e-01
-4.33718383e-01 -3.85185182e-01 6.37210667e-01 -1.50566041e+00
2.08880186e-01 -8.03319097e-01 5.67621231e-01 -2.20025015e+00
-4.20631059e-02 -6.81257173e-02 5.70852220e-01 8.53121877e-02
-1.07195497e+00 -1.71337709e-01 1.50332749e-01 -1.60577312e-01
-7.84681201e-01 1.02808893e+00 1.88587761e+00 -5.38247153e-02
-1.04894385e-01 4.32082973e-02 -1.21958399e+00 2.71246701e-01
7.49037981e-01 -3.10830772e-01 -6.65874898e-01 -6.36418402e-01
5.90458632e-01 5.30489624e-01 2.48496935e-01 -6.63394570e-01
2.14227334e-01 -5.24571054e-02 -7.29161724e-02 -3.20089519e-01
1.02525935e-01 -1.23596080e-02 -7.38144934e-01 2.74488270e-01
-9.56531763e-01 2.20164046e-01 -2.36970350e-01 2.40454420e-01
-3.79120350e-01 -6.60168707e-01 3.93684030e-01 -4.05720353e-01
2.50214636e-02 -5.13318107e-02 -5.42267799e-01 1.09660733e+00
4.14475322e-01 1.70381844e-01 -7.73869872e-01 -1.23985696e+00
1.99657846e-02 7.70806491e-01 1.21702261e-01 5.31756878e-01
6.43797219e-01 -1.30530465e+00 -1.34802938e+00 -1.38034016e-01
2.85930872e-01 2.17904732e-01 2.02759191e-01 -1.31479487e-01
-4.87209707e-01 5.29546976e-01 1.54305294e-01 -2.67731529e-02
-6.93702519e-01 -4.85454574e-02 4.26526040e-01 -7.56056190e-01
-1.29420226e-02 1.04148328e+00 -1.34380668e-01 -1.20876884e+00
-7.39792064e-02 -4.09873843e-01 -4.19387668e-01 1.16091244e-01
7.71630764e-01 3.35349411e-01 -1.21039674e-02 1.82643887e-02
3.07043016e-01 -8.50113109e-02 -3.96484658e-02 -4.54675764e-01
8.25448096e-01 -7.61358961e-02 -2.01722249e-01 1.64140612e-01
9.64203596e-01 -1.19493969e-01 -9.53589559e-01 -2.22164586e-01
-1.60759035e-02 -2.55132318e-01 -6.93542421e-01 -1.08389783e+00
-4.71190721e-01 7.80693173e-01 -2.29913443e-01 6.31415367e-01
8.32069337e-01 1.85228303e-01 1.52347577e+00 9.72813427e-01
7.82471895e-02 -1.03440225e+00 9.93832409e-01 9.55061615e-01
1.35428131e+00 -1.25405407e+00 -6.66552365e-01 -5.94908409e-02
-6.75634444e-01 7.59656489e-01 1.08617973e+00 8.49755667e-03
-1.33337334e-01 -3.40729803e-01 4.45959121e-01 -1.25023127e-01
-1.25911999e+00 -1.39227837e-01 3.51692826e-01 6.88538134e-01
6.36508584e-01 -1.09575599e-01 -2.01719642e-01 7.60463119e-01
-9.15086567e-01 1.99738890e-02 7.16543078e-01 6.92060709e-01
-7.09004879e-01 -1.28529966e+00 -1.42297596e-01 9.33602989e-01
-1.73194230e-01 -4.40892607e-01 -7.84491420e-01 2.57106185e-01
-3.92689914e-01 1.73080194e+00 -1.01635262e-01 -1.04020536e-01
6.45121098e-01 3.80540133e-01 2.30699420e-01 -1.02920032e+00
-1.18399811e+00 -1.03810024e+00 6.38930798e-01 -3.97888958e-01
9.45600718e-02 -3.78638417e-01 -1.21415079e+00 -1.54388919e-01
-3.63541543e-01 8.12865794e-01 1.95808768e-01 1.23528206e+00
6.13480330e-01 5.36681950e-01 6.89793110e-01 -1.07673228e-01
-1.02566397e+00 -1.44518185e+00 -3.37984115e-02 6.69916928e-01
3.82227808e-01 1.75588682e-01 -4.12898868e-01 5.16166016e-02] | [11.545605659484863, 8.153206825256348] |
8410401a-a95d-4db4-a6a1-c4f2ff1cc29f | an-extensive-empirical-evaluation-of | null | null | https://aclanthology.org/E17-1048 | https://aclanthology.org/E17-1048.pdf | An Extensive Empirical Evaluation of Character-Based Morphological Tagging for 14 Languages | This paper investigates neural character-based morphological tagging for languages with complex morphology and large tag sets. Character-based approaches are attractive as they can handle rarely- and unseen words gracefully. We evaluate on 14 languages and observe consistent gains over a state-of-the-art morphological tagger across all languages except for English and French, where we match the state-of-the-art. We compare two architectures for computing character-based word vectors using recurrent (RNN) and convolutional (CNN) nets. We show that the CNN based approach performs slightly worse and less consistently than the RNN based approach. Small but systematic gains are observed when combining the two architectures by ensembling. | ['Guenter Neumann', 'Josef van Genabith', 'Georg Heigold'] | 2017-04-01 | null | null | null | eacl-2017-4 | ['morphological-tagging'] | ['natural-language-processing'] | [-8.90504420e-02 -2.51407772e-01 -8.42913147e-03 -3.20455879e-01
-9.73245621e-01 -1.04878068e+00 3.03826571e-01 5.07948875e-01
-1.17766309e+00 4.89050478e-01 4.46924537e-01 -6.62918866e-01
3.16530257e-01 -7.71372199e-01 -2.73267776e-01 -4.23153877e-01
-2.16526777e-01 4.72331166e-01 1.87296048e-01 -3.73836428e-01
-1.01013251e-01 5.95031857e-01 -9.86359417e-01 1.92934006e-01
6.06777608e-01 2.53040373e-01 3.72697338e-02 6.94829643e-01
-5.07057071e-01 5.14896154e-01 -5.24098754e-01 -9.39563513e-01
2.19906524e-01 6.50667399e-02 -7.92009473e-01 -4.69665319e-01
7.90307820e-01 1.46613911e-01 -3.23596299e-01 1.24540865e+00
6.53266013e-01 2.02734843e-01 2.70102620e-01 -2.41477042e-01
-1.16906583e+00 1.18920863e+00 -1.82105795e-01 3.86213273e-01
1.29909739e-01 -4.50514304e-03 1.38185215e+00 -9.07718956e-01
7.73532271e-01 1.28644705e+00 1.22466612e+00 8.78299356e-01
-1.03979886e+00 -1.77606240e-01 3.30350280e-01 -2.52855331e-01
-1.53370214e+00 -4.97605354e-01 2.71329641e-01 -2.56780744e-01
1.74242163e+00 -8.48197471e-03 4.55395639e-01 8.48483801e-01
-9.47217364e-03 7.54673302e-01 8.67475986e-01 -6.22257113e-01
3.76442894e-02 -1.16852567e-01 4.54353064e-01 1.06582463e+00
6.41542017e-01 9.73539650e-02 -3.10933352e-01 -3.64674628e-01
5.70929229e-01 -2.72461683e-01 2.22788364e-01 2.07954302e-01
-1.21812010e+00 8.13142896e-01 8.39621127e-02 6.06479704e-01
-3.67459595e-01 3.23783129e-01 6.27984762e-01 3.01826179e-01
6.56555593e-01 6.78341389e-01 -9.96726632e-01 -3.27433258e-01
-1.08581197e+00 -6.38864413e-02 1.05310214e+00 1.09051669e+00
4.07331467e-01 8.51029217e-01 -3.77450623e-02 1.17586219e+00
-8.85829628e-02 4.40342396e-01 7.97343254e-01 -3.48128706e-01
4.07776892e-01 3.36862475e-01 -2.76579887e-01 -3.02603632e-01
-6.12529993e-01 -6.14887118e-01 -5.79427004e-01 -1.07121788e-01
5.08263707e-01 -4.23967481e-01 -1.31338096e+00 1.86454129e+00
-1.60781950e-01 -1.72304481e-01 2.48840362e-01 3.32415223e-01
8.06540191e-01 5.80033660e-01 5.21771789e-01 -2.87764706e-02
1.62219179e+00 -7.75718808e-01 -5.06418526e-01 -5.85313439e-01
7.68731833e-01 -8.80002201e-01 9.13769960e-01 3.12736064e-01
-1.36772859e+00 -3.95641506e-01 -9.87013817e-01 -1.31006047e-01
-9.04298604e-01 4.21605378e-01 9.64487910e-01 1.04044318e+00
-1.48018110e+00 8.63428533e-01 -9.59390223e-01 -7.61627376e-01
6.19172789e-02 6.20643020e-01 -3.80035251e-01 3.77959341e-01
-1.00504243e+00 9.37417567e-01 5.83920181e-01 8.59674662e-02
-6.67953908e-01 -3.82786304e-01 -1.06432950e+00 -1.54118165e-01
-4.45813872e-02 -3.51278782e-01 1.39659119e+00 -6.27313137e-01
-1.38529205e+00 1.24063301e+00 -1.05305344e-01 -6.08037412e-01
-1.01613566e-01 -3.73262107e-01 -7.55699456e-01 -2.43552566e-01
-2.71767735e-01 7.36113131e-01 1.98510706e-01 -8.32795799e-01
-6.32599831e-01 1.25977416e-02 -3.23633879e-01 -6.17817566e-02
-5.89273870e-01 8.25708628e-01 -3.68455648e-01 -1.09953463e+00
3.43329161e-02 -8.61402154e-01 -4.63088125e-01 -4.66671079e-01
-3.65777344e-01 -3.11941385e-01 4.74253781e-02 -9.03472602e-01
1.27691019e+00 -1.92945075e+00 -1.54222444e-01 -1.12589218e-01
-4.81471084e-02 6.68286502e-01 -5.28327107e-01 3.27373207e-01
-3.31458226e-02 4.77467686e-01 -2.54409790e-01 -5.95100105e-01
4.31785099e-02 5.42035282e-01 -9.75862294e-02 5.93955398e-01
5.47527552e-01 1.08816516e+00 -8.76320302e-01 -2.35219747e-01
-2.23964512e-01 5.02587378e-01 -4.19376552e-01 -3.56975850e-03
-1.30768076e-01 -2.53213644e-01 7.45886639e-02 1.03752470e+00
3.97167295e-01 2.83807307e-01 5.01923025e-01 1.74213171e-01
-3.21081609e-01 7.68482745e-01 -9.01015520e-01 1.77482986e+00
-6.53381228e-01 6.41572773e-01 9.34049189e-02 -5.35010159e-01
1.13695765e+00 4.21286017e-01 -2.41430670e-01 -2.39198297e-01
1.77059904e-01 6.15042508e-01 2.70006478e-01 4.95843880e-04
9.90852058e-01 -2.94898540e-01 -4.00798976e-01 2.82037497e-01
5.48770666e-01 1.60982460e-01 1.58267900e-01 -1.40912563e-01
1.37044466e+00 9.54598933e-02 5.75923860e-01 -5.45994103e-01
2.94869989e-01 8.66504479e-03 8.39791179e-01 8.52325201e-01
-2.50512064e-02 3.69303346e-01 5.36828376e-02 -6.08979344e-01
-1.16232169e+00 -1.22428584e+00 -5.76097406e-02 1.65232205e+00
-5.33497989e-01 -3.75152886e-01 -7.35572517e-01 -7.39137113e-01
-2.77223866e-02 4.97688681e-01 -5.13584971e-01 2.06069216e-01
-1.21632683e+00 -9.29336667e-01 1.37160695e+00 1.05407572e+00
1.07413739e-01 -1.51254857e+00 -1.34216353e-01 6.28947914e-01
2.97855884e-01 -1.06669986e+00 -5.35217643e-01 8.10090363e-01
-1.05271637e+00 -3.82864743e-01 -5.60239553e-01 -1.35260236e+00
4.17545825e-01 -1.20677270e-01 1.43680644e+00 2.29986012e-01
-2.24350691e-01 6.96368665e-02 -5.64290345e-01 -2.54556060e-01
-6.33126438e-01 6.22816503e-01 3.15240681e-01 -4.50217992e-01
7.19646811e-01 -4.68725860e-01 -3.40771861e-02 -3.51374656e-01
-7.39676058e-01 -7.15397835e-01 8.76584232e-01 7.68182874e-01
4.50374544e-01 -3.54291826e-01 2.65585750e-01 -1.13065779e+00
6.30050778e-01 -2.21675783e-01 -5.94967544e-01 2.68150091e-01
-5.02152801e-01 2.60847628e-01 6.96847320e-01 -7.70315886e-01
-8.33753467e-01 2.51586199e-01 -6.38380468e-01 7.23286048e-02
-4.34359252e-01 5.33315361e-01 -2.53458768e-01 -5.55807129e-02
4.55365121e-01 -3.45046120e-03 -5.90156734e-01 -1.13568294e+00
5.40955305e-01 7.06389427e-01 1.02179229e+00 -6.38045609e-01
8.00356448e-01 2.58564770e-01 -4.54532206e-01 -8.05461228e-01
-6.11849308e-01 -6.90354884e-01 -9.77937818e-01 4.68770444e-01
9.87393439e-01 -8.65452826e-01 -3.83807957e-01 7.26542175e-01
-1.41295588e+00 -4.31540787e-01 -4.56364036e-01 3.64771098e-01
-1.83273375e-01 3.83484721e-01 -1.52301824e+00 -7.28307009e-01
-7.61809766e-01 -6.45493984e-01 9.10098374e-01 1.29953265e-01
-2.81561106e-01 -1.38586032e+00 2.62175918e-01 -4.72732872e-01
4.42228168e-01 -2.31870234e-01 1.15942991e+00 -1.44401264e+00
-4.76216339e-02 -3.37295592e-01 1.80837773e-02 2.18911603e-01
-1.81703102e-02 -1.40514135e-01 -1.00375772e+00 -3.74374449e-01
-4.19758052e-01 -1.49404019e-01 1.12749708e+00 3.97814512e-02
5.26406944e-01 -3.47111762e-01 5.18600643e-02 7.84898221e-01
1.62525845e+00 6.93350881e-02 5.92632771e-01 4.39906031e-01
6.74816430e-01 3.88922721e-01 1.98916361e-01 8.03297460e-02
2.78526455e-01 3.26854467e-01 -9.99851823e-02 -1.68920532e-01
-3.56068403e-01 -3.69852275e-01 6.81994855e-01 1.56812632e+00
8.10547248e-02 -5.95266342e-01 -1.05748296e+00 1.09082615e+00
-1.61517906e+00 -7.18585789e-01 -6.28565401e-02 2.01581478e+00
1.19403744e+00 3.12053353e-01 9.49158370e-02 -1.87370002e-01
7.78991401e-01 9.88128707e-02 -3.83212939e-02 -1.01602542e+00
-6.23255610e-01 9.33149815e-01 9.16882455e-01 5.59140623e-01
-1.28512430e+00 1.86577976e+00 7.59959555e+00 9.35602367e-01
-9.34552372e-01 4.53426659e-01 2.87278682e-01 1.79578260e-01
-3.80236328e-01 6.00744560e-02 -1.15989590e+00 -9.39808190e-02
1.30077696e+00 2.96752900e-01 3.19678098e-01 7.22512484e-01
-3.98537427e-01 2.39288464e-01 -7.32487559e-01 6.48910701e-01
1.86808333e-01 -1.15006423e+00 1.65894151e-01 1.78613439e-01
5.71895182e-01 6.50493026e-01 -1.51861995e-01 3.40189457e-01
1.09063029e+00 -8.96056414e-01 9.95772123e-01 3.27064991e-01
1.11305535e+00 -7.69920230e-01 9.55259502e-01 -1.54082000e-01
-1.56489754e+00 1.65175080e-01 -7.64947653e-01 -4.26668264e-02
2.52611071e-01 4.12200600e-01 -5.62366724e-01 1.73314333e-01
3.53445351e-01 4.86584604e-01 -9.04732883e-01 1.18015647e+00
-4.50471550e-01 1.09074640e+00 -2.02680573e-01 -4.78270501e-01
4.74521935e-01 9.41612944e-02 6.12809420e-01 2.18657398e+00
1.75299138e-01 -8.91925767e-02 2.86912411e-01 4.68767107e-01
-2.77281970e-01 4.21147943e-01 -5.13128936e-01 -2.86736846e-01
5.68673730e-01 1.44559884e+00 -1.06477737e+00 -5.53858101e-01
-4.39689040e-01 1.10882163e+00 8.58355582e-01 1.27287716e-01
-1.94303647e-01 -8.13505888e-01 9.76855338e-01 -3.11686248e-01
7.04527318e-01 -7.09486365e-01 -3.39532256e-01 -1.09681582e+00
-2.62225896e-01 -7.73767650e-01 6.30546212e-01 -2.95383483e-01
-1.64578426e+00 1.01913583e+00 -4.47930455e-01 -7.49404848e-01
-1.37136966e-01 -1.26877558e+00 -5.16263962e-01 7.68878460e-01
-1.28701568e+00 -1.27798450e+00 3.85823429e-01 1.57153592e-01
4.55641985e-01 -5.20248950e-01 1.36503768e+00 2.30943501e-01
-4.76036459e-01 1.03714550e+00 1.24896005e-01 8.83291602e-01
5.48808753e-01 -1.61600375e+00 1.41687787e+00 1.15821016e+00
7.54151046e-01 8.91529500e-01 2.01748893e-01 -9.75826621e-01
-1.42658186e+00 -1.25917172e+00 1.55358362e+00 -5.47831059e-01
9.88574624e-01 -6.47322536e-01 -9.16731894e-01 7.02055573e-01
4.72845733e-01 8.61065835e-02 8.20201993e-01 5.00359237e-01
-8.50800097e-01 3.36391747e-01 -9.22257245e-01 6.82552695e-01
1.38500893e+00 -8.45331430e-01 -8.39298964e-01 1.91185381e-02
8.93483698e-01 -9.62226838e-02 -6.94029391e-01 1.74621746e-01
4.89839822e-01 -3.69176567e-01 7.30907142e-01 -9.92791057e-01
-2.91530788e-01 -4.97983359e-02 -3.74265671e-01 -1.35166621e+00
-7.75484383e-01 -9.25485194e-01 2.42993563e-01 1.54769015e+00
9.40820336e-01 -5.36371052e-01 8.40929329e-01 -1.00114867e-01
-3.98471415e-01 -4.01767790e-01 -8.74969363e-01 -1.27242231e+00
7.50208676e-01 -6.96409941e-01 6.38620019e-01 1.00460517e+00
1.37991875e-01 2.51636297e-01 -1.38039544e-01 7.44316131e-02
8.71427134e-02 -3.05528313e-01 -9.46180597e-02 -1.10570168e+00
-3.29308778e-01 -7.87677288e-01 -8.46934378e-01 -8.51310670e-01
4.96031642e-01 -1.22388256e+00 4.38772887e-01 -1.36640000e+00
1.74639389e-01 -3.25092942e-01 -4.61013585e-01 1.10839665e+00
-2.33074620e-01 5.58996737e-01 1.40167683e-01 -1.26963735e-01
-5.74600697e-01 -2.64460407e-02 2.72036761e-01 -1.92774504e-01
-8.13247114e-02 -3.82155895e-01 -5.81485093e-01 8.34073782e-01
9.93516862e-01 -7.19685316e-01 4.37575489e-01 -8.26592386e-01
6.28624916e-01 -5.76206565e-01 -1.06972568e-01 -9.53224540e-01
3.05013895e-01 1.79943830e-01 2.85769641e-01 -5.31498551e-01
1.48729280e-01 -1.35557830e-01 -3.68197501e-01 4.34848011e-01
-2.20359296e-01 8.51395726e-01 5.99741697e-01 1.94618225e-01
8.59910920e-02 -6.26992404e-01 6.37204528e-01 -4.36354280e-01
-7.45361745e-01 2.40089849e-01 -9.94528830e-01 4.79457080e-01
1.73360363e-01 -1.43943131e-01 -2.98900962e-01 -1.20574702e-02
-7.11629331e-01 -4.93395209e-01 3.94607484e-01 4.42814320e-01
5.86486161e-01 -1.19025457e+00 -8.42884660e-01 1.97421357e-01
9.53075960e-02 -7.39600480e-01 -5.66292822e-01 2.52086192e-01
-7.08309829e-01 4.13002551e-01 2.42890902e-02 1.87352568e-01
-1.16678762e+00 4.86843675e-01 3.73083115e-01 -4.80206579e-01
-4.20503885e-01 1.32252538e+00 -4.57590044e-01 -8.70048821e-01
3.94809037e-01 -4.57283020e-01 -1.18230641e-01 1.38777792e-01
4.00895864e-01 2.83103198e-01 4.14101124e-01 -7.46839404e-01
-6.06248617e-01 4.71301347e-01 -2.05970407e-01 -3.52373064e-01
1.39019358e+00 2.97612786e-01 -3.08557063e-01 4.58693057e-01
9.48327243e-01 5.21820307e-01 -4.88041013e-01 -2.93442905e-01
6.17443860e-01 1.98633671e-01 -1.38763726e-01 -8.44903171e-01
-1.02723014e+00 9.65675712e-01 4.33915466e-01 -5.07735461e-02
6.48324788e-01 5.97917400e-02 1.16298962e+00 7.01725662e-01
4.08522904e-01 -1.19072437e+00 -4.10582244e-01 1.24618959e+00
7.83237666e-02 -6.68651998e-01 -1.96241856e-01 -2.95444459e-01
-4.56621975e-01 1.11392403e+00 2.57042468e-01 -2.08239749e-01
6.57308340e-01 7.60699034e-01 4.19642091e-01 -1.43809661e-01
-7.29535580e-01 -8.78921628e-01 2.47468308e-01 7.06854641e-01
8.55484784e-01 4.46554482e-01 -3.20259660e-01 8.23065221e-01
-5.20399749e-01 -7.68084884e-01 3.45142663e-01 1.12966299e+00
-5.79878032e-01 -1.56386220e+00 -2.88432781e-02 4.23134267e-01
-1.02021027e+00 -1.13408446e+00 -7.05988109e-01 7.74346113e-01
2.27288321e-01 8.40381384e-01 2.40548044e-01 -4.22377557e-01
1.65780380e-01 5.18820047e-01 5.13061821e-01 -1.12033236e+00
-1.48268569e+00 3.50359604e-02 7.11488545e-01 -1.69156358e-01
3.47277001e-02 -9.46710706e-01 -1.31846929e+00 -3.08790237e-01
-4.62701797e-01 3.51678543e-02 5.95846832e-01 7.09473729e-01
6.53770193e-02 9.45057571e-02 9.90290791e-02 -4.74577546e-01
-6.50400758e-01 -1.03298819e+00 -7.40179181e-01 1.78155273e-01
1.69471093e-02 -4.47578356e-02 -1.06704459e-01 -5.00227213e-02] | [10.305227279663086, 10.03116512298584] |
75dffd6e-f9ca-444b-a8a8-7a1eb4d1d146 | business-entity-matching-with-siamese-graph | 2105.03701 | null | https://arxiv.org/abs/2105.03701v1 | https://arxiv.org/pdf/2105.03701v1.pdf | Business Entity Matching with Siamese Graph Convolutional Networks | Data integration has been studied extensively for decades and approached from different angles. However, this domain still remains largely rule-driven and lacks universal automation. Recent developments in machine learning and in particular deep learning have opened the way to more general and efficient solutions to data-integration tasks. In this paper, we demonstrate an approach that allows modeling and integrating entities by leveraging their relations and contextual information. This is achieved by combining siamese and graph neural networks to effectively propagate information between connected entities and support high scalability. We evaluated our approach on the task of integrating data about business entities, demonstrating that it outperforms both traditional rule-based systems and other deep learning approaches. | ['Anton Zorin', 'Christoph Miksovic', 'Paolo Scotton', 'Katsiaryna Mirylenka', 'Mattia Atzeni', 'Evgeny Krivosheev'] | 2021-05-08 | null | null | null | null | ['data-integration'] | ['knowledge-base'] | [-4.56850469e-01 2.73323417e-01 -6.53449714e-01 -1.76949888e-01
-2.60225594e-01 -5.38981438e-01 8.54338348e-01 9.20687675e-01
-5.11335611e-01 7.72375584e-01 1.52238831e-01 -3.78952801e-01
-4.51258540e-01 -1.30234361e+00 -4.92448002e-01 1.30811498e-01
-4.45839226e-01 9.02896285e-01 2.79561847e-01 -5.82352281e-01
-2.18515933e-01 6.10343695e-01 -1.08267915e+00 1.93470895e-01
1.16410840e+00 1.00965202e+00 -5.28469801e-01 4.05163944e-01
-6.94757104e-01 1.37513149e+00 -5.61280549e-01 -9.99283195e-01
1.60738036e-01 8.51427987e-02 -1.11773837e+00 -6.01715863e-01
4.50594634e-01 7.25629330e-02 -4.27131951e-01 1.00731134e+00
1.62911922e-01 1.83744967e-01 5.86370230e-01 -1.62570882e+00
-1.09059274e+00 1.09648120e+00 -2.52974391e-01 2.18359008e-01
4.67585385e-01 -2.95562804e-01 1.49604774e+00 -6.55276418e-01
1.21825981e+00 1.15747643e+00 9.33173120e-01 1.27061397e-01
-1.22791219e+00 -6.02567315e-01 2.48574600e-01 4.45742935e-01
-1.18135405e+00 -1.82237267e-01 6.34793639e-01 -3.94942135e-01
1.61918104e+00 -1.18918680e-01 8.33081603e-01 6.71414137e-01
1.26658633e-01 7.32434750e-01 2.62536258e-01 -4.66779292e-01
1.01859741e-01 4.89862002e-02 2.92100608e-01 8.01735461e-01
8.04181099e-01 -3.77682224e-02 -1.55662924e-01 -9.66830254e-02
3.59547645e-01 1.20902985e-01 3.04303646e-01 -6.47894323e-01
-1.15521514e+00 9.59801674e-01 9.44522560e-01 7.47766376e-01
-5.61123371e-01 3.07347029e-01 5.14273643e-01 4.71419156e-01
3.05227816e-01 9.27112460e-01 -6.11084342e-01 1.95813164e-01
-8.63960028e-01 4.64958251e-01 1.56246495e+00 1.12172699e+00
7.66344845e-01 -6.91857710e-02 1.50707215e-01 5.09153426e-01
4.63984877e-01 1.60729080e-01 6.45926446e-02 -1.06082749e+00
7.44617164e-01 1.20306098e+00 6.88117892e-02 -1.35351062e+00
-7.01814353e-01 -2.88089275e-01 -8.72114122e-01 1.43625900e-01
3.54476631e-01 -4.71180588e-01 -6.97394490e-01 1.54173934e+00
4.55505520e-01 -1.39351547e-01 1.78915471e-01 4.33542997e-01
9.60320055e-01 3.26764226e-01 3.99424881e-01 2.70447761e-01
1.12785542e+00 -9.35004652e-01 -1.00271904e+00 -4.05883323e-03
8.80544066e-01 -3.62840772e-01 3.06849837e-01 2.56157219e-01
-1.17157781e+00 -4.06552345e-01 -1.24496412e+00 -3.66334468e-01
-1.35608304e+00 -4.13602680e-01 9.79276359e-01 4.26916897e-01
-1.06180620e+00 7.59453416e-01 -7.04161584e-01 -6.32449746e-01
8.74579549e-01 6.62322104e-01 -7.15574265e-01 2.06246629e-01
-1.69601440e+00 1.31829810e+00 7.74513185e-01 -9.92695987e-02
-2.15544626e-01 -7.34728217e-01 -1.15333533e+00 3.12817097e-01
8.36993814e-01 -1.19560635e+00 1.12417996e+00 -4.95387018e-01
-1.13097167e+00 4.74767536e-01 2.23782390e-01 -1.10813653e+00
3.51060212e-01 -2.86292791e-01 -8.68444085e-01 -3.47793221e-01
4.30382267e-02 4.57920521e-01 1.88283101e-01 -9.98388886e-01
-9.28624392e-01 -4.49572742e-01 2.29218096e-01 -3.11218929e-02
-4.65333253e-01 1.57015175e-01 -3.06070715e-01 -4.39100385e-01
-3.74876499e-01 -5.19644439e-01 -4.41611379e-01 -1.21288814e-01
-4.74688917e-01 -5.06960094e-01 6.44656301e-01 -5.71752906e-01
1.42245638e+00 -1.46202505e+00 2.34263837e-01 4.10219073e-01
7.02033460e-01 5.24373114e-01 2.56095175e-02 8.70709062e-01
5.59153557e-02 3.87687773e-01 -1.34508714e-01 -2.48206869e-01
4.55939800e-01 2.93536782e-01 4.07021269e-02 -4.97715957e-02
4.44260567e-01 1.55124593e+00 -9.38518107e-01 -6.74963534e-01
1.42745718e-01 5.14909446e-01 -4.84649777e-01 1.05784461e-02
-5.41861176e-01 -8.63111615e-02 -3.14321816e-01 7.99400091e-01
3.72421592e-01 -4.46580827e-01 7.27983117e-01 -2.65693098e-01
9.98318493e-02 2.54999429e-01 -1.33413255e+00 1.79411328e+00
-4.18974221e-01 6.08909428e-01 6.73049241e-02 -1.21060646e+00
1.00611436e+00 2.18475446e-01 7.68812120e-01 -8.02478433e-01
1.27441004e-01 1.35152727e-01 4.48642820e-02 -7.56873786e-01
7.23379910e-01 1.00565843e-01 -1.26684934e-01 3.96475255e-01
4.32611853e-01 -1.40292436e-01 6.14353478e-01 3.53239834e-01
1.02583349e+00 2.07256138e-01 7.32937932e-01 1.94513425e-01
4.44673538e-01 2.99411803e-01 4.74725932e-01 7.26083159e-01
-1.64356127e-01 -1.82904154e-01 3.78659397e-01 -6.09466732e-01
-8.29333186e-01 -8.70484769e-01 1.94291279e-01 1.13048458e+00
-8.00740272e-02 -6.19146347e-01 -4.28446949e-01 -1.03477025e+00
7.20093369e-01 7.21480250e-01 -6.32323146e-01 -8.48660469e-02
-7.66733587e-01 -3.78698677e-01 6.53506517e-01 9.24201190e-01
2.56251752e-01 -1.11330450e+00 3.93680343e-03 5.40304422e-01
6.50314689e-02 -1.27115047e+00 1.16273627e-01 1.32255062e-01
-8.86435151e-01 -1.25963020e+00 -7.91914910e-02 -7.71216393e-01
1.97122276e-01 -2.58748919e-01 1.63693738e+00 7.89892748e-02
-2.72668153e-01 2.10612878e-01 -7.70243481e-02 -5.55087090e-01
-5.66687346e-01 7.70452499e-01 -2.15011165e-01 -3.18749487e-01
7.65651047e-01 -5.00026286e-01 -3.12897414e-01 -1.62378117e-01
-8.62036467e-01 -6.29389048e-01 5.99445641e-01 4.96857077e-01
2.46253043e-01 -2.79659647e-02 1.11674750e+00 -1.34671056e+00
9.94426191e-01 -8.73598158e-01 -7.17766881e-01 6.20767236e-01
-1.13456786e+00 1.99642673e-01 6.26194358e-01 1.25962704e-01
-9.55462575e-01 -2.34474778e-01 -3.26651521e-02 -1.30815491e-01
-3.36865902e-01 1.06984401e+00 -2.78575391e-01 5.93627663e-03
7.06595242e-01 -5.35595179e-01 -1.15099356e-01 -3.72753978e-01
1.19845700e+00 4.43446636e-01 5.62588573e-01 -4.64197427e-01
5.87606430e-01 3.99412155e-01 6.30196035e-02 -3.71236384e-01
-5.67354739e-01 -2.85553336e-01 -9.12240982e-01 1.76489741e-01
9.15090144e-01 -7.13002026e-01 -8.32299531e-01 1.72688905e-03
-1.21564496e+00 3.46387662e-02 -6.87425613e-01 3.62756640e-01
-2.88720101e-01 2.32271105e-01 -6.85041964e-01 -3.97324443e-01
-5.52993417e-01 -6.69503033e-01 4.81049776e-01 2.80536294e-01
-3.81946385e-01 -1.74590898e+00 4.64250147e-01 3.69104266e-01
6.61300898e-01 2.98248649e-01 9.70920444e-01 -1.33540606e+00
-7.46098280e-01 -6.30380869e-01 -3.79579067e-01 5.70447221e-02
3.54329854e-01 2.44740829e-01 -5.45892954e-01 3.43009643e-02
-6.47236288e-01 -6.37427866e-02 7.89470255e-01 1.08495511e-01
5.18037975e-01 -3.12412888e-01 -8.04858923e-01 4.57006902e-01
1.42003536e+00 1.76206231e-01 2.96338677e-01 4.76311654e-01
9.28430021e-01 6.92047477e-01 3.38735551e-01 6.07844032e-02
1.06745923e+00 5.99288940e-01 4.99034911e-01 -2.52064884e-01
-7.67052025e-02 -2.92730182e-01 -3.89104694e-01 6.34187281e-01
-3.82977396e-01 -2.92168856e-01 -1.27534020e+00 7.38799214e-01
-2.32045197e+00 -1.02853000e+00 -1.50079310e-01 1.68221331e+00
7.85854042e-01 7.56082535e-02 3.18595141e-01 -1.82593971e-01
5.18768907e-01 -5.23581095e-02 -5.39150596e-01 -4.30888236e-01
-2.30076060e-01 3.35900009e-01 3.63124758e-01 4.51403856e-01
-1.42190099e+00 1.09880471e+00 7.20509863e+00 3.41196686e-01
-5.51190972e-01 -4.30278033e-02 2.49433778e-02 1.33047789e-01
-3.06486726e-01 1.54989576e-02 -8.27731848e-01 1.60655379e-02
8.71950746e-01 -5.42395115e-01 2.14301556e-01 6.33577645e-01
-3.06458861e-01 1.79643884e-01 -1.35322058e+00 4.50132728e-01
-1.21462367e-01 -1.86536288e+00 1.30651072e-01 8.85106325e-02
9.85572040e-01 1.82430759e-01 -2.76702523e-01 5.89096844e-01
1.23721421e+00 -1.11746264e+00 2.51701325e-01 9.05709088e-01
8.73326734e-02 -6.96470559e-01 9.61033165e-01 1.08411778e-02
-1.33291125e+00 -3.74845564e-01 -7.96485543e-02 8.00747201e-02
1.43979430e-01 5.60325921e-01 -1.02652729e+00 1.41431069e+00
6.28613532e-01 1.17143536e+00 -5.87186217e-01 9.94535744e-01
-1.82815745e-01 1.21529631e-01 -2.50652701e-01 -4.65058275e-02
1.83530509e-01 -3.36346358e-01 2.65938640e-01 1.51147377e+00
1.14522472e-01 -5.59380949e-01 2.16554448e-01 1.02487946e+00
-6.95836842e-01 1.71722814e-01 -1.13415825e+00 -2.25352958e-01
6.05185688e-01 1.37073076e+00 -4.24343526e-01 -5.64423442e-01
-8.14090073e-01 1.80369228e-01 8.93282890e-01 4.76061374e-01
-7.14563727e-01 -6.70898855e-01 5.16159058e-01 -7.36302957e-02
5.64605057e-01 -3.59379411e-01 -3.62998277e-01 -1.17205858e+00
-1.00854024e-01 -7.05016971e-01 1.11262739e+00 -3.46740276e-01
-1.53514707e+00 5.22219896e-01 -6.40649348e-02 -7.11068988e-01
-4.40935433e-01 -3.73571992e-01 -4.75479096e-01 6.80904865e-01
-1.76732981e+00 -1.34832644e+00 -1.72755644e-01 6.12631083e-01
-1.98005944e-01 -3.81174535e-01 7.96560109e-01 6.41154349e-01
-6.01921618e-01 5.73723614e-01 1.09329969e-02 6.16014719e-01
5.42302847e-01 -1.58105385e+00 7.46737003e-01 6.94676578e-01
6.16316855e-01 7.67529011e-01 2.45554790e-01 -7.06093192e-01
-1.28641772e+00 -1.09886563e+00 1.14516866e+00 -6.32149279e-01
1.20059621e+00 -1.85716048e-01 -9.43395019e-01 1.05051231e+00
7.77757943e-01 7.89657235e-02 7.69582450e-01 5.26340008e-01
-6.63368642e-01 -3.74464065e-01 -1.04166722e+00 5.18889308e-01
1.08802736e+00 -5.04399359e-01 -9.92098451e-01 1.60979420e-01
7.99424827e-01 6.03543408e-02 -1.63089907e+00 6.13989532e-01
3.12372029e-01 -6.78308845e-01 1.10062587e+00 -1.43879485e+00
1.85673252e-01 -2.85937399e-01 4.81128804e-02 -1.34517360e+00
-3.35912317e-01 -5.89386344e-01 -9.81973529e-01 1.24338043e+00
7.97993958e-01 -7.19521344e-01 7.11983562e-01 7.89081991e-01
2.10561186e-01 -6.37125194e-01 -5.79041481e-01 -7.59199560e-01
3.68491620e-01 -3.24871719e-01 1.03835261e+00 1.36434007e+00
4.27822948e-01 5.60344458e-01 7.06171319e-02 -8.69587734e-02
3.60293180e-01 4.87017691e-01 6.75408721e-01 -1.69017529e+00
2.02712044e-01 -9.76274073e-01 -5.84783912e-01 -5.65534472e-01
3.27399701e-01 -1.32441449e+00 -5.57490528e-01 -2.36330152e+00
-3.54131073e-01 -6.11128569e-01 -5.05891621e-01 6.44643903e-01
-9.45787206e-02 -1.88842446e-01 1.47083223e-01 -6.70133978e-02
-1.00447214e+00 2.78434455e-01 9.18916285e-01 -5.21944642e-01
-2.10302368e-01 -2.15753108e-01 -7.55223870e-01 4.68808800e-01
7.28320360e-01 -3.25327367e-01 -1.66582242e-01 -4.04953986e-01
8.81726801e-01 -2.21534252e-01 3.31935398e-02 -8.04436922e-01
8.41249168e-01 -1.65828429e-02 1.82241112e-01 -4.99799460e-01
-7.02179894e-02 -1.07627976e+00 2.11594626e-01 3.51494372e-01
-4.41758215e-01 1.16598986e-01 1.57715186e-01 6.59496188e-01
-6.15091145e-01 3.70208770e-02 2.26342246e-01 -1.14594527e-01
-7.17459321e-01 5.11070430e-01 -7.95316473e-02 3.35414141e-01
1.12696552e+00 7.09338039e-02 -6.42810285e-01 -2.27362067e-01
-1.16277230e+00 6.14072859e-01 1.91950262e-01 6.50211930e-01
3.04861724e-01 -1.35672116e+00 -7.48339593e-01 2.12109964e-02
1.08738616e-01 6.20196499e-02 -2.82756776e-01 7.91220665e-01
-5.54594696e-01 5.79237640e-01 -2.22617760e-01 -2.76508510e-01
-8.30272198e-01 1.08800697e+00 6.61145031e-01 -7.34961569e-01
-4.44435209e-01 3.66987646e-01 -6.40035987e-01 -8.17162216e-01
3.20408583e-01 -5.00124812e-01 -6.71484768e-01 5.08375585e-01
1.05131224e-01 7.00543284e-01 2.64342576e-01 -2.75392801e-01
-3.52701753e-01 2.66826212e-01 -2.76257724e-01 3.08123767e-01
1.45584548e+00 1.55973196e-01 -4.00417864e-01 2.05787256e-01
8.96031201e-01 -1.02385938e-01 -5.31706512e-01 -5.14417350e-01
9.30332184e-01 -5.28638586e-02 -2.73829162e-01 -8.98236871e-01
-1.30933058e+00 7.28247821e-01 -2.28769537e-02 1.06689036e+00
6.17406189e-01 1.16956234e-01 7.97500849e-01 1.00038838e+00
2.57833034e-01 -1.28252006e+00 -3.09831768e-01 6.20324671e-01
5.41605234e-01 -1.45596933e+00 9.03098956e-02 -2.34063029e-01
-3.97993147e-01 1.22086215e+00 6.58793986e-01 -1.25524908e-01
1.03845513e+00 5.54001808e-01 7.58922100e-02 -5.45968831e-01
-9.58604813e-01 -5.17053545e-01 3.91833067e-01 8.38900685e-01
4.38449770e-01 1.02298453e-01 2.88983621e-02 5.99494874e-01
3.79671156e-02 3.85614008e-01 1.68238431e-01 9.93954062e-01
-1.84233084e-01 -1.42325425e+00 1.05868049e-01 5.24844289e-01
-4.12071407e-01 -5.23616374e-02 -6.43785179e-01 1.22883117e+00
1.60774574e-01 9.05768156e-01 -7.12458370e-03 -2.84737706e-01
7.36797571e-01 2.66378522e-01 3.48762006e-01 -5.42185485e-01
-9.62362885e-01 -5.34216523e-01 4.37563866e-01 -5.48978448e-01
-6.32747471e-01 -4.57526416e-01 -1.37250745e+00 -4.06544387e-01
-4.03250396e-01 1.78340346e-01 5.01050770e-01 8.45860541e-01
8.66674185e-01 7.94751167e-01 1.13729618e-01 -1.27144501e-01
-3.21434200e-01 -7.29103267e-01 -3.83421600e-01 4.05946016e-01
5.72862029e-02 -5.64203203e-01 1.36478782e-01 -2.19320163e-01] | [9.190417289733887, 8.343289375305176] |
cf23de7e-b4a6-4587-a44b-179ec2cf385a | docstruct-a-multimodal-method-to-extract | 2010.11685 | null | https://arxiv.org/abs/2010.11685v1 | https://arxiv.org/pdf/2010.11685v1.pdf | DocStruct: A Multimodal Method to Extract Hierarchy Structure in Document for General Form Understanding | Form understanding depends on both textual contents and organizational structure. Although modern OCR performs well, it is still challenging to realize general form understanding because forms are commonly used and of various formats. The table detection and handcrafted features in previous works cannot apply to all forms because of their requirements on formats. Therefore, we concentrate on the most elementary components, the key-value pairs, and adopt multimodal methods to extract features. We consider the form structure as a tree-like or graph-like hierarchy of text fragments. The parent-child relation corresponds to the key-value pairs in forms. We utilize the state-of-the-art models and design targeted extraction modules to extract multimodal features from semantic contents, layout information, and visual images. A hybrid fusion method of concatenation and feature shifting is designed to fuse the heterogeneous features and provide an informative joint representation. We adopt an asymmetric algorithm and negative sampling in our model as well. We validate our method on two benchmarks, MedForm and FUNSD, and extensive experiments demonstrate the effectiveness of our method. | ['Ding Liang', 'Xuebo Liu', 'Mingjie Zhan', 'Zilong Wang'] | 2020-10-15 | null | https://aclanthology.org/2020.findings-emnlp.80 | https://aclanthology.org/2020.findings-emnlp.80.pdf | findings-of-the-association-for-computational | ['table-detection'] | ['miscellaneous'] | [ 3.58026892e-01 -3.86440575e-01 -2.59877086e-01 -4.09077555e-01
-5.74678838e-01 -9.65472460e-01 6.19253933e-01 4.10661370e-01
-6.00874387e-02 3.47327441e-01 3.49019080e-01 -6.91162422e-02
-2.48094052e-01 -8.49489868e-01 -6.09990954e-01 -2.33640224e-01
2.31326416e-01 1.35063827e-01 2.78299332e-01 -3.36185515e-01
4.99003798e-01 4.10501510e-01 -1.63462973e+00 8.14018667e-01
1.13418555e+00 1.06264889e+00 2.24726990e-01 3.24417621e-01
-7.26978660e-01 5.71137249e-01 -5.33832252e-01 -7.35158861e-01
1.29218549e-01 -3.03930968e-01 -7.30569065e-01 5.25265872e-01
5.43889701e-01 -4.46846873e-01 -4.47207540e-01 9.46155906e-01
2.59864569e-01 -1.87578201e-01 7.65825808e-01 -1.21501207e+00
-7.36077487e-01 6.11897469e-01 -8.53873909e-01 -3.82906079e-01
8.79753709e-01 -7.55750760e-02 1.23494101e+00 -1.06379485e+00
6.78030550e-01 1.40352285e+00 3.11035812e-01 6.95993528e-02
-1.09234202e+00 -3.75070900e-01 4.84573066e-01 3.39617461e-01
-1.37152350e+00 -2.56820709e-01 9.26440716e-01 -4.29498494e-01
8.15947413e-01 3.54321808e-01 8.06369543e-01 7.05560088e-01
-2.70377914e-03 1.32886410e+00 9.02410030e-01 -6.50623560e-01
-5.19141890e-02 1.94596380e-01 2.63405442e-01 9.35921133e-01
3.78527790e-01 -5.87627769e-01 -7.66064525e-01 -1.75558001e-01
6.52638257e-01 1.09569341e-01 -3.21753025e-01 -4.97672021e-01
-1.25270140e+00 5.49812615e-01 3.26973468e-01 1.94641650e-01
-4.68670875e-02 -1.71263024e-01 1.66641697e-01 1.12740770e-01
-7.37742856e-02 3.22304845e-01 -3.12731326e-01 -1.22131296e-01
-7.79111326e-01 3.40086013e-01 8.01216722e-01 1.35235572e+00
1.17870533e+00 -5.00688851e-01 -3.00817221e-01 9.89972830e-01
3.22499245e-01 4.68393475e-01 2.89158076e-01 -6.64798677e-01
7.89338648e-01 1.38648152e+00 -1.26840591e-01 -1.15919256e+00
-2.40290150e-01 -2.13510603e-01 -5.30889630e-01 -5.02205491e-01
2.94169307e-01 8.15619603e-02 -8.41565132e-01 1.33133054e+00
2.24374622e-01 -4.53372806e-01 -2.88278162e-01 8.18982959e-01
9.32560802e-01 4.58161563e-01 -1.93687946e-01 5.78819662e-02
1.59968281e+00 -7.96341240e-01 -9.45020556e-01 -2.78093487e-01
6.42618418e-01 -1.12440193e+00 1.21678650e+00 4.30286556e-01
-8.19260240e-01 -4.09352064e-01 -1.10881233e+00 -3.23500901e-01
-7.08614111e-01 6.37763619e-01 6.31244361e-01 6.73837543e-01
-5.28357506e-01 3.71381938e-01 -5.16487002e-01 -4.39589888e-01
2.34017715e-01 3.45076472e-01 -5.54491639e-01 -2.98988849e-01
-9.14608777e-01 2.95022815e-01 3.63392383e-01 2.30819166e-01
-2.80491441e-01 -3.37140918e-01 -1.07455540e+00 5.27440049e-02
7.39914417e-01 -7.67729282e-01 9.70434248e-01 -7.98842669e-01
-1.13366163e+00 5.59513569e-01 -2.88939804e-01 3.23428303e-01
1.53914109e-01 -4.46424365e-01 -3.91144097e-01 3.24800700e-01
-9.35622007e-02 4.19385523e-01 7.21820891e-01 -1.41345608e+00
-7.96438277e-01 -6.75886750e-01 2.85526335e-01 3.36549520e-01
-8.15378129e-01 1.57428272e-02 -1.08059919e+00 -6.38015389e-01
5.76720119e-01 -7.36776829e-01 2.29275212e-01 1.33287564e-01
-8.14947724e-01 -1.67527288e-01 8.11803758e-01 -6.44828498e-01
1.89797342e+00 -2.01384306e+00 9.51047987e-02 4.23453748e-01
2.98186868e-01 -1.72361463e-01 -1.47055477e-01 9.34067070e-01
2.12780759e-01 2.04054281e-01 -2.40838498e-01 -1.93702459e-01
3.25480133e-01 1.22698478e-01 -2.24387154e-01 1.66380972e-01
3.03641289e-01 6.61381543e-01 -6.92384243e-01 -8.25896919e-01
1.53450415e-01 2.00602159e-01 -5.83456814e-01 2.83853233e-01
-2.71885991e-01 -3.94425720e-01 -7.97722280e-01 1.05794871e+00
8.10059726e-01 -3.27146769e-01 5.21588027e-01 -6.60410404e-01
-2.90517621e-02 1.91534653e-01 -1.50022399e+00 1.77216589e+00
-3.86500508e-01 1.91606715e-01 -5.35645559e-02 -5.34470379e-01
9.03778493e-01 -2.64426351e-01 1.11592665e-01 -4.38511044e-01
-2.99629588e-02 1.72492996e-01 -5.28122857e-02 -7.44758010e-01
7.82373369e-01 2.90794343e-01 -3.05208564e-01 2.38777265e-01
-2.78791389e-03 -1.32109791e-01 5.43638647e-01 6.01172090e-01
9.24704790e-01 3.91471893e-01 5.30088842e-01 1.51433586e-03
5.48768938e-01 -2.11554065e-01 5.60816884e-01 4.85981345e-01
2.15107813e-01 7.37300515e-01 8.88539016e-01 -2.08571970e-01
-4.68698829e-01 -1.03288043e+00 8.06475952e-02 1.13107991e+00
4.96839732e-01 -1.25405943e+00 -6.97347105e-01 -8.17760766e-01
4.80302237e-02 3.92012894e-01 -5.70486903e-01 -4.16830033e-02
-5.67806900e-01 -5.19559085e-01 2.46202067e-01 6.57552600e-01
2.67404228e-01 -6.51416183e-01 -3.47138107e-01 2.95685064e-02
-1.91668347e-01 -1.14137518e+00 -6.26440525e-01 -6.91519529e-02
-5.43511748e-01 -1.16527748e+00 -5.05050778e-01 -7.36877322e-01
6.88775837e-01 3.79926562e-01 1.10650206e+00 2.54784971e-01
-1.50729030e-01 3.84307146e-01 -6.39023542e-01 -1.86644644e-01
8.97554755e-02 2.08125889e-01 -4.27244306e-01 4.27355736e-01
2.41287053e-01 -5.27895391e-01 -6.82227135e-01 3.58930290e-01
-1.14806676e+00 2.01850429e-01 1.02205825e+00 7.86333203e-01
5.95160306e-01 -1.24452643e-01 -1.15052253e-01 -8.69820654e-01
6.27518833e-01 -9.03797001e-02 -3.96739542e-01 8.33987534e-01
-3.79229069e-01 3.60645354e-01 6.07483387e-01 -1.94266081e-01
-1.18263447e+00 3.88761193e-01 3.14612627e-01 -2.05513611e-01
-7.96365812e-02 6.97223723e-01 -9.09323394e-01 1.44696087e-01
2.30672866e-01 3.89804482e-01 -2.88887739e-01 -7.33903825e-01
5.12039959e-01 8.23170364e-01 3.11281532e-01 -1.02585351e+00
8.13091040e-01 4.83274937e-01 -5.25478497e-02 -9.50521946e-01
-5.70468605e-01 -3.46734673e-01 -7.85305381e-01 -1.68487340e-01
5.29566109e-01 -7.29733765e-01 -5.83132267e-01 3.22772920e-01
-1.31336379e+00 3.26972097e-01 1.57133296e-01 1.62958562e-01
-2.90194541e-01 7.62470901e-01 -5.71830392e-01 -6.92055345e-01
-6.17521517e-02 -1.09486389e+00 1.49456704e+00 2.81366885e-01
-4.09651808e-02 -6.01704657e-01 -1.48602694e-01 2.69038349e-01
-6.35903254e-02 1.26609802e-01 1.34134746e+00 -4.59487259e-01
-8.62232566e-01 -2.35560149e-01 -4.69484985e-01 -8.48291740e-02
2.65818924e-01 4.14825231e-01 -7.30011284e-01 1.68152824e-01
-5.75581729e-01 -3.02349269e-01 1.04913414e+00 -1.64060876e-01
1.31908858e+00 -2.17050195e-01 -5.12729585e-01 4.45281059e-01
1.37310100e+00 2.61090267e-02 5.90426445e-01 2.32819498e-01
9.68761146e-01 9.84359264e-01 7.44968414e-01 7.88091600e-01
5.96392274e-01 7.50587881e-01 3.32328737e-01 1.73578203e-01
6.79592341e-02 -6.61700130e-01 3.43723089e-01 9.12711978e-01
7.73459747e-02 -2.28197366e-01 -8.81739199e-01 4.97197807e-01
-1.93646848e+00 -6.47008777e-01 -3.27053785e-01 2.06030607e+00
8.72195959e-01 9.59082972e-03 7.67671391e-02 1.75140411e-01
6.78050160e-01 5.54890633e-02 -2.80027390e-01 -3.61074246e-02
-2.40817010e-01 -2.34818101e-01 6.53549880e-02 7.56048113e-02
-1.06980014e+00 9.33830917e-01 5.48889494e+00 1.21494234e+00
-7.89121091e-01 -4.19090778e-01 3.99491847e-01 1.28427949e-02
-7.09451377e-01 2.77355254e-01 -8.52701724e-01 2.42307872e-01
6.26264066e-02 -3.92316170e-02 3.79900962e-01 7.15962708e-01
-9.73972976e-02 -1.00537851e-01 -1.22851479e+00 1.30835164e+00
1.63371578e-01 -1.06305146e+00 5.62685966e-01 9.07194987e-02
4.50175136e-01 -8.17688286e-01 2.67919269e-03 7.11411983e-02
-1.11012407e-01 -9.15560424e-01 1.04601717e+00 5.36337674e-01
7.89423168e-01 -7.06698656e-01 4.37454700e-01 1.27720594e-01
-1.63202345e+00 2.32184790e-02 -3.44608963e-01 1.86052158e-01
3.32975388e-02 5.97118735e-01 -3.88392597e-01 9.64104414e-01
5.30857205e-01 7.57814229e-01 -1.07986748e+00 9.40331936e-01
-3.08880061e-01 2.48370662e-01 -3.39147359e-01 -3.65367889e-01
-4.91305515e-02 -2.44265541e-01 1.43251091e-01 1.30537093e+00
3.62033695e-01 -9.40494686e-02 3.81869853e-01 8.40749800e-01
-1.04730748e-01 4.68854219e-01 -4.89476800e-01 -3.50369781e-01
6.75379336e-01 1.64634705e+00 -7.90543377e-01 -2.30853498e-01
-8.11138511e-01 7.31121004e-01 3.02605540e-01 2.93278843e-01
-5.81933200e-01 -8.00823450e-01 4.10463214e-01 2.96795547e-01
4.74261016e-01 -2.26988718e-01 -3.47504318e-01 -1.34602261e+00
4.90723342e-01 -1.05432057e+00 4.83302057e-01 -6.85470879e-01
-1.33595824e+00 6.09225452e-01 6.11037724e-02 -1.57488000e+00
-7.67491534e-02 -7.44355619e-01 -4.03920382e-01 4.60547656e-01
-1.24007058e+00 -1.55719113e+00 -4.39347208e-01 7.11962998e-01
4.09696788e-01 -4.45171027e-03 5.80722868e-01 9.47812945e-02
-7.76339531e-01 7.87798166e-01 -1.18655242e-01 3.50764066e-01
7.34146714e-01 -1.37300193e+00 4.25552800e-02 7.22021997e-01
3.02303433e-01 1.07837760e+00 4.00538683e-01 -6.41587257e-01
-2.05362415e+00 -6.93451345e-01 6.36209428e-01 -2.30410472e-01
6.95440590e-01 -6.70288086e-01 -8.04934084e-01 5.00980377e-01
3.48153323e-01 -1.78205162e-01 7.99249172e-01 1.81784689e-01
-7.58186936e-01 -2.92312235e-01 -7.25139976e-01 7.54358768e-01
1.35936987e+00 -5.18694162e-01 -6.69021368e-01 -6.90056710e-03
5.78219235e-01 -1.75879285e-01 -7.97417462e-01 3.20262194e-01
8.95365238e-01 -1.13847637e+00 5.78345478e-01 -4.07549083e-01
6.89289153e-01 -5.19185901e-01 -5.01043558e-01 -9.44737136e-01
-7.58338422e-02 -6.79010808e-01 -2.27493107e-01 1.66105497e+00
5.54547846e-01 -1.84841305e-01 7.44549394e-01 5.25222301e-01
2.36799079e-03 -9.17527556e-01 -4.70298916e-01 -6.26689851e-01
-3.83067757e-01 -4.08699781e-01 8.86127532e-01 7.10915446e-01
3.12448084e-01 5.92739701e-01 -1.73052147e-01 -9.24022570e-02
3.70395482e-01 7.21696973e-01 9.12558079e-01 -1.08430481e+00
-4.60608155e-01 -4.33923036e-01 -3.42957377e-01 -1.36822093e+00
-9.30571929e-02 -6.48189485e-01 -2.61714637e-01 -1.56982589e+00
5.58770478e-01 1.89162735e-02 -1.16655566e-01 4.56273228e-01
-2.61324018e-01 -2.15894341e-01 4.07329053e-01 1.52131960e-01
-7.92168379e-01 7.59225249e-01 1.27150297e+00 -3.73250246e-01
-1.96940318e-01 -3.92120421e-01 -9.50587273e-01 8.03009987e-01
3.72355610e-01 -6.77571222e-02 -4.21931893e-01 -5.76583087e-01
5.16321361e-01 -2.70831138e-02 9.52142477e-02 -6.96030676e-01
1.67611197e-01 -2.32318088e-01 6.95243120e-01 -8.30158889e-01
3.85337681e-01 -9.08098161e-01 -1.80126816e-01 1.01592027e-01
-2.42846608e-01 1.80092111e-01 -3.11758276e-02 5.24934947e-01
-4.17672008e-01 -2.36446962e-01 1.44044697e-01 -5.43389171e-02
-7.43618131e-01 1.90837026e-01 -9.27567407e-02 -1.16682231e-01
6.65556550e-01 -3.40136617e-01 -5.06730080e-01 -4.45424050e-01
-3.35217267e-01 1.14360303e-01 7.54532158e-01 5.66442311e-01
8.86111617e-01 -1.53802657e+00 -3.82509619e-01 2.71551847e-01
6.75415456e-01 -1.69538289e-01 1.37634918e-01 6.45372629e-01
-4.52912420e-01 2.19652861e-01 -1.19984150e-02 -5.27218163e-01
-1.38561153e+00 4.76847351e-01 -3.46699357e-02 -2.33487800e-01
-1.14962056e-01 4.25092876e-01 5.27458072e-01 -3.45458120e-01
1.43190414e-01 -4.21648055e-01 -3.66468906e-01 2.77020484e-01
6.22673631e-01 1.32677540e-01 8.43554288e-02 -4.92691457e-01
-3.23904276e-01 9.17615473e-01 -3.16100717e-01 -1.09132081e-01
1.19646311e+00 -1.12075709e-01 -3.86810243e-01 3.24006796e-01
1.15549481e+00 2.72640824e-01 -9.42749143e-01 -2.65376717e-01
2.30271459e-01 -6.56147480e-01 -2.26923108e-01 -6.86675608e-01
-8.25209200e-01 1.03196502e+00 2.50392228e-01 2.27674589e-01
1.36655021e+00 7.81321451e-02 8.09515595e-01 5.21786511e-01
3.97638500e-01 -1.20815670e+00 2.06388071e-01 3.87335211e-01
1.02850771e+00 -8.14186156e-01 2.63567954e-01 -1.21740496e+00
-4.64815825e-01 1.40930951e+00 7.59514093e-01 3.20622832e-01
4.72525984e-01 3.68727058e-01 -1.97189122e-01 -1.27449438e-01
-8.27975869e-01 -4.03791308e-01 6.48065388e-01 4.36699450e-01
7.12323546e-01 2.40351297e-02 -5.16077518e-01 1.09684050e+00
-9.94311348e-02 -2.66403437e-01 3.69414032e-01 1.20193422e+00
-4.26873118e-01 -1.37275839e+00 -5.40844798e-01 5.69636226e-01
-2.67209202e-01 -6.76269829e-02 -9.52029169e-01 9.63247836e-01
1.98481932e-01 9.17656064e-01 -9.60979387e-02 -7.47277379e-01
4.84665930e-01 1.74698327e-02 7.76143610e-01 -5.57505310e-01
-4.28139091e-01 5.02469778e-01 1.93900108e-01 -6.14607036e-01
-2.77556270e-01 -7.25449324e-01 -1.20481670e+00 -1.99765012e-01
-3.42918813e-01 -2.67343163e-01 4.94692087e-01 8.09447289e-01
3.81176025e-01 3.57594788e-01 6.20603025e-01 -5.34682155e-01
-5.34029126e-01 -8.23687494e-01 -6.77808344e-01 6.45605147e-01
-9.60868374e-02 -6.92759812e-01 -9.30510648e-03 1.59060672e-01] | [11.535667419433594, 2.4460361003875732] |
e42dd7f6-d2a1-4881-8fbf-b9e7ab520e19 | a-comparative-study-of-pre-trained-encoders | null | null | https://openreview.net/forum?id=tJPQtbiO6jv | https://openreview.net/pdf?id=tJPQtbiO6jv | A Comparative Study of Pre-trained Encoders for Low-Resource Named Entity Recognition | Pre-trained language models (PLM) are effective components of few-shot named entity recognition (NER) approaches when augmented with continued pre-training on task-specific out-of-domain data or fine-tuning on in-domain data. However, their performance in low-resource scenarios, where such data is not available, remains an open question. We introduce an encoder evaluation framework, and use it to systematically compare the performance of state-of-the-art pre-trained representations on the task of low-resource NER. We analyze a wide range of encoders pre-trained with different strategies, model architectures, intermediate-task fine-tuning, and contrastive learning. Our experimental results across ten benchmark NER datasets in English and German show that encoder performance varies significantly, suggesting that the choice of encoder for a specific low-resource scenario needs to be carefully evaluated. | ['Anonymous'] | 2021-11-16 | null | null | null | acl-arr-november-2021-11 | ['low-resource-named-entity-recognition'] | ['natural-language-processing'] | [ 7.45608360e-02 1.70370508e-02 -1.34229332e-01 -4.40484554e-01
-9.91578698e-01 -5.98664463e-01 8.53226483e-01 1.27932221e-01
-1.22566617e+00 8.64494383e-01 6.90788686e-01 -1.34816647e-01
3.73186618e-02 -7.49445379e-01 -4.14861739e-01 -8.30613598e-02
-3.42740417e-02 6.04831219e-01 2.92932183e-01 -4.57931936e-01
-7.96855241e-02 5.53485632e-01 -1.29704165e+00 1.65054098e-01
6.66757822e-01 5.31463206e-01 2.40080923e-01 7.73392677e-01
-5.61573386e-01 8.91445875e-01 -7.16537893e-01 -7.94026077e-01
5.47428876e-02 -3.42057735e-01 -1.05844462e+00 -4.03314233e-01
1.33628950e-01 1.60358809e-02 -4.01929915e-01 7.57690191e-01
1.06098795e+00 5.05243480e-01 5.90235591e-01 -4.86688167e-01
-1.02217638e+00 7.48544753e-01 2.85380810e-01 7.38654673e-01
1.76098160e-02 1.18612908e-01 9.92413104e-01 -9.78482127e-01
9.74642038e-01 8.39937389e-01 1.07110059e+00 9.58691001e-01
-1.05337703e+00 -4.69365448e-01 -2.52337962e-01 9.39742625e-02
-1.30315351e+00 -8.11409771e-01 2.42541060e-01 -1.90358192e-01
1.68845594e+00 -1.79146096e-01 -9.58486646e-02 1.50726128e+00
2.16379669e-02 4.36237067e-01 7.53404677e-01 -5.58464050e-01
3.20366085e-01 2.15078015e-02 2.28894353e-01 4.17556077e-01
4.34785813e-01 1.99391395e-01 -2.89407313e-01 -1.52972430e-01
4.27643865e-01 -3.11030298e-01 -2.83838540e-01 9.55028366e-03
-1.07975733e+00 8.91205847e-01 3.04564267e-01 9.52674448e-01
-6.71280444e-01 -1.08072802e-01 6.67276025e-01 2.78675973e-01
5.84798038e-01 9.99271750e-01 -1.01068974e+00 -5.08141816e-01
-1.26671720e+00 -2.27811664e-01 1.19239068e+00 1.04487717e+00
7.44069874e-01 2.31687889e-01 -5.48724890e-01 1.07088852e+00
-3.42379987e-01 -6.10017069e-02 7.99829543e-01 -5.64001977e-01
6.82240784e-01 2.59322941e-01 1.90975383e-01 -7.90744051e-02
-5.20801187e-01 -2.11325243e-01 -8.72182608e-01 -1.31095499e-01
2.62880743e-01 -8.02727580e-01 -1.23523486e+00 1.69881785e+00
9.88683291e-03 3.59335482e-01 5.78687310e-01 5.03733754e-01
1.08440256e+00 6.53393149e-01 7.24213660e-01 4.11111005e-02
1.48012614e+00 -1.04445398e+00 -7.80810058e-01 -5.33137739e-01
7.80251980e-01 -5.29175818e-01 8.76966000e-01 -3.39595616e-01
-8.61940384e-01 -7.35834599e-01 -8.24799418e-01 -3.75708073e-01
-1.03399384e+00 3.45171124e-01 2.93833107e-01 6.97685778e-01
-8.57658327e-01 9.06865299e-01 -7.53910899e-01 -8.17188144e-01
2.68825114e-01 1.12991422e-01 -5.36373556e-01 -2.53499061e-01
-1.76560831e+00 1.47734356e+00 8.79404306e-01 -1.34293988e-01
-9.25171912e-01 -1.07252574e+00 -1.16764319e+00 4.83731985e-01
2.16895506e-01 -5.67630827e-01 1.33614397e+00 -3.48062873e-01
-1.38386095e+00 9.59982514e-01 1.35331556e-01 -5.57205677e-01
3.11562866e-01 -3.91957313e-01 -9.03920054e-01 -1.49143547e-01
2.72071123e-01 6.51812494e-01 3.43091458e-01 -8.34098876e-01
-4.37031686e-01 2.31424138e-01 2.84594715e-01 4.26264182e-02
-4.43282127e-01 5.13042748e-01 -1.81876317e-01 -5.73463023e-01
-1.05792642e+00 -4.93410677e-01 -5.98841131e-01 -5.62262774e-01
-1.61929220e-01 -2.94829488e-01 3.73164654e-01 -7.07194149e-01
1.36914325e+00 -2.18479943e+00 -2.49779671e-01 -4.62864637e-01
-3.63318741e-01 7.39301682e-01 -5.57647705e-01 8.44656289e-01
-2.18439594e-01 5.12473702e-01 -2.46653050e-01 -4.68345076e-01
1.25684947e-01 2.06959635e-01 2.39478331e-03 -1.57734659e-02
7.40206838e-01 9.64627206e-01 -1.14832926e+00 -5.34226179e-01
1.91555426e-01 7.10201621e-01 -2.33837321e-01 6.64390445e-01
1.17375173e-01 1.46814823e-01 -2.86267787e-01 3.45598340e-01
2.23537356e-01 -2.80673265e-01 1.08523853e-03 -1.71994686e-01
-3.63805056e-01 6.18854284e-01 -1.06775832e+00 1.95089209e+00
-7.20074773e-01 5.10252297e-01 -1.94504514e-01 -6.86795413e-01
8.43802989e-01 6.51682556e-01 6.21009842e-02 -6.90903902e-01
1.18070692e-01 1.78340718e-01 -4.91523109e-02 -4.36624408e-01
7.05376208e-01 -4.12357897e-01 -2.98843205e-01 2.26623744e-01
1.09745407e+00 3.46599668e-01 8.17553163e-01 -1.87176734e-01
1.59656751e+00 6.53756736e-03 8.06201279e-01 -1.69841528e-01
2.33725294e-01 2.05062658e-01 5.83628774e-01 9.45053816e-01
-3.90494198e-01 7.30503559e-01 4.57745306e-02 -4.25956309e-01
-1.21547103e+00 -7.42069960e-01 -3.78513813e-01 1.45902109e+00
-2.96315700e-01 -4.92268860e-01 -7.43046939e-01 -9.80138779e-01
-5.60435593e-01 1.25536668e+00 -7.68007696e-01 -8.22766200e-02
-6.13876760e-01 -9.71738160e-01 1.03228128e+00 7.12234735e-01
4.42451566e-01 -1.55321205e+00 -9.27593291e-01 6.08569264e-01
7.15842173e-02 -1.36207902e+00 -3.23053330e-01 8.70113611e-01
-6.33208871e-01 -9.14973140e-01 -1.05649531e+00 -6.73400581e-01
2.22843170e-01 -5.71838431e-02 1.62852430e+00 -2.54108161e-01
-2.18211055e-01 5.15607238e-01 -7.09005177e-01 -2.94284105e-01
-4.56161916e-01 6.53281510e-01 -1.88504279e-01 -5.50606549e-01
7.03565359e-01 -3.32380861e-01 -2.46242315e-01 3.57689261e-02
-8.55119169e-01 -4.95864689e-01 8.82801712e-01 1.22138035e+00
2.58433878e-01 3.09121003e-03 5.37869930e-01 -1.12034404e+00
7.60419011e-01 -7.13839054e-01 -1.60445616e-01 6.41690671e-01
-1.93208888e-01 3.82736623e-01 7.42078304e-01 -4.72077608e-01
-1.52273703e+00 -1.42015100e-01 -4.87271637e-01 -3.80330384e-01
-5.97003102e-01 6.52062774e-01 -1.72980636e-01 3.59997958e-01
1.10814404e+00 -6.69770613e-02 -8.94950449e-01 -9.62652802e-01
5.83366454e-01 6.22765064e-01 5.16803801e-01 -5.11302233e-01
5.30015647e-01 2.99976896e-02 -5.96793473e-01 -9.29362953e-01
-1.08214974e+00 -6.44645035e-01 -9.68859732e-01 3.16303611e-01
1.12890220e+00 -1.09545898e+00 3.48246276e-01 -1.08037800e-01
-1.43625855e+00 -5.05647004e-01 -6.88680768e-01 5.03553092e-01
-2.15830445e-01 3.09414063e-02 -7.96608388e-01 -7.05371141e-01
-6.09763503e-01 -8.46379817e-01 9.87537086e-01 3.55144024e-01
-2.76066571e-01 -1.22422945e+00 6.14166260e-01 -1.47714898e-01
5.95214784e-01 -1.48139089e-01 6.94843233e-01 -1.51731718e+00
2.29466274e-01 -2.17898831e-01 -1.36094823e-01 3.06720078e-01
-5.17173149e-02 -4.15414631e-01 -1.20211387e+00 -1.85098663e-01
-3.46646816e-01 -5.68764091e-01 8.87426138e-01 -9.61770676e-03
7.21405566e-01 -2.17332896e-02 -2.45975405e-01 4.37877268e-01
1.54816997e+00 -7.64023662e-02 7.03970671e-01 5.08171439e-01
3.78707290e-01 4.48177218e-01 4.32406396e-01 3.58955324e-01
2.10574716e-01 3.41005474e-01 -9.82779488e-02 -4.58292253e-02
-3.81355643e-01 -2.78302789e-01 3.07357490e-01 8.17099214e-01
-3.04305911e-01 -4.50334698e-01 -1.11881638e+00 8.83246422e-01
-1.41741014e+00 -1.17247605e+00 3.59414101e-01 1.81010020e+00
9.84616816e-01 6.97293580e-02 -1.06287926e-01 -3.84021163e-01
9.78623033e-01 2.98059434e-01 -4.84344780e-01 -5.38962960e-01
-1.44071057e-02 7.89009333e-01 6.58700764e-01 5.52288955e-03
-1.34900653e+00 1.15380073e+00 6.85504770e+00 8.76667142e-01
-7.89771199e-01 7.40392268e-01 3.24217886e-01 1.94555223e-01
3.44269201e-02 1.49466619e-02 -1.14745128e+00 3.01431298e-01
1.78402078e+00 -2.05700368e-01 1.79820463e-01 1.05999959e+00
-3.20737869e-01 4.59032983e-01 -1.04713583e+00 6.18370414e-01
6.93970174e-02 -1.34063041e+00 -1.93180040e-01 -1.94788679e-01
7.59924114e-01 8.31660211e-01 -6.77908659e-01 1.23375762e+00
7.59407997e-01 -9.69174683e-01 2.63347447e-01 4.59325969e-01
1.06279469e+00 -5.95805883e-01 1.12060893e+00 3.42395216e-01
-1.21081197e+00 1.28334031e-01 -7.61600435e-01 2.65347540e-01
5.44408381e-01 3.87484938e-01 -6.71787739e-01 6.61537409e-01
7.84983158e-01 4.14184839e-01 -6.29463077e-01 1.08343625e+00
-2.91214675e-01 6.92702830e-01 -1.46085978e-01 -3.41530442e-02
4.16726410e-01 4.60339576e-01 2.32775122e-01 2.02660561e+00
3.52885306e-01 9.93343517e-02 -1.26149267e-01 6.95544124e-01
-4.91168082e-01 2.19253346e-01 -5.96497953e-01 -4.63483393e-01
5.56453764e-01 1.42659259e+00 -4.60072935e-01 -5.15578985e-01
-7.96839356e-01 1.13553405e+00 8.87456000e-01 2.62682021e-01
-6.88274026e-01 -7.83139288e-01 6.98644757e-01 -9.47955027e-02
6.77691936e-01 -1.31922126e-01 4.88600582e-02 -1.53060627e+00
-5.65068841e-01 -5.20178735e-01 7.88032353e-01 -6.34403944e-01
-1.79396069e+00 1.06362569e+00 -1.13841735e-01 -9.68943119e-01
-3.71827841e-01 -7.76003957e-01 -8.45351338e-01 7.69551456e-01
-1.84475636e+00 -1.04209244e+00 -5.59066646e-02 4.33510005e-01
7.09671378e-01 -2.72382408e-01 1.37048292e+00 5.24042547e-01
-5.59881687e-01 5.97080588e-01 1.74572274e-01 6.51083589e-01
8.01792204e-01 -1.23923469e+00 8.99186552e-01 9.50640976e-01
3.67671460e-01 5.97531140e-01 3.80032063e-01 -5.38090110e-01
-8.46498132e-01 -1.33237040e+00 1.45345247e+00 -6.63569331e-01
6.84332252e-01 -4.50903505e-01 -1.05115461e+00 6.87602758e-01
5.82448959e-01 3.14703882e-01 1.09648752e+00 4.82522517e-01
-3.91166270e-01 3.64267081e-01 -1.03376842e+00 4.36299413e-01
1.15974987e+00 -6.37827396e-01 -1.28897321e+00 -1.63363852e-02
7.42571533e-01 -2.10497633e-01 -1.23781407e+00 4.45962042e-01
1.29260495e-01 -6.89374626e-01 9.66850817e-01 -1.05488300e+00
3.95809591e-01 1.36826649e-01 -6.75849617e-02 -1.54768455e+00
-6.19132936e-01 -3.84307772e-01 -1.06708057e-01 1.71917343e+00
6.07429564e-01 -1.06353886e-01 2.66306728e-01 6.44065619e-01
-2.92390168e-01 -3.64803612e-01 -9.00901735e-01 -9.74130929e-01
2.57324040e-01 -3.65775287e-01 4.41701859e-01 1.21841705e+00
1.46112712e-02 8.06997895e-01 -4.06405509e-01 -4.81444411e-03
1.54949099e-01 -4.87990648e-01 4.04806703e-01 -1.18358910e+00
-8.11305493e-02 -2.24503115e-01 -2.63019353e-01 -4.61792171e-01
5.08260429e-01 -8.01721632e-01 2.09151864e-01 -1.75796747e+00
1.82205245e-01 -2.30533883e-01 -7.70919085e-01 8.26867759e-01
-3.29740196e-01 7.03146076e-03 2.39208087e-01 1.22280428e-02
-8.21090162e-01 6.19273067e-01 6.05382323e-01 6.01354465e-02
-1.39997572e-01 -5.39868057e-01 -6.15884781e-01 5.33657134e-01
6.41115010e-01 -8.54717195e-01 -8.76846388e-02 -5.13687551e-01
-3.65065485e-02 -1.78262025e-01 -8.71336982e-02 -1.18412662e+00
2.50035465e-01 -1.20376227e-02 4.71549720e-01 -1.64304301e-01
2.24381700e-01 -5.50129533e-01 -1.07472986e-01 2.07743704e-01
-5.10279894e-01 1.84807509e-01 2.70315766e-01 3.89121592e-01
-3.20190340e-01 -8.13705087e-01 8.97121727e-01 -4.72701430e-01
-1.40102363e+00 2.57386297e-01 -3.06777447e-01 8.57239544e-01
7.00010657e-01 7.39888102e-02 -1.86501667e-01 4.82445583e-03
-6.61323130e-01 -1.45258546e-01 1.38764456e-01 6.63664639e-01
1.64224640e-01 -1.30950272e+00 -8.83393943e-01 -1.95548967e-01
4.80592459e-01 -4.36270684e-01 3.49257380e-01 3.07551265e-01
-9.27221105e-02 4.35580134e-01 -2.78276682e-01 1.57591924e-01
-6.23513401e-01 6.22380018e-01 2.87314266e-01 -9.04648781e-01
-6.32700562e-01 6.78408682e-01 -1.17767230e-01 -7.44012237e-01
-8.45193863e-02 -5.44575509e-04 -5.03056824e-01 2.80742675e-01
7.51584947e-01 3.74008089e-01 2.77273327e-01 -6.89591885e-01
-3.75937730e-01 4.47525717e-02 -3.27295139e-02 -8.72087106e-02
1.58812344e+00 6.64718524e-02 4.18454677e-01 4.86489892e-01
1.24395800e+00 -3.20780545e-01 -9.55604613e-01 -2.86033034e-01
5.19756377e-01 1.00829862e-01 1.03949457e-02 -8.89791965e-01
-6.06631160e-01 1.06090498e+00 3.98676455e-01 -1.41848579e-01
8.87064815e-01 -1.09008722e-01 7.87848592e-01 7.38785505e-01
4.51347947e-01 -1.32008922e+00 -1.59789443e-01 1.10443258e+00
4.82213825e-01 -1.19098067e+00 -3.01858783e-01 1.15013145e-01
-8.93279970e-01 1.11705971e+00 6.63729429e-01 -2.63430387e-01
7.96228647e-01 3.79932135e-01 -1.73917692e-02 -2.84030735e-02
-7.69898295e-01 -9.25104022e-01 3.18861932e-01 5.99475384e-01
8.11350942e-01 -2.56675422e-01 -3.36972326e-01 1.08879924e+00
-4.03060652e-02 2.49574751e-01 3.76064330e-01 1.06254375e+00
-3.21378917e-01 -1.05226183e+00 5.89313842e-02 3.55691403e-01
-9.02916551e-01 -6.92018390e-01 -1.74954712e-01 9.55613375e-01
2.95447350e-01 8.49126399e-01 -1.40924882e-02 -6.27416298e-02
6.12395406e-01 7.04613686e-01 1.83377981e-01 -1.04655409e+00
-9.54871535e-01 -4.60011393e-01 7.94210374e-01 -3.47789049e-01
-5.34454346e-01 -3.93433899e-01 -9.83357191e-01 1.28614530e-01
-3.28916192e-01 3.87277931e-01 4.80685800e-01 1.07201064e+00
6.53213263e-01 5.84192753e-01 -1.56486575e-02 -9.18991923e-01
-5.75471163e-01 -1.24224305e+00 -3.77100825e-01 5.66879630e-01
-8.68695155e-02 -7.02565014e-01 -2.74428457e-01 -3.08268936e-03] | [9.656146049499512, 9.313446998596191] |
e4a89db1-01e7-42b6-939e-761e778850d0 | a-holistic-approach-to-cross-channel-image | null | null | http://openaccess.thecvf.com/content_cvpr_2016/html/Nam_A_Holistic_Approach_CVPR_2016_paper.html | http://openaccess.thecvf.com/content_cvpr_2016/papers/Nam_A_Holistic_Approach_CVPR_2016_paper.pdf | A Holistic Approach to Cross-Channel Image Noise Modeling and Its Application to Image Denoising | Modelling and analyzing noise in images is a fundamental task in many computer vision systems. Traditionally, noise has been modelled per color channel assuming that the color channels are independent. Although the color channels can be considered as mutually independent in camera RAW images, signals from different color channels get mixed during the imaging process inside the camera due to gamut mapping, tone-mapping, and compression. We show the influence of the in-camera imaging pipeline on noise and propose a new noise model in the 3D RGB space to accounts for the color channel mix-ups. A data-driven approach for determining the parameters of the new noise model is introduced as well as its application to image denoising. The experiments show that our noise model represents the noise in regular JPEG images more accurately compared to the previous models and is advantageous in image denoising. | ['Youngbae Hwang', 'Seonghyeon Nam', 'Seon Joo Kim', 'Yasuyuki Matsushita'] | 2016-06-01 | null | null | null | cvpr-2016-6 | ['tone-mapping'] | ['computer-vision'] | [ 3.14552546e-01 -6.49897099e-01 6.96803987e-01 -1.79089069e-01
-2.30247498e-01 -5.45041382e-01 4.37089026e-01 -1.83867007e-01
-7.26880729e-01 2.28048116e-01 -4.16464061e-02 -5.90729900e-02
1.37369648e-01 -7.67451048e-01 -5.76853991e-01 -9.63468552e-01
3.71019602e-01 4.96791229e-02 5.76814890e-01 -1.07510544e-01
-6.82165399e-02 4.56289172e-01 -1.64806068e+00 2.16819540e-01
5.50354540e-01 9.54199374e-01 3.03670436e-01 1.12085664e+00
-4.32166964e-01 9.72633064e-01 -7.92418599e-01 -3.90512913e-01
5.70025325e-01 -7.14530468e-01 -1.57355487e-01 4.79071200e-01
1.99409783e-01 -4.18609709e-01 -4.35996085e-01 1.63754106e+00
2.43678063e-01 -1.10916205e-01 3.56746435e-01 -1.09460831e+00
-1.90665260e-01 4.02069151e-01 -5.28229594e-01 -1.85196936e-01
-1.09417975e-01 -5.39845377e-02 3.53400886e-01 -4.31627125e-01
5.22837639e-01 1.12600005e+00 7.16700017e-01 2.90513635e-01
-1.38347089e+00 -3.86027694e-01 -1.87800288e-01 4.29010034e-01
-1.31619954e+00 -9.22186300e-02 9.28670466e-01 -2.03311324e-01
2.68966585e-01 3.93576741e-01 1.03931999e+00 8.62123966e-01
1.05765574e-01 4.38853890e-01 1.43131280e+00 -7.40759075e-01
4.68381941e-01 5.54953776e-02 3.89810316e-02 1.14516050e-01
3.79109561e-01 1.78641558e-01 -4.62025464e-01 1.09611429e-01
1.01337528e+00 -9.74218622e-02 -2.14951113e-01 -2.49006346e-01
-8.75742614e-01 6.29937649e-01 2.01645955e-01 2.76893049e-01
-4.23137337e-01 4.85716403e-01 2.06063047e-01 2.74169594e-01
3.10104221e-01 -7.76182041e-02 -2.96691895e-01 -2.56808072e-01
-1.05124426e+00 -2.70901732e-02 8.36525142e-01 8.72090876e-01
8.23058367e-01 1.55889928e-01 2.41285190e-01 9.41031694e-01
2.34747916e-01 8.25710773e-01 1.12205379e-01 -1.36254990e+00
-2.49448828e-02 2.32141972e-01 5.74718080e-02 -1.01218116e+00
-2.17079759e-01 -2.42644340e-01 -1.09966302e+00 7.38430083e-01
5.31066060e-01 1.65963545e-01 -1.10777223e+00 1.19141400e+00
1.17776478e-02 2.58062601e-01 -4.92594279e-02 6.93523288e-01
4.75685954e-01 6.19985461e-01 -1.69073641e-01 -2.74860322e-01
1.12504470e+00 -4.73942757e-01 -1.14804673e+00 -8.64147618e-02
-1.29794315e-01 -1.24598873e+00 6.55620337e-01 1.16444147e+00
-1.09561145e+00 -7.15346813e-01 -9.17819381e-01 -1.25161946e-01
-2.10109621e-01 8.33006669e-03 2.13547260e-01 1.13637280e+00
-1.21009433e+00 4.17857617e-01 -9.30426419e-01 -1.39997914e-01
1.67669222e-01 -1.09159835e-02 -2.39213392e-01 -4.91659909e-01
-8.52427483e-01 9.50602174e-01 5.91203980e-02 5.63564241e-01
-6.15934253e-01 -1.53828993e-01 -4.98166233e-01 -1.39516741e-01
4.40919667e-01 -4.16784853e-01 8.97171855e-01 -1.18951702e+00
-1.57608366e+00 5.50081730e-01 -1.61563709e-01 -2.35853612e-01
8.66940439e-01 -2.71688968e-01 -2.25589573e-01 2.82988459e-01
-3.78485709e-01 9.92772803e-02 1.25208580e+00 -1.65736592e+00
-4.41606730e-01 -1.68351442e-01 -1.65658116e-01 8.62118304e-02
-2.95747928e-02 2.84971409e-02 -1.34724331e+00 -7.19520628e-01
4.97119457e-01 -9.02212679e-01 -4.46783185e-01 2.10283026e-01
-4.13595736e-01 8.83084834e-01 8.33106339e-01 -8.17152023e-01
9.67970312e-01 -2.38931561e+00 1.52813211e-01 5.99317908e-01
7.94043243e-02 1.61815643e-01 -4.65431958e-02 1.58567339e-01
-6.99899197e-02 -6.48604557e-02 -3.83965701e-01 -5.80350697e-01
-3.71322483e-01 6.99218392e-01 2.03433156e-01 5.07772446e-01
-1.59009937e-02 2.38882735e-01 -5.48580766e-01 -1.94726780e-01
7.63890266e-01 9.13994193e-01 -2.93235302e-01 1.26324385e-01
7.55423233e-02 4.00303334e-01 1.89924259e-02 4.30630684e-01
1.35677516e+00 4.06984389e-01 7.41225109e-02 -6.72246516e-01
-9.81893539e-02 -4.10994112e-01 -1.77446222e+00 1.36965036e+00
-4.58256215e-01 6.73280537e-01 6.23544097e-01 -5.89990556e-01
8.05921793e-01 9.05728713e-02 3.32452714e-01 -6.84919000e-01
2.54954100e-01 2.54804075e-01 5.16564921e-02 -3.80542308e-01
6.44905329e-01 -1.79722980e-01 3.51833433e-01 6.51816055e-02
-2.32482459e-02 -7.30849504e-01 2.80573934e-01 8.11525285e-02
9.11860526e-01 7.11758286e-02 -2.47143153e-02 -2.15193689e-01
6.01826131e-01 -3.11718266e-02 5.06078064e-01 9.84157503e-01
-9.19177383e-02 1.23087978e+00 5.76877892e-01 -1.65935174e-01
-1.10924411e+00 -9.39668238e-01 1.52520128e-02 4.55345988e-01
2.57824361e-01 -3.40297967e-01 -1.10144389e+00 1.83194205e-02
-3.33549559e-01 4.76318121e-01 -5.31863332e-01 1.73029434e-02
-4.38421369e-01 -1.05027914e+00 3.83170038e-01 1.69381157e-01
6.28811479e-01 -5.49335778e-01 -6.20149255e-01 2.04565898e-01
-9.88560170e-02 -1.31593478e+00 -1.46624461e-01 6.49392068e-01
-9.22942638e-01 -1.41571712e+00 -5.15793562e-01 -1.65698886e-01
6.75993443e-01 5.05545676e-01 1.09574986e+00 1.59943625e-01
-6.25747979e-01 7.66897738e-01 -6.15850151e-01 -4.32364970e-01
-7.90558755e-01 -7.96554744e-01 -2.13361070e-01 3.33744347e-01
2.28899285e-01 -3.22186261e-01 -6.25821054e-01 1.44937217e-01
-1.49613965e+00 3.76643194e-03 4.09880430e-01 7.48559058e-01
6.18551850e-01 8.30085397e-01 -6.71529174e-01 -1.00912654e+00
3.68694663e-01 5.52352797e-03 -1.00652051e+00 1.13147860e-02
-2.49989152e-01 -9.09412950e-02 5.17688870e-01 -3.87513757e-01
-1.18562055e+00 5.03286242e-01 -1.42038867e-01 -4.22868013e-01
-2.87391990e-01 1.95662871e-01 -4.76883382e-01 -2.50848800e-01
4.17980373e-01 1.73214138e-01 3.33795846e-01 -6.87774658e-01
3.92992675e-01 2.94893742e-01 6.76571369e-01 -1.68207511e-01
1.04459536e+00 8.57924104e-01 3.43436927e-01 -1.38692963e+00
-1.12602487e-01 -4.93408561e-01 -7.09910452e-01 -4.87174630e-01
8.51355195e-01 -9.42208052e-01 -4.02715027e-01 1.18293595e+00
-1.18531048e+00 -2.45645180e-01 -2.70503551e-01 3.70840639e-01
-2.10628420e-01 5.96210539e-01 -8.19787264e-01 -9.32846427e-01
2.82709360e-01 -1.66184700e+00 7.16977596e-01 2.77166963e-01
3.32317412e-01 -1.02113438e+00 -1.90495923e-01 2.41956916e-02
7.23285437e-01 -1.65936664e-01 7.99999177e-01 1.26447350e-01
-7.52649486e-01 -2.86606044e-01 -1.45401329e-01 1.08863020e+00
9.96181071e-02 4.52869326e-01 -1.08275342e+00 1.58397108e-01
5.14717102e-01 3.44469696e-01 9.99305725e-01 7.15522647e-01
1.02869570e+00 3.46174926e-01 3.20253223e-01 7.55667508e-01
2.02498126e+00 1.64384767e-01 1.34624219e+00 5.37055850e-01
7.47800589e-01 5.46515226e-01 1.30345941e-01 3.64326060e-01
-1.30762488e-01 6.91262126e-01 7.20199406e-01 -5.02502441e-01
-4.08304960e-01 2.58101553e-01 3.44845295e-01 5.91889322e-01
-5.30151241e-02 -2.47693896e-01 -6.56757772e-01 1.00862458e-01
-1.37625051e+00 -7.57863939e-01 -8.90921652e-01 2.33369398e+00
5.78745842e-01 6.20373636e-02 -2.13966802e-01 4.63036418e-01
7.09287465e-01 -2.43607759e-01 -1.51875719e-01 -2.42561564e-01
-6.62661314e-01 3.89819086e-01 1.03131711e+00 7.37516999e-01
-1.07880461e+00 5.54654419e-01 6.93154573e+00 6.88818812e-01
-1.08969414e+00 -2.00787578e-02 6.73746109e-01 3.00408676e-02
-7.00944737e-02 1.95435304e-02 -3.51166099e-01 5.30051172e-01
4.70882922e-01 4.21052814e-01 6.83305621e-01 4.87436414e-01
5.01402915e-01 -8.19252312e-01 -4.58931327e-01 1.22866058e+00
6.64979145e-02 -6.30365312e-01 -9.59060937e-02 1.32540464e-02
6.53685749e-01 -4.77240719e-02 3.11926529e-02 -5.73661268e-01
3.03464741e-01 -6.65240943e-01 1.07255602e+00 7.89945602e-01
3.23481679e-01 -6.93400204e-01 1.01972175e+00 1.09383143e-01
-7.14159191e-01 -3.17143761e-02 -5.35450876e-01 -8.54216702e-03
1.97814107e-01 1.13382864e+00 -1.10350437e-01 4.87261474e-01
9.52686012e-01 4.76918131e-01 -8.33906710e-01 1.28432715e+00
-3.39007378e-01 6.77496493e-01 -4.98564005e-01 5.04224360e-01
-1.51899442e-01 -8.16786051e-01 3.61784458e-01 1.11610365e+00
5.57442009e-01 -1.24286860e-01 -3.27117920e-01 7.05881119e-01
3.09346408e-01 -2.09537327e-01 -1.37978449e-01 3.11860323e-01
1.21552527e-01 1.25414085e+00 -1.14191210e+00 -9.39310119e-02
-6.70014501e-01 1.42488551e+00 -7.04546928e-01 6.15138888e-01
-5.33399761e-01 -7.31739476e-02 6.62483096e-01 7.63840303e-02
5.22858024e-01 -2.73707420e-01 -5.09454310e-01 -9.31673169e-01
6.40820563e-02 -8.87853444e-01 -1.45860896e-01 -9.48609591e-01
-1.31456423e+00 6.58255219e-01 -1.31283030e-01 -1.44047046e+00
2.24169031e-01 -9.70995426e-01 -4.70192969e-01 8.88290405e-01
-1.46869779e+00 -1.12286687e+00 -5.08413494e-01 7.64463603e-01
8.16786885e-02 5.55478409e-02 6.94093943e-01 3.62265438e-01
-4.09476221e-01 3.39152152e-03 6.49150550e-01 3.65045416e-04
5.80477953e-01 -1.25475574e+00 2.08260641e-01 1.38114965e+00
1.00574881e-01 4.98023659e-01 1.01347482e+00 -5.49834669e-01
-1.37658405e+00 -7.44759262e-01 1.82682291e-01 -7.52962194e-03
4.54238892e-01 -3.79488051e-01 -9.68540490e-01 1.39692649e-01
2.64433891e-01 1.95357442e-01 4.39044595e-01 -4.02777761e-01
-3.78853351e-01 -3.72157127e-01 -1.02929115e+00 1.99234679e-01
3.01303178e-01 -4.38500553e-01 5.60182594e-02 -1.08160272e-01
1.83428116e-02 -3.56793225e-01 -3.74793500e-01 -2.11019814e-01
3.50833118e-01 -1.57971454e+00 1.09887683e+00 3.05058986e-01
1.66047424e-01 -8.05469036e-01 -3.08618963e-01 -1.27242780e+00
-7.37136900e-02 -3.71839315e-01 3.05463046e-01 1.13030338e+00
2.02164464e-02 -2.62168467e-01 6.02187574e-01 6.47572756e-01
2.26283386e-01 3.93365622e-01 -1.05335283e+00 -4.97270226e-01
-2.19279915e-01 -9.46174324e-01 1.29266113e-01 5.60650110e-01
-8.00774455e-01 -1.86111480e-01 -6.35423541e-01 2.29950279e-01
1.00567377e+00 -6.55133784e-01 6.62583709e-01 -1.30982268e+00
-4.27969515e-01 -4.00525838e-01 -5.04868090e-01 -6.96091652e-01
-4.98449475e-01 -1.35264173e-01 2.53501445e-01 -1.43468893e+00
1.64435357e-01 -3.77140999e-01 -6.69450834e-02 -1.49260044e-01
-1.72828108e-01 6.84919059e-01 5.22712231e-01 1.86391082e-02
-1.36762366e-01 9.53039750e-02 7.93235302e-01 -8.22503939e-02
1.65292323e-02 1.87715497e-02 -2.58840531e-01 8.64672840e-01
4.16513562e-01 -4.39749569e-01 -1.35030434e-01 -6.99288130e-01
4.77730751e-01 -2.94193268e-01 6.16842628e-01 -1.13510716e+00
3.87104273e-01 2.20395952e-01 4.90123987e-01 -5.01763880e-01
4.93402779e-01 -1.68749535e+00 6.71808243e-01 4.57910091e-01
5.40456995e-02 -1.03854090e-01 2.20720589e-01 7.03847706e-01
-5.40084958e-01 -4.83951300e-01 1.00611401e+00 -4.92451996e-01
-6.77785933e-01 -3.05257201e-01 -7.52194583e-01 -4.90762472e-01
6.37558639e-01 -4.21820849e-01 -2.82473564e-02 -6.97362483e-01
-7.95040071e-01 -4.40850556e-01 7.58994162e-01 -3.50004770e-02
6.33931339e-01 -9.00072396e-01 -5.10206759e-01 5.52548349e-01
-2.24836513e-01 -1.04811959e-01 5.61530769e-01 5.47937155e-01
-1.22495627e+00 -4.39887822e-01 -1.02604717e-01 -6.59352899e-01
-1.44301915e+00 2.42953286e-01 4.39344049e-01 4.17356044e-02
-4.61242765e-01 8.14282537e-01 -1.01630598e-01 2.40715649e-02
3.89332056e-01 -4.41921234e-01 2.87255086e-02 2.41240244e-02
6.99538887e-01 6.17399395e-01 3.03145707e-01 -7.84869075e-01
4.50060777e-02 7.26061702e-01 2.80671626e-01 -3.48863959e-01
1.40936077e+00 -5.39146245e-01 -7.30404615e-01 5.69163799e-01
1.07924294e+00 1.68599665e-01 -1.11304379e+00 -9.82634500e-02
-2.05912098e-01 -6.72908664e-01 5.37611187e-01 -8.33287418e-01
-1.42747390e+00 9.55681026e-01 9.99967098e-01 3.32370877e-01
1.72784090e+00 -6.04647815e-01 2.57454336e-01 -5.88150732e-02
2.93567181e-01 -1.39383674e+00 -1.41633347e-01 4.13570076e-01
4.51659650e-01 -1.01555693e+00 2.27443486e-01 -7.34549582e-01
-5.54611385e-01 1.52071404e+00 6.78337514e-02 -2.50058919e-02
8.35806966e-01 7.90826976e-01 6.71045780e-01 2.61624843e-01
-8.23750198e-02 -2.59151101e-01 -2.12628886e-01 8.50364566e-01
1.92378491e-01 -1.63336769e-01 -1.83867686e-03 3.77198368e-01
1.73584029e-01 4.17715199e-02 1.16685903e+00 7.76319027e-01
-1.96060687e-01 -1.50596464e+00 -1.18150711e+00 3.20999175e-02
-5.63166499e-01 -2.04900712e-01 -2.06269830e-01 6.76416814e-01
5.07703125e-01 1.17217946e+00 -4.64185849e-02 -2.74938196e-01
4.82461274e-01 -1.17283799e-01 4.92148906e-01 -6.51923344e-02
-6.22455776e-01 7.44133055e-01 -3.02942961e-01 -6.60178721e-01
-6.85966790e-01 -6.22573733e-01 -8.20275187e-01 -1.05687253e-01
-3.96233171e-01 -1.63465932e-01 1.19602823e+00 4.99826729e-01
-7.46434703e-02 7.10873008e-01 4.87312376e-01 -9.58464265e-01
-9.82939154e-02 -6.90226912e-01 -1.13832843e+00 6.10846937e-01
3.79226923e-01 -3.88630658e-01 -7.30864167e-01 5.34791768e-01] | [11.259976387023926, -2.4548513889312744] |
c5262aac-f6f6-4bd7-a04e-dce3e8a110ab | linear-algebra-with-transformers | null | null | https://openreview.net/forum?id=L2a_bcarHcF | https://openreview.net/pdf?id=L2a_bcarHcF | Linear algebra with transformers | Most applications of transformers to mathematics, from integration to theorem proving, focus on symbolic computation. In this paper, we show that transformers can be trained to perform numerical calculations with high accuracy. We consider problems of linear algebra: matrix transposition, addition, multiplication, eigenvalues and vectors, singular value decomposition, and inversion. Training small transformers (up to six layers) over datasets of random matrices, we achieve high accuracies (over 90%) on all problems. We also show that trained models can generalize out of their training distribution, and that out-of-domain accuracy can be greatly improved by working from more diverse datasets (in particular, by training from matrices with non-independent and identically distributed coefficients). Finally, we show that few-shot learning can be leveraged to retrain models to solve larger problems.
| ['Francois Charton'] | 2021-09-29 | null | null | null | null | ['automated-theorem-proving', 'automated-theorem-proving'] | ['miscellaneous', 'reasoning'] | [ 3.67436707e-02 -1.83376357e-01 -2.18292370e-01 -1.03604168e-01
-8.97654653e-01 -7.68843770e-01 5.81478000e-01 -1.50131018e-04
1.30864382e-02 7.82088399e-01 -1.22061931e-01 -8.01754653e-01
-2.00152725e-01 -1.03093314e+00 -1.01158345e+00 -3.06415111e-01
-4.44781750e-01 6.93223357e-01 -1.63233593e-01 -5.62193513e-01
2.75055826e-01 5.53648829e-01 -1.12607729e+00 4.02438462e-01
9.60571826e-01 9.01022077e-01 -5.76280415e-01 8.18129957e-01
-1.84062541e-01 1.53305519e+00 -5.28406203e-01 -4.27179009e-01
3.39973956e-01 -5.85757121e-02 -9.06069875e-01 -2.29047239e-01
5.16022146e-01 -5.31808853e-01 -2.34051257e-01 1.06593823e+00
1.53466120e-01 2.82719493e-01 9.44000244e-01 -1.61391473e+00
-7.31537521e-01 8.40264976e-01 -2.60086954e-01 -2.29105484e-02
4.42144334e-01 3.22968751e-01 1.37320352e+00 -1.16409731e+00
5.26911736e-01 1.20385492e+00 1.00042474e+00 1.45225093e-01
-1.73821747e+00 -1.04395735e+00 -3.69873613e-01 4.20476496e-01
-1.41529548e+00 -5.82324684e-01 5.35664618e-01 -6.03600562e-01
1.44597518e+00 1.97117850e-01 7.86951482e-01 6.73060179e-01
1.72766075e-01 5.83705783e-01 6.34942412e-01 -1.67360216e-01
3.39398235e-01 -2.42831692e-01 -1.56250205e-02 8.80773246e-01
1.45954907e-01 -1.29911095e-01 -3.75747830e-01 -2.50307173e-01
6.80818081e-01 -7.71005377e-02 1.27904862e-01 -8.89889121e-01
-1.47957933e+00 9.76845741e-01 1.94010645e-01 2.27274269e-01
6.69552339e-03 5.81195414e-01 6.55567527e-01 7.11597979e-01
1.07773341e-01 1.05953550e+00 -5.67033112e-01 -3.95880967e-01
-8.39258790e-01 4.95390743e-01 1.17531335e+00 1.18120039e+00
7.45143712e-01 5.95125914e-01 -3.17940712e-02 5.36840379e-01
-4.20542449e-01 7.76258528e-01 1.36421844e-01 -1.22460532e+00
6.85679734e-01 3.75759691e-01 5.57083376e-02 -9.58110094e-01
-3.27708125e-01 -3.72692049e-01 -9.95198488e-01 1.52490288e-01
5.01317441e-01 -2.40012497e-01 -8.22943985e-01 1.61652279e+00
4.70724776e-02 3.51939142e-01 1.11206971e-01 2.49966502e-01
3.53695929e-01 8.90787542e-01 -2.81571120e-01 3.70484218e-02
8.04226875e-01 -5.55077910e-01 -3.56670618e-01 -1.21908888e-01
1.26705372e+00 -5.86076796e-01 9.23100829e-01 8.26875448e-01
-1.22662747e+00 -2.69230157e-01 -1.05594873e+00 -4.64603603e-01
-3.10445338e-01 -1.68799847e-01 1.21374798e+00 4.56313491e-01
-9.63490903e-01 1.00261927e+00 -8.97386849e-01 2.26310313e-01
6.39562845e-01 4.96869236e-01 -3.01701158e-01 -1.34444520e-01
-1.14509141e+00 9.82427478e-01 1.40804365e-01 -1.20892920e-01
-1.13148832e+00 -1.57863975e+00 -9.89128113e-01 4.91230071e-01
3.50631893e-01 -6.07415199e-01 1.44262719e+00 -5.23111284e-01
-1.30534387e+00 4.54428643e-01 2.53137350e-02 -7.21816123e-01
2.02070937e-01 4.40963870e-03 -1.08081505e-01 -1.17785268e-01
1.01238415e-01 6.03134707e-02 8.25516105e-01 -5.11629939e-01
-7.20890239e-02 -1.76352501e-01 1.68719158e-01 -4.48830009e-01
-3.22210878e-01 -2.45577648e-01 3.30038846e-01 -4.38059360e-01
7.12711290e-02 -7.22249687e-01 -3.07737261e-01 -1.05691120e-01
-4.13841903e-01 -3.52049530e-01 1.55026838e-01 -6.47228539e-01
1.01944137e+00 -1.94593513e+00 4.45883691e-01 4.94994849e-01
5.07919431e-01 1.51438072e-01 -3.03462334e-02 5.87826133e-01
-4.21991676e-01 6.62347972e-02 -7.96192288e-02 1.02837205e-01
4.64668006e-01 7.89504722e-02 -6.26507044e-01 5.07249713e-01
5.11681080e-01 1.14125717e+00 -9.70898151e-01 -4.01643336e-01
1.94423556e-01 6.29504994e-02 -9.75990415e-01 -2.42191136e-01
-4.94150311e-01 -1.23810926e-02 -1.19229257e-01 7.24794269e-01
5.43965042e-01 -4.91859078e-01 5.22374809e-01 -2.16067567e-01
4.11343873e-01 3.62718523e-01 -1.41743827e+00 1.63786983e+00
-8.40733707e-01 6.45627916e-01 -2.13886872e-02 -1.53562272e+00
5.55312455e-01 1.19663879e-01 3.38300437e-01 -5.39344668e-01
3.31431419e-01 1.31460577e-01 3.45283777e-01 -3.15149665e-01
4.16044742e-01 -3.43803257e-01 -3.05171490e-01 6.58962786e-01
5.06144822e-01 -7.94238091e-01 5.09850919e-01 5.38224220e-01
1.11359525e+00 -1.92233175e-01 2.83334225e-01 -2.40046438e-02
2.18229473e-01 1.11667261e-01 4.45712030e-01 6.45661116e-01
2.72081584e-01 3.08011342e-02 1.04313922e+00 -4.71444607e-01
-1.49988616e+00 -1.34321189e+00 7.97562748e-02 1.39922547e+00
-7.36028492e-01 -7.56579757e-01 -3.61977845e-01 2.48239040e-02
3.91212285e-01 9.95315373e-01 -3.59419972e-01 -4.82763857e-01
-5.18525481e-01 -3.96565318e-01 7.07370043e-01 8.63296211e-01
1.74169570e-01 -3.45608354e-01 6.21278696e-02 2.51423389e-01
2.41406724e-01 -1.11846364e+00 1.28709394e-02 3.67909312e-01
-1.10715377e+00 -1.04734540e+00 -4.21344578e-01 -6.81309462e-01
3.83932114e-01 -2.28072703e-01 1.28691113e+00 -1.52125899e-02
-4.09405023e-01 3.08346361e-01 3.21921319e-01 -2.33980298e-01
-6.82792842e-01 4.56575640e-02 2.11619958e-01 -4.55602407e-01
4.57961261e-02 -1.13428533e+00 2.43636847e-01 -2.12463036e-01
-4.87488449e-01 -1.49555132e-01 3.14391732e-01 1.02400446e+00
-6.37004301e-02 3.01224757e-02 3.14463288e-01 -8.74580085e-01
6.50993168e-01 -3.82869363e-01 -9.56674457e-01 3.83835107e-01
-4.31133628e-01 4.26089317e-01 1.03011644e+00 -6.57782674e-01
-5.09777904e-01 -1.27643093e-01 2.33940095e-01 -7.74456561e-01
5.46452403e-01 5.33300042e-01 1.30640581e-01 -2.65645891e-01
1.03895330e+00 2.61850089e-01 2.15769142e-01 -7.57898092e-02
9.15439427e-01 7.74199218e-02 5.32951474e-01 -1.15266037e+00
9.84070957e-01 5.73380850e-02 6.58650279e-01 -5.08721709e-01
-6.90961301e-01 -2.62436271e-03 -5.40755689e-01 3.74120474e-01
1.62426233e-01 -1.07166636e+00 -1.02053344e+00 3.65776211e-01
-9.44131136e-01 -6.05719209e-01 -3.07707101e-01 5.10694802e-01
-5.74341655e-01 1.94146335e-01 -8.45362961e-01 -7.81621635e-01
-2.11599857e-01 -8.43532264e-01 6.42683148e-01 -3.33016336e-01
-3.50806773e-01 -1.14657652e+00 -1.34165466e-01 6.11778460e-02
4.12749231e-01 4.41456176e-02 1.48749650e+00 -8.58498573e-01
-9.13659751e-01 -2.38743693e-01 -1.24667518e-01 5.48221886e-01
-1.77648753e-01 1.77435070e-01 -6.12955809e-01 -9.71943140e-02
-2.13198870e-01 -7.45190382e-01 7.69531727e-01 -3.35643487e-03
1.21976566e+00 -5.27864099e-01 -2.40417153e-01 7.46959805e-01
9.12035704e-01 -2.78676391e-01 6.11427009e-01 -5.36601357e-02
6.45080447e-01 -8.03350806e-02 3.32012981e-01 6.28009260e-01
7.87197798e-02 1.30493224e-01 1.12470396e-01 4.17974323e-01
3.32288474e-01 -1.66596249e-01 2.53435284e-01 6.60625100e-01
-1.14109665e-01 5.47214389e-01 -1.12378597e+00 6.49548471e-01
-1.51077759e+00 -1.20162833e+00 6.07206933e-02 2.08584070e+00
1.23466384e+00 2.64215529e-01 5.91131262e-02 4.39439178e-01
1.67399853e-01 -9.93100926e-02 -6.49350286e-01 -6.41005278e-01
2.06033885e-01 1.10829318e+00 4.64370310e-01 5.91533959e-01
-9.01197851e-01 9.39926445e-01 7.45646477e+00 1.05562222e+00
-9.87829804e-01 -6.88025579e-02 3.21528047e-01 -2.73340255e-01
-4.38396037e-01 1.25277445e-01 -4.73730356e-01 6.99074380e-03
1.09521616e+00 -5.89492381e-01 9.73196983e-01 1.12111640e+00
-5.17925382e-01 3.02133411e-01 -1.64508653e+00 1.08574569e+00
-5.57627343e-02 -1.59194946e+00 1.31707937e-02 -1.60494015e-01
1.12862384e+00 -2.56444931e-01 3.04950953e-01 9.83456194e-01
9.14497018e-01 -1.51947284e+00 2.16042757e-01 4.48747396e-01
8.75866234e-01 -1.00766563e+00 3.12655926e-01 2.39466384e-01
-1.07106709e+00 -2.96283871e-01 -5.19877672e-01 -5.74664295e-01
-4.12965029e-01 6.94017708e-01 -1.37101340e+00 2.78941005e-01
9.11712497e-02 9.71462727e-01 -4.38757360e-01 7.04223812e-01
-1.62509769e-01 4.73413825e-01 -4.32592452e-01 -3.64879370e-01
8.45955778e-03 -3.48725379e-01 3.87059599e-02 8.25406730e-01
3.88310224e-01 1.34954125e-01 4.56550941e-02 1.31329310e+00
-2.37886354e-01 -9.87182558e-02 -8.48767936e-01 -5.07826984e-01
4.61067855e-01 1.03747642e+00 -9.67247412e-02 -7.52959371e-01
-3.11048418e-01 4.84566092e-01 5.99193394e-01 4.30346340e-01
-8.89303744e-01 -7.36655235e-01 8.74417186e-01 -2.26804875e-02
6.84581637e-01 -5.98790050e-01 -5.60646594e-01 -1.29934049e+00
-2.35506713e-01 -1.21266413e+00 1.66853979e-01 -8.84785712e-01
-1.24037313e+00 -2.85541534e-01 1.23126306e-01 -1.00147653e+00
-8.61169696e-01 -8.69874060e-01 -5.66775441e-01 7.01791942e-01
-7.55055130e-01 -9.28093612e-01 2.23208874e-01 5.69469035e-01
-1.34310439e-01 -4.09765840e-01 8.52873623e-01 1.67903781e-01
-1.83457032e-01 7.38559067e-01 4.04828340e-01 6.92529380e-01
2.83760995e-01 -1.22449219e+00 5.51778913e-01 6.55196905e-01
2.58393675e-01 1.02331948e+00 7.08719671e-01 -2.60077059e-01
-2.20127726e+00 -8.92000735e-01 8.18197131e-01 -4.62731063e-01
1.44708157e+00 -5.16420305e-01 -8.33664894e-01 1.22986329e+00
-2.62712061e-01 2.47049585e-01 4.95901048e-01 7.77356148e-01
-9.52321053e-01 -2.56732136e-01 -1.05021024e+00 6.83151066e-01
1.01492691e+00 -1.01710737e+00 -7.28467226e-01 7.48474002e-01
7.50002325e-01 -7.30068564e-01 -1.22389841e+00 4.87162359e-02
4.57918584e-01 -7.00525224e-01 1.53542900e+00 -1.37835455e+00
7.93965578e-01 1.72534779e-01 -2.76254237e-01 -1.29398787e+00
-2.60259897e-01 -6.10221505e-01 -4.57256138e-01 9.06821370e-01
4.09057796e-01 -5.81551969e-01 6.48878813e-01 6.62978411e-01
1.81359611e-02 -5.88157237e-01 -8.90199542e-01 -9.66946661e-01
6.96973860e-01 -9.36854661e-01 8.95735085e-01 1.14632738e+00
5.41599095e-01 7.11526155e-01 -3.11610371e-01 -1.26787469e-01
4.72183645e-01 4.92599428e-01 1.17786002e+00 -1.18773258e+00
-3.86806756e-01 -4.53814358e-01 -6.88222766e-01 -8.41810822e-01
5.62019050e-01 -1.46187866e+00 -4.22188073e-01 -1.23198378e+00
8.01522061e-02 -3.85085493e-01 5.78258373e-02 6.13144040e-01
2.72670209e-01 -2.16092989e-02 1.79019466e-01 -3.48154545e-01
-5.61582267e-01 6.27901375e-01 1.14895356e+00 -4.70317364e-01
1.58281565e-01 -1.84583694e-01 -5.69214523e-01 6.05857015e-01
6.15339160e-01 -1.81375116e-01 -3.09458345e-01 -1.68561861e-01
7.45556593e-01 2.78747737e-01 5.25654376e-01 -1.27169335e+00
3.17594856e-01 -3.47881913e-01 5.00017524e-01 -4.25615966e-01
6.12692773e-01 -4.10444498e-01 -1.46521643e-01 5.83603084e-01
-3.07227790e-01 6.84587210e-02 3.23468268e-01 2.09864527e-01
1.99598029e-01 -2.84211963e-01 4.70876932e-01 -2.88916171e-01
-5.29687703e-01 1.02055460e-01 -3.58334392e-01 6.91755533e-01
6.59130096e-01 3.36460531e-01 -1.31365031e-01 -5.67671001e-01
-7.64293551e-01 2.21601829e-01 2.47360855e-01 -1.68965489e-01
5.39088666e-01 -1.60225582e+00 -5.36773980e-01 1.66766867e-01
-1.58708408e-01 5.01820035e-02 -2.50435490e-02 1.01397324e+00
-4.23933268e-01 5.46492219e-01 -1.44318357e-01 -6.18766546e-01
-9.98636365e-01 8.33304822e-01 3.47696275e-01 -4.09966350e-01
-2.15736896e-01 7.56932139e-01 -3.28465194e-01 -8.19878340e-01
2.42405355e-01 -1.12329113e+00 6.01741433e-01 -4.92063835e-02
5.29061973e-01 5.69919050e-01 9.81395543e-02 1.75435141e-01
-2.69611061e-01 4.31491673e-01 3.65261175e-02 2.02613138e-02
1.51108301e+00 6.87012553e-01 -3.81066799e-01 5.65208554e-01
1.42359591e+00 -6.63784444e-02 -5.26655912e-01 -3.87997478e-01
-2.66603529e-01 -3.32611769e-01 -3.45487803e-01 -4.29456204e-01
-8.49224806e-01 1.29152668e+00 -1.19827129e-01 2.05145031e-02
6.98081255e-01 -3.66943687e-01 6.67009056e-01 1.38651276e+00
4.16246831e-01 -9.40123975e-01 2.46055648e-01 9.45375144e-01
5.60772300e-01 -1.12070513e+00 3.21767688e-01 -2.55933523e-01
-4.33534145e-01 1.32567656e+00 1.71961948e-01 -4.46250319e-01
7.35138118e-01 5.45723319e-01 -7.10773349e-01 -2.05862802e-03
-1.08822787e+00 1.63790941e-01 1.37427986e-01 7.88486123e-01
3.77978683e-01 4.39971238e-02 4.98517156e-01 6.37588143e-01
-7.32267976e-01 4.85582858e-01 6.79927289e-01 8.48003864e-01
-3.70998949e-01 -8.16810608e-01 -4.36926872e-01 7.67142653e-01
-8.05257633e-02 -5.35079658e-01 -5.96314520e-02 8.56769264e-01
-2.93162227e-01 5.60401142e-01 1.25364766e-01 -4.35232967e-01
-7.44898291e-03 3.90118092e-01 1.03189945e+00 -6.00669086e-01
-3.34730834e-01 -5.54559708e-01 4.45919394e-01 -5.75758398e-01
1.86765715e-01 -6.26823068e-01 -1.34825945e+00 -1.14202905e+00
1.15474269e-01 1.13318957e-01 1.94534913e-01 9.14039612e-01
2.18843549e-01 6.65418506e-01 4.45059747e-01 -5.88162124e-01
-1.46280563e+00 -9.97485816e-01 -7.51870215e-01 1.05517641e-01
3.70201826e-01 -5.62481403e-01 -4.38913584e-01 -1.19780913e-01] | [9.140740394592285, 7.1045451164245605] |
98f6765e-5953-41ab-93cb-667a4ccb16bc | user-assisted-video-reflection-removal | 2009.03281 | null | https://arxiv.org/abs/2009.03281v1 | https://arxiv.org/pdf/2009.03281v1.pdf | User-assisted Video Reflection Removal | Reflections in videos are obstructions that often occur when videos are taken behind reflective surfaces like glass. These reflections reduce the quality of such videos, lead to information loss and degrade the accuracy of many computer vision algorithms. A video containing reflections is a combination of background and reflection layers. Thus, reflection removal is equivalent to decomposing the video into two layers. This, however, is a challenging and ill-posed problem as there is an infinite number of valid decompositions. To address this problem, we propose a user-assisted method for video reflection removal. We rely on both spatial and temporal information and utilize sparse user hints to help improve separation. The key idea of the proposed method is to use motion cues to separate the background layer from the reflection layer with minimal user assistance. We show that user-assistance significantly improves the layer separation results. We implement and evaluate the proposed method through quantitative and qualitative results on real and synthetic videos. Our experiments show that the proposed method successfully removes reflection from video sequences, does not introduce visual distortions, and significantly outperforms the state-of-the-art reflection removal methods in the literature. | ['Mohamed Hefeeda', 'Mohamed Elgharib', 'Amgad Ahmed', 'Suhong Kim'] | 2020-09-07 | null | null | null | null | ['reflection-removal'] | ['computer-vision'] | [ 6.85694158e-01 -2.90806919e-01 2.63234615e-01 1.53535634e-01
-5.29134214e-01 -5.02472281e-01 4.12236392e-01 -3.75065953e-01
-1.24377020e-01 3.50011051e-01 2.95388341e-01 -1.59903571e-01
9.82796475e-02 -4.91099387e-01 -6.44143105e-01 -8.78810763e-01
1.25713721e-01 -5.36379814e-01 5.85579276e-01 -4.93405797e-02
3.73547226e-01 2.79253930e-01 -1.73257923e+00 4.69935775e-01
9.35067415e-01 8.26607943e-01 3.52183372e-01 6.82082117e-01
1.66621521e-01 7.67797291e-01 -4.39935178e-01 -1.75516736e-02
6.07604325e-01 -3.41654420e-01 -8.70540366e-02 4.69050735e-01
8.32404196e-01 -6.58104062e-01 -4.49055165e-01 1.13875592e+00
2.92906463e-01 3.74510646e-01 5.47159195e-01 -9.11403656e-01
-1.99356437e-01 -8.66666809e-03 -9.95666802e-01 1.95942551e-01
7.14858413e-01 -1.31614298e-01 4.42555815e-01 -9.88177896e-01
3.47864717e-01 1.15117908e+00 7.15232611e-01 3.82764429e-01
-7.95436800e-01 -4.88076001e-01 2.12495014e-01 3.77451509e-01
-1.19797528e+00 -8.72107863e-01 9.69456315e-01 -5.08322954e-01
5.38839698e-01 5.78222096e-01 5.60403109e-01 7.45826840e-01
4.23984751e-02 6.72985971e-01 8.33397210e-01 -5.27317166e-01
7.87433982e-02 1.65521964e-01 1.80082291e-01 7.61937439e-01
5.82313478e-01 -8.10756758e-02 -4.07337636e-01 -8.32739323e-02
7.02093422e-01 3.96501601e-01 -9.20532227e-01 -2.43349865e-01
-8.50055754e-01 1.40414327e-01 1.44693069e-02 8.35408345e-02
-3.98392975e-01 -1.53490767e-01 2.39097849e-02 1.75052494e-01
3.83990079e-01 -7.09987730e-02 1.64075658e-01 1.43734396e-01
-1.13808227e+00 7.87582546e-02 7.22109377e-01 8.04706514e-01
5.27970016e-01 1.60217553e-01 -1.55534213e-02 9.74253714e-01
5.50664246e-01 6.86057389e-01 1.14112072e-01 -9.76682127e-01
5.94041884e-01 4.32053775e-01 4.69297677e-01 -1.44784367e+00
-4.98608239e-02 -1.39955223e-01 -8.04394007e-01 4.38423991e-01
3.37077081e-01 -3.55995856e-02 -8.38704586e-01 1.18791831e+00
4.86922771e-01 5.93703389e-01 -6.40902445e-02 1.23744845e+00
9.19288695e-01 7.45987535e-01 -4.76172447e-01 -6.34746313e-01
1.13189089e+00 -1.01603484e+00 -9.76561546e-01 -2.84169346e-01
2.96799429e-02 -1.22616637e+00 7.67171502e-01 8.19832265e-01
-1.21836197e+00 -5.15258431e-01 -1.22130036e+00 1.70091689e-01
4.05563772e-01 3.02492917e-01 1.04023643e-01 7.59065628e-01
-7.84095407e-01 2.96824366e-01 -8.76223683e-01 -2.45354310e-01
-1.48021383e-03 4.84385304e-02 -3.08207035e-01 -6.47964299e-01
-7.17400312e-01 6.23642862e-01 -2.57288754e-01 5.85380971e-01
-6.98142886e-01 -4.12406683e-01 -7.09939122e-01 -1.90089673e-01
7.38910317e-01 -4.64764535e-01 9.49648261e-01 -9.86660063e-01
-1.41344619e+00 3.38747352e-01 -4.18902248e-01 -4.16385792e-02
6.08355224e-01 -6.15400791e-01 -4.00897563e-01 4.68602687e-01
-3.00433010e-01 -2.68104464e-01 1.26044357e+00 -1.69251370e+00
-5.79557776e-01 -1.75991192e-01 2.00643644e-01 4.21466678e-01
-2.75908768e-01 -1.34207562e-01 -8.89177978e-01 -4.87230599e-01
4.78176981e-01 -9.17146146e-01 -2.54184823e-03 7.75862187e-02
-3.13840389e-01 5.45502365e-01 1.03978872e+00 -1.13752401e+00
1.16654122e+00 -2.27544069e+00 2.12398544e-02 8.61944556e-02
2.88503677e-01 5.82076073e-01 -7.74922892e-02 2.92843342e-01
1.59845009e-01 -2.66224176e-01 -1.22863263e-01 -3.30387503e-01
-4.42429811e-01 -5.52903786e-02 -6.85135275e-02 8.74285996e-01
-2.68630207e-01 -5.42121604e-02 -7.15359747e-01 -2.03364700e-01
5.23873210e-01 9.14687812e-01 -6.98628366e-01 2.98450381e-01
2.66814768e-01 4.00387436e-01 -2.99214125e-01 6.78119719e-01
1.07740700e+00 2.18910620e-01 2.19020247e-01 -5.87404728e-01
-2.88435012e-01 1.28897680e-02 -1.57984996e+00 1.19304025e+00
-3.31911296e-01 8.39182556e-01 4.81690854e-01 -8.72330546e-01
7.19599068e-01 2.74067014e-01 4.40877765e-01 -6.58521175e-01
-4.16767634e-02 1.59706950e-01 -9.33273584e-02 -9.45971608e-01
4.79331195e-01 1.26335388e-02 5.49128294e-01 9.26020667e-02
-5.24222016e-01 3.36542994e-01 2.30049744e-01 -1.12216864e-02
1.30830014e+00 2.10754663e-01 2.25352675e-01 1.12946406e-01
8.86558950e-01 -5.32967448e-01 8.65193665e-01 6.23632133e-01
1.20813986e-02 8.88342261e-01 7.83418119e-03 -1.37798503e-01
-7.09289432e-01 -1.02558446e+00 3.32674682e-01 5.10426044e-01
5.45358360e-01 -3.88158739e-01 -7.91127264e-01 -1.61279857e-01
-3.79197180e-01 3.36529970e-01 -2.32178017e-01 -7.74540529e-02
-7.38435209e-01 -4.57900554e-01 1.87438931e-02 6.15140200e-02
6.11441791e-01 -6.62959278e-01 -8.05026293e-01 6.14399537e-02
-7.34385788e-01 -1.47780061e+00 -4.24449742e-01 -6.30996048e-01
-1.00124204e+00 -1.46676075e+00 -8.91891539e-01 -6.20569944e-01
9.86774921e-01 1.29436338e+00 8.48054886e-01 4.42127943e-01
-5.18453240e-01 6.68397903e-01 -5.11592567e-01 -3.11942212e-03
-2.26951897e-01 -7.17769623e-01 3.88469286e-02 4.76683825e-01
6.21722313e-03 -5.67261815e-01 -9.25642252e-01 5.54036677e-01
-9.33237255e-01 2.44566366e-01 5.44901490e-01 3.94812137e-01
2.51021862e-01 4.40824032e-01 -2.51861066e-01 -7.53056288e-01
3.70838940e-01 -2.00613290e-01 -6.63607717e-01 1.13082945e-01
-1.86256375e-02 -2.50523508e-01 6.93411171e-01 -3.51116210e-01
-1.36781299e+00 -4.47717980e-02 3.38209778e-01 -6.05096161e-01
-9.22062695e-02 1.86818197e-01 -1.78437412e-01 -2.13270605e-01
3.52265567e-01 2.00175479e-01 -1.80453256e-01 -5.65403879e-01
1.83551069e-02 6.63393617e-01 4.62812513e-01 -1.44112229e-01
9.04440522e-01 8.05865467e-01 6.88949451e-02 -1.66913795e+00
-5.59076965e-01 -7.57847488e-01 -3.02010953e-01 -6.71740592e-01
6.51243925e-01 -8.62307787e-01 -6.63802326e-01 4.50735807e-01
-1.11121762e+00 -7.57442564e-02 5.05221665e-01 5.74203193e-01
-2.04670072e-01 8.73478115e-01 -4.53958601e-01 -1.32914686e+00
-1.11170456e-01 -1.12468576e+00 6.77267373e-01 2.70438433e-01
1.89732537e-01 -6.98983550e-01 -4.92123663e-02 6.79809451e-01
2.09304035e-01 2.04342172e-01 4.28392678e-01 2.44078368e-01
-9.84178662e-01 -1.15988582e-01 -1.70120031e-01 4.96095508e-01
4.82959181e-01 9.75103676e-02 -9.94558156e-01 -4.14722055e-01
3.39810789e-01 3.96204829e-01 1.00290477e+00 3.34211707e-01
5.75806499e-01 -3.68009090e-01 -1.62550524e-01 5.71195483e-01
1.50637484e+00 2.75550038e-01 9.57513154e-01 2.75913775e-01
9.80252564e-01 6.70799971e-01 9.06833470e-01 5.30586123e-01
2.21907385e-02 7.64506400e-01 5.37284553e-01 -1.87759101e-01
-3.86137009e-01 2.15281412e-01 7.15827525e-01 9.54840899e-01
-6.41290784e-01 -4.20238853e-01 -6.27391219e-01 2.78403193e-01
-1.97236466e+00 -1.20382798e+00 -4.45473641e-01 2.66114831e+00
3.56947154e-01 -3.71663831e-02 -1.51762709e-01 5.37339747e-01
7.16222405e-01 7.15167969e-02 -2.20934041e-02 1.34982048e-02
1.55025870e-01 -2.10612863e-01 5.00381947e-01 7.98085153e-01
-9.52357054e-01 5.10388017e-01 5.63109970e+00 2.91260183e-01
-1.12870491e+00 -8.84012952e-02 2.29074843e-02 -2.58269072e-01
-2.03523725e-01 -1.41689837e-01 -5.23339391e-01 3.71645212e-01
3.96332830e-01 3.59947920e-01 4.71009254e-01 3.02567303e-01
7.35588133e-01 -4.39019233e-01 -9.21792209e-01 1.24683797e+00
5.26960254e-01 -7.92168081e-01 -1.55593425e-01 -1.06505498e-01
5.64198434e-01 -4.82018292e-01 -1.34507213e-02 -1.99827269e-01
-3.80902857e-01 -6.64430320e-01 4.57247794e-01 6.77644789e-01
3.93471450e-01 -6.90945208e-01 4.74947989e-01 2.52626091e-01
-1.23059666e+00 -6.98551759e-02 -3.72710735e-01 -4.63308133e-02
2.87052095e-01 8.37114692e-01 -4.74021196e-01 5.83849967e-01
7.04538286e-01 7.43903041e-01 -1.46531209e-01 1.42592907e+00
-1.80282936e-01 4.77411777e-01 -3.04926813e-01 4.58617270e-01
-1.24552429e-01 -6.87590063e-01 8.91346633e-01 1.18045568e+00
3.85707468e-01 4.50309783e-01 2.59120166e-01 3.72974187e-01
2.83367932e-01 -6.57628402e-02 -6.73759699e-01 1.60814613e-01
1.12840831e-01 1.09512663e+00 -5.57940900e-01 -1.32806510e-01
-8.41874838e-01 1.13840544e+00 -3.29628468e-01 7.72068441e-01
-8.47247601e-01 -3.14822793e-01 6.69160843e-01 4.97310191e-01
3.24303150e-01 -3.90579253e-01 2.15570986e-01 -1.38161039e+00
4.39825833e-01 -1.02751338e+00 3.54504474e-02 -7.07041979e-01
-6.56972468e-01 3.28747571e-01 -2.45444015e-01 -1.72888267e+00
7.30938390e-02 -5.59031785e-01 -4.36142415e-01 5.92936575e-01
-1.64270961e+00 -7.62198448e-01 -8.67278516e-01 6.38613164e-01
7.92893231e-01 6.07144572e-02 3.29432726e-01 9.43608940e-01
-5.56443512e-01 3.33160728e-01 2.66039550e-01 -5.16645871e-02
7.44625449e-01 -6.47132993e-01 -1.34829953e-01 1.48872054e+00
1.12623982e-02 6.06175125e-01 1.13039255e+00 -6.39866829e-01
-1.75473881e+00 -8.00129175e-01 2.86073536e-01 2.15415448e-01
7.95140639e-02 -2.18799368e-01 -8.80945563e-01 4.93060768e-01
-1.26156798e-02 -3.70161794e-02 6.39581859e-01 -2.66672909e-01
-2.46393397e-01 -2.17389002e-01 -1.03898239e+00 7.75051355e-01
1.00801015e+00 -2.81619430e-01 -4.73867595e-01 8.59331414e-02
2.63833642e-01 -3.46636772e-01 -2.43360758e-01 2.76247948e-01
7.43632376e-01 -1.57820714e+00 1.24098849e+00 1.90379947e-01
3.55519354e-01 -6.34964406e-01 -3.12587023e-01 -1.10813260e+00
-1.86595619e-01 -7.73718834e-01 -2.97436178e-01 9.81685638e-01
6.36913814e-03 -5.04447520e-01 6.46320045e-01 6.31604433e-01
-1.36149853e-01 -1.47343397e-01 -3.88518035e-01 -6.26232922e-01
-8.79545331e-01 -3.32837313e-01 -2.23694444e-01 7.45657086e-01
-1.31803423e-01 -2.52551697e-02 -7.95548797e-01 5.96548080e-01
1.17728245e+00 9.75858718e-02 9.42931116e-01 -1.02653730e+00
-4.80962038e-01 4.88268211e-03 -4.59153354e-01 -1.27716053e+00
-3.52731973e-01 -1.33821517e-01 3.69726896e-01 -1.70370996e+00
1.87284857e-01 -1.90823361e-01 -1.71064079e-01 5.62008843e-02
-2.15125337e-01 4.62806880e-01 3.33049417e-01 1.98207229e-01
-4.85790044e-01 3.60812664e-01 1.08890963e+00 2.49781590e-02
-3.66366625e-01 1.35574862e-01 -4.68169659e-01 1.15577424e+00
5.21542609e-01 -2.94600874e-01 -6.33289993e-01 -5.31532824e-01
1.50332540e-01 2.33775035e-01 2.06046909e-01 -1.21943629e+00
1.32394463e-01 -1.82199910e-01 2.46057034e-01 -5.66196144e-01
6.35635078e-01 -1.15083480e+00 1.81037948e-01 3.66324008e-01
1.61336541e-01 -3.65566850e-01 1.90990433e-01 7.79734910e-01
-1.30801469e-01 -2.57251531e-01 9.01266754e-01 -2.02549577e-01
-5.12060702e-01 -2.68230382e-02 -5.10168791e-01 -2.12717146e-01
8.64830434e-01 -4.88127232e-01 -2.61944652e-01 -5.59697449e-01
-4.19122577e-01 -1.34370476e-01 4.97254491e-01 3.53727430e-01
1.02065194e+00 -9.88057435e-01 -6.16184652e-01 2.56868690e-01
-1.96066990e-01 -1.05351083e-01 4.73239690e-01 1.04104388e+00
-7.90006638e-01 5.13557084e-02 -1.24615446e-01 -6.03834391e-01
-2.04038191e+00 3.77751261e-01 1.58217609e-01 1.89630300e-01
-9.21221137e-01 4.61612672e-01 5.21955967e-01 2.86775976e-01
3.83398980e-01 -2.30589151e-01 -4.73071277e-01 -1.31704479e-01
1.01081538e+00 8.80824566e-01 -2.11139712e-02 -6.27392828e-01
-3.46727282e-01 9.59269345e-01 -3.46768945e-02 -2.42951930e-01
1.18840253e+00 -5.05388856e-01 5.66337220e-02 3.63259137e-01
9.82501149e-01 4.89362240e-01 -1.18033719e+00 -2.66986966e-01
-5.34509480e-01 -1.07893288e+00 1.73071995e-01 -3.08158040e-01
-1.17097986e+00 8.60835493e-01 5.62529445e-01 1.48931161e-01
1.30588591e+00 -7.23077774e-01 8.30429494e-01 3.15813810e-01
2.81905770e-01 -9.17938769e-01 2.58067459e-01 1.90236345e-01
7.13701248e-01 -1.15112269e+00 3.71934623e-01 -9.59671199e-01
-2.52552390e-01 1.10430503e+00 4.68795955e-01 -1.93160102e-01
4.43061769e-01 2.92934507e-01 2.45319933e-01 1.12041190e-01
-4.04189527e-01 -1.23941958e-01 4.73259419e-01 5.46890318e-01
4.82882828e-01 -3.50397140e-01 -2.91740745e-01 1.83110118e-01
3.41933668e-01 -4.05810475e-02 9.09890294e-01 1.04901683e+00
-7.15815723e-01 -7.88278937e-01 -9.36363339e-01 1.30786315e-01
-5.51221907e-01 -2.15250384e-02 -1.06754243e-01 4.91043001e-01
-1.75557688e-01 1.43723953e+00 -2.43035451e-01 -2.64090300e-01
3.41429442e-01 -4.20230269e-01 7.47286379e-01 -3.52131039e-01
-9.50412005e-02 5.62793136e-01 1.47359118e-01 -7.69377291e-01
-6.84103072e-01 -5.18694520e-01 -1.04700983e+00 -2.49600410e-01
-3.59741241e-01 -1.25228047e-01 6.86396360e-01 8.09562445e-01
1.32613674e-01 6.72440469e-01 6.45388842e-01 -1.09817123e+00
-1.52066633e-01 -6.80690646e-01 -5.35391629e-01 5.53500891e-01
7.41219878e-01 -7.28843331e-01 -7.45051563e-01 3.17584306e-01] | [10.389802932739258, -2.6144344806671143] |
f72c8a03-94d3-4d90-b56e-31ff6b27aec7 | chain-of-thought-prompt-distillation-for | 2306.14122 | null | https://arxiv.org/abs/2306.14122v2 | https://arxiv.org/pdf/2306.14122v2.pdf | Chain-of-Thought Prompt Distillation for Multimodal Named Entity and Multimodal Relation Extraction | Multimodal Named Entity Recognition (MNER) and Multimodal Relation Extraction (MRE) necessitate the fundamental reasoning capacity for intricate linguistic and multimodal comprehension. In this study, we explore distilling the reasoning ability of large language models (LLMs) into a more compact student model by generating a \textit{chain of thought} (CoT) -- a sequence of intermediate reasoning steps. Specifically, we commence by exemplifying the elicitation of such reasoning ability from LLMs through CoT prompts covering multi-grain (noun, sentence, multimodality) and data-augmentation (style, entity, image) dimensions. Subsequently, we present a novel conditional prompt distillation method to assimilate the commonsense reasoning ability from LLMs, thereby enhancing the utility of the student model in addressing text-only inputs without the requisite addition of image and CoT knowledge. Extensive experiments reveal that our approach attains state-of-the-art accuracy and manifests a plethora of advantages concerning interpretability, data efficiency, and cross-domain generalization on MNER and MRE datasets. | ['Yujian Feng', 'Feng Chen'] | 2023-06-25 | null | null | null | null | ['domain-generalization', 'relation-extraction'] | ['methodology', 'natural-language-processing'] | [ 6.45941019e-01 4.81148064e-01 1.02589980e-01 -5.75902283e-01
-9.20657218e-01 -8.38666975e-01 8.16776335e-01 3.92346054e-01
-4.88595575e-01 5.47294259e-01 2.45747015e-01 -9.07119811e-01
-1.29805669e-01 -6.66247070e-01 -7.05132246e-01 -1.46936476e-01
2.53008664e-01 4.88240093e-01 -2.02349767e-01 -3.71478766e-01
4.41464223e-02 4.88097399e-01 -1.32655489e+00 8.01561892e-01
1.27397299e+00 1.14867795e+00 -9.88147035e-02 8.94768476e-01
-5.25671721e-01 1.30343807e+00 -4.79248405e-01 -1.04073262e+00
-2.96826124e-01 -4.42590296e-01 -1.13207030e+00 2.72186816e-01
4.58815277e-01 -2.49756217e-01 -9.51642767e-02 7.80396998e-01
2.61699706e-01 1.51300833e-01 9.27240729e-01 -1.21349633e+00
-1.10862899e+00 7.90721953e-01 -2.45200828e-01 5.88658974e-02
7.20026195e-01 1.78459421e-01 1.16634440e+00 -1.03263879e+00
4.74768311e-01 1.30687952e+00 3.17236513e-01 5.60208797e-01
-1.52887809e+00 -3.59300971e-01 2.93787003e-01 1.49684757e-01
-1.13392699e+00 -3.48796666e-01 7.88201809e-01 -5.21946013e-01
9.43037510e-01 4.96080428e-01 1.68815747e-01 1.42710364e+00
-3.58633220e-01 1.32029855e+00 1.35048103e+00 -6.59883857e-01
2.51798391e-01 4.96245205e-01 2.37420946e-01 8.16860855e-01
-1.19695514e-01 -3.13271791e-01 -7.64974177e-01 5.93713485e-02
6.63569808e-01 -2.96581239e-01 -2.34262452e-01 -1.60004944e-01
-1.36372137e+00 6.85404360e-01 1.98656425e-01 1.79095760e-01
-3.11660707e-01 -4.00777280e-01 2.94246495e-01 4.12533224e-01
6.30854145e-02 7.74433136e-01 -6.21526122e-01 -1.57725006e-01
-6.91685557e-01 -1.25578016e-01 1.00223219e+00 9.71625805e-01
4.48213607e-01 -1.37945622e-01 -2.82836318e-01 7.64532983e-01
1.31027058e-01 6.97311342e-01 5.27343273e-01 -6.13285780e-01
8.87947321e-01 1.09887505e+00 -5.88254631e-02 -8.43405187e-01
-3.20369661e-01 -1.54549107e-01 -9.76846576e-01 -3.08475822e-01
3.89345050e-01 -2.38494590e-01 -7.25885570e-01 1.89429545e+00
2.17503980e-01 -3.82899344e-01 5.14844060e-01 5.76257110e-01
1.10100615e+00 4.44454044e-01 6.52104020e-01 4.63078357e-02
1.67113817e+00 -6.61083937e-01 -6.32390499e-01 -4.14012522e-01
6.94867074e-01 -4.55274224e-01 1.42295563e+00 5.00333250e-01
-1.04524684e+00 -5.92267156e-01 -7.11622834e-01 -3.48835140e-01
-6.64102733e-01 4.01852846e-01 7.73473680e-01 5.80414534e-01
-7.11856127e-01 2.96602715e-02 -3.77288193e-01 -3.12668294e-01
3.19432288e-01 2.65450120e-01 -6.28656745e-01 -2.84776628e-01
-1.29276383e+00 1.04464793e+00 5.80406904e-01 1.94555700e-01
-4.40084636e-01 -5.14770329e-01 -1.14775097e+00 -8.36669095e-03
5.98678589e-01 -7.55802333e-01 1.31244993e+00 -9.07846570e-01
-1.42956102e+00 1.03778577e+00 -5.49266040e-02 -1.81637123e-01
3.43595415e-01 -2.16436073e-01 -5.83701313e-01 3.21514070e-01
-2.23479420e-01 9.56507921e-01 6.91377521e-01 -1.29135060e+00
-4.75291878e-01 -4.14580047e-01 3.45393568e-01 3.56347889e-01
-6.01782978e-01 -4.60903198e-02 -1.55223653e-01 -6.35225058e-01
2.51911301e-02 -7.22788036e-01 -1.21693658e-02 -3.35231632e-01
-7.44429350e-01 -3.57567281e-01 2.88647890e-01 -6.87626481e-01
1.20099235e+00 -2.17102504e+00 4.83768493e-01 1.88678339e-01
4.18463320e-01 1.81128159e-01 -4.45694715e-01 4.62953180e-01
-2.13651046e-01 1.70089319e-01 -2.30223000e-01 -3.11913133e-01
4.14329141e-01 3.03596020e-01 -3.25432956e-01 -2.23990977e-01
7.91818321e-01 1.40959096e+00 -6.50275350e-01 -6.47134244e-01
6.36839941e-02 1.77040443e-01 -4.46771294e-01 4.43342865e-01
-4.86872464e-01 5.93445122e-01 -4.64454800e-01 8.13423336e-01
4.19770956e-01 -5.12140691e-01 2.86469400e-01 -4.00996447e-01
1.95240974e-01 1.62101090e-01 -1.13278592e+00 1.52274692e+00
-5.52229047e-01 4.77707326e-01 -8.69156346e-02 -6.83165252e-01
7.79515445e-01 2.20473796e-01 -4.68245819e-02 -7.32845068e-01
2.81991363e-01 3.90479784e-03 -3.45642157e-02 -9.02796388e-01
4.87949193e-01 -3.57639223e-01 -4.93038714e-01 3.89607370e-01
3.20777714e-01 1.32296551e-02 2.31268525e-01 4.56221163e-01
8.37853730e-01 -7.76866004e-02 2.85662323e-01 1.70936480e-01
6.57594144e-01 -3.61721404e-02 -4.32660133e-02 6.68554187e-01
6.02893718e-02 -6.19757595e-03 7.05424845e-01 3.05865472e-03
-7.53295183e-01 -1.12980413e+00 2.01149330e-01 1.54581404e+00
-2.26665244e-01 -2.06740290e-01 -5.96985579e-01 -7.18952179e-01
-1.15308046e-01 1.00193334e+00 -4.98681873e-01 -1.54241949e-01
-2.78213024e-01 -6.69967532e-01 9.22255695e-01 8.13850462e-01
4.74897295e-01 -1.03847802e+00 -4.52231735e-01 -8.43610149e-03
-5.10075271e-01 -1.77360165e+00 -1.64131038e-02 2.35028058e-01
-7.77044952e-01 -8.42024446e-01 -3.81944507e-01 -7.47557282e-01
8.71446550e-01 -4.04876202e-01 1.21541655e+00 -2.58175939e-01
-1.68826446e-01 7.83707678e-01 -4.95899737e-01 -2.28410006e-01
-5.42320728e-01 4.50873226e-02 -1.41802505e-01 6.97031021e-02
5.63241959e-01 -3.68564576e-01 -1.31368667e-01 -7.10619614e-02
-1.31121862e+00 5.95899343e-01 1.17099440e+00 6.65936232e-01
2.10008860e-01 -7.18365237e-02 5.47550499e-01 -8.74240398e-01
9.14188564e-01 -3.45205039e-01 -2.10463300e-01 6.64825320e-01
-2.49529287e-01 3.06513667e-01 5.29489934e-01 -7.94706583e-01
-1.48809230e+00 9.93801728e-02 -1.61544695e-01 -3.37404795e-02
-8.14663708e-01 8.41888428e-01 -4.91603255e-01 1.43670961e-01
5.96524596e-01 3.93992394e-01 -1.50353000e-01 -4.79513377e-01
9.67427671e-01 7.09177673e-01 9.14549947e-01 -1.23020363e+00
7.48399019e-01 1.62761331e-01 -7.40310475e-02 -6.54407918e-01
-1.10150838e+00 -2.51456797e-01 -8.50673199e-01 2.23300606e-02
1.05678117e+00 -8.58553827e-01 -1.09552181e+00 2.08403468e-01
-1.27699113e+00 -1.37841478e-01 -3.48445207e-01 3.25496942e-01
-3.40425402e-01 3.32766473e-01 -7.95683801e-01 -9.87617612e-01
-2.12226659e-01 -1.01267374e+00 9.85234737e-01 1.74743354e-01
-4.21381474e-01 -9.51291621e-01 -2.81681985e-01 8.71084988e-01
1.27602205e-01 7.15049431e-02 1.47130919e+00 -1.31834292e+00
-5.75977623e-01 -2.35206038e-01 -5.21052122e-01 3.11567903e-01
-2.82031298e-01 -3.90642732e-01 -1.15443981e+00 2.10401744e-01
-3.94845754e-02 -8.32217276e-01 5.47753513e-01 -3.90032828e-01
1.00244117e+00 -4.64640617e-01 2.31470823e-01 2.00176433e-01
1.16896296e+00 1.61356926e-01 4.10161495e-01 1.10609233e-01
7.66360939e-01 8.65744174e-01 3.17128032e-01 2.97422796e-01
1.00714839e+00 1.01174019e-01 2.52923876e-01 -3.83124501e-01
1.43498763e-01 -3.51057500e-01 3.17442745e-01 6.82323515e-01
1.25078708e-01 -2.44043827e-01 -1.16170883e+00 4.61783141e-01
-1.61797082e+00 -7.47181773e-01 -5.48120551e-02 1.77502668e+00
1.30923378e+00 -1.74094632e-01 -1.16736308e-01 6.31706193e-02
2.25993693e-01 -2.28873417e-01 -4.03236330e-01 -4.69163150e-01
-3.28943491e-01 -3.94121016e-04 -2.53501367e-02 3.86326998e-01
-9.56383109e-01 1.08340359e+00 5.49045181e+00 7.63501346e-01
-7.38991320e-01 -2.30962381e-01 7.25234807e-01 4.79249418e-01
-5.94520628e-01 -3.16397786e-01 -8.28472555e-01 -1.26211733e-01
9.79473472e-01 3.88572454e-01 4.29436684e-01 4.67712969e-01
-3.57846349e-01 -3.00700247e-01 -1.68738258e+00 9.05229807e-01
3.29643995e-01 -9.65412319e-01 5.45356393e-01 -1.70962840e-01
3.90183538e-01 -5.10615587e-01 9.72363204e-02 7.10695267e-01
2.24072248e-01 -1.26768196e+00 6.39130592e-01 6.22206032e-01
8.34937036e-01 -5.67623556e-01 6.21356547e-01 6.75678432e-01
-8.91267002e-01 -2.39383891e-01 3.62092853e-01 7.67256096e-02
3.12391985e-02 1.42873272e-01 -8.87780905e-01 7.21099436e-01
1.85496956e-01 -9.42074955e-02 -1.08666217e+00 3.90345633e-01
-3.41525942e-01 1.95925757e-01 -1.95070878e-01 -1.69152915e-01
1.42856717e-01 -2.59679798e-02 3.34320664e-01 1.49942613e+00
-2.31724858e-01 5.59044182e-01 2.39982456e-01 8.91222000e-01
-2.56356031e-01 2.18987897e-01 -3.74144346e-01 -5.99183500e-01
3.51460874e-01 1.22325039e+00 -5.52089095e-01 -6.43310249e-01
-5.24801970e-01 1.14136064e+00 5.57797551e-01 4.73294139e-01
-5.69889724e-01 -3.29777986e-01 1.27598614e-01 -3.36785913e-01
2.43074238e-01 -2.63336986e-01 -5.34637988e-01 -1.48361003e+00
2.34825406e-02 -1.32295918e+00 5.46484530e-01 -9.98494446e-01
-1.52343583e+00 7.09825873e-01 1.18421629e-01 -6.03844285e-01
-3.56339097e-01 -1.03775525e+00 -1.30667016e-01 8.19761336e-01
-1.39305699e+00 -1.65950942e+00 -1.72206789e-01 7.48475492e-01
5.73967695e-01 -4.00076136e-02 1.09374940e+00 2.61207819e-01
-6.45487368e-01 6.71500564e-01 -5.84372938e-01 4.85897303e-01
4.54853863e-01 -1.34688008e+00 -7.83144906e-02 7.08720744e-01
2.30978370e-01 8.99274886e-01 4.69963759e-01 -5.46394765e-01
-1.62520611e+00 -7.73477793e-01 1.23626018e+00 -8.59223902e-01
1.08247972e+00 -4.18239832e-01 -7.83035755e-01 9.04090047e-01
3.10171068e-01 -5.16008496e-01 1.07831430e+00 3.78312469e-01
-7.13500142e-01 2.87211478e-01 -1.04580891e+00 9.75552678e-01
6.36623204e-01 -9.78971243e-01 -1.07995057e+00 6.43988848e-02
6.43187940e-01 -4.10584122e-01 -1.14197147e+00 4.73228216e-01
2.86642224e-01 -5.50171375e-01 9.98796701e-01 -1.12880456e+00
7.33935058e-01 -2.82844659e-02 -4.32661802e-01 -8.96868944e-01
6.23220392e-02 -3.92068088e-01 -2.85304576e-01 1.58482611e+00
7.42785394e-01 -2.66914397e-01 3.61109436e-01 1.21041071e+00
2.38662899e-01 -7.63679028e-01 -6.27566695e-01 -3.33168268e-01
-6.80587813e-02 -8.17731917e-01 5.59642136e-01 1.15232265e+00
4.59631801e-01 9.10586059e-01 -3.57843153e-02 3.84020299e-01
3.42343301e-01 1.14546962e-01 6.09395564e-01 -1.19162309e+00
-3.54531884e-01 -3.00782084e-01 -9.78647098e-02 -1.24384224e+00
3.04730237e-01 -9.09964442e-01 -6.54645711e-02 -1.46927249e+00
4.34303254e-01 -7.54062235e-02 -4.15282995e-02 8.27280879e-01
-3.32836688e-01 -1.35981701e-02 3.28255802e-01 -1.28026903e-01
-9.39421296e-01 5.05938649e-01 1.33791018e+00 7.43283182e-02
-8.64353627e-02 -2.08464712e-01 -8.59601438e-01 8.15921128e-01
3.22310895e-01 6.17569499e-02 -5.34479320e-01 -5.31098723e-01
5.34577310e-01 4.20597166e-01 7.20882237e-01 -5.08240879e-01
3.80575478e-01 -1.21205531e-01 5.11771262e-01 -4.20861632e-01
4.00852114e-01 -8.81674886e-01 -4.02289689e-01 -1.66075692e-01
-8.54507923e-01 1.63751617e-01 4.47818786e-01 5.50965130e-01
-2.74924576e-01 -1.43617719e-01 2.97068924e-01 -3.33207054e-03
-8.51026952e-01 -1.67747587e-01 -4.72778946e-01 2.71574676e-01
6.53774679e-01 -1.05858026e-02 -3.67043823e-01 -2.36319199e-01
-1.11288774e+00 2.20755786e-01 -5.02890646e-02 4.28199112e-01
6.86316788e-01 -1.12610066e+00 -4.28716093e-01 3.22687149e-01
4.13768709e-01 -6.55863136e-02 3.10288846e-01 9.29223239e-01
3.18968892e-02 5.01388550e-01 1.61558576e-02 -4.51311082e-01
-1.17645931e+00 5.30509293e-01 2.99372561e-02 -3.96705121e-01
-2.51973897e-01 9.57830191e-01 1.73581183e-01 -8.60672355e-01
2.98009187e-01 -4.26471144e-01 -2.94559389e-01 2.40193531e-01
2.85967886e-01 -2.23300303e-03 -9.56658572e-02 -5.00934541e-01
-8.32400098e-02 6.81245029e-02 -6.40563220e-02 -4.32603538e-01
1.01047719e+00 -2.30256513e-01 -1.27137378e-01 5.30528128e-01
8.67802799e-01 -3.22127678e-02 -7.84223139e-01 -3.69285762e-01
4.10661787e-01 3.47080715e-02 -2.16568038e-01 -1.45578432e+00
-3.15859765e-01 9.08476174e-01 2.62027889e-01 1.77147254e-01
1.24198747e+00 3.96076918e-01 6.56354129e-01 8.32155943e-01
1.20121837e-02 -8.80857468e-01 2.00329885e-01 7.07810462e-01
7.77043104e-01 -1.50720835e+00 -4.10578609e-01 -4.78850871e-01
-1.21000767e+00 1.14251661e+00 7.69984841e-01 6.14862621e-01
1.51131541e-01 2.30661631e-01 8.00457671e-02 -3.28645945e-01
-7.26057529e-01 -3.13873023e-01 7.74963081e-01 4.24582630e-01
5.70938945e-01 6.27056062e-02 1.50881261e-01 1.09194076e+00
-2.38840640e-01 -9.72257257e-02 2.07089446e-02 1.00892377e+00
-2.46470705e-01 -8.78315687e-01 -3.70846748e-01 1.41104057e-01
-2.41231680e-01 -6.29752874e-01 -6.33655250e-01 8.54278743e-01
9.45406631e-02 1.01461291e+00 -7.16456473e-02 -3.03052276e-01
4.83374834e-01 5.14764428e-01 5.37919283e-01 -5.10225356e-01
-5.85460126e-01 -2.07520902e-01 4.54175532e-01 -2.53307372e-01
-2.17017800e-01 -3.18599015e-01 -1.26124382e+00 -7.57747293e-02
-1.43762499e-01 -1.59437321e-02 6.53751314e-01 1.44900310e+00
2.31954068e-01 5.45284510e-01 1.72910631e-01 -3.99820834e-01
-6.68971300e-01 -9.89383638e-01 -2.13128269e-01 5.78355491e-01
2.89372087e-01 -1.00881808e-01 -1.27994314e-01 4.09564197e-01] | [10.990202903747559, 8.132355690002441] |
55d3535a-0ba3-44a0-9a0c-eb9321a64aab | sinddm-a-single-image-denoising-diffusion | 2211.16582 | null | https://arxiv.org/abs/2211.16582v3 | https://arxiv.org/pdf/2211.16582v3.pdf | SinDDM: A Single Image Denoising Diffusion Model | Denoising diffusion models (DDMs) have led to staggering performance leaps in image generation, editing and restoration. However, existing DDMs use very large datasets for training. Here, we introduce a framework for training a DDM on a single image. Our method, which we coin SinDDM, learns the internal statistics of the training image by using a multi-scale diffusion process. To drive the reverse diffusion process, we use a fully-convolutional light-weight denoiser, which is conditioned on both the noise level and the scale. This architecture allows generating samples of arbitrary dimensions, in a coarse-to-fine manner. As we illustrate, SinDDM generates diverse high-quality samples, and is applicable in a wide array of tasks, including style transfer and harmonization. Furthermore, it can be easily guided by external supervision. Particularly, we demonstrate text-guided generation from a single image using a pre-trained CLIP model. | ['Tomer Michaeli', 'Matan Kleiner', 'Shahar Yadin', 'Vladimir Kulikov'] | 2022-11-29 | null | null | null | null | ['text-guided-image-editing', 'single-image-generation', 'text-guided-generation'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 4.09294218e-01 -8.51728916e-02 2.82413155e-01 -1.84203476e-01
-7.34426260e-01 -5.50828218e-01 8.64927888e-01 -5.12876153e-01
-1.78813234e-01 4.98194754e-01 3.67556572e-01 1.99672192e-01
2.27237016e-01 -8.49763989e-01 -8.09736013e-01 -8.27495992e-01
5.16457558e-01 3.80186051e-01 -9.97304246e-02 -3.43246609e-01
3.14857662e-02 4.29620534e-01 -1.16160679e+00 3.62301528e-01
1.01353920e+00 8.65347505e-01 5.64271688e-01 8.95647526e-01
-5.33966273e-02 5.97227693e-01 -7.75808692e-01 -5.69904983e-01
2.67099410e-01 -7.76402295e-01 -3.93556058e-01 3.56184274e-01
5.49140692e-01 -4.57935214e-01 -3.53083640e-01 9.96767044e-01
7.46631801e-01 5.31978719e-02 8.24526548e-01 -9.64419663e-01
-1.24149239e+00 4.05725271e-01 -6.35989487e-01 -2.23192230e-01
1.15939669e-01 3.48954767e-01 9.45716858e-01 -9.12930608e-01
9.39772427e-01 1.42085540e+00 4.86371577e-01 8.45046818e-01
-1.78903890e+00 -5.83962262e-01 -1.01631403e-01 -1.54502451e-01
-1.02179801e+00 -4.75855380e-01 9.86680031e-01 -5.26419461e-01
2.98667669e-01 5.36737852e-02 7.81149745e-01 1.48717248e+00
1.54472411e-01 8.82607579e-01 1.30304432e+00 -4.30802554e-01
2.65541106e-01 -2.25173205e-01 -6.70245528e-01 4.42571133e-01
-1.26269639e-01 6.60647592e-03 -6.90284312e-01 1.90000921e-01
1.33054459e+00 -2.50290990e-01 -2.32583404e-01 -2.80625731e-01
-1.23971832e+00 8.57692659e-01 4.41028118e-01 1.87521398e-01
-3.64072084e-01 3.26134413e-01 1.40281782e-01 4.38095868e-01
6.69301093e-01 4.38584179e-01 -1.82238072e-01 5.74318022e-02
-1.13952875e+00 4.03170615e-01 5.22556186e-01 8.95529985e-01
7.04599500e-01 2.33307406e-01 -6.16265953e-01 1.15638876e+00
5.90352565e-02 7.37173140e-01 4.46753263e-01 -1.31554163e+00
2.86489755e-01 -3.36067006e-02 1.17116064e-01 -8.07767034e-01
-1.30612567e-01 -3.47221941e-01 -1.33006644e+00 6.97576582e-01
3.27959150e-01 -2.84636289e-01 -1.23995924e+00 1.99056458e+00
3.39151889e-01 1.24265805e-01 -9.33505446e-02 1.00564241e+00
5.41115344e-01 7.19110906e-01 -2.16520771e-01 3.98491621e-02
9.76882935e-01 -1.01323915e+00 -7.73205638e-01 -2.31753096e-01
2.44284924e-02 -1.03846037e+00 1.24682009e+00 5.60845375e-01
-1.45828938e+00 -7.01279581e-01 -9.29606736e-01 -4.87233043e-01
-6.20763376e-02 2.88861156e-01 3.41072410e-01 2.36814335e-01
-1.24095297e+00 8.12689185e-01 -8.04945111e-01 -1.24449901e-01
5.90688109e-01 -5.99528179e-02 -1.44004628e-01 -1.48125947e-01
-8.86763632e-01 7.64195919e-01 -4.27228846e-02 1.89990148e-01
-1.16826022e+00 -7.12962508e-01 -7.74401367e-01 -1.45586893e-01
-6.71736822e-02 -1.18889201e+00 1.00513744e+00 -1.09426725e+00
-2.07386208e+00 1.00279748e+00 -7.06059188e-02 -2.99592972e-01
8.14953268e-01 -3.83257747e-01 -2.91286677e-01 3.92364059e-03
1.54602304e-01 9.71071422e-01 1.56814098e+00 -1.34856033e+00
-2.96331197e-01 -1.37959927e-01 -1.75993502e-01 1.46140233e-01
-2.92583019e-01 -1.37120172e-01 -6.63566113e-01 -1.35787094e+00
-2.39279181e-01 -8.05393755e-01 -2.88414896e-01 3.25796813e-01
-5.23499608e-01 3.86367023e-01 7.32698381e-01 -8.30110073e-01
1.04351485e+00 -2.29417396e+00 8.00328970e-01 1.11482859e-01
2.81896919e-01 1.64371565e-01 -5.67284524e-01 2.82349706e-01
8.93998221e-02 -5.58336899e-02 -5.11490643e-01 -7.63175726e-01
1.70946389e-01 1.59201816e-01 -4.77556765e-01 1.22316040e-01
4.03555095e-01 1.02452123e+00 -7.84081817e-01 -2.67132014e-01
1.91599339e-01 8.51839542e-01 -6.98106945e-01 5.37605345e-01
-4.43686426e-01 8.32332075e-01 -1.23282462e-01 2.98378885e-01
8.02118063e-01 -1.46876663e-01 -1.56425744e-01 -3.15353900e-01
6.14796951e-02 -1.19598500e-01 -1.17011189e+00 2.15524197e+00
-7.75558770e-01 7.89024413e-01 3.61335516e-01 -5.22113442e-01
9.52117503e-01 8.77979770e-02 1.30037874e-01 -6.24498248e-01
-8.03184584e-02 2.78936654e-01 -3.27929795e-01 -1.95399582e-01
4.94629234e-01 -2.50581473e-01 1.53550506e-01 4.98417169e-01
2.89005905e-01 -7.23029792e-01 3.85178745e-01 2.22782269e-01
8.59289467e-01 3.53607386e-01 -2.03473061e-01 -9.39452499e-02
1.96851999e-01 -3.77444297e-01 5.68651319e-01 5.49690723e-01
3.38926643e-01 1.28624964e+00 5.00557423e-01 -2.21439630e-01
-1.44236934e+00 -1.31085193e+00 4.17580493e-02 8.48919928e-01
-3.44399437e-02 -3.39871138e-01 -1.08164442e+00 -2.17945755e-01
-3.79615240e-02 5.14661551e-01 -7.06073940e-01 -1.73976198e-01
-5.62162638e-01 -6.09519482e-01 3.92355144e-01 3.31341267e-01
6.20160460e-01 -1.22317553e+00 -1.72604412e-01 3.08194846e-01
-1.28021225e-01 -9.86816704e-01 -9.86075163e-01 -7.21085966e-02
-7.70206034e-01 -5.59647799e-01 -1.25008059e+00 -8.12138438e-01
6.82139277e-01 9.29772407e-02 1.28954470e+00 -4.10501473e-02
-3.34027708e-01 1.38594940e-01 2.79456261e-03 -3.13605011e-01
-6.97693825e-01 1.19510613e-01 -1.57729670e-01 2.80817837e-01
-4.12120610e-01 -9.70650434e-01 -8.90361428e-01 1.50034294e-01
-1.21354663e+00 4.24615949e-01 6.85053766e-01 9.89366651e-01
7.46426702e-01 -1.88470945e-01 5.65703273e-01 -8.91412854e-01
9.81759548e-01 -7.80418813e-02 -5.66893220e-01 1.73576623e-01
-4.68265504e-01 5.94006181e-02 6.69504583e-01 -6.28466785e-01
-1.29869926e+00 -6.53853863e-02 -2.70869166e-01 -5.62174618e-01
-8.82955119e-02 4.90894206e-02 -2.24434659e-01 -1.60179194e-02
8.03960681e-01 1.82053655e-01 1.10003643e-01 -6.17182016e-01
9.22247946e-01 3.32228065e-01 7.88477838e-01 -6.77043736e-01
1.03535831e+00 5.78161538e-01 -9.36925188e-02 -7.18331754e-01
-8.67597461e-01 1.63404882e-01 -6.48748040e-01 -1.27038792e-01
9.38019574e-01 -1.02806103e+00 -2.12297156e-01 1.01295102e+00
-1.17962182e+00 -1.10196662e+00 -6.00343168e-01 1.65507700e-02
-7.53647745e-01 1.78443361e-02 -1.00263596e+00 -3.12487990e-01
-4.11765158e-01 -9.75404501e-01 1.33323205e+00 1.99303314e-01
-2.28125528e-01 -1.15349472e+00 1.54629782e-01 9.24242437e-02
7.53356934e-01 3.01717848e-01 7.72055387e-01 3.01614642e-01
-5.56642354e-01 2.20510185e-01 -1.40286967e-01 6.67532444e-01
1.72220126e-01 1.55891255e-01 -9.10853326e-01 -3.53192657e-01
-2.39993501e-02 -5.01350522e-01 1.06805944e+00 5.69519341e-01
1.22920656e+00 -7.44613484e-02 5.30301884e-04 1.10048950e+00
1.20983970e+00 -3.17219108e-01 7.09976852e-01 2.81165928e-01
1.00968838e+00 4.01100814e-01 1.30918235e-01 4.22483861e-01
2.58022040e-01 6.48729444e-01 1.56316683e-01 -6.34617746e-01
-8.20816576e-01 -3.93083990e-01 2.51314133e-01 8.02717209e-01
-9.72245634e-02 -2.60301620e-01 -4.16353256e-01 4.18149441e-01
-1.51356936e+00 -9.22662854e-01 8.17897692e-02 1.92131793e+00
1.32679856e+00 -2.01113243e-02 -5.89179248e-03 -1.45931125e-01
6.54801250e-01 3.99053246e-01 -7.82806873e-01 -2.19380528e-01
-4.98630315e-01 4.73953098e-01 1.44733191e-01 5.85566044e-01
-8.43531668e-01 1.09008849e+00 6.69753981e+00 9.92058694e-01
-1.26493597e+00 5.75483218e-02 8.12014937e-01 -1.93224177e-01
-6.82094991e-01 -2.52787352e-01 -4.73267913e-01 4.27686751e-01
5.23143649e-01 -1.15162186e-01 7.81683981e-01 4.02351916e-01
4.21826988e-01 1.82533145e-01 -9.47927296e-01 1.04432166e+00
8.52909461e-02 -1.48450673e+00 3.55186462e-01 -5.84190078e-02
1.24840927e+00 -1.29380375e-01 3.81028384e-01 -5.31210043e-02
6.23549283e-01 -1.00858748e+00 8.91866744e-01 6.76227450e-01
1.17435300e+00 -5.12393415e-01 4.03441414e-02 1.73777491e-01
-8.43987226e-01 7.34520927e-02 -3.74630243e-01 3.31278235e-01
4.45455045e-01 1.07832479e+00 -1.81668662e-02 2.72891223e-01
6.06419384e-01 8.25056672e-01 -3.53428423e-01 8.19471657e-01
-6.98175609e-01 4.58021104e-01 -2.00138148e-02 6.32477105e-01
-1.54791683e-01 -6.85173810e-01 4.29472148e-01 1.20796120e+00
6.19770110e-01 -9.71379355e-02 -1.29475027e-01 1.22311783e+00
-5.38041949e-01 -1.47399828e-01 -3.55938733e-01 1.29304916e-01
1.53541431e-01 1.47335362e+00 -4.45531636e-01 -3.36077720e-01
-1.32508695e-01 1.69495881e+00 4.71718311e-01 6.55418575e-01
-7.23267257e-01 -3.80722195e-01 7.10030317e-01 -4.57728803e-02
5.18386304e-01 -2.82919437e-01 -4.60511625e-01 -1.30806112e+00
-1.35218841e-03 -9.90377307e-01 -9.14533138e-02 -1.13510787e+00
-1.45984888e+00 5.70463479e-01 -4.74463701e-01 -1.06848919e+00
-1.04640156e-01 -4.19186234e-01 -8.73402178e-01 1.09773397e+00
-1.65476775e+00 -1.31944370e+00 -5.04049182e-01 6.47610962e-01
5.69130957e-01 -2.90691573e-02 7.43498623e-01 3.58611971e-01
-4.86749113e-01 4.08316910e-01 2.56353766e-01 1.44241199e-01
1.16733563e+00 -1.41060841e+00 7.62792349e-01 8.01565051e-01
2.03619003e-01 3.28070581e-01 7.62882054e-01 -6.35049224e-01
-1.12157619e+00 -1.23815477e+00 4.75223690e-01 -2.15350255e-01
4.98865902e-01 -5.06460667e-01 -8.63307893e-01 5.63078403e-01
5.17081439e-01 7.70068914e-02 3.31142962e-01 -2.00676501e-01
-3.33576351e-01 -2.59672701e-01 -1.00565839e+00 8.09845269e-01
1.21234775e+00 -6.27054095e-01 -1.89930931e-01 2.95832604e-01
6.21846557e-01 -7.47022212e-01 -9.38447654e-01 1.24416500e-02
4.73524064e-01 -9.94400799e-01 9.71598744e-01 -2.80311197e-01
9.93607938e-01 -2.63272494e-01 1.12638935e-01 -1.94726980e+00
-6.89497232e-01 -9.16556835e-01 -1.35962099e-01 1.49251485e+00
4.11230356e-01 -2.35798076e-01 5.35158157e-01 3.27794135e-01
-4.42725569e-02 -5.48172176e-01 -5.32715201e-01 -5.04081070e-01
2.87696630e-01 -2.11290032e-01 6.29715621e-01 7.03322828e-01
-6.91479146e-01 4.43407804e-01 -7.35913515e-01 -6.93736225e-02
7.82388449e-01 2.40543261e-01 9.48767722e-01 -9.76853788e-01
-5.02325952e-01 -5.76044917e-01 1.10451587e-01 -1.44138694e+00
1.10284425e-02 -7.39247262e-01 1.05641514e-01 -1.59909642e+00
1.16836771e-01 -4.01800215e-01 9.87635255e-02 7.97038227e-02
-3.04875374e-01 5.06650209e-01 2.88330853e-01 3.47720504e-01
-2.27596089e-01 8.80610585e-01 1.85608828e+00 -2.02719197e-01
-1.15119219e-01 -1.55611262e-01 -7.63996661e-01 6.35991037e-01
6.53021932e-01 -2.52577275e-01 -4.11056072e-01 -9.49138105e-01
2.06906483e-01 -1.30084395e-01 2.94137985e-01 -8.55688632e-01
2.23414619e-02 -1.83060423e-01 7.28423297e-01 -1.44988894e-01
4.24651295e-01 -4.52627152e-01 3.34178388e-01 7.98811615e-02
-5.66064477e-01 -1.37403205e-01 -3.16329822e-02 5.21704495e-01
-3.11846733e-01 8.67027491e-02 1.03873312e+00 -1.22170597e-01
-3.72688413e-01 5.39783001e-01 -4.32876032e-03 2.75435597e-01
7.14817822e-01 2.38758586e-02 -1.85327917e-01 -4.95134056e-01
-7.80613780e-01 3.26330252e-02 7.53402054e-01 3.82745147e-01
5.63803852e-01 -1.67747509e+00 -8.57980132e-01 3.95077527e-01
-1.81622460e-01 2.26608112e-01 3.31488907e-01 5.20490408e-01
-4.15451765e-01 -4.40125883e-01 -2.99834490e-01 -4.54513878e-01
-8.37622225e-01 3.71461809e-01 3.63052696e-01 -1.60080358e-01
-8.58087301e-01 9.07818675e-01 3.99143010e-01 -3.90610278e-01
5.28592058e-02 -1.97260529e-01 1.86081290e-01 4.41601779e-03
6.94464922e-01 1.23921283e-01 -1.90287217e-01 -4.54892218e-01
3.20769131e-01 8.41128290e-01 1.43706173e-01 -4.59999979e-01
1.51160085e+00 -2.01863304e-01 -2.03147709e-01 3.80467415e-01
1.06567383e+00 2.09802702e-01 -2.00462365e+00 -2.62648463e-01
-5.10617197e-01 -5.21856844e-01 1.24950148e-01 -7.77151823e-01
-1.49826837e+00 9.36203718e-01 4.02207077e-01 8.79326090e-02
1.28736973e+00 -2.59153396e-01 1.05667758e+00 -4.57362793e-02
1.37706259e-02 -1.30831265e+00 4.61480945e-01 3.19775343e-01
1.31322074e+00 -1.14654398e+00 -2.18164802e-01 -1.90059736e-01
-6.63347960e-01 1.11312592e+00 3.51511896e-01 -2.80061454e-01
5.16704559e-01 4.34241474e-01 3.27878773e-01 3.35265882e-02
-7.17329800e-01 5.00104204e-02 2.67657429e-01 7.47663975e-01
3.67928356e-01 1.40365371e-02 6.83329850e-02 2.07604051e-01
-3.03449243e-01 1.84699029e-01 4.12302285e-01 4.70814675e-01
-2.11569697e-01 -1.33684111e+00 -4.35514271e-01 2.61725545e-01
-3.42141241e-01 -2.35491499e-01 -2.21156090e-01 2.62401611e-01
1.24034710e-01 6.62505388e-01 3.72865051e-02 -1.48694873e-01
4.16244119e-01 -2.05569535e-01 6.07414663e-01 -4.02167559e-01
-3.36237133e-01 2.77909398e-01 -1.35207072e-01 -5.41871607e-01
-2.92985141e-01 -5.97810984e-01 -9.15015697e-01 -3.99407804e-01
-2.68313363e-02 -3.18850011e-01 5.92797220e-01 7.15058029e-01
5.22153199e-01 6.98909700e-01 6.93961442e-01 -1.28492260e+00
-3.87740880e-01 -1.00292146e+00 -7.94568360e-01 7.99755275e-01
4.38628018e-01 -3.56200218e-01 -3.02167475e-01 5.15338123e-01] | [11.465901374816895, -0.44852444529533386] |
231507f0-240a-4c27-a9c1-3c5ae840d9bc | constrained-causal-bayesian-optimization | 2305.20011 | null | https://arxiv.org/abs/2305.20011v1 | https://arxiv.org/pdf/2305.20011v1.pdf | Constrained Causal Bayesian Optimization | We propose constrained causal Bayesian optimization (cCBO), an approach for finding interventions in a known causal graph that optimize a target variable under some constraints. cCBO first reduces the search space by exploiting the graph structure and, if available, an observational dataset; and then solves the restricted optimization problem by modelling target and constraint quantities using Gaussian processes and by sequentially selecting interventions via a constrained expected improvement acquisition function. We propose different surrogate models that enable to integrate observational and interventional data while capturing correlation among effects with increasing levels of sophistication. We evaluate cCBO on artificial and real-world causal graphs showing successful trade off between fast convergence and percentage of feasible interventions. | ['Silvia Chiappa', 'Ira Ktena', 'Alan Malek', 'Virginia Aglietti'] | 2023-05-31 | null | null | null | null | ['gaussian-processes', 'bayesian-optimization'] | ['methodology', 'methodology'] | [ 5.67255974e-01 4.83951062e-01 -6.64471030e-01 -2.09087417e-01
-6.19043648e-01 -5.62013149e-01 8.95837247e-01 5.28623581e-01
-3.33846599e-01 1.10102570e+00 5.86554527e-01 -5.83926857e-01
-1.06918502e+00 -8.25532734e-01 -8.07434797e-01 -4.70588058e-01
-6.79197609e-01 8.13777268e-01 -2.55618058e-02 5.52340508e-01
1.90895140e-01 4.54647303e-01 -8.47389936e-01 -2.54958987e-01
1.29336476e+00 2.34517589e-01 1.74670562e-01 7.69093454e-01
7.07305193e-01 6.56027019e-01 -2.33649373e-01 -2.20809400e-01
5.53424582e-02 -5.21918893e-01 -5.53874493e-01 -1.64784491e-01
-8.38299394e-02 1.28316030e-01 -3.16051632e-01 8.82766008e-01
6.37668550e-01 1.38356268e-01 1.03821445e+00 -1.17955387e+00
-5.98836124e-01 8.16455662e-01 -5.43537021e-01 2.67117649e-01
5.10435402e-01 3.10879678e-01 1.04553962e+00 -2.68252462e-01
6.45808578e-01 1.69218302e+00 4.67124254e-01 1.78662866e-01
-1.98777771e+00 -3.95361066e-01 2.60152131e-01 6.12843335e-02
-1.09765351e+00 -1.92057848e-01 4.87024933e-01 -6.93966508e-01
7.21966505e-01 3.27390671e-01 5.67811012e-01 1.31923878e+00
4.65797603e-01 3.02334696e-01 1.37910116e+00 -3.48002881e-01
5.50166130e-01 -2.51981229e-01 3.09789300e-01 7.84752607e-01
5.10806084e-01 9.49990809e-01 -5.14831424e-01 -7.55940974e-01
5.80772698e-01 -8.07185620e-02 -3.66575927e-01 -4.19309378e-01
-1.17231131e+00 1.11073911e+00 1.76415935e-01 -3.89171630e-01
-9.89043951e-01 3.95226866e-01 1.00120172e-01 2.79683862e-02
3.25348139e-01 5.09639800e-01 -3.93629640e-01 1.53207660e-01
-7.19734073e-01 2.85686016e-01 8.49493444e-01 7.86063194e-01
1.98256776e-01 -2.38913432e-01 -6.35984063e-01 4.39888418e-01
5.59240580e-01 7.54141510e-01 -3.12660903e-01 -8.65431190e-01
2.99986809e-01 4.48266745e-01 3.30289304e-01 -9.04967248e-01
-5.69523811e-01 -2.60640502e-01 -8.14683378e-01 7.22243413e-02
3.56883317e-01 -5.38851619e-01 -1.04542255e+00 2.03958964e+00
6.41441047e-01 4.65181768e-01 -4.84527111e-01 6.11836314e-01
2.95184791e-01 4.38921332e-01 5.78884542e-01 -8.85216117e-01
1.17170656e+00 -3.87377828e-01 -8.15940559e-01 -1.47125095e-01
1.77504256e-01 -2.92559654e-01 9.39627588e-01 4.46269482e-01
-1.09364212e+00 -7.28294998e-02 -7.86242902e-01 5.96970499e-01
-7.05632418e-02 -2.19614580e-01 7.71804750e-01 9.43123817e-01
-8.93397331e-01 7.33746529e-01 -8.58146667e-01 -1.66346341e-01
5.85677326e-01 6.49662018e-01 -5.90553768e-02 -1.40354231e-01
-1.12062335e+00 8.11781466e-01 3.53338122e-01 -4.55533043e-02
-1.64493024e+00 -1.33387136e+00 -6.69811487e-01 1.83689892e-01
9.00364935e-01 -1.24314737e+00 7.03787506e-01 -2.41098031e-01
-1.43805981e+00 2.93723822e-01 -1.06207252e-01 -5.13152897e-01
5.03659725e-01 -1.75861910e-01 -2.40399808e-01 -9.62439999e-02
5.99899478e-02 2.15271771e-01 9.37156796e-01 -1.12043869e+00
-4.57359642e-01 -5.28266370e-01 -4.69036102e-02 2.04190090e-02
1.66853189e-01 2.30438948e-01 -1.68389574e-01 -5.48518896e-01
-4.34104294e-01 -8.75271738e-01 -8.37854922e-01 -5.85347950e-01
-7.35261261e-01 -1.97825566e-01 2.93514073e-01 -7.88982630e-01
1.50549638e+00 -1.51940870e+00 7.40050316e-01 5.98087728e-01
3.65457892e-01 -3.77918541e-01 -1.62317097e-01 4.50628161e-01
-2.16227859e-01 3.01797926e-01 -5.21668971e-01 4.96767946e-02
-1.20922811e-02 2.38537148e-01 -3.46091725e-02 8.67710829e-01
3.55601579e-01 9.39312696e-01 -1.14471555e+00 -5.84036231e-01
2.16342553e-01 1.09443113e-01 -8.62275422e-01 4.27631944e-01
-4.79080200e-01 6.30858243e-01 -7.44887590e-01 3.92283976e-01
3.44901294e-01 -2.32197225e-01 5.51125824e-01 1.90070480e-01
-8.96053165e-02 7.26561770e-02 -1.58686185e+00 1.15250933e+00
-3.02897632e-01 1.01887919e-01 1.01871058e-01 -1.34440804e+00
3.92012239e-01 4.44150925e-01 5.68015814e-01 -1.91140413e-01
1.75439954e-01 -2.33107597e-01 5.45232818e-02 -4.70257998e-01
-2.54735231e-01 -2.41199493e-01 -5.18835410e-02 3.02064896e-01
2.46382430e-01 -1.21512182e-01 1.35408249e-03 1.76069289e-01
1.46953785e+00 2.78691426e-02 6.12535417e-01 -6.65152133e-01
1.75745964e-01 4.90042046e-02 4.73236829e-01 1.39090729e+00
-7.07523373e-04 1.37238413e-01 9.75987315e-01 2.98850182e-02
-7.47534990e-01 -1.30722487e+00 -1.12454355e-01 8.10345173e-01
-4.34724987e-01 -8.58898908e-02 -4.30210412e-01 -7.05903113e-01
1.69883415e-01 8.45302582e-01 -1.05421889e+00 -2.23422036e-01
-5.55789471e-01 -1.43325317e+00 1.44732773e-01 9.87180322e-02
-1.76061243e-01 -8.55122864e-01 -6.51551306e-01 2.92454869e-01
3.69670361e-01 -4.28661704e-01 -4.39684600e-01 3.45393628e-01
-1.03005183e+00 -1.51276743e+00 -3.18691552e-01 -4.06244583e-02
6.45098925e-01 -4.31518763e-01 1.25938058e+00 -3.68032932e-01
-3.55218589e-01 5.41194320e-01 -2.19179019e-02 -5.48049688e-01
-3.32437158e-01 -3.91206175e-01 1.18427863e-02 -1.52037904e-01
-2.39672869e-01 -6.64690435e-01 -6.59391820e-01 -5.93513399e-02
-5.85771978e-01 -3.11945468e-01 3.88141155e-01 1.09873700e+00
4.88432497e-01 5.55815473e-02 6.01206124e-01 -1.03167975e+00
9.00233746e-01 -8.71368587e-01 -1.20467496e+00 3.68384689e-01
-1.00992346e+00 2.75437653e-01 1.41819507e-01 -6.50887370e-01
-1.15384042e+00 -9.04993862e-02 5.81842184e-01 -1.54845744e-01
4.74180467e-02 9.17933941e-01 -2.12031603e-01 3.52546364e-01
6.48881972e-01 -5.53638816e-01 -2.30206713e-01 -3.92260760e-01
8.35542977e-01 1.97384864e-01 3.75806391e-01 -8.31074655e-01
3.79399776e-01 4.22361076e-01 5.96237302e-01 -2.97727287e-01
-6.40565336e-01 -1.65869012e-01 -3.96157175e-01 -2.20116645e-01
9.13144469e-01 -4.92528051e-01 -1.35543203e+00 -2.60896564e-01
-9.98527527e-01 -5.19473195e-01 -3.39137286e-01 8.56725276e-01
-5.86203098e-01 -9.19270702e-03 -1.11557439e-01 -1.15292609e+00
-1.29504755e-01 -1.08514392e+00 8.06181550e-01 -4.21611704e-02
-4.13854718e-01 -1.26055312e+00 4.72639412e-01 -3.62917371e-02
-1.32605270e-01 7.14678705e-01 1.17423701e+00 -3.50899845e-01
-5.52222252e-01 -1.20930225e-01 -1.96081087e-01 -3.11139017e-01
1.99541762e-01 1.17779523e-01 -5.15401185e-01 -2.00793311e-01
-3.66676301e-02 2.64304221e-01 5.79650164e-01 1.42957187e+00
8.90126646e-01 -4.94940162e-01 -7.83500195e-01 3.71861577e-01
1.41419590e+00 4.89438266e-01 1.06970251e-01 -9.13686901e-02
6.32970154e-01 7.81638026e-01 4.12575364e-01 5.11973083e-01
2.09143475e-01 6.72238350e-01 5.30898094e-01 -1.25270234e-02
1.69114638e-02 -5.11026442e-01 4.87081148e-02 6.75224066e-02
-2.04989001e-01 -3.05169374e-01 -1.02356005e+00 6.49936676e-01
-2.16631842e+00 -8.23271453e-01 -6.13083720e-01 2.47534347e+00
1.01827383e+00 1.79180712e-01 3.87667209e-01 -3.20049316e-01
8.24744463e-01 -3.20035458e-01 -5.19391477e-01 -2.66215116e-01
3.48848216e-02 5.08443534e-01 1.06678069e+00 1.01911664e+00
-8.98343563e-01 4.92299616e-01 8.12637997e+00 6.94983840e-01
-4.03772831e-01 4.24052984e-01 6.84784889e-01 -1.49759948e-01
-3.41901124e-01 3.30885112e-01 -4.28549320e-01 3.22983414e-01
1.30649686e+00 -3.62982541e-01 6.56789541e-01 1.76464260e-01
1.03916156e+00 -2.38140985e-01 -1.32818902e+00 2.29399130e-01
-6.24231875e-01 -1.36029088e+00 -3.40616435e-01 3.20056289e-01
1.06632519e+00 -9.66275409e-02 -1.84516326e-01 -2.83829998e-02
1.47302020e+00 -1.32059383e+00 5.23922443e-01 9.60467815e-01
4.99236643e-01 -5.94769001e-01 4.15818036e-01 2.77393550e-01
-5.63419223e-01 -3.76278400e-01 4.47753221e-02 -1.23757953e-02
4.46324259e-01 7.82022893e-01 -9.18041825e-01 7.48337746e-01
5.09096324e-01 5.32010496e-01 -1.67862996e-01 1.11965477e+00
-5.94614267e-01 1.15571356e+00 -4.28002626e-01 -2.34525483e-02
6.17829822e-02 -5.81788659e-01 8.93233240e-01 1.06560659e+00
1.41067252e-01 2.41210625e-01 2.21911192e-01 1.09991181e+00
3.09953153e-01 -6.43544048e-02 -6.77406311e-01 -4.32482921e-02
6.06814981e-01 6.60224915e-01 -5.56906879e-01 -1.99476480e-01
-8.77621695e-02 2.66245455e-01 1.55105308e-01 6.65164948e-01
-8.63273025e-01 3.13398778e-01 1.19108923e-01 -1.31197289e-01
-5.22655025e-02 -5.22392169e-02 -5.39786339e-01 -5.43232620e-01
-5.65702379e-01 -9.23794806e-01 1.04193211e+00 -5.26712477e-01
-1.03316092e+00 -1.52927652e-01 7.74364233e-01 -3.31693619e-01
-1.62082583e-01 -3.73210102e-01 -4.42852467e-01 1.21455944e+00
-7.42981434e-01 -9.63035762e-01 2.97878832e-01 4.73452419e-01
3.58683109e-01 1.44473478e-01 5.17028987e-01 -4.52901870e-02
-8.34064364e-01 -5.64807467e-02 4.04236093e-03 -6.93674386e-01
4.38884050e-01 -1.61630464e+00 -1.88292459e-01 7.59991646e-01
-2.95567960e-01 6.12707794e-01 1.12503958e+00 -1.15164089e+00
-1.45229971e+00 -8.49152267e-01 5.50170898e-01 -5.72347403e-01
1.02185094e+00 -2.42016435e-01 -4.97351915e-01 6.11003518e-01
2.69203991e-01 -3.94274235e-01 3.01438838e-01 6.35141790e-01
1.08775720e-01 1.13748081e-01 -1.15747762e+00 9.82703030e-01
1.20014882e+00 -1.63933545e-01 -6.25484824e-01 6.09269142e-01
9.35513675e-01 -7.68304393e-02 -1.08685112e+00 5.34239590e-01
1.61508322e-01 -3.09411824e-01 1.14967489e+00 -1.15556538e+00
3.26309949e-01 -2.63155550e-02 1.26569986e-01 -1.68441892e+00
-4.07588392e-01 -1.15518177e+00 -2.80022472e-01 1.04929161e+00
5.20640075e-01 -6.36042953e-01 4.90742028e-01 5.61242461e-01
1.38139725e-01 -4.41955507e-01 -1.08037019e+00 -5.19302368e-01
-4.74932007e-02 -4.88143057e-01 4.83583510e-01 9.18603718e-01
3.29160653e-02 4.72158611e-01 -5.30999959e-01 6.32164896e-01
1.05133557e+00 -1.28638715e-01 4.71816212e-01 -1.17697561e+00
-6.34796143e-01 -5.38346589e-01 7.59025663e-02 -2.58459568e-01
1.38297439e-01 -6.61497474e-01 -4.31922451e-02 -1.37036884e+00
5.11330128e-01 -3.56635451e-01 -1.18325964e-01 2.96369731e-01
-7.92314112e-01 -5.97099125e-01 -2.38212317e-01 -3.08251232e-01
-5.92486635e-02 5.51468849e-01 1.08482695e+00 -1.31666213e-01
-5.63776791e-01 1.53707564e-01 -5.34748912e-01 5.90691745e-01
4.38285321e-01 -1.01990652e+00 -7.09908485e-01 1.28987983e-01
3.73951018e-01 7.50695050e-01 6.39538765e-01 -1.84681922e-01
9.55449417e-02 -8.35298240e-01 -1.25839889e-01 -3.80649000e-01
-2.26287246e-01 -6.76167607e-01 8.03296030e-01 7.92135060e-01
-5.61064482e-01 2.13765334e-02 1.82039976e-01 1.14444542e+00
2.67455995e-01 -3.33026439e-01 5.33326328e-01 2.04131931e-01
5.56895658e-02 1.91647068e-01 -3.54846299e-01 -6.27861768e-02
9.22230721e-01 3.21105391e-01 -3.29240739e-01 -3.31819177e-01
-1.15078223e+00 5.54342091e-01 -5.42010628e-02 1.09043814e-01
3.36776614e-01 -1.10608351e+00 -9.99046326e-01 -4.48602289e-01
-2.75266945e-01 -3.45903546e-01 7.52862170e-02 1.11634946e+00
-4.59089465e-02 4.39830124e-01 3.04384589e-01 -5.09883285e-01
-1.12104154e+00 1.13873827e+00 3.70319307e-01 -6.18326724e-01
-3.73487443e-01 6.50462568e-01 2.98513114e-01 -3.84938449e-01
7.06014484e-02 -3.99228297e-02 -2.04956412e-01 -2.44035982e-02
5.83534092e-02 7.89283276e-01 -3.77866596e-01 9.95746553e-02
-3.06223273e-01 6.52385950e-02 5.61377108e-01 -4.97045219e-01
1.55738103e+00 -1.32670775e-01 -3.69094461e-01 1.62966505e-01
6.38519585e-01 -5.29783405e-02 -1.42153442e+00 -5.56182154e-02
5.62924504e-01 -3.89196157e-01 3.80017191e-01 -1.09041190e+00
-5.69650471e-01 2.66464531e-01 6.40449584e-01 3.64521891e-01
1.18650222e+00 2.24346016e-02 -5.46134472e-01 -2.80388445e-01
8.99388492e-02 -7.18419075e-01 -9.73648131e-02 -1.65843382e-01
1.09581554e+00 -9.53993797e-01 3.59179974e-01 -6.06331825e-01
-2.52731949e-01 4.75616425e-01 -1.66330144e-01 -3.84583294e-01
9.80231285e-01 2.85070390e-01 -6.70217574e-01 -6.37177050e-01
-9.19493675e-01 -1.25490069e-01 6.99389338e-01 7.96256006e-01
1.13383226e-01 4.98461574e-01 -9.82516110e-01 5.00614643e-01
2.03020811e-01 3.15543041e-02 3.07768732e-01 5.69360197e-01
5.38258925e-02 -1.02571678e+00 -7.41550207e-01 7.08026409e-01
-5.00424981e-01 -3.66053790e-01 -1.52258009e-01 8.92797053e-01
-1.26597911e-01 1.26910973e+00 -2.85692900e-01 2.31483325e-01
6.16719127e-01 -2.18143061e-01 4.59498614e-01 -6.02545857e-01
-4.36913520e-01 5.15832126e-01 4.99040544e-01 -6.10888481e-01
-3.76957238e-01 -1.09007311e+00 -5.97801328e-01 -1.39119372e-01
-4.68318313e-01 1.52445450e-01 2.96094388e-01 9.79373097e-01
2.79319525e-01 9.52475488e-01 6.43370628e-01 -5.19918501e-01
-6.63938701e-01 -9.40369010e-01 -3.85863572e-01 -1.75021701e-02
2.92266488e-01 -8.97473633e-01 -2.03387812e-01 1.77181214e-01] | [7.79578161239624, 5.303860664367676] |
709c051c-5309-4e30-b281-565b2465e789 | knowledge-graph-for-nlg-in-the-context-of | 2307.01548 | null | https://arxiv.org/abs/2307.01548v1 | https://arxiv.org/pdf/2307.01548v1.pdf | Knowledge Graph for NLG in the context of conversational agents | The use of knowledge graphs (KGs) enhances the accuracy and comprehensiveness of the responses provided by a conversational agent. While generating answers during conversations consists in generating text from these KGs, it is still regarded as a challenging task that has gained significant attention in recent years. In this document, we provide a review of different architectures used for knowledge graph-to-text generation including: Graph Neural Networks, the Graph Transformer, and linearization with seq2seq models. We discuss the advantages and limitations of each architecture and conclude that the choice of architecture will depend on the specific requirements of the task at hand. We also highlight the importance of considering constraints such as execution time and model validity, particularly in the context of conversational agents. Based on these constraints and the availability of labeled data for the domains of DAVI, we choose to use seq2seq Transformer-based models (PLMs) for the Knowledge Graph-to-Text Generation task. We aim to refine benchmark datasets of kg-to-text generation on PLMs and to explore the emotional and multilingual dimensions in our future work. Overall, this review provides insights into the different approaches for knowledge graph-to-text generation and outlines future directions for research in this area. | ['Christophe Cruz', 'Massinissa Atmani', 'Hussam Ghanem'] | 2023-07-04 | null | null | null | null | ['knowledge-graphs', 'text-generation', 'kg-to-text'] | ['knowledge-base', 'natural-language-processing', 'natural-language-processing'] | [ 2.08723232e-01 8.48166108e-01 2.58731954e-02 -4.41462249e-01
-6.52343690e-01 -5.25703490e-01 8.08293879e-01 1.27775326e-01
-1.26147643e-01 1.08922803e+00 6.18436337e-01 -2.02177301e-01
-1.35175869e-01 -8.80159795e-01 -3.50375265e-01 -3.19861382e-01
1.00373410e-01 1.04319477e+00 -4.89678413e-01 -6.95218801e-01
1.07529342e-01 2.35662431e-01 -1.42220557e+00 6.63482249e-01
9.85755622e-01 6.99869871e-01 -1.76385492e-02 7.79667139e-01
-4.16965216e-01 1.23311877e+00 -1.10713768e+00 -9.79700804e-01
-1.95629120e-01 -7.98985600e-01 -1.25670528e+00 -1.32005349e-01
9.38645303e-02 -1.98277101e-01 -7.29899704e-02 7.00927377e-01
8.83581161e-01 6.15668833e-01 5.77422917e-01 -1.44968748e+00
-9.66261387e-01 1.14921510e+00 3.42661679e-01 -9.20534879e-02
8.28080416e-01 1.31081209e-01 1.02585936e+00 -6.06382668e-01
1.01406002e+00 1.38769674e+00 4.91400748e-01 9.34468389e-01
-8.47926557e-01 -3.72756988e-01 2.86210567e-01 4.65325683e-01
-1.09256005e+00 -5.07299304e-01 8.21841180e-01 -2.35346764e-01
1.65912855e+00 2.94209957e-01 7.79676139e-01 1.61685491e+00
4.85285670e-02 8.37389886e-01 1.00811148e+00 -4.58486170e-01
7.46958107e-02 2.92446554e-01 -1.73120424e-01 3.98682982e-01
-1.94747820e-01 9.68172327e-02 -1.04167533e+00 -2.11498737e-01
3.03002805e-01 -9.30980086e-01 -1.96742490e-01 -9.28614587e-02
-1.16953611e+00 1.32344234e+00 2.14614719e-01 2.83545762e-01
-3.21619272e-01 -1.03608742e-01 6.26284540e-01 4.68475640e-01
6.22995138e-01 7.50239432e-01 -2.83693731e-01 -4.78360713e-01
-6.92148685e-01 7.80239284e-01 1.29839742e+00 1.05548203e+00
4.63455886e-01 4.80971694e-01 -5.10319591e-01 9.60581422e-01
3.13671194e-02 3.60709757e-01 4.72854197e-01 -7.68670738e-01
6.11975312e-01 5.68975806e-01 5.89691624e-02 -8.93082261e-01
-6.21746242e-01 -3.40709299e-01 -6.20521128e-01 -1.11508362e-01
2.16634035e-01 -6.96792841e-01 -5.79566598e-01 1.80664110e+00
1.82261094e-01 -3.72391611e-01 5.93918324e-01 7.54691124e-01
1.52951312e+00 6.00275755e-01 8.27202424e-02 -2.84301460e-01
1.14330423e+00 -9.63593006e-01 -1.05738020e+00 -3.00447822e-01
7.75370598e-01 -5.97164512e-01 7.53408194e-01 4.51653525e-02
-1.25091994e+00 -3.05694520e-01 -6.27086520e-01 -1.09649219e-01
-6.24257505e-01 -1.35980379e-02 7.48165905e-01 5.66742539e-01
-1.46994293e+00 2.90981382e-01 -2.75658876e-01 -6.95265710e-01
-1.89589009e-01 3.66161346e-01 -1.79322302e-01 1.60185546e-01
-1.86772799e+00 1.37064016e+00 5.14684677e-01 2.62899280e-01
-4.66481894e-01 -3.55859578e-01 -1.06324804e+00 -1.55890569e-01
2.43090391e-01 -1.07046425e+00 1.47530293e+00 -8.90977442e-01
-1.91564274e+00 7.21616328e-01 -1.48607325e-02 -5.89485705e-01
4.79928553e-01 1.79292545e-01 -4.10162508e-01 -6.82594478e-02
-4.47671600e-02 9.37349260e-01 3.98396313e-01 -8.89743030e-01
-3.96139205e-01 6.30471334e-02 5.23548424e-01 6.16664231e-01
4.96210530e-02 6.80644736e-02 4.38781604e-02 -5.82298458e-01
-6.41875625e-01 -1.08717752e+00 2.10359003e-02 -1.00704300e+00
-3.66482437e-01 -6.83925450e-01 3.77758145e-01 -8.17013443e-01
1.06261063e+00 -1.47890580e+00 3.51975560e-01 5.39919063e-02
-7.26055279e-02 1.91568255e-01 -3.55153084e-01 1.10593534e+00
4.34561968e-02 5.23635149e-02 1.31061792e-01 -2.90177524e-01
7.76769221e-02 1.39743999e-01 -3.30252409e-01 -3.52637708e-01
3.51588160e-01 1.45418990e+00 -1.04207230e+00 -3.90761614e-01
1.30583793e-01 5.41978657e-01 -4.51795161e-01 3.90528888e-01
-5.68015277e-01 4.52071548e-01 -4.19782728e-01 1.82098731e-01
8.91642794e-02 -5.27239069e-02 3.90947461e-01 -1.12060256e-01
6.76955879e-02 6.55279040e-01 -6.65060341e-01 1.64289033e+00
-5.69920659e-01 5.45380473e-01 -1.22661993e-01 -7.45355844e-01
1.23458922e+00 6.03924513e-01 8.24646875e-02 -6.49666905e-01
1.37365147e-01 1.41863272e-01 1.32492632e-01 -4.69827831e-01
9.65748310e-01 -5.17863989e-01 -2.38664895e-01 6.89147174e-01
2.13505000e-01 -4.33975428e-01 5.96333265e-01 4.74340528e-01
6.83948040e-01 1.07859030e-01 9.87780690e-02 4.02799547e-02
4.20049906e-01 1.16126493e-01 1.31987482e-01 5.13256848e-01
7.94481263e-02 1.74467042e-01 6.22695148e-01 -2.77717352e-01
-6.76983416e-01 -5.70093930e-01 5.17879248e-01 1.11506236e+00
-4.56475735e-01 -5.18355608e-01 -9.10240352e-01 -8.20518434e-01
-3.47575873e-01 1.33253443e+00 -6.76931977e-01 -2.70600647e-01
-5.63292623e-01 -7.75801539e-01 6.80197239e-01 5.07327020e-01
2.18395606e-01 -1.72478104e+00 -3.56622815e-01 4.04337019e-01
-9.33053374e-01 -1.31287146e+00 -2.03320220e-01 -1.91557348e-01
-5.36372483e-01 -8.14415991e-01 -5.68372667e-01 -7.46229768e-01
3.59528810e-01 -1.72409207e-01 1.42916369e+00 -2.25525424e-01
1.72559932e-01 7.06281006e-01 -8.50628793e-01 -5.03111780e-01
-9.20330167e-01 3.86764199e-01 -2.54434437e-01 -1.81868315e-01
3.03322762e-01 -3.70225132e-01 -2.41249204e-01 -4.49054092e-02
-6.78048253e-01 4.95634407e-01 3.80095392e-01 7.42221534e-01
-1.81385987e-02 -5.81920370e-02 1.24556124e+00 -8.45092535e-01
1.59711111e+00 -3.20014924e-01 -1.26168162e-01 3.66720587e-01
-4.28113878e-01 8.08626264e-02 5.46961546e-01 -1.19606862e-02
-1.39831269e+00 -4.12474155e-01 -2.64254272e-01 -1.12718800e-02
1.82611682e-02 8.41444731e-01 6.84986860e-02 -1.55864935e-02
6.92780733e-01 7.06356019e-02 1.87213406e-01 6.05776869e-02
7.62728870e-01 6.64962411e-01 2.19364509e-01 -5.66452324e-01
3.28204751e-01 -1.42060176e-01 -2.73329854e-01 -6.94901466e-01
-6.87842429e-01 1.18378527e-01 -1.80721655e-01 -6.17146015e-01
9.12573636e-01 -6.50188327e-01 -8.75210106e-01 3.37887079e-01
-1.38161647e+00 -7.46740699e-01 -3.58420610e-01 2.41244748e-01
-8.25523019e-01 1.80725843e-01 -6.66005313e-01 -9.59852636e-01
-8.58436763e-01 -1.01047063e+00 7.52233207e-01 2.28609949e-01
-7.16836512e-01 -1.39300668e+00 8.84930640e-02 9.08694327e-01
7.00062633e-01 4.78384227e-01 1.04541898e+00 -8.38619709e-01
-2.72772014e-01 -4.94902604e-04 1.03602827e-01 3.01187068e-01
5.34356087e-02 -2.15239629e-01 -8.91372323e-01 -1.14891633e-01
-1.74826786e-01 -9.33944821e-01 3.90907466e-01 1.58265740e-01
3.98290724e-01 -5.42063475e-01 6.61848336e-02 -1.11480411e-02
7.91259170e-01 2.56485969e-01 4.44176257e-01 1.17499702e-01
5.56933224e-01 1.14125347e+00 6.31136715e-01 3.85737330e-01
9.68222201e-01 7.53540874e-01 2.56883740e-01 2.69877434e-01
-3.96809846e-01 -3.66729349e-01 6.07807815e-01 1.01000452e+00
-1.58706129e-01 -7.69708157e-01 -6.98654056e-01 7.05071092e-01
-1.82736278e+00 -9.48763907e-01 -9.76271406e-02 1.81030846e+00
9.40714955e-01 -2.82692403e-01 2.63420165e-01 -2.23844454e-01
7.89932549e-01 2.54566938e-01 -3.42467666e-01 -1.06215525e+00
-2.57554024e-01 1.29390955e-01 -1.57036632e-01 9.23218369e-01
-4.78265822e-01 1.14861846e+00 6.55372810e+00 6.34702504e-01
-9.95105326e-01 -1.06824555e-01 5.34191668e-01 -1.62912175e-01
-4.65898395e-01 -2.98071772e-01 -7.78621078e-01 1.22038513e-01
1.15708232e+00 -4.72968102e-01 8.47647190e-01 5.07919431e-01
3.15429956e-01 1.99450046e-01 -1.15267122e+00 7.33941436e-01
4.46240962e-01 -1.11574805e+00 4.41419691e-01 -3.41925323e-01
8.10536444e-01 6.50094748e-02 -9.56149921e-02 6.52002454e-01
6.69209123e-01 -1.16834795e+00 6.34549260e-01 4.10405189e-01
6.08750224e-01 -8.33652973e-01 9.77097511e-01 2.25738123e-01
-8.22912812e-01 2.43353531e-01 4.35034782e-02 -2.25845948e-01
4.81953174e-01 3.49188358e-01 -1.48248434e+00 1.03803313e+00
2.69441485e-01 3.74420136e-01 -2.52197295e-01 3.72786313e-01
-5.32025456e-01 3.54264975e-01 2.97786929e-02 -4.62053090e-01
3.73066962e-01 -9.55028236e-02 4.18227911e-01 1.33819389e+00
4.14183766e-01 1.48072422e-01 5.07689975e-02 8.84216547e-01
2.43436024e-02 4.05732960e-01 -6.57020509e-01 -4.24758255e-01
3.04295778e-01 1.32058668e+00 -3.46442670e-01 -3.01595241e-01
-2.00313076e-01 8.35267603e-01 7.58045435e-01 4.64591742e-01
-5.08010745e-01 -5.00367105e-01 3.44476193e-01 -1.13890104e-01
8.53193551e-03 -7.57691488e-02 3.32046039e-02 -9.15190935e-01
-7.82563016e-02 -1.23182952e+00 5.12457311e-01 -1.17017019e+00
-1.34055412e+00 9.69276130e-01 2.70065099e-01 -6.65642142e-01
-1.13758910e+00 -3.86558473e-01 -5.62206686e-01 9.53952074e-01
-1.43107498e+00 -1.44848120e+00 -2.96128541e-01 4.52582836e-01
5.37116826e-01 -8.04048926e-02 1.19861722e+00 -1.35504529e-01
-2.30249360e-01 5.46266139e-01 -3.42335522e-01 2.32334863e-02
6.31077528e-01 -1.05848145e+00 6.43420100e-01 1.78624332e-01
-6.16986584e-03 4.79068846e-01 7.69929171e-01 -8.25411856e-01
-1.17988753e+00 -1.10962021e+00 1.53185391e+00 -5.08236229e-01
6.09346449e-01 -3.30850393e-01 -6.64906919e-01 8.57741237e-01
8.25351179e-01 -9.88911808e-01 9.47710812e-01 1.71190366e-01
6.67946115e-02 1.83219805e-01 -1.06760764e+00 7.31054544e-01
9.70298529e-01 -4.80349034e-01 -5.08890629e-01 6.20371819e-01
6.84718847e-01 -5.75987399e-01 -9.43025291e-01 2.18696415e-01
2.55322814e-01 -9.73983765e-01 4.94084597e-01 -7.26408720e-01
2.92195886e-01 5.64073622e-02 3.19706321e-01 -1.94519794e+00
-1.68685004e-01 -1.08303022e+00 9.87647399e-02 1.44784427e+00
5.61265826e-01 -7.25208521e-01 5.98434687e-01 7.42847979e-01
-3.12412649e-01 -7.48329818e-01 -8.10703635e-01 -5.36208868e-01
6.21966012e-02 -4.32083696e-01 7.21808910e-01 9.55640495e-01
6.39531136e-01 8.60015988e-01 -6.27634168e-01 -4.43095595e-01
1.51223481e-01 1.07814513e-01 8.03101718e-01 -9.47632134e-01
-1.41144902e-01 -3.45048696e-01 -5.88772558e-02 -6.36033595e-01
6.48025274e-01 -1.28485525e+00 -1.71086982e-01 -2.13294005e+00
-1.19711109e-01 -3.95370200e-02 3.09544295e-01 5.64241648e-01
-1.64485961e-01 8.56316835e-02 2.91794926e-01 -4.12438273e-01
-4.34329182e-01 7.99282014e-01 1.31222582e+00 -1.01755083e-01
-3.11222643e-01 -1.37419224e-01 -9.52842593e-01 3.14664543e-01
1.18906534e+00 -3.55071314e-02 -8.92131269e-01 -4.06670928e-01
6.75521255e-01 3.08327019e-01 1.43577695e-01 -4.97495800e-01
6.55491650e-02 -1.86488003e-01 -9.72563103e-02 -3.98283303e-01
5.86215913e-01 -1.37672260e-01 2.68874943e-01 2.56484091e-01
-6.25197649e-01 2.97421426e-01 4.59657609e-01 2.00328141e-01
-2.65940309e-01 -2.92625576e-01 3.38258594e-01 -2.76950926e-01
-5.08691430e-01 -1.61366969e-01 -7.75873423e-01 3.42404664e-01
8.47349584e-01 -9.55917686e-03 -3.71813536e-01 -1.24201536e+00
-7.90248156e-01 4.19840187e-01 -1.12882098e-02 7.97461271e-01
6.94057345e-01 -1.41030538e+00 -1.11621630e+00 -8.35343301e-02
2.71824300e-01 -3.61386627e-01 5.17702878e-01 5.10005474e-01
-2.17043996e-01 8.90283942e-01 -2.26977780e-01 5.70851974e-02
-1.32770729e+00 3.08090895e-01 3.02723974e-01 -6.90300286e-01
-2.09319293e-01 7.86860347e-01 5.10084592e-02 -7.19103992e-01
-4.00623158e-02 -1.18913047e-01 -7.02208459e-01 2.53453374e-01
2.70804435e-01 4.74229872e-01 3.20399165e-01 -6.94311380e-01
-1.38917461e-01 7.04597756e-02 -2.00028762e-01 -2.85325736e-01
1.02473736e+00 -3.92735340e-02 -2.54044801e-01 3.00261766e-01
6.56232417e-01 -2.60058820e-01 -5.15088022e-01 -4.10302766e-02
-2.45108694e-01 1.61095560e-01 -2.66616553e-01 -1.27669692e+00
-9.98621225e-01 6.48983061e-01 -1.85427200e-02 2.96938837e-01
7.51335919e-01 -4.14574072e-02 7.54038215e-01 4.05920535e-01
3.01728308e-01 -1.35313964e+00 1.83723629e-01 9.24096823e-01
1.52669191e+00 -7.89410412e-01 -3.08790684e-01 -3.53274494e-01
-1.15399706e+00 1.06848121e+00 6.84448004e-01 4.10900980e-01
-1.20344847e-01 -4.89311107e-02 3.30311388e-01 -2.25789726e-01
-1.17042506e+00 -1.64662749e-01 2.11756483e-01 1.01855159e+00
6.79951429e-01 1.95985690e-01 -4.14180517e-01 4.80205446e-01
-1.03984296e+00 -4.11902554e-02 6.37863398e-01 6.13888323e-01
-1.82092171e-02 -1.34881520e+00 -1.52199894e-01 4.50740308e-01
-1.99877903e-01 -2.36292481e-01 -1.12870860e+00 6.37153566e-01
-2.84619153e-01 1.59811473e+00 -3.07150304e-01 -4.86930788e-01
5.00458241e-01 4.64874327e-01 4.90497351e-01 -5.89181900e-01
-1.13289309e+00 -4.84899968e-01 1.11014450e+00 -1.67110234e-01
-5.04596889e-01 -5.54482400e-01 -1.08187890e+00 -4.48580384e-01
-3.65640134e-01 5.24749339e-01 6.52864039e-01 7.47147501e-01
5.76478660e-01 6.20842397e-01 1.22834638e-01 -7.36333013e-01
-4.41968769e-01 -1.32044506e+00 -3.24437320e-01 4.07462001e-01
-2.26495072e-01 -4.59473580e-01 -1.44202292e-01 -5.06109744e-02] | [12.440619468688965, 8.353795051574707] |
3abd8811-c624-4e30-95f3-6dc2ddbe57ed | cross-lingual-information-retrieval-and | null | null | https://aclanthology.org/R13-1063 | https://aclanthology.org/R13-1063.pdf | Cross-Lingual Information Retrieval and Semantic Interoperability for Cultural Heritage Repositories | null | ['Maria Pia di Buono', 'Federica Marano', 'Johanna Monti', 'Mario Monteleone'] | 2013-09-01 | cross-lingual-information-retrieval-and-1 | https://aclanthology.org/R13-1063 | https://aclanthology.org/R13-1063.pdf | ranlp-2013-9 | ['cross-lingual-information-retrieval'] | ['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] | [-7.383035659790039, 3.59415340423584] |
fcd905a5-086e-458f-a02d-a65bed1dcc84 | cdnet-a-cascaded-decoupling-architecture-for | null | null | https://openreview.net/forum?id=DmKu5T2gEqc | https://openreview.net/pdf?id=DmKu5T2gEqc | CDNet: A cascaded decoupling architecture for video prediction | Video prediction is an essential task in the computer vision community, helping to solve many downstream vision tasks by predicting and modeling future motion dynamics and appearance. In the deterministic video prediction task, current methods mainly employ variants of stacked Recurrent Neural Networks (RNN) to capture spatiotemporal coherence, overlooking the conflict between long-term motion dynamics modeling and legible appearance generation. In this work, we propose a Cascaded Decoupling Network (CDNet) to solve the video prediction problem through two modules: motion LSTM to capture the motion trend and variation in the temporal highway without considering the appearance details, and refine LSTM to recover the detailed appearance according to the predicted motion dynamics and historical appearance iteratively. The cascaded structure provides a preliminary solution for the above conflict. We verify the rationality of our model on two real-world challenging video prediction datasets and yield state-of-the-art performance. | ['Mingtao Pei', 'Chuanqi Zang'] | 2021-09-29 | null | null | null | null | ['video-prediction'] | ['computer-vision'] | [ 4.53069881e-02 -2.55623937e-01 -2.69794196e-01 -2.35547498e-01
4.34590541e-02 -1.50065482e-01 5.75438201e-01 -7.34415412e-01
-4.02701348e-02 4.16965157e-01 4.05831605e-01 -1.16507195e-01
1.05951414e-01 -4.13907021e-01 -7.85693467e-01 -7.90579855e-01
1.64852422e-02 -2.04284966e-01 5.53527594e-01 -1.09469846e-01
2.92866994e-02 2.96213061e-01 -1.43423259e+00 4.62615699e-01
5.83412886e-01 1.00033271e+00 5.89787900e-01 8.13059092e-01
4.33738232e-02 1.66569614e+00 -4.07561101e-02 -3.30048889e-01
2.14244500e-02 -3.76941204e-01 -6.49291575e-01 3.93324912e-01
5.13574898e-01 -6.87658727e-01 -1.01263869e+00 8.79221678e-01
4.52607051e-02 1.89055115e-01 4.41327512e-01 -1.19822049e+00
-8.36430192e-01 2.89703608e-01 -6.83987260e-01 4.32708055e-01
6.41148388e-02 3.85858417e-01 8.59601676e-01 -7.44442165e-01
7.83379138e-01 1.07767391e+00 6.10749066e-01 8.46762538e-01
-1.03899693e+00 -3.66344869e-01 8.23274851e-01 8.29819083e-01
-1.03207886e+00 -5.40376008e-01 8.47430885e-01 -6.44392371e-01
8.48629773e-01 8.02318975e-02 8.59297574e-01 1.52168143e+00
3.71380419e-01 1.14437521e+00 2.94142902e-01 6.49696514e-02
-2.33529508e-01 -2.72716492e-01 -6.81469515e-02 7.97459841e-01
-2.95161188e-01 1.07014649e-01 -4.45830017e-01 3.08211416e-01
1.05990016e+00 3.79953980e-01 -4.00754094e-01 -1.79216355e-01
-1.16503394e+00 2.52041787e-01 4.13718402e-01 2.53424287e-01
-6.77740455e-01 4.57320988e-01 1.80578053e-01 1.77829832e-01
7.22935319e-01 -1.68303400e-01 -5.40766895e-01 -5.89721575e-02
-1.07501268e+00 1.68918461e-01 3.04812253e-01 7.89273918e-01
5.49183249e-01 5.23563087e-01 -3.14767748e-01 7.85737813e-01
4.35746014e-01 3.34415317e-01 5.47706127e-01 -1.16811550e+00
3.59563142e-01 2.56450415e-01 3.59555721e-01 -1.23534834e+00
-2.63186872e-01 -2.93041348e-01 -1.20177674e+00 -7.05111548e-02
3.90066534e-01 -1.59185722e-01 -9.42712545e-01 1.78628182e+00
1.62333757e-01 9.05782282e-01 -1.91085353e-01 1.11183894e+00
8.47009420e-01 1.13918996e+00 2.47719601e-01 -5.01995981e-01
8.89638543e-01 -1.59762859e+00 -6.87609851e-01 -1.55643329e-01
3.82857770e-01 -5.11130214e-01 6.50940180e-01 2.18468860e-01
-1.18667054e+00 -9.28960204e-01 -6.92339599e-01 -7.81357810e-02
2.97525853e-01 3.87018651e-01 4.28160101e-01 -6.73479289e-02
-1.43806708e+00 7.64209330e-01 -9.67966080e-01 -3.19642544e-01
1.19085260e-01 1.93451688e-01 -1.55166119e-01 -4.35472727e-02
-1.04746592e+00 6.07146859e-01 5.39601408e-02 7.37823308e-01
-1.03681827e+00 -4.54456955e-01 -5.77468693e-01 -1.39777353e-02
1.54992133e-01 -1.13375986e+00 1.15036511e+00 -1.59107518e+00
-1.63337886e+00 4.87333208e-01 -6.96009934e-01 -6.02275610e-01
7.00007439e-01 -3.48984063e-01 -5.40125430e-01 -1.13913342e-01
-3.25986713e-01 8.26569796e-01 1.09220409e+00 -1.13247991e+00
-7.71789372e-01 -4.70562343e-04 -3.53856653e-01 1.66578904e-01
-3.93860132e-01 -1.61590680e-01 -9.26074982e-01 -9.67976332e-01
-1.73622165e-02 -9.77546215e-01 -5.56280613e-01 -1.28805250e-01
-2.24173099e-01 -2.21195355e-01 1.06456196e+00 -1.00472367e+00
1.50183988e+00 -2.06261468e+00 4.60624129e-01 -3.92954826e-01
3.22358489e-01 2.94532746e-01 -4.79484439e-01 -9.33861136e-02
-5.02746291e-02 -2.38899678e-01 1.38561562e-01 -5.30786633e-01
-3.31519574e-01 1.27781153e-01 -6.60797000e-01 2.69197792e-01
2.71029681e-01 1.17058969e+00 -9.00973678e-01 -3.55064899e-01
2.87880659e-01 4.85102057e-01 -3.28348994e-01 4.96197313e-01
-5.62362373e-01 6.54725850e-01 -4.38147783e-01 3.98899853e-01
4.06149596e-01 -5.73459625e-01 2.89572507e-01 -3.65065753e-01
-1.97851360e-01 -8.49355757e-02 -7.48542726e-01 1.63193631e+00
-2.10701033e-01 8.99661422e-01 -1.26033649e-01 -9.02404249e-01
6.89644694e-01 3.59110832e-01 7.46650040e-01 -7.28737950e-01
-1.49718434e-01 -2.25266874e-01 -9.40315649e-02 -9.27821636e-01
8.27684283e-01 2.78481711e-02 4.49084073e-01 2.40236819e-01
-6.30172715e-02 6.97245359e-01 -1.00536846e-01 -1.75282098e-02
9.30420160e-01 6.35194421e-01 -1.36601478e-01 1.97768018e-01
5.47176003e-01 -1.28237933e-01 8.88760805e-01 3.86670172e-01
-3.74357849e-01 7.70527482e-01 2.52153575e-01 -1.03260672e+00
-1.22114742e+00 -9.29631054e-01 5.24590671e-01 1.10200477e+00
3.86881173e-01 -2.42399290e-01 -4.08003271e-01 -4.83129323e-01
-5.24248242e-01 4.52231646e-01 -6.82872355e-01 -2.29407907e-01
-8.48077893e-01 -8.08428228e-01 1.32484334e-02 6.23601377e-01
3.67703348e-01 -1.22018957e+00 -3.70962828e-01 4.33577120e-01
-6.67958081e-01 -1.32988048e+00 -6.16972327e-01 -5.51101983e-01
-8.82122219e-01 -8.20229888e-01 -8.70237470e-01 -9.41232026e-01
3.99181485e-01 6.38283253e-01 1.04217255e+00 2.13461682e-01
-1.99204348e-02 2.01727331e-01 -2.67940342e-01 2.56045580e-01
-3.06566209e-01 -2.06344411e-01 -3.90591845e-02 4.58809108e-01
-5.69803081e-02 -6.99982166e-01 -9.64581013e-01 2.75400966e-01
-7.40582526e-01 7.34669268e-01 5.27243316e-01 7.87330866e-01
4.65784371e-01 -1.28346726e-01 1.84421405e-01 -5.25288522e-01
8.67773518e-02 -7.54120231e-01 -5.19510090e-01 5.54874063e-01
-2.92364269e-01 -4.88673113e-02 7.66153872e-01 -6.92656159e-01
-1.46684158e+00 3.35204393e-01 -6.63457736e-02 -1.06780636e+00
-1.26974076e-01 1.46658853e-01 4.68975268e-02 2.04345241e-01
-1.21622011e-01 8.02466094e-01 -1.58304825e-01 -3.82433176e-01
3.73512864e-01 7.69691765e-02 7.43179560e-01 -6.61760569e-02
4.61346596e-01 5.81304729e-01 -1.42649233e-01 -6.43508375e-01
-8.31056476e-01 -2.14648321e-01 -7.13866830e-01 -6.42237544e-01
1.07407951e+00 -1.20803714e+00 -6.45034850e-01 7.53301620e-01
-1.46945381e+00 -6.72119796e-01 1.49266362e-01 2.96634912e-01
-5.83335161e-01 4.18536484e-01 -1.08833277e+00 -7.61024714e-01
-3.47868413e-01 -1.10613179e+00 7.30035186e-01 1.63463444e-01
-1.43542662e-02 -1.11483729e+00 1.72451705e-01 1.76389381e-01
3.00174743e-01 4.12954725e-02 7.56911695e-01 2.46428803e-01
-1.00636387e+00 1.32996857e-01 -3.60223442e-01 1.18270598e-01
-3.90779935e-02 5.47835052e-01 -7.71088004e-01 -4.31883708e-02
2.77831480e-02 1.69791937e-01 1.22458792e+00 8.86457801e-01
1.31446350e+00 -4.24262494e-01 -2.37911329e-01 8.66413713e-01
1.24174607e+00 2.00787351e-01 7.73697674e-01 3.11149597e-01
1.24153244e+00 6.99948490e-01 4.11353856e-01 5.42264223e-01
7.40201652e-01 8.22537124e-01 4.35390323e-01 -8.33447501e-02
-3.67697090e-01 -5.43807685e-01 6.71285689e-01 1.14372444e+00
-3.86646628e-01 -4.02700633e-01 -6.74289405e-01 5.22188365e-01
-2.45320010e+00 -1.48942387e+00 -4.62419361e-01 1.82667530e+00
2.95693606e-01 -8.09575766e-02 2.28161618e-01 -5.86326361e-01
8.17573428e-01 5.19847751e-01 -6.44555688e-01 3.00672594e-02
-2.89244771e-01 -6.73235357e-01 2.36614287e-01 4.35271919e-01
-1.23125613e+00 1.30393934e+00 6.30466175e+00 7.05699742e-01
-1.38853312e+00 6.65188283e-02 1.19612050e+00 -2.50110656e-01
-1.89881474e-01 -7.74647370e-02 -6.03041351e-01 7.18506634e-01
7.77716100e-01 6.62446693e-02 3.96034271e-01 5.35156012e-01
8.51861596e-01 1.91501170e-01 -8.62510920e-01 1.07693779e+00
-6.31088391e-03 -1.66859496e+00 4.78016049e-01 -1.39246941e-01
9.69741583e-01 1.71067476e-01 2.57992119e-01 7.43678063e-02
3.33278269e-01 -9.55487192e-01 1.01296568e+00 1.16427100e+00
4.11010921e-01 -2.75511801e-01 4.18696821e-01 4.63854909e-01
-1.47316289e+00 -2.81466275e-01 -3.17601740e-01 -2.44859308e-01
6.38139307e-01 3.32020521e-01 -2.38408610e-01 3.95934045e-01
7.23939002e-01 1.58066797e+00 -4.74994689e-01 9.75245118e-01
-5.97554967e-02 6.36782110e-01 1.46581426e-01 2.42115334e-01
3.93858194e-01 -4.96626884e-01 5.39201140e-01 1.01059330e+00
4.54681307e-01 9.23975110e-02 -2.58004777e-02 8.11358809e-01
2.56993063e-02 -2.55531758e-01 -3.87273610e-01 1.37180418e-01
9.27905291e-02 1.20235872e+00 -5.76896846e-01 -5.12990832e-01
-5.64477146e-01 1.29337227e+00 4.93036807e-01 7.99434721e-01
-1.01873934e+00 6.10620022e-01 8.63994002e-01 -8.59739706e-02
6.14191413e-01 -3.59820247e-01 -2.54132622e-03 -1.54687393e+00
1.67393014e-01 -4.78700131e-01 2.17719734e-01 -8.63904715e-01
-1.23272634e+00 7.12938190e-01 -4.90460485e-01 -1.64588511e+00
-3.95177960e-01 -4.22026485e-01 -9.72023308e-01 4.84608740e-01
-1.24061775e+00 -1.28199947e+00 -3.00259054e-01 6.32461369e-01
1.07790029e+00 1.94801409e-02 2.61069179e-01 3.80139649e-01
-1.04942691e+00 1.64568156e-01 7.41880834e-02 -2.44744215e-02
5.41589081e-01 -7.45930910e-01 6.30648375e-01 1.13733840e+00
2.80825198e-01 2.27000251e-01 7.76359320e-01 -6.21035278e-01
-1.34130180e+00 -1.54996121e+00 7.80250907e-01 -3.90325367e-01
8.17981541e-01 2.31186245e-02 -9.64059830e-01 7.31851578e-01
1.22059762e-01 1.73337102e-01 2.09537402e-01 -2.54188716e-01
-1.80502906e-01 -1.23752490e-01 -2.39649177e-01 8.35194707e-01
1.29128897e+00 -3.30098271e-01 -1.48311764e-01 -1.16575453e-02
7.65566766e-01 -1.58765107e-01 -4.51753020e-01 4.15428728e-01
8.26203227e-01 -1.03344667e+00 9.47511911e-01 -7.51100004e-01
1.00566387e+00 -3.95676941e-01 -5.18813467e-05 -1.02671921e+00
-7.81337261e-01 -5.42121649e-01 -6.19167924e-01 1.03966224e+00
2.88228422e-01 6.95436820e-02 1.03401375e+00 7.36421227e-01
-1.02872744e-01 -8.80058408e-01 -5.74683428e-01 -3.22927237e-01
-2.74885267e-01 -4.30150837e-01 2.39709228e-01 9.08443868e-01
-4.24514681e-01 4.09521937e-01 -1.36305106e+00 1.51115462e-01
4.30979967e-01 3.21411550e-01 6.92661464e-01 -7.41835535e-01
-5.12776196e-01 -7.27821112e-01 -3.61583054e-01 -1.70491886e+00
4.41093475e-01 -2.91207522e-01 1.02808692e-01 -1.54236329e+00
5.26420236e-01 1.38113052e-01 -4.12890643e-01 1.45810053e-01
-4.44032371e-01 1.34950295e-01 4.71333176e-01 5.73868334e-01
-1.09106159e+00 9.03557301e-01 1.37251413e+00 -2.67515004e-01
-2.98452497e-01 1.24203414e-01 -2.13343129e-01 9.06700253e-01
3.99127603e-01 -5.33470288e-02 -5.04227817e-01 -8.97855282e-01
3.12094558e-02 3.97373974e-01 6.31118417e-01 -7.67850339e-01
4.64416921e-01 -4.82167602e-01 6.53753459e-01 -6.38689458e-01
4.00739133e-01 -6.75279856e-01 4.30373132e-01 4.99895245e-01
-3.41646194e-01 3.84293467e-01 -1.66614093e-02 8.62525642e-01
-3.32990408e-01 3.88897777e-01 4.86063063e-01 -8.05767775e-02
-1.53328955e+00 9.24862802e-01 -6.30832493e-01 -4.33245599e-01
9.92452860e-01 -3.49152625e-01 -2.55510867e-01 -6.52861714e-01
-1.02062285e+00 3.81998509e-01 4.89171892e-01 7.94265091e-01
9.07203138e-01 -1.38204491e+00 -6.14201307e-01 -6.33886456e-03
-2.42575526e-01 -2.56497204e-01 9.61948216e-01 1.06276739e+00
-4.37736779e-01 2.53051043e-01 -1.75493717e-01 -5.60740232e-01
-1.27105820e+00 6.74520075e-01 4.94423777e-01 -2.07705259e-01
-7.92299807e-01 9.91955996e-01 7.68658340e-01 2.16105670e-01
2.28971690e-01 -1.74871489e-01 -4.39743847e-01 -1.37639776e-01
6.49310172e-01 3.07960689e-01 -5.48155248e-01 -8.80411685e-01
-3.37436646e-02 4.70382512e-01 -5.64056635e-02 2.13411808e-01
1.47463048e+00 -7.27160513e-01 -1.35457486e-01 5.86959660e-01
1.06345558e+00 -6.31864071e-01 -2.10845637e+00 -1.64793596e-01
4.31137942e-02 -2.91333616e-01 -4.47138585e-02 -3.24472576e-01
-1.41135228e+00 8.61035585e-01 3.26597810e-01 2.81280801e-02
1.16860557e+00 -1.70691296e-01 1.07788205e+00 1.31730393e-01
1.64699763e-01 -7.30971634e-01 2.57388145e-01 5.82023680e-01
6.51085854e-01 -1.15421700e+00 -2.25676343e-01 -2.43282616e-01
-8.74195814e-01 1.28602052e+00 1.00024104e+00 -1.06380887e-01
5.89009106e-01 -1.33678973e-01 2.53481176e-02 1.25577390e-01
-1.42855251e+00 2.07201615e-02 5.49647987e-01 2.65084296e-01
4.19811904e-01 -7.85724148e-02 1.89620286e-01 4.97750491e-01
2.73070663e-01 1.02855898e-01 3.31068933e-01 2.74299949e-01
-2.57192940e-01 -8.25506866e-01 -2.38931589e-02 3.43158275e-01
-2.70256490e-01 -5.35128228e-02 -2.80217171e-01 2.12514564e-01
1.58473939e-01 7.69300044e-01 2.43222237e-01 -7.25131214e-01
-9.40243676e-02 -2.28087008e-01 4.61896926e-01 -7.20703136e-03
-1.29344165e-01 3.09515595e-01 -9.22596753e-02 -8.92850757e-01
-6.55091107e-01 -6.34645700e-01 -8.03182960e-01 -4.17034924e-01
7.37318993e-02 -3.61133814e-01 2.54241042e-02 1.12806821e+00
4.75907475e-01 6.62796617e-01 6.68026805e-01 -1.11934412e+00
-9.47370455e-02 -8.01447868e-01 -4.24819142e-01 6.39474511e-01
6.64145470e-01 -3.66611302e-01 1.03488248e-02 6.58524036e-01] | [8.802618026733398, 0.17750431597232819] |
680e6e60-7e69-4054-8ebc-8eb149182f92 | detecting-and-classifying-lesions-in | 1707.08401 | null | http://arxiv.org/abs/1707.08401v3 | http://arxiv.org/pdf/1707.08401v3.pdf | Detecting and classifying lesions in mammograms with Deep Learning | In the last two decades Computer Aided Diagnostics (CAD) systems were
developed to help radiologists analyze screening mammograms. The benefits of
current CAD technologies appear to be contradictory and they should be improved
to be ultimately considered useful. Since 2012 deep convolutional neural
networks (CNN) have been a tremendous success in image recognition, reaching
human performance. These methods have greatly surpassed the traditional
approaches, which are similar to currently used CAD solutions. Deep CNN-s have
the potential to revolutionize medical image analysis. We propose a CAD system
based on one of the most successful object detection frameworks, Faster R-CNN.
The system detects and classifies malignant or benign lesions on a mammogram
without any human intervention. The proposed method sets the state of the art
classification performance on the public INbreast database, AUC = 0.95 . The
approach described here has achieved the 2nd place in the Digital Mammography
DREAM Challenge with AUC = 0.85 . When used as a detector, the system reaches
high sensitivity with very few false positive marks per image on the INbreast
dataset. Source code, the trained model and an OsiriX plugin are availaible
online at https://github.com/riblidezso/frcnn_cad . | ['István Csabai', 'Péter Pollner', 'Zsuzsa Unger', 'Anna Horváth', 'Dezső Ribli'] | 2017-07-26 | null | null | null | null | ['breast-cancer-detection', 'breast-cancer-detection'] | ['knowledge-base', 'medical'] | [ 1.84988379e-01 4.24813271e-01 -4.54175562e-01 -5.40739477e-01
-8.96254241e-01 -1.79883078e-01 4.70446587e-01 4.24366057e-01
-6.02683783e-01 3.28245342e-01 -2.52118111e-01 -7.93251157e-01
-1.35330409e-01 -7.77555585e-01 -4.54939961e-01 -6.31565809e-01
-1.74746796e-01 7.23444462e-01 5.32867253e-01 -1.84555262e-01
-1.94602594e-01 5.88809848e-01 -1.41202807e+00 6.46739960e-01
3.39118838e-01 1.27470458e+00 8.56689215e-02 1.06572139e+00
3.57733816e-01 8.27906728e-01 -2.06665933e-01 -6.74117386e-01
-4.23927382e-02 -3.24987173e-01 -7.53004372e-01 -4.59558696e-01
2.87455916e-01 -6.01786613e-01 -4.09863114e-01 7.76353598e-01
8.61116648e-01 -8.29611003e-01 4.66745913e-01 -3.51376623e-01
-4.91673589e-01 4.95017022e-01 -5.15921295e-01 7.98984289e-01
-1.65301234e-01 -5.01812696e-02 4.35709864e-01 -7.94623852e-01
3.50887328e-01 8.71423244e-01 1.03165317e+00 6.39381051e-01
-7.79524148e-01 -5.93916774e-01 -9.33417380e-01 2.78910697e-01
-1.34531653e+00 -4.44175005e-01 4.19333875e-02 -3.92297447e-01
7.54475474e-01 5.86027682e-01 8.51808667e-01 7.23306179e-01
7.66722739e-01 7.52631307e-01 7.70415783e-01 -5.01713812e-01
1.62855163e-02 2.72974223e-01 1.93477973e-01 1.03243220e+00
7.09409714e-01 3.30841810e-01 1.05558611e-01 -2.83081770e-01
6.90960050e-01 4.88491207e-02 1.85777143e-01 -1.85848325e-01
-9.77317214e-01 9.86628652e-01 8.20333302e-01 7.68193364e-01
-3.31293523e-01 1.84883885e-02 6.92797899e-01 5.56659400e-02
3.12247425e-01 2.97802687e-01 -4.24830243e-02 9.43505764e-02
-8.53877604e-01 1.57710552e-01 4.75639522e-01 2.89112568e-01
-1.52736425e-01 -3.45194608e-01 -2.12494820e-01 7.76934743e-01
8.82004350e-02 3.73192191e-01 7.78216958e-01 -5.39796233e-01
-2.32539088e-01 6.87389016e-01 -4.42758277e-02 -1.01042151e+00
-9.08004701e-01 -6.79981828e-01 -1.21138382e+00 8.19667801e-02
3.54362071e-01 5.38723059e-02 -1.19293118e+00 8.68396759e-01
1.48177683e-01 -3.71728480e-01 -8.31006691e-02 7.48122275e-01
1.22624445e+00 1.68723404e-01 1.23642109e-01 2.78809130e-01
1.80247176e+00 -2.49591500e-01 -5.48803866e-01 -1.00153834e-01
9.55891371e-01 -6.48729026e-01 3.31478953e-01 5.59720099e-01
-1.08557153e+00 -5.35231173e-01 -1.33801317e+00 -1.85617097e-02
-2.84821749e-01 7.23792374e-01 8.42087746e-01 1.16858757e+00
-1.07426202e+00 4.17217970e-01 -1.51319075e+00 -5.77258229e-01
1.00298142e+00 6.44308031e-01 -3.37044835e-01 -2.63747633e-01
-9.68532503e-01 1.12943649e+00 4.92709756e-01 3.26488614e-01
-9.45502698e-01 -4.97642726e-01 -4.94024515e-01 -2.51820296e-01
2.86496103e-01 -5.89779615e-01 1.78636718e+00 -7.53728747e-01
-9.10826325e-01 1.48694026e+00 1.63338065e-01 -9.18897629e-01
8.58292282e-01 -3.16307098e-02 -5.06569326e-01 3.01427007e-01
1.31085455e-01 6.90293789e-01 6.68632463e-02 -6.45203352e-01
-9.13020074e-01 -4.99510646e-01 -3.20961654e-01 -3.07381690e-01
-5.28772101e-02 1.45473301e-01 -4.71783876e-01 -1.74347222e-01
1.95896655e-01 -8.24181736e-01 -4.35450226e-01 3.70946437e-01
-1.58262372e-01 -1.09136321e-01 5.75927377e-01 -6.91904187e-01
9.76597071e-01 -2.04401541e+00 -5.36296546e-01 6.27551228e-03
3.79190743e-01 7.63685703e-01 5.08934915e-01 -1.71260536e-01
-3.27648401e-01 1.30308852e-01 -1.14630526e-02 1.96736872e-01
-5.85022151e-01 9.28642228e-02 6.47039831e-01 7.15603828e-01
4.78851467e-01 1.20738852e+00 -8.08445573e-01 -4.57301080e-01
4.23295677e-01 5.23300529e-01 -2.86299944e-01 -4.74000014e-02
3.25774968e-01 9.86236632e-02 -3.58420491e-01 7.45128155e-01
7.11346269e-01 -6.41491055e-01 3.44185650e-01 -2.94500768e-01
-8.66503827e-03 -1.44394904e-01 -8.26676488e-01 1.22943234e+00
-7.36671537e-02 5.64929485e-01 -1.19426101e-01 -1.21534634e+00
9.28150952e-01 4.79049921e-01 4.93746340e-01 -8.21115136e-01
4.62905854e-01 6.12812042e-01 5.83800912e-01 -7.45276034e-01
7.21260812e-03 -2.58431911e-01 4.37759936e-01 -2.26097301e-01
5.13414703e-02 1.02530606e-03 1.13203943e-01 1.16496094e-01
1.45725608e+00 -4.12999988e-01 9.81711149e-01 -4.10478592e-01
4.66481000e-01 3.28168005e-01 1.20321363e-01 9.93639112e-01
-1.93995655e-01 5.88783324e-01 5.92335403e-01 -7.52239287e-01
-9.81626391e-01 -8.07384193e-01 -8.51306021e-01 4.67806011e-01
-3.45932752e-01 5.58205731e-02 -4.19454753e-01 -4.29210216e-01
-1.78779975e-01 2.30905294e-01 -1.12904644e+00 -5.45212440e-02
-6.06070638e-01 -1.11197317e+00 8.11794758e-01 5.70975423e-01
6.43472016e-01 -8.95173907e-01 -1.12638843e+00 1.91543117e-01
1.19820833e-01 -6.89268231e-01 5.76917887e-01 4.11192119e-01
-8.77912283e-01 -1.36822653e+00 -1.20161963e+00 -7.81199694e-01
5.86047292e-01 1.55491620e-01 1.10082316e+00 2.19900727e-01
-1.15198219e+00 -5.51079288e-02 -2.95465827e-01 -8.14326525e-01
-7.70345986e-01 1.18291385e-01 -4.45293993e-01 -4.94129479e-01
7.31122494e-01 1.20733015e-01 -9.84710753e-01 1.19625896e-01
-8.56104612e-01 4.19158861e-02 1.15159345e+00 9.08311486e-01
5.89480400e-01 -2.95982659e-01 5.71316957e-01 -1.17279732e+00
9.93746966e-02 -6.49662137e-01 -4.00950074e-01 -6.13927059e-02
-5.24535179e-01 -5.19932747e-01 1.99785411e-01 2.72550397e-02
-8.35666597e-01 2.60498315e-01 -5.99586725e-01 1.00467257e-01
-2.34998941e-01 6.54672205e-01 5.32314420e-01 4.76973578e-02
1.20610654e+00 -1.88711911e-01 1.95052862e-01 -4.24587190e-01
-2.28176355e-01 8.32883179e-01 7.17595160e-01 1.81421399e-01
2.14043200e-01 7.18686283e-01 2.72314340e-01 -7.61409461e-01
-8.91261995e-01 -5.84166765e-01 -5.53231537e-01 -1.04948640e-01
9.31376815e-01 -9.17746067e-01 -5.62125146e-01 2.47940511e-01
-8.01639974e-01 1.82306245e-02 -2.70910859e-01 4.67488974e-01
-4.74659689e-02 9.72390473e-02 -5.99860907e-01 -6.39191449e-01
-5.14727831e-01 -1.05044651e+00 9.04663086e-01 4.27959561e-01
-2.45157778e-01 -6.20793641e-01 -1.07934866e-02 9.60607380e-02
6.11726642e-01 5.45847535e-01 6.72756314e-01 -8.28379154e-01
-8.37671235e-02 -1.01694357e+00 -6.12530828e-01 3.46923858e-01
2.20833957e-01 -1.29484134e-02 -9.77789521e-01 -1.97706938e-01
-1.69524267e-01 -2.29892522e-01 1.10600448e+00 8.80504012e-01
1.39860606e+00 2.47776657e-01 -7.81091809e-01 4.88435924e-01
1.30686438e+00 4.99663830e-01 7.81536102e-01 4.18279231e-01
7.33300596e-02 1.56047747e-01 4.47226197e-01 1.82865307e-01
-4.16142382e-02 3.34434360e-01 5.37881136e-01 -5.73986769e-01
-2.62374073e-01 1.41856432e-01 -3.91214341e-01 1.85732707e-01
-3.71794760e-01 2.77717393e-02 -1.55700982e+00 5.70461750e-01
-1.45732594e+00 -6.42721653e-01 -5.33147156e-01 1.96571541e+00
4.96604145e-01 5.23863673e-01 5.19286841e-02 2.24270225e-01
4.41785693e-01 -4.09940004e-01 -3.19308341e-01 -3.25160027e-01
-1.63670182e-02 6.40419245e-01 6.99243665e-01 3.08433473e-02
-1.45679653e+00 3.39316368e-01 6.36580992e+00 5.92517614e-01
-1.33985221e+00 4.58632290e-01 1.16259456e+00 1.35198265e-01
4.56427276e-01 -5.71019173e-01 -6.46491408e-01 6.44884855e-02
1.21954131e+00 1.40118245e-02 -6.12812817e-01 1.02908230e+00
1.29604369e-01 -5.14289021e-01 -9.77930129e-01 7.05849946e-01
-4.50919047e-02 -1.68043530e+00 -2.12491989e-01 3.27251628e-02
3.72749716e-01 3.23010743e-01 1.90958738e-01 2.36342788e-01
-5.55290934e-03 -1.34902227e+00 8.41337666e-02 5.13019085e-01
1.25572348e+00 -6.14986360e-01 1.56533194e+00 1.11839801e-01
-7.02079773e-01 -5.61329201e-02 -4.58725333e-01 1.56404763e-01
-2.81903505e-01 6.42185211e-01 -1.33953631e+00 5.16340911e-01
1.02620447e+00 6.12060905e-01 -1.03137279e+00 1.42471993e+00
3.33205521e-01 8.89251053e-01 -1.76711246e-01 -1.68014705e-01
2.46588632e-01 6.27364993e-01 5.12147099e-02 1.40946507e+00
4.13281322e-01 3.41741025e-01 -2.81110913e-01 3.89966577e-01
5.28346747e-02 1.63496315e-01 -3.81736577e-01 -3.50755174e-03
-4.67907786e-02 1.40549612e+00 -1.12701547e+00 -5.00240326e-01
-4.21922326e-01 4.76675153e-01 -2.22096145e-01 -4.53341901e-01
-7.34642148e-01 -2.38905936e-01 -2.43246108e-01 4.22569752e-01
1.87298909e-01 4.54549611e-01 -2.47761607e-01 -5.22274017e-01
-1.92923203e-01 -8.57530951e-01 5.40157557e-01 -4.61697131e-01
-9.17501211e-01 5.64979434e-01 -1.17219791e-01 -1.22708166e+00
-2.26083323e-01 -1.03722167e+00 -3.04094076e-01 5.27423620e-01
-1.21855724e+00 -1.06756616e+00 -5.18088162e-01 2.55338363e-02
2.96305954e-01 -2.89882123e-01 1.17434478e+00 5.95302761e-01
-3.20162773e-01 7.00806975e-01 3.46199930e-01 3.92296255e-01
5.68081379e-01 -1.05334198e+00 1.96760461e-01 4.64520246e-01
-5.26029408e-01 1.38200551e-01 3.11456591e-01 -5.09312093e-01
-1.09306586e+00 -9.72553134e-01 6.45170927e-01 -3.45905662e-01
4.81467783e-01 1.35095313e-01 -7.60798931e-01 6.55307353e-01
-1.56440169e-01 3.62430036e-01 7.48865366e-01 -2.01984361e-01
5.27510867e-02 -1.11356318e-01 -1.24235272e+00 3.27774324e-03
3.55040371e-01 2.46487558e-01 -2.79711068e-01 4.60055828e-01
4.43965718e-02 -6.06022835e-01 -7.71883070e-01 8.96942139e-01
8.37771297e-01 -1.19315004e+00 8.76883030e-01 -4.59739089e-01
6.88824117e-01 1.42161667e-01 3.96331623e-02 -6.39682770e-01
-3.85274172e-01 -2.97623128e-03 9.74636674e-02 4.80914116e-01
5.21476865e-01 -4.72035050e-01 9.55514371e-01 1.35219634e-01
-1.41927391e-01 -1.12499273e+00 -1.25444388e+00 -4.42795038e-01
1.52650788e-01 -2.24167407e-01 8.95036981e-02 6.66918039e-01
-2.14818522e-01 -4.83893156e-02 8.99087414e-02 -6.11979291e-02
3.50826770e-01 -5.22212744e-01 4.27328706e-01 -1.34007990e+00
-3.80194962e-01 -6.01071000e-01 -8.71616721e-01 -5.30236721e-01
-8.68253589e-01 -1.09215879e+00 -2.26552397e-01 -1.57662082e+00
6.92921042e-01 -4.53574747e-01 -3.91937882e-01 4.86916006e-01
-5.15448004e-02 7.14091659e-01 -2.67838895e-01 1.86924428e-01
-3.81046712e-01 -3.56298655e-01 1.17335153e+00 -2.15474620e-01
7.90755898e-02 4.12557840e-01 -7.49245882e-01 7.55357385e-01
1.01367378e+00 -5.12455463e-01 3.13671768e-01 -1.65340945e-01
2.90566862e-01 3.13431084e-01 5.20359159e-01 -1.46035802e+00
-3.89085934e-02 4.27754611e-01 9.41182971e-01 -7.50451386e-01
2.33497649e-01 -5.91197968e-01 3.74012917e-01 1.39599633e+00
-2.88998455e-01 -3.35685432e-01 4.92189527e-01 2.79561341e-01
-1.80559903e-01 -2.73941875e-01 1.13229823e+00 -4.10351694e-01
-4.19624954e-01 1.04448691e-01 -5.57393253e-01 -5.43867826e-01
1.01175141e+00 -2.91624755e-01 -2.17106625e-01 6.38056220e-03
-9.87358093e-01 -7.11097196e-03 -1.89611256e-01 2.53099382e-01
4.76767570e-01 -1.11774468e+00 -1.18835938e+00 -3.98793183e-02
2.99793780e-01 -3.48290019e-02 4.53963339e-01 1.12015903e+00
-1.05522442e+00 8.60085189e-01 -3.98400843e-01 -9.73187625e-01
-1.47769535e+00 3.04402441e-01 9.31332588e-01 -4.55664843e-01
-6.38141155e-01 1.00033569e+00 5.28763346e-02 -8.36729407e-02
4.97635007e-02 -4.02256072e-01 -4.93093848e-01 2.58862823e-02
8.72340798e-01 2.84773350e-01 6.75970316e-01 -2.61214703e-01
-4.53161687e-01 3.86190973e-02 -7.29741395e-01 3.80447119e-01
1.50970721e+00 4.59479809e-01 7.59591684e-02 3.36038709e-01
1.02946889e+00 -5.41573584e-01 -2.63252795e-01 -9.94879454e-02
-8.99078697e-03 -2.87667990e-01 3.80228490e-01 -1.18416798e+00
-9.56566870e-01 9.60541129e-01 1.57585812e+00 4.08995032e-01
9.87342834e-01 3.89217228e-01 3.03251415e-01 4.18044806e-01
-3.34516615e-02 -5.45482278e-01 -1.06590297e-02 -2.10640971e-02
6.59526467e-01 -1.73379755e+00 3.21484476e-01 -2.48635948e-01
-4.10263598e-01 1.37189996e+00 4.84599710e-01 -1.98242173e-01
9.18188989e-01 3.60881299e-01 1.23525396e-01 -5.90297818e-01
-4.37119067e-01 -5.10371774e-02 3.17601740e-01 4.77810115e-01
1.09639871e+00 4.19518709e-01 -6.46643817e-01 6.57402813e-01
-3.20407562e-03 4.72749889e-01 4.43601012e-01 9.86043632e-01
-6.77954435e-01 -6.99790657e-01 -2.99338043e-01 1.20966363e+00
-1.03072405e+00 2.92024851e-01 -3.09570611e-01 1.08238256e+00
2.27369398e-01 6.26391649e-01 -2.63147615e-02 -8.33967254e-02
5.10013938e-01 -2.43048280e-01 2.64850825e-01 -5.25509775e-01
-5.98606706e-01 1.15671113e-01 1.24440089e-01 -4.74685341e-01
-3.85476530e-01 -6.03426576e-01 -9.76534784e-01 -1.45735696e-01
-4.66294378e-01 -9.01520252e-02 6.55248225e-01 6.10465348e-01
8.16811994e-02 8.37499261e-01 1.59997314e-01 -7.15869188e-01
-3.99773717e-01 -1.12466955e+00 -4.14675981e-01 -2.17641234e-01
3.40226173e-01 -3.99667442e-01 1.43212527e-01 1.00134499e-02] | [15.252618789672852, -2.5111420154571533] |
41a4d4d5-ae1e-41ec-8903-84db8bec9577 | speaker-and-age-invariant-training-for-child | 2210.10231 | null | https://arxiv.org/abs/2210.10231v2 | https://arxiv.org/pdf/2210.10231v2.pdf | Speaker- and Age-Invariant Training for Child Acoustic Modeling Using Adversarial Multi-Task Learning | One of the major challenges in acoustic modelling of child speech is the rapid changes that occur in the children's articulators as they grow up, their differing growth rates and the subsequent high variability in the same age group. These high acoustic variations along with the scarcity of child speech corpora have impeded the development of a reliable speech recognition system for children. In this paper, a speaker- and age-invariant training approach based on adversarial multi-task learning is proposed. The system consists of one generator shared network that learns to generate speaker- and age-invariant features connected to three discrimination networks, for phoneme, age, and speaker. The generator network is trained to minimize the phoneme-discrimination loss and maximize the speaker- and age-discrimination losses in an adversarial multi-task learning fashion. The generator network is a Time Delay Neural Network (TDNN) architecture while the three discriminators are feed-forward networks. The system was applied to the OGI speech corpora and achieved a 13% reduction in the WER of the ASR. | ['Julien Epps', 'Beena Ahmed', 'Mostafa Shahin'] | 2022-10-19 | null | null | null | null | ['acoustic-modelling'] | ['speech'] | [ 3.56619537e-01 2.79664367e-01 3.73830914e-01 -4.87806171e-01
-9.00366724e-01 -2.93377250e-01 2.43327558e-01 -8.62943456e-02
-3.77432913e-01 3.90117854e-01 1.71333641e-01 -1.38369100e-02
4.36421074e-02 -5.16326129e-01 -5.36812901e-01 -7.54637837e-01
7.55270496e-02 5.09920299e-01 2.18068808e-01 -6.73502460e-02
-3.04314464e-01 2.73697048e-01 -1.90112972e+00 1.16374373e-01
8.74308467e-01 7.58813798e-01 2.38256529e-01 1.08562028e+00
5.38236052e-02 3.03738713e-01 -9.52506721e-01 -3.37518573e-01
-2.13422962e-02 -3.05369079e-01 -3.10920745e-01 -3.49990636e-01
5.27430773e-01 -3.18382263e-01 -3.41598809e-01 7.75359094e-01
1.07600188e+00 9.04654190e-02 8.30016673e-01 -1.08911741e+00
-6.36682332e-01 1.02331662e+00 -1.72418222e-01 -4.23735380e-03
-1.18228770e-03 -3.17812175e-01 6.18497670e-01 -8.34700048e-01
-8.91983360e-02 1.49204087e+00 3.16382855e-01 1.20787656e+00
-1.04930246e+00 -7.89755344e-01 -6.37630895e-02 -2.64951915e-01
-1.37673867e+00 -7.31395841e-01 6.87924385e-01 -5.58635831e-01
9.53363240e-01 -3.97104695e-02 4.25027221e-01 1.20490956e+00
1.78246759e-02 4.96973902e-01 7.80714452e-01 -6.49883151e-01
1.02209508e-01 3.06944083e-02 -2.89493412e-01 2.62299836e-01
-2.24695385e-01 5.42064428e-01 -5.53045154e-01 2.40960672e-01
4.64763850e-01 -4.28843081e-01 3.82387906e-01 -5.63175455e-02
-5.28301835e-01 7.40401268e-01 -1.90255851e-01 5.20266473e-01
-8.99190009e-02 1.92459032e-01 4.96066928e-01 5.07416189e-01
7.29773045e-01 -1.57980761e-03 -5.16449094e-01 -1.56190977e-01
-8.14763486e-01 2.50437200e-01 6.04577839e-01 8.20256352e-01
2.70412594e-01 6.74370527e-01 1.66591033e-01 1.53471720e+00
5.82127571e-01 8.58011484e-01 8.33050847e-01 -4.50810105e-01
4.38084722e-01 1.64943188e-01 -6.35980666e-01 -4.64210659e-01
-2.21485287e-01 -3.01403582e-01 -5.35441637e-01 5.26703119e-01
5.13949633e-01 -5.05485117e-01 -1.15039349e+00 2.17878819e+00
2.76625514e-01 1.30374938e-01 1.87689796e-01 1.87648162e-01
7.36788213e-01 6.57993257e-01 4.14899975e-01 -5.51810265e-02
8.03265870e-01 -7.06168711e-01 -4.95689452e-01 -5.57535291e-01
2.43569627e-01 -8.61888647e-01 4.59097624e-01 3.46807957e-01
-1.46665263e+00 -6.93844914e-01 -1.17678773e+00 2.13791072e-01
-4.37221706e-01 -6.36637509e-02 2.42386296e-01 1.26235747e+00
-1.10569310e+00 3.21600199e-01 -7.52311409e-01 -4.98338640e-02
-4.51361649e-02 7.47623205e-01 -1.52377963e-01 3.57561171e-01
-1.20839000e+00 9.86036062e-01 4.44240570e-01 -7.55342692e-02
-1.20450187e+00 -7.56231010e-01 -9.73608911e-01 3.02775130e-02
-3.38576734e-01 -2.55730957e-01 1.35332930e+00 -1.05933714e+00
-1.93881512e+00 8.46057773e-01 3.75069916e-01 -4.71123636e-01
3.59148353e-01 -8.04677010e-02 -7.42328405e-01 -1.17065251e-01
-1.09443322e-01 6.86077058e-01 1.20507896e+00 -8.41387928e-01
-8.06903780e-01 -5.40449917e-01 -6.47816658e-01 4.13610071e-01
-3.50967616e-01 4.60402876e-01 3.84372585e-02 -8.68264377e-01
-2.33620973e-04 -9.25966084e-01 -5.53085208e-02 -4.03199762e-01
1.93854168e-01 -2.31293365e-01 7.20763862e-01 -8.95005882e-01
1.11098409e+00 -2.45972943e+00 2.40113199e-01 6.39784411e-02
-1.55088857e-01 4.87959445e-01 -2.79195160e-01 1.66251436e-01
-2.72610128e-01 -2.06246302e-01 -2.08558962e-02 -5.22816837e-01
-1.94561541e-01 1.63189217e-01 7.26487339e-02 2.73830056e-01
2.46902391e-01 4.10544991e-01 -8.88622344e-01 -1.03867762e-01
3.21485400e-01 7.11907029e-01 -3.70597839e-01 5.65610647e-01
-7.58744553e-02 5.36763966e-01 -1.44875005e-01 3.99677843e-01
4.01031554e-01 9.39113259e-01 -2.24738702e-01 4.84300762e-01
-1.82071045e-01 3.46831799e-01 -9.88381922e-01 1.46137261e+00
-7.27179170e-01 2.93305367e-01 4.27703649e-01 -9.92631733e-01
1.33909547e+00 7.86573887e-01 2.55491287e-01 -5.24124265e-01
1.69225082e-01 7.04856575e-01 5.70580661e-01 -1.17385216e-01
2.04062775e-01 -3.28494668e-01 -1.69762447e-01 1.85150549e-01
4.01778668e-01 -6.33382261e-01 -2.04134017e-01 -3.07455987e-01
7.33764648e-01 -3.66466880e-01 -1.83718219e-01 -3.39877933e-01
6.16314292e-01 -9.70954061e-01 5.75526297e-01 3.22863847e-01
-1.73198342e-01 7.30478406e-01 -1.76592723e-01 -1.41427398e-01
-1.09882867e+00 -1.47165775e+00 9.22873616e-02 1.41624057e+00
-6.67300701e-01 3.51317137e-01 -7.50039577e-01 -1.05946034e-01
6.59238100e-02 7.58848369e-01 -3.45624238e-01 -4.78389949e-01
-9.17103291e-01 -1.47584364e-01 8.76218259e-01 5.75335979e-01
-1.34116381e-01 -1.22984803e+00 -1.06460631e-01 5.11100471e-01
5.28738081e-01 -8.32220852e-01 -7.23871827e-01 2.48312846e-01
-4.18872267e-01 -5.22536516e-01 -1.00230491e+00 -1.14412951e+00
5.16789913e-01 -4.97877747e-01 6.00392342e-01 -4.78873223e-01
-1.96570575e-01 3.75684202e-01 -2.75669008e-01 -1.10221303e+00
-1.15062141e+00 2.33131215e-01 5.00057876e-01 -1.14072539e-01
3.08582366e-01 -9.66597736e-01 -2.82956749e-01 1.18862875e-01
-5.92651486e-01 -2.79162109e-01 2.59176344e-01 8.95535111e-01
2.30627224e-01 -6.94988295e-02 1.34824443e+00 -4.52549189e-01
3.85003835e-01 -3.42749894e-01 -6.08972013e-01 1.21801272e-01
-6.20991230e-01 -2.34036259e-02 5.05266011e-01 -7.82030523e-01
-1.10494637e+00 1.65701777e-01 -6.63935781e-01 -1.41213745e-01
-2.91763097e-01 2.72082016e-02 -3.81245673e-01 1.01835392e-02
4.26862240e-01 1.73923131e-02 2.02141777e-01 -3.95582765e-01
3.36776793e-01 1.05370581e+00 8.81984234e-01 -5.91248691e-01
7.95250714e-01 -4.18968201e-01 -2.84693003e-01 -1.09081626e+00
-4.89971846e-01 -1.26591831e-01 -5.02252579e-01 -2.43095934e-01
8.68382156e-01 -1.02184892e+00 -9.29317325e-02 1.45328772e+00
-9.35985804e-01 -4.08985317e-01 -2.53583521e-01 6.65726721e-01
-6.63633227e-01 -1.94863662e-01 -4.08226550e-01 -1.00265479e+00
-8.91925693e-01 -1.13584304e+00 4.36173916e-01 5.13352096e-01
-1.24658406e-01 -1.02777731e+00 1.48973256e-01 3.72057796e-01
5.76499104e-01 1.18853368e-01 1.00945640e+00 -1.18472886e+00
1.59520999e-01 -1.15123153e-01 4.08342868e-01 1.02642655e+00
3.06400537e-01 1.77889884e-01 -1.22801304e+00 -5.08944511e-01
3.61098908e-02 -4.91039783e-01 2.67298609e-01 6.42335117e-01
8.66556406e-01 7.38314167e-02 8.97446126e-02 4.63308722e-01
9.23838496e-01 8.35828364e-01 1.42856538e-01 -2.72827566e-01
6.39810979e-01 7.83411086e-01 1.26643091e-01 5.22385649e-02
3.63576651e-01 5.23310006e-01 2.16548234e-01 1.97142079e-01
-4.90992516e-01 -3.57913941e-01 4.99927014e-01 1.50206161e+00
8.45695361e-02 -2.15015501e-01 -9.84599948e-01 1.09810770e+00
-1.17526722e+00 -6.72981024e-01 3.27035695e-01 2.39429641e+00
8.62025857e-01 1.22593842e-01 3.23181659e-01 2.58914292e-01
1.09279501e+00 6.21999837e-02 -3.55039477e-01 -1.12074661e+00
-1.18834913e-01 6.66022658e-01 2.23628640e-01 4.69697267e-01
-8.36919487e-01 8.53944898e-01 6.80781746e+00 6.98593020e-01
-1.22444654e+00 1.57437176e-01 6.18740439e-01 -1.62540838e-01
-1.40766233e-01 -6.23192072e-01 -8.97895336e-01 5.31445324e-01
1.49843276e+00 -3.40468943e-01 5.28815925e-01 7.30546176e-01
3.24722491e-02 2.50394940e-01 -1.27121747e+00 7.30712473e-01
1.84457988e-01 -4.04424608e-01 -3.29379320e-01 -2.60546893e-01
6.52842343e-01 1.95999175e-01 4.30090696e-01 3.51379693e-01
2.57952988e-01 -1.10395837e+00 1.09253812e+00 2.16269102e-02
1.17898810e+00 -1.01624894e+00 2.34457448e-01 2.30254605e-01
-1.05942166e+00 -1.18767247e-01 1.21799782e-02 -4.48498242e-02
8.65536481e-02 1.53151169e-01 -9.58479226e-01 1.01728933e-02
3.81540060e-01 -7.24160522e-02 -1.79223001e-01 8.18666935e-01
-2.49347597e-01 9.11953390e-01 -3.65898639e-01 -1.14443138e-01
-1.54418275e-01 1.16078928e-01 5.56304574e-01 1.11681843e+00
5.90173841e-01 -7.32929111e-02 -2.22067043e-01 3.69215250e-01
5.15273996e-02 2.09167302e-01 -5.70389569e-01 -1.70195267e-01
8.41087997e-01 7.35582769e-01 -1.55509487e-01 1.41009301e-01
-5.10093212e-01 7.63665020e-01 3.21841031e-01 -6.08363077e-02
-3.87954563e-01 -4.86105084e-01 6.52656555e-01 -4.08248380e-02
2.02626422e-01 -1.19954646e-01 -1.93749934e-01 -5.00078738e-01
-1.32509425e-01 -8.65862131e-01 1.47467211e-01 -9.84239429e-02
-1.24314344e+00 5.68918347e-01 4.95391712e-02 -6.06005132e-01
-9.38414395e-01 -5.61687231e-01 -9.47282612e-01 1.35723960e+00
-9.52610373e-01 -1.41273284e+00 3.39258194e-01 3.53784949e-01
8.21999729e-01 -6.89505100e-01 1.01800585e+00 7.40266144e-01
-5.60446978e-01 1.25882578e+00 3.10791619e-02 1.19212352e-01
5.73305547e-01 -1.29500222e+00 6.38329685e-01 8.20287526e-01
-2.02014416e-01 1.55811325e-01 6.26134574e-01 -3.85583013e-01
-7.66308248e-01 -9.66886401e-01 1.00345063e+00 -2.68688556e-02
6.88586533e-01 -6.19497240e-01 -7.92256355e-01 4.70090687e-01
7.23787993e-02 -3.07511032e-01 8.13032329e-01 -4.74288082e-03
-2.48149797e-01 -3.54909718e-01 -1.29282880e+00 2.33780742e-01
7.47706652e-01 -6.32338345e-01 -7.02389836e-01 -2.43080854e-01
8.24169099e-01 -5.67176759e-01 -1.06696963e+00 4.78381276e-01
8.23289037e-01 -4.64532554e-01 9.09172356e-01 -1.68561265e-01
1.38824314e-01 2.09376216e-01 -3.20942253e-02 -1.75904298e+00
1.44623229e-02 -5.60651302e-01 -1.62385274e-02 1.87105203e+00
7.95344353e-01 -6.91259146e-01 5.13677657e-01 6.97492659e-01
-5.10918140e-01 -6.38304830e-01 -1.29106355e+00 -7.24087000e-01
6.93869710e-01 -3.04132104e-01 6.59576774e-01 3.36571217e-01
-3.08147758e-01 2.43615061e-01 -3.39424074e-01 2.89188236e-01
4.12298679e-01 -6.66955650e-01 1.64386675e-01 -1.21054029e+00
-3.68078291e-01 -2.74190307e-01 -5.32702386e-01 -3.63608390e-01
1.80662692e-01 -6.47896051e-01 3.36706549e-01 -9.05908108e-01
-3.99830937e-01 -7.10849106e-01 -4.66888666e-01 3.24447453e-01
-1.16414554e-01 -3.37045789e-01 5.23472540e-02 -5.80655277e-01
3.01663697e-01 4.09185946e-01 7.68255532e-01 -1.41790837e-01
-3.59002143e-01 6.64624095e-01 -2.96233565e-01 7.25471497e-01
7.97540784e-01 -5.34137785e-01 -7.32983470e-01 -7.72975981e-01
-1.72119066e-01 1.87333718e-01 -4.12798762e-01 -1.13720071e+00
1.62359744e-01 2.50387371e-01 2.04159901e-01 -4.79723096e-01
4.12064761e-01 -5.37701666e-01 1.47294849e-01 5.73306501e-01
-3.12607467e-01 -2.42304262e-02 2.01174974e-01 -5.98287620e-02
-1.93246692e-01 -3.67926389e-01 1.15553498e+00 2.75359184e-01
-2.18557328e-01 4.17093158e-01 -4.92654115e-01 1.45833552e-01
9.47372913e-01 3.34426835e-02 5.62932529e-02 -2.33704552e-01
-7.71696270e-01 1.53785035e-01 -1.68367177e-02 9.12127435e-01
5.93862236e-01 -1.19126201e+00 -1.03756022e+00 7.27148235e-01
-8.80327523e-02 2.46809408e-01 2.42252111e-01 3.44749773e-03
-3.04402739e-01 1.59661189e-01 -3.19922298e-01 -2.13329405e-01
-1.53624845e+00 1.47264034e-01 3.79281342e-01 1.12134613e-01
7.42270891e-03 1.35892498e+00 6.71302453e-02 -5.24889171e-01
5.10302663e-01 -1.91670150e-01 -2.37907961e-01 1.57276969e-02
5.43713331e-01 6.47799313e-01 7.75310444e-03 -8.52143943e-01
-1.84162602e-01 3.84155095e-01 -2.87407577e-01 -3.95497084e-01
1.30434608e+00 -2.81315595e-02 2.05575496e-01 6.65568531e-01
1.05041897e+00 3.45525295e-02 -9.90245461e-01 -6.23405501e-02
-2.15824232e-01 -1.12728968e-01 -8.45019296e-02 -7.87790120e-01
-1.00698221e+00 9.58675265e-01 8.37321758e-01 2.98873067e-01
1.20965695e+00 8.92171413e-02 7.71396697e-01 -4.04506743e-01
1.99478731e-01 -1.24306703e+00 -1.59227997e-01 6.46077991e-01
8.58865619e-01 -9.25738215e-01 -6.18268132e-01 -2.75583804e-01
-1.47506192e-01 8.49767387e-01 9.09025490e-01 4.44498025e-02
6.38248920e-01 5.09505689e-01 3.05715978e-01 2.75270343e-01
-6.97870791e-01 2.06792485e-02 3.11914742e-01 1.00183570e+00
6.25533044e-01 4.32813466e-01 -4.66993034e-01 6.74006343e-01
-5.35558879e-01 -5.31669736e-01 1.40887752e-01 3.89732003e-01
-3.96905780e-01 -1.44142616e+00 -1.97800949e-01 3.77251744e-01
-8.15149248e-01 1.17118265e-02 -3.63347866e-02 3.39345813e-01
5.15360713e-01 1.10349393e+00 1.67907134e-01 -4.21973079e-01
3.19416404e-01 2.47359514e-01 5.81920505e-01 -1.07290697e+00
-7.39983201e-01 -9.69882980e-02 2.11760372e-01 5.00666499e-02
1.01032965e-02 -9.72037435e-01 -1.27629983e+00 2.13283181e-01
-4.43375707e-01 5.37311174e-02 1.25840664e+00 8.49045455e-01
-2.75377512e-01 6.61429763e-01 1.02190995e+00 -4.87058431e-01
-8.41701984e-01 -1.07381165e+00 -5.73558569e-01 1.75611116e-02
4.34110254e-01 -4.58786011e-01 -5.58071554e-01 -1.93678156e-01] | [14.437618255615234, 6.480188846588135] |
e510dc5c-07cc-4bbc-ac36-45a1a0351f10 | cqr-sql-conversational-question-reformulation | 2205.07686 | null | https://arxiv.org/abs/2205.07686v3 | https://arxiv.org/pdf/2205.07686v3.pdf | CQR-SQL: Conversational Question Reformulation Enhanced Context-Dependent Text-to-SQL Parsers | Context-dependent text-to-SQL is the task of translating multi-turn questions into database-related SQL queries. Existing methods typically focus on making full use of history context or previously predicted SQL for currently SQL parsing, while neglecting to explicitly comprehend the schema and conversational dependency, such as co-reference, ellipsis and user focus change. In this paper, we propose CQR-SQL, which uses auxiliary Conversational Question Reformulation (CQR) learning to explicitly exploit schema and decouple contextual dependency for SQL parsing. Specifically, we first present a schema enhanced recursive CQR method to produce domain-relevant self-contained questions. Secondly, we train CQR-SQL models to map the semantics of multi-turn questions and auxiliary self-contained questions into the same latent space through schema grounding consistency task and tree-structured SQL parsing consistency task, which enhances the abilities of SQL parsing by adequately contextual understanding. At the time of writing, our CQR-SQL achieves new state-of-the-art results on two context-dependent text-to-SQL benchmarks SParC and CoSQL. | ['Yunbo Cao', 'Zhoujun Li', 'Zhao Yan', 'Qian-Wen Zhang', 'Linzheng Chai', 'Dongling Xiao'] | 2022-05-16 | null | null | null | null | ['text-to-sql'] | ['computer-code'] | [ 1.26320392e-01 4.24934775e-01 -7.54231885e-02 -9.54117298e-01
-1.43478823e+00 -8.52009833e-01 5.17077208e-01 5.76097667e-01
-1.22952741e-02 2.85445601e-01 7.63318419e-01 -7.96309769e-01
1.05231598e-01 -1.20538366e+00 -9.13080513e-01 3.49841356e-01
4.13298607e-01 7.52345622e-01 5.04599094e-01 -6.65209293e-01
6.70153275e-02 -1.33376777e-01 -1.21609378e+00 1.15529573e+00
9.64475274e-01 7.09349275e-01 1.71479464e-01 9.03800786e-01
-1.06712687e+00 1.55094862e+00 -6.90890431e-01 -7.44014442e-01
-4.05463517e-01 -4.60111558e-01 -1.35539281e+00 -2.67316908e-01
4.47435379e-01 -2.02781916e-01 2.06731111e-02 5.99875152e-01
1.41134843e-01 -2.06222773e-01 2.92034051e-03 -1.01513219e+00
-5.28230727e-01 1.02258313e+00 1.68661192e-01 1.26141068e-02
9.23669755e-01 9.31838825e-02 1.43714845e+00 -6.28059149e-01
7.30556846e-01 1.49483645e+00 4.07770783e-01 5.10733485e-01
-1.21580815e+00 -1.67868972e-01 3.02138984e-01 -2.56803762e-02
-7.03549266e-01 -4.06025529e-01 7.71715343e-01 -2.94362515e-01
1.50766087e+00 7.35468090e-01 1.89879805e-01 9.47760463e-01
2.24368836e-04 8.50541174e-01 9.51078236e-01 -4.36344117e-01
1.34632409e-01 3.50493491e-01 5.98000288e-01 6.87223017e-01
-2.60056973e-01 -3.42931628e-01 -5.57818830e-01 -2.01995805e-01
2.45042786e-01 -1.15234174e-01 1.33838654e-01 -1.45983621e-01
-1.18304801e+00 9.16767716e-01 1.02161653e-01 5.95304221e-02
-1.66473091e-01 -1.33755252e-01 6.33971751e-01 5.11135697e-01
6.05729185e-02 7.11606979e-01 -8.51940870e-01 -2.29732126e-01
-3.89117122e-01 4.01797354e-01 1.46410835e+00 1.56854224e+00
9.41640913e-01 -4.02746558e-01 -3.19653094e-01 7.22883105e-01
1.89396694e-01 6.43576801e-01 2.23678723e-01 -1.29463327e+00
1.09324455e+00 1.14283347e+00 -5.91166690e-02 -6.91706955e-01
-3.20010662e-01 1.58380680e-02 -2.72712618e-01 -8.50691974e-01
1.06562279e-01 2.44425926e-02 -4.07751471e-01 1.61738658e+00
4.23277020e-01 -4.85450447e-01 6.20146513e-01 4.21895325e-01
1.25474668e+00 8.32267523e-01 3.78806919e-01 -2.06570134e-01
1.78162599e+00 -9.39638138e-01 -8.15716624e-01 -5.16768277e-01
8.50519478e-01 -7.29658604e-01 1.69270909e+00 -1.17336750e-01
-1.26818287e+00 -6.71164632e-01 -5.62136114e-01 -6.37500107e-01
-4.20944214e-01 -2.28945851e-01 5.75357854e-01 3.24326694e-01
-6.76617026e-01 -6.84515312e-02 -7.95257270e-01 -5.08564711e-01
-2.18357429e-01 -2.55949795e-01 -1.48887873e-01 -4.13549468e-02
-1.36553133e+00 6.07093990e-01 3.86605680e-01 -2.03374594e-01
-4.56897527e-01 -9.86027420e-01 -1.10779881e+00 1.17878802e-01
9.58283961e-01 -7.58248210e-01 1.69437802e+00 -2.84127623e-01
-1.46260655e+00 1.01664782e+00 -6.41971409e-01 -2.55099654e-01
1.75876990e-01 -4.35831964e-01 -4.16606575e-01 -8.44535977e-02
4.87503320e-01 4.20229256e-01 2.51057088e-01 -9.29924130e-01
-5.85338295e-01 -5.34688592e-01 4.08446819e-01 1.31424576e-01
2.22058669e-01 1.20465241e-01 -6.59052312e-01 -2.49485850e-01
3.28736037e-01 -5.36238730e-01 1.95150506e-02 -7.39462733e-01
-6.56504571e-01 -5.95352769e-01 4.55129653e-01 -7.41531730e-01
1.49154699e+00 -1.92707193e+00 5.48233129e-02 -2.57802308e-01
2.20137268e-01 -1.78498566e-01 -1.78353056e-01 9.94326472e-01
-2.14529186e-02 2.78819740e-01 -2.01474413e-01 -6.13185279e-02
1.58195332e-01 4.78260130e-01 -9.07269180e-01 -5.26646376e-01
6.64098144e-01 1.33059430e+00 -6.07321620e-01 -7.23944485e-01
-5.51707984e-04 -1.66539609e-01 -8.26881647e-01 1.02635872e+00
-1.22082496e+00 1.72589242e-01 -6.25824630e-01 6.97745740e-01
4.10001785e-01 -4.38336313e-01 5.01395404e-01 -4.35937762e-01
-7.15139583e-02 1.17841816e+00 -8.09550285e-01 1.77511418e+00
-7.89924860e-01 6.38417453e-02 -1.85516804e-01 -7.03738332e-01
9.25414622e-01 1.49517640e-01 5.38392626e-02 -1.20054471e+00
-5.84370255e-01 5.31613454e-02 -5.96061110e-01 -1.02494788e+00
4.78243351e-01 -4.49198373e-02 -7.41865277e-01 5.65880120e-01
-9.21871439e-02 -4.19420838e-01 1.13869347e-02 6.00662291e-01
1.10556006e+00 1.70377567e-01 3.83791953e-01 6.24929555e-02
7.78730214e-01 3.40271056e-01 4.08667117e-01 8.11414123e-01
1.65612459e-01 2.21294403e-01 1.04572868e+00 -3.48638624e-01
-6.30611300e-01 -1.38305867e+00 1.84556544e-01 1.74372029e+00
-4.57817540e-02 -8.56411934e-01 -6.67084694e-01 -9.77722824e-01
1.17360288e-02 1.41988361e+00 -3.46144348e-01 1.25266790e-01
-1.15375531e+00 -1.87697798e-01 5.15059292e-01 5.14517903e-01
4.03754115e-01 -1.33230758e+00 -6.48387730e-01 5.14073968e-01
-8.51773322e-01 -1.51093698e+00 -3.47809166e-01 2.77282059e-01
-8.29880416e-01 -1.31922662e+00 3.39589298e-01 -6.43977344e-01
3.15942871e-03 -6.65648952e-02 1.89371252e+00 -8.05411302e-03
-3.77817638e-02 6.49873495e-01 -2.93662459e-01 -1.71758696e-01
-8.34206402e-01 4.77451771e-01 -1.03507721e+00 -4.23136532e-01
6.62024856e-01 -2.67460048e-01 -1.96684688e-01 2.22214535e-01
-9.83024001e-01 3.90850782e-01 3.35555613e-01 4.46511030e-01
5.56606650e-01 -5.02578437e-01 5.10933220e-01 -1.77254963e+00
6.61380470e-01 -4.39662397e-01 -5.11889637e-01 8.52951944e-01
-4.52472985e-01 5.79571247e-01 6.32072210e-01 5.55737391e-02
-1.48721123e+00 -2.92223096e-01 -4.09612924e-01 3.33267152e-01
-3.34049344e-01 6.90777540e-01 -5.35586894e-01 9.04580772e-01
6.01594210e-01 3.21170986e-01 -3.02964330e-01 -3.94360274e-01
8.04523230e-01 6.03134274e-01 8.92769754e-01 -1.13003528e+00
4.63466227e-01 1.52839556e-01 -4.14285064e-01 -4.28987980e-01
-1.25989723e+00 -5.59623063e-01 -3.22955251e-01 2.60784864e-01
1.15301275e+00 -8.56673837e-01 -9.39451277e-01 1.15271062e-01
-1.38717735e+00 -3.76336068e-01 -3.14061821e-01 -2.40547895e-01
-7.02250302e-01 2.70266324e-01 -7.32234418e-01 -5.90819120e-01
-3.65945458e-01 -9.30273116e-01 1.40630054e+00 -6.64132461e-02
-3.15054148e-01 -9.54677761e-01 2.86583096e-01 8.68211806e-01
5.06901443e-01 8.26104060e-02 1.74560845e+00 -9.91911352e-01
-9.54384506e-01 4.71152067e-02 -2.37052143e-01 -3.71573679e-02
7.52737150e-02 -2.35722959e-01 -6.90128505e-01 1.21271156e-01
3.05542290e-01 -7.21030533e-01 3.36050063e-01 -3.08506161e-01
9.12497163e-01 -6.29845917e-01 1.63355861e-02 6.05193198e-01
1.34125543e+00 1.08214550e-01 5.34166396e-01 1.50044501e-01
5.64216852e-01 1.00532424e+00 6.30363107e-01 1.79083720e-01
1.32878554e+00 4.85426694e-01 3.05748314e-01 2.35015571e-01
7.95487836e-02 -8.09197307e-01 1.53905034e-01 1.06193209e+00
9.51411784e-01 -1.02369674e-01 -1.08900559e+00 2.61514395e-01
-1.78637242e+00 -5.04839957e-01 -3.33781242e-01 1.78546166e+00
1.27206516e+00 9.43774804e-02 -2.10194886e-01 -6.08483613e-01
4.73520249e-01 4.09723133e-01 -6.28772676e-01 -4.31634992e-01
-8.99295211e-02 1.59774885e-01 -1.55064598e-01 8.18673313e-01
-7.85217285e-01 1.31000435e+00 5.56359434e+00 1.11222006e-01
-8.61753881e-01 5.39657176e-02 3.10731560e-01 1.64963737e-01
-9.12205994e-01 3.96498084e-01 -1.03275633e+00 1.18875138e-01
1.16260743e+00 -7.35976314e-03 4.69309956e-01 8.45141172e-01
-2.77453333e-01 -9.84372273e-02 -1.60480249e+00 6.05293930e-01
-9.20634791e-02 -1.35464668e+00 2.13717937e-01 -6.07194960e-01
8.68042856e-02 -2.13500068e-01 -3.56533676e-01 9.79285061e-01
4.14606452e-01 -7.97361910e-01 5.04032910e-01 5.01963973e-01
5.55050850e-01 -3.83554459e-01 5.65729320e-01 4.51907158e-01
-1.17181993e+00 2.96784248e-02 -1.20233387e-01 2.60790676e-01
1.16617754e-01 2.73679972e-01 -7.42607713e-01 6.21793509e-01
6.53473675e-01 3.12717289e-01 -1.01890326e+00 -1.85601026e-01
-2.97991186e-01 6.01240754e-01 -1.22178651e-01 -1.87917426e-01
2.96844449e-02 -1.49186666e-03 2.74657071e-01 1.37995386e+00
-1.48609444e-01 3.29649597e-01 2.40375668e-01 1.05405450e+00
-9.27993953e-02 1.83486819e-01 -2.94160157e-01 -2.02106059e-01
6.42853618e-01 9.19081807e-01 -6.25342652e-02 -6.12707615e-01
-6.81913793e-01 6.84554040e-01 5.23890376e-01 3.73142511e-01
-6.61005974e-01 -3.19169551e-01 4.55535501e-01 3.50735575e-01
2.70804852e-01 1.85413565e-02 -3.12038302e-01 -1.49198246e+00
4.70266968e-01 -1.38810623e+00 8.78168702e-01 -9.24650192e-01
-1.18491650e+00 5.35827577e-01 6.53860271e-02 -2.75474250e-01
-5.80414116e-01 -2.30291605e-01 -5.31513870e-01 8.47835422e-01
-1.57709444e+00 -1.26342463e+00 -3.80732685e-01 6.46979332e-01
7.69244254e-01 -3.92042696e-02 1.00157273e+00 -8.09011457e-04
-4.50078309e-01 4.12539780e-01 -6.12914741e-01 4.10525709e-01
9.08359349e-01 -1.52032793e+00 8.50415468e-01 7.29434907e-01
-1.26298219e-02 1.16190398e+00 5.45365572e-01 -7.59587288e-01
-1.90909863e+00 -1.20762110e+00 1.48616672e+00 -1.05511069e+00
7.11850703e-01 -7.18658328e-01 -1.40639484e+00 1.02216649e+00
2.79175103e-01 -2.31227219e-01 7.57991135e-01 4.83624637e-01
-8.87675583e-01 -3.65906060e-01 -6.38174534e-01 5.17254889e-01
8.48274171e-01 -1.26223493e+00 -1.26355493e+00 3.45587015e-01
1.63788342e+00 -6.18465960e-01 -9.36745882e-01 5.26797175e-01
1.62968799e-01 -1.09296715e+00 8.95344734e-01 -1.24411523e+00
5.77858746e-01 -6.91068321e-02 -6.58546805e-01 -4.81064677e-01
1.22388661e-01 -7.15394974e-01 -2.27152899e-01 1.54196990e+00
5.25340736e-01 -4.92622405e-01 8.44241142e-01 9.62337255e-01
-1.59134045e-01 -5.43594360e-01 -5.94980896e-01 -2.23745778e-01
7.40529820e-02 -7.07727611e-01 8.91383171e-01 8.19780946e-01
1.72639161e-01 8.53065610e-01 2.63321936e-01 2.17699572e-01
3.45206708e-01 8.05674851e-01 1.08148205e+00 -8.02736402e-01
-6.13177598e-01 7.33804097e-03 4.72910464e-01 -1.33601046e+00
1.68450117e-01 -8.47043276e-01 1.27113640e-01 -1.53288901e+00
1.12692840e-01 -3.42594743e-01 2.22934708e-01 3.74387980e-01
-5.37962496e-01 -9.58784580e-01 2.53550351e-01 9.30183083e-02
-1.04951596e+00 2.99505949e-01 8.91733289e-01 -7.68645257e-02
-3.46123517e-01 1.55745611e-01 -1.01750243e+00 2.08171695e-01
4.40228820e-01 -5.61399221e-01 -5.77400744e-01 -6.84811890e-01
7.71742821e-01 1.11234605e+00 1.85387805e-01 -5.64099014e-01
4.38132584e-01 -4.03564900e-01 -3.34926695e-01 -7.86022127e-01
-2.74106394e-02 -4.83900964e-01 -1.52408211e-02 2.54364222e-01
-1.07003391e+00 4.77619529e-01 1.24109887e-01 4.90098208e-01
-5.23831069e-01 -1.15171567e-01 3.63189995e-01 -4.64674950e-01
-5.06710112e-01 -8.19983929e-02 -2.36481458e-01 1.10113251e+00
4.85375702e-01 4.18126315e-01 -6.65306151e-01 -4.22615230e-01
-5.85861325e-01 5.94824910e-01 7.65643045e-02 6.39459074e-01
4.75084364e-01 -9.05420482e-01 -6.79874241e-01 3.49270821e-01
5.77651858e-01 3.96362245e-01 2.41457686e-01 3.17216963e-01
-5.43057561e-01 9.26388621e-01 2.99768031e-01 -8.31762493e-01
-1.01056135e+00 6.60827935e-01 3.27624917e-01 -6.20278180e-01
-3.68611485e-01 6.51751161e-01 2.82347918e-01 -1.30929172e+00
3.06706190e-01 -7.50386834e-01 4.74374220e-02 -1.80425435e-01
2.65258312e-01 -1.63349971e-01 2.52837509e-01 1.08290195e-01
-3.56589079e-01 3.90854686e-01 -1.99427232e-01 -1.58690542e-01
1.04620075e+00 -2.72889167e-01 -5.58764100e-01 6.23809159e-01
1.28404701e+00 1.21388048e-01 -8.76523256e-01 -6.63185179e-01
6.02731884e-01 -1.13263642e-02 -6.26611471e-01 -1.12784183e+00
-3.59391689e-01 1.04953957e+00 -9.44081843e-02 4.31468040e-01
9.19903278e-01 4.81724709e-01 9.88727272e-01 1.01608515e+00
1.98108152e-01 -8.51838529e-01 3.79162252e-01 8.26350093e-01
9.58228648e-01 -1.35627282e+00 -3.90132576e-01 -6.17669821e-01
-8.00900996e-01 1.15787065e+00 9.90205169e-01 3.65404487e-01
4.31354672e-01 3.71664822e-01 4.21458602e-01 -3.80381674e-01
-1.33048832e+00 -4.80382703e-02 3.24277170e-02 3.54463905e-01
6.42800212e-01 -2.82237772e-02 1.91860944e-01 9.16577578e-01
-5.50748110e-01 -2.29627207e-01 2.22433805e-01 1.13864338e+00
-3.88001382e-01 -1.18545806e+00 1.69572625e-02 1.56302392e-01
-3.67374122e-01 -3.65305722e-01 -4.98955011e-01 8.80943954e-01
-4.17107910e-01 1.01610410e+00 1.97960645e-01 -1.43224552e-01
8.61220837e-01 5.57646394e-01 1.90080464e-01 -9.31354225e-01
-1.03645480e+00 -1.08790047e-01 4.40539479e-01 -9.35780048e-01
1.16545342e-01 -5.68397462e-01 -1.37625813e+00 -1.60430610e-01
1.58654332e-01 3.61094475e-01 5.64253092e-01 8.96541059e-01
5.92447400e-01 4.62453336e-01 4.62807685e-01 5.81877708e-01
-8.58848989e-01 -6.28599644e-01 2.85877913e-01 6.53234482e-01
2.69816011e-01 3.18313390e-01 -3.38849165e-02 3.52699161e-02] | [10.019991874694824, 7.851653575897217] |
4cd97e9f-53d2-495d-a4ad-ae83d5ecac85 | strictly-breadth-first-amr-parsing | 2211.03922 | null | https://arxiv.org/abs/2211.03922v1 | https://arxiv.org/pdf/2211.03922v1.pdf | Strictly Breadth-First AMR Parsing | AMR parsing is the task that maps a sentence to an AMR semantic graph automatically. We focus on the breadth-first strategy of this task, which was proposed recently and achieved better performance than other strategies. However, current models under this strategy only \emph{encourage} the model to produce the AMR graph in breadth-first order, but \emph{cannot guarantee} this. To solve this problem, we propose a new architecture that \emph{guarantees} that the parsing will strictly follow the breadth-first order. In each parsing step, we introduce a \textbf{focused parent} vertex and use this vertex to guide the generation. With the help of this new architecture and some other improvements in the sentence and graph encoder, our model obtains better performance on both the AMR 1.0 and 2.0 dataset. | ['Daniel Gildea', 'Chen Yu'] | 2022-11-08 | null | null | null | null | ['amr-parsing'] | ['natural-language-processing'] | [ 5.59224784e-01 9.44862187e-01 2.63899192e-02 -7.22220540e-01
-6.33642137e-01 -7.07275391e-01 4.71996069e-01 1.81856577e-03
-6.36932701e-02 5.03773212e-01 2.90448368e-01 -6.44201875e-01
2.63034016e-01 -1.21500444e+00 -6.33919120e-01 8.38371590e-02
3.33987266e-01 5.80868542e-01 3.89057875e-01 -4.02613461e-01
8.96029472e-02 2.05330923e-01 -9.48553443e-01 6.26914263e-01
6.66791856e-01 5.89500487e-01 5.60549974e-01 7.34430373e-01
-6.52288377e-01 1.22128046e+00 -6.38825178e-01 -9.11217749e-01
1.29693687e-01 -6.91300392e-01 -1.24444020e+00 -1.04604594e-01
2.07741007e-01 -3.58724475e-01 -2.58481055e-01 9.88076270e-01
1.25578657e-01 1.32312877e-02 2.74574429e-01 -8.42780769e-01
-1.04900491e+00 1.57517397e+00 -5.59068322e-01 2.37145275e-01
5.88212669e-01 -2.44297251e-01 1.41934001e+00 -4.27112341e-01
8.91563714e-01 1.23754358e+00 3.14023584e-01 9.85845208e-01
-8.79297316e-01 -3.99843246e-01 6.46983147e-01 -3.07766914e-01
-9.75338757e-01 -2.28321075e-01 9.44773495e-01 -6.92772493e-02
1.38418233e+00 2.82600641e-01 4.91924137e-01 8.67400765e-01
8.86174291e-02 8.12486529e-01 7.30708122e-01 -5.47491074e-01
-7.38772079e-02 -2.07608566e-01 4.74747062e-01 8.05150926e-01
-5.17905839e-02 -2.85166740e-01 -3.38040769e-01 3.16277027e-01
6.31956458e-01 -4.03604537e-01 -4.71428260e-02 2.80734748e-01
-7.63242900e-01 9.49559212e-01 4.62947130e-01 4.30155933e-01
-4.27160524e-02 4.26214159e-01 7.41757676e-02 2.14128017e-01
4.70529139e-01 5.59611022e-01 -3.89020503e-01 -8.13452750e-02
-7.90267766e-01 -7.85079971e-02 7.71975338e-01 1.29169297e+00
6.84468687e-01 1.10073179e-01 -2.69024223e-01 7.70782113e-01
4.27245110e-01 2.10288167e-01 8.31381381e-02 -1.05039632e+00
9.15430427e-01 8.20684612e-01 -1.29931793e-01 -8.58175278e-01
-5.47543764e-01 -6.18597567e-01 -5.80671608e-01 -1.91347197e-01
2.63414413e-01 -3.10455747e-02 -1.10203493e+00 1.88594627e+00
1.37343660e-01 -1.73260272e-01 3.25344086e-01 6.84987605e-01
1.15475500e+00 8.01123023e-01 1.41234696e-01 -1.91523075e-01
1.39015758e+00 -1.21964097e+00 -8.04905772e-01 -5.63403666e-01
1.00344610e+00 -6.96291268e-01 1.26524627e+00 1.62944615e-01
-1.23720980e+00 -6.13372803e-01 -9.65201855e-01 -3.57418746e-01
-1.81558311e-01 3.01733240e-02 8.02918375e-01 5.98740458e-01
-1.38607657e+00 5.58754206e-01 -6.70558691e-01 -1.44094646e-01
1.22132987e-01 1.44158512e-01 -4.02962804e-01 -9.78183299e-02
-1.32949972e+00 7.47610390e-01 3.42939138e-01 2.48777002e-01
-4.17243600e-01 -4.09947485e-01 -1.23000109e+00 3.03795300e-02
4.02857482e-01 -8.02719772e-01 1.43922019e+00 -9.78152990e-01
-1.60257030e+00 9.55456734e-01 -2.20491379e-01 -6.69665515e-01
1.98503330e-01 -2.48109788e-01 -2.81270683e-01 1.84131026e-01
9.92660522e-02 8.12407970e-01 4.57789034e-01 -1.09239531e+00
-5.09544075e-01 -2.90388465e-01 6.42545879e-01 2.05942124e-01
-1.24364205e-01 2.90982544e-01 -6.04880154e-01 -6.69364035e-01
3.20514113e-01 -8.05694044e-01 -3.54550928e-01 -8.94391060e-01
-5.58240354e-01 -5.64901352e-01 2.09998861e-01 -7.62417972e-01
1.60721755e+00 -1.91894567e+00 2.55642354e-01 -1.07019491e-01
2.69295096e-01 6.24198131e-02 -2.67832905e-01 5.51552296e-01
-7.99515545e-02 3.72901410e-01 -2.40844727e-01 -4.96281177e-01
-3.75579335e-02 2.74468064e-01 -3.42206478e-01 -1.64818570e-01
2.62761682e-01 1.09331763e+00 -8.50324988e-01 -4.25883353e-01
-2.86633316e-02 1.08905457e-01 -5.88193893e-01 4.05670673e-01
-6.28588259e-01 2.23078787e-01 -5.42060435e-01 2.57679403e-01
5.78910649e-01 -2.89736986e-01 4.53511417e-01 2.50478517e-02
-7.86936656e-02 6.78243935e-01 -8.85525942e-01 1.79355597e+00
-5.22052109e-01 1.77373350e-01 -7.65595585e-02 -8.70781302e-01
1.28663695e+00 1.12236992e-01 1.79341391e-01 -7.00948358e-01
2.47078650e-02 1.23642460e-01 -2.98853684e-02 -6.05667718e-02
5.98205328e-01 1.04829237e-01 -2.54017413e-01 4.47746098e-01
-8.53458419e-02 -2.65229851e-01 4.09095317e-01 5.96864700e-01
1.32636690e+00 3.66402835e-01 2.95579523e-01 -1.30070150e-01
6.18147373e-01 1.33248508e-01 5.44687510e-01 9.01146114e-01
2.87391990e-01 8.14107478e-01 7.84170330e-01 -5.41057706e-01
-8.74298632e-01 -7.41914988e-01 3.19153577e-01 9.94694352e-01
-5.50048165e-02 -9.23292458e-01 -1.00123835e+00 -1.16857529e+00
-5.29019594e-01 1.25591791e+00 -4.93112624e-01 -7.70979151e-02
-1.18592966e+00 -5.71776628e-01 5.59852242e-01 6.91920757e-01
3.84091139e-01 -1.40776527e+00 -4.56185639e-01 4.15308297e-01
-2.89889365e-01 -1.48425186e+00 -4.96294498e-01 -1.61700957e-02
-5.77099085e-01 -8.26464653e-01 -2.68998832e-01 -9.44846809e-01
7.31369913e-01 -5.24987504e-02 1.54062569e+00 4.09867406e-01
2.42446199e-01 2.56930571e-02 -8.72726083e-01 -1.47388577e-01
-8.22630584e-01 4.48037863e-01 -7.74769843e-01 -3.22051108e-01
2.59453446e-01 -3.10643375e-01 -2.91671723e-01 -5.80558591e-02
-7.67681956e-01 6.33342028e-01 3.08252394e-01 3.47492695e-01
6.18510365e-01 1.04251057e-01 6.29692018e-01 -1.43903017e+00
5.73515415e-01 -1.16093725e-01 -6.20699227e-01 3.21425974e-01
-5.02234221e-01 7.90891945e-02 1.12707067e+00 1.49196431e-01
-1.01167893e+00 1.80720359e-01 -8.08993876e-01 9.83681306e-02
-1.42535090e-01 4.34023499e-01 -3.27220052e-01 4.92232502e-01
2.73798734e-01 8.57011378e-02 -3.93968910e-01 -5.48647285e-01
5.87144315e-01 5.34975052e-01 4.37786967e-01 -5.85384130e-01
6.63128734e-01 1.57740206e-01 1.10503688e-01 -3.49530876e-01
-1.25036526e+00 -2.24829763e-02 -3.66905451e-01 -1.06145591e-02
1.20015371e+00 -7.34758437e-01 -4.96928841e-01 1.04768150e-01
-1.50732994e+00 -6.29677832e-01 -3.05782378e-01 1.36701509e-01
-5.25664985e-01 5.07943213e-01 -8.45834732e-01 -9.47169304e-01
-7.02608049e-01 -9.94509518e-01 1.10069382e+00 1.22886054e-01
-2.86883503e-01 -7.67204285e-01 -1.89261720e-01 4.72584665e-01
2.67929465e-01 8.49675294e-03 1.13009048e+00 -8.65984142e-01
-4.68430072e-01 3.26011255e-02 -3.10625494e-01 3.71735305e-01
2.70440672e-02 3.26843821e-02 -7.01743364e-01 -1.18158283e-02
6.20690221e-03 5.77116981e-02 8.39190304e-01 8.66801888e-02
1.07409203e+00 -2.96088994e-01 -1.12649143e-01 4.79078948e-01
1.15029383e+00 3.32345814e-01 6.53169870e-01 2.58362770e-01
8.48446310e-01 7.44443893e-01 4.19807076e-01 -8.56921896e-02
9.05745387e-01 6.85164809e-01 6.23283088e-01 -1.20245047e-01
-5.36563456e-01 -6.01316631e-01 4.85735267e-01 1.02546906e+00
1.30479902e-01 -5.38890183e-01 -8.57708097e-01 2.77516842e-01
-1.86145985e+00 -6.92883372e-01 -4.80413735e-01 1.83091712e+00
8.26817572e-01 4.59780306e-01 -1.04079708e-01 -6.40492812e-02
6.60070658e-01 3.30732673e-01 6.82235807e-02 -9.02699888e-01
-2.25539356e-01 3.77267361e-01 2.01136157e-01 9.60105360e-01
-7.93227613e-01 1.68737233e+00 6.39271450e+00 5.51068842e-01
-7.99810529e-01 -2.35239137e-03 4.90248203e-01 3.13429475e-01
-7.93081045e-01 3.10372502e-01 -1.23767960e+00 2.03116968e-01
8.82162035e-01 8.22513476e-02 5.22349954e-01 8.46849859e-01
8.70670006e-02 7.56091923e-02 -1.15395308e+00 4.47397202e-01
1.22738265e-01 -1.28136897e+00 3.96382064e-01 -2.82808602e-01
3.02611887e-01 -3.60246927e-01 -2.65871346e-01 5.28149605e-01
6.07073545e-01 -1.06851161e+00 8.71722698e-01 1.39425635e-01
6.21719837e-01 -7.35622048e-01 7.27977693e-01 3.99302930e-01
-1.52170300e+00 7.60757849e-02 -4.45681751e-01 -1.97292194e-01
5.15513361e-01 7.25857198e-01 -7.58318007e-01 1.06644523e+00
3.33333522e-01 6.25801086e-01 -6.84602380e-01 2.18320623e-01
-9.54955876e-01 6.21152282e-01 -7.69260898e-03 -3.42187807e-02
2.10192427e-01 -3.34723175e-01 4.81812686e-01 1.27979696e+00
2.56541491e-01 2.05459103e-01 2.87678868e-01 8.52015138e-01
-1.38342693e-01 2.35688239e-01 -6.17292047e-01 -9.25902650e-02
3.24272722e-01 1.23392510e+00 -9.21395183e-01 -3.55131865e-01
-5.47648549e-01 1.09324038e+00 7.48669863e-01 1.52731359e-01
-8.79387558e-01 -3.40406269e-01 1.13065027e-01 2.36479729e-01
2.30247498e-01 -2.65621692e-01 -3.09674799e-01 -9.99491453e-01
1.48667067e-01 -8.02547514e-01 8.17478836e-01 -1.07198310e+00
-9.92019832e-01 1.16540802e+00 2.82487087e-02 -3.35354805e-01
-3.18237901e-01 -6.14438474e-01 -5.60369015e-01 9.18460548e-01
-1.59317553e+00 -1.19718337e+00 -1.69320200e-02 3.41697454e-01
8.57534170e-01 1.20843932e-01 9.11410093e-01 1.31412325e-02
-5.07223904e-01 5.10986805e-01 -1.00298965e+00 4.20679331e-01
2.57887304e-01 -1.57136559e+00 1.05539894e+00 1.34760559e+00
4.43051875e-01 7.16498494e-01 6.57396376e-01 -9.59714711e-01
-1.23612523e+00 -1.16612697e+00 1.30842245e+00 -6.19412661e-01
5.91569185e-01 -5.72672606e-01 -8.04891050e-01 1.04398334e+00
4.34725553e-01 -3.68866593e-01 4.13963050e-01 4.27504666e-02
-4.35048014e-01 -5.51044643e-02 -8.55792344e-01 5.62218010e-01
1.47100663e+00 -2.22079918e-01 -9.24380302e-01 4.05029476e-01
1.49515593e+00 -5.96284866e-01 -7.35200286e-01 6.04777455e-01
-3.68413851e-02 -8.64713788e-01 5.30461133e-01 -6.67266726e-01
6.26682162e-01 -4.39611644e-01 -2.17986077e-01 -1.16407204e+00
-4.14194882e-01 -9.44453061e-01 -2.16220051e-01 1.52087343e+00
8.93911004e-01 -5.66340089e-01 7.55911231e-01 2.91753143e-01
-6.25694573e-01 -5.76712132e-01 -7.03404069e-01 -7.43277848e-01
1.24005556e-01 -5.64876974e-01 7.76501834e-01 6.24228776e-01
-9.81651843e-02 8.55683804e-01 -2.17823744e-01 2.43685827e-01
4.40499693e-01 4.29433078e-01 5.95772803e-01 -9.15391624e-01
-5.31814039e-01 -2.59174794e-01 8.24745372e-02 -1.22481680e+00
4.17094678e-01 -1.25181746e+00 6.70487881e-02 -2.29716921e+00
2.90118977e-02 -2.93767333e-01 -3.23542096e-02 7.84363806e-01
-4.37164396e-01 -3.90392020e-02 3.81965578e-01 -1.31579310e-01
-5.23865402e-01 1.13935202e-01 1.16625488e+00 1.22631229e-01
-1.71878904e-01 -6.69423342e-02 -9.67471063e-01 6.39302731e-01
9.35749710e-01 -5.07989883e-01 -5.84159374e-01 -9.62193370e-01
7.50763416e-01 1.11751266e-01 -1.04607053e-01 -6.12899780e-01
-7.39963502e-02 -3.78863886e-02 1.28738070e-02 -5.63292801e-01
4.64127809e-02 -5.56138158e-01 2.43571743e-01 2.53752887e-01
-4.46238250e-01 5.14437079e-01 -1.21217713e-01 2.19554946e-01
-1.13218196e-01 -4.42909241e-01 6.96621954e-01 -4.03131992e-01
-5.29259801e-01 1.37395367e-01 -1.52051672e-01 2.19993755e-01
7.91182041e-01 4.82965633e-02 -6.16160929e-01 -3.68868053e-01
-7.40701437e-01 3.36855739e-01 3.18063617e-01 5.20141661e-01
4.15051550e-01 -9.36794341e-01 -8.00981700e-01 7.75325522e-02
-9.01321992e-02 3.42264146e-01 -6.10089637e-02 4.91035104e-01
-6.07905328e-01 3.05797994e-01 2.62408882e-01 -1.99025765e-01
-9.84595418e-01 7.43502676e-01 1.94080621e-01 -6.83239341e-01
-8.99698615e-01 8.58773828e-01 2.34341249e-01 -4.93238151e-01
-8.67991298e-02 -2.44856015e-01 -5.12725174e-01 -3.18090528e-01
4.47472602e-01 -1.17037380e-02 9.21819210e-02 -6.57447517e-01
-3.27806085e-01 5.05717814e-01 -7.00781569e-02 -1.49436906e-01
1.17233419e+00 -1.76836893e-01 -4.58543301e-01 1.58391014e-01
7.02835143e-01 3.74734223e-01 -8.49404752e-01 4.72901501e-02
1.89028814e-01 -5.70581704e-02 -3.59029144e-01 -8.06190789e-01
-1.03808558e+00 7.55505025e-01 -4.10692245e-01 7.30997980e-01
1.08844590e+00 3.09113979e-01 9.15364623e-01 1.69982880e-01
2.31296882e-01 -1.04509997e+00 5.97653165e-03 8.12026799e-01
7.65178621e-01 -6.21694505e-01 -2.96877563e-01 -1.11858642e+00
-6.91362619e-01 1.24421906e+00 8.74195039e-01 -6.12465367e-02
2.08081588e-01 4.25287664e-01 5.63101657e-02 -1.51836827e-01
-7.67984629e-01 -4.02518451e-01 1.49008840e-01 4.84449416e-01
7.11075246e-01 1.25860825e-01 -7.34072328e-01 1.07027209e+00
-6.66793644e-01 -1.95264772e-01 5.66438675e-01 7.25664914e-01
-6.75964653e-01 -1.42976439e+00 2.87653595e-01 1.47701398e-01
-7.41989791e-01 -4.08108979e-01 -8.02170992e-01 7.78259993e-01
-1.63326710e-01 1.31657088e+00 6.35715351e-02 -3.53409380e-01
6.17569387e-01 1.15814604e-01 4.89771634e-01 -1.21058321e+00
-8.09930503e-01 5.77468686e-02 7.49466896e-01 -6.14130914e-01
-1.17312960e-01 -1.41677782e-01 -1.83279455e+00 -7.07370043e-02
-1.25058383e-01 2.52979010e-01 5.17915606e-01 8.56031537e-01
3.34152877e-01 8.68428648e-01 7.03840792e-01 -1.01241822e-04
-4.46747452e-01 -8.38276923e-01 -1.87025279e-01 3.93644124e-01
-3.18002075e-01 9.23485681e-02 -3.02331656e-01 -8.28556195e-02] | [10.441776275634766, 9.312012672424316] |
0a779e47-26b1-47a4-b10f-dd8cc269cfba | stochastic-video-prediction-with-structure | 2203.10528 | null | https://arxiv.org/abs/2203.10528v2 | https://arxiv.org/pdf/2203.10528v2.pdf | Stochastic Video Prediction with Structure and Motion | While stochastic video prediction models enable future prediction under uncertainty, they mostly fail to model the complex dynamics of real-world scenes. For example, they cannot provide reliable predictions for scenes with a moving camera and independently moving foreground objects in driving scenarios. The existing methods fail to fully capture the dynamics of the structured world by only focusing on changes in pixels. In this paper, we assume that there is an underlying process creating observations in a video and propose to factorize it into static and dynamic components. We model the static part based on the scene structure and the ego-motion of the vehicle, and the dynamic part based on the remaining motion of the dynamic objects. By learning separate distributions of changes in foreground and background, we can decompose the scene into static and dynamic parts and separately model the change in each. Our experiments demonstrate that disentangling structure and motion helps stochastic video prediction, leading to better future predictions in complex driving scenarios on two real-world driving datasets, KITTI and Cityscapes. | ['Fatma Güney', 'Sadra Safadoust', 'Adil Kaan Akan'] | 2022-03-20 | null | null | null | null | ['video-prediction'] | ['computer-vision'] | [ 1.11967467e-01 -1.97966680e-01 -2.13869676e-01 -5.26831925e-01
-7.21436962e-02 -5.37766099e-01 9.19998467e-01 -3.05505633e-01
-9.49868094e-03 5.43217182e-01 4.23731774e-01 -1.79979235e-01
1.82630673e-01 -6.65735543e-01 -9.59849060e-01 -8.06779146e-01
-1.96097761e-01 4.90402192e-01 7.42688954e-01 -1.35394245e-01
-8.41960534e-02 1.80560291e-01 -1.77322340e+00 5.21946788e-01
5.40616930e-01 6.96792901e-01 5.22449195e-01 1.05683160e+00
-1.60172194e-01 1.51452029e+00 -1.40107706e-01 -3.23824465e-01
3.29417914e-01 -3.57349813e-01 -1.75365001e-01 4.64716613e-01
4.37825054e-01 -4.63933647e-01 -1.07084453e+00 9.19874668e-01
-1.17494412e-01 1.94908336e-01 4.25954700e-01 -1.33694100e+00
-1.28772438e-01 4.53648627e-01 -5.89501023e-01 4.46634352e-01
1.72746092e-01 4.05232072e-01 7.04947054e-01 -3.13855886e-01
6.92781746e-01 1.42637968e+00 4.66623336e-01 5.94128430e-01
-1.14558041e+00 -4.91963983e-01 1.02204514e+00 7.30516970e-01
-9.73303020e-01 -4.27681953e-01 9.28033233e-01 -8.40313554e-01
6.28729999e-01 1.96549922e-01 7.81871676e-01 1.36502135e+00
7.22343504e-01 1.18817961e+00 7.51534581e-01 2.00683653e-01
2.68676043e-01 -8.38078260e-02 7.77511522e-02 3.61660838e-01
9.79786590e-02 3.51322055e-01 -6.32958412e-01 8.67568105e-02
2.78418779e-01 3.09979677e-01 -1.89513460e-01 -6.77048802e-01
-1.25285292e+00 4.66720283e-01 -1.95189923e-01 -7.23952949e-02
-5.16598761e-01 5.03938973e-01 6.81998730e-02 4.71083857e-02
4.82071102e-01 -3.33407044e-01 -4.87318218e-01 -3.39693427e-01
-1.02993357e+00 4.82261688e-01 7.88246095e-01 1.09842539e+00
9.69574213e-01 1.54102832e-01 -1.79648086e-01 2.31316745e-01
1.73975229e-01 7.82523870e-01 3.32685590e-01 -1.24216104e+00
4.50640470e-01 8.79534930e-02 5.99280000e-01 -1.11812711e+00
-1.96393609e-01 -2.01617554e-01 -5.23495734e-01 -4.03774902e-02
5.13282180e-01 -2.96376467e-01 -1.19229591e+00 1.96279812e+00
2.18263716e-01 1.01453936e+00 -1.59994349e-01 6.50318563e-01
4.27823216e-01 1.02164876e+00 1.46722579e-02 -2.91415304e-01
9.01677907e-01 -1.13089836e+00 -7.87347674e-01 -6.16714656e-01
3.25035334e-01 -4.67538357e-01 4.58845466e-01 2.14691013e-01
-1.07282293e+00 -8.20841551e-01 -7.49284446e-01 2.32570156e-01
2.41516884e-02 -2.70619929e-01 4.27440524e-01 2.77916253e-01
-1.05525756e+00 2.89574087e-01 -1.28973508e+00 -1.28986895e-01
1.01566695e-01 -6.69986457e-02 -1.28882214e-01 -3.21831733e-01
-1.04036784e+00 7.49552071e-01 1.31470457e-01 1.26671731e-01
-1.43619561e+00 -7.19220757e-01 -7.52467513e-01 7.90412165e-03
6.03752673e-01 -7.42182195e-01 1.26592529e+00 -1.31788623e+00
-1.26051879e+00 3.56156766e-01 -7.90750206e-01 -7.08066404e-01
9.43868577e-01 -3.74612927e-01 -3.96179914e-01 -1.25440001e-01
-1.64724052e-01 6.14001572e-01 9.96809125e-01 -1.38884866e+00
-1.11217344e+00 -3.09347838e-01 1.83287442e-01 4.58831757e-01
1.87356248e-01 -3.90944451e-01 -7.82706082e-01 -4.49927866e-01
2.19107587e-02 -1.27860570e+00 -3.93597603e-01 -1.03304230e-01
-1.42672369e-02 4.07763779e-01 1.12726498e+00 -6.31608009e-01
1.10696423e+00 -2.35198808e+00 1.61906138e-01 -2.50970632e-01
2.13395670e-01 -1.58591419e-01 -9.92890000e-02 8.52920488e-02
1.49006680e-01 -1.67316943e-01 6.51611686e-02 -4.67875838e-01
-1.71140954e-01 7.22872496e-01 -6.15670621e-01 4.87077087e-01
5.99034503e-02 8.51215303e-01 -1.22167921e+00 -1.97904631e-01
3.94518405e-01 4.77704883e-01 -4.75960523e-01 4.86398116e-02
-5.59926450e-01 5.86213112e-01 -5.44487655e-01 1.56201109e-01
8.24411690e-01 7.33143091e-02 1.86007231e-01 2.91713417e-01
-5.57848141e-02 4.92089204e-02 -1.44083500e+00 1.16183627e+00
-1.55530959e-01 1.00920713e+00 7.27941766e-02 -7.20718980e-01
4.88963485e-01 1.65304124e-01 6.17883384e-01 -5.54827988e-01
-3.01844686e-01 -2.40950331e-01 1.97679877e-01 -6.82914078e-01
6.91133261e-01 -3.54867190e-01 1.36671662e-01 1.85264662e-01
-2.55183578e-01 1.79376286e-02 1.83210388e-01 2.83888221e-01
1.19631970e+00 4.12426829e-01 -2.56080717e-01 2.62750015e-02
2.79636472e-01 -4.14741151e-02 1.10329223e+00 8.73087347e-01
-6.28611922e-01 6.50936902e-01 5.61763525e-01 -7.98506916e-01
-9.61955011e-01 -1.07474077e+00 3.03137928e-01 7.51128674e-01
6.78847790e-01 -3.47249776e-01 -6.19964302e-01 -3.48913819e-01
-1.79579444e-02 1.02542210e+00 -8.45203578e-01 -3.20418477e-01
-6.51772082e-01 -8.75610590e-01 -2.66829222e-01 3.22343767e-01
2.86878914e-01 -7.30331123e-01 -6.84407592e-01 3.71854484e-01
-6.23036265e-01 -1.49949265e+00 -4.93046820e-01 -2.16606289e-01
-7.77454019e-01 -8.65845025e-01 -2.14619443e-01 -1.33031324e-01
1.90248773e-01 6.68666422e-01 1.16884160e+00 -2.72883475e-01
-1.92518100e-01 6.25693917e-01 -2.80787349e-02 -4.64047223e-01
-4.80829030e-01 -5.24855375e-01 8.43170658e-02 6.14964068e-01
2.68976092e-01 -3.51963997e-01 -7.15911090e-01 3.97938460e-01
-6.89773917e-01 5.08576930e-01 1.69930220e-01 4.75043207e-01
5.79246879e-01 5.70854962e-01 -2.42272675e-01 -6.80945814e-01
-3.25664490e-01 -8.07271004e-01 -4.55405772e-01 2.47546732e-01
-1.09337926e-01 -1.19343728e-01 3.49596620e-01 -6.86464369e-01
-1.46778798e+00 3.65750015e-01 3.04820567e-01 -5.82367480e-01
-3.49320143e-01 1.24294378e-01 -2.48944908e-01 6.30863667e-01
2.76323073e-02 4.56637174e-01 -3.27054352e-01 -8.24474841e-02
2.14847073e-01 -3.62282172e-02 4.82428610e-01 -4.75950718e-01
8.19910407e-01 1.08733928e+00 -2.93420926e-02 -9.03588772e-01
-5.55075407e-01 -4.03462470e-01 -6.66159093e-01 -6.38401508e-01
9.96352553e-01 -1.38513350e+00 -2.23623127e-01 6.67478502e-01
-1.07414055e+00 -6.83345556e-01 -4.50301707e-01 5.83956063e-01
-8.34908485e-01 2.42236584e-01 -4.37580228e-01 -1.15187657e+00
6.85434699e-01 -1.29525125e+00 1.07397819e+00 1.10537164e-01
1.25941291e-01 -1.14130521e+00 1.44731253e-01 1.76153585e-01
1.91500634e-01 2.58758336e-01 6.83484852e-01 3.85054424e-02
-1.15479207e+00 -3.05876315e-01 2.26749793e-01 1.28957992e-02
5.49366921e-02 4.71207112e-01 -9.99257684e-01 1.77512858e-02
1.68462202e-01 4.53389227e-01 1.16766798e+00 8.92224848e-01
9.53285873e-01 -1.68853462e-01 -5.24259806e-01 6.73475206e-01
1.28112757e+00 4.56073880e-01 6.18944943e-01 1.08982682e-01
7.86612332e-01 9.56605673e-01 6.35330021e-01 5.33721745e-01
8.35209310e-01 6.32298112e-01 6.63237929e-01 2.39255250e-01
-1.15771376e-01 -5.41248083e-01 8.34018230e-01 5.18170953e-01
-6.63541332e-02 -6.18271649e-01 -9.83010948e-01 7.21162081e-01
-2.26564169e+00 -1.54710448e+00 -3.33925158e-01 2.05227900e+00
1.26978695e-01 3.79126757e-01 5.38635068e-02 -3.93142104e-01
6.94914997e-01 4.93899971e-01 -9.00492728e-01 1.52539715e-01
-3.98933023e-01 -6.72449529e-01 5.05242288e-01 7.41722107e-01
-1.00291240e+00 9.38216865e-01 6.64248562e+00 4.34367180e-01
-1.00716352e+00 4.64485250e-02 9.12490368e-01 -5.54790735e-01
-5.27976155e-01 3.10648054e-01 -1.01318645e+00 6.70581758e-01
1.19033742e+00 -2.44418517e-01 1.98782548e-01 5.82230568e-01
6.52654827e-01 -1.81946352e-01 -1.12177980e+00 7.18638599e-01
-1.80446118e-01 -1.23658371e+00 2.62614220e-01 4.21255603e-02
9.13813233e-01 2.27785036e-01 1.59778938e-01 3.39683145e-01
6.60623550e-01 -6.91665351e-01 1.26698267e+00 1.07445931e+00
-3.13732550e-02 -4.27061051e-01 4.75130677e-01 7.85342813e-01
-1.15780008e+00 -3.13298106e-01 -1.78670675e-01 -3.53411227e-01
5.76187909e-01 5.44258356e-01 -1.52713165e-01 1.96287036e-01
7.33291805e-01 1.17307949e+00 -4.69717652e-01 7.00710058e-01
-2.25843005e-02 7.89934993e-01 -1.97306827e-01 4.36391771e-01
2.38506183e-01 -4.19189155e-01 8.58472228e-01 1.02667701e+00
3.38955045e-01 1.17175549e-01 4.05481100e-01 5.98273873e-01
3.98409665e-01 -5.00801265e-01 -5.56330442e-01 1.10567205e-01
-5.39736683e-03 8.19795728e-01 -5.49514532e-01 -5.45472801e-01
-7.63508856e-01 8.71173084e-01 -1.22192837e-01 8.36271524e-01
-1.26198101e+00 5.62097371e-01 1.25152493e+00 4.06922340e-01
5.95710695e-01 -5.38118958e-01 -7.57637843e-02 -1.69685638e+00
9.62141231e-02 -5.64073682e-01 1.69762298e-01 -7.76905775e-01
-8.63155842e-01 4.59878385e-01 1.26006886e-01 -1.16601825e+00
-3.24476331e-01 -4.43037271e-01 -6.83573425e-01 6.48038328e-01
-1.63712513e+00 -1.13252878e+00 -2.87083626e-01 6.21696174e-01
1.04839849e+00 2.04918280e-01 3.51105034e-02 -1.12928294e-01
-6.61062717e-01 -2.50800550e-01 3.97828430e-01 -3.45277011e-01
5.12782276e-01 -1.19779015e+00 6.70009971e-01 1.23312259e+00
2.82053679e-01 9.01935473e-02 1.21198881e+00 -7.44519651e-01
-1.46876192e+00 -1.15582252e+00 7.91501105e-01 -7.92751670e-01
8.87212455e-01 -5.20621955e-01 -9.80349004e-01 7.93839276e-01
-1.23992957e-01 2.52106458e-01 3.06490481e-01 -2.66437620e-01
-1.37425050e-01 -8.91876519e-02 -5.75181425e-01 7.52025187e-01
1.12087214e+00 -3.83800179e-01 -2.87657142e-01 1.80438384e-01
7.90451348e-01 -4.14884061e-01 -1.79504350e-01 4.37369168e-01
6.65236533e-01 -1.17712986e+00 8.56129408e-01 -6.73522174e-01
5.67161977e-01 -4.47415560e-01 -3.88507724e-01 -1.07275999e+00
-4.93302524e-01 -4.05640870e-01 -6.11279368e-01 8.61543536e-01
3.24828416e-01 -2.93985724e-01 1.00880766e+00 1.02154005e+00
-1.44365858e-02 -3.59866768e-01 -8.32744956e-01 -6.30881190e-01
1.17357438e-02 -8.84190619e-01 6.29399002e-01 5.12791872e-01
-5.09900331e-01 5.52717550e-03 -7.78263807e-01 4.39050794e-01
5.59764385e-01 1.86075866e-01 1.03548849e+00 -9.92790222e-01
-4.91234124e-01 -1.91178352e-01 -6.77702546e-01 -1.33010352e+00
3.71855795e-01 -1.36376396e-01 4.54449266e-01 -1.32799804e+00
6.66842043e-01 -1.23958878e-01 -2.05392227e-01 -2.78828770e-01
-5.38978338e-01 -2.47967064e-01 3.12254339e-01 4.22973096e-01
-7.56536365e-01 6.66868329e-01 1.01448715e+00 -2.50005603e-01
-8.42643008e-02 2.02098504e-01 -3.30849439e-01 9.28386033e-01
4.96305168e-01 -3.55238467e-01 -7.78454185e-01 -6.94588065e-01
6.65916428e-02 2.42928952e-01 4.15583670e-01 -1.06414402e+00
1.43056631e-01 -7.67496228e-01 2.59598434e-01 -7.20695734e-01
4.86975193e-01 -9.53546405e-01 6.48416102e-01 5.01982331e-01
-1.29137531e-01 -1.07126869e-02 2.80985922e-01 1.37489033e+00
-2.32819363e-01 2.33410716e-01 5.23040533e-01 -1.93300128e-01
-1.17855895e+00 8.20665717e-01 -8.89594197e-01 -1.42385930e-01
1.28657019e+00 -3.84891748e-01 -1.12392575e-01 -9.73735571e-01
-1.04019964e+00 5.03444731e-01 6.55358195e-01 6.99832737e-01
3.96100760e-01 -1.00885701e+00 -5.95280766e-01 2.00210452e-01
-6.65160120e-02 -2.28183955e-01 8.43620896e-01 7.16898918e-01
-3.60952169e-01 2.57701367e-01 7.11811259e-02 -9.10259008e-01
-1.33281457e+00 8.05676103e-01 4.37497735e-01 -2.46774331e-01
-6.45667374e-01 6.32124424e-01 1.04967117e+00 2.57977867e-03
1.03434943e-01 -4.56847548e-01 -1.31439835e-01 -8.44125524e-02
6.53798938e-01 3.43666762e-01 -4.99151587e-01 -1.20616674e+00
-3.89884375e-02 4.16822910e-01 4.57879938e-02 -1.13139883e-01
1.12622428e+00 -8.20561171e-01 3.59278828e-01 1.10452843e+00
7.17076242e-01 -1.23447418e-01 -2.20745659e+00 -2.38214508e-01
-3.57535630e-02 -6.05227530e-01 1.40082240e-01 -3.08097750e-01
-9.93799806e-01 1.02665019e+00 5.08726597e-01 1.80149786e-02
8.94263983e-01 -3.44094932e-02 7.92889297e-01 3.80669311e-02
6.35631919e-01 -9.82483506e-01 -1.33315131e-01 7.32590199e-01
5.36174893e-01 -1.21891642e+00 -1.53258771e-01 -4.85918999e-01
-8.89328003e-01 9.97871101e-01 5.47583163e-01 3.61936241e-02
9.69215870e-01 4.05721068e-01 -8.17412212e-02 1.62962124e-01
-1.45717645e+00 -1.90988094e-01 2.38628998e-01 5.81060350e-01
5.39182965e-03 2.79023230e-01 4.68699723e-01 3.75431925e-01
-5.50462790e-02 -5.44096120e-02 8.19093585e-01 7.72485256e-01
-4.96919662e-01 -6.88925624e-01 -2.47477785e-01 3.06158125e-01
-2.76684463e-01 1.45149276e-01 -1.38695613e-01 6.36523247e-01
4.00145769e-01 9.17080998e-01 4.80872095e-01 -4.48733956e-01
2.80174345e-01 5.61787784e-02 3.80160123e-01 -4.21470225e-01
1.90279499e-01 2.04777107e-01 -1.06880836e-01 -8.71137857e-01
-4.65272367e-01 -1.41550088e+00 -7.72153199e-01 -4.81630683e-01
-6.80385232e-02 -1.76192194e-01 4.60582018e-01 1.07612348e+00
3.23247284e-01 6.31149471e-01 4.27417636e-01 -9.42125320e-01
-1.47981167e-01 -5.75443149e-01 -6.88087642e-01 4.54164863e-01
7.00291097e-01 -7.36592650e-01 -5.54138243e-01 6.11171544e-01] | [8.487957000732422, 0.12461459636688232] |
61721afc-5a94-4631-8b49-abcb94abcb8d | elaborative-rehearsal-for-zero-shot-action | 2108.02833 | null | https://arxiv.org/abs/2108.02833v2 | https://arxiv.org/pdf/2108.02833v2.pdf | Elaborative Rehearsal for Zero-shot Action Recognition | The growing number of action classes has posed a new challenge for video understanding, making Zero-Shot Action Recognition (ZSAR) a thriving direction. The ZSAR task aims to recognize target (unseen) actions without training examples by leveraging semantic representations to bridge seen and unseen actions. However, due to the complexity and diversity of actions, it remains challenging to semantically represent action classes and transfer knowledge from seen data. In this work, we propose an ER-enhanced ZSAR model inspired by an effective human memory technique Elaborative Rehearsal (ER), which involves elaborating a new concept and relating it to known concepts. Specifically, we expand each action class as an Elaborative Description (ED) sentence, which is more discriminative than a class name and less costly than manual-defined attributes. Besides directly aligning class semantics with videos, we incorporate objects from the video as Elaborative Concepts (EC) to improve video semantics and generalization from seen actions to unseen actions. Our ER-enhanced ZSAR model achieves state-of-the-art results on three existing benchmarks. Moreover, we propose a new ZSAR evaluation protocol on the Kinetics dataset to overcome limitations of current benchmarks and demonstrate the first case where ZSAR performance is comparable to few-shot learning baselines on this more realistic setting. We will release our codes and collected EDs at https://github.com/DeLightCMU/ElaborativeRehearsal. | ['Dong Huang', 'ShiZhe Chen'] | 2021-08-05 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Chen_Elaborative_Rehearsal_for_Zero-Shot_Action_Recognition_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Chen_Elaborative_Rehearsal_for_Zero-Shot_Action_Recognition_ICCV_2021_paper.pdf | iccv-2021-1 | ['zero-shot-action-recognition'] | ['computer-vision'] | [ 4.27291632e-01 -1.18975371e-01 -4.50203031e-01 -3.29063714e-01
-7.37349331e-01 -3.69797647e-01 7.76946723e-01 -5.88944033e-02
-4.60056394e-01 4.70489293e-01 8.25756431e-01 2.25164339e-01
1.41058536e-02 -4.37496036e-01 -8.08801830e-01 -4.93871778e-01
-2.24572551e-02 2.82624394e-01 4.77005303e-01 -1.87710822e-01
1.78989127e-01 2.41994828e-01 -1.82101941e+00 7.75937080e-01
3.75229031e-01 8.76599312e-01 2.21250892e-01 4.72876668e-01
3.98817956e-02 1.10989189e+00 -5.06854057e-01 -2.98954487e-01
2.64028549e-01 -7.15066135e-01 -1.06224263e+00 3.09814036e-01
5.73292017e-01 -6.47817731e-01 -6.17194176e-01 7.62828231e-01
2.72882402e-01 8.19858968e-01 5.23044407e-01 -1.38838482e+00
-7.90950775e-01 5.65773308e-01 -3.51934940e-01 5.10298252e-01
6.02834523e-01 3.29610348e-01 9.84406114e-01 -8.98642242e-01
8.75699222e-01 1.19394565e+00 3.64552200e-01 9.20178175e-01
-8.56368482e-01 -6.40195072e-01 5.48124909e-01 7.55048931e-01
-1.28513324e+00 -6.29421353e-01 5.99665701e-01 -3.93722773e-01
1.16409302e+00 2.09687635e-01 7.41816819e-01 1.52002978e+00
-2.55293608e-01 1.21787524e+00 5.92991590e-01 -7.91841447e-02
3.91389698e-01 -3.39944869e-01 2.31515467e-01 5.45305550e-01
-1.06280826e-01 -2.33547129e-02 -7.79947460e-01 2.96734601e-01
6.46051228e-01 6.38704062e-01 -4.91789132e-01 -5.60161054e-01
-1.26389802e+00 6.89919949e-01 3.92371356e-01 2.72437572e-01
-5.20337462e-01 3.84015083e-01 6.41367197e-01 2.11074501e-02
1.69460118e-01 3.48646998e-01 -3.07565540e-01 -5.37968278e-01
-6.64323211e-01 9.52262878e-02 3.52490664e-01 1.01300049e+00
6.17130518e-01 -8.85463227e-03 -4.84326124e-01 7.44851887e-01
-1.82322070e-01 4.19125855e-01 7.21542239e-01 -1.15960300e+00
3.28320026e-01 4.99649554e-01 -7.48481527e-02 -7.28677690e-01
-8.47893432e-02 -2.87550598e-01 -4.11211073e-01 -1.63438842e-01
1.66333422e-01 3.80464554e-01 -1.02298880e+00 1.75883794e+00
2.07168162e-01 6.74501240e-01 2.27797076e-01 1.13180292e+00
9.72723961e-01 7.18311608e-01 5.11976421e-01 -1.76333889e-01
1.26101029e+00 -1.33887601e+00 -5.38283348e-01 -2.21868843e-01
8.16101253e-01 -1.91662878e-01 1.20717120e+00 2.94808477e-01
-8.09288561e-01 -7.09028304e-01 -7.81039238e-01 -1.05615474e-01
-3.88253957e-01 -1.10305749e-01 5.52766979e-01 5.60172945e-02
-7.79555738e-01 5.77006757e-01 -1.06509686e+00 -5.73346019e-01
6.75839603e-01 4.30445187e-03 -6.48334205e-01 -3.54902118e-01
-1.19355071e+00 5.84022343e-01 6.15435123e-01 -2.86280036e-01
-1.29976904e+00 -7.40918875e-01 -1.17065179e+00 8.34721997e-02
9.49288249e-01 -5.46398997e-01 1.24893224e+00 -1.11293197e+00
-1.34014857e+00 7.77149081e-01 -1.65251702e-01 -8.02920043e-01
1.40338570e-01 -6.10595107e-01 -4.86094952e-01 6.03886068e-01
9.29462314e-02 8.73876214e-01 8.29059720e-01 -1.03171504e+00
-6.01130784e-01 -2.90937036e-01 3.62555414e-01 3.21875840e-01
-3.75097096e-01 -7.09637925e-02 -5.47492146e-01 -7.96656609e-01
-6.59318641e-02 -8.17027628e-01 1.40630260e-01 -6.88416064e-02
-4.84484062e-02 -2.23594338e-01 8.42719316e-01 -6.41707778e-01
1.29006195e+00 -2.40706825e+00 4.68157470e-01 -4.55256075e-01
1.86341524e-01 5.45419872e-01 -4.46604252e-01 5.84374428e-01
-2.69823819e-01 -1.45410702e-01 -1.72215924e-01 -2.34279871e-01
-7.68685201e-03 3.31123710e-01 -4.42290604e-01 1.96501777e-01
9.18266997e-02 1.10937619e+00 -1.23859608e+00 -2.83626914e-01
3.61570448e-01 5.15494347e-01 -6.23206675e-01 1.98024958e-01
-2.47309253e-01 3.33696455e-01 -3.43429118e-01 7.76986480e-01
1.07387528e-01 -3.54590684e-01 -5.38263544e-02 -3.16783369e-01
2.49386579e-01 1.30811632e-01 -8.81945252e-01 1.98492634e+00
-2.66331345e-01 4.59503800e-01 -5.92941880e-01 -1.24215174e+00
6.73099220e-01 2.24254355e-01 4.88281816e-01 -8.23821187e-01
1.01113934e-02 -2.61808425e-01 -3.32898587e-01 -6.40430570e-01
3.58136594e-01 -1.73399612e-01 -1.85727239e-01 3.75781089e-01
3.81939441e-01 2.92989969e-01 3.03651780e-01 4.42573398e-01
1.36757255e+00 5.53781033e-01 5.84488750e-01 2.86099136e-01
4.20731455e-01 5.28879575e-02 7.18559146e-01 5.89195490e-01
-5.58169067e-01 6.59091771e-01 1.94274485e-01 -5.77387750e-01
-5.80796063e-01 -1.23016477e+00 3.51772726e-01 1.43109632e+00
3.78009290e-01 -7.71455288e-01 -6.43634617e-01 -9.41654921e-01
-1.23628020e-01 9.32727993e-01 -6.98453665e-01 -5.61066031e-01
-5.23886383e-01 -4.74406183e-02 2.73053288e-01 9.96663153e-01
7.03644335e-01 -1.23683596e+00 -8.56485188e-01 8.51678476e-02
-4.11472321e-01 -1.20830274e+00 -5.68237305e-01 -1.75795659e-01
-6.40884578e-01 -1.23205948e+00 -5.62292874e-01 -3.61136883e-01
3.62246811e-01 6.74007654e-01 1.14845586e+00 8.59662965e-02
-3.42706621e-01 9.25491512e-01 -9.76955116e-01 -2.05696985e-01
-9.26364660e-02 -3.70945394e-01 6.55294210e-02 1.96488202e-01
7.49770522e-01 -3.07249725e-01 -8.29576552e-01 3.19635719e-01
-9.80998695e-01 2.90048182e-01 5.81953228e-01 5.91074347e-01
7.14396656e-01 -6.20028153e-02 5.42840004e-01 -6.44426584e-01
4.23308957e-04 -6.30114257e-01 -5.58882132e-02 2.52251446e-01
-6.21745512e-02 -4.42447811e-02 4.08703059e-01 -5.35793483e-01
-1.06679952e+00 1.71991974e-01 1.02512725e-01 -8.91930342e-01
-4.23189193e-01 3.59376997e-01 -1.46077022e-01 3.78838003e-01
4.71916705e-01 4.98618990e-01 -1.17390797e-01 -3.58067393e-01
5.38245976e-01 5.80385685e-01 6.68408394e-01 -4.35464919e-01
4.23150510e-01 6.60011947e-01 -3.58398348e-01 -8.27177823e-01
-1.18561518e+00 -8.16153049e-01 -7.26191342e-01 -4.37788159e-01
1.20472109e+00 -1.05378413e+00 -3.91430676e-01 2.93961436e-01
-8.40829372e-01 -5.33461154e-01 -5.99793494e-01 7.05975294e-01
-9.29117143e-01 4.10090864e-01 -5.31500638e-01 -5.16084194e-01
-9.59823579e-02 -9.92560923e-01 1.17803359e+00 6.03228137e-02
-4.42187339e-01 -7.02411592e-01 8.36696289e-03 7.56408632e-01
1.06310122e-01 2.54054159e-01 4.86458600e-01 -8.11244845e-01
-6.06097996e-01 -1.51585728e-01 8.68926942e-03 3.89694750e-01
8.49226788e-02 -3.63097638e-01 -7.30420172e-01 -2.46136412e-01
-6.10298328e-02 -7.06006944e-01 1.16643310e+00 1.10722244e-01
1.33229554e+00 -2.07494706e-01 -2.35241562e-01 5.19309938e-01
1.07273567e+00 2.59178489e-01 8.66482794e-01 3.15708369e-01
7.03863621e-01 3.62451464e-01 9.75582421e-01 5.52254617e-01
3.23302567e-01 7.62540758e-01 3.81485254e-01 2.87998825e-01
-4.26086307e-01 -4.58991587e-01 6.12758934e-01 4.82805818e-01
-2.85931677e-01 -2.96105862e-01 -8.01513553e-01 4.74012673e-01
-2.02893448e+00 -1.60058260e+00 2.47079223e-01 2.10809302e+00
6.17927015e-01 -4.53877822e-02 2.09202170e-01 -1.00376666e-01
7.07198799e-01 3.51441830e-01 -6.68970823e-01 1.00797810e-01
1.01083435e-01 1.61661670e-01 1.28350303e-01 1.48610547e-01
-1.24856937e+00 1.18914378e+00 5.17286968e+00 7.74139524e-01
-8.36619258e-01 2.53669649e-01 2.91800201e-01 -4.13757116e-01
1.15999579e-01 -5.56854755e-02 -8.21061492e-01 4.03156489e-01
8.25823903e-01 -1.20655999e-01 3.14525336e-01 9.49333191e-01
1.60507530e-01 -6.46183938e-02 -1.39386976e+00 1.21703768e+00
5.93908250e-01 -1.36019027e+00 5.33970833e-01 -2.96413153e-01
5.22408307e-01 -1.53655589e-01 -1.93708882e-01 8.55950832e-01
8.40750784e-02 -8.22835207e-01 6.53828979e-01 6.78023219e-01
6.51274323e-01 -4.35819566e-01 4.87979054e-01 2.02571541e-01
-1.34502518e+00 -4.02867228e-01 -2.98653424e-01 -1.70206204e-01
2.65931845e-01 -2.33060986e-01 -4.16724741e-01 4.52981591e-01
7.74302602e-01 1.37359810e+00 -5.54111958e-01 8.48896086e-01
-2.79241860e-01 4.86843109e-01 2.14343503e-01 3.12325954e-01
3.99566472e-01 6.28876388e-02 4.83666718e-01 9.81563270e-01
2.81246483e-01 6.10329449e-01 3.06832194e-01 5.59938371e-01
-5.66375367e-02 -1.57637283e-01 -5.51415622e-01 -3.62945914e-01
4.61349070e-01 9.22302127e-01 -6.23282433e-01 -6.72810733e-01
-6.76428437e-01 1.37258482e+00 3.07249337e-01 4.39085156e-01
-9.90909457e-01 -2.31476147e-02 7.81623125e-01 2.24874601e-01
5.76313615e-01 -3.95552665e-02 5.36351562e-01 -1.38374972e+00
-1.17018290e-01 -9.10787821e-01 8.69900048e-01 -1.06397867e+00
-1.02699280e+00 3.54819745e-01 3.62520903e-01 -1.50766635e+00
-3.04809153e-01 -3.99356544e-01 -3.18875849e-01 9.86837447e-02
-1.26746953e+00 -1.08016622e+00 -6.45802200e-01 8.12736928e-01
1.29064524e+00 -2.24601209e-01 7.50729918e-01 1.59490719e-01
-4.68116045e-01 3.93299758e-01 -2.80159265e-01 1.70640320e-01
7.08560526e-01 -8.84658813e-01 8.92044678e-02 7.63671398e-01
4.38873589e-01 4.72211361e-01 5.63501954e-01 -7.57922590e-01
-1.39947736e+00 -1.25970709e+00 3.87193441e-01 -7.11947799e-01
7.60089159e-01 -2.83932865e-01 -1.19137955e+00 1.04317904e+00
4.12703343e-02 2.42943019e-01 8.15822661e-01 -1.58828288e-01
-6.38092041e-01 5.09103909e-02 -5.93135476e-01 5.98281622e-01
1.69829619e+00 -5.44310689e-01 -1.02892852e+00 3.13355744e-01
8.12385023e-01 -1.77164823e-01 -6.64004445e-01 3.96184623e-01
4.30752307e-01 -1.00601459e+00 1.10472524e+00 -9.46001112e-01
5.77753305e-01 -3.37499470e-01 -3.80348414e-01 -1.23616290e+00
-3.17881912e-01 -2.31749669e-01 -5.16390026e-01 9.76316631e-01
-9.68329459e-02 -1.61491439e-01 6.98919058e-01 4.70825404e-01
-4.42957282e-01 -6.53099179e-01 -6.43852532e-01 -1.15286338e+00
-3.77416402e-01 -3.41771334e-01 4.29584742e-01 9.56412733e-01
1.22254536e-01 3.62439334e-01 -5.00508368e-01 -8.31390619e-02
4.50691432e-01 1.63650617e-01 8.58933032e-01 -8.98037553e-01
-5.08768260e-01 -3.01281333e-01 -8.58542800e-01 -1.05795181e+00
4.02069747e-01 -9.14148927e-01 -1.21243251e-02 -1.57496524e+00
5.89326441e-01 1.42428920e-01 -5.11731863e-01 7.09487796e-01
-2.12233886e-01 3.40471208e-01 4.47939485e-01 2.76931673e-01
-1.34157753e+00 8.97932112e-01 1.00638950e+00 -2.46864066e-01
-1.09736845e-01 -2.53903419e-01 -4.56005305e-01 6.74348652e-01
6.04312658e-01 -1.73388764e-01 -8.65110040e-01 -3.62836361e-01
-2.71149248e-01 7.89828449e-02 6.87280178e-01 -1.20686495e+00
5.77473640e-02 -2.83834308e-01 2.57640272e-01 -4.97454107e-01
5.72626591e-01 -6.93034291e-01 1.64802656e-01 4.27679718e-01
-6.17294014e-01 -2.15326384e-01 3.16818021e-02 9.51250732e-01
-3.30339253e-01 -1.11380808e-01 6.40154839e-01 -2.07002848e-01
-1.71120441e+00 4.78205383e-01 -1.31124154e-01 2.94399709e-01
1.45708203e+00 -4.39225674e-01 -4.86955076e-01 -4.80105728e-01
-1.04171503e+00 1.95019960e-01 4.94945616e-01 7.59269118e-01
9.87549245e-01 -1.42012048e+00 -5.02842426e-01 1.49335181e-02
6.19391978e-01 -3.23341399e-01 6.86249495e-01 8.04218471e-01
-1.79538414e-01 2.66380966e-01 -4.32708830e-01 -4.34943527e-01
-1.41871417e+00 1.02143943e+00 9.29980651e-02 1.97483316e-01
-1.07710171e+00 7.94354439e-01 5.98613858e-01 8.43133554e-02
3.54049772e-01 4.39023152e-02 -3.28981161e-01 -4.25838009e-02
9.83667195e-01 4.48307455e-01 -2.62200743e-01 -8.96517992e-01
-3.92561227e-01 3.69882584e-01 -1.37139067e-01 1.95141897e-01
1.41890550e+00 -1.05545007e-01 4.08967763e-01 5.54897010e-01
1.16992867e+00 -5.81482887e-01 -1.57580221e+00 -4.21812117e-01
1.07412972e-03 -6.74782634e-01 -6.83523044e-02 -6.99594676e-01
-9.71467495e-01 8.82196188e-01 4.36220109e-01 -3.28460157e-01
1.12541425e+00 3.97876024e-01 9.55570936e-01 5.26091814e-01
3.75756949e-01 -1.14308095e+00 7.79384255e-01 3.38786632e-01
1.05775058e+00 -1.32488561e+00 -9.69673991e-02 -3.38790894e-01
-1.18336189e+00 9.20019388e-01 8.05941999e-01 4.66162688e-04
3.52438569e-01 -2.61042893e-01 -2.15128481e-01 -3.70179862e-01
-8.72246385e-01 -5.08230269e-01 2.91425198e-01 6.14859879e-01
1.64067730e-01 -4.87968214e-02 -1.95760205e-01 7.18374431e-01
2.97619075e-01 2.55889684e-01 4.43124473e-01 1.11071813e+00
-5.18793285e-01 -7.32005477e-01 9.13551524e-02 3.81637096e-01
-1.43854260e-01 5.59623502e-02 -3.10947776e-01 8.50856960e-01
-4.37346287e-02 7.20971406e-01 2.91155428e-01 -5.37620962e-01
3.82234424e-01 2.84341812e-01 4.28926378e-01 -9.86889422e-01
-3.73360515e-02 -8.56629610e-02 1.05329268e-02 -1.21049058e+00
-6.88255489e-01 -8.19269061e-01 -1.65020406e+00 -2.34022550e-02
1.45497769e-01 -1.68370660e-02 2.11309239e-01 1.13049531e+00
5.26065528e-01 5.39954543e-01 2.34282956e-01 -7.38702595e-01
-5.84374428e-01 -7.26296067e-01 -4.72610056e-01 9.57396448e-01
1.26979202e-01 -1.14332843e+00 -3.43173712e-01 4.41986233e-01] | [8.678227424621582, 0.8118352293968201] |
a3e6babf-3c18-4cc4-acfa-baacdad87d62 | a-financial-service-chatbot-based-on-deep | 2003.04987 | null | https://arxiv.org/abs/2003.04987v1 | https://arxiv.org/pdf/2003.04987v1.pdf | A Financial Service Chatbot based on Deep Bidirectional Transformers | We develop a chatbot using Deep Bidirectional Transformer models (BERT) to handle client questions in financial investment customer service. The bot can recognize 381 intents, and decides when to say "I don't know" and escalates irrelevant/uncertain questions to human operators. Our main novel contribution is the discussion about uncertainty measure for BERT, where three different approaches are systematically compared on real problems. We investigated two uncertainty metrics, information entropy and variance of dropout sampling in BERT, followed by mixed-integer programming to optimize decision thresholds. Another novel contribution is the usage of BERT as a language model in automatic spelling correction. Inputs with accidental spelling errors can significantly decrease intent classification performance. The proposed approach combines probabilities from masked language model and word edit distances to find the best corrections for misspelled words. The chatbot and the entire conversational AI system are developed using open-source tools, and deployed within our company's intranet. The proposed approach can be useful for industries seeking similar in-house solutions in their specific business domains. We share all our code and a sample chatbot built on a public dataset on Github. | ['Yuxin Chen', 'Hussain Zaidi', 'Shi Yu'] | 2020-02-17 | null | null | null | null | ['spelling-correction'] | ['natural-language-processing'] | [-3.58005315e-02 4.02369052e-01 2.37259731e-01 -5.67608356e-01
-9.86875057e-01 -7.68291235e-01 2.85380006e-01 -1.97988600e-01
-4.38441873e-01 8.53534758e-01 5.66360131e-02 -5.08248389e-01
-2.69519717e-01 -6.48777783e-01 -2.31644213e-01 -5.05072653e-01
4.21469092e-01 1.13909388e+00 2.35892922e-01 -5.69157004e-01
9.55955386e-01 2.60177404e-01 -9.27512586e-01 8.46605301e-01
7.60233462e-01 8.72656822e-01 5.03691912e-01 8.56842995e-01
-5.71868002e-01 1.62874198e+00 -9.72702205e-01 -1.30179369e+00
1.92143753e-01 -2.70092636e-01 -1.37883687e+00 -4.67057496e-01
-3.57686907e-01 -2.56378591e-01 1.68034378e-02 1.03568697e+00
4.13350523e-01 5.09863235e-02 6.42605841e-01 -1.55658603e+00
-8.17986488e-01 1.27905083e+00 -6.94100484e-02 3.05803925e-01
2.98028648e-01 4.00189459e-01 1.42823839e+00 -6.25138044e-01
3.74757618e-01 1.25931692e+00 7.25884855e-01 6.01205468e-01
-9.15163755e-01 -5.60197473e-01 -5.21639884e-02 7.62460291e-01
-1.02818131e+00 -1.20252438e-01 7.07832932e-01 -5.57742178e-01
1.50337207e+00 4.74928796e-01 6.26720041e-02 1.17870915e+00
5.83665431e-01 7.88200080e-01 9.67998981e-01 -3.58807176e-01
1.68139309e-01 9.07462537e-01 3.41951162e-01 4.26667184e-01
-2.28471115e-01 -2.96152920e-01 -4.63212758e-01 -2.47547701e-01
1.49301514e-01 -8.75015035e-02 2.74052262e-01 3.42684150e-01
-6.80310011e-01 1.09014606e+00 -3.73360026e-03 4.51748192e-01
-2.73029149e-01 8.02375451e-02 4.61607426e-01 7.32512891e-01
4.57611859e-01 6.90792203e-01 -5.34723878e-01 -8.05941284e-01
-3.59130740e-01 3.30743194e-01 1.43026149e+00 1.35203564e+00
4.70371425e-01 -4.09396499e-01 -6.62947476e-01 9.94574726e-01
3.00790220e-01 1.02566946e-02 3.97899061e-01 -9.60697293e-01
8.28958511e-01 4.93002653e-01 3.23212922e-01 -8.99395287e-01
-1.91894293e-01 -1.54370526e-02 -1.61504105e-01 -6.09964803e-02
4.89186496e-01 -3.08721036e-01 -1.81833178e-01 1.22164178e+00
-1.68412402e-01 -3.13952267e-01 -1.55604288e-01 6.05384946e-01
7.04204381e-01 4.55971658e-01 -2.97452211e-01 -9.20643210e-02
1.37713623e+00 -1.11859012e+00 -1.13272977e+00 -2.64452249e-01
6.21736169e-01 -1.12796772e+00 1.13548708e+00 7.28796005e-01
-1.01817274e+00 1.05569370e-01 -5.22089362e-01 -1.86284930e-01
-3.65066677e-01 -6.16172180e-02 4.85065013e-01 8.11123669e-01
-6.78463876e-01 7.73643255e-01 -2.98767626e-01 -1.33880302e-01
1.57050252e-01 2.79850304e-01 2.05409780e-01 1.36886895e-01
-1.31050837e+00 1.47154808e+00 2.28164135e-03 2.68493861e-01
-7.08168149e-01 -5.67106068e-01 -3.80114824e-01 1.55909881e-01
5.58971107e-01 -1.54617026e-01 2.02661610e+00 -6.64594591e-01
-1.99892259e+00 7.33554184e-01 -4.33192812e-02 -6.66621625e-01
8.28961968e-01 -8.66724104e-02 -2.31655966e-02 -3.86015177e-01
6.91173524e-02 9.00703892e-02 5.90247095e-01 -7.31130838e-01
-7.03125238e-01 -2.52480328e-01 1.46616355e-01 1.36463530e-02
2.30828077e-02 6.00654781e-01 6.28303364e-02 -4.42231834e-01
-4.69823539e-01 -6.32956982e-01 -1.20073788e-01 -7.12942243e-01
-4.01166886e-01 -4.58584756e-01 6.08628869e-01 -1.00716794e+00
1.31911206e+00 -1.57316959e+00 1.24422433e-02 -2.44789105e-02
-6.77519813e-02 2.04695389e-01 3.10808700e-02 7.75029957e-01
3.14865351e-01 2.38287762e-01 -6.07237965e-02 -3.63444746e-01
3.10342163e-01 7.65313208e-02 -4.04238194e-01 3.43394876e-02
2.68800437e-01 8.84461999e-01 -7.09194779e-01 -5.26496351e-01
1.88679069e-01 -1.82621367e-03 -7.35945165e-01 4.58666116e-01
-4.36309934e-01 1.15131445e-01 -2.17864275e-01 6.35874033e-01
6.60424888e-01 1.03211425e-01 6.00287952e-02 2.88236618e-01
-1.12517878e-01 8.75256121e-01 -9.40561831e-01 1.14701200e+00
-9.23716009e-01 7.01806307e-01 1.85695410e-01 -9.90653634e-01
1.22219372e+00 1.84654370e-01 2.82516032e-02 -3.22117418e-01
5.71208954e-01 2.98672974e-01 1.22264184e-01 -7.59229302e-01
6.33914948e-01 -6.09298684e-02 -1.84940010e-01 7.15696931e-01
2.27511168e-01 -4.45109427e-01 2.59812027e-01 1.05889447e-01
1.35117686e+00 -1.90423831e-01 1.49688020e-01 -3.83282937e-02
5.23940504e-01 1.03945360e-02 5.36380947e-01 1.08388078e+00
-3.48720908e-01 4.11738843e-01 1.07534671e+00 -4.26581532e-01
-8.33533704e-01 -5.53032458e-01 9.37036425e-02 1.28536296e+00
-2.07828641e-01 -3.04539561e-01 -9.04086411e-01 -9.70392585e-01
-1.87291875e-01 1.50589180e+00 -3.59383434e-01 -6.47234544e-02
-4.56301004e-01 -6.22251987e-01 5.82473814e-01 1.71197414e-01
4.36082691e-01 -1.59964967e+00 -3.76540273e-01 4.77909207e-01
-6.03288472e-01 -9.73168015e-01 -5.74027896e-01 4.10877585e-01
-3.45240623e-01 -8.66003931e-01 -3.19195151e-01 -5.54852664e-01
-3.89496498e-02 -1.90127686e-01 1.12950122e+00 5.17825931e-02
-4.66764957e-01 1.10088788e-01 -6.72342420e-01 -7.90394485e-01
-8.76306355e-01 2.50609815e-01 -3.40675145e-01 -9.50124711e-02
9.54605520e-01 -5.41009486e-01 -1.41058862e-01 6.72231078e-01
-3.83984625e-01 -4.15808827e-01 3.00756782e-01 9.43875968e-01
-2.98742265e-01 -2.65010536e-01 6.94913566e-01 -1.08298147e+00
1.31093216e+00 -6.39259279e-01 -6.39647603e-01 4.83584046e-01
-6.10813260e-01 1.63383022e-01 4.05823946e-01 -2.73466110e-01
-1.43377137e+00 -2.10837811e-01 -4.41481441e-01 1.25090346e-01
1.17625967e-01 2.31088489e-01 3.13366987e-02 5.33411093e-02
6.40990496e-01 -1.92977771e-01 -1.78575851e-02 -5.48103988e-01
1.96194589e-01 1.41062295e+00 8.83084759e-02 -4.30726469e-01
1.91524476e-01 -9.84964445e-02 -6.64057136e-01 -3.96626532e-01
-6.73690200e-01 -4.82195318e-01 -4.69713479e-01 -2.68655479e-01
5.49125075e-01 -2.14464217e-01 -1.56693125e+00 4.11157310e-01
-1.82379377e+00 -2.10538089e-01 -1.96988404e-01 7.92036578e-02
-6.39466882e-01 2.90528059e-01 -8.67816567e-01 -1.33347201e+00
-5.55058300e-01 -1.19389868e+00 6.51113331e-01 1.56772003e-01
-5.63090503e-01 -7.36317933e-01 -1.06606288e-02 8.74216497e-01
5.51079214e-01 -5.56083262e-01 1.00985122e+00 -1.36579943e+00
-4.79845762e-01 -3.68177921e-01 -7.51833245e-02 7.76543856e-01
-6.42775074e-02 5.11622354e-02 -9.32277143e-01 3.77996355e-01
4.16410059e-01 -2.43459612e-01 5.06101489e-01 9.19158682e-02
1.05403757e+00 -6.98839664e-01 -7.12507144e-02 -8.30406174e-02
8.71024370e-01 6.92998588e-01 7.93573618e-01 4.07152951e-01
2.32798010e-01 9.99996066e-01 8.13357651e-01 7.86803007e-01
2.86533743e-01 7.45522499e-01 5.68291187e-01 8.54573667e-01
4.12719071e-01 -1.91974983e-01 5.30816317e-01 8.37281227e-01
1.28405496e-01 -5.08185327e-01 -9.03304756e-01 5.77151477e-01
-2.02986598e+00 -1.20599616e+00 -3.04524601e-01 1.86085296e+00
8.24851096e-01 1.74331844e-01 4.56360765e-02 -3.78375314e-02
9.82090592e-01 -3.53443056e-01 -2.78248072e-01 -1.17643929e+00
2.47656494e-01 1.39676824e-01 6.13856494e-01 9.55892146e-01
-5.84119618e-01 1.07181954e+00 5.76676035e+00 1.05694461e+00
-5.55766642e-01 5.61964691e-01 7.98951507e-01 -1.22305378e-01
-4.44536418e-01 6.05882742e-02 -9.01693702e-01 6.46772504e-01
8.77086878e-01 -2.88469315e-01 7.95510590e-01 1.12494659e+00
2.12511569e-01 -1.85087472e-01 -1.28642309e+00 8.65855634e-01
1.09255813e-01 -1.53466249e+00 -5.23662984e-01 -3.50217372e-01
3.61943662e-01 -1.42583266e-01 7.56667480e-02 6.76151156e-01
9.05996025e-01 -1.04084277e+00 7.04437494e-01 6.31172657e-01
3.18950601e-02 -8.20993543e-01 1.24693537e+00 7.37208366e-01
-3.93219769e-01 -5.03190279e-01 -5.02609611e-01 -2.81175017e-01
2.26244494e-01 4.90568072e-01 -1.50974953e+00 2.29271725e-01
8.70915115e-01 1.84391081e-01 -2.73112148e-01 8.15943599e-01
-1.48612782e-01 5.58508933e-01 -1.97442740e-01 -9.24810052e-01
2.41699487e-01 -3.27780813e-01 6.03708923e-01 1.32080531e+00
2.45351240e-01 -1.38067566e-02 -5.51270127e-01 1.40744507e+00
9.81713682e-02 1.82248518e-01 -4.59270388e-01 -1.17308885e-01
6.68119311e-01 1.17861986e+00 -4.97650266e-01 1.63095832e-01
-6.05002046e-02 1.00391698e+00 3.36549073e-01 -1.28987148e-01
-9.77427363e-01 -8.93596113e-01 5.20968795e-01 -1.79041415e-01
1.09551832e-01 3.69233221e-01 -7.07149267e-01 -7.80986905e-01
2.30263487e-01 -1.14376175e+00 3.50709260e-01 -7.77552009e-01
-1.54622853e+00 7.21338868e-01 -3.88916098e-02 -7.92641282e-01
-4.24293935e-01 -7.29459047e-01 -9.98973787e-01 9.60820675e-01
-9.38275695e-01 -7.38618970e-01 3.03019639e-02 1.79613963e-01
8.79937410e-01 -6.04548633e-01 6.21446908e-01 3.31102043e-01
-5.07493138e-01 6.42061472e-01 1.03431098e-01 -2.96699833e-02
6.43447101e-01 -1.15501773e+00 4.48665470e-01 3.93892676e-01
4.59610783e-02 5.43528259e-01 9.09686446e-01 -5.15800714e-01
-9.51765597e-01 -6.60226047e-01 1.61821401e+00 -9.98377979e-01
8.06607902e-01 -6.57954395e-01 -6.16804123e-01 9.07226264e-01
4.25696582e-01 -7.83544958e-01 3.42342585e-01 4.84725535e-02
-6.20051958e-02 -9.40990746e-02 -1.73441160e+00 5.89193165e-01
6.78292871e-01 -4.94233370e-01 -8.47179055e-01 8.76182437e-01
9.94519651e-01 -2.70826608e-01 -3.16200376e-01 -1.79178476e-01
1.82068408e-01 -1.20291901e+00 3.92117918e-01 -7.31992662e-01
4.49647576e-01 3.10457617e-01 4.19927910e-02 -1.15542984e+00
-1.29398689e-01 -1.23340392e+00 4.58741426e-01 1.70241868e+00
8.00884545e-01 -7.24059939e-01 7.22845256e-01 1.03955555e+00
1.03359185e-02 -5.75566411e-01 -1.15614927e+00 -4.71452117e-01
4.61132564e-02 -9.24254477e-01 6.42931104e-01 3.67466211e-01
6.20370746e-01 3.18974733e-01 -5.72960377e-01 -2.50393212e-01
9.58184674e-02 -2.33218864e-01 4.04920548e-01 -9.32091355e-01
-3.30952764e-01 -2.94917017e-01 -1.11042827e-01 -7.88519204e-01
3.04652989e-01 -8.04513872e-01 3.36913615e-01 -1.12170315e+00
6.00164644e-02 -4.54611421e-01 1.83454584e-02 3.49793464e-01
3.07553768e-01 -1.85981899e-01 2.82958657e-01 -2.09043860e-01
-2.50702202e-01 4.20289159e-01 7.12856174e-01 -2.93955714e-01
-1.86633930e-01 7.29441524e-01 -8.89578760e-01 7.17788517e-01
9.75085735e-01 -9.86098707e-01 -2.25312367e-01 -1.86855450e-01
6.39826834e-01 1.78978533e-01 2.00913876e-01 -3.77646476e-01
4.10737365e-01 -2.48802856e-01 -7.10632324e-01 -6.19014382e-01
3.29706550e-01 -1.02933550e+00 -1.21488139e-01 3.01720649e-01
-8.41499865e-01 2.69830257e-01 -6.92607388e-02 3.64463329e-01
-6.62925914e-02 -1.34029627e+00 5.59070706e-01 -5.00829458e-01
-8.20301846e-02 -3.07574809e-01 -9.04946148e-01 1.66364416e-01
1.10120952e+00 1.04362071e-01 -2.85392612e-01 -8.03073883e-01
-5.64318359e-01 1.49244338e-01 -3.38140160e-01 5.93891203e-01
3.44679683e-01 -8.63206744e-01 -6.88037634e-01 -1.70467064e-01
-1.58875912e-01 -4.08583492e-01 1.08594388e-01 8.42044711e-01
-8.12176466e-01 6.02744639e-01 -5.34201190e-02 -2.46211499e-01
-1.37668538e+00 3.46472681e-01 4.99035418e-01 -5.56897342e-01
-1.04387358e-01 1.43702054e+00 -4.99410093e-01 -8.32571149e-01
6.01891935e-01 -5.38205266e-01 -2.25330904e-01 1.57738939e-01
4.24774587e-01 6.52379632e-01 4.78991598e-01 -8.19555894e-02
-4.23250914e-01 -1.05857365e-01 -3.88743103e-01 -4.19577509e-01
1.31975770e+00 -2.11231738e-01 -4.05775070e-01 5.11622608e-01
1.05201280e+00 -1.48333058e-01 -6.02917969e-01 -1.52014300e-01
4.16366249e-01 -6.63024247e-01 -3.09362978e-01 -1.10687673e+00
-6.19628012e-01 1.09227693e+00 2.76072204e-01 4.98260498e-01
5.48745453e-01 -1.42706290e-01 6.08382702e-01 1.03793621e+00
3.83859009e-01 -1.53428733e+00 1.94008842e-01 1.04001915e+00
1.22811043e+00 -1.25636005e+00 -4.69238669e-01 -2.91912526e-01
-1.51419163e+00 1.09465504e+00 5.85822284e-01 1.04062781e-01
3.93713623e-01 5.09284019e-01 2.06126675e-01 -7.65028298e-02
-1.14239061e+00 1.98216423e-01 -3.47867131e-01 6.73858523e-01
5.60734093e-01 -1.88941836e-01 -5.16701996e-01 1.17032444e+00
-5.33974886e-01 -7.27066100e-02 9.07362342e-01 6.16939127e-01
-2.67185718e-01 -1.19290102e+00 -2.26824194e-01 3.59274149e-01
-6.76793039e-01 -3.99378419e-01 -8.54314506e-01 3.87889475e-01
-4.37916443e-02 1.46333206e+00 -7.48241991e-02 -6.65078819e-01
3.09874177e-01 4.91159648e-01 7.61833042e-02 -7.51304924e-01
-1.41006446e+00 -6.55202031e-01 5.12016296e-01 -3.13144326e-01
1.25760972e-01 -7.48606205e-01 -8.10670257e-01 -6.23025119e-01
-7.32198834e-01 5.07088006e-01 7.91539788e-01 1.06339014e+00
2.06519023e-01 3.28484714e-01 7.65258074e-01 -3.69736910e-01
-1.17711616e+00 -1.29746544e+00 -3.31427127e-01 -3.18844207e-02
-3.38137522e-02 -3.74857306e-01 -6.24324679e-01 -2.49444649e-01] | [12.658655166625977, 7.721740245819092] |
5e48eca7-a333-4b45-ba0c-24c5ac379dc9 | kernel-based-distributed-q-learning-a | 2302.10434 | null | https://arxiv.org/abs/2302.10434v1 | https://arxiv.org/pdf/2302.10434v1.pdf | Kernel-Based Distributed Q-Learning: A Scalable Reinforcement Learning Approach for Dynamic Treatment Regimes | In recent years, large amounts of electronic health records (EHRs) concerning chronic diseases, such as cancer, diabetes, and mental disease, have been collected to facilitate medical diagnosis. Modeling the dynamic properties of EHRs related to chronic diseases can be efficiently done using dynamic treatment regimes (DTRs), which are a set of sequential decision rules. While Reinforcement learning (RL) is a widely used method for creating DTRs, there is ongoing research in developing RL algorithms that can effectively handle large amounts of data. In this paper, we present a novel approach, a distributed Q-learning algorithm, for generating DTRs. The novelties of our research are as follows: 1) From a methodological perspective, we present a novel and scalable approach for generating DTRs by combining distributed learning with Q-learning. The proposed approach is specifically designed to handle large amounts of data and effectively generate DTRs. 2) From a theoretical standpoint, we provide generalization error bounds for the proposed distributed Q-learning algorithm, which are derived within the framework of statistical learning theory. These bounds quantify the relationships between sample size, prediction accuracy, and computational burden, providing insights into the performance of the algorithm. 3) From an applied perspective, we demonstrate the effectiveness of our proposed distributed Q-learning algorithm for DTRs by applying it to clinical cancer treatments. The results show that our algorithm outperforms both traditional linear Q-learning and commonly used deep Q-learning in terms of both prediction accuracy and computation cost. | ['Shao-Bo Lin', 'Shaojie Tang', 'Yao Wang', 'Di Wang'] | 2023-02-21 | null | null | null | null | ['medical-diagnosis'] | ['medical'] | [-4.64563631e-02 1.30524859e-01 -6.11457825e-01 -1.60760835e-01
-1.15994155e+00 -1.69782862e-01 -6.34294078e-02 4.97774839e-01
-3.27757150e-01 1.07743895e+00 2.08854303e-01 -5.06617904e-01
-6.27631605e-01 -9.96533096e-01 -6.29614711e-01 -7.10423589e-01
-1.73895106e-01 5.33165216e-01 -3.55457842e-01 3.13310735e-02
-1.35803968e-01 3.88296127e-01 -1.02860224e+00 1.74142197e-02
1.35938799e+00 9.57852840e-01 -1.08571723e-01 6.11018956e-01
6.00507036e-02 1.00477648e+00 -5.29376328e-01 -3.85634869e-01
1.37199908e-01 -6.60338819e-01 -6.07833147e-01 -5.95669076e-02
-1.39149472e-01 -6.60650074e-01 -2.43266493e-01 8.39176357e-01
9.28880513e-01 -8.11256170e-02 6.44233346e-01 -1.25099397e+00
-4.05121595e-01 5.62617302e-01 -2.69817680e-01 -1.42401889e-01
3.62485439e-01 2.92737693e-01 9.67417061e-01 -2.07100406e-01
5.51194549e-01 1.03897524e+00 8.00193608e-01 7.16548324e-01
-1.04690635e+00 -7.08670318e-01 -1.49345607e-01 1.28336176e-01
-1.12265265e+00 -1.73902169e-01 5.86149871e-01 -4.15323079e-01
6.42907619e-01 3.48944776e-02 8.80076170e-01 8.78474176e-01
7.49761105e-01 1.09385300e+00 1.05948675e+00 -4.48755652e-01
9.74726439e-01 -3.15066166e-02 -1.22675635e-01 3.95676613e-01
3.49879652e-01 5.16955793e-01 -2.47401580e-01 -7.80055106e-01
7.27344513e-01 3.74446690e-01 2.22210065e-01 -5.09392977e-01
-7.67196715e-01 1.14130259e+00 2.34159246e-01 8.88822526e-02
-8.66141319e-01 2.37016767e-01 6.03977740e-01 2.84136325e-01
7.23349869e-01 1.30736619e-01 -7.84519851e-01 -1.92801818e-01
-7.98853576e-01 5.18092513e-01 7.38428771e-01 8.06847513e-01
3.50620478e-01 -3.84391844e-02 -5.61928034e-01 6.19185269e-01
3.07186663e-01 8.20796609e-01 5.04369080e-01 -9.98776615e-01
2.93070853e-01 4.98663217e-01 4.25077736e-01 -6.52848840e-01
-5.39569914e-01 -2.68721551e-01 -1.12553740e+00 -4.33630258e-01
2.34788701e-01 -6.70668840e-01 -3.70483398e-01 1.73702037e+00
7.11461604e-01 3.16814750e-01 3.54861826e-01 4.85655189e-01
4.70592588e-01 4.85453874e-01 4.60555524e-01 -7.99594402e-01
1.12408197e+00 -4.18410838e-01 -1.02830803e+00 4.28910792e-01
9.88126397e-01 -1.99448273e-01 8.45761478e-01 4.44281101e-01
-1.11074686e+00 -1.87916338e-01 -4.09636229e-01 4.31206018e-01
1.23591483e-01 -1.58174381e-01 6.90495253e-01 9.50555563e-01
-9.87719417e-01 5.92195272e-01 -9.43049908e-01 -3.35607171e-01
7.44258583e-01 3.95340472e-01 3.21242720e-01 -3.08068305e-01
-1.37583828e+00 4.73126769e-01 1.15617327e-01 -1.71603248e-01
-8.63442957e-01 -9.32700753e-01 -6.49095297e-01 1.64214298e-01
4.51119483e-01 -1.13070488e+00 1.45730340e+00 -8.98401260e-01
-1.74592578e+00 1.91788673e-01 -7.65312240e-02 -6.15354419e-01
6.38175428e-01 -1.43822402e-01 -2.45708928e-01 2.61096507e-01
-1.03003159e-01 2.41976902e-01 6.23996556e-01 -6.76783979e-01
-6.19573891e-01 -4.93087471e-01 -3.08412075e-01 1.04309134e-01
-5.61019182e-01 -9.28262398e-02 2.72312075e-01 -6.23754025e-01
-5.75523853e-01 -8.49790871e-01 -8.40472221e-01 -3.20450291e-02
-1.56488374e-01 -5.35626173e-01 3.06290001e-01 -4.93160516e-01
1.43493438e+00 -1.85035884e+00 -8.89944583e-02 2.20277160e-01
2.62433678e-01 2.88985282e-01 -1.21157013e-01 7.06506073e-01
2.93579668e-01 6.94731399e-02 -1.34872481e-01 6.78246319e-02
-8.15887749e-02 1.02957055e-01 -1.60809129e-01 2.73364425e-01
9.45653841e-02 1.20186579e+00 -1.22081041e+00 -5.87190628e-01
8.19553509e-02 2.87533045e-01 -8.55171621e-01 3.59128416e-01
-3.75582367e-01 5.14858484e-01 -8.79997909e-01 4.91776735e-01
5.48632443e-01 -5.41461229e-01 6.15226150e-01 2.28057936e-01
4.21118498e-01 -7.63258487e-02 -1.04100871e+00 1.23247182e+00
-4.52599376e-01 -1.75189689e-01 -2.05466390e-01 -1.27184749e+00
7.59002507e-01 6.62458777e-01 1.26182199e+00 -7.60325670e-01
-5.88542446e-02 1.72186375e-01 -9.57617313e-02 -6.52212918e-01
-3.34951244e-02 -3.82885247e-01 -1.80723757e-01 7.09585488e-01
-2.22267151e-01 1.45792484e-01 -8.77771601e-02 8.34729075e-02
1.38834155e+00 -1.04097165e-02 8.23634386e-01 -5.69342487e-02
1.42002985e-01 5.05291671e-02 8.61087382e-01 8.34291577e-01
-5.91291666e-01 -2.99323946e-01 6.17009521e-01 -5.84895015e-01
-8.76802802e-01 -1.02238250e+00 -1.88045576e-01 8.22749078e-01
-3.00591737e-01 -2.43755832e-01 -6.45938516e-01 -7.52655625e-01
4.13055629e-01 4.81400341e-01 -4.43888307e-01 -1.92870796e-01
-2.57232934e-01 -1.01318872e+00 4.62567419e-01 5.20157039e-01
2.28061080e-01 -1.32219303e+00 -7.08301187e-01 5.77473581e-01
-3.77933905e-02 -7.97133863e-01 -3.88094962e-01 -1.65002957e-01
-1.48258066e+00 -1.07791460e+00 -7.98637152e-01 -4.09184843e-01
4.63358670e-01 -5.01327999e-02 9.77445483e-01 -1.74632713e-01
-3.55207801e-01 6.90428078e-01 -3.07359517e-01 -5.97912908e-01
-6.10920668e-01 1.83131509e-02 3.01906113e-02 -1.28937531e-02
3.48629147e-01 -2.35173911e-01 -1.00323558e+00 1.87143609e-02
-1.11202502e+00 -3.07132274e-01 6.76823497e-01 1.06238127e+00
6.52765512e-01 7.29160085e-02 1.46046579e+00 -1.38237834e+00
8.93981338e-01 -8.03783119e-01 -6.46933556e-01 3.87742639e-01
-9.62493181e-01 1.07862718e-01 8.42482269e-01 -3.74600053e-01
-9.20058310e-01 1.30223542e-01 -2.09354967e-01 -2.71442205e-01
-9.34062004e-02 6.73820138e-01 1.05878368e-01 3.06709111e-01
6.79667711e-01 2.60300428e-01 2.98233151e-01 -1.88989595e-01
3.99098963e-01 9.88305748e-01 -4.13570732e-01 -4.92274761e-01
2.88730562e-01 3.26708734e-01 1.27550811e-01 -4.49806720e-01
-7.59236395e-01 -3.79462868e-01 -1.45692140e-01 -1.95527002e-02
4.23204660e-01 -1.05526268e+00 -1.09135199e+00 3.51471454e-01
-6.25660777e-01 -5.55736542e-01 -4.87524152e-01 4.66084123e-01
-9.08578753e-01 1.99848861e-01 -8.78376365e-01 -1.06470978e+00
-8.74187469e-01 -8.31407189e-01 9.96802330e-01 1.76116973e-02
-1.42779201e-01 -1.37262714e+00 4.41006184e-01 2.69857943e-01
3.60982448e-01 2.89753854e-01 1.06737709e+00 -6.04964316e-01
-3.61232966e-01 -7.11516440e-02 5.13611175e-02 8.58165249e-02
3.22094500e-01 -2.25263819e-01 -5.73978782e-01 -7.60096192e-01
-1.28833294e-01 -5.09460688e-01 2.57010967e-01 9.45752442e-01
1.31134570e+00 -5.15614450e-01 -4.33295906e-01 2.49332473e-01
1.63737643e+00 4.47269320e-01 5.57505369e-01 -3.31562869e-02
3.49235743e-01 4.54979807e-01 8.46642256e-01 1.14910066e+00
4.72612679e-01 4.51601058e-01 7.35754073e-02 -2.05710396e-01
4.00960803e-01 -3.18965644e-01 3.52573305e-01 8.83903325e-01
2.07717478e-01 -1.32493963e-02 -8.89019668e-01 4.46327567e-01
-2.11257529e+00 -8.08722675e-01 2.79285282e-01 2.39668107e+00
1.11333466e+00 -2.61452526e-01 6.25506818e-01 -6.22617751e-02
3.87209713e-01 -5.07799327e-01 -1.04636025e+00 -4.11512136e-01
3.34634513e-01 4.80522722e-01 4.80500847e-01 8.94569308e-02
-8.26493144e-01 5.96599817e-01 7.13227320e+00 9.04095948e-01
-9.76889491e-01 1.90959290e-01 8.95395398e-01 -4.65750545e-02
-3.21959019e-01 -2.58189768e-01 -6.25487387e-01 4.58644122e-01
1.48721564e+00 -3.50450099e-01 3.23638141e-01 7.62863398e-01
7.83595085e-01 1.46830291e-01 -9.88180697e-01 8.75495434e-01
-5.04434645e-01 -1.18623769e+00 -2.94523779e-02 2.65387088e-01
1.03004456e+00 -2.22166255e-01 5.79957142e-02 4.71154273e-01
7.07067013e-01 -9.37450528e-01 1.53060900e-02 4.84305143e-01
8.88586879e-01 -9.80997086e-01 8.02144289e-01 6.25812650e-01
-8.59576821e-01 -5.82356572e-01 -2.43857726e-01 7.31823295e-02
-1.05187424e-01 9.58153665e-01 -8.41987133e-01 6.11787558e-01
4.93620247e-01 7.76579857e-01 -1.39647767e-01 1.14992917e+00
5.22330366e-02 8.68689418e-01 7.03244656e-02 -2.60548115e-01
-8.64366442e-02 -2.06263229e-01 -9.29556265e-02 8.84200633e-01
4.09883201e-01 9.79248360e-02 2.99918860e-01 6.72322035e-01
-6.03062101e-02 5.32523096e-01 -6.56040490e-01 -1.13638580e-01
7.01251507e-01 1.05099857e+00 -3.58756512e-01 -4.63232666e-01
-3.67569149e-01 4.30281579e-01 2.13100582e-01 2.73775965e-01
-7.12177336e-01 -1.78097770e-01 4.04299647e-01 4.52652052e-02
2.19441175e-01 3.57257843e-01 -2.32845992e-01 -1.06198931e+00
-2.80359089e-01 -1.24285638e+00 6.68175995e-01 -2.52741367e-01
-1.62769926e+00 8.34744051e-03 -4.90332022e-02 -1.36239636e+00
-7.02043235e-01 -2.30366707e-01 -1.41271919e-01 4.88573074e-01
-1.57848358e+00 -8.46760869e-01 1.05588675e-01 7.79701054e-01
2.33265951e-01 -4.08665016e-02 1.04433179e+00 2.84712195e-01
-6.80869162e-01 7.39205897e-01 8.44353914e-01 4.39482741e-03
5.90924859e-01 -1.11469948e+00 -1.21368587e-01 1.29085854e-01
-2.57517964e-01 3.69146198e-01 3.36338013e-01 -7.56891668e-01
-1.60171592e+00 -1.33237553e+00 7.23664582e-01 -2.36730538e-02
4.20996577e-01 -1.07176147e-01 -6.40920639e-01 3.68685037e-01
-1.83545768e-01 1.16656579e-01 1.12457716e+00 1.81519628e-01
1.51094377e-01 -5.28774142e-01 -1.42430520e+00 3.88111562e-01
6.99599802e-01 -4.08666432e-01 -1.62861630e-01 5.74282289e-01
5.89369118e-01 -1.96981832e-01 -1.36122954e+00 3.41331780e-01
5.69050431e-01 -5.27333438e-01 8.40764523e-01 -9.10191298e-01
5.19730151e-01 3.14610898e-01 2.07717642e-01 -1.46566212e+00
-1.84808448e-01 -8.42429876e-01 -4.53195006e-01 5.56751430e-01
1.92127973e-02 -7.98063874e-01 9.32209015e-01 6.09629869e-01
4.46275890e-01 -1.05513299e+00 -1.00944698e+00 -7.85484850e-01
4.71685946e-01 -1.66945979e-01 6.22450948e-01 7.62884438e-01
1.95104331e-01 4.69467640e-02 -4.77959335e-01 -1.66594937e-01
7.40095794e-01 1.17189817e-01 5.95119596e-01 -1.11112046e+00
-5.23034275e-01 -8.94342214e-02 -1.24117143e-01 -7.54558802e-01
-3.78938392e-02 -6.12448752e-01 -2.16985434e-01 -1.46187472e+00
5.58601916e-01 -6.18237615e-01 -6.73736930e-01 4.91074324e-01
-4.56702530e-01 -3.69966805e-01 3.91793326e-02 2.78958350e-01
-7.05576956e-01 8.12104166e-01 1.36286056e+00 6.75206110e-02
-5.00875175e-01 3.51544887e-01 -6.29956126e-01 2.30412528e-01
7.48091221e-01 -7.07492828e-01 -5.81440210e-01 1.46734670e-01
3.24139237e-01 9.68010306e-01 1.18100464e-01 -5.18254519e-01
1.20625934e-02 -5.20810068e-01 2.24887133e-01 -3.40601861e-01
-3.15013945e-01 -7.15350330e-01 8.69941339e-02 1.13361716e+00
-6.04183972e-01 -6.31636456e-02 9.77493674e-02 8.24453235e-01
-5.21012880e-02 1.05149485e-01 6.14576697e-01 -1.81255601e-02
-1.34453708e-02 6.09684467e-01 -4.76181686e-01 2.26843640e-01
1.29222238e+00 1.74626529e-01 9.23909470e-02 -6.06359184e-01
-6.85942650e-01 4.59971011e-01 -1.22662798e-01 -7.04936981e-02
7.81068921e-01 -1.37169731e+00 -7.92839587e-01 9.23834741e-03
-4.30775769e-02 -1.85889006e-01 4.52248394e-01 8.00624549e-01
-2.81705171e-01 4.31153119e-01 -6.36807084e-02 -4.46980923e-01
-8.93838882e-01 7.64650881e-01 3.08326006e-01 -8.01039100e-01
-5.87164581e-01 3.05955112e-02 -6.12866580e-02 -5.95104277e-01
6.99174330e-02 -3.16250563e-01 -4.65055518e-02 -9.16450750e-03
4.32611644e-01 3.98518980e-01 4.30740742e-03 1.92225948e-01
-1.15432970e-01 2.70510346e-01 -2.22827774e-02 7.26147816e-02
1.53069615e+00 2.74194907e-02 4.39503156e-02 2.71121383e-01
1.01636624e+00 -3.57125044e-01 -1.26152027e+00 -4.03154969e-01
1.69365674e-01 -1.33643717e-01 7.07777974e-04 -7.51390934e-01
-1.05776811e+00 6.70005620e-01 9.53463972e-01 8.18150043e-02
1.38669837e+00 -2.60243118e-01 1.03905475e+00 3.35188955e-01
4.68670934e-01 -1.17820418e+00 2.23476544e-01 7.55867660e-02
2.95110941e-01 -1.25525916e+00 3.20989303e-02 -3.74710783e-02
-6.17869258e-01 8.62396836e-01 2.43824512e-01 -1.04822814e-01
8.65273714e-01 1.46515384e-01 -1.42737525e-03 2.41731144e-02
-1.08393168e+00 1.35487884e-01 -1.46060009e-02 5.75657904e-01
4.25687283e-01 3.77720475e-01 -5.50948381e-01 6.41866267e-01
2.50883996e-01 8.63350987e-01 3.62756819e-01 1.03546739e+00
-1.78167149e-01 -1.26040614e+00 -3.10265571e-01 6.72566473e-01
-5.59852839e-01 -1.02569750e-02 4.40809131e-02 4.71243441e-01
-1.90001592e-01 9.78256106e-01 -1.38425052e-01 8.14676564e-03
1.46318287e-01 9.41993403e-06 4.07183051e-01 -5.11875570e-01
-5.90786815e-01 1.56901449e-01 -3.39232951e-01 -5.98842919e-01
-5.59918344e-01 -6.42572105e-01 -1.17068613e+00 -2.33074725e-01
-2.71772563e-01 1.45689964e-01 4.29335296e-01 9.92909908e-01
6.87841833e-01 5.59947670e-01 9.80390012e-01 -3.24945003e-02
-1.37872708e+00 -6.90141380e-01 -7.97215521e-01 4.75383282e-01
2.46980801e-01 -3.04749638e-01 5.75089734e-03 -1.68842360e-01] | [4.073030948638916, 2.8453516960144043] |
3fb12a04-4f7f-4996-9e54-17cbf01973ed | leveraging-pre-trained-audioldm-for-sound | 2303.03857 | null | https://arxiv.org/abs/2303.03857v2 | https://arxiv.org/pdf/2303.03857v2.pdf | Leveraging Pre-trained AudioLDM for Text to Sound Generation: A Benchmark Study | Deep neural networks have recently achieved breakthroughs in sound generation with text prompts. Despite their promising performance, current text-to-sound generation models face issues on small-scale datasets (e.g., overfitting), significantly limiting their performance. In this paper, we investigate the use of pre-trained AudioLDM, the state-of-the-art model for text-to-audio generation, as the backbone for sound generation. Our study demonstrates the advantages of using pre-trained models for text-to-sound generation, especially in data-scarcity scenarios. In addition, experiments show that different training strategies (e.g., training conditions) may affect the performance of AudioLDM on datasets of different scales. To facilitate future studies, we also evaluate various text-to-sound generation systems on several frequently used datasets under the same evaluation protocols, which allow fair comparisons and benchmarking of these methods on the common ground. | ['Wenwu Wang', 'Mark D. Plumbley', 'Xubo Liu', 'Jinhua Liang', 'Haohe Liu', 'Yi Yuan'] | 2023-03-07 | null | null | null | null | ['audio-generation'] | ['audio'] | [-2.36733574e-02 -2.46242285e-01 1.61068022e-01 -1.05101146e-01
-1.11693990e+00 -4.91453528e-01 6.12187326e-01 -2.80955344e-01
-1.27926961e-01 7.33565807e-01 5.27924836e-01 -2.12675884e-01
1.95571765e-01 -9.31132555e-01 -5.93205869e-01 -4.64023262e-01
1.83948442e-01 3.69974285e-01 6.39555231e-02 -2.62798160e-01
3.52344848e-02 1.30999088e-01 -1.88623989e+00 4.72292811e-01
6.93258584e-01 8.17855418e-01 3.07025313e-01 1.03700936e+00
-3.33885550e-01 9.17757690e-01 -1.38280809e+00 -3.12177449e-01
1.17037393e-01 -6.37256920e-01 -7.89977968e-01 -4.56571430e-01
5.97382247e-01 -5.68377435e-01 -3.31517071e-01 6.14684880e-01
1.39379549e+00 2.86719412e-01 4.79727298e-01 -1.46943235e+00
-7.06204414e-01 9.83043194e-01 7.38186762e-02 1.18705351e-03
5.34631908e-01 2.81253248e-01 1.23139656e+00 -9.63307142e-01
4.21369880e-01 1.22538412e+00 6.94276989e-01 9.15605545e-01
-1.01619494e+00 -9.74037170e-01 -3.82794231e-01 -2.96736415e-02
-1.39735639e+00 -9.33969975e-01 7.11420715e-01 -1.86976001e-01
1.02922237e+00 4.74307656e-01 4.08683211e-01 1.48393154e+00
-2.04852726e-02 7.47973204e-01 8.85181367e-01 -5.60965598e-01
1.87969461e-01 -1.35919675e-01 -4.31732059e-01 1.67549238e-01
7.88263455e-02 3.00693274e-01 -1.21979058e+00 -3.41145039e-01
5.80936611e-01 -7.11443126e-01 -2.47221574e-01 6.95679724e-01
-1.15088880e+00 8.12969506e-01 5.93134947e-02 3.47722173e-01
-2.80403554e-01 5.02549946e-01 5.17548740e-01 1.77729473e-01
6.14638150e-01 7.29357660e-01 -3.36581916e-01 -6.47192597e-01
-1.34627807e+00 7.76784122e-01 1.02694511e+00 1.01497567e+00
3.62907976e-01 6.75907671e-01 -6.85717285e-01 1.05142033e+00
2.22869858e-01 6.37638748e-01 7.45312631e-01 -9.74956930e-01
6.84832752e-01 -1.96798086e-01 1.18636293e-02 -7.48743951e-01
-3.12414289e-01 -5.19799531e-01 -8.92573833e-01 -1.58872589e-01
2.38051474e-01 -6.16241932e-01 -8.02544117e-01 1.78650963e+00
-9.08052698e-02 5.01093626e-01 9.86894593e-02 7.87323773e-01
1.44842434e+00 9.25605118e-01 3.29715721e-02 1.68373343e-02
1.01335120e+00 -1.15394437e+00 -8.00613821e-01 -1.07305229e-01
4.70932931e-01 -1.19680345e+00 1.38486719e+00 3.05424988e-01
-1.22324252e+00 -8.55609596e-01 -6.68894470e-01 2.96836104e-02
-3.03339660e-01 1.42238915e-01 4.60358232e-01 6.40666366e-01
-1.28369415e+00 6.16736770e-01 -4.04840618e-01 -3.63452673e-01
7.45691210e-02 -5.45198619e-02 1.91859841e-01 3.97067219e-01
-1.62210596e+00 3.64167184e-01 1.92612931e-01 -1.42768845e-01
-1.16180897e+00 -1.01451266e+00 -4.70923036e-01 1.51398435e-01
-4.60488498e-02 -7.92154610e-01 1.91457939e+00 -3.76484066e-01
-1.78746641e+00 3.45398307e-01 1.92600545e-02 -4.67748761e-01
4.29261595e-01 -4.44392204e-01 -5.23186743e-01 -8.61623418e-03
1.73006773e-01 1.02556241e+00 8.25445354e-01 -1.14171207e+00
-7.01034248e-01 3.40290219e-01 1.25929490e-01 1.45094678e-01
-4.69317287e-01 1.02993198e-01 -2.26187631e-01 -9.26493824e-01
-4.84856606e-01 -7.58885860e-01 -7.49319568e-02 -5.73154867e-01
-5.87066710e-01 -3.87492418e-01 6.19568825e-01 -4.69300479e-01
1.57453525e+00 -2.02170753e+00 -3.48811865e-01 -3.01781178e-01
1.81231555e-02 3.69772434e-01 -6.13069057e-01 9.23474252e-01
1.52008548e-01 5.02030313e-01 1.59531489e-01 -7.61706293e-01
2.45648384e-01 6.10286407e-02 -7.16961086e-01 -3.71792346e-01
1.45575926e-01 7.55520403e-01 -9.42059934e-01 -6.03352189e-01
1.29419148e-01 4.05672342e-01 -5.98657191e-01 5.84518731e-01
-4.70335633e-01 4.47629064e-01 -3.49110931e-01 3.55883688e-01
3.30683678e-01 2.84664631e-02 -3.19213510e-01 2.45508309e-02
-1.33032724e-01 7.22926617e-01 -1.02435839e+00 1.81903279e+00
-9.72341716e-01 9.55600023e-01 -1.88480541e-01 -3.86510402e-01
8.47683430e-01 9.03135061e-01 3.59004050e-01 -6.82429016e-01
-5.49937272e-03 4.49580997e-01 2.36134455e-02 -4.28686768e-01
9.86598313e-01 -1.44237876e-01 1.12015128e-01 6.80973947e-01
3.00041974e-01 -5.98799765e-01 4.67268705e-01 1.31896749e-01
9.88017261e-01 -2.12567076e-02 -2.59901762e-01 1.61459167e-02
-1.68425590e-02 -2.86877185e-01 1.69228509e-01 1.18063915e+00
4.36450802e-02 1.16546774e+00 9.17000100e-02 -2.01308802e-01
-9.86819267e-01 -8.65142584e-01 -1.34640440e-01 1.26901293e+00
-4.15535688e-01 -7.80488074e-01 -9.97720540e-01 -4.70451742e-01
-2.43177280e-01 9.47454512e-01 -1.65053979e-01 2.82516554e-02
-4.54406232e-01 -7.56229103e-01 1.37072146e+00 3.79566908e-01
3.31053734e-01 -1.58359146e+00 -6.37849033e-01 6.95477009e-01
-5.80904484e-01 -1.03265607e+00 -2.48016760e-01 -7.47398436e-02
-6.57243311e-01 -4.51942444e-01 -9.39073861e-01 -5.78251123e-01
4.83810045e-02 3.17680568e-01 1.62456095e+00 2.40162387e-01
-3.04461718e-02 2.75101036e-01 -7.64123976e-01 -8.51016223e-01
-9.71716821e-01 5.75700343e-01 7.03861937e-02 -4.63204294e-01
3.35764550e-02 -7.69839227e-01 -6.39771581e-01 2.22626045e-01
-1.03603184e+00 1.47193179e-01 3.36875349e-01 6.31827831e-01
2.66649723e-01 1.86162546e-01 1.12613380e+00 -5.58845103e-01
1.31870306e+00 -3.85949373e-01 -2.56978214e-01 5.48154563e-02
-4.27659601e-01 -2.24090770e-01 9.55904186e-01 -4.07018334e-01
-1.07497728e+00 -6.88827038e-01 -6.15380645e-01 -3.10858220e-01
-3.25704396e-01 6.45480990e-01 1.20182268e-01 4.06595707e-01
9.36820447e-01 2.88967162e-01 -5.39623559e-01 -6.73137665e-01
5.14844477e-01 1.18287647e+00 3.95217657e-01 -8.16658080e-01
6.50841773e-01 -1.17204823e-01 -2.45036915e-01 -6.98181093e-01
-8.80047321e-01 -8.51277038e-02 -1.07829794e-02 -1.55641034e-01
4.18106407e-01 -1.09805262e+00 -2.57418543e-01 7.37434089e-01
-1.35762012e+00 -5.76244056e-01 -5.34395158e-01 3.02303106e-01
-5.77573359e-01 -5.96896047e-03 -7.30896890e-01 -8.12460840e-01
-9.32602346e-01 -8.96850109e-01 1.33260906e+00 2.54076689e-01
-4.25007492e-01 -8.90986204e-01 4.36525583e-01 2.47660473e-01
7.88812160e-01 -9.53190029e-02 6.46040201e-01 -6.31814301e-01
-2.08349034e-01 -8.44405815e-02 2.85230484e-02 4.18179005e-01
2.45452926e-01 2.98792064e-01 -1.47997272e+00 -1.66339114e-01
-3.70962709e-01 -6.02406442e-01 6.34968042e-01 4.15951997e-01
1.28174508e+00 -3.30644846e-01 2.90242489e-02 4.25724894e-01
8.71255219e-01 1.67585552e-01 5.60946047e-01 3.19800526e-02
6.06154144e-01 4.03245777e-01 5.70009828e-01 7.01574802e-01
3.23923320e-01 6.56945169e-01 6.27016053e-02 -2.29905590e-01
-7.87745774e-01 -5.89339972e-01 4.04280573e-01 1.29231918e+00
-5.74113168e-02 -8.98529351e-01 -8.10676575e-01 5.95680833e-01
-1.54326439e+00 -8.81595075e-01 -1.27699420e-01 2.01071525e+00
1.34950888e+00 -1.05204813e-01 1.29979610e-01 2.93896288e-01
4.94933337e-01 4.27688509e-01 -7.82790855e-02 -3.94129753e-01
-4.13906202e-02 6.14202678e-01 -2.80401140e-01 2.42415786e-01
-8.12371731e-01 1.10675561e+00 7.18386793e+00 1.28901184e+00
-1.38366544e+00 1.28325209e-01 4.74990636e-01 -3.48259896e-01
-3.30863118e-01 -2.91348785e-01 -8.82065058e-01 6.18030667e-01
1.32992649e+00 -2.57209033e-01 7.23948002e-01 5.64380169e-01
4.86732006e-01 2.19726786e-01 -1.04840338e+00 8.06627810e-01
-1.24667250e-01 -1.37343848e+00 3.11316222e-01 -1.73093557e-01
9.99398530e-01 2.29038358e-01 9.03019682e-02 4.83710527e-01
4.17455107e-01 -9.08773899e-01 1.07399130e+00 1.36542439e-01
1.10562396e+00 -5.94656289e-01 6.93599164e-01 2.21504793e-01
-1.22889793e+00 2.91551590e-01 -3.96423936e-01 -1.35310724e-01
3.65194291e-01 8.47630084e-01 -1.10076058e+00 6.59047604e-01
6.16520762e-01 3.74779701e-01 -4.62331146e-01 9.67918217e-01
-2.35766575e-01 1.10386324e+00 -2.84961313e-01 -1.06173187e-01
2.10346952e-01 3.11806500e-01 5.39899588e-01 1.28716290e+00
8.41844738e-01 -4.13642853e-01 -1.31008253e-01 1.08110309e+00
-2.64667630e-01 2.91056931e-01 -6.57983184e-01 -3.78960043e-01
8.12437832e-01 1.19655120e+00 -2.77639300e-01 -3.19909006e-01
-2.60853291e-01 5.46157718e-01 1.19190037e-01 5.85754216e-01
-7.86125481e-01 -5.18387616e-01 5.85652709e-01 1.77041844e-01
-1.76936150e-01 -9.84223709e-02 -1.64984524e-01 -7.72217929e-01
-1.84345797e-01 -1.18056357e+00 1.67793155e-01 -1.02716267e+00
-1.25268590e+00 9.18467820e-01 6.63438961e-02 -1.33832061e+00
-6.93433225e-01 -7.37063587e-02 -1.02326012e+00 8.87091339e-01
-1.42281580e+00 -1.15133345e+00 -3.12278777e-01 3.09614837e-01
8.04667652e-01 -3.89363408e-01 1.02529299e+00 4.46426988e-01
-4.71989602e-01 8.54789197e-01 2.20474601e-02 -4.56784554e-02
1.00456727e+00 -1.15306580e+00 1.03146446e+00 6.77598000e-01
6.55334294e-01 4.49385643e-01 7.18135178e-01 -4.77340370e-01
-1.02568781e+00 -1.29551041e+00 9.74449694e-01 -3.78510803e-01
5.52712083e-01 -5.23586631e-01 -6.26429021e-01 2.05375671e-01
4.80428606e-01 -4.24840003e-01 7.83807933e-01 2.99974620e-01
4.51877415e-02 -1.14133477e-01 -7.43012428e-01 8.53996694e-01
1.05840456e+00 -4.52193350e-01 -1.58881098e-01 2.61632413e-01
1.13760519e+00 -5.77770770e-01 -8.04825783e-01 2.46162191e-01
6.03733242e-01 -9.13114250e-01 8.38696659e-01 -5.55528462e-01
7.47100532e-01 -8.23902711e-02 -1.72235057e-01 -1.74879014e+00
-6.35146126e-02 -1.05803549e+00 -2.33327627e-01 1.76241398e+00
4.96116310e-01 -4.27207947e-01 4.93771642e-01 2.01089323e-01
-2.88468301e-01 -5.78818977e-01 -9.62354362e-01 -8.43015850e-01
2.68360972e-01 -7.64333904e-01 1.15809572e+00 6.26843512e-01
-2.21019685e-01 5.10620058e-01 -6.41637623e-01 -1.59561098e-01
1.11440875e-01 -3.78630273e-02 1.19006479e+00 -9.52825546e-01
-3.31028670e-01 -4.62089747e-01 1.32647291e-01 -9.52780068e-01
1.96158290e-01 -7.71398604e-01 3.73293608e-01 -1.49006081e+00
-5.74174039e-02 -6.38993025e-01 -2.55866110e-01 4.55079556e-01
-3.58452171e-01 3.26216668e-01 4.64529544e-01 1.50143459e-01
-3.20338488e-01 7.23309338e-01 1.47086060e+00 1.03592081e-02
-3.35680574e-01 1.04815222e-01 -7.31148481e-01 3.62686247e-01
1.13103676e+00 -6.85504913e-01 -5.73567212e-01 -7.38531888e-01
2.16816366e-01 7.65662938e-02 1.92077950e-01 -1.29116583e+00
2.14659031e-02 -8.69069695e-02 -7.12870359e-02 -5.23487628e-01
3.01812083e-01 -2.59134799e-01 2.01735333e-01 4.96522970e-02
-5.66253483e-01 -1.01288117e-01 4.13477510e-01 4.59382832e-02
-4.36219931e-01 -4.84230429e-01 3.86065274e-01 4.59022783e-02
-1.05936371e-01 1.72210023e-01 -4.98177290e-01 4.52359080e-01
1.89075872e-01 1.90127894e-01 -6.05333626e-01 -9.41599190e-01
-1.11824393e-01 -1.30490974e-01 -9.64645576e-03 6.82357371e-01
4.97526735e-01 -1.51110935e+00 -1.09372151e+00 1.40076593e-01
8.27957019e-02 1.21415041e-01 2.61398107e-01 3.61026436e-01
-4.12092119e-01 6.08045757e-01 1.91542313e-01 -3.26546997e-01
-1.04676008e+00 -1.03706263e-01 3.11174184e-01 -4.21322197e-01
-3.44720334e-01 8.31909359e-01 1.29090846e-01 -5.11890173e-01
3.26761156e-01 -4.26531315e-01 4.09130976e-02 -4.56863223e-03
5.70817709e-01 6.09944165e-01 1.61978200e-01 -3.09734017e-01
-2.76042614e-02 2.10803077e-01 2.33604670e-01 -5.74769080e-01
1.06035006e+00 2.69723922e-01 1.32103518e-01 4.06784624e-01
8.01588714e-01 1.65788665e-01 -8.10867131e-01 -1.57127991e-01
-5.46958208e-01 -4.16132212e-01 2.12509096e-01 -1.03511107e+00
-1.15778959e+00 1.17561197e+00 2.88540065e-01 5.75518310e-01
1.04424012e+00 -2.03337356e-01 1.26401722e+00 3.82878691e-01
3.32817405e-01 -1.16776049e+00 2.54145443e-01 6.54093027e-01
1.26081514e+00 -9.30643022e-01 -2.36258551e-01 5.37697673e-02
-4.49574798e-01 9.57934380e-01 6.55789852e-01 3.85623932e-01
3.59263420e-01 4.81138736e-01 4.46406603e-01 1.23474710e-01
-1.14869225e+00 -1.20387577e-01 3.36540937e-01 6.64056718e-01
8.47722292e-01 2.21707284e-01 -1.58337817e-01 6.85035884e-01
-8.46976280e-01 1.90680876e-01 5.01191199e-01 6.53689682e-01
-1.95943192e-01 -1.35729563e+00 -4.24529880e-01 4.89553481e-01
-6.67906284e-01 -4.98863578e-01 -4.96632576e-01 6.13271296e-01
1.32645130e-01 1.59285438e+00 9.30817891e-03 -5.27419984e-01
3.60538989e-01 7.34512508e-02 2.40933731e-01 -8.41457903e-01
-8.64805162e-01 2.44905829e-01 4.63762045e-01 -2.24981233e-01
-2.22535729e-01 -3.64652246e-01 -1.10808241e+00 -5.07966518e-01
-5.37771940e-01 2.07673192e-01 6.49397671e-01 5.37999570e-01
5.75035334e-01 8.73588562e-01 6.85757160e-01 -8.86195183e-01
-6.02437615e-01 -1.48231125e+00 -5.54270327e-01 4.07061219e-01
1.09678745e-01 -3.31264406e-01 -3.96462828e-01 2.13189676e-01] | [15.364167213439941, 6.119099140167236] |
3b68622c-a3c0-4532-8945-365e11360f12 | implementation-of-robust-face-recognition | 1811.07339 | null | http://arxiv.org/abs/1811.07339v1 | http://arxiv.org/pdf/1811.07339v1.pdf | Implementation of Robust Face Recognition System Using Live Video Feed Based on CNN | The way to accurately and effectively identify people has always been an
interesting topic in research and industry. With the rapid development of
artificial intelligence in recent years, facial recognition gains lots of
attention due to prompting the development of emerging identification methods.
Compared to traditional card recognition, fingerprint recognition and iris
recognition, face recognition has many advantages including non-contact
interface, high concurrency, and user-friendly usage. It has high potential to
be used in government, public facilities, security, e-commerce, retailing,
education and many other fields. With the development of deep learning and the
introduction of deep convolutional neural networks, the accuracy and speed of
face recognition have made great strides. However, the results from different
networks and models are very different with different system architecture.
Furthermore, it could take significant amount of data storage space and data
processing time for the face recognition system with video feed, if the system
stores images and features of human faces. In this paper, facial features are
extracted by merging and comparing multiple models, and then a deep neural
network is constructed to train and construct the combined features. In this
way, the advantages of multiple models can be combined to mention the
recognition accuracy. After getting a model with high accuracy, we build a
product model. The model will take a human face image and extract it into a
vector. Then the distance between vectors are compared to determine if two
faces on different picture belongs to the same person. The proposed approach
reduces data storage space and data processing time for the face recognition
system with video feed scientifically with our proposed system architecture. | ['Yang Li', 'Sangwhan Cha'] | 2018-11-18 | null | null | null | null | ['robust-face-recognition'] | ['computer-vision'] | [-1.14277415e-02 -7.61548340e-01 -1.11378189e-02 -5.86453080e-01
2.93608844e-01 -1.51490092e-01 3.31866205e-01 -2.10535392e-01
-5.70657969e-01 3.20482582e-01 -3.32251877e-01 2.18883492e-02
-2.62462258e-01 -1.15234256e+00 -1.11974783e-01 -7.30580866e-01
1.75573677e-01 2.81450778e-01 -8.82848352e-02 5.10436893e-02
4.46951687e-01 1.06233883e+00 -1.87467802e+00 3.54337133e-02
5.83846152e-01 1.50005043e+00 5.36275543e-02 1.20633505e-01
-5.38080692e-01 4.56362575e-01 -5.08835971e-01 -4.56386894e-01
3.57782781e-01 -1.99659079e-01 -4.07324255e-01 1.66762143e-01
1.17436767e-01 -5.74469566e-01 -3.63646954e-01 1.11652672e+00
6.05995774e-01 8.30365047e-02 2.99874932e-01 -1.16204977e+00
-5.67585051e-01 -4.41699028e-02 -8.06448400e-01 -7.27682188e-02
2.03988001e-01 -6.63656294e-02 7.36309066e-02 -9.52511430e-01
1.61177978e-01 1.42159891e+00 6.41157389e-01 3.97474259e-01
-7.65568316e-01 -1.21565485e+00 -4.12075818e-01 6.22040629e-01
-1.48396647e+00 -4.62553471e-01 4.87904102e-01 -2.92960763e-01
6.25733078e-01 1.72056913e-01 7.73548245e-01 4.03338701e-01
-6.53407872e-02 5.01347542e-01 1.00695980e+00 -3.95901859e-01
-1.66221157e-01 3.14518899e-01 2.97978193e-01 8.22022676e-01
4.00810629e-01 7.89391175e-02 -1.16164543e-01 3.33096713e-01
8.77846658e-01 7.27557778e-01 -1.26945019e-01 2.25007802e-01
-6.50309861e-01 6.99984670e-01 3.24442297e-01 6.53907299e-01
-4.46573496e-01 -2.77754664e-01 3.00340861e-01 1.31616056e-01
-1.69315100e-01 -4.64247577e-02 -1.26555815e-01 -1.00438341e-01
-8.45566034e-01 6.81998208e-02 7.05901980e-01 4.96593565e-01
8.34747970e-01 2.80765686e-02 2.62981474e-01 1.05433524e+00
2.64858037e-01 6.80730939e-01 6.80542707e-01 -4.57232028e-01
7.62245432e-02 1.02584171e+00 -1.97705850e-01 -1.51877785e+00
-4.86583799e-01 -4.28605638e-02 -1.22613811e+00 2.87612587e-01
3.78927410e-01 -6.10247403e-02 -8.28876972e-01 1.10389388e+00
2.88894325e-01 1.83496967e-01 -5.06011210e-02 8.93371701e-01
9.46829438e-01 8.70703399e-01 -6.89627603e-02 -1.36663496e-01
1.54378474e+00 -5.49916983e-01 -7.78017282e-01 1.40139163e-01
1.05959184e-01 -9.80678201e-01 5.12615085e-01 4.98849243e-01
-6.00929976e-01 -9.79394555e-01 -9.66833711e-01 7.90875405e-02
-4.39886570e-01 4.24149871e-01 6.83286011e-01 8.59886348e-01
-7.97004461e-01 5.15038371e-01 -5.27425110e-01 -5.13781428e-01
5.69345653e-01 8.63952756e-01 -5.66993713e-01 -2.82557338e-01
-1.04414833e+00 7.53960669e-01 4.56783712e-01 4.63442653e-01
-1.54852629e-01 -9.86886248e-02 -6.15910828e-01 1.65346220e-01
1.36809811e-01 -7.20494241e-02 8.67396832e-01 -1.22572529e+00
-1.32473719e+00 5.49253881e-01 -2.24280313e-01 8.12324323e-03
4.18275259e-02 8.20310041e-02 -9.23274517e-01 -1.97073445e-01
-1.68343097e-01 4.32336628e-01 7.12563634e-01 -5.27254164e-01
-9.49566364e-01 -8.02048028e-01 -2.63379633e-01 -4.65942957e-02
-5.32197058e-01 5.62387884e-01 -6.24176204e-01 1.92111749e-02
9.43569615e-02 -6.93445027e-01 2.07746282e-01 6.41850615e-03
4.15680520e-02 -3.78110498e-01 1.21565866e+00 -7.53983736e-01
9.64969516e-01 -2.32471967e+00 -2.12450936e-01 5.66277981e-01
1.92283140e-03 8.23888123e-01 6.13161922e-02 6.38123304e-02
6.18453249e-02 -5.97233735e-02 1.17599294e-01 1.47563651e-01
-4.49070752e-01 1.14821114e-01 2.89202094e-01 1.66180030e-01
2.27913752e-01 4.66586560e-01 -2.67101139e-01 -5.12449801e-01
3.61666918e-01 7.85595536e-01 -2.46026099e-01 2.83462405e-01
4.12135780e-01 1.57743663e-01 -4.61558402e-01 7.54198611e-01
1.10415518e+00 1.43494830e-02 8.67278054e-02 -4.25729901e-01
-1.42760769e-01 -3.46087575e-01 -1.55427408e+00 8.99193048e-01
-3.48740160e-01 6.17009282e-01 9.04327184e-02 -1.18366230e+00
1.34981084e+00 2.87981898e-01 3.37475568e-01 -1.03687739e+00
5.23162425e-01 3.60485137e-01 1.43594489e-01 -9.39880550e-01
2.58402079e-01 -1.22517392e-01 4.52649206e-01 3.46149743e-01
3.84523682e-02 5.83094597e-01 1.50106430e-01 -3.52064043e-01
3.29451412e-01 -1.33323967e-01 5.26442938e-02 2.06665859e-01
9.82208371e-01 -3.79062921e-01 6.66099310e-01 -7.05270544e-02
1.63443740e-02 5.88978492e-02 9.72823724e-02 -9.13594544e-01
-8.26097488e-01 -4.49954748e-01 -3.46144736e-01 5.09296060e-01
2.40858719e-01 7.82317594e-02 -6.89441621e-01 -3.21503907e-01
2.46998250e-01 -5.19983321e-02 -3.03336412e-01 -7.26305619e-02
-5.47580183e-01 -5.20276427e-01 5.46286345e-01 4.41181690e-01
1.17360079e+00 -1.31329560e+00 -4.67559516e-01 2.11144105e-01
1.42756999e-01 -8.18767607e-01 -6.00621887e-02 -3.83703291e-01
-7.03791082e-01 -1.16559494e+00 -7.54287422e-01 -1.17579937e+00
7.57716179e-01 3.39208126e-01 4.28160548e-01 5.61019003e-01
-4.99464452e-01 -2.20747635e-01 -8.15522969e-02 -5.81795156e-01
-6.96906522e-02 -1.91080734e-01 3.92348796e-01 4.46324527e-01
1.05663443e+00 -1.77792013e-01 -5.34251451e-01 3.76081228e-01
-9.02407110e-01 -1.54852360e-01 6.38424754e-01 7.92343378e-01
1.65859863e-01 4.01348442e-01 5.68746507e-01 -4.47346509e-01
6.12277448e-01 -2.36208111e-01 -8.77958357e-01 2.99495786e-01
-6.17363632e-01 -1.54312879e-01 6.44390702e-01 -2.35821754e-01
-1.07977414e+00 1.90700665e-02 -2.77407050e-01 -2.88943261e-01
-3.21896225e-01 5.32998979e-01 -3.81580114e-01 -2.87227094e-01
2.06796795e-01 3.19730103e-01 6.14633143e-01 -6.00090563e-01
-2.00456768e-01 1.29618704e+00 4.02162760e-01 -6.92434162e-02
5.06916046e-01 7.85331056e-02 1.04617223e-01 -9.47322369e-01
-6.22819923e-02 -2.65571296e-01 -5.09013057e-01 -4.10172045e-01
8.71044517e-01 -4.87232417e-01 -1.25388479e+00 1.11606598e+00
-1.30376542e+00 6.69744849e-01 4.46679384e-01 7.19863236e-01
3.28838348e-01 3.92200649e-01 -4.99175638e-01 -8.54886711e-01
-4.82622415e-01 -1.43255866e+00 4.69482213e-01 9.85961735e-01
1.21344067e-01 -6.48325622e-01 -3.74977440e-01 2.67909050e-01
5.85753143e-01 -1.90827847e-01 6.37661576e-01 -5.47378421e-01
-6.06755257e-01 -7.15600729e-01 -5.96505702e-01 4.61136311e-01
5.38848102e-01 2.72002965e-01 -9.04348195e-01 -1.33830026e-01
1.16566829e-01 -2.09861249e-01 5.78894794e-01 2.31835648e-01
1.36250091e+00 -4.52163756e-01 -4.45976526e-01 7.42493272e-01
1.46280766e+00 1.00389504e+00 8.73730659e-01 2.75183529e-01
6.45392239e-01 9.09767747e-01 3.48346204e-01 2.12186098e-01
-5.24242036e-03 5.87320745e-01 1.52669460e-01 -2.64172703e-01
3.85146976e-01 1.10345989e-01 1.01668485e-01 6.18878722e-01
-5.10010898e-01 1.86470464e-01 -7.13473618e-01 1.50207758e-01
-1.39621210e+00 -1.38640523e+00 4.09814641e-02 2.28426647e+00
4.57579792e-01 -3.51406574e-01 -4.98734415e-03 4.14293617e-01
9.25025403e-01 -3.82209241e-01 -5.11766374e-01 -4.51172858e-01
7.24664405e-02 4.04627562e-01 2.34723330e-01 1.28117740e-01
-9.64173317e-01 5.22949636e-01 4.92744112e+00 8.21637273e-01
-1.78336549e+00 -2.86819547e-01 7.70509541e-01 8.57239515e-02
2.19837353e-01 -4.14734751e-01 -1.06266129e+00 6.80648208e-01
6.64846003e-01 -3.55210789e-02 6.13131762e-01 7.76905835e-01
5.01213595e-02 -7.81250820e-02 -9.10485923e-01 1.60299802e+00
1.63844898e-01 -1.30133522e+00 4.94010970e-02 2.43059218e-01
2.79804111e-01 -4.34461802e-01 6.21678121e-02 2.01267228e-01
-3.41242939e-01 -1.31980777e+00 1.80480734e-01 5.30495465e-01
6.99535251e-01 -1.05145371e+00 1.23066413e+00 1.91193864e-01
-1.18684244e+00 -2.80794919e-01 -5.57485938e-01 -1.62400618e-01
-5.38667962e-02 4.07627702e-01 -6.35456502e-01 4.94575620e-01
8.65545750e-01 4.09535497e-01 -3.89343947e-01 1.09086609e+00
3.47349823e-01 1.95384502e-01 -4.71458405e-01 -3.20877761e-01
3.42957228e-02 -5.95461845e-01 -1.88750476e-01 8.46047759e-01
6.78267419e-01 2.85945922e-01 -1.06734894e-02 6.82642639e-01
-8.07730556e-02 4.23691720e-01 -5.38422942e-01 -1.67560592e-01
4.98023808e-01 1.48260593e+00 -5.58820009e-01 -4.36274946e-01
-5.91471732e-01 6.83307886e-01 8.08699429e-03 -2.41708811e-02
-6.35611892e-01 -7.83174038e-01 5.35279453e-01 1.16975546e-01
5.66602089e-02 -3.32542099e-02 -8.50947946e-02 -7.07187474e-01
1.45959733e-02 -8.96460950e-01 1.19864449e-01 -3.23736429e-01
-9.25178826e-01 6.56903446e-01 -3.51170182e-01 -1.14484382e+00
-1.32333398e-01 -8.57187152e-01 -6.14665508e-01 1.26815951e+00
-1.12031102e+00 -8.12234819e-01 -6.25194609e-01 7.27257848e-01
1.56440660e-01 -6.51361942e-01 8.83300185e-01 7.81734347e-01
-1.02933872e+00 7.31468737e-01 3.10423315e-01 6.92285240e-01
5.53466916e-01 -3.42515498e-01 1.29575655e-01 7.31916428e-01
-5.06705940e-02 7.90559411e-01 -2.74772644e-02 -4.15946543e-01
-1.44818783e+00 -8.42143059e-01 8.15181851e-01 2.75104791e-01
-5.54992296e-02 7.84095824e-02 -9.81244266e-01 1.84826642e-01
3.09786238e-02 1.34041347e-02 6.79273665e-01 -1.83981173e-02
-1.15112551e-01 -7.34105945e-01 -1.39206219e+00 2.56843239e-01
3.27019989e-01 -3.40540409e-01 -1.72427058e-01 -2.53034402e-02
9.07801762e-02 4.09996547e-02 -6.96431756e-01 2.62055963e-01
1.03377378e+00 -1.07519448e+00 6.29384041e-01 -3.46599042e-01
1.85418054e-01 -3.88019562e-01 1.43109456e-01 -7.68833935e-01
-4.92198586e-01 1.03753999e-01 4.45839643e-01 1.33589530e+00
1.09711796e-01 -8.33666801e-01 7.56360173e-01 8.01915050e-01
3.56416702e-01 -8.22184920e-01 -6.85929954e-01 -5.18654287e-01
-5.02977908e-01 -2.91389273e-03 1.08961546e+00 8.56904447e-01
-1.08970568e-01 2.07398430e-01 -3.70039016e-01 -8.64142086e-03
4.63053882e-01 7.52907768e-02 6.56107605e-01 -1.62350285e+00
4.06148434e-02 -4.76175994e-01 -8.18970680e-01 -7.27594376e-01
-9.47027504e-02 -5.83868980e-01 -4.44280714e-01 -1.48594117e+00
1.27339289e-01 -6.53650463e-01 -2.62239456e-01 6.14844739e-01
4.46159840e-02 4.03513789e-01 1.56446710e-01 3.21455657e-01
-2.96032038e-02 2.55815893e-01 1.10166812e+00 -2.49139741e-01
-2.12666199e-01 7.97819272e-02 -4.80709404e-01 6.55121803e-01
7.68743992e-01 -4.80973609e-02 -1.04826450e-01 -5.34525335e-01
-2.15737939e-01 -1.31561402e-02 8.12197551e-02 -1.29220402e+00
4.48137403e-01 3.04459920e-03 1.09429145e+00 -2.33495414e-01
3.79337698e-01 -1.21280003e+00 4.42900360e-01 6.29883468e-01
2.41492614e-01 4.99716867e-03 2.42915168e-01 -2.40795873e-02
-5.39767325e-01 -3.33634436e-01 7.41255701e-01 -6.40636608e-02
-1.02677560e+00 5.97249985e-01 -2.09020123e-01 -8.33455622e-01
1.18084586e+00 -8.16812932e-01 -4.78836074e-02 -1.02274276e-01
-3.22258741e-01 9.97383296e-02 2.59360433e-01 5.85873246e-01
8.62503111e-01 -1.44580066e+00 -5.09281516e-01 8.92085373e-01
-2.03509346e-01 -7.61915818e-02 4.29006130e-01 5.58731198e-01
-7.87759244e-01 3.47370416e-01 -7.81214714e-01 -5.10084450e-01
-1.70630944e+00 4.39129740e-01 3.32813889e-01 2.23774835e-01
-1.87741756e-01 6.41552925e-01 -1.07477039e-01 -1.70616329e-01
2.79344916e-01 9.11395401e-02 -7.08351254e-01 1.05966292e-01
1.03567338e+00 4.39682871e-01 1.88245714e-01 -9.30573583e-01
-4.30559039e-01 9.36158597e-01 -3.33026171e-01 1.74385369e-01
1.08979106e+00 2.93443143e-01 -5.24392903e-01 -1.00161240e-01
1.33522213e+00 -1.48389444e-01 -5.27037740e-01 -6.61842227e-02
-2.52703965e-01 -6.75482094e-01 7.72753954e-02 -6.06221437e-01
-1.51583910e+00 1.06492126e+00 1.11901307e+00 1.30781382e-01
1.26997662e+00 -5.52389085e-01 7.85822630e-01 4.71733809e-01
5.99814832e-01 -1.01613116e+00 -4.17032868e-01 3.24586838e-01
5.32804430e-01 -1.22279656e+00 -1.11380197e-01 -1.50595337e-01
-3.86594713e-01 1.50737309e+00 7.55286694e-01 1.21207558e-01
8.02184463e-01 2.43088692e-01 1.82428345e-01 -3.86865810e-02
3.64082530e-02 9.98975635e-02 2.12636411e-01 5.87559223e-01
5.74303567e-01 -8.88647735e-02 -3.73614252e-01 6.39428139e-01
-3.85809354e-02 5.17454088e-01 1.04828618e-01 6.37936652e-01
-6.91943824e-01 -1.38480318e+00 -7.00618267e-01 8.05921257e-01
-5.81094325e-01 2.68601686e-01 -1.18381873e-01 5.50893903e-01
4.55030710e-01 9.72201288e-01 4.92369622e-01 -7.86339164e-01
2.10264713e-01 6.99747130e-02 4.89478886e-01 -1.69163436e-01
-4.02134836e-01 -1.80370227e-01 -2.83611089e-01 -3.38761806e-01
-5.25081158e-01 -2.56400228e-01 -1.20934403e+00 -8.67387176e-01
-5.26946366e-01 3.02634299e-01 9.09037828e-01 9.37543035e-01
4.17373717e-01 5.67160584e-02 8.50086927e-01 -7.73627222e-01
-3.03466827e-01 -9.58834708e-01 -6.68271661e-01 4.11969095e-01
-7.53416792e-02 -6.67159319e-01 1.14564225e-01 -5.13114370e-02] | [13.264286994934082, 0.9022314548492432] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.