Datasets:
J0nasW
/
paperswithcode

Dataset card Data Studio Files Community
1
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
e888e6e5-e1d5-4fed-9649-06d43dab2091
accented-speech-recognition-a-survey
2104.10747
null
https://arxiv.org/abs/2104.10747v2
https://arxiv.org/pdf/2104.10747v2.pdf
Accented Speech Recognition: A Survey
Automatic Speech Recognition (ASR) systems generalize poorly on accented speech. The phonetic and linguistic variability of accents present hard challenges for ASR systems today in both data collection and modeling strategies. The resulting bias in ASR performance across accents comes at a cost to both users and providers of ASR. We present a survey of current promising approaches to accented speech recognition and highlight the key challenges in the space. Approaches mostly focus on single model generalization and accent feature engineering. Among the challenges, lack of a standard benchmark makes research and comparison especially difficult.
['Miguel Jette', 'Nishchal Bhandari', 'Ilya Pirkin', 'Jennifer Drexler', 'Joseph Palakapilly', 'Michelle Huang', 'Ryan Westerman', 'Joshua Dong', 'Quinten McNamara', 'Miguel Del Rio', 'Natalie Delworth', 'Arthur Hinsvark']
2021-04-21
null
null
null
null
['accented-speech-recognition']
['speech']
[ 1.56672329e-01 -1.52194977e-01 -3.78611058e-01 -7.32529461e-01 -1.04737568e+00 -7.80666709e-01 4.42663223e-01 -1.36973664e-01 -3.79888088e-01 5.93988359e-01 3.74616444e-01 -4.65662003e-01 2.46736571e-01 7.44691640e-02 -2.65192896e-01 -5.43729126e-01 1.28179848e-01 6.02626443e-01 -2.00874493e-01 -6.62737250e-01 -4.84207226e-03 8.28674555e-01 -1.35577130e+00 5.70085272e-02 5.61799943e-01 7.34940886e-01 3.82252395e-01 8.08860242e-01 -3.63517642e-01 5.91111362e-01 -1.23280978e+00 -3.22178006e-01 1.67405710e-01 -2.75794536e-01 -7.74246037e-01 1.82420835e-01 6.80791199e-01 2.30750367e-01 -4.96178031e-01 1.05100644e+00 8.07731986e-01 3.73924643e-01 3.29263330e-01 -1.04803467e+00 -9.28641737e-01 8.96291912e-01 -7.82325119e-02 5.28641164e-01 1.84581190e-01 -3.34566124e-02 8.60433996e-01 -9.99422193e-01 3.22765559e-01 1.25072896e+00 5.59751272e-01 9.77895379e-01 -1.08299339e+00 -5.06843686e-01 4.04256016e-01 1.73955455e-01 -1.57511127e+00 -1.24964654e+00 7.02512264e-01 -1.38631195e-01 1.22249413e+00 7.64815092e-01 2.62204915e-01 1.14311838e+00 -2.94319093e-01 8.37430179e-01 1.30268455e+00 -4.27711099e-01 3.64706933e-01 1.97172105e-01 3.76605600e-01 -1.19782940e-01 -3.18315864e-01 1.12772055e-01 -9.51200366e-01 1.78653345e-01 5.12681663e-01 -4.44678366e-01 -2.47295916e-01 3.12834173e-01 -9.33538616e-01 5.42163491e-01 1.46204848e-02 4.90074426e-01 -4.77022976e-01 -2.97461271e-01 4.65190053e-01 5.04556537e-01 4.57920820e-01 5.33056080e-01 -7.35139787e-01 -6.22428834e-01 -1.12050867e+00 2.37666637e-01 5.43049634e-01 1.23596728e+00 5.51294982e-01 7.94478953e-01 2.47256711e-01 1.53966153e+00 1.10718809e-01 8.34957659e-01 7.19749272e-01 -5.34775794e-01 4.10938144e-01 -2.34670322e-02 -2.29396775e-01 -3.56029034e-01 -1.99429214e-01 -4.36493516e-01 -6.67844355e-01 -2.67082378e-02 2.77970910e-01 -1.72365353e-01 -1.45647132e+00 1.58922589e+00 2.94741318e-02 -7.91713148e-02 2.59764552e-01 7.29233027e-01 7.01471984e-01 6.46020830e-01 2.70749331e-01 -3.92372966e-01 1.25827825e+00 -8.77131701e-01 -1.15732551e+00 -7.23823786e-01 4.74182606e-01 -1.38859749e+00 1.14124799e+00 5.07930219e-02 -1.29167175e+00 -5.84582150e-01 -8.42917621e-01 -2.90971305e-02 -7.14516759e-01 -1.38580099e-01 2.15757638e-01 1.21738017e+00 -1.26352870e+00 -6.58943132e-02 -6.95711970e-01 -3.76495034e-01 -7.99449254e-03 5.21697760e-01 -3.94135863e-01 5.43935783e-02 -1.26729620e+00 1.19356227e+00 3.77454787e-01 2.80669630e-01 -3.22543770e-01 -1.03183544e+00 -8.59579027e-01 -1.84638023e-01 7.25752041e-02 4.52865427e-03 1.76874077e+00 -7.14348733e-01 -1.83385658e+00 6.98116839e-01 -5.87104261e-01 -4.42078650e-01 1.12910494e-01 -3.96870136e-01 -1.10746217e+00 -4.85797554e-01 -4.03232783e-01 3.16801578e-01 7.18649745e-01 -1.20146608e+00 -7.89168239e-01 -4.53907192e-01 -5.90876520e-01 4.64639843e-01 9.26989503e-03 7.21881270e-01 -1.89372655e-02 -9.83739078e-01 7.13838115e-02 -9.99899507e-01 -3.14131409e-01 -1.03534269e+00 -4.29281831e-01 -2.57827967e-01 1.04536426e+00 -9.33872223e-01 1.38329029e+00 -2.19228625e+00 -7.24445051e-03 3.61744873e-02 -2.56896704e-01 7.13241994e-01 1.09481193e-01 3.23578000e-01 -2.99580216e-01 2.47539029e-01 5.31667061e-02 -4.04622793e-01 1.76919743e-01 3.88176739e-01 -7.68631101e-01 1.96046516e-01 2.74699032e-01 7.47197449e-01 -6.63827300e-01 -1.67902023e-01 6.65698826e-01 6.61164641e-01 -8.40930417e-02 6.17731988e-01 -4.54539359e-02 2.49397248e-01 -2.71629319e-02 7.98707783e-01 6.73952222e-01 6.68082833e-01 9.27876458e-02 4.92405593e-02 -4.13253397e-01 1.06162465e+00 -1.24065089e+00 1.29089808e+00 -7.28731275e-01 4.69272047e-01 4.43102658e-01 -8.33614290e-01 1.28819323e+00 6.61386847e-01 1.92729849e-02 -4.42942411e-01 -2.95005262e-01 6.41255558e-01 1.40158698e-01 8.76257103e-03 8.32847595e-01 -4.44254667e-01 -1.95384026e-01 3.26685747e-03 2.93512076e-01 -4.65408117e-01 -1.37213245e-01 -3.28709394e-01 6.68108344e-01 -4.81964916e-01 2.47240469e-01 -5.02737761e-01 4.72095817e-01 -4.99608833e-03 6.62472308e-01 8.59652877e-01 -4.52042639e-01 8.49112272e-01 -2.76037961e-01 -3.34204227e-01 -1.08365047e+00 -1.20530570e+00 -2.89916754e-01 1.46650827e+00 -5.06355524e-01 -3.90157521e-01 -8.31311882e-01 -5.23788393e-01 -2.57079750e-01 1.02563190e+00 -1.74823955e-01 1.21244453e-01 -1.12942946e+00 -6.36409581e-01 8.11423719e-01 7.74300516e-01 3.66094671e-02 -1.06532741e+00 2.83495396e-01 3.10528457e-01 -4.18031327e-02 -1.37087071e+00 -7.64312625e-01 4.75831389e-01 -5.23531735e-01 -1.65543944e-01 -6.87496364e-01 -1.03388870e+00 2.37455387e-02 1.32968634e-01 1.24652541e+00 -2.03704134e-01 7.38428459e-02 3.41279179e-01 -4.17014241e-01 -7.28857875e-01 -1.04501426e+00 5.59921265e-01 5.77273607e-01 1.51395211e-02 7.84750581e-01 -3.80356103e-01 -1.21997572e-01 5.71207106e-01 -5.19022226e-01 -6.34914398e-01 5.40396094e-01 6.86061978e-01 6.19079351e-01 -2.51685560e-01 1.03150141e+00 -7.86453366e-01 6.34962916e-01 -3.37158799e-01 -4.86730903e-01 2.96576589e-01 -6.62281752e-01 -1.77657291e-01 6.21586323e-01 -2.18870983e-01 -1.15998709e+00 5.53307571e-02 -8.55926931e-01 -3.45251888e-01 -8.36278200e-01 3.16308886e-01 -5.24850070e-01 1.17475465e-01 5.23643970e-01 4.75263417e-01 -2.59557404e-02 -9.24828291e-01 3.59977454e-01 1.28975666e+00 6.31724000e-01 -3.13547909e-01 5.84626436e-01 -2.98285365e-01 -5.81392467e-01 -1.86267376e+00 -6.12980425e-01 -7.36369491e-01 -6.12563670e-01 7.20013380e-02 5.53317249e-01 -8.11610699e-01 -7.26638734e-02 6.94703162e-01 -1.08828640e+00 -1.69940904e-01 -4.49951231e-01 5.18487692e-01 -6.02781594e-01 8.23611394e-02 -6.05154634e-01 -9.96787250e-01 -2.97562748e-01 -1.45653737e+00 1.03078198e+00 1.40407935e-01 -5.43351233e-01 -1.12192345e+00 1.02396160e-01 5.19411862e-01 9.98358905e-01 -7.43133128e-01 6.14079118e-01 -1.10369217e+00 -2.34830886e-01 -1.13011204e-01 3.67898226e-01 8.01565588e-01 5.30439675e-01 5.50053222e-03 -1.45821369e+00 -1.85467765e-01 1.10586002e-01 -1.30696997e-01 3.93691719e-01 6.08321488e-01 9.06430006e-01 -2.72527456e-01 4.70018350e-02 5.76366365e-01 7.25832999e-01 5.19669831e-01 5.79970300e-01 6.11642972e-02 6.59085035e-01 6.15747213e-01 5.00206172e-01 -2.49906108e-01 8.56556818e-02 9.65006113e-01 -2.98954219e-01 -1.41116992e-01 -5.48505008e-01 -1.18008912e-01 5.16923070e-01 1.64286494e+00 -3.07010598e-02 -1.92965314e-01 -1.10563266e+00 8.18390846e-01 -1.29930007e+00 -8.56883287e-01 2.24298477e-01 2.35779166e+00 1.10268855e+00 -9.00283381e-02 3.32393855e-01 1.31711975e-01 1.09858310e+00 4.55559105e-01 -1.14548899e-01 -8.72386158e-01 -6.54324472e-01 3.93661559e-01 4.56672072e-01 9.47433352e-01 -9.06255305e-01 1.35642314e+00 7.86472130e+00 1.02531672e+00 -1.36652529e+00 -1.14346202e-02 5.46273410e-01 2.56042406e-02 -8.12020004e-02 -2.68984169e-01 -1.32249129e+00 3.17390531e-01 1.55927825e+00 -2.49028519e-01 5.96387804e-01 9.01038706e-01 1.55559927e-01 5.79275429e-01 -1.04805791e+00 1.07003939e+00 -3.73349898e-02 -1.12241387e+00 -1.33100776e-02 -1.13258161e-01 4.93545085e-01 6.27368271e-01 3.58457327e-01 3.30861241e-01 2.08789155e-01 -1.33754885e+00 7.64351547e-01 -7.51732960e-02 7.21678913e-01 -7.99759865e-01 6.22427106e-01 -2.21269876e-01 -1.01403069e+00 1.98114067e-01 -2.98110515e-01 1.01834834e-01 5.26805282e-01 2.08139285e-01 -1.16312742e+00 3.00712913e-01 4.54622775e-01 1.71908602e-01 -2.97728211e-01 7.74199009e-01 7.10439011e-02 1.26903069e+00 -4.69808608e-01 -4.41411808e-02 1.50680374e-02 -7.63378963e-02 7.83232212e-01 1.95811534e+00 1.22255422e-01 -1.64981380e-01 2.99337149e-01 3.24259520e-01 7.85458833e-02 4.04052347e-01 -4.83580738e-01 -3.53343695e-01 9.03863251e-01 8.27776432e-01 -3.22852015e-01 -2.28677183e-01 -4.23609853e-01 6.60370111e-01 2.81090528e-01 5.03205895e-01 -1.76088631e-01 -2.06028640e-01 1.49447215e+00 5.75207584e-02 1.11233875e-01 -6.33812904e-01 -4.71563637e-01 -8.15785587e-01 -2.95501947e-02 -1.39572787e+00 2.80469149e-01 -2.68247306e-01 -1.44230330e+00 1.01126182e+00 -8.70587453e-02 -5.96466839e-01 -5.01301169e-01 -7.60755479e-01 -3.09905648e-01 1.26029813e+00 -1.47311914e+00 -1.21860313e+00 2.36372665e-01 4.04548287e-01 1.19782507e+00 -5.84835052e-01 1.09015644e+00 4.08396661e-01 -5.14503241e-01 9.53875363e-01 3.49170603e-02 3.18446755e-02 7.24414885e-01 -1.57243001e+00 1.13923109e+00 8.01288486e-01 3.63307446e-01 7.74632037e-01 8.09730947e-01 -3.30959767e-01 -1.38493299e+00 -9.91161287e-01 1.19063318e+00 -7.79823720e-01 8.03127706e-01 -6.97930992e-01 -1.26074839e+00 9.42272246e-01 3.53619218e-01 -1.53305113e-01 8.19187522e-01 7.22202361e-01 -3.20448935e-01 -2.17291310e-01 -7.08112061e-01 7.16838479e-01 8.74090254e-01 -9.57004428e-01 -7.86963880e-01 1.52453795e-01 7.93805361e-01 -4.96174246e-01 -9.85566378e-01 2.90122539e-01 2.95725226e-01 -4.97834265e-01 8.02588642e-01 -9.69164133e-01 -4.54784065e-01 -3.25727761e-01 -5.62154531e-01 -1.86265314e+00 -2.72959113e-01 -1.16362941e+00 1.56596676e-01 1.65222740e+00 6.94847882e-01 -7.21451342e-01 5.58030069e-01 7.05043316e-01 -5.52975416e-01 -2.59807169e-01 -1.10477006e+00 -1.13261306e+00 3.22646350e-01 -7.04992175e-01 7.28540003e-01 9.68364120e-01 -1.60086397e-02 5.14645994e-01 -2.73536861e-01 3.75208497e-01 3.07598233e-01 -2.63189077e-01 3.96328688e-01 -7.20256090e-01 -5.40182516e-02 -5.36717653e-01 -6.97588861e-01 -1.05098045e+00 3.04370761e-01 -7.01221943e-01 2.88943380e-01 -8.54055226e-01 -4.82352078e-01 -7.24603236e-01 -4.56216633e-01 2.64335454e-01 -3.27248693e-01 -1.96295027e-02 3.24614286e-01 -5.50527498e-02 -9.24876332e-02 4.34727848e-01 5.97523510e-01 -9.55828801e-02 -3.53988588e-01 2.61282355e-01 -4.98206168e-01 5.27335346e-01 1.34580088e+00 -2.94677138e-01 -4.10453528e-01 -6.00735843e-01 -4.23939407e-01 -2.06901312e-01 -2.69685149e-01 -8.37876439e-01 1.48942649e-01 -4.86272186e-01 -1.38303220e-01 -4.88911808e-01 5.49399436e-01 -4.62454438e-01 7.44820461e-02 -2.59518385e-01 -4.43202496e-01 5.72852492e-01 3.35204810e-01 2.41410229e-02 -5.82767248e-01 -1.63772941e-01 1.07883418e+00 4.00790237e-02 -8.04140329e-01 5.04254885e-02 -5.84785044e-01 4.09900546e-01 3.86176944e-01 -3.85774649e-03 -2.83717364e-01 -2.91377366e-01 -8.13035667e-01 -2.31505811e-01 3.14236879e-01 9.98216987e-01 3.17311466e-01 -1.08845842e+00 -1.00891066e+00 5.56158423e-01 1.10230833e-01 -7.12342709e-02 3.08552831e-01 3.94300789e-01 -2.09638327e-01 7.04501092e-01 1.14657800e-03 -3.59300047e-01 -1.41990423e+00 4.48667973e-01 5.82254767e-01 3.13908994e-01 -2.49104917e-01 1.05352139e+00 1.83577150e-01 -6.88177407e-01 3.74489903e-01 -4.82668467e-02 1.57540701e-02 -2.04053029e-01 7.30456591e-01 4.51841176e-01 6.11916900e-01 -1.25419700e+00 -5.02483010e-01 1.20118201e-01 -5.51274359e-01 -2.98005700e-01 1.02942324e+00 -2.92801023e-01 9.37256590e-02 7.88582444e-01 8.85668039e-01 3.75121325e-01 -6.63597107e-01 -3.03120594e-02 3.16680700e-01 -4.10459161e-01 2.03339726e-01 -9.21050608e-01 -7.11490810e-01 8.49859416e-01 6.29664063e-01 4.98122573e-01 7.38474905e-01 1.42524347e-01 7.87152767e-01 3.08307439e-01 3.43698710e-02 -1.55637109e+00 -6.73922360e-01 8.58096838e-01 1.07325184e+00 -1.02011299e+00 -6.60713196e-01 -4.49215055e-01 -7.36592531e-01 8.40293348e-01 4.61328238e-01 2.39021063e-01 8.29210699e-01 6.88419521e-01 1.01342809e+00 1.90079644e-01 -5.18785954e-01 -1.10969089e-01 1.38000980e-01 1.01408303e+00 9.45282757e-01 3.99099827e-01 -8.78463611e-02 3.50585818e-01 -9.45259511e-01 -5.69043338e-01 4.47955817e-01 8.23639393e-01 -3.90166670e-01 -1.45241857e+00 -4.80077416e-01 1.58279836e-01 -7.62633741e-01 -2.27611825e-01 -4.61084574e-01 5.12572408e-01 -3.46942902e-01 1.14149630e+00 1.00026868e-01 -3.26469332e-01 6.07778311e-01 5.32370746e-01 8.94992054e-02 -8.30940366e-01 -5.64943790e-01 7.75305852e-02 5.92188060e-01 -1.68691322e-01 -5.22189960e-02 -9.48208749e-01 -9.36825037e-01 -1.39929548e-01 -4.52548504e-01 3.78594905e-01 1.18053448e+00 8.25251281e-01 4.80000407e-01 3.01057726e-01 7.74947584e-01 -4.62403804e-01 -9.76874352e-01 -1.14533460e+00 -9.34789002e-01 3.86486769e-01 6.83383703e-01 -3.56404275e-01 -3.41219932e-01 2.32587114e-01]
[14.354887962341309, 6.717148780822754]
4cf7d560-f0a6-496e-a3f3-ea1c10f56d14
using-super-resolution-for-enhancing-visual
2306.11848
null
https://arxiv.org/abs/2306.11848v1
https://arxiv.org/pdf/2306.11848v1.pdf
Using super-resolution for enhancing visual perception and segmentation performance in veterinary cytology
The primary objective of this research was to enhance the quality of semantic segmentation in cytology images by incorporating super-resolution (SR) architectures. An additional contribution was the development of a novel dataset aimed at improving imaging quality in the presence of inaccurate focus. Our experimental results demonstrate that the integration of SR techniques into the segmentation pipeline can lead to a significant improvement of up to 25% in the mean average precision (mAP) segmentation metric. These findings suggest that leveraging SR architectures holds great promise for advancing the state of the art in cytology image analysis.
['Kazimierz Wiatr', 'Marcin Pietroń', 'Agnieszka Dąbrowska-Boruch', 'Sebastian Koryciak', 'Ernest Jamro', 'Anna Śmiech', 'Rafał Frączek', 'Szymon Mazurek', 'Michał Karwatowski', 'Jakub Grzeszczyk', 'Paweł Russek', 'Daria Łukasik', 'Maciej Wielgosz', 'Jakub Caputa']
2023-06-20
null
null
null
null
['super-resolution']
['computer-vision']
[ 9.17545140e-01 3.97966534e-01 2.11421341e-01 -4.48319793e-01 -1.35683429e+00 -2.59458184e-01 3.77962708e-01 2.81422347e-01 -2.68867880e-01 3.90137762e-01 -9.44696441e-02 -2.53208131e-01 -7.91234449e-02 -6.84921980e-01 -5.25134206e-01 -7.33903766e-01 3.23728532e-01 3.35749567e-01 7.46338010e-01 -9.05153304e-02 5.32345295e-01 9.66330647e-01 -1.27345335e+00 6.06716096e-01 8.61013472e-01 8.88645232e-01 3.51692766e-01 9.75154817e-01 -3.69182318e-01 5.99679887e-01 -6.31835222e-01 -1.48096174e-01 1.29541960e-02 -3.49521935e-01 -1.15254939e+00 9.67656262e-03 6.65302873e-01 -3.06609333e-01 2.47699738e-01 8.62988114e-01 5.54320097e-01 -1.34954095e-01 2.16697067e-01 -3.10887367e-01 -3.34286630e-01 5.01243830e-01 -4.87451375e-01 1.06154394e+00 2.43829519e-01 5.13780676e-02 4.55096304e-01 -4.66962904e-01 8.69136751e-01 9.74823713e-01 7.60547101e-01 4.45231497e-01 -1.28876901e+00 -5.35347819e-01 -3.24672520e-01 -2.43906856e-01 -1.33967578e+00 -5.86592257e-01 2.47707441e-01 -2.24967778e-01 9.46645498e-01 3.07468265e-01 2.97428340e-01 4.16905761e-01 4.73383874e-01 5.10320365e-01 1.50172591e+00 -4.81679022e-01 2.30874598e-01 2.22714558e-01 1.74514771e-01 5.31035841e-01 3.74905735e-01 -1.66198000e-01 -5.76520503e-01 2.35779673e-01 1.07903099e+00 -4.94591117e-01 1.43590659e-01 -2.22306401e-02 -9.78119850e-01 4.63340193e-01 4.65323389e-01 1.00012147e+00 -5.07217795e-02 -3.90837528e-02 1.85356483e-01 -1.60888672e-01 6.69226646e-01 8.47874641e-01 5.68559617e-02 -3.28301370e-01 -1.26457727e+00 -6.93717673e-02 1.56256959e-01 5.54252565e-01 2.60224372e-01 -1.62170425e-01 -1.96152538e-01 4.62484390e-01 9.00160335e-03 3.51967871e-01 4.06237632e-01 -1.42523384e+00 -7.86692873e-02 8.13827872e-01 1.59900710e-01 -9.41110432e-01 -6.81228459e-01 -8.02078545e-01 -1.60715058e-01 1.37255058e-01 4.49800730e-01 2.79280096e-01 -1.37265158e+00 9.01445329e-01 3.34460139e-01 3.73187333e-01 -1.32976010e-01 8.32709730e-01 9.70602989e-01 -2.80963667e-02 1.01047583e-01 -1.31682560e-01 1.21844983e+00 -6.36540353e-01 -9.76873934e-01 1.25021443e-01 4.87541556e-01 -9.30378675e-01 8.84880245e-01 3.38442773e-01 -1.40501928e+00 -5.09247124e-01 -1.07949328e+00 -1.93308219e-01 -2.13997722e-01 3.12725827e-02 5.06235480e-01 9.29059625e-01 -1.37956965e+00 5.85209489e-01 -1.38132989e+00 -5.58352232e-01 9.42812681e-01 6.55953407e-01 -2.86696941e-01 -1.11544356e-01 -4.58382487e-01 9.13266838e-01 5.13476357e-02 3.22986618e-02 -4.48714435e-01 -1.24088883e+00 -6.36396527e-01 -2.05755457e-01 3.96600552e-02 -3.60682189e-01 1.10915446e+00 -6.44272745e-01 -1.38497925e+00 1.23008978e+00 -4.38323766e-01 -5.30101359e-01 5.74421227e-01 3.32460105e-02 -3.56053621e-01 8.48358572e-01 1.69681221e-01 6.74862504e-01 3.51563603e-01 -1.25018680e+00 -9.22618508e-01 -7.47569203e-01 -2.58427709e-01 2.56994307e-01 -1.10391282e-01 2.52067119e-01 -4.36423421e-01 -2.87677675e-01 2.35840395e-01 -8.98657441e-01 -3.86059433e-01 -1.35893345e-01 5.15609384e-02 1.47911102e-01 7.81566739e-01 -7.13765085e-01 1.00575447e+00 -2.18282270e+00 -1.29499044e-02 1.39224097e-01 1.96384430e-01 4.81148809e-01 1.31801799e-01 -2.46531501e-01 3.49145263e-01 3.63564730e-01 -1.97658300e-01 -3.02963048e-01 -8.77690673e-01 8.57957006e-02 8.79729390e-02 4.09081131e-01 3.57898027e-01 1.16791630e+00 -9.40331876e-01 -8.18445683e-01 4.70663369e-01 7.30135083e-01 -1.32686431e-02 -8.67100433e-03 9.61575583e-02 9.30220842e-01 -4.24579889e-01 5.72488487e-01 8.33540499e-01 -7.35441744e-01 2.23257974e-01 -3.62644605e-02 -3.59295458e-01 2.85515755e-01 -1.12880671e+00 2.04357338e+00 -1.16669238e-01 5.52162051e-01 1.49007708e-01 -4.83836859e-01 6.77863896e-01 1.32988423e-01 8.37230325e-01 -8.02231193e-01 2.43752122e-01 2.73528486e-01 -2.09288243e-02 -2.53865749e-01 6.24896467e-01 -3.87640804e-01 5.48768461e-01 1.00339472e-01 -1.43329695e-01 -2.27879167e-01 1.82167038e-01 1.49872661e-01 1.06036794e+00 -1.01032116e-01 1.50508396e-02 -5.27314007e-01 7.65031695e-01 4.15124536e-01 2.07056329e-01 7.07828581e-01 -3.68374258e-01 9.55264032e-01 2.61977792e-01 -1.90401673e-01 -9.59840953e-01 -1.04666579e+00 -6.31732881e-01 6.62470877e-01 2.64780998e-01 1.79030016e-01 -9.27173257e-01 -6.89907670e-01 -2.05146864e-01 3.85967672e-01 -8.90123010e-01 3.03982168e-01 -7.29138196e-01 -9.16710615e-01 6.79940701e-01 7.36584425e-01 4.60224837e-01 -7.74066567e-01 -9.77065146e-01 1.17067404e-01 -1.47764608e-01 -1.56517863e+00 9.62337255e-02 -1.45483643e-01 -1.27992320e+00 -9.82376218e-01 -9.17223632e-01 -6.88334405e-01 8.08245242e-01 4.55702335e-01 7.40233183e-01 1.77735731e-01 -6.91963077e-01 1.23014219e-01 -3.77327740e-01 -2.67989099e-01 -5.21439195e-01 3.01490724e-01 -5.28774142e-01 -3.77505898e-01 2.58779436e-01 9.41925421e-02 -9.25100148e-01 2.80178249e-01 -1.18589365e+00 -9.01017934e-02 5.00025630e-01 3.78067464e-01 9.75980163e-01 1.41284689e-01 4.71399337e-01 -1.27577186e+00 2.23603711e-01 2.04930436e-02 -5.58239758e-01 1.51727259e-01 -9.03999448e-01 3.43251750e-02 2.50619292e-01 -1.32520609e-02 -1.44092131e+00 -1.57108173e-01 -3.44214350e-01 -1.06464669e-01 -2.44440943e-01 1.57159358e-01 6.22817099e-01 -8.47735226e-01 7.41190970e-01 1.27348676e-01 4.86038864e-01 -2.39008710e-01 -1.02956608e-01 4.90858108e-01 5.63724101e-01 -4.83512022e-02 1.39376074e-01 1.20299792e+00 4.86941636e-01 -7.91217864e-01 -9.63653564e-01 -9.20975685e-01 -7.64010906e-01 -3.02934907e-02 8.53932202e-01 -7.54076898e-01 -1.97219491e-01 4.25761610e-01 -2.44846731e-01 -4.11804229e-01 -3.81946772e-01 1.95606023e-01 -3.12279433e-01 5.06765127e-01 -8.50137055e-01 -4.08309251e-01 -4.67780173e-01 -1.42086399e+00 1.46523607e+00 7.68267334e-01 -1.10276602e-01 -9.50037360e-01 -1.66693658e-01 1.00588453e+00 7.99796283e-01 1.68813378e-01 4.03129935e-01 -4.86992925e-01 -5.86372375e-01 -3.34882811e-02 -5.29559851e-01 1.15088578e-02 3.02532136e-01 1.33543238e-01 -9.09948945e-01 -3.94481182e-01 4.20575123e-03 7.28089064e-02 6.08384550e-01 8.21398675e-01 9.90545094e-01 7.01089323e-01 -5.03419340e-01 5.97504914e-01 1.59737718e+00 7.38331396e-03 8.18937421e-01 4.23630685e-01 3.44883859e-01 5.05993009e-01 9.90803897e-01 -7.29743242e-02 2.41924435e-01 7.02037215e-01 9.06441584e-02 -2.99329519e-01 -9.01215613e-01 1.46581396e-01 -4.12054658e-01 2.00817078e-01 -3.87681648e-02 2.46997073e-01 -1.11540270e+00 6.10778451e-01 -1.34685028e+00 -5.07452250e-01 -4.30618584e-01 1.90504682e+00 5.11886656e-01 5.52187935e-02 -8.73816758e-02 6.54991567e-02 4.06708211e-01 -3.15026432e-01 -4.59599376e-01 -3.62351924e-01 9.08375829e-02 4.98748094e-01 5.64006865e-01 6.24052644e-01 -9.81980801e-01 1.00800526e+00 8.02664280e+00 8.33219767e-01 -1.43793046e+00 2.79321019e-02 7.71325111e-01 1.03119731e-01 -1.68225586e-01 -4.44030553e-01 -1.11518204e+00 3.68846416e-01 1.22413945e+00 1.06994241e-01 1.04570977e-01 9.51717645e-02 3.39560688e-01 -6.39125288e-01 -6.33120120e-01 4.52826560e-01 2.29191631e-01 -1.71210766e+00 -8.41944963e-02 2.37894267e-01 8.12735617e-01 4.89188246e-02 4.51804042e-01 -3.49901855e-01 5.88936098e-02 -1.21555161e+00 -7.19168782e-02 3.97478253e-01 1.06151247e+00 -7.84998119e-01 1.18342769e+00 -2.79894788e-02 -9.14997458e-01 3.37771893e-01 -1.27039865e-01 3.83996964e-01 -4.10355888e-02 5.03208578e-01 -1.45993602e+00 6.84261262e-01 7.87171245e-01 4.80873466e-01 -9.63770092e-01 1.06309223e+00 1.86647728e-01 4.33040887e-01 -1.19979396e-01 4.50335115e-01 9.86080766e-02 1.35382086e-01 4.27495509e-01 1.57167721e+00 2.49415580e-02 2.25840315e-01 -2.69673407e-01 5.94943941e-01 2.49263003e-01 4.31294255e-02 -2.10023656e-01 -5.90439327e-02 2.87102163e-01 1.40998566e+00 -1.62301266e+00 -2.10357636e-01 -2.74186641e-01 7.00232506e-01 3.52488719e-02 3.99603397e-02 -4.65031356e-01 5.65002449e-02 1.82744443e-01 2.77988255e-01 3.99661869e-01 -3.44133601e-02 -1.08003891e+00 -4.94381100e-01 -2.34178931e-01 -7.19041526e-01 5.98156571e-01 -5.66112220e-01 -8.10985208e-01 4.26641047e-01 -3.79741520e-01 -6.32986426e-01 -9.71183088e-03 -4.76860017e-01 -1.55459885e-02 9.98019576e-01 -1.89471459e+00 -1.31411636e+00 -5.52889228e-01 5.44644780e-02 5.78849852e-01 2.44197503e-01 7.71274626e-01 1.05598733e-01 -3.24592561e-01 5.47712862e-01 2.24173561e-01 -3.02393854e-01 4.65849429e-01 -1.47095120e+00 4.26881127e-02 9.17481363e-01 -2.38386691e-01 5.14306903e-01 8.27262759e-01 -5.99823833e-01 -1.21250594e+00 -9.05174136e-01 5.63228011e-01 -7.44743228e-01 3.61082375e-01 2.21145451e-01 -8.54936004e-01 4.53871191e-01 -1.07337989e-01 2.62107253e-01 8.88988018e-01 -2.11939961e-01 2.95810223e-01 3.48314680e-02 -1.96091437e+00 2.16203347e-01 5.72706938e-01 -2.93901056e-01 -3.32033992e-01 -1.34374589e-01 6.41164005e-01 -9.98452187e-01 -1.24862754e+00 7.11821675e-01 4.80773032e-01 -9.65561271e-01 1.09999120e+00 -1.14505112e-01 6.03127122e-01 -2.57215083e-01 1.75248161e-01 -7.93171406e-01 -2.64698058e-01 -5.30207381e-02 -2.59723794e-02 9.58134472e-01 2.75253892e-01 -3.03963572e-01 1.40714014e+00 7.11416483e-01 -2.57277369e-01 -7.31826305e-01 -1.04038990e+00 -3.19863051e-01 2.14473054e-01 -9.26499814e-02 2.35905424e-01 5.97947001e-01 -2.44667396e-01 -2.76385278e-01 5.61704993e-01 3.37397486e-01 6.40350342e-01 9.98101681e-02 2.93210745e-01 -9.07984734e-01 1.20670438e-01 -4.49578226e-01 -5.94870389e-01 -4.38856184e-01 -2.69098848e-01 -8.92714202e-01 -1.49798989e-01 -1.91189718e+00 2.85494834e-01 -4.86290395e-01 -6.30329192e-01 -9.45013482e-03 -4.46750492e-01 8.11438680e-01 1.10943936e-01 7.98429325e-02 -7.55870104e-01 -3.27039748e-01 1.70267081e+00 2.39847362e-01 -8.51102769e-02 2.71293800e-03 -1.03956127e+00 4.36001509e-01 5.94278395e-01 -2.77883321e-01 -2.07290158e-01 -4.11430508e-01 -9.35195237e-02 -3.40702012e-02 2.40107272e-02 -1.04441738e+00 2.81835139e-01 5.03070354e-02 6.36725605e-01 -5.26936114e-01 2.21926823e-01 -7.52432883e-01 3.27056535e-02 6.22160673e-01 -4.49331880e-01 -4.22006309e-01 6.16743326e-01 4.61824715e-01 -1.58081293e-01 -5.08226492e-02 1.21268916e+00 -1.94987759e-01 -5.79980969e-01 -3.37909430e-01 -3.80591787e-02 -1.55668750e-01 1.18869376e+00 -5.27752459e-01 -3.87152016e-01 2.26079017e-01 -7.65589893e-01 2.26861104e-01 7.14952290e-01 1.14534833e-01 6.59466922e-01 -4.21165556e-01 -5.77000022e-01 1.23393178e-01 -2.02326968e-01 3.61526430e-01 4.41207409e-01 1.10784626e+00 -9.93098378e-01 9.62986887e-01 -2.58684576e-01 -9.98846829e-01 -1.66651440e+00 5.33444397e-02 6.69529915e-01 -6.78226113e-01 -6.96509063e-01 1.31514478e+00 -3.30977917e-01 -2.53750831e-01 -5.52498475e-02 -3.05368334e-01 -4.34244365e-01 -3.45446199e-01 7.74463117e-01 7.66464353e-01 5.52539110e-01 -4.74949569e-01 -5.77795267e-01 4.16036844e-01 -4.98807400e-01 -1.25044044e-02 1.22574866e+00 -3.98041815e-01 -4.19606119e-02 3.94038469e-01 6.49823010e-01 1.63552202e-02 -1.26914394e+00 -5.38053587e-02 1.31368205e-01 -6.92884684e-01 4.98149097e-01 -1.26469171e+00 -8.73843670e-01 6.53649569e-01 1.02699018e+00 -3.27362530e-02 1.06369293e+00 2.23266631e-01 7.59035707e-01 -2.69809842e-01 5.33859313e-01 -7.95956433e-01 -1.63612366e-01 1.22884162e-01 2.56682038e-01 -1.50672162e+00 2.69825608e-01 -8.26000690e-01 -5.95883310e-01 1.10262513e+00 3.73875558e-01 -1.48426890e-01 2.23902300e-01 7.23076999e-01 3.29406589e-01 -4.87915814e-01 -2.71348506e-01 -1.55387327e-01 3.74187410e-01 6.08750582e-01 8.93012285e-01 4.64556590e-02 -6.03204370e-01 -6.45419676e-03 3.31574562e-03 4.39952701e-01 6.44601047e-01 1.01893437e+00 -4.04385507e-01 -8.44426990e-01 -2.24539354e-01 7.38040328e-01 -1.10560369e+00 3.03958178e-01 -2.34505951e-01 4.86435413e-01 -1.83996595e-02 1.07646835e+00 2.82134533e-01 1.85657546e-01 2.00309202e-01 -3.34089786e-01 7.19483972e-01 -9.35239077e-01 -7.42496669e-01 6.22043572e-02 -8.10777992e-02 -8.42747092e-01 -5.91190696e-01 -8.02433610e-01 -1.82149053e+00 -4.88341153e-02 -1.02403425e-01 -5.20948283e-02 9.03840125e-01 9.96548474e-01 5.14390409e-01 1.13632691e+00 1.91258833e-01 -3.51058900e-01 -1.17722563e-01 -6.69151425e-01 -3.77399385e-01 2.03630939e-01 3.93066823e-01 -3.83553237e-01 -1.25509515e-01 1.43785462e-01]
[15.027816772460938, -2.551020860671997]
933061a7-d657-4bc7-b5dc-795d10cb3189
learning-intrinsic-images-for-clothing
2111.08521
null
https://arxiv.org/abs/2111.08521v1
https://arxiv.org/pdf/2111.08521v1.pdf
Learning Intrinsic Images for Clothing
Reconstruction of human clothing is an important task and often relies on intrinsic image decomposition. With a lack of domain-specific data and coarse evaluation metrics, existing models failed to produce satisfying results for graphics applications. In this paper, we focus on intrinsic image decomposition for clothing images and have comprehensive improvements. We collected CloIntrinsics, a clothing intrinsic image dataset, including a synthetic training set and a real-world testing set. A more interpretable edge-aware metric and an annotation scheme is designed for the testing set, which allows diagnostic evaluation for intrinsic models. Finally, we propose ClothInNet model with carefully designed loss terms and an adversarial module. It utilizes easy-to-acquire labels to learn from real-world shading, significantly improves performance with only minor additional annotation effort. We show that our proposed model significantly reduce texture-copying artifacts while retaining surprisingly tiny details, outperforming existing state-of-the-art methods.
['Xiaodong Yang', 'Zian Wang', 'Kuo Jiang']
2021-11-16
null
null
null
null
['intrinsic-image-decomposition']
['computer-vision']
[ 5.70987642e-01 -1.50152057e-01 1.87710971e-01 -5.37565351e-01 -8.40854883e-01 -4.75374579e-01 2.33559728e-01 -3.65995109e-01 -8.50568861e-02 7.50548601e-01 -3.18132974e-02 1.16012551e-01 3.74449790e-01 -6.20530486e-01 -9.08786058e-01 -7.33422458e-01 2.08817020e-01 1.69469431e-01 3.91948909e-01 -2.14139089e-01 -5.31827845e-02 -5.88053931e-03 -1.28029859e+00 3.46009046e-01 9.95936632e-01 1.09787810e+00 2.22827911e-01 6.61541224e-01 4.45933402e-01 7.18433321e-01 -2.52761006e-01 -7.72188246e-01 4.80723739e-01 -4.60910708e-01 -3.65277320e-01 4.11447823e-01 8.12252641e-01 -4.59237605e-01 -2.42316708e-01 1.06831658e+00 5.52248359e-01 -2.79715717e-01 4.91144389e-01 -1.03865230e+00 -9.25851405e-01 1.72828391e-01 -6.82430685e-01 -4.62506562e-01 3.68815690e-01 3.66616756e-01 9.24619317e-01 -8.62310112e-01 6.35321498e-01 9.99589145e-01 1.05603218e+00 4.88766462e-01 -1.50651133e+00 -3.86705369e-01 1.42693803e-01 -2.03160673e-01 -1.11268306e+00 -3.79373908e-01 1.16360319e+00 -2.06036136e-01 3.30539703e-01 5.74113786e-01 6.32604659e-01 1.49351799e+00 7.15196505e-02 8.34663868e-01 1.62572742e+00 -4.32012767e-01 1.33715004e-01 2.37390921e-01 -1.59794465e-02 1.04582715e+00 2.53688902e-01 7.74195418e-02 -2.72948802e-01 -1.41460270e-01 1.00909662e+00 -2.56968915e-01 -4.81526852e-01 -6.74554646e-01 -1.05769587e+00 3.60198408e-01 3.81928593e-01 -3.85380030e-01 -1.51302949e-01 3.28336179e-01 3.12307358e-01 1.85936779e-01 5.48923016e-01 9.18935388e-02 -5.34725010e-01 1.43930867e-01 -6.51554585e-01 1.98177114e-01 7.10415840e-01 1.06770265e+00 6.22232139e-01 1.85895428e-01 -1.88064009e-01 9.96469200e-01 1.36909142e-01 6.35392070e-01 -2.37488195e-01 -1.22563088e+00 4.20026183e-01 3.99501801e-01 2.64394313e-01 -1.29215252e+00 -5.90708964e-02 -4.44726169e-01 -9.19094801e-01 3.04573148e-01 5.38410187e-01 1.39195845e-01 -1.09974897e+00 1.80191624e+00 4.15857613e-01 1.19259834e-01 -2.83975959e-01 1.31343901e+00 6.74559355e-01 1.45546988e-01 -1.16182350e-01 1.21622734e-01 1.36651850e+00 -1.28275406e+00 -4.43310648e-01 -8.67030248e-02 -2.81645432e-02 -1.03764713e+00 1.89209390e+00 7.34146357e-01 -1.27498865e+00 -6.04141712e-01 -1.02657151e+00 -4.28057462e-01 7.30626360e-02 4.88801539e-01 9.57620323e-01 8.83942306e-01 -9.16441143e-01 5.95573843e-01 -8.98769855e-01 -2.88205773e-01 3.68036866e-01 1.93446532e-01 -4.50057060e-01 -2.97142893e-01 -7.13730156e-01 3.18458080e-01 -3.18805009e-01 2.47038618e-01 -1.25735688e+00 -7.93361485e-01 -1.03662336e+00 -3.84516746e-01 4.50464249e-01 -1.02667809e+00 7.96459258e-01 -1.02277160e+00 -1.63348973e+00 1.19741905e+00 1.33920938e-01 -8.88009816e-02 8.99755299e-01 -1.64265007e-01 -3.55765164e-01 2.74780355e-02 7.95312747e-02 3.09015572e-01 1.14467204e+00 -1.94498682e+00 -1.80031031e-01 -3.76745999e-01 3.09480876e-01 -4.27361056e-02 -2.00817034e-01 -2.28239104e-01 -9.06860590e-01 -1.17229772e+00 1.39224067e-01 -1.01802766e+00 -2.05775559e-01 5.22393584e-01 -7.01422930e-01 6.84904873e-01 5.10602415e-01 -9.17369187e-01 9.56883371e-01 -2.04306698e+00 1.20548084e-01 1.57340486e-02 2.40771413e-01 -1.12027191e-01 -2.99749285e-01 -1.50176594e-02 2.32736707e-01 -9.98270288e-02 -5.34166396e-01 -7.52998829e-01 7.78623745e-02 3.61446053e-01 -2.20936671e-01 5.62019765e-01 2.04603121e-01 6.44187868e-01 -8.85756254e-01 -4.57667768e-01 2.72794813e-01 6.67385578e-01 -8.19539309e-01 2.43277431e-01 -2.10908964e-01 5.68128824e-01 -1.97587356e-01 1.01816523e+00 9.85348463e-01 -2.62507021e-01 1.64348826e-01 -7.60500491e-01 3.53979141e-01 -1.22362360e-01 -1.14922345e+00 2.06157303e+00 -4.98492539e-01 3.33415568e-01 3.95031780e-01 -6.46730244e-01 7.20805883e-01 -6.02526516e-02 2.72205383e-01 -7.58746445e-01 2.24593878e-01 1.14308208e-01 -3.25028718e-01 -4.79273856e-01 4.07883346e-01 1.05620973e-01 -1.15385719e-01 8.95170048e-02 -7.21300840e-02 3.45188379e-02 -8.22757334e-02 6.23592027e-02 1.16413641e+00 6.83083415e-01 -2.51267642e-01 -3.57008904e-01 4.79519784e-01 -5.88372461e-02 7.82564878e-01 4.62301373e-01 -5.48555888e-02 1.26219380e+00 3.46915632e-01 -5.84083974e-01 -1.30178297e+00 -1.13888812e+00 -1.71255723e-01 9.54095006e-01 5.23183048e-01 -4.01108444e-01 -1.07345235e+00 -6.38150215e-01 1.13197103e-01 4.22231793e-01 -9.11412179e-01 1.53060406e-01 -5.18691063e-01 -8.40988338e-01 5.01446187e-01 6.71022594e-01 7.39343524e-01 -6.42217278e-01 -3.40338737e-01 -9.37176123e-02 2.85274419e-03 -1.41587245e+00 -8.48928154e-01 -1.00835681e-01 -6.95212424e-01 -1.08639276e+00 -7.14898944e-01 -7.53203452e-01 1.13652813e+00 2.74175793e-01 1.57191169e+00 1.41240224e-01 -6.04720354e-01 4.22967464e-01 -9.54924524e-02 -5.70747256e-03 -1.62967220e-01 -3.20446849e-01 -1.28562525e-01 2.55957544e-01 -1.86949745e-01 -7.13643193e-01 -1.07582021e+00 7.29484260e-01 -6.16620600e-01 5.30805051e-01 4.48353291e-01 1.14440596e+00 8.82333159e-01 -1.17281891e-01 4.10678759e-02 -1.14468849e+00 3.15461248e-01 7.08109587e-02 -4.87284750e-01 3.26711327e-01 -5.49942613e-01 -9.35134664e-02 8.50474417e-01 -4.59967583e-01 -1.25868547e+00 7.25350976e-02 -8.42526853e-02 -5.52499413e-01 -5.43515980e-02 -3.39683779e-02 -5.61168671e-01 -4.70692486e-01 6.41739309e-01 1.61272228e-01 -1.82491288e-01 -7.35650420e-01 4.25372481e-01 2.64749788e-02 7.65185177e-01 -1.22399652e+00 7.30278969e-01 7.06893921e-01 -1.21572830e-01 -6.17389202e-01 -8.48968208e-01 -8.98205861e-02 -5.79863489e-01 -1.28565833e-01 6.24566078e-01 -1.04256785e+00 -7.45173812e-01 5.61732650e-01 -9.12088096e-01 -7.11702168e-01 -2.27496043e-01 -1.92436706e-02 -6.69065416e-01 4.25963372e-01 -9.09071267e-01 -6.52750552e-01 -2.79375613e-01 -1.15831482e+00 1.42349374e+00 -1.59821972e-01 1.18574120e-01 -9.40612972e-01 -9.39692259e-02 6.69767320e-01 3.59412700e-01 9.00203109e-01 5.53055644e-01 4.59503621e-01 -7.73197353e-01 -2.88828462e-02 -4.89133030e-01 7.53067076e-01 1.99222267e-01 -3.16937119e-02 -1.15358734e+00 -4.78248239e-01 -1.21952444e-02 -5.33401072e-01 9.91410255e-01 3.00377786e-01 1.53507245e+00 -3.83580834e-01 -4.90535386e-02 1.19208062e+00 1.75347590e+00 -3.99994552e-01 7.60210454e-01 -4.31935042e-02 1.05786943e+00 3.95265609e-01 5.78473449e-01 3.47831249e-01 5.94507992e-01 6.49272323e-01 2.92132348e-01 -6.34853780e-01 -5.99754930e-01 -3.81133676e-01 1.34935454e-01 9.07928526e-01 -5.13945341e-01 -1.54012203e-01 -4.85905826e-01 4.79327679e-01 -1.75116515e+00 -5.61626732e-01 -1.60579681e-01 2.22343254e+00 1.00496304e+00 2.69114226e-01 1.39410840e-02 3.85099612e-02 2.53303170e-01 9.49848145e-02 -6.33893549e-01 -4.49164659e-02 -3.24184746e-01 2.45425567e-01 6.34951174e-01 6.57865405e-01 -1.11945045e+00 9.45759058e-01 6.41414833e+00 8.41664732e-01 -9.54908490e-01 3.11908364e-01 8.26907456e-01 1.28369510e-01 -6.96437240e-01 -7.67404214e-02 -3.73200893e-01 6.07161283e-01 1.71025798e-01 3.60466003e-01 6.20713651e-01 9.07713115e-01 2.28241812e-02 2.65595857e-02 -1.02028191e+00 1.07582676e+00 2.89178044e-01 -1.11185157e+00 1.22044962e-02 -5.01019508e-03 8.96471322e-01 -3.73131126e-01 2.63879985e-01 -2.53181662e-02 1.79889604e-01 -9.73446548e-01 8.19533348e-01 5.78638315e-01 1.20920646e+00 -3.91746998e-01 4.35696542e-01 -1.50616795e-01 -1.34804416e+00 3.27131987e-01 -5.05366504e-01 -5.64080998e-02 6.14775196e-02 6.15633786e-01 -2.60479748e-01 6.03273690e-01 6.22426867e-01 6.37792408e-01 -7.14249790e-01 6.35253310e-01 -2.88938463e-01 7.45749116e-01 -2.43272126e-01 3.97991031e-01 -1.99776098e-01 -3.43572050e-01 3.47645611e-01 1.21785188e+00 -3.76968756e-02 -2.95878779e-02 5.13337731e-01 9.93150651e-01 -1.41295359e-01 -7.81592429e-02 -5.36917448e-01 4.33129728e-01 -9.79212895e-02 1.51318359e+00 -6.87834382e-01 -7.06516430e-02 -3.76162320e-01 1.67924905e+00 2.92969346e-01 5.08904397e-01 -1.17168236e+00 -7.55331218e-02 6.45782411e-01 3.64330262e-01 1.36493832e-01 -2.60795325e-01 -5.46635389e-01 -1.46925676e+00 3.32713157e-01 -9.41327512e-01 -9.54774097e-02 -7.33994842e-01 -1.49101162e+00 6.20466411e-01 -4.08905178e-01 -1.25930548e+00 3.11410993e-01 -7.93712556e-01 -4.12298888e-01 6.98325694e-01 -1.42850149e+00 -1.84087956e+00 -9.08710063e-01 6.20447457e-01 4.12916064e-01 5.35071455e-02 8.96737933e-01 6.33342505e-01 -5.72282076e-01 1.02280033e+00 -2.84434929e-02 2.49934658e-01 8.38443816e-01 -1.42809737e+00 5.02448857e-01 7.36766994e-01 -1.25406817e-01 5.01107454e-01 8.76100242e-01 -5.17020583e-01 -1.76360261e+00 -1.14931750e+00 5.13907559e-02 -7.34732866e-01 4.07202065e-01 -7.96218216e-01 -7.06238747e-01 5.27183771e-01 1.69267580e-01 3.97346079e-01 6.33870363e-01 -6.23477474e-02 -5.11469960e-01 -4.56731826e-01 -1.33842027e+00 7.98421443e-01 1.62478149e+00 -4.19585437e-01 -1.83009089e-03 2.45629773e-01 7.29668140e-01 -7.02298999e-01 -9.85360682e-01 6.40863597e-01 9.48327899e-01 -1.17152822e+00 1.20234549e+00 -3.88462752e-01 6.60779893e-01 -5.42977750e-01 -5.05061030e-01 -9.58239913e-01 -4.26671386e-01 -6.14302039e-01 -1.24040052e-01 1.26645529e+00 2.31431589e-01 -2.59124875e-01 9.72242177e-01 7.57478297e-01 -1.04796469e-01 -1.01282442e+00 -3.73041749e-01 -7.51198649e-01 -4.29818988e-01 -4.67288822e-01 2.79663861e-01 9.84418988e-01 -7.77045548e-01 2.11435243e-01 -9.01833892e-01 2.38693014e-01 1.43515146e+00 3.12322825e-01 1.17683446e+00 -7.55236685e-01 -5.66159368e-01 -3.29767354e-02 -2.95350671e-01 -1.07002759e+00 -9.32984427e-02 -5.29526830e-01 -2.24444997e-02 -1.19024897e+00 4.08165306e-01 -6.70145333e-01 -2.11954623e-01 4.05048698e-01 -2.73990273e-01 9.60530519e-01 -3.15020122e-02 -5.84450066e-02 -6.33270562e-01 6.82996869e-01 1.60681283e+00 -3.38787258e-01 1.81155354e-01 -2.79406846e-01 -7.38454640e-01 8.66888642e-01 5.59547484e-01 -9.96984020e-02 -5.20284593e-01 -7.14308739e-01 8.14974010e-02 -2.96990424e-01 7.30501354e-01 -8.89602065e-01 -3.72263715e-02 -1.68139666e-01 6.17248237e-01 -2.06643775e-01 5.51801860e-01 -8.21332872e-01 3.93714100e-01 2.13661849e-01 -1.55987158e-01 -1.07917845e-01 1.23022608e-01 6.58042073e-01 6.97901770e-02 2.26014242e-01 8.45111907e-01 -1.39764994e-01 -5.71286142e-01 4.93314952e-01 5.98000348e-01 5.88429496e-02 8.32548082e-01 -2.87500352e-01 -1.83935732e-01 -1.71020940e-01 -6.90035045e-01 7.53246993e-02 1.14950395e+00 2.68279403e-01 6.86708987e-01 -1.48738420e+00 -7.06596971e-01 3.70207578e-01 1.44734129e-01 -2.05341782e-02 3.62646222e-01 6.47751153e-01 -8.66811037e-01 -2.78242081e-01 -3.54363620e-01 -7.40801215e-01 -1.19952726e+00 6.28897488e-01 1.97546050e-01 -1.66041374e-01 -7.17034996e-01 7.84491122e-01 5.48923969e-01 -5.29918015e-01 3.79239172e-01 -5.39843321e-01 5.52555919e-01 -6.28979564e-01 2.96915263e-01 2.19288588e-01 -1.02887042e-01 -4.16603833e-01 -2.48817518e-01 9.47723210e-01 7.14298859e-02 1.59157068e-01 1.12608874e+00 -2.17499450e-01 -1.17342127e-02 2.66545802e-01 1.04741168e+00 2.68060237e-01 -1.66603398e+00 -3.38420011e-02 -4.89994526e-01 -1.04015815e+00 -6.48284256e-02 -9.20198023e-01 -1.44801915e+00 7.63629496e-01 7.43168056e-01 -3.35828885e-02 1.39727747e+00 -3.87238592e-01 1.04730713e+00 -2.08838850e-01 8.95870388e-01 -9.89267886e-01 2.23953813e-01 -4.94716689e-02 1.10817635e+00 -1.58352172e+00 1.10796198e-01 -8.71919751e-01 -7.11838603e-01 6.62131190e-01 8.03201139e-01 -3.76452357e-01 5.88819563e-01 5.24289012e-01 1.24389023e-01 8.29157885e-03 -5.03265321e-01 2.52159610e-02 3.82882416e-01 7.06677794e-01 3.46270174e-01 1.26661405e-01 -1.84984162e-01 7.39839852e-01 -4.38976884e-02 -8.14908296e-02 1.43667366e-02 5.09321511e-01 9.52123702e-02 -1.28826511e+00 -2.76117653e-01 1.77313820e-01 -4.72452462e-01 -5.54958768e-02 -8.32514316e-02 6.21996641e-01 2.58543104e-01 5.19005597e-01 -2.81040937e-01 -4.99871552e-01 4.20526534e-01 -5.43058693e-01 9.52636421e-01 -2.49488622e-01 -4.01300788e-01 1.35933325e-01 2.57562429e-01 -7.39879310e-01 -3.94356459e-01 -2.54339755e-01 -6.89733922e-01 -4.63905692e-01 -2.02051997e-01 -3.84246171e-01 3.92749548e-01 3.66025060e-01 4.81301963e-01 6.57109022e-01 6.57222688e-01 -1.03611636e+00 -2.97201931e-01 -6.00935817e-01 -5.97925305e-01 9.17695224e-01 3.54550451e-01 -6.76809669e-01 -8.55308995e-02 4.59413081e-01]
[11.950610160827637, -0.8882811665534973]
e14508fe-f4f7-409d-ba0d-37c6d2cc3e63
differential-angular-imaging-for-material
1612.02372
null
http://arxiv.org/abs/1612.02372v2
http://arxiv.org/pdf/1612.02372v2.pdf
Differential Angular Imaging for Material Recognition
Material recognition for real-world outdoor surfaces has become increasingly important for computer vision to support its operation "in the wild." Computational surface modeling that underlies material recognition has transitioned from reflectance modeling using in-lab controlled radiometric measurements to image-based representations based on internet-mined images of materials captured in the scene. We propose to take a middle-ground approach for material recognition that takes advantage of both rich radiometric cues and flexible image capture. We realize this by developing a framework for differential angular imaging, where small angular variations in image capture provide an enhanced appearance representation and significant recognition improvement. We build a large-scale material database, Ground Terrain in Outdoor Scenes (GTOS) database, geared towards real use for autonomous agents. The database consists of over 30,000 images covering 40 classes of outdoor ground terrain under varying weather and lighting conditions. We develop a novel approach for material recognition called a Differential Angular Imaging Network (DAIN) to fully leverage this large dataset. With this novel network architecture, we extract characteristics of materials encoded in the angular and spatial gradients of their appearance. Our results show that DAIN achieves recognition performance that surpasses single view or coarsely quantized multiview images. These results demonstrate the effectiveness of differential angular imaging as a means for flexible, in-place material recognition.
['Ko Nishino', 'Jia Xue', 'Hang Zhang', 'Kristin Dana']
2016-12-07
differential-angular-imaging-for-material-1
http://openaccess.thecvf.com/content_cvpr_2017/html/Xue_Differential_Angular_Imaging_CVPR_2017_paper.html
http://openaccess.thecvf.com/content_cvpr_2017/papers/Xue_Differential_Angular_Imaging_CVPR_2017_paper.pdf
cvpr-2017-7
['material-recognition']
['computer-vision']
[ 8.30466807e-01 -4.01649863e-01 2.94261187e-01 -3.96699965e-01 -7.40759850e-01 -6.30652010e-01 5.64684808e-01 -3.72007638e-01 -7.11266920e-02 3.96981090e-01 -2.61400640e-01 -8.35397169e-02 -2.18620077e-01 -1.11354995e+00 -9.37039256e-01 -6.26725316e-01 1.41632864e-02 4.68064547e-01 2.74711370e-01 -4.55749005e-01 2.45045602e-01 1.01140773e+00 -1.93306971e+00 2.85298884e-01 3.97103488e-01 1.41366851e+00 3.25529695e-01 7.47323632e-01 6.13192432e-02 1.58111483e-01 -3.10948700e-01 1.60807446e-02 7.76662946e-01 3.82925361e-01 -3.54326546e-01 4.08573747e-01 1.09212399e+00 -5.66394866e-01 -3.24762285e-01 7.41861343e-01 2.45293066e-01 -8.37965086e-02 7.72769809e-01 -8.93401206e-01 -5.73950052e-01 -1.70834914e-01 -4.45330769e-01 -1.53137878e-01 7.06626117e-01 1.84071511e-01 1.07028270e+00 -9.51829493e-01 6.83625996e-01 1.11257565e+00 8.21385205e-01 3.35923076e-01 -1.44391418e+00 -3.51671159e-01 9.94834527e-02 7.68860206e-02 -1.42017174e+00 -6.00593507e-01 1.01179016e+00 -5.35257995e-01 9.85869467e-01 4.92293924e-01 1.01666105e+00 1.09619915e+00 1.75126210e-01 5.62471986e-01 1.32295680e+00 -5.15897274e-01 2.50720382e-01 -2.43419707e-02 -2.51614124e-01 9.92747724e-01 4.91284490e-01 1.47491515e-01 -8.57742250e-01 -4.77432944e-02 1.02077496e+00 2.43201286e-01 -3.72139275e-01 -5.51785350e-01 -1.06510162e+00 2.53476024e-01 4.75420535e-01 -2.42553487e-01 -5.13300359e-01 1.15286842e-01 -2.45305598e-01 5.36713898e-01 6.98663712e-01 3.61355901e-01 -2.72735894e-01 1.13831669e-01 -7.54351497e-01 1.03770748e-01 7.38785028e-01 8.51782203e-01 1.11593819e+00 2.80910760e-01 5.36952913e-01 9.67839062e-01 4.69156533e-01 1.29350388e+00 4.56869751e-02 -1.04099023e+00 2.69390374e-01 7.83912718e-01 2.15021789e-01 -1.20916259e+00 -3.09272319e-01 -4.21918221e-02 -4.84712750e-01 7.00107813e-01 2.20869645e-01 3.94262165e-01 -1.00444758e+00 1.34362411e+00 4.39867020e-01 -3.03588778e-01 5.33810966e-02 6.84791028e-01 6.45963848e-01 2.63378710e-01 -7.70508409e-01 2.96680540e-01 1.16465747e+00 -6.42874122e-01 1.55637860e-02 -1.97176546e-01 3.75541113e-02 -7.60107696e-01 1.29493403e+00 6.25605226e-01 -9.18206155e-01 -1.97960556e-01 -1.33543158e+00 2.08400905e-01 -5.55973053e-01 7.85522684e-02 5.77483118e-01 6.68589473e-01 -1.10690928e+00 3.75007242e-01 -7.85530269e-01 -4.80921924e-01 4.14314717e-01 3.66022319e-01 -5.28817117e-01 -4.32146043e-01 -4.60603952e-01 4.34617937e-01 -2.71057993e-01 1.43591270e-01 -8.25427055e-01 -6.61287010e-01 -8.10973346e-01 -5.94642043e-01 3.97042155e-01 -8.25306237e-01 7.70443439e-01 -1.14456034e+00 -1.67074823e+00 1.17041266e+00 1.11784928e-01 -1.85913771e-01 6.17820144e-01 -3.00084740e-01 -3.25166643e-01 5.15333951e-01 4.07157615e-02 5.17049551e-01 1.06890666e+00 -1.58971691e+00 -1.63497970e-01 -5.59758008e-01 1.47195220e-01 2.50244051e-01 -2.80046821e-01 -4.09507096e-01 -2.86467403e-01 -4.30211127e-01 5.92116296e-01 -1.05100429e+00 1.00595526e-01 6.63186431e-01 -4.71301943e-01 4.46883261e-01 1.01904511e+00 -4.37075913e-01 1.35908574e-01 -1.98882461e+00 -2.81373560e-01 5.26399970e-01 1.93848342e-01 -8.25940445e-02 -1.51804537e-01 4.88937944e-01 3.71082306e-01 -3.16357434e-01 -3.98440883e-02 -3.90157014e-01 4.73841246e-05 1.00380749e-01 -3.44229043e-01 5.78289866e-01 1.12734787e-01 5.88513672e-01 -4.20506567e-01 -1.80918023e-01 4.65318441e-01 5.64984798e-01 -5.09085059e-01 2.76572734e-01 -3.47086430e-01 2.96893537e-01 -4.00267065e-01 1.37993145e+00 9.02387321e-01 -2.20364764e-01 8.70936960e-02 -5.77507794e-01 -2.80717492e-01 -8.53747129e-02 -1.22340906e+00 1.68804443e+00 -5.53995013e-01 7.84556329e-01 5.65050960e-01 -5.40743828e-01 1.15290451e+00 -1.82373539e-01 7.24606931e-01 -8.42576385e-01 -7.17826784e-02 3.36830527e-01 -7.34403551e-01 -3.80042404e-01 7.03760445e-01 8.08513314e-02 3.34493101e-01 1.63732603e-01 -4.87426370e-01 -6.45216465e-01 -2.00995103e-01 -7.52137378e-02 1.19444442e+00 5.32241821e-01 -1.30587906e-01 -2.79483259e-01 2.59064913e-01 1.56986177e-01 4.08110842e-02 6.95888519e-01 2.58849174e-01 7.79815435e-01 -4.45125103e-01 -7.73092270e-01 -9.62477267e-01 -1.46805418e+00 -1.64435223e-01 9.17250335e-01 3.56737435e-01 -1.05657376e-01 -5.49148321e-01 -4.62016836e-02 3.10323626e-01 -1.84917286e-01 -6.85187221e-01 3.72351408e-02 -3.98748398e-01 -5.35297930e-01 3.43978733e-01 4.88251179e-01 9.09844160e-01 -7.64663756e-01 -8.20569813e-01 -8.12926441e-02 6.92541748e-02 -1.24288642e+00 6.59371391e-02 3.03686038e-02 -6.58354163e-01 -1.07078755e+00 -5.71140170e-01 -4.58753884e-01 6.73443198e-01 8.73041809e-01 1.30439568e+00 -5.50143272e-02 -8.92543972e-01 1.32891357e+00 -2.37801120e-01 -3.66160214e-01 -5.49605191e-02 -2.51693279e-01 2.57743478e-01 3.48657072e-01 -1.08428791e-01 -9.40658450e-01 -8.01264048e-01 5.45163810e-01 -8.44524145e-01 2.69430071e-01 7.79442668e-01 4.91824716e-01 7.97588348e-01 -3.90115768e-01 -2.04469915e-02 -5.72594523e-01 2.22416297e-01 -2.81780332e-01 -7.22500324e-01 3.30557823e-01 -5.09009540e-01 -2.49839500e-01 3.96595031e-01 -2.24377036e-01 -9.61722553e-01 -6.09679557e-02 1.43981591e-01 -4.44351017e-01 -1.58611476e-01 3.03080827e-01 -1.94170117e-01 -6.79192901e-01 7.05613494e-01 2.62642056e-01 1.58346236e-01 -3.28148305e-01 1.16119660e-01 7.05033839e-01 5.59973419e-01 -1.01082110e+00 9.52133298e-01 1.06029856e+00 2.08336502e-01 -1.49243212e+00 -5.33851743e-01 -3.66033226e-01 -4.75975424e-01 -6.03814960e-01 5.70010662e-01 -9.43333089e-01 -8.48237574e-01 7.98628867e-01 -7.48332024e-01 -6.59548223e-01 -3.04058850e-01 2.51799613e-01 -7.20855951e-01 2.21457034e-01 -4.25577104e-01 -9.19292033e-01 -3.76447290e-01 -9.56327081e-01 1.62805128e+00 5.58615103e-02 1.79000497e-01 -8.68733227e-01 5.31943440e-02 6.27449751e-01 7.03885913e-01 4.28897887e-01 6.43160343e-01 2.67600685e-01 -1.20938778e+00 -1.67948082e-01 -4.47008580e-01 2.62706131e-01 4.18143034e-01 1.52186766e-01 -1.29491568e+00 -3.25070649e-01 -3.29152912e-01 -4.74432379e-01 8.56813431e-01 1.95008457e-01 8.06700945e-01 1.08498342e-01 -4.37128842e-02 9.24867570e-01 1.68512774e+00 -1.19233958e-01 4.94053304e-01 6.34286702e-01 9.58024502e-01 6.58124089e-01 4.34107333e-01 4.46430594e-01 3.40929985e-01 9.48633254e-01 7.80520558e-01 -1.93729892e-01 -2.95911908e-01 -9.88453627e-03 5.80164850e-01 5.37405074e-01 -5.80784380e-01 -1.10443495e-01 -8.42117548e-01 1.97239161e-01 -1.20379269e+00 -7.57965803e-01 7.68967420e-02 2.29605055e+00 3.71132582e-01 1.22578256e-02 -1.04325354e-01 -8.48026425e-02 3.15346003e-01 1.92604259e-01 -7.98301160e-01 -5.18362075e-02 -5.16648352e-01 1.32192150e-01 6.98738754e-01 3.71522486e-01 -8.18673193e-01 7.56277680e-01 6.50621748e+00 4.21206713e-01 -1.43511069e+00 -4.95416641e-01 1.84588149e-01 2.75509357e-01 -5.76575279e-01 -3.97802323e-01 -6.16774738e-01 -7.40225539e-02 6.71315491e-01 3.65155846e-01 7.12436259e-01 9.55353677e-01 1.45806298e-01 -1.68967336e-01 -8.38581502e-01 1.13795626e+00 3.30748528e-01 -1.43037879e+00 2.32199296e-01 3.00930440e-01 6.37271643e-01 4.73984718e-01 4.04539883e-01 -5.01686335e-01 3.28663558e-01 -7.19104290e-01 9.82795656e-01 8.50038826e-01 1.10819995e+00 6.47854209e-02 1.23470820e-01 -2.98427250e-02 -1.58027792e+00 1.80079266e-01 -3.13173145e-01 1.22224189e-01 -1.58892035e-01 4.95183170e-01 -1.00117075e+00 6.16466224e-01 8.99106264e-01 9.55894589e-01 -6.03898168e-01 6.47635937e-01 4.26443428e-01 2.61026144e-01 -6.97352111e-01 1.32783011e-01 -5.93518429e-02 -4.98188794e-01 5.79977870e-01 9.31329072e-01 3.53137314e-01 -3.74096304e-01 3.99020433e-01 8.56915534e-01 -9.40638874e-03 -6.48419261e-02 -1.17092788e+00 8.79180580e-02 3.75917077e-01 1.24361658e+00 -7.89949059e-01 -2.50853360e-01 -2.54947245e-01 9.05657411e-01 1.25733986e-01 2.49054298e-01 -3.81014585e-01 3.31332348e-02 6.22276843e-01 3.34163994e-01 2.48836666e-01 -7.61032045e-01 -2.20794249e-02 -1.15537417e+00 4.03177023e-01 -8.13612342e-01 -2.03155860e-01 -1.19761193e+00 -1.35413516e+00 5.39957464e-01 -3.75944376e-02 -1.45485485e+00 2.16288552e-01 -1.15344203e+00 -2.80831516e-01 6.21108413e-01 -1.51530170e+00 -1.61732674e+00 -9.94317770e-01 7.40338147e-01 6.76926196e-01 -1.56567276e-01 9.88306940e-01 1.44876972e-01 -2.47330338e-01 1.71765640e-01 3.85365874e-01 -3.51763815e-01 5.76630414e-01 -1.16116798e+00 1.93842143e-01 4.82429147e-01 2.31054023e-01 5.47876358e-01 5.75165212e-01 -5.33152640e-01 -2.11395264e+00 -9.87186074e-01 -2.30607703e-01 -5.06632030e-01 4.54678833e-01 -3.82859379e-01 -5.64946711e-01 5.58141053e-01 -1.03751741e-01 -9.65881273e-02 6.19593680e-01 -1.35291219e-01 -7.23277271e-01 -5.29585063e-01 -1.25285184e+00 4.96149153e-01 1.40176308e+00 -7.53054500e-01 -1.38333961e-01 4.43317473e-01 3.74992818e-01 -3.17356646e-01 -8.03083599e-01 5.00430107e-01 1.11052251e+00 -1.05232668e+00 1.23437333e+00 -1.17172450e-02 2.06026107e-01 -2.27549151e-01 -1.02187347e+00 -1.07294083e+00 7.97616243e-02 -4.79036838e-01 3.39761615e-01 9.11517382e-01 3.20852816e-01 -1.03653407e+00 1.13218176e+00 5.73191524e-01 -4.38903809e-01 -4.99929726e-01 -6.67805076e-01 -9.09652352e-01 -5.19723594e-01 -6.09166622e-01 5.38367450e-01 6.11333907e-01 -8.08293760e-01 -1.28349930e-01 -1.85770303e-01 3.35439175e-01 9.54509318e-01 5.66694319e-01 1.09946907e+00 -1.34294009e+00 -3.61912489e-01 -1.79332137e-01 -7.08340049e-01 -1.13120759e+00 5.63970581e-02 -6.68646753e-01 7.05946311e-02 -1.33834994e+00 -3.01907826e-02 -7.36259162e-01 7.75281265e-02 2.28173077e-01 5.47372460e-01 8.20586085e-01 -1.44489437e-01 4.59846854e-01 -5.46823919e-01 5.34552336e-01 9.67269123e-01 -5.31247377e-01 -4.51444946e-02 -2.02838063e-01 -2.54202068e-01 7.69150555e-01 6.45457804e-01 2.55824238e-01 -3.13448250e-01 -6.69887483e-01 4.19671297e-01 -3.01457167e-01 6.58982933e-01 -1.43736148e+00 -2.25834846e-02 -3.93183738e-01 3.61527979e-01 -4.99603719e-01 1.09834397e+00 -1.12358475e+00 4.25861567e-01 1.07125148e-01 2.49266513e-02 5.27879223e-02 6.03330284e-02 7.52718210e-01 1.97939306e-01 3.52126926e-01 5.25115192e-01 -2.59439081e-01 -7.88916469e-01 3.04393977e-01 -1.21770300e-01 -3.64627451e-01 7.41624713e-01 -7.70042896e-01 -6.80488169e-01 -2.09736273e-01 -4.72845376e-01 -4.07407135e-01 1.16254723e+00 -1.83294341e-02 1.02710187e+00 -1.23970187e+00 -3.80035102e-01 7.21801639e-01 3.36379081e-01 -3.38875912e-02 2.62894034e-01 6.68429494e-01 -1.00831449e+00 2.38150749e-02 -5.00634074e-01 -9.25538778e-01 -1.44293237e+00 -1.50797352e-01 4.41105515e-01 2.04357117e-01 -8.43653381e-01 6.56046271e-01 1.30929649e-01 -4.49843913e-01 -1.79049924e-01 -3.76052618e-01 3.04260641e-01 -2.59769052e-01 2.68023670e-01 3.14551920e-01 4.39152330e-01 -6.63634777e-01 -1.37727186e-01 1.18958700e+00 9.05454159e-02 -2.75553524e-01 1.62598300e+00 -2.44433269e-01 -1.73357919e-01 6.44011378e-01 1.02967334e+00 2.27199093e-01 -1.54893005e+00 -3.56938362e-01 -4.85941440e-01 -7.67958403e-01 7.57275661e-03 -6.03962004e-01 -9.55383956e-01 6.50280476e-01 8.05042148e-01 4.18249905e-01 1.01678514e+00 -6.26433492e-02 5.52393615e-01 9.55584943e-01 1.07292986e+00 -1.10180914e+00 4.41547006e-01 2.63846248e-01 1.02220583e+00 -1.24217367e+00 1.19079247e-01 -6.52226269e-01 -1.40736839e-02 1.32354808e+00 3.21799368e-01 -1.86390549e-01 6.05900586e-01 3.20968062e-01 2.61268020e-01 -6.92422986e-01 -5.15956283e-01 -5.19970544e-02 3.26871157e-01 7.35309005e-01 -1.45373404e-01 3.90694231e-01 7.25560963e-01 -2.96508551e-01 -3.32665622e-01 -4.34549689e-01 2.69361556e-01 1.26102614e+00 -6.37867153e-01 -8.80072057e-01 -6.40063822e-01 5.70325494e-01 1.00347020e-01 9.85630825e-02 -5.60628235e-01 5.78406870e-01 -1.41926497e-01 7.63750494e-01 -4.15631570e-02 -6.04348183e-01 4.89153594e-01 -1.41582027e-01 7.77600765e-01 -3.59594852e-01 6.45184368e-02 -2.06627950e-01 3.60731125e-01 -8.59653234e-01 -7.94096887e-01 -7.16637075e-01 -6.90605223e-01 -2.34355614e-01 -2.61137635e-01 -4.65015322e-01 1.09546804e+00 7.54572570e-01 3.25093806e-01 3.00324678e-01 8.32930803e-01 -1.63388658e+00 -1.79384679e-01 -5.40000796e-01 -8.32046032e-01 2.85085976e-01 3.80974710e-01 -9.76392686e-01 -4.16638553e-01 2.32099861e-01]
[9.597152709960938, -2.800967216491699]
a0e9ea56-8689-420c-aacf-247ffb1cfd14
progressive-training-of-a-two-stage-framework
2204.09924
null
https://arxiv.org/abs/2204.09924v2
https://arxiv.org/pdf/2204.09924v2.pdf
Progressive Training of A Two-Stage Framework for Video Restoration
As a widely studied task, video restoration aims to enhance the quality of the videos with multiple potential degradations, such as noises, blurs and compression artifacts. Among video restorations, compressed video quality enhancement and video super-resolution are two of the main tacks with significant values in practical scenarios. Recently, recurrent neural networks and transformers attract increasing research interests in this field, due to their impressive capability in sequence-to-sequence modeling. However, the training of these models is not only costly but also relatively hard to converge, with gradient exploding and vanishing problems. To cope with these problems, we proposed a two-stage framework including a multi-frame recurrent network and a single-frame transformer. Besides, multiple training strategies, such as transfer learning and progressive training, are developed to shorten the training time and improve the model performance. Benefiting from the above technical contributions, our solution wins two champions and a runner-up in the NTIRE 2022 super-resolution and quality enhancement of compressed video challenges. Code is available at https://github.com/ryanxingql/winner-ntire22-vqe.
['Ying Chen', 'Huaida Liu', 'Lai Jiang', 'Mai Xu', 'Minglang Qiao', 'Qunliang Xing', 'Meisong Zheng']
2022-04-21
null
null
null
null
['video-super-resolution', 'video-restoration']
['computer-vision', 'computer-vision']
[ 3.29304159e-01 -3.93088400e-01 -1.52249068e-01 -1.90314233e-01 -8.94611180e-01 1.06468305e-01 1.09773621e-01 -4.79560465e-01 -2.66323984e-02 7.16340423e-01 4.54026699e-01 -1.69351436e-02 -1.80268034e-01 -5.20073891e-01 -6.69350207e-01 -5.85360110e-01 -3.03524230e-02 -2.43339524e-01 1.85338527e-01 -3.02867144e-01 1.71170175e-01 -5.29331975e-02 -1.50061274e+00 4.05883461e-01 1.16689777e+00 1.13347793e+00 6.72680140e-01 6.32017732e-01 1.33531034e-01 1.30974126e+00 -1.88296065e-01 -4.03809875e-01 1.32522747e-01 -4.86682445e-01 -5.58802724e-01 1.25011489e-01 2.47475192e-01 -6.59400225e-01 -8.51597846e-01 1.30890405e+00 6.56786144e-01 1.44957855e-01 -2.57939287e-02 -9.52759862e-01 -7.70508230e-01 5.70408463e-01 -6.97779655e-01 4.87128377e-01 2.30916470e-01 1.25161424e-01 8.31912816e-01 -1.09052086e+00 3.15748006e-01 1.20694125e+00 6.13679767e-01 6.03032112e-01 -7.30854094e-01 -6.95866585e-01 -4.24938761e-02 9.54978764e-01 -1.13447022e+00 -8.37002277e-01 6.84168398e-01 -1.99955136e-01 6.63834214e-01 1.26106843e-01 6.88491404e-01 1.08181107e+00 2.19261218e-02 9.02289689e-01 7.63208628e-01 -2.24886760e-02 -7.35632703e-02 -2.53699392e-01 -3.82989436e-01 3.59507054e-01 -2.33907953e-01 8.52016807e-02 -5.04044056e-01 3.04486632e-01 1.14745128e+00 1.93973795e-01 -7.47651756e-01 2.59280670e-02 -9.22119081e-01 4.73466516e-01 5.33025265e-01 2.46763885e-01 -5.22194743e-01 2.24158525e-01 5.46648562e-01 2.47173414e-01 5.64692557e-01 8.87633190e-02 -1.93462372e-01 -4.61738527e-01 -1.07108569e+00 1.75109223e-01 7.30185434e-02 7.71890283e-01 3.80736470e-01 4.41943318e-01 -1.83454212e-02 1.12434435e+00 9.35149565e-02 3.37013662e-01 5.85852563e-01 -1.34229469e+00 7.06743300e-01 1.21367589e-01 3.64709884e-01 -1.18725455e+00 -8.31162706e-02 -6.51233613e-01 -1.41749084e+00 8.54308717e-03 1.41636282e-01 -4.40132171e-02 -5.25686085e-01 1.67282379e+00 1.13423653e-01 6.95303619e-01 -2.15761736e-01 1.21383822e+00 7.14357793e-01 1.01202261e+00 -5.75355925e-02 -5.73827982e-01 1.17448485e+00 -1.21117616e+00 -1.10485709e+00 -1.63880289e-01 1.37339281e-02 -8.27192128e-01 8.49746168e-01 5.78327477e-01 -1.58034968e+00 -8.47878397e-01 -1.09792340e+00 -2.78584510e-01 3.71824682e-01 5.85911162e-02 3.26705724e-01 1.45906314e-01 -1.10135996e+00 8.92539382e-01 -7.62058318e-01 1.20154865e-01 5.95532417e-01 1.64495170e-01 2.66035087e-02 -3.73730361e-01 -1.36679411e+00 6.42980814e-01 1.51108921e-01 4.41437155e-01 -9.27091420e-01 -6.78955972e-01 -5.38077235e-01 1.93437621e-01 7.22117484e-01 -5.71712673e-01 1.15750480e+00 -9.96791959e-01 -1.52631712e+00 3.41376871e-01 -1.56266049e-01 -3.92003030e-01 6.22792542e-01 -6.06943011e-01 -6.83903992e-01 2.16837302e-01 -6.28115609e-02 2.72423953e-01 1.09011066e+00 -9.48635399e-01 -8.82033706e-01 -2.52640873e-01 -7.95242637e-02 2.88250566e-01 -4.81523007e-01 2.58656323e-01 -6.16339445e-01 -7.90635049e-01 3.77430581e-02 -4.92476672e-01 -2.97962785e-01 7.15096947e-03 -6.21947907e-02 6.11709012e-03 8.33207190e-01 -1.29854870e+00 1.44599116e+00 -2.18813753e+00 3.90244961e-01 -5.04550934e-01 4.43960041e-01 6.16478741e-01 -2.42947727e-01 4.59307758e-03 -1.06027260e-01 1.21949553e-01 -6.02677874e-02 -2.23574400e-01 -4.05074865e-01 -9.89513695e-02 -4.18850243e-01 2.69239813e-01 1.38641357e-01 8.02808225e-01 -9.39629674e-01 -3.61937255e-01 2.86262840e-01 8.87361825e-01 -4.58600372e-01 3.77203733e-01 -3.56567353e-02 6.34989321e-01 -3.69069785e-01 5.12868941e-01 7.46782959e-01 -5.49391091e-01 1.00974329e-01 -5.00273049e-01 -2.10268185e-01 1.26943633e-01 -1.10323703e+00 1.65751159e+00 -4.88850474e-01 4.58597213e-01 3.80197376e-01 -1.07145786e+00 6.59301817e-01 6.09039962e-01 6.23566687e-01 -1.17833567e+00 1.90095920e-02 4.06423330e-01 -2.70805150e-01 -8.07701766e-01 7.90084183e-01 -1.11021265e-01 5.62008977e-01 2.48601986e-03 -2.88927615e-01 3.12529236e-01 1.79977268e-01 9.93084684e-02 9.71071661e-01 3.47396523e-01 7.69133493e-02 2.30698600e-01 6.84071064e-01 -6.08377814e-01 1.07682586e+00 1.79808855e-01 -3.08685958e-01 8.53705645e-01 2.42528856e-01 -3.68003547e-01 -1.44118416e+00 -7.53206551e-01 1.44465655e-01 7.95325994e-01 3.16022664e-01 -3.40676725e-01 -6.59660101e-01 -7.22808689e-02 -5.67924261e-01 3.68448108e-01 -1.72409549e-01 -1.65346861e-01 -8.93318713e-01 -5.48692226e-01 1.15418859e-01 4.05525208e-01 7.92607725e-01 -1.04671812e+00 -5.50766945e-01 4.54894304e-01 -8.19283903e-01 -1.40853751e+00 -5.76666892e-01 -4.02241409e-01 -1.05036938e+00 -9.43842649e-01 -1.24312162e+00 -7.81278491e-01 7.63680711e-02 6.45470917e-01 9.54055011e-01 1.93580762e-01 -3.33237909e-02 -9.64681655e-02 -4.34361637e-01 1.65340841e-01 -3.20048451e-01 -1.81211099e-01 4.68744151e-02 1.41360223e-01 -1.43807679e-01 -9.03739750e-01 -9.16944444e-01 4.05763656e-01 -9.49339747e-01 3.33308518e-01 5.48945665e-01 8.99480879e-01 5.79889238e-01 2.09279761e-01 6.95043623e-01 -3.68459463e-01 4.86076623e-01 -4.61712331e-01 -6.18504822e-01 1.39220849e-01 -5.51507771e-01 -3.33725452e-01 9.81231689e-01 -3.46661061e-01 -1.24721694e+00 -4.06081617e-01 -2.06414565e-01 -9.36537981e-01 3.58226061e-01 4.96278167e-01 -2.63519078e-01 1.19231157e-01 3.60778540e-01 4.67623979e-01 1.00897392e-02 -6.59722030e-01 1.40415981e-01 6.31389022e-01 6.86380982e-01 -7.39956498e-02 7.81072080e-01 2.68845528e-01 -2.80439913e-01 -6.91375852e-01 -8.30359876e-01 -3.07268173e-01 -9.35085416e-02 -4.11335856e-01 6.49912238e-01 -1.24883759e+00 -5.73038578e-01 7.57952809e-01 -1.15446818e+00 -1.88166276e-01 -1.48033306e-01 4.82833654e-01 -5.54181695e-01 7.42167711e-01 -1.05279505e+00 -7.68011570e-01 -6.06026113e-01 -1.28162861e+00 4.91513014e-01 5.74712217e-01 5.05283117e-01 -5.84975600e-01 -2.64652729e-01 5.53435802e-01 9.66130257e-01 -1.34722978e-01 6.53115511e-01 1.85525447e-01 -8.80321741e-01 1.59480497e-01 -5.58790743e-01 7.70493209e-01 2.12007880e-01 -2.50138789e-01 -6.76176548e-01 -4.47970897e-01 4.21812862e-01 -3.25714320e-01 8.00090551e-01 5.72487652e-01 1.34006846e+00 -4.56403464e-01 6.58223554e-02 9.10433769e-01 1.40470469e+00 3.68861943e-01 1.04020047e+00 3.54574144e-01 7.19682455e-01 2.73393601e-01 5.98570406e-01 6.97129607e-01 3.69482130e-01 8.84510219e-01 6.03866816e-01 -6.57905191e-02 -3.85572314e-01 -2.78569192e-01 4.39485788e-01 1.14233375e+00 -4.73631620e-01 -8.28762800e-02 -5.51438272e-01 4.92474318e-01 -1.84015799e+00 -1.23630023e+00 -8.77107084e-02 2.10847688e+00 8.25350881e-01 -1.36699509e-02 1.92520339e-02 1.48039669e-01 1.03531444e+00 3.86840791e-01 -7.40478635e-01 1.59244984e-01 -3.16702515e-01 -1.45351514e-01 1.37449369e-01 3.92877460e-01 -9.35468137e-01 7.97854125e-01 5.11228514e+00 1.01981688e+00 -1.20694256e+00 1.80184126e-01 9.25329447e-01 -2.08880574e-01 -5.69272786e-02 -1.31126598e-01 -4.15136963e-01 7.35437512e-01 9.61114526e-01 -1.81549698e-01 8.94920886e-01 6.59984112e-01 7.48557627e-01 2.40342125e-01 -4.73119497e-01 1.32298541e+00 -1.78179902e-03 -1.35946250e+00 -8.15886597e-05 -1.55700102e-01 7.04890490e-01 7.32689574e-02 2.95148432e-01 1.88441247e-01 -2.40956008e-01 -1.00998116e+00 6.55627251e-01 4.83438879e-01 9.74137008e-01 -7.40462184e-01 6.22969985e-01 1.90033987e-01 -1.34070492e+00 -3.57926399e-01 -4.64614958e-01 -8.47174972e-02 6.65833414e-01 7.32986331e-01 2.10795432e-01 6.70790792e-01 9.21050191e-01 1.19669628e+00 -1.57261044e-01 1.32123220e+00 -2.52090573e-01 5.28612018e-01 1.41894415e-01 5.87500632e-01 -2.18354072e-02 -3.69286031e-01 6.87734008e-01 8.06333899e-01 6.14452302e-01 2.91326106e-01 -3.99127379e-02 6.91235840e-01 -1.59091532e-01 -7.31623024e-02 8.21500085e-03 2.55151652e-02 3.13727885e-01 1.27496004e+00 -8.55103731e-02 -3.99237156e-01 -4.98532265e-01 9.11416054e-01 3.36050875e-02 4.97042239e-01 -1.08991253e+00 -1.73150122e-01 6.63368344e-01 3.19985628e-01 4.43694025e-01 -1.70556709e-01 -2.87267454e-02 -1.48193872e+00 2.44170725e-01 -1.31161284e+00 -9.76592302e-03 -9.45373654e-01 -1.03887260e+00 8.10888231e-01 -4.40257400e-01 -1.58822119e+00 -1.32096037e-01 -2.24069566e-01 -2.92581975e-01 7.79524982e-01 -1.92031193e+00 -8.45031142e-01 -5.07927358e-01 6.43614888e-01 9.38087642e-01 -3.22680846e-02 2.56480426e-01 9.48053658e-01 -7.54123449e-01 4.57354963e-01 3.53726476e-01 -7.56957307e-02 5.60558915e-01 -6.40188158e-01 3.00573826e-01 1.10162103e+00 -3.59461099e-01 2.88848042e-01 7.62873828e-01 -4.77241814e-01 -1.42974126e+00 -1.05010283e+00 5.69420874e-01 3.33627760e-01 6.76312208e-01 7.06151426e-02 -1.23228121e+00 3.89494985e-01 1.75069809e-01 4.02043387e-02 9.45964083e-02 -2.80352831e-01 -1.65552825e-01 -2.48472795e-01 -7.82839656e-01 5.40213346e-01 9.86945331e-01 -5.23915529e-01 -1.12647526e-01 2.45875657e-01 7.41602659e-01 -5.59711397e-01 -9.17879939e-01 4.67044175e-01 4.03644770e-01 -1.35582888e+00 1.08988070e+00 -1.30182371e-01 1.16223669e+00 -3.06405544e-01 -2.27223977e-01 -1.04590929e+00 -4.41946894e-01 -8.69357288e-01 -4.75629061e-01 1.20095634e+00 1.97098497e-02 -2.43743956e-01 6.66648746e-01 3.72232229e-01 -2.46115223e-01 -8.41623545e-01 -9.55221236e-01 -5.48969090e-01 -3.40403050e-01 -2.64340937e-01 3.74198973e-01 7.33303130e-01 -2.38940477e-01 4.96745318e-01 -1.25665832e+00 6.46418259e-02 8.08982730e-01 -2.02864215e-01 4.06053782e-01 -6.99790359e-01 -4.69997138e-01 -3.87955934e-01 -8.30242708e-02 -1.55474520e+00 -4.18457091e-01 -5.37449539e-01 -3.88639271e-02 -1.59266019e+00 3.67898643e-01 -1.62838593e-01 -3.51745546e-01 1.69111997e-01 -3.88979882e-01 1.48226589e-01 4.42834169e-01 5.28401434e-01 -8.64832878e-01 1.02876115e+00 1.52153468e+00 -8.09102654e-02 -1.89603474e-02 1.07071020e-01 -6.13965094e-01 5.62027931e-01 6.72176182e-01 -2.10081473e-01 -4.21708792e-01 -8.15327764e-01 2.81755924e-01 7.93360174e-01 3.47389936e-01 -1.14807725e+00 2.60278344e-01 -4.54567745e-03 2.25030020e-01 -4.52752709e-01 6.29770458e-01 -6.03200316e-01 2.16977566e-01 4.24204677e-01 -2.70840168e-01 1.61901921e-01 -7.52327964e-02 4.82689291e-01 -5.57464063e-01 -1.94174051e-02 1.01668525e+00 -3.12975943e-01 -7.96635091e-01 6.43373787e-01 -1.04627624e-01 4.63464558e-02 5.61371446e-01 -2.49528065e-01 -3.44072282e-01 -7.78584480e-01 -5.54952979e-01 3.85785401e-01 3.09037358e-01 5.28037965e-01 1.00661159e+00 -1.20938516e+00 -1.11709261e+00 -8.46570060e-02 -5.43101490e-01 -7.84161910e-02 1.10550058e+00 8.88207853e-01 -2.91955322e-01 2.14108393e-01 -2.92729169e-01 -4.00326550e-01 -1.12932467e+00 6.21571302e-01 4.09604222e-01 -4.85159576e-01 -7.71956205e-01 6.25764489e-01 7.07005262e-02 1.82879135e-01 1.91590488e-01 1.27145588e-01 -4.27443624e-01 -1.86800882e-01 1.01488972e+00 7.65460551e-01 -1.92804694e-01 -6.24793351e-01 -1.03037385e-02 5.71820796e-01 -2.17837200e-01 2.35161766e-01 1.51026213e+00 -6.40438974e-01 -3.72026227e-02 7.77591765e-02 9.96349037e-01 -2.26112336e-01 -1.54458141e+00 -4.39642370e-01 -2.16154158e-01 -6.66495502e-01 2.96912849e-01 -7.05639482e-01 -1.46329045e+00 9.01272595e-01 7.86973417e-01 8.97530690e-02 1.59134352e+00 -4.81359512e-01 1.31658149e+00 -2.91842550e-01 3.00563425e-01 -9.52163160e-01 3.63114446e-01 4.69298840e-01 1.01677585e+00 -1.25205362e+00 4.40604053e-02 -3.30848068e-01 -5.52766919e-01 9.81463909e-01 6.02342486e-01 -9.64403674e-02 2.66835153e-01 1.07634582e-01 1.32690798e-02 2.87826836e-01 -8.13096642e-01 2.35870242e-01 6.10235520e-02 4.17134434e-01 4.95436400e-01 -2.48183101e-01 -3.77852082e-01 5.63222289e-01 7.91623145e-02 5.11700749e-01 6.60822093e-01 4.20707703e-01 -3.40194702e-01 -9.46031630e-01 -2.03458458e-01 3.88368130e-01 -6.67708039e-01 -3.94096762e-01 6.56237662e-01 2.90603995e-01 -2.18341965e-02 1.07500374e+00 -3.19288254e-01 -5.71989238e-01 2.18182802e-01 -5.15816450e-01 2.75199682e-01 -5.39499074e-02 -3.19133788e-01 3.21093857e-01 -1.92968443e-01 -8.11429143e-01 -4.57801223e-01 -4.30297256e-01 -9.38278854e-01 -4.41336185e-01 -2.68107235e-01 9.03294533e-02 4.39124078e-01 8.20368290e-01 4.74266171e-01 8.39487731e-01 8.90233040e-01 -9.43313837e-01 -7.27838695e-01 -9.40110743e-01 -5.56609869e-01 2.65396237e-01 4.38297004e-01 -2.99063712e-01 -2.53155798e-01 4.37549688e-02]
[11.159753799438477, -1.8987257480621338]
0134ba87-7a4c-476c-863b-2b18d050cab6
permutation-invariant-strategy-using
null
null
https://aclanthology.org/2022.findings-naacl.59
https://aclanthology.org/2022.findings-naacl.59.pdf
Permutation Invariant Strategy Using Transformer Encoders for Table Understanding
Representing text in tables is essential for many business intelligence tasks such as semantic retrieval, data exploration and visualization, and question answering. Existing methods that leverage pretrained Transformer encoders range from a simple construction of pseudo-sentences by concatenating text across rows or columns to complex parameter-intensive models that encode table structure and require additional pretraining. In this work, we introduce a novel encoding strategy for Transformer encoders that preserves the critical property of permutation invariance across rows or columns. Unlike existing state-of-the-art methods for Table Understanding, our proposed approach does not require any additional pretraining and still substantially outperforms existing methods in almost all instances. We demonstrate the effectiveness of our proposed approach on three table interpretation tasks: column type annotation, relation extraction, and entity linking through extensive experiments on existing tabular datasets.
['Alfio Gliozzo', 'Nandana Mihindukulasooriya', 'Sugato Bagchi', 'Sarthak Dash']
null
null
null
null
findings-naacl-2022-7
['semantic-retrieval', 'column-type-annotation']
['natural-language-processing', 'natural-language-processing']
[ 3.93827587e-01 4.62721467e-01 -5.00037074e-01 -5.48458397e-01 -8.91115725e-01 -7.64028132e-01 5.67624986e-01 8.78394902e-01 -6.99596033e-02 9.30484772e-01 3.98886800e-01 -7.43404865e-01 -2.99902469e-01 -1.13843942e+00 -1.23099542e+00 8.00518319e-02 1.14155710e-01 7.20785677e-01 2.71880597e-01 -3.24878812e-01 8.41683522e-02 9.38866660e-02 -1.27888465e+00 7.52323866e-01 8.96222472e-01 1.27224326e+00 -3.00861418e-01 2.23260090e-01 -6.89523637e-01 1.12057436e+00 -4.50818509e-01 -9.93534386e-01 1.43235788e-01 -2.65567631e-01 -1.07178628e+00 -1.32574327e-03 3.69787395e-01 -3.06170106e-01 -4.81679171e-01 9.93729949e-01 7.14627504e-02 -1.25425786e-01 3.74503613e-01 -1.05595338e+00 -9.00200427e-01 1.34747708e+00 -3.43071997e-01 -3.29616405e-02 3.50739300e-01 -4.09275204e-01 1.43841445e+00 -6.61716461e-01 7.13854849e-01 1.36939108e+00 5.73481679e-01 9.51498598e-02 -1.22164547e+00 -4.82825011e-01 2.36452982e-01 2.42846519e-01 -1.13554108e+00 -4.36336339e-01 6.57832086e-01 -3.01356554e-01 1.00351608e+00 2.08063737e-01 1.33951053e-01 8.02058518e-01 -2.22714543e-01 8.60115588e-01 7.35844374e-01 -1.70465171e-01 8.12522694e-02 4.61718202e-01 2.66740620e-01 8.23787391e-01 8.16770852e-01 -7.21261442e-01 -5.88076591e-01 -3.05274175e-03 5.61349332e-01 -1.76767319e-01 -2.18957160e-02 -6.68990374e-01 -1.43886697e+00 7.25483477e-01 4.02967215e-01 -1.36097267e-01 -1.03695430e-01 1.78160548e-01 7.86997437e-01 1.96793690e-01 1.30972013e-01 4.53966379e-01 -7.49579251e-01 -3.83533053e-02 -4.23670173e-01 1.82817385e-01 1.01065612e+00 1.52335668e+00 7.19722688e-01 -3.73167902e-01 -2.59476751e-01 7.06239641e-01 3.26023698e-02 3.82908583e-01 1.96127519e-01 -7.65612364e-01 1.19277453e+00 1.07066119e+00 2.26944219e-02 -9.84019756e-01 -1.66161478e-01 -3.11612189e-01 -7.28322029e-01 -7.01798439e-01 2.84947306e-01 1.67622063e-02 -7.91861773e-01 1.54636490e+00 1.92993447e-01 -4.21197891e-01 4.21054602e-01 3.47052336e-01 1.03056908e+00 4.59699720e-01 -1.19306356e-01 1.09569333e-03 1.60920286e+00 -7.74408817e-01 -1.13186145e+00 -3.82096052e-01 7.64148593e-01 -4.56845224e-01 1.12062013e+00 -5.70508316e-02 -1.09746110e+00 -2.63229340e-01 -1.22509527e+00 -6.34972155e-01 -9.01945531e-01 2.17766047e-01 1.11221576e+00 6.39617085e-01 -4.51723367e-01 4.18467373e-01 -5.67019761e-01 -3.62794966e-01 7.48335481e-01 3.66474956e-01 -5.40214241e-01 4.57215384e-02 -1.22133195e+00 5.56380689e-01 6.00138664e-01 1.78262696e-01 -2.81343520e-01 -7.99797595e-01 -1.27973235e+00 5.98916769e-01 8.54527891e-01 -7.39301622e-01 1.23412514e+00 -2.31634840e-01 -1.15271688e+00 6.12961709e-01 -3.12966436e-01 -8.54428709e-01 2.70055532e-01 -4.99596119e-01 -2.74290562e-01 -1.16714895e-01 1.47862598e-01 4.19474632e-01 9.04372111e-02 -1.19747627e+00 -2.32999623e-01 -4.02250350e-01 2.82615066e-01 1.66571930e-01 -4.51243371e-01 -1.58557162e-01 -6.26451254e-01 -6.63703263e-01 3.09063524e-01 -4.53583419e-01 1.31914273e-01 -9.11594629e-02 -9.84377205e-01 -1.54234350e-01 6.12508237e-01 -6.41765058e-01 1.45872056e+00 -1.87516117e+00 -6.13617189e-02 2.21997350e-01 7.69743472e-02 -1.56537779e-02 2.88712233e-01 7.14519382e-01 2.80028228e-02 3.76625866e-01 -4.58120167e-01 3.61322649e-02 5.03525496e-01 1.65960103e-01 -5.84036469e-01 -9.88180935e-02 3.63033742e-01 1.04289436e+00 -6.44380987e-01 -6.90986097e-01 9.56855193e-02 2.16357857e-01 -7.96562672e-01 3.48025709e-02 -5.46476066e-01 -2.05117628e-01 -2.94918299e-01 8.09723437e-01 5.28946400e-01 -6.74865544e-01 7.56119370e-01 -4.73612815e-01 3.06983292e-01 1.06692314e+00 -1.33494735e+00 1.67886543e+00 -3.48151624e-01 4.91027832e-01 -4.03602749e-01 -1.11454761e+00 8.70253146e-01 8.40915740e-02 3.67671698e-01 -7.82776237e-01 5.53114936e-02 2.43340328e-01 -9.96929258e-02 -2.79014587e-01 6.34457767e-01 3.01069230e-01 -3.83546472e-01 3.79277617e-01 -4.39068042e-02 8.13832507e-02 6.47550762e-01 4.21152711e-01 1.03334296e+00 1.47852242e-01 4.82019931e-01 2.28439015e-03 6.50974154e-01 3.51549014e-02 5.48027337e-01 6.69680834e-01 4.19004291e-01 2.23874971e-01 1.02201653e+00 -4.34850544e-01 -9.80984747e-01 -1.00011706e+00 -1.73891783e-01 8.73400509e-01 6.85395226e-02 -9.96513367e-01 -7.13538349e-01 -9.28880095e-01 3.85268688e-01 7.30147123e-01 -6.76977158e-01 4.82518524e-02 -6.84790075e-01 -6.80355668e-01 4.91725266e-01 1.03188908e+00 7.89764941e-01 -8.81870270e-01 -2.02633992e-01 2.13484138e-01 -3.62570524e-01 -1.59696472e+00 -3.13665628e-01 4.79534477e-01 -8.22239399e-01 -1.24668682e+00 1.76255628e-01 -8.08592439e-01 7.10750103e-01 -1.45479575e-01 1.23436999e+00 -1.85249776e-01 1.24168973e-02 -1.47260159e-01 -4.75731045e-02 -4.80055004e-01 -3.12997907e-01 4.21984881e-01 -4.23861146e-01 -2.05955580e-01 4.92431909e-01 -2.72042036e-01 -3.02610636e-01 2.29220152e-01 -8.50339353e-01 3.83226126e-01 7.48767972e-01 8.35763693e-01 6.76509142e-01 9.39332992e-02 5.84092796e-01 -1.80691719e+00 5.45870245e-01 -3.97583157e-01 -5.82145989e-01 7.43843079e-01 -6.30053222e-01 7.02569246e-01 7.11340308e-01 1.10471398e-01 -1.08932495e+00 3.49040292e-02 -2.14038212e-02 -2.58382708e-02 3.76172774e-02 7.49982834e-01 -5.88220775e-01 5.15544772e-01 1.91866577e-01 2.05982383e-02 -3.10065001e-01 -5.18498540e-01 5.36224484e-01 5.41591525e-01 6.78705096e-01 -6.83730364e-01 8.93618405e-01 4.01961714e-01 1.08366899e-01 -1.07125193e-01 -1.03600073e+00 -6.07485175e-02 -8.34747493e-01 5.29104292e-01 6.91715121e-01 -7.65458643e-01 -1.02007186e+00 -1.31567508e-01 -9.35035229e-01 -9.25499275e-02 -2.13155136e-01 6.43313229e-02 -4.32182550e-01 1.68809101e-01 -5.19702077e-01 -4.49965954e-01 -3.13667476e-01 -9.45801437e-01 1.10477018e+00 -1.27448037e-01 -2.14650020e-01 -9.67285633e-01 -3.45130950e-01 5.80465198e-01 2.28892535e-01 1.39423192e-01 1.57478261e+00 -8.85495424e-01 -9.21222866e-01 -1.93988085e-01 -5.62608123e-01 -1.03810452e-01 3.95985395e-01 -3.01567495e-01 -6.45125389e-01 1.84967712e-01 -6.56023800e-01 -4.60122079e-01 8.73188555e-01 -1.97848111e-01 1.70749760e+00 -6.43926561e-01 -4.34817940e-01 8.18385065e-01 1.27245164e+00 3.87517095e-01 5.49643934e-01 6.38800740e-01 8.71654630e-01 6.64276242e-01 6.02751195e-01 3.18425983e-01 8.19562614e-01 4.64645028e-01 3.30511063e-01 5.21204760e-03 6.67759180e-02 -8.34770381e-01 -1.72397196e-01 6.22514069e-01 5.39956570e-01 -4.18340176e-01 -7.58468926e-01 5.31064689e-01 -1.99117744e+00 -7.85364091e-01 3.04689676e-01 2.09581518e+00 1.04595983e+00 4.68416870e-01 -2.01537162e-01 4.88463432e-01 4.72585082e-01 9.52285007e-02 -4.67930675e-01 -4.53587890e-01 -9.09566730e-02 2.56660968e-01 7.55886197e-01 2.99874038e-01 -1.27820861e+00 9.04934943e-01 6.06678915e+00 2.69849509e-01 -5.47088981e-01 -3.96438420e-01 7.37121761e-01 1.58825532e-01 -7.32736468e-01 1.11835897e-01 -8.98298979e-01 2.76286662e-01 6.96341991e-01 -3.45343798e-01 3.07236075e-01 7.98114300e-01 -2.91491657e-01 6.39485493e-02 -1.52016544e+00 8.66867602e-01 -1.24888578e-02 -1.71656358e+00 4.94539946e-01 -2.80741841e-01 3.52956504e-01 -7.12745965e-01 -1.14568714e-02 2.97903419e-01 3.72930884e-01 -1.11886489e+00 5.63672900e-01 7.51576647e-02 9.35339391e-01 -7.04149663e-01 8.41081917e-01 -2.20434561e-01 -1.25836265e+00 -1.73288107e-01 -1.82412490e-01 1.60178795e-01 -4.74685468e-02 3.21491957e-01 -9.39630568e-01 6.76036716e-01 6.49649203e-01 7.67369807e-01 -9.01419103e-01 4.94512647e-01 -1.28236100e-01 4.18843985e-01 -1.72301739e-01 -8.34945142e-02 1.03856111e-02 3.48117426e-02 -5.49014471e-03 1.17470813e+00 7.36060143e-02 -4.89259064e-02 -8.51820856e-02 7.47459590e-01 -7.08157539e-01 7.49974251e-02 -6.47069871e-01 -4.02248949e-01 7.99979687e-01 8.62501681e-01 -7.35044599e-01 -6.83872223e-01 -6.84026659e-01 5.64980686e-01 4.61176068e-01 2.61256844e-01 -7.66897738e-01 -7.49815524e-01 4.87979531e-01 2.02602521e-01 6.79042816e-01 -1.04118653e-01 -8.65275681e-01 -1.17455614e+00 4.14249837e-01 -9.62405145e-01 7.98372030e-01 -6.12926602e-01 -8.35901797e-01 5.95904589e-01 2.37578422e-01 -8.76684308e-01 -4.37455058e-01 -6.62357152e-01 -5.44635132e-02 4.95051026e-01 -1.52540350e+00 -1.10905910e+00 -2.66311467e-01 4.45494503e-01 3.80375832e-01 -6.81274310e-02 8.10890853e-01 3.34164560e-01 -7.51072645e-01 8.33866656e-01 -2.00279150e-02 5.71994007e-01 6.92859232e-01 -1.60399306e+00 7.27874875e-01 8.73266280e-01 1.98470697e-01 1.05654275e+00 5.56202888e-01 -6.15770400e-01 -1.82884181e+00 -9.99266624e-01 1.18733847e+00 -3.43140453e-01 7.99784958e-01 -8.98729920e-01 -9.54374373e-01 1.17592871e+00 4.03031707e-01 -1.05308682e-01 7.38856614e-01 3.89238685e-01 -7.41161823e-01 -5.99666655e-01 -8.96000922e-01 4.37428176e-01 1.13152087e+00 -6.18742228e-01 -6.10975802e-01 3.89020354e-01 8.57190549e-01 -7.08627164e-01 -1.03745973e+00 5.90793490e-01 6.33455992e-01 -6.00112081e-01 9.59367037e-01 -7.63330758e-01 6.95246100e-01 -2.98703671e-01 -2.46039465e-01 -9.24547791e-01 -3.70355397e-02 -6.33528709e-01 -4.80241865e-01 1.34789765e+00 7.17822015e-01 -4.84082103e-01 8.98081422e-01 6.40980721e-01 -1.00204699e-01 -7.64366865e-01 -3.39360386e-01 -4.65641290e-01 -1.83051899e-01 -2.42538631e-01 1.06355226e+00 9.07912254e-01 2.69279629e-01 5.28656840e-01 -6.84726089e-02 1.64929882e-01 5.32119095e-01 5.77619016e-01 8.86554003e-01 -1.11024415e+00 -2.05037728e-01 -2.34231398e-01 -1.31739736e-01 -1.05594277e+00 8.06968138e-02 -9.35172498e-01 -1.62694842e-01 -1.95868087e+00 3.43502730e-01 -4.07492191e-01 -2.12766722e-01 7.18391180e-01 -2.37882912e-01 -8.24915692e-02 -1.07826956e-01 -1.46283388e-01 -6.98249102e-01 3.09230059e-01 1.05674934e+00 -3.29636186e-01 1.00975342e-01 -4.01221126e-01 -1.21584821e+00 4.20943201e-01 5.83963335e-01 -4.32082236e-01 -7.58145332e-01 -6.71065271e-01 6.74611151e-01 1.96060687e-01 5.36818057e-02 -7.17783093e-01 1.86592817e-01 -1.45105273e-01 4.69613969e-01 -8.58973265e-01 6.23493418e-02 -7.78201878e-01 -7.93430656e-02 1.40609980e-01 -6.50879443e-01 3.62835765e-01 4.35643047e-01 4.26788062e-01 -4.66266304e-01 1.23633549e-01 2.75593728e-01 -1.19584883e-02 -5.16876519e-01 2.01673135e-01 5.81632815e-02 2.96939164e-01 6.57016575e-01 1.05118148e-01 -7.09606528e-01 -2.36690968e-01 -3.50318700e-01 3.75965953e-01 3.04439902e-01 5.58247983e-01 4.68801439e-01 -1.20488012e+00 -2.83675045e-01 3.60904396e-01 2.87899345e-01 3.23444843e-01 -2.01947287e-01 3.99965972e-01 -6.54883623e-01 1.01582801e+00 -1.68201908e-01 -1.67423323e-01 -1.04063475e+00 7.12372541e-01 8.16422403e-02 -6.23061180e-01 -5.73388994e-01 5.04646301e-01 3.15487266e-01 -5.01755595e-01 4.54118401e-01 -7.41094291e-01 -1.41524360e-01 1.25249783e-02 2.12785259e-01 -5.96755557e-02 4.33410466e-01 -2.66400781e-02 -5.20085037e-01 1.79052308e-01 -4.60081160e-01 1.57754734e-01 1.19598019e+00 3.87940109e-02 -1.01637058e-01 4.55506772e-01 1.00144994e+00 5.35891168e-02 -7.29098320e-01 -3.11479092e-01 3.72716695e-01 -3.45254809e-01 -3.29920799e-01 -9.09302115e-01 -8.08538198e-01 7.50284791e-01 -1.20489873e-01 1.09229766e-01 1.06452203e+00 -6.62856996e-02 9.37410295e-01 1.02693641e+00 1.58321470e-01 -8.80696535e-01 1.18710008e-02 4.50540274e-01 6.14289343e-01 -1.34201264e+00 1.38749138e-01 -9.81501520e-01 -6.10399365e-01 1.04648757e+00 7.14751840e-01 3.83612573e-01 5.25136828e-01 6.03022575e-01 -1.91422045e-01 -2.62263656e-01 -1.12041104e+00 -2.85598397e-01 2.96662092e-01 2.93108851e-01 7.67156482e-01 -1.20020062e-01 -2.93918878e-01 6.87406480e-01 -4.51282740e-01 -2.33150665e-02 3.98541749e-01 1.14866710e+00 -1.07208230e-01 -1.24331069e+00 1.21102639e-01 7.51166761e-01 -4.57274377e-01 -3.45854849e-01 -5.53740323e-01 1.01363444e+00 -2.40233660e-01 5.77483952e-01 3.14078689e-01 -2.27453738e-01 3.74203473e-01 3.31020981e-01 3.83444101e-01 -6.38517141e-01 -4.81953233e-01 -2.94167399e-01 6.25435174e-01 -3.80448729e-01 -2.50995815e-01 -7.51226306e-01 -1.33351505e+00 -4.21736449e-01 -1.22987725e-01 2.76806295e-01 4.25081342e-01 7.45246470e-01 5.63110054e-01 8.66783798e-01 2.68346518e-01 2.72577018e-01 -3.12715620e-01 -1.00029969e+00 -1.86252519e-01 5.17499626e-01 1.34803474e-01 -7.41669357e-01 1.71623006e-01 1.42187119e-01]
[9.574139595031738, 7.852672576904297]
c0af878f-1e22-4931-af8e-fea1f93f740a
ultra-high-definition-image-hdr
null
null
http://openaccess.thecvf.com//content/ICCV2021/html/Zheng_Ultra-High-Definition_Image_HDR_Reconstruction_via_Collaborative_Bilateral_Learning_ICCV_2021_paper.html
http://openaccess.thecvf.com//content/ICCV2021/papers/Zheng_Ultra-High-Definition_Image_HDR_Reconstruction_via_Collaborative_Bilateral_Learning_ICCV_2021_paper.pdf
Ultra-High-Definition Image HDR Reconstruction via Collaborative Bilateral Learning
Existing single image high dynamic range (HDR) reconstruction attempt to expand the range of luminance. They are not effective to generate plausible textures and colors in the reconstructed results, especially for high-density pixels in ultra-high-definition (UHD) images.To address these problems, we propose a new HDR reconstruction network for UHD images by collaboratively learning color and texture details. First, we propose a dual-path network to extract content and chromatic features at a reduced resolution of the low dynamic range (LDR) input. These two types features are used to fit bilatera-space affine models for real-time HDR reconstruction. To extract the main data structure of the LDR input, we propose to use 3D Tucker decomposition and reconstruction to prevents false edges and noise amplification in the learned bilateral grid. As a result, the high-quality content and chromatic features can be reconstructed capitalized on guided bilateral upsampling. Finally, we fuse these two full-resolution feature maps into the HDR reconstructed results.Our proposed method can achieve real-time processing for UHD image (about 160 fps).Experimental results demonstrate that the proposed algorithm performs favorably against the state-of-the-art HDR reconstruction approaches on public benchmarks and real-world UHD images.
['Xiuyi Jia', 'Tao Wang', 'Xiaochun Cao', 'Wenqi Ren', 'Zhuoran Zheng']
2021-01-01
null
null
null
iccv-2021-1
['hdr-reconstruction']
['computer-vision']
[ 2.05794826e-01 -3.21032465e-01 1.29884690e-01 -1.69249758e-01 -6.96650088e-01 8.83072615e-02 2.92900056e-01 -6.03781521e-01 -2.01981962e-01 8.18587422e-01 3.43349755e-01 -5.35985231e-02 9.03209448e-02 -1.28422773e+00 -8.05219710e-01 -8.71406198e-01 1.48792788e-01 -2.18669310e-01 2.33830795e-01 -3.62167597e-01 6.49811849e-02 6.45177186e-01 -1.84900296e+00 5.70558012e-01 8.43593597e-01 1.02027583e+00 4.94261265e-01 7.25842535e-01 -6.28907084e-02 9.20962274e-01 -1.83522016e-01 -5.79336435e-02 6.76044464e-01 -3.25225204e-01 -3.93550307e-01 1.31874263e-01 4.39375103e-01 -1.02911377e+00 -7.79299676e-01 1.05864859e+00 7.11790979e-01 1.31618410e-01 3.39652538e-01 -6.11110091e-01 -1.11398065e+00 2.55309075e-01 -1.13009501e+00 3.88534442e-02 4.20493454e-01 4.48179424e-01 5.42840540e-01 -1.11320686e+00 8.05434167e-01 1.50146949e+00 5.44950724e-01 3.73470962e-01 -1.42949533e+00 -7.97493160e-01 -1.93242475e-01 1.32156432e-01 -1.44162989e+00 -3.57513666e-01 9.33395445e-01 1.53302535e-01 8.69055212e-01 2.64131397e-01 7.28641748e-01 8.23183835e-01 3.55148971e-01 5.06999493e-01 1.58559680e+00 -3.53035301e-01 -1.30974829e-01 -2.74868637e-01 -3.99698555e-01 7.12797582e-01 1.72934070e-01 4.83164072e-01 -6.34117901e-01 2.51657933e-01 1.68790269e+00 3.75412405e-02 -6.10259295e-01 -4.83505800e-02 -1.34240294e+00 3.49318475e-01 5.98762035e-01 8.83642882e-02 -3.30516189e-01 8.26144665e-02 8.39368775e-02 2.90293872e-01 5.54335713e-01 3.78106721e-02 -8.97907987e-02 1.59717739e-01 -6.18139207e-01 -6.69302279e-03 1.30775884e-01 9.91579711e-01 9.79695320e-01 1.36423022e-01 -3.22015852e-01 1.03884089e+00 2.09780023e-01 7.44040489e-01 9.28378627e-02 -1.12166464e+00 4.03713912e-01 2.32517719e-01 2.80287772e-01 -1.05924094e+00 -2.22461522e-01 -1.89239383e-01 -1.63468969e+00 5.18615425e-01 1.93322480e-01 2.13888049e-01 -9.29046810e-01 1.15289927e+00 4.72913176e-01 4.14173275e-01 8.95224959e-02 1.33478177e+00 9.29391921e-01 1.29076874e+00 -3.34784985e-01 -3.91664863e-01 1.11273777e+00 -6.98674738e-01 -8.62322211e-01 2.18356475e-01 -3.83671112e-02 -9.95327115e-01 1.29985619e+00 4.51075286e-01 -1.38108194e+00 -1.05043542e+00 -1.12259686e+00 -6.40086174e-01 7.05497488e-02 -8.27165041e-03 5.77258825e-01 3.53585392e-01 -1.00377297e+00 6.07514262e-01 -3.52111548e-01 2.06180289e-01 2.70006239e-01 6.40870333e-02 -2.40505680e-01 -7.54215240e-01 -1.31238008e+00 4.69792575e-01 3.01717639e-01 3.45590711e-01 -6.72874629e-01 -9.42649662e-01 -6.49488568e-01 -2.34861478e-01 1.32063985e-01 -8.05103958e-01 4.83146429e-01 -7.22358167e-01 -1.69064713e+00 8.40916634e-01 2.04744428e-01 -1.61164194e-01 6.27712667e-01 1.96362450e-03 -4.69767094e-01 2.58332849e-01 -2.22275794e-01 5.93975186e-01 1.01700199e+00 -1.55342484e+00 -7.06028640e-01 -1.92987546e-01 -2.78619051e-01 4.28277373e-01 1.03905220e-02 -2.23519176e-01 -7.40734816e-01 -8.23816121e-01 2.79047370e-01 -4.71863061e-01 -8.84355232e-02 3.68608177e-01 -2.23684549e-01 2.10552648e-01 9.34522986e-01 -8.35950553e-01 1.00554001e+00 -2.33402467e+00 -3.52898464e-02 1.29279330e-01 3.56008261e-01 1.22466750e-01 -2.12514609e-01 -2.62248099e-01 3.19073163e-02 -2.38609031e-01 -7.84569755e-02 -1.59065902e-01 -2.48784274e-01 5.48991859e-02 -4.08882588e-01 7.18974054e-01 4.70402613e-02 7.86425173e-01 -6.93728030e-01 -5.39519787e-01 8.06250989e-01 1.07123888e+00 -4.26169276e-01 5.43747485e-01 5.40458784e-02 5.18643081e-01 -1.97104990e-01 5.08347631e-01 1.34981585e+00 2.04457734e-02 -3.13280523e-02 -8.78632128e-01 -3.97416085e-01 -2.36343607e-01 -1.45553815e+00 1.71564853e+00 -6.27509952e-01 4.41722602e-01 -5.62247075e-03 -3.58314365e-01 1.31742358e+00 -1.91935077e-01 5.43256760e-01 -1.42415357e+00 3.51191871e-02 2.04181701e-01 -5.16768873e-01 -2.71649688e-01 6.53821766e-01 -1.76278830e-01 1.38039410e-01 3.67069542e-01 -2.51542717e-01 -2.10691392e-01 -2.01862976e-01 -1.18082669e-02 6.13766372e-01 2.47610033e-01 -2.50525344e-02 -1.33941427e-01 6.09726369e-01 -5.61025918e-01 6.76188827e-01 6.41780317e-01 1.90426037e-02 1.06098068e+00 1.43506780e-01 -6.24332070e-01 -1.74938154e+00 -1.38013577e+00 -2.57574677e-01 6.33950114e-01 6.23190701e-01 -9.02799424e-03 -3.90399188e-01 5.22793569e-02 -2.86437541e-01 3.40630054e-01 -5.14655292e-01 -9.98064280e-02 -7.72656441e-01 -7.43555307e-01 1.76747262e-01 2.07388550e-01 1.28837287e+00 -9.25772905e-01 -4.69001979e-01 1.90767661e-01 -3.12578052e-01 -1.11066294e+00 -3.58446658e-01 -3.74075584e-02 -6.00979865e-01 -6.69248879e-01 -1.07005310e+00 -7.39632845e-01 3.93290341e-01 6.54947460e-01 1.24975586e+00 6.20835535e-02 -5.88148773e-01 4.65808362e-02 -3.64001334e-01 2.84943789e-01 -2.85389721e-01 -5.27155817e-01 -1.53743982e-01 1.81217432e-01 -1.52985975e-01 -4.82117742e-01 -1.04629445e+00 3.80877882e-01 -1.05917084e+00 9.27878201e-01 7.43489563e-01 9.75533485e-01 1.22230256e+00 4.18265849e-01 2.51201630e-01 -6.85424089e-01 3.67856622e-01 -1.40381949e-02 -6.01644278e-01 1.44467294e-01 -4.29079711e-01 -4.79978397e-02 8.91219735e-01 -4.53788966e-01 -1.66973782e+00 -4.57051136e-02 -2.06903979e-01 -7.35695839e-01 -8.76754969e-02 -3.39972585e-01 -2.69015402e-01 -2.59678453e-01 4.15600300e-01 4.67823178e-01 -2.79029459e-02 -3.96026313e-01 6.04154170e-01 5.36622822e-01 6.92935944e-01 -5.92783511e-01 9.62936580e-01 7.84559190e-01 6.32798895e-02 -9.42483604e-01 -6.84108973e-01 -1.20200001e-01 -3.06064039e-01 -4.36285853e-01 9.43524897e-01 -1.28807175e+00 -7.19081402e-01 7.45438874e-01 -8.55507851e-01 -6.22330546e-01 -4.08308893e-01 3.69155228e-01 -7.97007680e-01 4.41413879e-01 -1.11117089e+00 -6.05148435e-01 -5.30066609e-01 -1.07385254e+00 1.27504134e+00 4.12446350e-01 5.91732740e-01 -4.23185647e-01 -2.20631212e-01 1.81052625e-01 6.16997302e-01 2.67474055e-01 9.70430553e-01 9.28444266e-01 -1.15351295e+00 4.27008837e-01 -1.01936793e+00 3.23438227e-01 -1.73773114e-02 1.14029562e-02 -8.34874809e-01 -3.03100914e-01 5.19251004e-02 -3.23515803e-01 9.04384971e-01 5.69609642e-01 1.55394471e+00 -9.27200094e-02 1.87977791e-01 1.28119385e+00 1.94224918e+00 -1.74984783e-01 1.27772081e+00 3.63293827e-01 1.01739919e+00 3.64045739e-01 1.00114810e+00 7.19720244e-01 2.45007604e-01 7.22730935e-01 2.78448313e-01 -7.89698124e-01 -8.29556644e-01 -3.52919370e-01 3.07414323e-01 6.56794906e-01 -3.29073012e-01 4.66008782e-02 -3.90298337e-01 1.12838775e-01 -1.54116428e+00 -9.70590293e-01 -3.29798371e-01 2.13493776e+00 9.06078160e-01 -1.50599435e-01 -1.37885034e-01 -7.44331256e-03 7.79469669e-01 3.96400362e-01 -6.86654329e-01 -8.40796530e-02 -7.56618857e-01 1.83964729e-01 6.99020922e-01 3.23700249e-01 -7.70719469e-01 8.11906755e-01 5.59483528e+00 1.15611041e+00 -1.05354702e+00 -8.04671645e-02 1.20553935e+00 -1.49416178e-01 -5.84633350e-01 -3.56861413e-01 -7.12136924e-01 4.08684611e-01 5.28513968e-01 1.31763428e-01 8.47007275e-01 3.06395173e-01 3.61886472e-01 -1.59998655e-01 -5.39561272e-01 1.60287452e+00 -1.53051037e-02 -1.50968969e+00 2.04085186e-01 9.01252031e-02 1.06300426e+00 -2.73631185e-01 4.33422387e-01 7.03050941e-03 2.46706754e-01 -9.80490923e-01 6.57325745e-01 7.54826069e-01 1.48330927e+00 -1.11216068e+00 4.39898491e-01 -2.24822871e-02 -1.35393727e+00 -2.30862759e-02 -9.62690651e-01 4.23038214e-01 1.54012620e-01 8.76815796e-01 -1.67435855e-01 5.41498721e-01 1.13196611e+00 9.56412792e-01 -3.92077237e-01 6.23317122e-01 1.22371670e-02 -7.25963339e-03 -1.95790023e-01 4.30056155e-01 -1.45578953e-02 -4.68521357e-01 1.93597272e-01 9.35878396e-01 4.65692282e-01 4.92415041e-01 -8.65622088e-02 9.85954523e-01 -1.92941979e-01 6.25279769e-02 -6.78380132e-01 4.31610674e-01 1.34556621e-01 1.23538208e+00 -5.95487177e-01 -2.13857546e-01 -5.34561038e-01 1.32899272e+00 2.53395796e-01 5.17583549e-01 -1.04206407e+00 -2.81985104e-01 5.62551200e-01 2.43013784e-01 1.60029590e-01 -2.89190322e-01 -2.53343403e-01 -1.28923798e+00 -1.77637205e-01 -9.16803777e-01 1.39072642e-01 -1.16185951e+00 -1.38614070e+00 4.54967856e-01 -4.01127189e-01 -1.45844245e+00 4.24463212e-01 -4.07074571e-01 -1.25588745e-01 1.05011189e+00 -1.96826911e+00 -1.20029831e+00 -6.99055851e-01 1.04746842e+00 5.73875785e-01 1.65494576e-01 4.06223357e-01 6.22990489e-01 -3.37292910e-01 2.20774680e-01 3.15052450e-01 -7.75474906e-02 9.66197431e-01 -8.79166305e-01 2.44849876e-01 8.53611767e-01 -3.26055646e-01 2.82614440e-01 4.86496210e-01 -6.28603041e-01 -1.90951562e+00 -1.26232910e+00 -9.08189639e-02 3.48739147e-01 5.95803522e-02 -3.46627802e-01 -1.00364494e+00 2.38334745e-01 -1.02201076e-02 4.36404794e-01 2.46556491e-01 -4.32699561e-01 -3.51677865e-01 -5.02562940e-01 -1.36201644e+00 6.65420353e-01 1.15689826e+00 -5.49655795e-01 2.55497843e-02 3.49234380e-02 9.70019460e-01 -5.68295240e-01 -1.12384307e+00 3.90207142e-01 6.84979618e-01 -1.55725706e+00 1.43762326e+00 2.88089216e-01 8.38891506e-01 -7.25198686e-01 -3.96539092e-01 -9.40472543e-01 -5.64291775e-01 -4.35455501e-01 -1.15268216e-01 1.18727684e+00 -1.57359034e-01 -3.45372617e-01 4.42845166e-01 3.47311050e-01 -8.76665637e-02 -7.77780950e-01 -7.29947984e-01 -3.22201908e-01 -2.09006026e-01 -1.54587775e-01 7.60544479e-01 8.89613390e-01 -7.52010703e-01 5.97024560e-02 -9.54790533e-01 1.40395224e-01 1.21843553e+00 4.73240644e-01 8.95117104e-01 -6.52958870e-01 -3.04730415e-01 -8.53356048e-02 -1.39662340e-01 -1.19786644e+00 -1.40874282e-01 -4.96193796e-01 8.35939869e-02 -1.50414121e+00 3.28010350e-01 -7.54935384e-01 -2.46742532e-01 -6.17590174e-02 -1.22415729e-01 7.89052069e-01 1.44503191e-01 2.06475571e-01 -4.41535890e-01 7.42015839e-01 1.92351222e+00 2.34498661e-02 -2.44337633e-01 -5.83539903e-01 -4.20292526e-01 4.14479733e-01 5.11045396e-01 1.45366505e-01 -2.42293999e-01 -5.67525387e-01 1.92256615e-01 3.06566060e-01 3.74711394e-01 -1.04589033e+00 -3.56661491e-02 -3.17886412e-01 1.12453651e+00 -7.01856792e-01 3.62781107e-01 -8.94135237e-01 5.25826871e-01 1.96590230e-01 -1.21626750e-01 -4.37960505e-01 -1.18066750e-01 6.16554618e-01 -1.39466509e-01 5.30417323e-01 1.49581909e+00 -1.43365666e-01 -9.77635264e-01 6.08962476e-01 4.07973789e-02 -3.04255754e-01 1.08275414e+00 -4.35635835e-01 -5.66854119e-01 -2.33759880e-01 -1.56581640e-01 -1.16247624e-01 8.13277364e-01 3.61367762e-01 1.21191311e+00 -1.51986182e+00 -8.27475250e-01 7.46170759e-01 -1.22237235e-01 2.19951734e-01 1.00554836e+00 5.44612408e-01 -9.30033863e-01 -1.37937352e-01 -7.73302436e-01 -5.65983593e-01 -1.03840661e+00 5.79150736e-01 3.54769856e-01 -1.80844754e-01 -1.34786296e+00 5.53715229e-01 3.61620337e-01 -1.00107603e-01 7.66185578e-03 -1.38336375e-01 9.48952734e-02 -3.83309275e-01 7.76718140e-01 3.45540941e-01 -2.81946182e-01 -6.69581354e-01 2.02184364e-01 8.98045719e-01 2.24202294e-02 8.17337930e-02 1.61151481e+00 -6.73926950e-01 -2.03398585e-01 3.13777059e-01 1.33815765e+00 -8.17278549e-02 -1.61229873e+00 -1.61979169e-01 -9.75001454e-01 -1.20698106e+00 4.49004441e-01 -3.90160531e-01 -1.37089884e+00 7.49320984e-01 1.08378375e+00 -5.39293513e-02 1.65563667e+00 -3.29181373e-01 1.20788836e+00 -8.91626254e-02 5.96101046e-01 -1.15363419e+00 1.31386951e-01 2.15464905e-01 8.17367315e-01 -1.12522936e+00 2.23966479e-01 -5.58310866e-01 -4.80824679e-01 1.27087665e+00 7.83995748e-01 -3.65438968e-01 3.34703028e-01 4.22940791e-01 -1.31958753e-01 1.73124284e-01 -4.89441484e-01 -1.46466404e-01 8.81768689e-02 7.73030281e-01 2.97785997e-01 -9.77302417e-02 -1.27211049e-01 9.42729414e-03 1.96635723e-01 -3.43861766e-02 6.66434228e-01 3.84828985e-01 -4.50672656e-01 -6.06992781e-01 -5.14448464e-01 4.43363667e-01 -1.80313393e-01 -1.12253860e-01 3.67012233e-01 4.47485030e-01 1.56579837e-01 5.77445269e-01 2.17619568e-01 -4.17412609e-01 3.39694887e-01 -6.70574605e-01 7.54200935e-01 -2.20497902e-02 -4.91120331e-02 3.59777510e-01 -1.75121546e-01 -1.00456107e+00 -3.99512023e-01 -1.24319427e-01 -1.13961935e+00 -6.50863051e-01 1.96784884e-02 -5.10147929e-01 4.70625609e-01 2.30096012e-01 2.79520839e-01 6.10265315e-01 9.42475855e-01 -1.09083807e+00 4.23998572e-02 -5.27355433e-01 -1.15029001e+00 6.05726600e-01 3.79475743e-01 -4.75644469e-01 -3.99618626e-01 -9.02865897e-04]
[10.815096855163574, -2.1235055923461914]
1dcc415b-c4f9-4127-a12b-b906d4b41ec8
a-generative-model-for-relation-extraction
2202.13229
null
https://arxiv.org/abs/2202.13229v1
https://arxiv.org/pdf/2202.13229v1.pdf
A Generative Model for Relation Extraction and Classification
Relation extraction (RE) is an important information extraction task which provides essential information to many NLP applications such as knowledge base population and question answering. In this paper, we present a novel generative model for relation extraction and classification (which we call GREC), where RE is modeled as a sequence-to-sequence generation task. We explore various encoding representations for the source and target sequences, and design effective schemes that enable GREC to achieve state-of-the-art performance on three benchmark RE datasets. In addition, we introduce negative sampling and decoding scaling techniques which provide a flexible tool to tune the precision and recall performance of the model. Our approach can be extended to extract all relation triples from a sentence in one pass. Although the one-pass approach incurs certain performance loss, it is much more computationally efficient.
['Radu Florian', 'Alfio Gliozzo', 'Gaetano Rossiello', 'Jian Ni']
2022-02-26
null
null
null
null
['knowledge-base-population']
['natural-language-processing']
[ 6.16942346e-01 2.89676368e-01 -4.03096408e-01 -2.89042115e-01 -1.07436895e+00 -6.94012284e-01 5.94152331e-01 4.42412317e-01 -3.22535574e-01 9.89739776e-01 3.55014689e-02 -7.24839330e-01 4.48678993e-02 -1.20953798e+00 -7.68065155e-01 -4.52655345e-01 -1.03865072e-01 6.23592377e-01 4.76317257e-01 -3.27443421e-01 1.91982780e-02 3.71749401e-01 -1.36831725e+00 5.34153402e-01 1.03982520e+00 7.47113705e-01 1.42480228e-02 7.87799418e-01 -4.81425762e-01 7.80482650e-01 -6.97448730e-01 -9.39562976e-01 -8.07998255e-02 -7.03063726e-01 -1.24079013e+00 -2.51305938e-01 -3.73033524e-01 3.14398967e-02 -1.34207025e-01 1.03357577e+00 4.14139539e-01 -6.40899986e-02 7.71050990e-01 -1.11584365e+00 -4.95168179e-01 9.21635807e-01 -5.84694207e-01 2.73332596e-01 6.97953463e-01 -3.31436694e-01 1.12677336e+00 -6.95013165e-01 7.28695273e-01 1.15098155e+00 4.28589314e-01 5.39501727e-01 -1.10087454e+00 -5.25192022e-01 -6.98560700e-02 4.33327764e-01 -1.38651204e+00 -5.16838193e-01 4.51434374e-01 -2.30585515e-01 1.36540222e+00 5.51258981e-01 4.89631414e-01 9.47264910e-01 1.27250746e-01 9.10236299e-01 7.55530357e-01 -5.95571280e-01 2.84736753e-01 -5.80057055e-02 3.14932734e-01 7.12123036e-01 4.65449631e-01 -2.95658320e-01 -5.36753118e-01 -4.53823894e-01 2.85911560e-01 -4.15481329e-01 -3.99650127e-01 -3.15277204e-02 -8.57590914e-01 9.42195058e-01 1.16460443e-01 4.88115363e-02 -3.65336359e-01 6.00788705e-02 4.26940173e-01 2.99636453e-01 3.75768572e-01 3.50041598e-01 -5.33421397e-01 -4.01529223e-01 -3.86174768e-01 3.78282219e-01 1.21562374e+00 1.29761899e+00 5.59955180e-01 -4.38236564e-01 -4.93933678e-01 8.09787512e-01 1.08462945e-01 2.63217866e-01 2.70426691e-01 -3.56349975e-01 8.04623723e-01 6.79680109e-01 2.11004410e-02 -8.24997783e-01 -3.19915235e-01 -3.09901297e-01 -8.42718601e-01 -6.41925991e-01 1.12409569e-01 -2.57581174e-01 -8.97174597e-01 1.46961844e+00 5.59046090e-01 6.61025867e-02 3.03699374e-01 4.95059341e-01 8.97193789e-01 7.96968281e-01 8.53449777e-02 -5.24996936e-01 1.74398184e+00 -8.05306792e-01 -8.85605872e-01 -3.77213031e-01 8.91074002e-01 -5.90001166e-01 5.86753964e-01 1.14590675e-01 -9.47990179e-01 -1.46209719e-02 -1.10489905e+00 -3.21685642e-01 -3.80120784e-01 2.45181900e-02 8.99794161e-01 7.43862152e-01 -5.26220739e-01 3.95982951e-01 -9.45873976e-01 -3.34857643e-01 3.64391476e-01 3.00106972e-01 -4.03227240e-01 -1.74502686e-01 -1.26688552e+00 9.12266910e-01 8.15316379e-01 1.17813155e-01 -3.92786026e-01 -4.52573538e-01 -1.02278662e+00 2.06629589e-01 7.74654388e-01 -7.17125893e-01 1.24932873e+00 -6.34125480e-03 -1.41584456e+00 6.01236522e-01 -5.76017380e-01 -7.86647975e-01 7.97820911e-02 -3.65434915e-01 -4.12667960e-01 -1.71637014e-02 -6.68194890e-02 2.74835557e-01 3.39734286e-01 -9.75239575e-01 -5.67736149e-01 -1.53375253e-01 -1.23994932e-01 -2.92180851e-03 9.85514224e-02 4.33833331e-01 -6.93468034e-01 -5.70744872e-01 9.26270932e-02 -8.26615572e-01 -1.73686504e-01 -6.38087153e-01 -7.93308914e-01 -4.73746240e-01 4.00880516e-01 -7.24942267e-01 1.63489616e+00 -1.71508753e+00 9.29812565e-02 3.99161875e-01 -7.60343112e-03 6.60824895e-01 -7.19193071e-02 8.98383617e-01 -2.72538196e-02 3.35648507e-01 -4.44974989e-01 3.84419831e-03 -6.65313601e-02 2.55907834e-01 -2.52807766e-01 -1.42241805e-03 7.12996781e-01 1.33494699e+00 -9.30993021e-01 -4.91933078e-01 -3.61887902e-01 3.47689420e-01 -4.28238750e-01 2.22017348e-01 -4.58431214e-01 -4.51799594e-02 -5.14186561e-01 5.61362565e-01 5.24651229e-01 -4.95175213e-01 6.40279651e-01 -5.50309084e-02 4.85225797e-01 8.14666212e-01 -9.73798096e-01 1.28013074e+00 -2.06051812e-01 2.81328410e-01 -5.34228861e-01 -9.94225860e-01 1.07859290e+00 3.16343755e-01 2.23450319e-04 -3.78406078e-01 1.26631811e-01 1.34771779e-01 1.61878705e-01 -6.10842705e-01 6.09504580e-01 -2.04329006e-02 -1.00957267e-01 3.71741712e-01 8.66973698e-02 1.35519817e-01 5.53601682e-01 3.08928519e-01 1.33325541e+00 8.32549706e-02 1.00617445e+00 4.42115031e-02 5.52138448e-01 -6.83615357e-02 8.27883303e-01 6.22967184e-01 3.06372523e-01 3.95844519e-01 9.07415986e-01 -1.41515166e-01 -7.65606701e-01 -6.74639225e-01 1.14599124e-01 6.49331450e-01 -1.21267594e-01 -1.00300717e+00 -6.48491323e-01 -9.92719650e-01 -1.34374663e-01 7.80295372e-01 -4.79000807e-01 -3.33392203e-01 -5.72633505e-01 -1.18051875e+00 6.71885669e-01 6.30771399e-01 3.96707088e-01 -9.39643204e-01 -2.60226756e-01 4.67726022e-01 -5.91261327e-01 -1.29277492e+00 -4.20099676e-01 3.95305097e-01 -5.71682632e-01 -1.17096162e+00 -3.29224974e-01 -7.87863553e-01 3.88378769e-01 -4.33345065e-02 1.22381878e+00 -3.45891379e-02 -1.76226228e-01 -3.04586351e-01 -7.37914145e-01 -4.64592397e-01 -5.48265159e-01 5.53601682e-01 -3.46409500e-01 -1.40064403e-01 7.53575504e-01 -3.61997873e-01 -2.05121204e-01 -1.48135545e-02 -8.34741533e-01 1.48188293e-01 4.83368278e-01 7.75421023e-01 7.28324115e-01 1.35514895e-02 8.93962383e-01 -1.50678563e+00 8.76617968e-01 -5.52144468e-01 -4.87986684e-01 6.57835066e-01 -6.74914360e-01 4.53070968e-01 2.87204504e-01 -2.75187612e-01 -9.85701680e-01 1.84014961e-01 -5.06907642e-01 2.08603323e-01 1.88691765e-01 8.11735213e-01 -5.02448976e-01 2.23994359e-01 5.47901928e-01 3.23866695e-01 -2.15864599e-01 -6.08432174e-01 3.63669842e-01 8.14739227e-01 4.99342412e-01 -6.71264291e-01 6.39048755e-01 -9.21904743e-02 5.62095195e-02 -6.30703390e-01 -1.08383358e+00 -4.42131549e-01 -3.13876033e-01 3.60744625e-01 6.36405647e-01 -6.40460908e-01 -9.38286722e-01 2.08385766e-01 -1.31405711e+00 -9.13883671e-02 -7.22207278e-02 2.07218602e-01 -3.36228460e-01 2.45529294e-01 -8.54879022e-01 -1.01854587e+00 -6.55242503e-01 -9.65820670e-01 1.03960502e+00 1.99652851e-01 -3.08343560e-01 -6.66696191e-01 1.56522449e-02 2.09127098e-01 -3.18647586e-02 1.64394677e-01 1.20593166e+00 -1.06711495e+00 -6.71416044e-01 -1.72038287e-01 -2.89541095e-01 1.25371724e-01 1.36171848e-01 -1.38781056e-01 -7.08982229e-01 -1.91752929e-02 -5.00757277e-01 -2.36835301e-01 9.08109665e-01 -2.58673877e-01 1.27635252e+00 -5.62286675e-01 -6.56022668e-01 4.95819539e-01 1.26951361e+00 4.69699830e-01 1.07655883e+00 1.86213672e-01 6.11520648e-01 4.87764359e-01 7.56920457e-01 1.36326075e-01 5.49261630e-01 6.42559826e-01 -7.01648965e-02 2.37150475e-01 -4.01541451e-03 -6.00293756e-01 1.49944931e-01 7.90535390e-01 -1.55102164e-01 -6.49980247e-01 -9.02016163e-01 4.91756976e-01 -1.90960109e+00 -9.62218165e-01 -2.76624769e-01 2.04031277e+00 1.39126921e+00 4.44062576e-02 7.38098323e-02 4.41191345e-01 5.56266189e-01 -1.81285515e-01 -1.93357319e-01 -4.36947435e-01 -7.62504339e-02 7.78102279e-01 4.37305987e-01 5.70962846e-01 -1.01255584e+00 1.11849082e+00 6.62719297e+00 1.03878319e+00 -6.06411934e-01 -1.44345909e-01 5.09791791e-01 2.71436870e-01 -4.73112732e-01 1.44264042e-01 -1.25972867e+00 3.35138410e-01 1.10379887e+00 -2.05865622e-01 2.42655545e-01 6.05962038e-01 -3.51604134e-01 -1.93977624e-01 -1.20564842e+00 8.91995907e-01 4.57327329e-02 -1.51996410e+00 1.82442084e-01 1.06548868e-01 4.44839418e-01 -4.41871315e-01 -5.88914871e-01 4.06283200e-01 4.76100266e-01 -1.02567005e+00 1.38828620e-01 3.45643640e-01 9.64605987e-01 -1.06361270e+00 1.07375693e+00 4.10400152e-01 -1.15870523e+00 2.08841294e-01 -2.06542701e-01 1.22488260e-01 3.50768507e-01 8.47324550e-01 -1.05663848e+00 1.01901102e+00 4.45399940e-01 3.41618925e-01 -5.45377374e-01 8.82566154e-01 -4.99864489e-01 6.98535144e-01 -3.06368381e-01 -4.66664493e-01 -1.85239568e-01 -1.41368136e-01 2.25042835e-01 1.49964249e+00 2.59821802e-01 3.88507992e-01 2.54920591e-02 4.71640110e-01 -3.16260368e-01 1.07514568e-01 -5.67889988e-01 -1.95905745e-01 8.98724198e-01 1.03639865e+00 -7.36302376e-01 -4.29066509e-01 -3.02214354e-01 1.01710880e+00 7.21418798e-01 3.84180754e-01 -6.95203960e-01 -8.34265053e-01 4.14562285e-01 -3.14688414e-01 5.85313797e-01 3.90181169e-02 -9.11652148e-02 -1.26444304e+00 2.67623216e-01 -9.99783993e-01 3.98130327e-01 -3.34460676e-01 -1.07028961e+00 6.82970643e-01 1.55628636e-01 -7.49061942e-01 -7.30346024e-01 -3.34662646e-01 -2.25028336e-01 8.74865234e-01 -1.40604901e+00 -7.68049598e-01 7.61084184e-02 2.25506768e-01 1.03593655e-01 6.53190166e-03 1.03246427e+00 2.37943381e-01 -8.14107180e-01 9.52470541e-01 -1.82432294e-01 4.91508722e-01 2.84476131e-01 -1.09970820e+00 9.08177257e-01 9.92660224e-01 3.55878174e-01 8.85598183e-01 5.61335564e-01 -7.32532203e-01 -1.39163423e+00 -1.19711971e+00 1.47257388e+00 -2.27581367e-01 5.12542188e-01 -6.10752583e-01 -9.83640194e-01 8.82681727e-01 -3.63357179e-02 -1.70522764e-01 9.25514519e-01 4.20894742e-01 -4.27117854e-01 2.47861091e-02 -1.02732432e+00 5.42613029e-01 1.17339361e+00 -5.92863202e-01 -6.01595223e-01 1.30416363e-01 8.86810184e-01 -6.29322231e-01 -9.59291220e-01 5.32672822e-01 3.83161753e-01 -5.43670654e-01 9.40018415e-01 -7.87525833e-01 4.04757887e-01 -3.19888264e-01 -4.43423018e-02 -1.15594447e+00 -2.90170640e-01 -8.64393771e-01 -7.86910892e-01 1.60813856e+00 8.55912447e-01 -7.99101770e-01 6.45191848e-01 3.60645682e-01 2.20463112e-01 -1.14695644e+00 -6.47231698e-01 -7.44744956e-01 -1.49379566e-01 -4.69506860e-01 1.01477039e+00 5.74200869e-01 2.23869458e-01 1.04637825e+00 -2.52412528e-01 1.30344227e-01 3.19068819e-01 2.86468297e-01 5.99322677e-01 -1.02086449e+00 -4.53628808e-01 2.30964795e-02 -2.51269042e-01 -1.23413241e+00 -2.64263875e-03 -9.96742845e-01 1.50922284e-01 -1.59562552e+00 3.72668445e-01 -4.17389542e-01 1.16368525e-01 6.35015428e-01 -5.72002351e-01 5.82215190e-02 -2.67048795e-02 -1.80987492e-01 -3.77476066e-01 5.46521604e-01 9.28743839e-01 5.86865377e-03 -1.59551695e-01 8.92150477e-02 -1.02380657e+00 3.05128872e-01 8.71340632e-01 -5.80648363e-01 -5.97163498e-01 -1.94170997e-01 5.07680357e-01 1.97806284e-01 5.69013730e-02 -4.72646385e-01 1.81265756e-01 -1.59180298e-01 1.30474567e-01 -6.75901830e-01 2.80658275e-01 -3.04393977e-01 1.23513229e-01 2.29143843e-01 -4.18335587e-01 -5.60601167e-02 4.44265977e-02 5.63458920e-01 -2.10141227e-01 -4.46687400e-01 4.13961530e-01 9.92932022e-02 -3.36744606e-01 2.29562640e-01 -9.31270570e-02 2.69994944e-01 8.40272069e-01 3.07735503e-01 -6.25436664e-01 -3.11021715e-01 -3.69870871e-01 2.64210105e-01 6.43465221e-02 3.57871979e-01 5.90224624e-01 -1.24768639e+00 -8.24990630e-01 1.66886494e-01 3.55487436e-01 2.46304169e-01 -2.22612575e-01 4.71996754e-01 -4.72419590e-01 5.92479527e-01 5.25101960e-01 -1.64892241e-01 -1.34396315e+00 7.14856684e-01 1.22877005e-02 -6.78043306e-01 -6.09162450e-01 9.80099142e-01 -8.68315995e-02 -3.54959697e-01 8.32578689e-02 -2.58460134e-01 -2.26626962e-01 -1.49688110e-01 8.43408823e-01 9.97305512e-02 3.81225705e-01 -2.14226723e-01 -5.39260209e-01 1.78356573e-01 -3.89881670e-01 -1.44116348e-02 1.10353506e+00 1.91859469e-01 -2.13860065e-01 1.79374486e-01 1.14893496e+00 -1.65067296e-02 -6.54042244e-01 -2.94336945e-01 5.14989316e-01 -1.72804892e-01 -2.93109149e-01 -7.85739124e-01 -7.73480654e-01 5.26980400e-01 -7.01260716e-02 1.72382280e-01 1.09787834e+00 2.35566095e-01 9.83121932e-01 5.64761937e-01 4.70219821e-01 -7.27844536e-01 -5.38967848e-01 6.08084023e-01 6.52461827e-01 -1.05532348e+00 1.16832092e-01 -1.07709110e+00 -6.10376179e-01 6.89523697e-01 3.37716132e-01 2.56384134e-01 4.48823988e-01 6.38037145e-01 -3.84920895e-01 -1.47869112e-02 -1.19126940e+00 -5.64819634e-01 3.35586458e-01 7.31336474e-01 7.11145163e-01 1.25296310e-01 -7.37705410e-01 7.87805080e-01 -4.72591102e-01 9.24127921e-02 2.03155473e-01 1.03942156e+00 -2.73950100e-01 -1.65803897e+00 6.84308261e-02 7.17154980e-01 -6.56930983e-01 -3.59651446e-01 -4.57582980e-01 6.30491972e-01 -1.96946874e-01 1.17398214e+00 -9.18980688e-02 -5.79412520e-01 2.43357956e-01 3.74484658e-01 5.73876381e-01 -7.88626969e-01 -4.78568077e-01 -9.62174237e-02 8.76254261e-01 -3.03093106e-01 -3.58839124e-01 -5.70718825e-01 -1.28566515e+00 -3.03232789e-01 -5.64430833e-01 5.46544492e-01 2.49333367e-01 1.01646674e+00 5.50418735e-01 5.64067841e-01 4.14554894e-01 3.33946310e-02 -4.14421678e-01 -1.16004109e+00 -2.95578808e-01 1.36252061e-01 1.36248559e-01 -5.42140007e-01 2.37492412e-01 8.66704807e-02]
[9.391840934753418, 8.696677207946777]
40243fd5-6d6a-4756-b37e-9613e1f0d5a2
ambicoref-evaluating-human-and-model
2302.00762
null
https://arxiv.org/abs/2302.00762v2
https://arxiv.org/pdf/2302.00762v2.pdf
AmbiCoref: Evaluating Human and Model Sensitivity to Ambiguous Coreference
Given a sentence "Abby told Brittney that she upset Courtney", one would struggle to understand who "she" refers to, and ask for clarification. However, if the word "upset" were replaced with "hugged", "she" unambiguously refers to Abby. We study if modern coreference resolution models are sensitive to such pronominal ambiguity. To this end, we construct AmbiCoref, a diagnostic corpus of minimal sentence pairs with ambiguous and unambiguous referents. Our examples generalize psycholinguistic studies of human perception of ambiguity around particular arrangements of verbs and their arguments. Analysis shows that (1) humans are less sure of referents in ambiguous AmbiCoref examples than unambiguous ones, and (2) most coreference models show little difference in output between ambiguous and unambiguous pairs. We release AmbiCoref as a diagnostic corpus for testing whether models treat ambiguity similarly to humans.
['Mark Yatskar', 'Chaitanya Malaviya', 'Yuewei Yuan']
2023-02-01
null
null
null
null
['coreference-resolution']
['natural-language-processing']
[ 2.65053242e-01 3.68180007e-01 1.58876285e-01 -7.37752318e-01 -7.25282550e-01 -1.17109692e+00 7.14886189e-01 4.93039042e-01 -5.32643795e-01 8.06730449e-01 5.57039499e-01 -7.35870838e-01 -1.55436590e-01 -5.60347021e-01 -3.85202259e-01 -1.86038300e-01 5.23361683e-01 9.38503325e-01 1.84119284e-01 -8.08498621e-01 4.17325646e-01 3.54900062e-01 -1.30717409e+00 5.02645910e-01 7.52112150e-01 1.17958270e-01 5.11186898e-01 4.81476068e-01 -1.60634771e-01 4.98673469e-01 -8.09314609e-01 -7.32306182e-01 -8.52779076e-02 -4.42714781e-01 -1.52644813e+00 -2.93870300e-01 5.22305369e-01 4.40947898e-02 1.55736551e-01 1.28895843e+00 2.42604300e-01 -1.11797698e-01 5.02091110e-01 -1.01970983e+00 -4.98110443e-01 1.30033314e+00 -2.59167790e-01 5.29351652e-01 9.04197395e-01 1.70377288e-02 1.48695529e+00 -6.97664082e-01 8.03585589e-01 1.86237061e+00 4.63298202e-01 6.96180284e-01 -1.50769782e+00 -3.43588084e-01 2.62424529e-01 1.28350109e-01 -1.06494641e+00 -4.86758262e-01 5.22350788e-01 -2.58137703e-01 1.07075846e+00 8.60546887e-01 4.70301509e-01 1.30650330e+00 3.46860021e-01 2.84259826e-01 1.29234600e+00 -7.60333061e-01 -2.15590581e-01 7.24843564e-03 7.86299586e-01 -1.41433086e-02 6.13184810e-01 2.37341121e-01 -5.28612733e-01 -3.14662129e-01 2.62534648e-01 -6.67460501e-01 -3.72544408e-01 1.11459211e-01 -1.28634894e+00 7.41817236e-01 2.51165450e-01 7.81024277e-01 -1.97787851e-01 -1.58536151e-01 2.75876284e-01 5.63705027e-01 -2.30004787e-01 1.04365647e+00 -4.11554337e-01 -8.16285759e-02 1.81220304e-02 6.60142839e-01 1.01253355e+00 8.10157835e-01 3.34963739e-01 -4.10080492e-01 2.97169209e-01 8.32418025e-01 4.06737983e-01 4.93860364e-01 4.37955648e-01 -1.36363614e+00 5.89644969e-01 4.25637633e-01 6.85312390e-01 -9.40136731e-01 -7.56352723e-01 -5.85238822e-02 -4.77736071e-02 8.68052095e-02 6.37380362e-01 1.13934472e-01 -2.37380221e-01 2.21249056e+00 4.98133563e-02 -8.44583929e-01 6.19304419e-01 1.10191810e+00 6.77429914e-01 4.29574639e-01 4.83410120e-01 -4.35211718e-01 2.01254773e+00 -6.55421317e-02 -8.35985065e-01 -8.08858752e-01 7.93866873e-01 -1.11647594e+00 1.65855932e+00 1.33022383e-01 -1.29044318e+00 -2.52593905e-01 -1.11205816e+00 -4.07196492e-01 8.80162045e-03 -4.68682647e-01 8.08450639e-01 5.26929498e-01 -6.97632432e-01 5.24050176e-01 -5.83671391e-01 -4.76487607e-01 -4.83775407e-01 1.76818848e-01 -4.13164318e-01 1.38069555e-01 -1.61871898e+00 1.49272847e+00 5.97364724e-01 1.97659984e-01 -3.85292500e-01 -1.55136481e-01 -1.14822745e+00 5.69539964e-02 3.48185748e-01 -6.61901116e-01 1.84995890e+00 -1.26738417e+00 -7.73021042e-01 1.53149581e+00 -4.26078320e-01 -3.67727846e-01 2.19394013e-01 -4.70658094e-01 -6.85220838e-01 -4.09864746e-02 7.06968069e-01 4.83892500e-01 3.16297591e-01 -1.23503077e+00 -6.24318600e-01 -7.80849040e-01 5.42949855e-01 5.58541596e-01 4.23639148e-01 7.92746484e-01 5.16299486e-01 -4.55660194e-01 8.08235526e-01 -7.31076956e-01 2.91184545e-01 -4.37758833e-01 -7.18882680e-01 -6.01092339e-01 4.91902083e-01 -1.49540067e-01 1.22438717e+00 -2.04760933e+00 -1.96374029e-01 6.53712749e-02 2.18756244e-01 -8.24887156e-02 -5.73280863e-02 6.16350710e-01 -6.77898228e-01 4.47383851e-01 -2.37759843e-01 3.22740883e-01 4.53999698e-01 6.20641708e-01 -4.13991183e-01 2.65319526e-01 1.05796285e-01 7.90243208e-01 -7.40638137e-01 -3.88552994e-01 -1.64191678e-01 -4.35718186e-02 -4.25119489e-01 -8.89193267e-04 -5.30383363e-02 -4.24580500e-02 -1.27400503e-01 5.42603016e-01 5.62079668e-01 -1.36885196e-01 7.40608811e-01 -3.17835897e-01 -2.82578826e-01 1.05673718e+00 -1.03548539e+00 1.01517200e+00 -4.24001217e-02 5.90809226e-01 3.38091969e-01 -4.99950379e-01 6.04437232e-01 3.97319824e-01 -8.18907857e-01 -6.31134391e-01 2.31441662e-01 5.42986512e-01 8.90303671e-01 -4.96378273e-01 4.97628033e-01 -7.05590367e-01 -6.49224997e-01 3.25613201e-01 -6.03351116e-01 -6.04419827e-01 3.07127863e-01 5.62908113e-01 7.30801165e-01 -1.03162870e-01 7.32571065e-01 -7.83880413e-01 6.59537256e-01 1.42649218e-01 9.05595422e-01 6.36279166e-01 -1.97490200e-01 1.62664011e-01 8.00239444e-01 -5.63883960e-01 -8.11374664e-01 -1.63547814e+00 -4.41720903e-01 1.05821681e+00 4.89257455e-01 -4.95880038e-01 -9.08151448e-01 -3.10075939e-01 -2.60375768e-01 1.57056999e+00 -4.04855400e-01 1.74035519e-01 -1.11312640e+00 -5.52168906e-01 5.80776751e-01 4.51803446e-01 -1.68984465e-03 -1.25767040e+00 -1.14671338e+00 2.29532331e-01 -6.83961451e-01 -8.68805408e-01 -4.83443558e-01 2.30994359e-01 -4.42074120e-01 -1.31576979e+00 4.70086724e-01 -6.79928899e-01 3.92265409e-01 6.30489364e-02 1.34089470e+00 4.94125128e-01 6.64186925e-02 1.89295799e-01 -2.25666091e-01 -5.68437874e-01 -7.72040069e-01 -4.82340485e-01 1.37917787e-01 -7.90024102e-01 7.03125238e-01 -4.43606943e-01 -1.03945313e-02 2.56018430e-01 -6.72697842e-01 -1.25842318e-01 1.46308690e-01 7.83713043e-01 1.58350259e-01 -7.36055613e-01 3.77310723e-01 -1.26684439e+00 1.02927423e+00 -1.94981769e-01 -5.80396533e-01 3.32034469e-01 -4.10252154e-01 2.03955829e-01 3.05745810e-01 -1.20495252e-01 -1.23321486e+00 -5.10831892e-01 -9.07933041e-02 4.38072354e-01 -2.10231021e-01 3.66141021e-01 -3.90717328e-01 5.78602910e-01 1.01881707e+00 -3.66974533e-01 -5.81988227e-03 -2.80472279e-01 8.84084255e-02 5.22006214e-01 1.11357391e+00 -1.25831187e+00 5.30723214e-01 -1.58462506e-02 -1.76251665e-01 -5.79747736e-01 -9.62722600e-01 -1.12756401e-01 -4.46151525e-01 3.13316584e-01 6.96814239e-01 -5.80710769e-01 -1.12053442e+00 5.65782897e-02 -1.78580284e+00 3.28064933e-02 -1.08595841e-01 4.81850028e-01 -8.13905597e-01 7.53997982e-01 -7.02647507e-01 -6.93197489e-01 -1.96196526e-01 -9.85711873e-01 6.95510268e-01 9.06702727e-02 -1.46511841e+00 -7.92329192e-01 -1.28119409e-01 2.41862983e-01 1.66651383e-02 -3.23200673e-02 1.52855682e+00 -1.32458508e+00 1.33982122e-01 1.25116169e-01 -9.13996547e-02 -7.12984577e-02 1.85578778e-01 -7.50141814e-02 -5.88091612e-01 8.90835002e-02 3.75816703e-01 -3.25838029e-01 4.72777672e-02 -4.55233268e-02 3.59583765e-01 -4.24601674e-01 -3.55379164e-01 1.07211508e-01 1.18380857e+00 3.82094979e-01 6.26926720e-01 3.05687129e-01 1.11058585e-01 9.63534594e-01 7.47463942e-01 -9.65333208e-02 5.28607726e-01 7.22260833e-01 2.17720881e-01 6.09589815e-01 1.21649511e-01 -1.85574949e-01 2.90501922e-01 3.59943986e-01 3.74524295e-01 -4.57048059e-01 -1.05669034e+00 6.31840050e-01 -1.67085767e+00 -1.11744714e+00 -6.73528135e-01 1.86716151e+00 1.14842319e+00 4.29442137e-01 -4.47941512e-01 2.42395490e-01 1.05816174e+00 1.07337952e-01 -6.09782524e-02 -8.48937988e-01 -4.19649780e-01 2.43230030e-01 -2.89872497e-01 1.21463954e+00 -7.51946568e-01 1.27607048e+00 6.54147768e+00 1.58866927e-01 -5.40447891e-01 -7.47721940e-02 1.33671850e-01 3.05266142e-01 -7.01208591e-01 4.80020940e-01 -7.17604220e-01 3.12282175e-01 8.98104906e-01 -2.89377600e-01 3.77599411e-02 3.90813857e-01 2.10799783e-01 -3.74817312e-01 -1.70085740e+00 6.15343034e-01 -1.18767485e-01 -9.51099098e-01 1.22161172e-01 -6.05040371e-01 -1.01274021e-01 -6.16829097e-01 -3.37732881e-01 -2.08792631e-02 3.68406802e-01 -9.89699900e-01 9.77696896e-01 8.20832476e-02 2.36410901e-01 -4.52392101e-01 8.85607243e-01 4.38423842e-01 -6.61192358e-01 5.46030514e-02 -4.03505445e-01 -6.58793807e-01 4.83135939e-01 -2.55072325e-01 -6.35110319e-01 3.51762883e-02 5.54492235e-01 -2.21431285e-01 -2.81892061e-01 3.95087183e-01 -8.93555343e-01 2.12975547e-01 -3.89830410e-01 -1.50533304e-01 1.42544135e-01 -1.96829975e-01 9.60201502e-01 1.03645313e+00 8.83326307e-02 8.39089632e-01 -7.98469782e-02 9.53974426e-01 2.85188317e-01 7.27181956e-02 -5.99067271e-01 1.29728943e-01 1.18562877e+00 9.04764295e-01 -5.81953168e-01 -3.09489101e-01 -5.18459268e-02 7.54123271e-01 2.87520170e-01 -5.22452220e-02 -4.25473005e-01 -3.72055829e-01 6.99507117e-01 1.31180644e-01 -3.97309482e-01 1.84440330e-01 -2.56151736e-01 -7.87172198e-01 7.07778558e-02 -9.28284824e-01 5.42504013e-01 -1.34555137e+00 -1.28148985e+00 5.39976418e-01 4.14951086e-01 -6.29779577e-01 -6.06661856e-01 -8.16437066e-01 -7.12804615e-01 1.23403740e+00 -7.77244270e-01 -8.32142770e-01 1.38306290e-01 3.11789423e-01 5.42077601e-01 4.00514245e-01 1.19572508e+00 -4.72896487e-01 -5.97296134e-02 4.15855110e-01 -7.99924612e-01 2.16662109e-01 7.70583391e-01 -1.27562547e+00 3.70968461e-01 9.42517042e-01 1.32674187e-01 1.23580778e+00 1.43105316e+00 -6.79868102e-01 -7.91789353e-01 -9.04708803e-02 1.87854624e+00 -7.99371898e-01 7.74519622e-01 -5.27409427e-02 -1.14803708e+00 1.10704625e+00 6.36596143e-01 -2.82953113e-01 6.19208276e-01 3.53492856e-01 -7.07485080e-01 2.63564676e-01 -1.28989601e+00 8.42657924e-01 1.29174924e+00 -5.75655937e-01 -1.98169482e+00 3.46920252e-01 9.96513188e-01 -5.07464051e-01 -5.62993646e-01 3.78786922e-01 3.23863506e-01 -1.17711139e+00 6.74337268e-01 -1.14089537e+00 1.49675027e-01 -3.09524626e-01 -5.27432859e-01 -1.26109660e+00 -2.71416038e-01 -5.31820059e-01 6.67501092e-01 9.74549949e-01 6.93876326e-01 -8.24455082e-01 1.15226535e-02 9.24571455e-01 -5.24036348e-01 -6.55344082e-03 -1.19328892e+00 -4.69548672e-01 3.07060480e-01 -4.77696091e-01 4.86776084e-01 9.97246325e-01 8.23182464e-01 9.09299612e-01 3.53357106e-01 8.88385475e-02 2.96813339e-01 3.46519679e-01 8.80702212e-02 -1.55565882e+00 -1.50031969e-01 -2.53581494e-01 -3.28616165e-02 -6.69389129e-01 3.95252824e-01 -8.15304935e-01 2.27585211e-01 -1.12257934e+00 2.01491982e-01 -3.31890255e-01 3.29714149e-01 4.46988612e-01 -9.40036848e-02 -1.08876295e-01 3.20753485e-01 1.81699738e-01 -1.62375465e-01 -2.86019221e-03 9.90549207e-01 1.36927590e-01 -8.73891916e-03 -1.82520002e-01 -1.36757541e+00 1.33367419e+00 6.88505828e-01 -4.85969067e-01 -3.61628473e-01 -5.63205481e-01 6.86476350e-01 3.53919774e-01 5.99989295e-01 -2.83507347e-01 5.37644662e-02 -4.97792691e-01 1.27618983e-01 -2.43147910e-01 2.37290815e-01 -6.26248598e-01 1.40920788e-01 5.82047641e-01 -4.94256198e-01 9.82130885e-01 3.83330941e-01 -2.41558049e-02 -1.13124937e-01 -7.30073810e-01 5.66523254e-01 -3.73881549e-01 -7.60317862e-01 -6.75814033e-01 -5.94328463e-01 5.08206546e-01 6.46435559e-01 -1.25194594e-01 -7.89233446e-01 -1.26308322e-01 -1.08910942e+00 1.53285936e-01 5.66375554e-01 3.21478397e-01 6.58154249e-01 -1.09424222e+00 -7.13214099e-01 -2.01943710e-01 1.14650443e-01 -1.60164326e-01 -1.17782988e-01 6.71305418e-01 -5.69240689e-01 4.44286078e-01 -1.55193329e-01 -1.21487156e-01 -1.53122687e+00 5.74727595e-01 4.67902601e-01 2.56097198e-01 -2.47478634e-01 8.80922794e-01 4.43653822e-01 -5.75774908e-01 -2.54404902e-01 -3.66970569e-01 -1.52083531e-01 8.11227262e-02 4.93339092e-01 -1.39474511e-01 -2.28368580e-01 -9.74371850e-01 -7.79998839e-01 3.12074453e-01 -3.68039221e-01 -4.27023172e-01 6.45077944e-01 -4.40623760e-01 -6.78242743e-01 7.45245457e-01 7.13679850e-01 5.58700800e-01 -2.92462766e-01 5.51349968e-02 3.01476717e-01 -4.01771218e-01 -9.06898558e-01 -1.04541981e+00 2.33885631e-01 5.74028611e-01 -5.40660471e-02 4.69065309e-01 6.23565197e-01 5.35181046e-01 1.46069750e-01 7.57272661e-01 4.79032308e-01 -8.77792299e-01 -3.62166822e-01 7.51113653e-01 1.31416345e+00 -8.07877541e-01 -1.35511801e-01 -8.75592887e-01 -6.30112767e-01 1.03235877e+00 9.57340479e-01 2.85546407e-02 -8.22379813e-02 2.99575508e-01 3.94134045e-01 -4.95458186e-01 -1.02965462e+00 8.04574564e-02 -2.24556535e-01 5.36093295e-01 8.48486245e-01 3.44243258e-01 -1.26695049e+00 9.27058578e-01 -1.10334563e+00 -8.81862044e-01 9.66465652e-01 7.53070354e-01 -6.81148946e-01 -1.18512833e+00 -8.31694961e-01 4.36524488e-02 -4.92532909e-01 -4.09258425e-01 -7.78454065e-01 1.37778342e+00 -5.39754108e-02 1.05122316e+00 4.48320836e-01 1.49056911e-01 4.23649251e-01 1.74177200e-01 7.11646438e-01 -8.73811185e-01 -6.30403638e-01 -7.45533183e-02 6.77531302e-01 -3.00790966e-01 -3.74236733e-01 -9.83336687e-01 -1.81171024e+00 -3.41937423e-01 -3.39868754e-01 6.57708228e-01 -9.45346896e-03 1.26209497e+00 -1.68809488e-01 -4.30946760e-02 -3.45120169e-02 -2.32901022e-01 -9.21661675e-01 -1.01621747e+00 -3.79716605e-01 6.63661540e-01 1.41566172e-01 -5.05343020e-01 -6.25550389e-01 -3.34921300e-01]
[10.148037910461426, 9.2455472946167]
2d3f3d3c-022f-460e-989c-e8b65ff8c7ef
probing-inter-modality-visual-parsing-with-1
null
null
https://openreview.net/forum?id=e0nZIFEpmYh
https://openreview.net/pdf?id=e0nZIFEpmYh
Probing Inter-modality: Visual Parsing with Self-Attention for Vision-and-Language Pre-training
Vision-Language Pre-training (VLP) aims to learn multi-modal representations from image-text pairs and serves for downstream vision-language tasks in a fine-tuning fashion. The dominant VLP models adopt a CNN-Transformer architecture, which embeds images with a CNN, and then aligns images and text with a Transformer. Visual relationship between visual contents plays an important role in image understanding and is the basic for inter-modal alignment learning. However, CNNs have limitations in visual relation learning due to local receptive field's weakness in modeling long-range dependencies. Thus the two objectives of learning visual relation and inter-modal alignment are encapsulated in the same Transformer network. Such design might restrict the inter-modal alignment learning in the Transformer by ignoring the specialized characteristic of each objective. To tackle this, we propose a fully Transformer visual embedding for VLP to better learn visual relation and further promote inter-modal alignment. Specifically, we propose a metric named Inter-Modality Flow (IMF) to measure the interaction between vision and language modalities (i.e., inter-modality). We also design a novel masking optimization mechanism named Masked Feature Regression (MFR) in Transformer to further promote the inter-modality learning. To the best of our knowledge, this is the first study to explore the benefit of Transformer for visual feature learning in VLP. We verify our method on a wide range of vision-language tasks, including Visual Question Answering (VQA), Visual Entailment and Visual Reasoning. Our approach not only outperforms the state-of-the-art VLP performance, but also shows benefits on the IMF metric.
['Jiebo Luo', 'Houqiang Li', 'Jianlong Fu', 'Houwen Peng', 'Bei Liu', 'Yupan Huang', 'Hongwei Xue']
2021-05-21
null
null
null
neurips-2021-12
['visual-entailment']
['reasoning']
[ 6.37896582e-02 -1.46590490e-02 -3.46511960e-01 -4.09686923e-01 -4.96661872e-01 -5.23040473e-01 8.64919424e-01 -1.28778622e-01 -4.11171913e-01 1.62563279e-01 3.82440418e-01 -4.12683547e-01 9.17263702e-02 -7.90882647e-01 -9.37947631e-01 -6.11427486e-01 5.15501380e-01 -2.42345911e-02 2.14197025e-01 -2.21800923e-01 1.58224195e-01 1.82699010e-01 -1.54873109e+00 8.57088089e-01 7.55726039e-01 1.10251510e+00 5.35854578e-01 4.82165515e-01 -4.74088997e-01 1.24359632e+00 -1.32418796e-01 -5.20803511e-01 1.59713048e-02 -4.73920733e-01 -9.68677700e-01 1.21329412e-01 4.90871012e-01 -1.57906458e-01 -4.12468582e-01 1.00420535e+00 3.64542276e-01 -6.71525076e-02 7.89083064e-01 -1.55035710e+00 -1.37113166e+00 5.03255904e-01 -7.27368295e-01 1.93677202e-01 2.36659274e-01 4.74086523e-01 1.43101716e+00 -1.04169917e+00 3.02431226e-01 1.39419293e+00 5.09271562e-01 5.99349022e-01 -1.11353028e+00 -4.69162911e-01 3.56541991e-01 4.81369525e-01 -1.18929636e+00 -3.57665777e-01 9.58582580e-01 -5.06517828e-01 1.06682849e+00 1.94502488e-01 6.03278399e-01 1.05821109e+00 4.28226404e-02 1.04023373e+00 1.24724138e+00 -3.33787858e-01 -2.90203452e-01 3.18800986e-01 4.34037410e-02 9.40730095e-01 -3.31840277e-01 2.03800440e-01 -6.96743667e-01 4.83535618e-01 6.99598432e-01 2.08478197e-02 -4.68371063e-01 -2.55202681e-01 -1.40989053e+00 8.33089292e-01 7.05385864e-01 2.72042483e-01 -1.75890893e-01 1.72769949e-01 3.01019341e-01 2.98485249e-01 6.03686459e-02 6.03544973e-02 -3.78466576e-01 1.63882479e-01 -6.28031850e-01 -2.68759221e-01 4.66213346e-01 7.85503626e-01 8.74378979e-01 -4.01316844e-02 -6.12280786e-01 8.36773574e-01 7.24885046e-01 5.11111081e-01 4.67655063e-01 -7.58784056e-01 7.19832838e-01 8.18794429e-01 -4.56606269e-01 -8.52294743e-01 -3.08754116e-01 -1.63943097e-01 -1.00177062e+00 2.28939191e-01 4.32137668e-01 3.68572325e-01 -9.05304313e-01 1.97441781e+00 1.97037116e-01 -4.19780686e-02 2.91669965e-01 1.05711484e+00 1.43098414e+00 7.30448365e-01 2.24192336e-01 -7.57085830e-02 1.73679209e+00 -1.27835166e+00 -4.82148975e-01 -4.08960223e-01 3.40923220e-01 -9.05779064e-01 1.63056290e+00 -1.04794271e-01 -1.04652965e+00 -8.94948184e-01 -7.78864443e-01 -6.98418260e-01 -3.47129345e-01 1.67370841e-01 5.26044309e-01 2.84849614e-01 -1.00876892e+00 -2.68584453e-02 -5.14989495e-01 -3.09415430e-01 4.75287169e-01 1.92467168e-01 -4.56246734e-01 -1.88062474e-01 -1.30597126e+00 9.68870044e-01 2.34972030e-01 1.83961660e-01 -9.42755461e-01 -7.33279049e-01 -1.10302103e+00 -5.81904408e-03 3.08767587e-01 -1.09375513e+00 9.13505435e-01 -1.06898248e+00 -1.36676681e+00 1.20920646e+00 -3.31754029e-01 -2.21460238e-01 3.25757772e-01 7.91082829e-02 -2.98933566e-01 1.70894489e-01 7.28802308e-02 9.78810847e-01 1.11407721e+00 -1.39258182e+00 -5.23793817e-01 -2.61050731e-01 5.33027112e-01 4.22868520e-01 -4.15707231e-01 -2.07618669e-01 -8.30336094e-01 -5.51347256e-01 -7.85601884e-02 -5.64905107e-01 2.38258943e-01 1.53936937e-01 -3.85356098e-01 -3.95610988e-01 7.41673112e-01 -5.49084783e-01 8.83161545e-01 -2.10185266e+00 2.77166098e-01 -1.56298891e-01 3.39471847e-01 1.52964205e-01 -5.02813876e-01 2.32103333e-01 -5.78302294e-02 1.14969527e-02 -1.02818593e-01 -5.46662211e-01 2.38531288e-02 4.20614332e-01 -3.97446752e-01 3.17971438e-01 4.86535847e-01 1.49353576e+00 -7.75417745e-01 -7.67265141e-01 3.80832076e-01 7.06603646e-01 -5.42430758e-01 3.71027708e-01 -3.28096926e-01 5.46349406e-01 -2.98452288e-01 8.39914262e-01 5.49027622e-01 -6.02252305e-01 -1.71487302e-01 -8.72065544e-01 -1.51414663e-01 3.37031819e-02 -6.16790831e-01 1.79100180e+00 -7.47015655e-01 6.61128640e-01 -1.19139783e-01 -1.27441180e+00 7.66273975e-01 1.23591401e-01 1.73935786e-01 -1.17811537e+00 1.49015114e-01 -2.26460233e-01 -6.49589151e-02 -8.65682006e-01 2.58267879e-01 -1.84016168e-01 2.14153558e-01 3.45811695e-01 2.36594558e-01 1.12456515e-01 -3.90595553e-04 1.44035369e-01 5.65015793e-01 1.88156024e-01 2.97985464e-01 5.83702289e-02 9.92112279e-01 -2.43702590e-01 2.63382465e-01 6.10978365e-01 -3.63991648e-01 5.67304969e-01 6.22563124e-01 -1.69017211e-01 -8.40835452e-01 -1.25269902e+00 4.37393598e-02 1.28725350e+00 3.83086681e-01 -3.25920969e-01 -3.12331408e-01 -8.79097760e-01 -2.84798443e-02 5.36014915e-01 -7.72973478e-01 -2.73909599e-01 -3.78367394e-01 -5.05532622e-01 5.77334225e-01 6.95021391e-01 7.66923606e-01 -1.25978827e+00 -2.68151224e-01 -3.46346498e-01 -5.19964099e-01 -1.53757179e+00 -7.52374947e-01 6.36797473e-02 -4.69313890e-01 -1.16261292e+00 -5.83085239e-01 -1.16484916e+00 5.74695230e-01 3.63180399e-01 1.16541350e+00 4.04792652e-02 1.39610311e-02 7.44170666e-01 -3.14586043e-01 -7.42666274e-02 -2.59402573e-01 -1.12584025e-01 -3.40155602e-01 2.42239907e-01 4.16952074e-01 -5.40617764e-01 -6.85014367e-01 3.20356071e-01 -8.97238731e-01 3.61033916e-01 9.17588234e-01 1.05791092e+00 7.49932230e-01 -2.06091657e-01 2.18179271e-01 -4.41643387e-01 5.17547131e-01 -2.43049622e-01 -4.43106979e-01 7.26215184e-01 -4.32318330e-01 3.21551859e-01 6.96646154e-01 -5.93980849e-01 -1.01964855e+00 -9.18103456e-02 -1.15903459e-01 -7.08537400e-01 -1.04992375e-01 4.78574216e-01 -5.06222069e-01 -1.90364644e-01 2.93552458e-01 5.52996278e-01 1.48367118e-02 -1.07110523e-01 9.23798263e-01 4.15576756e-01 6.62196755e-01 -4.37026560e-01 8.24734867e-01 6.32550776e-01 1.10875696e-01 -7.15974629e-01 -1.07989371e+00 -4.26740378e-01 -5.17598927e-01 -1.25129879e-01 1.29070783e+00 -9.72175717e-01 -1.10362875e+00 3.73792708e-01 -1.35396540e+00 -3.14983219e-01 -1.81108028e-01 4.12281841e-01 -5.44644475e-01 5.54768443e-01 -4.61219728e-01 -5.22382438e-01 -3.68197680e-01 -1.26950586e+00 1.03874373e+00 2.11526871e-01 2.88058877e-01 -1.10038459e+00 -1.37200043e-01 7.99985766e-01 2.40046084e-01 -2.10758910e-01 9.76096630e-01 -2.75357127e-01 -6.78896248e-01 5.07911623e-01 -8.71942341e-01 6.12632096e-01 -3.51352990e-02 -2.24656284e-01 -1.19717515e+00 -1.03326790e-01 2.61564348e-02 -5.60690701e-01 1.27985930e+00 3.30588400e-01 1.21290684e+00 -1.68675721e-01 -8.37031100e-03 8.72901857e-01 1.42635441e+00 -1.11769326e-01 7.18632400e-01 3.74280483e-01 1.23365366e+00 7.55544245e-01 4.93842930e-01 5.05162738e-02 9.06120002e-01 6.24740064e-01 6.61998928e-01 -3.76719862e-01 -5.48130751e-01 -3.18158954e-01 5.83270013e-01 8.28236103e-01 -5.05656786e-02 -1.78652003e-01 -9.06884491e-01 4.64884967e-01 -1.92766631e+00 -9.43113387e-01 -4.69858944e-02 1.83990455e+00 8.66377056e-01 -1.58490002e-01 -4.18211557e-02 -1.21610448e-01 5.42520940e-01 3.84326011e-01 -3.99366617e-01 -2.40737453e-01 -3.27727258e-01 2.77465656e-02 2.26911694e-01 5.60918868e-01 -1.15835142e+00 1.06297386e+00 4.92950296e+00 7.69121230e-01 -1.26527143e+00 3.14443350e-01 4.73888844e-01 1.40088439e-01 -4.95058656e-01 5.82700074e-02 -7.24648118e-01 2.83334523e-01 3.12715888e-01 3.02686006e-01 5.26663899e-01 5.74906468e-01 4.61668661e-03 1.74707159e-01 -1.24965262e+00 1.36883950e+00 4.18118894e-01 -1.32754886e+00 3.86475623e-01 -4.14047018e-02 4.28077251e-01 5.27354926e-02 3.25478941e-01 3.30516249e-01 -1.76913232e-01 -1.28273773e+00 8.14028025e-01 5.75143099e-01 8.41542780e-01 -5.50384402e-01 5.07239699e-01 9.01090503e-02 -1.57122171e+00 -8.43644366e-02 -2.88294584e-01 2.06456974e-01 2.72590488e-01 3.27948451e-01 -2.96222061e-01 5.72263479e-01 7.92298079e-01 1.02349472e+00 -7.48079419e-01 5.37737727e-01 -4.50379163e-01 2.85742909e-01 1.48412526e-01 2.21261635e-01 3.33582669e-01 -1.41040236e-01 4.32076484e-01 1.11133206e+00 -5.61047308e-02 -2.46788099e-01 1.07028946e-01 1.22406054e+00 -6.74713776e-02 2.33165100e-02 -6.42141342e-01 -1.67153254e-01 1.85297102e-01 1.29031253e+00 -3.06909293e-01 -1.25137329e-01 -9.47238028e-01 1.07620215e+00 4.27092582e-01 5.43203592e-01 -1.02993846e+00 -6.60770312e-02 6.29445732e-01 -2.05636844e-01 5.39146483e-01 6.80610829e-04 -2.81159282e-01 -1.34321082e+00 2.43379787e-01 -8.85518909e-01 4.19448584e-01 -9.67216015e-01 -1.63809323e+00 6.26976788e-01 -1.74359128e-01 -1.27713168e+00 6.45765290e-02 -8.87291312e-01 -5.53489089e-01 9.17410135e-01 -2.04345298e+00 -1.98391068e+00 -4.75614995e-01 1.31617868e+00 4.33797002e-01 -2.26986766e-01 5.52498996e-01 2.09341154e-01 -3.86247396e-01 7.27024794e-01 -4.39831972e-01 2.64745981e-01 6.85581446e-01 -1.13588417e+00 -8.18027705e-02 8.69370937e-01 4.29292142e-01 5.84747970e-01 2.52627701e-01 -2.61840582e-01 -1.51662016e+00 -1.19481015e+00 9.07224000e-01 -4.50589836e-01 8.08718324e-01 -2.78359205e-01 -9.24981058e-01 6.02394581e-01 4.48636979e-01 2.05996647e-01 5.80342293e-01 3.20284590e-02 -8.97638500e-01 -2.19046503e-01 -6.60502970e-01 6.98423982e-01 1.00974059e+00 -1.23391593e+00 -8.25949490e-01 1.57172024e-01 9.64054048e-01 -2.07717299e-01 -7.86948204e-01 4.72937047e-01 5.55934012e-01 -1.05825686e+00 1.34351969e+00 -5.14794230e-01 6.86603010e-01 -4.81001526e-01 -4.05942202e-01 -9.37950671e-01 -2.66415238e-01 -1.80455416e-01 -1.93048075e-01 1.53328025e+00 3.18948895e-01 -5.72105467e-01 2.95270234e-01 1.77916616e-01 -6.48807138e-02 -8.41010213e-01 -6.99860752e-01 -5.30782938e-01 1.00414723e-01 -5.18931866e-01 3.11420679e-01 1.05584168e+00 -2.98265398e-01 7.00320542e-01 -5.01628757e-01 3.19977492e-01 5.58399320e-01 3.28439772e-01 5.35584807e-01 -7.18962193e-01 -5.08968651e-01 -6.92731559e-01 -2.68890053e-01 -1.17800891e+00 4.64221507e-01 -1.14739180e+00 -2.19914526e-01 -1.68211079e+00 4.76205707e-01 -1.01678304e-01 -4.37921762e-01 6.76282525e-01 -2.06126973e-01 3.54596496e-01 3.87904972e-01 2.35313207e-01 -7.27984011e-01 7.40727842e-01 1.70621514e+00 -4.25492585e-01 -5.49551621e-02 -3.78980190e-01 -7.61506379e-01 6.59076512e-01 6.60273373e-01 -6.06281199e-02 -7.20804513e-01 -6.91183805e-01 2.29750410e-01 -8.92246962e-02 8.80753815e-01 -5.63171029e-01 3.37740242e-01 -2.17727870e-01 3.69615763e-01 -6.39607787e-01 3.20498496e-01 -7.35725880e-01 -3.17322731e-01 1.68530390e-01 -4.42669362e-01 8.34840313e-02 8.84458274e-02 4.98212337e-01 -5.56258321e-01 1.33256853e-01 7.00009882e-01 -1.44375235e-01 -1.10095096e+00 4.75838333e-01 1.11178169e-02 3.19790483e-01 8.32018733e-01 -1.95186689e-01 -7.46173680e-01 -2.40047321e-01 -5.50738931e-01 4.15280670e-01 4.07563090e-01 6.15882635e-01 9.20344353e-01 -1.40232360e+00 -5.17739177e-01 2.31074631e-01 5.19520819e-01 -7.01502189e-02 3.95201892e-01 1.27512026e+00 -1.92838162e-01 4.05463457e-01 -2.85324842e-01 -8.87625515e-01 -1.31909323e+00 7.75587201e-01 4.81891334e-01 -2.59401441e-01 -5.91649055e-01 9.24286902e-01 8.77300262e-01 -3.13236833e-01 2.96988368e-01 -2.54290909e-01 -4.91338164e-01 3.58854048e-02 5.44293642e-01 -2.02251464e-01 -2.78675795e-01 -9.74015176e-01 -4.63925540e-01 1.00288582e+00 3.72842848e-02 -4.64860648e-02 8.96495879e-01 -4.32211012e-01 -3.96316588e-01 4.62969363e-01 1.48143888e+00 -2.11193740e-01 -1.25026488e+00 -4.57402766e-01 -4.22993839e-01 -2.19006911e-01 3.27737965e-02 -6.74611926e-01 -1.35012555e+00 1.29147887e+00 5.03608823e-01 -4.26249877e-02 1.33777237e+00 3.95391285e-01 6.88169360e-01 1.86478809e-01 -2.18472555e-01 -7.27088928e-01 6.49644852e-01 6.41517341e-01 1.02881002e+00 -1.57677650e+00 -2.85505116e-01 -2.95462608e-01 -1.04085100e+00 1.06543314e+00 8.99491906e-01 2.81975865e-01 6.20509565e-01 -1.07036948e-01 2.20788479e-01 -3.14903051e-01 -6.76686883e-01 -7.88730025e-01 8.64346325e-01 7.90554523e-01 3.92901778e-01 -1.95475340e-01 3.01571749e-02 4.55561280e-01 -2.33253073e-02 -2.13006929e-01 -1.51099831e-01 5.22314906e-01 -1.15219258e-01 -1.08937252e+00 -1.89444453e-01 9.82473791e-02 -1.87207997e-01 -4.66411948e-01 -2.94921517e-01 8.45208585e-01 4.69714940e-01 9.52117622e-01 9.10937116e-02 -4.62691188e-01 2.06000090e-01 -9.42557529e-02 7.00802088e-01 -2.29410365e-01 -5.36834002e-01 1.85381118e-02 -1.83385029e-01 -7.75856197e-01 -8.64492357e-01 -1.79845870e-01 -1.21632993e+00 -1.68572992e-01 5.92005998e-03 -3.39448929e-01 3.31031024e-01 1.09768879e+00 1.16296396e-01 6.71623230e-01 4.82261270e-01 -5.75189292e-01 -2.70667166e-01 -6.17577493e-01 -3.45489830e-01 7.19514489e-01 5.24681509e-01 -6.63810134e-01 -3.12136650e-01 9.85568836e-02]
[10.746261596679688, 1.496458888053894]
0509bf57-56e6-4962-9b34-dfaab6e0d312
implementing-measurement-error-models-in-a
2307.01539
null
https://arxiv.org/abs/2307.01539v1
https://arxiv.org/pdf/2307.01539v1.pdf
Implementing measurement error models in a likelihood-based framework for estimation, identifiability analysis, and prediction in the life sciences
Throughout the life sciences we routinely seek to interpret measurements and observations using parameterised mechanistic mathematical models. A fundamental and often overlooked choice in this approach involves relating the solution of a mathematical model with noisy and incomplete measurement data. This is often achieved by assuming that the data are noisy measurements of the solution of a deterministic mathematical model, and that measurement errors are additive and normally distributed. While this assumption of additive Gaussian noise is extremely common and simple to implement and interpret, it is often unjustified and can lead to poor parameter estimates and non-physical predictions. One way to overcome this challenge is to implement a different measurement error model. In this review, we demonstrate how to implement a range of measurement error models in a likelihood-based framework for estimation, identifiability analysis, and prediction. We focus our implementation within a frequentist profile likelihood-based framework, but our approach is directly relevant to other approaches including sampling-based Bayesian methods. Case studies, motivated by simple caricature models routinely used in the systems biology and mathematical biology literature, illustrate how the same ideas apply to different types of mathematical models. Open-source Julia code to reproduce results is available on GitHub.
['Matthew J. Simpson', 'Oliver J. Maclaren', 'Ryan J. Murphy']
2023-07-04
null
null
null
null
['caricature']
['computer-vision']
[ 5.62220991e-01 -1.68308467e-01 4.50247154e-02 -2.09561363e-01 -4.32522863e-01 -5.85375249e-01 6.63455427e-01 2.66824752e-01 -2.78428406e-01 1.15630126e+00 -2.99778711e-02 -4.82734561e-01 -7.29937136e-01 -6.84714496e-01 -6.85378015e-01 -8.73301208e-01 1.58211187e-01 5.54070354e-01 2.56877225e-02 5.76914549e-02 3.31679255e-01 5.01630008e-01 -1.47873104e+00 -4.91499990e-01 5.28357267e-01 6.01671159e-01 3.35232466e-01 8.43203127e-01 1.60838217e-02 3.03356498e-01 -5.01562417e-01 -1.90634623e-01 -7.16709644e-02 -5.36734760e-01 -4.01073694e-01 -1.04308456e-01 -2.79658973e-01 7.40709528e-02 8.00673142e-02 8.25029135e-01 4.37179536e-01 -1.02391012e-01 1.04848337e+00 -1.07095981e+00 -5.30512810e-01 3.58485192e-01 -1.57635570e-01 4.34099399e-02 3.41132522e-01 6.67983294e-02 4.41780686e-01 -4.08771068e-01 2.79823631e-01 1.41545141e+00 9.04868007e-01 1.40080303e-01 -1.76832569e+00 -3.08034271e-01 -1.03269838e-01 -2.10810259e-01 -1.64854980e+00 -5.66146612e-01 2.85814106e-01 -6.83768153e-01 4.97919172e-01 4.21113431e-01 6.56163335e-01 1.25749373e+00 6.87682986e-01 2.21483156e-01 1.37384856e+00 -4.55172956e-01 5.61724126e-01 1.61303088e-01 -6.56941161e-03 2.03035921e-01 8.41215611e-01 5.63582063e-01 -2.83423543e-01 -5.39837778e-01 8.25631678e-01 1.19962230e-01 -1.71634898e-01 -3.31860095e-01 -9.85559046e-01 8.01053107e-01 -2.24553555e-01 2.27741569e-01 -4.70498502e-01 5.44094205e-01 8.56202543e-02 1.46256968e-01 4.08527970e-01 2.75613040e-01 -4.93049204e-01 -2.07612857e-01 -9.06517327e-01 4.73990232e-01 1.13507128e+00 6.00508213e-01 5.87006450e-01 3.97801958e-03 2.44437844e-01 7.13407636e-01 8.35249066e-01 7.80519068e-01 1.39320314e-01 -1.17601466e+00 -4.65644479e-01 1.12017229e-01 5.26700497e-01 -8.45842481e-01 -3.72676998e-01 -3.74092162e-01 -7.88188279e-01 1.88311130e-01 8.00412834e-01 -6.03823643e-03 -8.91952872e-01 1.64689982e+00 4.71092671e-01 3.12027931e-01 -7.49745443e-02 6.37452960e-01 4.43022341e-01 5.77543676e-01 2.34729052e-01 -5.94300508e-01 1.06659281e+00 1.33480728e-01 -8.12038541e-01 -1.42876983e-01 3.79928529e-01 -8.58331382e-01 6.36369467e-01 5.20665944e-01 -8.85215998e-01 2.89698336e-02 -8.96057963e-01 1.58599243e-01 -4.07062232e-01 -1.99068487e-01 4.31006700e-01 9.27420080e-01 -7.65808046e-01 6.80703580e-01 -1.01988614e+00 -6.62202895e-01 7.36206248e-02 8.45882520e-02 -1.08526334e-01 -3.89904045e-02 -9.07219470e-01 1.24591804e+00 1.00505399e-02 2.92628437e-01 -1.04789233e+00 -7.98662066e-01 -6.63664877e-01 -1.21802405e-01 4.16041732e-01 -7.59271681e-01 1.24028647e+00 -3.25700492e-01 -1.58369601e+00 5.48601151e-01 -3.64748746e-01 -2.00010225e-01 4.81823772e-01 8.21607485e-02 -5.90393133e-02 -3.27780724e-01 -1.37867793e-01 -1.26689479e-01 4.57914740e-01 -1.35686815e+00 -1.18175391e-02 -2.92212605e-01 -2.49605671e-01 -1.88086689e-01 5.28565347e-01 1.19680487e-01 3.58247131e-01 -3.97753805e-01 3.47926259e-01 -7.26726115e-01 -4.87123311e-01 2.21319437e-01 -2.89755404e-01 1.25877842e-01 3.46551389e-01 -5.00899315e-01 1.12781799e+00 -1.75208008e+00 1.40824407e-01 3.04169029e-01 -9.19090882e-02 -2.47894805e-02 1.30781516e-01 8.47800076e-01 2.26391867e-01 3.33996236e-01 -7.63712823e-01 -3.23152930e-01 1.31661326e-01 4.79606897e-01 -2.32912436e-01 9.25302327e-01 6.23782985e-02 7.01840937e-01 -1.02208698e+00 -1.54239178e-01 5.98467410e-01 8.19770753e-01 -9.18593258e-02 2.62703951e-02 -1.08884409e-01 5.86889565e-01 -3.55686516e-01 5.93609095e-01 5.79329550e-01 -1.32081077e-01 3.20313454e-01 2.98943520e-01 -4.28562403e-01 2.97875702e-01 -1.58265209e+00 1.11764085e+00 -2.52140969e-01 3.08631897e-01 1.24035984e-01 -1.23703969e+00 9.18843865e-01 3.59372497e-01 3.15420121e-01 -9.93267866e-04 3.21531981e-01 4.40409184e-01 1.84950694e-01 -4.08935279e-01 -2.86454037e-02 -7.29236126e-01 9.27004870e-03 5.00776887e-01 -2.75906622e-02 -7.18552709e-01 -4.10727337e-02 -2.58004546e-01 9.69031692e-01 4.17647749e-01 8.70324373e-01 -5.75194001e-01 2.57113844e-01 -7.52857104e-02 5.91387093e-01 9.56756473e-01 -4.83390354e-02 4.50904846e-01 4.53680813e-01 -2.56965727e-01 -1.08750212e+00 -1.11138129e+00 -7.55001426e-01 3.76260698e-01 -1.52104616e-01 -2.54132748e-01 -4.80810016e-01 3.18997145e-01 1.53966501e-01 8.11295509e-01 -8.45832229e-01 -1.94534227e-01 -7.91823044e-02 -1.27947176e+00 4.66937095e-01 1.23573177e-01 -2.63495892e-01 -5.29600561e-01 -6.41958296e-01 5.34362137e-01 1.75683483e-01 -4.84901756e-01 3.47269386e-01 4.71223533e-01 -7.36552179e-01 -1.03891432e+00 -6.23189509e-01 8.54548588e-02 3.95267397e-01 5.40153049e-02 8.55805457e-01 1.59494236e-01 -5.55935204e-01 5.16042113e-01 -2.00031251e-01 -9.58018303e-01 -7.26754844e-01 -5.04681945e-01 2.54100740e-01 -3.23548764e-01 4.69860941e-01 -7.05579281e-01 -4.18671846e-01 4.41499174e-01 -1.11922920e+00 -5.18141747e-01 2.99753129e-01 8.76045227e-01 6.39406204e-01 -1.46100566e-01 5.45577824e-01 -5.70613384e-01 5.22996545e-01 -7.28194296e-01 -9.32056129e-01 2.33256489e-01 -5.99704802e-01 -1.22586703e-02 3.28673899e-01 -4.43959326e-01 -8.51090610e-01 1.97050106e-02 -4.63343710e-02 -6.65560067e-02 -5.25857687e-01 9.24100339e-01 -1.01319626e-01 -4.45351042e-02 6.58934653e-01 2.60764152e-01 3.53095621e-01 -5.12516379e-01 1.86008021e-01 6.88106179e-01 1.16672330e-01 -7.74063110e-01 4.16287839e-01 5.16495764e-01 6.20310783e-01 -1.19325197e+00 -5.41991413e-01 -3.27839434e-01 -5.34011543e-01 -1.98217556e-01 4.94938582e-01 -5.00356734e-01 -8.13665092e-01 4.20111656e-01 -8.52436244e-01 -5.38098395e-01 -3.85417789e-01 7.30675101e-01 -9.27306056e-01 3.56873751e-01 -1.08007267e-01 -1.58174634e+00 4.38882768e-01 -1.17459035e+00 1.08544874e+00 2.45088041e-02 -4.28175837e-01 -1.45057893e+00 4.96262163e-01 -1.00792736e-01 5.89786947e-01 3.02300721e-01 7.26972938e-01 -4.06424999e-01 -3.25508237e-01 -4.01191205e-01 1.12131715e-01 1.16811000e-01 2.01663509e-01 4.51090574e-01 -9.40599859e-01 4.99888025e-02 3.22934240e-01 -2.12092716e-02 6.00200772e-01 1.00070071e+00 7.17565835e-01 4.47890498e-02 -3.96416664e-01 3.61242950e-01 1.49168038e+00 1.48552030e-01 6.14469647e-01 3.95629816e-02 8.06383193e-02 9.99260485e-01 4.34187949e-01 6.62043154e-01 2.55382806e-01 6.87362432e-01 4.21510249e-01 3.80039245e-01 3.57651204e-01 -1.31392241e-01 9.07273367e-02 4.82161760e-01 8.29546601e-02 -4.69801933e-01 -8.64474654e-01 4.89626616e-01 -2.01292133e+00 -1.13306916e+00 -6.12566411e-01 2.64754248e+00 8.04104030e-01 -3.19586277e-01 1.37272999e-01 1.33000121e-01 4.94953990e-01 -3.92284155e-01 -2.91617960e-01 -3.51447880e-01 -2.61875123e-01 -4.16214503e-02 6.63540244e-01 7.89227307e-01 -7.56231487e-01 3.81204635e-01 8.12092686e+00 5.83661735e-01 -8.31792474e-01 9.21649113e-02 4.32710797e-01 -6.75198659e-02 -3.68776649e-01 3.94319028e-01 -8.05081546e-01 6.77967370e-01 1.55547082e+00 -3.56637448e-01 4.01589483e-01 1.12112045e-01 9.75443602e-01 -7.45404005e-01 -1.12178028e+00 5.55857360e-01 -3.40492576e-01 -9.83238161e-01 -3.98369849e-01 3.87359887e-01 4.26232845e-01 -1.87565640e-01 -1.26349628e-01 -1.05609357e-01 5.61289549e-01 -1.25532675e+00 6.02467895e-01 1.03196931e+00 4.24289107e-01 -2.43959188e-01 7.54496276e-01 5.38092852e-01 -6.44355834e-01 1.33683652e-01 -7.99133658e-01 -5.56806028e-01 6.26956463e-01 1.06714904e+00 -5.21490514e-01 4.23958242e-01 4.78235185e-01 4.10911500e-01 3.01105231e-02 1.56587636e+00 -1.74817964e-01 8.40589106e-01 -8.70755732e-01 -1.62167460e-01 -1.42797351e-01 -5.57829499e-01 6.96508110e-01 1.13498390e+00 7.02923000e-01 9.65414718e-02 -2.32195809e-01 1.12311947e+00 7.68824875e-01 -6.65969849e-02 -7.92300403e-01 -1.41303599e-01 6.27958000e-01 8.53743553e-01 -5.74768424e-01 -1.93243459e-01 -4.38944399e-01 2.71897495e-01 -1.23604782e-01 4.01268244e-01 -6.19337559e-01 1.89261705e-01 7.49810934e-01 2.55652398e-01 3.55811417e-02 -2.74330139e-01 -3.54693234e-01 -8.98284316e-01 -3.49944919e-01 -5.28439522e-01 1.47709474e-01 -7.61979699e-01 -1.41536963e+00 -3.03952068e-01 7.21989572e-01 -8.57196331e-01 -3.66004556e-01 -8.83336246e-01 -3.57046634e-01 1.16278422e+00 -1.11801183e+00 -8.98939013e-01 1.29686847e-01 -1.02045558e-01 1.10290766e-01 4.22754824e-01 8.90645683e-01 -8.93732384e-02 -5.20641029e-01 -9.78176743e-02 7.03462124e-01 -7.14294970e-01 5.22048175e-01 -1.24170578e+00 -1.42363580e-02 5.80493212e-01 -2.62986273e-01 1.05761731e+00 1.43486750e+00 -9.31173384e-01 -1.34765363e+00 -9.09352541e-01 7.37082839e-01 -5.70951045e-01 9.26871181e-01 -2.65927345e-01 -9.20671046e-01 6.82618976e-01 -2.48624310e-01 -1.68150008e-01 1.02839613e+00 -2.63825785e-02 3.36712338e-02 3.54144216e-01 -1.24303520e+00 4.94868815e-01 6.37060404e-01 -2.40214854e-01 -5.49053311e-01 3.65580589e-01 1.82351410e-01 -1.15622962e-02 -1.14178753e+00 3.03962976e-01 8.94430339e-01 -6.87318683e-01 9.07085657e-01 -6.22370660e-01 -3.27739865e-02 -6.64469898e-01 -4.63324934e-01 -1.38549674e+00 -1.85158908e-01 -6.57346010e-01 9.87058058e-02 9.98428226e-01 2.58624434e-01 -9.58114862e-01 1.83165640e-01 6.82574868e-01 1.17522843e-01 -5.45975447e-01 -1.24045336e+00 -9.88905966e-01 4.97660607e-01 -5.66174567e-01 4.43125993e-01 6.09135330e-01 -4.96566482e-02 -1.00177780e-01 -4.80416626e-01 1.67877704e-01 9.95160341e-01 -3.14822465e-01 7.57316649e-01 -1.60881221e+00 -4.04800117e-01 -2.20656112e-01 -5.21625459e-01 -7.39678025e-01 -1.44316658e-01 -2.89891154e-01 4.91260707e-01 -1.44512272e+00 2.60509700e-01 -2.96117485e-01 -1.02359811e-02 -1.17485024e-01 -7.24086463e-02 1.42352566e-01 -2.77398348e-01 4.25261967e-02 8.18740800e-02 3.77442151e-01 7.95757890e-01 3.17894101e-01 -6.42317440e-03 1.75871700e-01 -7.18202233e-01 6.34180427e-01 6.66079223e-01 -7.94835925e-01 -3.64729911e-01 -5.83157167e-02 5.29120445e-01 8.89552906e-02 9.59269285e-01 -4.19534594e-01 -1.53027372e-02 -6.14810705e-01 2.43911058e-01 -2.46666372e-01 4.06840980e-01 -7.68803298e-01 1.10973346e+00 4.79540974e-01 -1.20421119e-01 -2.11980581e-01 2.58621097e-01 8.92412066e-01 2.66914040e-01 -7.36945391e-01 7.05661654e-01 -3.73451740e-01 6.71885982e-02 -5.66877984e-02 -9.66803014e-01 -4.13588434e-01 8.79880786e-01 -3.60564768e-01 -4.62275416e-01 -6.00615978e-01 -1.06741261e+00 -5.54858856e-02 8.05632770e-01 -1.58034772e-01 3.78613055e-01 -7.79096305e-01 -7.43489623e-01 -1.35803834e-01 -5.17760776e-03 -2.20215037e-01 1.39399558e-01 1.37557864e+00 -3.96562666e-01 5.11261582e-01 1.97031423e-01 -7.06428826e-01 -7.49421835e-01 4.69112307e-01 6.04243159e-01 1.86459452e-01 -9.42394510e-02 4.50966388e-01 2.30037078e-01 -4.74627018e-01 -2.60398805e-01 -3.19888532e-01 1.39637187e-01 -2.58274347e-01 5.15575528e-01 6.60278678e-01 -2.21319661e-01 -6.27262533e-01 -3.81451517e-01 5.30557632e-01 5.45486212e-01 -3.19185734e-01 1.35607886e+00 -6.23902500e-01 -2.43976563e-01 1.14676464e+00 7.38805830e-01 -1.66074067e-01 -1.17142928e+00 -3.08583360e-02 -1.46907950e-02 -4.02167439e-01 9.89670008e-02 -8.45093548e-01 -8.04407522e-02 7.82646120e-01 3.41177791e-01 5.32428801e-01 5.79796672e-01 -1.93836372e-02 -2.16092303e-01 1.08589418e-01 3.01109642e-01 -6.52049839e-01 -7.21239209e-01 1.21543422e-01 1.07031083e+00 -1.00782180e+00 4.03482765e-01 -4.98187274e-01 8.17675292e-02 9.40361977e-01 -5.38383611e-02 -1.62314817e-01 1.01463687e+00 5.78899205e-01 -4.90326695e-02 -1.30830199e-01 -7.84490526e-01 -2.32908398e-01 -4.90525030e-02 8.57026458e-01 6.22804165e-01 -4.24325652e-02 -7.43702292e-01 4.33977723e-01 -2.28664428e-02 2.39455476e-01 7.89838433e-01 9.83174562e-01 -5.23623645e-01 -1.25476444e+00 -7.48056054e-01 5.42461753e-01 -5.54486811e-01 6.19132966e-02 -3.98501188e-01 7.36412883e-01 -2.83833921e-01 1.24580741e+00 -4.38177288e-02 2.13458851e-01 5.40316924e-02 1.07286200e-01 5.59875786e-01 -6.06328905e-01 -8.38339925e-02 3.94096494e-01 4.63702902e-02 -2.97411203e-01 -7.28385985e-01 -1.18025732e+00 -6.42064512e-01 -4.38646048e-01 -5.48921347e-01 1.24795660e-01 9.08281624e-01 1.18215168e+00 8.57003182e-02 3.65139037e-01 7.77872950e-02 -8.76171827e-01 -8.54397953e-01 -1.08696663e+00 -8.24384928e-01 -1.81158930e-01 3.30568552e-01 -1.14227784e+00 -7.75278270e-01 9.42855179e-02]
[6.560003280639648, 3.9903030395507812]
17be832d-fb35-4411-b383-f1aef56e4efe
exploiting-explicit-paths-for-multi-hop
1811.01127
null
https://arxiv.org/abs/1811.01127v2
https://arxiv.org/pdf/1811.01127v2.pdf
Exploiting Explicit Paths for Multi-hop Reading Comprehension
We propose a novel, path-based reasoning approach for the multi-hop reading comprehension task where a system needs to combine facts from multiple passages to answer a question. Although inspired by multi-hop reasoning over knowledge graphs, our proposed approach operates directly over unstructured text. It generates potential paths through passages and scores them without any direct path supervision. The proposed model, named PathNet, attempts to extract implicit relations from text through entity pair representations, and compose them to encode each path. To capture additional context, PathNet also composes the passage representations along each path to compute a passage-based representation. Unlike previous approaches, our model is then able to explain its reasoning via these explicit paths through the passages. We show that our approach outperforms prior models on the multi-hop Wikihop dataset, and also can be generalized to apply to the OpenBookQA dataset, matching state-of-the-art performance.
['Souvik Kundu', 'Ashish Sabharwal', 'Tushar Khot', 'Peter Clark']
2018-11-02
exploiting-explicit-paths-for-multi-hop-1
https://aclanthology.org/P19-1263
https://aclanthology.org/P19-1263.pdf
acl-2019-7
['multi-hop-reading-comprehension', 'implicit-relations']
['natural-language-processing', 'natural-language-processing']
[ 1.69072852e-01 9.45839882e-01 -3.06324333e-01 -3.58612299e-01 -9.84996319e-01 -6.52188778e-01 7.10931718e-01 1.02515376e+00 -1.98451459e-01 8.57348442e-01 5.90114772e-01 -7.44611681e-01 -7.50110686e-01 -1.45442390e+00 -1.01340759e+00 2.80238420e-01 -2.64805313e-02 9.22720671e-01 8.85588586e-01 -6.92678154e-01 3.97462249e-01 -1.34161487e-01 -1.34887075e+00 6.56455338e-01 1.40426624e+00 4.42302227e-01 8.56978223e-02 9.62609708e-01 -7.80313134e-01 1.48152828e+00 -3.89880657e-01 -9.17707384e-01 -2.76577115e-01 -6.05342090e-01 -1.89275086e+00 -4.23245251e-01 5.67089617e-01 -2.58049190e-01 -5.39285362e-01 7.11025357e-01 -8.29540715e-02 5.38007557e-01 6.58340394e-01 -9.29293573e-01 -9.67659593e-01 1.12150848e+00 1.09672544e-06 4.17751551e-01 1.18809962e+00 -4.11175400e-01 1.72397137e+00 -6.17343783e-01 8.49140584e-01 1.21212244e+00 5.69547594e-01 2.94670790e-01 -9.59781647e-01 9.39903688e-03 4.15752918e-01 8.80067706e-01 -8.48785222e-01 1.50120541e-01 4.59259748e-01 -1.01004124e-01 1.27883148e+00 2.32424870e-01 5.71904957e-01 8.68335605e-01 -5.20616025e-02 8.75903130e-01 7.53361225e-01 -4.95408267e-01 -1.18411049e-01 -9.86336172e-02 1.03220618e+00 1.15070188e+00 8.30856711e-02 -4.16601717e-01 -7.59068489e-01 -2.59306610e-01 3.58092636e-01 -3.39477032e-01 -6.79521859e-01 -1.87081203e-01 -1.22422874e+00 6.79170847e-01 6.96794152e-01 1.41807705e-01 -4.53334451e-01 -8.82741585e-02 5.50440401e-02 4.62125033e-01 -2.97206473e-02 8.20823371e-01 -6.77749157e-01 -1.35316193e-01 -7.23085463e-01 6.51400864e-01 1.49407136e+00 1.14292765e+00 8.34573865e-01 -8.39049637e-01 -7.04656959e-01 5.70461452e-01 2.60701656e-01 5.23483679e-02 1.02114148e-01 -1.03352284e+00 9.90712345e-01 9.53088880e-01 2.40464240e-01 -1.20826411e+00 -4.88259882e-01 -2.45275468e-01 -3.06629926e-01 -4.97281343e-01 5.88467419e-01 5.94203733e-02 -6.97550178e-01 1.51876044e+00 3.50466907e-01 4.07287896e-01 4.35820282e-01 6.67824507e-01 1.24935257e+00 7.57555664e-01 1.84358880e-01 4.41139750e-02 1.43752110e+00 -1.57127964e+00 -5.42667925e-01 -6.77153049e-03 5.94053626e-01 -3.79997969e-01 9.87873852e-01 2.50766724e-01 -1.37274098e+00 -3.28089565e-01 -8.60767901e-01 -5.52965343e-01 -5.80818295e-01 -3.10213000e-01 3.40767622e-01 -1.55738994e-01 -1.09173167e+00 7.28582740e-01 -4.17504728e-01 -5.42890489e-01 1.17925674e-01 -4.20634672e-02 -1.54795423e-01 -4.81671482e-01 -1.68706167e+00 1.02997243e+00 6.71163857e-01 -1.39182776e-01 -6.09896421e-01 -9.68363822e-01 -9.67503488e-01 4.46439624e-01 7.92830706e-01 -1.42135501e+00 1.32403779e+00 -9.39188972e-02 -1.32173181e+00 3.47396761e-01 -5.42158604e-01 -5.93317330e-01 2.44788542e-01 -4.86432016e-01 -4.59531724e-01 6.10230863e-01 2.72862166e-01 3.87467414e-01 3.88412803e-01 -1.17634404e+00 -5.55289090e-01 9.66988057e-02 8.37445557e-01 3.48037422e-01 1.19092271e-01 -1.77172720e-01 -6.92487478e-01 -3.19713503e-01 1.74625412e-01 -5.44620752e-01 -3.14118117e-01 -2.83831984e-01 -6.85011387e-01 -5.75482786e-01 2.40557715e-01 -9.58901584e-01 1.33689356e+00 -1.17280054e+00 5.11370301e-01 2.58473545e-01 4.40698236e-01 7.61314034e-02 -3.39018077e-01 8.34072888e-01 3.58847201e-01 1.62049279e-01 -4.39776599e-01 9.97021887e-03 5.54405823e-02 3.96603733e-01 -3.54774266e-01 -5.07741630e-01 3.65674496e-01 1.08607101e+00 -1.37610734e+00 -5.85911572e-01 -3.03342760e-01 2.12435216e-01 -7.00086832e-01 2.48273507e-01 -8.43451083e-01 6.26667514e-02 -7.73939371e-01 3.61728489e-01 4.35601175e-01 -5.84196568e-01 2.32211083e-01 8.94555598e-02 3.77748817e-01 8.12792003e-01 -8.96660030e-01 1.95173657e+00 -5.49138486e-01 3.36562812e-01 -4.67320502e-01 -7.70522654e-01 8.39530230e-01 2.75231957e-01 -3.88541929e-02 -6.77800834e-01 -3.58589381e-01 5.04078828e-02 -2.39017606e-01 -8.51600289e-01 9.39795315e-01 2.55254000e-01 -3.22882719e-02 5.59440792e-01 3.65510434e-01 -8.94895718e-02 5.83510756e-01 9.71019030e-01 1.52807653e+00 4.39397812e-01 3.35427165e-01 8.20900500e-02 1.02568293e+00 4.83208150e-01 9.38324183e-02 1.10105860e+00 3.55067968e-01 3.39622587e-01 6.85012758e-01 -2.00379044e-01 -6.09079778e-01 -1.27114069e+00 2.35594183e-01 1.01244593e+00 4.53308284e-01 -9.44600999e-01 -5.42951405e-01 -1.21795011e+00 9.11000222e-02 1.07507455e+00 -5.18411756e-01 8.64985362e-02 -7.07424521e-01 -9.23053548e-02 5.95002294e-01 6.38160348e-01 5.23534715e-01 -1.03743100e+00 -1.69879094e-01 5.43290854e-01 -8.47771883e-01 -1.21165121e+00 8.63903239e-02 -2.54165560e-01 -7.38900840e-01 -1.57533240e+00 -2.83630490e-01 -7.11607873e-01 5.32953203e-01 8.15393217e-03 1.72297490e+00 5.62343240e-01 1.05174072e-01 7.52554417e-01 -8.52322161e-01 -1.74324363e-01 -4.09280986e-01 2.91250765e-01 -5.91927707e-01 -4.36378777e-01 4.55997527e-01 -5.18518567e-01 -5.45594573e-01 1.46603316e-01 -6.42267287e-01 5.19598350e-02 3.78120452e-01 8.10211837e-01 7.06944644e-01 1.13025799e-01 7.27606714e-01 -1.21603310e+00 9.93369758e-01 -9.31760371e-01 -1.35333151e-01 8.81809533e-01 -5.85931659e-01 4.24624920e-01 8.24443579e-01 1.26911923e-01 -1.37387800e+00 -6.59600258e-01 -2.34034806e-01 1.64933845e-01 -1.31396264e-01 1.02992558e+00 1.40360326e-01 1.69653758e-01 6.25998974e-01 5.98947890e-02 -4.66383696e-01 -4.59687859e-01 8.58821571e-01 3.93085666e-02 6.39006734e-01 -8.86386991e-01 9.08923924e-01 1.77772865e-02 3.44999917e-02 -4.04054642e-01 -1.44205105e+00 -5.63907862e-01 -7.63950825e-01 1.81246419e-02 9.51441944e-01 -6.78486168e-01 -7.79950142e-01 -2.57735997e-01 -1.41339886e+00 -1.77643359e-01 -1.43568531e-01 2.70082057e-01 -5.88653386e-01 4.24171805e-01 -7.14236856e-01 -4.91372228e-01 -3.67290854e-01 -4.48135287e-01 6.23116851e-01 3.81604344e-01 -4.72922981e-01 -1.37476051e+00 2.21339330e-01 8.31953943e-01 1.99593738e-01 -1.53399995e-02 1.45905495e+00 -9.77171838e-01 -9.59918380e-01 3.60320718e-03 -3.39240372e-01 -1.64152071e-01 -2.43559927e-01 -1.51788533e-01 -5.39348602e-01 2.45218351e-01 -5.37301779e-01 -4.59713906e-01 1.03677595e+00 -2.59953022e-01 1.15425611e+00 -6.68420076e-01 -3.32381070e-01 1.00343302e-01 1.41928494e+00 -3.16359162e-01 6.91194117e-01 5.17823040e-01 6.79896474e-01 7.45181441e-01 5.98441601e-01 4.65419814e-02 1.20180809e+00 2.22823679e-01 3.39810938e-01 3.01756561e-01 -2.17518240e-01 -8.84037316e-01 -5.43496720e-02 9.41475391e-01 -1.98775098e-01 -4.96048719e-01 -1.01028776e+00 8.40715706e-01 -1.94648838e+00 -1.13663316e+00 -6.77933216e-01 1.63887143e+00 7.23411858e-01 1.95793472e-02 -9.45572853e-02 6.29830956e-02 2.09003717e-01 9.34993848e-02 -3.22859764e-01 -5.49269617e-01 2.99636424e-02 6.39407218e-01 -2.71727191e-03 9.59903121e-01 -5.44619977e-01 9.78295088e-01 6.73680830e+00 2.05693305e-01 -5.41539788e-02 -9.71456990e-02 -6.49902821e-02 3.43589574e-01 -8.03087115e-01 2.78823674e-01 -7.76715636e-01 -2.05602035e-01 9.50695813e-01 -4.89374429e-01 3.98184448e-01 4.03958708e-01 -2.88579077e-01 -1.29366204e-01 -1.26329780e+00 1.11539274e-01 2.66770482e-01 -1.57921922e+00 5.56550503e-01 -6.40251517e-01 6.19627237e-01 -2.67050833e-01 -4.56254542e-01 6.63832545e-01 6.86433017e-01 -9.82051373e-01 3.33480120e-01 9.08308208e-01 2.73245014e-02 -6.54882669e-01 6.56166673e-01 5.36747098e-01 -1.25249696e+00 -2.54069239e-01 -3.41201603e-01 -1.50564224e-01 4.24812347e-01 2.82786697e-01 -1.12711823e+00 1.46156621e+00 5.38177431e-01 7.50804663e-01 -8.44802499e-01 1.10663521e+00 -9.56502974e-01 7.32505620e-01 8.62349048e-02 -3.28131944e-01 3.54991108e-01 -2.33469866e-02 6.29511714e-01 1.15269184e+00 3.12682211e-01 5.02194762e-01 1.06762096e-01 1.00104928e+00 -2.26540476e-01 2.21206456e-01 -2.17768073e-01 -1.25630414e-02 4.79168028e-01 9.53547478e-01 -1.97749570e-01 -6.75769806e-01 -6.16707146e-01 1.03818166e+00 9.80673492e-01 5.40835559e-01 -6.70802891e-01 -6.44547403e-01 3.36077005e-01 -3.76224294e-02 3.92450601e-01 -7.98685253e-02 -6.82451725e-02 -1.25399315e+00 2.85291880e-01 -6.68786585e-01 1.05760741e+00 -1.08657694e+00 -1.41941154e+00 7.62175500e-01 2.27485299e-01 -8.25869024e-01 -3.81422997e-01 -4.20690507e-01 -8.41569543e-01 1.02985001e+00 -2.24211049e+00 -1.00475311e+00 -5.01261055e-01 7.61500835e-01 5.16944587e-01 2.53710240e-01 1.01576269e+00 -1.28842860e-01 -1.30738303e-01 3.37453842e-01 -3.91066104e-01 1.52732715e-01 3.54779124e-01 -1.59771073e+00 5.86405754e-01 7.88415194e-01 5.08379102e-01 8.42656672e-01 7.31587172e-01 -6.85785472e-01 -1.19586182e+00 -1.02450395e+00 1.42650795e+00 -7.88487971e-01 9.87066925e-01 2.61233032e-01 -1.52196717e+00 9.52426374e-01 5.08455396e-01 -3.72718781e-01 6.81533754e-01 4.81425434e-01 -7.60720491e-01 3.05836111e-01 -9.26193178e-01 5.11142612e-01 1.31441951e+00 -4.64685559e-01 -1.64818048e+00 3.80261600e-01 1.16446006e+00 -5.31892359e-01 -1.07119656e+00 1.31207570e-01 1.44772395e-01 -7.55732000e-01 1.01815569e+00 -1.03604567e+00 9.91348505e-01 -3.14298153e-01 1.79183275e-01 -1.53688252e+00 -3.68973762e-01 -5.14312685e-01 -6.51103377e-01 1.09582984e+00 1.15427017e+00 -5.44111729e-01 6.70875549e-01 4.97019500e-01 -1.83967918e-01 -8.18982780e-01 -5.05520701e-01 -5.09767711e-01 2.32614428e-01 -2.54226714e-01 8.81442070e-01 7.20841765e-01 5.65011740e-01 5.36906540e-01 5.86125031e-02 6.37100697e-01 6.44536674e-01 3.65631521e-01 5.17842650e-01 -1.40027463e+00 -5.08646309e-01 -2.14659393e-01 4.87817600e-02 -1.44324791e+00 5.00246882e-01 -1.25439835e+00 -3.12031489e-02 -2.55108714e+00 4.12046164e-02 -3.83030087e-01 -2.83641368e-01 3.94419402e-01 -6.99989915e-01 -3.00448149e-01 2.19210327e-01 -1.50696188e-01 -9.08629656e-01 3.30501735e-01 1.63373721e+00 -2.35811964e-01 -9.65440050e-02 -1.24395236e-01 -9.08801258e-01 5.62012076e-01 6.22800648e-01 -3.62605006e-01 -8.53262305e-01 -7.27092028e-01 7.60395050e-01 6.08793080e-01 4.92953777e-01 -8.35309684e-01 7.90260792e-01 -9.39266607e-02 1.14709340e-01 -5.93730032e-01 2.63841867e-01 -6.26311362e-01 -2.87595779e-01 1.59929823e-02 -7.77800202e-01 1.57900348e-01 2.51528919e-02 8.87877524e-01 -5.23562253e-01 -3.78550261e-01 -6.67191073e-02 -3.66214484e-01 -8.15658927e-01 2.18600273e-01 -1.64107695e-01 4.81861502e-01 7.22232461e-01 1.62249982e-01 -8.38667095e-01 -4.91236180e-01 -9.49224710e-01 8.75357330e-01 4.59449030e-02 3.93870056e-01 8.06712389e-01 -1.10350454e+00 -7.99035966e-01 -2.69269377e-01 3.03572506e-01 2.79532820e-01 3.78076285e-01 6.39180660e-01 -4.76778597e-01 5.52236915e-01 4.76289093e-02 -2.19118074e-01 -1.05872011e+00 5.60539365e-01 1.00745030e-01 -8.13248813e-01 -9.21465218e-01 8.48262548e-01 -4.93878335e-01 -6.67196155e-01 5.02521778e-03 -4.97795641e-01 -8.20894957e-01 -3.52826826e-02 7.36604333e-01 4.42249000e-01 1.03958480e-01 -1.68518975e-01 -5.10903308e-04 5.49963057e-01 -2.22284839e-01 -1.21973261e-01 1.20045781e+00 -2.38854781e-01 -2.80340165e-01 2.96243787e-01 8.41379702e-01 -2.00964957e-02 -6.05478883e-01 -5.20175278e-01 3.34702969e-01 -3.63672197e-01 -4.55121517e-01 -1.25258291e+00 -3.57921213e-01 6.51013732e-01 -6.89906240e-01 3.67350757e-01 9.06296015e-01 2.75260478e-01 1.12413204e+00 9.89988506e-01 4.17486072e-01 -6.84834421e-01 2.09997267e-01 1.00911236e+00 9.61953640e-01 -8.84810150e-01 -1.60335109e-01 -8.11898172e-01 -6.47297800e-01 1.23572266e+00 7.78457582e-01 1.42229065e-01 3.60374480e-01 -3.03993195e-01 -1.98823050e-01 -5.37235141e-01 -1.16802478e+00 -3.25923949e-01 5.91621459e-01 6.38427675e-01 3.34762275e-01 -7.06075430e-02 -4.12361115e-01 6.10348403e-01 -5.28367519e-01 8.68692473e-02 7.13393748e-01 8.25568140e-01 -4.81238693e-01 -1.14848554e+00 -8.74073133e-02 5.31558394e-01 -6.02211505e-02 -2.84414798e-01 -5.35514116e-01 7.26953924e-01 -1.76794782e-01 1.15971398e+00 -1.01578072e-01 -1.77123800e-01 7.04125166e-01 5.45496762e-01 5.30986905e-01 -6.37129784e-01 -7.34951198e-01 -1.13934839e+00 6.80000246e-01 -6.83030844e-01 -3.84669334e-01 -4.39028352e-01 -1.60043252e+00 -3.06504786e-01 -8.27290118e-02 5.69618165e-01 1.58975258e-01 1.06165421e+00 3.39015722e-01 7.91962504e-01 2.40078866e-01 1.36314303e-01 -3.03919673e-01 -7.33023167e-01 -1.64778337e-01 5.44739366e-01 3.58573675e-01 -4.84445840e-01 -1.90127134e-01 -8.51867273e-02]
[10.784125328063965, 7.925074100494385]
728fe9f9-d6ca-4a59-9770-4fd634ba9e80
anisotropic-diffusion-for-details-enhancement
1307.2818
null
http://arxiv.org/abs/1307.2818v1
http://arxiv.org/pdf/1307.2818v1.pdf
Anisotropic Diffusion for Details Enhancement in Multi-Exposure Image Fusion
We develop a multiexposure image fusion method based on texture features, which exploits the edge preserving and intraregion smoothing property of nonlinear diffusion filters based on partial differential equations (PDE). With the captured multiexposure image series, we first decompose images into base layers and detail layers to extract sharp details and fine details, respectively. The magnitude of the gradient of the image intensity is utilized to encourage smoothness at homogeneous regions in preference to inhomogeneous regions. Then, we have considered texture features of the base layer to generate a mask (i.e., decision mask) that guides the fusion of base layers in multiresolution fashion. Finally, well-exposed fused image is obtained that combines fused base layer and the detail layers at each scale across all the input exposures. Proposed algorithm skipping complex High Dynamic Range Image (HDRI) generation and tone mapping steps to produce detail preserving image for display on standard dynamic range display devices. Moreover, our technique is effective for blending flash/no-flash image pair and multifocus images, that is, images focused on different targets.
['Vinay Kumar', 'Harbinder Singh', 'Sunil Bhooshan']
2013-07-10
null
null
null
null
['multi-exposure-image-fusion', 'tone-mapping']
['computer-vision', 'computer-vision']
[ 7.15326488e-01 -4.68359947e-01 2.91661263e-01 -2.22308323e-01 -5.73010921e-01 -6.28389418e-01 2.58781463e-01 -2.90245235e-01 -3.19810718e-01 8.53810966e-01 1.48491815e-01 2.51169980e-01 -1.96002662e-01 -6.90454006e-01 -4.15733844e-01 -1.12067950e+00 4.57881391e-01 -5.87345600e-01 6.22885704e-01 -1.91784531e-01 5.53981900e-01 5.80249608e-01 -1.70419872e+00 5.47433496e-01 1.13904190e+00 1.13731551e+00 5.26011229e-01 8.13021779e-01 6.47315150e-03 6.81575894e-01 -4.20087576e-01 -6.95439875e-02 3.42748046e-01 -4.00034964e-01 -2.61853784e-01 2.90339291e-01 6.95245504e-01 -5.48058391e-01 -3.63648742e-01 1.28549647e+00 4.21341091e-01 3.01419944e-01 7.67883718e-01 -7.43063211e-01 -9.74064291e-01 -7.35399202e-02 -1.21301687e+00 6.82114959e-01 3.05452585e-01 1.45929053e-01 1.63209423e-01 -7.86738217e-01 5.60750961e-01 9.28959846e-01 3.04576188e-01 1.56317055e-01 -1.19950819e+00 -5.86184978e-01 -3.82944457e-02 -1.42742231e-01 -1.15602839e+00 -6.09410167e-01 1.12819886e+00 -2.17462450e-01 5.50522864e-01 4.77392852e-01 5.18720031e-01 3.08789343e-01 9.27389860e-01 1.90382347e-01 1.95070159e+00 -3.49325091e-01 -1.38607815e-01 2.41220415e-01 1.51007816e-01 8.07578802e-01 2.38542140e-01 3.90667111e-01 -4.62729841e-01 -5.03423475e-02 1.18243730e+00 2.12319702e-01 -9.33864117e-01 1.51967794e-01 -9.56317067e-01 1.77961335e-01 1.39536798e-01 4.29295272e-01 -7.26311445e-01 -4.53330994e-01 -3.29146922e-01 2.07015917e-01 3.27277094e-01 1.11043334e-01 2.15038031e-01 3.42674613e-01 -1.04598308e+00 -6.18629903e-02 1.96728647e-01 6.30870640e-01 1.00238705e+00 -5.63054830e-02 -2.95548588e-01 9.16397750e-01 1.65383235e-01 5.82760692e-01 2.50948042e-01 -1.06712675e+00 1.11955829e-01 3.37510139e-01 3.06086272e-01 -9.85171318e-01 4.60283905e-02 -1.44926876e-01 -1.01657474e+00 6.94338441e-01 8.25067163e-02 -2.10332081e-01 -1.03115511e+00 1.46280169e+00 4.27116454e-01 1.76891521e-01 1.54423967e-01 1.22549486e+00 6.14009976e-01 8.32889259e-01 -2.56522000e-01 -7.14140654e-01 1.34873569e+00 -6.15815938e-01 -1.11599934e+00 6.76365895e-03 -3.01708072e-01 -1.04918575e+00 9.01164114e-01 5.25577128e-01 -1.75610471e+00 -7.97785819e-01 -1.29866767e+00 -2.04634205e-01 -1.66092977e-01 -1.84323601e-02 2.75709718e-01 4.45865303e-01 -1.21976101e+00 3.85994166e-01 -3.00055921e-01 1.67953849e-01 2.80175637e-02 3.00611705e-01 -3.84646684e-01 -3.26878458e-01 -9.35533822e-01 6.37839317e-01 4.44235131e-02 -2.78450754e-02 -3.76494378e-01 -8.91034305e-01 -5.30182719e-01 -1.51652902e-01 -6.50360808e-02 -5.85262597e-01 4.90253836e-01 -9.74056423e-01 -1.58358192e+00 8.61016333e-01 -3.10881525e-01 -4.46472690e-02 1.35441795e-01 1.06866762e-01 -7.27412224e-01 6.22044802e-01 -1.63737789e-01 4.45311755e-01 1.35493660e+00 -1.51252878e+00 -8.62652957e-01 -3.39460850e-01 -2.86812991e-01 6.21861279e-01 -1.32057555e-02 3.32840644e-02 -3.97888601e-01 -5.38273454e-01 7.83541650e-02 -3.93334210e-01 2.34131709e-01 1.62743859e-03 -2.51746386e-01 4.11335588e-01 1.23560286e+00 -9.03677940e-01 1.51591957e+00 -2.49995828e+00 2.28051525e-02 7.27542415e-02 4.14654940e-01 5.92123419e-02 -2.61118393e-02 -6.91180304e-02 1.08775739e-02 -2.71176249e-01 -4.68283385e-01 1.28283367e-01 -5.88205934e-01 -3.92015725e-01 -2.45421290e-01 4.95529354e-01 8.58013108e-02 6.64598465e-01 -4.98177677e-01 -6.90879345e-01 4.50227767e-01 7.18222499e-01 -1.47617117e-01 3.20289433e-01 2.52898425e-01 6.21317565e-01 -3.28663200e-01 6.43062115e-01 1.35100126e+00 -3.33581455e-02 -2.88539529e-01 -7.40369797e-01 -6.67963445e-01 -5.62028646e-01 -1.20818484e+00 1.39052975e+00 -4.42632705e-01 5.84565997e-01 4.86819178e-01 -3.27073306e-01 1.15918005e+00 1.03014715e-01 4.38422382e-01 -7.25720048e-01 1.44448981e-01 1.93229869e-01 -2.91652501e-01 -4.34354484e-01 7.12416768e-01 -1.89603746e-01 2.26810887e-01 5.21757483e-01 -3.97970527e-01 -2.79195249e-01 1.72758568e-02 -7.46864527e-02 6.12237155e-01 -1.60202637e-01 -1.04282256e-02 -3.31276119e-01 8.35451722e-01 -3.64012331e-01 1.57906517e-01 5.63069940e-01 -1.47929221e-01 9.54025984e-01 -1.90547239e-02 -1.88881099e-01 -1.04717672e+00 -1.23443878e+00 -1.83407560e-01 6.40929461e-01 7.12692261e-01 3.47252816e-01 -6.90042496e-01 -2.62904819e-02 -3.00774455e-01 2.73843050e-01 -5.49506187e-01 -7.37781972e-02 -5.43440998e-01 -5.68586051e-01 3.48838903e-02 -5.82304411e-02 1.18396473e+00 -8.24283183e-01 -7.20480680e-01 2.02923462e-01 -1.11705281e-01 -8.96154523e-01 -9.60994422e-01 9.97415464e-03 -6.68806612e-01 -1.00059640e+00 -1.19472659e+00 -9.44712818e-01 5.58230102e-01 6.52196705e-01 7.67491519e-01 -1.84004202e-01 -4.35413241e-01 2.40093470e-01 1.33355036e-01 1.96685463e-01 -1.52204156e-01 -6.51003242e-01 -1.02327131e-01 5.46837330e-01 -2.02814102e-01 -6.54414237e-01 -1.03938460e+00 3.63877773e-01 -1.23932540e+00 3.21342409e-01 7.59868443e-01 6.30415320e-01 8.42547417e-01 4.86134529e-01 2.02206478e-01 -4.63280618e-01 8.70523274e-01 2.26123817e-02 -6.59511685e-01 5.55983424e-01 -2.97243625e-01 -1.17126465e-01 4.50890929e-01 -7.16303289e-01 -1.81538165e+00 -3.61300081e-01 4.10282522e-01 -4.67899352e-01 -1.27758488e-01 -1.20420456e-01 -4.85688299e-02 -4.99948710e-01 4.72143978e-01 6.80495858e-01 4.74366955e-02 -2.10880741e-01 2.80011445e-01 9.57884610e-01 8.95067155e-01 -3.45013201e-01 4.79526848e-01 7.75374055e-01 -4.05350812e-02 -9.82343733e-01 -4.15328622e-01 -4.66987230e-02 -2.59960562e-01 -4.57912832e-01 1.27002585e+00 -6.95313811e-01 -7.95764089e-01 7.14846671e-01 -8.69844615e-01 -2.10104644e-01 6.11100160e-02 3.87241781e-01 -3.15686554e-01 3.03882599e-01 -8.83230925e-01 -7.52516568e-01 -4.35007185e-01 -1.08980405e+00 1.08204305e+00 1.03519392e+00 3.58160943e-01 -8.68090212e-01 -1.10382572e-01 1.21257767e-01 7.06881821e-01 3.27353656e-01 9.83392000e-01 5.28300107e-01 -9.81851757e-01 1.71915069e-01 -4.73281801e-01 3.46121460e-01 5.55951834e-01 5.52004948e-02 -9.85048115e-01 -1.95194215e-01 4.26697820e-01 -1.74232237e-02 8.33851457e-01 9.22002494e-01 8.31236422e-01 -1.50146596e-02 -2.41252333e-01 7.81298041e-01 1.77215111e+00 5.46332240e-01 8.23516011e-01 6.34273440e-02 4.63164359e-01 5.86403608e-01 6.56071484e-01 9.93134305e-02 -6.59178663e-03 5.31226456e-01 -3.20298821e-01 -7.49630570e-01 -4.80083793e-01 3.31932940e-02 1.97592929e-01 3.64619493e-01 2.49157604e-02 -3.12219828e-01 -1.79859489e-01 2.62939990e-01 -1.06699216e+00 -8.74751806e-01 -1.12440787e-01 2.16153073e+00 8.15358937e-01 -1.57637253e-01 -2.56309569e-01 -9.75686759e-02 1.15920186e+00 2.12493688e-01 -5.92173040e-01 -3.16050649e-01 -6.52326524e-01 2.24475250e-01 4.65915501e-01 9.19340730e-01 -7.48316407e-01 6.05803668e-01 5.75432968e+00 9.86333489e-01 -1.51544785e+00 -1.46155193e-01 8.91959369e-01 1.16494969e-02 -6.70994282e-01 -1.54703021e-01 -5.78274250e-01 6.84285879e-01 2.73436010e-01 -8.71739089e-02 6.19814932e-01 -2.02199742e-01 2.36506209e-01 -7.20856369e-01 -4.38444614e-01 1.21227217e+00 5.05327694e-02 -1.13118756e+00 -2.41855308e-02 5.87504953e-02 9.44997907e-01 -5.54444551e-01 6.97205305e-01 -4.10353780e-01 -8.21073726e-02 -7.96942830e-01 3.60772163e-01 1.03745615e+00 9.92794275e-01 -8.10086668e-01 1.79903299e-01 1.22448960e-02 -1.26700187e+00 -6.34677187e-02 -2.98178822e-01 4.35567170e-01 3.31056833e-01 6.86286271e-01 1.41003564e-01 6.41491473e-01 7.59665370e-01 4.53007847e-01 -3.10441673e-01 6.21924758e-01 2.91293800e-01 -6.17536344e-02 -3.20201397e-01 5.28967679e-01 -7.26320744e-02 -7.57078648e-01 5.81954956e-01 8.44964623e-01 5.31766295e-01 6.38969898e-01 -1.62997022e-01 9.03557599e-01 3.54105324e-01 -2.21830890e-01 -6.37381792e-01 2.33728305e-01 4.63276297e-01 1.40580142e+00 -6.71411455e-01 -4.53228533e-01 -4.89833891e-01 1.41144621e+00 -4.35004905e-02 8.72285903e-01 -6.74797714e-01 -6.10026777e-01 3.31285805e-01 1.60278127e-01 2.94540107e-01 4.31474037e-02 -3.13938618e-01 -1.08591926e+00 1.95779763e-02 -7.60543406e-01 3.30952466e-01 -1.21091712e+00 -9.93077338e-01 7.94502020e-01 -1.21298619e-01 -1.12366915e+00 2.97348231e-01 -1.21704206e-01 -7.05172598e-01 1.46957135e+00 -1.75513661e+00 -9.11554217e-01 -5.27565241e-01 9.65058386e-01 2.54150212e-01 2.48092398e-01 1.09618157e-01 2.25649878e-01 -3.37232798e-01 2.88062394e-01 1.34577647e-01 -5.41229129e-01 9.26426411e-01 -8.48751485e-01 -2.99014688e-01 1.00506949e+00 -6.04060233e-01 5.36179245e-01 4.14431870e-01 -8.62678289e-01 -1.18180931e+00 -6.20967031e-01 3.58509660e-01 2.31118337e-03 8.44673514e-02 -1.64566878e-02 -1.12106884e+00 1.32111877e-01 5.88583767e-01 -1.93827257e-01 3.07955176e-01 -9.26713467e-01 1.82387456e-01 -2.75333166e-01 -1.51309586e+00 5.99037945e-01 5.75576484e-01 -5.07321775e-01 -1.97864622e-01 -2.63632149e-01 5.88802636e-01 -6.42016947e-01 -1.10884368e+00 5.59225678e-01 7.33131349e-01 -1.45914042e+00 8.59959543e-01 2.98052311e-01 4.75233227e-01 -7.11673796e-01 -2.00240523e-01 -9.06176329e-01 -5.41318953e-01 -9.20736313e-01 8.04985762e-02 1.39869225e+00 2.23839417e-01 -8.44774604e-01 3.81441504e-01 5.67590177e-01 -1.38680398e-01 -6.17327273e-01 -4.09650296e-01 -1.15278870e-01 -4.11795825e-01 5.12135148e-01 4.87943351e-01 7.85669267e-01 -5.02630413e-01 -4.24992293e-02 -4.99596626e-01 4.49729353e-01 1.01140022e+00 4.47080582e-01 2.47984514e-01 -7.23582923e-01 -3.56762826e-01 -2.49980286e-01 1.06569808e-02 -9.99363124e-01 -2.85990000e-01 -4.50863123e-01 -1.35780260e-01 -1.18062055e+00 2.72860080e-01 -3.79901499e-01 -3.68317395e-01 -2.26389125e-01 -4.87577498e-01 6.51270449e-01 4.04583886e-02 4.57476974e-01 -4.94856983e-02 1.97412059e-01 1.75648415e+00 6.60417750e-02 -4.96215075e-01 -2.56360590e-01 -7.61476338e-01 3.28451216e-01 4.88778025e-01 8.96637663e-02 -4.71463352e-01 -3.62261087e-01 -3.86669904e-01 6.24199271e-01 3.46850663e-01 -9.61358070e-01 3.25553775e-01 -1.93772569e-01 9.15363133e-01 -5.26163757e-01 4.31208342e-01 -6.87239587e-01 4.88652945e-01 1.30036771e-01 -2.96278805e-01 -4.41593453e-02 1.91102877e-01 5.45996666e-01 -3.25965524e-01 1.48074478e-01 1.23686266e+00 4.50364277e-02 -6.00095093e-01 3.50369781e-01 -2.38637537e-01 -2.33059302e-01 1.08473015e+00 -6.75023019e-01 -5.32578945e-01 -3.00669581e-01 -5.18743575e-01 -1.87800288e-01 8.85742784e-01 1.04839236e-01 9.45257783e-01 -1.17130649e+00 -6.20487869e-01 6.97841108e-01 -2.72510797e-01 -3.62064779e-01 1.08654451e+00 1.08491349e+00 -5.55908680e-01 -6.56271577e-02 -5.00398636e-01 -5.20000517e-01 -1.40518332e+00 5.40273845e-01 5.39185882e-01 -5.72679639e-02 -7.64102817e-01 7.10680366e-01 7.49573708e-01 5.72341979e-01 -1.31952688e-01 -2.41773218e-01 -3.02512825e-01 -3.54116410e-01 9.15670931e-01 3.38201791e-01 -2.60497779e-01 -5.63083231e-01 -5.40478863e-02 1.28206885e+00 -2.52744973e-01 -4.73521471e-01 1.03287446e+00 -7.54956663e-01 -1.72794193e-01 2.90348917e-01 1.21623683e+00 4.12833214e-01 -1.64072609e+00 -7.96858892e-02 -7.66946733e-01 -7.90787458e-01 5.14806986e-01 -7.97989249e-01 -1.02298963e+00 7.30786622e-01 1.10448849e+00 1.54393271e-01 1.78437710e+00 -2.57959753e-01 8.17098618e-01 -4.35123414e-01 -3.11247073e-02 -8.43179226e-01 7.79590085e-02 -1.98944375e-01 7.28663206e-01 -9.20739174e-01 -2.78521534e-02 -4.07846540e-01 -6.63733602e-01 1.06778789e+00 3.94546390e-01 -9.61338654e-02 5.10534286e-01 5.61890006e-01 1.20054036e-01 -1.88358560e-01 -5.60494840e-01 -7.86559470e-03 4.64102834e-01 6.48794889e-01 2.45002657e-01 -3.41864079e-01 -1.69242963e-01 1.34047300e-01 3.28858167e-01 9.15027857e-02 5.23608327e-01 7.78451204e-01 -7.73869097e-01 -5.45820951e-01 -6.82907701e-01 3.94798130e-01 -5.39512873e-01 -2.21325845e-01 -5.48967943e-02 4.33004647e-01 1.52627215e-01 1.05389750e+00 1.91792727e-01 -1.32326767e-01 1.52458772e-01 -3.28318447e-01 8.34135473e-01 3.01900450e-02 -3.51931214e-01 4.86267298e-01 -3.09711277e-01 -3.23067635e-01 -4.42373753e-01 -2.54634291e-01 -9.31011677e-01 -2.24643052e-01 -2.53479153e-01 4.44170870e-02 4.67791826e-01 4.97783989e-01 2.54358560e-01 4.09962237e-01 8.30459177e-01 -9.33547974e-01 1.52923688e-01 -7.68179774e-01 -1.04503000e+00 3.76581669e-01 7.38285363e-01 -4.52403784e-01 -6.45943522e-01 1.78719416e-01]
[10.919646263122559, -2.4125216007232666]
04a5062c-746a-4292-8463-e3c5577c1f33
autonomous-vision-based-rapid-aerial-grasping
2211.13093
null
https://arxiv.org/abs/2211.13093v2
https://arxiv.org/pdf/2211.13093v2.pdf
Autonomous Marker-less Rapid Aerial Grasping
In a future with autonomous robots, visual and spatial perception is of utmost importance for robotic systems. Particularly for aerial robotics, there are many applications where utilizing visual perception is necessary for any real-world scenarios. Robotic aerial grasping using drones promises fast pick-and-place solutions with a large increase in mobility over other robotic solutions. Utilizing Mask R-CNN scene segmentation (detectron2), we propose a vision-based system for autonomous rapid aerial grasping which does not rely on markers for object localization and does not require the appearence of the object to be previously known. Combining segmented images with spatial information from a depth camera, we generate a dense point cloud of the detected objects and perform geometry-based grasp planning to determine grasping points on the objects. In real-world experiments on a dynamically grasping aerial platform, we show that our system can replicate the performance of a motion capture system for object localization up to 94.5% of the baseline grasping success rate. With our results, we show the first use of geometry-based grasping techniques with a flying platform and aim to increase the autonomy of existing aerial manipulation platforms, bringing them further towards real-world applications in warehouses and similar environments.
['Robert K. Katzschmann', 'Barnabas Gavin Cangan', 'Erik Bauer']
2022-11-23
null
null
null
null
['scene-segmentation']
['computer-vision']
[ 1.88439656e-02 -1.87418148e-01 3.34163725e-01 -2.88864374e-01 -8.82700160e-02 -1.17207336e+00 1.57534644e-01 -3.35464552e-02 -3.67693990e-01 1.87303841e-01 -6.25276804e-01 -1.39420062e-01 -3.14366430e-01 -8.11855078e-01 -8.91987562e-01 -4.33655381e-01 -5.21200836e-01 6.45700991e-01 6.85426235e-01 -6.43243849e-01 3.20620090e-01 9.81697738e-01 -1.86583662e+00 3.34270060e-01 5.17948627e-01 1.23160267e+00 1.12257731e+00 9.69051301e-01 3.67045045e-01 1.45087093e-01 -5.05320072e-01 3.73787880e-01 9.53673840e-01 3.47213835e-01 -5.91189682e-01 2.73507744e-01 3.60144883e-01 -1.05920899e+00 -2.23318458e-01 6.70371056e-01 9.52958167e-02 -6.60993233e-02 3.60109299e-01 -1.48620093e+00 -2.71522492e-01 4.23931599e-01 -3.97122294e-01 -2.65698612e-01 4.19502169e-01 5.10627925e-01 4.95431721e-01 -5.28034031e-01 4.37897056e-01 1.23150206e+00 3.27537209e-01 4.94753718e-01 -4.96908069e-01 -1.98201954e-01 2.77296215e-01 -2.75321007e-01 -1.05326223e+00 -9.93651375e-02 3.95223141e-01 -3.88825476e-01 7.69333780e-01 -4.21416461e-02 6.98843837e-01 6.92428768e-01 9.85626280e-02 6.86791778e-01 5.54720342e-01 -5.14942586e-01 2.75864840e-01 -3.21268350e-01 -2.60200113e-01 1.03796184e+00 4.45042431e-01 -5.21158352e-02 -2.74409242e-02 1.75099015e-01 1.12954056e+00 5.77999473e-01 -3.79115999e-01 -9.73625124e-01 -1.85333633e+00 3.03047031e-01 1.03902900e+00 1.67517602e-01 -5.85460842e-01 5.92125237e-01 -2.15620324e-01 -4.84903269e-02 -1.81610137e-01 6.76240742e-01 -5.22654176e-01 1.95854247e-01 -4.75730002e-01 3.29408437e-01 6.55901670e-01 1.77331388e+00 5.20209491e-01 -1.19575061e-01 2.63262510e-01 2.18062654e-01 2.62198091e-01 9.73805845e-01 1.66777208e-01 -1.35149992e+00 5.33047676e-01 9.32959914e-01 8.85835886e-01 -8.81919801e-01 -5.91936350e-01 2.45794952e-01 -1.16781123e-01 6.40717685e-01 5.68088472e-01 -1.67968154e-01 -1.16543269e+00 9.72933948e-01 4.01771307e-01 -5.47742367e-01 9.10388771e-03 1.26237428e+00 4.05203074e-01 5.03528714e-01 -4.05209124e-01 5.10263324e-01 1.26524961e+00 -7.67458677e-01 -1.35184169e-01 -3.28357816e-01 3.62728685e-01 -5.05770326e-01 8.70537281e-01 7.09652066e-01 -7.95785964e-01 -4.32956100e-01 -1.25690401e+00 2.57410761e-02 -4.29204792e-01 8.78939807e-01 9.28795636e-01 6.45800903e-02 -1.11271632e+00 6.18241906e-01 -1.24254644e+00 -8.46749187e-01 3.58481467e-01 7.17026234e-01 -4.44007546e-01 -3.66606951e-01 -2.45888144e-01 8.46485376e-01 5.09244740e-01 2.59186268e-01 -1.20587075e+00 -1.09938405e-01 -5.62013865e-01 -5.15646413e-02 5.08090913e-01 -4.21976328e-01 1.30816412e+00 -4.11813140e-01 -1.41478384e+00 5.63362718e-01 3.53967190e-01 -2.73155540e-01 3.95472795e-01 -5.87470055e-01 4.40188885e-01 8.12142432e-01 1.20301902e-01 1.25545883e+00 1.04432034e+00 -1.64589465e+00 -9.17275310e-01 -4.87205744e-01 5.93360484e-01 2.44855672e-01 -8.91340747e-02 -4.51762825e-01 -2.97679424e-01 5.47004417e-02 6.16611183e-01 -1.26895809e+00 -2.51270175e-01 8.51214468e-01 -2.72768512e-02 -1.73785329e-01 1.40568328e+00 -2.26712555e-01 -2.55708732e-02 -1.95878112e+00 2.98044056e-01 -5.11863008e-02 -1.50590315e-01 3.67722452e-01 -1.44194081e-01 7.67100036e-01 6.98198974e-01 -3.36451352e-01 -2.33795196e-01 9.60207358e-03 -1.46243319e-01 1.97062209e-01 -7.07608700e-01 3.91354948e-01 4.71104011e-02 8.32098365e-01 -1.16900790e+00 -2.88661178e-02 5.54080427e-01 1.38264835e-01 -3.99288327e-01 3.40661675e-01 -7.27184236e-01 3.43507558e-01 -6.47861123e-01 1.28507006e+00 6.16461396e-01 4.75307330e-02 2.40660310e-01 -4.05779444e-02 -6.28343225e-01 -3.52609396e-01 -9.14282382e-01 1.98255193e+00 -4.03977871e-01 3.26982647e-01 6.55143678e-01 -5.29179573e-01 8.41491818e-01 -4.02502976e-02 4.41460341e-01 2.16333419e-01 1.97642207e-01 4.10434306e-01 -1.28150731e-01 -7.09488809e-01 6.79706693e-01 4.97502327e-01 1.04222707e-02 -2.52918843e-02 5.01884744e-02 -9.31208432e-01 1.20117776e-01 6.66711181e-02 1.15717816e+00 6.55277491e-01 -1.73242673e-01 -3.40532482e-01 -1.89240098e-01 7.56230652e-01 -7.70351142e-02 6.22719109e-01 -5.42145073e-02 7.54905581e-01 -1.53342322e-01 -4.51326698e-01 -9.76183355e-01 -9.76128459e-01 1.10962979e-01 6.56385958e-01 8.58406663e-01 3.38975340e-02 -7.23882437e-01 -4.30197567e-01 5.12238503e-01 2.91182011e-01 -3.01469982e-01 2.84730107e-01 -5.06132185e-01 -1.18722945e-01 1.10517442e-01 7.20214963e-01 7.06578672e-01 -1.12202334e+00 -1.88791251e+00 1.12564243e-01 -7.07966760e-02 -1.41800082e+00 2.20098346e-01 2.37689421e-01 -9.31682110e-01 -1.34688032e+00 -9.10231650e-01 -8.02649617e-01 1.01307440e+00 1.22918701e+00 4.64255989e-01 2.93662906e-01 -7.25003421e-01 1.10823858e+00 -8.38503003e-01 -6.55613661e-01 -3.73415975e-03 3.98814082e-02 3.43512595e-01 -6.32471323e-01 -3.37102972e-02 -1.68930426e-01 -9.25030529e-01 5.10842323e-01 -6.89638793e-01 -3.43777984e-01 8.59604001e-01 3.42918843e-01 1.80366829e-01 9.18983761e-03 1.78997666e-01 4.66078579e-01 1.61885247e-01 -2.57073015e-01 -1.05634034e+00 2.18521714e-01 6.82317093e-02 -3.84961396e-01 3.24446112e-01 -4.79095012e-01 -4.19462472e-01 6.25150681e-01 7.10611463e-01 -7.91138470e-01 -2.45403096e-01 6.51338175e-02 1.25467300e-01 -3.08303535e-01 5.65250516e-01 -4.80592288e-02 2.26909399e-01 -4.99707721e-02 3.65152925e-01 9.23299015e-01 4.68520015e-01 -4.50686991e-01 7.11075306e-01 1.03821385e+00 2.09858507e-01 -1.10480297e+00 -4.08721030e-01 -5.67968190e-01 -1.35892475e+00 -3.16525578e-01 8.82344782e-01 -9.01842296e-01 -1.21140206e+00 4.10301208e-01 -1.43184185e+00 -5.67540824e-01 -4.13268246e-03 5.81314325e-01 -7.04427302e-01 2.97942430e-01 -1.32394791e-01 -9.23933148e-01 -2.26948842e-01 -1.23927045e+00 1.71244299e+00 2.78603911e-01 2.67024308e-01 -1.84299722e-01 -7.63214946e-01 1.62198439e-01 2.44851291e-01 1.90352112e-01 3.29539150e-01 -1.40420068e-03 -1.39486241e+00 -3.37182194e-01 -2.01168671e-01 1.02688022e-01 3.19596112e-01 1.94321778e-02 -5.04862607e-01 -4.97045547e-01 -4.20424283e-01 -3.57694447e-01 6.68948531e-01 3.46024156e-01 8.30891252e-01 8.42159986e-02 -8.78105998e-01 1.74651757e-01 1.37609768e+00 5.03621876e-01 2.84250826e-01 2.30364755e-01 5.75055301e-01 6.32315636e-01 1.31177771e+00 5.55619419e-01 1.98652104e-01 6.71590507e-01 1.34210324e+00 2.20862359e-01 9.80041251e-02 -1.00085534e-01 2.75935203e-01 -8.91681481e-03 -3.81149083e-01 -3.86730224e-01 -1.02238321e+00 7.94231951e-01 -1.82204521e+00 -4.92031693e-01 -2.43246675e-01 2.33038878e+00 8.00360590e-02 -2.81578571e-01 -8.50833207e-02 -1.24295041e-01 5.49885988e-01 -4.57321614e-01 -5.85280359e-01 1.33315876e-01 4.68153328e-01 -1.41123846e-01 9.81006861e-01 2.70364821e-01 -1.21485639e+00 1.36660147e+00 5.48975229e+00 -1.38100445e-01 -1.11127973e+00 -4.53794986e-01 -5.80792129e-01 8.08955804e-02 5.13537407e-01 8.06645155e-02 -7.33548462e-01 9.23143923e-02 1.49467096e-01 6.06278121e-01 6.47212625e-01 1.38004220e+00 -1.16301589e-01 -6.46022737e-01 -1.19604254e+00 7.45829582e-01 4.86679897e-02 -7.80774415e-01 -8.66190195e-02 1.33023456e-01 2.63352126e-01 2.32265562e-01 -1.30843341e-01 -2.20833942e-02 4.66599375e-01 -6.27078831e-01 9.75701630e-01 4.97876197e-01 5.84383667e-01 -3.70962620e-01 6.38702631e-01 7.39597678e-01 -1.13376081e+00 -6.73445940e-01 -8.02000165e-01 -2.18816996e-01 1.52087003e-01 2.18777478e-01 -1.45212448e+00 3.59036952e-01 9.51067924e-01 3.07028502e-01 -1.54806763e-01 1.08296943e+00 -1.99805647e-01 -1.52522996e-01 -5.96413374e-01 -4.43090498e-01 4.52284425e-01 -1.15978859e-01 6.41340852e-01 4.63560075e-01 7.12198555e-01 3.86765003e-01 6.02486372e-01 6.72407925e-01 1.93310797e-01 -5.12588918e-01 -1.07713687e+00 -8.85907188e-02 4.35239106e-01 1.32641625e+00 -1.13270056e+00 -1.19717382e-01 1.49769381e-01 1.25267291e+00 8.68633687e-02 -6.27409667e-02 -5.05996525e-01 -7.58802772e-01 4.48524147e-01 2.33694851e-01 5.80399454e-01 -1.25281549e+00 3.58913362e-01 -9.75585043e-01 2.52368599e-01 -4.14338559e-01 -5.24468064e-01 -1.40917253e+00 -5.36729038e-01 5.25223196e-01 1.60014614e-01 -1.53812158e+00 1.70533615e-03 -1.31215274e+00 -4.07130904e-02 4.65893388e-01 -1.40378988e+00 -1.51929176e+00 -1.11951244e+00 2.25424334e-01 7.97819912e-01 2.36651246e-02 8.55140507e-01 -4.81799781e-01 3.23464304e-01 -4.18883383e-01 -2.22442910e-01 8.23539158e-04 5.43609142e-01 -1.07014823e+00 1.48336619e-01 6.94472015e-01 -8.06735978e-02 6.99696839e-01 3.72723490e-01 -9.27885473e-01 -2.08315325e+00 -9.25156236e-01 -1.84542134e-01 -8.83232057e-01 2.83653349e-01 -5.62098920e-01 -2.64120102e-01 8.44845116e-01 6.08512387e-02 1.49751930e-02 -1.93643361e-01 -5.66023111e-01 2.75899261e-01 -1.62674591e-01 -1.43859720e+00 5.26231885e-01 1.18677163e+00 3.75731349e-01 -5.59109390e-01 7.64215887e-01 7.83820033e-01 -8.46726537e-01 -5.69374681e-01 8.29526126e-01 8.21830451e-01 -6.42468393e-01 9.89382684e-01 -3.35199833e-01 2.26857707e-01 -7.09597528e-01 -5.39847910e-01 -1.14075756e+00 -1.79563954e-01 -4.84105855e-01 2.14882404e-01 6.31844521e-01 -7.24432291e-03 -3.67574781e-01 6.81749284e-01 4.35595542e-01 -4.10435259e-01 -3.16079378e-01 -6.87966883e-01 -9.58992600e-01 -4.16377604e-01 1.26616180e-01 5.24148345e-01 4.10255373e-01 1.62334159e-01 -3.38150173e-01 5.84483258e-02 1.06125629e+00 4.48049366e-01 5.14293969e-01 1.08281565e+00 -1.36997700e+00 1.67907059e-01 9.15592909e-02 -5.97286105e-01 -1.36660635e+00 -1.18720882e-01 -6.78239405e-01 7.05836535e-01 -2.05343866e+00 -4.27521497e-01 -9.01111305e-01 4.73829716e-01 8.26133430e-01 4.27113086e-01 -2.86780614e-02 5.09899735e-01 4.58484441e-01 -5.34429669e-01 4.30579573e-01 1.19084895e+00 -1.29535064e-01 -2.59593248e-01 1.27829507e-01 -9.23607200e-02 6.03488743e-01 7.81993449e-01 -7.09154904e-02 -4.26652312e-01 -8.38519037e-01 -2.73356527e-01 2.23866329e-01 7.52012908e-01 -1.55214548e+00 3.22639525e-01 -2.29147539e-01 4.63207990e-01 -6.68833554e-01 8.55155766e-01 -1.46024013e+00 -4.66093779e-01 8.47001076e-01 1.46731734e-01 2.18753278e-01 3.19033563e-01 6.89694345e-01 4.29041535e-01 -5.67279518e-01 3.22027028e-01 -6.66631162e-01 -8.48644376e-01 1.22511737e-01 -5.56003511e-01 -7.34234035e-01 1.71442974e+00 -1.25360489e-01 -4.49366510e-01 -1.88017473e-01 -4.62805241e-01 5.17050564e-01 7.75044441e-01 5.31141281e-01 9.38397646e-01 -6.40537977e-01 -2.75454015e-01 1.69594541e-01 8.10483322e-02 6.02636456e-01 -1.75530761e-01 3.96750063e-01 -1.11890650e+00 6.54553235e-01 -3.67883891e-01 -1.14964318e+00 -1.08359838e+00 6.67529643e-01 1.35715744e-02 6.55620992e-01 -7.57694066e-01 7.00782478e-01 -2.34008089e-01 -4.71635610e-01 1.45054281e-01 -1.02739918e+00 2.09541142e-01 -4.72119242e-01 2.87076384e-01 3.13207358e-01 -4.46449295e-02 -7.96099901e-02 -4.29618388e-01 7.08600104e-01 2.26181582e-01 -1.01888172e-01 1.41114020e+00 4.66529205e-02 -1.56720251e-01 9.61477906e-02 3.30093890e-01 -1.08847737e-01 -1.73518431e+00 4.13858533e-01 -3.31356585e-01 -5.35536468e-01 -1.24415092e-01 -9.36428368e-01 -7.04572797e-01 8.73652279e-01 5.44829845e-01 2.25388661e-01 7.25867271e-01 1.36469454e-01 4.79118109e-01 1.22169304e+00 1.54777288e+00 -8.78867865e-01 2.71994352e-01 4.71622646e-01 1.24702108e+00 -1.51869798e+00 1.32442921e-01 -6.70842171e-01 -5.26123762e-01 1.49245226e+00 7.04372048e-01 -4.58197534e-01 3.20243299e-01 2.26802438e-01 3.31534259e-02 -3.20885837e-01 -2.22489476e-01 -2.75755942e-01 -1.63461566e-01 9.18170214e-01 -5.17183363e-01 1.08435951e-01 3.28192323e-01 8.18170086e-02 -2.50655450e-02 3.23201006e-04 6.83141172e-01 1.62578511e+00 -9.69704270e-01 -7.73637056e-01 -4.09989327e-01 9.88612510e-03 1.44270286e-01 2.12197393e-01 -4.76408035e-01 1.09345472e+00 -2.30267495e-02 1.12924898e+00 8.74715596e-02 -3.25759292e-01 4.93712366e-01 -2.07588434e-01 9.59220767e-01 -8.38481009e-01 -1.93692386e-01 -4.17373419e-01 -2.67159909e-01 -7.66042888e-01 -3.82578105e-01 -5.68721116e-01 -1.58817744e+00 4.40418273e-01 -5.22087812e-01 -3.11529994e-01 1.53959310e+00 6.61830485e-01 7.34311700e-01 2.35455945e-01 4.49013591e-01 -1.70748448e+00 -7.16725349e-01 -7.78864622e-01 -3.86577129e-01 -3.08367908e-01 3.12653124e-01 -9.13144588e-01 -1.79421917e-01 -1.22299474e-02]
[5.776187896728516, -0.8430099487304688]
063dcac9-9e8d-4ed7-ada8-ffafe09f5a92
unprocessing-images-for-learned-raw-denoising
1811.11127
null
http://arxiv.org/abs/1811.11127v1
http://arxiv.org/pdf/1811.11127v1.pdf
Unprocessing Images for Learned Raw Denoising
Machine learning techniques work best when the data used for training resembles the data used for evaluation. This holds true for learned single-image denoising algorithms, which are applied to real raw camera sensor readings but, due to practical constraints, are often trained on synthetic image data. Though it is understood that generalizing from synthetic to real data requires careful consideration of the noise properties of image sensors, the other aspects of a camera's image processing pipeline (gain, color correction, tone mapping, etc) are often overlooked, despite their significant effect on how raw measurements are transformed into finished images. To address this, we present a technique to "unprocess" images by inverting each step of an image processing pipeline, thereby allowing us to synthesize realistic raw sensor measurements from commonly available internet photos. We additionally model the relevant components of an image processing pipeline when evaluating our loss function, which allows training to be aware of all relevant photometric processing that will occur after denoising. By processing and unprocessing model outputs and training data in this way, we are able to train a simple convolutional neural network that has 14%-38% lower error rates and is 9x-18x faster than the previous state of the art on the Darmstadt Noise Dataset, and generalizes to sensors outside of that dataset as well.
['Jonathan T. Barron', 'Ben Mildenhall', 'Dillon Sharlet', 'Tim Brooks', 'Jiawen Chen', 'Tianfan Xue']
2018-11-27
unprocessing-images-for-learned-raw-denoising-1
http://openaccess.thecvf.com/content_CVPR_2019/html/Brooks_Unprocessing_Images_for_Learned_Raw_Denoising_CVPR_2019_paper.html
http://openaccess.thecvf.com/content_CVPR_2019/papers/Brooks_Unprocessing_Images_for_Learned_Raw_Denoising_CVPR_2019_paper.pdf
cvpr-2019-6
['tone-mapping']
['computer-vision']
[ 8.42851520e-01 -1.64976269e-01 6.81316137e-01 -6.24334455e-01 -7.97636211e-01 -6.67564511e-01 4.55067337e-01 1.12861529e-01 -8.41972649e-01 3.15796643e-01 -5.95443249e-02 -3.12364668e-01 1.04325645e-01 -9.65475440e-01 -1.20399094e+00 -7.70245135e-01 4.38562632e-02 1.51430834e-02 1.75306529e-01 -2.76740223e-01 7.64054060e-03 5.92842579e-01 -1.66109228e+00 2.78360695e-01 3.49931747e-01 1.21285546e+00 -2.16369913e-03 1.15071464e+00 4.85678352e-02 6.92810893e-01 -8.00628781e-01 -4.47625965e-01 6.59272611e-01 -2.57845968e-01 -3.47508997e-01 3.99249315e-01 7.66117632e-01 -6.33075416e-01 -3.15135747e-01 1.27435994e+00 4.26126957e-01 -1.40566364e-01 2.68494934e-01 -9.70063150e-01 -1.87389359e-01 4.53692168e-01 -2.68338025e-01 -5.77583723e-02 8.17770138e-02 3.66416812e-01 3.53835076e-01 -3.92126888e-01 2.49075204e-01 1.11198187e+00 8.82829428e-01 3.38278145e-01 -1.52509236e+00 -6.44780457e-01 -1.78874940e-01 -1.78755999e-01 -1.07295585e+00 -4.67236400e-01 5.49141586e-01 -3.23182046e-01 9.24929917e-01 1.69675976e-01 5.27275264e-01 1.23430288e+00 1.00947827e-01 3.31751436e-01 1.13421071e+00 -4.37394977e-01 2.95556068e-01 1.11462153e-01 -9.66609195e-02 2.15217173e-01 2.20534801e-01 3.03188562e-01 -1.83707923e-01 9.97038931e-02 7.07482040e-01 -2.89387435e-01 -3.63858253e-01 -1.44479185e-01 -1.22172987e+00 5.79095125e-01 4.25234646e-01 -2.30023824e-02 -3.85642409e-01 5.94602406e-01 5.16469359e-01 6.93662763e-01 2.50700206e-01 3.32349032e-01 -4.88984793e-01 -1.27278283e-01 -1.08051622e+00 1.79028943e-01 7.90445864e-01 8.63771737e-01 1.10118103e+00 2.32323453e-01 2.77841538e-01 5.03901303e-01 -1.27343321e-02 7.42639422e-01 3.62875402e-01 -1.46432018e+00 3.96595478e-01 2.35461235e-01 1.58256263e-01 -7.12063372e-01 -2.28777781e-01 -2.18820363e-01 -9.87682045e-01 8.16771507e-01 5.57204843e-01 -4.39180821e-01 -1.10897994e+00 1.38122547e+00 -3.96482758e-02 1.41209349e-01 2.42651880e-01 7.32500255e-01 3.33562344e-01 5.98936021e-01 -4.68608700e-02 5.86685315e-02 1.08019149e+00 -3.86666387e-01 -4.62665856e-01 -4.57214832e-01 3.23474199e-01 -9.48568940e-01 1.07220328e+00 9.47320282e-01 -1.01439047e+00 -7.60910690e-01 -1.45488358e+00 1.00291604e-02 -4.22203094e-01 6.63047731e-02 4.44963008e-01 9.15553033e-01 -1.13920188e+00 9.81168926e-01 -9.51103926e-01 -4.48828131e-01 1.65608570e-01 5.07147551e-01 -4.57088977e-01 -2.42650345e-01 -8.42825472e-01 9.21181560e-01 4.48725939e-01 9.24226120e-02 -8.72361660e-01 -6.71458900e-01 -9.32160735e-01 7.17505291e-02 2.96781987e-01 -4.03667927e-01 1.44277227e+00 -1.46858132e+00 -1.55483544e+00 6.59740150e-01 1.62509799e-01 -6.26019895e-01 6.25207722e-01 -1.96390599e-01 -4.07650977e-01 1.01903640e-02 -2.74766982e-01 6.27170503e-01 1.22660339e+00 -1.28109741e+00 -4.70751822e-01 -1.16964370e-01 1.27115488e-01 -8.18667710e-02 -8.75835866e-02 -1.01902664e-01 -6.00599051e-01 -6.08816445e-01 2.49982905e-02 -9.00460482e-01 -4.24263716e-01 1.26584679e-01 -1.86467186e-01 7.44831979e-01 7.91728199e-01 -6.73217177e-01 4.70618367e-01 -2.40165997e+00 -1.97556973e-01 3.94916266e-01 -2.40501538e-01 2.64614642e-01 -2.15358198e-01 3.32272202e-01 -4.43535417e-01 -9.83141810e-02 -4.08617049e-01 -5.42533815e-01 -1.62657499e-01 4.14265543e-01 -3.25338393e-01 3.87161285e-01 3.13737959e-01 3.88378739e-01 -5.76278925e-01 1.89746525e-02 6.52371287e-01 6.27523541e-01 -4.11480546e-01 2.94779211e-01 -2.18005404e-01 4.71074671e-01 1.38725966e-01 3.66639823e-01 8.61867249e-01 1.58730581e-01 3.68371829e-02 -5.56014955e-01 2.43352930e-04 2.21315352e-03 -1.54867041e+00 1.58345604e+00 -6.64003789e-01 8.22363615e-01 4.61590230e-01 -1.04808879e+00 7.31782436e-01 1.58449739e-01 3.66749406e-01 -8.02958786e-01 2.54991025e-01 1.45224318e-01 -2.02155590e-01 -5.27696431e-01 3.64786536e-01 -1.39816746e-01 6.89471886e-02 2.03382075e-01 1.65558103e-02 -6.72607720e-01 2.34566450e-01 -4.08162326e-02 1.37065959e+00 2.35420704e-01 -1.68448046e-01 -1.26330912e-01 2.74555027e-01 1.54358998e-01 2.60125190e-01 8.41174364e-01 1.63521335e-01 1.02088141e+00 4.22862351e-01 -4.69147325e-01 -1.44812620e+00 -1.13008952e+00 -3.27172466e-02 6.02101743e-01 -1.07560851e-01 -2.21240103e-01 -1.05975389e+00 -1.34171411e-01 -1.47999570e-01 7.60593653e-01 -5.02153397e-01 -1.31167784e-01 -5.12832582e-01 -1.03427482e+00 6.54671907e-01 5.00595272e-01 7.14009106e-01 -7.93622732e-01 -7.79970765e-01 3.17475677e-01 2.47893646e-01 -1.35287309e+00 9.08621680e-03 5.58120310e-01 -8.29787552e-01 -1.16651118e+00 -3.07675302e-01 -3.35393608e-01 6.80054307e-01 9.50074494e-02 1.28881836e+00 1.28894717e-01 -4.50648755e-01 5.48445940e-01 -2.66572028e-01 -6.81769490e-01 -7.84016907e-01 -3.57217461e-01 -1.95030510e-01 2.27180328e-02 2.28467524e-01 -6.56801641e-01 -5.63849807e-01 -4.22115140e-02 -1.55793619e+00 -6.41936064e-02 6.91348851e-01 6.73355699e-01 5.55015564e-01 5.19690037e-01 -3.38867366e-01 -9.34340179e-01 5.12435794e-01 1.19315265e-02 -1.17038321e+00 5.19250371e-02 -4.20002759e-01 2.01862752e-01 7.82685161e-01 -4.05364037e-01 -9.61958408e-01 5.98091245e-01 -4.11660284e-01 -4.02371258e-01 -4.61157918e-01 2.07684681e-01 -2.44673833e-01 -2.68392891e-01 7.77332544e-01 3.41672190e-02 2.05318257e-01 -3.63632679e-01 2.29027227e-01 5.69887459e-01 8.25686514e-01 -2.93298990e-01 1.07242084e+00 7.08770275e-01 1.27133876e-01 -9.63968575e-01 -5.74100912e-01 -2.20393836e-01 -3.92246217e-01 -5.72150722e-02 8.61490667e-01 -1.12021661e+00 -7.16622472e-01 7.46163845e-01 -1.01612151e+00 -4.93648350e-01 -5.22418678e-01 4.93728876e-01 -3.69441330e-01 2.53847867e-01 -5.68895876e-01 -5.06629348e-01 -2.00581662e-02 -1.50202727e+00 1.12413943e+00 3.56645361e-02 2.80239433e-02 -8.10852647e-01 -2.69514740e-01 4.80669141e-02 5.99657834e-01 3.44693691e-01 6.61806822e-01 -1.48809105e-01 -6.42613292e-01 -3.71348172e-01 -2.70603806e-01 1.35077929e+00 -6.22795708e-02 1.88113585e-01 -1.21700656e+00 -2.55218059e-01 3.15857917e-01 -2.62790710e-01 9.02304053e-01 3.17823321e-01 1.18158722e+00 5.62934354e-02 1.83099300e-01 7.21796095e-01 1.94072199e+00 -1.41947702e-01 1.06060946e+00 5.01462638e-01 4.03608382e-01 4.29524004e-01 2.28141293e-01 2.74586290e-01 3.75594236e-02 4.61897701e-01 9.30198789e-01 -3.49601001e-01 -3.78075382e-03 1.67418018e-01 5.83942950e-01 2.96443462e-01 2.60126770e-01 -7.57375956e-02 -6.39943302e-01 3.92691851e-01 -1.39207840e+00 -7.54087150e-01 -3.22679669e-01 2.44859195e+00 6.50973916e-01 3.87478113e-01 -2.85619467e-01 3.37972790e-01 4.06155229e-01 -2.13592779e-02 -5.30191064e-01 -5.19034624e-01 -7.37715065e-02 6.92492843e-01 1.33382058e+00 5.96702814e-01 -1.18733215e+00 4.53948200e-01 6.82206202e+00 3.92473012e-01 -1.25075769e+00 -5.50309382e-02 6.26936972e-01 -1.18893057e-01 2.14206837e-02 8.06599408e-02 -3.42984557e-01 3.21603864e-01 1.21504629e+00 3.45133156e-01 9.98930454e-01 6.59163117e-01 3.69038582e-01 -5.14451563e-01 -1.21593976e+00 1.06374109e+00 6.31886050e-02 -1.10371864e+00 -1.36620909e-01 -2.29594871e-01 4.77012038e-01 3.66339713e-01 -1.03818335e-01 9.87220928e-02 4.65121984e-01 -1.09271908e+00 7.02746212e-01 4.90541279e-01 5.45401812e-01 -5.42436242e-01 8.20028245e-01 2.36457855e-01 -6.94742620e-01 -1.10575370e-01 -5.41893721e-01 -2.39963606e-01 5.06401062e-02 1.03233337e+00 -6.68150663e-01 3.48152608e-01 1.03688323e+00 2.73849308e-01 -7.94813633e-01 9.66911376e-01 -2.77781904e-01 7.27567255e-01 -6.69403791e-01 4.35153306e-01 1.12122923e-01 -2.30373651e-01 3.81963998e-02 1.16749847e+00 6.04833603e-01 -1.19679935e-01 -1.60571441e-01 5.65961003e-01 -9.53301787e-02 -3.68503332e-01 -5.37557065e-01 2.30220884e-01 4.92528342e-02 1.27329612e+00 -6.71813965e-01 -3.31757188e-01 -6.29190445e-01 1.02919769e+00 -2.40595549e-01 4.93643254e-01 -8.29916775e-01 -4.67735440e-01 8.05751264e-01 2.53184438e-01 4.40355450e-01 -3.13683301e-01 -3.67318541e-01 -1.09112668e+00 2.09537655e-01 -1.17763567e+00 -5.28024808e-02 -1.21129906e+00 -1.02002680e+00 4.12847757e-01 -5.57177030e-02 -1.15972924e+00 -2.51497567e-01 -9.76081908e-01 -5.81289113e-01 8.51901114e-01 -1.42532229e+00 -9.50846255e-01 -6.53605402e-01 5.61259866e-01 2.84299940e-01 1.52748078e-01 8.63998413e-01 4.11337584e-01 -2.33619764e-01 1.84615195e-01 3.12321007e-01 2.58281827e-01 6.88235223e-01 -1.24102271e+00 6.99203014e-01 1.12450612e+00 7.08545744e-02 2.90602416e-01 1.05189788e+00 -1.67884484e-01 -1.53366470e+00 -1.19501030e+00 2.10267320e-01 -2.52753168e-01 7.14488566e-01 -4.10945714e-01 -7.37623155e-01 8.16793919e-01 4.78770137e-01 7.83306435e-02 1.49500802e-01 -2.98538566e-01 -3.88328671e-01 -6.26751363e-01 -1.12901068e+00 4.43314493e-01 6.37369692e-01 -3.94819260e-01 -1.80206060e-01 2.63882190e-01 4.75054651e-01 -5.71284413e-01 -8.24651718e-01 3.39546323e-01 2.61447608e-01 -1.33502400e+00 9.92739856e-01 -6.01775832e-02 2.85840064e-01 -4.47529554e-01 -3.70056242e-01 -1.40336788e+00 1.62058622e-01 -5.11475921e-01 4.56760347e-01 1.14455354e+00 3.91302258e-01 -5.08705676e-01 7.31868148e-01 6.43521190e-01 2.00372301e-02 2.02098815e-03 -7.80427694e-01 -5.54473281e-01 -1.27228154e-02 -1.01555204e+00 5.53417623e-01 4.36935365e-01 -7.18042672e-01 -2.95855086e-02 -3.04827362e-01 5.66387475e-01 7.30908215e-01 -2.77562767e-01 1.01890957e+00 -9.52297270e-01 -5.38684487e-01 -1.46458611e-01 -5.82738101e-01 -7.88322568e-01 -1.44899845e-01 -1.71200275e-01 2.48522714e-01 -1.20063806e+00 -2.81212628e-01 -2.41437927e-01 -3.21010090e-02 2.19263196e-01 2.00509325e-01 7.25590050e-01 1.03588350e-01 -1.39625371e-01 -1.00684516e-01 -2.63539515e-02 7.61901498e-01 -2.85484105e-01 2.63577014e-01 -2.04988718e-02 -5.35245121e-01 7.54878402e-01 7.45575666e-01 -4.70680803e-01 -3.82956922e-01 -9.62534845e-01 3.62943828e-01 -1.65430486e-01 7.21937180e-01 -1.56593513e+00 2.70395786e-01 7.93607309e-02 6.06207728e-01 -6.85974434e-02 5.00619411e-01 -1.28774846e+00 5.30321896e-01 3.07286531e-01 -2.50836402e-01 1.52305111e-01 3.51947635e-01 2.73479372e-01 -3.55507582e-01 -4.60845590e-01 9.58881199e-01 -3.82656693e-01 -7.80390799e-01 -6.87189549e-02 -4.73625481e-01 -2.26901397e-01 8.25570226e-01 -2.32806787e-01 -8.69717821e-02 -6.30695164e-01 -7.72624552e-01 -2.87413567e-01 8.50963891e-01 1.23053208e-01 3.25376898e-01 -9.37271118e-01 -5.39635479e-01 5.43816268e-01 -3.24210636e-02 2.96319336e-01 1.15469389e-01 5.17729402e-01 -1.07546318e+00 -2.72534937e-01 -7.70892724e-02 -5.95177412e-01 -1.08860910e+00 5.09796441e-01 5.69467425e-01 -3.85210216e-02 -6.04334831e-01 6.37452543e-01 -2.81495780e-01 -3.35484594e-01 4.74455386e-01 -7.75374353e-01 4.15925145e-01 -2.56387442e-01 5.19128561e-01 3.01022511e-02 6.33585632e-01 -3.15402538e-01 -3.76429185e-02 5.43693900e-01 2.43853033e-01 -1.41373828e-01 1.55382597e+00 -9.94583890e-02 -1.61646038e-01 1.98168501e-01 1.27984118e+00 -8.97653848e-02 -1.67940891e+00 5.09349406e-02 -1.08981974e-01 -2.64766365e-01 2.99737006e-01 -7.56967366e-01 -1.33805346e+00 1.00157702e+00 1.07460654e+00 3.46544206e-01 1.68634987e+00 -5.35357475e-01 6.28246367e-01 6.03907824e-01 2.72523195e-01 -1.21525109e+00 -2.09935844e-01 3.17462325e-01 5.61476111e-01 -1.35501945e+00 1.66849285e-01 -3.24343920e-01 -2.86111325e-01 1.31422782e+00 2.47565404e-01 -3.38568568e-01 6.31463647e-01 9.99985218e-01 5.69062829e-01 4.36554849e-02 -4.48147297e-01 -2.24251524e-02 -3.01405996e-01 7.08054125e-01 2.20155030e-01 -3.59988153e-01 2.88482100e-01 -5.96762300e-02 -1.66360945e-01 9.32829455e-02 9.33748245e-01 9.65471148e-01 -2.46172890e-01 -1.29623806e+00 -7.68826962e-01 3.69423866e-01 -6.11021399e-01 -1.44256011e-01 -1.10523365e-01 8.97261441e-01 3.20753306e-01 9.27236199e-01 1.73657492e-01 -3.77257198e-01 6.92193747e-01 -1.83114395e-01 4.73769754e-01 -3.96794736e-01 -7.17012048e-01 -5.23438752e-02 1.97435886e-01 -8.53817403e-01 -5.76468825e-01 -5.35167873e-01 -9.43646073e-01 -3.11660796e-01 2.42479630e-02 -2.01403543e-01 1.23392570e+00 9.42448318e-01 7.58647993e-02 6.53640509e-01 3.89346451e-01 -1.28286958e+00 -7.89434791e-01 -8.03958476e-01 -5.04017830e-01 7.56460905e-01 4.99555498e-01 -1.47147208e-01 -5.02985656e-01 5.46655476e-01]
[11.003206253051758, -2.3672523498535156]
96ba781b-f468-407c-aa3a-d1caabd89903
language-identification-using-deep
1708.04811
null
http://arxiv.org/abs/1708.04811v1
http://arxiv.org/pdf/1708.04811v1.pdf
Language Identification Using Deep Convolutional Recurrent Neural Networks
Language Identification (LID) systems are used to classify the spoken language from a given audio sample and are typically the first step for many spoken language processing tasks, such as Automatic Speech Recognition (ASR) systems. Without automatic language detection, speech utterances cannot be parsed correctly and grammar rules cannot be applied, causing subsequent speech recognition steps to fail. We propose a LID system that solves the problem in the image domain, rather than the audio domain. We use a hybrid Convolutional Recurrent Neural Network (CRNN) that operates on spectrogram images of the provided audio snippets. In extensive experiments we show, that our model is applicable to a range of noisy scenarios and can easily be extended to previously unknown languages, while maintaining its classification accuracy. We release our code and a large scale training set for LID systems to the community.
['Tom Herold', 'Christian Bartz', 'Haojin Yang', 'Christoph Meinel']
2017-08-16
null
null
null
null
['spoken-language-identification']
['speech']
[ 3.30451012e-01 -2.73923993e-01 4.88194339e-02 -4.50391471e-01 -1.20620787e+00 -8.04949284e-01 4.30996597e-01 -2.17226535e-01 -4.56678063e-01 3.78157765e-01 2.04841062e-01 -6.00497842e-01 3.10102493e-01 -2.93326139e-01 -3.45892936e-01 -4.66758132e-01 -1.99429076e-02 4.82865304e-01 -1.81069579e-02 -4.67106178e-02 -9.87554565e-02 5.81612766e-01 -1.70715773e+00 4.08707410e-01 2.16836706e-01 1.03153527e+00 1.76327541e-01 1.25033021e+00 -2.20908225e-01 9.68813896e-01 -6.95441425e-01 1.90461352e-01 1.00739323e-01 -3.91702354e-01 -9.94119763e-01 3.31688821e-01 1.99757189e-01 -3.52451354e-01 -3.24034989e-01 9.30414796e-01 5.91147721e-01 1.64511472e-01 2.98171103e-01 -1.03196168e+00 -3.63854289e-01 7.47648358e-01 2.25774199e-02 1.46874100e-01 5.90396285e-01 -6.97120354e-02 9.47340906e-01 -1.12524617e+00 2.47857347e-01 1.51413488e+00 4.79108512e-01 7.37219274e-01 -1.19000936e+00 -7.42985904e-01 2.56419212e-01 -8.90510976e-02 -1.56536829e+00 -1.01230669e+00 5.22740841e-01 -1.45074189e-01 1.12707627e+00 2.71904409e-01 2.12619737e-01 1.16681695e+00 -3.18550974e-01 1.03894722e+00 8.80730152e-01 -5.77875555e-01 3.09630930e-01 1.16587654e-01 3.23581338e-01 4.09337848e-01 -4.02891695e-01 -1.95479974e-01 -6.00477934e-01 -1.94488019e-01 3.99534404e-01 -3.97652715e-01 -3.62327039e-01 5.75175844e-02 -1.00126863e+00 8.52034867e-01 -6.06958084e-02 2.84396052e-01 -2.61187911e-01 4.02771082e-04 5.88702679e-01 5.89163065e-01 4.10200924e-01 1.19700246e-01 -2.50932902e-01 -3.40759188e-01 -1.10194385e+00 -1.76158864e-02 9.48545337e-01 6.28573418e-01 5.44312119e-01 2.34936610e-01 9.90295410e-02 1.30546832e+00 2.37231061e-01 4.25554663e-01 8.25622559e-01 -7.51083493e-01 1.70001477e-01 1.68258533e-01 -1.20930977e-01 -4.26201701e-01 -4.15070117e-01 -1.80955186e-01 -7.35900044e-01 -1.24394409e-01 2.54608095e-01 -1.54090121e-01 -1.11474061e+00 1.56422412e+00 2.76893917e-02 4.87380698e-02 5.13090491e-01 9.33357298e-01 9.96797502e-01 8.00376832e-01 -2.19185010e-01 -2.75979877e-01 1.36377108e+00 -7.06026196e-01 -6.69304371e-01 -4.84018385e-01 5.50396740e-01 -9.46014404e-01 1.18221271e+00 6.32215381e-01 -9.32893336e-01 -3.33338261e-01 -8.74495566e-01 -8.46573412e-02 -1.73982650e-01 4.00414944e-01 1.92694455e-01 6.05290294e-01 -1.42102766e+00 -4.20183204e-02 -7.59545267e-01 -7.20111072e-01 -2.60164171e-01 5.12682915e-01 -3.83631408e-01 1.03764795e-02 -1.12368810e+00 4.97758418e-01 1.54760629e-01 2.25972936e-01 -9.71347809e-01 -4.82939370e-02 -1.03466249e+00 -2.29106261e-03 2.49040961e-01 4.09500971e-02 1.85299885e+00 -1.11490786e+00 -1.78872001e+00 8.34273815e-01 -5.17881632e-01 -7.11663246e-01 1.45148501e-01 -4.98294309e-02 -4.66301888e-01 2.26636171e-01 -1.34226575e-01 5.31201780e-01 1.03070939e+00 -9.39607739e-01 -5.72863340e-01 -9.18254629e-02 -2.90153772e-01 1.79672018e-01 -2.17672467e-01 6.12827480e-01 -4.56892312e-01 -5.36216378e-01 3.23086232e-01 -1.01253664e+00 8.48180950e-02 -4.51124400e-01 -4.21865046e-01 -3.96616459e-01 9.68629003e-01 -7.13628471e-01 1.07731712e+00 -2.41868258e+00 -1.22145660e-01 1.90705687e-01 -1.80924237e-01 3.50378752e-01 -2.89586246e-01 2.94803709e-01 -1.02904119e-01 6.71862662e-02 -1.91407070e-01 -6.92054629e-01 -1.81861445e-01 2.51256227e-01 -7.41092682e-01 3.30522001e-01 2.39600137e-01 3.85838568e-01 -6.22774303e-01 -1.80927232e-01 3.57981659e-02 4.22772884e-01 -3.15835476e-01 5.22087336e-01 -2.07939878e-01 2.10352898e-01 1.97838317e-03 5.26753426e-01 2.70618796e-01 3.22986059e-02 1.03451595e-01 2.68090934e-01 -1.50028482e-01 9.51297224e-01 -1.34243643e+00 1.27691233e+00 -7.38415837e-01 8.86472583e-01 6.81847572e-01 -1.00061619e+00 1.14089894e+00 7.22888350e-01 8.72683823e-02 -3.85742098e-01 3.13320793e-02 4.09370005e-01 1.24931917e-01 -2.68333793e-01 5.51873028e-01 2.06982553e-01 -1.03410184e-01 9.08896625e-01 2.41233647e-01 -2.23872453e-01 5.78416213e-02 2.50434935e-01 1.02832091e+00 -6.21765494e-01 1.38737753e-01 4.62922901e-02 6.14728808e-01 -2.51660824e-01 3.12484205e-01 8.52322519e-01 -2.26431012e-01 8.24835956e-01 2.52985299e-01 -3.02306980e-01 -9.87629056e-01 -8.93850386e-01 -1.51406869e-01 1.35502434e+00 -4.14965153e-01 -3.13173026e-01 -7.14256942e-01 -3.52786839e-01 -2.88059473e-01 3.67540658e-01 -9.24782827e-03 -7.11123571e-02 -4.64104116e-01 -3.51541221e-01 1.03780484e+00 4.41978514e-01 2.62500614e-01 -1.27614784e+00 -6.94584176e-02 4.34469908e-01 -2.12505206e-01 -1.33974338e+00 -6.66489065e-01 4.61037546e-01 -2.47515738e-01 -7.19052315e-01 -5.84774673e-01 -1.06105399e+00 3.13237488e-01 3.76310974e-01 7.63398767e-01 5.94262891e-02 -2.07078353e-01 6.13483906e-01 -3.70136589e-01 -2.23875508e-01 -1.05083776e+00 2.17382073e-01 6.40873551e-01 3.82844329e-01 4.57037598e-01 -2.47152939e-01 -1.32602900e-01 2.82205403e-01 -7.09793746e-01 -1.91610113e-01 2.01545984e-01 9.92459893e-01 3.41294348e-01 -1.22805044e-01 7.85426140e-01 -3.35813046e-01 8.87353301e-01 -6.59712926e-02 -7.17954457e-01 2.21900865e-01 -1.30688325e-01 -8.28340501e-02 7.44435906e-01 -6.46508574e-01 -7.03500330e-01 4.17864591e-01 -5.67133904e-01 -4.74705905e-01 -4.68846440e-01 6.27884746e-01 -1.28269196e-01 8.68992954e-02 4.11553383e-01 6.76968098e-01 1.68295577e-01 -6.88696384e-01 7.33282939e-02 1.45360255e+00 7.83838511e-01 -2.86680937e-01 2.79684186e-01 1.20747000e-01 -6.53847694e-01 -1.43502665e+00 -5.80117762e-01 -7.63347626e-01 -3.92799884e-01 -8.87937099e-02 5.28864861e-01 -1.18474460e+00 -7.99232543e-01 5.63605309e-01 -1.18057871e+00 -3.36476207e-01 8.75708163e-02 7.01144993e-01 -5.14071465e-01 2.36108661e-01 -8.14139903e-01 -1.45858467e+00 -4.56210345e-01 -1.28655708e+00 1.34702265e+00 4.38233949e-02 -4.01112616e-01 -6.13373518e-01 7.99567252e-03 1.31510973e-01 3.12298089e-01 -8.84299099e-01 6.27013326e-01 -1.19612610e+00 -3.95865858e-01 -2.05202684e-01 1.27383946e-02 6.56154990e-01 3.07204783e-01 2.56673992e-01 -1.56212759e+00 -4.45349693e-01 -1.48381963e-01 -7.40105987e-01 1.15328693e+00 3.02544862e-01 1.01011980e+00 -3.59399259e-01 4.49032858e-02 3.15231055e-01 6.45934701e-01 3.09997320e-01 3.85718971e-01 -1.86513849e-02 4.23318505e-01 6.36087418e-01 1.32369131e-01 2.08799809e-01 3.09979051e-01 6.42395139e-01 -1.30186662e-01 -6.24984168e-02 -1.18287370e-01 -1.55334696e-01 9.01022196e-01 1.14581120e+00 7.04954088e-01 -3.90904099e-01 -1.13425672e+00 8.24051380e-01 -1.60739005e+00 -8.66646111e-01 3.31903785e-01 2.21702957e+00 8.26344371e-01 -1.93016115e-03 3.30120713e-01 3.54083031e-01 8.33380401e-01 -7.95183703e-02 -3.85057658e-01 -7.79790759e-01 -1.88899532e-01 2.28241280e-01 1.31625488e-01 8.38945150e-01 -1.25216413e+00 1.05150735e+00 7.25328207e+00 6.42773390e-01 -1.53935289e+00 -2.03227744e-01 5.15870035e-01 -2.20659412e-02 1.37949421e-03 -2.08244234e-01 -8.98070395e-01 7.70866349e-02 1.35299456e+00 -8.31923932e-02 6.99490190e-01 7.34873712e-01 4.37454402e-01 -6.12079501e-02 -1.24454534e+00 1.24463236e+00 9.75031629e-02 -9.29296792e-01 -1.51143959e-02 -2.05831990e-01 3.45618725e-02 3.76547337e-01 1.28870234e-01 3.54560643e-01 5.33197939e-01 -1.20971334e+00 5.67286015e-01 7.46927410e-02 8.66013587e-01 -6.49213672e-01 5.02590001e-01 4.69444692e-01 -1.17826688e+00 -1.26170814e-01 -6.04731552e-02 -1.12355329e-01 -5.61448075e-02 2.80204684e-01 -1.45804405e+00 -4.39657122e-02 7.16722488e-01 4.32328194e-01 -3.22747588e-01 7.98083007e-01 -1.19698919e-01 1.00917971e+00 -6.68121040e-01 -6.74780831e-02 2.36863345e-01 1.86693430e-01 5.41043043e-01 1.25298548e+00 4.21801984e-01 -7.97970220e-02 4.20981497e-01 5.73632717e-01 -2.53151834e-01 2.10167304e-01 -7.62674153e-01 -4.46278065e-01 6.55644238e-01 1.09753323e+00 -6.42128825e-01 -2.26611495e-01 -3.45823079e-01 9.00671124e-01 2.41557047e-01 5.32523036e-01 -8.95085260e-02 -3.26713949e-01 1.00887036e+00 -3.26787859e-01 2.38130465e-01 -4.59863365e-01 3.58052701e-02 -1.18900049e+00 6.35051355e-02 -1.06358397e+00 3.02119493e-01 -6.64492965e-01 -1.20038831e+00 8.49125326e-01 -3.75495672e-01 -1.06615627e+00 -8.49234581e-01 -5.89831114e-01 -5.44082940e-01 8.75218630e-01 -1.48651278e+00 -8.11203539e-01 1.91967621e-01 5.31801045e-01 8.55667233e-01 -4.33636457e-01 1.18601191e+00 2.21640438e-01 -6.97845161e-01 6.75488710e-01 1.64671794e-01 5.25932014e-01 8.24946940e-01 -9.52884316e-01 5.94292819e-01 8.80590558e-01 5.31817794e-01 5.78650355e-01 5.88789642e-01 -3.25831234e-01 -1.26271081e+00 -1.15904355e+00 1.07502925e+00 1.11488523e-02 6.89609826e-01 -9.37001467e-01 -1.13176370e+00 6.08075738e-01 2.05506578e-01 -1.40200078e-01 6.84853852e-01 9.53955278e-02 -5.27899027e-01 -6.41816780e-02 -7.16383278e-01 4.19400513e-01 4.91846859e-01 -1.23792529e+00 -3.91117126e-01 3.04158688e-01 8.32510769e-01 -2.95722783e-01 -3.77997845e-01 1.30194977e-01 5.25777936e-01 -4.79072511e-01 6.25111282e-01 -4.58305299e-01 -1.96988821e-01 -3.63030910e-01 -3.70885938e-01 -1.17049313e+00 1.99009374e-01 -8.11511576e-01 1.35062605e-01 1.34209406e+00 4.89680499e-01 -5.76002181e-01 3.71422738e-01 4.54848051e-01 6.79611936e-02 -5.06284796e-02 -1.28061998e+00 -8.31014276e-01 -1.99524090e-01 -8.93333018e-01 4.64925528e-01 5.92805505e-01 7.24449828e-02 6.20272636e-01 -4.35556740e-01 4.50899541e-01 2.05495819e-01 1.60190221e-02 6.59441829e-01 -1.00284564e+00 -1.82883143e-01 -3.03310871e-01 -3.46511006e-01 -1.10562587e+00 5.12120962e-01 -7.07006574e-01 4.99470890e-01 -1.14905202e+00 -1.56628251e-01 -3.86266887e-01 -6.33780509e-02 7.49890029e-01 2.65162498e-01 3.84830147e-01 9.42602009e-02 2.84610718e-01 -4.89984721e-01 3.58667433e-01 3.61786962e-01 -4.10109639e-01 -3.84071261e-01 2.49924347e-01 -3.23924303e-01 6.55552745e-01 6.62919462e-01 -2.11534604e-01 -2.29084745e-01 -3.04918110e-01 -1.51618347e-01 8.66725668e-02 2.78237313e-01 -8.67912054e-01 4.77711350e-01 2.71625936e-01 -6.14205655e-03 -5.46979606e-01 5.21853805e-01 -4.78910536e-01 -2.96512455e-01 9.07272100e-02 -7.46824324e-01 -7.28513971e-02 2.64458448e-01 1.33423030e-01 -5.97960114e-01 -8.54232386e-02 8.06285560e-01 4.07153592e-02 -5.89192092e-01 4.14241105e-02 -1.06831729e+00 -2.47910649e-01 4.68454570e-01 3.32184322e-02 -6.55914471e-02 -9.96171236e-01 -9.09691155e-01 2.58030266e-01 3.00598979e-01 5.84096670e-01 7.66077459e-01 -1.10734165e+00 -7.29664564e-01 6.87105477e-01 1.74233332e-01 -1.22166209e-01 -3.27822194e-02 5.82163632e-01 -2.91253865e-01 6.31289005e-01 4.15956318e-01 -7.91612983e-01 -1.58665490e+00 3.27715904e-01 5.82783222e-01 1.48035422e-01 -5.45440078e-01 7.79531062e-01 1.52495384e-01 -5.87985933e-01 8.73362482e-01 -5.36725461e-01 -1.99763834e-01 1.49947470e-02 9.90077794e-01 -2.43753940e-01 5.55911958e-01 -9.62920666e-01 -5.35280943e-01 1.25271350e-01 -3.06919664e-01 -4.92658377e-01 1.01108348e+00 -3.05117428e-01 -1.66657176e-02 8.52409124e-01 1.20724261e+00 -1.50207609e-01 -7.69179285e-01 -4.88947928e-01 -2.88167577e-02 5.90354651e-02 2.34699637e-01 -6.24530673e-01 -6.21420920e-01 1.14288282e+00 6.58609092e-01 5.71096063e-01 1.21644819e+00 1.39720872e-01 7.74238288e-01 7.48734057e-01 1.63239285e-01 -1.16794586e+00 -1.24770835e-01 1.01914966e+00 8.41317356e-01 -1.23745596e+00 -7.14525878e-01 -1.23959616e-01 -6.23851478e-01 1.23264229e+00 2.26231337e-01 9.31097269e-02 5.88427722e-01 6.73493266e-01 5.81648409e-01 2.66069651e-01 -1.02407897e+00 -4.23116207e-01 1.71116933e-01 5.11711776e-01 6.64341748e-01 4.08406928e-02 2.42309660e-01 6.43444657e-01 -2.66152442e-01 -3.59174043e-01 6.81980908e-01 8.06084812e-01 -4.89964664e-01 -1.15449262e+00 -8.02465737e-01 2.09472060e-01 -5.08547366e-01 -1.78191379e-01 -8.79416943e-01 1.76490799e-01 -4.84776616e-01 1.32792091e+00 3.92810941e-01 -4.01548296e-01 8.35166052e-02 5.42462170e-01 -6.22898787e-02 -8.80399227e-01 -4.36071604e-01 5.03763855e-01 1.91173241e-01 -3.45481545e-01 -3.05233061e-01 -6.40692174e-01 -1.36614823e+00 2.77604777e-02 -2.53257245e-01 3.06512475e-01 7.26632237e-01 1.01245868e+00 4.07387912e-01 1.95917591e-01 9.23187971e-01 -6.79703414e-01 -5.21916330e-01 -1.00560212e+00 -7.16297269e-01 -8.57380703e-02 1.03742027e+00 -2.37144440e-01 -3.35286766e-01 7.91051909e-02]
[14.179137229919434, 6.537304878234863]
c0e43291-50a7-4209-af92-73f227619d3d
bayesian-neural-network-language-modeling-for
2208.13259
null
https://arxiv.org/abs/2208.13259v1
https://arxiv.org/pdf/2208.13259v1.pdf
Bayesian Neural Network Language Modeling for Speech Recognition
State-of-the-art neural network language models (NNLMs) represented by long short term memory recurrent neural networks (LSTM-RNNs) and Transformers are becoming highly complex. They are prone to overfitting and poor generalization when given limited training data. To this end, an overarching full Bayesian learning framework encompassing three methods is proposed in this paper to account for the underlying uncertainty in LSTM-RNN and Transformer LMs. The uncertainty over their model parameters, choice of neural activations and hidden output representations are modeled using Bayesian, Gaussian Process and variational LSTM-RNN or Transformer LMs respectively. Efficient inference approaches were used to automatically select the optimal network internal components to be Bayesian learned using neural architecture search. A minimal number of Monte Carlo parameter samples as low as one was also used. These allow the computational costs incurred in Bayesian NNLM training and evaluation to be minimized. Experiments are conducted on two tasks: AMI meeting transcription and Oxford-BBC LipReading Sentences 2 (LRS2) overlapped speech recognition using state-of-the-art LF-MMI trained factored TDNN systems featuring data augmentation, speaker adaptation and audio-visual multi-channel beamforming for overlapped speech. Consistent performance improvements over the baseline LSTM-RNN and Transformer LMs with point estimated model parameters and drop-out regularization were obtained across both tasks in terms of perplexity and word error rate (WER). In particular, on the LRS2 data, statistically significant WER reductions up to 1.3% and 1.2% absolute (12.1% and 11.3% relative) were obtained over the baseline LSTM-RNN and Transformer LMs respectively after model combination between Bayesian NNLMs and their respective baselines.
['Helen Meng', 'Xunying Liu', 'Mengzhe Geng', 'Junhao Xu', 'Shoukang Hu', 'Boyang Xue']
2022-08-28
null
null
null
null
['lipreading']
['computer-vision']
[ 2.19956696e-01 4.25525457e-01 -2.83850972e-02 -4.61850911e-01 -1.40326786e+00 -1.52293444e-01 6.40381575e-01 -5.25591135e-01 -6.21730030e-01 7.75238395e-01 4.60713267e-01 -5.21296620e-01 -7.21139833e-02 1.86913814e-02 -7.33665168e-01 -9.91123855e-01 3.10621619e-01 7.60766268e-01 -4.09049951e-02 2.06731901e-01 -9.31256190e-02 2.83314168e-01 -1.28898382e+00 2.12095246e-01 4.76419449e-01 9.97141004e-01 4.68080252e-01 9.74295855e-01 -7.45342299e-02 5.36641181e-01 -6.18809879e-01 -1.42048463e-01 -2.60659248e-01 -2.59061866e-02 -4.24264073e-01 -1.10724904e-01 1.55418694e-01 -1.47602916e-01 -3.42650473e-01 7.98351049e-01 9.22379076e-01 3.96879286e-01 1.04078996e+00 -6.97181225e-01 -2.92803675e-01 1.02940667e+00 -4.62018639e-01 1.81464463e-01 -1.53517365e-01 2.04995021e-01 6.89800382e-01 -1.20999944e+00 -2.54765570e-01 1.79841328e+00 6.62974894e-01 8.39778721e-01 -1.39724267e+00 -7.73923039e-01 1.28156438e-01 4.96132188e-02 -1.54374838e+00 -1.14035022e+00 4.22105312e-01 -3.27956676e-01 1.58550644e+00 -2.96599306e-02 9.48746577e-02 1.56217718e+00 3.72017883e-02 6.70502305e-01 8.83094311e-01 -8.34644794e-01 3.40735391e-02 3.98575246e-01 8.88179615e-02 4.11857575e-01 -2.15095699e-01 1.62941039e-01 -9.63203728e-01 -1.43389940e-01 4.68832761e-01 -5.90841711e-01 7.28703961e-02 2.46183366e-01 -1.01199389e+00 6.85846150e-01 -3.25333297e-01 3.05617392e-01 -3.69843274e-01 4.54374105e-01 5.52720845e-01 -1.60006317e-03 5.97122252e-01 -2.26231068e-01 -8.10155213e-01 -3.46095979e-01 -1.21109211e+00 -1.84508458e-01 5.81009388e-01 7.64003098e-01 2.38933623e-01 7.19156027e-01 -3.08568448e-01 1.50108242e+00 1.02610862e+00 9.25492525e-01 7.09176421e-01 -6.94601476e-01 5.68670690e-01 -2.36770496e-01 -1.15596525e-01 -3.40685427e-01 -2.20798537e-01 -5.58303356e-01 -9.53224480e-01 -1.59373820e-01 1.31010488e-02 -2.27062270e-01 -1.43926203e+00 2.00107646e+00 -9.15385410e-02 1.65965348e-01 3.17131430e-01 2.76691943e-01 6.44439995e-01 1.16531134e+00 1.31571159e-01 -4.45741773e-01 1.27061939e+00 -7.09873199e-01 -1.13640630e+00 -3.30876529e-01 4.76715505e-01 -1.04419243e+00 1.01949513e+00 5.86317599e-01 -1.36275923e+00 -6.39954150e-01 -8.82965863e-01 2.23679066e-01 -6.28567487e-02 6.75097167e-01 -8.46787021e-02 1.00547349e+00 -1.12301636e+00 2.76404649e-01 -1.08710229e+00 -1.19506642e-01 1.31166771e-01 8.83633196e-01 -2.12988295e-02 2.63084233e-01 -1.26600730e+00 1.05342674e+00 4.25854325e-01 5.20446241e-01 -1.22516394e+00 -4.83840138e-01 -6.51399016e-01 -2.75967140e-02 6.15276620e-02 -5.12929797e-01 1.54194224e+00 -7.24280596e-01 -2.24775076e+00 5.15868127e-01 -6.69414878e-01 -6.98813200e-01 4.00426127e-02 -3.19243103e-01 -4.94909346e-01 -3.32365185e-01 -5.63725770e-01 7.75703728e-01 1.22534382e+00 -9.80211496e-01 -4.51681852e-01 -2.76722461e-01 -8.83346021e-01 3.73577774e-01 -1.80479228e-01 3.90133500e-01 -9.82201248e-02 -6.39473617e-01 2.92193145e-01 -8.68536115e-01 1.25291407e-01 -7.70203829e-01 -5.84253669e-01 -5.28504014e-01 7.18798995e-01 -9.31740701e-01 1.20221686e+00 -2.00670719e+00 1.60399675e-01 2.29765847e-01 -4.33642149e-01 3.77902150e-01 1.03423689e-02 8.51420909e-02 1.09927438e-01 1.01595573e-01 3.65438834e-02 -9.08189833e-01 2.13099211e-01 3.32933873e-01 -4.56567526e-01 4.02162194e-01 -1.13640368e-01 5.73741376e-01 -5.58010377e-02 -3.50045443e-01 4.85448420e-01 1.11397648e+00 -1.51060849e-01 2.60363549e-01 -1.63716674e-01 3.59723359e-01 2.87061661e-01 4.71511245e-01 2.81079173e-01 3.66027027e-01 -4.18022089e-02 -4.32431161e-01 -8.16693623e-03 6.46197021e-01 -1.19019973e+00 1.44221759e+00 -7.70670831e-01 9.62812245e-01 1.58331245e-01 -8.49719524e-01 8.94654512e-01 9.42501843e-01 -1.52040273e-01 -2.17829898e-01 1.47868782e-01 4.91558522e-01 1.94439307e-01 -3.39084774e-01 1.19077489e-01 -4.31185067e-01 2.74641573e-01 6.59596324e-02 6.42435610e-01 5.41871972e-02 -3.48912507e-01 -3.29483300e-01 3.25513452e-01 3.41112278e-02 -5.76168522e-02 -2.09231794e-01 6.00620091e-01 -7.86168993e-01 4.74235743e-01 8.88988972e-01 1.66843623e-01 4.59815949e-01 9.20402110e-02 1.32267296e-01 -9.66222584e-01 -1.15566003e+00 -3.12996596e-01 1.25588918e+00 -9.85151052e-01 -4.29800339e-02 -8.07329059e-01 1.60203934e-01 -5.83613873e-01 1.23103595e+00 -2.33295843e-01 -4.19955589e-02 -5.73437691e-01 -8.08534086e-01 1.09272122e+00 4.03076470e-01 1.46798864e-01 -9.93760765e-01 -1.24707989e-01 4.03788298e-01 -2.32132256e-01 -1.17603874e+00 -3.17590714e-01 4.51133758e-01 -9.79426920e-01 -2.09009543e-01 -8.79044533e-01 -5.67395389e-01 2.89649397e-01 -4.05850589e-01 6.48946047e-01 -9.14609551e-01 1.30923167e-01 2.61788696e-01 2.00261459e-01 -6.73506200e-01 -6.40067518e-01 1.45525232e-01 6.55680418e-01 6.41154945e-02 3.82019639e-01 -5.81609488e-01 -1.38427868e-01 2.97104180e-01 -3.76914859e-01 -3.57529372e-02 7.91660488e-01 1.17026389e+00 4.73892897e-01 -1.45882189e-01 5.44586837e-01 -3.59335095e-01 6.35695279e-01 -7.39860237e-02 -6.57593489e-01 3.76812518e-01 -6.16855025e-01 3.37484270e-01 1.32384812e-02 -7.71526456e-01 -1.56471145e+00 -2.29760334e-01 -3.57908309e-01 -6.27743661e-01 -2.23510951e-01 4.57115442e-01 -2.97104061e-01 3.73495400e-01 3.74636889e-01 3.99050355e-01 -3.20883803e-02 -6.08570576e-01 1.82953313e-01 1.05751419e+00 4.14461732e-01 -4.34250563e-01 1.58904880e-01 -1.08041987e-01 -2.71736860e-01 -1.18246973e+00 -5.91243446e-01 -3.30249906e-01 -4.83240426e-01 -5.50787859e-02 8.32455933e-01 -9.37645257e-01 -7.30640233e-01 7.75415778e-01 -1.44294906e+00 -3.42496008e-01 8.32252502e-02 9.27309811e-01 -6.10104322e-01 -6.51321933e-02 -5.55913448e-01 -1.65500998e+00 -6.07036531e-01 -1.48799431e+00 8.51837873e-01 -9.02642310e-02 -4.01562274e-01 -1.08558762e+00 -3.13228555e-02 4.71691579e-01 7.90340006e-01 -5.22405326e-01 1.10888469e+00 -8.97224009e-01 -7.19256327e-02 -6.55417144e-02 2.05681026e-01 9.43313658e-01 2.25041341e-03 1.16527840e-01 -1.54683375e+00 -2.18825430e-01 1.13350324e-01 -2.63233215e-01 8.02079201e-01 1.19152081e+00 7.48674631e-01 -4.43491191e-01 -2.79315680e-01 3.27540994e-01 9.77348208e-01 3.12085211e-01 4.04152125e-01 -2.31304452e-01 5.11400938e-01 5.17955899e-01 9.16070677e-03 2.18523279e-01 -1.46104112e-01 6.83145642e-01 -8.25859327e-03 3.01675856e-01 -2.04279929e-01 -3.20094854e-01 8.49976778e-01 1.35154712e+00 1.96626157e-01 -4.84373063e-01 -1.01328635e+00 4.80860770e-01 -1.57650256e+00 -8.27992141e-01 5.77629320e-02 2.41057348e+00 8.17504883e-01 4.48730975e-01 -8.52140114e-02 2.60772765e-01 7.52245367e-01 -1.44215571e-02 -4.57428187e-01 -5.91156065e-01 -2.28752315e-01 1.86869800e-01 6.32455826e-01 9.74503875e-01 -7.13883102e-01 1.00895298e+00 5.75528717e+00 1.24305153e+00 -9.64847565e-01 4.93421137e-01 8.42515528e-01 -5.21037281e-01 3.07670888e-03 -2.76194394e-01 -1.43117619e+00 1.69827700e-01 1.92222536e+00 5.63124120e-01 4.03238922e-01 3.57300997e-01 7.48050690e-01 -1.75627992e-01 -8.56043756e-01 1.22402275e+00 -3.90766636e-02 -1.00132084e+00 2.43767649e-02 1.19867273e-01 4.03852135e-01 6.20981634e-01 4.41560298e-01 4.71923023e-01 1.45888150e-01 -1.45181549e+00 8.49856079e-01 7.53549695e-01 8.99465382e-01 -8.74392033e-01 8.00835967e-01 4.61156458e-01 -8.19724202e-01 -7.90820047e-02 -3.83870453e-01 4.66438562e-01 3.43731850e-01 5.17095625e-01 -1.07677758e+00 -1.27732279e-02 5.86827636e-01 -1.34597179e-02 2.28998870e-01 6.45109534e-01 1.28704635e-02 1.32531965e+00 -8.22743416e-01 -1.27745047e-01 3.11070859e-01 1.71920300e-01 7.80440509e-01 1.42160225e+00 5.02415538e-01 -2.58716732e-01 -4.29268837e-01 7.93313086e-01 1.73670594e-02 -5.79379126e-02 -3.35607022e-01 4.91418131e-02 7.11146057e-01 8.67429316e-01 -1.59251720e-01 -2.21930683e-01 1.45945484e-02 5.86020947e-01 -3.52264345e-02 7.88597763e-01 -7.22603679e-01 -1.31746411e-01 5.34903288e-01 -7.32564658e-04 2.84877896e-01 -2.27931648e-01 -4.79222648e-02 -7.73492515e-01 -2.02980876e-01 -7.89153695e-01 -5.07380329e-02 -7.55167902e-01 -9.68602955e-01 8.97105575e-01 3.62489343e-01 -4.67882663e-01 -8.22129428e-01 -7.72385955e-01 -3.17803711e-01 1.40024388e+00 -1.33193469e+00 -1.32137656e+00 5.77233613e-01 4.30453748e-01 9.47089911e-01 -7.02139616e-01 1.01087332e+00 1.99512348e-01 -7.88733363e-01 8.89799476e-01 4.87940311e-01 -2.26502284e-01 4.79644507e-01 -8.47477436e-01 1.02194361e-01 5.56125164e-01 2.37023950e-01 7.54213691e-01 8.70482266e-01 -2.94601411e-01 -1.01915264e+00 -8.17867398e-01 1.05489790e+00 -2.36072138e-01 3.99357855e-01 -6.66042507e-01 -7.92611361e-01 6.81220889e-01 3.84763479e-01 -2.80760437e-01 6.37915850e-01 2.54824460e-01 -8.78522247e-02 -1.40376300e-01 -9.22587097e-01 4.85647291e-01 3.37961107e-01 -6.92486167e-01 -5.99901557e-01 2.29254082e-01 5.74972272e-01 -2.47507989e-01 -6.48889244e-01 5.43892562e-01 6.20110750e-01 -6.85329914e-01 8.92707586e-01 -3.23594511e-01 -5.67914009e-01 -6.11798726e-02 -7.71842539e-01 -1.03412712e+00 1.90084189e-01 -1.10042763e+00 -3.02908242e-01 1.56589293e+00 8.68019581e-01 -5.21405339e-01 4.88163233e-01 4.82983679e-01 -2.61473745e-01 -7.04087913e-01 -1.48193479e+00 -6.71140373e-01 3.38702165e-02 -1.18369102e+00 -7.53853023e-02 -4.04077880e-02 -4.08646911e-01 5.75200081e-01 -5.72937310e-01 3.16173106e-01 6.95400655e-01 -1.18966746e+00 3.04050565e-01 -9.96865988e-01 -2.68768728e-01 -4.79043305e-01 7.88337924e-03 -1.18417096e+00 5.42405903e-01 -4.64359075e-01 3.30009192e-01 -1.45554292e+00 -4.42961529e-02 -1.19543776e-01 -4.53077585e-01 4.54790235e-01 3.66595984e-01 -2.45364383e-01 -1.60099402e-01 1.79312587e-01 4.11321223e-02 7.67904162e-01 2.86405295e-01 -7.07984120e-02 -3.19816470e-01 5.35369158e-01 1.75771154e-02 8.95087481e-01 9.12511885e-01 -6.84440076e-01 -6.43538535e-01 -3.30819041e-01 4.48131524e-02 2.45710865e-01 2.01083854e-01 -9.84256804e-01 3.79293919e-01 2.44569913e-01 1.58237696e-01 -9.46539223e-01 1.02759826e+00 -4.17381287e-01 1.84529543e-01 1.94359317e-01 -6.74386859e-01 -1.90222234e-01 5.01354575e-01 5.62898040e-01 -1.21380046e-01 -4.32214677e-01 9.80910838e-01 1.98630560e-02 2.54898779e-02 -1.55678883e-01 -1.10081029e+00 -2.46604055e-01 3.27535510e-01 -2.95187950e-01 4.37465400e-01 -5.63080609e-01 -9.56274867e-01 -3.07952851e-01 -5.07725596e-01 2.40428150e-01 7.57650971e-01 -1.09792948e+00 -8.63925636e-01 3.39624405e-01 -3.95729035e-01 -1.16354994e-01 2.39304796e-01 1.08688712e+00 2.24752530e-01 1.11686838e+00 4.21583086e-01 -7.18074024e-01 -1.50091171e+00 -1.88224271e-01 8.07925820e-01 -2.89687634e-01 -5.12632057e-02 1.34451425e+00 1.07541934e-01 -5.76625288e-01 8.94275546e-01 -5.24953544e-01 -3.67343649e-02 -1.51080787e-01 2.76354909e-01 5.78541994e-01 1.97420418e-01 -7.37944841e-01 -3.22919995e-01 3.62437546e-01 -1.00787729e-01 -9.26245928e-01 1.06770027e+00 -5.32242239e-01 4.85240631e-02 1.15389419e+00 1.05463028e+00 -3.02857041e-01 -1.10498405e+00 -4.08948123e-01 1.11440875e-01 3.16336423e-01 7.28718221e-01 -1.03442025e+00 -6.06417716e-01 1.42382383e+00 9.94533479e-01 -3.77494395e-01 7.03819931e-01 -5.47484830e-02 5.14750719e-01 4.40983802e-01 -2.01049939e-01 -1.32564712e+00 -2.72559315e-01 7.88734376e-01 8.67665946e-01 -1.05559361e+00 -4.43864077e-01 2.98494637e-01 -4.68805909e-01 1.26186681e+00 2.62746960e-01 3.93404871e-01 7.78759181e-01 3.39332670e-01 5.42901307e-02 2.36827955e-01 -1.10379016e+00 2.41711855e-01 6.78707659e-01 4.57032531e-01 6.84868991e-01 8.38941038e-02 4.14252669e-01 5.80507457e-01 -8.74917284e-02 -2.74282604e-01 6.66732416e-02 2.24549398e-01 -5.13081014e-01 -8.37614059e-01 -6.52108133e-01 3.87670368e-01 -7.47321665e-01 -5.69690943e-01 3.26353610e-01 4.93978888e-01 -2.65906811e-01 1.27314258e+00 -2.13122088e-03 -1.19139329e-01 7.45163485e-02 6.75865114e-01 4.06247318e-01 -5.16581178e-01 -3.84542078e-01 7.91456640e-01 3.15437824e-01 4.35577938e-03 -3.16407055e-01 -7.86969125e-01 -1.23791504e+00 5.97102381e-02 -8.12180281e-01 3.85229401e-02 1.32687008e+00 1.20011389e+00 2.13192284e-01 6.34006560e-01 4.06429684e-03 -8.75250578e-01 -1.08114707e+00 -1.75009799e+00 -3.89160246e-01 -6.57322764e-01 3.72640580e-01 -5.35851836e-01 -5.62840343e-01 1.12838531e-02]
[14.554274559020996, 6.313554286956787]
877df515-930a-4018-8c6e-5a45b742b1a7
generating-textual-explanations-for-machine
null
null
https://aclanthology.org/2022.lrec-1.379
https://aclanthology.org/2022.lrec-1.379.pdf
Generating Textual Explanations for Machine Learning Models Performance: A Table-to-Text Task
Numerical tables are widely employed to communicate or report the classification performance of machine learning (ML) models with respect to a set of evaluation metrics. For non-experts, domain knowledge is required to fully understand and interpret the information presented by numerical tables. This paper proposes a new natural language generation (NLG) task where neural models are trained to generate textual explanations, analytically describing the classification performance of ML models based on the metrics’ scores reported in the tables. Presenting the generated texts along with the numerical tables will allow for a better understanding of the classification performance of ML models. We constructed a dataset comprising numerical tables paired with their corresponding textual explanations written by experts to facilitate this NLG task. Experiments on the dataset are conducted by fine-tuning pre-trained language models (T5 and BART) to generate analytical textual explanations conditioned on the information in the tables. Furthermore, we propose a neural module, Metrics Processing Unit (MPU), to improve the performance of the baselines in terms of correctly verbalising the information in the corresponding table. Evaluation and analysis conducted indicate, that exploring pre-trained models for data-to-text generation leads to better generalisation performance and can produce high-quality textual explanations.
['Noura Al Moubayed', 'Amir Enshaei', 'James Burton', 'Isaac Ampomah']
null
null
null
null
lrec-2022-6
['data-to-text-generation']
['natural-language-processing']
[ 5.25831759e-01 8.99955869e-01 -1.35657638e-02 -6.64779067e-01 -1.01483059e+00 -5.26551425e-01 9.41442907e-01 6.09643757e-01 1.50335252e-01 9.83957648e-01 6.00192249e-01 -6.05589807e-01 -6.37073116e-03 -9.48439300e-01 -6.98317945e-01 -3.11722662e-02 1.62183255e-01 7.29790211e-01 -5.59287429e-01 -2.08825737e-01 7.18421578e-01 1.25486493e-01 -1.71003723e+00 1.19288039e+00 1.18380857e+00 1.21716428e+00 -2.06334027e-03 9.38919425e-01 -6.20959044e-01 8.48023176e-01 -1.22488523e+00 -7.96065509e-01 2.80971471e-02 -7.54997611e-01 -7.82686949e-01 -5.40079502e-03 4.06392455e-01 -1.78809375e-01 3.75981301e-01 5.74174464e-01 4.24710987e-03 -1.97888538e-02 1.31499910e+00 -1.46895194e+00 -1.11939919e+00 1.29664218e+00 1.36005461e-01 -2.79084623e-01 8.11447442e-01 1.36884287e-01 1.29565525e+00 -9.14117277e-01 4.13040221e-01 1.52200663e+00 4.64054734e-01 6.33467674e-01 -1.16217327e+00 -4.86449093e-01 1.77517254e-02 4.61214595e-02 -7.48994350e-01 -9.55664217e-02 4.25912589e-01 -4.50468987e-01 1.11101508e+00 4.60510164e-01 3.68996859e-01 1.12035942e+00 2.52608567e-01 4.65056568e-01 8.55629027e-01 -6.12041891e-01 2.10566431e-01 6.96335793e-01 -1.90881938e-01 4.82116878e-01 5.26214361e-01 -9.84692201e-02 -6.90704226e-01 1.95751414e-01 3.81578654e-01 -4.89899784e-01 -1.55265957e-01 6.87029504e-04 -1.43326485e+00 1.18309391e+00 6.25921428e-01 3.07901293e-01 -6.01482570e-01 -6.09757751e-03 5.39842010e-01 2.07897529e-01 4.63067442e-01 1.13580000e+00 -5.06553411e-01 -6.35562316e-02 -7.95118570e-01 5.31234384e-01 9.09772515e-01 1.08990359e+00 5.22602558e-01 8.57338160e-02 -7.57806242e-01 4.72430438e-01 3.94887477e-01 3.87943566e-01 7.94869721e-01 -7.96164572e-01 1.21834099e+00 1.15789068e+00 2.39955023e-01 -1.15168476e+00 -2.02088207e-01 -2.96476215e-01 -8.99099648e-01 1.09770492e-01 4.01761711e-01 -2.74860770e-01 -6.38918161e-01 1.50173855e+00 -5.14493287e-02 -6.91524923e-01 6.61575556e-01 6.48766458e-01 1.04167747e+00 1.03346896e+00 -1.91665441e-02 3.87111232e-02 1.20844758e+00 -5.81482768e-01 -7.84983039e-01 -2.02942297e-01 9.95813966e-01 -5.57749271e-01 1.58830178e+00 4.60775882e-01 -9.30420518e-01 -9.51902032e-01 -1.30721796e+00 -1.79851845e-01 -8.59096825e-01 5.72845876e-01 4.09319192e-01 6.51185572e-01 -9.65624034e-01 6.86390758e-01 -3.57036293e-01 -1.43286631e-01 2.46440530e-01 1.99350432e-01 -4.94432300e-02 3.70835155e-01 -1.36425447e+00 9.79724348e-01 7.91637003e-01 1.09168403e-02 -2.98614115e-01 -8.39450657e-01 -1.23384142e+00 2.52902448e-01 2.97747757e-02 -6.88630402e-01 1.50328529e+00 -6.42916262e-01 -1.12379718e+00 5.68360269e-01 2.99547929e-02 -7.90250659e-01 6.21603847e-01 -4.53466251e-02 -2.45535895e-01 8.65658186e-03 3.01485896e-01 1.15300453e+00 2.62658000e-01 -1.37236226e+00 -5.57794631e-01 2.15761270e-02 4.77553345e-02 1.12816557e-01 -2.33828947e-01 -5.24824202e-01 2.92319000e-01 -8.79260778e-01 -6.23100623e-02 -4.27794069e-01 -8.36639777e-02 -2.99311996e-01 -7.62441099e-01 -4.77907270e-01 3.08771402e-01 -8.63250494e-01 1.38964760e+00 -1.56726122e+00 -2.64407992e-01 2.61264056e-01 9.25575290e-03 -4.09042947e-02 3.68942916e-02 5.84374845e-01 -1.07588448e-01 4.67358887e-01 -3.22730511e-01 -2.36877039e-01 4.43009436e-01 2.26730350e-02 -5.84558427e-01 -5.26306987e-01 6.66722238e-01 8.27745080e-01 -6.52256906e-01 -3.52652818e-01 9.59718749e-02 1.48047715e-01 -5.85744083e-01 4.71745968e-01 -7.45853126e-01 1.03894807e-01 -1.49122119e-01 1.65461347e-01 1.48645788e-01 -2.40246788e-01 -5.25943749e-02 -1.94044307e-01 2.04683751e-01 8.05074573e-01 -1.10501909e+00 1.25450552e+00 -8.26871336e-01 7.90908039e-01 -8.80588770e-01 -6.17064416e-01 1.40345776e+00 3.04651439e-01 -3.16602916e-01 -7.48755872e-01 1.43557206e-01 1.78625986e-01 6.01330511e-02 -4.54846114e-01 7.45802104e-01 -1.27358466e-01 -3.18302393e-01 8.04098427e-01 -1.43324181e-01 -6.03992045e-01 7.22893655e-01 4.85517621e-01 5.63236952e-01 -7.31539652e-02 3.19853663e-01 6.48935065e-02 8.21947336e-01 2.81873584e-01 -2.57994294e-01 8.49769115e-01 7.32614994e-01 6.42522395e-01 8.04603159e-01 -3.85998547e-01 -1.18244433e+00 -8.17018926e-01 2.51716495e-01 8.08271050e-01 -5.46570063e-01 -7.05035985e-01 -8.57839048e-01 -7.30638087e-01 -1.30983055e-01 1.75395727e+00 -1.05130112e+00 -2.94590682e-01 -1.76679045e-01 -4.25147444e-01 3.28313142e-01 7.99171269e-01 3.40512574e-01 -1.47251916e+00 -8.44129324e-01 2.36312211e-01 -4.87788022e-01 -1.00801027e+00 -2.28823081e-01 3.28462660e-01 -1.01234269e+00 -8.88170481e-01 -2.72839218e-01 -3.52643609e-01 8.05222034e-01 -4.11017299e-01 1.22030294e+00 2.22352296e-01 3.06956302e-02 1.81757975e-02 -5.78321338e-01 -8.59121084e-01 -1.23245490e+00 1.60102248e-01 -2.70971268e-01 -1.84318557e-01 1.21487476e-01 -9.24209505e-02 -6.55383840e-02 -1.79510601e-02 -1.19916761e+00 5.37399292e-01 7.76643336e-01 6.68607295e-01 2.00432375e-01 -1.04062505e-01 6.55851603e-01 -9.77348566e-01 1.40421569e+00 -2.89315522e-01 -3.98175091e-01 4.55740750e-01 -8.70311439e-01 8.92403007e-01 1.08758771e+00 -3.67505372e-01 -1.06899071e+00 -4.28686976e-01 1.38150960e-01 2.52745211e-01 -2.64211744e-01 7.58470416e-01 -2.87613362e-01 7.18127668e-01 1.03750622e+00 4.02776599e-02 -2.31682714e-02 -1.37993291e-01 6.80179596e-01 8.02124321e-01 7.02744782e-01 -6.53538585e-01 8.17497849e-01 -2.93366373e-01 -4.98428904e-02 -6.93747401e-02 -1.07044566e+00 1.78815067e-01 -7.34288573e-01 -1.75545275e-01 6.36017621e-01 -3.77666235e-01 -4.77379054e-01 -2.24858463e-01 -1.38686287e+00 -3.30766141e-01 -2.61895835e-01 3.45862180e-01 -6.80709600e-01 -2.18922421e-01 -1.79660946e-01 -8.21661711e-01 -5.76634884e-01 -1.01401973e+00 1.03778028e+00 1.05043307e-01 -9.82774258e-01 -1.18580294e+00 -1.92118198e-01 4.85476077e-01 4.79213238e-01 4.43614751e-01 1.60663855e+00 -1.25438225e+00 -2.53523111e-01 -2.15422213e-01 -3.65695029e-01 4.12828833e-01 1.16121463e-01 1.89975679e-01 -8.65048885e-01 4.61931050e-01 -2.87258714e-01 -4.93495375e-01 4.32375342e-01 6.04865290e-02 1.36039984e+00 -1.12729108e+00 1.27896229e-02 2.66639087e-02 1.03131855e+00 3.14632297e-01 4.77693081e-01 4.43124861e-01 3.89464021e-01 1.04911053e+00 7.74161398e-01 5.20066321e-01 5.70909739e-01 3.49302590e-01 2.58325547e-01 2.93450598e-02 -2.16495879e-02 -5.41375458e-01 3.47794563e-01 5.56913912e-01 3.42202485e-01 -4.70793307e-01 -8.63222718e-01 3.53769839e-01 -1.43073738e+00 -7.52627015e-01 -2.85907149e-01 1.91975808e+00 1.02805281e+00 6.00965381e-01 -6.02947101e-02 6.51228487e-01 3.01234096e-01 -2.81681985e-01 -2.82815039e-01 -1.19940972e+00 6.87221885e-02 1.58221796e-01 -1.15239404e-01 5.21379769e-01 -5.43505251e-01 5.40028095e-01 5.57983398e+00 4.24220800e-01 -8.79617929e-01 -5.81525326e-01 9.52883959e-01 2.34920144e-01 -6.57460213e-01 -3.14109236e-01 -7.14122951e-01 1.66195855e-01 1.39554989e+00 -5.82410336e-01 1.94154456e-01 7.73521662e-01 2.97799975e-01 -9.90212709e-02 -1.72610807e+00 7.80490458e-01 3.05058330e-01 -1.50410628e+00 1.11907029e+00 -1.84499711e-01 7.26255596e-01 -8.99300396e-01 1.96895197e-01 4.95486706e-01 2.15255946e-01 -1.37998712e+00 9.84844029e-01 4.90186661e-01 7.77219951e-01 -8.60455871e-01 1.09721410e+00 3.58828753e-01 -7.78103173e-01 -7.85397515e-02 -3.90699863e-01 -3.66392255e-01 -2.71665990e-01 3.02444845e-01 -1.80638731e+00 6.46787465e-01 3.13034445e-01 4.23053622e-01 -1.14372611e+00 5.12902856e-01 -6.16061866e-01 4.30336416e-01 1.19109340e-01 -5.95100582e-01 4.10407871e-01 5.20570613e-02 -9.79658812e-02 1.24985087e+00 6.34006917e-01 -1.44605059e-02 -2.98971981e-01 1.53729212e+00 -2.85540491e-01 3.11169147e-01 -5.30866742e-01 -3.63526732e-01 2.21312672e-01 1.07802117e+00 -7.30250359e-01 -8.00856054e-01 -6.44194931e-02 4.33095872e-01 8.97289664e-02 3.46816629e-02 -6.43333554e-01 -6.66564941e-01 1.47336394e-01 2.14590088e-01 -3.05197258e-02 2.18389511e-01 -1.09188318e+00 -7.27766633e-01 2.41518289e-01 -1.19254494e+00 4.69478846e-01 -1.33377290e+00 -1.12115383e+00 9.56204832e-01 2.87912518e-01 -1.19031703e+00 -1.09014773e+00 -7.31268823e-01 -6.58465505e-01 1.10030842e+00 -9.52150643e-01 -7.25631118e-01 -5.28462529e-01 3.54645848e-02 6.95875227e-01 -2.53821343e-01 9.32480931e-01 -4.19582188e-01 -7.02905878e-02 7.25989282e-01 -4.08706754e-01 2.45497748e-01 4.41481769e-01 -1.68074834e+00 7.39738345e-01 7.04531908e-01 4.11029398e-01 7.01359212e-01 1.04564393e+00 -7.46048212e-01 -6.83764637e-01 -1.14581084e+00 1.32270515e+00 -5.60910761e-01 4.86810714e-01 -4.58192676e-01 -9.80233133e-01 5.00679135e-01 3.42494965e-01 -9.18860197e-01 7.46741951e-01 -1.97317809e-01 -3.16869557e-01 5.98068722e-02 -1.17810452e+00 5.62741756e-01 4.11902159e-01 -3.13178778e-01 -1.08469248e+00 3.83505225e-01 8.38601291e-01 -3.85986775e-01 -6.17914438e-01 2.40022764e-01 3.24601233e-01 -7.96032906e-01 6.43789351e-01 -7.99679518e-01 1.30240774e+00 -2.03257516e-01 -6.99596703e-02 -1.68789470e+00 3.27192783e-01 -3.11458230e-01 4.36523966e-02 1.39418757e+00 1.21558368e+00 -3.16332161e-01 5.49945951e-01 8.36719692e-01 -3.82319055e-02 -7.44444430e-01 -3.81816864e-01 -4.92647976e-01 4.99720909e-02 -6.95753276e-01 1.00831938e+00 5.50017953e-01 2.73306489e-01 5.33193350e-01 1.31764740e-01 -6.10228442e-02 2.82811403e-01 8.18261504e-02 9.11497653e-01 -1.16513741e+00 1.25079125e-01 -6.59552217e-01 -1.45435840e-01 -5.97594380e-01 -3.87659892e-02 -1.18251240e+00 1.58336207e-01 -2.11376953e+00 -6.92483038e-02 -1.10578194e-01 2.09664643e-01 5.25937796e-01 -3.34274918e-01 -4.88338098e-02 3.91558588e-01 -2.79791981e-01 -2.99918264e-01 5.21328270e-01 1.17937326e+00 -1.30665302e-01 -2.54141122e-01 4.57177535e-02 -1.16268086e+00 3.63529563e-01 8.83115709e-01 -5.51095009e-01 -5.17064691e-01 -2.03267708e-01 5.02959788e-01 2.90251762e-01 2.96220571e-01 -1.27497888e+00 5.60264885e-02 -6.96258396e-02 9.27543283e-01 -8.71383429e-01 1.28064388e-02 -4.86171484e-01 -1.58124283e-01 4.78144377e-01 -1.29377449e+00 6.38667464e-01 3.74978960e-01 -1.40959360e-02 -2.77271837e-01 -3.29683781e-01 2.68535018e-01 -1.17199898e-01 4.01967391e-03 -4.26642239e-01 -3.75532120e-01 2.16612324e-01 5.71020305e-01 -2.66575366e-01 -3.45683217e-01 -7.14029610e-01 -5.18057227e-01 2.36120850e-01 -2.09037550e-02 6.52862310e-01 7.79149711e-01 -1.36269200e+00 -9.60353673e-01 1.62240386e-01 3.64924401e-01 -5.47365323e-02 -2.47840032e-01 1.28649771e-01 -5.79839051e-01 1.01529586e+00 -3.34189683e-01 -2.92423368e-01 -8.11987281e-01 4.19836193e-01 1.80251077e-01 -6.63321555e-01 -3.24702501e-01 6.55172408e-01 -1.20679915e-01 -7.66223669e-01 4.21481729e-01 -1.20936942e+00 -4.86089647e-01 2.28744730e-01 7.20954061e-01 2.94795603e-01 3.03480089e-01 -1.98294938e-01 2.01924458e-01 5.67874797e-02 5.81632107e-02 -5.23469865e-01 1.18255496e+00 1.99716195e-01 2.22256362e-01 7.42370844e-01 9.17762399e-01 -1.91919997e-01 -8.40428829e-01 3.72690968e-02 4.71115172e-01 -8.68392885e-02 -4.88728285e-01 -1.46984184e+00 -4.74217385e-01 1.16447115e+00 -5.00924736e-02 3.00838172e-01 8.50581706e-01 -1.46443620e-01 4.53849792e-01 7.43380427e-01 -1.23441562e-01 -7.94004500e-01 3.52919966e-01 3.90521139e-01 1.52568889e+00 -1.22892714e+00 -9.37961787e-02 -2.23821267e-01 -9.86795783e-01 1.62327254e+00 8.79127681e-01 3.23776990e-01 -2.28317082e-01 1.25211123e-02 3.38152021e-01 -1.12130105e-01 -1.17977953e+00 2.79186875e-01 8.82820368e-01 6.69520795e-01 7.62266576e-01 1.19334310e-01 -9.66973379e-02 1.00279891e+00 -1.27608800e+00 -3.23775440e-01 8.54065716e-01 3.63336384e-01 -2.95152068e-01 -9.54564810e-01 -6.56916678e-01 8.44568491e-01 -1.29515454e-01 -2.45974317e-01 -9.89050806e-01 8.42942655e-01 -7.68946111e-02 1.25330603e+00 -7.69818574e-02 -3.30017716e-01 3.83776277e-01 4.21543866e-01 6.60676062e-02 -9.02424574e-01 -8.80703926e-01 -6.42692089e-01 3.46424967e-01 -2.19287306e-01 1.92556903e-02 -4.87346798e-01 -1.52470803e+00 -1.27436727e-01 -8.64567794e-03 5.98517776e-01 8.84057581e-01 1.08095217e+00 -1.18966075e-02 8.56594265e-01 3.43683451e-01 -4.98880476e-01 -8.16051066e-01 -1.42164576e+00 5.47999777e-02 5.51759779e-01 1.11519203e-01 -3.20678830e-01 -4.47039515e-01 2.35783592e-01]
[11.495943069458008, 8.80200481414795]
13b78ab2-e67e-485f-a324-2670066b715f
iterative-thresholded-bi-histogram
1508.05704
null
http://arxiv.org/abs/1508.05704v1
http://arxiv.org/pdf/1508.05704v1.pdf
Iterative Thresholded Bi-Histogram Equalization for Medical Image Enhancement
Enhancement of human vision to get an insight to information content is of vital importance. The traditional histogram equalization methods have been suffering from amplified contrast with the addition of artifacts and a surprising unnatural visibility of the processed images. In order to overcome these drawbacks, this paper proposes interative, mean, and multi-threshold selection criterion with plateau limits, which consist of histogram segmentation, clipping and transformation modules. The histogram partition consists of multiple thresholding processes that divide the histogram into two parts, whereas the clipping process nicely enhances the contrast by having a check on the rate of enhancement that could be tuned. Histogram equalization to each segmented sub-histogram provides the output image with preserved brightness and enhanced contrast. Results of the present study showed that the proposed method efficiently handles the noise amplification. Further, it also preserves the brightness by retaining natural look of targeted image.
['Muhammad Ali Qadar', 'Li Hua', 'Yan Zhaowen']
2015-08-24
null
null
null
null
['medical-image-enhancement']
['computer-vision']
[ 4.08436835e-01 -2.69666582e-01 4.15652066e-01 -2.66447067e-01 -2.44866908e-01 -3.31265211e-01 3.70216161e-01 3.97645414e-01 -6.90877318e-01 6.45658553e-01 -1.79520790e-02 -9.36548188e-02 -7.55154788e-02 -7.60951042e-01 -1.82268098e-01 -1.20993400e+00 4.05469202e-02 -3.91518742e-01 6.99239969e-01 -1.54341415e-01 5.73728323e-01 4.97919053e-01 -2.10317826e+00 1.71580985e-01 9.73994792e-01 1.03049827e+00 2.40457490e-01 7.06970155e-01 6.27071261e-02 5.92146873e-01 -5.38083732e-01 -1.42323509e-01 3.27875257e-01 -6.11476004e-01 -3.76024723e-01 4.01514262e-01 2.53822058e-01 -4.12218839e-01 6.42270176e-03 1.41858995e+00 6.39019251e-01 2.17637703e-01 5.74579835e-01 -1.11386836e+00 -5.05941808e-01 1.03074148e-01 -8.80356908e-01 5.33193648e-01 6.61405325e-02 2.64466345e-01 1.83695421e-01 -6.16671205e-01 2.80091316e-01 8.60126913e-01 3.86946648e-01 1.25542149e-01 -1.10502708e+00 -4.71606582e-01 -5.02539873e-01 3.78945380e-01 -1.28400338e+00 -3.13089043e-01 7.05655992e-01 -2.54115075e-01 4.91106004e-01 4.97012287e-01 8.72101367e-01 9.21578333e-02 5.93398035e-01 5.60913682e-02 1.63255012e+00 -6.30484164e-01 -6.55333931e-03 5.00532389e-01 1.97527602e-01 4.30908829e-01 5.00439286e-01 1.70625463e-01 1.10972570e-02 1.86722949e-01 5.77702105e-01 -1.20929055e-01 -5.34348011e-01 -2.02034578e-01 -8.79210889e-01 3.91681224e-01 3.00255567e-01 5.96897542e-01 -5.09195924e-01 -5.44331551e-01 3.52695882e-01 7.50778243e-02 -6.73045814e-02 1.94714397e-01 -9.58325714e-02 1.55005977e-01 -1.04494143e+00 -9.86350998e-02 3.19445640e-01 4.75293934e-01 7.52848983e-01 3.25855352e-02 -2.01560423e-01 4.76132333e-01 -3.36510502e-03 5.42393088e-01 6.33926928e-01 -6.40642583e-01 -6.57887533e-02 5.76280832e-01 1.82230413e-01 -1.21713066e+00 -3.33130002e-01 -3.58173847e-01 -9.06138778e-01 8.05247486e-01 5.24899542e-01 1.61589682e-02 -1.05884695e+00 1.38640094e+00 3.81844193e-01 -5.70676029e-01 3.77006195e-02 1.07712233e+00 7.29027212e-01 8.92248094e-01 2.61739612e-01 -6.31631672e-01 1.51278985e+00 -6.10474110e-01 -1.14286757e+00 1.69529408e-01 -2.49130607e-01 -1.09138000e+00 1.01900411e+00 5.10340035e-01 -1.37771201e+00 -9.23560202e-01 -1.16936457e+00 6.99748844e-03 -4.94672030e-01 6.83592185e-02 1.30979791e-01 6.99345767e-01 -9.95459259e-01 2.40906253e-01 -3.55772585e-01 -3.03093970e-01 -8.66111070e-02 3.07931751e-01 -3.32806736e-01 1.64654508e-01 -1.03577638e+00 9.80961025e-01 8.22658777e-01 3.10974628e-01 -3.49571854e-01 -2.03187272e-01 -7.39368260e-01 3.26360971e-01 1.35523705e-02 -3.76354039e-01 6.83341265e-01 -1.14483809e+00 -1.18602085e+00 7.92185783e-01 -9.41933468e-02 -9.74729583e-02 4.94157881e-01 2.53900826e-01 -4.79172140e-01 3.72755617e-01 -1.88896716e-01 4.79862094e-01 9.82842088e-01 -1.24202335e+00 -9.01721179e-01 -4.36082631e-01 -5.84906042e-01 3.93929631e-01 -3.50085825e-01 5.28434105e-02 -4.01218683e-01 -3.09435725e-01 1.74432218e-01 -2.60938376e-01 -1.04709290e-01 -3.59172463e-01 -9.98790562e-02 3.11771810e-01 1.00466311e+00 -9.44924355e-01 1.41263938e+00 -2.39855218e+00 -4.20492560e-01 4.22029883e-01 1.30307883e-01 2.64144331e-01 3.49349976e-01 1.64296538e-01 -1.01028392e-02 -1.57771975e-01 -3.56652558e-01 3.36931229e-01 -3.34092557e-01 -1.63233131e-01 1.59957081e-01 6.03406668e-01 -1.49038777e-01 1.55119702e-01 -5.38062334e-01 -1.04015815e+00 4.64995980e-01 7.98400164e-01 -1.82121515e-01 2.07950190e-01 4.49702322e-01 3.29582423e-01 1.88896488e-02 5.95483303e-01 1.11771798e+00 1.75351590e-01 5.25606237e-02 -4.88548726e-01 -6.57990813e-01 -5.76713800e-01 -1.36185396e+00 5.57265341e-01 1.51853397e-01 6.58100605e-01 2.43698582e-01 -5.08729100e-01 1.04099262e+00 2.10163653e-01 3.50854844e-01 -9.87879813e-01 6.60672307e-01 2.68004149e-01 7.59151056e-02 -7.28824615e-01 7.97875524e-01 -3.83257747e-01 3.86849016e-01 -8.82967114e-02 -3.18916172e-01 2.74856836e-02 3.26272100e-01 -2.80947268e-01 2.56289393e-01 -2.07139298e-01 5.88108838e-01 -2.81204611e-01 8.67664754e-01 -1.09888107e-01 4.10872966e-01 4.26732183e-01 -5.38491964e-01 3.67436707e-01 2.16383621e-01 -3.28911282e-02 -1.24754846e+00 -9.82778430e-01 -4.75540996e-01 7.38096893e-01 6.39692247e-01 3.80186915e-01 -8.62664044e-01 3.13983597e-02 -2.33313411e-01 5.14991403e-01 -4.17499453e-01 -1.57610044e-01 -4.15822327e-01 -9.99683321e-01 1.57036409e-01 7.37138987e-02 1.16750598e+00 -1.19895828e+00 -1.12694550e+00 6.16551824e-02 -2.41660967e-01 -7.16308236e-01 -3.93127650e-01 1.41580179e-01 -8.90952468e-01 -9.16667998e-01 -8.24125230e-01 -1.23498905e+00 8.14254403e-01 3.97956014e-01 5.15413523e-01 2.27525353e-01 -3.87399882e-01 -1.04976498e-01 -2.21005157e-01 -1.87674195e-01 -3.50791961e-01 -3.92974019e-01 -3.94500703e-01 1.55099913e-01 3.18442374e-01 -4.54798996e-01 -9.38117981e-01 3.14638019e-01 -1.15112805e+00 -2.37170592e-01 9.20159996e-01 6.33142710e-01 6.10003173e-01 7.33186126e-01 2.79715061e-01 -4.85012680e-01 7.43124962e-01 -8.12402740e-02 -7.09560692e-01 1.93530768e-01 -8.75843108e-01 -1.47488147e-01 6.56263351e-01 -1.96990505e-01 -1.55935526e+00 -8.83547142e-02 1.04389504e-01 1.66975737e-01 -4.03039455e-01 7.78502375e-02 -1.98851302e-01 -3.37232351e-01 5.70363402e-01 6.93913817e-01 3.07878792e-01 -1.76605374e-01 2.34716684e-01 7.93283343e-01 1.02222049e+00 9.30478275e-02 6.80095434e-01 3.67305100e-01 1.31595165e-01 -1.01976967e+00 -1.22000650e-01 -6.73864961e-01 -5.45826137e-01 -6.46800280e-01 9.97947752e-01 -3.07059675e-01 -7.70958543e-01 7.58783221e-01 -8.34442556e-01 1.22198857e-01 8.56545940e-02 5.30729353e-01 -2.22215503e-01 8.07630301e-01 -5.78299522e-01 -9.74894702e-01 -4.25563157e-01 -1.01322925e+00 3.76451880e-01 9.49032068e-01 4.11996812e-01 -7.69588530e-01 -1.77907005e-01 2.14978367e-01 6.85091615e-01 4.07096326e-01 9.03489232e-01 -2.25445017e-01 -4.18742865e-01 -1.81601867e-01 -5.06255031e-01 3.66653115e-01 1.89581677e-01 1.57386616e-01 -9.34447825e-01 -2.81116039e-01 2.60447174e-01 1.81460679e-02 8.50159049e-01 7.39278316e-01 8.83300960e-01 -1.70419618e-01 -2.42140759e-02 5.80403149e-01 1.92628229e+00 7.96840072e-01 1.21219552e+00 8.57066751e-01 -6.85554072e-02 6.32702768e-01 8.14621687e-01 4.11047012e-01 -7.77470991e-02 1.70614943e-01 4.18624520e-01 -7.95754254e-01 -1.53279215e-01 8.71666968e-02 1.21992128e-03 4.50922221e-01 -2.16222897e-01 -2.19207585e-01 -4.01080191e-01 5.97814620e-01 -1.19776452e+00 -1.24834323e+00 -4.12864566e-01 2.41230392e+00 8.57460022e-01 2.81620324e-01 1.64166942e-01 4.82658654e-01 1.17273033e+00 -6.63874578e-03 -1.81541011e-01 -6.59875393e-01 -3.68701011e-01 8.91277418e-02 7.30796218e-01 5.42009830e-01 -1.09464049e+00 4.00756210e-01 6.25530005e+00 6.95707440e-01 -1.19118655e+00 -2.35470623e-01 5.38017035e-01 4.81440216e-01 2.87981494e-03 -1.16824307e-01 -4.97205079e-01 6.55987322e-01 4.35574859e-01 -1.70589462e-01 1.27483130e-01 5.15045702e-01 4.26462144e-01 -8.72201264e-01 -7.11974651e-02 8.04213166e-01 4.43049073e-02 -4.62806880e-01 2.01908648e-01 1.18329473e-01 5.01943588e-01 -8.34846258e-01 3.90078604e-01 -1.29967630e-01 -4.47907120e-01 -7.95590580e-01 5.58916032e-01 7.17069566e-01 4.73505676e-01 -1.02964318e+00 1.14838839e+00 1.81428805e-01 -1.14880979e+00 -1.76952109e-01 -3.47959965e-01 -1.46663666e-01 9.45742652e-02 4.85089004e-01 -6.54916644e-01 3.43320787e-01 7.12156057e-01 -1.54056877e-01 -8.41775239e-01 1.65063012e+00 -7.28123784e-02 2.59902239e-01 -1.82277352e-01 7.07109272e-02 1.68037921e-01 -6.11354947e-01 4.96929079e-01 1.28420138e+00 2.68572032e-01 2.63100922e-01 -1.06745586e-01 6.20058000e-01 4.98703480e-01 4.74170566e-01 -3.71828198e-01 3.08885962e-01 4.48999584e-01 1.27665865e+00 -1.09050202e+00 -3.91234368e-01 -3.11668903e-01 9.38833356e-01 -3.72384042e-01 3.65989059e-01 -8.37878942e-01 -1.04359293e+00 -1.14684150e-01 1.64769083e-01 2.79141724e-01 9.91043355e-03 -4.36602205e-01 -4.96693045e-01 -2.66818777e-02 -7.25500524e-01 5.09348214e-01 -7.65147209e-01 -5.68582594e-01 6.25589967e-01 6.57648370e-02 -1.08899176e+00 3.19879085e-01 -3.62169147e-01 -5.61195493e-01 9.91004229e-01 -1.44864047e+00 -8.70925844e-01 -6.51854098e-01 5.59272528e-01 3.89701962e-01 1.87117934e-01 1.81446344e-01 4.03012127e-01 -3.95547032e-01 4.51888591e-01 2.01083720e-01 -1.02566563e-01 7.45150566e-01 -1.28015888e+00 -6.04466021e-01 1.20068169e+00 -8.02329481e-01 3.41486782e-01 1.22973061e+00 -5.94605625e-01 -4.31241423e-01 -4.91589397e-01 9.73023176e-01 3.14504415e-01 1.10517733e-01 2.83195376e-01 -1.09899402e+00 1.77842513e-01 6.11493766e-01 -4.38922256e-01 3.72029096e-01 -5.79652846e-01 2.20228180e-01 -2.29710519e-01 -1.38355148e+00 4.59591180e-01 5.86120039e-02 -2.76918802e-02 -6.88768506e-01 -3.03433418e-01 2.01683827e-02 -1.38822362e-01 -7.74508178e-01 3.37372810e-01 7.85566568e-01 -1.60734773e+00 7.66255736e-01 4.36173938e-02 2.32027367e-01 -6.35266125e-01 1.65433139e-01 -8.51580083e-01 -5.00036836e-01 -2.28658125e-01 8.65938663e-01 1.41238022e+00 3.32944393e-01 -5.31814396e-01 3.52765143e-01 4.61536497e-01 -5.22690602e-02 -4.31670785e-01 -4.73869860e-01 -4.73658711e-01 -4.77167875e-01 3.84843022e-01 2.01281741e-01 6.75429583e-01 1.60812065e-01 -1.38457268e-01 -3.74505877e-01 4.18308616e-01 1.01959229e+00 -2.31453199e-02 4.05824661e-01 -9.44784224e-01 8.90552253e-02 -6.29554510e-01 -4.98372465e-01 -4.34717059e-01 -6.39404237e-01 -3.25661451e-01 1.91663742e-01 -1.56724370e+00 5.30742407e-01 2.66924381e-01 -3.74762297e-01 2.13045269e-01 -4.65576738e-01 4.53801572e-01 4.44061160e-02 2.54010141e-01 -3.48721296e-01 3.48650038e-01 1.54681754e+00 1.09313101e-01 -4.57038522e-01 -7.17145801e-02 -4.10594851e-01 5.67033529e-01 8.96422446e-01 -1.49507254e-01 -3.74116153e-01 2.11734772e-01 -1.82529524e-01 -1.86816156e-02 3.68248910e-01 -1.15906453e+00 3.60028952e-01 -6.62097633e-02 6.65925622e-01 -6.71790481e-01 3.22874300e-02 -9.49016809e-01 1.30966336e-01 6.30989790e-01 -1.30573943e-01 2.06281975e-01 2.43872449e-01 3.70524734e-01 -4.87803698e-01 -3.04044068e-01 1.42878640e+00 -1.17781885e-01 -1.05759621e+00 -3.28933507e-01 -7.29733825e-01 -4.55751717e-01 1.22704613e+00 -9.67364013e-01 -3.84933561e-01 -3.25905412e-01 -6.67013407e-01 2.19630077e-02 5.93254626e-01 -3.28374296e-01 6.91469848e-01 -9.52695966e-01 -3.80470932e-01 4.36239213e-01 -1.36143997e-01 -5.45464993e-01 5.60983539e-01 1.13975370e+00 -8.83641779e-01 1.92365479e-02 -9.99751270e-01 -3.17763180e-01 -1.76508987e+00 8.75917554e-01 4.39291716e-01 -5.26475012e-02 -5.14854133e-01 3.62587601e-01 -6.85253888e-02 4.04573470e-01 2.58464068e-01 -1.57277912e-01 -7.41608441e-01 5.53918779e-02 5.47986329e-01 8.61953378e-01 -1.09577879e-01 -7.14891493e-01 -4.98958640e-02 8.05535316e-01 1.07464612e-01 -1.77452654e-01 8.38848650e-01 -6.35601103e-01 -4.48652148e-01 1.89195454e-01 1.01420820e+00 3.52952987e-01 -1.08038044e+00 3.28180015e-01 -2.94709891e-01 -6.68590426e-01 1.98956206e-01 -1.00393283e+00 -8.15793753e-01 7.27954090e-01 9.85432982e-01 6.73233926e-01 1.85660470e+00 -4.75074202e-01 7.90172219e-01 -2.76356310e-01 -4.72525731e-02 -1.16109288e+00 -2.45715693e-01 -7.35302269e-02 4.57601756e-01 -1.08857274e+00 8.58237669e-02 -5.00466049e-01 -6.35959208e-01 1.44871676e+00 5.31890035e-01 2.16941878e-01 4.30789590e-01 2.29282320e-01 2.37705812e-01 -1.90336123e-01 -1.75781831e-01 -5.27281761e-01 2.88986623e-01 6.01280272e-01 2.73229629e-01 -2.93721437e-01 -1.05167270e+00 2.48718500e-01 -1.53179213e-01 -8.76231566e-02 6.01988971e-01 8.31391096e-01 -1.34354937e+00 -5.12308419e-01 -9.02742088e-01 9.27267820e-02 -5.69685042e-01 1.09694906e-01 -1.96735471e-01 1.14169955e+00 4.99458015e-01 1.12648439e+00 -1.84817016e-02 -2.42291734e-01 4.06242549e-01 -2.94362251e-02 3.69960934e-01 2.18008220e-01 -5.29315054e-01 4.58987445e-01 -3.10309827e-01 -1.40603989e-01 -4.85602260e-01 -3.52242678e-01 -1.30026019e+00 -2.42972970e-01 -3.30491543e-01 4.91148472e-01 6.09608173e-01 5.22054791e-01 -2.51635581e-01 4.68642712e-01 5.83742440e-01 -4.72929567e-01 -4.32350367e-01 -9.47468340e-01 -1.01617312e+00 6.17232561e-01 5.01792550e-01 -4.19027328e-01 -8.37471247e-01 3.57142657e-01]
[10.91064167022705, -2.3980486392974854]
94ccbfcf-7341-4a72-9f2f-7971ebe2b837
prediction-of-adverse-biological-effects-of
2112.04605
null
https://arxiv.org/abs/2112.04605v2
https://arxiv.org/pdf/2112.04605v2.pdf
Prediction of Adverse Biological Effects of Chemicals Using Knowledge Graph Embeddings
We have created a knowledge graph based on major data sources used in ecotoxicological risk assessment. We have applied this knowledge graph to an important task in risk assessment, namely chemical effect prediction. We have evaluated nine knowledge graph embedding models from a selection of geometric, decomposition, and convolutional models on this prediction task. We show that using knowledge graph embeddings can increase the accuracy of effect prediction with neural networks. Furthermore, we have implemented a fine-tuning architecture that adapts the knowledge graph embeddings to the effect prediction task and leads to better performance. Finally, we evaluate certain characteristics of the knowledge graph embedding models to shed light on the individual model performance.
['Knut Erik Tollefsen', 'Raoul Wolf', 'Jiaoyan Chen', 'Ernesto Jiménez-Ruiz', 'Erik B. Myklebust']
2021-12-08
null
null
null
null
['knowledge-graph-embeddings', 'knowledge-graph-embeddings']
['graphs', 'methodology']
[ 4.89725955e-02 2.12904945e-01 -1.80922985e-01 -2.12820247e-02 -1.86774075e-01 -6.30951643e-01 5.11265874e-01 8.63822937e-01 -3.53926688e-01 4.21671212e-01 6.48540914e-01 -4.46927100e-01 -6.09694600e-01 -1.25103426e+00 -7.82837391e-01 -3.79726797e-01 -4.06851321e-01 1.98716670e-01 3.74777108e-01 -3.08740735e-01 1.28593624e-01 7.18511462e-01 -9.50754285e-01 3.51937145e-01 4.91084486e-01 7.09763348e-01 -1.97928250e-01 6.13214612e-01 1.67290658e-01 6.40855849e-01 -4.41811532e-01 -6.63227439e-01 -2.42529940e-02 5.06453589e-02 -7.22818851e-01 -9.45819676e-01 4.01072562e-01 -5.27190939e-02 -5.90622663e-01 6.37567043e-01 7.39271760e-01 2.24959806e-01 1.30826020e+00 -1.01948023e+00 -1.13727057e+00 5.25558710e-01 -6.25712126e-02 3.19371223e-01 3.36503655e-01 2.03546166e-01 1.10583115e+00 -6.55158579e-01 6.33304179e-01 1.29885542e+00 1.08735967e+00 5.03815830e-01 -1.35725772e+00 -5.99696040e-01 7.52010047e-02 6.25303626e-01 -1.28122163e+00 -1.94787066e-02 5.54849386e-01 -7.53429830e-01 1.48904347e+00 1.74772646e-02 7.75487900e-01 1.02825391e+00 7.47973859e-01 3.60910773e-01 6.77955210e-01 -3.19140702e-02 3.18395168e-01 -2.17411205e-01 3.94171268e-01 7.42569625e-01 6.15053713e-01 5.16490161e-01 -4.59939688e-01 -3.14254880e-01 4.39580113e-01 -1.41831450e-02 -4.55668449e-01 -5.06564602e-02 -5.67072988e-01 1.24513352e+00 9.24959421e-01 1.24855988e-01 -1.55452907e-01 9.32912946e-01 4.78442520e-01 2.53147393e-01 7.94316649e-01 8.89536917e-01 -8.98796856e-01 4.93685663e-01 -3.13047737e-01 -4.31135334e-02 1.02204740e+00 4.22409981e-01 8.46463978e-01 1.29337162e-01 -3.85930389e-01 7.73754179e-01 5.71920156e-01 4.94023561e-01 1.37090757e-01 -1.18418954e-01 8.85840226e-03 8.14908266e-01 -1.51926726e-01 -1.29575920e+00 -8.58563304e-01 -2.61839569e-01 -1.40501916e-01 2.09221348e-01 1.27499118e-01 -2.42500737e-01 -1.11478388e+00 1.41324246e+00 2.03548953e-01 2.61671394e-01 9.96581763e-02 4.50920522e-01 1.21926439e+00 7.16138184e-01 8.81572187e-01 4.08453852e-01 1.19291127e+00 -6.87242925e-01 -3.90685409e-01 1.56791151e-01 9.14525151e-01 -2.39424050e-01 6.72410011e-01 1.94288164e-01 -3.00523609e-01 -3.19222987e-01 -1.22630715e+00 1.79403089e-02 -1.29220831e+00 -3.07339370e-01 9.48389351e-01 9.33261871e-01 -7.84859240e-01 9.79954600e-01 -6.06983542e-01 -4.76208746e-01 5.10023952e-01 5.42519391e-01 -5.37652493e-01 -1.92932412e-01 -1.64078665e+00 1.38155401e+00 5.84134519e-01 -1.66814968e-01 -1.24517834e+00 -1.29876566e+00 -1.06345439e+00 3.45541924e-01 -5.21231182e-02 -7.47871697e-01 6.92607999e-01 -1.06071822e-01 -1.26102877e+00 1.20025113e-01 5.86744606e-01 -4.05349761e-01 -1.09579623e-01 2.04702187e-02 -6.78677320e-01 5.23925312e-02 -3.18873227e-01 5.83872318e-01 3.65417510e-01 -1.01441646e+00 -1.77244782e-01 -2.77291179e-01 9.21744257e-02 2.95189191e-02 -8.18499625e-01 -2.17714995e-01 -1.58929512e-01 -4.86072570e-01 -7.37763286e-01 -6.94897354e-01 -3.98262084e-01 -7.23007545e-02 -3.32452469e-02 -2.32266337e-01 5.05956888e-01 -6.48059785e-01 1.23523629e+00 -1.97725999e+00 1.01801947e-01 3.58826369e-01 4.34048474e-01 2.29940534e-01 -6.42809749e-01 9.35430527e-01 -3.07041019e-01 6.00942969e-01 -2.11996734e-01 3.31648082e-01 1.47777289e-01 6.08723201e-02 -2.01227784e-01 5.45115352e-01 2.61221200e-01 1.20956397e+00 -1.11055958e+00 -1.40325800e-01 2.29473874e-01 7.90161729e-01 -6.78870857e-01 -2.11175494e-02 -6.38701379e-01 -1.33177489e-01 -6.50236785e-01 3.15763295e-01 5.31393707e-01 -7.10379481e-02 2.89788395e-01 -4.69832033e-01 1.15349092e-01 -1.22311823e-02 -6.33610904e-01 1.42710149e+00 -6.49452269e-01 4.41695184e-01 -6.42906308e-01 -5.13920009e-01 6.27363622e-01 2.65135854e-01 4.84581888e-01 -5.68342388e-01 2.59748623e-02 -1.16010346e-01 1.08042978e-01 -4.13411915e-01 3.84071469e-01 -2.68661201e-01 5.10199554e-02 3.20926815e-01 5.31045437e-01 6.53052032e-02 3.74696180e-02 5.45845218e-02 1.50730503e+00 -1.33892000e-01 2.07514584e-01 -3.93898815e-01 1.22138917e-01 1.66991264e-01 1.25779048e-01 3.42165887e-01 4.75662760e-02 2.01535299e-01 6.77706420e-01 -7.49764860e-01 -4.63775694e-01 -9.91715550e-01 -3.54644179e-01 1.29535186e+00 -1.34947881e-01 -6.95288897e-01 -3.45170617e-01 -9.75736320e-01 5.52259028e-01 6.34037614e-01 -1.24926829e+00 -6.49053752e-01 -4.19355696e-03 -1.25514030e+00 9.55732405e-01 8.32135558e-01 -6.45300746e-02 -9.54950929e-01 -3.69256288e-02 3.05145353e-01 6.04439020e-01 -5.97629786e-01 -2.68177807e-01 7.01258063e-01 -5.19312203e-01 -1.62110233e+00 -2.39979595e-01 -6.57707036e-01 3.30383509e-01 -1.96062356e-01 8.89938354e-01 1.34783998e-01 -2.57963628e-01 7.59734690e-01 -5.22220075e-01 -8.19856465e-01 -4.00543749e-01 8.83548707e-03 -8.77633020e-02 -3.31657231e-01 3.08454901e-01 -3.65391046e-01 -6.15257323e-01 1.61398590e-01 -1.23265409e+00 -7.27839828e-01 5.65356195e-01 5.36927640e-01 5.32950282e-01 1.88618481e-01 8.29319954e-01 -1.02560103e+00 8.71728957e-01 -6.82658195e-01 -3.52085263e-01 3.68325263e-01 -7.95139432e-01 5.10608196e-01 6.69489145e-01 -2.91152328e-01 -6.96863532e-01 1.35151297e-01 -4.66123343e-01 -2.28881806e-01 1.70606256e-01 1.10884833e+00 -6.52143806e-02 -5.90983212e-01 1.14445853e+00 -2.15532184e-01 -2.71510154e-01 -4.40471500e-01 7.85529077e-01 2.02047631e-01 -2.05336973e-01 -3.46874416e-01 6.27136230e-01 -1.71178803e-02 3.96998584e-01 -8.51980209e-01 -4.02077138e-01 -2.25833192e-01 -4.26401436e-01 3.77160870e-02 1.17622316e+00 -8.05908740e-01 -7.93346465e-01 8.03728774e-02 -1.27514505e+00 -6.31814063e-01 -9.81853250e-03 3.28512222e-01 -1.90664455e-02 3.19201261e-01 -6.59083486e-01 -2.73504853e-01 -4.56386894e-01 -7.52634704e-01 6.41404808e-01 -3.64345461e-02 -1.50826452e-02 -1.72055197e+00 6.35738909e-01 -3.94322217e-01 4.37023938e-01 6.35080040e-01 1.38706100e+00 -8.43863666e-01 -1.53417051e-01 -2.31572539e-01 -2.54198134e-01 -1.02109471e-02 1.04498953e-01 3.06680109e-02 -1.12809181e+00 -6.02217503e-02 -6.82538569e-01 -2.21921638e-01 1.52568996e+00 3.68739247e-01 1.01409304e+00 -2.19159648e-01 -6.07478499e-01 6.64043069e-01 1.70636666e+00 2.13057622e-01 7.92163551e-01 4.41874228e-02 9.98283565e-01 4.84300733e-01 -6.84537664e-02 1.26248389e-01 2.41326317e-01 5.37627220e-01 6.81023657e-01 2.55727749e-02 -3.79033118e-01 -4.79065895e-01 3.25256169e-01 3.59258741e-01 -8.08823109e-02 -5.14024198e-01 -1.06539643e+00 5.71760476e-01 -1.67489874e+00 -8.49453509e-01 -2.39350095e-01 1.80542874e+00 8.56014132e-01 -2.69832730e-01 -2.68930104e-02 -2.96193302e-01 3.69040757e-01 1.52900100e-01 -4.73082155e-01 -7.79021263e-01 1.56003967e-01 6.27896249e-01 1.13467550e+00 6.02940321e-01 -1.18505883e+00 1.16138482e+00 7.74601507e+00 8.03518176e-01 -1.03847969e+00 5.56419231e-02 2.83760309e-01 2.77731031e-01 -6.15208387e-01 -2.23903000e-01 -6.07698858e-01 2.42069915e-01 1.35498190e+00 -3.23394150e-01 2.98422188e-01 7.03887045e-01 -1.26040921e-01 2.68859088e-01 -1.52796459e+00 4.03384447e-01 1.36872932e-01 -1.51790059e+00 5.46223879e-01 8.06156471e-02 6.34244382e-01 1.68265432e-01 4.96423990e-02 3.64271432e-01 8.54760349e-01 -1.55369699e+00 8.04195479e-02 4.40623283e-01 8.78001809e-01 -8.24080706e-01 6.03791535e-01 -3.70385677e-01 -1.26553714e+00 -2.62811661e-01 -7.57854342e-01 8.62319842e-02 -1.87770531e-01 5.06463170e-01 -1.22541404e+00 8.30915511e-01 4.25797731e-01 1.03178978e+00 -1.08878350e+00 1.15852439e+00 -6.33837759e-01 6.28542900e-01 -8.86686966e-02 -3.61584604e-01 1.15860134e-01 2.88519830e-01 9.91253778e-02 1.52257204e+00 5.02259210e-02 -7.15501457e-02 -1.59922302e-01 7.46960640e-01 -2.48758212e-01 2.03079283e-01 -1.09611845e+00 -2.26286888e-01 2.10539818e-01 1.06982398e+00 -6.11232638e-01 1.35295659e-01 -4.07214105e-01 6.76172078e-01 3.37856025e-01 3.67669344e-01 -9.21623647e-01 -5.94626725e-01 9.37085807e-01 -1.13881879e-01 5.07351696e-01 -1.91024244e-02 -1.09448604e-01 -6.79044306e-01 -8.66243422e-01 -3.12515199e-01 6.87877297e-01 -6.34584248e-01 -1.49049425e+00 3.34095120e-01 6.35090470e-02 -8.49625885e-01 4.32589293e-01 -1.30252516e+00 -7.98763454e-01 6.81649148e-01 -1.62353194e+00 -1.36503637e+00 5.93942404e-02 3.87661994e-01 -1.28285676e-01 -4.63426970e-02 1.25507557e+00 3.58555287e-01 -5.63897192e-01 6.33818805e-01 1.45876408e-01 1.43559784e-01 6.91293478e-01 -1.42747521e+00 3.84404600e-01 3.75827104e-01 1.72564477e-01 5.73578358e-01 3.75172496e-01 -8.47047865e-01 -1.53850937e+00 -1.68291128e+00 5.38905621e-01 -9.86892343e-01 1.00276887e+00 -4.95003872e-02 -6.79534316e-01 6.15058124e-01 1.38521880e-01 -3.08316648e-02 1.43907964e+00 4.40934747e-01 -7.61370957e-01 1.44939413e-02 -1.09467590e+00 4.43404496e-01 8.72901797e-01 -5.43773115e-01 -3.38331908e-01 3.37939650e-01 1.01332927e+00 5.14016598e-02 -1.45412242e+00 2.18122154e-01 6.53183281e-01 -1.13178298e-01 1.21668422e+00 -1.21583056e+00 8.11753750e-01 -3.08024734e-01 -1.51851058e-01 -1.99666822e+00 -9.92883086e-01 3.08223277e-01 -1.62789196e-01 6.62859082e-01 9.43334818e-01 -4.97694165e-01 6.14749134e-01 5.18825114e-01 -3.28575343e-01 -6.86097383e-01 -6.97377920e-01 -8.17328215e-01 7.30968237e-01 -3.29516888e-01 7.50038445e-01 1.04658997e+00 2.72413999e-01 7.05346763e-01 -2.87560403e-01 4.14382637e-01 3.84475529e-01 -2.20450595e-01 2.16380700e-01 -1.40134120e+00 -2.19845131e-01 -5.09658992e-01 -9.27040637e-01 -4.38378662e-01 1.88382044e-01 -1.53189433e+00 -4.78959143e-01 -1.98223269e+00 1.58960089e-01 1.27893493e-01 -9.53287303e-01 9.32392716e-01 -2.95870721e-01 3.08644116e-01 -4.65802811e-02 -3.95570546e-01 -4.80801255e-01 6.51341796e-01 1.07088864e+00 -7.40068495e-01 -2.11408287e-01 -6.36433125e-01 -1.14428127e+00 3.70136708e-01 8.77901137e-01 -6.17524445e-01 -4.62423742e-01 -7.51892328e-01 5.88905632e-01 -2.62749761e-01 2.47148439e-01 -1.02445328e+00 4.89828400e-02 -2.31438294e-01 8.55041385e-01 -4.43014689e-02 6.57410920e-02 -8.75729978e-01 3.54673743e-01 7.41671026e-01 -3.57682049e-01 -3.40245128e-01 7.75979102e-01 1.20089769e+00 2.37283096e-01 -1.58605725e-01 6.27177715e-01 2.14117587e-01 -9.32783782e-01 5.80933750e-01 -4.19069022e-01 -3.42218190e-01 9.16660666e-01 -1.40977912e-02 -7.39215910e-01 1.15626611e-01 -8.57347190e-01 3.17612439e-01 1.92711189e-01 3.38393807e-01 7.19258130e-01 -1.53443050e+00 -5.81588864e-01 -2.43660495e-01 5.95427096e-01 -6.47939742e-01 5.70088886e-02 4.14560050e-01 -8.59000266e-01 4.72414285e-01 -2.90824711e-01 2.09093690e-01 -9.52946186e-01 8.26117218e-01 3.80573303e-01 -4.93826658e-01 -1.38242587e-01 7.22384155e-01 3.89468253e-01 -5.79660296e-01 -7.70922601e-02 -5.36541343e-01 -8.52099955e-01 2.05708236e-01 4.02358890e-01 6.44440174e-01 1.60712123e-01 -2.39920095e-01 -5.65088391e-01 5.98468363e-01 6.23892136e-02 4.44542289e-01 1.64403474e+00 7.23704338e-01 3.33128907e-02 -6.69449195e-02 1.55326366e+00 -5.48655912e-02 -9.52336550e-01 2.64456183e-01 7.32226148e-02 -2.38028988e-02 2.70537049e-01 -1.31573951e+00 -1.08334839e+00 7.39853084e-01 5.99476099e-01 7.25467056e-02 7.28095472e-01 -8.86073187e-02 4.29212302e-01 8.23487520e-01 4.66070622e-02 -9.79831815e-01 2.23483130e-01 7.36258566e-01 1.03718495e+00 -1.00501394e+00 4.55608994e-01 -3.98194045e-01 -3.61317873e-01 1.13505983e+00 4.42788273e-01 -1.75217733e-01 1.27298784e+00 -1.19433731e-01 -2.59111702e-01 -8.06917250e-01 -7.97386944e-01 -2.66972899e-01 4.80167449e-01 1.08380818e+00 5.40453255e-01 2.26496324e-01 -3.12767863e-01 8.68151128e-01 1.95366502e-01 -2.44312435e-01 4.25283968e-01 5.24385095e-01 -5.84844232e-01 -1.21865153e+00 -2.27473071e-03 7.35006809e-01 -4.87521619e-01 -4.20845538e-01 -9.51264679e-01 7.09430397e-01 1.12028599e-01 8.99384797e-01 -5.55539668e-01 -8.78827393e-01 5.27129710e-01 1.17331006e-01 5.67112505e-01 -8.88360858e-01 -8.44575822e-01 -6.10288322e-01 3.25320452e-01 -3.90442073e-01 -9.44527909e-02 6.97174147e-02 -1.12040973e+00 -3.27762336e-01 -4.09004629e-01 1.47077560e-01 5.07743239e-01 5.38553059e-01 3.80149215e-01 7.70659924e-01 4.10163462e-01 -5.50259531e-01 -1.16826355e-01 -9.57271576e-01 -7.57034659e-01 2.70905346e-01 -5.82587421e-02 -8.29286933e-01 -1.53060928e-01 3.49176228e-02]
[7.879035949707031, 7.340205192565918]
88b7ba53-0350-4c22-a499-107ed7e3d22b
melt-mutual-enhancement-of-long-tailed-user
2304.08382
null
https://arxiv.org/abs/2304.08382v1
https://arxiv.org/pdf/2304.08382v1.pdf
MELT: Mutual Enhancement of Long-Tailed User and Item for Sequential Recommendation
The long-tailed problem is a long-standing challenge in Sequential Recommender Systems (SRS) in which the problem exists in terms of both users and items. While many existing studies address the long-tailed problem in SRS, they only focus on either the user or item perspective. However, we discover that the long-tailed user and item problems exist at the same time, and considering only either one of them leads to sub-optimal performance of the other one. In this paper, we propose a novel framework for SRS, called Mutual Enhancement of Long-Tailed user and item (MELT), that jointly alleviates the long-tailed problem in the perspectives of both users and items. MELT consists of bilateral branches each of which is responsible for long-tailed users and items, respectively, and the branches are trained to mutually enhance each other, which is trained effectively by a curriculum learning-based training. MELT is model-agnostic in that it can be seamlessly integrated with existing SRS models. Extensive experiments on eight datasets demonstrate the benefit of alleviating the long-tailed problems in terms of both users and items even without sacrificing the performance of head users and items, which has not been achieved by existing methods. To the best of our knowledge, MELT is the first work that jointly alleviates the long-tailed user and item problems in SRS.
['Chanyoung Park', 'Sukwon Yun', 'Dongmin Hyun', 'Kibum Kim']
2023-04-17
null
null
null
null
['sequential-recommendation']
['miscellaneous']
[-1.17326498e-01 -2.67405689e-01 -3.76317888e-01 -2.26456180e-01 -3.68785083e-01 -5.68152428e-01 4.72464822e-02 -1.57312810e-01 -2.21312895e-01 5.46159089e-01 2.32395187e-01 -5.21271229e-01 -4.10397112e-01 -6.08445466e-01 -5.62775016e-01 -6.65277302e-01 1.45466477e-01 4.52332377e-01 4.29396778e-01 -5.27129233e-01 1.48560002e-01 -5.48171587e-02 -1.70484841e+00 2.11422145e-01 1.32373333e+00 9.81159031e-01 5.32979906e-01 3.12071055e-01 -2.48646289e-01 4.95637566e-01 -3.48781407e-01 -2.56894171e-01 3.67413759e-01 -3.25106204e-01 -3.91386032e-01 3.56022976e-02 4.75474328e-01 -4.27096725e-01 -6.19020835e-02 8.06066453e-01 5.20523369e-01 2.86027402e-01 4.22433019e-01 -1.31637645e+00 -8.39040339e-01 6.38079226e-01 -7.54436016e-01 2.00426266e-01 1.92628399e-01 -4.26305354e-01 1.56454980e+00 -1.01843786e+00 1.05346926e-01 9.76188183e-01 6.36991441e-01 2.98574209e-01 -9.99374092e-01 -8.83419991e-01 7.95950174e-01 -6.70388632e-04 -1.10490251e+00 6.12043776e-02 4.41525906e-01 -4.70939666e-01 6.58950210e-01 4.14405227e-01 5.28250754e-01 6.49485707e-01 -1.52393252e-01 1.20485294e+00 7.71768630e-01 -1.71061695e-01 4.38667983e-02 2.86233783e-01 5.75466752e-01 1.37141511e-01 1.58759221e-01 1.06152892e-01 -3.40435237e-01 -1.66423380e-01 5.84660888e-01 4.80279148e-01 -1.42105818e-01 -4.70472604e-01 -8.04693341e-01 8.50322008e-01 2.10267842e-01 1.39788345e-01 -3.02309453e-01 -4.46388692e-01 -1.88996103e-02 4.60434616e-01 4.71074909e-01 4.53224003e-01 -1.01358092e+00 2.31741592e-02 -9.14593339e-01 4.41601455e-01 8.62347126e-01 1.16347337e+00 5.17336845e-01 -1.33470848e-01 -3.98256540e-01 9.00101542e-01 2.73776829e-01 4.14413780e-01 4.31239098e-01 -3.60425919e-01 2.42080465e-01 6.54230237e-01 4.51545566e-01 -7.49142349e-01 -4.13008064e-01 -1.20612967e+00 -6.11688793e-01 -2.37164050e-01 3.63087326e-01 -1.80913851e-01 -6.62371993e-01 2.05413151e+00 4.63910550e-01 3.44417185e-01 -2.61782855e-01 1.05269122e+00 8.90415430e-01 5.88606298e-01 -3.18213701e-02 -4.11519378e-01 1.24600065e+00 -1.47223794e+00 -6.12780392e-01 -1.36395767e-01 6.01316631e-01 -1.01549721e+00 1.26261258e+00 4.75774616e-01 -1.07868564e+00 -4.66822952e-01 -8.38269770e-01 -2.97129638e-02 -1.42219737e-01 1.50001913e-01 4.96433765e-01 6.09970570e-01 -7.34646082e-01 4.03482348e-01 -2.64803082e-01 1.21739451e-02 6.31503463e-02 4.87146139e-01 1.82489067e-01 -1.55664468e-02 -1.42214513e+00 7.23411858e-01 -8.33111778e-02 -1.72103330e-01 -2.00495288e-01 -1.10253060e+00 -4.24051672e-01 3.62788767e-01 9.07561243e-01 -6.55113220e-01 1.64979839e+00 -9.97043610e-01 -1.32332122e+00 8.96280110e-02 -1.30747318e-01 3.01037170e-02 4.14398789e-01 -5.18656850e-01 -6.31579518e-01 -6.72497153e-01 1.33421198e-01 -8.10544491e-02 6.32734835e-01 -1.10115361e+00 -1.10347271e+00 -3.70646775e-01 3.65759045e-01 5.37279963e-01 -8.04333568e-01 -4.50002328e-02 -7.70237327e-01 -8.11312914e-01 -1.19630247e-01 -9.29765701e-01 -1.91311315e-01 -2.31773838e-01 4.91753668e-02 -4.91054416e-01 7.72379994e-01 -3.76559824e-01 1.83294606e+00 -2.10135150e+00 2.88287103e-01 3.21623124e-02 1.88426062e-01 5.80021024e-01 -4.11748528e-01 6.81296051e-01 -7.05299992e-03 -4.51555178e-02 3.02159309e-01 -5.18727005e-01 1.12945717e-02 4.00473028e-01 -3.22850734e-01 7.67466500e-02 -2.59316146e-01 6.34497941e-01 -1.06152320e+00 -9.48597863e-02 -1.36402652e-01 2.82831401e-01 -8.56708527e-01 4.46815848e-01 -2.56665647e-01 3.57958168e-01 -3.94204408e-01 2.82167047e-01 8.17222536e-01 -3.74631226e-01 1.12742528e-01 -1.10755458e-01 -1.85133174e-01 4.23225373e-01 -1.38950443e+00 1.21226871e+00 -5.67020237e-01 -3.29243809e-01 1.27174109e-01 -6.64086759e-01 6.82286501e-01 2.14960843e-01 5.90739787e-01 -9.79072392e-01 -2.28570253e-01 3.09416801e-01 1.43641233e-01 -3.95657599e-01 8.66811097e-01 -3.42973650e-01 4.63055186e-02 7.39246845e-01 -8.55768919e-02 5.54054499e-01 2.58207381e-01 4.89539951e-01 6.06597722e-01 4.50798646e-02 2.88850039e-01 -2.06471547e-01 3.24818492e-01 -6.27516448e-01 7.04389751e-01 8.27163517e-01 1.96671173e-01 3.35013032e-01 3.30396205e-01 -2.04064846e-02 -6.88003480e-01 -8.38059604e-01 4.38475385e-02 2.03326607e+00 5.83677649e-01 -6.42491221e-01 -3.01882327e-01 -9.69203591e-01 3.06982726e-01 7.56008565e-01 -5.31502366e-01 -6.70094267e-02 -3.05050224e-01 -5.63075364e-01 -1.15319349e-01 6.10342264e-01 -2.26717219e-01 -6.73895597e-01 -2.44705394e-01 3.08205545e-01 -3.09186667e-01 -9.47017670e-01 -1.13277006e+00 6.98086694e-02 -7.24969983e-01 -9.01566386e-01 -8.07091415e-01 -4.81688589e-01 4.77221310e-01 1.07554340e+00 1.12312889e+00 3.66004825e-01 2.83843488e-01 -1.84314810e-02 -7.05097258e-01 -4.66008842e-01 2.68134266e-01 1.09272279e-01 3.35575826e-02 2.19039440e-01 1.03178412e-01 -6.66137576e-01 -7.94025242e-01 1.05962300e+00 -8.96320105e-01 -5.57909235e-02 6.06051087e-01 8.90376627e-01 3.65520418e-01 1.64672941e-01 8.80744934e-01 -1.19986498e+00 5.31005144e-01 -1.01124775e+00 -3.38980228e-01 2.72148311e-01 -1.03397512e+00 -2.13660121e-01 9.10470665e-01 -6.86293185e-01 -9.61223662e-01 -3.09507757e-01 -2.87420958e-01 -1.93921044e-01 1.60130724e-01 7.58663356e-01 -2.05995575e-01 2.38336757e-01 9.21020508e-02 1.35469660e-01 -2.92329192e-01 -9.82960224e-01 3.58411849e-01 8.38898063e-01 2.16596618e-01 -4.02372420e-01 6.33503854e-01 7.46697709e-02 -4.12552059e-01 -4.60020006e-01 -1.52893794e+00 -1.10745358e+00 -2.02309281e-01 -3.67031768e-02 1.44694224e-01 -1.05471337e+00 -5.50707161e-01 3.44119787e-01 -5.57669640e-01 -4.89682034e-02 -3.17455649e-01 3.81728023e-01 -1.46279797e-01 3.82701755e-01 -3.96599323e-01 -7.31198668e-01 -3.68180156e-01 -1.10015011e+00 7.91769028e-01 4.20406371e-01 3.89638543e-02 -9.12557662e-01 -2.15615071e-02 5.23898900e-01 5.17038584e-01 -6.20512187e-01 8.63023221e-01 -9.84728634e-01 -2.28150532e-01 -2.19680965e-01 -2.12181062e-01 2.65054435e-01 1.12565316e-01 -3.64496261e-01 -3.70112598e-01 -6.65779352e-01 -4.37992886e-02 -3.84975709e-02 8.07827175e-01 1.99007034e-01 9.33972418e-01 -2.78938472e-01 -1.21815592e-01 2.64079273e-01 1.14927912e+00 3.44959907e-02 4.10559773e-01 5.05230874e-02 7.09248066e-01 3.80713403e-01 8.60286355e-01 7.62771904e-01 8.52271497e-01 1.02907658e+00 5.12030900e-01 -3.16762686e-01 8.70797634e-02 -3.31003219e-01 3.22061241e-01 8.64535928e-01 8.57127309e-02 -5.03952920e-01 -3.03310066e-01 5.54650486e-01 -2.25318646e+00 -7.12839663e-01 -4.34701830e-01 2.24937224e+00 6.87637150e-01 -1.93618640e-01 6.60498381e-01 1.53748570e-02 4.97624159e-01 -1.18693441e-01 -5.12061536e-01 -1.78810045e-01 8.11882839e-02 3.76508646e-02 2.71449536e-01 2.64835656e-01 -8.52551818e-01 7.45074093e-01 5.56424999e+00 9.55136776e-01 -8.85831416e-01 1.65505797e-01 3.72327380e-02 -4.22550499e-01 -4.68107373e-01 1.03202909e-01 -1.11576200e+00 7.35294163e-01 4.99582708e-01 -2.93271244e-01 3.99976403e-01 8.62311006e-01 2.07354680e-01 8.74046236e-02 -1.05781579e+00 5.56782603e-01 1.48280874e-01 -7.52435088e-01 -1.16217211e-01 2.47717410e-01 9.92084265e-01 -1.13192052e-01 1.86120063e-01 7.28277862e-01 3.63277763e-01 -6.23530447e-01 7.77151585e-01 3.52001518e-01 3.66797060e-01 -8.46921325e-01 7.78554976e-01 8.91580045e-01 -1.38397622e+00 -3.96803290e-01 -1.70361087e-01 -2.95981199e-01 1.27865970e-01 6.92609906e-01 -2.27696136e-01 8.12302530e-01 5.24186790e-01 7.70705760e-01 -2.93438315e-01 1.24131835e+00 -1.24693364e-01 6.94741786e-01 -1.18182041e-01 8.85777920e-02 2.77055144e-01 -2.61319399e-01 4.56197143e-01 9.92856681e-01 5.38078189e-01 2.14360267e-01 8.36821318e-01 4.27215695e-01 1.15609088e-03 4.37847674e-01 -5.13347909e-02 2.37197652e-01 4.77144837e-01 1.36586821e+00 -1.72363713e-01 -3.65803719e-01 -7.64810085e-01 5.83905518e-01 4.35709089e-01 3.24290544e-01 -7.81205893e-01 -8.30528066e-02 6.19741023e-01 4.05281574e-01 8.33714724e-01 8.91325995e-02 -2.75350332e-01 -1.33541238e+00 3.66919935e-02 -1.01258540e+00 7.14208782e-01 -2.72522509e-01 -1.63644254e+00 2.10924044e-01 -3.00863057e-01 -1.40892804e+00 8.42453092e-02 -1.78638741e-01 -6.39970481e-01 9.10388887e-01 -1.72440624e+00 -1.15858316e+00 -6.59902468e-02 7.81126142e-01 5.41144729e-01 2.90996656e-02 6.20929122e-01 7.37880230e-01 -7.15320885e-01 9.62714076e-01 4.74786580e-01 -5.65923750e-01 8.18355381e-01 -1.29019690e+00 -8.33643973e-02 6.40915215e-01 -1.20849952e-01 9.04942572e-01 7.20928133e-01 -5.80131531e-01 -1.36755896e+00 -1.05871165e+00 1.07613754e+00 -1.49528548e-01 6.21660650e-01 -2.32242405e-01 -1.01160538e+00 5.24599254e-01 -3.04200444e-02 -2.55173266e-01 1.16018677e+00 8.09282959e-01 -6.22920036e-01 5.47109097e-02 -7.57023990e-01 5.66719592e-01 9.70271051e-01 -1.75272182e-01 -4.95803893e-01 2.90634751e-01 7.31722474e-01 -3.73542398e-01 -8.37782443e-01 4.23164964e-01 7.08024204e-01 -9.84647155e-01 9.56620216e-01 -8.33470881e-01 4.04613912e-01 -2.71445662e-01 -2.06446692e-01 -1.39138758e+00 -6.76881731e-01 -4.26502585e-01 -6.25003636e-01 1.13411331e+00 4.60769653e-01 -4.26691413e-01 6.85045838e-01 5.18511713e-01 -3.66911024e-01 -1.06670094e+00 -4.10947680e-01 -1.04371214e+00 1.80391267e-01 -1.47246376e-01 8.28466654e-01 8.75911593e-01 2.00213864e-02 7.71721303e-01 -1.09749901e+00 1.29373595e-01 3.61990750e-01 7.46283412e-01 8.26827645e-01 -1.27435720e+00 -6.63986027e-01 -4.16709751e-01 4.54975188e-01 -1.64900172e+00 -2.10790873e-01 -8.27211440e-01 7.79436305e-02 -1.56756878e+00 4.47694391e-01 -7.83050418e-01 -7.72943199e-01 3.50173891e-01 -6.68114901e-01 4.66806777e-02 3.03753108e-01 2.80190676e-01 -9.95586574e-01 5.59395909e-01 1.35612702e+00 4.43749040e-01 -4.86928523e-01 6.55744553e-01 -1.40555036e+00 6.98799789e-01 6.24217868e-01 -4.34032112e-01 -6.80492878e-01 -4.40516323e-01 4.60505456e-01 2.08908349e-01 -4.75359596e-02 -4.88700569e-01 3.90851587e-01 -3.75654787e-01 -1.17063724e-01 -8.78517687e-01 1.33528933e-01 -9.63619769e-01 6.44452646e-02 1.90948233e-01 -3.41782242e-01 -1.10538237e-01 -1.38159260e-01 8.30048084e-01 1.90174896e-02 -1.42108023e-01 7.44293392e-01 8.55386481e-02 -4.04158860e-01 5.87660253e-01 -2.44780421e-01 7.74847046e-02 1.06920338e+00 9.75262895e-02 -2.31713623e-01 -5.62175274e-01 -5.77321231e-01 7.39996493e-01 1.68150917e-01 6.73774183e-01 3.83391142e-01 -1.21611428e+00 -6.35293186e-01 2.38364384e-01 3.18460584e-01 -2.87830710e-01 6.95826471e-01 1.02952266e+00 5.22527337e-01 2.78460979e-01 8.94669443e-02 -1.26338482e-01 -1.47591019e+00 7.95578778e-01 -7.54271597e-02 -8.91838491e-01 -2.35452414e-01 8.92305195e-01 4.66823190e-01 -6.90801084e-01 4.59403306e-01 5.79574332e-02 -5.46700299e-01 3.16318423e-01 6.30261958e-01 3.62943590e-01 -6.87613860e-02 -5.95737278e-01 -1.92518800e-01 5.39635777e-01 -6.01541936e-01 3.88121545e-01 1.40998602e+00 -4.43842858e-01 6.30089715e-02 3.31139207e-01 8.28722537e-01 3.04153174e-01 -8.58181834e-01 -8.71984422e-01 -2.07262486e-01 -6.08927548e-01 2.67835725e-02 -1.09840548e+00 -1.21011782e+00 6.64572597e-01 2.03387350e-01 3.86712879e-01 1.28423369e+00 -9.27293375e-02 1.34454632e+00 -1.28006572e-02 3.31119835e-01 -1.03977740e+00 1.86253592e-01 8.00323844e-01 6.77396953e-01 -1.02342784e+00 5.84296696e-02 -5.67418277e-01 -8.23892057e-01 6.50387883e-01 8.51683795e-01 -3.31719182e-02 8.44468951e-01 1.70919687e-01 -2.26633877e-01 1.43710926e-01 -1.09897530e+00 -3.72220814e-01 5.45923293e-01 2.54403383e-01 5.20834148e-01 2.70759851e-01 -7.06205666e-01 1.25535989e+00 6.33895621e-02 1.63685590e-01 2.71204859e-01 9.70465541e-01 -4.85098988e-01 -1.57883334e+00 -5.07418290e-02 7.61482358e-01 -4.44796205e-01 -2.94892311e-01 -1.88807666e-01 3.63701493e-01 3.50877225e-01 1.13242459e+00 -2.97985077e-01 -7.19623983e-01 6.28470778e-01 -3.06297928e-01 6.72581270e-02 -6.88174665e-01 -9.32548583e-01 5.08991599e-01 6.50716349e-02 -3.62078398e-01 -1.32731274e-01 -5.25053263e-01 -9.20382798e-01 -4.33129162e-01 -8.24232817e-01 3.09756160e-01 4.50441092e-01 9.76597309e-01 5.79880059e-01 6.63484514e-01 9.54426408e-01 -4.36360329e-01 -1.07630706e+00 -7.82337427e-01 -9.96135771e-01 4.65179652e-01 2.74829179e-01 -8.80533457e-01 -1.82543546e-01 -4.97385710e-01]
[10.142111778259277, 5.550002574920654]
a25612e9-0211-46a4-9d0a-7cac7830d33c
diabetic-retinopathy-diagnosis-based-on
2008.00148
null
https://arxiv.org/abs/2008.00148v1
https://arxiv.org/pdf/2008.00148v1.pdf
Diabetic Retinopathy Diagnosis based on Convolutional Neural Network
Diabetic Retinopathy DR is a popular disease for many people as a result of age or the diabetic, as a result, it can cause blindness. therefore, diagnosis of this disease especially in the early time can prevent its effect for a lot of patients. To achieve this diagnosis, eye retina must be examined continuously. Therefore, computer-aided tools can be used in the field based on computer vision techniques. Different works have been performed using various machine learning techniques. Convolutional Neural Network is one of the promise methods, so it was for Diabetic Retinopathy detection in this paper. Also, the proposed work contains visual enhancement in the pre-processing phase, then the CNN model is trained to be able for recognition and classification phase, to diagnosis the healthy and unhealthy retina image. Three public dataset DiaretDB0, DiaretDB1 and DrimDB were used in practical testing. The implementation of this work based on Matlab- R2019a, deep learning toolbox and deep network designer to design the architecture of the convolutional neural network and train it. The results were evaluated to different metrics; accuracy is one of them. The best accuracy that was achieved: for DiaretDB0 is 100%, DiaretDB1 is 99.495% and DrimDB is 97.55%.
['Lamia Abed Noor Muhammed', 'Mohammed hamzah abed', 'Sarah Hussein Toman']
2020-08-01
null
null
null
null
['diabetic-retinopathy-detection']
['medical']
[-2.13622138e-01 -1.52334034e-01 2.58007854e-01 -2.06954464e-01 2.86925554e-01 1.11605786e-01 2.58017838e-01 -2.08168790e-01 -3.40953231e-01 7.26708055e-01 2.89097995e-01 -2.79540122e-01 -1.37262791e-01 -8.86114299e-01 -1.80581748e-01 -8.26854765e-01 6.73011988e-02 8.38974491e-02 3.16420794e-01 -4.61558178e-02 4.63654280e-01 7.38502502e-01 -1.79699099e+00 4.18970704e-01 9.86898839e-01 7.78249264e-01 1.74910322e-01 7.47456431e-01 -1.03777409e-01 9.05250490e-01 -4.30513114e-01 2.38864776e-02 4.23165858e-01 -3.97543401e-01 -5.15033722e-01 2.99826767e-02 7.18022659e-02 -6.19133949e-01 -2.22611770e-01 9.68480766e-01 9.50498521e-01 -1.55330032e-01 8.41656625e-01 -6.45897686e-01 -5.85978746e-01 -4.03665900e-02 -7.37062454e-01 5.09466469e-01 -9.83029231e-02 4.38237011e-01 -1.29874215e-01 -5.01244485e-01 1.46074712e-01 1.34874582e+00 4.74635601e-01 5.49034595e-01 -7.54948735e-01 -6.30650699e-01 -6.30537450e-01 5.76184452e-01 -1.04184377e+00 -1.92096889e-01 2.92866111e-01 -9.41453874e-01 6.73400164e-01 6.34335801e-02 7.96827614e-01 4.56054032e-01 4.55011487e-01 3.87115687e-01 1.62045324e+00 -4.39499408e-01 -2.59246789e-02 1.36250794e-01 6.12015486e-01 5.37691772e-01 5.64872205e-01 4.28480685e-01 2.46584326e-01 5.04173577e-01 8.79096806e-01 1.00484475e-01 -3.52880716e-01 3.07668775e-01 -6.30558193e-01 6.04491651e-01 5.46530724e-01 3.68306130e-01 -4.71606106e-01 -2.42803335e-01 4.01863396e-01 3.52700949e-01 -7.95263126e-02 -8.08144882e-02 -2.87173182e-01 5.62860891e-02 -3.55938166e-01 -5.27622504e-03 4.33911085e-01 3.82461816e-01 5.70760727e-01 -3.82178426e-02 -3.20367634e-01 7.52083898e-01 6.46657765e-01 3.17149758e-01 6.54061258e-01 -4.97083813e-01 -7.24399760e-02 1.05454195e+00 9.45199057e-02 -9.13780093e-01 -5.06720960e-01 -5.25018036e-01 -1.17929602e+00 9.84908938e-01 3.34535867e-01 -3.20096940e-01 -1.59271801e+00 7.25415766e-01 2.06482530e-01 9.28696543e-02 4.14822608e-01 1.08817017e+00 1.31627035e+00 8.12860310e-01 -1.84760895e-02 -1.50706008e-01 1.47754383e+00 -6.76542163e-01 -6.32870138e-01 2.81959474e-02 2.70321816e-01 -1.22493446e+00 7.69230664e-01 7.00852811e-01 -8.20369065e-01 -6.53836548e-01 -9.58679795e-01 -3.05498932e-02 -2.37859264e-01 8.40472043e-01 5.72820604e-01 6.20792925e-01 -9.83597040e-01 1.61970958e-01 -5.17738104e-01 -7.71147072e-01 7.39741623e-01 3.38652402e-01 -2.45855272e-01 -5.27565122e-01 -7.05821991e-01 1.22839427e+00 4.62472290e-01 6.31337166e-01 -6.70447111e-01 -2.48681650e-01 -7.72066787e-02 -2.51579314e-01 -3.52759689e-01 -6.97457790e-01 9.36141670e-01 -1.04630446e+00 -1.15515363e+00 1.01521873e+00 1.03463642e-01 -4.84126151e-01 5.08307934e-01 -1.79721028e-01 -6.91002190e-01 7.27054626e-02 -2.68640399e-01 2.59989232e-01 2.46717930e-01 -9.56198215e-01 -7.99157679e-01 -5.05290389e-01 -1.00746423e-01 -3.96693638e-03 1.25248894e-01 3.10791492e-01 -1.58823967e-01 -1.31494746e-01 4.56769168e-02 -5.48685908e-01 1.20792789e-02 5.03523089e-02 -4.68738168e-01 -2.54952759e-01 8.39436650e-01 -9.09318864e-01 8.68861496e-01 -2.07140446e+00 -2.38360718e-01 2.59424180e-01 1.53728381e-01 1.07304883e+00 2.27173075e-01 1.01151735e-01 -1.66131407e-01 9.06338766e-02 2.28576005e-01 4.11835104e-01 -5.99428177e-01 7.06923753e-02 4.39161211e-01 4.42321122e-01 2.36968324e-01 2.37029165e-01 -1.26412407e-01 -4.24453825e-01 4.32238132e-01 7.96866357e-01 -1.57852069e-01 2.26820439e-01 9.56677794e-02 4.33453232e-01 -4.68164802e-01 7.80195892e-01 9.20281649e-01 9.96789113e-02 -3.59071463e-01 -6.60495818e-01 -5.61254382e-01 -5.08776367e-01 -1.09663284e+00 6.49602473e-01 -6.80774972e-02 9.22100425e-01 -1.66274384e-01 -1.10873592e+00 1.21334481e+00 3.50597233e-01 3.89532745e-02 -8.83382797e-01 6.45027757e-01 3.25367272e-01 4.00368214e-01 -1.51195431e+00 -3.24509442e-01 1.09379873e-01 1.12030518e+00 -1.49904147e-01 -5.29246449e-01 6.88052595e-01 3.27806383e-01 -3.70886326e-01 8.94530475e-01 -1.21537536e-01 3.28232706e-01 4.14249822e-02 8.08869421e-01 1.71042755e-01 5.96939325e-01 1.09954953e-01 -3.06497604e-01 4.54129875e-01 4.63007092e-01 -7.83353686e-01 -1.05933094e+00 -5.72019517e-01 -5.11761546e-01 -1.15931250e-01 -7.14657456e-02 6.88535273e-01 -6.15893483e-01 -1.24131456e-01 -4.36607637e-02 2.67252266e-01 -4.12249774e-01 -1.32503524e-01 -3.58477950e-01 -9.72227454e-01 2.63684958e-01 1.32232100e-01 1.31908405e+00 -1.11232209e+00 -5.55040538e-01 1.82972532e-02 5.37086785e-01 -5.24064839e-01 3.15132111e-01 -4.51478273e-01 -1.08502579e+00 -1.65100336e+00 -9.67191696e-01 -1.16592598e+00 6.89036608e-01 2.46828958e-01 6.80007279e-01 4.10953134e-01 -8.96417439e-01 -1.67767256e-01 -2.94230729e-01 -6.89163387e-01 -4.74352509e-01 -5.11140585e-01 -5.00977993e-01 2.25987896e-01 8.50230277e-01 -5.92911541e-01 -1.09523749e+00 1.75547302e-01 -5.80385208e-01 1.44522369e-01 1.43672609e+00 4.89915818e-01 3.04921836e-01 2.05996037e-01 4.79068696e-01 -7.14957297e-01 4.49387550e-01 -2.72775590e-01 -8.99437249e-01 8.14878866e-02 -7.46012211e-01 -2.88553447e-01 1.39815584e-01 -9.64932740e-02 -9.33081090e-01 -3.90565656e-02 -8.42902903e-03 -1.33439079e-01 -4.91008908e-01 5.42983353e-01 -2.48451233e-01 -9.44284424e-02 8.01524043e-01 4.81397286e-02 4.38853472e-01 -7.99168706e-01 -2.74311572e-01 1.27100766e+00 3.35243374e-01 2.26084083e-01 3.13397616e-01 3.56439268e-03 8.60578269e-02 -9.35633481e-01 -4.80934978e-01 -5.83717048e-01 -1.82858318e-01 -5.82239211e-01 1.28554583e+00 -9.38095391e-01 -9.49916661e-01 1.00043249e+00 -1.22801983e+00 -2.08839457e-02 5.34289658e-01 9.93603826e-01 1.12922475e-01 9.87923145e-03 -2.32150152e-01 -8.77197027e-01 -6.05576158e-01 -1.07233202e+00 1.76032692e-01 9.93821084e-01 4.42696542e-01 -7.61535704e-01 -4.32061255e-02 4.20373380e-01 5.87386370e-01 6.23233080e-01 1.15603936e+00 -2.98098296e-01 -7.83257365e-01 -3.10865939e-01 -8.39659572e-01 8.89932930e-01 2.10117862e-01 5.41672885e-01 -8.38200867e-01 -1.23767955e-02 -2.56618083e-01 -1.38263153e-02 1.00142944e+00 6.95508540e-01 9.02272642e-01 -2.89084524e-01 -2.46579707e-01 4.22538847e-01 1.97327375e+00 8.12547326e-01 1.54552984e+00 6.30495310e-01 4.12542522e-01 3.67772341e-01 5.38370252e-01 1.44167259e-01 1.02808475e-01 2.13040829e-01 7.42276192e-01 -5.58087468e-01 -6.74392045e-01 5.26425660e-01 1.02243103e-01 2.74866074e-01 -6.16353035e-01 -5.90552688e-02 -1.04549181e+00 5.23375750e-01 -1.48975062e+00 -1.12355447e+00 -9.79597807e-01 2.21401262e+00 7.16966212e-01 -5.56440540e-02 3.52717787e-02 4.35104549e-01 8.80151153e-01 -6.44063115e-01 -2.97359675e-01 -3.78684610e-01 -1.06296077e-01 4.33678001e-01 5.14574885e-01 1.62145019e-01 -9.97329414e-01 3.73975009e-01 5.04338598e+00 5.08554950e-02 -1.48567712e+00 -1.63386747e-01 6.35380924e-01 1.17344156e-01 3.91425043e-01 -1.32270649e-01 -7.38592207e-01 6.68490052e-01 5.27242184e-01 1.45501748e-01 2.48192891e-01 5.36527634e-01 8.94926846e-01 -4.78380710e-01 -5.19051611e-01 1.07090640e+00 -1.38767242e-01 -1.13954198e+00 -8.51378143e-02 -3.76303606e-02 6.27942502e-01 -7.35089332e-02 -1.66248694e-01 -4.46739979e-03 4.32137772e-03 -1.37803316e+00 -3.08081478e-01 1.28503704e+00 4.88262504e-01 -9.47219193e-01 1.39881766e+00 9.01810303e-02 -4.90110338e-01 -2.02332750e-01 -6.81291163e-01 -1.20707294e-02 -3.14352661e-01 9.67404544e-01 -7.93196261e-01 2.54889041e-01 8.44737649e-01 8.69575620e-01 -7.27553546e-01 2.06862426e+00 -6.25298396e-02 7.97419012e-01 8.02426711e-02 -1.94720998e-01 -6.27643317e-02 -3.86390835e-01 3.38103473e-01 9.39987838e-01 5.95733225e-01 2.24095777e-01 -1.51016086e-01 7.04477549e-01 3.88162673e-01 3.75534832e-01 -3.35075617e-01 1.46933094e-01 1.13653474e-01 1.09523964e+00 -2.51349390e-01 -1.77797116e-02 -3.82005125e-01 4.28844213e-01 -1.98294714e-01 3.85905147e-01 -6.03387892e-01 -6.86570585e-01 4.25662369e-01 3.01110685e-01 1.45637244e-02 1.04248643e-01 -2.32149601e-01 -5.90316355e-01 -1.23235770e-01 -8.75654995e-01 4.53629382e-02 -1.10201061e+00 -9.60495412e-01 4.25150275e-01 -4.35673773e-01 -1.09424448e+00 4.45134908e-01 -1.04553390e+00 -8.96032631e-01 1.47065735e+00 -1.69002903e+00 -1.09790015e+00 -9.46166635e-01 5.84796011e-01 2.26815686e-01 -7.19681859e-01 5.62283695e-01 7.00647652e-01 -1.03590727e+00 4.97553051e-02 2.60724038e-01 2.47198865e-01 8.16776335e-01 -1.00277698e+00 -6.47001207e-01 9.67928112e-01 -8.47619951e-01 3.54990810e-01 7.44495273e-01 -5.60757279e-01 -1.03661895e+00 -1.21669888e+00 5.42688668e-01 3.45943004e-01 -8.37300271e-02 5.71867585e-01 -7.43121624e-01 4.33059096e-01 3.45212907e-01 -1.70106869e-02 3.54721487e-01 -2.26034790e-01 2.51840830e-01 -5.20771921e-01 -1.37108028e+00 4.47672367e-01 3.16551059e-01 2.82120705e-01 -4.09656435e-01 3.69073391e-01 5.42412512e-02 -3.50743771e-01 -8.41459036e-01 3.03593665e-01 5.20129502e-01 -1.50605714e+00 7.23399162e-01 -4.60015595e-01 6.38852477e-01 -7.70114601e-01 2.04022765e-01 -8.40788603e-01 -1.40084684e-01 -7.60613829e-02 1.73032984e-01 1.37883902e+00 3.59815389e-01 -7.98029065e-01 6.81334913e-01 2.61096150e-01 -2.85751611e-01 -6.90097868e-01 -4.28070247e-01 -3.80644470e-01 -1.26723558e-01 2.85626322e-01 2.59424567e-01 5.79815567e-01 -1.00025284e+00 2.99157560e-01 -9.70003232e-02 4.49849337e-01 4.65266287e-01 -3.95702183e-01 8.20028067e-01 -1.49193764e+00 1.24487743e-01 -2.69304544e-01 -1.01795578e+00 -3.84570658e-01 -7.67480195e-01 -4.61282343e-01 -4.50170726e-01 -2.35931778e+00 5.65147698e-02 -4.53246772e-01 -9.54936296e-02 5.29803097e-01 7.93951154e-02 1.23751312e-01 -3.77077430e-01 2.28068590e-01 4.25219685e-01 1.25871657e-03 1.64437020e+00 -1.29468963e-01 -3.99610549e-01 4.21894431e-01 -6.01440549e-01 5.42794347e-01 1.10462105e+00 -8.23662709e-03 -5.03711700e-01 -4.00800616e-01 -3.03150224e-03 -2.88401186e-01 7.55352736e-01 -1.40217590e+00 2.03018785e-01 -9.02748257e-02 6.90879226e-01 -7.18932569e-01 -1.30684990e-02 -8.76098692e-01 2.54897535e-01 6.68525875e-01 1.06377685e-02 -4.42081302e-01 1.19236372e-01 3.12233657e-01 -1.50380775e-01 -2.50912279e-01 1.24738479e+00 -7.23037943e-02 -8.87632847e-01 1.63108110e-01 -4.01700765e-01 -2.33034045e-01 1.29521787e+00 -7.12653399e-01 -6.33320153e-01 4.07362916e-02 -6.61955118e-01 2.46751949e-01 1.07397884e-01 -2.94892918e-02 7.09088147e-01 -1.09152424e+00 -8.98635089e-01 5.71740642e-02 1.05920702e-01 1.57427549e-01 3.89226377e-01 1.20018196e+00 -1.30293643e+00 9.62311924e-02 -8.32829952e-01 -4.70290571e-01 -1.76697195e+00 3.28539222e-01 9.33514118e-01 3.04138839e-01 -6.39251947e-01 4.66255933e-01 -3.48034829e-01 3.15955311e-01 3.64972264e-01 -4.85596150e-01 -1.18981957e+00 -1.42121047e-01 7.82900214e-01 6.87590957e-01 2.89478675e-02 -3.99410188e-01 1.23810684e-02 8.71905088e-01 -1.75232604e-01 3.86612236e-01 1.35145712e+00 -1.52213015e-02 -5.53605855e-01 1.67661235e-02 9.64650631e-01 -1.90466180e-01 -7.62706995e-01 2.34383419e-01 -2.81405956e-01 -4.70254064e-01 4.37092602e-01 -1.50344968e+00 -1.34531128e+00 1.04025114e+00 1.62919939e+00 6.46405965e-02 1.41883838e+00 -5.63827574e-01 4.92960691e-01 3.09324980e-01 -5.92631511e-02 -7.56038785e-01 -2.96333879e-01 1.93324015e-01 8.35634887e-01 -1.39832091e+00 8.20619836e-02 -2.88315594e-01 -3.96310627e-01 1.52078569e+00 6.52736306e-01 -2.91764826e-01 7.71661818e-01 -1.36183947e-01 4.64715362e-01 -3.20618927e-01 -3.14720958e-01 -4.18078244e-01 2.68467039e-01 8.71235490e-01 7.74355412e-01 -1.99038714e-01 -7.82212198e-01 4.07556295e-01 3.67106125e-02 7.94548392e-01 7.36389637e-01 7.11208165e-01 -1.05654967e+00 -9.33951735e-01 -4.88408118e-01 8.19965720e-01 -5.15126944e-01 2.03405127e-01 -2.76240885e-01 1.03432941e+00 6.38252616e-01 1.02431953e+00 5.36286943e-02 -2.10746720e-01 4.30203110e-01 -3.29424948e-01 2.67032444e-01 -3.86033744e-01 -3.52080941e-01 -1.67589933e-01 3.16400051e-01 -1.99107185e-01 -7.32507765e-01 -3.47349703e-01 -1.39293706e+00 -4.45847631e-01 -4.57789600e-02 -1.57001838e-01 7.94611037e-01 8.96322608e-01 2.36235678e-01 7.27642417e-01 5.32768905e-01 -6.05552122e-02 -1.09294996e-01 -1.23976362e+00 -6.09508157e-01 -5.21563590e-02 3.41937989e-01 -5.77446759e-01 -1.47931099e-01 1.78972751e-01]
[15.836468696594238, -3.9965524673461914]
b9d60398-9883-4248-aaed-6667e46a2457
know-what-and-know-where-an-object-and-room
2104.04167
null
https://arxiv.org/abs/2104.04167v2
https://arxiv.org/pdf/2104.04167v2.pdf
The Road to Know-Where: An Object-and-Room Informed Sequential BERT for Indoor Vision-Language Navigation
Vision-and-Language Navigation (VLN) requires an agent to find a path to a remote location on the basis of natural-language instructions and a set of photo-realistic panoramas. Most existing methods take the words in the instructions and the discrete views of each panorama as the minimal unit of encoding. However, this requires a model to match different nouns (e.g., TV, table) against the same input view feature. In this work, we propose an object-informed sequential BERT to encode visual perceptions and linguistic instructions at the same fine-grained level, namely objects and words. Our sequential BERT also enables the visual-textual clues to be interpreted in light of the temporal context, which is crucial to multi-round VLN tasks. Additionally, we enable the model to identify the relative direction (e.g., left/right/front/back) of each navigable location and the room type (e.g., bedroom, kitchen) of its current and final navigation goal, as such information is widely mentioned in instructions implying the desired next and final locations. We thus enable the model to know-where the objects lie in the images, and to know-where they stand in the scene. Extensive experiments demonstrate the effectiveness compared against several state-of-the-art methods on three indoor VLN tasks: REVERIE, NDH, and R2R. Project repository: https://github.com/YuankaiQi/ORIST
['Qi Wu', 'Anton Van Den Hengel', 'Ming-Hsuan Yang', 'Yicong Hong', 'Zizheng Pan', 'Yuankai Qi']
2021-04-09
null
http://openaccess.thecvf.com//content/ICCV2021/html/Qi_The_Road_To_Know-Where_An_Object-and-Room_Informed_Sequential_BERT_for_ICCV_2021_paper.html
http://openaccess.thecvf.com//content/ICCV2021/papers/Qi_The_Road_To_Know-Where_An_Object-and-Room_Informed_Sequential_BERT_for_ICCV_2021_paper.pdf
iccv-2021-1
['vision-language-navigation']
['computer-vision']
[-6.88534603e-02 -4.97990519e-01 -4.69124643e-03 -6.44202828e-01 -1.96967512e-01 -7.26428509e-01 6.29551053e-01 -3.25354785e-02 -4.14152980e-01 4.67513382e-01 3.88029367e-01 -5.55518389e-01 -6.21298403e-02 -8.61996651e-01 -8.43586326e-01 -4.21657532e-01 1.71760187e-01 3.16175193e-01 1.67977527e-01 -5.03501058e-01 4.64121997e-01 5.08653045e-01 -1.75870824e+00 3.97991478e-01 5.48304021e-01 9.72730339e-01 1.13100040e+00 6.73865378e-01 5.73951565e-02 7.73648202e-01 -1.76632822e-01 1.62283525e-01 2.13136792e-01 -2.16572583e-01 -5.90913653e-01 1.58519760e-01 5.84379792e-01 -5.73697746e-01 -5.16249418e-01 8.70785236e-01 2.21438408e-01 6.25148654e-01 5.48370540e-01 -1.10658240e+00 -8.38187575e-01 2.82888234e-01 -3.98283571e-01 9.38795656e-02 8.77458990e-01 4.76832509e-01 9.75356221e-01 -9.80441093e-01 6.31621301e-01 1.25489426e+00 -3.36665176e-02 3.30913305e-01 -9.39472854e-01 -4.57159698e-01 6.89522386e-01 5.56475937e-01 -1.37868595e+00 -5.40746152e-01 5.96876144e-01 -3.06007177e-01 8.92292917e-01 4.46045816e-01 6.56050622e-01 1.09368181e+00 1.64204985e-01 5.79173505e-01 1.16954327e+00 -4.51557577e-01 2.69870311e-01 -4.80373539e-02 -8.10660422e-02 9.43943083e-01 4.71284054e-02 3.10711145e-01 -7.61958122e-01 3.43845338e-01 1.09003544e+00 2.22671047e-01 -5.98794103e-01 -5.29407799e-01 -1.54946721e+00 5.78124166e-01 7.83290148e-01 3.24920267e-01 -5.08203864e-01 1.39324620e-01 -1.01504438e-01 -1.22991212e-01 -2.82629132e-01 2.04564735e-01 -3.21185052e-01 -4.91656363e-02 -4.31538552e-01 1.13412380e-01 4.31901395e-01 1.31414425e+00 9.66362596e-01 -1.24402799e-01 -1.24295354e-01 3.37595880e-01 6.04904115e-01 6.12908602e-01 1.80882275e-01 -1.12269723e+00 7.74850667e-01 2.97024727e-01 4.87139314e-01 -1.24922419e+00 -4.75730091e-01 -6.91644028e-02 -5.20843923e-01 2.40443945e-01 5.63099861e-01 1.90558463e-01 -1.08687484e+00 1.78827655e+00 3.77747774e-01 -1.94080900e-02 -4.54806425e-02 1.21745956e+00 7.11810112e-01 7.11283147e-01 -1.21055335e-01 1.99934497e-01 1.62691200e+00 -1.14965761e+00 -7.25386918e-01 -9.12151158e-01 4.27223086e-01 -6.25518858e-01 1.50691056e+00 2.21538782e-01 -5.94992816e-01 -8.58544827e-01 -8.66297245e-01 -3.91375124e-01 -6.46519721e-01 3.30352902e-01 6.85986161e-01 1.92692161e-01 -1.06877315e+00 -7.66064823e-02 -6.89206421e-01 -5.70873022e-01 -9.79043692e-02 -2.69483123e-03 -5.35695732e-01 -4.51695651e-01 -9.73238170e-01 1.02548361e+00 2.81758130e-01 5.52890897e-01 -1.03585267e+00 -6.77414984e-02 -1.28387237e+00 -6.70427456e-02 5.24918795e-01 -7.76721179e-01 1.21631289e+00 -6.92625821e-01 -1.30632043e+00 8.16879511e-01 -5.67834914e-01 -5.79771325e-02 2.73849994e-01 -2.45851785e-01 -4.36437458e-01 2.93531176e-02 3.97375852e-01 9.89716947e-01 5.72124481e-01 -1.60812402e+00 -9.90781307e-01 -5.78743935e-01 4.95812386e-01 7.38349378e-01 3.81992042e-01 -3.52003664e-01 -7.22805798e-01 -2.02458382e-01 5.42759299e-01 -8.26740146e-01 -2.99364746e-01 2.31540963e-01 -6.09644353e-01 1.73553322e-02 4.40005392e-01 -6.78082168e-01 9.41368341e-01 -2.11071825e+00 4.66186851e-02 2.16294765e-01 -9.02016684e-02 -4.16482925e-01 -4.22165282e-02 4.76172149e-01 1.36402711e-01 -1.78672746e-01 1.63808405e-01 -3.38450223e-01 6.79001510e-02 4.58585978e-01 -4.07237589e-01 4.54083681e-01 -5.91127753e-01 8.58843863e-01 -1.07038045e+00 -2.10159644e-01 6.99432850e-01 4.84259158e-01 -3.38763326e-01 1.39960274e-01 -3.14569801e-01 7.71850705e-01 -6.03161335e-01 4.32997823e-01 4.22658622e-01 -1.35327965e-01 2.49413606e-02 -1.44320786e-01 -5.04028022e-01 5.27730525e-01 -1.27586293e+00 2.08173966e+00 -7.52757132e-01 6.87233686e-01 -8.39948431e-02 -3.54617000e-01 6.76510811e-01 -6.22643642e-02 -2.87969291e-01 -1.19433296e+00 -8.45669881e-02 1.53856516e-01 -1.20462649e-01 -5.21924078e-01 6.02639616e-01 4.48067963e-01 -2.29195803e-01 2.54757911e-01 -5.20124853e-01 -3.52070332e-02 1.37731135e-01 5.77617995e-02 6.99199617e-01 4.00784105e-01 6.75679624e-01 -6.65190965e-02 4.61674362e-01 -2.35473197e-02 2.66005248e-01 1.08110535e+00 -9.51039493e-02 6.00760818e-01 2.04603206e-02 -6.11578166e-01 -6.59344018e-01 -1.00398350e+00 3.41139138e-01 1.35850716e+00 7.80433536e-01 -2.67881960e-01 -3.69880617e-01 -4.73739564e-01 -3.22143644e-01 1.43251669e+00 -7.12679386e-01 1.80844128e-01 -6.79159880e-01 1.22233890e-01 -1.21537663e-01 4.68932658e-01 5.22838414e-01 -1.29187500e+00 -1.30435145e+00 -1.08717009e-01 -6.78011775e-01 -1.18393803e+00 -8.40906024e-01 2.46288404e-01 -4.00171131e-01 -9.33689058e-01 -2.88663000e-01 -9.04031277e-01 9.86755192e-01 7.13898957e-01 1.00376427e+00 -2.26665251e-02 7.66221657e-02 6.13091290e-01 -3.08012486e-01 -6.49505183e-02 7.73292854e-02 -4.98331696e-01 1.21358912e-02 -1.99034577e-03 1.94414645e-01 -3.52033883e-01 -9.90138590e-01 6.31615996e-01 -4.32435900e-01 6.11347675e-01 4.36314911e-01 3.92076731e-01 7.52143979e-01 -1.63060948e-01 -3.19760084e-01 -3.90637845e-01 3.15547526e-01 -2.11754903e-01 -7.59941041e-01 3.90297592e-01 -2.66468793e-01 2.30304264e-02 4.80841190e-01 -2.40612641e-01 -1.05994141e+00 8.59202221e-02 5.17378300e-02 -8.73173848e-02 -6.73031390e-01 3.77816647e-01 -4.29022074e-01 2.03694284e-01 5.20018816e-01 5.78437030e-01 -5.79028487e-01 -7.63520822e-02 8.33401561e-01 3.46170843e-01 5.86555183e-01 -4.73939359e-01 5.23229599e-01 7.59731948e-01 -1.28999040e-01 -5.16001463e-01 -7.78424323e-01 -5.10072947e-01 -6.89058542e-01 -2.91977733e-01 9.98781681e-01 -7.75989890e-01 -9.40942585e-01 1.33549809e-01 -1.31023741e+00 -5.64473987e-01 -2.74113435e-02 5.49167573e-01 -7.15436816e-01 1.17298067e-01 -9.55817550e-02 -6.87324524e-01 2.63587743e-01 -1.27103043e+00 9.78454173e-01 4.14968938e-01 -2.46521741e-01 -7.68093884e-01 -3.56395304e-01 2.34631509e-01 1.76161632e-01 -5.27995713e-02 7.87784636e-01 -2.70087987e-01 -9.77200806e-01 1.54223278e-01 -3.11307818e-01 -3.39619458e-01 3.97080719e-01 -2.88532972e-01 -7.17128694e-01 6.29271418e-02 1.33685589e-01 6.46408647e-03 7.13141680e-01 6.12400472e-01 1.10538089e+00 -3.49477142e-01 -5.01570880e-01 7.04787493e-01 1.39103186e+00 7.80820549e-01 6.85748816e-01 5.99257112e-01 7.42853105e-01 6.85602009e-01 9.58514750e-01 3.98835868e-01 1.00890899e+00 9.34177041e-01 7.64664948e-01 1.03728406e-01 -8.48147422e-02 -5.66493869e-01 3.32411647e-01 2.88462311e-01 -9.46874395e-02 -5.29678762e-01 -9.85870361e-01 4.23611879e-01 -1.90122139e+00 -9.43165004e-01 1.69867612e-02 2.25546336e+00 4.87626702e-01 7.28053451e-02 -4.87564027e-01 -4.41370040e-01 4.22193766e-01 4.95466113e-01 -5.73726892e-01 -1.86271012e-01 5.48870973e-02 -4.13087815e-01 4.65434462e-01 8.39801550e-01 -8.71128142e-01 1.04675746e+00 4.91405916e+00 6.06925130e-01 -9.12365973e-01 -9.42075327e-02 4.50016201e-01 1.81273427e-02 -2.73446053e-01 8.77094865e-02 -1.04081595e+00 2.90990204e-01 3.84403765e-01 2.73687124e-01 9.42103922e-01 6.91980779e-01 5.68650186e-01 -7.03371823e-01 -1.46558917e+00 1.21730411e+00 2.42030561e-01 -1.12661946e+00 -2.70996951e-02 1.40528160e-03 3.08664113e-01 -8.37877244e-02 2.78114349e-01 2.08568588e-01 3.15876335e-01 -1.02986217e+00 1.15746224e+00 6.70265079e-01 6.56268656e-01 -3.79249036e-01 1.72599018e-01 5.88631570e-01 -1.44146514e+00 -1.66920185e-01 -2.07974449e-01 -2.97208309e-01 4.48152006e-01 -7.90283531e-02 -7.60075331e-01 4.92863625e-01 7.94546366e-01 4.94831473e-01 -4.73255247e-01 7.57353544e-01 -7.48541951e-01 -9.12125260e-02 -2.17742324e-01 -2.44532123e-01 4.58984911e-01 -3.67891580e-01 3.88778955e-01 7.34720290e-01 5.41572273e-01 3.05363774e-01 2.76022911e-01 9.06792402e-01 4.09188628e-01 -2.13237450e-01 -7.57806003e-01 5.17780364e-01 5.64432323e-01 8.94053340e-01 -8.54773641e-01 -5.04186824e-02 -4.55078721e-01 1.04560053e+00 2.75346786e-01 8.33793283e-01 -8.58011961e-01 -1.22558504e-01 7.27172673e-01 1.77435845e-01 4.55341756e-01 -6.01702332e-01 -1.42189220e-01 -8.02131414e-01 1.84228823e-01 -7.03221798e-01 1.64369851e-01 -1.40549660e+00 -6.89298093e-01 6.51806295e-01 2.35667787e-02 -1.38383663e+00 -1.95760220e-01 -6.96967185e-01 -3.87579829e-01 9.23747420e-01 -1.52268612e+00 -1.22521400e+00 -6.96172297e-01 7.03824162e-01 8.45127523e-01 2.28971511e-01 8.93260360e-01 1.38290506e-02 -2.15336606e-01 4.52105068e-02 -1.78280309e-01 1.25668645e-01 5.39895773e-01 -1.09740114e+00 4.90000248e-01 8.97375226e-01 4.96976942e-01 8.96913886e-01 8.06907594e-01 -5.56086004e-01 -1.38503289e+00 -6.11548305e-01 9.02308583e-01 -7.07784653e-01 3.70014131e-01 -6.35749221e-01 -4.74253923e-01 1.00439394e+00 1.87224030e-01 -3.13572213e-02 3.59245926e-01 1.08189695e-02 -3.98905665e-01 -5.92921712e-02 -7.21012950e-01 1.25501156e+00 1.29300308e+00 -5.23920715e-01 -5.57943165e-01 4.46042955e-01 7.63193786e-01 -8.06592405e-01 8.14305618e-02 -9.60229859e-02 6.59372807e-01 -1.28686213e+00 1.08179808e+00 -3.22466463e-01 2.14759499e-01 -7.17597723e-01 -6.76423192e-01 -1.30504632e+00 -3.85581613e-01 -2.23127887e-01 7.56576285e-02 7.78598547e-01 3.64661813e-01 -5.65592468e-01 3.61238956e-01 6.78281009e-01 -2.29148299e-01 -5.63871920e-01 -9.48623776e-01 -4.11939830e-01 -8.37185860e-01 -5.89363813e-01 6.70436740e-01 4.90133196e-01 -2.37434283e-01 2.84241468e-01 -4.15107906e-01 6.04949057e-01 2.35771477e-01 4.44832116e-01 8.40721130e-01 -7.27592707e-01 -2.37418618e-03 -4.24899310e-01 -2.15898260e-01 -1.87930942e+00 -8.88119936e-02 -5.51809609e-01 3.61864269e-01 -2.21946430e+00 -1.62728637e-01 -4.24624383e-01 -1.92052260e-01 5.15631199e-01 5.41110300e-02 -3.78061086e-01 3.08722794e-01 2.97916923e-02 -7.78938949e-01 5.03192604e-01 1.46209061e+00 -7.37298727e-02 -2.98613310e-01 3.81303728e-02 -6.85338557e-01 7.54060209e-01 6.97059631e-01 -1.58962801e-01 -5.86127698e-01 -8.78903091e-01 3.22280586e-01 2.49023333e-01 6.39793634e-01 -8.80278230e-01 6.40877366e-01 -5.74846506e-01 4.54376578e-01 -8.19746614e-01 7.90253699e-01 -1.17460918e+00 5.67859150e-02 3.52167159e-01 -2.87359983e-01 3.98823828e-01 1.16830669e-01 7.08070636e-01 8.88106748e-02 -6.14528582e-02 3.08217615e-01 -4.01864469e-01 -1.53184450e+00 1.88536659e-01 -2.87438720e-01 -2.14540452e-01 9.84494686e-01 -5.35483956e-01 -4.79555398e-01 -6.39844596e-01 -5.27688265e-01 5.23628891e-01 4.89060581e-01 7.32375443e-01 1.15683532e+00 -1.25269616e+00 -2.81111151e-01 3.67599428e-01 3.30192566e-01 1.41266659e-01 5.46081960e-01 7.86796093e-01 -5.13842463e-01 7.04828262e-01 -3.76563847e-01 -5.83774090e-01 -1.05694294e+00 7.14189470e-01 3.47842813e-01 9.76176858e-02 -6.56284034e-01 9.38191414e-01 8.81643951e-01 -5.02901554e-01 3.33326757e-01 -6.48867309e-01 -4.91714269e-01 -1.90464810e-01 5.96132934e-01 -4.24096361e-03 -2.85962313e-01 -8.58915865e-01 -4.48249102e-01 7.24686325e-01 1.80945963e-01 -3.26375961e-01 8.55878830e-01 -6.66808665e-01 -1.06871434e-01 7.56821334e-01 6.71971440e-01 2.44175643e-01 -1.55912268e+00 -3.48073661e-01 -3.02074730e-01 -7.02342689e-01 -1.62125722e-01 -9.49377060e-01 -6.46050036e-01 8.06250095e-01 5.11868715e-01 -1.58758223e-01 1.01517379e+00 1.22016191e-01 2.85870880e-01 5.28507888e-01 8.76433253e-01 -7.78538883e-01 -4.93684458e-03 7.17341423e-01 1.11986661e+00 -1.34305847e+00 -1.81309119e-01 -2.50526190e-01 -7.27016389e-01 9.91223693e-01 9.03445542e-01 4.56099808e-01 3.48662019e-01 -1.63887754e-01 1.47899389e-01 -3.27105552e-01 -6.24724925e-01 -3.60805094e-01 4.33705658e-01 7.08993375e-01 4.67975996e-02 3.63076895e-01 3.11907291e-01 3.63273740e-01 -3.80425215e-01 -5.66310525e-01 2.41538882e-01 8.95763636e-01 -6.11951888e-01 -4.74779934e-01 -3.63601685e-01 7.61948004e-02 3.74609292e-01 -2.36496761e-01 -9.89796370e-02 8.28138888e-01 2.60851234e-01 1.17693043e+00 7.53166005e-02 -2.71325111e-01 4.80461746e-01 -2.84306109e-01 4.49207395e-01 -7.39299178e-01 -1.41060770e-01 -9.90067422e-02 -3.16206105e-02 -9.13707376e-01 -3.04303169e-01 -6.04787111e-01 -1.56506133e+00 -7.12682381e-02 1.19814381e-01 -1.41183212e-01 6.72162592e-01 9.82729018e-01 2.99820870e-01 5.53649008e-01 3.47202867e-01 -1.09763718e+00 -7.68878460e-02 -5.74624836e-01 -3.32152873e-01 2.79032290e-01 4.70280915e-01 -8.06615651e-01 -2.34451145e-01 -3.49163823e-02]
[4.499335289001465, 0.4764355421066284]
964828ac-5db8-4437-a372-dba60cd3c5ad
convolutional-neural-networks-for-automatic
1902.09600
null
http://arxiv.org/abs/1902.09600v1
http://arxiv.org/pdf/1902.09600v1.pdf
Convolutional Neural Networks for Automatic Meter Reading
In this paper, we tackle Automatic Meter Reading (AMR) by leveraging the high capability of Convolutional Neural Networks (CNNs). We design a two-stage approach that employs the Fast-YOLO object detector for counter detection and evaluates three different CNN-based approaches for counter recognition. In the AMR literature, most datasets are not available to the research community since the images belong to a service company. In this sense, we introduce a new public dataset, called UFPR-AMR dataset, with 2,000 fully and manually annotated images. This dataset is, to the best of our knowledge, three times larger than the largest public dataset found in the literature and contains a well-defined evaluation protocol to assist the development and evaluation of AMR methods. Furthermore, we propose the use of a data augmentation technique to generate a balanced training set with many more examples to train the CNN models for counter recognition. In the proposed dataset, impressive results were obtained and a detailed speed/accuracy trade-off evaluation of each model was performed. In a public dataset, state-of-the-art results were achieved using less than 200 images for training.
['Gabriel R. Gonçalves', 'Rayson Laroca', 'William Robson Schwartz', 'Victor Barroso', 'David Menotti', 'Matheus A. Diniz']
2019-02-25
null
null
null
null
['image-based-automatic-meter-reading']
['computer-vision']
[ 1.56231940e-01 -2.50757784e-01 -3.74871105e-01 -1.44503757e-01 -8.64998639e-01 -2.04307497e-01 6.92971408e-01 2.45799154e-01 -5.93156099e-01 7.38286495e-01 6.64699590e-03 -4.32545960e-01 2.40610018e-01 -7.92897940e-01 -5.32251596e-01 -4.00001317e-01 4.76860553e-02 3.08992058e-01 -5.04479259e-02 -1.88220948e-01 5.40058553e-01 4.95518446e-01 -1.54037058e+00 2.71948546e-01 7.17242897e-01 1.55319941e+00 1.16297975e-02 8.94071341e-01 6.35767654e-02 1.26233876e+00 -1.01808214e+00 -3.45906734e-01 4.16721702e-01 -3.78308296e-01 -5.53787112e-01 5.14164716e-02 5.49745142e-01 -4.91213739e-01 -4.58057195e-01 8.93085718e-01 7.14169145e-01 -1.61236688e-01 4.78865832e-01 -1.02217579e+00 -1.07652795e+00 7.30329454e-01 -5.06622493e-01 6.39456213e-01 3.66884291e-01 1.16138086e-01 1.00796390e+00 -6.60697460e-01 7.29354518e-03 6.67305470e-01 6.29665434e-01 3.84614617e-01 -9.02133405e-01 -7.05295205e-01 -5.09162657e-02 4.82688785e-01 -1.55601287e+00 -2.20067695e-01 7.98218310e-01 -3.82853448e-01 1.07557464e+00 4.79784220e-01 7.63129354e-01 1.21035671e+00 -1.94232002e-01 1.07519078e+00 1.29613531e+00 -6.67897046e-01 3.84200633e-01 3.24469805e-01 1.44108474e-01 4.06872153e-01 4.87956762e-01 -1.62366126e-02 -3.75450522e-01 7.67412111e-02 5.41428030e-01 -1.33335948e-01 -1.51062250e-01 6.36333376e-02 -1.04706824e+00 8.63646150e-01 2.52999455e-01 8.08909714e-01 -3.09525579e-01 1.63435712e-01 6.41851723e-01 5.87480739e-02 4.58101392e-01 3.60452831e-01 -3.85985941e-01 -1.16619453e-01 -1.15912414e+00 2.19397128e-01 8.98320675e-01 8.61346722e-01 1.92001328e-01 4.11887139e-01 -2.87916452e-01 6.22910500e-01 3.51287052e-03 4.63326335e-01 8.47873926e-01 -4.95095283e-01 8.41657460e-01 8.20366740e-01 2.46357456e-01 -1.08290386e+00 -5.35005093e-01 -6.07555032e-01 -1.14683211e+00 6.57948386e-03 5.11899531e-01 2.20296979e-01 -9.03163850e-01 9.99757051e-01 -1.29597634e-01 1.47513419e-01 2.01803550e-01 6.72979712e-01 9.44249511e-01 3.87456805e-01 -5.22230640e-02 -1.40623208e-02 1.37527800e+00 -1.01215196e+00 -7.64855862e-01 -4.95650545e-02 6.76511049e-01 -5.55436850e-01 1.05230582e+00 7.09142268e-01 -8.66558135e-01 -7.34350920e-01 -1.44537759e+00 1.57965064e-01 -7.58811235e-01 5.37500620e-01 5.30100286e-01 1.10319686e+00 -6.72499955e-01 5.25094926e-01 -3.28623980e-01 -1.94774792e-01 6.63337886e-01 1.16484821e-01 -1.89055793e-03 5.43858893e-02 -1.19731498e+00 1.15877044e+00 4.86405551e-01 2.96943545e-01 -8.93524051e-01 -3.04768354e-01 -7.38053679e-01 9.08560380e-02 3.11791956e-01 -1.49944499e-01 1.46301532e+00 -1.11551034e+00 -1.37257743e+00 8.94477129e-01 4.72277254e-01 -1.06446135e+00 7.53463864e-01 -2.24289790e-01 -7.58885026e-01 7.80569240e-02 -3.35419402e-02 2.73342550e-01 7.40868151e-01 -1.05550838e+00 -7.42604554e-01 -5.75601421e-02 5.17230444e-02 -2.25450188e-01 -3.87843430e-01 5.55116460e-02 -1.74822450e-01 -8.46106708e-01 -4.56815869e-01 -5.24901867e-01 -1.07562773e-01 -6.90881610e-01 -5.09152532e-01 -1.22638032e-01 6.10077024e-01 -7.48856544e-01 1.40191865e+00 -1.88246036e+00 -5.88169217e-01 7.13476017e-02 3.74355018e-01 7.87560463e-01 6.03314750e-02 1.20029137e-01 -2.74879515e-01 6.97311088e-02 -8.05320814e-02 -3.93972486e-01 3.49350959e-01 -2.36470643e-02 -1.90899909e-01 6.98415458e-01 1.02680467e-01 1.03566301e+00 -6.89779699e-01 -3.10379624e-01 6.03846610e-01 3.65675867e-01 -1.06166676e-01 1.54705271e-01 -6.10286035e-02 2.47082412e-01 -4.12084490e-01 9.67526734e-01 7.18910873e-01 -2.42661282e-01 -7.34997988e-02 -1.43256545e-01 -1.19015671e-01 8.63552094e-02 -1.36596656e+00 1.08138490e+00 -4.03670132e-01 6.69839203e-01 -5.03324151e-01 -1.23600829e+00 1.06415570e+00 3.83017689e-01 5.75602233e-01 -1.32543266e+00 5.80936253e-01 4.22934175e-01 -1.22541867e-01 -4.25527453e-01 9.85363960e-01 3.14848602e-01 -1.59817949e-01 1.62696809e-01 -1.89932540e-01 2.81497061e-01 4.34941411e-01 -2.34595403e-01 1.00679088e+00 -2.00326219e-01 5.04692674e-01 -6.26377687e-02 8.18388700e-01 1.29306898e-01 2.15179145e-01 1.15145791e+00 -4.21485901e-01 6.21042967e-01 6.42416999e-02 -7.44743645e-01 -1.23127007e+00 -6.11148953e-01 -2.93193728e-01 6.76029801e-01 -2.58943498e-01 -2.34151155e-01 -8.25960875e-01 -3.85988742e-01 -8.88942629e-02 6.08199298e-01 -5.97757220e-01 2.43873626e-01 -7.20734417e-01 -1.01061320e+00 8.80338728e-01 9.67155635e-01 1.04856694e+00 -1.34571815e+00 -8.75435174e-01 4.13452923e-01 -8.15503672e-02 -1.51555407e+00 -2.44460851e-01 2.80517578e-01 -6.38622940e-01 -1.44802201e+00 -8.56547058e-01 -4.91061270e-01 2.39514112e-01 6.93912804e-02 1.41535366e+00 3.30485374e-01 -4.58416998e-01 3.00745487e-01 -6.16837204e-01 -6.40223145e-01 -3.55221897e-01 2.45469376e-01 -2.99441278e-01 -3.77637856e-02 6.65292978e-01 -1.88058972e-01 -6.73366010e-01 1.13665164e-01 -7.15424001e-01 -3.30031693e-01 6.47694826e-01 5.55845976e-01 3.18383396e-01 6.02802783e-02 7.38736928e-01 -6.62619293e-01 6.04118943e-01 -2.29547575e-01 -8.92072737e-01 -4.73994650e-02 -8.26545119e-01 -4.13791090e-01 7.66644359e-01 -6.08456612e-01 -4.99940783e-01 5.01276068e-02 -2.90812045e-01 -1.71018869e-01 -3.33961219e-01 2.67086506e-01 3.93421464e-02 -3.42875980e-02 6.35483265e-01 4.06154603e-01 -6.66031122e-01 -6.88489616e-01 4.04552788e-01 1.00006604e+00 7.38569856e-01 -2.01384619e-01 6.34250462e-01 2.87033021e-01 -1.12397917e-01 -6.73313200e-01 -1.01108253e+00 -6.51208103e-01 -5.57248950e-01 -1.83223143e-01 1.03319430e+00 -1.08227563e+00 -7.80896485e-01 8.99404526e-01 -7.22747326e-01 -3.49827528e-01 -3.70532751e-01 4.34002668e-01 -5.02698660e-01 1.96209580e-01 -5.22091687e-01 -1.17828882e+00 -6.93019092e-01 -1.04114819e+00 8.36703479e-01 1.14100255e-01 8.16588998e-02 -7.67755866e-01 -4.50914018e-02 5.34571767e-01 7.60133803e-01 3.76927465e-01 3.61341029e-01 -9.34225738e-01 -5.89402020e-01 -6.62319720e-01 -4.00505871e-01 7.49733806e-01 -1.90314069e-01 -3.40706289e-01 -1.31343436e+00 -1.89012647e-01 -2.96986233e-02 -3.74823540e-01 1.01893556e+00 2.57715553e-01 1.56166911e+00 -2.32279003e-01 7.84736723e-02 2.42779851e-01 1.50478697e+00 3.15094948e-01 1.16930568e+00 9.20049548e-01 7.63041615e-01 -2.30522454e-02 3.64296854e-01 6.57380819e-01 5.21318495e-01 5.60591340e-01 4.70695168e-01 -3.43737990e-01 1.28161944e-02 -9.89672635e-03 1.69519380e-01 7.32283652e-01 -2.21164435e-01 -4.26925689e-01 -7.07708240e-01 5.79076111e-01 -1.60635841e+00 -9.24727798e-01 -3.74183297e-01 2.08659935e+00 5.40474355e-01 3.29837054e-01 3.26912135e-01 8.53055120e-01 5.26401818e-01 2.32325956e-01 -1.56287000e-01 -3.20394069e-01 -5.21989074e-03 4.27521735e-01 9.02839661e-01 1.45249069e-01 -1.49889350e+00 5.76695442e-01 7.05609989e+00 8.23369205e-01 -1.03683233e+00 1.77790269e-01 9.14286435e-01 1.92207158e-01 3.60770196e-01 -4.88497674e-01 -1.02062130e+00 4.46745872e-01 1.19171166e+00 2.05063194e-01 3.50889295e-01 1.05252862e+00 2.04312384e-01 -4.01237160e-01 -8.38059545e-01 1.11486554e+00 4.50966150e-01 -1.42471004e+00 -3.54401261e-01 5.30395769e-02 6.60053074e-01 5.28710783e-02 5.66911250e-02 5.07209063e-01 5.46367429e-02 -1.29708540e+00 9.02907014e-01 5.94040811e-01 6.58819497e-01 -8.63137364e-01 1.00854957e+00 3.23183447e-01 -1.39389420e+00 -4.38803494e-01 -3.56917620e-01 -2.91984558e-01 -2.44434074e-01 5.93680620e-01 -5.21058977e-01 6.74585581e-01 7.16859460e-01 4.54774737e-01 -1.04223955e+00 1.07110989e+00 -4.46366966e-02 6.47227466e-01 -6.91534057e-02 -2.35332906e-01 1.90462455e-01 1.93439677e-01 1.81797706e-02 1.35377097e+00 2.61206657e-01 -2.12738261e-01 2.53486663e-01 8.32137108e-01 -2.24066690e-01 3.46518427e-01 -4.63310957e-01 -1.12750240e-01 1.92367733e-01 1.47918701e+00 -6.75672889e-01 -7.02554047e-01 -5.12765765e-01 6.27171457e-01 1.01882607e-01 -4.35399041e-02 -9.89205241e-01 -1.81702048e-01 6.13585720e-03 6.54988587e-02 4.95989859e-01 -6.44768551e-02 -5.14155924e-01 -1.15850329e+00 1.32366195e-01 -1.03962731e+00 3.74298513e-01 -5.64446330e-01 -1.29392838e+00 5.08869290e-01 9.42982882e-02 -1.28819585e+00 -3.42256241e-02 -9.10874665e-01 -6.24474943e-01 7.13598907e-01 -1.92509520e+00 -8.94332767e-01 -6.69892788e-01 3.77710849e-01 5.01214206e-01 -4.66948986e-01 9.04508173e-01 8.27268302e-01 -7.15849519e-01 7.19671190e-01 1.35911450e-01 4.97969329e-01 1.91493124e-01 -1.47917914e+00 5.17037749e-01 7.55040646e-01 1.50195107e-01 4.52630706e-02 4.59925771e-01 -2.44703948e-01 -1.17194486e+00 -1.29236829e+00 8.54376793e-01 -4.62154329e-01 6.72019422e-01 -4.50999886e-01 -7.08841741e-01 6.73769653e-01 3.39073658e-01 1.20742030e-01 4.89361227e-01 -3.12422603e-01 -2.71861076e-01 -2.24010110e-01 -1.23681724e+00 2.41155475e-01 4.91840422e-01 -3.68489325e-01 -4.18789268e-01 1.53360486e-01 1.56114817e-01 -4.97693449e-01 -9.61764097e-01 3.62260520e-01 3.50042522e-01 -1.02148306e+00 9.53074574e-01 -7.41515532e-02 3.56909454e-01 -2.85647541e-01 -4.34354067e-01 -8.84219289e-01 -5.16269393e-02 -2.23730534e-01 -5.25479496e-01 1.19944906e+00 3.07348669e-01 -6.13524616e-01 6.84511125e-01 5.20836040e-02 -5.55352047e-02 -8.39474559e-01 -8.44800234e-01 -8.40311170e-01 9.29563493e-02 -7.83817708e-01 7.60081112e-01 8.28187943e-01 -2.44850263e-01 1.57803044e-01 -6.26145840e-01 -6.12366796e-02 4.15299147e-01 5.53681776e-02 8.31261456e-01 -8.55184674e-01 -2.74656147e-01 -4.71521556e-01 -5.90269804e-01 -8.79790127e-01 -1.64719999e-01 -7.63892174e-01 -1.57361865e-01 -1.44635320e+00 2.86140203e-01 -2.40427598e-01 -4.14241374e-01 3.88138145e-01 -2.49349345e-02 8.87903929e-01 3.97808313e-01 8.31896141e-02 -5.94038546e-01 3.19838881e-01 1.03432918e+00 -4.02671903e-01 7.90099725e-02 -3.73127051e-02 -7.06826627e-01 7.18021154e-01 1.29351366e+00 -3.99386078e-01 -4.16074060e-02 -4.04184014e-02 1.73329283e-02 -2.74755001e-01 5.35691082e-01 -1.31237042e+00 -2.14366585e-01 1.80413574e-01 9.20863509e-01 -1.11934495e+00 7.66534731e-02 -6.00836754e-01 -1.49905115e-01 4.82502669e-01 -2.22230375e-01 1.94940448e-01 9.98153910e-02 2.66379476e-01 -5.93965873e-02 -4.23379838e-01 9.10560310e-01 -4.48733687e-01 -7.59547651e-01 -5.42230122e-02 -4.51619387e-01 6.06986471e-02 9.49969888e-01 -3.06536168e-01 -4.26118702e-01 -1.82705462e-01 -4.49924409e-01 -8.04437175e-02 -3.62456441e-02 3.99984539e-01 4.49739605e-01 -1.52479148e+00 -6.12001061e-01 7.45935831e-03 3.70263308e-01 -2.32964665e-01 -2.22872701e-02 9.50639188e-01 -5.10698915e-01 5.86019635e-01 -9.16076899e-02 -4.68682706e-01 -9.40430582e-01 7.68153489e-01 6.18765533e-01 -7.30074108e-01 -8.72492552e-01 1.53398231e-01 -2.91157812e-01 -1.59937873e-01 4.55194563e-01 -6.22042120e-01 -6.55327797e-01 5.90218976e-02 9.27839100e-01 4.85086590e-01 4.13222224e-01 -6.56722605e-01 -1.70768052e-01 1.80113733e-01 5.37081882e-02 2.50172079e-01 1.25600076e+00 1.10856436e-01 2.62917191e-01 5.41758597e-01 1.00576007e+00 -1.78529009e-01 -8.65489602e-01 -2.87262034e-02 2.11599335e-01 -5.50878525e-01 1.40411332e-01 -8.19885790e-01 -1.29841912e+00 7.12473214e-01 1.11113000e+00 5.75794995e-01 1.17376888e+00 -3.86489272e-01 8.87510121e-01 3.83587569e-01 1.74294248e-01 -1.52926493e+00 3.82249415e-01 5.07877231e-01 7.00647831e-01 -1.27014732e+00 -2.78281420e-02 1.21033555e-02 -4.81119394e-01 1.07483172e+00 4.94532079e-01 -2.80890971e-01 3.10622543e-01 4.14831370e-01 1.25982687e-01 -1.55499175e-01 -2.27795854e-01 -5.19283414e-01 2.53325105e-01 6.19340181e-01 4.09617782e-01 2.53215909e-01 -4.91464466e-01 6.92957163e-01 -2.23871157e-01 3.18450451e-01 7.17818499e-01 1.02742577e+00 -4.89351064e-01 -8.77082109e-01 -4.96703863e-01 5.50258100e-01 -8.61180961e-01 -1.29355595e-01 -2.00446144e-01 1.08447695e+00 2.92912602e-01 9.84539986e-01 -6.95842365e-03 -3.44962150e-01 5.48574328e-01 -9.55524072e-02 5.36411345e-01 -2.40347952e-01 -7.23653674e-01 -1.01431042e-01 2.07029641e-01 -3.15845698e-01 -5.06274760e-01 -3.61391693e-01 -8.70539069e-01 -4.85217839e-01 -4.49650198e-01 -2.94269830e-01 5.93537986e-01 1.07128370e+00 -2.59191632e-01 8.12016964e-01 7.51452565e-01 -9.51201260e-01 -7.67239153e-01 -1.25329435e+00 -7.65858710e-01 3.94493699e-01 1.11104548e-01 -4.83928293e-01 -1.75289139e-01 -1.18565299e-02]
[11.343738555908203, 2.625261068344116]
dfe635be-d35a-4ef1-95b7-c0a950444332
gam-changer-editing-generalized-additive
2112.03245
null
https://arxiv.org/abs/2112.03245v1
https://arxiv.org/pdf/2112.03245v1.pdf
GAM Changer: Editing Generalized Additive Models with Interactive Visualization
Recent strides in interpretable machine learning (ML) research reveal that models exploit undesirable patterns in the data to make predictions, which potentially causes harms in deployment. However, it is unclear how we can fix these models. We present our ongoing work, GAM Changer, an open-source interactive system to help data scientists and domain experts easily and responsibly edit their Generalized Additive Models (GAMs). With novel visualization techniques, our tool puts interpretability into action -- empowering human users to analyze, validate, and align model behaviors with their knowledge and values. Built using modern web technologies, our tool runs locally in users' computational notebooks or web browsers without requiring extra compute resources, lowering the barrier to creating more responsible ML models. GAM Changer is available at https://interpret.ml/gam-changer.
['Rich Caruana', 'Jennifer Wortman Vaughan', 'Mihaela Vorvoreanu', 'Duen Horng Chau', 'Mark Nunnally', 'Peter Stella', 'Harsha Nori', 'Alex Kale', 'Zijie J. Wang']
2021-12-06
null
null
null
null
['additive-models']
['methodology']
[-1.34111613e-01 5.49510121e-01 -2.73853183e-01 -5.42821050e-01 -3.29782873e-01 -7.14130819e-01 1.67642906e-01 1.91053480e-01 -3.44265178e-02 2.72778481e-01 1.37447178e-01 -1.05880296e+00 -8.94372389e-02 -5.36939144e-01 -4.82578993e-01 -8.07846244e-03 7.83202350e-02 4.14758146e-01 -2.39008084e-01 -4.30785380e-02 4.38148677e-01 9.84038636e-02 -1.33312237e+00 5.21237493e-01 9.24200714e-01 4.41212326e-01 -1.14743158e-01 6.57251418e-01 1.88505985e-02 1.35003138e+00 -4.18615907e-01 -5.41413665e-01 4.70642358e-01 -2.36306876e-01 -6.30038440e-01 -5.81197500e-01 2.78141558e-01 -4.75457519e-01 2.96519250e-01 6.82649910e-01 2.23363295e-01 -1.38240412e-01 3.23223293e-01 -1.85096514e+00 -9.46870327e-01 8.04298759e-01 -4.88258898e-01 -3.40293273e-02 2.69900948e-01 8.92617941e-01 7.76147008e-01 -5.74344873e-01 3.00359070e-01 1.18718457e+00 1.01716542e+00 4.31073487e-01 -1.37622833e+00 -8.52408111e-01 1.31662875e-01 -7.11041735e-03 -1.20736921e+00 -5.39222777e-01 4.20648366e-01 -8.02043319e-01 1.10563362e+00 8.94139647e-01 7.15683043e-01 9.29980814e-01 1.16349265e-01 5.11518061e-01 1.18796837e+00 -4.53930289e-01 4.44395810e-01 4.75083560e-01 4.76095408e-01 8.72451782e-01 3.44087780e-01 -1.08782217e-01 -5.98041534e-01 -5.22448778e-01 7.37918794e-01 6.54691532e-02 1.41070470e-01 -9.75317433e-02 -6.97556674e-01 6.45225286e-01 1.45094037e-01 -9.38745737e-02 -5.58918118e-01 3.71553510e-01 -1.80427060e-01 2.70258635e-01 7.13792682e-01 7.73118198e-01 -7.43873239e-01 -8.46366525e-01 -6.29019976e-01 2.61807144e-01 9.96300817e-01 7.67128885e-01 7.55231738e-01 7.00854585e-02 1.37799397e-01 5.71884036e-01 6.31660223e-01 2.35704482e-01 2.23904233e-02 -1.11975932e+00 9.16771069e-02 1.05981493e+00 4.31608588e-01 -1.25981796e+00 -4.09529179e-01 -2.61657417e-01 -2.01387987e-01 6.04691446e-01 4.36000198e-01 -3.58933389e-01 -4.76243854e-01 1.29110956e+00 2.78221130e-01 1.40776217e-01 -4.90882874e-01 6.58099234e-01 4.91834670e-01 2.21467659e-01 6.73757613e-01 2.78320760e-01 9.39783573e-01 -6.03311718e-01 -3.98941159e-01 -4.57261652e-01 1.01390898e+00 -6.28437102e-01 1.84069526e+00 5.35120249e-01 -1.08008492e+00 -1.59493506e-01 -1.00484490e+00 -4.38034050e-02 -4.27592576e-01 -1.69480950e-01 9.86028612e-01 1.04441762e+00 -1.02263939e+00 6.93729162e-01 -1.15655625e+00 -1.92282632e-01 7.25686789e-01 2.56125838e-01 5.50105609e-02 3.68416816e-01 -7.76111662e-01 1.01420152e+00 4.54526618e-02 -1.57757089e-01 -3.96401972e-01 -1.40629303e+00 -4.38724607e-01 9.22051668e-02 3.76690835e-01 -8.86941433e-01 1.57265186e+00 -1.05671871e+00 -1.06912041e+00 5.24859607e-01 2.09581020e-04 -1.75992981e-01 7.73303568e-01 -4.43346560e-01 -1.82000279e-01 -7.87056684e-01 -4.11315441e-01 3.64776641e-01 4.09611642e-01 -1.10873306e+00 -3.65461469e-01 -3.21648002e-01 3.43721926e-01 4.16288078e-02 -3.76059592e-01 3.33651364e-01 -5.15010841e-02 -2.23053068e-01 -3.23660195e-01 -7.09587097e-01 -4.58418816e-01 1.93942204e-01 -2.36001879e-01 -6.45575393e-03 6.87608242e-01 -1.02757704e+00 1.79769313e+00 -1.81141591e+00 -4.09905285e-01 3.03159237e-01 6.34705901e-01 2.77559072e-01 7.06797987e-02 4.26158458e-01 -1.84169054e-01 9.81531203e-01 1.92402020e-01 -3.03844959e-01 4.22569066e-01 -1.94195554e-01 -1.54555455e-01 -1.16342939e-02 6.63204715e-02 8.48191202e-01 -9.22129989e-01 -1.74147472e-01 3.59715670e-01 3.18977594e-01 -7.88451791e-01 3.82209718e-02 -4.52932656e-01 2.66366005e-01 -2.20485419e-01 5.95780194e-01 6.71722412e-01 -3.00911784e-01 4.85056370e-01 1.42715141e-01 -2.29379505e-01 5.56641400e-01 -1.16213298e+00 1.11581802e+00 -5.91490269e-01 6.65907085e-01 -1.24213308e-01 -3.28050345e-01 7.66258776e-01 -2.69177735e-01 1.94339067e-01 -4.51793760e-01 -1.34524181e-01 -1.34630665e-01 -9.65652168e-02 -5.75930119e-01 4.80326086e-01 4.58712459e-01 2.24285036e-01 1.08848608e+00 -6.00905120e-01 1.16370924e-01 -1.47367731e-01 2.06499472e-01 1.22460949e+00 1.90749764e-01 5.33536553e-01 -2.60654420e-01 -4.63787705e-01 1.27903357e-01 4.48327720e-01 8.84075105e-01 2.96875387e-01 2.78704381e-03 6.88027859e-01 -7.35097229e-01 -1.01718414e+00 -1.08504367e+00 2.21411109e-01 1.43757486e+00 -5.13315082e-01 -9.34103072e-01 -8.17053258e-01 -6.50773942e-01 2.21747950e-01 1.44960093e+00 -4.99224901e-01 -2.56481379e-01 -4.75475304e-02 -6.73397422e-01 5.64311922e-01 5.65395296e-01 1.38218030e-01 -7.68806279e-01 -9.03524458e-01 -7.65823722e-02 4.17159609e-02 -2.20610514e-01 -3.72174948e-01 -2.55967110e-01 -8.43881130e-01 -1.10084391e+00 4.68927175e-01 1.65395722e-01 6.35471284e-01 1.26622438e-01 1.36669421e+00 5.30549645e-01 -5.07020414e-01 5.59902966e-01 -2.75017619e-01 -9.57829595e-01 -6.37173951e-01 -4.20699492e-02 -1.15846343e-01 -6.41751826e-01 8.78493130e-01 -7.68772900e-01 -4.51684207e-01 4.06075239e-01 -6.59928441e-01 7.74439275e-01 1.78480506e-01 2.57793993e-01 9.87539366e-02 -2.69145161e-01 4.51075584e-01 -1.29549146e+00 1.09356916e+00 -7.69904375e-01 -6.42769158e-01 3.87979299e-01 -1.40804374e+00 -2.11356103e-01 3.41090262e-01 -3.07851255e-01 -8.31693769e-01 -1.57733425e-01 1.58675089e-01 -1.93416372e-01 -1.70515001e-01 7.50661612e-01 -1.77768424e-01 3.60098928e-02 1.13240063e+00 -2.65336871e-01 -7.40084564e-04 -6.81339860e-01 5.95798552e-01 9.96175289e-01 1.42248899e-01 -5.14889300e-01 6.52887464e-01 -1.30766004e-01 -5.53117692e-01 -4.21296149e-01 -3.15587819e-01 3.27397399e-02 -5.32608390e-01 -4.00983244e-01 2.14460284e-01 -7.00368643e-01 -9.99852657e-01 1.56814530e-01 -7.40272224e-01 -9.09512520e-01 -1.70469597e-01 1.63993379e-03 -3.67769599e-01 -5.82482219e-02 -1.03111692e-01 -1.09424281e+00 -3.89088213e-01 -6.56627297e-01 3.83174628e-01 3.93156350e-01 -1.19786584e+00 -1.16662598e+00 -8.61339867e-02 6.16674840e-01 8.05894732e-01 3.54033589e-01 9.61500108e-01 -7.26026773e-01 -6.66157722e-01 -2.22416028e-01 -1.17429420e-01 9.89856720e-02 1.49021044e-01 6.19911969e-01 -1.04092324e+00 6.23785853e-02 -5.33847451e-01 -5.98265836e-03 6.25292026e-03 1.27305493e-01 1.45840311e+00 -9.95820880e-01 -2.28838116e-01 4.29406285e-01 9.07097220e-01 3.67829144e-01 6.36272967e-01 3.73417646e-01 7.17557847e-01 4.45612431e-01 6.03132606e-01 8.67570579e-01 5.91472685e-01 4.34866518e-01 3.00997555e-01 -3.82221669e-01 3.31944048e-01 -7.20564723e-01 2.54178256e-01 2.54267991e-01 -2.61457354e-01 -5.97050153e-02 -1.43568587e+00 1.54208764e-01 -2.17506433e+00 -8.33073258e-01 -3.24183166e-01 2.25986028e+00 8.49073529e-01 1.18793100e-01 2.22826228e-01 -3.87057692e-01 1.34073526e-01 -1.50832087e-01 -8.23386014e-01 -8.30532491e-01 5.11960506e-01 1.51722625e-01 5.40902853e-01 6.06817901e-01 -8.19418967e-01 8.15778852e-01 6.56427717e+00 4.69800472e-01 -1.10792172e+00 2.97871947e-01 8.49809885e-01 -5.60056508e-01 -7.51378179e-01 4.17319715e-01 -3.15739036e-01 5.12870550e-01 1.16469145e+00 -7.79973388e-01 8.48344743e-01 1.37828016e+00 9.91109371e-01 -1.98911116e-01 -1.33709097e+00 8.00812662e-01 -3.40394348e-01 -1.51924455e+00 -2.32575968e-01 1.33239597e-01 3.13005775e-01 -1.42373731e-02 4.00256574e-01 2.42922977e-01 1.20496416e+00 -1.34726536e+00 8.35522413e-01 9.23503458e-01 5.55904388e-01 -6.71998560e-01 1.17582545e-01 5.21923840e-01 -5.52143097e-01 -1.27051830e-01 -1.07588865e-01 -8.37065160e-01 -1.73919901e-01 4.09511834e-01 -1.46755552e+00 9.52311680e-02 7.05109835e-01 6.44841969e-01 -1.06516039e+00 7.93275893e-01 3.82749364e-02 1.09657788e+00 -3.26933265e-01 -6.21712059e-02 -4.17861104e-01 -9.10718516e-02 3.21769029e-01 1.23410296e+00 2.01271102e-02 1.48131568e-02 4.62819189e-02 1.37704921e+00 1.96701556e-01 -7.17633069e-02 -4.44154203e-01 -4.26366717e-01 6.97908700e-01 1.30134439e+00 -4.11650240e-01 -1.34334683e-01 -8.77138525e-02 3.28236490e-01 1.82537138e-01 2.91697651e-01 -8.72013867e-01 -2.36898884e-02 1.07865179e+00 6.43477798e-01 -7.27777183e-01 -3.18771303e-01 -1.03316236e+00 -9.12458181e-01 -9.33839567e-03 -1.23905468e+00 2.73030609e-01 -1.13939893e+00 -1.07600939e+00 1.19069099e-01 9.57194790e-02 -8.47402334e-01 -4.13082600e-01 -3.64788711e-01 -6.86227202e-01 9.39331770e-01 -5.73614240e-01 -1.27305722e+00 -5.54619789e-01 1.80981502e-01 1.65628448e-01 1.22329809e-01 8.14376831e-01 3.66467386e-02 -5.69999874e-01 7.39898324e-01 -4.30997759e-02 -3.21873516e-01 6.29071772e-01 -1.19709253e+00 1.03921008e+00 5.31439245e-01 4.43095230e-02 1.26311755e+00 8.75347495e-01 -8.74683380e-01 -1.04676127e+00 -8.64943564e-01 6.48558795e-01 -1.14192331e+00 7.71505117e-01 -5.35974324e-01 -8.81307006e-01 1.17108047e+00 1.72554046e-01 -5.76832712e-01 1.21278095e+00 6.48206830e-01 -3.74197394e-01 1.01575717e-01 -1.10692859e+00 8.87388527e-01 1.07426989e+00 -5.31917512e-01 -1.41831115e-01 4.30176705e-01 4.43990737e-01 -3.25464725e-01 -7.76470721e-01 2.42725257e-02 7.77238250e-01 -1.01057994e+00 6.97555780e-01 -1.20091343e+00 3.67284477e-01 -2.88509905e-01 2.30198666e-01 -1.35229957e+00 -1.36438906e-01 -8.84288609e-01 -1.09081149e-01 1.18632960e+00 8.23400617e-01 -9.65715647e-01 5.91811478e-01 2.00466800e+00 5.49846031e-02 -5.71720541e-01 -2.53247231e-01 -3.14308733e-01 -1.38152570e-01 -9.19937313e-01 9.75711644e-01 1.41228414e+00 3.84428769e-01 -2.66399711e-01 -3.96867573e-01 1.75582036e-01 4.45137560e-01 -3.10606420e-01 1.20835006e+00 -1.23093939e+00 -7.05398381e-01 -3.57120693e-01 -1.55919597e-01 -5.97288907e-01 -3.18396807e-01 -7.94275343e-01 -6.18373811e-01 -1.36626542e+00 2.73331672e-01 -8.91089141e-01 -9.09176841e-02 1.44512022e+00 -1.90119192e-01 -1.04245499e-01 6.06635690e-01 2.45141625e-01 -6.20592713e-01 -1.56651452e-01 4.60153311e-01 1.48664936e-01 -4.88129377e-01 1.37632936e-01 -1.40017056e+00 1.13565302e+00 1.14645123e+00 -6.07197702e-01 -5.77610493e-01 -5.63490152e-01 6.47820532e-01 -4.57370788e-01 8.09123039e-01 -7.77627647e-01 1.84107155e-01 -6.64723039e-01 3.43136162e-01 -1.20313734e-01 -2.30136011e-02 -6.56606615e-01 9.79903460e-01 2.72286057e-01 -4.89636987e-01 2.64264971e-01 5.11781633e-01 -7.54596060e-03 6.42140865e-01 -2.20130160e-01 3.53135943e-01 -5.57065047e-02 -3.33445847e-01 -6.72332495e-02 -4.86558706e-01 -1.42523900e-01 1.12326884e+00 -1.02249481e-01 -7.97846794e-01 -7.28415906e-01 -6.92539036e-01 2.34968632e-01 8.84994864e-01 6.95120692e-01 2.87775606e-01 -6.88261092e-01 -3.37343812e-01 1.29889354e-01 -1.16610169e-01 -3.38742435e-01 4.10043687e-01 6.44623756e-01 -8.08247566e-01 5.71223423e-02 -1.45904690e-01 -1.47998348e-01 -1.40729153e+00 2.33702883e-01 4.55878049e-01 7.12427199e-02 -3.52946609e-01 6.70590580e-01 -2.86652118e-01 -5.80329657e-01 -6.62341416e-02 -2.61039555e-01 1.89029589e-01 -3.48972678e-01 8.34350407e-01 6.72111928e-01 -7.05884025e-02 1.55195132e-01 -2.07501501e-01 -2.08093196e-01 -1.43925041e-01 8.10715109e-02 1.43942070e+00 -5.41812442e-02 -1.61853675e-02 6.05675161e-01 4.24326509e-01 1.14718862e-01 -1.49996543e+00 3.84581089e-01 1.66714355e-01 -8.11152279e-01 1.53898504e-02 -1.45301056e+00 -5.87275684e-01 6.92725241e-01 5.75672209e-01 3.07438344e-01 8.90936136e-01 -1.97031066e-01 1.78799406e-02 2.65355021e-01 3.93377483e-01 -9.03852224e-01 -3.49896103e-01 -1.56056270e-01 8.24271798e-01 -1.09697795e+00 1.60023540e-01 -3.21970582e-01 -7.69647658e-01 9.64283168e-01 9.05505002e-01 4.45369899e-01 6.76762760e-01 6.52907431e-01 3.40541244e-01 -1.08319283e-01 -1.24312556e+00 4.58763391e-01 -1.26947075e-01 8.85422826e-01 5.67669451e-01 4.89911646e-01 -1.92782551e-01 1.13293755e+00 -4.11395818e-01 5.50600410e-01 6.90677464e-01 7.95029104e-01 -1.71288028e-01 -1.10631740e+00 -3.82616341e-01 7.50204206e-01 -1.87178850e-01 -3.65798205e-01 -6.84404612e-01 7.51693547e-01 -2.14978145e-03 1.14447665e+00 1.25094265e-01 -7.38723040e-01 2.09663495e-01 1.43083587e-01 -1.00935735e-01 -5.29857457e-01 -7.72531450e-01 -4.56647396e-01 4.70616192e-01 -8.24961483e-01 3.49585980e-01 -7.74155378e-01 -1.02039599e+00 -1.00513935e+00 2.48313299e-03 -1.86702624e-01 7.44130731e-01 5.57707608e-01 9.87214744e-01 3.01123232e-01 3.32053393e-01 -3.88220221e-01 -4.44888711e-01 -8.86374712e-01 -1.82136729e-01 1.46143243e-01 -3.22833568e-01 -4.24570560e-01 -2.71187305e-01 2.98840463e-01]
[8.761786460876465, 5.924224853515625]
eeb91fdd-1f35-451a-ace8-8b6062343b85
joint-estimation-of-clustered-user-activity
2212.00116
null
https://arxiv.org/abs/2212.00116v1
https://arxiv.org/pdf/2212.00116v1.pdf
Joint Estimation of Clustered User Activity and Correlated Channels with Unknown Covariance in mMTC
This paper considers joint user identification and channel estimation (JUICE) in grant-free access with a \emph{clustered} user activity pattern. In particular, we address the JUICE in massive machine-type communications (mMTC) network under correlated Rayleigh fading channels with unknown channel covariance matrices. We formulate the JUICE problem as a maximum \emph{a posteriori} probability (MAP) problem with properly chosen priors to incorporate the partial knowledge of the UEs' clustered activity and the unknown covariance matrices. We derive a computationally-efficient algorithm based on alternating direction method of multipliers (ADMM) to solve the MAP problem iteratively via a sequence of closed-form updates. Numerical results highlight the significant improvements brought by the proposed approach in terms of channel estimation and activity detection performances for clustered user activity patterns.
['Markku Juntti', 'Markus Leinonen', 'Hamza Djelouat']
2022-11-30
null
null
null
null
['activity-detection']
['computer-vision']
[ 4.40860540e-01 2.64025956e-01 -1.47381693e-01 3.01348478e-01 -8.79295468e-01 -5.95094040e-02 1.26377150e-01 -3.42299908e-01 -4.63029504e-01 1.22814071e+00 -9.34309214e-02 -7.35153198e-01 -3.87089342e-01 -2.99334407e-01 -3.44569117e-01 -1.12116456e+00 -6.19815767e-01 2.08593115e-01 -5.48722446e-01 1.42569602e-01 1.32396534e-01 1.14455588e-01 -5.62467873e-01 -3.85856390e-01 8.36056590e-01 1.10842681e+00 3.37418169e-01 9.93606448e-01 4.53368753e-01 7.70601213e-01 -6.88682318e-01 -9.14313570e-02 2.70870417e-01 -6.14760220e-01 -3.92972320e-01 8.72103631e-01 -6.46454096e-01 -3.13934714e-01 -5.40474951e-01 8.08799028e-01 7.97514021e-01 1.19096674e-02 7.85451651e-01 -1.35465837e+00 1.76617011e-01 3.86839092e-01 -9.84638691e-01 2.74389058e-01 1.97348207e-01 -3.99410874e-01 7.27317691e-01 -1.07377577e+00 8.62627178e-02 8.17070723e-01 7.50382602e-01 3.86906452e-02 -8.16684008e-01 -8.10322404e-01 -8.02078843e-02 2.46084318e-01 -2.11440396e+00 -4.24039066e-01 1.55393779e-01 -1.97039768e-01 4.58910763e-01 4.33840096e-01 4.37281668e-01 5.46810865e-01 -2.63949752e-01 7.99558342e-01 6.70698464e-01 -6.56446278e-01 5.12957335e-01 1.65756732e-01 -1.83716595e-01 6.78395391e-01 6.23647809e-01 -2.50323385e-01 -3.51115584e-01 -6.23236775e-01 1.05061746e+00 -4.13860261e-01 -5.21062791e-01 -2.11002156e-01 -1.26080787e+00 6.61207259e-01 -5.73067248e-01 1.68217525e-01 -9.11516488e-01 3.17199707e-01 -6.17097855e-01 4.42087919e-01 2.55818814e-01 -1.14174813e-01 -4.17924404e-01 -1.54875681e-01 -8.50766063e-01 -1.00192495e-01 1.13065577e+00 1.55620182e+00 6.55426681e-01 3.16283017e-01 -2.02093899e-01 6.98638558e-01 7.35114396e-01 8.93440843e-01 -3.15218031e-01 -8.96689177e-01 7.55668819e-01 -3.75528306e-01 6.58024848e-01 -7.69187033e-01 -4.05347705e-01 -1.50715160e+00 -1.17067885e+00 -4.15746450e-01 5.93367755e-01 -1.39215374e+00 -3.82466838e-02 1.65650761e+00 1.90448061e-01 9.02780235e-01 2.34685555e-01 5.41542947e-01 -1.70686439e-01 6.52649283e-01 -3.54536563e-01 -1.10673988e+00 9.83660638e-01 -4.14485276e-01 -1.11879456e+00 -7.15270936e-02 8.74908566e-01 -6.59302711e-01 -2.26718545e-01 5.95966518e-01 -1.12611771e+00 9.09843296e-02 -1.26674879e+00 1.13224459e+00 4.21964347e-01 3.91817272e-01 4.46647942e-01 1.48595059e+00 -9.03871953e-01 1.50877452e-02 -4.37204897e-01 -1.05412908e-01 4.16557640e-01 5.91255069e-01 1.71853781e-01 3.43722776e-02 -9.79180276e-01 2.42958307e-01 1.81927785e-01 3.90242845e-01 -5.18437922e-01 -5.69659054e-01 -5.01443088e-01 1.29670069e-01 3.98774832e-01 -3.99103582e-01 1.21478367e+00 -5.32367229e-01 -1.51021361e+00 1.20672777e-01 -3.55567008e-01 -3.41839015e-01 5.65056860e-01 -7.00804144e-02 -6.47259295e-01 3.99124920e-02 -2.43583530e-01 -4.19920325e-01 8.34243119e-01 -1.03700328e+00 -1.02270448e+00 -1.69206187e-01 -2.72706121e-01 2.79071599e-01 -4.44341153e-01 -1.64750353e-01 -7.81510353e-01 -8.14878345e-01 2.43037969e-01 -1.09826434e+00 -6.08659685e-01 -2.64267981e-01 -5.12466371e-01 5.14689326e-01 5.31764388e-01 -1.07734203e+00 1.88403976e+00 -2.05494523e+00 1.58504426e-01 9.96748805e-01 3.87811773e-02 -1.10358424e-01 3.13991725e-01 5.78483999e-01 5.03245294e-01 -3.34383845e-01 -3.02176267e-01 -3.13843846e-01 -1.40115529e-01 -7.02951988e-03 3.91223103e-01 8.30482841e-01 -4.52464551e-01 1.60122871e-01 -8.97737503e-01 -2.46236399e-01 -1.35762123e-02 3.23025256e-01 -5.99859059e-01 2.55494982e-01 3.09976071e-01 7.76086569e-01 -5.65936387e-01 4.66163337e-01 1.12290549e+00 -5.22567272e-01 9.66644108e-01 1.33681476e-01 5.30510955e-02 -5.98030865e-01 -1.90488803e+00 1.21331561e+00 -7.09641874e-01 5.41917026e-01 8.80718946e-01 -1.06009638e+00 2.30470628e-01 1.08561456e+00 7.43359983e-01 -3.70494694e-01 2.99739540e-01 4.03377593e-01 9.14093480e-02 -2.82950640e-01 1.96706116e-01 3.62449628e-03 -1.75447240e-01 5.89261353e-01 1.43910229e-01 7.55742848e-01 -2.09310930e-02 3.58498722e-01 1.14415300e+00 -5.79048216e-01 8.62033844e-01 -4.23506528e-01 7.29285002e-01 -7.53977835e-01 6.63303673e-01 1.17253160e+00 -4.50861938e-02 7.32211396e-02 3.07930797e-01 5.89685917e-01 -6.18661463e-01 -9.60803449e-01 -2.26250201e-01 6.03036344e-01 9.15018022e-02 -7.64885694e-02 -6.82287753e-01 -2.13929832e-01 -1.94434568e-01 4.44137871e-01 -2.91112781e-01 3.72737199e-01 -2.53065288e-01 -1.56882179e+00 4.58586156e-01 1.11162663e-01 7.71205008e-01 5.77081293e-02 -2.72567254e-02 7.09149837e-01 -3.80948812e-01 -1.34657562e+00 -3.51573199e-01 2.60751247e-01 -5.57463229e-01 -8.58176351e-01 -1.16027451e+00 -7.16192305e-01 8.44841838e-01 5.29182017e-01 5.08848488e-01 -3.38624343e-02 8.12527537e-02 7.49835730e-01 -1.72719449e-01 -1.94867998e-01 -1.67405009e-01 -5.29175773e-02 1.46650568e-01 8.26247811e-01 1.68574434e-02 -6.46509349e-01 -5.39415598e-01 6.67052269e-01 -2.65031576e-01 -1.67667344e-01 8.85089338e-01 6.51624441e-01 5.38296327e-02 4.94883746e-01 7.22603977e-01 -8.04189622e-01 6.61822498e-01 -1.10220397e+00 -5.68860114e-01 2.97940373e-01 -5.54824114e-01 -2.73824930e-01 3.68644185e-02 -1.05083019e-01 -1.46988821e+00 3.63216400e-02 -1.33504167e-01 2.75863111e-01 1.89061716e-01 4.98033434e-01 -5.47149539e-01 -4.42726076e-01 2.53027976e-01 3.55509341e-01 -2.84429044e-01 -3.38645220e-01 3.41655433e-01 1.23175132e+00 3.44643205e-01 -5.53731740e-01 1.02105367e+00 4.73484546e-01 1.90199479e-01 -1.39828491e+00 -4.97719347e-01 -9.18798745e-01 -6.21192932e-01 -3.26734871e-01 2.52044499e-01 -1.44622076e+00 -7.49772847e-01 6.41636133e-01 -7.21459627e-01 -2.88283885e-01 5.09656906e-01 9.16153610e-01 -4.07933265e-01 8.97069752e-01 -5.61202526e-01 -1.39969635e+00 -5.07127158e-02 -4.20640588e-01 5.20844877e-01 6.12974539e-02 3.27774920e-02 -1.23855567e+00 -2.82023162e-01 1.81304067e-01 5.60682595e-01 2.68996675e-02 5.26201844e-01 7.27903992e-02 -6.31680787e-01 -2.88442284e-01 -1.48725465e-01 -4.11315449e-03 4.36589234e-02 -7.22039402e-01 -7.02287614e-01 -6.85239494e-01 -1.30393744e-01 6.61574662e-01 -1.72391664e-02 7.48710752e-01 9.28993285e-01 -5.93977273e-01 -6.37372434e-01 5.11431634e-01 1.39820075e+00 3.29621464e-01 7.12944269e-01 -1.01155035e-01 2.85839647e-01 -7.50705823e-02 7.92446613e-01 1.61099494e+00 1.89522848e-01 8.46021175e-01 1.94611356e-01 -1.77947953e-02 4.42339808e-01 3.20187122e-01 6.70252368e-02 8.64146948e-01 -1.96541667e-01 -7.84653068e-01 -4.10387635e-01 3.80826026e-01 -1.92957604e+00 -9.10075068e-01 -6.53685093e-01 2.32828856e+00 4.32089984e-01 -1.41714931e-01 7.44727477e-02 5.95865488e-01 9.51027632e-01 -3.45382214e-01 -1.59185708e-01 2.90326327e-01 -3.42066050e-01 6.36520833e-02 1.19938099e+00 6.90397561e-01 -9.87640023e-01 1.21802129e-01 5.63957024e+00 1.21113324e+00 -3.26447636e-01 4.85954285e-01 4.59794134e-01 1.34480104e-01 1.28937699e-02 -9.00274068e-02 -5.99164546e-01 5.17616332e-01 1.00362396e+00 -1.02178184e-02 2.34176636e-01 2.09843591e-01 6.31865025e-01 -4.63082701e-01 -4.71931159e-01 1.38486874e+00 -1.79711223e-01 -1.19037187e+00 -4.97108787e-01 7.45387256e-01 1.00393951e+00 -3.55453432e-01 -1.38872825e-02 1.87959418e-01 1.71075135e-01 -3.83082658e-01 4.53074515e-01 5.61827958e-01 7.32719243e-01 -6.46246016e-01 7.14800954e-01 3.33142042e-01 -1.33474243e+00 -5.68502843e-01 -1.89744157e-03 -3.71674836e-01 6.00807965e-01 1.00366271e+00 -8.16637337e-01 6.28793299e-01 -5.25650121e-02 2.92159706e-01 1.63524702e-01 1.66584265e+00 -1.72065899e-01 1.11034417e+00 -3.42267364e-01 -3.70561145e-02 1.54698119e-01 -4.36182141e-01 6.95992708e-01 1.29044640e+00 9.04623747e-01 4.32599396e-01 -5.24920179e-03 -3.56079303e-02 2.78951251e-04 3.21990669e-01 7.00671598e-02 4.66927528e-01 9.30060029e-01 1.07261515e+00 -4.87075478e-01 -2.15233803e-01 -6.99589431e-01 1.04173589e+00 -3.97565812e-01 8.99982274e-01 -6.35046899e-01 -4.60927427e-01 5.22138655e-01 8.43483433e-02 4.46348548e-01 -6.00144565e-01 -4.50266331e-01 -1.07439446e+00 -4.78665791e-02 -5.22132456e-01 3.52710128e-01 -1.70701459e-01 -5.12766302e-01 -2.24026516e-01 -3.37898493e-01 -1.55995095e+00 -2.45498374e-01 -2.14307055e-01 -4.66545790e-01 6.98125422e-01 -1.39240658e+00 -5.85646987e-01 -1.89851359e-01 8.21178198e-01 1.00605525e-01 -3.12419623e-01 8.18046331e-01 1.00784433e+00 -9.27594483e-01 9.48891580e-01 9.50681090e-01 1.38226867e-01 -1.69628069e-01 -1.14819276e+00 -4.38704759e-01 1.16389287e+00 -2.68131763e-01 2.81994462e-01 8.45472217e-01 -3.04472089e-01 -1.40977919e+00 -1.01973653e+00 7.39870489e-01 2.07360566e-01 4.93449569e-01 -3.07683498e-01 -2.45662734e-01 5.38564444e-01 1.70505941e-01 -2.77931273e-01 1.26228178e+00 -5.21195903e-02 3.93536359e-01 3.19357246e-01 -1.04612386e+00 4.51748937e-01 9.14047718e-01 -2.00653642e-01 5.00355959e-01 4.14496601e-01 -2.37060547e-01 -1.85120001e-01 -1.09671748e+00 1.07118294e-01 5.70727587e-01 -2.80410051e-01 8.12120855e-01 2.91656610e-02 -9.03740942e-01 -3.28106701e-01 -5.01977026e-01 -1.02385199e+00 -5.82927108e-01 -1.55357289e+00 -7.32558608e-01 8.67282212e-01 8.11218262e-01 -3.09146881e-01 1.03027844e+00 2.28731409e-01 2.92912245e-01 -3.84435803e-01 -1.11442995e+00 -6.62459016e-01 -3.95130694e-01 -7.02981234e-01 1.07880540e-01 7.42830098e-01 5.63693881e-01 3.51602793e-01 -1.18928039e+00 7.99413025e-01 1.12387812e+00 -5.82369804e-01 8.91239882e-01 -1.22416103e+00 -8.44746828e-01 9.20175388e-02 -2.13101104e-01 -1.64735711e+00 -2.60580599e-01 -3.90492529e-01 -8.39099661e-02 -1.14748847e+00 9.25053433e-02 -7.50456810e-01 -3.41053635e-01 -8.09713900e-02 -2.11876974e-01 1.28698304e-01 -1.71597913e-01 2.57024579e-02 -9.25503731e-01 4.28594142e-01 6.89748347e-01 2.82863319e-01 -1.01374283e-01 1.00448453e+00 -5.96276343e-01 3.36610645e-01 7.74919868e-01 -3.00861329e-01 -3.74604791e-01 -1.22650348e-01 3.03732604e-01 7.95991480e-01 1.25750713e-02 -1.25287378e+00 1.06462158e-01 4.40807492e-02 2.09593624e-01 -5.62878191e-01 5.47991216e-01 -9.14172411e-01 3.28740954e-01 7.16108143e-01 1.69310853e-01 -6.49833083e-01 -1.56628519e-01 1.16319382e+00 1.74751520e-01 -1.07388049e-01 6.66454911e-01 3.87412220e-01 -6.79488480e-01 2.83758014e-01 -1.46896303e+00 -2.27775916e-01 1.11137497e+00 -3.03780079e-01 3.29105258e-01 -1.17985249e+00 -1.11023545e+00 5.17178833e-01 -3.96233946e-01 -2.78585672e-01 3.45149666e-01 -1.09598505e+00 -9.17920411e-01 -2.44800691e-02 -1.25522852e-01 -7.63615549e-01 5.05705953e-01 1.38006771e+00 -6.97890520e-02 7.56383717e-01 2.59039044e-01 -3.79814088e-01 -1.13624012e+00 -1.49699017e-01 4.52711970e-01 -1.76841944e-01 5.57628386e-02 8.91574621e-01 -3.06466848e-01 -6.64678514e-02 3.99493605e-01 4.08131808e-01 -1.44148827e-01 -1.47857577e-01 3.70276868e-01 9.76199389e-01 1.31720081e-01 -6.64656341e-01 -2.35452391e-02 1.93497345e-01 2.38845542e-01 -4.64478970e-01 8.64082217e-01 -9.21236634e-01 3.00012022e-01 -2.25522414e-01 1.13680065e+00 3.41705501e-01 -1.32566810e+00 -8.53525579e-01 3.61670256e-02 -6.14485145e-01 2.47201785e-01 -4.80762452e-01 -1.07247996e+00 3.33343327e-01 7.28274584e-01 -1.56978264e-01 7.95132399e-01 -4.30964828e-01 5.54143846e-01 4.71346408e-01 9.17355418e-01 -1.25672197e+00 -2.05550760e-01 3.26714605e-01 1.33217558e-01 -6.91130698e-01 -4.52725403e-02 -7.23755836e-01 -1.00762442e-01 9.00951564e-01 2.06819866e-04 4.68989640e-01 1.36737800e+00 2.17404395e-01 -2.17117339e-01 1.97392255e-01 -4.72205788e-01 -4.58489895e-01 -1.90979466e-01 8.01830113e-01 3.16605896e-01 4.37907666e-01 -7.01657951e-01 6.55435026e-01 2.18958482e-01 -1.43401086e-01 1.02055979e+00 1.03834224e+00 -7.10518479e-01 -1.02908671e+00 -7.44874537e-01 7.71343887e-01 -7.20207512e-01 -1.24177992e-01 2.55290627e-01 3.63847703e-01 -1.67418808e-01 1.61410022e+00 -3.27306837e-01 -1.78965881e-01 -4.83400971e-02 -4.04808760e-01 3.75051439e-01 -3.66729826e-01 2.22489506e-01 4.78065193e-01 4.96761203e-01 -6.66067302e-02 -1.49909109e-01 -9.20129955e-01 -1.07895839e+00 -2.41630465e-01 -6.00594580e-01 6.79365456e-01 5.65639853e-01 1.26095366e+00 1.99020743e-01 6.18414819e-01 1.07470953e+00 -4.80846286e-01 -4.64565158e-01 -8.63623261e-01 -1.28202701e+00 -2.19945148e-01 4.70717959e-02 -4.59251791e-01 -4.33141738e-01 -2.43891165e-01]
[6.205172061920166, 1.4123210906982422]
97ec728f-d4fa-4a5b-8318-8a063b3cd01b
automated-essay-scoring-for-swedish
null
null
https://aclanthology.org/W13-1705
https://aclanthology.org/W13-1705.pdf
Automated Essay Scoring for Swedish
null
['Erik H{\\"o}glin', 'Bj{\\"o}rn Tyrefors Hinnerich', 'Robert {\\"O}stling', 'Andre Smolentzov']
2013-06-01
null
null
null
ws-2013-6
['automated-essay-scoring']
['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]
[-7.332584381103516, 3.5991404056549072]
14d96d7f-d8c7-46bf-96c2-ba6d3f3dd3f4
comparing-machines-and-children-using
2305.11243
null
https://arxiv.org/abs/2305.11243v1
https://arxiv.org/pdf/2305.11243v1.pdf
Comparing Machines and Children: Using Developmental Psychology Experiments to Assess the Strengths and Weaknesses of LaMDA Responses
Developmental psychologists have spent decades devising experiments to test the intelligence and knowledge of infants and children, tracing the origin of crucial concepts and capacities. Moreover, experimental techniques in developmental psychology have been carefully designed to discriminate the cognitive capacities that underlie particular behaviors. We propose that using classical experiments from child development is a particularly effective way to probe the computational abilities of AI models, in general, and LLMs in particular. First, the methodological techniques of developmental psychology, such as the use of novel stimuli to control for past experience or control conditions to determine whether children are using simple associations, can be equally helpful for assessing the capacities of LLMs. In parallel, testing LLMs in this way can tell us whether the information that is encoded in text is sufficient to enable particular responses, or whether those responses depend on other kinds of information, such as information from exploration of the physical world. In this work we adapt classical developmental experiments to evaluate the capabilities of LaMDA, a large language model from Google. We propose a novel LLM Response Score (LRS) metric which can be used to evaluate other language models, such as GPT. We find that LaMDA generates appropriate responses that are similar to those of children in experiments involving social understanding, perhaps providing evidence that knowledge of these domains is discovered through language. On the other hand, LaMDA's responses in early object and action understanding, theory of mind, and especially causal reasoning tasks are very different from those of young children, perhaps showing that these domains require more real-world, self-initiated exploration and cannot simply be learned from patterns in language input.
['Danielle Krettek Cobb', 'Alison Gopnik', 'Leslie Lai', 'Emily Rose Reagan', 'Eliza Kosoy']
2023-05-18
null
null
null
null
['action-understanding']
['computer-vision']
[-6.14565797e-02 1.40067816e-01 1.12968564e-01 -2.67131627e-01 3.49975824e-01 -5.38712382e-01 7.35746026e-01 5.34793854e-01 -5.65086663e-01 2.83626795e-01 5.30322194e-02 -4.18246925e-01 -3.30066383e-01 -1.12321723e+00 -8.44652236e-01 -3.92952055e-01 -1.95161909e-01 5.17182708e-01 4.11481738e-01 -3.40182483e-01 5.83451331e-01 5.56502938e-01 -1.83029795e+00 1.38326183e-01 1.03764939e+00 2.05314726e-01 5.77449322e-01 5.13333976e-01 -2.96063691e-01 9.04647171e-01 -2.46089563e-01 -3.30802381e-01 -1.39935106e-01 -6.58797979e-01 -8.55937123e-01 -4.93639737e-01 2.67315924e-01 -6.59984350e-01 -5.70338313e-03 9.89886522e-01 1.90513417e-01 1.18610002e-01 6.38248920e-01 -7.47424126e-01 -8.92105758e-01 1.06337762e+00 -1.85103849e-01 3.75785440e-01 8.85630429e-01 3.12891185e-01 5.80967665e-01 -5.96502483e-01 3.40455949e-01 1.58312106e+00 2.14090049e-01 8.90235305e-01 -1.43554950e+00 -7.03374684e-01 2.34671980e-01 1.43075436e-01 -9.45402145e-01 -3.85836005e-01 4.81883079e-01 -7.24846244e-01 8.23095083e-01 -6.31723776e-02 1.09483182e+00 1.09820688e+00 2.83970078e-03 5.09562016e-01 1.56630278e+00 -6.67241633e-01 3.17949891e-01 3.88554394e-01 5.47869802e-02 5.92956245e-01 4.44762856e-01 7.00330436e-01 -1.03234732e+00 1.48345962e-01 8.99457753e-01 -4.63449806e-01 -1.10166028e-01 -1.88692093e-01 -1.08281159e+00 7.05075681e-01 1.04874857e-01 8.93440187e-01 -2.74000764e-01 -5.65422093e-03 5.63180111e-02 5.63148916e-01 3.50617990e-02 1.09742439e+00 -5.16694546e-01 -2.82225847e-01 -6.78136706e-01 2.03807354e-01 6.27793789e-01 6.66231453e-01 5.64905107e-01 -3.07214390e-02 1.47782743e-01 8.53831053e-01 4.95231718e-01 4.26907748e-01 8.26589048e-01 -9.15262520e-01 7.44467601e-02 7.11414993e-01 -2.92480171e-01 -9.54526603e-01 -4.65919971e-01 -8.33847933e-03 2.35979874e-02 5.19279361e-01 6.36064172e-01 -3.43569815e-02 -5.66890776e-01 2.32274437e+00 4.30787414e-01 -1.62755340e-01 8.04480538e-02 6.91181898e-01 6.97581112e-01 4.51122314e-01 5.18228650e-01 -2.90508717e-01 8.48507702e-01 -1.65792540e-01 -3.99225652e-02 -4.21516895e-01 9.35362339e-01 -2.72795886e-01 1.28314018e+00 5.58831334e-01 -1.46924531e+00 -5.88250160e-01 -1.06916416e+00 7.00222850e-02 -5.66532552e-01 -6.87691748e-01 1.07397318e+00 9.46321070e-01 -1.35084283e+00 1.02618051e+00 -7.96315074e-01 -7.29955256e-01 2.33870909e-01 2.55241275e-01 -3.87968987e-01 -7.54854754e-02 -1.20738339e+00 1.23450768e+00 7.34615803e-01 -3.58848512e-01 -1.13975763e+00 -4.96288717e-01 -6.77714407e-01 1.55334845e-01 2.01173156e-01 -2.52867669e-01 1.13928032e+00 -1.04201925e+00 -1.40530670e+00 1.16095686e+00 1.87269822e-01 -5.25024720e-02 -4.90413159e-02 -2.69629732e-02 -2.89767414e-01 2.92910457e-01 1.23482697e-01 6.85427666e-01 4.87420410e-01 -8.02829802e-01 -2.50933439e-01 -5.81690967e-01 4.45515215e-01 1.56909838e-01 -3.55797470e-01 4.87887830e-01 2.29779452e-01 -5.27077436e-01 3.48821640e-01 -7.42145360e-01 3.09296716e-02 1.03172027e-01 3.11444074e-01 -4.46556717e-01 -1.62140682e-01 -4.54362005e-01 8.87642980e-01 -2.10018682e+00 2.12056801e-01 2.52239227e-01 2.61180282e-01 1.76007256e-01 -2.98399895e-01 6.25811994e-01 -2.08566606e-01 4.41080838e-01 1.80641949e-01 4.42557305e-01 3.04630082e-02 1.24011971e-01 -2.40463894e-02 1.66083768e-01 2.26573914e-01 9.99715984e-01 -1.00166476e+00 -3.84029478e-01 -2.44032443e-02 6.76928908e-02 -6.09831154e-01 2.64510155e-01 -4.16109204e-01 6.33960009e-01 -3.55047345e-01 1.08717851e-01 2.83646047e-01 -6.75174966e-02 2.41573170e-01 7.42909849e-01 -3.17576706e-01 8.37198555e-01 -7.44410157e-01 1.41460192e+00 -4.51689780e-01 7.77967751e-01 -4.73522156e-01 -9.26233828e-01 6.34595096e-01 4.43876326e-01 -2.61241674e-01 -1.25910974e+00 1.26938030e-01 1.56854615e-01 9.19606268e-01 -7.54769921e-01 -1.59626976e-01 -2.50384986e-01 2.76194006e-01 8.25032651e-01 1.79516390e-01 -4.54450130e-01 4.08041149e-01 2.20786244e-01 9.38660443e-01 1.65114209e-01 2.08632812e-01 -6.18085802e-01 4.05758202e-01 -2.92516947e-01 3.82928938e-01 8.57011497e-01 3.21290761e-01 -6.47733435e-02 6.18714869e-01 -2.49944508e-01 -7.82474101e-01 -1.12941968e+00 -2.46095926e-01 1.45588505e+00 -1.61958486e-01 -3.69832627e-02 -7.75251567e-01 -2.08343953e-01 -3.19818258e-01 1.41195560e+00 -7.36552060e-01 -4.88462836e-01 -5.74187040e-01 -4.42302823e-01 2.34089673e-01 5.93315423e-01 1.95645094e-01 -1.54023027e+00 -1.10334527e+00 2.91522413e-01 5.81629574e-01 -6.91158175e-01 2.80752182e-01 8.11330276e-04 -1.08026600e+00 -9.25242364e-01 -3.31525207e-01 -4.95048016e-01 7.01336324e-01 1.52024012e-02 1.03682804e+00 7.32742012e-01 2.44510034e-03 7.60798991e-01 -5.72974205e-01 -6.37966752e-01 -8.19142401e-01 -2.87561506e-01 2.58544356e-01 -7.96395600e-01 5.66987813e-01 -1.11675537e+00 -2.97447711e-01 1.65620312e-01 -8.86806130e-01 2.40875602e-01 5.82024157e-01 3.70400101e-01 -2.79419780e-01 -1.23849064e-01 5.20307839e-01 -7.10760593e-01 5.97791135e-01 -6.45005763e-01 -6.58567607e-01 3.46841097e-01 -7.41436481e-01 3.26038837e-01 3.45681190e-01 -9.30120707e-01 -1.04614818e+00 -5.47340512e-01 -1.36655662e-02 3.00489783e-01 -5.29625773e-01 7.39474297e-01 -3.42073627e-02 -1.45885393e-01 8.23559701e-01 3.81511271e-01 -1.24469973e-01 -5.28133154e-01 1.45762578e-01 2.40538642e-02 1.62928924e-01 -1.47392488e+00 7.61420190e-01 -1.99431285e-01 -1.90735027e-01 -9.25957620e-01 -6.20064020e-01 2.40838259e-01 -5.43765068e-01 -2.77562767e-01 8.10892642e-01 -6.54509902e-01 -8.11799049e-01 4.65901494e-01 -9.69830036e-01 -5.90780437e-01 -1.67662755e-01 8.00298274e-01 -5.69976211e-01 4.35037538e-02 -5.25765955e-01 -7.62580276e-01 1.21274933e-01 -9.10239577e-01 2.63361156e-01 4.79503751e-01 -5.38901925e-01 -9.89902079e-01 2.37040907e-01 -5.71960546e-02 4.03435320e-01 -6.49246499e-02 1.46289086e+00 -1.10753858e+00 -4.71473604e-01 2.53841996e-01 1.57042846e-01 6.76047802e-02 -2.50482202e-01 1.75311744e-01 -8.07084620e-01 -5.20527773e-02 3.25935274e-01 -6.98315024e-01 4.95764136e-01 -1.30074158e-01 7.96636283e-01 -1.52761564e-01 -6.71076104e-02 3.33558083e-01 1.22774684e+00 6.76856160e-01 5.87343752e-01 1.51954100e-01 2.17379510e-01 1.24190962e+00 4.12312567e-01 -1.36076033e-01 4.85798329e-01 4.67019171e-01 -1.47206485e-01 5.26211262e-01 1.06170416e-01 -5.95944166e-01 4.35713112e-01 7.94411540e-01 -1.71692148e-01 -1.46050025e-02 -1.28424418e+00 7.05395699e-01 -1.37492394e+00 -8.80324304e-01 -7.04297274e-02 2.40503478e+00 1.35458982e+00 3.37421596e-01 6.91999542e-03 5.17498516e-02 3.34431767e-01 -2.37120628e-01 -4.80138838e-01 -8.29496503e-01 1.98892038e-03 7.18743265e-01 -1.82702959e-01 2.65049845e-01 -1.31399706e-01 1.03459394e+00 6.78952646e+00 3.79609436e-01 -9.93947208e-01 -5.84157258e-02 2.88950711e-01 1.10084914e-01 -5.94684184e-01 -1.92877296e-02 -3.57592940e-01 4.17828768e-01 1.22487390e+00 -3.38923931e-01 6.31009519e-01 6.08894289e-01 -8.51634070e-02 -6.62492514e-01 -1.69422376e+00 3.45302492e-01 -1.65250525e-01 -6.84127212e-01 -4.71067950e-02 -1.01321988e-01 3.98943067e-01 -3.32833081e-01 6.89218193e-02 3.36677819e-01 4.33526307e-01 -1.13927519e+00 9.49355960e-01 3.75868648e-01 5.40670335e-01 -1.66680977e-01 1.57923460e-01 7.92202950e-01 -4.62597668e-01 9.22604203e-02 -4.18301135e-01 -7.99917102e-01 -4.60866421e-01 -2.09115781e-02 -7.25562394e-01 -4.67682749e-01 6.47236526e-01 2.56463647e-01 -7.74132848e-01 7.91882217e-01 -7.59618223e-01 8.20313096e-01 -3.51052076e-01 -4.93381947e-01 -1.90375507e-01 8.39103237e-02 3.55677247e-01 6.60306156e-01 1.42493248e-01 5.71377993e-01 -5.56155562e-01 1.24546623e+00 4.53315109e-01 4.74764526e-01 -7.66765714e-01 -2.89428055e-01 6.06419742e-01 6.87532008e-01 -7.95288086e-01 -1.57812208e-01 -6.12910748e-01 3.57380182e-01 4.49323326e-01 6.84933662e-02 -3.61683428e-01 4.22580123e-01 4.68488038e-01 2.94984192e-01 1.06018066e-01 -3.82508129e-01 -1.45545542e-01 -1.05521977e+00 -3.65910083e-01 -8.70069444e-01 -1.43204242e-01 -8.51918757e-01 -9.39809859e-01 -3.89339365e-02 5.15055120e-01 -2.88798600e-01 -5.00953972e-01 -7.26070404e-01 -8.07739258e-01 7.69688249e-01 -8.66275012e-01 -6.28345966e-01 1.01387398e-02 5.17839909e-01 4.34648037e-01 1.28010988e-01 7.31512427e-01 -2.06468210e-01 -5.84119976e-01 4.59236652e-01 -4.80148137e-01 9.33290794e-02 2.76558667e-01 -1.24547565e+00 5.93807578e-01 8.67268741e-01 3.42462659e-01 1.05105567e+00 7.36331880e-01 -6.67719781e-01 -9.99265313e-01 -1.19469641e-02 7.20193744e-01 -4.88230109e-01 5.49931228e-01 -4.56969589e-01 -1.36234844e+00 4.96297032e-01 9.21825990e-02 -6.30927444e-01 6.43524647e-01 1.93131402e-01 -4.38559383e-01 2.22198814e-01 -9.60119963e-01 6.54120445e-01 1.47533953e+00 -4.12705332e-01 -1.20473957e+00 1.42669408e-02 8.12411308e-01 -7.21021593e-02 -6.93912745e-01 2.18863353e-01 6.72590375e-01 -1.26780248e+00 7.89421618e-01 -8.55665863e-01 7.36658335e-01 1.13713875e-01 6.61945865e-02 -1.28385973e+00 -4.76479203e-01 -1.09270595e-01 2.17321932e-01 1.44704461e+00 5.81259429e-01 -9.76069033e-01 2.38568276e-01 1.07219195e+00 1.65314570e-01 -5.33274293e-01 -6.16147935e-01 -4.44317281e-01 6.42816186e-01 -6.67854607e-01 7.12991297e-01 1.03100014e+00 1.46878883e-01 1.53656080e-01 4.43671376e-01 1.20219558e-01 4.27722096e-01 -1.88311338e-01 4.21626687e-01 -1.53746986e+00 -4.73302245e-01 -4.72980052e-01 -2.68114209e-01 -4.64187920e-01 2.38855273e-01 -6.67804778e-01 -9.32946876e-02 -1.10872185e+00 7.24151656e-02 -6.45367384e-01 -1.13838904e-01 3.28526378e-01 8.33114889e-03 -5.19385219e-01 4.07565862e-01 -1.50775537e-01 -7.76404664e-02 1.18214272e-01 9.64052200e-01 3.70772332e-01 -1.70766607e-01 -2.16114208e-01 -1.02406585e+00 1.31654406e+00 7.68483520e-01 -5.81921577e-01 -6.70250952e-01 -5.71449935e-01 9.64897096e-01 5.36755510e-02 4.51024443e-01 -1.12644243e+00 3.05807769e-01 -6.05461419e-01 5.25580049e-01 -3.37072345e-03 -1.74959451e-01 -6.35643899e-01 5.78969717e-03 5.42277753e-01 -4.82531160e-01 6.33372441e-02 4.17813331e-01 7.27037154e-03 2.80695945e-01 -6.71936154e-01 6.18379593e-01 -4.43846524e-01 -8.82809460e-01 -2.49999881e-01 -6.21904194e-01 3.34260345e-01 9.02807415e-01 -3.21183294e-01 -3.18372577e-01 -1.81492105e-01 -6.22774601e-01 1.69795021e-01 7.52046466e-01 5.33200443e-01 7.25763798e-01 -8.78390789e-01 -6.53820038e-01 2.70786464e-01 6.28382564e-02 -2.24815562e-01 4.40380275e-02 8.19842577e-01 -6.46608293e-01 1.97968259e-01 -3.35381061e-01 -2.00872168e-01 -8.96863580e-01 9.01474953e-01 1.93284765e-01 3.43018472e-01 -2.23732397e-01 1.22739291e+00 8.08293581e-01 3.58526111e-02 -1.81382269e-01 -3.81726831e-01 -5.38131058e-01 1.20776422e-01 8.17284226e-01 1.45899191e-01 -4.84276772e-01 -2.87138909e-01 -2.58952469e-01 7.31187999e-01 -1.22147955e-01 -2.60472000e-01 1.40266848e+00 3.62257175e-02 -4.23215359e-01 7.32338548e-01 5.84653616e-01 2.83866286e-01 -8.18176091e-01 -1.86761945e-01 8.16396773e-02 -5.07039070e-01 -3.77035350e-01 -9.73495841e-01 -5.84185421e-01 1.05534112e+00 2.85812706e-01 2.46383667e-01 1.20661855e+00 5.42511880e-01 -1.17583260e-01 2.59760469e-01 7.08837271e-01 -9.43744659e-01 8.46699327e-02 1.96941808e-01 1.06127751e+00 -9.12840068e-01 4.04477753e-02 -1.11314148e-01 -2.96316624e-01 1.01321435e+00 9.66963887e-01 -3.05647049e-02 3.46184582e-01 9.80811790e-02 -4.20200884e-01 -2.10300654e-01 -1.26049030e+00 -2.00546116e-01 2.26622298e-01 6.64720058e-01 6.32374287e-01 4.85637709e-02 -6.21844411e-01 4.68286127e-01 -5.81460416e-01 -1.89064249e-01 5.81048250e-01 7.32320368e-01 -5.90381622e-01 -1.09250963e+00 -3.60838890e-01 3.71828824e-01 -4.07167375e-01 -3.97489369e-01 -6.13197803e-01 9.94161189e-01 3.91354918e-01 7.85093725e-01 1.74799620e-03 -2.59554654e-01 1.17258109e-01 2.84675598e-01 1.00218153e+00 -1.21202564e+00 -4.89907503e-01 -4.57248002e-01 6.40404373e-02 -5.92522383e-01 -4.40332681e-01 -1.12375617e+00 -1.30452216e+00 -3.53323966e-01 -2.88378686e-01 -2.63609570e-02 4.88913387e-01 1.12419379e+00 -2.91612625e-01 1.33022685e-02 1.54140219e-01 -3.74719858e-01 -1.74755543e-01 -9.57152545e-01 -4.94282991e-01 2.08867326e-01 7.01103806e-02 -8.09295297e-01 -4.27538604e-01 -3.51716816e-01]
[10.241897583007812, 8.604592323303223]
b2914663-215f-4ec9-99ad-a433ceb84351
metric-oriented-speech-enhancement-using
2302.11989
null
https://arxiv.org/abs/2302.11989v1
https://arxiv.org/pdf/2302.11989v1.pdf
Metric-oriented Speech Enhancement using Diffusion Probabilistic Model
Deep neural network based speech enhancement technique focuses on learning a noisy-to-clean transformation supervised by paired training data. However, the task-specific evaluation metric (e.g., PESQ) is usually non-differentiable and can not be directly constructed in the training criteria. This mismatch between the training objective and evaluation metric likely results in sub-optimal performance. To alleviate it, we propose a metric-oriented speech enhancement method (MOSE), which leverages the recent advances in the diffusion probabilistic model and integrates a metric-oriented training strategy into its reverse process. Specifically, we design an actor-critic based framework that considers the evaluation metric as a posterior reward, thus guiding the reverse process to the metric-increasing direction. The experimental results demonstrate that MOSE obviously benefits from metric-oriented training and surpasses the generative baselines in terms of all evaluation metrics.
['Eng Siong Chng', 'Weiwei Weng', 'Yuchen Hu', 'Chen Chen']
2023-02-23
null
null
null
null
['speech-enhancement']
['speech']
[ 1.47410214e-01 2.00976692e-02 1.58160239e-01 -5.67587018e-01 -1.17118239e+00 -2.02901945e-01 6.93571687e-01 -3.06694210e-01 -4.87197310e-01 6.95080042e-01 5.87036133e-01 -3.64595056e-01 -2.55004764e-01 -5.62885642e-01 -4.99231070e-01 -8.79363418e-01 2.75023937e-01 3.95374522e-02 -5.52319130e-03 -1.80800021e-01 -4.89297658e-02 -1.65380146e-02 -9.61449206e-01 -2.37915516e-01 1.22614276e+00 9.25713956e-01 3.55260402e-01 5.22383332e-01 7.61317238e-02 6.97376728e-01 -7.82481432e-01 -5.27343154e-01 1.41950101e-01 -7.66558707e-01 -3.44653815e-01 1.90010236e-03 3.82220075e-02 -3.84511709e-01 -5.45725882e-01 1.30340731e+00 8.16810668e-01 3.09944183e-01 5.76208293e-01 -1.16442561e+00 -1.04583764e+00 7.59888291e-01 -5.05630791e-01 1.31159574e-01 -1.72150537e-01 1.40147507e-01 1.14640749e+00 -8.99257004e-01 1.68825716e-01 1.22393060e+00 5.54825664e-01 5.84808290e-01 -1.05810344e+00 -4.47987765e-01 2.71744430e-01 1.66393414e-01 -1.16242898e+00 -5.22284150e-01 1.25202179e+00 -1.75437063e-01 5.22456586e-01 1.07142448e-01 2.65004188e-01 1.18044984e+00 -1.32193342e-01 9.12009358e-01 1.26391530e+00 -2.94685960e-01 4.20335829e-01 1.28629237e-01 -7.05357045e-02 3.53255808e-01 -1.42361060e-01 6.93147838e-01 -5.04884601e-01 2.13210240e-01 6.07183218e-01 -3.51696104e-01 -3.51469457e-01 -1.57425165e-01 -9.87329185e-01 6.95528030e-01 4.10935640e-01 1.89372286e-01 -6.38413310e-01 1.70187935e-01 2.80073285e-01 2.73333430e-01 5.17893732e-01 4.76355910e-01 -3.65000516e-01 -4.50529337e-01 -1.11604154e+00 5.82979396e-02 3.72407168e-01 6.30881071e-01 5.97483754e-01 5.43918550e-01 -6.80335581e-01 1.04036319e+00 5.11857808e-01 3.84790838e-01 4.15047944e-01 -1.08809805e+00 5.79288840e-01 1.63544387e-01 1.95542425e-01 -6.51249349e-01 -2.43621599e-03 -8.63784611e-01 -7.83897996e-01 3.26345801e-01 1.89120725e-01 -5.57468295e-01 -8.60490620e-01 2.14036274e+00 2.80027032e-01 2.98403352e-02 5.11824265e-02 1.13002121e+00 4.79190230e-01 7.55184054e-01 2.43150070e-01 -2.44909033e-01 1.03646863e+00 -1.36169100e+00 -1.06542385e+00 -1.85266733e-01 1.98922440e-01 -6.81598127e-01 1.33317482e+00 3.38443249e-01 -1.14195406e+00 -4.93992448e-01 -1.37589002e+00 2.24808469e-01 -6.23414665e-02 2.20784351e-01 2.29159907e-01 7.77845919e-01 -1.18277812e+00 7.52910316e-01 -9.13995802e-01 1.49096996e-01 2.98451871e-01 1.78022370e-01 9.78357568e-02 2.38460265e-02 -1.38771594e+00 8.70243549e-01 3.40967357e-01 2.42724344e-01 -1.06673145e+00 -5.15703142e-01 -6.76464915e-01 2.54096776e-01 3.73261869e-01 -6.18492484e-01 1.38091063e+00 -9.89427388e-01 -2.19815588e+00 9.57766622e-02 -2.77480055e-02 -1.67816177e-01 4.34889644e-01 -1.83026299e-01 -5.16778469e-01 -1.09343238e-01 -3.24135184e-01 7.07079768e-01 1.10618460e+00 -1.40159810e+00 -5.54881990e-01 -1.16102539e-01 2.44140223e-01 4.27068055e-01 -8.67177904e-01 1.36117503e-01 -2.84498483e-01 -1.05341160e+00 -6.73951805e-02 -5.95267296e-01 -2.49747157e-01 -1.73010021e-01 -4.11200792e-01 -2.29341522e-01 6.97408319e-01 -1.01692557e+00 1.57098222e+00 -2.17581558e+00 1.76783293e-01 2.96461955e-02 2.21967757e-01 3.79310429e-01 -2.47511402e-01 1.93098262e-01 5.00736199e-03 9.32185799e-02 -4.75321263e-01 -7.54077077e-01 3.61523837e-01 1.41449645e-01 1.15222987e-02 2.04824105e-01 5.63515723e-01 7.92116225e-01 -1.16797459e+00 -4.34490472e-01 5.07986099e-02 8.85864079e-01 -6.75237060e-01 4.11495030e-01 7.97731876e-02 4.25952345e-01 -4.77191061e-01 5.12386560e-01 7.71182299e-01 -4.49033938e-02 1.20600499e-01 -2.29296818e-01 -9.95454863e-02 4.41766977e-01 -9.38244522e-01 1.70838189e+00 -7.26846218e-01 6.38687193e-01 1.32368177e-01 -9.59703684e-01 1.10703218e+00 4.14161116e-01 3.70675862e-01 -5.79276204e-01 2.08078884e-03 1.66201368e-01 1.33761957e-01 -2.61521190e-01 6.48658991e-01 -2.24534705e-01 4.02880758e-01 4.22094017e-01 1.53040467e-02 1.28038019e-01 -7.24635422e-02 -5.32603078e-03 1.12400162e+00 4.60558057e-01 7.13385120e-02 -1.30944848e-01 4.55965787e-01 -4.94947016e-01 9.79154289e-01 4.88146871e-01 -6.21077120e-01 6.79573894e-01 3.78710538e-01 2.98323482e-01 -1.10001945e+00 -1.21243417e+00 1.23623451e-02 1.04227161e+00 2.01428868e-02 -2.39866495e-01 -1.01999712e+00 -9.06535268e-01 -4.84148830e-01 7.90718317e-01 -5.09427309e-01 -4.06743437e-01 -5.65428078e-01 -8.67399752e-01 4.70755160e-01 6.58315837e-01 6.12827063e-01 -9.73091543e-01 -1.92133307e-01 3.99800509e-01 -1.85337856e-01 -8.50414574e-01 -8.94677341e-01 2.38575757e-01 -7.44310260e-01 -3.13580960e-01 -1.03477681e+00 -6.42582655e-01 6.45676613e-01 1.39939830e-01 9.37303722e-01 -3.55709106e-01 6.88233197e-01 1.40454069e-01 -4.24714595e-01 -2.96884805e-01 -5.42288125e-01 -1.97408646e-02 -1.16190881e-01 2.72021562e-01 3.18519950e-01 -7.86056340e-01 -9.15722728e-01 2.71652579e-01 -7.35203683e-01 -1.36917546e-01 7.42115259e-01 1.15269423e+00 3.68637115e-01 1.38877168e-01 1.08257711e+00 -3.21374863e-01 9.59919333e-01 -3.85632217e-01 -3.95015687e-01 3.40855151e-01 -9.99680161e-01 2.28247702e-01 6.10155880e-01 -6.18375003e-01 -1.39783216e+00 -1.87637761e-01 -3.43017459e-01 -4.97870535e-01 1.39742076e-01 6.27738237e-01 -5.08806646e-01 1.94858655e-01 4.94929194e-01 1.84968501e-01 -2.44583026e-01 -4.12198782e-01 3.87588501e-01 8.04790735e-01 4.46559608e-01 -5.80302417e-01 9.36552227e-01 1.08062327e-01 -5.28811574e-01 -3.63114059e-01 -6.78935230e-01 -1.73549175e-01 -1.19073078e-01 -3.47310364e-01 7.90426791e-01 -7.69776404e-01 -3.28427166e-01 4.42298084e-01 -1.34108329e+00 -3.86345416e-01 -4.54808652e-01 7.24590659e-01 -4.83128637e-01 3.96544158e-01 -5.43828189e-01 -1.11142528e+00 -4.92852300e-01 -1.31432998e+00 9.15265381e-01 3.38700116e-01 1.26432970e-01 -9.33097780e-01 1.37704641e-01 2.16837555e-01 9.02907312e-01 8.70387908e-03 7.36072659e-01 -3.88688505e-01 -4.91554290e-01 3.89157236e-02 -4.00221884e-01 9.17839587e-01 1.71120510e-01 1.56343058e-02 -1.21872616e+00 -1.82126448e-01 4.06462967e-01 -2.04364777e-01 7.13595629e-01 5.06469667e-01 1.01611602e+00 -1.88463002e-01 2.72395372e-01 5.48690140e-01 1.13808072e+00 3.54205817e-01 6.75471246e-01 2.59682775e-01 5.17247558e-01 5.94697654e-01 5.43393850e-01 3.80608171e-01 4.40459013e-01 7.77075529e-01 2.61275828e-01 -1.75824553e-01 -4.36780959e-01 -4.19479579e-01 7.22064734e-01 1.40749884e+00 7.48253316e-02 -3.20160121e-01 -4.88081783e-01 4.01380390e-01 -1.69723034e+00 -8.20308447e-01 4.92023677e-01 2.17190576e+00 1.13018787e+00 1.24009311e-01 -7.14012459e-02 3.09348941e-01 9.94819403e-01 3.79713118e-01 -6.50379241e-01 -3.01520437e-01 -1.77536219e-01 -2.57736333e-02 1.35157451e-01 5.45610726e-01 -1.07396686e+00 7.47268319e-01 6.07552910e+00 1.09780228e+00 -1.10519159e+00 3.92368615e-01 6.92721307e-01 7.66186416e-02 -6.37919009e-01 -1.79016292e-02 -3.91604900e-01 4.72942233e-01 1.13690460e+00 -2.23634437e-01 7.17921853e-01 6.37682140e-01 6.24329388e-01 3.48789871e-01 -7.79404223e-01 8.70347798e-01 -1.33771807e-01 -8.64296556e-01 -3.18933606e-01 2.13386878e-01 7.57144392e-01 4.81216982e-03 3.66443515e-01 6.85126424e-01 4.70406353e-01 -6.57107651e-01 8.89431894e-01 4.78842437e-01 5.75962424e-01 -7.40927339e-01 5.13208985e-01 9.95208994e-02 -1.03036761e+00 -1.38673499e-01 -2.63984591e-01 4.06477332e-01 5.57983518e-01 7.26335168e-01 -6.97593927e-01 5.01406312e-01 3.63124371e-01 5.95614135e-01 -4.99302633e-02 1.05118537e+00 -5.08217216e-01 9.82695103e-01 -3.11314575e-02 5.35386130e-02 3.62824768e-01 -5.29958069e-01 7.62395620e-01 1.19697499e+00 5.66471815e-01 -1.75417840e-01 -4.78689969e-02 1.13180554e+00 -2.59746253e-01 -4.94324975e-02 -1.78685933e-01 -1.49662301e-01 5.47762156e-01 1.27869868e+00 -2.48516217e-01 -1.38564855e-01 -3.83793861e-01 9.06297505e-01 3.45533401e-01 7.72849441e-01 -1.22705436e+00 -5.96461892e-01 6.17059350e-01 -3.27408731e-01 4.76363361e-01 -2.00260252e-01 -4.12838876e-01 -9.10089910e-01 2.12332159e-01 -9.79275286e-01 -1.57895550e-01 -6.61678672e-01 -1.34239793e+00 9.90342379e-01 -2.85224676e-01 -1.25605297e+00 -2.48949289e-01 -1.95527747e-01 -8.23964834e-01 9.95090008e-01 -1.95230961e+00 -1.03172731e+00 -6.49622530e-02 3.80156964e-01 5.63370883e-01 -8.56782589e-03 4.67752427e-01 6.97441101e-01 -9.19175029e-01 1.02165568e+00 3.54727179e-01 -2.01772284e-02 7.19381928e-01 -1.42840576e+00 2.98068404e-01 1.10046864e+00 4.84603345e-02 3.98594320e-01 6.34323418e-01 -4.16831464e-01 -9.17330980e-01 -1.07681465e+00 8.16022098e-01 -1.14383802e-01 7.35523164e-01 -1.04654402e-01 -9.50860500e-01 3.73077057e-02 4.82369930e-01 -2.62759954e-01 5.46495259e-01 4.63979058e-02 -2.78895676e-01 -3.31453174e-01 -1.00481915e+00 7.19970942e-01 1.07058883e+00 -6.87335134e-01 -3.72593731e-01 8.17312896e-02 1.07266271e+00 -1.46963894e-01 -1.00133038e+00 4.37799513e-01 2.05119342e-01 -7.47172236e-01 8.32825065e-01 -5.17068744e-01 4.36380804e-01 -3.69798303e-01 -3.58018965e-01 -1.66669488e+00 -2.86557436e-01 -9.47543561e-01 -5.02843559e-01 1.57150269e+00 5.92259347e-01 -5.22242308e-01 5.86387277e-01 5.78220248e-01 -4.83465314e-01 -9.96314228e-01 -1.00797606e+00 -1.05680406e+00 1.35593057e-01 -4.44452107e-01 8.65788519e-01 7.06725180e-01 -1.85859501e-01 3.15290600e-01 -6.51678681e-01 2.33909026e-01 6.14857078e-01 -2.37209409e-01 3.32978487e-01 -6.65969014e-01 -6.39027953e-01 -8.10139596e-01 -2.45157573e-02 -1.34160292e+00 8.92753303e-02 -6.39882982e-01 4.57305133e-01 -1.40143800e+00 -2.88986620e-02 -4.91170168e-01 -9.07164812e-01 2.62645602e-01 -5.92461050e-01 -1.30352691e-01 2.44554117e-01 -1.62909534e-02 -4.91738737e-01 1.34431231e+00 1.35307813e+00 -2.61452287e-01 -2.50615716e-01 1.37749359e-01 -9.00430560e-01 4.85794276e-01 8.14826012e-01 -5.59033215e-01 -5.69857717e-01 -6.43930256e-01 3.47635075e-02 2.27080788e-02 7.20395371e-02 -8.83386016e-01 2.73826003e-01 -1.97656617e-01 -2.15836391e-01 -2.10067347e-01 3.97370338e-01 -5.46849310e-01 -3.64788830e-01 9.99774411e-02 -5.48272014e-01 5.66227846e-02 -2.74047732e-01 7.66953528e-01 -5.36848962e-01 -2.57710934e-01 7.44591534e-01 4.60731119e-01 -3.29972506e-01 4.19837445e-01 -3.39357033e-02 1.08944960e-01 5.33300579e-01 -9.07589644e-02 -2.58389473e-01 -5.86467683e-01 -5.38608193e-01 7.90639669e-02 -1.08055212e-02 4.25186932e-01 6.50562823e-01 -1.60610735e+00 -1.05870342e+00 -2.16798544e-01 -2.14818716e-01 -3.26784998e-01 2.51266897e-01 9.20097291e-01 1.10085323e-01 1.07995749e-01 2.52733558e-01 -3.49096239e-01 -7.21919000e-01 3.63023311e-01 4.63708222e-01 -6.12394691e-01 -2.54965603e-01 6.44041955e-01 2.34921768e-01 -5.20251393e-01 5.61333418e-01 -1.14616007e-01 -2.69248635e-01 -2.66322434e-01 4.11728412e-01 4.78256792e-01 3.94274294e-02 -3.81576419e-01 -6.67595789e-02 1.51475757e-01 5.38647398e-02 -6.17072225e-01 1.30388474e+00 -4.56300408e-01 3.60342056e-01 4.04606499e-02 1.25792074e+00 -3.93712107e-05 -1.87477577e+00 -5.37662983e-01 -2.79701725e-02 -3.65913808e-01 5.96650958e-01 -1.13321006e+00 -1.48664224e+00 9.88047659e-01 7.13006079e-01 2.41479985e-02 1.36698198e+00 -4.08651024e-01 1.03832579e+00 1.72319159e-01 -5.40853664e-02 -1.37725711e+00 2.70472080e-01 5.70907056e-01 9.69420016e-01 -1.33471179e+00 -3.62890124e-01 -6.72581345e-02 -7.80635357e-01 8.18418026e-01 5.68990648e-01 1.29499719e-01 7.20628798e-01 3.08556855e-01 2.58360654e-01 1.20748766e-01 -6.47659361e-01 -1.99421376e-01 4.63127911e-01 8.56773913e-01 4.05275196e-01 1.20015934e-01 -3.69790971e-01 6.57100141e-01 -8.97648628e-04 -2.27821782e-01 6.25067055e-02 6.87311172e-01 -2.59612352e-01 -1.40046036e+00 -2.19435722e-01 1.55999109e-01 -2.74547219e-01 -4.54225808e-01 1.44481048e-01 2.99837291e-01 -6.43069521e-02 1.14277971e+00 -2.74772346e-01 -5.67304313e-01 3.47730786e-01 -1.85362041e-01 9.74077061e-02 -1.20088086e-01 -5.61595261e-01 2.51305491e-01 1.03498854e-01 -2.99017280e-01 -4.03955489e-01 -5.67397714e-01 -9.83477294e-01 -5.60262650e-02 -5.86423516e-01 1.78452671e-01 9.34012771e-01 8.84853423e-01 4.00147885e-01 8.53116155e-01 1.17627108e+00 -5.70712984e-01 -1.13059831e+00 -1.11403453e+00 -4.16224658e-01 2.81814545e-01 2.92967558e-01 -5.57684243e-01 -4.47430551e-01 -2.11672232e-01]
[14.833250045776367, 5.9426960945129395]
c92c6f4a-b9d4-46e5-9f8e-3cbd18be0f47
explainability-of-the-implications-of
2112.04827
null
https://arxiv.org/abs/2112.04827v1
https://arxiv.org/pdf/2112.04827v1.pdf
Explainability of the Implications of Supervised and Unsupervised Face Image Quality Estimations Through Activation Map Variation Analyses in Face Recognition Models
It is challenging to derive explainability for unsupervised or statistical-based face image quality assessment (FIQA) methods. In this work, we propose a novel set of explainability tools to derive reasoning for different FIQA decisions and their face recognition (FR) performance implications. We avoid limiting the deployment of our tools to certain FIQA methods by basing our analyses on the behavior of FR models when processing samples with different FIQA decisions. This leads to explainability tools that can be applied for any FIQA method with any CNN-based FR solution using activation mapping to exhibit the network's activation derived from the face embedding. To avoid the low discrimination between the general spatial activation mapping of low and high-quality images in FR models, we build our explainability tools in a higher derivative space by analyzing the variation of the FR activation maps of image sets with different quality decisions. We demonstrate our tools and analyze the findings on four FIQA methods, by presenting inter and intra-FIQA method analyses. Our proposed tools and the analyses based on them point out, among other conclusions, that high-quality images typically cause consistent low activation on the areas outside of the central face region, while low-quality images, despite general low activation, have high variations of activation in such areas. Our explainability tools also extend to analyzing single images where we show that low-quality images tend to have an FR model spatial activation that strongly differs from what is expected from a high-quality image where this difference also tends to appear more in areas outside of the central face region and does correspond to issues like extreme poses and facial occlusions. The implementation of the proposed tools is accessible here [link].
['Naser Damer', 'Biying Fu']
2021-12-09
null
null
null
null
['face-image-quality', 'face-image-quality-assessment']
['computer-vision', 'computer-vision']
[ 1.22974932e-01 4.39340770e-01 3.39638591e-01 -8.29533875e-01 -1.26812026e-01 -4.13215578e-01 5.50058067e-01 -2.89507538e-01 9.47158784e-03 4.03932244e-01 1.78341120e-01 1.06400132e-01 -6.85878277e-01 -7.49500573e-01 -5.35212219e-01 -6.95812762e-01 -4.05142978e-02 1.96926326e-01 -2.12031469e-01 -3.89618635e-01 1.85888097e-01 9.79598701e-01 -1.94241166e+00 7.63347507e-01 5.87211609e-01 9.16611552e-01 -2.72167504e-01 5.25806546e-01 -9.37781185e-02 5.30800700e-01 -6.77672267e-01 -6.35632694e-01 4.13722485e-01 -8.33792508e-01 -7.85934031e-01 -5.64222857e-02 1.10066080e+00 -2.51889735e-01 4.06354927e-02 1.13502038e+00 3.40378255e-01 -1.39774606e-01 9.90635216e-01 -1.58010256e+00 -7.71588326e-01 2.64264911e-01 -3.79022747e-01 4.50529665e-01 4.07718986e-01 3.84043604e-01 7.13755548e-01 -9.19405937e-01 5.48199475e-01 1.79465413e+00 7.57102430e-01 8.41843963e-01 -1.44618511e+00 -5.36427736e-01 -2.53782216e-02 7.53018707e-02 -1.38861489e+00 -5.47755897e-01 6.76952899e-01 -5.15219808e-01 8.90675604e-01 4.11306113e-01 5.36379874e-01 9.76617873e-01 2.97551066e-01 1.64298475e-01 1.62597585e+00 -3.67598802e-01 1.93585262e-01 3.37345958e-01 2.64694899e-01 8.16375613e-01 1.88775897e-01 2.17862532e-01 -5.39737403e-01 7.04886615e-02 8.27761769e-01 -2.57157952e-01 -2.41373837e-01 -1.27867684e-01 -5.87416589e-01 7.69150078e-01 6.55804574e-01 8.57043087e-01 -4.16049570e-01 3.04997981e-01 -9.20382962e-02 5.96708059e-01 2.89269090e-01 3.99117351e-01 -3.90336215e-01 2.57945269e-01 -1.15764928e+00 1.53262421e-01 4.52342153e-01 4.72098023e-01 1.23330200e+00 2.55048752e-01 -3.37098420e-01 5.22503257e-01 3.00488353e-01 4.09258008e-01 1.68476537e-01 -1.12278998e+00 -2.13940106e-02 6.33957565e-01 -1.39153540e-01 -1.37829661e+00 -4.18789834e-01 -3.69552165e-01 -6.35227978e-01 7.22498596e-01 6.78346455e-01 1.46697432e-01 -8.27764571e-01 1.70982265e+00 1.46515435e-02 -1.96566403e-01 -5.48726432e-02 9.68927324e-01 9.20003951e-01 3.52428824e-01 2.72372127e-01 -3.84405077e-01 1.41250169e+00 -4.27022606e-01 -1.01088941e+00 2.31009647e-01 4.95093048e-01 -5.45697391e-01 1.46069777e+00 4.72820371e-01 -1.14302611e+00 -8.72386813e-01 -1.03261495e+00 1.22295894e-01 -4.90746647e-01 2.75451064e-01 4.17878538e-01 1.10820401e+00 -1.61705315e+00 9.53585446e-01 -3.19631338e-01 -6.29206717e-01 7.73950636e-01 5.77807426e-01 -7.15138376e-01 1.08868152e-01 -8.24735820e-01 1.10691798e+00 -8.16519186e-02 4.35274363e-01 -1.02320695e+00 -6.52965546e-01 -4.23537046e-01 2.10039750e-01 -9.74451229e-02 -5.05814433e-01 5.12147188e-01 -1.92347562e+00 -1.32497811e+00 9.29268718e-01 -1.61563948e-01 -1.63903877e-01 2.94893414e-01 -5.18313050e-02 -6.71587706e-01 2.13292301e-01 -2.45058939e-01 8.96815479e-01 1.06284654e+00 -1.52220345e+00 6.72567189e-02 -5.53288460e-01 1.57697171e-01 -1.53518721e-01 -3.14350307e-01 1.68483719e-01 2.18995020e-01 -2.98114717e-01 2.24081166e-02 -5.17642140e-01 1.69743851e-01 2.62781918e-01 -1.65167943e-01 -1.61755085e-01 7.66564727e-01 -5.01882672e-01 9.20121372e-01 -2.17622185e+00 -5.55975251e-02 5.26324749e-01 3.26194376e-01 6.73978105e-02 -4.27155852e-01 1.08849473e-01 -5.56744814e-01 5.44681311e-01 -2.75203705e-01 -2.56191324e-02 -2.66065844e-03 3.24557155e-01 5.04109599e-02 7.19573915e-01 8.05029571e-01 6.26075506e-01 -5.47754645e-01 -4.58182514e-01 1.07481219e-01 7.86118209e-01 -7.29281187e-01 6.23860769e-02 5.67050576e-02 5.59710860e-01 -7.84807131e-02 3.57533902e-01 1.02927613e+00 1.45014867e-01 -4.19719033e-02 -6.20680511e-01 2.69028265e-02 -2.22339466e-01 -1.10224283e+00 1.31785250e+00 -2.10409790e-01 8.20228338e-01 7.54527897e-02 -9.37043786e-01 1.02029097e+00 2.67273009e-01 1.82546616e-01 -8.55372131e-01 2.93826967e-01 1.77210510e-01 5.12747049e-01 -4.54918593e-01 1.72784850e-02 -3.52877468e-01 7.26374090e-01 4.33091611e-01 4.97461945e-01 5.40500954e-02 -9.22702253e-02 1.17147844e-02 9.55048978e-01 -5.78809604e-02 8.24855734e-03 -9.29952979e-01 9.00838554e-01 -4.98132855e-01 2.68202513e-01 4.58121419e-01 -4.32767659e-01 6.68650270e-01 9.22172070e-01 -5.49436152e-01 -8.86092365e-01 -1.23135185e+00 -4.51956421e-01 6.06698215e-01 -1.42341197e-01 -1.67346016e-01 -1.16012692e+00 -8.55129719e-01 -1.07901111e-01 5.31692207e-01 -1.17557037e+00 -2.12619066e-01 -5.03999174e-01 -5.60775578e-01 5.03456295e-01 5.53048328e-02 5.65933883e-01 -1.36528540e+00 -5.27914464e-01 -5.13589978e-01 2.31375098e-01 -7.27090061e-01 1.53033614e-01 -2.06242606e-01 -9.68377113e-01 -1.05446386e+00 -4.60565656e-01 -3.75934958e-01 1.02342534e+00 -1.74663335e-01 1.31666529e+00 6.69833362e-01 -3.67003620e-01 6.46023095e-01 -1.23414315e-01 -2.45163456e-01 -5.50929904e-01 -2.85600692e-01 1.62881285e-01 3.25423241e-01 3.74583066e-01 -3.74495834e-01 -7.49653757e-01 4.88977671e-01 -1.05716217e+00 -3.92475545e-01 4.34121817e-01 5.67574739e-01 3.58946979e-01 4.32318673e-02 5.14330685e-01 -6.90307915e-01 6.57251239e-01 -3.47415209e-01 -2.60264158e-01 2.67530501e-01 -8.33487809e-01 3.16645801e-01 3.19219619e-01 -1.59502253e-01 -1.13704872e+00 -1.93802625e-01 -1.06916189e-01 -2.80007273e-01 -5.68164766e-01 -1.83018401e-01 -3.90287966e-01 -2.93494850e-01 1.20847809e+00 -2.40032002e-01 4.17731941e-01 -2.54650682e-01 3.86990666e-01 3.89216781e-01 1.35830566e-01 -4.47769016e-01 9.73849654e-01 6.37159705e-01 1.32412344e-01 -8.16473246e-01 -5.34886718e-01 1.55265778e-01 -6.48808599e-01 -7.52504408e-01 1.08722889e+00 -3.90063345e-01 -8.10836554e-01 -1.11353965e-02 -1.19491422e+00 -1.99675798e-01 -4.61337924e-01 2.47962832e-01 -6.15946651e-01 1.21524148e-01 -3.24008673e-01 -9.86841500e-01 -1.39699236e-01 -1.32867503e+00 9.22818959e-01 3.75839472e-01 -3.44166905e-01 -9.59424317e-01 -1.20628200e-01 2.24050492e-01 6.10971570e-01 2.06862137e-01 1.05052173e+00 -4.65772510e-01 -3.66975576e-01 4.01788533e-01 -3.48519981e-01 5.45520902e-01 1.90613061e-01 4.21753466e-01 -1.45875514e+00 -2.19155729e-01 1.05017737e-01 -1.65115878e-01 6.58658087e-01 4.56601024e-01 1.03527105e+00 -3.42044562e-01 6.34191930e-02 4.08674121e-01 1.59878373e+00 -7.08512291e-02 1.11309028e+00 -2.80617736e-02 3.33074540e-01 1.30189431e+00 5.03875613e-01 -1.27618462e-01 -5.43296576e-01 8.10221195e-01 6.24103665e-01 -4.91519719e-01 -4.78636801e-01 9.11984369e-02 7.46643722e-01 2.04055563e-01 -3.10795337e-01 7.26517662e-02 -6.11292362e-01 3.30198854e-01 -1.33267009e+00 -1.14167488e+00 -5.20930290e-01 2.02679539e+00 4.79105860e-01 -2.88070440e-01 1.76217407e-01 2.32399151e-01 6.01497352e-01 -1.48072958e-01 -6.93041682e-02 -9.08697188e-01 -2.77606815e-01 5.87974548e-01 -2.60474328e-02 6.82074308e-01 -5.27908802e-01 7.37248361e-01 7.17961979e+00 6.62582517e-01 -1.15582514e+00 2.68031985e-01 9.98522580e-01 -1.11215696e-01 -6.08273983e-01 -7.77921900e-02 -4.70199049e-01 2.00532690e-01 1.15371692e+00 3.72606993e-01 4.82710212e-01 6.64081216e-01 3.45771581e-01 -1.02438092e-01 -1.26394713e+00 1.02119005e+00 1.61287442e-01 -1.05488873e+00 2.66124815e-01 6.88616782e-02 5.35181284e-01 -5.26018202e-01 3.64069432e-01 -8.96182582e-02 -1.51031926e-01 -1.72029459e+00 5.94142020e-01 6.89253211e-01 8.52542639e-01 -6.33124113e-01 7.10858881e-01 -3.16655725e-01 -7.65234113e-01 -1.73869297e-01 -5.88693321e-01 -6.02981895e-02 -5.28582931e-01 5.77017069e-01 -5.49740791e-01 3.47746998e-01 9.85888362e-01 2.87138730e-01 -8.67356122e-01 3.88507575e-01 -2.11475611e-01 4.34048146e-01 -4.26383503e-02 2.63632476e-01 -1.26594320e-01 -2.21647859e-01 3.22908640e-01 1.07223642e+00 3.64330918e-01 -5.50223961e-02 -9.01225507e-01 1.55564392e+00 3.46613616e-01 3.43842447e-01 -8.72785628e-01 2.64400780e-01 -7.21209273e-02 1.49767005e+00 -8.35846126e-01 -1.55411020e-01 -2.92003274e-01 9.12123859e-01 1.07355177e-01 4.27531451e-01 -7.96777785e-01 5.79432137e-02 6.59253836e-01 4.58927304e-01 -5.53161278e-02 1.90077513e-01 -2.83844650e-01 -7.55700767e-01 -1.75072160e-02 -1.03968012e+00 2.49740511e-01 -1.06310225e+00 -1.32547438e+00 1.06538677e+00 2.91914284e-01 -8.89655113e-01 -2.11250540e-02 -8.65560234e-01 -4.89961743e-01 1.09888411e+00 -1.26751769e+00 -8.92033517e-01 -6.02477908e-01 9.04445052e-01 1.52724370e-01 -3.30854625e-01 7.29123652e-01 3.09366107e-01 -4.19039965e-01 7.25287259e-01 -6.83600783e-01 2.25106906e-02 4.65682358e-01 -9.57874954e-01 -1.12343304e-01 8.28253090e-01 3.79227936e-01 8.72735918e-01 8.19969594e-01 -2.98003882e-01 -9.17598248e-01 -6.98802233e-01 6.61263108e-01 -6.78745151e-01 9.45699811e-02 -2.35763147e-01 -9.59362268e-01 3.22049409e-01 4.90968853e-01 1.35781974e-01 5.62110186e-01 1.69582486e-01 -4.22285229e-01 -5.05692184e-01 -1.61451495e+00 4.44263607e-01 1.12818837e+00 -5.49874246e-01 -4.33560342e-01 1.69415146e-01 -1.29332545e-03 4.62727636e-01 -9.03712273e-01 4.01316673e-01 4.85742718e-01 -1.70556879e+00 7.17733741e-01 -5.25178015e-01 3.95433873e-01 -5.29513419e-01 -4.94200066e-02 -1.13408434e+00 -3.47451240e-01 -2.49582410e-01 3.80784482e-01 1.22585189e+00 4.63380784e-01 -7.47935653e-01 7.58060932e-01 5.85499048e-01 1.40554979e-01 -3.70389968e-01 -1.12982106e+00 -7.27244675e-01 1.44911125e-01 -3.36966366e-01 5.80938995e-01 9.32929218e-01 -3.10792834e-01 -2.46578567e-02 1.49986356e-01 2.38062292e-01 4.77598131e-01 -3.67343336e-01 6.81413114e-01 -1.33537400e+00 -1.92748278e-01 -5.23060560e-01 -7.43820488e-01 -1.05471551e-01 6.73563257e-02 -8.08253288e-01 -1.21817924e-01 -1.27543783e+00 1.99570134e-01 -4.33546215e-01 -2.59794623e-01 5.89966297e-01 2.79060513e-01 5.18024445e-01 2.31870443e-01 2.61200190e-01 2.83516850e-02 3.45596224e-01 1.28359056e+00 2.37413310e-02 8.63554105e-02 -5.88056743e-01 -5.54522216e-01 6.59769058e-01 6.75470293e-01 -5.38452804e-01 -6.11628234e-01 -1.29443437e-01 3.62989932e-01 -2.81483620e-01 6.43654883e-01 -1.40535378e+00 -5.42353094e-01 -2.35865526e-02 7.42188156e-01 6.30306005e-02 1.83729216e-01 -1.16073430e+00 5.71250319e-01 6.92041814e-01 -2.56813735e-01 1.39296472e-01 3.03560287e-01 3.34751010e-02 -2.19250947e-01 -2.25889757e-01 1.03824234e+00 -1.35072991e-01 -2.92976052e-01 6.55735508e-02 -3.71377379e-01 -4.62741464e-01 8.63509476e-01 -7.46326327e-01 -3.29700083e-01 -5.82224190e-01 -9.83009815e-01 -4.17760849e-01 5.65925062e-01 2.92296112e-01 6.29914165e-01 -1.37313974e+00 -7.30616868e-01 5.46370983e-01 -1.36014119e-01 -6.40848160e-01 4.60725814e-01 1.04722190e+00 -5.16787112e-01 1.34832934e-01 -8.53626072e-01 -7.24618077e-01 -1.35224104e+00 4.41336215e-01 9.57752526e-01 7.83734471e-02 -5.50920926e-02 7.38883615e-01 5.92688560e-01 -2.05144510e-01 -2.45730639e-01 -2.63941228e-01 -5.91702700e-01 1.64663196e-02 4.94764715e-01 2.79766172e-01 1.53627306e-01 -9.82811034e-01 -4.01101440e-01 8.96366179e-01 2.43486837e-01 -2.07028594e-02 1.00195014e+00 1.83411047e-01 -2.96449721e-01 2.16983557e-01 1.17025518e+00 1.72164261e-01 -1.07848787e+00 5.83465219e-01 -2.18411922e-01 -5.21910608e-01 -6.80179670e-02 -8.66860271e-01 -1.56378388e+00 1.17990971e+00 1.43908894e+00 1.77839980e-01 1.46559930e+00 1.80754624e-02 -2.43744165e-01 -8.62645432e-02 2.69291431e-01 -9.59707618e-01 2.69408643e-01 -3.10956966e-02 1.38904965e+00 -9.10038292e-01 4.79502380e-02 -5.92797816e-01 -4.16395426e-01 1.33877504e+00 7.56742775e-01 -2.17347503e-01 6.94145501e-01 3.98650728e-02 2.44329602e-01 -7.60745704e-01 -4.12674665e-01 -1.72742203e-01 5.00375807e-01 8.75538707e-01 6.07171774e-01 -2.19194144e-01 -4.29696858e-01 4.76280510e-01 -2.19322771e-01 -6.06473275e-02 4.56464797e-01 4.23087060e-01 -1.50121465e-01 -1.10342038e+00 -5.00287712e-01 4.01151210e-01 -4.35990602e-01 5.23843197e-03 -7.16422915e-01 1.07859099e+00 6.32600546e-01 9.32857394e-01 4.07415867e-01 -4.55089748e-01 3.91569167e-01 2.26736069e-01 8.02878797e-01 -3.37045103e-01 -1.01656055e+00 -3.00859839e-01 -6.29375502e-02 -9.00378466e-01 -8.26858222e-01 -5.63809156e-01 -1.23727012e+00 -5.50568640e-01 -1.76904306e-01 -5.06165773e-02 5.22053838e-01 8.72158170e-01 4.09780800e-01 3.57333392e-01 3.74006987e-01 -6.11114323e-01 2.11451910e-02 -1.08213925e+00 -5.89410663e-01 7.79750705e-01 1.65034905e-01 -8.64092171e-01 -7.04821348e-01 4.83114682e-02]
[12.964550018310547, 0.9538121819496155]
f513f2f4-f8e5-450d-9638-14e76140f66a
data-driven-grammatical-error-detection-in
null
null
https://aclanthology.org/D14-1106
https://aclanthology.org/D14-1106.pdf
Data Driven Grammatical Error Detection in Transcripts of Children's Speech
null
['Anna Eva Hallin', 'Eric Morley', 'Brian Roark']
2014-10-01
null
null
null
emnlp-2014-10
['grammatical-error-detection']
['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]
[-7.22912073135376, 3.7041378021240234]
b7c077f8-1256-48d5-b7c6-4f0e636eb8e1
recurrent-instance-segmentation
1511.08250
null
http://arxiv.org/abs/1511.08250v3
http://arxiv.org/pdf/1511.08250v3.pdf
Recurrent Instance Segmentation
Instance segmentation is the problem of detecting and delineating each distinct object of interest appearing in an image. Current instance segmentation approaches consist of ensembles of modules that are trained independently of each other, thus missing opportunities for joint learning. Here we propose a new instance segmentation paradigm consisting in an end-to-end method that learns how to segment instances sequentially. The model is based on a recurrent neural network that sequentially finds objects and their segmentations one at a time. This net is provided with a spatial memory that keeps track of what pixels have been explained and allows occlusion handling. In order to train the model we designed a principled loss function that accurately represents the properties of the instance segmentation problem. In the experiments carried out, we found that our method outperforms recent approaches on multiple person segmentation, and all state of the art approaches on the Plant Phenotyping dataset for leaf counting.
['Bernardino Romera-Paredes', 'Philip H. S. Torr']
2015-11-25
null
null
null
null
['plant-phenotyping', 'occlusion-handling']
['computer-vision', 'computer-vision']
[ 5.99768043e-01 5.10125995e-01 -2.97349155e-01 -4.95559305e-01 -7.02870548e-01 -6.26917899e-01 3.04242104e-01 2.90359944e-01 -1.92482740e-01 4.40541416e-01 -6.04953110e-01 -3.54321778e-01 -1.19290784e-01 -8.53866994e-01 -1.01788259e+00 -6.86074257e-01 -9.60759595e-02 9.23110902e-01 3.12097400e-01 4.66244161e-01 2.12630749e-01 4.86436158e-01 -1.44639146e+00 3.63537520e-01 7.31759548e-01 8.83591473e-01 7.01492369e-01 9.72657621e-01 -3.85515302e-01 8.66730928e-01 -8.45792651e-01 1.03030764e-02 1.05619743e-01 -3.92548740e-01 -1.18975127e+00 6.65819764e-01 6.76201761e-01 -1.21359430e-01 1.46077797e-01 7.87181318e-01 2.36915976e-01 -1.50257006e-01 5.55339813e-01 -9.68726814e-01 -5.78956544e-01 8.58248711e-01 -1.07175756e+00 8.55759159e-02 -1.96090147e-01 3.46243121e-02 1.11243522e+00 -4.71957713e-01 6.62412703e-01 1.01905680e+00 7.02882707e-01 5.01428545e-01 -1.78148019e+00 -1.73637241e-01 6.18617117e-01 -1.24795355e-01 -9.69784260e-01 -4.50200364e-02 4.56003428e-01 -3.43561858e-01 6.91908479e-01 2.49402255e-01 7.41169155e-01 5.64037859e-01 -3.31696779e-01 1.31236839e+00 7.97012508e-01 -5.64470649e-01 3.64603519e-01 1.43795893e-01 5.08187294e-01 8.85187984e-01 2.14227792e-02 -1.48300752e-01 -3.28687672e-03 1.48714289e-01 8.35436285e-01 -1.74536511e-01 5.58182299e-02 -8.54771554e-01 -9.86900270e-01 6.66848302e-01 6.71624362e-01 4.33915854e-01 -4.12982523e-01 2.25842759e-01 3.76724690e-01 -3.33155245e-01 5.56446373e-01 4.33822364e-01 -9.22254801e-01 5.26469648e-01 -1.52117121e+00 1.97554469e-01 9.60462391e-01 8.08970928e-01 6.45976067e-01 -3.66269886e-01 -6.82721555e-01 6.53576851e-01 1.99659497e-01 1.51413023e-01 2.76487786e-02 -1.21544194e+00 8.94964784e-02 9.40195501e-01 8.88371617e-02 -4.60416257e-01 -3.36355448e-01 -5.34134984e-01 -4.95115429e-01 5.03871560e-01 8.39217603e-01 -1.70488358e-02 -1.22546649e+00 1.76628625e+00 3.93427074e-01 3.04228216e-01 -3.27209115e-01 4.74537760e-01 6.65606022e-01 6.20086491e-01 6.14550114e-01 8.64603370e-02 1.30726564e+00 -1.08181155e+00 -3.96457613e-01 -3.78374845e-01 6.31392479e-01 -4.05005246e-01 7.97496319e-01 3.78127038e-01 -1.39416540e+00 -7.38599420e-01 -8.07817221e-01 3.76460813e-02 -6.81957245e-01 8.62750173e-01 5.96763670e-01 5.54783762e-01 -1.07711709e+00 9.69885230e-01 -8.45095634e-01 -2.70819932e-01 8.59054804e-01 5.10183394e-01 -8.44940096e-02 2.89215058e-01 -3.47938329e-01 7.01261699e-01 8.46086502e-01 4.37485695e-01 -1.13635135e+00 -5.87773025e-01 -6.04577184e-01 1.95308030e-01 5.69571733e-01 -5.22937298e-01 1.36369300e+00 -1.26706123e+00 -1.37455916e+00 1.46676981e+00 -3.25733304e-01 -7.28964508e-01 2.43222505e-01 -1.12809494e-01 1.01491690e-01 2.17029043e-02 3.37060541e-01 1.13311982e+00 1.11074042e+00 -1.45171177e+00 -9.30287063e-01 -7.34109223e-01 -7.94954970e-02 -1.27669707e-01 1.34315342e-01 -9.84383970e-02 -4.32084352e-01 -1.45959094e-01 1.98179126e-01 -7.66876638e-01 -6.07691884e-01 1.58897430e-01 -7.67832220e-01 -1.24403253e-01 1.15804636e+00 -8.41875494e-01 8.66334558e-01 -1.91443181e+00 4.74273860e-01 1.11135393e-02 1.66488230e-01 6.54265642e-01 -2.41242036e-01 -1.21197455e-01 -3.17111552e-01 1.97777495e-01 -6.60892844e-01 -6.21098638e-01 -5.63031211e-02 2.26866364e-01 -9.91406888e-02 2.69837916e-01 4.19698536e-01 9.88256156e-01 -8.36344540e-01 -5.22819757e-01 3.47821623e-01 3.80141526e-01 -1.24999195e-01 3.74268889e-01 -8.82152021e-01 5.33375680e-01 -3.54949743e-01 7.68714786e-01 8.81622493e-01 -4.96846020e-01 2.03753054e-01 3.38833518e-02 -3.66456389e-01 -1.17639236e-01 -1.13558197e+00 1.84648860e+00 -2.78565794e-01 4.63503927e-01 2.69542664e-01 -1.42743683e+00 7.22749829e-01 3.13154876e-01 4.33527797e-01 -2.77398825e-01 5.78356050e-02 2.53823787e-01 -2.11283177e-01 -2.38319576e-01 3.90516594e-02 5.59667587e-01 2.95066983e-01 2.46867016e-01 2.46742725e-01 -1.20837748e-01 5.93188643e-01 1.07698534e-02 1.04645216e+00 7.16565669e-01 2.30075896e-01 -2.33045116e-01 6.50781453e-01 2.96621442e-01 5.95170557e-01 9.92542982e-01 -2.77639806e-01 6.24966264e-01 6.02144957e-01 -5.62736452e-01 -9.43832338e-01 -9.04876351e-01 -1.90117836e-01 1.16965067e+00 1.00881997e-02 -1.84473488e-02 -1.18676031e+00 -9.41308916e-01 -9.60449651e-02 7.81664014e-01 -8.22202146e-01 2.89274782e-01 -5.27214646e-01 -7.29809701e-01 3.60476583e-01 6.22727990e-01 5.20610332e-01 -1.51344657e+00 -1.06869066e+00 4.00441587e-01 1.31622359e-01 -1.09787393e+00 4.84404750e-02 7.69097447e-01 -1.05927312e+00 -1.14263284e+00 -8.18184316e-01 -8.94017577e-01 7.94348478e-01 1.00525253e-01 1.58700407e+00 1.17782563e-01 -9.04177964e-01 1.56348065e-01 -4.56750616e-02 -5.81838191e-01 -2.65138388e-01 6.24258757e-01 -9.20590460e-01 -1.43360287e-01 3.41434121e-01 -3.98978829e-01 -4.37274218e-01 2.34704036e-02 -7.96580970e-01 -7.99135119e-02 6.43695533e-01 7.70624042e-01 8.65305901e-01 2.40489338e-02 1.01394609e-01 -1.39185679e+00 2.26186756e-02 -2.26149142e-01 -1.09076309e+00 6.10697746e-01 -1.10398419e-01 1.19553814e-02 4.44394797e-01 -1.87163606e-01 -9.55012381e-01 8.52952898e-01 2.43022859e-01 -1.10175528e-01 -8.48269939e-01 1.33795828e-01 -2.30909079e-01 2.38172337e-03 4.26058918e-01 5.97950630e-02 -1.54479429e-01 -5.60818017e-01 6.15325928e-01 1.76487699e-01 6.17619991e-01 -4.60353225e-01 6.03206515e-01 4.76982415e-01 1.34738475e-01 -9.33210194e-01 -1.35125303e+00 -5.40465295e-01 -1.21508682e+00 -1.79642841e-01 1.20814812e+00 -6.33345664e-01 -7.37903953e-01 5.07941186e-01 -1.37448239e+00 -7.41173267e-01 -6.74800694e-01 -2.31002588e-02 -6.40217125e-01 1.56799331e-01 -4.84241694e-01 -9.10404027e-01 -2.25148201e-01 -1.07457089e+00 1.55240929e+00 5.27629614e-01 -3.43624614e-02 -9.66428041e-01 8.28632265e-02 2.95965403e-01 1.69132620e-01 2.38803208e-01 8.32401454e-01 -4.95528579e-01 -8.63969088e-01 -3.85982066e-01 -6.29018366e-01 3.00813109e-01 -1.30307689e-01 4.53585863e-01 -1.16030812e+00 -1.07056975e-01 -2.42005870e-01 -3.25924546e-01 1.03455329e+00 1.06053185e+00 1.83363700e+00 8.72920901e-02 -6.47906244e-01 5.49651980e-01 1.57771397e+00 1.97117496e-02 5.66083312e-01 2.27640778e-01 7.21086562e-01 8.12630773e-01 3.62676054e-01 1.17059946e-01 2.80047003e-02 5.16661465e-01 8.44657063e-01 -6.10504568e-01 -4.36824188e-03 1.19900011e-01 -1.54282495e-01 -3.11768711e-01 1.05084650e-01 -3.52890342e-01 -8.14191103e-01 7.59884536e-01 -2.11790729e+00 -1.05779469e+00 -2.57571369e-01 2.10083055e+00 5.60844004e-01 1.63824305e-01 2.89567292e-01 2.33301520e-01 8.01660597e-01 1.84609920e-01 -7.40192354e-01 -4.02511418e-01 1.86151732e-02 4.66918766e-01 7.51435518e-01 4.22642916e-01 -1.50679874e+00 1.25345075e+00 6.67256260e+00 5.36436796e-01 -6.98738217e-01 -1.88314214e-01 9.79253113e-01 3.81442338e-01 2.64564872e-01 3.25586379e-01 -9.52204764e-01 8.09791610e-02 6.57023132e-01 7.22402930e-01 3.30127656e-01 9.56441820e-01 5.21618314e-02 -4.22092497e-01 -1.13691020e+00 2.57532358e-01 -1.43862814e-01 -1.05006731e+00 -1.35802463e-01 -1.04283608e-01 6.03217423e-01 -2.58135974e-01 1.55220628e-02 1.09774388e-01 5.96237957e-01 -1.13699067e+00 5.42441726e-01 4.73762304e-01 1.92909807e-01 -4.10352528e-01 2.55463511e-01 6.04663014e-01 -1.17749918e+00 -2.30579644e-01 -4.91185993e-01 2.68381864e-01 -2.04658024e-02 7.45071173e-01 -8.98517370e-01 3.44619572e-01 4.51994330e-01 6.76356316e-01 -9.08542216e-01 1.25415182e+00 -3.19140941e-01 6.60487950e-01 -3.18450958e-01 3.00363451e-01 5.23918867e-01 -2.02245399e-01 3.57700706e-01 1.21787035e+00 -1.67852808e-02 -3.44614267e-01 4.88886982e-01 1.49189544e+00 6.71991110e-02 -2.07005531e-01 -5.31113386e-01 1.09027466e-02 2.20930353e-01 1.54003465e+00 -1.34643400e+00 -3.55798125e-01 -1.93677276e-01 1.15902233e+00 5.80373645e-01 1.75541878e-01 -6.80220962e-01 -2.84761488e-01 3.67572829e-02 -8.68904665e-02 8.77949715e-01 -1.27488058e-02 -4.80852753e-01 -6.60753071e-01 -9.17854384e-02 -5.57677269e-01 3.45893264e-01 -6.32017195e-01 -9.89496052e-01 2.74330527e-01 -1.81684643e-01 -5.00509322e-01 -8.05804133e-02 -9.05886233e-01 -8.37087989e-01 9.83480036e-01 -1.38251531e+00 -1.49336159e+00 -3.50744486e-01 3.67283933e-02 7.37912357e-01 8.97658914e-02 1.07450342e+00 -5.90997143e-03 -1.05297029e+00 8.72033834e-02 -6.55736104e-02 2.34858885e-01 4.50962447e-02 -1.66721714e+00 5.19990563e-01 8.94165039e-01 5.52374721e-01 3.00928533e-01 6.51602924e-01 -5.40934563e-01 -6.64813578e-01 -1.12336600e+00 8.95796597e-01 -5.71774185e-01 2.23817617e-01 -2.90607601e-01 -9.53687549e-01 8.54883015e-01 1.55847639e-01 2.45609954e-01 3.23992312e-01 1.28569573e-01 -1.75388336e-01 2.33791973e-02 -1.10669827e+00 3.00004184e-01 7.84069121e-01 -3.51488382e-01 -1.38318866e-01 6.05296195e-01 4.26890880e-01 -2.95010149e-01 -4.05953020e-01 4.12715554e-01 3.11825395e-01 -9.64708447e-01 1.12994754e+00 -8.27107787e-01 3.90044987e-01 -4.23779607e-01 5.82232215e-02 -9.51053143e-01 -5.21188498e-01 -2.47365266e-01 -2.44526103e-01 1.27527416e+00 6.85812891e-01 4.39258032e-02 1.18852091e+00 4.98216629e-01 1.50451809e-01 -6.57501459e-01 -4.24498111e-01 -5.79956472e-01 -3.69686931e-02 2.42558997e-02 4.13017452e-01 4.25614059e-01 -5.87608278e-01 2.22122177e-01 3.08718085e-02 3.45090777e-01 9.47710633e-01 5.32472908e-01 4.77044672e-01 -1.41061032e+00 -4.22960252e-01 -5.90059221e-01 -2.32815772e-01 -1.00350451e+00 5.14570773e-01 -7.59350419e-01 2.49890029e-01 -1.59284925e+00 3.31745297e-01 -3.48034233e-01 -8.35977644e-02 5.74481010e-01 -3.23208272e-01 -1.85420793e-02 -9.61803459e-03 -1.23531148e-01 -6.07024252e-01 -1.51132941e-01 9.88079846e-01 -3.39758337e-01 -4.01340544e-01 4.20015395e-01 -4.25027460e-01 9.01558220e-01 8.81830513e-01 -5.25014460e-01 -2.88697511e-01 -6.91781223e-01 -2.43673101e-01 -1.02644339e-01 7.43564427e-01 -1.05946267e+00 9.72634330e-02 1.70859411e-01 7.81345189e-01 -9.20948863e-01 2.82075047e-01 -8.01848173e-01 -2.16713697e-01 5.14458537e-01 -6.37934327e-01 -3.51552516e-01 1.35313526e-01 5.24936855e-01 7.27035552e-02 -7.53001213e-01 1.03495717e+00 -7.27837026e-01 -8.09882104e-01 3.92578691e-01 -1.90374017e-01 -7.48235174e-03 1.19862282e+00 -1.91657022e-01 1.42440930e-01 1.69357464e-01 -1.05727613e+00 2.62033314e-01 1.62197456e-01 -5.75479828e-02 9.62709934e-02 -6.71529531e-01 -6.88067973e-01 -4.17843424e-02 -1.45511359e-01 2.61861324e-01 -1.35000497e-01 4.01191056e-01 -5.19919634e-01 4.47279930e-01 -2.30704039e-01 -8.33700836e-01 -1.15655661e+00 5.22983372e-01 5.63687801e-01 -7.01750755e-01 -5.40981889e-01 1.03811347e+00 1.56163750e-02 -7.02680826e-01 6.50725007e-01 -3.50046247e-01 -4.66528833e-01 1.01468854e-01 3.14033657e-01 2.09571570e-01 -2.02479362e-02 -4.55169946e-01 -9.05578732e-02 4.93387222e-01 -1.06951401e-01 -1.99845564e-02 1.34728730e+00 2.48352468e-01 -2.32691213e-01 6.48357689e-01 8.49250853e-01 -6.67790115e-01 -1.44059050e+00 -3.01855411e-02 5.54413438e-01 -2.90203333e-01 3.52968238e-02 -1.07050097e+00 -1.32025170e+00 8.86576772e-01 6.93540156e-01 6.08196497e-01 1.09920049e+00 -7.85849020e-02 5.28997481e-01 2.96848267e-01 4.49352302e-02 -1.17331111e+00 -2.58643180e-01 3.89159113e-01 3.01731825e-01 -1.33792746e+00 -1.02316417e-01 -5.57725191e-01 -2.78561264e-01 1.08555853e+00 5.91410458e-01 -2.99996585e-01 5.75589180e-01 3.62661779e-01 -1.55299425e-01 -4.24152404e-01 -4.46181983e-01 -5.99368930e-01 1.52795002e-01 7.79124677e-01 7.32304692e-01 1.61119163e-01 -3.85735044e-03 2.78063387e-01 3.91968608e-01 5.78519814e-02 6.92265900e-03 1.01163900e+00 -5.47599077e-01 -1.27714133e+00 -2.34487548e-01 4.71365780e-01 -3.56727570e-01 2.85274863e-01 -7.92072773e-01 6.44991994e-01 3.89125109e-01 5.89829683e-01 -9.10942256e-02 2.93201983e-01 3.01366061e-01 2.50945240e-01 6.79185867e-01 -8.43625307e-01 -7.97994375e-01 -2.09739387e-01 -2.78894901e-01 -6.54545486e-01 -3.99105579e-01 -7.52223015e-01 -1.12177336e+00 3.73462528e-01 -5.90084255e-01 -1.53288335e-01 7.76126683e-01 9.45211709e-01 7.49727488e-02 9.06740308e-01 4.47847068e-01 -1.12159324e+00 -4.43853199e-01 -7.00183094e-01 -7.98637331e-01 2.54646301e-01 1.95400000e-01 -3.57046604e-01 2.71389224e-02 2.66080141e-01]
[9.480244636535645, 0.213429257273674]
2a3510b1-555f-4e6f-953c-75c52673e078
seeing-out-of-the-box-end-to-end-pre-training
2104.03135
null
https://arxiv.org/abs/2104.03135v2
https://arxiv.org/pdf/2104.03135v2.pdf
Seeing Out of tHe bOx: End-to-End Pre-training for Vision-Language Representation Learning
We study joint learning of Convolutional Neural Network (CNN) and Transformer for vision-language pre-training (VLPT) which aims to learn cross-modal alignments from millions of image-text pairs. State-of-the-art approaches extract salient image regions and align regions with words step-by-step. As region-based visual features usually represent parts of an image, it is challenging for existing vision-language models to fully understand the semantics from paired natural languages. In this paper, we propose SOHO to "See Out of tHe bOx" that takes a whole image as input, and learns vision-language representation in an end-to-end manner. SOHO does not require bounding box annotations which enables inference 10 times faster than region-based approaches. In particular, SOHO learns to extract comprehensive yet compact image features through a visual dictionary (VD) that facilitates cross-modal understanding. VD is designed to represent consistent visual abstractions of similar semantics. It is updated on-the-fly and utilized in our proposed pre-training task Masked Visual Modeling (MVM). We conduct experiments on four well-established vision-language tasks by following standard VLPT settings. In particular, SOHO achieves absolute gains of 2.0% R@1 score on MSCOCO text retrieval 5k test split, 1.5% accuracy on NLVR$^2$ test-P split, 6.7% accuracy on SNLI-VE test split, respectively.
['Jianlong Fu', 'Dongmei Fu', 'Bei Liu', 'Yupan Huang', 'Zhaoyang Zeng', 'Zhicheng Huang']
2021-04-07
null
http://openaccess.thecvf.com//content/CVPR2021/html/Huang_Seeing_Out_of_the_Box_End-to-End_Pre-Training_for_Vision-Language_Representation_CVPR_2021_paper.html
http://openaccess.thecvf.com//content/CVPR2021/papers/Huang_Seeing_Out_of_the_Box_End-to-End_Pre-Training_for_Vision-Language_Representation_CVPR_2021_paper.pdf
cvpr-2021-1
['visual-entailment']
['reasoning']
[-9.63592250e-03 -1.30711824e-01 -3.75944465e-01 -4.27637398e-01 -1.09070623e+00 -6.07124507e-01 8.46835375e-01 2.56912529e-01 -8.11627686e-01 3.44345033e-01 2.12754071e-01 -2.87334323e-01 4.51208532e-01 -4.60847586e-01 -1.20649087e+00 -2.99629152e-01 3.79698217e-01 3.67880225e-01 1.30974308e-01 -2.02430680e-01 1.94677353e-01 2.69082934e-01 -1.41194391e+00 7.63819575e-01 5.14432549e-01 1.07876277e+00 5.60779870e-01 7.38041222e-01 -2.72305787e-01 8.69891882e-01 -2.12451905e-01 -4.43195432e-01 3.41324121e-01 -1.82876140e-01 -9.03515816e-01 5.23146391e-02 1.27259123e+00 -3.62343788e-01 -5.63905835e-01 1.06779695e+00 2.73659766e-01 6.79873079e-02 7.32875645e-01 -1.30722368e+00 -1.20585394e+00 5.03475845e-01 -9.97944236e-01 3.09466928e-01 4.31208946e-02 4.05208141e-01 1.26782489e+00 -1.48131049e+00 6.78849339e-01 1.23648727e+00 3.51344943e-01 4.44901317e-01 -1.30280590e+00 -6.45308554e-01 2.15722725e-01 1.81501016e-01 -1.61667407e+00 -4.01864082e-01 5.56431711e-01 -4.37774658e-01 1.24410319e+00 8.02505314e-02 7.03191578e-01 8.74676406e-01 1.73401520e-01 1.19030225e+00 1.05996418e+00 -4.60947037e-01 -2.26970345e-01 2.69329578e-01 6.05428927e-02 1.07220149e+00 6.91834167e-02 -1.94937587e-02 -8.36371124e-01 2.82377958e-01 8.00587058e-01 2.06469983e-01 -2.36977577e-01 -5.61303794e-01 -1.44235849e+00 7.89583743e-01 9.14009631e-01 1.89918384e-01 -2.09475368e-01 4.91111904e-01 3.21556002e-01 2.61261761e-01 3.11687708e-01 1.07588597e-01 -2.35150024e-01 2.72243500e-01 -1.26211011e+00 2.22736686e-01 1.71866685e-01 1.08174860e+00 8.92794311e-01 1.66599639e-02 -5.44897079e-01 8.08745205e-01 4.45875525e-01 9.86784637e-01 3.18643868e-01 -5.91116071e-01 6.51999235e-01 6.12038612e-01 -1.66945174e-01 -7.95952439e-01 6.62687197e-02 -5.37123263e-01 -1.00830698e+00 6.77009523e-02 2.46819928e-01 4.17632490e-01 -1.32152283e+00 1.74945235e+00 5.28784208e-02 2.83716377e-02 3.27981383e-01 9.60357368e-01 1.09849131e+00 7.26260602e-01 1.70538232e-01 1.35143712e-01 1.57647395e+00 -1.36700046e+00 -1.46009669e-01 -6.29282832e-01 4.84206080e-01 -9.20267999e-01 1.47898257e+00 1.43481484e-02 -1.18799841e+00 -8.31409991e-01 -7.80265272e-01 -6.16846561e-01 -4.97283727e-01 3.15056831e-01 1.85033396e-01 5.60937170e-03 -1.23207176e+00 -7.85393938e-02 -5.44635475e-01 -3.44975710e-01 7.92663872e-01 1.50276333e-01 -5.70255756e-01 -3.90098661e-01 -8.33450913e-01 7.92232037e-01 3.90003324e-01 -1.15083531e-01 -1.35277283e+00 -9.11242008e-01 -1.11255622e+00 2.55257655e-02 2.03383654e-01 -8.17695558e-01 1.14924240e+00 -1.15153790e+00 -8.26412559e-01 1.50039852e+00 -4.84330446e-01 -7.56631196e-01 5.16708553e-01 -3.74456763e-01 -2.24716261e-01 3.25736552e-01 3.27646554e-01 1.37120819e+00 1.02534580e+00 -1.39617658e+00 -6.40304685e-01 -2.93500721e-01 2.21801596e-03 2.69232213e-01 -1.52634740e-01 -5.74745201e-02 -1.10902870e+00 -7.39634156e-01 -1.72527090e-01 -7.17727840e-01 -1.25248641e-01 3.96823555e-01 -4.40672696e-01 -8.88998806e-02 6.80067182e-01 -7.82934070e-01 7.43073940e-01 -2.00421000e+00 7.75446519e-02 1.89297274e-02 4.81520981e-01 1.86265126e-01 -5.20802259e-01 1.28164440e-01 6.99880347e-02 -1.65208384e-01 -1.73784032e-01 -5.82831323e-01 -9.32422131e-02 2.45781057e-02 -6.76128805e-01 4.01187420e-01 2.49017626e-01 1.57538021e+00 -7.39558578e-01 -7.76942194e-01 4.00325090e-01 4.35939103e-01 -5.50684810e-01 2.28012234e-01 -4.07564461e-01 1.11819245e-01 1.42810810e-02 8.02964866e-01 5.42650878e-01 -5.02646148e-01 -3.13869745e-01 -5.86346805e-01 6.77367747e-02 -2.70062119e-01 -5.82428753e-01 2.09963632e+00 -5.93271136e-01 1.08402252e+00 -2.23437086e-01 -1.07728958e+00 7.39000022e-01 -1.59139499e-01 4.29204805e-03 -1.17350554e+00 -2.24168245e-02 -6.47346973e-02 -4.39389020e-01 -2.25310072e-01 6.44610822e-01 1.26294494e-01 -2.13296577e-01 1.20606646e-01 1.96121737e-01 -8.75928923e-02 -7.04393163e-02 5.30968964e-01 5.44741213e-01 -1.35708943e-01 3.70999187e-01 -1.42133549e-01 5.48908293e-01 3.69511664e-01 9.55206230e-02 8.40946436e-01 -1.05884232e-01 6.51315629e-01 6.90052509e-02 -2.60525346e-01 -1.15497303e+00 -1.49590373e+00 1.08544372e-01 1.17873514e+00 3.81173313e-01 -3.63018155e-01 -4.47614282e-01 -5.37149131e-01 1.39686644e-01 7.35353947e-01 -6.39400542e-01 -7.01970458e-02 -3.59584689e-01 -3.93414050e-02 6.17435217e-01 7.58590996e-01 7.07101643e-01 -9.88432825e-01 -5.76744258e-01 -2.96871454e-01 -2.49171346e-01 -1.48913825e+00 -8.84472728e-01 -4.26356792e-02 -4.90392953e-01 -8.23033988e-01 -9.50820923e-01 -1.31447113e+00 8.96341622e-01 8.36733520e-01 1.40364981e+00 2.57140808e-02 -5.77012539e-01 6.81542397e-01 4.43011569e-03 -3.49825501e-01 -2.03858733e-01 -2.40748033e-01 -1.04477622e-01 -2.26857010e-02 4.45735812e-01 -2.35024884e-01 -7.93990731e-01 4.41532023e-02 -7.61543810e-01 4.40649033e-01 8.83864641e-01 1.01465917e+00 1.22627985e+00 -4.73003626e-01 5.23097403e-02 -4.30775881e-01 2.80561447e-01 -2.46637046e-01 -7.81010389e-01 6.68616235e-01 -4.67065185e-01 1.69215247e-01 5.06588638e-01 -3.59471381e-01 -6.26694977e-01 1.91125005e-01 3.77191663e-01 -1.18229628e+00 -3.12409624e-02 3.77193242e-01 -1.09800976e-02 8.84615183e-02 4.93871331e-01 7.62146771e-01 9.32918116e-03 -2.11839631e-01 1.03140414e+00 5.95433354e-01 1.11813104e+00 -4.23430115e-01 9.76003349e-01 6.82037175e-01 -3.05537909e-01 -8.06533158e-01 -1.09484470e+00 -7.34507680e-01 -6.34414256e-01 -9.63199958e-02 1.32315946e+00 -1.56157303e+00 -4.43113685e-01 2.12779969e-01 -1.09287012e+00 -4.49110955e-01 -1.11815050e-01 2.77806342e-01 -4.99988079e-01 2.74467111e-01 -2.85759240e-01 -4.02447909e-01 -8.46219659e-01 -1.10498655e+00 1.37715077e+00 2.52719611e-01 -7.70342490e-03 -7.89946079e-01 -2.07750097e-01 7.23751903e-01 2.21282393e-01 -1.35969967e-01 8.20852399e-01 -3.68938267e-01 -8.07137132e-01 9.79767069e-02 -8.30902934e-01 4.53860611e-01 -2.24356577e-01 -2.99782008e-01 -9.64562953e-01 -6.22708797e-01 -5.75551271e-01 -6.87084079e-01 1.13719165e+00 3.80832464e-01 1.27739990e+00 -1.65433317e-01 -3.19056690e-01 7.81111538e-01 1.79704607e+00 -3.19446236e-01 4.70247716e-01 1.71847090e-01 1.11087310e+00 3.47270519e-01 6.86215103e-01 -1.97801483e-03 5.69151938e-01 7.40170181e-01 4.68958974e-01 -5.57290971e-01 -3.98930103e-01 -4.86629933e-01 4.49806780e-01 5.42984605e-01 4.19894993e-01 -3.29474360e-02 -1.09098256e+00 1.00670254e+00 -1.79083872e+00 -9.72549140e-01 1.21611051e-01 1.96787703e+00 9.46550429e-01 3.36324447e-03 -1.08675860e-01 -5.46720505e-01 4.81084019e-01 2.94029742e-01 -5.93997717e-01 -8.48244205e-02 -3.15222055e-01 2.49464273e-01 7.10907519e-01 5.68608940e-01 -1.05798018e+00 1.46868491e+00 5.17971802e+00 9.46921945e-01 -1.25920725e+00 1.80351868e-01 7.49522269e-01 -3.76395285e-01 -4.19719309e-01 -7.80237606e-03 -8.70234132e-01 3.71387824e-02 4.76426095e-01 -5.75570352e-02 3.56505930e-01 8.56705964e-01 4.85656969e-02 -1.40265852e-01 -1.13874626e+00 1.58533692e+00 5.31234741e-01 -1.83416057e+00 5.91903687e-01 -1.67824551e-01 8.71030867e-01 5.58217049e-01 3.28962207e-01 2.56180614e-01 3.66065174e-01 -1.39613211e+00 1.01205730e+00 5.67702711e-01 1.12163651e+00 -7.23433316e-01 4.86785799e-01 1.76680565e-01 -1.45119691e+00 1.16128050e-01 -4.53134179e-01 4.27364618e-01 6.92026690e-02 2.93180674e-01 -7.79311001e-01 4.13691550e-01 9.63259280e-01 7.74887621e-01 -8.25860500e-01 7.46969581e-01 -3.97695377e-02 2.96660453e-01 -1.69643201e-02 1.29359975e-01 6.20360672e-01 1.86183259e-01 3.31063867e-01 1.31287813e+00 5.85738011e-02 -3.59659284e-01 3.90635341e-01 1.11431265e+00 -3.57320756e-01 8.84361938e-02 -7.04237401e-01 -1.61858723e-02 4.90909666e-01 1.08129323e+00 -5.21080196e-01 -5.95729768e-01 -6.48565948e-01 1.34967542e+00 5.62927783e-01 3.95974547e-01 -9.90268171e-01 -1.87061533e-01 7.31818974e-01 -1.37941048e-01 6.42480016e-01 -1.58295691e-01 -8.73203278e-02 -1.25778031e+00 1.31218567e-01 -8.89797926e-01 2.62313753e-01 -1.20199442e+00 -1.28507757e+00 4.91948724e-01 4.10990007e-02 -1.09993243e+00 -5.12200780e-02 -6.32020891e-01 -3.79802138e-01 9.98971283e-01 -1.65359139e+00 -1.74432123e+00 -6.03183687e-01 1.03056037e+00 9.72921371e-01 -3.07727575e-01 5.64007521e-01 -4.07168120e-02 -2.26505250e-01 8.30213428e-01 3.08088418e-02 5.26452899e-01 7.71977365e-01 -1.04254329e+00 5.88030338e-01 9.85000730e-01 9.46681678e-01 5.52381516e-01 4.27519172e-01 -3.94566894e-01 -1.41214108e+00 -1.44640529e+00 8.92718613e-01 -6.60193443e-01 5.62509656e-01 -4.92323041e-01 -8.28176022e-01 6.06568635e-01 4.47017074e-01 4.00207639e-01 3.41387749e-01 -2.94467211e-01 -1.02428007e+00 -2.12752104e-01 -6.96346998e-01 8.87474537e-01 9.80005264e-01 -1.16705287e+00 -7.74610460e-01 2.80244023e-01 9.66413140e-01 -4.44158882e-01 -5.75597644e-01 1.88809857e-01 5.14121830e-01 -6.31298542e-01 1.42990744e+00 -6.87981308e-01 5.92552841e-01 -5.08162379e-01 -5.92204869e-01 -9.47636545e-01 -1.48850068e-01 -1.75322309e-01 5.31549081e-02 1.06872952e+00 4.30621833e-01 -1.27904490e-01 4.92141098e-01 6.93647414e-02 7.54537806e-02 -8.06255996e-01 -7.41865933e-01 -5.80806255e-01 8.20411593e-02 -6.06705904e-01 1.58196628e-01 8.96300852e-01 -4.42641705e-01 5.23453414e-01 -3.26920271e-01 3.07683557e-01 7.87278533e-01 4.68988419e-01 9.49653447e-01 -7.51181901e-01 -3.14330697e-01 -5.75462818e-01 -3.69318873e-01 -1.22908854e+00 2.76235938e-01 -1.18884456e+00 -5.05991168e-02 -1.61735773e+00 6.47805631e-01 2.55609676e-02 -4.74652529e-01 6.42819822e-01 -1.74477443e-01 5.17558694e-01 4.34387475e-01 2.42448494e-01 -8.29774559e-01 5.40553153e-01 1.09946501e+00 -7.24467456e-01 1.58545487e-02 -6.66875362e-01 -7.19725311e-01 5.51276803e-01 6.37597620e-01 -2.48037249e-01 -7.01108694e-01 -7.44362056e-01 -8.75406489e-02 -1.86946228e-01 9.93466973e-01 -8.15865934e-01 2.20351145e-01 -2.21129119e-01 6.36627138e-01 -1.03639531e+00 3.28755468e-01 -6.94258928e-01 -2.78758138e-01 3.93530458e-01 -5.58621705e-01 2.89102644e-01 4.09674227e-01 7.85769582e-01 -3.41805130e-01 3.07247698e-01 8.25473964e-01 -1.83421239e-01 -1.22814178e+00 3.47923487e-01 1.67291149e-01 4.06014413e-01 9.21028316e-01 -1.31378740e-01 -4.16955590e-01 -3.06682229e-01 -3.41161996e-01 4.83838528e-01 6.19530916e-01 6.91991687e-01 1.07506824e+00 -1.28013456e+00 -7.03636527e-01 7.51511604e-02 7.83673108e-01 1.63394049e-01 4.00053918e-01 7.30255902e-01 -5.25420308e-01 6.00623906e-01 -2.40020156e-01 -1.10442567e+00 -1.70523560e+00 6.81931973e-01 3.56182009e-01 -1.81045040e-01 -7.26236403e-01 1.16039014e+00 6.66770697e-01 -3.36630680e-02 3.41570288e-01 -4.43027526e-01 6.26727939e-02 -1.33282170e-01 5.66857636e-01 -3.81257176e-01 -2.10668772e-01 -9.60027277e-01 -4.43402082e-01 7.86017776e-01 -4.51915264e-01 -3.29916179e-01 9.79257762e-01 -1.66658625e-01 -5.17943986e-02 3.66340905e-01 1.44870257e+00 -9.97130424e-02 -1.28991520e+00 -8.18824410e-01 -3.70699644e-01 -3.88612628e-01 2.48976126e-01 -7.58816481e-01 -1.19661570e+00 1.12024033e+00 7.81672418e-01 -6.66233957e-01 9.60555851e-01 4.24549729e-01 7.95029879e-01 4.47995186e-01 1.28654376e-01 -9.51804578e-01 4.58307981e-01 3.55470359e-01 1.13234770e+00 -1.57254922e+00 -1.04224190e-01 -3.03206854e-02 -9.69504952e-01 7.72616982e-01 8.47654939e-01 -5.37315868e-02 4.13252592e-01 -1.44174099e-01 1.22715890e-01 -2.94160724e-01 -8.12092185e-01 -5.21207452e-01 8.88020456e-01 4.72932756e-01 1.56655148e-01 8.07554647e-02 3.75451803e-01 2.47131810e-01 -1.48302719e-01 -2.87604272e-01 -2.99843214e-02 7.18970358e-01 -4.95267123e-01 -6.03614807e-01 -2.60470390e-01 3.03144157e-01 -1.11440532e-01 -6.73466742e-01 -3.62592667e-01 8.70878994e-01 -3.71376500e-02 6.67889297e-01 4.15703297e-01 -2.94385612e-01 6.62025586e-02 3.89564447e-02 4.99872595e-01 -4.33649868e-01 -3.51340830e-01 -6.55721352e-02 -2.34372035e-01 -6.76469803e-01 -3.71652961e-01 -2.81374484e-01 -1.44641685e+00 -1.19293161e-01 6.22865818e-02 -1.84114963e-01 4.63351727e-01 9.77955222e-01 4.37248796e-01 3.58703345e-01 2.10742250e-01 -6.14857793e-01 -2.55259424e-01 -6.68979287e-01 -1.19101159e-01 5.74347496e-01 5.53331196e-01 -4.33342487e-01 4.99789566e-02 3.47879142e-01]
[10.552867889404297, 1.5991672277450562]
d7f1deab-8a2e-49e6-9105-36c7691c9011
aco-tagger-a-novel-method-for-part-of-speech
2303.16760
null
https://arxiv.org/abs/2303.16760v1
https://arxiv.org/pdf/2303.16760v1.pdf
ACO-tagger: A Novel Method for Part-of-Speech Tagging using Ant Colony Optimization
Swarm Intelligence algorithms have gained significant attention in recent years as a means of solving complex and non-deterministic problems. These algorithms are inspired by the collective behavior of natural creatures, and they simulate this behavior to develop intelligent agents for computational tasks. One such algorithm is Ant Colony Optimization (ACO), which is inspired by the foraging behavior of ants and their pheromone laying mechanism. ACO is used for solving difficult problems that are discrete and combinatorial in nature. Part-of-Speech (POS) tagging is a fundamental task in natural language processing that aims to assign a part-of-speech role to each word in a sentence. In this research paper, proposed a high-performance POS-tagging method based on ACO called ACO-tagger. This method achieved a high accuracy rate of 96.867%, outperforming several state-of-the-art methods. The proposed method is fast and efficient, making it a viable option for practical applications.
['Mohammad bahrani', 'Sara Hajiaghajani', 'Amirhossein Mohammadi']
2023-03-27
null
null
null
null
['part-of-speech-tagging']
['natural-language-processing']
[ 1.49883389e-01 -2.56840706e-01 7.23658726e-02 1.58126447e-02 2.24969491e-01 -4.16237801e-01 5.70504785e-01 6.21991634e-01 -8.54788840e-01 8.22844684e-01 -1.38991028e-02 1.32408245e-02 -1.13899902e-01 -8.67050111e-01 5.56998327e-02 -9.30522382e-01 -2.22535014e-01 7.97960103e-01 5.09148538e-01 -4.37424034e-01 4.84272540e-01 4.14477050e-01 -1.68159020e+00 -8.16996545e-02 1.13737237e+00 5.93067884e-01 8.61739874e-01 4.41200137e-01 -5.55282891e-01 5.56729317e-01 -9.93441880e-01 -3.08885515e-01 -9.93577018e-02 -6.65104210e-01 -4.35426533e-01 -9.77731049e-02 -1.06668174e+00 9.69645143e-01 2.52220482e-01 1.43124545e+00 4.26435471e-01 3.84519696e-01 4.82298315e-01 -1.09495652e+00 -5.53363144e-01 5.00470877e-01 -5.28932571e-01 4.55030501e-01 1.91608846e-01 -2.64589071e-01 1.04453325e+00 -2.04233304e-01 3.29399973e-01 1.09249508e+00 4.00751263e-01 5.08522332e-01 -4.71216619e-01 -6.37501001e-01 9.03295428e-02 4.06500936e-01 -1.19729722e+00 1.27415568e-01 6.01130962e-01 -2.17990037e-02 1.03171003e+00 4.53134716e-01 1.06928635e+00 4.03166234e-01 7.11501658e-01 8.33036721e-01 1.12397015e+00 -5.84980071e-01 6.00361407e-01 -1.05229050e-01 -2.00542390e-01 6.52177572e-01 7.32165575e-01 -2.59449571e-01 -2.87750453e-01 -4.34678853e-01 2.78096408e-01 8.27686787e-02 8.72064307e-02 1.73510775e-01 -1.27039969e+00 9.88782346e-01 5.12737036e-01 8.43783677e-01 -6.54366553e-01 -2.97869205e-01 6.56288564e-02 -3.58411878e-01 1.63823128e-01 8.80838633e-01 -4.47183043e-01 -4.92490292e-01 -1.20822959e-01 -7.08444715e-02 9.66182232e-01 2.95198709e-01 3.25838953e-01 -1.05133429e-01 9.47710648e-02 1.01084256e+00 5.76079309e-01 6.71626925e-01 1.08879411e+00 -4.46743101e-01 1.43724784e-01 1.04248774e+00 5.84554113e-02 -1.11922097e+00 -6.94862008e-01 -5.74651897e-01 -9.32660341e-01 -3.94819155e-02 -1.04359500e-01 -2.66201079e-01 -6.81975663e-01 1.30744636e+00 5.49552381e-01 -1.04602054e-02 3.35562587e-01 7.98476458e-01 5.70223570e-01 9.19501722e-01 3.42436373e-01 -5.61406732e-01 1.92002714e+00 -1.08153749e+00 -8.56732309e-01 -5.39606512e-01 1.97934330e-01 -9.55477715e-01 6.88389897e-01 4.26283240e-01 -5.77319145e-01 -8.07060897e-02 -9.45830822e-01 7.79883921e-01 -5.20382285e-01 -6.90089241e-02 1.00523877e+00 6.41880155e-01 -6.39951289e-01 2.94772476e-01 -9.36018407e-01 -6.31657660e-01 8.58417153e-02 5.31169653e-01 -3.49227339e-02 4.70147312e-01 -9.73120511e-01 7.14896977e-01 4.84110624e-01 7.10184500e-02 3.55281159e-02 3.90013397e-01 -5.15663445e-01 1.37412369e-01 4.82996643e-01 -4.51117396e-01 1.20711207e+00 -1.08701384e+00 -1.60230863e+00 6.68927848e-01 -2.30160892e-01 -4.24739778e-01 -1.21844850e-01 2.72988617e-01 -6.97750688e-01 -7.47530460e-02 2.56443083e-01 2.07029179e-01 5.78169405e-01 -9.21293437e-01 -9.81981337e-01 -3.81768107e-01 -3.35778207e-01 4.92648661e-01 -4.65390563e-01 4.95734245e-01 -2.08367467e-01 -7.96448767e-01 2.64207721e-01 -1.13372600e+00 -5.58868706e-01 -8.07868600e-01 -7.20708817e-02 -5.56209326e-01 4.78699535e-01 -2.60761350e-01 1.32283962e+00 -1.81415045e+00 -5.01831770e-02 4.45741899e-02 -1.94052055e-01 8.66082251e-01 -1.37665600e-01 6.27295017e-01 3.49333763e-01 1.23384982e-01 -4.91105795e-01 4.04096618e-02 -2.15518937e-01 4.54562098e-01 4.24840331e-01 1.35051653e-01 8.71875510e-03 5.05630553e-01 -1.20823145e+00 -6.53242826e-01 -3.67567018e-02 1.88628078e-01 -2.66270429e-01 2.93074369e-01 -3.37109804e-01 2.20407397e-01 -8.86499345e-01 7.86723375e-01 4.85929251e-01 -2.12240413e-01 4.72276866e-01 4.20764148e-01 -3.74145031e-01 2.66686641e-02 -1.18690431e+00 1.18446553e+00 -2.67552197e-01 3.43314528e-01 1.70635208e-01 -9.22214806e-01 1.20591819e+00 1.18164830e-01 6.30311430e-01 -7.60524631e-01 4.96180236e-01 3.90919119e-01 6.14268780e-01 -6.36806488e-01 4.85396862e-01 -1.05400234e-02 -5.42661920e-02 5.67529380e-01 -5.41992426e-01 -1.43881857e-01 6.76018178e-01 -2.17795432e-01 1.39424121e+00 -3.77619505e-01 8.72014999e-01 -2.09341973e-01 8.60818803e-01 3.99523288e-01 1.04875839e+00 4.42226499e-01 -2.01397285e-01 2.03194410e-01 3.33921351e-02 -5.33617973e-01 -5.01087546e-01 -5.37505567e-01 1.80458695e-01 7.83707619e-01 3.89876962e-01 -2.38048196e-01 -7.73808479e-01 -4.39380854e-01 -2.82971144e-01 5.69149017e-01 -4.53921705e-01 -2.58990116e-02 -6.41189575e-01 -1.29327679e+00 2.27183446e-01 -4.86995131e-02 9.78945315e-01 -1.84443915e+00 -9.15497780e-01 6.55219972e-01 -1.90624967e-01 -1.08553421e+00 -1.35845691e-01 2.66024917e-01 -6.57022178e-01 -1.14048874e+00 -4.75277513e-01 -1.22938478e+00 8.64279449e-01 4.55719471e-01 8.35323274e-01 5.79798222e-01 -9.26288366e-02 -2.64903992e-01 -1.13666415e+00 -8.73496175e-01 -6.32676780e-01 3.70934337e-01 9.52072367e-02 2.73014635e-01 4.80666548e-01 -3.08677882e-01 -4.22554016e-01 4.28789258e-01 -9.20271218e-01 -1.89603180e-01 1.10389864e+00 7.64828086e-01 6.07566059e-01 6.83964849e-01 2.18883932e-01 -6.80154622e-01 1.15422177e+00 -4.15561587e-01 -6.23269439e-01 4.64936644e-01 -5.75532913e-01 5.95133975e-02 9.88635778e-01 -3.71015519e-01 -9.10166025e-01 3.11006188e-01 -1.31893784e-01 3.75170290e-01 -1.31988779e-01 5.11649728e-01 -9.02287513e-02 -1.23850636e-01 1.65957615e-01 3.60616505e-01 -1.67544447e-02 -4.83882397e-01 -4.14725393e-01 1.14665675e+00 2.86603332e-01 -3.14179718e-01 3.98730963e-01 3.56067240e-01 1.23863682e-01 -9.30442989e-01 -6.51542366e-01 -7.59014606e-01 -1.20333508e-01 -7.22626969e-02 1.15269434e+00 -3.66101444e-01 -1.00740528e+00 5.53139448e-01 -1.14737046e+00 1.02955297e-01 9.93741676e-02 5.87272644e-01 1.07619181e-01 3.12449723e-01 -1.15230553e-01 -1.24722326e+00 -6.19863391e-01 -9.46409702e-01 5.90054750e-01 9.91367579e-01 5.44613861e-02 -8.77158344e-01 2.31607586e-01 3.10290396e-01 4.03713882e-01 4.20729481e-02 5.92522979e-01 -1.03129852e+00 -1.37302354e-01 -2.16460839e-01 1.95144758e-01 -9.54938084e-02 3.52420956e-01 -3.75382975e-02 -3.46416086e-01 -5.44100106e-02 2.31287837e-01 3.91356051e-01 2.97773927e-01 1.70113876e-01 6.85444713e-01 -3.76514822e-01 -5.17095208e-01 1.34735361e-01 1.20898104e+00 1.08888674e+00 3.94969523e-01 1.07253480e+00 1.62197515e-01 4.59612280e-01 1.09743881e+00 5.94336689e-01 3.85003209e-01 6.47348583e-01 6.48415506e-01 4.17195261e-02 2.23650545e-01 -2.05585361e-02 1.39556840e-01 1.20154679e+00 -2.34751880e-01 -6.71092987e-01 -1.19336331e+00 3.27822775e-01 -2.06613922e+00 -1.03590965e+00 -3.65462542e-01 1.79322946e+00 5.63269317e-01 1.69579446e-01 6.02299049e-02 2.32472539e-01 1.10802948e+00 5.32776378e-02 -1.62757277e-01 -6.67222559e-01 -2.65868723e-01 2.50299841e-01 4.54725027e-01 2.21455753e-01 -1.04609501e+00 1.05803180e+00 5.04834652e+00 8.49766731e-01 -1.03408241e+00 1.28357545e-01 2.75094151e-01 5.89824855e-01 1.41958699e-01 -4.22403403e-02 -5.74382663e-01 7.99108386e-01 6.77134931e-01 -1.47554964e-01 5.97287297e-01 5.44939280e-01 3.08829993e-01 -5.49988449e-01 -5.76186739e-02 8.63747060e-01 1.04732990e-01 -1.09372830e+00 -2.78619468e-01 1.15926005e-02 8.47557366e-01 -4.27921256e-03 -3.14721137e-01 -1.35848582e-01 5.02612233e-01 -8.81384313e-01 3.56517971e-01 1.57962263e-01 -1.01056315e-01 -7.36728787e-01 1.22400045e+00 7.08336711e-01 -1.40776145e+00 -3.85988116e-01 -4.58418190e-01 -4.47117269e-01 2.04102576e-01 6.26077771e-01 -9.05820251e-01 3.60553473e-01 9.49745238e-01 3.70715916e-01 -3.44529390e-01 1.55531144e+00 -5.10114074e-01 5.32990873e-01 -3.37487608e-01 -1.36011958e+00 3.83824915e-01 -5.57177186e-01 6.93146884e-01 9.04398143e-01 4.02314901e-01 4.45239216e-01 1.85191497e-01 2.30629519e-01 1.82020113e-01 3.38711113e-01 -9.54215303e-02 -4.21489924e-01 8.70425045e-01 1.49361992e+00 -1.61281824e+00 -9.96499509e-02 7.93721154e-03 6.92649901e-01 1.76103376e-02 -2.72974789e-01 -7.43450403e-01 -8.35965633e-01 4.76586610e-01 -2.40994170e-01 3.63463610e-01 -3.33220243e-01 -2.86460429e-01 -7.59326577e-01 -1.48156747e-01 -8.84766400e-01 4.98313278e-01 -6.44009531e-01 -9.15715277e-01 8.93032551e-01 -3.80762160e-01 -1.28912139e+00 -9.40447077e-02 -4.39820439e-01 -7.25071251e-01 3.71433318e-01 -1.41924334e+00 -5.68671525e-01 -4.56367165e-01 2.43168041e-01 7.40451813e-01 -5.99011421e-01 9.90493238e-01 -2.37885803e-01 -6.65795863e-01 4.85081933e-02 3.13643336e-01 1.48451969e-01 1.93797410e-01 -1.01050818e+00 3.26035082e-01 9.98304546e-01 4.70042974e-01 5.55914104e-01 8.00648689e-01 -7.70728886e-01 -1.52154946e+00 -6.01435840e-01 1.25985241e+00 1.48014277e-01 6.06829703e-01 -1.05497450e-01 -3.53695124e-01 -1.29446715e-01 2.72883236e-01 -2.63531566e-01 6.35289550e-01 -3.23868752e-01 6.05922341e-01 -1.53961644e-01 -1.24541545e+00 5.11547327e-01 8.53462279e-01 3.93433094e-01 -7.02721894e-01 4.51064259e-01 4.16759819e-01 -2.75378644e-01 -4.04062212e-01 1.31231278e-01 1.27229556e-01 -1.00350404e+00 6.33315325e-01 -1.25620842e-01 -2.48939991e-01 -6.23081505e-01 2.39068508e-01 -1.55887270e+00 -5.16371906e-01 -8.97102773e-01 2.49665618e-01 1.15860605e+00 4.24852610e-01 -1.06932974e+00 8.88923109e-01 8.80932584e-02 -1.29811466e-01 -6.41641617e-01 -8.57753396e-01 -9.10814404e-01 -5.86550772e-01 9.14514661e-02 9.33735728e-01 8.03642154e-01 -4.92797047e-03 4.53265339e-01 2.43537650e-01 -6.56091981e-03 4.10108060e-01 4.20968682e-02 3.10373396e-01 -1.49599540e+00 -2.69740671e-01 -5.87804139e-01 -4.69417840e-01 -5.75949371e-01 1.06770493e-01 -6.05279267e-01 3.44071627e-01 -1.69510281e+00 1.01573817e-01 -8.28846812e-01 -2.30309933e-01 6.04754686e-01 -2.62919754e-01 2.35029593e-01 2.56631076e-01 5.05548954e-01 -7.36581445e-01 5.19020319e-01 1.24700093e+00 -1.66411951e-01 -4.27612633e-01 3.56818646e-01 -6.86162353e-01 7.98907101e-01 1.28206563e+00 -9.55122054e-01 -1.33881539e-01 -3.96704048e-01 3.59189600e-01 -2.08580419e-01 -4.60036308e-01 -1.12184620e+00 5.60691237e-01 -4.96585220e-01 -1.82424024e-01 -1.14262938e-01 2.67839134e-01 -9.95593131e-01 1.49006248e-01 1.15175450e+00 2.22196966e-01 7.03534067e-01 -1.04337268e-01 5.03794730e-01 -3.86355221e-01 -6.86990380e-01 5.29061198e-01 -3.49654913e-01 -9.57441568e-01 -1.17896497e-01 -5.71881294e-01 1.43054575e-01 1.50413036e+00 -1.18871421e-01 -4.27460581e-01 -4.88835648e-02 -2.57985443e-01 5.43576963e-02 4.38226163e-01 5.13121724e-01 4.51914012e-01 -9.50988948e-01 -5.30143201e-01 1.97024364e-02 -8.19440931e-02 -1.80365831e-01 -3.73125970e-01 6.59393847e-01 -1.08330965e+00 4.33819830e-01 -2.88599819e-01 -3.20104092e-01 -1.55512142e+00 2.03620061e-01 -7.80814365e-02 -4.96611685e-01 -1.93066850e-01 7.89832532e-01 -5.46647251e-01 -3.27432692e-01 -4.73458320e-02 6.68742228e-03 -8.34566712e-01 -1.97218508e-01 3.12070936e-01 2.65967250e-01 1.23347297e-01 -7.94463575e-01 -9.06820953e-01 6.95214152e-01 3.34430039e-01 -2.08743080e-01 1.50591469e+00 2.16315910e-01 -6.25430524e-01 8.41145664e-02 6.50807500e-01 3.47758353e-01 -3.82418931e-01 2.09871948e-01 2.25621402e-01 -3.45739692e-01 -2.04219773e-01 -1.05114555e+00 -7.89604247e-01 3.82508546e-01 2.58811712e-01 8.11942399e-01 1.14779568e+00 -1.54104412e-01 9.79526639e-01 5.46871066e-01 6.31582916e-01 -1.31959283e+00 -9.31842998e-03 8.21562946e-01 4.41278160e-01 -1.24210477e+00 1.06780417e-02 -4.37436163e-01 -7.91998982e-01 1.00310659e+00 5.12431920e-01 1.42012462e-02 5.14548242e-01 2.46771619e-01 1.57733619e-01 -5.88028925e-03 -5.98149836e-01 -4.64232326e-01 1.38994735e-02 7.75447071e-01 3.71512383e-01 2.84533978e-01 -1.00871789e+00 5.78370869e-01 -3.92685384e-01 -3.86923820e-01 2.92184740e-01 1.29398859e+00 -8.33765745e-01 -1.55905890e+00 -6.07372642e-01 3.32152933e-01 -5.25558531e-01 1.01025179e-01 -6.72357440e-01 5.06494105e-01 3.40513229e-01 1.56112587e+00 1.72683094e-02 -5.75573683e-01 7.04453140e-02 -5.85221469e-01 5.04294336e-02 -6.50420487e-01 -9.67006326e-01 -1.42332554e-01 4.25437018e-02 -1.12828173e-01 -7.42514312e-01 -9.37146902e-01 -1.90594280e+00 -9.20834094e-02 -3.09170216e-01 1.20921552e+00 1.02543128e+00 8.63488555e-01 4.61888909e-01 4.57818985e-01 8.48376811e-01 -5.24779737e-01 -2.78167754e-01 -7.98413455e-01 -5.39260447e-01 2.55544275e-01 -4.02451813e-01 -7.47489631e-01 -2.50296921e-01 -4.41130489e-01]
[5.701745510101318, 3.4989864826202393]
c3ae2641-ef4c-4c54-bb80-dc1b123992d0
image-cropping-on-twitter-fairness-metrics
2105.08667
null
https://arxiv.org/abs/2105.08667v2
https://arxiv.org/pdf/2105.08667v2.pdf
Image Cropping on Twitter: Fairness Metrics, their Limitations, and the Importance of Representation, Design, and Agency
Twitter uses machine learning to crop images, where crops are centered around the part predicted to be the most salient. In fall 2020, Twitter users raised concerns that the automated image cropping system on Twitter favored light-skinned over dark-skinned individuals, as well as concerns that the system favored cropping woman's bodies instead of their heads. In order to address these concerns, we conduct an extensive analysis using formalized group fairness metrics. We find systematic disparities in cropping and identify contributing factors, including the fact that the cropping based on the single most salient point can amplify the disparities because of an effect we term argmax bias. However, we demonstrate that formalized fairness metrics and quantitative analysis on their own are insufficient for capturing the risk of representational harm in automatic cropping. We suggest the removal of saliency-based cropping in favor of a solution that better preserves user agency. For developing a new solution that sufficiently address concerns related to representational harm, our critique motivates a combination of quantitative and qualitative methods that include human-centered design.
['Shubhanshu Mishra', 'Uthaipon Tantipongpipat', 'Kyra Yee']
2021-05-18
null
null
null
null
['image-cropping']
['computer-vision']
[ 2.57201403e-01 7.69213676e-01 -5.40871441e-01 -4.08377826e-01 -5.06762445e-01 -5.46677351e-01 7.22179174e-01 6.31871045e-01 -5.40970862e-01 5.52639484e-01 1.04661322e+00 -4.23932016e-01 2.73957461e-01 -5.75051367e-01 -4.88940448e-01 -6.51258007e-02 5.23146868e-01 -3.12769949e-01 -3.08921248e-01 -2.70549119e-01 7.65731394e-01 1.69278756e-01 -1.47779679e+00 3.15019518e-01 1.13013875e+00 4.50693369e-01 -4.29799706e-01 1.52499244e-01 -1.48531288e-01 1.17432213e+00 -6.60165250e-01 -8.01161289e-01 3.79256904e-01 -6.69784427e-01 -5.43078601e-01 -9.96260047e-02 8.51118982e-01 -7.13249505e-01 4.28161621e-02 1.16953051e+00 4.80320007e-01 -1.70887634e-01 7.14528978e-01 -1.61337113e+00 -9.66647267e-01 6.97297335e-01 -1.20513058e+00 9.34778973e-02 3.55109453e-01 4.11960065e-01 9.97980475e-01 -7.35314190e-01 7.73128450e-01 1.52878892e+00 6.57968104e-01 5.77204883e-01 -1.10244584e+00 -9.85307932e-01 2.09044978e-01 -3.33645046e-01 -1.19066489e+00 -7.82771051e-01 5.38856745e-01 -7.65880108e-01 1.89968958e-01 6.68025076e-01 7.79796004e-01 7.05664456e-01 1.18955687e-01 4.41034704e-01 1.34588683e+00 -5.19498646e-01 2.50016153e-01 4.54705030e-01 7.70609602e-02 6.96403742e-01 1.07377231e+00 -1.62292540e-01 -5.75508714e-01 -6.64378822e-01 3.82973492e-01 -1.48065463e-01 1.93564981e-01 -1.99978441e-01 -1.12125170e+00 1.20428097e+00 5.05303919e-01 1.20924458e-01 -4.18568134e-01 1.43252164e-01 3.13428968e-01 -4.19616282e-01 8.97452474e-01 8.78867865e-01 2.62992412e-01 -6.90049827e-02 -1.27693450e+00 5.93427718e-01 3.97377908e-01 5.54752588e-01 8.44833970e-01 -1.16844401e-01 -6.05114996e-01 4.75163758e-01 3.52784783e-01 4.68803227e-01 6.35451078e-02 -1.44950914e+00 3.06929111e-01 8.16222131e-01 4.52530116e-01 -1.47326422e+00 -4.00202096e-01 -1.64701343e-01 -2.59965569e-01 5.54984689e-01 4.97643590e-01 -6.61724687e-01 -6.52391076e-01 1.65089118e+00 8.37385505e-02 -6.10782146e-01 -5.19282520e-01 1.18208122e+00 6.21299803e-01 2.33573660e-01 6.86642647e-01 -8.05437099e-03 1.46128321e+00 -5.73130488e-01 -9.55240667e-01 -4.43302542e-01 7.23614037e-01 -5.28074443e-01 1.20389152e+00 -1.34294435e-01 -1.25946689e+00 6.10655844e-02 -1.08632767e+00 -2.95789480e-01 -3.07071030e-01 -2.47363612e-01 5.92617154e-01 1.13289261e+00 -1.01295650e+00 3.36653173e-01 -1.57948643e-01 -6.76902115e-01 8.63026440e-01 -1.74677655e-01 -1.00007147e-01 2.84760594e-01 -1.06229889e+00 1.11383033e+00 -4.23003286e-01 -2.13990822e-01 -3.78878236e-01 -6.79931939e-01 -9.34234321e-01 3.49928141e-02 1.09221809e-01 -6.12688303e-01 1.10353947e+00 -1.72608507e+00 -5.43172896e-01 1.11581147e+00 -1.46312624e-01 -5.56219637e-01 8.81176233e-01 -1.84398308e-01 8.42753053e-03 1.62325263e-01 6.95520580e-01 1.10864592e+00 8.83768082e-01 -1.41746020e+00 -7.27730393e-01 -3.74735028e-01 1.91488102e-01 5.66855073e-01 -6.85224175e-01 4.96515960e-01 4.18085843e-01 -5.38369358e-01 -4.70017701e-01 -7.63168395e-01 -2.63038397e-01 1.60616890e-01 -4.59656835e-01 1.92287073e-01 6.48061991e-01 -7.93179095e-01 1.38280988e+00 -2.08862424e+00 -8.12566936e-01 1.29406512e-01 6.42437160e-01 -2.58363169e-02 2.36550868e-01 1.40489459e-01 4.99793477e-02 9.90037978e-01 8.40925649e-02 -1.65583029e-01 7.35136643e-02 -5.36900699e-01 -1.94102138e-01 8.26009035e-01 2.94367403e-01 7.11620033e-01 -9.72501397e-01 -8.97597671e-01 3.39496508e-02 2.87051499e-01 -6.26924872e-01 -4.58804697e-01 2.12627202e-01 -1.99468657e-02 -1.95747867e-01 7.17883050e-01 8.46221805e-01 -4.50377092e-02 3.15420032e-02 -1.36090428e-01 -6.04071498e-01 1.64018646e-01 -3.95630211e-01 1.05557716e+00 1.10019274e-01 8.22191775e-01 1.58488512e-01 -2.66980350e-01 6.92430317e-01 -1.42012358e-01 4.99452353e-01 -8.26041818e-01 9.59938616e-02 -5.53379674e-03 -5.76469712e-02 -3.56143266e-01 1.17923164e+00 -4.55517501e-01 -2.87636757e-01 7.48782158e-01 -7.76321530e-01 -2.04708308e-01 -2.70295650e-01 5.65448642e-01 7.93822885e-01 2.72498443e-03 2.00595945e-01 -7.07589149e-01 -2.92788327e-01 3.64749163e-01 4.87005264e-01 7.78245091e-01 -9.89173114e-01 7.76735842e-01 7.88078666e-01 -2.07295567e-01 -1.12600374e+00 -5.33408940e-01 2.22977683e-01 1.40592957e+00 3.61861318e-01 -3.34083050e-01 -1.26502562e+00 -5.98402083e-01 4.41752560e-02 1.27742195e+00 -8.89784336e-01 -1.00977160e-01 -5.02534956e-02 -6.32787108e-01 6.54283583e-01 -1.11845070e-02 2.76198626e-01 -7.85389721e-01 -1.51893401e+00 -3.00070822e-01 -4.41721559e-01 -7.20934153e-01 -5.93790233e-01 -4.73894775e-01 -3.97175342e-01 -1.03791356e+00 -1.12353444e+00 -2.35531315e-01 8.77580047e-01 6.39541388e-01 8.47367704e-01 2.03887179e-01 -1.06209569e-01 5.07353783e-01 -2.38962978e-01 -9.29135501e-01 -2.70911247e-01 -2.00177401e-01 -2.62779981e-01 -1.66703701e-01 4.31015134e-01 4.18370403e-02 -8.01952779e-01 -7.65736327e-02 -7.38826573e-01 3.85388583e-01 2.26068914e-01 2.11188421e-01 -1.34608656e-01 -3.59679788e-01 7.34530747e-01 -1.07259703e+00 8.95832896e-01 -6.47491038e-01 -8.73696730e-02 1.04489133e-01 -7.74868846e-01 -4.40726012e-01 -1.19592488e-01 -1.46522447e-01 -1.20551562e+00 -2.49525219e-01 6.99087858e-01 1.14123479e-01 2.44086117e-01 2.40459904e-01 2.35239729e-01 8.66647437e-03 1.21784079e+00 -6.79862320e-01 3.76662999e-01 7.95159042e-02 4.89887714e-01 5.34659624e-01 1.22988939e-01 -4.63290274e-01 6.70183420e-01 8.04562151e-01 -4.62459326e-01 -5.78834772e-01 -7.84204066e-01 5.14842616e-03 6.64863437e-02 -7.00502157e-01 8.62567484e-01 -1.13006532e+00 -5.60169101e-01 2.97253951e-02 -1.01950800e+00 -2.43354678e-01 -4.06804800e-01 1.46142766e-01 -3.19893658e-01 1.88279659e-01 -3.75200123e-01 -1.11887932e+00 -4.16894346e-01 -7.94577479e-01 7.16538250e-01 4.21971112e-01 -9.60818052e-01 -3.74767870e-01 -1.53047860e-01 7.12872386e-01 6.36548460e-01 7.49236822e-01 6.63637102e-01 -3.78787160e-01 -5.55863120e-02 -1.81603208e-01 -6.86140239e-01 -1.65131360e-01 2.90107787e-01 3.47813129e-01 -1.00029135e+00 -6.70562685e-02 -2.46403068e-01 -2.41525114e-01 4.62358177e-01 5.91611683e-01 7.48691738e-01 -6.24329388e-01 -2.29358912e-01 -1.92661196e-01 1.30665159e+00 -1.18323013e-01 6.05235279e-01 5.15440404e-01 6.57640815e-01 1.24101174e+00 7.42175102e-01 7.72296965e-01 7.59133875e-01 1.36549503e-01 3.99892151e-01 -5.99015594e-01 5.52955978e-02 -5.97687960e-01 4.14122671e-01 -3.38793039e-01 2.43711382e-01 1.50050446e-01 -1.09542251e+00 7.50029504e-01 -1.68125856e+00 -1.08913112e+00 -1.19104117e-01 2.21102762e+00 5.12576103e-01 2.59466559e-01 6.56942904e-01 -2.16152400e-01 1.31106424e+00 4.40304220e-01 -4.39209640e-01 -7.96273947e-01 -1.40645817e-01 -2.40770251e-01 9.44970548e-01 6.14231884e-01 -8.64985585e-01 7.80560434e-01 7.43495703e+00 3.06919694e-01 -1.00963902e+00 2.05439627e-01 1.46083140e+00 -2.65045315e-01 -9.46687520e-01 1.92284212e-01 -3.21680933e-01 4.97470617e-01 7.07084298e-01 -6.81007326e-01 8.67101923e-02 6.57512009e-01 9.22098219e-01 -4.14673805e-01 -5.58555126e-01 6.37027502e-01 2.30669960e-01 -1.31595993e+00 -7.57661238e-02 2.86055624e-01 8.20510745e-01 -4.43001717e-01 4.64525074e-01 -1.04857041e-02 3.49905461e-01 -1.29712331e+00 1.34388506e+00 3.18551332e-01 8.05089653e-01 -6.82435632e-01 3.35134000e-01 -1.24003002e-02 -5.30839741e-01 6.37132581e-03 -6.25527650e-02 -6.70221329e-01 -3.11727449e-02 7.17393935e-01 -8.35056365e-01 -1.13485210e-01 4.36843306e-01 2.75506467e-01 -8.51812363e-01 7.84307539e-01 1.18650988e-01 3.97313774e-01 2.09461823e-01 -2.86427904e-02 2.05263600e-01 3.37271541e-02 3.77797097e-01 1.31478083e+00 2.66816258e-01 -5.31239584e-02 -3.22770804e-01 1.02179623e+00 -1.17086232e-01 3.67693722e-01 -8.22419584e-01 -2.16479778e-01 6.93592668e-01 1.23621285e+00 -8.53716433e-01 -3.64542842e-01 -5.20683587e-01 3.99914891e-01 4.70788628e-02 2.60401517e-01 -8.92689586e-01 -7.30572268e-02 5.20101190e-01 8.02337050e-01 -5.13132453e-01 3.13965678e-01 -1.03162229e+00 -7.47448206e-01 -2.65712798e-01 -1.06182444e+00 1.39538929e-01 -9.93929446e-01 -8.37827742e-01 -9.12578627e-02 1.39361709e-01 -8.95007968e-01 1.38271749e-01 7.89358914e-02 -5.31657517e-01 8.83536339e-01 -1.17573714e+00 -9.90458548e-01 -3.34614486e-01 7.49756321e-02 3.59625854e-02 4.67171401e-01 2.54432172e-01 1.23437367e-01 -2.59968400e-01 5.91181159e-01 -6.39927387e-01 -1.57414854e-01 1.14833856e+00 -1.06133986e+00 2.81507045e-01 9.03752983e-01 -6.58635020e-01 5.42851090e-01 1.14378309e+00 -8.54230165e-01 -7.29772389e-01 -8.91388774e-01 1.11020851e+00 -4.66050625e-01 1.85081199e-01 2.39538830e-02 -4.37091708e-01 5.28288960e-01 4.61989224e-01 -4.31024462e-01 8.11193705e-01 -5.76855876e-02 -2.79434979e-01 2.37606943e-01 -1.72349596e+00 1.06713080e+00 7.45925605e-01 -3.94282877e-01 -4.34180051e-01 1.39555782e-01 6.24984801e-01 -3.24500017e-02 -2.73263961e-01 2.77685989e-02 7.19081104e-01 -1.20657516e+00 5.37645102e-01 -2.96832293e-01 7.49415338e-01 -5.43645620e-02 9.71770212e-02 -1.01273739e+00 -4.53383237e-01 -7.14897156e-01 5.57028532e-01 1.33340192e+00 5.35579026e-01 -4.88772362e-01 9.40030515e-01 1.61491978e+00 1.33961380e-01 -2.55381018e-01 -4.41262275e-01 -7.64444098e-02 2.62476832e-01 -6.79712370e-02 5.40253162e-01 1.41307127e+00 4.78686243e-01 2.58179065e-02 -5.40303051e-01 -1.83733404e-01 6.31556690e-01 -3.24453920e-01 7.09881663e-01 -9.73580778e-01 5.94680429e-01 -5.51665545e-01 -1.62356481e-01 3.86001281e-02 -1.07917622e-01 -2.77633935e-01 5.59338406e-02 -1.57546425e+00 7.78857768e-01 -2.50227273e-01 -5.94688579e-02 5.89361608e-01 -3.45459193e-01 2.72293329e-01 6.52308404e-01 1.82021990e-01 -3.86238962e-01 4.25580367e-02 1.20367646e+00 -9.05234888e-02 -2.36298323e-01 -5.76855838e-01 -2.01445127e+00 8.05713177e-01 7.60520160e-01 -3.68119001e-01 -2.99765408e-01 -2.35452279e-01 6.01438582e-01 -3.93282175e-01 6.34543419e-01 -6.70874476e-01 -5.75013198e-02 -6.32549524e-01 3.80142450e-01 -2.22455189e-01 1.00521304e-01 -5.73777616e-01 1.02945194e-01 6.22170687e-01 -8.17295730e-01 7.86150061e-03 2.14405343e-01 1.52274698e-01 1.78838640e-01 -2.60471731e-01 7.56498814e-01 -2.52104580e-01 -2.88602918e-01 -1.97881311e-01 -6.15962207e-01 1.15045957e-01 1.05308867e+00 -3.75942349e-01 -7.86726594e-01 -9.65939820e-01 -1.22628674e-01 1.32597938e-01 1.15362942e+00 4.17929024e-01 2.05023512e-01 -1.17560565e+00 -8.29930127e-01 -3.46625149e-01 1.55702710e-01 -5.51368773e-01 7.49394968e-02 7.32460976e-01 -5.64432144e-01 4.93649021e-02 -4.64750856e-01 1.46416917e-01 -1.15593767e+00 4.18000698e-01 2.07280666e-01 2.94239432e-01 -3.10760707e-01 4.43580508e-01 5.11209488e-01 7.40348995e-02 4.93477546e-02 -2.99038216e-02 -2.78282702e-01 4.83115822e-01 7.15856254e-01 7.29688048e-01 -4.30281639e-01 -9.29395020e-01 -5.25074124e-01 -4.89753559e-02 1.64063394e-01 -5.89203000e-01 8.40344727e-01 -4.99547452e-01 -7.59278312e-02 1.78691328e-01 8.39764297e-01 5.75246274e-01 -1.22559476e+00 3.55718315e-01 -9.19413865e-02 -9.94525373e-01 2.14957204e-02 -8.54447067e-01 -6.15615249e-01 4.85409349e-01 5.99485576e-01 5.04227519e-01 8.56961846e-01 -4.16657835e-01 3.48247200e-01 -3.30652982e-01 9.03979093e-02 -1.39209473e+00 -8.74850377e-02 -2.91329384e-01 9.40386891e-01 -1.22778058e+00 5.13494670e-01 -4.23720390e-01 -1.18540275e+00 5.80803752e-01 6.53865218e-01 -5.24021983e-02 2.90786237e-01 -1.17746778e-02 1.54441386e-01 -3.24206263e-01 -3.67675960e-01 -1.03981709e-02 -1.03301466e-01 5.58477461e-01 8.25388610e-01 5.06538391e-01 -9.64005649e-01 3.95734221e-01 -4.16119546e-01 4.77590933e-02 9.40261066e-01 9.79237378e-01 -6.69283688e-01 -4.08148974e-01 -8.93686593e-01 7.65242577e-01 -7.43772209e-01 -4.60721739e-02 -9.06662583e-01 7.47823179e-01 2.84136981e-01 1.28493321e+00 2.37849221e-01 -2.00129971e-01 1.10774942e-01 -1.29896820e-01 8.55235085e-02 -5.49489021e-01 -8.55492473e-01 -1.60461918e-01 5.61737359e-01 -4.44220603e-01 -6.22198641e-01 -8.63723993e-01 -1.03222847e+00 -6.98625147e-01 -1.77406505e-01 -4.98535559e-02 6.10158980e-01 7.37389982e-01 5.91468036e-01 -5.61946258e-02 5.37008464e-01 -6.72380984e-01 -1.67972952e-01 -6.75893068e-01 -3.68704796e-01 6.19622707e-01 4.15211558e-01 -4.36667651e-01 -3.30668181e-01 -8.72813985e-02]
[12.678815841674805, 1.344490885734558]
36ee1943-fb0a-4eae-855d-13c5798a291d
hierarchical-latent-structure-for-multi-modal
2207.04624
null
https://arxiv.org/abs/2207.04624v1
https://arxiv.org/pdf/2207.04624v1.pdf
Hierarchical Latent Structure for Multi-Modal Vehicle Trajectory Forecasting
Variational autoencoder (VAE) has widely been utilized for modeling data distributions because it is theoretically elegant, easy to train, and has nice manifold representations. However, when applied to image reconstruction and synthesis tasks, VAE shows the limitation that the generated sample tends to be blurry. We observe that a similar problem, in which the generated trajectory is located between adjacent lanes, often arises in VAE-based trajectory forecasting models. To mitigate this problem, we introduce a hierarchical latent structure into the VAE-based forecasting model. Based on the assumption that the trajectory distribution can be approximated as a mixture of simple distributions (or modes), the low-level latent variable is employed to model each mode of the mixture and the high-level latent variable is employed to represent the weights for the modes. To model each mode accurately, we condition the low-level latent variable using two lane-level context vectors computed in novel ways, one corresponds to vehicle-lane interaction and the other to vehicle-vehicle interaction. The context vectors are also used to model the weights via the proposed mode selection network. To evaluate our forecasting model, we use two large-scale real-world datasets. Experimental results show that our model is not only capable of generating clear multi-modal trajectory distributions but also outperforms the state-of-the-art (SOTA) models in terms of prediction accuracy. Our code is available at https://github.com/d1024choi/HLSTrajForecast.
['Kyoungwook Min', 'Dooseop Choi']
2022-07-11
null
null
null
null
['trajectory-forecasting']
['computer-vision']
[-2.78299928e-01 -1.78484693e-01 -3.30615968e-01 -2.29196936e-01 -5.28599262e-01 -2.32229397e-01 8.57339323e-01 -5.20057440e-01 2.32327208e-01 6.27018929e-01 4.09541398e-01 -4.63231802e-01 1.40916288e-01 -8.78794134e-01 -8.87952030e-01 -1.12714028e+00 7.49745741e-02 4.45233166e-01 2.80407183e-02 -1.22685425e-01 -1.61445484e-01 3.37846667e-01 -1.64394331e+00 2.15958387e-01 9.92119074e-01 7.40934193e-01 4.95007634e-01 4.47391689e-01 -3.11566085e-01 9.10806417e-01 -3.22313488e-01 -2.48216137e-01 -1.91986479e-03 -4.12206262e-01 -1.89531013e-01 2.12675586e-01 1.49191037e-01 -4.25231725e-01 -7.22180247e-01 9.12015736e-01 4.06606570e-02 4.46123451e-01 1.11880159e+00 -1.54717374e+00 -7.43701637e-01 3.31008315e-01 -5.74493110e-01 -8.09487924e-02 -1.13172196e-01 1.97287902e-01 7.25424111e-01 -9.68685865e-01 6.37444139e-01 1.38346314e+00 4.06995475e-01 6.41480744e-01 -1.24238455e+00 -7.54187226e-01 2.66118735e-01 4.34329420e-01 -1.46412325e+00 -4.33241606e-01 9.63284552e-01 -7.74857342e-01 5.79804182e-01 1.18840471e-01 5.08547723e-01 1.42577958e+00 4.19607222e-01 9.50471640e-01 7.72757113e-01 -5.20674400e-02 1.32869154e-01 3.35570246e-01 1.21712135e-02 5.37154496e-01 -3.90748173e-01 2.81112820e-01 -4.63735089e-02 -1.63807392e-01 8.09398353e-01 2.41996169e-01 -2.25777328e-01 -3.61051202e-01 -1.06675756e+00 1.06717610e+00 3.45099360e-01 1.10168405e-01 -5.02558589e-01 2.73932993e-01 -2.39292197e-02 -1.31713793e-01 6.34351432e-01 -3.51614863e-01 -2.79941801e-02 -3.65432091e-02 -1.02866256e+00 4.12791371e-01 4.20973837e-01 9.60585773e-01 8.88958514e-01 4.43305612e-01 -4.03763652e-01 1.02337193e+00 6.53271616e-01 6.26499116e-01 3.13899606e-01 -1.06781435e+00 3.75311762e-01 1.78655356e-01 3.30540031e-01 -1.07954836e+00 -6.69919923e-02 -3.20048153e-01 -1.15401328e+00 5.84275238e-02 2.85199404e-01 -2.76215732e-01 -1.12649548e+00 1.88142121e+00 2.18704447e-01 7.15145826e-01 -1.25350608e-02 9.76491868e-01 6.83723867e-01 1.46310866e+00 1.45023301e-01 -1.57009006e-01 1.04007876e+00 -1.15200937e+00 -9.08070087e-01 1.15128011e-01 2.65357167e-01 -7.21478403e-01 8.30237091e-01 2.56470740e-02 -8.34393442e-01 -7.20588267e-01 -7.63901353e-01 8.94390270e-02 -3.14802021e-01 3.36111724e-01 4.57863837e-01 1.96475804e-01 -9.54976618e-01 2.77302802e-01 -8.77235353e-01 -1.08514100e-01 1.80294812e-01 -1.97283477e-01 -4.30022478e-02 -8.28892663e-02 -1.30454588e+00 8.30381393e-01 7.08943382e-02 2.17318147e-01 -1.15848422e+00 -6.50695026e-01 -1.01113248e+00 1.61078155e-01 9.80578945e-04 -7.26615727e-01 9.22298849e-01 -7.58438408e-01 -1.50692213e+00 1.05899610e-01 -6.41379774e-01 -2.44595200e-01 4.14542347e-01 1.96951881e-01 -7.20625937e-01 -2.03159619e-02 4.44377773e-02 9.11768973e-01 1.23946607e+00 -1.68490565e+00 -7.09622443e-01 1.19488336e-01 -2.28374287e-01 6.95514679e-02 -1.01705939e-01 -3.57161641e-01 -5.74608803e-01 -7.48333693e-01 -8.47204477e-02 -1.08633447e+00 -1.11015409e-01 -1.23044945e-01 -3.88559222e-01 -2.52564877e-01 1.08410561e+00 -7.72947848e-01 1.42514634e+00 -2.30499864e+00 1.90379292e-01 1.22329086e-01 2.45394468e-01 -9.32543445e-03 -1.34827316e-01 5.23357928e-01 -9.81701463e-02 -8.65311623e-02 -3.17596525e-01 -6.10635221e-01 1.74103335e-01 4.26518172e-01 -7.01904595e-01 3.90016407e-01 9.10053998e-02 7.94867635e-01 -8.39124680e-01 -2.69712657e-01 6.60922468e-01 8.12826991e-01 -3.90125632e-01 2.30425969e-01 -2.09115490e-01 5.31823158e-01 -3.08526278e-01 3.90415281e-01 9.54896629e-01 -1.56951651e-01 -2.56681796e-02 -2.30715781e-01 -1.63488805e-01 -8.09305012e-02 -1.17085397e+00 1.13861227e+00 -5.12605608e-01 9.07018125e-01 -1.20228589e-01 -7.53793538e-01 7.63151169e-01 4.06003028e-01 4.83699322e-01 -3.10263604e-01 8.67283121e-02 -1.33104056e-01 -2.07501784e-01 -5.14723659e-01 6.39418066e-01 -1.56024709e-01 2.87619755e-02 1.38808519e-01 8.25185180e-02 1.00548871e-01 1.07542969e-01 2.10592985e-01 3.77185404e-01 2.15047106e-01 -3.63277167e-01 -1.08349770e-01 5.16689301e-01 1.57236196e-02 6.40245438e-01 3.63473892e-01 -9.51566473e-02 6.08886003e-01 3.74728054e-01 -3.65029901e-01 -1.22462773e+00 -1.37908101e+00 -1.55549660e-01 6.22874022e-01 2.75237888e-01 -1.15947209e-01 -7.49436617e-01 -2.51231760e-01 -1.17813721e-02 1.37130415e+00 -6.41580641e-01 -1.81381151e-01 -3.63220662e-01 -6.70658529e-01 2.64282763e-01 4.14789200e-01 3.06019396e-01 -8.49962234e-01 -4.46223430e-02 2.30982020e-01 -4.92992818e-01 -9.85356927e-01 -6.04220033e-01 -5.13982594e-01 -6.10179901e-01 -7.23748803e-01 -8.85192454e-01 -5.97808003e-01 6.66116059e-01 3.62076163e-01 9.36495841e-01 -2.13571161e-01 2.48399943e-01 2.43230104e-01 -1.35068849e-01 -1.68833479e-01 -5.85286856e-01 -2.04972997e-01 1.75299227e-01 4.45202708e-01 3.94575089e-01 -5.00094056e-01 -5.44509888e-01 2.97419280e-01 -9.23293293e-01 3.03327590e-01 3.02017599e-01 9.15197730e-01 6.02149189e-01 1.95202097e-01 4.64772791e-01 -5.54891646e-01 5.20410359e-01 -9.95462239e-01 -5.51963210e-01 2.68943701e-02 -3.45585495e-01 -1.13978602e-01 6.66031420e-01 -6.80491209e-01 -1.25637901e+00 -9.53321159e-02 -2.93520600e-01 -1.08157361e+00 -3.81417304e-01 5.44201136e-01 5.17424271e-02 5.12718260e-01 1.17818922e-01 5.53484559e-01 1.71607643e-01 -3.28875542e-01 5.26966095e-01 6.29107296e-01 3.99526745e-01 -4.09822792e-01 9.26657081e-01 4.79773730e-01 -2.17942476e-01 -1.13659024e+00 -3.95224839e-01 -2.46221408e-01 -3.32630962e-01 -5.31212628e-01 1.04065824e+00 -1.12639511e+00 -4.62561220e-01 5.98036110e-01 -1.08792174e+00 -5.18940926e-01 -1.10800102e-01 5.92550159e-01 -5.70306897e-01 1.04557514e-01 -6.60656571e-01 -1.02479768e+00 1.50918201e-01 -1.40038085e+00 9.78591800e-01 2.23056421e-01 6.70549944e-02 -1.21290302e+00 -7.62055104e-04 1.92590639e-01 3.95661265e-01 2.60562330e-01 1.08991921e+00 9.73529369e-02 -7.56920576e-01 -3.77525240e-02 -8.84457976e-02 2.81565249e-01 5.43906502e-02 3.91366959e-01 -9.21490133e-01 -1.40815958e-01 -1.92650259e-01 2.01010436e-01 9.95983660e-01 7.73981392e-01 1.24290562e+00 -2.16605797e-01 -4.39428836e-01 6.55449748e-01 1.25654709e+00 1.68339804e-01 8.15397680e-01 -3.88750620e-02 9.99750078e-01 6.19664192e-01 3.42266023e-01 2.84910798e-01 8.72866988e-01 7.27879941e-01 2.80789256e-01 -2.31360905e-02 -1.97110400e-01 -5.44968247e-01 4.78200585e-01 1.10610807e+00 2.82299127e-02 -4.40429688e-01 -6.86096787e-01 7.55843341e-01 -2.08104253e+00 -1.26959801e+00 -3.01826090e-01 1.96983230e+00 3.36824983e-01 -1.50793612e-01 1.81840941e-01 -3.48290831e-01 6.59692705e-01 4.40983981e-01 -4.16554689e-01 -2.96587825e-01 2.21863315e-02 -4.93292153e-01 3.06823909e-01 6.46712244e-01 -1.10051525e+00 9.92403805e-01 6.00358677e+00 1.20382071e+00 -1.17710972e+00 2.09707424e-01 5.57027817e-01 3.29180956e-02 -6.31892204e-01 -1.37115583e-01 -7.08684742e-01 9.26368058e-01 1.06150913e+00 7.89008141e-02 5.42389393e-01 6.84589446e-01 5.28920889e-01 1.03156842e-01 -7.44148135e-01 8.75930309e-01 -2.17525214e-01 -1.47297573e+00 2.62013197e-01 1.87480509e-01 8.26323807e-01 6.15732046e-03 3.86174172e-01 6.13806486e-01 2.26560354e-01 -9.80161607e-01 9.58571494e-01 9.30153728e-01 8.01629722e-01 -9.62277889e-01 5.26826441e-01 7.15921164e-01 -1.34757292e+00 -7.12925941e-03 -5.21728754e-01 1.01698130e-01 5.10589540e-01 3.85625392e-01 -4.82123971e-01 4.19279903e-01 5.31340182e-01 8.68204653e-01 -7.67846927e-02 7.25786626e-01 -1.12944484e-01 8.96620512e-01 -1.59327522e-01 2.13527605e-01 4.79936749e-01 -7.19285429e-01 7.08620667e-01 1.13914549e+00 7.25621402e-01 -6.71210513e-02 1.65247992e-01 1.31013560e+00 2.29805291e-01 -3.05382818e-01 -6.96127415e-01 1.03594184e-01 5.16275823e-01 1.22560334e+00 -2.42440268e-01 -4.71848339e-01 -4.83224005e-01 8.87815893e-01 1.02445215e-01 9.15583372e-01 -1.26595914e+00 1.50991045e-02 1.00329494e+00 7.28342012e-02 5.25010943e-01 -4.64977890e-01 4.72131371e-02 -1.29935086e+00 -2.09973052e-01 -4.10886526e-01 9.59340110e-03 -8.92183006e-01 -1.32800746e+00 7.21931517e-01 3.04177791e-01 -1.63349152e+00 -6.49449706e-01 -3.92997712e-01 -8.12530935e-01 1.02535737e+00 -1.58315444e+00 -1.35454369e+00 -1.41093716e-01 5.69701135e-01 7.12170780e-01 -2.83733249e-01 6.94876850e-01 5.28626382e-01 -8.37355256e-01 5.41837335e-01 4.81243372e-01 1.90895214e-03 3.29178244e-01 -1.09396839e+00 2.44640499e-01 8.91327977e-01 -1.96013059e-02 5.08793533e-01 9.21404660e-01 -5.24660885e-01 -1.28297544e+00 -1.45072782e+00 7.91067362e-01 -4.87836689e-01 6.41626477e-01 -3.81681859e-01 -9.56087410e-01 7.76886463e-01 1.94587827e-01 -1.84424534e-01 5.16223013e-01 -2.71817476e-01 -5.27730845e-02 4.64593917e-02 -9.35037553e-01 7.22070038e-01 5.18239319e-01 -4.64846551e-01 -3.40476394e-01 1.09129712e-01 6.07019365e-01 -4.00264680e-01 -7.70293593e-01 1.28461450e-01 6.02571189e-01 -8.08164835e-01 1.00170374e+00 -1.95803791e-01 7.08743870e-01 -5.12698114e-01 -1.96606174e-01 -1.62801051e+00 -6.48494601e-01 -2.15698466e-01 -6.37436509e-01 1.21517825e+00 3.72764200e-01 -6.45599186e-01 6.02151036e-01 5.50455272e-01 -2.36620292e-01 -8.95962358e-01 -8.13716590e-01 -6.47609353e-01 2.46226564e-01 -5.31451046e-01 8.05913985e-01 8.46873760e-01 -4.83439386e-01 2.44489178e-01 -8.60292196e-01 4.65824604e-01 7.05231071e-01 2.14534581e-01 7.68511534e-01 -8.63668978e-01 -2.06722513e-01 -5.44288456e-01 -7.64838979e-02 -1.42495143e+00 3.66731673e-01 -6.84979677e-01 1.53047204e-01 -1.54105210e+00 5.37788868e-02 -3.64416480e-01 -1.62920401e-01 -1.83478873e-02 -1.64287806e-01 8.05393532e-02 3.71637791e-01 3.72239918e-01 -1.10159442e-01 1.04744589e+00 1.30547559e+00 -2.46416628e-01 -1.98610425e-01 1.84251666e-01 -3.00518215e-01 6.46636248e-01 6.48097336e-01 -3.06703031e-01 -6.22545302e-01 -3.97931129e-01 -3.05967271e-01 3.16796482e-01 4.56024557e-01 -7.95555592e-01 1.03545696e-01 -3.48017812e-01 3.95921350e-01 -1.00920308e+00 5.89382589e-01 -8.70373905e-01 5.19014716e-01 5.57461902e-02 -2.54267361e-02 -1.65458262e-01 1.31922647e-01 6.74808383e-01 -3.58275622e-01 2.20643561e-02 5.80450535e-01 9.42074955e-02 -8.49821806e-01 7.35119164e-01 -8.96749258e-01 -4.18349326e-01 9.26941931e-01 -1.73911244e-01 -1.74876288e-01 -8.31361055e-01 -6.24152064e-01 5.01050353e-01 4.23772871e-01 6.48639441e-01 7.07510591e-01 -1.76384366e+00 -7.26681828e-01 3.14517587e-01 1.40355542e-01 -1.03798769e-01 7.88632691e-01 7.62190759e-01 -2.99491435e-01 3.38900000e-01 -1.16864138e-03 -7.82471478e-01 -5.98511457e-01 6.43255472e-01 3.15541118e-01 -6.40071332e-02 -6.07530892e-01 5.95135927e-01 5.42488754e-01 -4.90609199e-01 1.01412751e-01 -2.70277679e-01 -4.05701786e-01 4.84796427e-02 5.80241203e-01 4.81027931e-01 -5.73359847e-01 -1.25545466e+00 -9.13496017e-02 4.86877799e-01 2.53605515e-01 -9.23819691e-02 1.16639745e+00 -4.52496707e-01 1.80955231e-02 7.14312375e-01 1.18752086e+00 -2.21116304e-01 -1.66284502e+00 -5.10349944e-02 -5.10806441e-01 -4.04072881e-01 3.02521318e-01 -3.57840896e-01 -1.13823032e+00 1.04381192e+00 6.87075257e-01 2.61409670e-01 8.43847156e-01 -1.66936085e-01 1.13965201e+00 -1.73942775e-01 2.44390607e-01 -9.52973783e-01 -3.06233913e-01 5.65519750e-01 9.15396273e-01 -1.16136992e+00 -3.92013431e-01 -2.50555515e-01 -9.84065831e-01 9.12463248e-01 5.46319842e-01 -2.94357806e-01 9.25173879e-01 -4.80335616e-02 1.89809188e-01 -2.91769014e-04 -8.45275402e-01 -5.28805926e-02 5.77491462e-01 5.73468626e-01 1.65030912e-01 4.93745387e-01 1.52303472e-01 5.02173126e-01 -1.02543563e-01 -1.46576494e-01 5.10983229e-01 2.91640997e-01 -1.45952359e-01 -8.64661336e-01 -3.85193586e-01 4.32367176e-01 1.11205159e-02 7.42133632e-02 3.87200862e-01 6.07416093e-01 1.72154397e-01 1.01260436e+00 4.40277487e-01 -5.71238875e-01 7.22897751e-03 1.18987449e-01 6.95818141e-02 -1.85505986e-01 1.31579503e-01 4.02913779e-01 -1.91613883e-01 -4.34245974e-01 -3.14740121e-01 -7.18121886e-01 -1.13193429e+00 -6.31184459e-01 -8.98344964e-02 4.34323885e-02 5.18050075e-01 8.52715373e-01 4.53620076e-01 7.53877044e-01 7.71432042e-01 -1.24081635e+00 -3.65810424e-01 -1.06830537e+00 -7.06756294e-01 4.11455303e-01 6.92173302e-01 -1.09259665e+00 -3.93785626e-01 1.56873822e-01]
[6.544538974761963, 1.0588490962982178]
eaa66539-ffb0-4814-97e3-0ea91ebad00e
source-free-domain-adaptation-via-1
2101.10842
null
https://arxiv.org/abs/2101.10842v1
https://arxiv.org/pdf/2101.10842v1.pdf
Source-free Domain Adaptation via Distributional Alignment by Matching Batch Normalization Statistics
In this paper, we propose a novel domain adaptation method for the source-free setting. In this setting, we cannot access source data during adaptation, while unlabeled target data and a model pretrained with source data are given. Due to lack of source data, we cannot directly match the data distributions between domains unlike typical domain adaptation algorithms. To cope with this problem, we propose utilizing batch normalization statistics stored in the pretrained model to approximate the distribution of unobserved source data. Specifically, we fix the classifier part of the model during adaptation and only fine-tune the remaining feature encoder part so that batch normalization statistics of the features extracted by the encoder match those stored in the fixed classifier. Additionally, we also maximize the mutual information between the features and the classifier's outputs to further boost the classification performance. Experimental results with several benchmark datasets show that our method achieves competitive performance with state-of-the-art domain adaptation methods even though it does not require access to source data.
['Masashi Sugiyama', 'Masato Ishii']
2021-01-19
source-free-domain-adaptation-via
https://openreview.net/forum?id=HWqv5Pm3E3
https://openreview.net/pdf?id=HWqv5Pm3E3
null
['source-free-domain-adaptation']
['computer-vision']
[ 2.80196100e-01 3.84364463e-03 -6.04349315e-01 -7.65170932e-01 -7.30464101e-01 -6.94638252e-01 4.86060739e-01 9.65723321e-02 -5.55010736e-01 8.87722254e-01 6.04763627e-02 9.93406326e-02 1.54504091e-01 -8.12191129e-01 -8.51719916e-01 -6.84602976e-01 3.55228215e-01 5.72871506e-01 1.86169192e-01 -3.72292362e-02 -1.32351769e-02 1.60790324e-01 -1.23588097e+00 1.59827143e-01 8.41617048e-01 1.09628260e+00 2.40504846e-01 2.16931641e-01 -2.83015251e-01 5.99000275e-01 -6.14139020e-01 -4.39359933e-01 3.69057328e-01 -5.90430200e-01 -5.10291755e-01 3.14554155e-01 1.12101786e-01 -2.68825680e-01 -4.81404334e-01 1.21767652e+00 1.63936853e-01 2.91123748e-01 8.31968367e-01 -1.34372795e+00 -9.77444828e-01 5.43958187e-01 -4.29005355e-01 2.63945878e-01 1.00748926e-01 -2.08744779e-01 7.54191160e-01 -9.32471395e-01 6.02591932e-01 9.19800103e-01 1.98425755e-01 7.58368492e-01 -1.32204127e+00 -8.69828105e-01 4.28323179e-01 1.39763355e-01 -1.40643370e+00 -5.83580613e-01 7.87885010e-01 -3.07954937e-01 5.39354980e-01 -3.93889338e-01 2.01379493e-01 1.34488440e+00 -1.83851153e-01 6.09531164e-01 7.33592093e-01 -5.26894748e-01 5.55142999e-01 6.91430986e-01 9.21896175e-02 4.40248251e-01 9.75513756e-02 -3.45966220e-02 -6.22858942e-01 -3.17836195e-01 7.03644156e-01 1.64317891e-01 -1.37515947e-01 -8.87722731e-01 -1.17706287e+00 8.63535225e-01 2.16201648e-01 1.79409042e-01 -3.73892337e-01 -3.44295830e-01 1.53266326e-01 5.96037030e-01 4.49248910e-01 3.66252512e-02 -8.71850669e-01 1.12403519e-01 -6.51904166e-01 -7.46282041e-02 7.67985523e-01 1.39615703e+00 1.09316194e+00 -2.06025735e-01 -4.67627943e-02 9.63496506e-01 1.45165727e-01 6.35400355e-01 8.41228485e-01 -7.20867276e-01 7.98242390e-01 6.57419145e-01 1.88764811e-01 -4.76692200e-01 1.41081167e-02 -3.17851841e-01 -7.72346616e-01 -3.00952911e-01 6.82363510e-01 -2.77421236e-01 -1.03933966e+00 2.04631352e+00 4.51669991e-01 2.85623610e-01 3.55014563e-01 7.30507612e-01 2.73138911e-01 5.83934546e-01 1.23033404e-01 -2.03237310e-01 8.84642839e-01 -8.54464352e-01 -5.44914186e-01 -6.05542243e-01 5.26007235e-01 -4.51475352e-01 9.64584410e-01 8.30873102e-02 -6.30514205e-01 -5.99279761e-01 -1.21056187e+00 2.06135049e-01 -4.53739107e-01 3.51201028e-01 3.03897262e-01 5.08612335e-01 -3.47413272e-01 3.10621530e-01 -9.40442681e-01 -4.10251051e-01 5.06271958e-01 3.98806274e-01 -5.18995106e-01 -4.18300301e-01 -1.22429132e+00 6.74116910e-01 6.38434291e-01 -5.11772752e-01 -8.00571501e-01 -7.73611069e-01 -9.42683101e-01 1.25599533e-01 4.53346997e-01 -4.91016716e-01 1.40202570e+00 -1.33071959e+00 -1.59490490e+00 6.60082817e-01 -4.31802422e-01 -3.38906765e-01 2.91268080e-01 -7.07171485e-02 -6.41443551e-01 -2.66157955e-01 2.49400884e-01 4.72699612e-01 9.17533934e-01 -1.11095560e+00 -8.90435457e-01 -3.32115442e-01 -1.70039043e-01 1.07793622e-01 -8.29726160e-01 -1.67857349e-01 -6.32020175e-01 -4.71813232e-01 4.60003354e-02 -7.41593182e-01 -2.22097948e-01 -1.35031482e-02 -2.15379313e-01 -6.41790256e-02 7.31365681e-01 -2.79639453e-01 1.11176836e+00 -2.50849509e+00 8.45482666e-03 4.68020380e-01 -1.02279931e-01 2.40429297e-01 -3.09716821e-01 2.26637781e-01 -1.08901106e-01 -3.02301943e-01 -5.36783934e-01 -3.13103467e-01 -4.61597554e-02 4.29083347e-01 -4.68315601e-01 5.17655015e-01 2.80859560e-01 5.27726293e-01 -9.62485373e-01 -3.74724269e-01 3.77923064e-02 2.83990920e-01 -6.21562064e-01 5.94828427e-01 -2.03596592e-01 5.93597233e-01 -6.58749044e-01 3.38616222e-01 7.73189306e-01 -4.62838143e-01 3.80658925e-01 1.63715884e-01 4.27117527e-01 4.00352061e-01 -1.40588939e+00 1.88942695e+00 -5.40750086e-01 1.89947978e-01 -1.71321198e-01 -1.41288257e+00 1.06793094e+00 2.94757038e-01 5.08212507e-01 -6.95565760e-01 -1.60362180e-02 2.58719683e-01 -7.88054913e-02 -1.77784488e-01 6.04910590e-02 -1.17594145e-01 -2.43640050e-01 4.48444813e-01 5.96775115e-01 3.16763818e-01 1.23864308e-01 7.29243457e-03 8.68535399e-01 1.59649938e-01 5.60218811e-01 1.44213483e-01 6.37268245e-01 9.16380882e-02 8.72084737e-01 6.70364141e-01 -8.27774853e-02 4.65728611e-01 3.37578952e-01 -2.09114805e-01 -1.06682825e+00 -1.17081738e+00 -8.33348408e-02 1.39122295e+00 9.21043828e-02 -2.40274921e-01 -7.50922203e-01 -1.15583360e+00 1.10127248e-01 6.46830320e-01 -7.59934843e-01 -4.93803859e-01 -4.24608797e-01 -7.56305695e-01 1.85535237e-01 7.25472629e-01 5.45286834e-01 -6.78861201e-01 -6.12045601e-02 3.91016066e-01 -1.22464843e-01 -1.26644886e+00 -6.29358768e-01 4.30138052e-01 -9.01520312e-01 -9.76264477e-01 -5.35048366e-01 -9.18994665e-01 9.24464345e-01 5.32979295e-02 8.88458729e-01 -4.50306147e-01 2.57831991e-01 1.45151652e-03 -3.45083416e-01 -3.76887023e-01 -4.88275915e-01 5.02239645e-01 1.65291399e-01 2.75586843e-01 9.18996751e-01 -5.42139113e-01 -1.20545849e-01 3.84527981e-01 -1.04282570e+00 -4.60913539e-01 6.29185915e-01 9.63785887e-01 6.90728486e-01 1.21876694e-01 8.93044949e-01 -1.21090949e+00 2.86177278e-01 -9.41325545e-01 -6.83394551e-01 3.72587562e-01 -6.39677107e-01 2.74219662e-01 9.55876470e-01 -9.17085111e-01 -1.39443493e+00 4.47790146e-01 2.55841523e-01 -4.73765016e-01 -3.41828555e-01 4.85257924e-01 -6.43146336e-01 3.31719071e-01 8.07985127e-01 2.60501176e-01 -1.54057786e-01 -6.66955173e-01 2.65705675e-01 9.95506167e-01 5.90851486e-01 -4.85508651e-01 1.03284276e+00 3.24537247e-01 -4.33906853e-01 -3.95082742e-01 -1.22049034e+00 -5.64988554e-01 -9.63541567e-01 5.21756470e-01 4.74507868e-01 -1.11607039e+00 9.25476197e-03 3.59170854e-01 -8.88629138e-01 -3.71980309e-01 -4.86205578e-01 7.20812440e-01 -5.71536660e-01 9.66528133e-02 -1.04527101e-01 -5.75922608e-01 1.60856977e-01 -8.23380589e-01 7.81095088e-01 3.28011304e-01 5.72813973e-02 -1.14127934e+00 1.96471199e-01 -9.94532853e-02 3.72973293e-01 -1.67663395e-01 9.34090853e-01 -1.33385777e+00 -2.52500653e-01 -3.09721589e-01 -1.34095788e-01 5.05574286e-01 6.37009382e-01 -4.62852627e-01 -9.60838020e-01 -2.48682633e-01 3.59528442e-03 -2.59031892e-01 7.03639209e-01 1.23720296e-01 1.14262533e+00 -3.55088860e-01 -5.39397299e-01 6.54497623e-01 1.23534501e+00 1.72579810e-01 2.70565629e-01 3.28844488e-01 5.15646935e-01 4.35751021e-01 7.48979151e-01 6.61413908e-01 3.85665238e-01 6.67039454e-01 -5.63509129e-02 3.78811099e-02 -1.00262454e-02 -6.72270477e-01 5.09346187e-01 4.84034687e-01 5.05876541e-01 -2.76220709e-01 -6.74685240e-01 7.31764436e-01 -1.74438441e+00 -7.06891835e-01 5.02633035e-01 2.55585527e+00 1.11434770e+00 9.49729457e-02 1.15700200e-01 -1.78515270e-01 8.71105850e-01 -2.11429149e-01 -1.00765634e+00 5.08005843e-02 1.61783304e-02 2.56457686e-01 7.12096393e-01 3.20998460e-01 -1.26659358e+00 9.95135128e-01 6.07770348e+00 6.18196905e-01 -1.01090980e+00 1.96792617e-01 3.62446278e-01 -1.59731433e-01 -8.08947310e-02 -1.24989357e-02 -1.05598152e+00 6.37822032e-01 1.17700326e+00 -3.95299762e-01 4.90998626e-01 1.11013186e+00 -2.60075599e-01 1.20718069e-01 -1.48881507e+00 6.98004305e-01 1.14716021e-02 -7.62960374e-01 -1.17701739e-01 1.05562218e-01 7.90571392e-01 5.74759506e-02 8.88974071e-02 5.50901413e-01 5.87148130e-01 -5.74165940e-01 5.84541321e-01 3.37442189e-01 8.46760333e-01 -7.93285131e-01 6.58222556e-01 5.82873642e-01 -1.03038299e+00 -2.73119032e-01 -7.61819899e-01 1.84144109e-01 -9.99535173e-02 6.89568520e-01 -8.67223501e-01 3.14359933e-01 4.78919476e-01 8.97982538e-01 -4.31506157e-01 9.37764883e-01 -3.11911136e-01 7.04394162e-01 -4.45343554e-01 3.38700354e-01 -1.69431895e-01 -1.16490172e-02 2.72458196e-01 1.01122487e+00 3.85415465e-01 2.05965135e-02 4.56018865e-01 7.45919406e-01 -3.87470603e-01 1.90549240e-01 -6.94927692e-01 -1.25243619e-01 8.73425722e-01 9.31205332e-01 -2.49119118e-01 -5.57654262e-01 -7.76798248e-01 1.22531211e+00 5.47741413e-01 6.31420553e-01 -7.54653156e-01 -2.97418356e-01 9.22717094e-01 3.24356183e-02 5.30189753e-01 -3.87699045e-02 -1.16885692e-01 -1.56850839e+00 1.72736034e-01 -6.95696354e-01 6.46578074e-01 -2.23511651e-01 -1.73841786e+00 5.02443254e-01 1.07644849e-01 -1.34129274e+00 -5.14221787e-01 -4.90351409e-01 -2.39968002e-01 1.06620407e+00 -1.71567297e+00 -9.95525777e-01 -7.58304521e-02 9.24227178e-01 4.63350087e-01 -3.88682902e-01 1.02473271e+00 2.16943234e-01 -6.40451729e-01 1.02452028e+00 6.06407642e-01 4.09790814e-01 1.23226118e+00 -1.03954720e+00 4.28294539e-01 7.33630598e-01 1.43963367e-01 7.11402595e-01 3.30605567e-01 -6.57478631e-01 -1.09666920e+00 -1.30118883e+00 8.39471519e-01 -4.09280956e-01 5.39586902e-01 -5.08500636e-01 -1.18440211e+00 9.21917737e-01 -3.15159187e-02 3.85001004e-01 1.00876141e+00 9.77127701e-02 -6.99745119e-01 -2.40215510e-01 -1.33023000e+00 9.04191881e-02 8.32465410e-01 -4.59996790e-01 -6.75360203e-01 2.28945851e-01 5.22473872e-01 -3.65442038e-01 -8.74662519e-01 6.24686442e-02 3.26751709e-01 -3.89604360e-01 8.47694099e-01 -1.05017364e+00 1.69349998e-01 -1.90955192e-01 -4.25763100e-01 -1.68618762e+00 -3.16925615e-01 -1.92081511e-01 -3.06114972e-01 1.52057135e+00 6.92602038e-01 -8.06113064e-01 8.82244647e-01 8.93551767e-01 2.11340159e-01 -1.55965075e-01 -9.14188445e-01 -1.00780010e+00 2.36513615e-01 -8.59930739e-02 9.25299406e-01 1.04702294e+00 1.14387549e-01 3.36975396e-01 -1.90808907e-01 3.55416149e-01 5.92866242e-01 1.26498193e-01 7.72840619e-01 -1.31347024e+00 -4.17300493e-01 6.40440360e-02 -1.57219261e-01 -1.16031182e+00 5.91929257e-01 -8.54213059e-01 9.80643481e-02 -1.01580179e+00 3.44480306e-01 -6.29452705e-01 -6.20447278e-01 7.10185111e-01 -3.55765790e-01 -8.00588354e-02 -8.87106657e-02 3.51834536e-01 -4.96693969e-01 5.61588168e-01 8.22201669e-01 -2.54977103e-02 -3.91223341e-01 1.62533596e-01 -9.73769844e-01 5.26453495e-01 8.13642323e-01 -9.66774046e-01 -6.38914883e-01 -4.59657699e-01 -2.43418053e-01 -1.52194396e-01 8.67695287e-02 -9.36053038e-01 2.56008506e-01 -5.23502171e-01 8.16564500e-01 -2.23614573e-02 2.01372504e-01 -1.14391315e+00 -1.94714636e-01 -7.61031955e-02 -5.73049545e-01 -3.09906691e-01 -5.55156767e-02 7.84546375e-01 -3.05131316e-01 -2.87306070e-01 1.02328634e+00 1.17631130e-01 -6.19085550e-01 4.22697425e-01 -2.91957349e-01 2.14036077e-01 1.03754115e+00 1.18361458e-01 -5.90679720e-02 -3.51371020e-01 -6.65655613e-01 1.84159219e-01 7.12728441e-01 5.23478329e-01 3.54012042e-01 -1.50681567e+00 -5.59508502e-01 6.63725019e-01 5.77273428e-01 5.21883667e-02 -7.44042844e-02 3.05035859e-01 4.57904637e-01 4.24971461e-01 -1.65956065e-01 -2.79415786e-01 -6.96088791e-01 9.91195917e-01 2.47926399e-01 -2.66680807e-01 -2.60565192e-01 5.66714585e-01 4.54708457e-01 -7.80329823e-01 1.98957145e-01 -1.15057468e-01 -8.28646198e-02 -2.06002649e-02 6.23710334e-01 3.89489830e-02 1.87738851e-01 -4.83199447e-01 -4.71220762e-01 2.81280935e-01 -4.28706437e-01 -2.45404318e-01 1.31265092e+00 -2.72126794e-01 4.25374597e-01 4.57217336e-01 1.41282785e+00 -1.20019011e-01 -1.53756201e+00 -1.04585207e+00 6.15513101e-02 -5.69389105e-01 -1.73428446e-01 -8.45265269e-01 -1.14849687e+00 7.43908703e-01 5.72027385e-01 -2.60337353e-01 1.32726526e+00 1.05742298e-01 6.49197459e-01 5.94929278e-01 3.49706054e-01 -1.30045211e+00 -1.67342037e-01 5.17124951e-01 2.08471596e-01 -1.40536118e+00 -2.94031650e-01 -3.33420068e-01 -8.12877595e-01 8.61372888e-01 8.04695070e-01 -1.67939991e-01 7.10476041e-01 1.33558363e-01 1.19379863e-01 3.08839172e-01 -8.11668754e-01 -2.51899779e-01 1.29821047e-01 9.40130889e-01 3.80345702e-01 -1.02345794e-01 9.19686705e-02 8.66319001e-01 1.62772257e-02 2.35063240e-01 4.10184264e-01 9.46114421e-01 -3.82482499e-01 -1.65212691e+00 -2.16365084e-01 3.98752987e-01 -3.96701217e-01 -4.74605476e-04 -3.38258237e-01 6.26307905e-01 4.75508124e-02 9.61800694e-01 1.45087004e-01 -2.59003222e-01 4.97587800e-01 4.29757565e-01 3.04450870e-01 -9.09314454e-01 -1.50092214e-01 -5.68423010e-02 -2.42295519e-01 -3.89230371e-01 -2.40871906e-01 -7.86960900e-01 -1.15599203e+00 1.24834247e-01 -2.20982388e-01 4.58173186e-01 6.80073082e-01 1.01862156e+00 4.48339164e-01 2.24010289e-01 9.00143683e-01 -3.15557569e-01 -9.64231849e-01 -1.14897537e+00 -6.17756724e-01 4.38094586e-01 4.60279852e-01 -8.51877570e-01 -1.90717995e-01 5.13419032e-01]
[10.407031059265137, 3.118748664855957]
a985bbd3-ee0e-4761-875b-c7397ddc6404
bias-against-93-stigmatized-groups-in-masked
2306.05550
null
https://arxiv.org/abs/2306.05550v1
https://arxiv.org/pdf/2306.05550v1.pdf
Bias Against 93 Stigmatized Groups in Masked Language Models and Downstream Sentiment Classification Tasks
The rapid deployment of artificial intelligence (AI) models demands a thorough investigation of biases and risks inherent in these models to understand their impact on individuals and society. This study extends the focus of bias evaluation in extant work by examining bias against social stigmas on a large scale. It focuses on 93 stigmatized groups in the United States, including a wide range of conditions related to disease, disability, drug use, mental illness, religion, sexuality, socioeconomic status, and other relevant factors. We investigate bias against these groups in English pre-trained Masked Language Models (MLMs) and their downstream sentiment classification tasks. To evaluate the presence of bias against 93 stigmatized conditions, we identify 29 non-stigmatized conditions to conduct a comparative analysis. Building upon a psychology scale of social rejection, the Social Distance Scale, we prompt six MLMs: RoBERTa-base, RoBERTa-large, XLNet-large, BERTweet-base, BERTweet-large, and DistilBERT. We use human annotations to analyze the predicted words from these models, with which we measure the extent of bias against stigmatized groups. When prompts include stigmatized conditions, the probability of MLMs predicting negative words is approximately 20 percent higher than when prompts have non-stigmatized conditions. In the sentiment classification tasks, when sentences include stigmatized conditions related to diseases, disability, education, and mental illness, they are more likely to be classified as negative. We also observe a strong correlation between bias in MLMs and their downstream sentiment classifiers (r =0.79). The evidence indicates that MLMs and their downstream sentiment classification tasks exhibit biases against socially stigmatized groups.
['Aylin Caliskan', 'Sonia Fereidooni', 'Katelyn X. Mei']
2023-06-08
null
null
null
null
['sentiment-analysis']
['natural-language-processing']
[ 4.11823481e-01 4.56824660e-01 -6.30531251e-01 -7.51965523e-01 -2.46287197e-01 -5.33521950e-01 6.69817984e-01 6.69598937e-01 -7.49553800e-01 8.04950356e-01 9.75608349e-01 -5.22002935e-01 -1.77993011e-02 -7.79873252e-01 -3.75091702e-01 -3.43470454e-01 1.91700071e-01 2.39610210e-01 -5.37316322e-01 -5.57230055e-01 3.83501232e-01 4.11188513e-01 -9.54033732e-01 2.30364650e-01 1.03548527e+00 4.64230865e-01 -5.48440337e-01 1.11798689e-01 2.88751870e-01 6.05136514e-01 -6.97735786e-01 -9.66020048e-01 9.81615409e-02 -2.88699090e-01 -5.20303607e-01 -1.43174976e-01 7.35807657e-01 -3.68932247e-01 1.63545817e-01 1.33740544e+00 6.96418822e-01 -3.46332341e-01 1.08447838e+00 -1.19532716e+00 -1.03172219e+00 7.63900578e-01 -4.72502321e-01 2.78715670e-01 4.67220783e-01 1.28693953e-01 9.49737012e-01 -6.95983052e-01 9.19302881e-01 1.78332698e+00 8.37877095e-01 8.24625075e-01 -1.32254386e+00 -1.27266920e+00 5.10634422e-01 -5.50319672e-01 -9.32898879e-01 -3.91249061e-01 2.86005169e-01 -7.83760190e-01 6.13054991e-01 2.73246288e-01 5.84318459e-01 1.54816103e+00 3.75053316e-01 5.46254925e-02 1.54524362e+00 -4.71102148e-02 3.33838105e-01 6.69365108e-01 5.48538983e-01 5.58627248e-01 7.18149424e-01 1.36410192e-01 -4.91146743e-01 -9.79463756e-01 1.07675090e-01 2.36915156e-01 -5.71520440e-02 1.52967662e-01 -1.19384444e+00 1.11809158e+00 7.77989507e-01 2.03170061e-01 -2.09619731e-01 -3.96403521e-01 3.46968651e-01 2.60339379e-01 1.05507600e+00 5.58707058e-01 -4.66810405e-01 3.54261458e-01 -5.99709570e-01 1.81975514e-01 5.71313560e-01 4.43687379e-01 5.56516290e-01 -7.91072249e-02 -2.07973570e-01 1.05491626e+00 2.00168863e-01 9.88534391e-01 5.31424105e-01 -5.13824642e-01 3.77687722e-01 6.87696755e-01 8.30831453e-02 -1.47668910e+00 -7.90693283e-01 -3.65281880e-01 -9.48602796e-01 7.31565729e-02 1.81821853e-01 -6.24478877e-01 -9.14633334e-01 2.10039067e+00 2.35149205e-01 -2.42005840e-01 2.41567954e-01 9.24880683e-01 5.37243724e-01 1.28028169e-01 8.18940461e-01 -4.62264381e-02 1.45401251e+00 -2.76240647e-01 -5.71996927e-01 -7.18871951e-01 1.19880974e+00 -6.32313490e-01 9.88582909e-01 2.49094311e-02 -7.10874140e-01 -2.61489958e-01 -8.42986703e-01 7.19813537e-03 -6.23706996e-01 -3.38899046e-01 7.37095833e-01 9.81116831e-01 -9.14964020e-01 5.58638871e-01 -1.14043571e-01 -8.06408107e-01 6.34200931e-01 1.13438830e-01 -3.47443491e-01 -3.24832574e-02 -1.59250116e+00 1.12356889e+00 -6.99160025e-02 -2.70126283e-01 -6.99086726e-01 -7.80633330e-01 -8.53848696e-01 -2.87047416e-01 -1.62544399e-01 -5.52299678e-01 6.21048987e-01 -1.67323661e+00 -4.14701194e-01 1.62220252e+00 -2.50856042e-01 -3.09800416e-01 2.89567620e-01 -1.74461812e-01 -1.15619850e+00 -2.68113434e-01 5.17022431e-01 8.17532718e-01 7.95592368e-01 -1.09589565e+00 -4.61339414e-01 -7.21348524e-01 1.79444030e-01 3.43343884e-01 -7.51509190e-01 7.05570042e-01 7.62865484e-01 -8.03201139e-01 -2.76411235e-01 -1.11073041e+00 -3.16505313e-01 -5.32124005e-03 -5.15720725e-01 -1.17325820e-01 1.96268886e-01 -5.17225981e-01 1.14437544e+00 -2.27184129e+00 -1.49943277e-01 2.80384183e-01 3.06294680e-01 3.63504320e-01 2.97239795e-02 5.11204377e-02 -3.55161935e-01 7.62262285e-01 -1.96749851e-01 -5.11223935e-02 -5.31512611e-02 1.59587398e-01 -4.47715491e-01 5.63804150e-01 5.22467732e-01 4.92989004e-01 -1.05854630e+00 -3.64599138e-01 -1.78131387e-01 2.62981683e-01 -8.11750054e-01 -2.68309504e-01 4.04376924e-01 -1.32546157e-01 -2.35183939e-01 8.98004949e-01 5.94243467e-01 4.33365960e-04 1.44774675e-01 1.93828717e-01 1.20990686e-02 5.01936674e-01 -4.70870435e-01 7.67429173e-01 -1.26341268e-01 3.30896348e-01 3.39152217e-01 -6.19873464e-01 9.03824687e-01 1.60082906e-01 -1.79433569e-01 -4.07150805e-01 2.57649630e-01 3.04463327e-01 3.86435747e-01 -2.18285486e-01 3.74723673e-01 -8.04324448e-01 -3.38528275e-01 5.61152279e-01 -3.35187733e-01 -1.01959705e-01 -2.05650911e-01 5.10657430e-01 9.54138577e-01 -5.99644959e-01 3.31316411e-01 -8.31220806e-01 3.29343945e-01 2.86991068e-04 8.66953492e-01 6.25537336e-01 -7.53586471e-01 2.61884660e-01 6.95281386e-01 -3.17095697e-01 -7.57642508e-01 -1.14115465e+00 -4.67461139e-01 1.32532060e+00 -1.30574837e-01 -9.37632993e-02 -5.64548194e-01 -1.13320875e+00 5.18165171e-01 1.02858233e+00 -9.53502059e-01 -6.61284387e-01 4.78242040e-02 -1.23211193e+00 6.17963731e-01 2.69921601e-01 1.19606517e-01 -9.98434186e-01 -2.11339202e-02 -5.18688738e-01 5.02865128e-02 -8.48535657e-01 -3.01631898e-01 -1.67531759e-01 -5.99588633e-01 -8.99000883e-01 -7.32627451e-01 -6.37610853e-01 1.07519197e+00 -2.70768292e-02 1.12305307e+00 3.79215300e-01 -3.90177406e-02 1.45362452e-01 -5.56849651e-02 -9.21561182e-01 -5.13929307e-01 -2.07523420e-01 5.56332946e-01 -8.59567970e-02 7.97788918e-01 -1.22839607e-01 -4.77291137e-01 2.86251009e-01 -8.79783928e-01 -2.25928068e-01 3.45487952e-01 3.47910404e-01 -1.82649076e-01 -6.17014587e-01 9.69927371e-01 -1.40159941e+00 8.09988976e-01 -1.03249776e+00 3.80064510e-02 -1.38490777e-02 -8.75618637e-01 -7.24892080e-01 2.71405995e-01 -5.67017972e-01 -9.80074525e-01 -6.00093126e-01 -1.48192663e-02 4.98519093e-01 -5.68995893e-01 2.75015444e-01 1.62789226e-01 2.69154936e-01 1.04786837e+00 -7.51448214e-01 -1.74456105e-01 -5.86776622e-02 -8.77834707e-02 1.16684043e+00 -2.85239398e-01 -3.11412573e-01 7.49357462e-01 7.21537471e-01 -4.64820355e-01 -7.99577594e-01 -1.51313055e+00 -1.26432508e-01 -2.84982443e-01 -9.88572016e-02 1.05445194e+00 -1.19407451e+00 -3.98810923e-01 2.57363856e-01 -6.71688378e-01 -2.90306032e-01 1.73213705e-01 6.99197590e-01 1.70925066e-01 1.32624609e-02 -5.08087099e-01 -8.94259095e-01 -4.37011212e-01 -6.95411861e-01 9.23840702e-01 3.31441909e-02 -1.14239299e+00 -1.21118200e+00 -2.13384688e-01 5.24868011e-01 3.00411731e-01 3.56009752e-01 1.26468849e+00 -9.57809329e-01 5.62613308e-01 -7.90891796e-02 -4.62484062e-01 5.49167991e-01 4.19550836e-01 -9.12640542e-02 -7.07975566e-01 -3.34361255e-01 -1.41240165e-01 -6.50462508e-01 5.90742111e-01 1.63833126e-01 6.15053058e-01 -2.64742821e-01 -4.22590584e-01 2.52636582e-01 1.05239058e+00 7.73885697e-02 1.15911290e-01 1.71129093e-01 5.42465568e-01 1.02340305e+00 7.33117640e-01 1.87062114e-01 2.77640581e-01 -5.23645021e-02 1.74995944e-01 -5.06283939e-01 2.14286640e-01 -2.03606576e-01 8.23690176e-01 3.91579211e-01 2.85051167e-01 -1.66803196e-01 -1.19303000e+00 7.71692693e-01 -1.11176860e+00 -6.81655407e-01 -1.31724730e-01 2.07198668e+00 8.03797424e-01 4.38849896e-01 -1.36153907e-01 -9.47939605e-02 1.02439857e+00 3.32384020e-01 -6.56054676e-01 -1.05369246e+00 -3.80514055e-01 -1.50012728e-02 5.99238038e-01 6.80934310e-01 -9.15452421e-01 9.51395273e-01 7.01358223e+00 3.52769822e-01 -1.05131614e+00 -1.55339077e-01 1.20783675e+00 -3.49897653e-01 -6.88852012e-01 -9.37335119e-02 -9.63293970e-01 4.65900660e-01 8.48089635e-01 -2.85121441e-01 2.04867527e-01 6.76505208e-01 2.84861386e-01 -1.07977919e-01 -8.60931039e-01 2.71160126e-01 4.48658317e-01 -5.84828854e-01 7.37989098e-02 1.35066479e-01 1.05513573e+00 2.13511407e-01 2.98441023e-01 3.22791785e-01 7.30567098e-01 -1.03141963e+00 6.48497522e-01 2.93938249e-01 7.85288632e-01 -7.02253282e-01 7.91216612e-01 2.01951027e-01 5.62217869e-02 -3.20649147e-01 -4.11680311e-01 -7.49083817e-01 1.67548344e-01 1.08974314e+00 -7.71781981e-01 -1.15421489e-01 7.98866570e-01 8.93472791e-01 -5.79547048e-01 -9.30430740e-02 -3.64492118e-01 9.07869041e-01 -1.48207508e-02 -1.14033267e-01 2.93709606e-01 -1.31540358e-01 3.26475799e-01 1.29507911e+00 1.56104222e-01 2.12144479e-01 -4.62482683e-02 5.62711298e-01 -2.94143140e-01 2.41467714e-01 -9.93015707e-01 -4.52811956e-01 1.02178574e-01 1.41174710e+00 -6.16510987e-01 -7.21196175e-01 -4.04156834e-01 7.07346022e-01 3.47000450e-01 2.98644751e-01 -5.59200406e-01 -3.89236882e-02 1.31664205e+00 2.52681822e-01 -6.78001702e-01 2.83365548e-01 -4.50092196e-01 -1.15473676e+00 -5.36519408e-01 -1.13877034e+00 4.58275169e-01 -8.20876598e-01 -1.79985213e+00 2.61108756e-01 -2.01131612e-01 -5.15136003e-01 7.69249052e-02 -6.66023135e-01 -1.70097843e-01 9.52817917e-01 -1.39761651e+00 -8.07091534e-01 8.49485248e-02 6.16951764e-01 -1.16739310e-01 -1.15066074e-01 1.05005896e+00 3.19692314e-01 -6.95754290e-01 3.67312491e-01 -3.22461337e-01 4.40148473e-01 1.53889418e+00 -9.86381233e-01 3.51081043e-01 5.19768119e-01 -5.96531689e-01 9.31943655e-01 5.10567605e-01 -1.25816131e+00 -3.30853939e-01 -1.12520325e+00 1.52483010e+00 -5.65257251e-01 8.49939406e-01 -5.36639333e-01 -5.53292692e-01 9.32989836e-01 -3.27778272e-02 -3.53972584e-01 1.38413346e+00 5.47867358e-01 -6.31808460e-01 1.55715093e-01 -1.59733951e+00 9.25408721e-01 1.51752472e+00 -7.57445931e-01 -8.90846908e-01 6.43198729e-01 6.93420291e-01 1.58273876e-01 -8.01637530e-01 5.24406672e-01 4.03613538e-01 -1.00312936e+00 8.74881029e-01 -1.27332747e+00 9.10898328e-01 4.53421563e-01 -1.21977359e-01 -1.65284121e+00 -5.77386737e-01 -7.93570057e-02 8.04479480e-01 1.33956349e+00 6.58836961e-01 -1.22543132e+00 3.89879316e-01 1.02375853e+00 9.01530981e-02 -3.79149348e-01 -5.59631050e-01 -2.38406435e-01 5.01995504e-01 -1.75129473e-01 4.38162625e-01 1.84063148e+00 1.80213130e-03 2.83740401e-01 -1.35196477e-01 2.72754282e-01 4.61010993e-01 -4.05888967e-02 3.37025821e-01 -1.40910959e+00 3.19900692e-01 -3.48015249e-01 -1.51362985e-01 -6.02687299e-02 4.66557354e-01 -1.04803324e+00 -4.83128488e-01 -1.43473756e+00 3.37168783e-01 -6.21090233e-01 -4.86208409e-01 5.23953974e-01 -4.59693581e-01 4.38371092e-01 2.44575024e-01 -1.83689281e-01 -3.99776585e-02 4.23272289e-02 9.69260693e-01 -2.04289537e-02 2.03347340e-01 -2.48093173e-01 -1.51348805e+00 1.39463758e+00 9.14530218e-01 -7.53757417e-01 -5.62837720e-02 -4.80828911e-01 7.14884043e-01 -5.97997427e-01 3.52760583e-01 -5.54395020e-01 -5.06259680e-01 -3.83348465e-01 5.87464392e-01 6.83259219e-02 2.66933113e-01 -7.18215048e-01 -3.39832872e-01 8.25745702e-01 -8.72268975e-01 1.47910252e-01 4.70169559e-02 2.22631156e-01 1.10843174e-01 -1.49890438e-01 7.05586970e-01 -6.84256554e-02 -3.42233181e-02 1.15633896e-02 -6.94807708e-01 4.14848924e-01 8.87151241e-01 2.84157395e-01 -5.52568674e-01 -4.15111393e-01 -7.05396473e-01 8.01139325e-02 6.85233414e-01 6.55999422e-01 3.43844265e-01 -1.18728244e+00 -7.83192635e-01 2.26299405e-01 3.40318948e-01 -7.57283747e-01 -7.09861591e-02 8.22831392e-01 -5.58830239e-02 7.62851611e-02 -1.99398711e-01 8.47097859e-02 -1.44957340e+00 6.52583539e-01 2.62387861e-02 1.37239397e-01 2.25438908e-01 8.98572028e-01 5.18739045e-01 -7.22755551e-01 2.67393347e-02 -2.79994845e-01 -1.97540849e-01 5.18904984e-01 4.79929268e-01 5.37210345e-01 -3.02966446e-01 -1.06234050e+00 -5.00515699e-01 4.95413780e-01 -2.23926809e-02 3.25132459e-02 1.03115499e+00 -5.93121536e-02 -4.61756468e-01 4.11203414e-01 1.00454235e+00 5.79591990e-01 -2.71009862e-01 -4.49569337e-02 1.27067901e-02 -3.82365584e-01 1.41226267e-02 -1.13951898e+00 -6.67776585e-01 6.75807774e-01 4.90014285e-01 2.44640663e-01 8.33500445e-01 -2.29346007e-01 4.30742890e-01 1.89476516e-02 2.46243492e-01 -1.20793366e+00 -2.79066771e-01 5.19169271e-01 9.82757568e-01 -1.40068138e+00 1.27333716e-01 -4.64686841e-01 -1.07135332e+00 2.25097477e-01 6.38854504e-01 -2.45903626e-01 9.15720582e-01 -8.31847265e-03 5.64269483e-01 -2.05976382e-01 -5.27936161e-01 -6.78693457e-03 1.60040051e-01 5.19683242e-01 6.82502270e-01 4.10608232e-01 -7.79193521e-01 5.59420824e-01 -4.51754272e-01 -4.32691187e-01 4.54071730e-01 5.50424278e-01 -2.40915969e-01 -7.32940137e-01 -5.92771590e-01 8.71502221e-01 -8.49812508e-01 -4.20147061e-01 -1.18136883e+00 5.00557721e-01 5.19714236e-01 1.13178325e+00 3.00213456e-01 -1.87921420e-01 2.32738927e-01 2.80814886e-01 -3.06243986e-01 -9.28604424e-01 -1.29489446e+00 -3.40692341e-01 6.64493203e-01 -4.32292610e-01 -7.12209284e-01 -7.59650230e-01 -1.09019351e+00 -5.02317309e-01 -1.54451966e-01 -2.12654501e-01 5.36054790e-01 7.94582129e-01 5.28586686e-01 1.67445526e-01 3.42309147e-01 -2.94064641e-01 -3.80123973e-01 -1.03690112e+00 -6.63802922e-01 1.07949030e+00 4.64002252e-01 -5.19439578e-01 -7.75570869e-01 -4.43263084e-01]
[9.242762565612793, 10.213057518005371]
b10b9d87-e34b-4b42-b0c3-d6a15cd7cd50
sim2real-3d-object-classification-using
2103.06134
null
https://arxiv.org/abs/2103.06134v1
https://arxiv.org/pdf/2103.06134v1.pdf
Sim2Real 3D Object Classification using Spherical Kernel Point Convolution and a Deep Center Voting Scheme
While object semantic understanding is essential for most service robotic tasks, 3D object classification is still an open problem. Learning from artificial 3D models alleviates the cost of annotation necessary to approach this problem, but most methods still struggle with the differences existing between artificial and real 3D data. We conjecture that the cause of those issue is the fact that many methods learn directly from point coordinates, instead of the shape, as the former is hard to center and to scale under variable occlusions reliably. We introduce spherical kernel point convolutions that directly exploit the object surface, represented as a graph, and a voting scheme to limit the impact of poor segmentation on the classification results. Our proposed approach improves upon state-of-the-art methods by up to 36% when transferring from artificial objects to real objects.
['Markus Vincze', 'Timothy Patten', 'Jean-Baptiste Weibel']
2021-03-10
null
null
null
null
['3d-object-classification']
['computer-vision']
[-4.61599045e-02 5.85819602e-01 -2.94606239e-02 -4.31370199e-01 -3.34315747e-01 -6.59873843e-01 6.25317097e-01 1.81018099e-01 -2.49203101e-01 2.89930552e-01 -3.14081699e-01 -1.57371476e-01 1.24368794e-01 -6.93646371e-01 -8.69152844e-01 -6.01018667e-01 1.25782713e-01 8.19263816e-01 7.23297894e-01 1.20621197e-01 3.92233580e-01 9.72015500e-01 -1.57200015e+00 -2.52038836e-02 7.64110506e-01 1.23939848e+00 2.97584474e-01 3.65265161e-01 -6.79841101e-01 3.22127402e-01 -5.86840034e-01 -1.62667736e-01 5.87433636e-01 1.06711350e-01 -8.50056410e-01 3.70674849e-01 6.38116181e-01 -8.86578932e-02 -3.57404113e-01 1.18128943e+00 6.56264648e-02 -1.05856080e-03 1.06857193e+00 -1.56529367e+00 -6.90005481e-01 -1.13011748e-01 -5.02852738e-01 -4.12508607e-01 3.99532095e-02 2.03314051e-03 6.61596775e-01 -9.66056705e-01 6.44713163e-01 1.48836851e+00 8.63450527e-01 5.12434900e-01 -1.01486766e+00 -2.91091383e-01 1.63600504e-01 1.11165578e-02 -1.14079595e+00 -2.48005271e-01 9.26964819e-01 -6.01356268e-01 8.78114939e-01 -6.74839467e-02 7.28546500e-01 6.26625061e-01 7.82373548e-02 7.84599006e-01 9.28356588e-01 -4.73505080e-01 4.46846664e-01 1.82519227e-01 2.84482986e-01 8.75652850e-01 3.92104059e-01 -2.99170047e-01 -8.73313099e-02 -7.08130971e-02 1.18315184e+00 5.96634299e-02 -1.63336750e-02 -1.21213102e+00 -1.08087897e+00 7.08863020e-01 7.20202386e-01 9.59530920e-02 -2.92604268e-01 1.82359740e-01 9.77224335e-02 3.68948095e-02 4.73088712e-01 5.01873255e-01 -6.56003594e-01 8.31879377e-02 -4.52624768e-01 2.44803771e-01 8.93122733e-01 1.21918941e+00 1.01054454e+00 -2.68594176e-01 2.61913568e-01 6.68459415e-01 3.63511592e-01 4.80244815e-01 1.56784691e-02 -1.13680518e+00 1.31462649e-01 1.09376359e+00 1.89592525e-01 -8.81954968e-01 -5.21639526e-01 -2.91715682e-01 -4.13503349e-01 6.93275094e-01 9.84914899e-01 3.28690797e-01 -1.28957856e+00 1.34206712e+00 5.55089116e-01 1.11527063e-01 -9.92980525e-02 1.01577008e+00 8.72286201e-01 2.12132394e-01 -1.04307055e-01 3.68789464e-01 1.14222205e+00 -1.03737712e+00 -3.05248886e-01 -3.77165020e-01 7.04643488e-01 -5.93028128e-01 9.79892671e-01 -2.00179294e-02 -8.82877946e-01 -4.83178496e-01 -1.00825393e+00 -2.62250721e-01 -5.32552481e-01 6.30617365e-02 8.11892927e-01 5.63480258e-01 -7.57178962e-01 4.58154887e-01 -1.07246768e+00 -6.60253167e-01 7.54661500e-01 4.09598172e-01 -6.28088057e-01 8.27460922e-03 -5.11480510e-01 1.25562561e+00 4.21033144e-01 -6.96575344e-02 -3.64230573e-01 -7.80665636e-01 -8.46154511e-01 -2.22949117e-01 3.77026498e-01 -6.24072671e-01 1.17035294e+00 -6.81199610e-01 -1.39103520e+00 1.23256278e+00 -3.78978695e-03 -3.23483348e-01 6.89421296e-01 -3.03445995e-01 2.45330557e-01 9.83115062e-02 -8.50898772e-02 8.01933646e-01 7.68221855e-01 -1.65558124e+00 -4.59220260e-01 -8.40500653e-01 3.71624738e-01 2.76894569e-01 2.35359166e-02 -4.98114347e-01 -6.35278106e-01 -3.27590257e-01 8.46755624e-01 -1.13984585e+00 -3.09370995e-01 7.15795934e-01 -1.49015054e-01 -5.25874078e-01 1.30113506e+00 -5.29081821e-01 1.44030869e-01 -2.10797763e+00 5.44049703e-02 1.08227856e-01 1.74056083e-01 1.48139462e-01 -4.18388322e-02 -5.85549623e-02 1.42320454e-01 -9.40744579e-02 -2.50984102e-01 -3.07804734e-01 1.10863604e-01 4.71945018e-01 -5.24668358e-02 8.06139052e-01 3.51255774e-01 8.91165733e-01 -8.09869826e-01 -4.84723210e-01 5.15304685e-01 5.57188809e-01 -3.92065227e-01 1.24072842e-01 -5.06749451e-01 5.50291359e-01 -6.43651783e-01 6.73367798e-01 9.32593048e-01 -1.16263457e-01 -2.49034181e-01 -3.32316697e-01 1.09431610e-01 2.48639941e-01 -1.33831525e+00 2.15995765e+00 -2.32018128e-01 4.20996875e-01 9.76568535e-02 -1.33627605e+00 1.15986967e+00 1.08487211e-01 7.22324967e-01 -2.43039444e-01 2.11806387e-01 2.76060164e-01 -2.38183007e-01 -4.43361551e-01 2.25680813e-01 1.24537170e-01 1.30482048e-01 1.66813433e-01 1.00010671e-01 -8.66440594e-01 -2.04314500e-01 -2.65545696e-02 9.43966687e-01 6.10379696e-01 1.01025179e-01 -3.54227453e-01 3.37005794e-01 3.46379101e-01 4.14539754e-01 6.19239807e-01 -2.97786444e-01 7.95657873e-01 5.06161630e-01 -5.92329681e-01 -1.20902336e+00 -1.01900458e+00 -2.21057966e-01 5.25475860e-01 6.83668613e-01 1.51858240e-01 -8.53474021e-01 -9.80525434e-01 3.93079817e-01 6.81921661e-01 -5.67365289e-01 -1.93796262e-01 -5.14826715e-01 -4.21207756e-01 2.08584726e-01 6.60867393e-01 4.49856192e-01 -7.01539099e-01 -7.34759271e-01 1.16877742e-01 1.38278589e-01 -1.50975728e+00 -6.00826293e-02 2.59441406e-01 -1.19577801e+00 -1.30585229e+00 -4.73276407e-01 -9.00159299e-01 9.65386212e-01 5.15510976e-01 1.14005530e+00 9.08801407e-02 -3.51046681e-01 4.40697372e-01 -2.99862146e-01 -7.99916089e-01 -5.97647011e-01 1.57621667e-01 -2.50120070e-02 -2.20874920e-01 4.92802799e-01 -5.01701176e-01 -3.98890704e-01 4.55282897e-01 -7.33979106e-01 1.05976596e-01 5.10740995e-01 5.59107304e-01 6.58915102e-01 -5.74245788e-02 2.31677309e-01 -7.24284887e-01 -4.76941727e-02 -9.49565768e-02 -6.64384723e-01 1.96224198e-01 -3.45939398e-01 1.88195035e-01 1.02035686e-01 -5.25575399e-01 -8.57962132e-01 5.85386217e-01 4.84286770e-02 -6.40635014e-01 -6.24806046e-01 -1.77639142e-01 -1.86538488e-01 -4.39055741e-01 5.81026554e-01 -1.83137074e-01 4.04593945e-01 -6.11869156e-01 5.52037477e-01 6.40167892e-01 4.21513975e-01 -5.11910379e-01 8.80857766e-01 8.85090411e-01 2.59061515e-01 -9.55623090e-01 -1.00105512e+00 -5.06901622e-01 -1.21526134e+00 -2.07897320e-01 9.36443210e-01 -7.00929046e-01 -5.40292978e-01 3.80526006e-01 -1.48044431e+00 -2.69479334e-01 -5.61667502e-01 3.55147839e-01 -8.75581264e-01 2.27223009e-01 -4.56214339e-01 -8.99747670e-01 1.73736643e-02 -1.01366901e+00 1.34885430e+00 1.01052649e-01 3.32441293e-02 -8.01588714e-01 -3.78578007e-01 3.85660172e-01 1.05211467e-01 3.64955097e-01 1.17263961e+00 -4.56915975e-01 -7.57995427e-01 -4.94677871e-01 -5.69927514e-01 2.17689276e-01 1.61127761e-01 -2.43758112e-01 -1.13561809e+00 5.84134925e-03 1.68919697e-01 -2.45713830e-01 4.77719575e-01 3.65160435e-01 1.17084026e+00 2.00392097e-01 -5.57400823e-01 3.53449166e-01 1.22769237e+00 -5.53278141e-02 5.76342642e-01 2.36669466e-01 8.37282896e-01 7.52065241e-01 5.58622956e-01 3.92561890e-02 3.50065500e-01 7.66905606e-01 8.88422072e-01 -1.77828148e-01 -4.77329582e-01 -2.24326849e-01 -1.42892644e-01 4.79027122e-01 -1.49443910e-01 8.66589621e-02 -1.00070131e+00 4.19840634e-01 -1.89655447e+00 -3.88608634e-01 -4.52947080e-01 2.13212442e+00 3.90820265e-01 3.86909723e-01 -9.08143669e-02 2.26605833e-01 6.40150666e-01 -2.70855367e-01 -7.51230955e-01 1.07140519e-01 -8.33598748e-02 -7.60985389e-02 6.75999045e-01 3.43440175e-01 -1.12945199e+00 1.02959192e+00 6.21585655e+00 5.51099598e-01 -1.05595160e+00 4.94479612e-02 2.64328986e-01 4.29950297e-01 2.22599402e-01 1.14886038e-01 -6.29271030e-01 1.13783060e-02 1.21919207e-01 3.80660057e-01 7.46116042e-02 1.22109067e+00 -7.11539090e-02 -3.84001955e-02 -1.29519939e+00 1.11254597e+00 8.79090652e-02 -1.07118917e+00 2.68407222e-02 6.36013672e-02 6.37594879e-01 1.50924847e-01 -3.20302278e-01 9.06858072e-02 2.03170970e-01 -8.11581433e-01 9.38415885e-01 4.63067234e-01 2.98198074e-01 -3.18040252e-01 6.49647772e-01 5.29315829e-01 -1.00657856e+00 9.65707004e-02 -5.75883627e-01 -3.18572596e-02 -5.19427620e-02 5.92110276e-01 -1.08782017e+00 2.92622238e-01 6.82598233e-01 4.99273032e-01 -6.28991127e-01 1.25860667e+00 -1.73374996e-01 5.14462143e-02 -6.68178976e-01 2.05805432e-02 1.66157186e-01 -2.24545836e-01 7.34222591e-01 8.69332016e-01 2.17928991e-01 -6.19780039e-03 4.49988663e-01 9.61214542e-01 1.31329954e-01 -5.08973859e-02 -6.91761672e-01 1.44269601e-01 2.09351227e-01 1.12679720e+00 -1.22688615e+00 -8.60822275e-02 -5.21698475e-01 9.66957927e-01 3.69326949e-01 4.11780089e-01 -5.76978207e-01 -1.67768896e-01 7.25750029e-01 2.15785891e-01 2.42266059e-01 -8.17637146e-01 -7.49548078e-01 -1.00186515e+00 2.06992418e-01 -2.51395673e-01 4.21902388e-02 -9.21302319e-01 -1.19385052e+00 3.90357137e-01 -1.49490342e-01 -1.11404121e+00 1.46688491e-01 -1.04449069e+00 -3.78094912e-02 5.89306593e-01 -1.56147528e+00 -1.33143461e+00 -5.04969954e-01 2.49570146e-01 7.05491543e-01 1.94142237e-01 7.33752608e-01 -3.07193007e-02 2.16190457e-01 9.20470878e-02 -1.38803676e-01 1.94011733e-01 3.11355352e-01 -1.34259510e+00 5.29029071e-01 3.24382275e-01 2.66305387e-01 8.74541551e-02 5.80658615e-01 -4.72609907e-01 -1.52675521e+00 -8.45670223e-01 5.28159320e-01 -9.66325462e-01 3.36223722e-01 -6.43390834e-01 -1.04672551e+00 3.51887971e-01 -4.03717428e-01 3.60788703e-01 -3.05679161e-02 -2.30761021e-01 -4.62942004e-01 2.50429124e-01 -1.41056836e+00 5.02879500e-01 1.44925225e+00 -3.77893865e-01 -6.30145252e-01 3.83184463e-01 9.30667520e-01 -7.50194490e-01 -7.70655394e-01 5.34929633e-01 4.91456062e-01 -7.04777181e-01 1.10115337e+00 -6.78284705e-01 -1.96492523e-02 -4.77452517e-01 -3.14408034e-01 -1.16325569e+00 -9.16643515e-02 -1.93704158e-01 -3.36138219e-01 8.18763554e-01 1.64621964e-01 -3.99336725e-01 1.24482715e+00 8.55970323e-01 -3.44243169e-01 -5.21073341e-01 -8.64003122e-01 -9.13753152e-01 1.14250511e-01 -5.65596759e-01 5.77057183e-01 9.76700842e-01 -5.36878288e-01 1.90822914e-01 3.34371060e-01 3.75777781e-01 8.26919556e-01 2.10804746e-01 1.07449007e+00 -1.77416539e+00 1.56688780e-01 -4.38912719e-01 -9.96188700e-01 -1.23288608e+00 2.85395503e-01 -8.75913143e-01 1.97205856e-01 -1.76202655e+00 -2.07788870e-01 -9.86290991e-01 1.16795272e-01 4.76540834e-01 2.33567044e-01 3.56681257e-01 1.04280494e-01 1.20683655e-01 -3.18828940e-01 5.20526648e-01 1.51648378e+00 -1.91019431e-01 -2.81696320e-01 1.77995283e-02 -1.72151208e-01 1.16013384e+00 5.61866701e-01 -3.73092979e-01 -1.47661582e-01 -7.25191355e-01 -2.27437004e-01 -3.16709727e-01 6.09709799e-01 -9.92800295e-01 2.20307812e-01 -2.26760730e-01 2.96334624e-01 -7.14064419e-01 5.52372038e-01 -1.25871754e+00 -1.29245549e-01 2.93251544e-01 1.02028763e-02 -3.06992203e-01 1.29121944e-01 5.59674203e-01 1.55644774e-01 -3.44700694e-01 9.27012324e-01 -2.64764100e-01 -7.22471297e-01 3.55868071e-01 1.36749357e-01 -1.04624137e-01 1.00100374e+00 -4.90653247e-01 2.18211096e-02 -9.27524343e-02 -7.11953700e-01 2.67918762e-02 8.38168502e-01 6.89391851e-01 4.07908589e-01 -1.17783368e+00 -4.22959983e-01 2.63163298e-01 2.40041986e-01 6.45638287e-01 -1.87105790e-01 5.24714351e-01 -7.64954448e-01 2.15392083e-01 -2.18533836e-02 -1.07643914e+00 -9.27396774e-01 4.31130439e-01 4.59364384e-01 3.31306130e-01 -9.66098726e-01 6.73122287e-01 2.86525100e-01 -7.23432541e-01 6.50777996e-01 -5.67238510e-01 1.57344136e-02 -2.93766975e-01 4.69923653e-02 2.47440428e-01 2.68058121e-01 -7.19786942e-01 -3.03903073e-01 9.39740598e-01 1.94499120e-01 2.66784817e-01 1.35868227e+00 -6.81925192e-02 -4.17525396e-02 5.59892654e-01 1.02709627e+00 -5.42327821e-01 -1.55784798e+00 -4.12638247e-01 1.66595340e-01 -5.31688452e-01 6.54445440e-02 -4.31669772e-01 -9.93613183e-01 8.98084760e-01 7.25284874e-01 4.19149518e-01 4.63739604e-01 4.34789777e-01 7.20697641e-01 7.34831333e-01 5.93777716e-01 -1.06625724e+00 6.86999336e-02 5.58622420e-01 8.34001780e-01 -1.38152289e+00 -1.75343954e-03 -1.04578197e+00 -2.05445752e-01 1.30020857e+00 6.13034308e-01 -3.64938229e-01 9.00909960e-01 1.53977737e-01 1.32364303e-01 -3.33666354e-01 -4.41809557e-02 -2.60886908e-01 2.74736673e-01 8.95017922e-01 1.26117319e-01 -9.50081050e-02 1.17538571e-01 1.80055082e-01 -6.79400638e-02 -1.99031457e-01 1.28801078e-01 1.03944528e+00 -5.13531029e-01 -9.75930393e-01 -4.07921255e-01 3.10998321e-01 -1.14583060e-01 5.79581380e-01 -3.56889367e-01 8.80445719e-01 1.75136432e-01 6.49792552e-01 3.62510175e-01 -3.83130200e-02 5.44684649e-01 2.88136989e-01 8.90350223e-01 -6.27115905e-01 6.78840578e-02 -1.39730215e-01 -2.13211432e-01 -4.38894719e-01 -7.26707935e-01 -5.76373279e-01 -1.61083114e+00 2.04403654e-01 -6.59182966e-01 -1.36555567e-01 1.27768457e+00 1.09866118e+00 2.48334527e-01 2.44047850e-01 2.87506014e-01 -1.18378007e+00 -5.23974955e-01 -9.60067928e-01 -4.91368979e-01 6.90704882e-01 1.52796939e-01 -1.14208651e+00 -3.28554511e-01 2.05836117e-01]
[7.973675727844238, -2.944370746612549]
b2336a7b-b7fe-4ff6-a9d0-de0c5b52b745
quantum-state-tomography-with-conditional
2008.03240
null
https://arxiv.org/abs/2008.03240v2
https://arxiv.org/pdf/2008.03240v2.pdf
Quantum State Tomography with Conditional Generative Adversarial Networks
Quantum state tomography (QST) is a challenging task in intermediate-scale quantum devices. Here, we apply conditional generative adversarial networks (CGANs) to QST. In the CGAN framework, two duelling neural networks, a generator and a discriminator, learn multi-modal models from data. We augment a CGAN with custom neural-network layers that enable conversion of output from any standard neural network into a physical density matrix. To reconstruct the density matrix, the generator and discriminator networks train each other on data using standard gradient-based methods. We demonstrate that our QST-CGAN reconstructs optical quantum states with high fidelity orders of magnitude faster, and from less data, than a standard maximum-likelihood method. We also show that the QST-CGAN can reconstruct a quantum state in a single evaluation of the generator network if it has been pre-trained on similar quantum states.
['Anton Frisk Kockum', 'Carlos Sánchez Muñoz', 'Shahnawaz Ahmed', 'Franco Nori']
2020-08-07
null
null
null
null
['quantum-state-tomography']
['medical']
[ 6.82842374e-01 2.95343250e-01 2.17225596e-01 -1.94026947e-01 -1.33432603e+00 -6.20587170e-01 6.02859020e-01 -4.99514759e-01 -4.90164548e-01 1.06337142e+00 -1.69789597e-01 -5.54469705e-01 2.88254976e-01 -1.28274739e+00 -1.09379232e+00 -1.01477599e+00 2.14554489e-01 8.55447650e-01 -7.65532628e-02 -1.97651282e-01 1.73169449e-01 3.68014336e-01 -8.06429148e-01 2.62389243e-01 8.25161695e-01 8.52404118e-01 -2.52900720e-01 1.10759890e+00 4.34597999e-01 9.75008547e-01 -7.01276004e-01 -3.85713816e-01 4.89187926e-01 -1.03598368e+00 -9.40132737e-01 -7.67790377e-01 2.45264500e-01 -7.49503493e-01 -1.33866227e+00 1.50702691e+00 5.11854172e-01 1.28182814e-01 8.00554812e-01 -1.00205874e+00 -9.89271283e-01 6.28318727e-01 2.88261592e-01 -1.78574324e-01 1.82234481e-01 6.50050104e-01 9.32566345e-01 -3.48985314e-01 7.91952252e-01 1.17874300e+00 3.62783134e-01 1.11859119e+00 -1.97715938e+00 -9.12403405e-01 -1.17767811e+00 2.72028372e-02 -1.30665708e+00 -6.59383893e-01 4.87719536e-01 -3.18089761e-02 1.33421183e+00 -1.21819995e-01 5.59719682e-01 1.28776085e+00 4.58096564e-01 2.66246676e-01 1.54169226e+00 -4.97263759e-01 3.81346107e-01 -4.84782457e-02 -2.89223075e-01 9.22394454e-01 -5.74422181e-02 7.07984686e-01 -4.96317625e-01 -2.91918278e-01 9.56362009e-01 -2.77130306e-01 -2.91528236e-02 -1.95779249e-01 -1.15252233e+00 9.31785166e-01 7.44924307e-01 -1.23324335e-01 -1.81801125e-01 8.69933724e-01 4.21409048e-02 5.62192261e-01 -2.48525932e-01 7.57381082e-01 2.41223648e-01 3.04168724e-02 -7.99189746e-01 9.60451737e-02 9.65093493e-01 7.45708823e-01 1.25229573e+00 2.00126216e-01 -3.10792178e-01 -2.54440513e-02 1.20156594e-01 1.19471049e+00 1.48608506e-01 -1.20835388e+00 1.04250714e-01 -2.04142332e-01 1.55999869e-01 -8.94830227e-02 5.34517132e-03 4.18075286e-02 -9.92868721e-01 3.72123539e-01 2.24289715e-01 -6.10683501e-01 -1.32455790e+00 2.03138995e+00 -5.28860604e-03 2.67945707e-01 5.02511501e-01 8.11978817e-01 6.48881495e-01 9.31264758e-01 -3.55558157e-01 1.00123003e-01 6.32689834e-01 -5.75073183e-01 -5.19640028e-01 -2.98329502e-01 6.43017232e-01 -5.75630546e-01 4.86310065e-01 2.59033501e-01 -1.26068783e+00 -4.04089779e-01 -1.32895100e+00 -4.50739205e-01 -2.82520056e-01 -1.95570633e-01 7.55961120e-01 7.57321596e-01 -1.28521955e+00 1.27969062e+00 -1.08207214e+00 1.31692767e-01 4.23911721e-01 7.83302069e-01 -4.06528771e-01 -1.25686660e-01 -1.35713291e+00 9.03051257e-01 7.91561007e-01 3.45478505e-02 -1.63405490e+00 -1.78558260e-01 -6.70540214e-01 -1.30397910e-02 -4.12508845e-02 -1.10466933e+00 1.34799206e+00 -3.57878417e-01 -2.21382689e+00 6.54901326e-01 -1.73116297e-01 -6.00658715e-01 -2.18218684e-01 4.71406847e-01 -4.31090951e-01 3.03516060e-01 5.07766455e-02 5.77998757e-01 8.83543193e-01 -7.37845898e-01 4.30670641e-02 -9.52968150e-02 1.03398047e-01 -3.90735179e-01 3.26679200e-01 -3.04142982e-01 4.94698770e-02 1.93997175e-01 2.79049695e-01 -1.30208755e+00 -1.15237884e-01 -3.56148869e-01 -1.02244651e+00 2.21801773e-01 4.49652970e-01 -2.43172392e-01 5.06671429e-01 -1.75068390e+00 3.55475962e-01 3.13544124e-01 2.89228201e-01 3.61060351e-01 -3.49435508e-01 5.50770402e-01 -2.07120851e-02 -1.13749184e-01 -4.25253272e-01 -4.04707223e-01 3.71087939e-01 3.21475476e-01 -3.44216466e-01 4.55112755e-01 4.59141344e-01 1.51664805e+00 -1.02694798e+00 -2.26455573e-02 -3.29886675e-02 4.67441767e-01 -8.58547986e-01 4.14989263e-01 -4.03592467e-01 9.11126018e-01 -2.66620457e-01 4.97734725e-01 7.85700619e-01 -5.72314739e-01 3.23403865e-01 -3.64571363e-01 1.71776488e-01 1.00473952e+00 -6.78944230e-01 2.03610253e+00 -3.97206217e-01 6.11641526e-01 -1.96448535e-01 -9.39921498e-01 4.62888896e-01 3.45923543e-01 5.23503684e-02 -8.60161245e-01 3.51812869e-01 5.42712569e-01 4.37909275e-01 -1.51502371e-01 5.45322299e-01 -9.55720723e-01 -4.57835525e-01 9.52599168e-01 9.11593556e-01 -9.34290588e-01 2.14324798e-02 5.29441178e-01 1.25764942e+00 1.42919868e-01 -5.85859120e-02 1.55459195e-01 -1.28226042e-01 -2.21565664e-01 3.10364455e-01 1.18085515e+00 -1.67072937e-01 5.23308039e-01 6.12118304e-01 -3.46861519e-02 -1.58340943e+00 -1.57125711e+00 -1.08039483e-01 3.49717081e-01 -2.16169897e-02 -4.48024631e-01 -7.00484812e-01 -4.10026193e-01 -1.87106997e-01 7.73228168e-01 -4.11935747e-01 -8.00489664e-01 -2.55733818e-01 -9.14685786e-01 1.04966307e+00 1.96247756e-01 6.88924670e-01 -1.07114875e+00 1.27089411e-01 3.36694568e-01 2.02592909e-01 -1.26528823e+00 -1.30340010e-01 6.84043944e-01 -6.45157933e-01 -6.97254062e-01 -2.92494118e-01 -5.05965173e-01 6.15258515e-01 -4.41808969e-01 1.04596353e+00 -4.11646456e-01 1.58246618e-03 -1.68518707e-01 3.02171648e-01 -2.12860312e-02 -1.34055257e+00 1.33516505e-01 2.44032830e-01 -2.05253556e-01 4.61768746e-01 -9.58918095e-01 -4.42449182e-01 -3.49134624e-01 -9.74617660e-01 7.10077360e-02 6.50564194e-01 1.03868639e+00 5.33554912e-01 -2.39811644e-01 1.33863091e-01 -1.13825428e+00 5.40457368e-01 -1.87986806e-01 -8.86409223e-01 3.08594517e-02 -5.06341219e-01 7.77659416e-01 7.77353168e-01 -1.29307717e-01 -8.29399347e-01 -9.50189084e-02 -4.27015036e-01 -4.27201718e-01 -1.88920721e-02 2.88942844e-01 1.37728363e-01 -6.01087630e-01 1.20891297e+00 4.95332450e-01 -8.03754330e-02 1.42843693e-01 6.02906525e-01 5.59320509e-01 8.36272299e-01 -7.74995208e-01 1.33241832e+00 4.19012249e-01 6.62658870e-01 -2.18923911e-01 -1.08401775e+00 1.33873269e-01 -6.79153502e-01 4.65738684e-01 1.09190869e+00 -9.68305826e-01 -1.05137217e+00 7.67959356e-01 -1.20071232e+00 -6.57692194e-01 -5.23095310e-01 5.16658068e-01 -7.10597634e-01 7.83555359e-02 -1.15058756e+00 -7.82885134e-01 -3.83228332e-01 -1.25584602e+00 1.11590648e+00 4.61296588e-01 4.74671572e-01 -9.30824041e-01 3.26136529e-01 2.27128327e-01 5.73368430e-01 2.26743102e-01 8.93890858e-01 -2.04165250e-01 -1.31899011e+00 -4.25629646e-01 -2.20711842e-01 7.21730411e-01 -1.95621744e-01 -4.04780179e-01 -1.35414088e+00 -5.40681720e-01 1.70192420e-02 -1.22738397e+00 1.00059390e+00 1.67359458e-03 8.89802277e-01 -2.94132024e-01 -9.91570652e-02 1.13937402e+00 1.49665534e+00 1.33931637e-01 9.75029051e-01 -3.26007485e-01 9.25338328e-01 -3.40269089e-01 -3.83229703e-01 -5.95071772e-03 -5.51838428e-02 2.10367054e-01 4.00699526e-01 2.25030258e-01 -3.31598856e-02 -5.39747953e-01 7.78849542e-01 9.24721658e-01 -2.16017216e-01 -1.36418775e-01 -5.36109984e-01 -6.47841990e-02 -1.36817122e+00 -1.17069340e+00 1.37634769e-01 2.14069510e+00 9.92676139e-01 1.91825122e-01 -2.82706469e-01 -2.64263064e-01 3.45596075e-01 1.79879397e-01 -9.27884102e-01 -5.03434658e-01 -1.96213022e-01 1.24821305e+00 7.04332590e-01 5.98375738e-01 -8.97891164e-01 1.23001933e+00 7.09057236e+00 7.12033272e-01 -1.23629916e+00 3.71579409e-01 2.54460245e-01 -1.70813516e-01 -5.19394338e-01 4.89734620e-01 -4.93674338e-01 3.04108620e-01 1.67961025e+00 6.61846995e-02 1.28313148e+00 4.25214291e-01 -5.30627191e-01 3.75016220e-02 -1.22183585e+00 9.59269643e-01 -2.14234993e-01 -1.69062102e+00 6.23047398e-03 3.49539310e-01 1.16988623e+00 7.09061682e-01 2.77629733e-01 6.33700311e-01 8.42439711e-01 -1.41766155e+00 4.59875375e-01 6.96930945e-01 1.48244452e+00 -7.07214713e-01 5.80381155e-01 3.95288050e-01 -6.17061615e-01 1.71824038e-01 -6.74267828e-01 3.12380735e-02 4.89137352e-01 4.52917367e-01 -8.30369413e-01 4.48586583e-01 7.17190579e-02 2.79686362e-01 -8.58601630e-02 4.01335895e-01 -6.08064890e-01 6.88659310e-01 -3.78998697e-01 1.06526166e-02 3.86551738e-01 -6.87495708e-01 4.57539052e-01 6.18539929e-01 4.85336095e-01 5.79834022e-02 -1.46823183e-01 1.89632297e+00 -5.89934051e-01 -9.42874849e-01 -7.35737324e-01 -7.92537212e-01 4.31623638e-01 1.23003340e+00 -2.06827782e-02 -5.39030492e-01 -5.31492494e-02 1.55502498e+00 5.09956002e-01 4.28441375e-01 -5.91546655e-01 -5.68994522e-01 4.31808442e-01 -3.14241499e-01 1.19586371e-01 -3.16207588e-01 2.73808151e-01 -1.64114773e+00 -4.28648889e-01 -7.26807296e-01 -3.19772005e-01 -1.07295763e+00 -1.35338497e+00 5.55170536e-01 -4.34108526e-01 -7.47475743e-01 -7.53255844e-01 -8.74534965e-01 -5.97042263e-01 1.47849309e+00 -1.40312946e+00 -9.34154570e-01 1.93512082e-01 6.52238071e-01 -8.36862206e-01 -2.19736487e-01 1.45323265e+00 -6.66174712e-03 -5.72293162e-01 4.84908521e-01 5.13693273e-01 5.42555153e-01 3.78429294e-01 -1.50546491e+00 8.53575826e-01 1.03085995e+00 3.02850515e-01 1.06976461e+00 5.29895186e-01 -5.66340446e-01 -1.72276795e+00 -8.86742890e-01 4.72858340e-01 -5.33131659e-01 8.44087720e-01 -6.76556110e-01 -5.98375201e-01 1.00131631e+00 3.24249268e-01 5.77830017e-01 6.12448573e-01 -1.43302485e-01 -6.02656603e-01 2.85694152e-01 -1.25548935e+00 1.82244167e-01 8.43718767e-01 -1.81645739e+00 -3.34100902e-01 5.94293058e-01 5.59620619e-01 -7.77490377e-01 -9.30613339e-01 -6.02171645e-02 5.32591105e-01 -1.04725087e+00 8.84464145e-01 -8.08140516e-01 3.73755097e-01 -3.17270130e-01 -2.94259757e-01 -1.46084690e+00 -1.49345994e-01 -1.28336823e+00 -2.84740806e-01 3.25498015e-01 4.29388463e-01 -8.37854207e-01 9.70137477e-01 4.62683797e-01 -1.98118031e-01 -1.50435358e-01 -1.23942494e+00 -7.60680497e-01 5.64708769e-01 -1.85236901e-01 5.55039287e-01 5.04956305e-01 -1.93090782e-01 8.52862179e-01 -5.74150741e-01 3.60210508e-01 8.16550195e-01 1.97028890e-01 6.82142794e-01 -7.90307999e-01 -8.29590857e-01 -3.30959894e-02 -4.42569941e-01 -1.17333102e+00 2.50988156e-01 -1.43088329e+00 2.52602041e-01 -1.14450431e+00 3.22162658e-01 -1.42410278e-01 -3.34884316e-01 2.66843051e-01 3.41972597e-02 5.89875042e-01 1.16087422e-01 1.32892981e-01 -4.62299764e-01 7.61097789e-01 1.48315072e+00 -2.78427482e-01 2.11903259e-01 -1.55744717e-01 -3.40291589e-01 2.30947480e-01 6.04085147e-01 -1.04279172e+00 -1.82089955e-01 -2.53083348e-01 5.70850968e-01 3.69413763e-01 8.03581893e-01 -1.44055295e+00 4.37360138e-01 1.86333194e-01 2.95298785e-01 -2.52586752e-02 7.62073219e-01 8.21554437e-02 1.97840124e-01 5.40419161e-01 -1.47955507e-01 -4.89323884e-01 -1.61918685e-01 4.39389557e-01 5.07903770e-02 -4.28856343e-01 1.00832725e+00 -3.66273195e-01 -1.83516920e-01 5.75991988e-01 -1.79343104e-01 3.48310098e-02 1.95619896e-01 3.22202504e-01 -6.32474363e-01 -4.36276019e-01 -7.25660264e-01 -4.26416636e-01 6.51932776e-01 -5.45575380e-01 4.09336925e-01 -1.37937021e+00 -2.80146062e-01 4.72211093e-01 -1.62135094e-01 1.31293505e-01 2.12272525e-01 6.26042128e-01 -6.12778962e-01 5.22440434e-01 -2.19769344e-01 -3.79023880e-01 -1.55391529e-01 3.57602119e-01 9.31507170e-01 -3.26158166e-01 -1.49566799e-01 9.31135595e-01 -1.73662826e-01 -9.53924060e-01 -6.98700786e-01 -2.45067060e-01 7.27492809e-01 -7.91051388e-01 2.29008988e-01 -1.71334222e-01 -5.21246269e-02 -6.55436814e-01 8.86262879e-02 1.33547589e-01 1.96913570e-01 -6.19988978e-01 9.93406713e-01 5.33045053e-01 -3.09394360e-01 3.24004591e-01 1.55269861e+00 -5.65619580e-02 -1.27298319e+00 -3.26617181e-01 -6.41467154e-01 1.56315893e-01 2.56252021e-01 -7.72232473e-01 -8.34813297e-01 1.12772298e+00 5.89876950e-01 4.10181433e-01 7.16864049e-01 -2.99033895e-02 1.20311284e+00 1.06652033e+00 6.70845509e-01 -8.02026391e-01 2.43775155e-02 6.29841924e-01 1.67121187e-01 -1.41184878e+00 -3.91341388e-01 2.27222443e-01 3.95115577e-02 1.07256019e+00 1.50849000e-01 -5.28867185e-01 4.61847007e-01 1.78548008e-01 -1.35714054e-01 -4.14969087e-01 -5.01735926e-01 -2.07634583e-01 2.91856945e-01 6.66266680e-01 2.36361995e-02 3.20903033e-01 6.54365599e-01 7.90726617e-02 -5.44699132e-01 9.35510024e-02 7.78545320e-01 8.69211674e-01 -1.68073863e-01 -1.51016843e+00 -5.79833277e-02 5.20053804e-01 -3.46375465e-01 -4.93730724e-01 -1.09091356e-01 4.10093009e-01 2.81931460e-02 6.09938204e-01 8.99548922e-03 -7.31090188e-01 -5.09333387e-02 1.96863726e-01 1.06115842e+00 -9.33853626e-01 -1.35363162e-01 -2.60093242e-01 -2.48225816e-02 -8.75055075e-01 -4.82390612e-01 -4.14667845e-01 -1.38709509e+00 -7.61622667e-01 -3.68894190e-01 1.80549875e-01 6.88928068e-01 1.20515525e+00 3.63214463e-01 4.78908241e-01 6.11629665e-01 -8.76578212e-01 -9.90904748e-01 -1.13817716e+00 -7.49939322e-01 1.79993615e-01 6.08000576e-01 -8.00052136e-02 -4.09923851e-01 -4.18265671e-01]
[5.608818531036377, 4.960163116455078]
bdac3244-8b3c-496a-93e3-8cf760c17f45
improve-chit-chat-and-qa-sentence
null
null
https://aclanthology.org/2021.rocling-1.19
https://aclanthology.org/2021.rocling-1.19.pdf
Improve Chit-Chat and QA Sentence Classification in User Messages of Dialogue System using Dialogue Act Embedding
In recent years, dialogue system is booming and widely used in customer service system, and has achieved good results. Viewing the conversation records between users and real customer service, we can see that the user’s sentences are mixed with questions about products and services, and chat with customer service. According to the experience of professionals, it is helpful in improving the user experience to mix some chats in customer service conversations. However, users’ questions are expected to be answered, while chatting is expected to interact with customer service. In order to produce an appropriate response, the dialogue system must be able to distinguish these two intentions effectively. Dialog act is a classification that linguists define according to its function. We think this information will help distinguishing questioning sentences and chatting sentences. In this paper, we combine a published COVID-19 QA dataset and a COVID-19-topic chat dataset to form our experimental data. Based on the BERT (Bidirectional Encoder Representation from Transformers) model, we build a question-chat classifier model. The experimental results show that the accuracy of the configuration with dialog act embedding is 16% higher than that with only original statement embedding. In addition, it is found that conversation behavior types such as “Statement-non-opinion”, “Signal-non-understanding” and “Appreciation” are more related to question sentences, while “Wh-Question”, “Yes-No-Question” and “Rhetorical-Question” questions are more related to chat sentences.
['Yu Ching Chiu', 'Xi Jie Hou', 'Chi Hsiang Chao']
null
null
null
null
rocling-2021-10
['sentence-classification']
['natural-language-processing']
[-1.51315406e-01 2.40059510e-01 6.44637570e-02 -7.39793301e-01 -4.87337738e-01 -5.48782647e-01 6.51340604e-01 -5.60663566e-02 -1.75445721e-01 5.97396135e-01 8.53886306e-01 -5.89851856e-01 2.48737112e-01 -7.08297849e-01 2.34992757e-01 -4.02299374e-01 4.83062744e-01 6.70817614e-01 2.20778719e-01 -9.97187316e-01 3.32912803e-01 -2.54190445e-01 -1.01727176e+00 9.62394059e-01 7.14070559e-01 9.09176648e-01 5.44530571e-01 6.10084772e-01 -1.06773186e+00 1.27886498e+00 -8.59794438e-01 -5.94821870e-01 -3.35918814e-01 -8.16009998e-01 -1.45889378e+00 2.30879724e-01 -3.88323247e-01 -5.04923642e-01 -9.47781429e-02 8.50428522e-01 5.55398524e-01 -9.10763294e-02 5.21879852e-01 -1.30790603e+00 -6.55556679e-01 7.54638553e-01 3.87717821e-02 -2.14220099e-02 9.22912180e-01 1.94598883e-01 1.27274406e+00 -6.76706553e-01 3.92748743e-01 1.53710449e+00 3.78264278e-01 7.24257827e-01 -9.09149289e-01 -4.59980041e-01 4.95237894e-02 3.57979774e-01 -7.17177689e-01 -3.38628769e-01 7.92639196e-01 -3.72345090e-01 1.00840592e+00 6.84376597e-01 4.47083235e-01 1.22915328e+00 1.70830265e-01 1.03573728e+00 8.71351838e-01 -3.75715435e-01 -1.99263971e-02 9.12269652e-01 6.42414033e-01 7.75782540e-02 -7.77032316e-01 -5.02659440e-01 1.13598490e-02 -1.88951001e-01 3.09132695e-01 -2.84640081e-02 -4.66958463e-01 4.34700966e-01 -7.96233416e-01 1.29602122e+00 1.25024319e-01 7.85975754e-01 -3.70992899e-01 -6.19339585e-01 6.51992381e-01 6.12471342e-01 1.80226937e-01 3.72602612e-01 -5.86276889e-01 -7.81056702e-01 -1.55611530e-01 1.41360030e-01 1.63356984e+00 1.03850055e+00 3.19323450e-01 -2.16901258e-01 -2.13150784e-01 1.37497461e+00 4.72031862e-01 2.41600290e-01 7.60683298e-01 -7.78129280e-01 4.36343074e-01 1.03321266e+00 2.47132003e-01 -9.41722751e-01 -3.95449519e-01 1.33727267e-01 -6.65291071e-01 -2.48226792e-01 2.98242122e-01 -3.82566541e-01 -1.74468875e-01 1.23661184e+00 3.47049721e-02 -5.73060155e-01 5.08302629e-01 9.79286015e-01 1.45876598e+00 9.66178238e-01 -1.97714809e-02 -2.86155403e-01 2.12882280e+00 -1.03211522e+00 -1.46838939e+00 -8.16977769e-02 6.51485205e-01 -1.10628402e+00 1.49120426e+00 1.48028851e-01 -8.38090301e-01 -4.65128213e-01 -7.61774004e-01 -1.96802229e-01 -3.50427300e-01 1.08174644e-01 3.14739645e-01 5.62692583e-01 -5.78841507e-01 -1.25018810e-03 3.24747488e-02 -6.01544917e-01 -4.92681146e-01 9.72322971e-02 -2.92991638e-01 3.25068980e-01 -1.76461971e+00 1.19600773e+00 1.06085874e-01 1.99888766e-01 -3.99258405e-01 -1.50942400e-01 -8.25075746e-01 1.14067622e-01 3.89440268e-01 -1.65778100e-01 1.77059615e+00 -9.27026391e-01 -1.81667924e+00 5.35025418e-01 -3.35005462e-01 -1.81424782e-01 9.31923464e-02 1.50772724e-02 -8.46802413e-01 7.75766596e-02 6.32183452e-04 2.38175482e-01 3.40680748e-01 -1.15454996e+00 -7.92130709e-01 -2.60241538e-01 6.38441145e-01 4.59834367e-01 -1.66365147e-01 5.16540170e-01 -1.94716319e-01 8.50069895e-02 4.21241261e-02 -6.77988052e-01 2.72914439e-01 -5.22571802e-01 -4.68842864e-01 -7.28409886e-01 1.20684934e+00 -9.93167400e-01 1.21569788e+00 -1.97728300e+00 -2.61654377e-01 -2.78689057e-01 7.97890052e-02 1.62694648e-01 1.80948138e-01 1.09746480e+00 8.18644613e-02 1.22950725e-01 7.28993192e-02 -1.04164653e-01 2.25025132e-01 4.33249027e-01 -3.46332699e-01 -1.76261857e-01 2.79968232e-01 6.59666836e-01 -5.48703134e-01 -6.44593716e-01 1.84184074e-01 2.54211426e-02 -4.47031289e-01 8.24061215e-01 -1.89849392e-01 3.31110209e-01 -5.72084606e-01 5.20996630e-01 5.06020784e-01 -1.66501909e-01 3.07477415e-01 -3.34832817e-01 -1.00064531e-01 7.94627786e-01 -8.64273489e-01 8.90785694e-01 -8.64431024e-01 6.09344184e-01 5.21259129e-01 -9.57361877e-01 1.19315267e+00 9.02057648e-01 1.56575114e-01 -6.20890081e-01 4.41598654e-01 3.28272469e-02 4.53391224e-01 -1.13930511e+00 6.23261154e-01 -1.94770962e-01 -1.24264874e-01 4.63178664e-01 3.27118821e-02 -3.29153717e-01 2.15745755e-02 3.13254148e-01 8.96408200e-01 -3.68707538e-01 2.42321849e-01 -1.38148054e-01 1.10305250e+00 1.80358943e-02 1.23336442e-01 1.20184362e-01 -2.67104298e-01 2.71876633e-01 9.89129901e-01 1.59559585e-02 -6.95084035e-01 -7.45872319e-01 -1.66525930e-01 1.10778105e+00 1.13729894e-01 -3.57407361e-01 -7.19434619e-01 -6.99822426e-01 -3.69867951e-01 1.14879704e+00 -2.65581936e-01 4.53823917e-02 -5.72565913e-01 -1.90191805e-01 4.07981515e-01 2.72602111e-01 8.57847989e-01 -1.48751390e+00 -6.83431923e-02 5.00818849e-01 -7.60475636e-01 -1.17802131e+00 -7.66903758e-01 2.49555837e-02 -3.57096821e-01 -9.77708399e-01 -4.98344511e-01 -1.10310531e+00 2.44701669e-01 1.13105245e-01 1.00673330e+00 1.54415503e-01 2.31898710e-01 1.99145600e-01 -8.30304682e-01 -1.53793082e-01 -8.55802834e-01 1.02793528e-02 -3.50268513e-01 1.82940349e-01 6.56784773e-01 -2.13897303e-01 -4.84553963e-01 6.51991367e-01 -8.36160243e-01 2.12509115e-03 2.94411868e-01 1.03710878e+00 -5.32880425e-01 -4.21479583e-01 8.49861920e-01 -8.14481437e-01 1.51764691e+00 -6.88487947e-01 2.21971478e-02 2.52041012e-01 -4.62281853e-01 -1.44397587e-01 8.39358389e-01 -4.28012878e-01 -1.31331336e+00 -5.53103745e-01 -9.20746684e-01 3.48123729e-01 -3.55021179e-01 5.20312250e-01 -4.49212939e-01 5.29563487e-01 3.08753014e-01 3.34475160e-01 4.85718876e-01 -4.86134112e-01 9.92609560e-02 1.69955969e+00 1.85862333e-02 -3.75848085e-01 -2.76333224e-02 -4.52252999e-02 -1.04799581e+00 -9.72746193e-01 -3.70015264e-01 -8.33974898e-01 2.69908458e-02 -5.06770074e-01 1.18617964e+00 -4.39961225e-01 -1.45373309e+00 2.85145074e-01 -1.53887367e+00 5.25650643e-02 9.72603187e-02 4.59865481e-01 -3.77018809e-01 5.19282043e-01 -1.01226866e+00 -1.23785782e+00 -4.27498430e-01 -1.49365342e+00 7.46699929e-01 4.06740695e-01 -6.53144896e-01 -1.06021690e+00 -2.42177039e-01 8.07640195e-01 5.40102839e-01 -5.44639468e-01 1.19773626e+00 -1.20258129e+00 1.05440788e-01 -7.03660678e-03 -8.95328671e-02 6.97726369e-01 4.48011041e-01 -1.56469882e-01 -8.80805016e-01 3.20920408e-01 7.42645264e-01 -4.53618646e-01 6.87353760e-02 -1.14724874e-01 6.00227892e-01 -7.11692572e-01 1.39553830e-01 -2.85519540e-01 8.35049152e-01 7.25361109e-01 8.66025388e-01 3.31358202e-02 4.36988696e-02 1.19286907e+00 7.06712782e-01 2.39133373e-01 8.84458005e-01 8.46182227e-01 3.22657287e-01 1.48456857e-01 1.65135056e-01 -1.88208565e-01 6.67057753e-01 1.20729661e+00 4.38917160e-01 -3.64165515e-01 -6.64850473e-01 4.35004830e-01 -1.78564537e+00 -1.07234693e+00 -6.95654869e-01 1.54991984e+00 1.15942621e+00 -6.52965531e-03 6.95690662e-02 1.15146451e-01 6.63189292e-01 1.60530716e-01 -1.37549818e-01 -1.01381063e+00 6.77641109e-02 -1.16909117e-01 -1.93908229e-01 7.88532615e-01 -5.85677922e-01 8.65317345e-01 5.38370037e+00 6.89635217e-01 -9.70620751e-01 1.76360220e-01 5.24079204e-01 5.95675528e-01 -5.42582452e-01 1.32151425e-01 -8.20375085e-01 7.34963119e-01 8.95469844e-01 -1.51643947e-01 1.69701532e-01 7.73990452e-01 4.26391453e-01 -2.65109181e-01 -1.10362816e+00 8.87066543e-01 4.86094542e-02 -1.13266945e+00 -1.42141223e-01 -2.60167599e-01 -1.08432390e-01 -6.73211277e-01 -3.68766785e-01 8.62919807e-01 2.70995945e-01 -8.50179076e-01 3.01105827e-01 3.03152680e-01 2.33633429e-01 -2.78286368e-01 1.37534988e+00 8.33381474e-01 -1.08153844e+00 -2.23801155e-02 -2.73093849e-01 -3.08125287e-01 7.05927074e-01 2.15770185e-01 -1.21819675e+00 5.11819899e-01 5.59458554e-01 1.49057314e-01 1.40161356e-02 3.60998303e-01 -2.28992745e-01 7.74346530e-01 8.51585343e-03 -8.48704338e-01 1.91651776e-01 -6.40295088e-01 2.36817867e-01 1.11616206e+00 1.43786266e-01 4.59983826e-01 3.96585390e-02 7.55388379e-01 1.60165861e-01 6.04327202e-01 -4.82587695e-01 -2.05864191e-01 3.73822182e-01 1.36335599e+00 -1.87030002e-01 -4.03649718e-01 -5.20119667e-01 1.04628015e+00 -2.07531467e-01 1.75924242e-01 -8.93947899e-01 -5.49028456e-01 5.82585394e-01 -9.16078612e-02 -2.00127333e-01 9.86602828e-02 -1.02028824e-01 -7.85381675e-01 5.12621850e-02 -1.14904034e+00 1.33841127e-01 -1.02147603e+00 -1.32873237e+00 8.53404462e-01 -2.77063698e-01 -1.09560573e+00 -4.66833651e-01 -4.61660564e-01 -9.00423944e-01 1.14261270e+00 -1.10772228e+00 -8.62267375e-01 -2.11416438e-01 5.15459061e-01 1.03184819e+00 -1.45147189e-01 1.07318020e+00 5.25017917e-01 -2.88458228e-01 3.69352788e-01 -3.90578151e-01 4.42654729e-01 7.77914166e-01 -1.13194549e+00 -3.30525376e-02 -6.06982112e-02 -2.41571620e-01 6.67014718e-01 8.75034153e-01 -2.32690573e-01 -1.21959853e+00 -3.83596629e-01 1.34737480e+00 -2.61749297e-01 7.45368481e-01 -4.03517723e-01 -1.06708813e+00 4.53792304e-01 1.03194761e+00 -9.63679612e-01 8.94505799e-01 2.17618495e-01 1.17639937e-01 -5.94093576e-02 -1.17230320e+00 7.72341013e-01 3.67529958e-01 -7.07577109e-01 -1.22545135e+00 5.03067672e-01 1.04120922e+00 -3.03823620e-01 -7.82107234e-01 -2.70592365e-02 3.92986327e-01 -1.08721626e+00 4.27633733e-01 -7.51088858e-01 4.84099627e-01 1.62106641e-02 -1.68045968e-01 -1.18494153e+00 1.46796659e-01 -5.34592450e-01 4.80182558e-01 1.57867086e+00 6.35188699e-01 -9.21426654e-01 4.57061827e-01 7.99502313e-01 -2.54858285e-01 -5.23975194e-01 -7.80219674e-01 -3.19138378e-01 -1.37873152e-02 -2.30653241e-01 6.45311594e-01 9.84799743e-01 7.81414688e-01 1.11954868e+00 -4.25053567e-01 -2.72910446e-01 -3.93735528e-01 9.45682302e-02 7.72202790e-01 -9.04566050e-01 -2.38927603e-01 -3.15208197e-01 6.30394667e-02 -1.72962117e+00 6.86113583e-03 -4.78918910e-01 1.14773721e-01 -1.78874981e+00 -7.89687634e-02 -1.50299907e-01 4.61997241e-01 1.30808726e-01 -6.33390993e-03 -5.02819180e-01 1.93583459e-01 7.13072568e-02 -5.25871068e-02 6.49622560e-01 1.35626161e+00 -2.26139754e-01 -1.17589913e-01 2.81548560e-01 -6.96749151e-01 5.52265286e-01 8.32487106e-01 -7.24569112e-02 -3.31610829e-01 -1.23699307e-02 1.41900420e-01 7.49300659e-01 -1.64879456e-01 -3.30221921e-01 1.81425840e-01 -9.42123905e-02 -4.72703397e-01 -6.67212605e-01 7.37665951e-01 -1.09924281e+00 -4.05646652e-01 4.21454668e-01 -6.13544047e-01 2.21515298e-01 -2.26695359e-01 -2.22518910e-02 -6.61545992e-01 -7.30856717e-01 4.97384042e-01 -9.74779204e-02 -4.61811692e-01 -3.08101028e-01 -8.93383920e-01 1.76172540e-01 8.50228012e-01 -1.43215179e-01 -5.41804910e-01 -1.41672277e+00 -6.37152076e-01 6.75186098e-01 -2.44988024e-01 6.97349250e-01 5.99081814e-01 -1.20219433e+00 -6.99572504e-01 -1.57315787e-02 6.43464550e-02 -4.46700603e-01 3.55249107e-01 9.33854640e-01 -5.06133020e-01 4.91187304e-01 4.24886681e-02 -4.23119396e-01 -1.32914114e+00 -2.44222321e-02 2.84768730e-01 -2.13102803e-01 -1.45806402e-01 6.95080400e-01 6.00476153e-02 -8.86379302e-01 1.99807227e-01 -4.06395018e-01 -8.46853912e-01 3.15225482e-01 4.31155086e-01 2.00802281e-01 -1.48022532e-01 -5.52553654e-01 -1.91817269e-01 -9.81690362e-02 -1.95683613e-01 -4.17677730e-01 6.92716777e-01 -2.29047388e-01 -2.74121612e-01 7.63877690e-01 1.47416520e+00 1.17838547e-01 -4.04580534e-01 -1.35078430e-01 1.13732040e-01 -2.18650565e-01 -4.73639488e-01 -9.26796913e-01 -5.86525142e-01 1.09274209e+00 4.17819053e-01 1.29474592e+00 6.81848228e-01 2.50439823e-01 1.22125483e+00 3.70519817e-01 1.31388664e-01 -1.30119944e+00 2.29592174e-01 1.08284152e+00 1.33403111e+00 -1.40809238e+00 -5.63716948e-01 -3.63098741e-01 -1.32051313e+00 1.25524509e+00 9.19025779e-01 3.05285722e-01 6.29489541e-01 2.06496760e-01 5.94174922e-01 -2.56081760e-01 -1.06565714e+00 -1.70842096e-01 -1.42049909e-01 3.07960302e-01 8.15242410e-01 2.82428926e-03 -8.10127258e-01 1.08643997e+00 -5.67588091e-01 -4.66599584e-01 6.69839799e-01 5.73824227e-01 -6.27157986e-01 -1.35931361e+00 -2.32822165e-01 5.08115232e-01 -3.80120277e-01 -1.31583959e-01 -7.01566994e-01 7.48324573e-01 -2.61493266e-01 1.77459311e+00 1.29059270e-01 -6.18766129e-01 5.60301125e-01 4.99147743e-01 -2.50710279e-01 -8.11952710e-01 -1.08766091e+00 -4.35189204e-03 7.71904171e-01 -2.87433505e-01 -2.56692797e-01 -3.85651588e-01 -1.47270048e+00 -3.23497951e-01 -6.69007838e-01 7.74465144e-01 7.16408253e-01 1.08313537e+00 2.82154847e-02 4.35442239e-01 9.51767683e-01 -4.52161487e-03 -1.03445542e+00 -1.57144785e+00 -4.34056908e-01 5.01527309e-01 1.36071676e-02 -3.31124276e-01 -6.54432654e-01 -2.02768072e-01]
[12.761570930480957, 7.736016750335693]
806ece4d-837b-4bb6-9b6a-20b96f0990b1
subgraph-neighboring-relations-infomax-for
2208.00850
null
https://arxiv.org/abs/2208.00850v3
https://arxiv.org/pdf/2208.00850v3.pdf
Subgraph Neighboring Relations Infomax for Inductive Link Prediction on Knowledge Graphs
Inductive link prediction for knowledge graph aims at predicting missing links between unseen entities, those not shown in training stage. Most previous works learn entity-specific embeddings of entities, which cannot handle unseen entities. Recent several methods utilize enclosing subgraph to obtain inductive ability. However, all these works only consider the enclosing part of subgraph without complete neighboring relations, which leads to the issue that partial neighboring relations are neglected, and sparse subgraphs are hard to be handled. To address that, we propose Subgraph Neighboring Relations Infomax, SNRI, which sufficiently exploits complete neighboring relations from two aspects: neighboring relational feature for node feature and neighboring relational path for sparse subgraph. To further model neighboring relations in a global way, we innovatively apply mutual information (MI) maximization for knowledge graph. Experiments show that SNRI outperforms existing state-of-art methods by a large margin on inductive link prediction task, and verify the effectiveness of exploring complete neighboring relations in a global way to characterize node features and reason on sparse subgraphs.
['Chaoyang Yan', 'Chengpeng Chao', 'Yongquan He', 'Peng Zhang', 'Xiaohan Xu']
2022-07-28
null
null
null
null
['inductive-link-prediction']
['graphs']
[-3.27453136e-01 7.20468283e-01 -7.64866889e-01 -3.18553567e-01 2.30102614e-02 -4.47316349e-01 4.31772679e-01 3.88826758e-01 3.46156470e-02 9.20039892e-01 3.87823254e-01 -1.37965500e-01 -5.77713847e-01 -1.33398342e+00 -7.01336324e-01 -3.71602297e-01 -4.79566455e-01 3.51105779e-01 3.29566628e-01 -1.59086093e-01 -3.42044771e-01 4.12465543e-01 -1.20146656e+00 -2.73719802e-02 1.02141535e+00 4.24693882e-01 -1.50309786e-01 -5.57579249e-02 -3.46500218e-01 8.11443031e-01 1.02004101e-02 -5.80513656e-01 -3.51617709e-02 1.15416888e-02 -1.04166186e+00 -3.64014268e-01 8.27373266e-02 -1.25105277e-01 -9.18036044e-01 8.44003856e-01 2.56765515e-01 9.32237804e-02 6.85501873e-01 -1.35185325e+00 -8.63130748e-01 1.29592133e+00 -5.05350351e-01 3.25535655e-01 7.03742564e-01 -4.60938096e-01 1.64203441e+00 -9.32807684e-01 9.19008911e-01 1.13848972e+00 7.05458939e-01 2.66419910e-02 -1.04962814e+00 -8.05657983e-01 6.31159186e-01 2.40935653e-01 -1.99676299e+00 -1.11905695e-03 8.22552264e-01 -1.72412097e-01 9.40077305e-01 2.59645253e-01 5.77534020e-01 8.94970953e-01 -1.55671760e-01 7.68926978e-01 4.50195432e-01 2.21852269e-02 -3.42727244e-01 4.62836355e-01 3.02567840e-01 8.14643621e-01 6.66794181e-01 -2.76049338e-02 -4.62864697e-01 -1.06715798e-01 5.45632184e-01 1.19725361e-01 -5.52618146e-01 -6.75077975e-01 -1.36430085e+00 6.89181447e-01 1.03683376e+00 5.71150839e-01 -1.99445844e-01 -1.66847780e-01 1.37686403e-02 4.11661446e-01 4.29966003e-01 4.57136571e-01 -7.15755820e-01 3.71482074e-01 -6.42963648e-01 -2.91068256e-01 1.02981758e+00 1.34124601e+00 1.20084321e+00 -2.66842335e-01 -1.31317079e-01 4.15959746e-01 5.91530323e-01 2.82470524e-01 1.22828506e-01 -2.40200624e-01 6.59438848e-01 1.14343715e+00 -2.63059974e-01 -1.53531766e+00 -5.49754262e-01 -8.05606782e-01 -8.63971829e-01 -8.65828097e-01 -1.00811198e-01 -2.64885932e-01 -6.51114941e-01 1.65967274e+00 6.42306745e-01 4.59276974e-01 1.51017845e-01 8.05748880e-01 1.61899745e+00 5.85115552e-01 1.80405721e-01 -2.82305986e-01 9.62707996e-01 -7.84596264e-01 -5.82635522e-01 -1.16743054e-02 1.24308622e+00 -4.08538133e-01 4.49672908e-01 -1.67137995e-01 -6.49195313e-01 -3.67788166e-01 -6.98234200e-01 -5.11097759e-02 -7.40682006e-01 1.02621935e-01 1.14323843e+00 2.64922678e-01 -9.86371696e-01 5.20822525e-01 -5.30522168e-01 -4.68494147e-01 3.00343335e-01 5.62818527e-01 -7.82143414e-01 -9.04497057e-02 -1.79482651e+00 6.92321956e-01 9.18107986e-01 2.24565566e-01 -4.76010889e-01 -8.62825274e-01 -1.24891162e+00 4.01389688e-01 6.80278897e-01 -6.99371517e-01 1.10510767e-01 -2.74234891e-01 -7.49714315e-01 5.58169723e-01 -2.34057024e-01 -1.13803998e-01 3.24824490e-02 -1.77366380e-02 -9.48288023e-01 -8.26819465e-02 4.85416390e-02 6.74799740e-01 2.98264474e-01 -1.35629439e+00 -3.73440325e-01 -2.45930493e-01 2.34132797e-01 3.04281384e-01 -9.06038463e-01 -7.52085865e-01 -7.10677505e-01 -5.06681800e-01 5.61686397e-01 -6.01695418e-01 -9.10470709e-02 -5.46484351e-01 -9.72022355e-01 -5.09479761e-01 8.45937192e-01 -3.50972414e-01 1.77874625e+00 -1.98601031e+00 1.65842026e-01 8.11781943e-01 6.18210733e-01 1.50737107e-01 -1.86995164e-01 5.99584579e-01 -2.33333901e-01 3.22017878e-01 2.06298769e-01 1.08013935e-01 -1.33168116e-01 3.65648091e-01 -1.96099922e-01 2.31004909e-01 1.56396523e-01 1.21082079e+00 -1.04343283e+00 -8.10242891e-01 -5.96568659e-02 4.59524155e-01 -4.82563496e-01 1.29936179e-02 5.97146573e-03 3.42534721e-01 -1.10162103e+00 7.02451646e-01 8.31088066e-01 -4.83810276e-01 5.30705214e-01 -5.76760352e-01 2.16056526e-01 1.92843586e-01 -1.30856371e+00 1.54782295e+00 -3.08242470e-01 2.79720277e-01 -2.88471282e-01 -1.06643867e+00 1.05142355e+00 1.61949873e-01 6.14236474e-01 -3.06857616e-01 -4.27098386e-02 3.62925678e-02 -1.72511742e-01 -5.50201714e-01 4.28124428e-01 3.82423520e-01 1.26870006e-01 6.06328100e-02 2.46095687e-01 5.68968832e-01 2.74594963e-01 8.50973070e-01 1.39635706e+00 4.43547815e-02 2.22780362e-01 -1.87243208e-01 6.03712559e-01 -1.27270937e-01 7.23835111e-01 4.31941211e-01 1.73736528e-01 3.34995210e-01 6.97706699e-01 -2.07399830e-01 -4.34107691e-01 -1.20827162e+00 -1.76732987e-01 6.92236066e-01 6.86824262e-01 -8.08356524e-01 -7.90341198e-02 -1.21222579e+00 3.69777292e-01 4.62298155e-01 -7.95543849e-01 -3.32851022e-01 -3.75023723e-01 -5.73748291e-01 2.62208372e-01 5.88815749e-01 2.44176015e-01 -7.31365025e-01 6.82007492e-01 6.55519962e-02 -5.05849235e-02 -1.28599393e+00 -2.09853202e-01 5.11626760e-03 -8.76566172e-01 -1.30732369e+00 -3.16177040e-01 -1.03037024e+00 1.06764293e+00 2.78599799e-01 1.23224282e+00 3.51445317e-01 -1.41405210e-01 4.02821809e-01 -6.05393350e-01 2.77552277e-01 2.78324783e-01 5.18890858e-01 -2.10047001e-03 -4.18668725e-02 5.63371539e-01 -8.97957087e-01 -5.56287944e-01 3.13394994e-01 -6.25779688e-01 -1.80267379e-01 8.61901999e-01 6.50630534e-01 5.34994245e-01 4.01885360e-01 6.96931303e-01 -1.30743742e+00 2.12172911e-01 -1.05404854e+00 -1.88807636e-01 6.45282030e-01 -7.05924571e-01 2.55176187e-01 5.08503854e-01 -1.86563268e-01 -1.01963162e+00 -1.32980436e-01 1.24936491e-01 -2.88752258e-01 8.40070397e-02 9.44967031e-01 -4.15039182e-01 -9.09463093e-02 3.32137316e-01 3.88857611e-02 -5.46142220e-01 -3.13515306e-01 5.13448238e-01 2.55863726e-01 3.38938311e-02 -4.00678724e-01 1.28080559e+00 3.70901614e-01 2.86123544e-01 -5.51678479e-01 -8.86957169e-01 -7.86620557e-01 -7.24952281e-01 1.41618222e-01 6.01448536e-01 -1.17928851e+00 -7.78189242e-01 -4.58191067e-01 -8.93918991e-01 2.14197204e-01 -3.81047986e-02 7.95323610e-01 2.65838206e-01 4.65846688e-01 -4.41192806e-01 -5.23224235e-01 -1.04704417e-01 -5.64311266e-01 7.00481176e-01 2.38132924e-01 -9.28254500e-02 -1.28038609e+00 2.43277811e-02 1.73218131e-01 1.05160058e-01 2.80926138e-01 9.34925258e-01 -1.03857136e+00 -8.91013741e-01 -1.71805695e-01 -5.67789018e-01 -1.03126980e-01 1.88402936e-01 -3.24885137e-02 -4.99501199e-01 -2.60938585e-01 -9.36707139e-01 -2.16811836e-01 1.24095917e+00 -8.95990282e-02 7.92295575e-01 -2.94312507e-01 -1.15776610e+00 5.93914747e-01 1.32375157e+00 -3.26741189e-01 5.77214360e-01 1.83312576e-02 1.20721388e+00 6.38002276e-01 7.18515754e-01 2.65479296e-01 9.42683935e-01 2.35420361e-01 4.33406204e-01 -2.06343576e-01 3.19350287e-02 -7.17348754e-01 1.20897386e-02 9.36112583e-01 -8.18585455e-02 -4.95223105e-01 -6.75723374e-01 8.53187621e-01 -1.82685471e+00 -7.64994204e-01 -4.00577754e-01 1.99946725e+00 7.98779726e-01 1.05027653e-01 -2.28812143e-01 -1.86944634e-01 9.14925992e-01 1.47932470e-01 -3.06940079e-01 2.14783356e-01 -1.54654413e-01 -1.74647808e-01 4.06575143e-01 4.10018682e-01 -9.72973406e-01 1.13739944e+00 5.00741625e+00 9.40328956e-01 -3.86409283e-01 -3.98137048e-02 2.47011885e-01 2.53083169e-01 -1.06256258e+00 5.14445305e-01 -1.03362930e+00 2.18939781e-01 3.54805261e-01 -1.18016869e-01 3.41884382e-02 8.52383494e-01 -2.47779772e-01 1.45285770e-01 -1.11530721e+00 6.81769133e-01 -1.36629075e-01 -1.04915154e+00 2.31181666e-01 9.74869877e-02 1.08167207e+00 2.36604013e-04 -1.73535526e-01 6.65170133e-01 4.76690233e-01 -9.74597335e-01 -2.37727016e-01 7.77743101e-01 4.03378904e-01 -7.81283081e-01 8.97625685e-01 1.41161442e-01 -1.76699269e+00 1.47728458e-01 -5.73842406e-01 3.09426904e-01 1.28462180e-01 1.08506930e+00 -8.24435234e-01 1.23717356e+00 5.58482409e-01 1.09733725e+00 -6.52963996e-01 9.19999719e-01 -4.45261687e-01 5.77870846e-01 -6.10630095e-01 -1.33989066e-01 2.04369038e-01 -1.15967922e-01 5.35773993e-01 1.02466667e+00 4.09958452e-01 2.22301573e-01 3.82438481e-01 9.44126070e-01 -5.01071393e-01 3.94812137e-01 -7.58886456e-01 -1.49097785e-01 8.51467490e-01 1.38575661e+00 -5.83597064e-01 -3.08114458e-02 -7.15440750e-01 7.74938703e-01 6.67784989e-01 6.06941283e-01 -8.45643222e-01 -5.06288946e-01 2.93084174e-01 1.39163032e-01 4.90011513e-01 -4.18081693e-02 2.23721847e-01 -1.46403301e+00 1.96797043e-01 -7.56577775e-02 6.84013009e-01 -4.05590326e-01 -1.32277298e+00 3.81982893e-01 5.89208771e-03 -1.20762634e+00 2.12659448e-01 -1.06789105e-01 -6.90085113e-01 4.64410514e-01 -1.92362893e+00 -1.46184838e+00 -4.29099947e-01 6.60241187e-01 -3.61427307e-01 3.66530404e-03 6.69971168e-01 4.32081491e-01 -7.22506523e-01 7.63056636e-01 3.91759686e-02 3.46967757e-01 5.80520988e-01 -1.15559697e+00 -7.92245343e-02 5.10742843e-01 4.96969312e-01 9.08670068e-01 2.92080909e-01 -1.02423275e+00 -1.42712998e+00 -1.23838282e+00 1.32819331e+00 -3.25124890e-01 7.95597196e-01 -1.12980358e-01 -1.00221670e+00 9.85116720e-01 -1.27104864e-01 5.28455734e-01 8.05322647e-01 8.76780689e-01 -4.72261637e-01 -1.77408278e-01 -9.39367652e-01 5.25067925e-01 1.54023814e+00 -3.91939312e-01 -5.98443687e-01 3.13906133e-01 9.42876041e-01 -8.66655037e-02 -1.53470647e+00 9.47558761e-01 2.24161953e-01 -5.93326032e-01 1.09017313e+00 -5.34407794e-01 2.36409158e-01 -5.44075489e-01 2.53916353e-01 -1.18161941e+00 -4.57611978e-01 -4.22730654e-01 -6.28012061e-01 1.68061686e+00 9.69934523e-01 -7.16736794e-01 1.02914214e+00 3.72507244e-01 7.28547852e-03 -8.69536877e-01 -6.25142694e-01 -7.43523121e-01 -2.77936697e-01 -1.01468777e-02 7.50639081e-01 1.48090386e+00 5.08301735e-01 5.66187143e-01 -3.00493777e-01 6.54171646e-01 4.09960955e-01 4.37427998e-01 5.69619477e-01 -1.51047552e+00 -9.79309306e-02 -1.82331949e-02 -8.46784174e-01 -1.20468163e+00 4.97183055e-01 -1.16370940e+00 -5.43675065e-01 -1.85291791e+00 3.57960075e-01 -8.16829264e-01 -4.79547739e-01 5.09749591e-01 -3.98682684e-01 6.91316202e-02 -3.56264621e-01 1.51089191e-01 -1.04945910e+00 9.13391411e-01 1.35147476e+00 -2.35416025e-01 -3.03203553e-01 -1.85408607e-01 -6.96443617e-01 6.43455684e-01 4.47189242e-01 -2.90860981e-01 -7.27794766e-01 -2.96616912e-01 7.02721953e-01 -7.47243315e-02 2.51161188e-01 -8.34830582e-01 5.52122951e-01 1.01817369e-01 4.12823617e-01 -6.48884416e-01 8.70928466e-02 -1.03640568e+00 2.82530189e-01 -3.25806104e-02 -2.72589892e-01 -5.49181044e-01 -2.63672203e-01 1.00959504e+00 -4.90789294e-01 -5.43954931e-02 -2.11280137e-02 1.56844899e-01 -9.83513951e-01 7.35411704e-01 3.49421918e-01 1.68975160e-01 1.15284288e+00 -1.36624813e-01 -3.18614125e-01 -2.34525427e-01 -1.26222408e+00 8.94437909e-01 5.60022816e-02 3.69367957e-01 5.92014551e-01 -1.62662542e+00 -6.29809499e-01 8.56586695e-02 4.41914529e-01 1.39413953e-01 4.31266367e-01 9.97839272e-01 -2.69983280e-02 4.39126462e-01 5.32655716e-01 -2.47228086e-01 -1.04010713e+00 7.67671347e-01 -5.14442660e-02 -5.50448060e-01 -5.08264601e-01 1.07781279e+00 2.35966504e-01 -6.58922195e-01 8.88367817e-02 -2.57004797e-02 -6.47451699e-01 1.41575888e-01 -9.81004462e-02 2.02064067e-01 -1.81175113e-01 -7.61096597e-01 -5.49711466e-01 5.91047525e-01 -3.63605827e-01 7.61616230e-01 1.18041146e+00 -3.82151902e-01 -1.85171157e-01 4.43567373e-02 1.44987810e+00 2.46951714e-01 -6.88750744e-01 -6.05248809e-01 1.31244240e-02 -4.49271530e-01 6.36546761e-02 -3.82817775e-01 -1.38202894e+00 4.63422805e-01 -9.87063721e-02 3.42459679e-01 8.21201265e-01 4.53123987e-01 8.24292362e-01 6.71112657e-01 3.71396780e-01 -8.06607127e-01 -2.74077922e-01 4.20659900e-01 3.99148643e-01 -1.27605009e+00 3.63278538e-01 -1.35075796e+00 -3.66428733e-01 9.84848320e-01 1.00891984e+00 2.02600643e-01 1.18648767e+00 -7.98555762e-02 -6.09752536e-01 -5.74465096e-01 -5.94491005e-01 -7.13108778e-01 6.55839503e-01 4.92558777e-01 5.05432904e-01 6.28246441e-02 -3.89399558e-01 5.65297604e-01 -3.03495172e-02 -4.50864732e-01 2.40165088e-02 6.34952843e-01 -3.39557767e-01 -1.18727875e+00 2.46521711e-01 6.68499529e-01 -2.10333541e-01 -4.14673388e-01 -3.89819562e-01 1.00311565e+00 3.06840032e-01 8.45171273e-01 -1.26467571e-01 -6.70235574e-01 1.92514449e-01 -2.74450630e-01 1.13770187e-01 -6.49498343e-01 -1.69527471e-01 -2.97492862e-01 3.55011374e-01 -3.19099098e-01 -4.22776878e-01 -2.84133196e-01 -1.38463473e+00 -3.79829377e-01 -9.54366922e-01 5.85807621e-01 -1.05277658e-01 9.38002408e-01 6.19515479e-01 5.49255550e-01 7.19515920e-01 -1.54851869e-01 6.35412186e-02 -8.82405400e-01 -9.52727139e-01 4.79145646e-01 -3.98301408e-02 -8.13505590e-01 -4.97727275e-01 -4.94526803e-01]
[8.776371002197266, 7.952267646789551]
ef87eb7f-071a-4da4-97b3-3dd382e0bcd6
toward-sensor-based-sleep-monitoring-with
1901.11440
null
http://arxiv.org/abs/1901.11440v1
http://arxiv.org/pdf/1901.11440v1.pdf
Toward Sensor-based Sleep Monitoring with Electrodermal Activity Measures
We use self-report and electrodermal activity (EDA) wearable sensor data from 77 nights of sleep on six participants to test the efficacy of EDA data for sleep monitoring. We used factor analysis to find latent factors in the EDA data, and causal model search to find the most probable graphical model accounting for self-reported sleep efficiency (SE), sleep quality (SQ), and the latent EDA factors. Structural equation modeling was used to confirm fit of the extracted graph. Based on the generated graph, logistic regression and naive Bayes models were used to test the efficacy of the EDA data in predicting SE and SQ. Six EDA features extracted from the total signal over a night's sleep could be explained by two latent factors, EDA Magnitude and EDA Storms. EDA Magnitude performed as a strong predictor for SE to aid detection of substantial changes in time asleep. The performance of EDA Magnitured and SE in classifying SQ showed promise for wearable sleep monitoring applications. However, our data suggest that obtaining a more accurate sensor-based measure of SE will be necessary before smaller changes in SQ can be detected from EDA sensor data alone.
['Tanvi Banerjee', 'Garrett Goodman', 'William Romine']
2019-01-31
null
null
null
null
['sleep-quality-prediction']
['medical']
[-1.19769223e-01 -1.20427810e-01 -2.29228809e-01 -6.46590829e-01 -2.94873536e-01 -3.37644041e-01 -1.17157940e-02 4.41988140e-01 -4.29154336e-01 5.31265199e-01 7.73692071e-01 -5.81696510e-01 -5.00167191e-01 -5.57680726e-01 -1.76584497e-01 -2.99492478e-01 -6.08436346e-01 -1.87327608e-01 -1.28431901e-01 1.91217382e-02 1.00236624e-01 2.71638542e-01 -1.20135534e+00 -1.81802735e-01 8.71714890e-01 1.01603794e+00 2.65693665e-01 3.67583394e-01 2.58412451e-01 3.31121832e-01 -8.34409356e-01 2.11542681e-01 -1.14651723e-03 -8.44424546e-01 -1.25532057e-02 -9.92456153e-02 3.83612700e-02 -2.07415894e-01 -1.19498350e-01 3.50350261e-01 6.83715701e-01 8.71217921e-02 3.30002099e-01 -1.40159082e+00 6.00042082e-02 4.05006856e-01 -3.40260953e-01 1.00919974e+00 8.36438239e-01 2.11108401e-01 8.94096494e-01 -4.47834849e-01 7.00573549e-02 8.50494087e-01 9.95056152e-01 2.08573088e-01 -1.69175923e+00 -1.11915588e+00 -5.03252447e-01 5.74547723e-02 -1.32853186e+00 -8.07550550e-01 6.90193534e-01 -2.75051028e-01 1.34261477e+00 4.33943123e-01 1.51753235e+00 8.49362135e-01 9.34715331e-01 -3.17391418e-02 1.59743011e+00 -8.34970549e-02 5.53672731e-01 1.86837278e-02 3.88107240e-01 4.84528273e-01 6.34560645e-01 6.92803860e-02 -1.07112193e+00 -5.60763419e-01 2.28395283e-01 1.55534714e-01 -1.78249832e-02 3.56143177e-01 -7.16113746e-01 7.48211920e-01 2.57723451e-01 3.67945552e-01 -5.70643008e-01 -1.40421495e-01 -1.94485113e-02 1.17047355e-01 3.70447338e-01 6.69747174e-01 -1.66268826e-01 -5.65024555e-01 -1.59360635e+00 -1.40069991e-01 5.83470285e-01 9.91800055e-02 4.42674458e-01 9.83787999e-02 -1.85557842e-01 5.61396480e-01 6.10813439e-01 7.21329272e-01 6.51187360e-01 -8.26490760e-01 -1.91519648e-01 7.72827387e-01 5.04588336e-02 -1.31271315e+00 -1.29182684e+00 -5.24722278e-01 -3.93800467e-01 -2.01703742e-01 -4.38564317e-03 -6.39008507e-02 -5.66112399e-01 1.50064051e+00 -2.95045018e-01 -5.24966531e-02 -4.20708209e-01 6.12810075e-01 5.51789701e-01 4.17269349e-01 2.12957636e-01 -8.36106241e-01 1.45357943e+00 1.19565390e-02 -1.03154767e+00 -7.39932001e-01 3.12538475e-01 -3.89139920e-01 9.68614221e-01 2.24761277e-01 -1.12960708e+00 -5.16135395e-01 -1.42103779e+00 2.12791577e-01 2.71313954e-02 8.93191900e-03 4.52682853e-01 8.31103444e-01 -1.23612595e+00 5.53867280e-01 -1.55929983e+00 -6.62777126e-01 3.97848696e-01 6.63034976e-01 -9.90017653e-02 1.99435174e-01 -9.22256768e-01 8.68578136e-01 -2.33787164e-01 1.66270167e-01 -6.00035489e-01 -3.96436572e-01 -6.72256172e-01 1.94868878e-01 -2.64050871e-01 -7.54121661e-01 6.26558542e-01 -3.79650474e-01 -8.55901599e-01 5.08642852e-01 -6.13321245e-01 -4.83749509e-01 -3.77116382e-01 1.81390792e-01 -8.58762085e-01 9.68704149e-02 4.08862382e-01 -2.71815043e-02 8.56389403e-01 -5.86543262e-01 -1.02369450e-01 -1.00564504e+00 -5.45379341e-01 1.50265694e-02 -3.22007626e-01 1.38779301e-02 4.69210774e-01 -5.01735359e-02 5.08713007e-01 -8.38726223e-01 4.06212956e-02 -3.51051480e-01 -6.00907318e-02 -1.77424192e-01 2.32089207e-01 -8.51570129e-01 1.75910830e+00 -2.31916523e+00 -6.65487707e-01 4.67283279e-01 7.15819657e-01 -3.73171538e-01 1.63774207e-01 4.85150665e-01 2.64791280e-01 1.43401533e-01 4.51212786e-02 -3.68751913e-01 -8.58708657e-03 2.13141605e-01 6.63198158e-02 8.12016428e-01 -9.38850716e-02 9.18471038e-01 -7.63883233e-01 -2.42422998e-01 1.34284198e-01 3.69950622e-01 -2.74884731e-01 2.57117361e-01 4.66055810e-01 1.18952990e-01 2.68136454e-03 2.91834295e-01 3.05546194e-01 -2.90578067e-01 1.91931799e-01 -1.09122373e-01 -4.00693387e-01 9.84244645e-01 -5.18688500e-01 1.44617951e+00 -1.55303642e-01 7.28225052e-01 -4.43993434e-02 -2.24433497e-01 1.28795755e+00 -2.64284294e-02 5.70261776e-01 -1.44370890e+00 4.52406481e-02 2.80615482e-02 4.58192557e-01 -5.25239587e-01 2.71574855e-01 -5.13967872e-01 1.02714412e-01 6.92884266e-01 -1.03467129e-01 2.37794235e-01 -2.05942951e-02 2.41560236e-01 1.69163382e+00 -3.85850847e-01 5.48292160e-01 -6.12774134e-01 -3.41958165e-01 -1.31021097e-01 5.77566147e-01 7.70458102e-01 -3.57718915e-01 -1.03519605e-02 3.02705199e-01 -1.40116617e-01 -5.76202750e-01 -1.07355249e+00 -1.93525016e-01 6.22762442e-01 5.21628447e-02 -1.10591483e+00 -1.73586547e-01 -1.07069924e-01 -1.48295360e-02 1.13131332e+00 -5.78716397e-01 -6.53676093e-01 2.99863547e-01 -1.13905334e+00 2.90667385e-01 5.65577924e-01 3.51106763e-01 -6.17900193e-01 -1.01866508e+00 3.07777226e-01 -2.12235421e-01 -6.82254791e-01 -2.97920972e-01 5.49471796e-01 -1.07599020e+00 -9.96294975e-01 1.89423546e-01 -2.98805232e-03 3.02649945e-01 5.72718024e-01 9.20618296e-01 -4.51433621e-02 -5.25258780e-01 7.90793478e-01 1.06468685e-01 -2.34042838e-01 6.44427836e-02 -4.27792937e-01 4.46381122e-01 -1.70371220e-01 7.20070720e-01 -1.30441153e+00 -1.19615746e+00 4.47875410e-01 -2.79367626e-01 -3.08671445e-01 5.09177923e-01 1.41875938e-01 4.61326420e-01 2.31589645e-01 5.63772798e-01 5.41385263e-02 1.10548234e+00 -8.21092725e-01 -3.53965648e-02 -3.60393256e-01 -1.50940454e+00 -3.20394039e-01 1.06901325e-01 -1.72845349e-01 -5.76885045e-01 -3.82176727e-01 7.35971183e-02 5.31966202e-02 -8.56845081e-02 5.52512288e-01 4.65885960e-02 3.41562271e-01 9.23558652e-01 2.61680871e-01 1.27994165e-01 -2.07449660e-01 -5.01548648e-01 6.92557931e-01 3.18371415e-01 1.66876957e-01 1.71505705e-01 5.04582047e-01 1.80002630e-01 -1.21398795e+00 -6.66765749e-01 -8.19688380e-01 -3.53936702e-01 -2.50112951e-01 1.10694087e+00 -1.11090338e+00 -9.20705676e-01 -2.04893932e-01 -4.13249284e-01 -3.14638317e-01 -1.10772528e-01 9.65221524e-01 -1.88147813e-01 -1.64197609e-01 1.16121151e-01 -1.17214549e+00 -5.14589965e-01 -6.53594196e-01 5.65408707e-01 3.71086866e-01 -1.10186231e+00 -8.37606251e-01 4.93983835e-01 4.56240863e-01 5.87601066e-01 2.29749277e-01 5.86639583e-01 -6.00546002e-01 -7.24927261e-02 -2.49997169e-01 3.27382594e-01 -2.13157013e-02 6.14201784e-01 -3.62571716e-01 -8.10119927e-01 -3.83485228e-01 5.72467804e-01 1.36004284e-03 2.92669624e-01 6.23438776e-01 4.27429825e-01 -4.13588166e-01 -3.52495700e-01 3.28561395e-01 1.02320778e+00 3.68101805e-01 4.51270491e-01 3.82196486e-01 2.87577450e-01 2.67334525e-02 1.44526158e-02 5.58259070e-01 6.75099790e-01 6.45498693e-01 1.19124480e-01 6.50344566e-02 -1.22364894e-01 -3.06098908e-01 8.24961841e-01 8.62229526e-01 7.41872936e-02 1.58855692e-01 -7.27134943e-01 3.61685634e-01 -1.11501062e+00 -9.21978652e-01 -6.35288894e-01 2.28018951e+00 1.64511532e-01 4.48694795e-01 7.84504175e-01 1.41456202e-01 7.43668079e-02 1.29322916e-01 -6.58142745e-01 -5.34726858e-01 -6.98490143e-02 2.53053278e-01 5.67337036e-01 5.78882620e-02 -1.98317971e-02 -1.09696239e-01 7.55353785e+00 -5.71425185e-02 -8.70315671e-01 1.41131461e-01 1.91322997e-01 -6.60010040e-01 -4.42520827e-01 2.67490268e-01 -6.38717949e-01 8.71615827e-01 1.73723221e+00 -3.36750060e-01 6.77411020e-01 4.45769221e-01 1.30908275e+00 -7.81494200e-01 -7.20483482e-01 1.08356929e+00 5.35069481e-02 -8.52317989e-01 -9.01624024e-01 4.48511899e-01 1.49116516e-01 2.05990478e-01 -3.52198124e-01 1.06499018e-02 -1.25766829e-01 -8.31824124e-01 2.92763889e-01 8.63954127e-01 6.82679534e-01 -2.70915419e-01 2.74819553e-01 2.35041186e-01 -9.00435507e-01 -2.94081509e-01 -2.44308077e-02 -7.09090114e-01 -3.88549827e-02 6.34961784e-01 -1.19540954e+00 -1.51656419e-01 9.51289356e-01 6.98190689e-01 -1.05617249e+00 9.51674581e-01 -2.20981851e-01 1.42976689e+00 -6.82672262e-01 -8.34786072e-02 -2.37197936e-01 -9.09564644e-02 5.22062123e-01 8.41676593e-01 3.27955246e-01 1.79422796e-01 -3.81591111e-01 1.22370863e+00 2.49421433e-01 -1.34910956e-01 -4.81939256e-01 -3.68306279e-01 5.38299978e-01 1.28961563e+00 -1.15050483e+00 1.48207396e-01 -6.21504724e-01 6.33009613e-01 -1.78044349e-01 -5.47642994e-04 -3.59629065e-01 1.51168806e-02 5.01511276e-01 7.93881238e-01 -1.88878089e-01 -5.71716964e-01 -5.49975812e-01 -8.68989885e-01 -1.72652990e-01 -4.41857338e-01 4.86862391e-01 -1.14649355e+00 -1.06204832e+00 2.47371510e-01 -9.40511897e-02 -7.97177613e-01 7.90987629e-03 2.45724812e-01 -1.03385830e+00 8.70054662e-01 -7.63669848e-01 -4.75341469e-01 -5.44068992e-01 6.35638773e-01 1.25546560e-01 2.88155764e-01 7.78157413e-01 -8.36002734e-03 -5.30466974e-01 2.68480748e-01 -1.46124527e-01 -4.92344439e-01 6.08183265e-01 -1.24461317e+00 1.46745071e-01 5.98946750e-01 1.48236245e-01 1.02608693e+00 8.08512330e-01 -1.03424644e+00 -1.55693984e+00 -5.05172074e-01 1.03541720e+00 -5.98285675e-01 7.55085528e-01 -4.51147288e-01 -5.22231758e-01 6.15525186e-01 1.43680535e-02 -5.30254900e-01 1.28382742e+00 3.88874561e-01 3.06943327e-01 -4.85414445e-01 -9.91481245e-01 1.00840971e-01 8.68438482e-01 -5.35101235e-01 -7.60389328e-01 -2.46350005e-01 4.95853461e-02 3.02102059e-01 -1.21800482e+00 1.18388638e-01 8.88111353e-01 -1.14583540e+00 4.53467041e-01 -7.81733990e-02 -8.31239522e-02 -1.48249120e-01 1.53904453e-01 -1.34896147e+00 -5.54703772e-01 -6.70713961e-01 -1.19380038e-02 1.07920027e+00 2.96242058e-01 -4.81305152e-01 4.86901969e-01 1.19561160e+00 -2.40935355e-01 -3.95569861e-01 -8.72157812e-01 -5.97154081e-01 -7.02538490e-01 -5.06423533e-01 3.40576559e-01 6.02467120e-01 3.13748240e-01 8.13917041e-01 -2.00949356e-01 1.15382321e-01 4.53930229e-01 -5.09626158e-02 3.59633118e-01 -1.24850392e+00 -1.48231328e-01 1.90701500e-01 -6.75317645e-01 -2.97047555e-01 -3.72515559e-01 -1.00709987e+00 -1.91065893e-01 -2.06547022e+00 3.03019136e-01 -9.37815905e-02 -4.25005436e-01 7.08164811e-01 8.29416290e-02 2.15043738e-01 -3.63899320e-02 2.07353204e-01 -5.28130949e-01 4.25980300e-01 4.64976907e-01 1.77271768e-01 -1.02859175e+00 2.37033591e-01 -9.34175372e-01 3.97321433e-01 9.62308705e-01 -7.19770491e-01 -8.30677927e-01 1.19941503e-01 5.75462699e-01 3.67223382e-01 4.87491608e-01 -1.15302694e+00 3.01069707e-01 1.38223439e-01 9.53310132e-01 -4.68473166e-01 4.28211242e-01 -8.06678772e-01 4.14995760e-01 3.60710680e-01 -3.82340513e-02 2.81445324e-01 5.38773954e-01 3.33296955e-01 4.03253078e-01 3.32738817e-01 2.50467300e-01 1.71627671e-01 -1.46590680e-01 -1.46605566e-01 -8.35750639e-01 -1.95081607e-01 3.74784410e-01 -5.05207837e-01 -2.32791379e-01 -5.90232790e-01 -1.12724507e+00 2.20281258e-01 2.30437174e-01 2.53251731e-01 7.02818274e-01 -1.01865458e+00 -1.51506558e-01 4.05202031e-01 -8.99273306e-02 -8.76115263e-01 3.55560809e-01 1.40705872e+00 2.20786199e-01 2.61184245e-01 -5.24502397e-01 -4.99811381e-01 -1.07921851e+00 4.38131958e-01 2.08475560e-01 2.46039927e-01 -6.10288620e-01 3.44727606e-01 -5.13041615e-01 7.08263636e-01 -1.42958388e-01 -2.33782485e-01 3.82086821e-02 1.91830471e-01 6.88596070e-01 9.16753173e-01 1.23930074e-01 -3.69373024e-01 -7.00035334e-01 -1.94723886e-02 5.00591576e-01 -8.17089900e-02 1.51235533e+00 -7.78624594e-01 -2.67142951e-01 9.71194565e-01 8.98227930e-01 3.30823332e-01 -9.18516934e-01 5.83267212e-01 -5.10919541e-02 -1.45460099e-01 3.73181522e-01 -1.03753710e+00 -4.00399655e-01 5.34981668e-01 1.18687940e+00 5.85601091e-01 1.25255692e+00 1.16635207e-02 5.41137457e-01 -1.71875563e-02 1.28807455e-01 -7.48245776e-01 -2.22739801e-01 -5.75814545e-01 5.57313144e-01 -8.26109707e-01 4.74350542e-01 2.87743062e-01 -3.39155406e-01 9.24918473e-01 1.84946999e-01 -6.58958927e-02 8.84628296e-01 1.07269026e-01 -3.57899785e-01 -7.32336700e-01 -6.47139907e-01 -1.18430302e-01 2.55730838e-01 3.03284973e-01 3.03665847e-01 2.67315984e-01 -8.75612497e-01 1.07687914e+00 -8.08258593e-01 1.77293390e-01 5.72642744e-01 6.99272037e-01 -4.17588532e-01 -6.70787036e-01 -2.55619913e-01 1.30899966e+00 -4.61271733e-01 1.03844047e-01 -4.20214653e-01 6.02964222e-01 2.01623693e-01 1.65631831e+00 3.75442892e-01 -7.18689740e-01 3.66048187e-01 4.13848937e-01 3.50461543e-01 -4.83001202e-01 -4.43335623e-01 2.12892562e-01 3.40709627e-01 -9.62066352e-01 -5.00097573e-01 -9.46887016e-01 -1.08057415e+00 -8.68474618e-02 -2.32267138e-02 3.74727875e-01 9.21926856e-01 9.88408804e-01 7.18135238e-01 3.19996357e-01 6.54461145e-01 -2.57652044e-01 2.80262202e-01 -1.15060186e+00 -1.05798018e+00 3.38862017e-02 5.20144165e-01 -5.87364912e-01 -5.73605180e-01 -1.07152700e-01]
[13.555607795715332, 3.4138896465301514]
9c8599d6-a66d-4b7b-bfc9-cc8c37001dec
semi-supervised-skin-lesion-segmentation-via
1808.03887
null
http://arxiv.org/abs/1808.03887v1
http://arxiv.org/pdf/1808.03887v1.pdf
Semi-supervised Skin Lesion Segmentation via Transformation Consistent Self-ensembling Model
Automatic skin lesion segmentation on dermoscopic images is an essential component in computer-aided diagnosis of melanoma. Recently, many fully supervised deep learning based methods have been proposed for automatic skin lesion segmentation. However, these approaches require massive pixel-wise annotation from experienced dermatologists, which is very costly and time-consuming. In this paper, we present a novel semi-supervised method for skin lesion segmentation by leveraging both labeled and unlabeled data. The network is optimized by the weighted combination of a common supervised loss for labeled inputs only and a regularization loss for both labeled and unlabeled data. In this paper, we present a novel semi-supervised method for skin lesion segmentation, where the network is optimized by the weighted combination of a common supervised loss for labeled inputs only and a regularization loss for both labeled and unlabeled data. Our method encourages a consistent prediction for unlabeled images using the outputs of the network-in-training under different regularizations, so that it can utilize the unlabeled data. To utilize the unlabeled data, our method encourages the consistent predictions of the network-in-training for the same input under different regularizations. Aiming for the semi-supervised segmentation problem, we enhance the effect of regularization for pixel-level predictions by introducing a transformation, including rotation and flipping, consistent scheme in our self-ensembling model. With only 300 labeled training samples, our method sets a new record on the benchmark of the International Skin Imaging Collaboration (ISIC) 2017 skin lesion segmentation challenge. Such a result clearly surpasses fully-supervised state-of-the-arts that are trained with 2000 labeled data.
['Pheng-Ann Heng', 'Chi-Wing Fu', 'Lequan Yu', 'Hao Chen', 'Xiaomeng Li']
2018-08-12
null
null
null
null
['skin-lesion-segmentation']
['medical']
[ 6.61536455e-01 3.75028282e-01 -6.23772144e-01 -4.94759858e-01 -9.22850966e-01 -4.76102084e-01 2.22534835e-01 4.03952822e-02 -6.98702395e-01 5.51500618e-01 -2.84006685e-01 -2.39504695e-01 2.79500544e-01 -6.68344617e-01 -5.77385128e-01 -9.18449938e-01 4.59116638e-01 1.41943246e-01 3.40038627e-01 1.75072789e-01 -1.96157172e-01 3.38070899e-01 -1.15879774e+00 3.13993096e-01 1.22413838e+00 1.07846189e+00 7.23825842e-02 5.80057740e-01 -1.99508592e-01 6.67451501e-01 7.11200014e-02 -3.82248253e-01 4.34578329e-01 -5.40099323e-01 -9.23395991e-01 5.48696101e-01 7.26373613e-01 -1.52447253e-01 -1.19301565e-01 1.26924956e+00 4.14551646e-01 -3.99843156e-01 7.88330913e-01 -8.22817802e-01 -2.93136507e-01 3.91826391e-01 -9.29548621e-01 -2.82416940e-01 -2.05058381e-01 3.03145856e-01 7.42199481e-01 -4.37200963e-01 8.21042657e-01 5.14296055e-01 7.68124223e-01 9.48898196e-01 -1.19083989e+00 -2.30236366e-01 1.79414660e-01 -1.28793344e-01 -1.30121350e+00 -5.99858202e-02 6.29097104e-01 -5.12209594e-01 3.83813143e-01 3.17280442e-01 7.22604454e-01 8.05207133e-01 -9.28405300e-02 8.66393089e-01 1.29644299e+00 -6.65397525e-01 3.63937765e-01 4.58388329e-01 1.82704538e-01 9.53833222e-01 5.34556981e-04 8.83333981e-02 -1.11794062e-02 8.82267207e-02 7.64475048e-01 6.77871406e-02 -1.49871096e-01 -3.51617724e-01 -6.65647030e-01 5.87121487e-01 5.22482336e-01 1.39242932e-01 -3.66036773e-01 -7.61940470e-03 2.61807323e-01 1.49421785e-02 6.62072361e-01 1.44326597e-01 -3.14440370e-01 3.40883106e-01 -1.23298347e+00 -3.14233631e-01 6.37080133e-01 4.22768891e-01 8.73241305e-01 -3.22416514e-01 -3.41253042e-01 9.39245641e-01 3.65613341e-01 3.20609301e-01 5.48535109e-01 -6.78795576e-01 8.36075321e-02 9.82440770e-01 -1.83968320e-01 -2.76751578e-01 -4.19353575e-01 -2.92854697e-01 -9.53006804e-01 5.52368760e-01 7.57225931e-01 -4.08682525e-01 -1.52431178e+00 1.63892567e+00 5.80395937e-01 2.64108866e-01 -1.50551707e-01 9.78391826e-01 6.01699293e-01 1.67382538e-01 3.17315906e-01 -2.31382936e-01 1.09568441e+00 -1.37101614e+00 -4.45550591e-01 -1.20348006e-01 8.86747777e-01 -6.12804592e-01 9.54779863e-01 4.89991039e-01 -1.04518032e+00 -3.61296147e-01 -9.53334153e-01 3.03492714e-02 -1.54532090e-01 5.25689244e-01 5.47360659e-01 8.22300673e-01 -9.81945872e-01 5.67299545e-01 -9.41253006e-01 -5.94122350e-01 6.99670017e-01 3.12911034e-01 -5.19721746e-01 -1.54000625e-01 -7.51775980e-01 7.63725817e-01 1.58800408e-01 1.22754335e-01 -7.23587275e-01 -6.81905091e-01 -7.23820329e-01 -3.44667882e-01 4.94238377e-01 -3.34624946e-01 1.05407858e+00 -1.46874416e+00 -1.60366154e+00 1.33553088e+00 -8.97220075e-02 -5.99063635e-01 9.09232438e-01 8.49405155e-02 -3.44520509e-01 3.26619387e-01 -4.29937877e-02 7.85745442e-01 9.39929307e-01 -1.25574303e+00 -5.70415556e-01 -3.88759673e-01 -1.74123615e-01 2.02008680e-01 -4.27399993e-01 -4.09669727e-01 -7.99881756e-01 -3.88647497e-01 2.75847558e-02 -1.02894223e+00 -6.43209577e-01 6.02466524e-01 -8.76957119e-01 1.39074087e-01 5.36555886e-01 -6.77951396e-01 1.12301481e+00 -2.19238520e+00 -1.12563841e-01 4.78059411e-01 1.91205621e-01 5.89723051e-01 -3.66583467e-01 -1.65405981e-02 -1.89045429e-01 1.89790875e-01 -7.17164040e-01 -5.61317563e-01 -4.68932211e-01 -2.33128108e-02 7.45983869e-02 6.16606057e-01 4.30853546e-01 8.19495440e-01 -8.58674705e-01 -7.88498819e-01 2.96493292e-01 4.49671775e-01 -3.32246661e-01 1.11493371e-01 -4.03156400e-01 5.82705498e-01 -2.88869143e-01 6.36800289e-01 7.71865249e-01 -4.59731728e-01 3.61084372e-01 -3.04658324e-01 7.06308335e-02 -2.21349761e-01 -9.91451800e-01 1.90737247e+00 -4.28967088e-01 3.57581794e-01 1.80589795e-01 -7.95535803e-01 5.95501482e-01 3.12392533e-01 6.26530707e-01 -3.75999570e-01 1.24544479e-01 3.25083822e-01 -1.38267547e-01 -5.59136987e-01 -1.11132383e-01 -2.15557039e-01 4.76297796e-01 4.49252009e-01 8.53278190e-02 -1.91024676e-01 3.11203063e-01 1.10102572e-01 1.03968585e+00 3.00895065e-01 1.90267608e-01 7.34050944e-02 6.17464185e-01 2.51532972e-01 3.98102373e-01 4.38644826e-01 -2.81856418e-01 7.77122915e-01 4.49787140e-01 -3.35269421e-01 -9.38139439e-01 -9.28591490e-01 -4.27372903e-01 6.79684937e-01 1.83454320e-01 -1.36998501e-02 -1.15642989e+00 -1.39872837e+00 -6.13468513e-02 2.44548738e-01 -7.92874813e-01 7.20338672e-02 -1.15333818e-01 -8.93734932e-01 5.67897379e-01 3.53325635e-01 4.90471065e-01 -8.40879619e-01 -1.78699344e-01 2.41768695e-02 2.15174392e-01 -1.00160897e+00 -4.87500966e-01 2.46526554e-01 -8.22587073e-01 -1.45090640e+00 -1.08450425e+00 -9.19145823e-01 1.46712577e+00 6.61043227e-02 5.97179413e-01 2.40589723e-01 -8.47584546e-01 1.66756883e-01 -1.17906339e-01 -2.11910784e-01 -4.35853362e-01 5.03961183e-02 -3.84425312e-01 3.95658761e-01 2.06503361e-01 -1.62093580e-01 -6.70300961e-01 3.18386972e-01 -1.09930253e+00 2.34717637e-01 7.82198310e-01 1.10733271e+00 1.04613006e+00 -1.16760217e-01 4.83879209e-01 -1.57529533e+00 1.67686015e-01 -3.09969366e-01 -5.24839818e-01 5.83621085e-01 -5.88795483e-01 -5.22021465e-02 6.35630429e-01 -4.69523758e-01 -1.26858580e+00 7.22961426e-01 -2.73704320e-01 -2.80276895e-01 -2.59667635e-01 3.57122034e-01 1.68048233e-01 -5.25923908e-01 9.14417684e-01 5.45782968e-02 3.26639652e-01 -3.11322063e-01 5.23096263e-01 7.45920897e-01 3.19277078e-01 -1.30497351e-01 7.64303029e-01 7.52896130e-01 8.76627341e-02 -5.22369087e-01 -1.09579968e+00 -5.82974255e-01 -7.69855022e-01 -1.73203394e-01 9.53966618e-01 -7.42563248e-01 -1.69743285e-01 8.71898949e-01 -7.77998209e-01 -5.89244425e-01 -6.76830590e-01 4.06022966e-01 -3.54522556e-01 7.13988900e-01 -5.60462296e-01 -7.14128733e-01 -4.04598176e-01 -1.21548438e+00 8.91777277e-01 5.34405053e-01 -2.31563468e-02 -1.31000447e+00 2.47443095e-01 5.27009785e-01 2.25586101e-01 3.37956250e-01 6.70117557e-01 -7.15682864e-01 -1.48887500e-01 -5.37234485e-01 -4.16878521e-01 9.28208649e-01 4.07337755e-01 2.40859091e-01 -1.01439202e+00 -2.01904207e-01 -3.22504371e-01 -6.96860313e-01 1.27172375e+00 4.58358258e-01 1.46184015e+00 -3.84050980e-02 -3.78678292e-01 8.66280854e-01 1.63883400e+00 -2.81524479e-01 5.67181468e-01 -1.86858937e-01 7.88621485e-01 8.26226175e-01 4.85760152e-01 -1.09922644e-02 2.16425896e-01 1.35436609e-01 4.73088592e-01 -9.23939049e-01 -2.94788897e-01 -2.65672207e-01 -3.30004282e-02 2.31164485e-01 -1.02829605e-01 -4.25747372e-02 -7.62482285e-01 5.63539982e-01 -1.72686791e+00 -4.44387287e-01 -1.25205159e-01 2.37801862e+00 1.26764309e+00 4.05807607e-03 1.08813234e-01 5.28224967e-02 8.77410591e-01 -3.26947169e-03 -9.08773780e-01 -1.98720187e-01 3.20786946e-02 5.21370411e-01 7.31192410e-01 5.23602247e-01 -1.34149158e+00 9.13277566e-01 5.84492826e+00 1.16198051e+00 -1.53786802e+00 -1.00737624e-02 1.00769043e+00 1.81262044e-03 -2.22049445e-01 -5.18252179e-02 -6.55527234e-01 4.69212383e-01 5.54746747e-01 3.12588692e-01 1.60509527e-01 8.36918592e-01 1.05221599e-01 -4.94187474e-01 -1.04569197e+00 5.92270315e-01 1.34921774e-01 -1.35358703e+00 -6.45389855e-02 7.40270689e-02 1.09891903e+00 -1.01877078e-01 1.48133010e-01 -1.93446055e-01 1.42895564e-01 -1.09309483e+00 7.03751668e-02 4.86948341e-01 1.18662345e+00 -2.33587041e-01 7.98828304e-01 2.88142025e-01 -7.92395949e-01 2.80578703e-01 -5.25032356e-02 4.14206147e-01 4.18329462e-02 7.63597786e-01 -9.90403116e-01 4.29123074e-01 9.63926166e-02 7.30743408e-01 -5.38880825e-01 1.12407935e+00 -4.63224143e-01 7.70445049e-01 -2.45543346e-01 1.71830028e-01 3.22140872e-01 -2.88842112e-01 8.44575241e-02 1.09530652e+00 -7.80454502e-02 -3.05378824e-01 2.26611823e-01 6.93490446e-01 -1.13370344e-01 3.64784628e-01 -1.67715594e-01 -1.03274010e-01 9.26003605e-02 1.68199480e+00 -6.29387498e-01 -1.98899090e-01 -2.30244100e-01 1.05378532e+00 3.09371859e-01 4.10535544e-01 -6.83323205e-01 -1.41049102e-01 -1.41002582e-02 3.74134004e-01 -1.86758727e-01 3.18805456e-01 -5.31550050e-01 -9.08585012e-01 -1.07583322e-01 -4.48926896e-01 3.67170960e-01 -3.45171303e-01 -1.47257149e+00 4.72660601e-01 -4.61810052e-01 -1.35701823e+00 -3.74019183e-02 -6.65823698e-01 -9.88891423e-01 1.00799990e+00 -2.04837871e+00 -1.43626773e+00 -5.01532793e-01 6.22366309e-01 1.56475037e-01 -5.51228598e-02 7.69678473e-01 1.84319347e-01 -8.11667264e-01 7.27308452e-01 -1.50536159e-02 2.74052143e-01 1.08386540e+00 -1.39858770e+00 4.85535748e-02 6.97819114e-01 -3.90774687e-04 2.66443282e-01 8.10113735e-03 -5.31212509e-01 -9.08394635e-01 -1.30097210e+00 5.83402038e-01 8.65960568e-02 6.22767091e-01 1.11624680e-01 -8.09187591e-01 5.22216499e-01 5.70838973e-02 4.59464639e-01 1.04386306e+00 -1.94556370e-01 -2.88301498e-01 -1.82568669e-01 -1.64187062e+00 5.39547563e-01 5.25032938e-01 -3.99087638e-01 6.93111718e-02 7.36063182e-01 4.41114902e-01 -4.83400792e-01 -6.73360646e-01 4.67051268e-01 4.97999251e-01 -7.93059051e-01 6.75993145e-01 -5.84103942e-01 6.33900106e-01 -1.93637609e-01 3.27135295e-01 -1.13199615e+00 6.17374964e-02 -3.78109068e-01 1.57233804e-01 8.99252295e-01 6.60495758e-01 -5.72387755e-01 1.29787946e+00 6.86333835e-01 -3.34146991e-02 -1.34614515e+00 -8.31985116e-01 -4.51196253e-01 2.28308305e-01 -1.47453427e-01 -1.11665003e-01 7.54441321e-01 -9.80764553e-02 -1.81645200e-01 -2.78399915e-01 -2.11617146e-02 7.82813370e-01 -2.08187446e-01 5.17000437e-01 -8.89473438e-01 -3.37771446e-01 -2.19635740e-01 -1.81720838e-01 -8.12013447e-01 8.39780364e-03 -9.21763897e-01 4.00505438e-02 -1.57905066e+00 3.59949112e-01 -5.18697739e-01 -2.75512844e-01 8.51292491e-01 -3.46810490e-01 6.36527598e-01 -2.42569717e-03 1.39527872e-01 -4.72645193e-01 1.08871730e-02 1.59846127e+00 -3.55571181e-01 -2.66316175e-01 1.36565521e-01 -5.90670943e-01 9.97205973e-01 7.74275362e-01 -3.12704980e-01 -2.95186669e-01 -2.16010034e-01 -1.58125550e-01 -1.09235734e-01 3.89432371e-01 -8.39287043e-01 5.29064238e-01 -2.98481852e-01 4.36345398e-01 -3.14897031e-01 2.71451414e-01 -7.13464141e-01 -2.19424203e-01 6.42939448e-01 -4.55461442e-01 -9.61786509e-01 5.84562216e-03 5.16760051e-01 -3.04310679e-01 -5.08057833e-01 1.27102876e+00 -1.95097268e-01 -6.42341077e-01 5.24919748e-01 1.00706946e-02 -2.12429464e-02 1.41849840e+00 -3.56755853e-01 -3.23660225e-01 -3.68910059e-02 -1.00019336e+00 3.94755155e-01 6.20354235e-01 -1.63029090e-01 4.07711685e-01 -9.54289794e-01 -7.72581279e-01 2.29549110e-01 1.76529810e-01 2.70082235e-01 5.91505229e-01 1.03359783e+00 -6.32335544e-01 -4.82913703e-02 -1.76694706e-01 -6.63301170e-01 -1.20956957e+00 1.72534972e-01 5.99224746e-01 -6.60448134e-01 -1.33057550e-01 1.12929893e+00 1.48427291e-02 -6.22175217e-01 2.91676730e-01 -9.21772867e-02 -5.64968474e-02 -1.34365991e-01 3.38823140e-01 1.53965309e-01 7.16560856e-02 -2.69241214e-01 -1.11850552e-01 7.09704399e-01 -4.14812684e-01 5.99934906e-02 9.31219399e-01 1.89393878e-01 -1.44161999e-01 1.04748681e-01 1.19235170e+00 9.22739133e-03 -1.53232932e+00 -3.58783692e-01 -3.17401201e-01 -2.54019052e-01 1.34793162e-01 -1.19738019e+00 -1.34375942e+00 9.61258054e-01 7.86533296e-01 -1.40028849e-01 1.29877543e+00 -2.56961167e-01 7.61535943e-01 4.34586033e-02 1.98229298e-01 -1.29613984e+00 6.91782385e-02 6.88011646e-02 3.87357026e-01 -1.59899414e+00 1.15567468e-01 -8.29603255e-01 -9.71148968e-01 1.03819048e+00 7.14681864e-01 -3.34556848e-01 6.23323977e-01 3.00127894e-01 4.55041319e-01 1.91907719e-01 -3.95396054e-01 -3.65155816e-01 5.56518376e-01 6.42207205e-01 4.52862710e-01 1.16679683e-01 -5.69659233e-01 5.46589375e-01 4.50727612e-01 2.41331220e-01 2.74149358e-01 7.07695365e-01 -2.64596611e-01 -1.46035063e+00 4.84820548e-03 7.29167521e-01 -4.33474749e-01 -9.76337269e-02 -5.20268083e-01 5.87980628e-01 3.83412570e-01 7.46408463e-01 -1.16734557e-01 -2.51582623e-01 -6.16035685e-02 -7.66618848e-02 4.85945702e-01 -9.46016312e-01 -7.50413060e-01 1.22216120e-01 -1.51500106e-01 -4.24919248e-01 -4.81215477e-01 -3.71423721e-01 -1.35005307e+00 3.00221682e-01 -4.33426231e-01 -1.12162933e-01 8.13174963e-01 8.10490191e-01 1.71376159e-03 3.70522380e-01 9.44934130e-01 -4.90360260e-01 -8.43735814e-01 -7.65971184e-01 -8.20461571e-01 4.55284387e-01 1.41819164e-01 -1.63694665e-01 -4.96220559e-01 2.60669768e-01]
[15.441121101379395, -2.772311210632324]
4f56cf04-adff-4b5a-9e8a-39f3b77e8386
dorabella-cipher-as-musical-inspiration
null
null
https://aclanthology.org/2021.smp-1.5
https://aclanthology.org/2021.smp-1.5.pdf
Dorabella Cipher as Musical Inspiration
The Dorabella cipher is an encrypted note of English composer Edward Elgar, which has defied decipherment attempts for more than a century. While most proposed solutions are English texts, we investigate the hypothe- sis that Dorabella represents enciphered music. We weigh the evidence in favor of and against the hypothesis, devise a simplified music nota- tion, and attempt to reconstruct a melody from the cipher. Our tools are n-gram models of mu- sic which we validate on existing music cor- pora enciphered using monoalphabetic substi- tution. By applying our methods to Dorabella, we produce a decipherment with musical qual- ities, which is then transformed via artful com- position into a listenable melody. Far from ar- guing that the end result represents the only true solution, we instead frame the process of decipherment as part of the composition pro- cess.
['Grzegorz Kondrak', 'Scott Smallwood', 'Abram Hindle', 'Colin Choi', 'Bradley Hauer']
null
null
null
null
smp-icon-2021-12
['decipherment']
['natural-language-processing']
[ 4.57040727e-01 -1.27084017e-01 2.85436571e-01 2.26218447e-01 -7.11700499e-01 -1.22126913e+00 6.51659429e-01 -2.40200367e-02 -5.13318717e-01 7.92289972e-01 6.05610371e-01 -5.82790852e-01 -8.69497508e-02 -7.28170395e-01 -3.74532908e-01 -6.24303341e-01 -1.92623921e-02 4.48421508e-01 -4.38139021e-01 -1.75255626e-01 4.53722715e-01 4.01117027e-01 -1.29668617e+00 6.13481820e-01 4.68293190e-01 7.23393321e-01 -1.36313275e-01 1.25871813e+00 2.05039278e-01 1.16924727e+00 -8.69183481e-01 -1.05962074e+00 4.35502112e-01 -9.86166060e-01 -7.39606440e-01 -4.23809409e-01 9.01597515e-02 -3.01729053e-01 -4.40325409e-01 1.12571597e+00 3.33057225e-01 -5.86431146e-01 6.66958272e-01 -7.62902379e-01 -6.03199974e-02 1.39190984e+00 -4.68800366e-02 -4.35136497e-01 4.41960275e-01 -2.32698992e-01 1.18504322e+00 -3.32374811e-01 5.79682231e-01 7.29714692e-01 1.03443897e+00 3.66668701e-01 -1.23770237e+00 -9.72796142e-01 -1.00973284e+00 2.04978555e-01 -1.60786211e+00 -5.07362366e-01 8.95537257e-01 -1.28402412e-01 3.36533993e-01 1.12690568e+00 1.24397469e+00 8.50469410e-01 3.59239399e-01 8.17311287e-01 1.47295153e+00 -5.58440447e-01 1.03112739e-02 -8.79039019e-02 -4.11215574e-01 4.80258316e-01 3.43570977e-01 3.35585266e-01 -7.75104582e-01 -3.96204084e-01 5.66091835e-01 -5.70724547e-01 -4.26195830e-01 3.67766500e-01 -1.45802581e+00 4.29898888e-01 -2.13463560e-01 4.13630873e-01 -2.90022105e-01 3.68116438e-01 5.72017789e-01 6.82693779e-01 -2.07762823e-01 8.97645712e-01 -1.35286242e-01 -8.11795235e-01 -1.43195951e+00 6.07805789e-01 1.32337630e+00 3.27276200e-01 3.10885191e-01 1.77483380e-01 5.06831765e-01 1.94780201e-01 -1.80747882e-02 3.78597349e-01 5.61481059e-01 -8.49258184e-01 2.12374717e-01 -6.55431971e-02 -4.36876267e-01 -8.78573775e-01 1.09783068e-01 -4.29571360e-01 -8.34846258e-01 3.06350827e-01 4.22366202e-01 -1.69134915e-01 -1.01438619e-01 1.27211916e+00 -3.99568200e-01 2.23965362e-01 4.14249927e-01 7.51416743e-01 3.65624666e-01 5.84893942e-01 -4.04297501e-01 -2.77032137e-01 1.52733338e+00 -4.44844782e-01 -7.08192170e-01 5.19825995e-01 5.05988240e-01 -1.27658904e+00 6.46323979e-01 1.14228034e+00 -1.18359017e+00 -3.09902579e-01 -1.49352813e+00 3.29859145e-02 3.34189355e-01 -7.04875886e-02 5.26531935e-01 9.31767702e-01 -5.18911183e-01 9.36280131e-01 -4.37868059e-01 3.44822466e-01 1.14655331e-01 3.20956439e-01 -3.95309448e-01 6.24279976e-01 -8.38672996e-01 5.80841362e-01 6.84945107e-01 3.96589525e-02 -7.62894630e-01 -5.43979824e-01 -2.07043201e-01 -5.25080077e-02 2.06777696e-02 -6.92531228e-01 1.31044567e+00 -1.23041153e+00 -1.77490175e+00 8.41635883e-01 2.72721410e-01 -7.20731795e-01 7.81796634e-01 2.20701531e-01 -8.35879087e-01 2.03307167e-01 -4.76995438e-01 -3.25503312e-02 9.90999877e-01 -1.34585035e+00 -5.11101067e-01 -1.54144362e-01 -1.16044439e-01 1.15193769e-01 -3.48221734e-02 7.92045519e-02 2.27207646e-01 -1.27845335e+00 2.16664419e-01 -1.01858461e+00 3.41185838e-01 -4.11185145e-01 -7.37484157e-01 5.63316286e-01 7.66658604e-01 -8.46120536e-01 1.71965718e+00 -2.21765018e+00 5.60369305e-02 4.51932877e-01 3.15169990e-01 1.68965772e-01 4.11476754e-02 7.86241710e-01 -3.91306430e-01 2.26580009e-01 -4.64118063e-01 -3.02232891e-01 1.90767109e-01 2.81728655e-01 -8.16443145e-01 5.65512836e-01 -4.56552714e-01 1.03006649e+00 -5.64723015e-01 -2.86284506e-01 -1.67559043e-01 2.12860838e-01 -6.60195887e-01 9.85959545e-02 -1.44960657e-01 1.29302993e-01 -1.52137175e-01 5.26923656e-01 3.42418551e-01 4.67879236e-01 5.57496190e-01 -2.17588902e-01 -4.63916898e-01 6.08604252e-01 -1.18782485e+00 1.78719473e+00 -2.03844771e-01 9.58048761e-01 1.80298164e-01 -5.37366688e-01 9.45759773e-01 5.00539482e-01 4.24095780e-01 -3.31598729e-01 5.52941680e-01 5.89077473e-01 4.15918946e-01 -2.87090689e-01 1.14950240e+00 -8.38870108e-01 -3.54492873e-01 8.96902323e-01 -3.11964363e-01 -5.78835547e-01 -1.57721668e-01 1.41972438e-01 1.05806375e+00 2.34678969e-01 6.60439074e-01 -1.02967598e-01 3.98397118e-01 -3.21286805e-02 1.68102518e-01 4.64130878e-01 3.22864830e-01 7.22419500e-01 4.52754468e-01 -3.51493984e-01 -1.32246435e+00 -9.56002891e-01 4.56828028e-02 3.46052617e-01 -1.90661326e-01 -1.39195240e+00 -8.92946661e-01 -1.63137153e-01 -1.60830423e-01 7.71986067e-01 -3.27793121e-01 -4.85605747e-02 -6.65930629e-01 -4.18014258e-01 1.53982377e+00 -2.68058300e-01 3.42828155e-01 -9.85974252e-01 -9.81238365e-01 4.44054753e-01 -2.12200120e-01 -6.51682377e-01 -3.68180990e-01 4.57262874e-01 -7.20464468e-01 -1.01547766e+00 -4.34330344e-01 -5.28210402e-01 -1.53320590e-02 -2.29641050e-01 8.83779526e-01 1.08720586e-01 -3.64610821e-01 5.25475256e-02 -4.20571327e-01 -5.98375678e-01 -1.29835248e+00 -5.05903289e-02 1.64572671e-01 -3.25646214e-02 1.74113616e-01 -1.04738081e+00 -4.80489939e-01 -2.61174172e-01 -1.10159659e+00 2.82037824e-01 4.61974293e-01 4.43082601e-01 4.38637197e-01 2.62793332e-01 1.88844442e-01 -7.95587838e-01 5.98093688e-01 5.20151854e-02 -1.97115079e-01 -2.85711717e-02 -4.66746092e-01 1.88980684e-01 9.15845394e-01 -5.59594810e-01 -5.04808545e-01 -1.27575904e-01 -4.77749437e-01 -1.81943685e-01 2.07839191e-01 4.65695769e-01 -7.49643892e-02 -1.52178809e-01 7.09681034e-01 5.37118375e-01 5.97244576e-02 -5.47930300e-01 4.07491624e-01 9.85064447e-01 1.27684820e+00 -7.36797631e-01 9.85270500e-01 4.31360811e-01 4.62296046e-02 -8.64001811e-01 -1.62866846e-01 2.51469225e-01 -3.86786371e-01 -1.99858442e-01 5.50952792e-01 -6.83064342e-01 -1.15968728e+00 3.55550736e-01 -1.10639668e+00 6.91174716e-02 -8.21824908e-01 5.44929624e-01 -9.70667183e-01 8.63812864e-01 -7.08103299e-01 -7.86382258e-01 -5.07479191e-01 -5.50164580e-01 5.21692514e-01 -3.57686162e-01 -9.20644462e-01 -5.88865280e-01 4.36448485e-01 2.82765985e-01 -8.69081635e-03 1.74632117e-01 1.17069530e+00 -5.28906763e-01 -5.18706739e-01 -5.61159372e-01 2.77414352e-01 3.55899006e-01 1.49192363e-01 -6.50659502e-02 -1.06742024e+00 -8.04924220e-02 3.19067359e-01 -1.24347620e-01 6.04841828e-01 -3.63601327e-01 8.68255913e-01 -6.62454784e-01 4.14411306e-01 1.00271142e+00 1.36302555e+00 1.72147945e-01 1.17226756e+00 4.71217155e-01 3.14467311e-01 2.23248780e-01 7.61157274e-02 8.30354333e-01 1.89470932e-01 4.14493740e-01 3.85898948e-01 4.64877665e-01 -3.70313197e-01 -9.29171920e-01 6.65894926e-01 1.67238057e+00 -6.20561898e-01 -5.84023707e-02 -4.09834892e-01 7.63150305e-02 -1.39634287e+00 -1.44233751e+00 -1.94541380e-01 2.12628841e+00 1.07985139e+00 2.59617846e-02 2.70334065e-01 9.58969355e-01 -1.16035398e-02 2.46665046e-01 1.00114942e-01 -5.80039978e-01 -5.68203270e-01 8.05707514e-01 6.62513793e-01 5.56939840e-01 -6.96020007e-01 7.53810585e-01 6.77815866e+00 9.23615813e-01 -1.16000092e+00 -2.15288758e-01 -5.30474782e-02 -1.43904528e-02 -6.95793867e-01 4.96146798e-01 -6.86789602e-02 4.89999712e-01 9.49603558e-01 -5.13026297e-01 9.91918862e-01 6.95988089e-02 -9.54537839e-02 5.01446500e-02 -9.81783450e-01 1.31823134e+00 4.06022459e-01 -1.40556753e+00 2.87893862e-01 1.31329224e-01 4.74677652e-01 -4.88557220e-01 -3.11338399e-02 -8.16017389e-02 1.10453211e-01 -1.26468825e+00 1.37454021e+00 7.18071222e-01 1.03997397e+00 -9.22553480e-01 3.91392797e-01 4.57065940e-01 -1.09567928e+00 -1.05381727e-01 -1.14995517e-01 -4.56388444e-01 1.64497420e-01 1.46670744e-01 -8.35288942e-01 9.25255001e-01 -2.46613119e-02 4.95754063e-01 -2.11929768e-01 9.51486528e-01 -2.25089788e-01 1.01760840e+00 -1.82037771e-01 -1.60785139e-01 -7.54737854e-02 -5.69666684e-01 9.23780501e-01 1.24149239e+00 9.64024723e-01 2.93245316e-01 -4.89613235e-01 6.01507187e-01 -2.98622977e-02 1.45382583e-01 -3.12494248e-01 -5.43696165e-01 3.43590260e-01 9.10633683e-01 -5.03294647e-01 -3.84480089e-01 1.97250247e-01 1.33477902e+00 -3.57077897e-01 -8.38399678e-03 -4.84955519e-01 -5.00308275e-01 4.61348265e-01 1.35320023e-01 2.60746002e-01 -4.73237932e-01 -6.76171303e-01 -1.25748610e+00 -3.26745301e-01 -1.67225695e+00 9.20254514e-02 -7.13298440e-01 -7.06657052e-01 6.80267632e-01 -4.15556461e-01 -1.51620221e+00 -4.70054597e-01 -2.18541637e-01 -5.51697373e-01 6.55405998e-01 -8.35707664e-01 -1.08745027e+00 4.43882883e-01 4.13762480e-01 8.17312077e-02 -4.23868150e-01 1.39930880e+00 2.22345278e-01 2.39365354e-01 5.56053162e-01 1.37203097e-01 1.45477191e-01 6.53678060e-01 -1.05659032e+00 1.35797545e-01 5.49326479e-01 9.47308064e-01 6.03565395e-01 1.27989888e+00 -5.73021114e-01 -1.84528649e+00 -3.47217292e-01 1.22871292e+00 -2.73606449e-01 9.04300570e-01 -4.43743467e-01 -4.94792759e-01 5.09229660e-01 2.71200269e-01 -9.01662648e-01 8.33418429e-01 -4.09648508e-01 -5.12736082e-01 -7.74129573e-03 -6.49807990e-01 8.00095677e-01 7.19664693e-01 -1.00513971e+00 -8.83151710e-01 -2.03428015e-01 3.62983435e-01 -1.17632605e-01 -8.75331521e-01 1.59159061e-02 1.27949679e+00 -1.01356399e+00 7.15186536e-01 -3.10381174e-01 6.16356134e-01 -5.68339288e-01 -4.93331522e-01 -9.72267926e-01 6.57335445e-02 -1.48267353e+00 1.53662696e-01 9.58017528e-01 1.96567491e-01 -1.79520726e-01 9.45617974e-01 -2.58859813e-01 -6.03685491e-02 -9.65491980e-02 -1.03870416e+00 -7.30363309e-01 1.01173267e-01 -9.19709623e-01 7.10329771e-01 1.22421741e+00 2.91218221e-01 2.51549095e-01 -9.49493051e-01 -3.13074946e-01 7.68167496e-01 2.70500094e-01 1.06922090e+00 -1.14169681e+00 -7.73202002e-01 -3.88230830e-01 -5.35228908e-01 -8.97833645e-01 1.09910823e-01 -1.37236071e+00 -2.04493448e-01 -6.66607261e-01 -1.78263169e-02 -2.75894940e-01 -6.50605280e-03 2.28090882e-01 4.31878895e-01 7.79726148e-01 6.76671505e-01 5.01725674e-01 1.55743942e-01 4.31953937e-01 1.13251531e+00 -2.11704791e-01 5.04656881e-03 -7.10352138e-03 -8.79797578e-01 6.98209822e-01 7.03636527e-01 -6.62063420e-01 -1.53081626e-01 1.84506495e-02 9.39601779e-01 3.15065026e-01 4.40370262e-01 -1.22251928e+00 3.17941248e-01 2.95457095e-01 1.07268699e-01 -5.54552495e-01 4.71651167e-01 -9.24695551e-01 9.70784247e-01 7.47528672e-01 -2.32519016e-01 2.33839080e-01 -9.92983505e-02 2.46511593e-01 -4.55319107e-01 -4.39348072e-01 4.41539586e-01 -8.79175216e-02 -2.82103688e-01 -4.03589606e-02 -5.63236952e-01 -4.16552067e-01 6.16305172e-01 -4.90497857e-01 -1.02569573e-02 -5.40515900e-01 -6.56484604e-01 -7.69853771e-01 7.25524902e-01 -7.62911141e-02 4.28034842e-01 -1.15759552e+00 -1.03015435e+00 3.81256342e-01 -1.91171333e-01 -8.41435194e-01 5.62481545e-02 4.66204196e-01 -1.34379065e+00 1.01884447e-01 -2.52469003e-01 4.83659506e-02 -1.83582282e+00 3.26093763e-01 -4.31339145e-02 7.28408247e-02 -8.85989666e-01 3.22517037e-01 -2.76598603e-01 -1.36987358e-01 -1.20073013e-01 -1.58720106e-01 2.46009827e-01 -1.44723454e-03 5.54805279e-01 3.25059146e-01 -1.74253374e-01 -7.44145274e-01 5.13035059e-02 4.53967929e-01 4.99393642e-01 -7.00589061e-01 1.25887334e+00 1.12151168e-01 -6.34724140e-01 6.02306247e-01 1.22474158e+00 9.06526566e-01 -5.17875910e-01 3.82921159e-01 -3.34856391e-01 -4.66725796e-01 -4.45444971e-01 -8.65341425e-01 -3.46280843e-01 6.38319910e-01 1.25353232e-01 2.60411382e-01 1.14505386e+00 -2.19918951e-01 9.62219834e-01 4.42495048e-01 3.46229404e-01 -7.84109712e-01 -3.28024626e-01 5.07248700e-01 8.30297530e-01 -1.31538481e-01 3.55148613e-01 1.57023907e-01 -5.18496275e-01 1.47931027e+00 -6.88798487e-01 -4.13968295e-01 5.29062688e-01 7.47089207e-01 1.92478880e-01 1.94174454e-01 -7.03718781e-01 2.91996688e-01 6.48737997e-02 2.14387417e-01 2.94008970e-01 4.67655748e-01 -8.99823129e-01 1.01179826e+00 -1.59375334e+00 1.06539041e-01 7.32785940e-01 7.04624534e-01 -3.56136680e-01 -1.49886084e+00 -7.83327520e-01 5.69265932e-02 -9.28250551e-01 -4.08076078e-01 -8.06466281e-01 8.48997772e-01 3.67011726e-01 6.83495820e-01 -7.71176815e-02 -1.12666285e+00 -5.07761501e-02 3.64409536e-02 8.15756738e-01 5.09325936e-02 -1.14499760e+00 2.08247855e-01 5.64472914e-01 5.12619205e-02 -2.85272092e-01 -4.85082686e-01 -1.05302846e+00 -1.07776082e+00 -6.15375340e-02 5.59758842e-01 8.48731279e-01 6.04191720e-01 -2.52536088e-01 3.45707685e-01 5.62525630e-01 -4.93025273e-01 -4.89076108e-01 -5.67761481e-01 -1.08797860e+00 2.63093978e-01 3.84800613e-01 5.13571739e-01 -4.44484234e-01 1.19312920e-01]
[15.997786521911621, 5.450129508972168]
b41e5459-2b6d-4c52-a61e-6ea2b77ae4d2
transformer-and-snowball-graph-convolution
2303.16132
null
https://arxiv.org/abs/2303.16132v2
https://arxiv.org/pdf/2303.16132v2.pdf
Transformer and Snowball Graph Convolution Learning for Brain functional network Classification
Advanced deep learning methods, especially graph neural networks (GNNs), are increasingly expected to learn from brain functional network data and identify the functional connections between brain disorder and health. In this paper, we proposed a novel Transformer and snowball encoding networks (TSEN) for brain functional network classification, which introduced Transformer architecture with graph snowball connection into GNNs for learning whole-graph representation. TSEN combined graph snowball connection with graph Transformer by snowball encoding layers, which enhanced the power to capture multi-scale information and global patterns of brain functional networks. TSEN also introduced snowball graph convolution as position embedding in Transformer structure, which was a simple yet effective method for capturing local patterns naturally. We evaluated the proposed model by two large-scale brain functional network datasets, and the results demonstrated that TSEN outperformed the state-of-the-art GNN models and the graph-transformer based GNN models.
['Shoubin Dong', 'Yangmin Huang', 'Jinlong Hu']
2023-03-28
null
null
null
null
['graph-classification']
['graphs']
[ 2.46340148e-02 1.96630836e-01 2.37069011e-01 -3.67002726e-01 4.90365267e-01 -1.27776578e-01 5.07281482e-01 -2.24615987e-02 -6.47385865e-02 5.81752896e-01 3.02179277e-01 -1.34557173e-01 -6.32260323e-01 -1.28732026e+00 -6.11430466e-01 -5.22304714e-01 -5.69177806e-01 4.77414668e-01 4.99440253e-01 -4.00156856e-01 -9.93150547e-02 5.39481819e-01 -8.99408698e-01 4.93066072e-01 8.18734586e-01 9.89978373e-01 5.05022764e-01 4.08078521e-01 -4.17640358e-01 1.02173841e+00 -3.02179903e-01 -3.64509612e-01 8.18840694e-03 -4.34473485e-01 -7.23076820e-01 -2.65054196e-01 2.58918196e-01 6.03603423e-02 -1.05807710e+00 1.13040841e+00 5.93582451e-01 2.96342815e-03 6.02254629e-01 -1.49520516e+00 -1.11746073e+00 9.33607757e-01 -2.05979332e-01 8.21175635e-01 1.77367628e-01 2.30172813e-01 1.08737099e+00 -5.54077804e-01 5.37355125e-01 1.42259371e+00 1.05080390e+00 5.16256511e-01 -9.72042739e-01 -9.24004197e-01 8.88318345e-02 4.78833854e-01 -8.67031515e-01 -1.96571257e-02 7.72374868e-01 -5.30756235e-01 1.31291163e+00 -1.10999942e-01 1.67808986e+00 1.23259354e+00 8.54657173e-01 6.43339336e-01 9.83157396e-01 1.83199629e-01 -3.43448222e-01 -8.53062868e-01 3.90677959e-01 1.32426202e+00 4.02517378e-01 6.48616254e-02 -5.02014339e-01 -5.98126166e-02 1.21020818e+00 4.81954128e-01 -4.67124283e-01 -1.98139369e-01 -1.43050337e+00 7.25351751e-01 1.43063498e+00 6.01374507e-01 -4.46358532e-01 5.90063453e-01 6.60340667e-01 5.82875669e-01 6.38340235e-01 2.16478765e-01 -1.49759352e-01 3.04206938e-01 -6.64478123e-01 -2.03620255e-01 5.07726729e-01 5.95907509e-01 4.28222328e-01 4.41983968e-01 -3.21356922e-01 7.99228907e-01 4.25977737e-01 2.39747405e-01 7.94596612e-01 -1.05521902e-01 3.66924882e-01 1.32233298e+00 -1.05665398e+00 -1.20053101e+00 -1.00748420e+00 -6.97838783e-01 -1.34819233e+00 -1.54233664e-01 1.15213975e-01 1.45312637e-01 -1.08007693e+00 1.58444309e+00 -1.19177774e-01 3.95468146e-01 -3.70445490e-01 6.32313550e-01 1.43672740e+00 3.75884891e-01 -1.37665570e-01 3.46313566e-01 1.34626710e+00 -9.50412869e-01 -7.44429588e-01 -3.54528636e-01 6.39504194e-01 -1.43116303e-02 1.00506783e+00 -1.11803666e-01 -7.63134778e-01 -4.69254464e-01 -9.18160975e-01 -5.13652563e-02 -5.14776289e-01 -3.83232206e-01 9.09385204e-01 2.96567023e-01 -1.66566837e+00 8.98246408e-01 -9.68039393e-01 -5.19495428e-01 1.02724814e+00 5.94361186e-01 -7.83593833e-01 -2.32991531e-01 -1.45497155e+00 7.63619423e-01 4.57758695e-01 3.83174717e-01 -1.26642525e+00 -7.93069422e-01 -9.81553197e-01 4.67779487e-01 -5.08482978e-02 -9.12603557e-01 5.28189600e-01 -8.37498486e-01 -1.14463174e+00 8.24054062e-01 2.02843279e-01 -4.58929002e-01 1.20250285e-01 3.19157869e-01 -5.48621058e-01 3.27893674e-01 -1.35285273e-01 5.32116771e-01 6.38327956e-01 -4.31359887e-01 3.96140367e-01 -5.23141384e-01 -8.52331147e-02 3.72970887e-02 -4.85403627e-01 -1.68080911e-01 2.65316069e-01 -6.68887436e-01 3.12957019e-01 -5.52636385e-01 6.68945387e-02 1.59215003e-01 -4.44080532e-01 -4.54910696e-01 8.69121194e-01 -7.13860393e-01 9.89160240e-01 -1.79410493e+00 2.93719888e-01 2.46677622e-01 1.28043008e+00 1.72508001e-01 -4.63909119e-01 5.00779450e-01 -6.39054894e-01 2.51488704e-02 -1.59767419e-01 5.98003455e-02 -8.84205550e-02 2.91417092e-01 2.54794478e-01 5.43079197e-01 -1.80617755e-03 1.70774603e+00 -1.14765048e+00 -1.16057679e-01 3.65520380e-02 6.46306872e-01 -3.29773575e-01 2.21575141e-01 -1.24147842e-02 -2.81248912e-02 -3.09777111e-01 4.75116313e-01 4.86886501e-01 -4.90952969e-01 1.83150813e-01 -5.22642553e-01 4.73867893e-01 3.15011293e-01 -2.86510766e-01 1.47803795e+00 -9.01978612e-02 8.68475616e-01 -1.58285886e-01 -1.44211257e+00 1.01733387e+00 3.63986552e-01 6.78358436e-01 -8.83050740e-01 3.63353670e-01 -1.29766436e-02 6.18687749e-01 -4.20622975e-01 -4.31101680e-01 -8.84689316e-02 4.24172342e-01 5.11346996e-01 7.36437380e-01 2.75432765e-01 -7.36239031e-02 2.78262556e-01 1.67745495e+00 -2.83731461e-01 2.39805982e-01 -8.02028656e-01 4.08317566e-01 -4.58863735e-01 1.89925388e-01 2.98919141e-01 -3.49300474e-01 2.95525521e-01 1.00031531e+00 -7.81393468e-01 -6.06868744e-01 -1.23792922e+00 2.02868581e-01 9.06163514e-01 -2.33002812e-01 -6.45553410e-01 -7.66366541e-01 -8.26235771e-01 -3.62238288e-02 -1.02846667e-01 -9.58591163e-01 -6.82135463e-01 -4.32269365e-01 -9.55115974e-01 8.08609903e-01 5.75670481e-01 9.52896953e-01 -1.44697165e+00 -2.31754151e-03 1.83768407e-01 5.64809069e-02 -8.18379104e-01 -8.06432724e-01 1.47091255e-01 -1.06697893e+00 -1.52437913e+00 -7.05239594e-01 -1.21318197e+00 8.37515593e-01 2.23961145e-01 1.11238444e+00 3.90740216e-01 -4.18955684e-01 2.03580618e-01 -2.09438711e-01 2.46290006e-02 1.43013410e-02 3.09321489e-02 -3.88777070e-03 -6.93903863e-02 1.14063077e-01 -1.14304459e+00 -8.04964781e-01 2.66130179e-01 -8.43566477e-01 1.52428448e-01 6.32373393e-01 1.11695850e+00 2.97584444e-01 7.22918659e-02 7.36404479e-01 -1.19467437e+00 1.10248637e+00 -6.54932618e-01 -2.12327436e-01 2.20779240e-01 -6.83760881e-01 1.84108302e-01 8.43602061e-01 -4.27738667e-01 -3.28978419e-01 -5.77304900e-01 -3.96541506e-01 -4.37501401e-01 4.32232291e-01 8.28671217e-01 -2.04820111e-01 -6.26367986e-01 5.92491746e-01 5.51179826e-01 2.84155279e-01 -3.67267221e-01 1.34085804e-01 1.65531784e-02 2.45106176e-01 -2.92439759e-01 2.93050706e-01 7.06405044e-02 2.79636264e-01 -5.54436803e-01 -5.38992047e-01 -4.96272072e-02 -7.14901984e-01 -2.87995666e-01 9.84332561e-01 -5.92396021e-01 -8.20603967e-01 8.21006238e-01 -9.77422535e-01 -6.06909752e-01 -1.70329779e-01 2.85016298e-01 -3.45069677e-01 4.13047016e-01 -1.03228676e+00 -5.67254610e-02 -7.19157517e-01 -9.53894734e-01 6.55384004e-01 -7.24727511e-02 2.81273663e-01 -1.64809644e+00 1.03634045e-01 6.80726115e-03 6.67708993e-01 4.53574777e-01 1.25962055e+00 -5.78516901e-01 -3.07595551e-01 2.43154801e-02 -6.74127102e-01 2.14072898e-01 2.14617148e-01 -5.15180707e-01 -5.96896350e-01 -4.81668562e-01 -1.84811428e-01 -1.24968812e-01 1.07292557e+00 4.48418021e-01 1.22847652e+00 -4.45845008e-01 -4.42502528e-01 9.40808117e-01 1.33181286e+00 -2.87021771e-02 8.13801706e-01 1.59607783e-01 1.22792602e+00 1.38876855e-01 -7.17434645e-01 -1.30038038e-01 4.71756637e-01 2.14338258e-01 6.08865976e-01 -2.54044414e-01 -6.52246714e-01 -2.58925319e-01 5.09940088e-01 1.14984035e+00 -2.16633022e-01 -3.97591174e-01 -1.06960821e+00 4.80436027e-01 -1.85759938e+00 -9.52698648e-01 -1.91498369e-01 1.41737342e+00 3.37034822e-01 2.98851430e-01 1.34730026e-01 1.44629730e-02 8.10457587e-01 4.19767201e-01 -6.81257069e-01 -1.25343591e-01 -2.65451461e-01 4.60190624e-01 3.15433413e-01 6.78160042e-02 -6.46505237e-01 9.07753289e-01 6.27789736e+00 6.15598321e-01 -1.08979118e+00 5.24634063e-01 3.30831259e-01 1.59201428e-01 -3.75614911e-01 -4.36008483e-01 -1.96577478e-02 4.73291904e-01 9.61056709e-01 -3.00626129e-01 7.15397656e-01 4.01226103e-01 -1.23936325e-01 7.61243999e-01 -1.03618860e+00 1.22710550e+00 -1.04206227e-01 -1.47567308e+00 4.74426031e-01 6.30490929e-02 4.28115368e-01 6.93559766e-01 -2.06541851e-01 2.17343479e-01 4.93837804e-01 -1.52832246e+00 1.88476801e-01 7.92617321e-01 8.87525082e-01 -4.75663215e-01 7.96611428e-01 9.85546857e-02 -1.66150653e+00 -1.89862698e-01 -6.40111029e-01 -2.05948532e-01 -2.56196350e-01 6.35596812e-01 -6.33524835e-01 6.21658921e-01 6.88251913e-01 1.51252115e+00 -7.78154492e-01 1.17408228e+00 -3.43367130e-01 6.89096153e-01 -5.97319081e-02 -1.57940894e-01 3.89413625e-01 -2.09167212e-01 3.11838478e-01 1.12956190e+00 2.36891314e-01 -1.10969730e-01 1.57208666e-01 1.06336915e+00 -5.09064317e-01 -8.14658105e-02 -8.71291399e-01 -5.21275640e-01 3.10047120e-02 1.52168572e+00 -1.06482518e+00 -5.11202365e-02 -3.66316497e-01 9.48453307e-01 8.08629632e-01 2.42446408e-01 -5.45747995e-01 -5.09922504e-01 5.54350078e-01 5.93853593e-01 -1.06585234e-01 -3.00392061e-01 5.83736338e-02 -1.20276952e+00 -2.01361433e-01 -5.20389676e-01 5.41457534e-01 -9.73501623e-01 -1.70324516e+00 1.18950784e+00 -3.14468235e-01 -6.35207832e-01 3.75913829e-01 -1.00475574e+00 -1.04236209e+00 7.77988851e-01 -1.38418269e+00 -1.36540699e+00 -5.19449294e-01 1.29180479e+00 1.33336801e-02 -4.99336123e-01 6.76199377e-01 3.65761817e-01 -4.96609747e-01 4.75281984e-01 -6.90865517e-02 5.62602937e-01 1.01620294e-01 -1.12037385e+00 6.86800539e-01 6.12099707e-01 1.59304529e-01 7.66408205e-01 -7.85485879e-02 -7.00990498e-01 -1.22254264e+00 -1.28342474e+00 5.93774736e-01 2.38960236e-02 1.09616935e+00 -7.89926052e-01 -8.80989134e-01 8.65249693e-01 3.41238618e-01 5.41402340e-01 3.63532156e-01 -7.12643415e-02 -3.92009377e-01 -3.38831067e-01 -1.30359793e+00 2.14064330e-01 1.82682145e+00 -7.45731294e-01 -7.33181715e-01 7.38777995e-01 9.14652526e-01 -8.29045475e-02 -1.03511369e+00 2.42051929e-01 4.63219255e-01 -7.16965139e-01 9.88878846e-01 -6.95798635e-01 3.49788278e-01 1.09372161e-01 1.48038611e-01 -1.95043981e+00 -1.03871346e+00 -3.41047615e-01 -2.20417842e-01 6.30364835e-01 2.17882946e-01 -1.19479716e+00 7.30576277e-01 -1.08578093e-01 -6.41468644e-01 -1.10789084e+00 -9.17274356e-01 -5.92252791e-01 -1.06557980e-01 -2.92387567e-02 7.62741148e-01 9.93026674e-01 1.97918102e-01 6.17511451e-01 -9.61243659e-02 -4.07205254e-01 4.93319929e-01 -3.78915399e-01 -1.66403815e-01 -1.45324028e+00 2.29853485e-03 -7.70723462e-01 -1.20553136e+00 -5.44254720e-01 4.64010060e-01 -1.78219593e+00 -6.81129158e-01 -2.14214587e+00 3.81102532e-01 9.21649188e-02 -8.46342146e-01 8.14422905e-01 9.23898537e-03 3.57211679e-01 -1.90441012e-02 -1.16064407e-01 -3.87350142e-01 7.15554237e-01 1.84539533e+00 -4.53115612e-01 2.61170059e-01 -1.79291710e-01 -7.25928783e-01 4.16994184e-01 6.92274153e-01 -2.74046093e-01 -1.04960120e+00 -6.25360072e-01 3.23522896e-01 -1.49042100e-01 6.76371574e-01 -1.19933772e+00 3.90030921e-01 4.27388161e-01 6.84149861e-01 -7.01050237e-02 9.41521376e-02 -6.97074831e-01 1.44895568e-01 8.20369959e-01 -1.62922382e-01 4.38052505e-01 1.43547818e-01 5.08758247e-01 -1.41095802e-01 5.02474427e-01 6.66307330e-01 -3.95666510e-01 -4.65203315e-01 1.22291219e+00 -2.78311729e-01 -1.65018979e-02 7.00026631e-01 -3.29414040e-01 -6.56154633e-01 -1.66141614e-01 -1.03490651e+00 2.14976728e-01 -2.06039444e-01 3.42162758e-01 1.26899958e+00 -1.56163180e+00 -5.68134844e-01 5.23156047e-01 -1.67003959e-01 -3.69653374e-01 3.17294180e-01 9.32312846e-01 -6.63972318e-01 3.16518962e-01 -8.39355290e-01 -4.02995229e-01 -9.72891927e-01 3.90986443e-01 8.16375732e-01 -5.57187080e-01 -1.20130718e+00 1.19742465e+00 4.68824297e-01 -8.55689466e-01 -1.33874103e-01 -6.15529418e-01 -4.40978259e-01 -1.83610350e-01 2.28859663e-01 1.69168890e-01 1.53377593e-01 -5.11187613e-01 -4.79578435e-01 3.58454168e-01 1.75437692e-03 5.22333145e-01 1.77343917e+00 2.17864871e-01 -8.08278441e-01 2.49039397e-01 1.29515004e+00 -8.04481149e-01 -8.54484379e-01 -3.74629796e-01 -1.15636081e-01 -5.51975481e-02 3.07475865e-01 -8.26565087e-01 -1.88677061e+00 1.03405917e+00 5.98674178e-01 3.03169638e-01 1.14419723e+00 -8.64498764e-02 1.19531906e+00 5.08742988e-01 4.81461883e-01 -2.95543045e-01 3.76370847e-01 6.79804444e-01 1.11721861e+00 -7.78063357e-01 -2.15712115e-01 -2.06125021e-01 -7.95559809e-02 1.56047213e+00 8.01982641e-01 -6.26348019e-01 1.25174952e+00 5.52401058e-02 -6.22679114e-01 -1.15856993e+00 -6.94014430e-01 8.83725137e-02 6.54469788e-01 7.11568594e-01 4.25394028e-01 1.96236193e-01 -4.80061248e-02 7.06657529e-01 -2.87471920e-01 4.18646075e-02 1.26999751e-01 4.34849411e-01 -4.55690503e-01 -8.29852581e-01 3.08748513e-01 1.27836406e+00 -1.41797945e-01 -6.21093035e-01 -6.34552717e-01 5.92588902e-01 1.17492184e-01 4.15281028e-01 -9.12574008e-02 -7.89030194e-01 2.78291345e-01 5.24822772e-02 8.37607801e-01 -6.56173229e-01 -7.67625630e-01 -4.27473068e-01 -9.14071202e-02 -8.75205457e-01 -1.20610297e-01 -1.97439075e-01 -1.29007149e+00 -4.82395440e-01 5.58333471e-03 -6.12132587e-02 3.93974222e-02 9.35796916e-01 4.98464316e-01 1.32191479e+00 2.54237384e-01 -6.62093282e-01 7.51434490e-02 -1.27908206e+00 -1.07016158e+00 2.83406824e-01 2.86143452e-01 -8.44790220e-01 -1.34420991e-01 -4.37875092e-01]
[12.371113777160645, 3.3952908515930176]
64a1b618-cb49-407e-98ae-09007e04ed98
boosting-weakly-supervised-temporal-action
2305.00607
null
https://arxiv.org/abs/2305.00607v1
https://arxiv.org/pdf/2305.00607v1.pdf
Boosting Weakly-Supervised Temporal Action Localization with Text Information
Due to the lack of temporal annotation, current Weakly-supervised Temporal Action Localization (WTAL) methods are generally stuck into over-complete or incomplete localization. In this paper, we aim to leverage the text information to boost WTAL from two aspects, i.e., (a) the discriminative objective to enlarge the inter-class difference, thus reducing the over-complete; (b) the generative objective to enhance the intra-class integrity, thus finding more complete temporal boundaries. For the discriminative objective, we propose a Text-Segment Mining (TSM) mechanism, which constructs a text description based on the action class label, and regards the text as the query to mine all class-related segments. Without the temporal annotation of actions, TSM compares the text query with the entire videos across the dataset to mine the best matching segments while ignoring irrelevant ones. Due to the shared sub-actions in different categories of videos, merely applying TSM is too strict to neglect the semantic-related segments, which results in incomplete localization. We further introduce a generative objective named Video-text Language Completion (VLC), which focuses on all semantic-related segments from videos to complete the text sentence. We achieve the state-of-the-art performance on THUMOS14 and ActivityNet1.3. Surprisingly, we also find our proposed method can be seamlessly applied to existing methods, and improve their performances with a clear margin. The code is available at https://github.com/lgzlIlIlI/Boosting-WTAL.
['Xinbo Gao', 'Xiaoyu Wang', 'Nannan Wang', 'Xinpeng Ding', 'De Cheng', 'Guozhang Li']
2023-05-01
null
http://openaccess.thecvf.com//content/CVPR2023/html/Li_Boosting_Weakly-Supervised_Temporal_Action_Localization_With_Text_Information_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Li_Boosting_Weakly-Supervised_Temporal_Action_Localization_With_Text_Information_CVPR_2023_paper.pdf
cvpr-2023-1
['weakly-supervised-temporal-action', 'action-localization', 'action-recognition']
['computer-vision', 'computer-vision', 'computer-vision']
[ 2.91619688e-01 -1.01996221e-01 -5.71364999e-01 -3.03269297e-01 -9.46536303e-01 -3.30647171e-01 5.07293046e-01 -9.01803225e-02 -4.83545691e-01 5.08596718e-01 4.50777739e-01 7.04759061e-02 1.06010824e-01 -4.23510104e-01 -5.73842108e-01 -8.48643005e-01 1.92387760e-01 5.04921526e-02 7.77446687e-01 8.46464187e-02 8.95657018e-02 -3.59517173e-03 -1.47714806e+00 5.14754534e-01 8.97694945e-01 1.05899024e+00 4.18110371e-01 7.81694651e-02 -2.60810435e-01 1.03650725e+00 -3.61207277e-01 -4.15655114e-02 -1.22190483e-01 -6.12132311e-01 -7.17744827e-01 2.73644209e-01 2.52106279e-01 -3.98308069e-01 -6.27880037e-01 1.05725586e+00 3.07314098e-01 2.85152584e-01 2.80914098e-01 -1.32231486e+00 -2.61746526e-01 5.66895187e-01 -8.02246988e-01 6.48032278e-02 3.34900588e-01 8.22298154e-02 1.05004227e+00 -1.05582583e+00 7.18028426e-01 1.16628754e+00 4.84822124e-01 5.16034484e-01 -8.82776082e-01 -6.00327253e-01 5.67252874e-01 5.17266035e-01 -1.58552623e+00 -4.09116536e-01 9.08698916e-01 -3.37973386e-01 6.04983926e-01 2.95887172e-01 6.03939593e-01 1.31907153e+00 -3.52828065e-03 1.27108324e+00 8.39257061e-01 -1.78861067e-01 1.85742855e-01 -2.42971808e-01 -3.96077223e-02 6.30358398e-01 -1.83321908e-01 -3.39228570e-01 -6.89959884e-01 1.72719538e-01 4.67876822e-01 4.00641590e-01 -3.53280187e-01 -2.43854582e-01 -1.35957766e+00 5.75643897e-01 2.79074818e-01 6.77167177e-01 -2.77802646e-01 1.36937112e-01 6.14300728e-01 -1.44292891e-01 5.79877019e-01 -1.70363158e-01 -4.49275345e-01 -3.66700828e-01 -1.21536493e+00 6.95503727e-02 3.89081448e-01 1.02329206e+00 8.88711274e-01 -1.65265709e-01 -5.28345406e-01 9.71535385e-01 2.54268914e-01 2.66984105e-01 6.56295896e-01 -8.61243188e-01 7.53240049e-01 6.95541203e-01 -9.33054462e-02 -9.01660502e-01 -2.35677004e-01 -4.61521864e-01 -6.78118765e-01 -2.84044683e-01 3.36452693e-01 2.15884093e-02 -9.28607464e-01 1.75400758e+00 3.65641594e-01 1.94610640e-01 -2.58271277e-01 1.00385678e+00 7.66574919e-01 7.10747600e-01 1.68373540e-01 -4.06910330e-01 1.26795805e+00 -1.36373222e+00 -8.21605980e-01 -3.25301409e-01 8.41929436e-01 -7.02685058e-01 1.13227391e+00 3.63217711e-01 -7.71758258e-01 -5.97174883e-01 -8.11846972e-01 -4.15250994e-02 -1.54970735e-01 5.13078809e-01 3.22434962e-01 6.55942708e-02 -7.11600900e-01 4.94782358e-01 -1.06698835e+00 -4.92813647e-01 5.28384745e-01 2.99656857e-02 -3.79607230e-01 -3.44657004e-01 -1.08866501e+00 3.55268657e-01 6.16755784e-01 6.44310191e-02 -1.05893779e+00 -3.69390368e-01 -8.26421976e-01 -1.52736217e-01 1.01154435e+00 -2.18198225e-01 1.11324751e+00 -1.23088014e+00 -1.07366753e+00 6.24558210e-01 -4.56657052e-01 -2.65078098e-01 6.50645673e-01 -2.51050115e-01 -3.11243922e-01 2.76376486e-01 4.45566893e-01 8.77098083e-01 7.46734679e-01 -9.44889307e-01 -8.66148174e-01 -3.97583097e-01 -5.09294588e-03 3.91045034e-01 -5.60321212e-01 -2.35198498e-01 -1.17125785e+00 -1.08147192e+00 2.97487199e-01 -9.63581920e-01 1.54541414e-02 -1.35687161e-02 -4.33651894e-01 -4.64233875e-01 1.09785211e+00 -8.31955075e-01 1.58004344e+00 -2.34081101e+00 2.18175292e-01 -2.37974241e-01 1.10817760e-01 1.17796458e-01 -2.15200722e-01 5.12746334e-01 5.24419621e-02 -9.02865380e-02 -3.00392091e-01 -6.38795495e-01 -1.88968211e-01 3.45008165e-01 -1.56302169e-01 4.04144853e-01 -1.03178341e-02 9.35521066e-01 -1.01973569e+00 -8.71095300e-01 2.92227209e-01 2.86573142e-01 -3.93269539e-01 -9.17303339e-02 -3.24667901e-01 5.83819747e-01 -7.66572237e-01 7.85113454e-01 5.50620019e-01 -1.80837750e-01 6.99078813e-02 -4.07762200e-01 -2.14431673e-01 2.24925697e-01 -1.14892340e+00 2.20375729e+00 -1.37964651e-01 4.85993773e-01 -4.59344536e-02 -1.17265821e+00 5.54688931e-01 1.30580992e-01 1.00559652e+00 -7.57625818e-01 -1.35912538e-01 2.37842530e-01 -2.84778833e-01 -6.72541738e-01 3.05023551e-01 1.21411249e-01 -4.95395586e-02 1.31789535e-01 1.69103488e-01 4.35981959e-01 4.09625232e-01 3.81804347e-01 1.14426553e+00 6.75492048e-01 1.12726472e-01 -5.01749367e-02 6.58451378e-01 -1.95855666e-02 8.35500419e-01 5.49378276e-01 -3.19330782e-01 6.72067642e-01 4.85874891e-01 -1.46980047e-01 -6.85994923e-01 -8.38452101e-01 4.00693677e-02 1.13924897e+00 5.30447841e-01 -6.77006781e-01 -7.56703496e-01 -1.05781019e+00 -3.43214989e-01 5.94604254e-01 -6.07769966e-01 -2.08248168e-01 -6.26199186e-01 -5.55807292e-01 5.23646474e-01 6.16692007e-01 7.12944806e-01 -1.09362149e+00 -3.22073370e-01 2.49448568e-01 -8.10080051e-01 -1.23651302e+00 -9.25618172e-01 1.19771548e-02 -8.26957524e-01 -9.06056285e-01 -9.08474982e-01 -7.50160992e-01 4.90562618e-01 4.87310171e-01 5.59521198e-01 -1.42483963e-02 -6.74180984e-02 1.51771918e-01 -7.64617145e-01 1.01375952e-01 -1.22084670e-01 7.28701279e-02 -2.28899315e-01 3.46294522e-01 3.00172597e-01 -4.64676827e-01 -7.76577532e-01 5.96936405e-01 -1.00931370e+00 3.15917343e-01 6.16118312e-01 6.90114260e-01 8.07616591e-01 1.31785855e-01 4.67793763e-01 -4.37314242e-01 -1.94849297e-02 -3.58529687e-01 -1.71143591e-01 2.28960291e-01 -3.40862900e-01 -1.52521074e-01 5.43954492e-01 -6.59869373e-01 -1.02305913e+00 1.58440039e-01 -1.93704009e-01 -6.82751298e-01 -1.10425428e-01 5.12004018e-01 -4.04513627e-01 2.00393170e-01 2.68304616e-01 6.77276552e-01 -3.22338879e-01 -5.72078586e-01 1.24671042e-01 5.03839195e-01 3.67887467e-01 -4.81083870e-01 4.75354254e-01 6.67431176e-01 -1.90503031e-01 -7.68682301e-01 -1.06348467e+00 -8.30993533e-01 -5.16110718e-01 -3.88844252e-01 1.04695320e+00 -9.39597845e-01 -2.39195183e-01 4.35838491e-01 -1.02455688e+00 -3.45455170e-01 -2.54164159e-01 7.15198398e-01 -5.15908062e-01 8.17326605e-01 -5.83561242e-01 -6.28688514e-01 -7.10895658e-02 -1.21063769e+00 1.26957989e+00 3.52810435e-02 1.07895015e-02 -6.17043555e-01 -1.42121151e-01 5.85241258e-01 -5.45471162e-02 3.06457728e-02 4.89683628e-01 -6.64834380e-01 -6.27297699e-01 -9.59520340e-02 -2.32190847e-01 4.31490541e-01 1.23016372e-01 -2.30790049e-01 -8.66833687e-01 -2.68277943e-01 -5.42909950e-02 -2.87688524e-01 1.17857039e+00 3.89132261e-01 1.35751438e+00 -2.81312704e-01 -4.69131768e-01 4.82425004e-01 1.13234651e+00 2.30117485e-01 7.70815790e-01 2.59114772e-01 8.99227083e-01 5.89461923e-01 1.17581058e+00 5.14231384e-01 3.06320965e-01 9.45683599e-01 4.11701202e-01 -2.96887718e-02 -2.58405268e-01 -6.03834748e-01 7.37387359e-01 7.61576295e-01 -8.57760012e-02 -3.60198289e-01 -6.76747203e-01 5.91625512e-01 -2.20081377e+00 -1.09259963e+00 -1.42456964e-01 2.11072135e+00 7.94904351e-01 1.35827273e-01 2.43180528e-01 7.49508739e-02 8.05163860e-01 5.19189477e-01 -5.82993865e-01 4.33937758e-01 -5.99405318e-02 -2.07521960e-01 3.91033173e-01 1.87768251e-01 -1.30395377e+00 9.72071767e-01 4.18344164e+00 1.64928317e+00 -9.22400713e-01 4.57837611e-01 5.58861852e-01 -1.92274004e-01 -5.28994165e-02 2.09885746e-01 -8.64842176e-01 7.20716536e-01 3.64240527e-01 2.17017904e-01 3.24854821e-01 7.45401919e-01 6.10236883e-01 -2.78401881e-01 -9.05953467e-01 9.85102594e-01 1.83886573e-01 -9.83829558e-01 1.14432663e-01 -1.96622536e-02 6.49561286e-01 5.58395877e-05 -2.97793806e-01 4.82956022e-01 -4.52641159e-01 -5.08427322e-01 1.03571534e+00 4.32737529e-01 7.77006686e-01 -5.48386812e-01 6.01569772e-01 5.96459627e-01 -1.58968043e+00 -1.34559408e-01 -1.18657082e-01 2.89394736e-01 2.45180845e-01 6.75841153e-01 -2.52049595e-01 7.52663136e-01 8.10897768e-01 1.20385206e+00 -4.70625967e-01 9.66667891e-01 -3.49698484e-01 7.05234826e-01 -2.27702245e-01 1.21084146e-01 5.05417228e-01 -2.71081060e-01 6.45558953e-01 1.16831183e+00 4.01088834e-01 -1.67409003e-01 6.19147837e-01 6.39318824e-01 -2.09132172e-02 2.85628855e-01 -2.32113719e-01 -8.54171440e-02 3.42820078e-01 1.21158385e+00 -9.02683318e-01 -4.68746394e-01 -6.21939957e-01 1.26245022e+00 1.61434822e-02 4.92169291e-01 -1.17968571e+00 -2.95682758e-01 1.91206127e-01 1.35347679e-01 3.45510066e-01 -1.52488545e-01 -7.86193274e-03 -1.42703867e+00 5.09389758e-01 -7.59571373e-01 5.89377105e-01 -7.68388629e-01 -8.92447948e-01 2.98901916e-01 1.00336120e-01 -1.63559973e+00 1.18572034e-01 -2.38407347e-02 -4.09034044e-01 4.29333806e-01 -1.28031421e+00 -1.33387101e+00 -4.97919410e-01 6.54617429e-01 1.12672710e+00 1.95104167e-01 1.44294336e-01 7.50913143e-01 -7.12610185e-01 5.07333398e-01 -7.50317648e-02 1.70785904e-01 8.55524957e-01 -8.53098869e-01 -1.49217933e-01 9.51707244e-01 1.84168339e-01 2.48197392e-01 4.37580675e-01 -8.34051371e-01 -1.09109533e+00 -1.37676406e+00 8.37281406e-01 -1.31119162e-01 7.68025041e-01 -4.07219619e-01 -9.37293231e-01 6.38454914e-01 -1.27984151e-01 -6.20869957e-02 1.97793230e-01 -2.91095883e-01 -1.18830346e-01 -3.08386803e-01 -6.84046805e-01 6.61156833e-01 1.35029233e+00 -4.51546222e-01 -4.47935998e-01 3.65610212e-01 9.44395423e-01 -1.47448733e-01 -6.25598192e-01 4.89444286e-01 4.24694568e-01 -8.48815024e-01 7.52577603e-01 -1.49168134e-01 5.44332147e-01 -5.57808876e-01 -2.01509207e-01 -7.76398957e-01 -9.66937765e-02 -5.13071060e-01 -1.33236587e-01 1.50041401e+00 1.09045133e-01 -3.44470233e-01 7.51381755e-01 6.57023787e-02 -4.51628089e-01 -8.59343350e-01 -1.17963779e+00 -8.51122856e-01 -4.11240429e-01 -6.69118881e-01 2.07256421e-01 8.94366801e-01 1.22748408e-03 2.50700623e-01 -6.34579659e-01 -8.77042189e-02 5.29363215e-01 6.57704026e-02 5.01299858e-01 -7.70196021e-01 -3.09964806e-01 -3.48170847e-01 -2.51242667e-01 -1.52215421e+00 1.00084610e-01 -8.53682160e-01 3.23598176e-01 -1.46064270e+00 5.36086977e-01 -3.07580471e-01 -3.12047184e-01 8.09803843e-01 -9.26369354e-02 2.49939278e-01 2.05129534e-01 5.36980927e-01 -1.17833948e+00 8.59277010e-01 1.45098794e+00 -1.82910576e-01 -2.63420343e-01 -4.65427488e-02 -3.67195487e-01 7.94023275e-01 6.09198451e-01 -5.79010069e-01 -4.84673232e-01 -2.98157811e-01 -1.77968666e-01 1.67513773e-01 4.66246456e-01 -1.02786481e+00 2.73145854e-01 -2.47538298e-01 2.28170648e-01 -8.55116606e-01 3.52550030e-01 -7.10618377e-01 -3.23634706e-02 4.65002894e-01 -3.14637065e-01 -3.59301656e-01 -8.61041900e-03 7.84364045e-01 -4.60321188e-01 -2.81339705e-01 6.30737722e-01 -6.99984208e-02 -9.56711054e-01 5.31720757e-01 -2.91527212e-01 1.14212953e-01 1.00670385e+00 -2.38660812e-01 -2.62233883e-01 -3.19338560e-01 -7.73657024e-01 4.11160201e-01 4.63016510e-01 5.30080438e-01 4.88098323e-01 -1.39675832e+00 -5.12026906e-01 2.42793909e-03 3.66760850e-01 -6.62670061e-02 6.64583802e-01 1.37005067e+00 -9.81316715e-02 4.68255550e-01 2.39464089e-01 -7.69953549e-01 -1.24750900e+00 5.66070795e-01 1.77811563e-01 -4.04736727e-01 -7.21022010e-01 4.80356455e-01 7.54831791e-01 4.60434407e-02 4.59444821e-01 -2.64182359e-01 -1.98427305e-01 1.87697306e-01 4.25014228e-01 3.40335071e-01 -6.41076267e-02 -7.40436554e-01 -5.20722628e-01 6.65993512e-01 2.62218192e-02 -6.65686652e-02 1.09201527e+00 -2.80596197e-01 6.13654628e-02 3.73096287e-01 1.38765812e+00 -9.40286741e-02 -1.41384816e+00 -3.18723649e-01 -1.16537912e-02 -4.17923898e-01 1.13684997e-01 -6.42251074e-01 -1.19456220e+00 8.43434334e-01 5.27410507e-01 -1.30349070e-01 1.33776188e+00 2.51879245e-01 1.01744008e+00 1.36196584e-01 3.18629056e-01 -1.27083564e+00 4.49661940e-01 3.87058496e-01 7.52447069e-01 -1.09195530e+00 1.94027647e-02 -4.53927785e-01 -7.85247326e-01 9.32679176e-01 6.06640279e-01 3.10220510e-01 3.29319954e-01 -7.31935576e-02 -3.42630863e-01 -4.50864621e-02 -5.58216155e-01 -4.55826133e-01 3.74955297e-01 2.73491710e-01 2.60096401e-01 -1.13404714e-01 -6.46619916e-01 6.08497798e-01 5.77679396e-01 1.81120306e-01 1.21843427e-01 8.83281052e-01 -5.06040871e-01 -1.11812556e+00 -1.32340014e-01 3.37461203e-01 -5.30181289e-01 -6.56126416e-04 -1.94034919e-01 6.91275120e-01 3.91848445e-01 9.34676588e-01 -1.54197350e-01 -3.83764863e-01 3.31834823e-01 -1.67935211e-02 1.90426603e-01 -4.48063612e-01 -5.31168841e-02 6.71823680e-01 8.33446234e-02 -8.07649910e-01 -6.83738232e-01 -8.49789619e-01 -1.44313407e+00 1.45909190e-01 -2.90200710e-01 4.85511310e-02 3.12062204e-01 9.53258157e-01 3.10801893e-01 4.80500340e-01 5.58213174e-01 -7.25468218e-01 -3.46825063e-01 -1.07665122e+00 -5.83813429e-01 5.43994725e-01 9.50332880e-02 -7.34328985e-01 -4.43925142e-01 2.01573700e-01]
[8.852846145629883, 0.6098569631576538]
b541927d-cd9b-4831-ac01-881124ef27c9
cross-view-action-recognition-understanding
2305.15699
null
https://arxiv.org/abs/2305.15699v1
https://arxiv.org/pdf/2305.15699v1.pdf
Cross-view Action Recognition Understanding From Exocentric to Egocentric Perspective
Understanding action recognition in egocentric videos has emerged as a vital research topic with numerous practical applications. With the limitation in the scale of egocentric data collection, learning robust deep learning-based action recognition models remains difficult. Transferring knowledge learned from the large-scale exocentric data to the egocentric data is challenging due to the difference in videos across views. Our work introduces a novel cross-view learning approach to action recognition (CVAR) that effectively transfers knowledge from the exocentric to the egocentric view. First, we introduce a novel geometric-based constraint into the self-attention mechanism in Transformer based on analyzing the camera positions between two views. Then, we propose a new cross-view self-attention loss learned on unpaired cross-view data to enforce the self-attention mechanism learning to transfer knowledge across views. Finally, to further improve the performance of our cross-view learning approach, we present the metrics to measure the correlations in videos and attention maps effectively. Experimental results on standard egocentric action recognition benchmarks, i.e., Charades-Ego, EPIC-Kitchens-55, and EPIC-Kitchens-100, have shown our approach's effectiveness and state-of-the-art performance.
['Khoa Luu', 'Thanh-Dat Truong']
2023-05-25
null
null
null
null
['action-recognition-in-videos']
['computer-vision']
[ 2.89892871e-03 -3.03397775e-01 -2.58545876e-01 -3.70639086e-01 -4.34910595e-01 -3.50082457e-01 5.45752406e-01 -7.81706810e-01 -2.86381930e-01 4.28051561e-01 7.21949577e-01 4.51241434e-01 -2.45759696e-01 -4.11827058e-01 -8.72112155e-01 -8.13096285e-01 1.10263281e-01 -4.23700847e-02 2.86457598e-01 -5.18447980e-02 2.39020050e-01 4.10184562e-01 -1.29994714e+00 4.10275191e-01 5.76679289e-01 9.35346007e-01 9.53317210e-02 7.07487285e-01 4.23680484e-01 1.31022620e+00 -2.63013005e-01 -3.32302183e-01 4.24334198e-01 -5.65638840e-01 -7.26343453e-01 2.72999138e-01 6.30412400e-01 -7.52574742e-01 -1.12856126e+00 9.77341235e-01 5.00463068e-01 3.88588041e-01 3.65527779e-01 -1.41374159e+00 -7.50745177e-01 2.48702571e-01 -9.21155155e-01 4.67636079e-01 4.14037138e-01 3.70166928e-01 7.29678214e-01 -7.45024920e-01 6.68893456e-01 1.31066775e+00 3.92193496e-01 4.95670080e-01 -7.63300836e-01 -6.72728002e-01 5.31835914e-01 8.53080690e-01 -1.18443346e+00 -4.28155929e-01 9.41482127e-01 -5.69519520e-01 9.12485361e-01 -2.85297986e-02 7.82242477e-01 1.37723207e+00 3.16976488e-01 1.11342156e+00 8.15582752e-01 -8.30328614e-02 1.48510292e-01 -4.16906834e-01 -3.23353142e-01 5.00241816e-01 3.09060216e-02 -5.80411665e-02 -6.71492100e-01 3.79850090e-01 1.14146900e+00 4.57919627e-01 -3.96455020e-01 -1.03760254e+00 -1.34483278e+00 7.55415976e-01 4.86047387e-01 -2.43136194e-02 -3.89690071e-01 2.68518507e-01 8.33146214e-01 -9.29790270e-03 4.74756509e-01 2.27887571e-01 -4.94062692e-01 -5.65255940e-01 -2.68256575e-01 -9.71639678e-02 1.20959908e-01 1.15400529e+00 3.53520602e-01 1.82914287e-01 -1.49627775e-01 6.28893137e-01 9.87180844e-02 5.36401868e-01 5.61039388e-01 -1.02252316e+00 8.23479950e-01 7.07127690e-01 -6.48735911e-02 -1.07230842e+00 -3.28041166e-01 -3.63765568e-01 -7.38554716e-01 1.94429066e-02 3.88035446e-01 -8.82951319e-02 -5.46629369e-01 1.76985824e+00 3.60070199e-01 3.86414438e-01 1.52206391e-01 1.12946165e+00 5.37219644e-01 2.52029836e-01 -4.35512476e-02 -1.02114975e-01 1.12727022e+00 -1.29962528e+00 -6.07459664e-01 -1.82733104e-01 7.43581235e-01 -3.68544608e-01 1.03405678e+00 1.65740296e-01 -9.69079256e-01 -6.25464261e-01 -8.27531934e-01 -1.44181952e-01 -1.48605168e-01 3.38989139e-01 5.90519190e-01 1.52700514e-01 -4.44115520e-01 3.37641656e-01 -1.15764236e+00 -5.73499799e-01 7.54505038e-01 1.13865890e-01 -8.77705216e-01 -2.55618632e-01 -9.89972532e-01 6.11208975e-01 3.83359492e-01 4.56324685e-03 -1.07117558e+00 -6.43127561e-01 -1.00113404e+00 1.42128900e-01 5.62845707e-01 -6.54315948e-01 9.28891838e-01 -1.15305173e+00 -1.53284371e+00 6.75130188e-01 1.61147341e-01 -2.36876741e-01 6.94910347e-01 -6.33076906e-01 -4.84993190e-01 5.26059985e-01 2.62625754e-01 5.00603795e-01 6.60786152e-01 -8.99972200e-01 -6.39892876e-01 -7.93442190e-01 5.10690749e-01 5.01121938e-01 -3.44749123e-01 -1.30369514e-01 -5.97904027e-01 -8.21729243e-01 2.38166615e-01 -1.02498043e+00 1.69735909e-01 4.56016697e-02 -1.83481574e-01 -1.02857843e-01 1.03644395e+00 -6.28930092e-01 8.64517629e-01 -2.23921084e+00 5.16215920e-01 -4.14511740e-01 5.06641530e-02 2.48540699e-01 -1.71014130e-01 2.85723627e-01 -4.93276864e-01 -5.49794257e-01 2.02541336e-01 1.15922615e-01 -2.87652820e-01 2.31276020e-01 -3.62155885e-01 7.47274041e-01 1.07903749e-01 1.00355089e+00 -1.20945883e+00 -2.66063124e-01 5.15693963e-01 4.36401486e-01 -7.83530176e-01 3.91647846e-01 1.88740388e-01 7.22689509e-01 -5.65234542e-01 5.41711807e-01 4.88429338e-01 -2.41216496e-01 -6.07926957e-02 -4.42121029e-01 1.98895514e-01 -1.17332898e-01 -9.30178106e-01 2.09950066e+00 -4.57430094e-01 6.34156048e-01 -3.21622699e-01 -1.12083697e+00 5.54951906e-01 1.42108217e-01 8.93842936e-01 -7.47980118e-01 1.49446204e-01 -3.20144743e-01 -9.10983011e-02 -9.76006269e-01 1.88327849e-01 3.26576158e-02 -2.43592262e-02 4.14624751e-01 3.26199830e-01 4.26174939e-01 1.65630475e-01 1.32392794e-01 1.07089126e+00 6.89404607e-01 4.12857920e-01 7.07835779e-02 8.15332651e-01 -2.17357919e-01 8.43816638e-01 3.14957708e-01 -5.58454692e-01 6.72160923e-01 6.11071646e-01 -5.27517796e-01 -8.19507420e-01 -1.03518879e+00 2.68922538e-01 1.10648787e+00 1.87107310e-01 -3.37113082e-01 -7.39363432e-01 -1.21969450e+00 -1.13875158e-01 5.02457857e-01 -9.20452118e-01 -6.30698502e-01 -6.85006440e-01 -3.02943081e-01 4.17804748e-01 1.17560351e+00 1.02815449e+00 -9.16090250e-01 -7.41630197e-01 -2.77870238e-01 -3.79453689e-01 -1.44174933e+00 -7.40293801e-01 -3.04175347e-01 -8.86923254e-01 -1.37008107e+00 -9.06948566e-01 -3.73893112e-01 7.20485747e-01 7.12503314e-01 7.16169655e-01 -5.92693806e-01 -1.90316275e-01 9.33353364e-01 -5.72512686e-01 -2.72024572e-02 4.34245974e-01 -3.11349422e-01 3.25612336e-01 4.98409748e-01 6.64432406e-01 -6.43913507e-01 -8.69112790e-01 5.62426329e-01 -6.65114224e-01 6.78459927e-02 6.36688232e-01 8.47426593e-01 4.76371020e-01 -2.21554622e-01 4.25523788e-01 -4.22302127e-01 -1.81871548e-01 -5.70232689e-01 -2.88041681e-01 2.12223917e-01 5.58284335e-02 -2.76363730e-01 6.82793260e-01 -3.33437711e-01 -1.11120117e+00 2.88214684e-01 2.80509114e-01 -1.14525390e+00 -1.17386200e-01 1.14374265e-01 -7.90242970e-01 -3.06495111e-02 2.49506131e-01 4.25278693e-01 2.70110741e-02 -2.98783213e-01 4.46815878e-01 4.27168936e-01 5.68692505e-01 -1.99645773e-01 5.46498239e-01 9.76589203e-01 -1.36246189e-01 -7.05887854e-01 -1.12356031e+00 -5.50669611e-01 -1.08923316e+00 -4.93880868e-01 1.34930193e+00 -1.23177600e+00 -8.38608921e-01 6.43425047e-01 -8.83975446e-01 -2.13315651e-01 -3.78416359e-01 1.01459539e+00 -1.18719471e+00 6.94421470e-01 -2.16707066e-01 -3.96200240e-01 2.82430463e-02 -1.09203982e+00 1.13934410e+00 1.84465617e-01 7.16616213e-02 -9.42140281e-01 2.48325303e-01 7.97133982e-01 -1.01099275e-01 2.06593335e-01 4.34034675e-01 -3.39035273e-01 -6.15796030e-01 -9.94259194e-02 -3.38552952e-01 4.61434752e-01 2.95787573e-01 -3.91081303e-01 -8.53454530e-01 -3.92899901e-01 1.61105633e-01 -4.68643844e-01 8.98025751e-01 3.47262532e-01 1.45307660e+00 -1.14733905e-01 -2.55814582e-01 9.82818365e-01 1.08768642e+00 1.99253082e-01 9.13122654e-01 3.40422451e-01 1.10042453e+00 5.53549230e-01 8.39103401e-01 5.52996099e-01 4.48705971e-01 9.03492093e-01 5.46516061e-01 2.11883605e-01 -2.11490672e-02 -4.85220462e-01 6.25169992e-01 7.57053494e-01 -3.42169017e-01 6.67534471e-02 -4.60269690e-01 4.85753477e-01 -2.16906977e+00 -1.45523775e+00 2.35209882e-01 2.06831121e+00 1.76783711e-01 -2.03048468e-01 1.28316209e-01 -2.24577695e-01 6.01105928e-01 3.82829338e-01 -9.50098634e-01 7.34282583e-02 9.02470276e-02 -4.17195469e-01 4.70350921e-01 3.17880511e-02 -1.56036258e+00 8.37756515e-01 5.48839951e+00 5.01019776e-01 -1.16051733e+00 8.25080648e-02 2.76748478e-01 -4.19728965e-01 4.22848970e-01 -2.65242994e-01 -5.38325667e-01 6.38736367e-01 4.00236696e-01 6.29029144e-03 2.26071790e-01 1.21392190e+00 2.85401344e-01 2.32753754e-02 -1.42022073e+00 1.45383596e+00 6.87220037e-01 -9.92606521e-01 9.33088884e-02 7.85180554e-02 9.20315385e-01 4.92944643e-02 -4.18810509e-02 4.66185600e-01 1.53714940e-01 -6.76665425e-01 3.79963189e-01 6.86799169e-01 6.62055433e-01 -7.55731344e-01 5.26754200e-01 1.27959356e-01 -1.23072791e+00 -4.25502449e-01 -5.04245639e-01 -1.28872693e-01 2.56862581e-01 -4.16641049e-02 -5.66959679e-02 7.17612743e-01 8.50868106e-01 1.63313723e+00 -5.50867379e-01 6.93356752e-01 -1.77924529e-01 1.27912760e-01 2.19287187e-01 4.60548609e-01 2.92243928e-01 -4.35695261e-01 6.33084714e-01 7.50278533e-01 1.56932816e-01 3.60365540e-01 1.96302347e-02 6.43493176e-01 1.03474893e-01 -1.73668087e-01 -8.65876019e-01 5.58181405e-02 -2.60456175e-01 1.04576302e+00 -3.60281169e-01 -2.77247012e-01 -6.98516309e-01 1.20820391e+00 5.02854526e-01 4.57846642e-01 -1.06036127e+00 -2.70440966e-01 1.05373728e+00 1.09934493e-03 7.27796733e-01 -2.45335251e-01 2.32304305e-01 -1.70995200e+00 1.73012331e-01 -8.58803093e-01 5.01301110e-01 -9.61420298e-01 -1.13335836e+00 1.89606935e-01 2.29525954e-01 -1.75119841e+00 -2.77461350e-01 -7.37070858e-01 -6.27828419e-01 3.43915641e-01 -1.12557244e+00 -1.25691473e+00 -6.86605275e-01 8.85613322e-01 8.66182864e-01 -5.78626633e-01 4.27040160e-01 3.96971911e-01 -6.67189360e-01 6.33977175e-01 3.16882610e-01 5.24913549e-01 8.08372974e-01 -9.89425242e-01 1.43324345e-01 8.13794971e-01 1.70248047e-01 5.19896626e-01 1.26894102e-01 -4.40273464e-01 -1.68824506e+00 -1.24038076e+00 1.60631001e-01 -7.75867462e-01 6.41022384e-01 -2.24535793e-01 -7.09727883e-01 1.15945208e+00 2.76586250e-03 3.37422252e-01 6.63051903e-01 -4.18445170e-02 -4.81441587e-01 -3.43055815e-01 -6.40043259e-01 5.99124014e-01 1.34340143e+00 -6.78668261e-01 -5.73997855e-01 4.06894892e-01 3.13152105e-01 -4.22655642e-01 -9.12232637e-01 3.84999216e-01 8.56697857e-01 -1.11286128e+00 1.00405538e+00 -1.12458205e+00 8.04691195e-01 -3.88603032e-01 -3.29267293e-01 -1.33568704e+00 -4.65414912e-01 -3.40711981e-01 -2.23890081e-01 1.02639902e+00 -4.58091468e-01 -6.30389571e-01 8.68168771e-01 1.96348876e-01 -1.49391264e-01 -8.01218569e-01 -1.05673516e+00 -8.69623840e-01 -8.12886581e-02 -1.92723095e-01 2.70005018e-01 1.00140917e+00 1.44437596e-01 2.00467646e-01 -6.74265206e-01 -1.28524555e-02 5.53679883e-01 1.08540662e-01 1.19467890e+00 -7.64790118e-01 -4.17200655e-01 -1.48371518e-01 -1.07195067e+00 -1.32706273e+00 2.91769505e-01 -6.47842944e-01 -2.56973296e-01 -1.37559986e+00 5.60416758e-01 2.52149433e-01 -4.96514440e-01 1.45749211e-01 -1.06322587e-01 2.59140998e-01 2.43295640e-01 1.30029857e-01 -8.62762630e-01 1.12879729e+00 1.54487848e+00 -1.00287825e-01 1.75691962e-01 -2.25546226e-01 -5.35182476e-01 1.02739871e+00 5.40518284e-01 -2.36442208e-01 -7.65074193e-01 -5.41289985e-01 -1.00508504e-01 -1.23069823e-01 4.90768671e-01 -1.19836867e+00 1.74701393e-01 -3.02974254e-01 6.02670908e-01 -6.37109876e-01 5.49100459e-01 -9.19687688e-01 -1.97547212e-01 4.25422162e-01 -3.09009790e-01 -8.32417700e-03 -8.27594250e-02 1.01926064e+00 -2.86452323e-01 2.53123909e-01 7.23870039e-01 -1.57783329e-01 -9.80104744e-01 4.78584260e-01 1.47066370e-01 4.84674513e-01 1.54075551e+00 -3.94400299e-01 -4.02387142e-01 -5.47550201e-01 -6.34302855e-01 3.91820341e-01 5.08011639e-01 8.52491796e-01 6.98691905e-01 -1.68617415e+00 -5.22184253e-01 4.15264040e-01 4.63527173e-01 -2.28482217e-01 7.95740187e-01 1.26739764e+00 -4.11612570e-01 6.21867180e-01 -8.03191781e-01 -8.23803246e-01 -1.37670779e+00 8.55932176e-01 4.28665847e-01 -1.76497437e-02 -9.59734321e-01 6.32797778e-01 1.07341373e+00 -2.66170144e-01 2.52342701e-01 -1.16980083e-01 -3.06418687e-01 -1.25124007e-01 6.73981845e-01 7.08623946e-01 -4.19583499e-01 -1.05513656e+00 -4.38750535e-01 9.57296789e-01 -2.35898778e-01 2.01632857e-01 1.27542210e+00 -1.08119823e-01 2.30876938e-01 3.93304527e-01 1.33037531e+00 -4.18851554e-01 -1.84296989e+00 -1.79410622e-01 -5.12227595e-01 -1.10620964e+00 -1.06267482e-01 -4.04765755e-01 -1.44334435e+00 1.15127790e+00 6.89399779e-01 -5.84722996e-01 1.00506413e+00 -1.17580198e-01 6.63325071e-01 4.84826326e-01 3.31621468e-01 -1.37511492e+00 6.87234282e-01 5.55553794e-01 1.21947789e+00 -1.47765601e+00 1.96476027e-01 -1.74226865e-01 -1.07848024e+00 9.73548293e-01 1.14082575e+00 -2.94332981e-01 5.43879509e-01 -3.77059758e-01 -7.04521537e-02 -2.38484532e-01 -5.58150530e-01 3.26343477e-02 2.43769303e-01 7.21800804e-01 1.45977914e-01 -1.23741679e-01 7.60051534e-02 4.37261492e-01 2.42146283e-01 7.56730419e-03 4.61290717e-01 8.93564343e-01 -8.20422620e-02 -5.52272439e-01 -8.06712285e-02 1.98967546e-01 -2.95662433e-01 4.78768378e-01 -4.75237101e-01 8.74297500e-01 8.06809496e-03 6.91324890e-01 2.72856385e-01 -4.34982270e-01 5.18218577e-01 -1.32946819e-01 7.12234974e-01 -4.14625108e-01 -1.80793889e-02 -7.31132403e-02 -2.59257853e-01 -1.06014442e+00 -6.36505961e-01 -8.39870393e-01 -9.39705908e-01 -9.10926908e-02 -6.97423741e-02 -2.34360293e-01 2.08932191e-01 1.00388992e+00 6.52098417e-01 6.86848521e-01 7.70671368e-01 -9.08061683e-01 -6.97077870e-01 -9.29080248e-01 -6.53169572e-01 6.15165949e-01 1.53840199e-01 -1.18460917e+00 -1.51768938e-01 1.02989018e-01]
[8.34745979309082, 0.5540667176246643]
0932f41e-d589-4efa-9b7f-a4fe2bb5b901
nprf-a-neural-pseudo-relevance-feedback
1810.12936
null
http://arxiv.org/abs/1810.12936v1
http://arxiv.org/pdf/1810.12936v1.pdf
NPRF: A Neural Pseudo Relevance Feedback Framework for Ad-hoc Information Retrieval
Pseudo-relevance feedback (PRF) is commonly used to boost the performance of traditional information retrieval (IR) models by using top-ranked documents to identify and weight new query terms, thereby reducing the effect of query-document vocabulary mismatches. While neural retrieval models have recently demonstrated strong results for ad-hoc retrieval, combining them with PRF is not straightforward due to incompatibilities between existing PRF approaches and neural architectures. To bridge this gap, we propose an end-to-end neural PRF framework that can be used with existing neural IR models by embedding different neural models as building blocks. Extensive experiments on two standard test collections confirm the effectiveness of the proposed NPRF framework in improving the performance of two state-of-the-art neural IR models.
['Kai Hui', 'Andrew Yates', 'Canjia Li', 'Ben He', 'Yingfei Sun', 'Le Wang', 'Jungang Xu', 'Le Sun']
2018-10-30
nprf-a-neural-pseudo-relevance-feedback-1
https://aclanthology.org/D18-1478
https://aclanthology.org/D18-1478.pdf
emnlp-2018-10
['ad-hoc-information-retrieval']
['natural-language-processing']
[ 2.24603072e-01 -1.33752733e-01 -6.54217303e-01 -3.07498872e-01 -9.67365682e-01 -4.36765611e-01 8.32391381e-01 2.34834045e-01 -6.64387345e-01 3.31022590e-01 4.08208400e-01 -3.04228872e-01 -4.94540840e-01 -5.53238928e-01 -5.96593797e-01 -1.12748798e-02 8.46326128e-02 6.09399259e-01 2.67782092e-01 -7.78073490e-01 5.23723543e-01 2.79610366e-01 -1.60648012e+00 6.68781221e-01 5.85944474e-01 1.07088017e+00 6.17082529e-02 3.34159017e-01 -2.60618120e-01 7.03524172e-01 -5.33602417e-01 -2.10483089e-01 4.83589172e-01 1.25261247e-01 -8.20466638e-01 -8.71629477e-01 8.18043470e-01 -5.61794579e-01 -7.52898395e-01 6.91764295e-01 5.57184935e-01 4.67047274e-01 7.88351417e-01 -6.72087729e-01 -1.59316051e+00 8.04218411e-01 -5.30162513e-01 3.27808291e-01 3.04942578e-01 -5.42767227e-01 1.33203995e+00 -1.11685991e+00 5.23084581e-01 1.48710799e+00 4.07414198e-01 7.30859578e-01 -9.44849730e-01 -6.40809596e-01 2.53551155e-01 7.81465620e-02 -1.42099440e+00 -4.10292566e-01 6.78410470e-01 9.47193578e-02 1.35793412e+00 3.44480842e-01 1.85883671e-01 8.76352131e-01 3.55482511e-02 1.18554926e+00 4.49981183e-01 -9.00836766e-01 -1.09877288e-01 2.31508315e-01 7.82116115e-01 1.97213247e-01 3.72402519e-02 4.26877826e-01 -4.77339774e-01 -4.40740615e-01 5.52713633e-01 1.85615420e-01 -3.31264794e-01 -1.72727063e-01 -7.73679376e-01 9.73569214e-01 9.45956886e-01 5.14609575e-01 -4.81626689e-01 2.16389120e-01 4.20300305e-01 4.71458346e-01 4.82522190e-01 9.31247413e-01 -3.12251478e-01 5.58839738e-01 -9.46532786e-01 5.75989842e-01 3.76373738e-01 6.57673478e-01 5.10966718e-01 -8.49633366e-02 -8.40933144e-01 1.33327675e+00 7.14130282e-01 3.23347926e-01 8.53048027e-01 -7.80027449e-01 1.80869520e-01 6.96952224e-01 4.04771743e-03 -9.74814415e-01 -1.78184673e-01 -5.02882719e-01 -5.73152602e-01 -2.72799700e-01 -2.47773126e-01 4.40387070e-01 -1.06246006e+00 1.53173959e+00 -1.10560298e-01 -1.96207643e-01 2.41426036e-01 9.77252781e-01 9.92858768e-01 8.94120276e-01 2.31744781e-01 1.08106062e-02 1.22549415e+00 -1.21480834e+00 -6.79100752e-01 -5.67295074e-01 4.60229695e-01 -7.49624312e-01 1.10979164e+00 1.02537833e-01 -9.13053274e-01 -5.99717677e-01 -1.08876514e+00 -4.15937454e-01 -5.76326847e-01 1.62731096e-01 7.78337121e-01 3.50163519e-01 -1.47752249e+00 2.80551285e-01 -2.72643417e-01 -1.37395680e-01 -1.31375371e-02 3.90691310e-01 -1.51928186e-01 -3.55602294e-01 -1.85898638e+00 9.19700265e-01 5.75608373e-01 3.08079302e-01 -7.22462237e-01 -7.22625434e-01 -5.50675392e-01 3.61847013e-01 2.53565013e-01 -8.12840104e-01 1.56600142e+00 -7.25134671e-01 -1.28109884e+00 3.73078167e-01 9.71907657e-03 -5.18113017e-01 -6.56733885e-02 -8.07782292e-01 -3.75445396e-01 1.33485105e-02 -2.95137316e-01 1.11253738e+00 7.21375108e-01 -1.18329751e+00 -2.89044619e-01 -3.59425366e-01 2.12345228e-01 5.38062215e-01 -7.88864136e-01 2.42211178e-01 -6.82392359e-01 -6.52966857e-01 -2.02645674e-01 -8.18696737e-01 -1.79962412e-01 -2.28254452e-01 -4.25971597e-02 -8.71907473e-01 7.02720046e-01 -3.99369776e-01 1.53818786e+00 -2.00141382e+00 -1.41661987e-01 1.78425804e-01 -3.92209627e-02 6.52137041e-01 -7.23051965e-01 5.56198776e-01 -2.89035067e-02 2.05547333e-01 3.23809505e-01 -1.00458622e-01 8.53841454e-02 -2.01206237e-01 -6.68486476e-01 -2.76472300e-01 1.36298433e-01 1.24966705e+00 -6.79879069e-01 -2.24010721e-01 -1.54732496e-01 7.94853866e-01 -4.14123774e-01 1.58259213e-01 -4.35948819e-01 -4.78573322e-01 -6.04778707e-01 4.78799552e-01 2.22747803e-01 -3.40656161e-01 -1.45595763e-02 -6.00808412e-02 4.23167855e-01 5.26232243e-01 -7.10506976e-01 1.81292021e+00 -2.96146065e-01 4.15885627e-01 -4.37221020e-01 -7.52563953e-01 1.06282759e+00 4.05730665e-01 1.97430462e-01 -1.36256027e+00 -1.17938668e-02 9.82555076e-02 -2.27746651e-01 -4.09020595e-02 1.22145283e+00 2.53750861e-01 6.26324713e-02 6.01912737e-01 5.90074202e-03 3.23954731e-01 3.01716685e-01 4.12908882e-01 9.67268407e-01 4.25653495e-02 5.22882538e-03 -7.73090944e-02 4.71022189e-01 -6.53444380e-02 2.36995928e-02 1.10821366e+00 8.14261064e-02 5.78111112e-01 -2.69742817e-01 -3.73054564e-01 -7.49213815e-01 -8.29918861e-01 -8.66232961e-02 1.76301587e+00 -1.03836963e-02 -2.85020381e-01 -2.58670986e-01 -7.22935438e-01 6.37755990e-02 5.95764637e-01 -7.89792955e-01 -6.37407422e-01 -5.31075478e-01 -6.84639096e-01 6.07884943e-01 5.55492163e-01 1.50709540e-01 -1.31413710e+00 -1.83998749e-01 2.59581596e-01 -1.01750411e-01 -7.40146101e-01 -4.53719735e-01 1.22686632e-01 -1.03139508e+00 -6.02377832e-01 -9.99768376e-01 -8.39175105e-01 2.21918270e-01 9.42701101e-01 1.52555704e+00 3.63472670e-01 -2.83147744e-03 3.54800820e-01 -4.44851398e-01 -4.42119032e-01 -3.62402558e-01 5.45450509e-01 6.15114830e-02 -4.56410229e-01 7.58720517e-01 -2.52514314e-02 -8.91531169e-01 1.01931408e-01 -1.42649245e+00 -4.52698261e-01 8.78543198e-01 9.21757877e-01 3.60568225e-01 -4.00611520e-01 9.65645373e-01 -6.83351099e-01 1.41924644e+00 -3.62526923e-01 -5.58864951e-01 9.39734876e-01 -1.28835297e+00 3.63899946e-01 2.03544065e-01 -5.25232255e-01 -9.98803735e-01 -4.68090475e-01 1.42901346e-01 -5.97195029e-01 3.83473933e-01 9.05220270e-01 6.30773306e-01 -2.71271206e-02 1.00965178e+00 9.26757157e-02 -1.32516384e-01 -4.97099400e-01 6.02406263e-01 8.81709456e-01 3.31296921e-01 -3.54033113e-01 5.10905921e-01 -1.31966799e-01 -4.47057903e-01 -2.82556355e-01 -9.26550865e-01 -8.79247427e-01 -2.30391622e-01 4.07386422e-02 3.18162292e-01 -1.10007465e+00 -2.83182740e-01 -1.03135012e-01 -1.13742995e+00 1.15566917e-01 -1.17023692e-01 4.47372377e-01 9.03238878e-02 3.26620191e-01 -8.33890855e-01 -7.19686270e-01 -1.12585390e+00 -9.87631500e-01 1.15491748e+00 2.70056516e-01 -1.74602494e-01 -8.60842824e-01 5.04303396e-01 4.60868090e-01 1.02224243e+00 -7.77369916e-01 1.14550364e+00 -9.86082315e-01 -2.89450973e-01 -5.75897872e-01 -4.47632372e-01 3.51347804e-01 -2.13344201e-01 -2.13865370e-01 -1.14384341e+00 -3.69107455e-01 -4.10894811e-01 -7.41185129e-01 1.38386393e+00 2.22958967e-01 7.62599826e-01 -2.77816236e-01 -3.66001070e-01 -7.69206807e-02 1.48854184e+00 2.44071707e-01 8.22131634e-01 4.99035805e-01 3.95039678e-01 4.72498447e-01 6.30618811e-01 -1.08119771e-01 2.17960775e-01 8.78402889e-01 1.43045530e-01 7.21958801e-02 -1.67643011e-01 -3.04763019e-01 3.38751048e-01 6.69229031e-01 3.14063638e-01 -4.24817324e-01 -8.68917525e-01 4.80057210e-01 -1.86316967e+00 -6.80075586e-01 4.13907170e-01 2.19429231e+00 9.60110664e-01 -2.30805092e-02 -2.76940018e-01 -2.06791952e-01 3.61275285e-01 1.12899020e-01 -4.92750883e-01 -3.75625014e-01 8.47605348e-04 2.78584421e-01 2.22424299e-01 5.16181588e-01 -1.04327488e+00 1.01168644e+00 7.08323622e+00 8.84415567e-01 -1.15985322e+00 -1.19030863e-01 4.29209858e-01 -2.17977509e-01 -4.42501277e-01 -2.57222235e-01 -1.07779765e+00 -2.67113328e-01 1.11445844e+00 -1.52739093e-01 5.29840410e-01 9.83277082e-01 -4.37430710e-01 3.98019969e-01 -1.18889749e+00 7.16074526e-01 3.75344694e-01 -1.25198662e+00 7.92051733e-01 -2.24997610e-01 5.43661892e-01 4.18821484e-01 2.70061016e-01 1.10172713e+00 4.06865209e-01 -9.94943917e-01 1.52564973e-01 5.14723361e-01 5.27316809e-01 -4.73306775e-01 8.24447036e-01 8.95262659e-02 -6.20762467e-01 -2.18194827e-01 -6.14684284e-01 1.56247690e-01 -2.34456852e-01 1.38949886e-01 -1.10844195e+00 5.96468329e-01 7.69181192e-01 3.18631142e-01 -8.54807496e-01 9.44541216e-01 9.09831896e-02 2.50783086e-01 -2.78826267e-01 -2.18482822e-01 4.20110554e-01 2.55188823e-01 2.51319975e-01 1.05471408e+00 5.64356633e-02 -3.57972115e-01 -6.29067793e-02 6.80369496e-01 -5.18672168e-01 4.21167731e-01 -7.19498098e-01 -2.38287210e-01 6.27737105e-01 1.19038045e+00 -7.29971752e-02 -2.81872600e-01 -4.02239352e-01 6.29742146e-01 5.87292373e-01 6.76974118e-01 -2.52464741e-01 -3.89652342e-01 2.74706841e-01 -5.79790175e-02 5.79085667e-04 2.62227595e-01 1.59944460e-01 -9.90233600e-01 1.52818665e-01 -9.74964738e-01 6.92210197e-01 -8.40152383e-01 -1.50095010e+00 9.03314531e-01 2.14359999e-01 -1.03652728e+00 -7.57219493e-01 -4.16494220e-01 6.59185871e-02 9.21400309e-01 -2.06182575e+00 -1.20360732e+00 3.11231390e-02 4.47638690e-01 4.93167996e-01 -4.16540414e-01 1.08401346e+00 3.91710132e-01 -2.98926562e-01 9.64075685e-01 4.50551987e-01 1.51955366e-01 9.95522380e-01 -9.45265710e-01 3.05370688e-01 5.97084343e-01 5.18473685e-01 1.19366884e+00 1.35824814e-01 -3.72189760e-01 -1.42582417e+00 -8.59936714e-01 8.46176744e-01 -3.61551106e-01 2.88877726e-01 -8.90770182e-02 -1.17847383e+00 4.06816095e-01 4.99101222e-01 -1.78985938e-01 6.62816584e-01 5.40991843e-01 -9.68276203e-01 -3.61265570e-01 -6.72169924e-01 6.39334083e-01 4.80354458e-01 -6.85910881e-01 -9.19082165e-01 2.20724016e-01 1.19042099e+00 -1.25853568e-02 -7.66780555e-01 9.08393443e-01 8.27979982e-01 -4.26499903e-01 1.40521896e+00 -7.24096119e-01 3.63593131e-01 -6.77679777e-02 -2.12443694e-01 -1.13427448e+00 -6.02378845e-01 -2.27457151e-01 -3.66587877e-01 8.68503213e-01 5.90668499e-01 -3.66740376e-01 4.93567288e-01 7.45672941e-01 -6.27375534e-03 -6.92050636e-01 -7.27821529e-01 -5.30217826e-01 3.70400578e-01 -2.41704136e-01 3.86944026e-01 8.18448842e-01 7.33756498e-02 9.80248630e-01 -2.32698798e-01 -2.24683300e-01 5.95199727e-02 -5.78820072e-02 4.67036694e-01 -1.43368292e+00 -3.06086987e-01 -7.09500849e-01 9.24960449e-02 -1.20835638e+00 2.06440851e-01 -1.01018965e+00 1.22298628e-01 -1.56566978e+00 4.05875921e-01 -4.85737056e-01 -1.10801172e+00 6.56239092e-01 -3.99058312e-01 3.64225626e-01 2.26164684e-01 5.53420067e-01 -7.87984252e-01 6.54153168e-01 9.68719900e-01 -4.76164192e-01 -2.59720355e-01 -3.92089427e-01 -1.05041623e+00 5.60996942e-02 4.85487401e-01 -4.73068625e-01 -7.42364109e-01 -8.89443278e-01 6.88085735e-01 3.82414423e-02 -1.40039632e-02 -6.40811682e-01 4.17355835e-01 2.28732467e-01 3.09963554e-01 -5.71062863e-01 3.16970229e-01 -7.77095258e-01 -4.46770936e-01 1.19126268e-01 -1.06973636e+00 3.08759391e-01 3.30832362e-01 6.86769724e-01 -6.22412801e-01 -3.65342379e-01 2.66037554e-01 -1.09093942e-01 -5.88015795e-01 1.86177626e-01 -1.17814057e-01 -2.23072022e-01 1.95510492e-01 1.82910398e-01 -6.06606960e-01 -4.09392565e-01 4.39088568e-02 3.68039310e-01 5.25614538e-04 1.19724703e+00 9.30628598e-01 -1.32749021e+00 -8.04138601e-01 1.10030808e-01 6.53571486e-01 -1.23890072e-01 9.00489315e-02 1.65997833e-01 -1.29537329e-01 1.23987293e+00 1.68861583e-01 -3.87364000e-01 -1.13101852e+00 8.69645536e-01 1.20083272e-01 -9.21298146e-01 -1.47086933e-01 7.41580129e-01 2.58464158e-01 -7.28141248e-01 6.69799924e-01 -2.18080990e-02 -6.68883860e-01 -6.75054640e-02 9.60946381e-01 -1.03807300e-01 5.06362319e-01 -1.12642929e-01 -9.49322730e-02 2.38767296e-01 -1.05134952e+00 -2.88796395e-01 1.15719521e+00 -2.62165014e-02 -4.04585935e-02 3.27210665e-01 1.24511516e+00 -4.65422988e-01 -3.02172512e-01 -8.06879163e-01 2.75345713e-01 -1.64886862e-01 6.63656592e-01 -1.27333140e+00 -8.73608172e-01 7.26288497e-01 8.95986080e-01 9.24216583e-03 1.22799587e+00 -2.72549331e-01 7.88040876e-01 1.14830279e+00 1.84580341e-01 -9.79553401e-01 1.26962572e-01 8.65281224e-01 1.06149602e+00 -1.18474162e+00 -7.04286471e-02 1.77918926e-01 -3.53131324e-01 1.02623582e+00 6.56102061e-01 -1.44209042e-01 5.56661665e-01 -2.28616506e-01 3.60589206e-01 -2.80026883e-01 -1.15690982e+00 4.21893410e-02 1.01291037e+00 1.51328102e-01 8.69781375e-01 -4.60614413e-01 -5.05033910e-01 2.70175159e-01 2.36641183e-01 3.65550488e-01 -5.74529283e-02 1.02246737e+00 -3.38856936e-01 -1.28204215e+00 -1.00329056e-01 3.44895095e-01 -7.36233830e-01 -6.69401407e-01 -6.22424901e-01 5.97791314e-01 -9.10481334e-01 7.49274969e-01 9.03557055e-03 -3.67577374e-01 2.26587310e-01 4.13896143e-01 1.98161870e-01 -5.54076135e-01 -1.00555873e+00 1.15037479e-01 -8.16094968e-03 -4.69680011e-01 -4.73677963e-01 -7.79477954e-02 -7.08031535e-01 3.44991952e-01 -6.38278425e-01 4.05908167e-01 5.88853717e-01 6.55760050e-01 7.16632545e-01 3.90248418e-01 4.94755983e-01 -4.61044908e-01 -9.82673526e-01 -1.52808464e+00 -1.12481102e-01 3.38085115e-01 1.84421644e-01 -4.06968325e-01 -8.73955041e-02 -5.68393528e-01]
[11.46475601196289, 7.6197614669799805]
11e7525e-96de-4240-a71c-192bba24af31
cross-modal-contrastive-attention-model-for
null
null
https://aclanthology.org/2022.coling-1.210
https://aclanthology.org/2022.coling-1.210.pdf
Cross-modal Contrastive Attention Model for Medical Report Generation
Medical report automatic generation has gained increasing interest recently as a way to help radiologists write reports more efficiently. However, this image-to-text task is rather challenging due to the typical data biases: 1) Normal physiological structures dominate the images, with only tiny abnormalities; 2) Normal descriptions accordingly dominate the reports. Existing methods have attempted to solve these problems, but they neglect to exploit useful information from similar historical cases. In this paper, we propose a novel Cross-modal Contrastive Attention (CMCA) model to capture both visual and semantic information from similar cases, with mainly two modules: a Visual Contrastive Attention Module for refining the unique abnormal regions compared to the retrieved case images; a Cross-modal Attention Module for matching the positive semantic information from the case reports. Extensive experiments on two widely-used benchmarks, IU X-Ray and MIMIC-CXR, demonstrate that the proposed model outperforms the state-of-the-art methods on almost all metrics. Further analyses also validate that our proposed model is able to improve the reports with more accurate abnormal findings and richer descriptions.
['Pengxu Wei', 'Ying Liu', 'Junzhong Ji', 'Xiaodan Zhang', 'Xiao Song']
null
null
null
null
coling-2022-10
['medical-report-generation']
['medical']
[ 3.01499069e-01 2.38876358e-01 -2.81417847e-01 -1.91299349e-01 -1.33892596e+00 -1.97919533e-01 5.88788509e-01 3.84270966e-01 -1.11648843e-01 8.22127759e-01 6.20867670e-01 -2.22883865e-01 -1.70903787e-01 -3.71186137e-01 -4.68901724e-01 -6.62127435e-01 5.33798188e-02 3.97291899e-01 2.35546321e-01 4.31776270e-02 6.11017227e-01 2.01868221e-01 -1.54065502e+00 7.44056821e-01 9.65759277e-01 1.00682294e+00 4.26996857e-01 6.04509473e-01 -2.91415542e-01 1.14467657e+00 -5.54550052e-01 -5.48582792e-01 -9.57095623e-02 -7.15511441e-01 -9.39716041e-01 3.43295902e-01 3.43311846e-01 -2.47149691e-01 -3.06163043e-01 1.14891446e+00 8.02495778e-01 2.70062499e-02 8.60478103e-01 -7.43497252e-01 -1.07757151e+00 3.64582539e-01 -1.07533896e+00 8.76414180e-01 5.37072062e-01 2.64784604e-01 8.61931801e-01 -8.40982676e-01 8.65421653e-01 1.10152793e+00 2.32168600e-01 5.91578186e-01 -7.70533144e-01 -5.91668904e-01 3.11720461e-01 5.31187415e-01 -1.32336450e+00 -1.08278587e-01 6.91572905e-01 -4.58390236e-01 8.91083479e-01 5.68293810e-01 7.14804292e-01 1.18923807e+00 4.29468066e-01 9.18430388e-01 1.18176723e+00 -2.79754139e-02 1.49306525e-02 2.73074538e-01 -1.13390456e-03 6.90540254e-01 2.35047176e-01 -1.46263048e-01 -3.48991275e-01 -1.08851492e-01 7.18147755e-01 2.46905997e-01 -7.59537697e-01 4.43574190e-02 -1.51544404e+00 7.46638238e-01 6.65128529e-01 6.72983110e-01 -7.32229054e-01 -1.50273740e-01 4.61303651e-01 -1.13432392e-01 5.90355992e-01 5.75532913e-01 6.28894195e-02 1.70746863e-01 -9.23645675e-01 3.09881359e-01 3.34044427e-01 7.59690285e-01 1.12139270e-01 -1.47648975e-01 -7.61156261e-01 8.28765154e-01 1.47789702e-01 4.22988117e-01 1.03520739e+00 -4.47425425e-01 7.08673418e-01 7.28544474e-01 1.37678506e-02 -1.19337249e+00 -5.21291912e-01 -7.33585298e-01 -1.13864362e+00 -1.71891838e-01 -1.05579510e-01 4.41965282e-01 -1.22802353e+00 1.28288078e+00 1.45050600e-01 6.09762147e-02 1.37778923e-01 1.25188673e+00 1.58747208e+00 6.56409323e-01 1.71931177e-01 -4.70701516e-01 1.55284810e+00 -1.13220608e+00 -1.13368642e+00 -1.61931768e-01 3.90104204e-01 -8.51708233e-01 9.53896701e-01 1.92143917e-01 -1.63868630e+00 -5.86504340e-01 -1.04330957e+00 -1.78102870e-02 -1.78311780e-01 3.93055767e-01 5.18820882e-01 -1.69767421e-02 -9.92569506e-01 2.04013839e-01 -5.30322671e-01 -2.54165024e-01 7.28333652e-01 7.14499280e-02 -3.67356509e-01 -2.31028631e-01 -1.02747321e+00 1.10959232e+00 5.15984774e-01 -1.58393364e-02 -7.31754243e-01 -9.50520098e-01 -7.54701257e-01 2.93776002e-02 4.98825550e-01 -9.10789549e-01 1.25224352e+00 -7.42906928e-01 -8.90839338e-01 1.31728947e+00 -1.02766089e-01 -3.01221877e-01 6.82991445e-01 1.00935273e-01 -4.48160142e-01 5.86252689e-01 3.90265793e-01 8.54078233e-01 7.40609109e-01 -1.52020693e+00 -5.95439672e-01 -3.70234013e-01 2.52939388e-02 2.64317095e-01 -5.27512431e-02 -2.08221257e-01 -7.09861875e-01 -1.25044155e+00 1.77053079e-01 -6.35652900e-01 -3.43541414e-01 -3.72279133e-03 -5.82781672e-01 -2.14316010e-01 4.32867199e-01 -8.37764859e-01 1.43407655e+00 -1.93381274e+00 4.24031503e-02 -1.27608981e-02 6.31757259e-01 1.55899391e-01 9.74716097e-02 -5.92267700e-02 -4.57499802e-01 1.84391618e-01 -4.73009765e-01 -2.53805071e-01 -5.19169092e-01 8.77335388e-03 -2.00666860e-01 2.96257555e-01 5.86456835e-01 1.08277154e+00 -9.83088374e-01 -9.71798003e-01 1.02994584e-01 2.42329136e-01 -3.15163910e-01 4.73117709e-01 1.03305988e-01 7.22828150e-01 -5.27669966e-01 8.32411885e-01 6.57559454e-01 -8.42067063e-01 -5.58477268e-02 -3.74242425e-01 6.62517175e-02 1.98945686e-01 -8.63398969e-01 1.63888133e+00 -4.23795134e-01 2.17601374e-01 -2.46263489e-01 -9.55999434e-01 6.23077750e-01 4.79623526e-01 5.32442868e-01 -1.02498591e+00 2.05165684e-01 1.81623474e-01 1.38172314e-01 -1.01266158e+00 3.23169351e-01 -3.39690953e-01 2.02459902e-01 5.62015414e-01 -1.49150901e-02 -1.49530530e-01 3.25918078e-01 4.80222523e-01 1.05848396e+00 -4.21398044e-01 7.72786975e-01 5.76127134e-02 8.23109269e-01 2.48048156e-02 4.05169278e-01 9.00006592e-01 -2.36030757e-01 1.20217168e+00 4.38602000e-01 -4.21182394e-01 -8.68319988e-01 -1.08400488e+00 -1.96281224e-01 4.33477432e-01 3.15366477e-01 -2.25754797e-01 -4.70420778e-01 -1.00691879e+00 -2.75102854e-01 6.68968260e-01 -1.08460701e+00 -2.55333930e-01 -4.13673669e-01 -8.90766621e-01 -6.24053646e-03 8.65330875e-01 4.81712192e-01 -1.31438637e+00 -5.88091731e-01 2.02574685e-01 -4.33686763e-01 -1.01625288e+00 -5.39386630e-01 -1.37298301e-01 -7.97799647e-01 -1.04518700e+00 -1.43887663e+00 -5.95911562e-01 9.79616344e-01 3.82641226e-01 1.27914238e+00 4.06194240e-01 -7.32862294e-01 5.26772797e-01 -4.70774323e-01 -5.57341456e-01 -5.61956584e-01 -9.25853625e-02 -4.85821992e-01 1.27574146e-01 -5.95171601e-02 -1.88052848e-01 -8.41113329e-01 -1.24201871e-01 -1.11655211e+00 5.21545529e-01 1.16343713e+00 1.02316868e+00 9.03900683e-01 -3.94063860e-01 5.88615656e-01 -1.09932899e+00 7.07273722e-01 -7.06190050e-01 -1.14523591e-02 3.91962737e-01 -6.00696266e-01 7.37589449e-02 3.46626610e-01 -2.29688168e-01 -1.27254105e+00 -2.03602135e-01 -1.84676439e-01 -4.94834691e-01 -1.27159059e-01 3.32974285e-01 3.37962240e-01 1.90082684e-01 3.75771016e-01 6.38711214e-01 -1.64912671e-01 -1.57087028e-01 1.26418576e-01 4.46522772e-01 8.01996410e-01 -6.35097399e-02 5.21166563e-01 6.53673589e-01 -5.52235246e-02 -2.53540665e-01 -1.25792766e+00 -7.59187341e-01 -4.16785985e-01 -3.21677387e-01 1.13291836e+00 -7.64068127e-01 -4.70553756e-01 -1.62462145e-01 -1.37415528e+00 5.61591148e-01 -2.45889157e-01 5.73243380e-01 -5.55848897e-01 3.83255035e-01 -5.15599251e-01 -6.63146853e-01 -6.92540109e-01 -1.60103166e+00 1.17027009e+00 3.29970956e-01 -1.40726835e-01 -7.68726647e-01 1.45508824e-02 4.61439371e-01 3.35214972e-01 2.41364926e-01 1.13878787e+00 -5.90836227e-01 -5.67444980e-01 2.12778300e-02 -6.70824468e-01 9.17611178e-03 1.27480283e-01 -2.77181387e-01 -1.08768654e+00 -1.58276498e-01 1.06605887e-01 -2.12546930e-01 1.21337879e+00 5.19721329e-01 1.69141996e+00 -3.11220527e-01 -4.98952717e-01 5.19731283e-01 1.16787159e+00 4.02546465e-01 6.06175542e-01 4.77343202e-02 5.55779219e-01 4.20009017e-01 7.85865128e-01 4.29117054e-01 3.73885989e-01 3.69903922e-01 7.40217626e-01 -6.38923645e-01 -3.06816518e-01 6.69489577e-02 -2.34266818e-01 1.00389600e+00 -3.50487918e-01 -1.92505211e-01 -9.35992897e-01 8.19432557e-01 -1.78479052e+00 -8.16481709e-01 -9.33804438e-02 1.74965537e+00 7.97548771e-01 6.15739673e-02 -2.01117143e-01 -3.50877941e-02 6.10521793e-01 2.95136571e-01 -4.71831709e-01 -1.73185289e-01 -5.64602576e-02 2.22257704e-01 1.30785614e-01 -1.05597928e-01 -1.20309520e+00 2.60881454e-01 6.21354198e+00 8.54686081e-01 -1.15071249e+00 3.95886123e-01 1.08217239e+00 -8.29816237e-02 -4.54959542e-01 -6.38093710e-01 -4.57976729e-01 4.79007930e-01 4.32949185e-01 -1.83710888e-01 -3.50378066e-01 8.86957765e-01 5.30472957e-02 -5.05385362e-02 -1.09147823e+00 1.29179645e+00 6.35476053e-01 -1.94823253e+00 6.33796573e-01 -7.72229359e-02 9.88456607e-01 -3.43150020e-01 2.80688912e-01 2.08187908e-01 -3.25806856e-01 -9.68669534e-01 5.87684691e-01 9.85650718e-01 9.53877389e-01 -6.26491845e-01 1.11413598e+00 -1.37838259e-01 -1.11111748e+00 1.49284611e-02 -2.30629981e-01 4.59670067e-01 8.06136988e-04 3.03213179e-01 -1.14044535e+00 8.25596452e-01 8.25124085e-01 8.98295820e-01 -9.10284758e-01 1.23448241e+00 -3.30001079e-02 4.55853283e-01 3.09424788e-01 -7.07465857e-02 4.72056597e-01 2.62400448e-01 6.59775555e-01 1.40508413e+00 3.90320808e-01 4.74314690e-01 1.20840482e-01 1.23351681e+00 -6.78257197e-02 4.40933168e-01 -6.23354018e-01 1.21701926e-01 -1.25333831e-01 1.29065752e+00 -9.21254873e-01 -8.11925352e-01 -6.63565755e-01 8.33353877e-01 1.48559704e-01 1.62587732e-01 -9.18902636e-01 -1.07862458e-01 7.99332634e-02 4.02620018e-01 2.28578791e-01 4.61436272e-01 -4.59962815e-01 -1.11567545e+00 1.54156327e-01 -8.59744012e-01 7.00483561e-01 -1.18169856e+00 -1.40511215e+00 8.41026783e-01 -1.48977295e-01 -1.64551091e+00 -2.28565246e-01 -3.79970282e-01 -6.43594384e-01 6.45048440e-01 -1.81892395e+00 -1.09175611e+00 -5.96402764e-01 4.65334177e-01 8.37799430e-01 -2.67352313e-01 6.67620182e-01 3.84206742e-01 -2.67420888e-01 5.83609402e-01 -3.09827894e-01 -4.22482938e-02 7.23384798e-01 -1.36741698e+00 -1.27787692e-02 6.94270968e-01 -9.78810191e-02 6.43127799e-01 2.38635629e-01 -5.85775793e-01 -9.50184226e-01 -1.09098113e+00 5.32109141e-01 -1.89012766e-01 3.28531802e-01 2.31961399e-01 -1.26616931e+00 3.74730766e-01 4.31591779e-01 8.58525932e-02 5.95539153e-01 -5.88545918e-01 -2.19504852e-02 2.64363512e-02 -1.04914021e+00 5.79330206e-01 7.94579685e-01 -1.46786004e-01 -7.50776947e-01 5.35396457e-01 7.86145568e-01 -6.08262360e-01 -8.36190641e-01 4.91524398e-01 2.50911564e-01 -8.35537493e-01 9.76042330e-01 -5.88918209e-01 9.91987824e-01 -3.72155160e-01 1.63826168e-01 -1.18197680e+00 -4.67066258e-01 -2.49750271e-01 -4.67050113e-02 9.72937346e-01 4.51893181e-01 -3.67826909e-01 2.26681322e-01 1.01607822e-01 -4.02248979e-01 -1.18054461e+00 -8.58409822e-01 -7.41777495e-02 -3.27050060e-01 -4.40038979e-01 3.90184373e-01 1.06610763e+00 -2.94786125e-01 2.38087699e-01 -1.34303242e-01 6.71965778e-02 3.15563142e-01 2.92770416e-01 1.83724955e-01 -9.00078475e-01 -1.77276671e-01 -6.74882591e-01 -5.71479321e-01 -6.29302025e-01 -1.57955989e-01 -1.11536717e+00 -8.21426585e-02 -1.79200840e+00 1.01845396e+00 -8.63635615e-02 -4.78862107e-01 3.31817508e-01 -7.93050945e-01 4.87987399e-01 1.98138356e-01 1.51027396e-01 -8.35373819e-01 4.00910974e-01 1.82001543e+00 -5.15149176e-01 1.69630066e-01 -5.86303249e-02 -9.94482279e-01 8.41032982e-01 3.21917504e-01 -4.67866391e-01 -1.87060907e-01 -1.72235370e-01 1.71426058e-01 2.66645908e-01 4.53284085e-01 -9.48083162e-01 -2.48959064e-02 -3.27810720e-02 6.56928897e-01 -1.11493266e+00 4.45346758e-02 -5.14319181e-01 -3.15874845e-01 5.38941145e-01 -5.31649292e-01 2.58799732e-01 2.41379127e-01 7.50232816e-01 -5.27960837e-01 -1.53608844e-01 9.24387991e-01 -6.51272416e-01 -5.76245427e-01 3.38097900e-01 -2.77444422e-01 2.70064443e-01 1.12251782e+00 -1.45808356e-02 -4.02484149e-01 -4.67198282e-01 -8.60912919e-01 1.97045103e-01 -8.62978473e-02 5.77874780e-01 1.11994219e+00 -1.34991288e+00 -1.10184431e+00 6.35806024e-02 5.52305222e-01 6.76543042e-02 7.33646095e-01 1.25747335e+00 -5.27498782e-01 6.54397905e-01 -4.19500023e-02 -9.07180786e-01 -1.24868357e+00 8.47802341e-01 2.60127991e-01 -8.09938848e-01 -8.09869409e-01 7.83786297e-01 9.06920016e-01 6.89435229e-02 1.80607155e-01 -6.77254617e-01 -5.26621580e-01 3.32797132e-02 1.04217923e+00 5.31267375e-02 3.27799946e-01 -6.47851944e-01 -3.71298999e-01 7.20941126e-01 -6.73945487e-01 1.65121421e-01 1.27900314e+00 -9.66260135e-02 1.49061363e-02 3.22761685e-01 9.73601878e-01 -3.12888473e-01 -6.00472748e-01 -2.27883875e-01 -3.03686470e-01 -4.23002690e-01 9.93988737e-02 -1.02394092e+00 -1.34872854e+00 9.34116066e-01 7.98702538e-01 1.03204213e-01 1.22118580e+00 3.11142176e-01 7.69692719e-01 -1.82782020e-02 -9.15276185e-02 -7.56524205e-01 5.18257797e-01 -8.76614004e-02 1.51682079e+00 -1.62292624e+00 2.63718098e-01 -3.72230470e-01 -1.12124336e+00 9.93044913e-01 7.41411328e-01 2.58450396e-02 3.52821827e-01 5.98087907e-02 5.93179911e-02 -4.05468911e-01 -9.06677783e-01 -2.54798770e-01 8.28194857e-01 5.88447928e-01 5.41856349e-01 -8.40066746e-02 -4.28371489e-01 8.66668761e-01 2.74383157e-01 -2.83303261e-01 5.91571748e-01 7.17095077e-01 -1.02663040e-01 -3.83585423e-01 -6.96066618e-01 8.39845002e-01 -9.01657939e-01 -1.83296278e-01 -2.80045390e-01 9.36068773e-01 1.46887437e-01 6.90675676e-01 1.03136696e-01 -1.10620238e-01 3.57791871e-01 -1.80983588e-01 4.03855979e-01 -6.69673562e-01 -7.07367480e-01 2.47802243e-01 -2.93336332e-01 -6.78580463e-01 -5.91225564e-01 -4.52866524e-01 -1.36791599e+00 2.31665000e-01 -2.24025056e-01 -5.08963726e-02 3.72333407e-01 8.33133936e-01 3.45667541e-01 1.32786965e+00 4.26696897e-01 -6.72354758e-01 -2.86069483e-01 -1.08818066e+00 -4.13721234e-01 7.87588954e-01 5.15709460e-01 -9.57499504e-01 -8.17473084e-02 2.84826964e-01]
[15.030667304992676, -1.4361791610717773]
ec1e83a2-89bf-4541-b4f5-2117e7f7aca6
cadge-context-aware-dialogue-generation
2305.06294
null
https://arxiv.org/abs/2305.06294v2
https://arxiv.org/pdf/2305.06294v2.pdf
CADGE: Context-Aware Dialogue Generation Enhanced with Graph-Structured Knowledge Aggregation
Commonsense knowledge is crucial to many natural language processing tasks. Existing works usually incorporate graph knowledge with conventional graph neural networks (GNNs), leading to the text and graph knowledge encoding processes being separated in a serial pipeline. We argue that these separate representation learning stages may be suboptimal for neural networks to learn the overall context contained in both types of input knowledge. In this paper, we propose a novel context-aware graph-attention model (Context-aware GAT), which can effectively incorporate global features of relevant knowledge graphs based on a context-enhanced knowledge aggregation process. Specifically, our framework leverages a novel representation learning approach to process heterogeneous features - combining flattened graph knowledge with text. To the best of our knowledge, this is the first attempt at hierarchically applying graph knowledge aggregation on a connected subgraph in addition to contextual information to support commonsense dialogue generation. This framework shows superior performance compared to conventional GNN-based language frameworks. Both automatic and human evaluation demonstrates that our proposed model has significant performance uplifts over state-of-the-art baselines.
['Chenghua Lin', 'Tyler Loakman', 'Hongbo Zhang', 'Stefan Goetze', 'Chen Tang']
2023-05-10
null
null
null
null
['dialogue-generation', 'dialogue-generation']
['natural-language-processing', 'speech']
[ 6.58812225e-01 7.66323924e-01 -1.78350478e-01 -2.28276521e-01 -5.17810345e-01 -5.39992452e-01 9.32449758e-01 7.48528302e-01 -2.79562443e-01 5.42590976e-01 7.47815728e-01 -3.87991101e-01 4.35967594e-02 -1.09744859e+00 -5.01356781e-01 -1.64182663e-01 1.65200770e-01 5.38808882e-01 1.23995498e-01 -6.52505934e-01 4.63752776e-01 8.26271176e-02 -1.18423522e+00 5.11587560e-01 1.09444368e+00 7.51515090e-01 3.13195288e-02 6.45828366e-01 -7.84083188e-01 1.58260298e+00 -6.26911402e-01 -7.65405655e-01 -1.14048861e-01 -6.14844024e-01 -1.35875618e+00 -1.58946097e-01 6.15194738e-01 -1.93076685e-01 -4.04766977e-01 1.07344460e+00 2.91091323e-01 5.12613595e-01 5.66212177e-01 -9.12569821e-01 -1.46530998e+00 1.26685047e+00 -2.93212324e-01 2.48031124e-01 6.32855654e-01 1.59229949e-01 1.63102806e+00 -3.95319343e-01 7.48919964e-01 1.42449844e+00 4.96438533e-01 4.66125846e-01 -1.05320084e+00 -2.45961681e-01 5.87668240e-01 4.28638101e-01 -8.97183895e-01 -8.42870399e-02 1.21808648e+00 -2.36999050e-01 1.74990606e+00 3.26691121e-02 9.86480355e-01 1.18574893e+00 -9.92264878e-03 9.59588230e-01 8.82024229e-01 -6.52133942e-01 9.63939354e-02 -3.34683627e-01 5.79982877e-01 1.09442234e+00 5.02059340e-01 -3.56575847e-01 -6.86565757e-01 -1.26532465e-01 6.02024317e-01 -1.27614588e-01 -1.25973254e-01 -2.42494449e-01 -1.14455235e+00 1.01813757e+00 8.04999769e-01 3.10590059e-01 -4.93982524e-01 6.55769646e-01 7.03150630e-01 2.67114639e-01 5.75129092e-01 7.50392854e-01 -2.03362674e-01 -3.86956125e-03 -6.87875688e-01 8.48092884e-02 1.05811810e+00 1.06263530e+00 7.47831702e-01 1.14363424e-01 -6.33754849e-01 5.61471283e-01 3.70853931e-01 7.59716183e-02 5.05794942e-01 -7.18544126e-01 6.77438617e-01 1.16017735e+00 -5.18405497e-01 -1.20902860e+00 -2.04889014e-01 -4.30503428e-01 -5.94090343e-01 -3.48605484e-01 2.35954374e-01 2.94379853e-02 -9.92466450e-01 1.69133842e+00 8.57551172e-02 2.69556195e-01 3.15471917e-01 6.08720541e-01 1.27558076e+00 3.90659899e-01 4.07868177e-01 1.25036135e-01 1.35801697e+00 -1.08100057e+00 -8.78693402e-01 -4.90605235e-01 5.90408206e-01 -3.47504497e-01 1.38678229e+00 1.22231781e-01 -9.82001126e-01 -1.38576165e-01 -1.02957177e+00 -6.01758063e-01 -8.06245506e-01 -3.68103832e-01 1.13013220e+00 4.76422757e-01 -1.35723245e+00 4.82646614e-01 -5.55655003e-01 -7.76139796e-01 5.51663101e-01 -2.67948378e-02 -3.94373499e-02 -2.18953833e-01 -1.54108858e+00 1.03875029e+00 7.50907481e-01 2.28670966e-02 -6.09110177e-01 -4.56760585e-01 -1.33136904e+00 1.01616301e-01 9.06320512e-01 -1.27098107e+00 1.13695765e+00 -7.05039978e-01 -1.41577458e+00 8.58508706e-01 -2.25304469e-01 -7.12269306e-01 1.31078064e-01 -2.93170035e-01 -8.96292254e-02 2.78898597e-01 7.96872452e-02 3.80999893e-01 6.14601672e-01 -1.14237463e+00 -8.07067566e-03 -3.23221087e-01 6.50163114e-01 4.68954712e-01 -2.51878709e-01 -6.32436946e-02 -1.99380413e-01 -7.56992102e-01 -1.28735349e-01 -5.20425737e-01 -1.85609996e-01 -4.80500549e-01 -6.56629622e-01 -4.34987992e-01 5.87308049e-01 -6.94636762e-01 1.38660812e+00 -1.42129970e+00 2.26007849e-01 1.18201099e-01 5.57859361e-01 1.77107170e-01 -1.59476057e-01 6.90001369e-01 2.67519116e-01 3.70099872e-01 -3.31609219e-01 -2.27050975e-01 2.49301657e-01 3.61248910e-01 -3.54026169e-01 -1.67725354e-01 5.67369998e-01 1.61967158e+00 -1.55810404e+00 -4.17811364e-01 1.55422976e-02 4.44480896e-01 -6.09310031e-01 4.89723161e-02 -6.35174155e-01 1.01303086e-02 -4.45176363e-01 6.17179930e-01 9.46553200e-02 -5.79141676e-01 3.44200164e-01 -2.60302335e-01 5.32659590e-01 6.59498274e-01 -7.77330339e-01 1.99720454e+00 -4.15033311e-01 5.12518942e-01 -2.89328992e-01 -1.05656159e+00 9.15356219e-01 1.43721744e-01 -9.20912251e-02 -3.90954554e-01 9.16445404e-02 -2.12728500e-01 -8.46041217e-02 -3.52920383e-01 9.88157928e-01 -2.27189541e-01 -2.09277064e-01 5.79879820e-01 4.82817203e-01 -4.11520422e-01 3.35231811e-01 8.86307299e-01 1.37349021e+00 2.29310453e-01 6.81979477e-01 7.58484798e-03 4.16434109e-01 -4.72233035e-02 1.14966907e-01 9.83017683e-01 -2.75991857e-01 1.64306387e-01 7.86492825e-01 -7.02967793e-02 -6.10724151e-01 -8.44322443e-01 7.55918205e-01 1.35554612e+00 -5.94351627e-02 -8.78579736e-01 -5.62350869e-01 -8.85455251e-01 -2.38549802e-02 1.14846408e+00 -7.87861109e-01 -4.43374366e-01 -4.90995675e-01 -4.88754719e-01 8.22140574e-01 7.59896994e-01 5.36575079e-01 -1.38351190e+00 -3.00083578e-01 3.46961349e-01 -8.15132782e-02 -1.23696339e+00 -3.78005266e-01 1.66546181e-01 -7.66703904e-01 -1.15120327e+00 -1.44495666e-01 -6.70549870e-01 4.11418200e-01 2.28842840e-01 1.56822741e+00 2.66235590e-01 -1.31292671e-01 8.85147572e-01 -6.62071228e-01 -3.12390447e-01 -4.69413966e-01 1.32000953e-01 -6.36570036e-01 -2.32614234e-01 5.41913688e-01 -6.94682777e-01 -4.02360588e-01 -6.47611201e-01 -8.85417461e-01 2.64410198e-01 5.43455184e-01 7.97432184e-01 3.58715266e-01 -1.87599450e-01 6.67074263e-01 -1.26711869e+00 1.35889113e+00 -6.81775808e-01 1.11442106e-02 5.73703289e-01 -6.95812464e-01 3.46065670e-01 6.07584000e-01 5.31850345e-02 -1.17000449e+00 -2.26843610e-01 3.05544257e-01 -1.97172478e-01 -8.84462819e-02 9.78624105e-01 -9.09124091e-02 1.03646621e-01 7.51529932e-01 2.83722013e-01 -2.23163784e-01 -1.27443701e-01 1.22410786e+00 2.58890033e-01 5.06580412e-01 -9.66270208e-01 4.93086785e-01 2.04793617e-01 6.06419519e-02 -6.48778558e-01 -1.10167277e+00 -5.03596365e-01 -7.64320612e-01 -7.07949325e-02 8.85050178e-01 -7.83400238e-01 -4.30521727e-01 1.76962465e-01 -1.35814893e+00 -3.66659433e-01 -5.10942876e-01 -3.70020680e-02 -5.85257769e-01 7.29261041e-01 -7.89480984e-01 -7.50581264e-01 -8.05323541e-01 -7.05431521e-01 1.10424852e+00 1.44626975e-01 -2.15548620e-01 -1.49897623e+00 -2.09021047e-02 6.66380346e-01 5.43651164e-01 5.59455693e-01 9.90393758e-01 -1.13533628e+00 -5.13157785e-01 -2.17663255e-02 -5.63975036e-01 4.07910766e-03 2.27748945e-01 -2.06951365e-01 -9.48286533e-01 4.29964848e-02 -4.29884911e-01 -7.65104771e-01 1.24720562e+00 8.47129058e-03 8.67809176e-01 -5.64020097e-01 -4.50190045e-02 1.97061285e-01 1.31145692e+00 -3.01336348e-01 4.89666998e-01 2.93210506e-01 1.23877323e+00 3.74903858e-01 1.36282966e-01 2.76955545e-01 1.00281990e+00 2.32593521e-01 4.43344057e-01 8.41322839e-02 -5.50949454e-01 -5.10180056e-01 4.41296637e-01 8.41659009e-01 -2.36560345e-01 -2.95394301e-01 -1.11547494e+00 7.59577274e-01 -1.97947145e+00 -1.17241263e+00 -1.70494232e-03 1.56924582e+00 1.11375904e+00 2.12902576e-02 -6.07207790e-02 -2.24692270e-01 7.94939339e-01 5.27942896e-01 -5.68237007e-01 -6.88535750e-01 -6.85037300e-02 4.53056693e-01 1.75990745e-01 6.72934651e-01 -9.73411977e-01 1.44732392e+00 5.90212345e+00 6.34125054e-01 -6.28455281e-01 2.28199828e-02 4.25660424e-02 1.29709601e-01 -7.13461459e-01 2.82489955e-01 -4.92498547e-01 -1.21991090e-01 9.56159115e-01 -5.38091600e-01 6.15995169e-01 7.52412558e-01 -1.66891575e-01 2.57691324e-01 -1.05361342e+00 7.34455407e-01 5.52505076e-01 -1.38712025e+00 7.54181325e-01 -1.73722014e-01 5.99419773e-01 4.81610782e-02 -3.52935821e-01 7.68964291e-01 1.00848711e+00 -1.05219603e+00 4.64316815e-01 6.36662364e-01 3.51804316e-01 -5.81821382e-01 7.51147211e-01 -1.03204772e-02 -1.40317976e+00 1.49949491e-01 -2.29956031e-01 -2.66438812e-01 1.79127157e-01 5.34817934e-01 -1.20064461e+00 9.98454750e-01 1.32167526e-02 1.01322603e+00 -9.14236963e-01 5.09162426e-01 -8.66706431e-01 7.63613522e-01 4.91189286e-02 -2.71270365e-01 3.88285398e-01 1.32022977e-01 6.20187521e-01 1.62392104e+00 -1.06601492e-01 1.05675712e-01 5.66266596e-01 1.27362728e+00 -5.22647858e-01 3.59997779e-01 -8.96049082e-01 -6.75120413e-01 3.55830640e-01 1.27721465e+00 -6.39676690e-01 -7.56861806e-01 -4.96471614e-01 1.02940702e+00 9.72248495e-01 3.45089227e-01 -4.89781260e-01 -6.16669476e-01 1.87312409e-01 -1.99152276e-01 2.64464110e-01 -2.81381398e-01 -3.17337245e-01 -1.44879460e+00 -2.00027257e-01 -5.54095447e-01 8.61932755e-01 -7.81968415e-01 -1.73883867e+00 5.78559637e-01 6.37006387e-02 -4.26441997e-01 -4.40646052e-01 -5.82327306e-01 -7.83542097e-01 7.38859117e-01 -1.77949572e+00 -1.81687474e+00 -3.54861796e-01 8.29710543e-01 4.30016160e-01 -1.05670385e-01 9.79626060e-01 -4.83324558e-01 -2.67086655e-01 5.09837210e-01 -5.15267134e-01 3.39286536e-01 3.97002608e-01 -1.79425025e+00 6.68883026e-01 1.02782369e+00 3.78454536e-01 9.87887263e-01 3.71403575e-01 -1.13294733e+00 -1.42564452e+00 -1.21144342e+00 8.57151151e-01 -5.59371173e-01 1.16435075e+00 -1.97629929e-01 -1.05881798e+00 9.56774652e-01 7.93601751e-01 -9.47308689e-02 1.04892683e+00 4.93308693e-01 -7.80703068e-01 4.66098458e-01 -9.76778150e-01 6.51133478e-01 1.33116257e+00 -7.99720645e-01 -1.33116364e+00 3.16647202e-01 1.21548378e+00 -2.30351672e-01 -9.24536228e-01 6.90634325e-02 4.21617702e-02 -5.63460410e-01 7.18584955e-01 -8.99652123e-01 6.05455637e-01 -2.28467092e-01 -2.65014261e-01 -1.39856064e+00 -3.50053400e-01 -8.05919647e-01 -7.52941012e-01 1.37552571e+00 3.33648115e-01 -6.17117643e-01 4.62976336e-01 6.06495261e-01 -2.69637704e-01 -6.38982654e-01 -5.34535050e-01 -5.16365469e-01 1.05919838e-02 -4.46776718e-01 5.27447939e-01 1.35102510e+00 5.34141481e-01 9.31810319e-01 -6.97587654e-02 -1.33664355e-01 6.34706020e-01 1.62564963e-01 4.67485368e-01 -1.23286867e+00 -3.69941205e-01 -6.41415894e-01 -4.69052911e-01 -6.02672338e-01 5.94403028e-01 -1.64258945e+00 -2.71859825e-01 -2.37758517e+00 3.83210689e-01 1.22154936e-01 -5.40724277e-01 9.35583651e-01 -6.89946771e-01 -1.33518457e-01 2.71064490e-01 -2.25986391e-01 -9.83821869e-01 6.52180254e-01 1.28749359e+00 -4.20118898e-01 -2.10986391e-01 -8.09113979e-01 -1.31447601e+00 5.16155958e-01 9.49828804e-01 -6.77835569e-02 -9.10625100e-01 -4.17471200e-01 4.26253259e-01 -4.80438918e-01 5.51883638e-01 -6.58012748e-01 4.36201364e-01 -2.67151505e-01 6.78777024e-02 -2.06268340e-01 8.27780068e-02 -3.20069194e-01 -4.01371628e-01 1.09023049e-01 -6.72564745e-01 -2.03355208e-01 3.11862439e-01 9.73429978e-01 -1.78091422e-01 5.45210093e-02 1.73336729e-01 -6.61523461e-01 -9.17484701e-01 -2.72372086e-03 -2.14319929e-01 4.82096434e-01 6.04069054e-01 -1.58647612e-01 -7.96267033e-01 -5.72423220e-01 -6.45099938e-01 6.68172017e-02 2.97736883e-01 4.22781914e-01 7.60500789e-01 -1.21763408e+00 -6.69729471e-01 -2.81003267e-01 3.86566520e-01 3.59997191e-02 -2.66259555e-02 4.76196498e-01 -3.00162852e-01 3.55552524e-01 5.23048490e-02 -1.19859889e-01 -9.34285641e-01 7.06722677e-01 1.08971402e-01 -8.36550593e-01 -7.22765744e-01 8.39131236e-01 -1.31668419e-01 -5.16619146e-01 -4.37592305e-02 -6.31912231e-01 -4.63189721e-01 8.40146393e-02 3.49929929e-01 1.22092567e-01 4.63769250e-02 -5.47640741e-01 -2.17372298e-01 2.71983743e-01 -3.73878866e-01 5.16155921e-02 1.27639532e+00 -6.76699653e-02 -4.78435516e-01 4.66793060e-01 7.17763662e-01 6.95674960e-03 -5.98864496e-01 -5.49152732e-01 2.62506634e-01 -1.69435635e-01 1.78150967e-01 -1.21326447e+00 -8.57233167e-01 8.13808858e-01 -2.37488464e-01 2.32075691e-01 9.35948193e-01 1.71695739e-01 7.96756744e-01 8.02730680e-01 3.04120004e-01 -9.73205566e-01 3.53996724e-01 9.47188497e-01 1.09859610e+00 -1.08464706e+00 1.36051819e-01 -4.51168001e-01 -8.88772905e-01 1.22633958e+00 7.75119901e-01 -1.76462919e-01 1.57687187e-01 -1.56591550e-01 -2.05749616e-01 -6.88154161e-01 -9.81813371e-01 -6.72957599e-01 5.74574947e-01 7.22544730e-01 8.35586429e-01 1.24558695e-01 -1.49060279e-01 7.84317136e-01 -2.77205348e-01 -7.20574185e-02 5.73443115e-01 1.11076808e+00 -5.51024854e-01 -9.82464969e-01 8.26485083e-02 6.17067695e-01 -2.06503749e-01 -7.41369605e-01 -1.03167069e+00 5.81047714e-01 -2.87102848e-01 1.06388164e+00 -3.60217303e-01 -2.54349113e-01 3.74217540e-01 4.84500527e-01 6.29567206e-01 -1.05742323e+00 -9.04813886e-01 -4.54534024e-01 4.85425830e-01 -6.42421186e-01 -5.29757082e-01 -2.25414410e-01 -1.52880228e+00 -1.61877215e-01 -1.97136089e-01 -5.87415621e-02 2.14856356e-01 9.90761101e-01 5.74563324e-01 9.73882735e-01 -2.42248401e-01 -6.36665642e-01 -3.32962453e-01 -1.04788315e+00 -4.25899446e-01 5.95447123e-01 7.57512525e-02 -5.66579401e-01 -2.27803051e-01 -5.40337944e-03]
[10.422201156616211, 8.08612060546875]
98c0008b-7bb1-4818-b2ca-ecacf65e385b
automatic-generation-of-personalized-comment
1907.10371
null
https://arxiv.org/abs/1907.10371v1
https://arxiv.org/pdf/1907.10371v1.pdf
Automatic Generation of Personalized Comment Based on User Profile
Comments on social media are very diverse, in terms of content, style and vocabulary, which make generating comments much more challenging than other existing natural language generation~(NLG) tasks. Besides, since different user has different expression habits, it is necessary to take the user's profile into consideration when generating comments. In this paper, we introduce the task of automatic generation of personalized comment~(AGPC) for social media. Based on tens of thousands of users' real comments and corresponding user profiles on weibo, we propose Personalized Comment Generation Network~(PCGN) for AGPC. The model utilizes user feature embedding with a gated memory and attends to user description to model personality of users. In addition, external user representation is taken into consideration during the decoding to enhance the comments generation. Experimental results show that our model can generate natural, human-like and personalized comments.
['Lei LI', 'Wenhuan Zeng', 'Pengcheng Yang', 'Abulikemu Abuduweili']
2019-07-24
automatic-generation-of-personalized-comment-1
https://aclanthology.org/P19-2032
https://aclanthology.org/P19-2032.pdf
acl-2019-7
['comment-generation']
['natural-language-processing']
[ 3.30480514e-03 2.77605623e-01 2.56940365e-01 -5.72139144e-01 -2.45112285e-01 -3.12456131e-01 6.27592146e-01 1.38200685e-01 -1.46175325e-01 7.79781759e-01 8.18167686e-01 -9.38447267e-02 6.57236636e-01 -9.12251413e-01 -2.41450131e-01 -3.86926740e-01 2.23710358e-01 4.26603913e-01 -1.41429320e-01 -7.79673636e-01 3.22881460e-01 -1.72729254e-01 -1.18264890e+00 4.19367224e-01 1.12130582e+00 6.38650835e-01 6.95574820e-01 9.29598391e-01 -4.69509661e-01 4.02069896e-01 -7.13972032e-01 -8.27124834e-01 -3.19871455e-01 -6.12656832e-01 -6.59061849e-01 2.43599400e-01 -8.16945657e-02 -3.52878958e-01 -1.91348493e-01 1.01723969e+00 9.01027739e-01 3.90251279e-01 1.05770648e+00 -1.04253721e+00 -1.33181775e+00 1.35430372e+00 -1.73553362e-01 9.18325596e-03 5.93512774e-01 -1.84825379e-02 1.29166675e+00 -9.94894862e-01 5.08953691e-01 1.29685569e+00 2.78523773e-01 1.09186542e+00 -9.26898479e-01 -5.58589518e-01 3.44502985e-01 -1.87441528e-01 -1.12433934e+00 -2.99098760e-01 8.70996892e-01 -4.87984508e-01 4.55189794e-01 2.33194828e-01 6.16064548e-01 1.48566186e+00 1.68754414e-01 1.00052595e+00 5.25183201e-01 -2.69835234e-01 4.39136736e-02 6.28562093e-01 -2.21713353e-02 4.57321167e-01 -7.96905607e-02 -6.03821039e-01 -3.11886013e-01 -2.89457470e-01 5.79568267e-01 -2.92761587e-02 -2.49927208e-01 3.95077318e-01 -1.12956488e+00 1.12837565e+00 3.10497224e-01 8.75743479e-02 -2.62339681e-01 -8.95080790e-02 2.59393096e-01 7.65188709e-02 7.91306674e-01 5.08564115e-01 -1.44014195e-01 2.40444895e-02 -3.66016954e-01 5.01182497e-01 8.39230359e-01 1.43440938e+00 8.79276812e-01 1.57312378e-01 -4.74508971e-01 1.11305571e+00 4.74026501e-01 7.24654198e-01 1.10814333e+00 -1.44680768e-01 5.44095874e-01 5.25472820e-01 4.02907997e-01 -1.49409795e+00 -2.49383360e-01 -4.26285475e-01 -1.15552747e+00 -7.36197889e-01 -1.78199902e-01 -9.32230949e-01 -3.03723276e-01 1.65288675e+00 -1.59445167e-01 9.20036882e-02 1.39528960e-01 6.42334700e-01 1.13144851e+00 1.02601957e+00 1.43229187e-01 -2.74369717e-01 9.89157975e-01 -8.73445928e-01 -7.33640432e-01 -2.80642778e-01 4.41305667e-01 -8.32791626e-01 1.29296231e+00 -7.69809783e-02 -1.00115418e+00 -7.04275787e-01 -4.70160872e-01 1.15213990e-01 -1.89006358e-01 3.04692388e-01 7.18136609e-01 4.99383241e-01 -9.69903946e-01 3.61426234e-01 -1.14759423e-01 -3.30963999e-01 1.05905540e-01 3.11599165e-01 2.85963211e-02 3.36820781e-01 -1.65245795e+00 1.49100497e-01 3.11585605e-01 1.63400531e-01 -3.71864527e-01 -2.53638417e-01 -8.64867926e-01 -1.38200726e-02 -2.04919234e-01 -8.38323474e-01 1.42001450e+00 -1.03960192e+00 -1.68279195e+00 2.28925630e-01 -4.42671508e-01 -2.14671995e-02 4.27711785e-01 7.09579373e-03 -7.94317842e-01 -2.24698633e-01 -4.65371124e-02 6.44647598e-01 9.32449281e-01 -1.05127180e+00 -5.96465528e-01 2.65670270e-01 1.14369102e-01 4.21625674e-01 -7.72344828e-01 2.42104084e-04 -4.44089353e-01 -7.73009479e-01 -3.48390877e-01 -9.56523359e-01 -5.81683695e-01 -8.67575943e-01 -7.63161242e-01 -4.67390269e-01 2.74308234e-01 -6.45175695e-01 1.61103642e+00 -1.98339963e+00 1.46833897e-01 2.44483337e-01 3.90063375e-01 4.59674112e-02 -2.76100338e-01 7.06096947e-01 4.11569566e-01 4.19143081e-01 2.86418885e-01 -5.15625060e-01 3.16539735e-01 -2.46768892e-01 -3.12959075e-01 -3.05047393e-01 -4.99680056e-04 1.13169014e+00 -1.13470256e+00 -4.56823766e-01 -3.86797041e-01 2.08730340e-01 -8.75164688e-01 6.17478013e-01 -5.15867412e-01 4.93334174e-01 -1.01312065e+00 2.99623869e-02 3.51951331e-01 -4.35433149e-01 -2.21646782e-02 7.29895234e-02 1.79173365e-01 1.88933000e-01 -7.80765474e-01 1.30575490e+00 -6.85144365e-01 3.83335561e-01 -5.20862639e-01 -1.63263202e-01 1.20955694e+00 5.14652550e-01 7.95008391e-02 -1.15197942e-01 5.79801857e-01 7.42917368e-03 4.50305566e-02 -6.12158656e-01 1.21103346e+00 -9.92890224e-02 -4.87891942e-01 1.02697706e+00 -7.42197707e-02 3.61283161e-02 3.29818547e-01 7.41160989e-01 6.41035259e-01 -5.82229733e-01 1.08293712e-01 -1.00516282e-01 5.24347901e-01 -6.32139385e-01 2.22145125e-01 6.57298803e-01 1.60133287e-01 7.41075575e-01 3.73717576e-01 1.32321939e-01 -8.64119470e-01 -7.17702091e-01 3.87843817e-01 1.35257435e+00 -4.19091396e-02 -5.56679726e-01 -8.34341288e-01 -4.65848625e-01 -1.93839327e-01 5.92772305e-01 -5.44138134e-01 -1.82099819e-01 -3.99217397e-01 -9.70170200e-01 2.46730343e-01 4.24649715e-01 2.01557472e-01 -1.36392105e+00 4.40042853e-01 6.32267714e-01 -6.37313426e-01 -8.29609871e-01 -1.11722600e+00 -6.69894934e-01 -5.25341749e-01 -3.62632364e-01 -8.27146471e-01 -9.60564137e-01 1.10036552e+00 1.48567468e-01 1.07024419e+00 5.91862164e-02 2.92730868e-01 2.10348099e-01 -9.90568340e-01 -5.98114610e-01 -5.50626338e-01 6.83443367e-01 1.69729695e-01 3.29295278e-01 1.99109346e-01 -6.51233375e-01 -4.93141174e-01 1.51379183e-01 -9.55636501e-01 4.56126541e-01 5.38194537e-01 8.02207112e-01 -2.04674661e-01 7.24028870e-02 1.15617716e+00 -1.39345157e+00 1.56519461e+00 -9.27246094e-01 1.32075816e-01 -6.40115375e-03 -4.75761592e-01 -1.37488604e-01 1.04669130e+00 -7.53993511e-01 -1.33889997e+00 -3.91745329e-01 -3.62066597e-01 3.97704303e-01 1.27683088e-01 7.66378880e-01 -3.27806979e-01 4.66062844e-01 6.67928755e-01 4.02457595e-01 -1.27398193e-01 -3.80613923e-01 4.12465870e-01 1.11265671e+00 1.29483670e-01 -3.88405532e-01 9.54305470e-01 -1.90010667e-01 -8.63514662e-01 -7.36574113e-01 -8.06159139e-01 -3.02177042e-01 -4.44181651e-01 -3.35929573e-01 7.39961326e-01 -9.02039111e-01 -7.39444137e-01 6.27433658e-01 -1.23264420e+00 -1.93889380e-01 2.67885476e-01 3.13673228e-01 -2.21597686e-01 2.38371313e-01 -9.06057358e-01 -7.94625282e-01 -6.73212588e-01 -7.78013110e-01 6.67424619e-01 5.94541907e-01 -4.25929487e-01 -1.46406627e+00 -3.69012915e-02 2.21279755e-01 6.71617746e-01 -5.27470969e-02 7.94691741e-01 -7.52877712e-01 -3.71776402e-01 -3.32852572e-01 -2.61542290e-01 4.45488662e-01 3.28327715e-01 -1.11881323e-01 -6.69937611e-01 -1.79702148e-01 -4.21433181e-01 -1.81222558e-01 3.60346198e-01 2.48698182e-02 1.01072860e+00 -8.61154079e-01 -2.12243497e-02 2.35014990e-01 1.05736840e+00 -9.72388685e-02 5.21900654e-01 -4.04823273e-01 1.00963056e+00 5.99439919e-01 4.47567940e-01 9.71977532e-01 9.05772626e-01 2.49944910e-01 1.79027066e-01 6.96978942e-02 4.20076519e-01 -5.44115007e-01 7.42867827e-01 1.54874969e+00 -7.01526999e-02 -1.03292406e+00 -4.56122279e-01 4.26854312e-01 -1.84694040e+00 -1.09138167e+00 -4.14262176e-01 1.80421674e+00 1.08899844e+00 7.74955470e-03 1.35324061e-01 -2.57642716e-01 9.15441930e-01 3.83144110e-01 -2.67148823e-01 -6.23137474e-01 5.63478321e-02 -6.54149987e-03 2.00923011e-01 5.53801179e-01 -6.05108738e-01 1.05839443e+00 5.64993715e+00 5.50635934e-01 -1.11449516e+00 6.70071319e-02 7.03931391e-01 2.01596215e-01 -9.47615027e-01 -5.43897390e-01 -1.25386786e+00 9.47042227e-01 9.16099787e-01 -8.85176480e-01 5.15350342e-01 6.44790411e-01 5.01587152e-01 4.77372527e-01 -8.34872425e-01 9.09824371e-01 5.23000479e-01 -1.08153868e+00 4.72018838e-01 -2.00536638e-01 1.12764204e+00 -2.26888895e-01 1.50371611e-01 5.27213693e-01 5.90938628e-01 -8.88369083e-01 5.19194126e-01 6.00591063e-01 5.74522674e-01 -8.54060531e-01 9.14051056e-01 6.34307444e-01 -1.01692605e+00 1.34342551e-01 -7.90473402e-01 -2.78981000e-01 5.91303229e-01 8.34747672e-01 -1.42929554e+00 2.89518893e-01 -4.95202206e-02 8.60094786e-01 -7.74791598e-01 6.81676745e-01 -2.51837581e-01 6.68566644e-01 2.01006249e-01 -8.68902087e-01 1.45750716e-01 -2.12968290e-01 2.36804768e-01 1.30978835e+00 5.55080116e-01 7.85177574e-02 3.06137711e-01 6.04728997e-01 -4.19831783e-01 7.66120255e-01 -2.53712356e-01 -5.45860827e-01 3.38990033e-01 1.36271858e+00 -7.90498480e-02 -3.77432168e-01 -1.96980819e-01 1.37601066e+00 3.44502568e-01 4.68426615e-01 -6.85631096e-01 -4.75811660e-01 4.75427628e-01 2.68861741e-01 -1.62157312e-01 3.22876386e-02 1.58293411e-01 -1.49414265e+00 -2.40255281e-01 -6.38930798e-01 -1.48638114e-01 -8.09487879e-01 -1.78136599e+00 9.06657219e-01 -4.26072031e-01 -1.22135818e+00 -6.62888765e-01 -2.49911606e-01 -1.04402924e+00 1.28561687e+00 -1.05576158e+00 -1.18710613e+00 -2.34525263e-01 5.60912430e-01 6.73044145e-01 -2.56854624e-01 8.36551189e-01 5.05966246e-01 -5.62193990e-01 8.84029210e-01 -5.05110882e-02 2.42881224e-01 8.15015078e-01 -1.38407362e+00 8.62838686e-01 5.21007299e-01 -3.18598151e-01 9.06135619e-01 7.76333630e-01 -8.60530078e-01 -9.03136849e-01 -1.56235325e+00 1.58422208e+00 -3.94869715e-01 7.41936207e-01 -5.32759845e-01 -5.26170850e-01 7.43624270e-01 3.05332065e-01 -6.19557917e-01 1.34652507e+00 3.04409295e-01 8.23844522e-02 2.91007996e-01 -7.23299742e-01 1.08460784e+00 1.03567457e+00 -4.80933458e-01 -3.63712966e-01 5.48151553e-01 8.90259802e-01 -2.28829041e-01 -6.43747091e-01 -2.73815721e-01 4.21535760e-01 -5.68037212e-01 4.73872572e-01 -7.77801871e-01 7.87826955e-01 -9.59650949e-02 1.67101473e-01 -1.90316260e+00 -6.28789961e-01 -9.29129660e-01 2.51058310e-01 1.65796340e+00 8.10522497e-01 -5.63493252e-01 7.44094908e-01 6.96390748e-01 -2.98720032e-01 -5.56870282e-01 1.43337682e-01 -9.80008319e-02 -2.00651929e-01 -2.17746094e-01 9.72815514e-01 8.75299692e-01 4.07240540e-01 9.17762995e-01 -1.12568212e+00 -4.50352617e-02 1.92071665e-02 -1.30790651e-01 1.11111701e+00 -1.37401593e+00 -5.89556873e-01 -2.69289106e-01 -3.02437674e-02 -1.38498557e+00 2.55842298e-01 -1.00463104e+00 2.65923113e-01 -1.54595232e+00 2.62261659e-01 -6.02069199e-01 9.31076780e-02 -5.73177859e-02 -7.18669295e-01 3.54880899e-01 8.12543407e-02 6.58798739e-02 -6.27105474e-01 7.27593899e-01 1.64297843e+00 4.69102263e-02 -2.58471876e-01 4.38087344e-01 -1.35287356e+00 5.77199340e-01 9.91160989e-01 -1.05466127e-01 -8.24700058e-01 -4.75661337e-01 8.31015289e-01 3.00403554e-02 -3.51719677e-01 -5.99315941e-01 -4.24870737e-02 -2.19247982e-01 1.61144406e-01 -2.52468050e-01 2.35526904e-01 -2.09505990e-01 -7.22253993e-02 -5.36262877e-02 -7.30745792e-01 3.42522562e-01 -4.32594419e-01 6.64680600e-01 -8.81532505e-02 -4.16343093e-01 3.30242634e-01 -2.54871130e-01 -3.84306252e-01 7.00271010e-01 -7.57633388e-01 4.03806381e-02 4.56603020e-01 2.55169094e-01 -3.93023610e-01 -1.12341559e+00 -7.91949570e-01 1.99338138e-01 3.93842548e-01 6.27307296e-01 6.58354461e-01 -1.52885485e+00 -1.05279934e+00 1.25291586e-01 3.75660062e-01 -8.00315067e-02 5.30728877e-01 2.76216984e-01 -2.15556219e-01 -3.95895019e-02 3.60385925e-02 9.64727327e-02 -1.07572734e+00 1.81964129e-01 -3.12007844e-01 -3.25000793e-01 -1.22683048e-01 1.16906416e+00 1.66502252e-01 -5.32897949e-01 -2.05416799e-01 -5.22369891e-02 -7.76241362e-01 3.35963368e-01 9.35346127e-01 -1.01781823e-02 -3.10372591e-01 -7.98471391e-01 4.79661107e-01 -1.35447264e-01 -4.73955065e-01 -7.05367774e-02 1.03803432e+00 -4.37249213e-01 -1.55419484e-01 5.90936840e-01 1.23546934e+00 5.12537003e-01 -7.52417028e-01 -3.01370025e-01 -4.47191298e-01 -4.03887272e-01 -5.04102767e-01 -4.41271424e-01 -9.26680326e-01 6.91128492e-01 -1.05374083e-01 4.21283066e-01 6.25712454e-01 -1.73070148e-01 1.44281411e+00 3.75984073e-01 3.58666182e-01 -1.12005186e+00 4.01995003e-01 8.93553674e-01 8.10766101e-01 -1.25693119e+00 -4.17791605e-01 -4.84997660e-01 -1.11980140e+00 8.07677686e-01 7.64373720e-01 -4.95513529e-02 7.57903874e-01 -3.89405489e-01 2.55809426e-01 3.32531095e-01 -9.57068920e-01 4.58296128e-02 3.35607767e-01 6.75831676e-01 8.04887474e-01 3.93039078e-01 -6.31690919e-01 1.09287608e+00 -9.35269773e-01 -3.18708509e-01 1.09008539e+00 4.73358929e-01 -6.65576041e-01 -1.41375518e+00 6.95061907e-02 8.38484168e-01 -3.81523520e-01 -4.97331858e-01 -3.21697354e-01 -1.48925319e-01 2.68654395e-02 1.20644832e+00 -2.98103988e-02 -8.26543570e-01 -1.44544065e-01 -1.00781554e-02 -1.93923354e-01 -1.28779018e+00 -6.92736804e-01 -1.80937499e-01 3.39288890e-01 1.60762504e-01 -1.75012625e-03 -6.73838198e-01 -1.19034457e+00 -6.51379883e-01 -4.39888030e-01 6.40477896e-01 4.85024422e-01 7.72057831e-01 3.61346334e-01 5.10157764e-01 1.11073625e+00 -9.49864805e-01 -3.02007675e-01 -1.40443325e+00 -8.18234384e-01 7.39798784e-01 -1.09139025e-01 -6.47887960e-02 -3.31934959e-01 3.29106927e-01]
[12.113656997680664, 9.024527549743652]
8ddba1e6-4705-4c40-86cd-c453c2a42d43
amigos-a-dataset-for-affect-personality-and
1702.02510
null
http://arxiv.org/abs/1702.02510v3
http://arxiv.org/pdf/1702.02510v3.pdf
AMIGOS: A Dataset for Affect, Personality and Mood Research on Individuals and Groups
We present AMIGOS-- A dataset for Multimodal research of affect, personality traits and mood on Individuals and GrOupS. Different to other databases, we elicited affect using both short and long videos in two social contexts, one with individual viewers and one with groups of viewers. The database allows the multimodal study of the affective responses, by means of neuro-physiological signals of individuals in relation to their personality and mood, and with respect to the social context and videos' duration. The data is collected in two experimental settings. In the first one, 40 participants watched 16 short emotional videos. In the second one, the participants watched 4 long videos, some of them alone and the rest in groups. The participants' signals, namely, Electroencephalogram (EEG), Electrocardiogram (ECG) and Galvanic Skin Response (GSR), were recorded using wearable sensors. Participants' frontal HD video and both RGB and depth full body videos were also recorded. Participants emotions have been annotated with both self-assessment of affective levels (valence, arousal, control, familiarity, liking and basic emotions) felt during the videos as well as external-assessment of levels of valence and arousal. We present a detailed correlation analysis of the different dimensions as well as baseline methods and results for single-trial classification of valence and arousal, personality traits, mood and social context. The database is made publicly available.
[]
2017-04-13
null
null
null
null
['continuous-affect-estimation']
['computer-vision']
[-2.39429519e-01 -1.28505349e-01 2.63833642e-01 -6.15608871e-01 -1.49292260e-01 -7.56223381e-01 3.77180040e-01 4.59917367e-01 -3.96806002e-01 6.12351179e-01 3.98125470e-01 7.88849473e-01 9.97073427e-02 -3.38524252e-01 -1.84650913e-01 -7.36666858e-01 -5.12755275e-01 -4.83312696e-01 -4.72704142e-01 -1.54601678e-01 1.18080109e-01 1.02646269e-01 -1.79978740e+00 4.13116068e-01 2.41319373e-01 1.55959356e+00 -4.21108842e-01 7.81269729e-01 4.64304090e-01 3.74587744e-01 -7.80612230e-01 -5.49128771e-01 -5.39624169e-02 -5.11614680e-01 -3.26303333e-01 7.54161999e-02 2.64224291e-01 -8.46496075e-02 -2.84762174e-01 8.38537753e-01 9.45894897e-01 1.43326312e-01 1.09411933e-01 -1.35894608e+00 -3.66452396e-01 1.79493949e-01 -4.05343533e-01 3.89049500e-01 1.35339928e+00 4.02904212e-01 4.91091341e-01 -6.72429800e-01 7.11347699e-01 1.07512045e+00 5.98044097e-01 5.67099392e-01 -1.18548489e+00 -6.71824574e-01 -6.61474243e-02 4.24772888e-01 -1.25998366e+00 -5.74496567e-01 6.89953148e-01 -6.13691747e-01 7.27961540e-01 4.41393673e-01 1.50463462e+00 1.66889369e+00 4.06913996e-01 1.27620772e-01 1.51901150e+00 3.91829088e-02 4.83747393e-01 7.18307793e-01 2.46825099e-01 7.10957423e-02 -2.07131475e-01 -6.88823089e-02 -8.82459402e-01 -2.98690528e-01 4.56755996e-01 -2.76838809e-01 -5.88982463e-01 -1.78635325e-02 -1.28865385e+00 4.24617589e-01 8.74959379e-02 4.29091841e-01 -7.78095126e-01 1.95575953e-02 8.15798938e-01 7.27482617e-01 4.92448032e-01 3.48543108e-01 -2.83544600e-01 -6.07138932e-01 -6.59607232e-01 -6.00230545e-02 7.98749506e-01 4.01052117e-01 3.72616053e-01 -1.35599121e-01 -2.70736367e-01 7.20412791e-01 3.94923612e-02 4.67814535e-01 6.24946177e-01 -6.96136296e-01 -1.87473610e-01 4.63564813e-01 3.93606722e-03 -1.52281415e+00 -7.19899654e-01 2.06341997e-01 -5.32986164e-01 5.03561571e-02 -2.00873949e-02 -6.70137405e-01 -1.44498691e-01 1.98807728e+00 4.41264927e-01 1.60504162e-01 -7.98899531e-02 1.15973520e+00 1.22950017e+00 4.70564455e-01 1.82602197e-01 -9.97456789e-01 1.83418667e+00 -8.65807943e-03 -1.33018923e+00 3.07739154e-03 1.19880080e-01 -4.76757079e-01 9.68889952e-01 6.73765659e-01 -1.44163167e+00 -7.28611708e-01 -8.52369785e-01 4.45090860e-01 -3.98586601e-01 -2.08410576e-01 4.19292182e-01 1.01275826e+00 -1.16461897e+00 6.30789876e-01 -3.45206797e-01 -6.18103981e-01 1.87234029e-01 3.41266274e-01 -8.00117850e-01 5.34084260e-01 -1.48518598e+00 7.67284095e-01 -2.21115008e-01 2.12236166e-01 -7.60551155e-01 -5.15408456e-01 -6.78567290e-01 -2.99104005e-01 -3.30817014e-01 -3.36379409e-01 7.75564969e-01 -1.66315937e+00 -1.82778466e+00 1.20491874e+00 -2.46377550e-02 1.13971725e-01 1.32439762e-01 -1.98841512e-01 -8.45343888e-01 5.72179019e-01 -3.14515710e-01 5.66068947e-01 7.99012065e-01 -6.51360393e-01 2.45586142e-01 -9.03298974e-01 -3.03704198e-02 5.30138075e-01 -6.03517056e-01 3.22415024e-01 -9.41927880e-02 -1.39477313e-01 -3.46959800e-01 -7.14145422e-01 1.48334593e-01 -1.71483636e-01 -9.83416960e-02 7.07357302e-02 4.33325231e-01 -6.09042764e-01 1.16454840e+00 -2.62415338e+00 4.92107272e-01 3.44194621e-01 2.48430774e-01 -9.20007154e-02 -7.70512223e-02 5.84148049e-01 -3.83140147e-01 -2.04321695e-03 4.45665926e-01 -3.40924799e-01 -4.17986214e-02 -2.62565076e-01 2.38405019e-01 8.90183151e-01 -2.27073863e-01 7.29732990e-01 -7.46493280e-01 -4.08450812e-01 2.59070724e-01 7.50803173e-01 -2.03566834e-01 4.79251713e-01 4.47703987e-01 8.39907169e-01 -1.60658620e-02 5.93607247e-01 5.45701504e-01 4.67421293e-01 1.59891129e-01 -2.91925997e-01 -2.55661726e-01 -1.02302469e-01 -1.13315237e+00 1.46061659e+00 -4.35348094e-01 1.04519689e+00 4.27713841e-01 -5.68780899e-01 9.36516106e-01 9.06667352e-01 5.15952289e-01 -8.53084147e-01 6.02419019e-01 -3.41344327e-01 -1.84025988e-01 -1.16074026e+00 1.69147477e-01 -1.92928150e-01 -2.76611090e-01 2.58121401e-01 8.76506492e-02 4.59913127e-02 1.39007330e-01 1.29376769e-01 8.87591064e-01 -2.46259660e-01 3.42595935e-01 -1.28583446e-01 3.60548377e-01 -7.11280286e-01 1.47220790e-01 9.82372165e-02 -5.12119234e-01 2.61031687e-01 9.49028730e-01 -3.14267814e-01 -2.58084238e-01 -8.53668094e-01 -1.39218122e-01 1.05603349e+00 3.37862074e-01 -6.12429142e-01 -8.41635644e-01 9.25461650e-02 -2.23613560e-01 4.67517972e-01 -9.69916880e-01 -4.21880543e-01 2.92129993e-01 -5.41474342e-01 3.09062749e-01 2.45148301e-01 2.05690145e-01 -1.54422235e+00 -1.03976715e+00 -1.88831300e-01 -3.40396315e-01 -1.11232841e+00 -4.38632220e-02 -2.05798466e-02 -6.12048984e-01 -8.53234410e-01 -2.33137533e-01 -2.28035167e-01 2.66011804e-01 -4.18905497e-01 1.03017187e+00 -4.80344832e-01 -6.06240153e-01 1.22569418e+00 -4.13998723e-01 -2.71544755e-01 3.07059854e-01 -7.49845982e-01 4.24964577e-01 5.04402757e-01 3.34084392e-01 -9.07171130e-01 -9.00249302e-01 1.47457987e-01 -5.54449260e-01 -4.86063719e-01 1.32754862e-01 -1.43271238e-02 3.84366900e-01 -4.98069406e-01 5.29165924e-01 -1.16958998e-01 9.98850882e-01 -7.44816542e-01 2.00437844e-01 -2.18483984e-01 -4.62983288e-02 -1.20596123e+00 2.92568475e-01 -7.42112637e-01 -9.29883540e-01 -1.96924925e-01 1.59647185e-02 -3.65992367e-01 -6.57258034e-01 3.17233682e-01 -1.23085618e-01 -7.62941763e-02 7.02249944e-01 -1.26810357e-01 -9.30294618e-02 1.17386729e-01 9.98763740e-02 9.29227889e-01 6.66063428e-01 -1.25809550e-01 -4.89176139e-02 2.89982021e-01 -2.52441525e-01 -1.36893082e+00 -3.86670858e-01 -3.03694785e-01 -6.52844608e-01 -1.14149201e+00 1.10298264e+00 -1.17777669e+00 -1.29567182e+00 5.84753752e-01 -7.73405492e-01 -1.40104502e-01 -1.37736484e-01 7.70414650e-01 -5.24441659e-01 2.06710249e-01 -7.24897861e-01 -9.48950052e-01 -5.61795175e-01 -7.37773657e-01 6.48644030e-01 4.55469787e-01 -9.75794077e-01 -8.97055745e-01 3.80990475e-01 1.56631067e-01 2.46132568e-01 8.97922456e-01 2.38953128e-01 -3.67677063e-01 5.77816784e-01 -4.03797656e-01 3.46692950e-01 2.73747742e-01 -1.21659487e-02 -6.36837631e-02 -1.20846403e+00 -1.36488706e-01 2.77757078e-01 -8.52085650e-01 7.74671063e-02 6.20214701e-01 8.94963562e-01 -5.71474619e-02 4.57040407e-02 3.15290511e-01 1.04026723e+00 3.44130188e-01 9.80006337e-01 -2.82956451e-01 2.89313644e-01 8.76103163e-01 3.90168518e-01 8.45643520e-01 -4.43030242e-03 5.21601915e-01 5.40895522e-01 -4.06987630e-02 5.46150982e-01 4.23749626e-01 9.41616535e-01 6.02675080e-01 -3.10566068e-01 -1.97606198e-02 -3.35492730e-01 2.62336820e-01 -1.33282793e+00 -1.13874233e+00 -5.59598446e-01 2.35757113e+00 5.72512567e-01 -3.83197427e-01 4.77579713e-01 1.73708692e-01 7.67703354e-01 9.65821519e-02 -4.73954111e-01 -9.97114182e-01 -2.00304925e-01 2.09384009e-01 -3.02656680e-01 1.51393339e-02 -8.41352940e-01 2.01593474e-01 6.61360407e+00 -3.21941935e-02 -1.47504973e+00 -1.12347128e-02 6.08717203e-01 -9.96557653e-01 8.83929338e-03 -4.40934479e-01 -8.17881674e-02 5.82462788e-01 1.51711738e+00 1.75938591e-01 6.16387248e-01 5.43318868e-01 5.19605875e-01 -5.46787202e-01 -1.21401012e+00 1.41887534e+00 4.07667518e-01 -5.10536730e-01 -7.61049807e-01 -1.86134413e-01 1.79487333e-01 -3.90044183e-01 4.27203663e-02 2.06253290e-01 -9.66010094e-01 -9.22945082e-01 6.53942883e-01 1.00190568e+00 1.10712707e+00 -6.80778265e-01 6.27168834e-01 -1.87495872e-01 -7.41349578e-01 -3.89523730e-02 1.33972704e-01 -4.46177185e-01 2.17983589e-01 5.40849924e-01 1.85033217e-01 9.56199020e-02 1.05431163e+00 8.47179711e-01 -3.60948920e-01 5.42017102e-01 -2.81447899e-02 4.66608167e-01 -6.35954514e-02 -2.56724477e-01 -3.43199223e-01 -3.21715534e-01 5.79929292e-01 1.44548810e+00 1.97543979e-01 6.02555692e-01 -2.82174796e-01 6.03756487e-01 2.96472102e-01 3.64428192e-01 -6.31396949e-01 -2.65312016e-01 3.42122525e-01 1.84272254e+00 -5.85419357e-01 -2.73501366e-01 -2.54890949e-01 1.09424984e+00 -1.95260391e-01 1.95817605e-01 -9.49861467e-01 -5.42639613e-01 7.97816634e-01 2.75119804e-02 -3.42801958e-01 1.76671833e-01 -8.80725458e-02 -1.10016668e+00 2.71925647e-02 -7.11898208e-01 1.97252274e-01 -1.12322247e+00 -1.23488879e+00 5.70767045e-01 4.34181914e-02 -8.63390028e-01 1.04982585e-01 -1.64074957e-01 -8.47346425e-01 7.03056097e-01 -6.25529170e-01 -4.59645957e-01 -8.62108529e-01 9.13781703e-01 2.57906705e-01 3.15055549e-01 9.46105719e-01 4.49502945e-01 -9.12836194e-01 3.38918179e-01 -6.66050792e-01 -4.80434448e-01 1.10784698e+00 -1.02031481e+00 -5.89540601e-01 -1.20742217e-01 -4.69922394e-01 3.38267446e-01 7.61571229e-01 -2.89705426e-01 -1.46426940e+00 -4.49502110e-01 4.77466226e-01 -2.38509759e-01 5.03896832e-01 -5.41793764e-01 -4.64316428e-01 3.87190878e-01 8.29164147e-01 -1.18541539e-01 1.30754375e+00 1.03433885e-01 1.75389245e-01 -1.25880644e-01 -1.31668723e+00 3.92806798e-01 7.85006762e-01 -5.14073491e-01 -4.30176139e-01 1.24730326e-01 9.76219699e-02 -2.18905449e-01 -1.49528265e+00 5.26741073e-02 1.12140000e+00 -1.50277257e+00 6.08578861e-01 -2.08973214e-01 4.88164365e-01 1.89800635e-01 -2.18641117e-01 -1.58329773e+00 -9.90718529e-02 -6.37948275e-01 1.75235957e-01 1.43647861e+00 -1.76300198e-01 -5.84836721e-01 -9.36151370e-02 7.38424897e-01 6.52612597e-02 -6.96276009e-01 -6.22104883e-01 -1.52085230e-01 -4.58972842e-01 -4.51145113e-01 6.17376678e-02 9.95320976e-01 1.07634485e+00 4.85450983e-01 -4.01748151e-01 -1.44594043e-01 -5.67148775e-02 -1.42724708e-01 6.05296552e-01 -1.25003600e+00 5.56892296e-03 -2.53302246e-01 -7.94117808e-01 2.06940964e-01 5.39136268e-02 -4.55803275e-01 -4.41654712e-01 -1.15403903e+00 2.98130423e-01 6.06547296e-01 -2.82968849e-01 2.17816040e-01 1.81135625e-01 5.07298768e-01 5.15095331e-02 -3.24038774e-01 -6.94263816e-01 4.51138973e-01 8.91978800e-01 4.62378621e-01 -5.32949090e-01 -3.79470825e-01 -4.48605031e-01 6.74489737e-01 6.71495140e-01 2.05106996e-02 -3.24504405e-01 3.32351387e-01 5.63796699e-01 5.17454982e-01 5.31546772e-01 -1.16061556e+00 -1.52022213e-01 3.41300726e-01 8.07141006e-01 -2.82473773e-01 9.17284071e-01 -5.57045043e-01 3.87814969e-01 3.69370341e-01 -3.22230309e-01 2.47916311e-01 3.23581755e-01 4.45659071e-01 -1.33226395e-01 1.87226921e-01 9.00557041e-01 -1.59488901e-01 -3.99024427e-01 -8.56984481e-02 -7.59879768e-01 -1.46032304e-01 1.56284440e+00 -4.60138500e-01 -2.11005360e-01 -8.02216947e-01 -1.51160347e+00 9.79826301e-02 3.42618346e-01 2.91827321e-01 6.16143465e-01 -1.31301260e+00 -3.62279117e-01 3.23103458e-01 3.92691866e-02 -9.68979180e-01 9.58877087e-01 1.66359174e+00 1.63230404e-01 -2.26793751e-01 -8.49059343e-01 -5.70417702e-01 -1.61838126e+00 5.56945622e-01 2.56653816e-01 4.01294500e-01 -2.09004462e-01 7.15388536e-01 1.70011103e-01 1.88119397e-01 3.54420960e-01 1.67490572e-01 -7.43589878e-01 1.07686913e+00 8.22833538e-01 5.66418409e-01 1.70520172e-02 -7.62774825e-01 -3.09818029e-01 5.87627053e-01 5.25374651e-01 -1.78634703e-01 1.12213659e+00 -6.24726713e-01 -3.19737643e-01 1.13871241e+00 1.19258070e+00 -2.91070668e-03 -5.90216517e-01 4.58805323e-01 -5.95805466e-01 -3.36791188e-01 2.64550708e-02 -9.49987710e-01 -1.02274859e+00 9.76050913e-01 1.03807056e+00 4.72554833e-01 1.49922836e+00 -2.27169380e-01 4.18510407e-01 -4.06102352e-02 5.48121296e-02 -1.29258740e+00 4.31994557e-01 -6.00436255e-02 1.04989707e+00 -8.15865993e-01 -1.67517617e-01 -9.57354307e-02 -1.11269295e+00 1.02649736e+00 5.62372208e-01 -1.53411940e-01 6.66220069e-01 1.36075780e-01 7.87601396e-02 -6.47974372e-01 -1.09104729e+00 6.26069605e-02 1.04937777e-01 6.00718379e-01 6.86209857e-01 1.67083055e-01 -4.97687340e-01 9.97218370e-01 -3.30767810e-01 -2.08137445e-02 6.40277624e-01 4.71810341e-01 6.67769611e-02 -2.00052366e-01 -5.65122187e-01 3.17401856e-01 -6.20904088e-01 3.24962616e-01 -7.87246823e-01 5.93639016e-01 4.65041041e-01 1.12850201e+00 3.15194458e-01 -5.66862702e-01 7.37087488e-01 3.61382574e-01 6.31248236e-01 -3.68608564e-01 -1.08805668e+00 8.81455615e-02 3.31372738e-01 -1.00153744e+00 -5.64828455e-01 -1.00889862e+00 -9.34633374e-01 -1.99701726e-01 8.20042342e-02 -6.48750663e-02 9.05467212e-01 5.33524871e-01 3.55904579e-01 4.79131967e-01 8.31031561e-01 -1.29262662e+00 1.65433154e-01 -1.27959716e+00 -1.07856488e+00 6.30482018e-01 2.40252525e-01 -5.32035053e-01 -6.16572857e-01 6.55017719e-02]
[13.463116645812988, 2.602109432220459]
d24cd5e2-6328-43d8-8029-29f6dcb1db44
gaze-estimation-using-transformer
2105.14424
null
https://arxiv.org/abs/2105.14424v1
https://arxiv.org/pdf/2105.14424v1.pdf
Gaze Estimation using Transformer
Recent work has proven the effectiveness of transformers in many computer vision tasks. However, the performance of transformers in gaze estimation is still unexplored. In this paper, we employ transformers and assess their effectiveness for gaze estimation. We consider two forms of vision transformer which are pure transformers and hybrid transformers. We first follow the popular ViT and employ a pure transformer to estimate gaze from images. On the other hand, we preserve the convolutional layers and integrate CNNs as well as transformers. The transformer serves as a component to complement CNNs. We compare the performance of the two transformers in gaze estimation. The Hybrid transformer significantly outperforms the pure transformer in all evaluation datasets with less parameters. We further conduct experiments to assess the effectiveness of the hybrid transformer and explore the advantage of self-attention mechanism. Experiments show the hybrid transformer can achieve state-of-the-art performance in all benchmarks with pre-training.To facilitate further research, we release codes and models in https://github.com/yihuacheng/GazeTR.
['Feng Lu', 'Yihua Cheng']
2021-05-30
null
null
null
null
['gaze-estimation']
['computer-vision']
[-2.58564293e-01 -1.67385023e-02 7.59432539e-02 -3.27146262e-01 -2.31579289e-01 -3.41443598e-01 5.04450560e-01 -5.94524980e-01 -3.75767976e-01 3.62067163e-01 9.57749113e-02 -2.82352895e-01 2.88158298e-01 -4.19384748e-01 -8.26368213e-01 -8.26731980e-01 4.37007129e-01 -1.88507006e-01 5.28581142e-01 -1.60020396e-01 4.27139401e-01 4.21160366e-03 -1.86949837e+00 -3.86217348e-02 1.08722365e+00 1.31411219e+00 1.98121935e-01 3.27339888e-01 -4.73160408e-02 1.06939697e+00 -4.21042591e-01 -7.73159504e-01 5.21696918e-02 -3.75975370e-01 -7.25496054e-01 -2.22972542e-01 5.48448563e-01 -4.38624442e-01 -2.28504494e-01 1.18502283e+00 6.47448003e-01 -2.05745265e-01 5.33551097e-01 -1.59472096e+00 -9.26394105e-01 3.59995723e-01 -9.18741107e-01 3.73844117e-01 3.54596525e-01 3.83017868e-01 8.33505094e-01 -9.09370363e-01 6.18116139e-03 1.14562917e+00 6.39964700e-01 4.33118671e-01 -6.14232719e-01 -1.28653026e+00 3.49736184e-01 6.59599900e-01 -1.30526602e+00 -7.27054894e-01 7.48551071e-01 -1.91326320e-01 7.09136724e-01 1.41637875e-02 6.03796721e-01 1.07236230e+00 8.44191387e-02 1.09226489e+00 1.39564502e+00 -3.04898262e-01 -2.86690772e-01 1.38070852e-01 2.45385140e-01 1.00096107e+00 -3.63417232e-04 -5.55863902e-02 -7.95250535e-01 4.18243229e-01 4.15286183e-01 1.90104455e-01 -7.11834550e-01 -1.29746154e-01 -1.08439410e+00 4.62597132e-01 9.34141874e-01 -4.84788902e-02 -2.17768908e-01 1.68080345e-01 8.80774930e-02 1.71919912e-01 6.21412516e-01 9.55101103e-02 2.98959482e-02 -1.31385088e-01 -8.09364498e-01 7.96735510e-02 5.15705407e-01 1.17694795e+00 6.61057532e-01 -1.32667765e-01 -5.42778075e-01 7.04988718e-01 5.03899395e-01 6.27370536e-01 5.65663159e-01 -7.94591784e-01 3.97898912e-01 6.84862792e-01 -1.11437373e-01 -6.27760828e-01 -1.02338031e-01 -2.54096270e-01 -6.38722181e-01 1.47772864e-01 4.01196808e-01 -3.85573544e-02 -9.07434106e-01 1.68047369e+00 3.04220170e-01 5.51559567e-01 -3.22950512e-01 9.42102909e-01 1.24104941e+00 3.94791633e-01 8.88976604e-02 -6.84127863e-03 1.55771017e+00 -1.45489478e+00 -8.48762929e-01 -6.16351962e-02 4.14027452e-01 -7.60776222e-01 1.49276364e+00 2.37439319e-01 -1.27993691e+00 -5.04620194e-01 -9.74376142e-01 -4.24411654e-01 -2.72639543e-01 3.90260309e-01 4.78453487e-01 6.86769843e-01 -1.38898671e+00 3.43038678e-01 -9.60262597e-01 -4.50764805e-01 7.23587811e-01 4.66283590e-01 -1.35113344e-01 1.42427579e-01 -9.83642578e-01 7.79295444e-01 -2.30184421e-01 2.98512608e-01 -8.99257660e-01 -6.23932302e-01 -9.08108950e-01 2.79177755e-01 3.64038497e-01 -6.56313241e-01 1.67411590e+00 -9.99917328e-01 -1.70187175e+00 8.57779324e-01 -7.49676883e-01 -2.67414778e-01 5.05664051e-01 -4.04562473e-01 -6.21081069e-02 -1.78327560e-02 -5.37523888e-02 7.74060190e-01 8.42015803e-01 -8.14970911e-01 -7.61188149e-01 -4.27567124e-01 4.92727935e-01 2.09913969e-01 -7.50662267e-01 1.81209296e-01 -8.00182819e-01 -1.15483157e-01 -2.82655656e-01 -1.02742958e+00 5.15777826e-01 1.99553609e-01 -5.75493455e-01 -6.71965659e-01 8.19754303e-01 -3.90282273e-01 1.35980546e+00 -2.13594055e+00 -6.86400244e-03 -3.94587606e-01 7.36172855e-01 3.37271363e-01 -2.24622642e-03 1.49143949e-01 5.26552424e-02 1.02536343e-01 8.62792805e-02 -9.40251231e-01 8.57747272e-02 -2.39644393e-01 -1.60692930e-01 4.16732490e-01 1.92937434e-01 1.24313867e+00 -5.80439031e-01 -6.41787291e-01 -3.32845785e-02 6.07149541e-01 -3.08267623e-01 3.79403442e-01 8.52957293e-02 2.69451678e-01 -5.32281697e-01 7.55015671e-01 7.25093305e-01 -6.22022629e-01 -2.47299448e-01 -4.61876541e-01 -2.85337627e-01 4.23427224e-01 -4.00559127e-01 1.38339865e+00 -1.40545607e-01 9.39124167e-01 -3.32914650e-01 -5.16764998e-01 6.61751270e-01 1.58580020e-01 -1.16835125e-01 -1.03721142e+00 5.32728195e-01 -5.62922172e-02 1.76029895e-02 -6.74713433e-01 4.74599481e-01 1.17322966e-01 4.57856029e-01 5.01179874e-01 1.46196499e-01 4.85162139e-01 -3.59444804e-02 1.30946934e-01 7.44896710e-01 3.46903652e-01 1.39217809e-01 -2.08644077e-01 5.79891205e-01 -4.78857994e-01 2.96543807e-01 2.74807572e-01 -5.75616837e-01 6.17239058e-01 7.84662545e-01 -3.04208416e-02 -5.55631220e-01 -8.94987404e-01 1.83842838e-01 1.41992331e+00 4.76866722e-01 -3.36483419e-01 -1.04597449e+00 -7.88698316e-01 -3.49194974e-01 4.71585870e-01 -9.93967056e-01 -1.77468300e-01 -2.83137172e-01 -5.30023336e-01 4.71032381e-01 7.55313993e-01 1.00233078e+00 -1.17351902e+00 -7.20259845e-01 -6.56247079e-01 -3.85378450e-01 -1.10257995e+00 -7.25793421e-01 -3.49260181e-01 -4.24651265e-01 -1.18237519e+00 -1.06074166e+00 -7.34513581e-01 5.62080681e-01 7.13356137e-01 1.08389878e+00 2.71025479e-01 4.53569353e-01 1.67749047e-01 -4.09105867e-01 -9.86559272e-01 3.42120707e-01 4.77026254e-01 -1.82978183e-01 1.23190790e-01 6.77228153e-01 -4.56733108e-01 -1.04454279e+00 4.90139574e-01 -4.14228082e-01 1.29434898e-01 5.51233947e-01 5.71076393e-01 1.33798078e-01 -4.84290928e-01 1.46223798e-01 -9.39113259e-01 7.58463442e-01 -3.95345747e-01 -6.32807553e-01 2.71882057e-01 -6.87890768e-01 1.01052672e-01 8.13817903e-02 -3.90669465e-01 -1.08970225e+00 -2.08588958e-01 -1.71006009e-01 -8.27664435e-01 1.72064617e-01 3.59593302e-01 -8.88580456e-02 -2.75051653e-01 3.00012082e-01 2.18116105e-01 1.14700519e-01 -5.06182671e-01 5.25165629e-03 7.94032276e-01 3.81247193e-01 -2.36527860e-01 5.84998369e-01 3.62820566e-01 -3.71854573e-01 -4.20345396e-01 -1.21806955e+00 -3.06370944e-01 -1.69279978e-01 -2.53809690e-01 8.88508618e-01 -9.76529300e-01 -1.39773715e+00 8.64243329e-01 -9.98380363e-01 -3.87687683e-01 3.82681698e-01 3.13271910e-01 -1.76513404e-01 1.30910918e-01 -5.16904533e-01 -7.07908988e-01 -6.55818284e-01 -1.53176773e+00 1.14274430e+00 7.89262712e-01 2.45083898e-01 -6.34556651e-01 1.68107208e-02 4.25600678e-01 5.31172693e-01 -1.50077671e-01 3.44182283e-01 -3.40115786e-01 -7.09906161e-01 1.81586742e-01 -6.47284448e-01 1.44769207e-01 -7.72979856e-02 1.56920984e-01 -1.64370775e+00 -3.25917363e-01 -1.27428696e-01 -4.25547034e-01 1.02529407e+00 4.09652174e-01 1.36008167e+00 -7.16607552e-03 -4.82712656e-01 9.10431921e-01 1.12131798e+00 -3.46479788e-02 1.00626051e+00 4.72772539e-01 8.31316769e-01 6.07795596e-01 5.08307636e-01 -9.01590660e-03 1.02575529e+00 4.68105584e-01 6.49443984e-01 -1.39854163e-01 -3.01766116e-02 -1.67641267e-01 4.37341869e-01 6.49511158e-01 -5.71779549e-01 -4.21727657e-01 -1.00418413e+00 5.33230126e-01 -1.76545477e+00 -7.76140928e-01 -8.47763345e-02 1.98053265e+00 7.27393329e-01 1.05722144e-01 3.57298017e-01 8.32127500e-03 5.77163279e-01 1.11140274e-01 -6.07031584e-01 -3.62794921e-02 2.84406990e-01 2.02656329e-01 4.16452795e-01 4.49639559e-02 -1.00962412e+00 9.93333220e-01 5.75026464e+00 5.45998156e-01 -1.43815184e+00 3.44809145e-01 5.35444677e-01 -2.54644930e-01 2.13764189e-03 -2.19750181e-01 -9.47323799e-01 7.44832039e-01 8.38529527e-01 -1.76723748e-01 2.92247444e-01 6.38423383e-01 -2.53785197e-02 -6.16836883e-02 -1.05002689e+00 1.13741338e+00 1.66980833e-01 -9.94613290e-01 -2.47860223e-01 1.08867005e-01 3.19350809e-01 3.92165631e-01 3.97673547e-01 4.58130270e-01 8.73113126e-02 -1.06324852e+00 8.53927255e-01 6.00905240e-01 1.00936651e+00 -5.83451390e-01 9.20218527e-01 6.59712851e-02 -1.19248331e+00 -1.37909174e-01 -3.43844354e-01 -1.00658052e-01 -1.07151464e-01 -1.03574805e-01 -4.19575512e-01 3.44107717e-01 1.18540442e+00 1.00650764e+00 -1.05733848e+00 1.28828657e+00 -4.55172479e-01 7.21413493e-01 -1.70619220e-01 -1.54045343e-01 1.21982396e-01 -2.15462856e-02 1.72375947e-01 7.52820194e-01 3.56345475e-01 -1.76819209e-02 -4.57989484e-01 8.87396693e-01 -4.00029898e-01 -2.01213270e-01 -3.03521305e-01 1.32611543e-01 7.00281382e-01 1.24629045e+00 -3.29790175e-01 -1.06607087e-01 -5.89325309e-01 7.58120954e-01 5.75710297e-01 5.17803371e-01 -1.12456310e+00 -4.19049710e-01 6.58402741e-01 2.52884835e-01 4.17116821e-01 1.36404485e-01 1.30741075e-02 -1.20327103e+00 2.37112746e-01 -8.59808803e-01 8.86463001e-02 -1.32177877e+00 -1.02043068e+00 8.48999321e-01 1.94882900e-02 -1.23886228e+00 -4.97330241e-02 -5.95858455e-01 -9.10858095e-01 1.00651956e+00 -1.98362303e+00 -1.47332239e+00 -1.02186799e+00 8.75786066e-01 2.91679054e-01 -2.82595232e-02 4.16563749e-01 2.11367175e-01 -9.45482254e-01 1.03392315e+00 -2.01962844e-01 2.92323291e-01 9.69639778e-01 -1.18700337e+00 4.40885186e-01 8.31705093e-01 -2.12453723e-01 7.88701117e-01 4.65556383e-01 -1.13003470e-01 -1.19859552e+00 -8.79173756e-01 7.82691360e-01 -6.38762236e-01 5.22649109e-01 -4.26440299e-01 -9.60740387e-01 1.07737923e+00 8.74704182e-01 -5.01532890e-02 3.54218245e-01 3.44285667e-01 -5.62859654e-01 -2.70205438e-01 -7.66908646e-01 6.43787920e-01 9.94728029e-01 -5.84810674e-01 -4.15388852e-01 -2.00649694e-01 8.27484250e-01 -5.58195829e-01 -5.07402658e-01 3.08016568e-01 8.01469743e-01 -1.42220151e+00 7.69263327e-01 -1.24013014e-01 6.65745080e-01 -2.21298248e-01 4.25382704e-01 -1.16908503e+00 -2.25119382e-01 -4.39909786e-01 -1.58667967e-01 1.43363845e+00 2.45420247e-01 -1.01308537e+00 6.09404087e-01 4.63036776e-01 -4.31865565e-02 -1.05482101e+00 -4.18766528e-01 -4.04271036e-01 -2.52468418e-02 -5.22347502e-02 1.00381482e+00 7.44480312e-01 -2.31735393e-01 6.52094662e-01 -3.04292321e-01 2.37380918e-02 6.20106101e-01 -1.27474926e-02 9.14425194e-01 -9.90588903e-01 1.06086195e-01 -6.16996586e-01 -3.31271231e-01 -1.17942464e+00 2.09948584e-01 -2.93090850e-01 -1.21576644e-01 -1.28775978e+00 4.38921541e-01 -1.83032677e-01 -5.56858599e-01 6.38061404e-01 -5.49281597e-01 5.65063238e-01 3.78103495e-01 4.03345048e-01 -8.37839365e-01 8.01638365e-01 1.40078580e+00 -2.10590400e-02 6.91414997e-03 5.00398800e-02 -1.14209962e+00 6.30627036e-01 9.28539217e-01 -2.98271358e-01 -6.67822897e-01 -7.51701117e-01 1.69663966e-01 -4.04477507e-01 5.11716425e-01 -9.54295933e-01 5.15618145e-01 1.31334752e-01 1.45443127e-01 -5.79529703e-01 2.62325108e-01 -4.64200318e-01 -2.66921133e-01 2.13383487e-03 -2.03158379e-01 2.49294057e-01 2.85798967e-01 3.56464446e-01 -2.67105877e-01 5.67518808e-02 6.95940375e-01 1.62913352e-01 -6.48432851e-01 5.26028514e-01 2.52560586e-01 4.23629507e-02 8.85484636e-01 -3.42284530e-01 -7.45961726e-01 -5.05193412e-01 -1.93947822e-01 5.14492333e-01 6.45507216e-01 5.18981040e-01 5.06035984e-01 -1.13729489e+00 -4.09002364e-01 2.15373069e-01 3.44821393e-01 -9.56613198e-02 8.97594243e-02 1.44231570e+00 -3.50516796e-01 4.26968187e-01 -4.17138129e-01 -7.69253433e-01 -1.49631321e+00 5.98372698e-01 6.23289645e-01 1.32202566e-01 -2.93060988e-01 1.08646560e+00 8.93932164e-01 -7.00913668e-02 3.79127502e-01 -4.13032562e-01 -7.03899324e-01 -3.95031981e-02 8.87741029e-01 1.65128678e-01 -1.08671956e-01 -6.54349804e-01 -3.66171181e-01 6.46096706e-01 -2.36041427e-01 2.77922451e-01 1.24411035e+00 -4.59097952e-01 -1.10110737e-01 4.33638990e-01 1.13526261e+00 -2.22437695e-01 -1.25370085e+00 -3.69178504e-01 -3.79123002e-01 -4.82987911e-01 2.13319078e-01 -5.74888885e-01 -1.48309302e+00 1.33861959e+00 7.71606743e-01 1.18908659e-01 1.60134828e+00 -4.35098447e-02 6.63802028e-01 6.33662343e-02 1.31292298e-01 -4.12541062e-01 9.25562624e-03 3.91690403e-01 7.44647980e-01 -1.39566433e+00 -2.11665109e-01 -3.78180683e-01 -8.39635968e-01 6.03080630e-01 9.19879377e-01 -1.42833412e-01 8.13858628e-01 7.26741776e-02 2.79503375e-01 -4.34408456e-01 -8.15322638e-01 -6.55209720e-01 6.25760257e-01 4.78288472e-01 7.09361255e-01 -2.42882684e-01 -5.75835481e-02 4.13230956e-01 -3.81825715e-01 4.21194464e-01 3.03193450e-01 6.86951756e-01 -8.06244388e-02 -7.44370162e-01 -1.27512172e-01 4.73060578e-01 -6.29538774e-01 -3.90790313e-01 -4.30615753e-01 7.92511880e-01 -1.02630615e-01 8.95830393e-01 7.71899372e-02 -4.75702345e-01 4.40162629e-01 -7.82686248e-02 5.39077401e-01 -2.24851832e-01 -6.68451905e-01 4.97647412e-02 -1.69590399e-01 -8.24985147e-01 -8.03150356e-01 -5.17363846e-01 -7.23284006e-01 -6.75148249e-01 -6.05922639e-01 -1.05969720e-01 3.34762543e-01 8.98803115e-01 4.86020088e-01 6.77739441e-01 3.73779088e-01 -8.23341608e-01 -3.76155138e-01 -1.32139277e+00 -3.77069682e-01 8.50021541e-02 5.67261636e-01 -1.17927647e+00 -2.18333602e-01 9.58837047e-02]
[14.112004280090332, 0.06112677603960037]
a2ba5dc0-8cdf-4660-8f0a-64bf53969ef6
a-variational-inequality-perspective-on
1802.10551
null
https://arxiv.org/abs/1802.10551v5
https://arxiv.org/pdf/1802.10551v5.pdf
A Variational Inequality Perspective on Generative Adversarial Networks
Generative adversarial networks (GANs) form a generative modeling approach known for producing appealing samples, but they are notably difficult to train. One common way to tackle this issue has been to propose new formulations of the GAN objective. Yet, surprisingly few studies have looked at optimization methods designed for this adversarial training. In this work, we cast GAN optimization problems in the general variational inequality framework. Tapping into the mathematical programming literature, we counter some common misconceptions about the difficulties of saddle point optimization and propose to extend techniques designed for variational inequalities to the training of GANs. We apply averaging, extrapolation and a computationally cheaper variant that we call extrapolation from the past to the stochastic gradient method (SGD) and Adam.
['Simon Lacoste-Julien', 'Gaëtan Vignoud', 'Pascal Vincent', 'Gauthier Gidel', 'Hugo Berard']
2018-02-28
a-variational-inequality-perspective-on-1
https://openreview.net/forum?id=r1laEnA5Ym
https://openreview.net/pdf?id=r1laEnA5Ym
iclr-2019-5
['misconceptions']
['miscellaneous']
[ 1.69065148e-01 4.77021903e-01 1.97028130e-01 -2.09406719e-01 -8.25712562e-01 -5.70381045e-01 7.54007816e-01 -5.08188903e-01 -7.18405321e-02 1.18925321e+00 1.17082335e-01 -3.77289951e-01 -6.67128712e-02 -9.62443888e-01 -8.44840825e-01 -8.94039989e-01 5.52901685e-01 4.91882116e-01 -4.89440709e-01 -4.43395972e-01 1.12093799e-01 5.65403998e-01 -8.65063846e-01 -2.12203264e-01 1.12559283e+00 7.74863303e-01 -5.04329920e-01 6.77092552e-01 -1.06238805e-01 8.78054738e-01 -6.46711171e-01 -9.98220265e-01 3.72963339e-01 -9.85580623e-01 -7.24094808e-01 1.14763223e-01 2.41758212e-01 -2.99841285e-01 -1.64807752e-01 1.03817213e+00 4.53748077e-01 2.70837873e-01 8.16606164e-01 -1.49656916e+00 -9.12325680e-01 3.50390971e-01 -1.67372361e-01 -1.08496308e-01 2.74079800e-01 7.69030079e-02 8.10258627e-01 -6.85712636e-01 6.69944167e-01 1.01918089e+00 9.31285858e-01 1.03484988e+00 -1.36124408e+00 -2.20298246e-01 -1.10131991e-03 -2.22301841e-01 -1.00371492e+00 -3.66065353e-01 8.88958156e-01 -4.55130488e-01 5.28942466e-01 5.33405423e-01 6.54585540e-01 1.45262516e+00 4.73264232e-02 9.92782593e-01 1.32518148e+00 -4.28984106e-01 3.32379013e-01 3.49764854e-01 -4.37197089e-01 4.98707861e-01 1.09226676e-02 1.43926874e-01 -2.84737825e-01 -1.17351428e-01 9.19964135e-01 -8.98902491e-03 -2.10350037e-01 -3.06998312e-01 -6.40452623e-01 1.42523479e+00 3.58944982e-01 1.59558028e-01 -4.07141656e-01 3.33438903e-01 2.45643511e-01 3.74328136e-01 9.48941588e-01 6.02902770e-01 -1.06399968e-01 -1.03490688e-01 -1.10279477e+00 5.85654616e-01 9.79389906e-01 7.08773196e-01 5.93945563e-01 5.46420157e-01 -2.47353837e-01 6.11160815e-01 5.16876519e-01 2.19557270e-01 1.23472676e-01 -1.06504881e+00 3.13640118e-01 2.16013208e-01 1.26672775e-01 -7.67363191e-01 9.31621268e-02 -4.22280967e-01 -9.56700921e-01 5.62953413e-01 5.44314623e-01 -4.18799400e-01 -8.51765215e-01 1.73877835e+00 3.98425221e-01 3.87236997e-02 -1.48256505e-02 7.43837237e-01 3.33912373e-01 7.04885662e-01 -1.71909407e-01 -3.37609142e-01 5.21914482e-01 -1.01188242e+00 -6.30310059e-01 -2.21787784e-02 2.58145452e-01 -5.90788662e-01 9.19747233e-01 6.05946004e-01 -1.51646852e+00 -3.38224202e-01 -9.62218404e-01 -7.56592974e-02 -4.59657401e-01 -1.87805265e-01 7.84125745e-01 1.22281277e+00 -1.15502799e+00 1.07770419e+00 -9.83718872e-01 -2.29063332e-02 5.45968413e-01 2.26318464e-01 1.06878705e-01 7.94314072e-02 -1.00020432e+00 1.05442405e+00 4.66376245e-02 4.54639137e-01 -1.01341724e+00 -7.58427978e-01 -7.81368673e-01 -3.04613948e-01 3.48088026e-01 -1.19051325e+00 9.94414866e-01 -1.41861439e+00 -2.11950111e+00 7.97412694e-01 -1.91759557e-01 -7.04995155e-01 1.16004372e+00 -4.77952957e-01 8.76224693e-03 -3.50590020e-01 -2.99308509e-01 2.75381386e-01 1.39327359e+00 -1.30602074e+00 1.78565368e-01 -1.81070819e-01 1.77872136e-01 -1.46127835e-01 -1.07608419e-02 8.23055487e-03 1.83774769e-01 -9.94803071e-01 -3.37420762e-01 -9.45864499e-01 -5.08606017e-01 -1.55796483e-01 -5.18696666e-01 -1.11698225e-01 6.45113170e-01 -8.80384266e-01 1.03473842e+00 -1.80579090e+00 6.77203953e-01 2.34949782e-01 1.37521356e-01 4.33360010e-01 2.37894014e-01 6.82244778e-01 -4.90774354e-03 2.82997489e-01 -5.78111172e-01 -7.53942966e-01 4.41015869e-01 5.54092586e-01 -7.19656348e-01 4.82107520e-01 3.74520272e-01 1.32824969e+00 -9.43589509e-01 -3.34114939e-01 1.47273302e-01 7.03170478e-01 -9.04537022e-01 4.47381496e-01 -4.09745812e-01 8.71870100e-01 -4.74582762e-01 4.39309448e-01 6.65808439e-01 1.77756697e-01 -6.09301180e-02 1.17603794e-01 1.48792751e-02 2.35975116e-01 -9.82367218e-01 1.67367578e+00 -5.36386549e-01 6.28106534e-01 1.10029519e-01 -1.39138818e+00 7.15351820e-01 3.61091524e-01 4.22079384e-01 -3.60267460e-02 1.23043753e-01 2.85491884e-01 -3.99337262e-01 -3.45590800e-01 2.82715619e-01 -6.45574152e-01 1.86764091e-01 2.25768939e-01 1.97879165e-01 -6.92075670e-01 6.10336997e-02 2.48609688e-02 8.27011168e-01 7.06192434e-01 1.44002527e-01 -1.02011390e-01 5.60242057e-01 -1.21505186e-01 2.79647946e-01 7.10388541e-01 2.05822617e-01 8.37796986e-01 6.26090050e-01 -4.47300792e-01 -1.22604477e+00 -1.30545568e+00 6.90835491e-02 7.36358285e-01 -5.82657397e-01 -3.51101011e-01 -1.08709443e+00 -7.89726555e-01 -1.73751697e-01 9.42799568e-01 -8.22536588e-01 -1.49062872e-01 -7.00379670e-01 -7.21680641e-01 5.86663544e-01 5.14858544e-01 4.00034368e-01 -9.33273494e-01 -2.17825472e-01 3.00151169e-01 1.09656140e-01 -7.23454475e-01 -3.69652927e-01 7.90243819e-02 -9.99239147e-01 -5.84053159e-01 -1.16847217e+00 -2.44190574e-01 6.08741164e-01 -7.03310788e-01 1.42555606e+00 -9.30128098e-02 2.01273933e-02 5.41789889e-01 -2.76913494e-01 -6.65082693e-01 -9.14086699e-01 1.08677939e-01 -1.29954651e-01 1.77256867e-01 -6.71955422e-02 -8.55177522e-01 -4.02776152e-01 6.14692979e-02 -1.01501322e+00 -5.31141944e-02 4.16622937e-01 1.09644008e+00 6.24072433e-01 -3.39682966e-01 4.81723875e-01 -1.24967766e+00 8.44728351e-01 -4.19763088e-01 -5.50089777e-01 1.22216366e-01 -7.96398818e-01 1.88782558e-01 9.24887598e-01 -4.12650019e-01 -9.62019444e-01 -1.44776806e-01 -7.19833970e-01 -7.48598635e-01 2.34346017e-02 3.42679918e-01 -1.36831030e-01 -2.97166675e-01 5.14194369e-01 3.13385546e-01 4.80807908e-02 -2.92418748e-01 5.82361460e-01 6.07020147e-02 3.80418986e-01 -7.43907750e-01 1.16037560e+00 5.12072682e-01 1.66610762e-01 -7.14456797e-01 -9.95118916e-01 2.47354612e-01 -3.74426305e-01 -1.48783356e-01 1.04850197e+00 -3.76600266e-01 -5.01476645e-01 3.99684161e-01 -1.02971232e+00 -6.19580626e-01 -8.85881186e-01 2.53128447e-02 -1.02041829e+00 2.32521772e-01 -4.57047641e-01 -1.10714042e+00 -3.86873245e-01 -9.91917908e-01 9.44045186e-01 2.27770939e-01 -1.87115267e-01 -1.58202255e+00 3.61405551e-01 1.83789149e-01 7.48624742e-01 8.83837104e-01 5.63873112e-01 -3.40602785e-01 -3.65651518e-01 -1.81594029e-01 1.95318148e-01 9.13825154e-01 -1.26244590e-01 7.39092603e-02 -9.56291616e-01 -2.16798604e-01 6.77891016e-01 -2.85904199e-01 6.98993504e-01 5.27021646e-01 1.14797294e+00 -7.24847853e-01 3.38040181e-02 9.33945715e-01 1.58232105e+00 -3.25249061e-02 9.91758525e-01 1.47106290e-01 7.02503264e-01 2.96112537e-01 1.15683183e-01 3.69190127e-01 -6.81105349e-03 5.43579936e-01 4.58789170e-01 -7.90367872e-02 1.25005871e-01 -5.32083988e-01 4.72382426e-01 6.73243403e-01 -5.99166811e-01 -2.23966748e-01 -5.76789856e-01 3.67026031e-01 -1.65524948e+00 -1.01413691e+00 -9.35124233e-03 1.90398467e+00 7.54680574e-01 1.43082634e-01 2.34876245e-01 1.41924009e-01 2.35891134e-01 1.75876185e-01 -3.73841256e-01 -7.91268051e-01 -1.52364105e-01 7.50591636e-01 3.99709553e-01 5.14074504e-01 -8.80509257e-01 7.70621955e-01 7.15229940e+00 8.75493407e-01 -1.00979352e+00 2.71280229e-01 7.19205260e-01 1.31323397e-01 -6.30018413e-01 1.41116112e-01 -5.84933102e-01 5.02608657e-01 8.89544725e-01 -3.24491337e-02 8.61085653e-01 8.73396277e-01 9.47014168e-02 1.58674493e-01 -1.23188794e+00 7.26716697e-01 6.32752627e-02 -1.29796910e+00 2.91973446e-02 2.17063904e-01 1.25891387e+00 -3.57878208e-01 4.69853997e-01 3.55317414e-01 4.51388478e-01 -1.51607704e+00 7.23068774e-01 7.56340563e-01 4.43124831e-01 -6.88246965e-01 5.66601396e-01 3.39810640e-01 -6.62969410e-01 2.24772125e-01 -2.09495544e-01 -1.62010834e-01 4.92227197e-01 4.72192615e-01 -5.37138164e-01 6.49454117e-01 2.53618091e-01 4.94260490e-01 -2.22796962e-01 6.13385975e-01 -4.46842492e-01 8.27551305e-01 -3.38728696e-01 9.21121016e-02 4.56916302e-01 -9.47998881e-01 7.62634456e-01 8.53571832e-01 4.00820553e-01 -2.58803278e-01 -4.11390275e-01 1.41853178e+00 -1.21896923e-01 -1.09828524e-01 -7.16945946e-01 -2.50027061e-01 -3.35224062e-01 9.52512681e-01 -4.51605618e-01 -8.26639757e-02 -3.45031321e-01 1.26045763e+00 2.77984917e-01 5.45494556e-01 -1.03328884e+00 -1.10997044e-01 6.14938259e-01 7.94567913e-02 2.32391313e-01 -3.49288940e-01 -5.03919363e-01 -1.31783533e+00 5.69779687e-02 -1.02417731e+00 1.60604134e-01 -7.29581118e-01 -1.46588898e+00 4.62406456e-01 1.37881748e-02 -8.03094268e-01 -6.80242777e-01 -6.83460534e-01 -1.10670102e+00 9.96083260e-01 -1.18103313e+00 -1.30561829e+00 -5.96154518e-02 6.45119429e-01 5.81297278e-01 6.18881285e-02 7.80002177e-01 1.01627849e-01 -4.91052330e-01 6.72632039e-01 3.03759307e-01 -4.87052053e-02 2.46049285e-01 -1.58034706e+00 4.16055799e-01 1.04512417e+00 3.46893549e-01 5.40960491e-01 1.16247439e+00 -3.29189330e-01 -1.41032612e+00 -8.66933644e-01 7.23129690e-01 -6.54774725e-01 5.59068561e-01 -2.92392164e-01 -8.54010284e-01 9.59425032e-01 4.53912169e-01 -1.72216550e-01 3.90543967e-01 -2.06505939e-01 3.73774990e-02 2.33602002e-01 -1.33008647e+00 5.73493242e-01 8.84442091e-01 -4.66278076e-01 -4.36802179e-01 3.64350230e-01 3.83750707e-01 -6.15468740e-01 -9.41736460e-01 6.50977343e-02 3.17981362e-01 -1.04084849e+00 1.04435062e+00 -8.94307911e-01 7.38348067e-01 5.40328920e-02 -9.82556269e-02 -1.75736403e+00 2.48809665e-01 -1.39175975e+00 -3.95306647e-01 1.48742580e+00 2.48758450e-01 -9.53403592e-01 9.52051401e-01 6.89543247e-01 -2.27734387e-01 -1.05816877e+00 -1.03975284e+00 -8.19510162e-01 7.16799736e-01 -4.01231796e-01 4.22679007e-01 7.73852348e-01 -4.68655050e-01 -5.05154245e-02 -7.75237441e-01 -2.43627518e-01 6.80184662e-01 -5.82035556e-02 1.00522363e+00 -9.20625031e-01 -6.16425157e-01 -5.75343490e-01 -1.32248610e-01 -1.03687763e+00 2.78563917e-01 -8.14782619e-01 -1.31734237e-01 -1.31458259e+00 -3.70199472e-01 -2.82098770e-01 1.37681365e-01 -1.56082675e-01 -1.40683696e-01 2.34037295e-01 1.27107576e-01 -1.04887918e-01 -5.36259301e-02 7.25848019e-01 1.24375343e+00 -2.25063320e-02 -6.49933144e-02 4.22799379e-01 -6.21890724e-01 7.69823015e-01 8.30113173e-01 -4.52595055e-01 -2.58210123e-01 -4.21470225e-01 6.79160476e-01 -9.16981921e-02 5.43151557e-01 -8.14794362e-01 -3.56042013e-02 -2.96153665e-01 2.67021030e-01 -1.82904914e-01 4.09905940e-01 -5.75522065e-01 5.18047214e-01 2.60705113e-01 -2.53436625e-01 -4.93190810e-02 -5.36865853e-02 3.71814162e-01 -1.99194089e-01 -5.70127726e-01 8.77040744e-01 -3.08547914e-01 -8.75773281e-03 4.18721497e-01 -2.04028234e-01 5.86883664e-01 8.25827301e-01 -8.51281881e-02 2.44527563e-01 -6.13348842e-01 -9.63544250e-01 -8.76596794e-02 6.04701996e-01 -1.11612216e-01 5.84379137e-01 -1.33802676e+00 -7.62877882e-01 1.48493513e-01 -6.57895267e-01 8.49209204e-02 2.23162100e-02 1.08163416e+00 -6.76529467e-01 2.39460766e-01 -1.65602431e-01 -1.20591730e-01 -6.05384529e-01 7.15707600e-01 7.32818663e-01 -6.84008598e-01 -3.42699885e-01 1.04212213e+00 -3.45093897e-03 -2.69057453e-01 -9.89320725e-02 -7.25417212e-02 2.82635927e-01 -2.83755511e-02 1.28526598e-01 7.41246819e-01 -1.04490392e-01 -4.57433701e-01 1.30337365e-02 4.42712218e-01 4.03230965e-01 -2.13834256e-01 1.45796895e+00 4.92961779e-02 -1.31627232e-01 4.42858517e-01 1.21800065e+00 2.20114470e-01 -1.37371635e+00 2.26026535e-01 -3.64762753e-01 -3.12358230e-01 -1.78964853e-01 -4.05179143e-01 -1.13179445e+00 1.05351090e+00 1.99804202e-01 8.25888038e-01 1.03170300e+00 -1.65791467e-01 7.78478205e-01 -9.08205062e-02 1.56375363e-01 -9.60927606e-01 -1.57474801e-01 3.68633777e-01 1.06494319e+00 -1.07145858e+00 -1.29711792e-01 -3.11973840e-01 -5.61219871e-01 1.10594630e+00 2.12489918e-01 -7.57776022e-01 5.87023675e-01 1.14425018e-01 -1.25880808e-01 5.24201654e-02 -2.74176836e-01 7.54947262e-03 3.34479421e-01 7.88212597e-01 3.56495231e-01 -1.44572645e-01 -4.60008562e-01 4.11099225e-01 -4.01271284e-01 6.39991686e-02 2.60828227e-01 6.74572110e-01 3.65370899e-01 -1.33900714e+00 -2.15827987e-01 2.98634350e-01 -9.00729418e-01 -8.85096043e-02 -4.11146462e-01 6.92354381e-01 4.98398840e-02 6.42281711e-01 -3.79309118e-01 -2.82285996e-02 9.55361873e-02 4.13472325e-01 7.42666543e-01 -2.63837397e-01 -8.14063668e-01 -2.21572489e-01 -1.24969587e-01 -3.95699114e-01 -5.76372087e-01 -6.68653607e-01 -3.90308887e-01 -4.14961904e-01 -2.57230759e-01 3.17952067e-01 6.38656855e-01 1.13682473e+00 -1.11636795e-01 6.08189523e-01 5.68504333e-01 -1.05879927e+00 -1.01354873e+00 -8.65873992e-01 -4.71117079e-01 4.75306809e-01 2.47129366e-01 -4.34637427e-01 -5.20943820e-01 4.52044047e-02]
[11.606589317321777, -0.043271370232105255]
782bbfda-3ed4-4be2-929e-5b0186d28650
pose-aware-attention-network-for-flexible
2306.08006
null
https://arxiv.org/abs/2306.08006v1
https://arxiv.org/pdf/2306.08006v1.pdf
Pose-aware Attention Network for Flexible Motion Retargeting by Body Part
Motion retargeting is a fundamental problem in computer graphics and computer vision. Existing approaches usually have many strict requirements, such as the source-target skeletons needing to have the same number of joints or share the same topology. To tackle this problem, we note that skeletons with different structure may have some common body parts despite the differences in joint numbers. Following this observation, we propose a novel, flexible motion retargeting framework. The key idea of our method is to regard the body part as the basic retargeting unit rather than directly retargeting the whole body motion. To enhance the spatial modeling capability of the motion encoder, we introduce a pose-aware attention network (PAN) in the motion encoding phase. The PAN is pose-aware since it can dynamically predict the joint weights within each body part based on the input pose, and then construct a shared latent space for each body part by feature pooling. Extensive experiments show that our approach can generate better motion retargeting results both qualitatively and quantitatively than state-of-the-art methods. Moreover, we also show that our framework can generate reasonable results even for a more challenging retargeting scenario, like retargeting between bipedal and quadrupedal skeletons because of the body part retargeting strategy and PAN. Our code is publicly available.
['Shihong Xia', 'Boyuan Jiang', 'Chongyang Zhong', 'Zihao Zhang', 'Lei Hu']
2023-06-13
null
null
null
null
['motion-retargeting']
['computer-vision']
[-3.73051316e-02 1.37403443e-01 -4.01964217e-01 -1.98289119e-02 -2.39864811e-01 -4.31685925e-01 4.01414812e-01 -5.77474356e-01 -2.64300883e-01 6.06150091e-01 5.73645949e-01 1.44328430e-01 1.94906682e-01 -8.31282854e-01 -7.93610990e-01 -7.19232500e-01 3.47606391e-01 1.66435465e-01 7.19913304e-01 -4.36725438e-01 1.28567606e-01 3.72919410e-01 -1.34183395e+00 -5.25150448e-02 7.24486828e-01 4.96411622e-01 3.63633007e-01 4.80333745e-01 1.96216092e-01 2.99447775e-01 -2.36312717e-01 -2.09455043e-01 1.82546034e-01 -4.62115973e-01 -7.76274979e-01 8.89453515e-02 3.72714669e-01 -3.09376150e-01 -6.28199697e-01 8.55880439e-01 6.74753189e-01 4.55588371e-01 5.66196740e-01 -1.16048574e+00 -6.10839725e-01 7.52810478e-01 -9.29427922e-01 5.30375727e-02 2.96097487e-01 2.87112504e-01 1.03453362e+00 -7.21574485e-01 7.04406738e-01 1.42575824e+00 4.26774055e-01 8.40839684e-01 -1.17162561e+00 -7.09598005e-01 3.72902751e-01 1.50009125e-01 -1.47379827e+00 -2.61760294e-01 9.39621568e-01 -4.60334510e-01 5.26238501e-01 2.70693839e-01 7.89451599e-01 1.15876806e+00 6.39275789e-01 7.75849640e-01 2.15136409e-01 -1.33264288e-01 5.41564114e-02 -7.02677190e-01 -1.98883906e-01 8.45897019e-01 2.72041291e-01 -7.90367573e-02 -5.80986142e-01 -6.20815940e-02 1.32478118e+00 3.74263488e-02 -4.78885025e-01 -9.09087896e-01 -1.81144905e+00 7.71061003e-01 5.82385898e-01 1.97374851e-01 -1.48968324e-01 6.17083669e-01 3.41993481e-01 -1.97776780e-01 1.51547194e-01 3.27228248e-01 -3.53062928e-01 2.46880576e-03 -6.96509719e-01 4.82281655e-01 4.33521092e-01 7.69345224e-01 8.30253482e-01 5.43296561e-02 -6.01925313e-01 6.49430633e-01 5.36564589e-01 2.59605438e-01 8.97483349e-01 -1.09272146e+00 5.52415967e-01 4.38174874e-01 7.83868954e-02 -1.23825026e+00 -5.41800320e-01 -3.23995113e-01 -8.92984390e-01 4.43306789e-02 2.46829256e-01 -2.72971392e-01 -9.70498443e-01 2.16140580e+00 4.92240995e-01 2.79289093e-02 -2.49943584e-01 1.11045170e+00 6.41844332e-01 4.94401485e-01 8.44136924e-02 9.36300531e-02 1.37555194e+00 -1.45253873e+00 -5.71988106e-01 -2.96543241e-01 6.37754560e-01 -5.60579836e-01 1.11789048e+00 -2.48163015e-01 -1.01842117e+00 -6.81728899e-01 -1.02222157e+00 -2.07926959e-01 1.54531091e-01 1.47848994e-01 6.18779898e-01 4.28550988e-01 -8.39668334e-01 6.11524701e-01 -1.01955819e+00 -5.99806309e-01 -4.48926464e-02 4.41298097e-01 -4.61984605e-01 1.57127112e-01 -1.13918030e+00 6.09445691e-01 5.16691864e-01 7.30802938e-02 -5.62231958e-01 -4.33074623e-01 -1.13602149e+00 7.32916668e-02 2.96742529e-01 -1.38304806e+00 1.17399526e+00 -8.74947608e-01 -1.92756486e+00 5.14033854e-01 2.58237310e-02 -4.51262929e-02 4.94474858e-01 -4.11787629e-01 -1.31647199e-01 1.16313389e-03 2.84362495e-01 1.10496318e+00 9.03272748e-01 -8.43900502e-01 -5.93456805e-01 -2.35691115e-01 -4.35735807e-02 4.10645157e-01 -1.70979217e-01 -4.40043807e-01 -1.02600634e+00 -1.30631781e+00 3.33280146e-01 -1.23856378e+00 -4.71847862e-01 2.17831537e-01 -5.71528018e-01 -6.00174814e-03 7.80311465e-01 -4.74911958e-01 1.33887351e+00 -2.10986781e+00 9.35280204e-01 -5.95246851e-02 2.02176407e-01 2.02440396e-02 -1.47208676e-01 3.20949852e-01 -1.45254061e-01 -2.27439944e-02 -2.47262329e-01 -1.02225140e-01 -8.22725296e-02 3.14143091e-01 -9.74386409e-02 4.52621013e-01 -9.54785496e-02 1.22786844e+00 -9.24171865e-01 -5.23249328e-01 1.86538026e-01 3.09490830e-01 -7.89964080e-01 -9.37389862e-03 -1.59532815e-01 9.15194273e-01 -5.30062795e-01 4.54276681e-01 4.00272816e-01 -1.87675923e-01 1.18336201e-01 -5.93018949e-01 -3.75181660e-02 -1.30840361e-01 -1.26091731e+00 2.21213913e+00 -1.94097832e-01 2.20076203e-01 -1.57355040e-01 -6.41891181e-01 6.00303113e-01 1.30017564e-01 6.23183370e-01 -5.03842354e-01 1.23749310e-02 1.59162521e-01 1.23838402e-01 -3.50962102e-01 7.12979674e-01 -8.74551311e-02 -3.41553599e-01 4.14929360e-01 -9.51598063e-02 3.92055511e-02 -6.60805106e-02 -6.90712407e-02 8.71863961e-01 5.20957410e-01 4.01513487e-01 -2.52648145e-01 4.04416651e-01 -2.66431093e-01 9.43146527e-01 4.02788728e-01 -1.30146101e-01 8.96142781e-01 1.74516961e-01 -4.39591467e-01 -9.76006627e-01 -1.09230757e+00 2.43506953e-01 1.10796916e+00 5.44387341e-01 -4.24134016e-01 -6.15250289e-01 -6.12653673e-01 -1.81529015e-01 3.57900083e-01 -6.90489411e-01 -5.79033017e-01 -1.02989614e+00 -6.93463266e-01 4.45156068e-01 8.28292549e-01 7.11950600e-01 -8.24611306e-01 -9.15735304e-01 4.18678075e-01 -4.24168438e-01 -7.74387836e-01 -1.05344045e+00 -2.20014274e-01 -9.39358056e-01 -8.84830952e-01 -1.33910584e+00 -7.53093481e-01 5.34899771e-01 3.57426226e-01 6.87860787e-01 3.95031869e-02 -4.15329002e-02 2.66997337e-01 -3.20954889e-01 3.85693997e-01 -1.13512399e-02 4.06462550e-01 3.27622779e-02 -2.70595122e-02 -3.48824263e-01 -6.60484731e-01 -9.31624949e-01 6.74796283e-01 -1.05494308e+00 5.30204773e-01 5.42663336e-01 8.03425491e-01 5.23360372e-01 -1.57081664e-01 2.46350572e-01 -4.65872824e-01 2.50947297e-01 -4.13128912e-01 -1.70711204e-01 1.66164920e-01 -1.10959545e-01 3.24225694e-01 3.17575216e-01 -6.07434213e-01 -8.99638355e-01 4.68309104e-01 -2.46684477e-01 -3.90022248e-01 1.03877641e-01 2.75506407e-01 -6.09109402e-01 -6.94839433e-02 3.67580146e-01 1.51735097e-01 -2.29694352e-01 -5.70257783e-01 6.80900514e-01 5.87276183e-02 7.28789032e-01 -6.25974596e-01 8.42532933e-01 6.25849068e-01 1.70871317e-01 -6.08110905e-01 -3.11426312e-01 -1.85076699e-01 -8.49630952e-01 -1.49307176e-01 1.01321757e+00 -7.96389461e-01 -5.28253973e-01 6.25030398e-01 -1.17813110e+00 -3.92720044e-01 -2.91703373e-01 4.98965293e-01 -7.12786794e-01 8.16188335e-01 -6.03383720e-01 -9.90037099e-02 -2.18883246e-01 -1.36690927e+00 1.35859227e+00 2.46120840e-01 -4.38238591e-01 -8.96382689e-01 4.17044669e-01 -1.25442266e-01 1.48912326e-01 4.00390297e-01 1.01870942e+00 -1.41854137e-01 -3.66589785e-01 1.27945185e-01 5.12568019e-02 -2.54387468e-01 5.31076312e-01 3.30318101e-02 -4.71702635e-01 -4.32284325e-01 -4.54331875e-01 4.28062230e-02 1.06554306e+00 4.47913527e-01 9.56723452e-01 -3.59065443e-01 -7.65007198e-01 7.13052332e-01 1.02313435e+00 -7.55477836e-03 7.57210135e-01 3.61287683e-01 1.22758377e+00 6.12052023e-01 5.28927624e-01 3.39937389e-01 6.17818058e-01 1.32896948e+00 5.46192586e-01 1.42735280e-02 -3.77544552e-01 -4.97602224e-01 5.23949623e-01 8.78045559e-01 -7.80892521e-02 -3.06756228e-01 -7.07348704e-01 5.29619098e-01 -2.21099424e+00 -7.60605752e-01 -6.69295620e-03 2.28817081e+00 7.97003329e-01 -1.80033773e-01 2.26262704e-01 -1.17828296e-02 8.79254520e-01 2.29960248e-01 -5.89876056e-01 -7.26710111e-02 6.23360239e-02 -1.73097715e-01 3.85749876e-01 4.86030310e-01 -1.06889474e+00 1.19968271e+00 6.15945148e+00 8.78969789e-01 -1.33426976e+00 4.11380231e-02 2.05996916e-01 -1.65910646e-02 -4.34993207e-01 -6.27915934e-03 -7.61255205e-01 6.41902983e-01 3.08164120e-01 -8.79466012e-02 4.78904955e-02 5.89387834e-01 7.09772483e-02 -5.37812188e-02 -9.94136989e-01 9.64831829e-01 -9.97614395e-03 -1.18688095e+00 4.25227493e-01 2.57801916e-02 7.13444233e-01 -2.50213861e-01 6.98920153e-03 1.71841279e-01 2.27276236e-01 -8.14897478e-01 1.01850414e+00 6.48632050e-01 6.98347807e-01 -6.02589607e-01 3.69072169e-01 2.51028538e-01 -1.66533661e+00 1.37175202e-01 -3.76958966e-01 3.28494348e-02 4.36547369e-01 3.53718370e-01 -1.58850804e-01 8.19494128e-01 6.06037378e-01 8.20759237e-01 -5.69455981e-01 1.21963561e+00 -3.01873714e-01 1.76300555e-01 -2.36453235e-01 3.40301812e-01 -5.78547642e-02 3.05213369e-02 7.65156269e-01 7.64919341e-01 4.85491216e-01 -1.48362011e-01 1.72115877e-01 7.03321397e-01 1.30326018e-01 1.44290969e-01 -5.47526360e-01 2.86937118e-01 2.16470823e-01 1.01942503e+00 -8.16538095e-01 -1.32642224e-01 -3.26609910e-01 1.54689610e+00 2.16942698e-01 4.89782393e-01 -1.05136597e+00 -2.89246351e-01 7.81243980e-01 9.42750201e-02 4.99824494e-01 -4.93464142e-01 -6.81218430e-02 -1.44910514e+00 -1.00889564e-01 -4.73711580e-01 2.80168951e-01 -6.82352841e-01 -9.62169528e-01 4.51146960e-01 8.52719694e-02 -1.35191107e+00 -2.50073731e-01 -3.69367748e-01 -5.90009153e-01 5.46872854e-01 -8.15345287e-01 -1.44807589e+00 -3.16504747e-01 6.09363973e-01 6.90355957e-01 1.69725016e-01 4.86994356e-01 3.00505280e-01 -7.43821084e-01 6.14430606e-01 -2.15645060e-01 -1.42741809e-02 8.32652807e-01 -8.43956470e-01 5.57249188e-01 8.97141039e-01 7.50593022e-02 6.77236140e-01 6.77580655e-01 -9.20718133e-01 -1.24674439e+00 -1.29988158e+00 5.20115733e-01 -3.35000783e-01 3.60056609e-01 -1.04451701e-01 -7.11785376e-01 8.44185472e-01 -1.85490679e-02 -9.64871943e-02 4.84772295e-01 -2.54337162e-01 -9.37017128e-02 1.94576100e-01 -6.96474433e-01 1.26249909e+00 1.36919188e+00 3.46166529e-02 -6.84313416e-01 -7.28325844e-02 1.02984834e+00 -7.25061476e-01 -5.45905054e-01 6.67780101e-01 8.35798383e-01 -6.94195390e-01 1.02414656e+00 -5.39047062e-01 3.44242185e-01 -6.83051705e-01 -2.51027435e-01 -1.23408628e+00 -7.88455069e-01 -5.68990469e-01 -2.72315323e-01 1.10310030e+00 1.99431144e-02 -3.25631976e-01 8.88204336e-01 4.26462263e-01 -1.47515461e-01 -5.98265588e-01 -1.15050948e+00 -8.05150867e-01 7.88145065e-02 -2.19214067e-01 6.78934693e-01 7.47044981e-01 -1.76168501e-01 5.69958568e-01 -7.67012179e-01 1.21797428e-01 2.89391160e-01 2.61300385e-01 9.93645787e-01 -9.67163086e-01 -4.41160560e-01 -4.16001797e-01 -6.06643796e-01 -1.49157572e+00 1.48515522e-01 -7.40850031e-01 8.74495730e-02 -1.57967913e+00 2.47185200e-01 -1.25391796e-01 -6.85460446e-03 8.13807845e-01 -2.92496771e-01 4.61657166e-01 1.95690647e-01 4.57490206e-01 -4.10965681e-01 1.09794950e+00 1.68901980e+00 -9.65882391e-02 -3.88635963e-01 -2.79133162e-03 -6.39350057e-01 8.43761563e-01 6.22418463e-01 -3.59332800e-01 -3.83522332e-01 -6.41207874e-01 2.06647813e-01 3.80682051e-02 6.12182796e-01 -1.20887244e+00 2.23634347e-01 -3.72321308e-01 2.60937333e-01 -5.01137078e-01 3.40828210e-01 -6.84876800e-01 4.61381465e-01 7.55457282e-01 1.59654990e-02 1.66316465e-01 1.63595125e-01 7.44474828e-01 3.36187929e-02 7.30374530e-02 7.70051777e-01 3.27363126e-02 -8.80591571e-01 4.59609956e-01 -5.67870617e-01 -1.82363182e-01 1.11265814e+00 -3.97964776e-01 -3.73269618e-02 -2.88227379e-01 -7.77873397e-01 8.91097263e-02 8.55348110e-01 7.61877775e-01 4.26886052e-01 -1.72459936e+00 -4.65088516e-01 1.51456790e-02 1.68371111e-01 7.09863827e-02 4.22181934e-01 8.43265235e-01 -5.75300455e-01 3.86528939e-01 -5.49415529e-01 -6.37134194e-01 -9.09180343e-01 5.70325732e-01 3.43386382e-01 -1.19470619e-01 -8.21022689e-01 7.59645820e-01 9.45863068e-01 -4.13563699e-01 -1.06877096e-01 -5.02680719e-01 -2.38881767e-01 -2.39667892e-01 2.12949112e-01 4.63133156e-01 -3.46912265e-01 -9.88050580e-01 -4.12290692e-01 1.17245913e+00 1.57706186e-01 -2.82245904e-01 8.53207886e-01 -3.91815156e-01 6.21223375e-02 3.34615737e-01 8.40604246e-01 2.10991472e-01 -1.47692358e+00 -5.60914204e-02 -3.41638595e-01 -5.15359700e-01 -3.12273532e-01 -3.45143646e-01 -1.35495985e+00 6.64934754e-01 5.20977437e-01 -4.84911054e-01 1.00414360e+00 6.57003447e-02 9.33467686e-01 1.37991115e-01 5.69713891e-01 -7.97224164e-01 2.95390695e-01 4.44772482e-01 1.07658935e+00 -7.16039479e-01 -5.85408956e-02 -5.52623391e-01 -6.08349621e-01 9.79876578e-01 9.04354274e-01 -4.43240255e-02 4.92271215e-01 -2.20860794e-01 -1.71471268e-01 -8.73517767e-02 -3.68297309e-01 2.45867316e-02 4.60889578e-01 5.74623644e-01 4.11576867e-01 -2.59536915e-02 -6.04846358e-01 7.99761713e-01 -1.26441315e-01 -1.00629687e-01 2.80113757e-01 8.85057688e-01 -4.22012478e-01 -1.38959205e+00 -3.57090175e-01 -1.07799336e-01 -2.57280320e-01 1.44159928e-01 -2.43287325e-01 7.93765008e-01 1.42522126e-01 5.39289594e-01 -1.48554221e-01 -6.02545142e-01 2.99025208e-01 -2.52319425e-01 5.24738610e-01 -8.73182118e-01 -2.44122609e-01 1.72649056e-01 -2.66381353e-01 -9.02539194e-01 -3.73453796e-01 -4.20188367e-01 -1.33216572e+00 -1.80515900e-01 -2.31482938e-01 -8.48698393e-02 7.28972778e-02 7.94377863e-01 4.78360087e-01 7.03137815e-01 2.74370730e-01 -1.26249516e+00 -1.68417662e-01 -7.83280373e-01 -5.70938468e-01 2.34222189e-01 1.34817958e-01 -1.08418739e+00 1.91248044e-01 1.77111343e-01]
[7.454733848571777, -0.4104195535182953]
cf73176f-5389-4da5-bf7d-2c84b567105c
towards-computational-architecture-of-liberty
2305.00510
null
https://arxiv.org/abs/2305.00510v2
https://arxiv.org/pdf/2305.00510v2.pdf
Towards Computational Architecture of Liberty: A Comprehensive Survey on Deep Learning for Generating Virtual Architecture in the Metaverse
3D shape generation techniques utilizing deep learning are increasing attention from both computer vision and architectural design. This survey focuses on investigating and comparing the current latest approaches to 3D object generation with deep generative models (DGMs), including Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), 3D-aware images, and diffusion models. We discuss 187 articles (80.7% of articles published between 2018-2022) to review the field of generated possibilities of architecture in virtual environments, limited to the architecture form. We provide an overview of architectural research, virtual environment, and related technical approaches, followed by a review of recent trends in discrete voxel generation, 3D models generated from 2D images, and conditional parameters. We highlight under-explored issues in 3D generation and parameterized control that is worth further investigation. Moreover, we speculate that four research agendas including data limitation, editability, evaluation metrics, and human-computer interaction are important enablers of ubiquitous interaction with immersive systems in architecture for computer-aided design Our work contributes to researchers' understanding of the current potential and future needs of deep learnings in generating virtual architecture.
['Pan Hui', 'Lik-Hang Lee', 'Jiachuan Shen', 'Jiahua Dong', 'Anqi Wang']
2023-04-30
null
null
null
null
['3d-shape-generation']
['computer-vision']
[-1.17665000e-01 5.97622871e-01 4.53315854e-01 2.06469506e-01 -4.89144325e-01 -4.88020182e-01 5.59439898e-01 -8.62285852e-01 3.77759010e-01 8.82215321e-01 4.86119598e-01 -3.64596844e-01 1.69510826e-01 -1.32818472e+00 -8.10922086e-01 -5.89043379e-01 -3.25606577e-02 4.79184061e-01 -3.36526275e-01 -6.84022129e-01 2.38195551e-03 8.60335231e-01 -1.55702174e+00 1.22451857e-01 7.48934865e-01 1.11079395e+00 2.86322087e-02 8.26105893e-01 -1.07763506e-01 6.20233119e-01 -1.05784094e+00 -5.92739880e-01 1.72345445e-01 -5.00822902e-01 -3.10730547e-01 -1.31666916e-03 3.88235003e-01 -5.91183960e-01 -4.38613802e-01 3.73677820e-01 1.27184212e+00 -2.04856709e-01 1.20774698e+00 -1.38019288e+00 -1.36477649e+00 3.94073606e-01 -1.50652705e-02 -2.69755989e-01 3.28378320e-01 5.91578126e-01 3.82747531e-01 -1.08707368e+00 1.01848876e+00 1.09536195e+00 7.59069324e-01 1.10745513e+00 -1.04612064e+00 -5.17165184e-01 -2.42249519e-01 -2.40689591e-01 -1.38141882e+00 -2.02119201e-01 1.25573909e+00 -8.58223975e-01 1.13846540e+00 3.73298794e-01 1.20527661e+00 1.68650699e+00 8.17360044e-01 7.84881830e-01 6.55108869e-01 -3.79652411e-01 5.04923999e-01 2.01879248e-01 -5.71131825e-01 5.03197074e-01 1.85585946e-01 5.84120989e-01 -3.48865479e-01 -3.34463775e-01 1.68237352e+00 -7.10913658e-01 1.17917977e-01 -7.96142459e-01 -9.49374437e-01 1.02858734e+00 5.53667903e-01 3.05874180e-02 -7.23958611e-01 6.93469942e-01 2.69020408e-01 1.36928648e-01 5.16090453e-01 6.84389770e-01 -9.03851762e-02 -9.12184268e-03 -5.56530058e-01 7.56107390e-01 6.95519924e-01 1.46337712e+00 1.55442253e-01 1.28491831e+00 -2.06806958e-01 6.83574021e-01 5.24121344e-01 6.29579961e-01 2.66201675e-01 -1.02370238e+00 5.21000177e-02 2.73090955e-02 5.47122546e-02 -6.75828815e-01 -1.70471132e-01 -3.19798589e-01 -1.07185364e+00 9.63609159e-01 -5.77276886e-01 -7.60338068e-01 -1.05361927e+00 1.29433894e+00 4.03179318e-01 -6.83153197e-02 -6.72062486e-02 7.39552915e-01 1.44471037e+00 4.05688196e-01 -6.88845068e-02 1.85809240e-01 9.69768167e-01 -4.22757357e-01 -7.39858806e-01 2.78513998e-01 -1.78081729e-02 -8.58284295e-01 7.35746861e-01 9.88445133e-02 -1.48494148e+00 -7.76237965e-01 -1.46861577e+00 4.29742783e-02 -2.12087378e-01 -2.34053388e-01 7.42123246e-01 1.26510847e+00 -1.47481227e+00 3.74024808e-01 -9.34962571e-01 3.60433534e-02 6.98347747e-01 3.63061935e-01 2.28628278e-01 3.30209285e-01 -1.16097677e+00 9.10791337e-01 -2.87974477e-01 -1.50662258e-01 -1.31773365e+00 -1.10669482e+00 -9.29951787e-01 -4.49264258e-01 -1.85168907e-01 -1.78482521e+00 1.33550155e+00 -3.00057352e-01 -2.04913139e+00 6.76237047e-01 4.25294846e-01 -4.28796023e-01 6.71878278e-01 -2.50087023e-01 -1.83535680e-01 -4.97633755e-01 -2.17977434e-01 8.94288778e-01 7.02181458e-01 -1.73893499e+00 -7.04846485e-03 -1.93418667e-01 7.48793334e-02 1.40872359e-01 6.84986338e-02 -7.16365576e-01 1.79863021e-01 -8.82362187e-01 -9.35684964e-02 -8.88195455e-01 -5.70855260e-01 2.16499358e-01 -7.36918449e-01 8.07359535e-03 9.41725433e-01 -6.16595924e-01 8.39291036e-01 -1.79698491e+00 2.50980079e-01 -3.74409510e-03 4.03998971e-01 -1.42500475e-01 8.33522305e-02 5.36835372e-01 1.62543163e-01 4.44652140e-01 2.24743951e-02 -2.87005246e-01 1.99829549e-01 -5.30459844e-02 -2.55565524e-01 -1.84296407e-02 1.55565560e-01 1.51438272e+00 -4.86315370e-01 -1.60997495e-01 7.75222003e-01 9.54107404e-01 -8.52633715e-01 2.53516257e-01 -3.69420379e-01 6.42583191e-01 -6.40947759e-01 8.30425739e-01 7.38040209e-01 1.89833343e-01 -1.30685106e-01 -3.45631868e-01 -1.78280361e-02 -1.52408341e-02 -1.03777778e+00 1.68020344e+00 -8.73116016e-01 5.87294102e-01 1.54801272e-02 -1.60672918e-01 1.25002265e+00 3.82394820e-01 6.79942727e-01 -5.82664549e-01 2.70997971e-01 2.04821661e-01 -2.52837479e-01 -3.78487498e-01 7.90820241e-01 -1.69962972e-01 -2.96420932e-01 2.89214581e-01 7.19936416e-02 -1.14175069e+00 -6.40466869e-01 8.48377421e-02 9.58230138e-01 5.09740055e-01 -7.65981525e-02 -1.67743772e-01 -1.27444685e-01 -1.49746254e-01 1.93444360e-02 5.58899283e-01 1.43522009e-01 9.92971122e-01 1.46384776e-01 -4.68646407e-01 -1.74559188e+00 -1.48655081e+00 4.17275652e-02 2.29040906e-01 5.08793779e-02 -2.23971918e-01 -7.66125023e-01 -2.45417610e-01 1.55761331e-01 1.08853412e+00 -9.65927780e-01 -2.75731474e-01 -6.81911290e-01 -4.41147238e-01 4.04396921e-01 8.65496814e-01 2.14123696e-01 -1.22837400e+00 -7.16626048e-01 3.25562984e-01 3.09838533e-01 -6.70083225e-01 -3.84400226e-02 -3.25515300e-01 -1.09538388e+00 -5.07621706e-01 -1.16673219e+00 -7.81489015e-01 3.94373447e-01 -1.55775517e-01 1.58097792e+00 -2.81488538e-01 -4.69426155e-01 9.56233442e-01 -2.00402364e-01 -6.68813169e-01 -9.30062354e-01 1.43964738e-01 1.74111634e-01 -8.36535335e-01 -2.27061838e-01 -9.78323340e-01 -9.32774007e-01 1.88338995e-01 -6.68908238e-01 2.09931701e-01 5.03953755e-01 7.94949293e-01 7.55053580e-01 -3.72849554e-01 4.36426312e-01 -4.95784253e-01 9.55924869e-01 -3.48038197e-01 -2.96374947e-01 -8.76793712e-02 -3.73079956e-01 -5.85121755e-03 1.37523606e-01 -3.17794621e-01 -1.24690807e+00 -1.81772307e-01 -6.81083083e-01 -4.82855260e-01 -4.97974604e-02 -9.82126519e-02 -5.24179518e-01 -1.27077967e-01 1.00233746e+00 1.88897531e-02 2.39004627e-01 -1.42430753e-01 7.43512511e-01 3.22522372e-01 1.87035933e-01 -5.76178551e-01 1.01762128e+00 4.08004105e-01 -1.25667304e-01 -9.32441294e-01 3.27629894e-01 8.14929962e-01 -4.84078348e-01 -4.92306471e-01 8.83315027e-01 -7.14986145e-01 -4.82458383e-01 5.48862100e-01 -1.30724645e+00 -5.46376526e-01 -9.39269125e-01 9.37252715e-02 -1.07657290e+00 -3.43463808e-01 -5.69186330e-01 -8.76232564e-01 -8.31030548e-01 -1.53988278e+00 1.29252720e+00 1.86498940e-01 -5.82382143e-01 -9.37419116e-01 1.14264809e-01 1.62777185e-01 9.26953912e-01 1.07094407e+00 1.16244054e+00 2.75953472e-01 -8.75113428e-01 -1.56239435e-01 4.95017469e-01 3.25477690e-01 1.09812759e-01 1.94515482e-01 -9.09082830e-01 2.26759128e-02 -1.07535504e-01 -2.72886399e-02 1.20266810e-01 1.30269003e+00 9.86751258e-01 -1.36454314e-01 -5.35425603e-01 5.67514181e-01 1.30026543e+00 5.49424946e-01 1.15687501e+00 1.30375803e-01 9.26153779e-01 2.23627850e-01 1.30474731e-01 7.30529904e-01 1.05483174e-01 5.72811782e-01 7.71079123e-01 -1.14794269e-01 -7.91049302e-01 -4.22550082e-01 6.93334714e-02 8.85076880e-01 -6.14627123e-01 -3.60832453e-01 -4.34955746e-01 3.06980729e-01 -1.05206215e+00 -6.95130825e-01 -9.04221237e-02 1.81277299e+00 2.40642101e-01 6.84571415e-02 -1.57358479e-02 -2.68979788e-01 6.69291675e-01 6.86236247e-02 -9.71655428e-01 -5.69111109e-01 -8.45436677e-02 7.07923830e-01 2.51486629e-01 2.33435169e-01 -5.92064083e-01 7.21403122e-01 7.69455147e+00 7.41476953e-01 -9.26642120e-01 8.14238861e-02 9.22872543e-01 -3.42915982e-01 -1.22164023e+00 -5.30640543e-01 -6.08557701e-01 2.16138721e-01 7.21734405e-01 -3.42178106e-01 1.13372959e-01 1.42193007e+00 1.19826533e-01 4.63952571e-01 -9.27916348e-01 1.00259519e+00 2.13156641e-01 -2.04840922e+00 2.54659206e-01 5.25587976e-01 1.21010745e+00 -2.42114633e-01 7.46640980e-01 3.83147150e-01 5.85287690e-01 -1.16944110e+00 1.17668724e+00 5.83347857e-01 1.34115946e+00 -9.88932490e-01 4.98856425e-01 -3.06122273e-01 -1.09187877e+00 2.88266599e-01 -2.54381865e-01 3.81394595e-01 6.93472028e-01 3.70540380e-01 -8.42824996e-01 4.32705790e-01 5.55530608e-01 1.34197354e-01 -7.38161802e-02 7.61171758e-01 -8.34798887e-02 8.05977955e-02 -1.65773369e-02 -4.19755876e-01 -2.04618156e-01 1.75369099e-01 9.83223855e-01 3.86910021e-01 8.55698109e-01 1.21812150e-01 -4.31742102e-01 1.53945243e+00 -4.52259518e-02 -2.31291264e-01 -8.51391375e-01 1.07095137e-01 5.81006527e-01 7.82148659e-01 -3.78543526e-01 3.14516388e-02 1.07944407e-01 1.04448760e+00 -3.69959265e-01 3.32671076e-01 -1.07027245e+00 -2.36498713e-01 9.90248919e-01 6.55635774e-01 2.53983825e-01 -6.14830077e-01 -8.33816707e-01 -4.95138645e-01 -1.43006444e-01 -6.52677536e-01 -2.63384253e-01 -1.33176339e+00 -1.32451677e+00 9.74945009e-01 -7.06783123e-03 -1.52441871e+00 -6.35525703e-01 -7.13807106e-01 -4.87070322e-01 8.35836649e-01 -5.22956550e-01 -1.47839129e+00 -5.14093816e-01 1.24636345e-01 8.76425982e-01 -5.22642672e-01 9.04088378e-01 6.25338480e-02 4.21154313e-02 6.58156872e-01 7.45781139e-02 -2.89107025e-01 1.29165024e-01 -9.47242498e-01 1.30256367e+00 1.81203008e-01 -2.55018800e-01 2.20773414e-01 7.68817544e-01 -9.22539711e-01 -1.78708220e+00 -8.20387423e-01 -2.01838821e-01 -8.23238969e-01 8.34091455e-02 -3.67677927e-01 -3.49219531e-01 4.69623774e-01 5.33928990e-01 -5.86747050e-01 7.65817404e-01 -2.39242613e-01 4.19124424e-01 3.99528921e-01 -1.34981823e+00 8.87973726e-01 1.46892071e+00 -2.16948286e-01 4.32927813e-03 -8.29038695e-02 1.12418926e+00 -1.06058240e+00 -1.08953357e+00 5.12378216e-01 7.78182864e-01 -8.74532163e-01 1.25540113e+00 -3.10310125e-01 9.29701269e-01 9.58780795e-02 -2.28440717e-01 -1.56545651e+00 -5.39504886e-01 -7.96918273e-01 -4.39340144e-01 9.20186102e-01 3.70002329e-01 -2.62602717e-01 1.20218790e+00 9.77770448e-01 -8.22735727e-01 -9.16491389e-01 -8.94826472e-01 -4.91425812e-01 5.28007448e-01 -5.30115366e-01 1.15518200e+00 6.11999273e-01 -8.14452887e-01 1.31997585e-01 -4.67442632e-01 -2.13706106e-01 4.68098313e-01 -2.41314724e-01 1.09490716e+00 -9.78312671e-01 -1.43775135e-01 -5.11608124e-01 -7.43658900e-01 -8.99091482e-01 -3.48546863e-01 -6.14181042e-01 -2.24419802e-01 -2.08446360e+00 -3.84743005e-01 -4.47692543e-01 4.06116843e-01 -3.05347174e-01 3.42983723e-01 2.69575536e-01 -8.71858075e-02 -1.90550819e-01 4.07206535e-01 1.18724465e+00 1.82797921e+00 -4.27510172e-01 -6.30557537e-01 5.42057455e-02 -7.78819382e-01 2.96908557e-01 7.01420546e-01 2.01279819e-01 -5.93541324e-01 -5.94626248e-01 1.65118203e-02 5.40128686e-02 4.99075979e-01 -1.02255952e+00 -1.97330073e-01 -1.60205166e-03 8.32883418e-01 -8.80977571e-01 5.78552306e-01 -6.04796410e-01 8.35403740e-01 5.07038534e-01 1.01025291e-01 2.01283798e-01 5.73541164e-01 1.83206096e-01 3.07610452e-01 7.89488405e-02 4.89197671e-01 -3.81535679e-01 -7.29787588e-01 5.25830984e-01 -7.16086745e-01 -2.76717216e-01 1.14328861e+00 -7.14463890e-01 3.16900834e-02 -6.63473248e-01 -1.06542504e+00 -3.30299467e-01 4.66742992e-01 6.42469943e-01 1.12451923e+00 -2.15712571e+00 -9.47113931e-01 1.78234413e-01 -2.46200502e-01 1.74276277e-01 9.75170791e-01 -1.41048834e-01 -1.00583005e+00 1.96490705e-01 -4.82026368e-01 -5.43443024e-01 -7.80883133e-01 3.90254438e-01 5.16195416e-01 9.40015391e-02 -7.05969453e-01 1.12126267e+00 2.47904584e-01 -4.06047881e-01 -1.82740241e-01 -1.23170279e-02 1.12350203e-01 -3.53532612e-01 1.57518368e-02 4.27685291e-01 2.04181686e-01 -3.66797477e-01 -9.18855146e-02 3.77870888e-01 1.37754679e-01 -3.75793934e-01 1.34715521e+00 1.69841751e-01 4.48860079e-01 2.29764789e-01 8.79852653e-01 -2.53978968e-01 -1.29532886e+00 5.26658416e-01 -9.96170104e-01 -1.21353582e-01 8.93104151e-02 -6.72464609e-01 -1.52600741e+00 7.12296486e-01 1.14579999e+00 1.68332025e-01 7.79401243e-01 2.83983797e-01 1.02251923e+00 -3.90715837e-01 7.45245457e-01 -9.04681385e-01 5.89409471e-01 2.71042556e-01 1.69744623e+00 -5.37411988e-01 -5.70310242e-02 -4.74006683e-01 -7.09438801e-01 8.22504640e-01 1.03750825e+00 -4.22115743e-01 1.04793298e+00 8.00356209e-01 -5.18396087e-02 -3.77975762e-01 -3.30620736e-01 3.66956621e-01 3.05089831e-01 1.38240790e+00 5.69564342e-01 2.18446761e-01 1.60151318e-01 5.91152251e-01 -1.05027282e+00 -3.63468796e-01 3.23844284e-01 9.76991415e-01 1.57530800e-01 -1.29380679e+00 -4.91830945e-01 5.27934432e-01 1.39657021e-01 -1.35870993e-01 -3.09655994e-01 8.36865664e-01 7.85052851e-02 4.86504912e-01 2.95209110e-01 -8.59650493e-01 6.07988596e-01 -2.82530636e-01 7.90364861e-01 -3.44906181e-01 -3.71514976e-01 -3.60736996e-02 9.55990180e-02 -3.03371072e-01 3.33479680e-02 -4.81117398e-01 -6.26636207e-01 -5.60296237e-01 -3.52846295e-01 -4.58486050e-01 1.04566431e+00 2.31937036e-01 8.76647592e-01 1.29040587e+00 5.18563867e-01 -1.38802958e+00 -7.99643099e-02 -8.24896693e-01 -4.75536615e-01 -7.40393698e-02 -1.81518644e-01 -8.99623692e-01 -6.91664666e-02 1.05856232e-01]
[5.834673881530762, 3.234663724899292]
40ac1c25-404e-4480-a8ad-f8661bf8c619
dual-gan-joint-bvp-and-noise-modeling-for
null
null
http://openaccess.thecvf.com//content/CVPR2021/html/Lu_Dual-GAN_Joint_BVP_and_Noise_Modeling_for_Remote_Physiological_Measurement_CVPR_2021_paper.html
http://openaccess.thecvf.com//content/CVPR2021/papers/Lu_Dual-GAN_Joint_BVP_and_Noise_Modeling_for_Remote_Physiological_Measurement_CVPR_2021_paper.pdf
Dual-GAN: Joint BVP and Noise Modeling for Remote Physiological Measurement
Remote photoplethysmography (rPPG) based physiological measurement has great application values in health monitoring, emotion analysis, etc. Existing methods mainly focus on how to enhance or extract the very weak blood volume pulse (BVP) signals from face videos, but seldom explicitly model the noises that dominate face video content. Thus, they may suffer from poor generalization ability in unseen scenarios. This paper proposes a novel adversarial learning approach for rPPG based physiological measurement by using Dual Generative Adversarial Networks (Dual-GAN) to model the BVP estimation and noise distribution jointly. The BVP-GAN aims to learn a noise-resistant mapping from input to ground-truth BVP, and the Noise-GAN aims to learn the noise distribution. The dual GANs can promote each other's capability, leading to improved feature disentanglement between BVP and noises. Besides, a plug-and-play block named ROI alignment and fusion (ROI-AF) block is proposed to alleviate the inconsistencies between different ROIs and exploit informative features from a wider receptive field in terms of ROIs. In comparison to state-of-the-art methods, our method achieves better performance in heart rate, heart rate variability, and respiration frequency estimation from face videos.
['S. Kevin Zhou', 'Hu Han', 'Hao Lu']
2021-06-19
null
null
null
cvpr-2021-1
['heart-rate-variability']
['medical']
[-1.11821508e-02 -5.68603575e-02 2.29394883e-01 -4.03933436e-01 -6.50949717e-01 -1.01238512e-01 1.53821886e-01 -6.20835900e-01 -1.02079287e-01 9.02353406e-01 3.85313272e-01 2.80113965e-01 1.61425859e-01 -6.16308630e-01 -2.92204589e-01 -1.39335990e+00 2.99863011e-01 -2.75047421e-01 -3.79539877e-01 -5.51528633e-02 -1.61645159e-01 4.92105007e-01 -1.05699706e+00 1.53879851e-01 8.10716867e-01 1.14159405e+00 -1.77468523e-01 5.11287034e-01 -1.23209292e-02 9.51053679e-01 -8.46064508e-01 -2.01078534e-01 2.48629555e-01 -9.63260353e-01 -1.16518915e-01 -3.77427906e-01 1.47370383e-01 -4.68150854e-01 -8.19833457e-01 1.03250611e+00 1.21185017e+00 8.86205211e-03 5.04450619e-01 -1.35448205e+00 -5.85799634e-01 3.25348854e-01 -8.30980241e-01 4.84867573e-01 8.46846178e-02 5.07784009e-01 1.39444858e-01 -6.08057201e-01 1.77653015e-01 1.00665319e+00 7.53670692e-01 9.33805406e-01 -1.27529001e+00 -1.10133457e+00 -3.51523608e-01 3.03636461e-01 -1.43059647e+00 -6.73191130e-01 1.27374732e+00 -2.02174768e-01 3.67734164e-01 4.36113805e-01 8.08988750e-01 1.30563772e+00 3.59350145e-01 4.32377815e-01 1.34462428e+00 8.70649740e-02 -1.22881100e-01 1.15440696e-01 -5.05501330e-01 4.05007631e-01 -1.04906768e-01 1.53278545e-01 -6.23580158e-01 -8.68439674e-02 1.01191509e+00 -5.13757020e-02 -8.70437443e-01 9.95516106e-02 -9.61970747e-01 6.21047616e-01 2.87938237e-01 4.11955923e-01 -6.40556574e-01 3.17265272e-01 5.36363244e-01 1.61552176e-01 4.11127746e-01 2.06050858e-01 -2.18416169e-01 -1.45450756e-01 -8.78366947e-01 -7.39230439e-02 6.85753107e-01 4.45547521e-01 4.87453848e-01 5.33827603e-01 -6.24409318e-01 8.71490955e-01 3.28625798e-01 8.05379510e-01 7.66437650e-01 -1.09846842e+00 7.41813779e-02 1.72817752e-01 -1.74485698e-01 -1.23374498e+00 -3.72165412e-01 -3.33946139e-01 -1.11772132e+00 -5.07448465e-02 1.96110010e-01 -3.76229823e-01 -7.76566207e-01 1.88117266e+00 5.66893578e-01 7.40412056e-01 1.17530972e-01 1.13280332e+00 1.32653379e+00 6.40634775e-01 2.10923225e-01 -6.28487289e-01 1.17305088e+00 -4.75401014e-01 -1.03055775e+00 -2.15164483e-01 9.27845165e-02 -4.81580079e-01 7.41022766e-01 2.78771371e-01 -1.02811527e+00 -7.40075469e-01 -8.12763155e-01 8.46100897e-02 8.89660642e-02 -2.72542443e-02 3.07253867e-01 8.46590281e-01 -8.74635339e-01 4.98473942e-01 -7.56989837e-01 2.28085577e-01 7.81691968e-01 3.09159577e-01 -3.99548858e-01 -4.20345552e-02 -1.40330005e+00 7.89156020e-01 -8.34788233e-02 6.95982635e-01 -9.51823711e-01 -1.08447993e+00 -8.27364683e-01 -3.08839865e-02 2.86239803e-01 -5.72072566e-01 5.79725325e-01 -9.56442535e-01 -1.90430295e+00 5.91214657e-01 -6.35084063e-02 -1.95122153e-01 3.97683620e-01 1.95122715e-02 -5.93941450e-01 3.75802785e-01 -3.00538778e-01 4.47335124e-01 1.17644417e+00 -9.26086783e-01 3.13577175e-01 -4.34893101e-01 -4.97224838e-01 1.81828126e-01 -2.37108320e-01 -4.75909039e-02 -8.60201195e-02 -6.48454607e-01 2.44764145e-02 -3.03222895e-01 4.15813476e-02 2.89651472e-02 -1.85739532e-01 1.84344381e-01 9.54279006e-01 -9.86172140e-01 9.43679869e-01 -2.13051891e+00 7.90569931e-02 8.85934159e-02 4.79834944e-01 4.88171190e-01 -2.60208845e-01 -8.27403292e-02 -1.07499920e-01 1.04217596e-01 -4.74760756e-02 -3.14363018e-02 -3.10066432e-01 1.75369203e-01 3.12007940e-03 8.89284730e-01 1.88568115e-01 1.22983956e+00 -7.86512911e-01 -6.25001550e-01 5.01569927e-01 1.04033697e+00 -1.76881850e-01 4.79051024e-01 3.53784144e-01 1.11116242e+00 -4.42581266e-01 6.23772979e-01 9.78549302e-01 3.26000661e-01 -3.47726829e-02 -6.29961610e-01 2.90654182e-01 -1.72878608e-01 -8.91176164e-01 1.49284708e+00 -3.39448273e-01 5.19673705e-01 3.90441716e-01 -1.16526604e+00 1.32923520e+00 6.51490748e-01 8.87789786e-01 -9.76190329e-01 5.40661097e-01 2.06840830e-03 1.93846032e-01 -9.18313205e-01 -3.36029232e-01 -7.21401274e-01 2.24544168e-01 1.60943031e-01 1.66531265e-01 -2.04956904e-01 -4.22046691e-01 -2.90359050e-01 7.96465993e-01 1.68957338e-01 2.26865426e-01 -1.69760838e-01 8.45369637e-01 -8.99755538e-01 9.99387681e-01 3.80484611e-01 -6.38248146e-01 7.11631298e-01 6.53585672e-01 -2.85488874e-01 -7.13682294e-01 -8.08838367e-01 -1.24195524e-01 3.81718457e-01 1.88507866e-02 8.33747461e-02 -7.94669509e-01 -5.92903972e-01 -6.99441358e-02 3.73236597e-01 -8.19112480e-01 -5.40783286e-01 -6.02050602e-01 -9.35446322e-01 8.38256240e-01 6.42874122e-01 7.75784850e-01 -1.32914329e+00 -4.29462910e-01 2.47482330e-01 -4.31405902e-01 -1.01385880e+00 -4.07713205e-01 -1.00663103e-01 -6.18144333e-01 -1.02761865e+00 -8.42616677e-01 -1.83353111e-01 4.11978424e-01 -1.03011630e-01 8.92012179e-01 -2.40029752e-01 -5.43332219e-01 4.79367942e-01 -1.42569676e-01 -6.48799717e-01 -4.11139637e-01 -3.92591804e-01 -8.73607304e-03 4.22556102e-01 5.67762077e-01 -8.54757667e-01 -9.82811749e-01 4.17417258e-01 -6.18606687e-01 -2.47719318e-01 3.59384060e-01 8.67879689e-01 5.74208200e-01 -2.33409822e-01 9.57429528e-01 -6.74509048e-01 5.56549430e-01 -6.04686141e-01 -1.72324717e-01 -8.51832516e-03 -2.69888252e-01 -3.60527843e-01 7.07158744e-01 -6.03006244e-01 -1.10682142e+00 -2.26886839e-01 -2.87165493e-01 -8.86299729e-01 -1.33122191e-01 1.84408680e-01 -5.24323583e-01 -3.17277849e-01 6.55413926e-01 5.08995116e-01 3.85899484e-01 -9.71323345e-03 1.53714061e-01 5.71135998e-01 7.31498063e-01 -3.41337234e-01 5.91969728e-01 2.69409895e-01 2.11134210e-01 -8.38620007e-01 -5.16482532e-01 -1.97919562e-01 -3.83909523e-01 -4.33308065e-01 8.13749671e-01 -1.03310168e+00 -9.83582258e-01 7.87804246e-01 -9.19056177e-01 -3.64909887e-01 -4.57081556e-01 6.80027544e-01 -5.64545333e-01 3.65696490e-01 -5.69303274e-01 -9.63423073e-01 -6.86737657e-01 -8.91679585e-01 6.03585899e-01 7.78355062e-01 1.27736256e-01 -9.05054033e-01 5.96518964e-02 4.88505602e-01 8.37056875e-01 7.69848645e-01 3.61802340e-01 -3.41585815e-01 -1.58104479e-01 -1.84481531e-01 -1.54180780e-01 8.90720606e-01 2.37690464e-01 -3.70317876e-01 -1.41108000e+00 -1.34517208e-01 8.17677498e-01 -2.66550809e-01 6.13984764e-01 7.83165753e-01 1.49952066e+00 -3.53664309e-01 7.18552172e-02 9.87414479e-01 1.30045187e+00 3.48540306e-01 1.19467080e+00 -4.39281940e-01 9.93989825e-01 5.45021951e-01 2.51998544e-01 4.17501748e-01 -4.37745675e-02 3.45982999e-01 4.29633945e-01 -3.68453801e-01 -2.52697945e-01 -5.24620153e-02 3.26542467e-01 5.33829093e-01 -1.47460818e-01 -2.26099581e-01 -3.47950280e-01 2.09557489e-01 -1.37298763e+00 -9.40599561e-01 -4.06687446e-02 2.08644700e+00 8.91700625e-01 -4.56064314e-01 -5.10637835e-03 8.38514268e-02 8.65241051e-01 3.73152912e-01 -8.94392192e-01 -2.95686156e-01 -1.49267450e-01 5.10874212e-01 2.80695140e-01 1.26158893e-01 -7.49763668e-01 2.69771338e-01 5.83215094e+00 6.86839104e-01 -1.51647592e+00 4.28135216e-01 9.22197640e-01 -1.93218380e-01 -3.17173719e-01 -4.94691283e-01 -4.56069708e-01 8.65381062e-01 1.09088421e+00 -6.60618246e-02 6.03187680e-01 5.02254665e-01 4.43524629e-01 4.62186299e-02 -7.52778530e-01 1.45196748e+00 4.05520886e-01 -8.71819735e-01 -4.27829981e-01 6.30281121e-02 2.80654520e-01 -2.31232777e-01 1.88829284e-02 2.56407827e-01 -4.84793752e-01 -1.34765697e+00 8.85326117e-02 9.86312807e-01 1.01506507e+00 -7.08898246e-01 9.83424246e-01 1.00813292e-01 -8.78618181e-01 2.16705538e-02 -4.44231480e-01 3.66607904e-01 -1.18555799e-02 8.15754771e-01 -4.08957630e-01 4.66252804e-01 3.64686131e-01 6.55717969e-01 -1.31872907e-01 8.14097643e-01 -3.59372824e-01 6.97893322e-01 -1.87045708e-01 1.79229394e-01 -3.15234840e-01 -2.77404308e-01 5.96143663e-01 9.18299913e-01 2.87159026e-01 5.54247200e-01 -2.15206951e-01 1.37756062e+00 -2.70124763e-01 1.24767080e-01 -4.99548703e-01 1.80057716e-02 2.93648332e-01 1.53536820e+00 -3.71951967e-01 -1.69759601e-01 -3.28098655e-01 7.69685507e-01 -2.43203983e-01 5.04641354e-01 -1.11646307e+00 -4.76165205e-01 6.24400795e-01 2.05547109e-01 -3.62622924e-02 3.16556990e-01 -1.98214561e-01 -1.18320096e+00 -2.31225099e-02 -8.75855029e-01 1.46309406e-01 -9.31721210e-01 -1.36270058e+00 5.46645522e-01 -2.04414785e-01 -9.73853052e-01 -2.79552359e-02 -1.80444539e-01 -9.73233998e-01 1.40135741e+00 -1.74584162e+00 -9.30375218e-01 -8.67528260e-01 7.17717707e-01 1.28633931e-01 1.70439914e-01 6.93934023e-01 5.41430056e-01 -7.54517198e-01 7.12225974e-01 -2.52192885e-01 2.73900777e-01 8.23571682e-01 -9.49746072e-01 -2.17752084e-01 7.68809080e-01 -3.60915005e-01 3.08257103e-01 4.87585902e-01 -4.83119726e-01 -1.31632710e+00 -1.06835616e+00 1.77882254e-01 -1.69292048e-01 4.56417836e-02 -2.96979874e-01 -1.10254359e+00 1.72230050e-01 1.46698475e-01 8.47678304e-01 6.83390737e-01 -5.59181929e-01 -8.87474939e-02 -6.92301989e-01 -1.68481004e+00 1.64535642e-01 4.87404436e-01 -5.31314552e-01 -3.04138809e-01 3.54184397e-02 3.54084432e-01 -6.10110581e-01 -1.21958590e+00 5.36475360e-01 6.84298515e-01 -9.26924169e-01 9.01418090e-01 -2.72477418e-01 3.88880134e-01 -2.48374611e-01 1.72812730e-01 -1.40417004e+00 -1.24990426e-01 -9.10877585e-01 -3.30721885e-01 1.50309086e+00 -3.05505663e-01 -9.35503066e-01 7.45763421e-01 5.32136798e-01 -1.04865216e-01 -7.28472233e-01 -8.79406571e-01 -3.80807340e-01 1.60311796e-02 -1.36415452e-01 6.33198500e-01 9.02840376e-01 -1.08656101e-01 1.13944717e-01 -6.94433510e-01 -9.15884972e-02 6.04382753e-01 -3.09141189e-01 6.20875895e-01 -9.79289651e-01 -2.37751782e-01 -2.12230489e-01 -5.05412042e-01 -4.56337661e-01 1.69479743e-01 -5.46872795e-01 1.85136646e-01 -1.20186114e+00 1.68508336e-01 -2.18274936e-01 -6.65896833e-01 4.58317399e-01 -3.75162244e-01 6.93868876e-01 1.45945018e-02 -2.13959947e-01 6.99621662e-02 7.32940912e-01 1.63215303e+00 -4.37656604e-02 -4.05695975e-01 -2.27744564e-01 -8.32839787e-01 4.09020454e-01 7.53130972e-01 -5.04304230e-01 -5.03586352e-01 9.25853401e-02 -2.23366693e-01 5.35046577e-01 5.51070511e-01 -1.01039183e+00 4.16759141e-02 -7.02795312e-02 9.28606749e-01 -2.10609704e-01 4.02546227e-01 -5.63983619e-01 3.26309472e-01 3.16419512e-01 -5.31178638e-02 -4.18801785e-01 2.40182072e-01 3.51991445e-01 -2.79651195e-01 1.96614731e-02 1.10428298e+00 -3.85407507e-01 -7.18045011e-02 6.05857611e-01 -1.56380370e-01 3.38331580e-01 9.82103169e-01 -2.76015520e-01 -4.67825502e-01 -4.83628124e-01 -6.99134767e-01 2.64211874e-02 -4.07172740e-02 1.89561889e-01 7.68912196e-01 -1.32562828e+00 -9.25191402e-01 5.75718701e-01 -2.55177796e-01 -7.10798725e-02 8.50098789e-01 1.41720891e+00 -2.47922540e-01 -8.16395134e-02 -4.77744311e-01 -4.73078549e-01 -1.02727091e+00 3.53536874e-01 9.55943346e-01 5.96404821e-03 -5.85697114e-01 8.07046294e-01 2.89834797e-01 6.64581135e-02 -5.59990183e-02 1.35857955e-01 -3.78413856e-01 -4.89528663e-02 7.15987146e-01 5.47907650e-01 -5.85906953e-02 -6.44130766e-01 -2.27555379e-01 5.70659339e-01 3.42009783e-01 2.33286202e-01 1.25098288e+00 -2.29216591e-01 -1.99215189e-01 2.48351395e-01 1.27888894e+00 -2.27837692e-04 -1.36107576e+00 4.48206887e-02 -9.08305228e-01 -6.88911319e-01 3.20717633e-01 -8.77191007e-01 -1.82548690e+00 1.02735782e+00 9.41581368e-01 -4.13152613e-02 1.60585165e+00 -4.00595665e-01 8.82494450e-01 -3.31263751e-01 -2.12083012e-02 -8.18506539e-01 1.60928905e-01 -3.35041612e-01 9.16152477e-01 -9.34994578e-01 -6.93359599e-02 -3.96689683e-01 -7.14617193e-01 1.24204028e+00 6.89225197e-01 -7.91668072e-02 6.23068750e-01 4.36467409e-01 3.25417101e-01 -1.20753773e-01 -3.49618912e-01 2.37650901e-01 2.38247514e-01 9.31109905e-01 2.83685505e-01 -1.87787354e-01 -3.13924730e-01 6.77954495e-01 2.12546468e-01 3.15320522e-01 4.38231230e-01 3.61014575e-01 -5.66333756e-02 -7.85952508e-01 -2.64680564e-01 6.57269955e-01 -8.19782972e-01 3.79349068e-02 1.28401577e-01 5.77597618e-01 3.47271144e-01 8.72131050e-01 -5.16713485e-02 -3.59984517e-01 2.37014741e-01 2.11752132e-01 6.14695549e-01 -2.66550332e-01 -6.20287657e-01 3.51178557e-01 -4.44000304e-01 -7.86850035e-01 -6.10271692e-01 -5.34341156e-01 -1.00093532e+00 -2.03432202e-01 -3.36450040e-01 1.22555368e-01 6.31377697e-01 8.12997580e-01 2.77755678e-01 7.49996364e-01 9.80749309e-01 -6.28052831e-01 -3.39979440e-01 -1.20917606e+00 -9.21693921e-01 3.53781164e-01 4.08720404e-01 -5.31742394e-01 -6.31476223e-01 -1.24742076e-01]
[13.896013259887695, 2.7274270057678223]
88d30329-4f7c-4689-842f-e3661d4186b4
ft-tdr-frequency-guided-transformer-and-top
2108.04424
null
https://arxiv.org/abs/2108.04424v2
https://arxiv.org/pdf/2108.04424v2.pdf
FT-TDR: Frequency-guided Transformer and Top-Down Refinement Network for Blind Face Inpainting
Blind face inpainting refers to the task of reconstructing visual contents without explicitly indicating the corrupted regions in a face image. Inherently, this task faces two challenges: (1) how to detect various mask patterns of different shapes and contents; (2) how to restore visually plausible and pleasing contents in the masked regions. In this paper, we propose a novel two-stage blind face inpainting method named Frequency-guided Transformer and Top-Down Refinement Network (FT-TDR) to tackle these challenges. Specifically, we first use a transformer-based network to detect the corrupted regions to be inpainted as masks by modeling the relation among different patches. We also exploit the frequency modality as complementary information for improved detection results and capture the local contextual incoherence to enhance boundary consistency. Then a top-down refinement network is proposed to hierarchically restore features at different levels and generate contents that are semantically consistent with the unmasked face regions. Extensive experiments demonstrate that our method outperforms current state-of-the-art blind and non-blind face inpainting methods qualitatively and quantitatively.
['Yu-Gang Jiang', 'Zuxuan Wu', 'Shaoxiang Chen', 'Junke Wang']
2021-08-10
null
null
null
null
['facial-inpainting']
['computer-vision']
[ 4.67502117e-01 -9.24976543e-02 1.02905020e-01 -1.58726051e-01 -6.46292508e-01 -2.88624614e-01 2.72531241e-01 -6.93953931e-01 1.12685464e-01 7.08327591e-01 6.75760150e-01 1.68733388e-01 2.51775570e-02 -5.59486151e-01 -7.37070501e-01 -6.84509575e-01 3.90839964e-01 -1.00983478e-01 1.25638127e-01 -2.57557303e-01 1.32533327e-01 7.34937608e-01 -1.90000069e+00 5.31943738e-01 9.81565416e-01 9.03460622e-01 1.48926750e-01 3.26787412e-01 -5.50819896e-02 7.61985242e-01 -3.72749746e-01 -1.58789560e-01 6.03055418e-01 -8.47998142e-01 -4.20897484e-01 5.86259663e-01 8.01805496e-01 -5.64243019e-01 -4.55969155e-01 1.35646057e+00 3.85329664e-01 9.14300308e-02 6.81207955e-01 -9.00720477e-01 -9.20580029e-01 1.02499858e-01 -1.14919114e+00 8.80427135e-04 5.82034886e-01 2.10658178e-01 3.88436794e-01 -1.30801308e+00 4.92954075e-01 1.52791202e+00 5.14136851e-01 8.36632192e-01 -1.31512105e+00 -8.16190124e-01 2.78360903e-01 2.81947285e-01 -1.41148591e+00 -1.02904606e+00 1.19359124e+00 -3.90576988e-01 3.82122785e-01 3.46116394e-01 4.67794478e-01 7.19630122e-01 -7.85642490e-02 3.13299716e-01 1.33701825e+00 -5.78108430e-01 -2.33517103e-02 -7.86152259e-02 -5.09002805e-01 8.16222131e-01 1.51583523e-01 4.31209773e-01 -7.22310543e-01 2.80323625e-02 1.12130392e+00 1.82725057e-01 -9.18805122e-01 -6.16048500e-02 -9.12788391e-01 5.39047062e-01 5.40446401e-01 3.33118707e-01 -6.46288037e-01 -1.73939675e-01 -2.85540342e-01 1.16793208e-01 7.87160993e-01 1.86111197e-01 -7.87968189e-02 8.06383491e-01 -1.27119434e+00 2.03896642e-01 2.98352301e-01 6.44892275e-01 1.02535605e+00 1.75023794e-01 -5.76974690e-01 1.11109161e+00 4.59222317e-01 2.10826397e-01 2.28870153e-01 -1.13312185e+00 3.61774862e-01 4.36405599e-01 4.43616688e-01 -9.90533710e-01 7.02404827e-02 -2.78356820e-01 -1.10907197e+00 6.46516621e-01 2.06921786e-01 -3.72699685e-02 -1.28086841e+00 1.65478408e+00 5.07397532e-01 5.75728834e-01 -2.45602503e-01 1.25941861e+00 8.31298411e-01 6.44978225e-01 -2.52183139e-01 -6.14757478e-01 1.55233753e+00 -1.12130797e+00 -1.06172907e+00 -5.39172411e-01 -4.20431703e-01 -1.03469241e+00 7.73255944e-01 2.06173465e-01 -1.45766234e+00 -6.91329122e-01 -9.05201614e-01 -1.85257524e-01 1.92499772e-01 2.19238281e-01 4.63490598e-02 4.53678131e-01 -1.42266905e+00 3.30006361e-01 -2.83587277e-01 -1.16951659e-01 7.17259288e-01 1.17671467e-01 -4.50139463e-01 -6.04229271e-01 -9.15842056e-01 6.82436287e-01 -3.03682089e-02 3.33950251e-01 -1.01852703e+00 -7.28289902e-01 -1.04842246e+00 -6.53272495e-02 2.29547635e-01 -8.31181407e-01 9.43414032e-01 -1.23826730e+00 -1.34050727e+00 9.58784461e-01 -8.32720339e-01 1.39350384e-01 5.73320091e-01 5.19215837e-02 -4.90600109e-01 4.10508305e-01 3.16078514e-01 6.31884277e-01 1.64544320e+00 -1.85418487e+00 -4.98232901e-01 -4.96582299e-01 -4.12343413e-01 4.21618134e-01 -2.07065061e-01 3.01776409e-01 -7.55335271e-01 -1.33900344e+00 4.60455596e-01 -3.08928967e-01 6.45448342e-02 5.56629717e-01 -3.68609995e-01 3.19664627e-01 1.02360404e+00 -1.33599925e+00 1.05520618e+00 -2.20907903e+00 3.75344455e-01 3.11509613e-02 5.20636141e-01 2.48179078e-01 -3.71233404e-01 1.40893355e-01 -1.73659325e-01 -2.65778959e-01 -7.37033188e-01 -6.95196807e-01 -3.65296215e-01 7.89711028e-02 -4.10259962e-01 6.93929493e-01 3.40524435e-01 6.14229858e-01 -7.05786824e-01 -5.53797364e-01 3.91801149e-01 9.00833786e-01 -5.43910801e-01 5.26304483e-01 -1.69487849e-01 8.90675366e-01 -2.13626958e-02 9.40415263e-01 1.26232851e+00 1.34876877e-01 -1.19623676e-01 -4.70295280e-01 -9.31694917e-03 -2.00218439e-01 -1.05532491e+00 1.62091386e+00 -3.67628783e-01 4.06529427e-01 9.54551160e-01 -6.36442602e-01 7.50866592e-01 3.21654618e-01 1.40900165e-01 -7.27587640e-01 6.09410964e-02 1.68031082e-01 -4.70081478e-01 -5.78188539e-01 3.25968951e-01 -4.55415964e-01 5.91275036e-01 3.63444120e-01 -1.03320993e-01 4.39581759e-02 -2.33706132e-01 -2.53072828e-01 7.20740318e-01 -1.19724125e-03 -1.25321478e-01 -5.67147434e-02 7.91928113e-01 -6.77040875e-01 7.09086418e-01 3.21721017e-01 -2.40495294e-01 1.23304152e+00 8.31193253e-02 -2.40707159e-01 -6.70860887e-01 -1.04681480e+00 2.55159345e-02 9.38597620e-01 4.35871184e-01 -3.04746684e-02 -1.07073379e+00 -4.41846192e-01 -5.21825626e-02 3.34790438e-01 -8.01248133e-01 5.00394730e-03 -6.17194355e-01 -3.51173222e-01 6.95863590e-02 1.98548838e-01 7.83827126e-01 -1.20434892e+00 -2.13240832e-02 -5.54839289e-03 -5.77271938e-01 -9.88239169e-01 -1.01601779e+00 -5.01306415e-01 -6.20030046e-01 -1.12248218e+00 -1.18449819e+00 -1.40869653e+00 1.17470968e+00 7.47124016e-01 8.63272667e-01 3.29035014e-01 -3.86266947e-01 7.23117171e-03 -1.53696641e-01 1.19761810e-01 -4.33831990e-01 -7.48756886e-01 3.89811136e-02 7.53647149e-01 -2.79080540e-01 -8.77705693e-01 -9.89123702e-01 4.60128367e-01 -1.13672435e+00 2.21978083e-01 6.07463241e-01 9.46673870e-01 6.28875732e-01 3.41944873e-01 3.54596615e-01 -5.96011460e-01 5.91378331e-01 -1.30254686e-01 -3.58924001e-01 3.97400200e-01 -2.62656957e-01 -1.17346868e-01 4.91025716e-01 -5.13477743e-01 -1.36078560e+00 6.68425187e-02 6.25391975e-02 -9.25127208e-01 -8.27834457e-02 -1.42006978e-01 -5.76543689e-01 -4.92141217e-01 4.96759087e-01 5.36725223e-01 1.62813827e-01 -6.97000921e-01 4.82988626e-01 5.59199810e-01 9.33805764e-01 -3.67887735e-01 1.02372134e+00 7.66577601e-01 -1.40812904e-01 -6.72489822e-01 -6.60451353e-01 -1.94568947e-01 -3.37071359e-01 -4.35240209e-01 8.13812137e-01 -9.81176078e-01 -3.98109645e-01 8.08432400e-01 -1.31019986e+00 -1.82067677e-01 -1.61057815e-01 -5.56515809e-03 -3.75826538e-01 7.03599453e-01 -7.17255712e-01 -7.56959081e-01 -3.81684214e-01 -1.14385056e+00 1.21351755e+00 3.42891753e-01 2.73416281e-01 -5.09886682e-01 -2.78581977e-01 5.92251897e-01 5.23312747e-01 1.81243047e-01 7.33057559e-01 3.72235626e-01 -6.25189483e-01 2.13664562e-01 -5.74098587e-01 4.60846245e-01 7.44206011e-01 -3.23614419e-01 -1.22651172e+00 -4.82255995e-01 2.63130397e-01 4.43670712e-02 1.08553910e+00 5.17398298e-01 1.15803528e+00 -7.75691986e-01 -2.39032984e-01 8.56720388e-01 1.18320596e+00 -4.98502329e-02 8.78866732e-01 -1.67869836e-01 7.94413984e-01 1.04667819e+00 3.04567605e-01 1.96595520e-01 2.39390600e-02 6.93753600e-01 4.63415474e-01 -4.14708108e-01 -7.80750692e-01 -5.43285012e-01 3.19752902e-01 4.22551423e-01 8.17277208e-02 -8.13065246e-02 -3.53424966e-01 6.50902092e-01 -1.73110735e+00 -9.42972541e-01 1.57822400e-01 2.10361147e+00 1.00507295e+00 -3.59660357e-01 -1.25525042e-01 8.89536962e-02 1.26654851e+00 3.19687366e-01 -5.29257715e-01 2.19240233e-01 -3.53428870e-01 3.65514815e-01 8.15410614e-02 9.00739968e-01 -1.00620437e+00 9.54275310e-01 5.81076527e+00 1.05353069e+00 -8.78472686e-01 2.86608875e-01 8.55597556e-01 5.78846410e-02 -3.79055411e-01 -2.79908683e-02 -3.16984594e-01 5.40911913e-01 -2.61604805e-02 2.35580772e-01 8.69187951e-01 1.35961100e-01 2.68219382e-01 -1.05317131e-01 -7.44688928e-01 1.23472309e+00 4.61466461e-01 -1.31507337e+00 3.29835683e-01 -1.51376724e-01 1.00462520e+00 -6.26914561e-01 1.55782178e-01 -3.45014304e-01 4.91977995e-03 -1.30594039e+00 9.51317489e-01 8.08773816e-01 1.23380184e+00 -7.52430499e-01 1.89691201e-01 3.77821550e-02 -1.37651801e+00 -1.55618876e-01 -1.96402311e-01 2.03610346e-01 1.79084852e-01 5.45873106e-01 -2.47092247e-01 5.10091007e-01 8.18468869e-01 8.35851371e-01 -3.78808230e-01 1.01354849e+00 -5.29897273e-01 2.17700467e-01 7.85747021e-02 8.70938957e-01 -4.06204700e-01 -2.39433259e-01 7.32523561e-01 8.10651183e-01 3.65839034e-01 1.91269979e-01 -2.16252238e-01 1.28954613e+00 -2.89130151e-01 -2.16901854e-01 -3.32864016e-01 3.52632791e-01 6.33808196e-01 1.23481536e+00 -5.82977593e-01 -1.22860432e-01 -4.02199477e-01 1.46765709e+00 1.41203720e-02 6.45384252e-01 -5.87769866e-01 -3.11720520e-01 9.05002713e-01 4.53900665e-01 3.84635240e-01 2.09738463e-01 -1.75943300e-01 -1.13795400e+00 3.19315612e-01 -1.00168383e+00 1.38296500e-01 -1.11058068e+00 -1.49680328e+00 7.10752428e-01 -4.61555004e-01 -1.41003299e+00 1.92651823e-01 -3.31795633e-01 -7.57748425e-01 1.37835503e+00 -1.90469909e+00 -1.28860462e+00 -6.25992179e-01 1.02389288e+00 5.72899699e-01 -8.95643085e-02 4.90519345e-01 3.51472795e-01 -5.60590148e-01 6.18643224e-01 -1.48911744e-01 8.53491575e-02 9.44606721e-01 -5.36576688e-01 1.57158449e-01 1.35673463e+00 -2.31350675e-01 4.35288280e-01 6.20353401e-01 -7.89826035e-01 -1.21972203e+00 -1.47013092e+00 6.64490581e-01 1.08716778e-01 -2.04253811e-02 -2.23518983e-01 -1.10583150e+00 3.83697659e-01 1.82126760e-01 3.22473854e-01 -9.11384970e-02 -7.20703185e-01 -7.40656078e-01 -2.97891140e-01 -1.57233942e+00 5.76705754e-01 1.11166573e+00 -7.56656945e-01 -4.34626758e-01 1.40620843e-01 8.03292692e-01 -3.37291002e-01 -1.83476180e-01 4.64937747e-01 3.37151498e-01 -1.39192688e+00 1.17387295e+00 -1.94292683e-02 3.60456973e-01 -6.72308505e-01 -1.01353221e-01 -1.20887387e+00 -2.81336725e-01 -1.00772059e+00 -2.30698198e-01 1.41448033e+00 -4.34609391e-02 -7.18114555e-01 6.60616517e-01 2.78853118e-01 -1.16513446e-01 -4.57909077e-01 -9.92791295e-01 -3.39322150e-01 -2.78061092e-01 5.72832637e-02 6.43314838e-01 1.03923023e+00 -5.13969481e-01 -1.33041635e-01 -6.06672287e-01 3.19588810e-01 9.15404022e-01 1.59191102e-01 2.85975277e-01 -1.05958474e+00 -1.61526233e-01 -3.79820496e-01 -9.49307010e-02 -9.71493185e-01 1.30269513e-01 -5.17258763e-01 3.34333837e-01 -1.62348092e+00 1.61069065e-01 -1.46132540e-02 3.40544954e-02 5.08721948e-01 -2.86544800e-01 6.97124839e-01 -1.33861139e-01 4.19373512e-01 -4.47827540e-02 7.32339084e-01 1.59017396e+00 -1.06273375e-01 -1.19749591e-01 -2.59011209e-01 -1.08519053e+00 6.07153237e-01 4.49252546e-01 -2.34278917e-01 -2.31077120e-01 -4.98108596e-01 -2.56744534e-01 3.86961520e-01 7.45676339e-01 -9.77364540e-01 2.48895317e-01 -2.11080626e-01 8.30306530e-01 -2.89380759e-01 5.65635622e-01 -7.58339703e-01 1.32911474e-01 2.07038507e-01 4.99005839e-02 -2.12269306e-01 3.18160266e-01 7.04077065e-01 -4.17913884e-01 1.61158964e-01 1.26802981e+00 -4.02592421e-02 -4.51809913e-01 5.18618643e-01 -1.60707682e-01 -1.46512017e-02 9.04579580e-01 -2.83349782e-01 -3.23602974e-01 -5.35775304e-01 -6.57444954e-01 -1.04553506e-01 7.49887884e-01 4.38274831e-01 1.14360726e+00 -1.33683395e+00 -9.78125155e-01 9.53792572e-01 -7.76111558e-02 7.98106007e-03 7.64251590e-01 5.49402773e-01 -5.28716683e-01 -1.53582409e-01 -2.41307348e-01 -2.22924963e-01 -1.34558892e+00 6.09184265e-01 6.20912969e-01 2.11304322e-01 -7.09295154e-01 1.16031706e+00 6.41662776e-01 -4.00909707e-02 4.76116210e-01 -1.27987582e-02 -2.04602093e-01 -7.47630978e-03 8.64556432e-01 1.46993682e-01 9.24109109e-03 -1.02197528e+00 -2.38579914e-01 9.08700526e-01 -4.54532541e-02 -2.45600849e-01 1.20249677e+00 -3.64047378e-01 -6.72844887e-01 -3.23074579e-01 9.52444732e-01 2.54771680e-01 -1.61848581e+00 -4.25781935e-01 -5.41742504e-01 -9.69667256e-01 1.85774669e-01 -7.88965762e-01 -1.64590371e+00 7.05690265e-01 7.56374419e-01 -1.89739034e-01 1.80207801e+00 -3.49546112e-02 7.47977257e-01 -5.81889570e-01 1.60596400e-01 -6.06384814e-01 2.76381433e-01 -5.45389727e-02 1.48604798e+00 -8.44198227e-01 -4.82763350e-02 -8.45907629e-01 -2.56023675e-01 8.98144066e-01 7.18325317e-01 -1.70753319e-02 6.60408616e-01 1.09083399e-01 1.47046670e-01 -5.35876155e-02 -2.66066462e-01 -2.74775684e-01 5.84534764e-01 8.43785107e-01 1.42163262e-01 -2.28681490e-01 4.52612117e-02 4.48444396e-01 9.58766192e-02 -1.20993771e-01 1.90876111e-01 7.58458555e-01 -4.29291934e-01 -9.05269921e-01 -1.03607202e+00 2.17081308e-01 -2.34454662e-01 -2.32909888e-01 -5.29533565e-01 2.68898547e-01 4.09488142e-01 1.38302672e+00 -9.48928148e-02 -3.82726133e-01 8.45633522e-02 -1.09691165e-01 7.27473199e-01 -5.56721568e-01 -2.96250105e-01 3.27950209e-01 -3.59095752e-01 -6.18567467e-01 -4.31162417e-01 -3.02181184e-01 -8.75761569e-01 -1.95541590e-01 -3.72721367e-02 -9.22492594e-02 1.76670432e-01 5.92310429e-01 5.56301534e-01 5.05514562e-01 8.23240399e-01 -1.39813411e+00 1.24025028e-02 -9.96037602e-01 -8.21491659e-01 2.83449084e-01 9.89829361e-01 -7.37619102e-01 -6.84316695e-01 1.53961316e-01]
[12.721569061279297, -0.14921171963214874]
f4475734-0276-4d5d-9d04-a465a54f8607
deep-neural-review-text-interaction-for
2003.07051
null
https://arxiv.org/abs/2003.07051v1
https://arxiv.org/pdf/2003.07051v1.pdf
Deep Neural Review Text Interaction for Recommendation Systems
Users' reviews contain valuable information which are not taken into account in most recommender systems. According to the latest studies in this field, using review texts could not only improve the performance of recommendation, but it can also alleviate the impact of data sparsity and help to tackle the cold start problem. In this paper, we present a neural recommender model which recommends items by leveraging user reviews. In order to predict user rating for each item, our proposed model, named MatchPyramid Recommender System (MPRS), represents each user and item with their corresponding review texts. Thus, the problem of recommendation is viewed as a text matching problem such that the matching score obtained from matching user and item texts could be considered as a good representative of their joint extent of similarity. To solve the text matching problem, inspired by MatchPyramid (Pang, 2016), we employed an interaction-based approach according to which a matching matrix is constructed given a pair of input texts. The matching matrix, which has the property of hierarchical matching patterns, is then fed into a Convolutional Neural Network (CNN) to compute the matching score for the given user-item pair. Our experiments on the small data categories of Amazon review dataset show that our proposed model gains from 1.76% to 21.72% relative improvement compared to DeepCoNN model, and from 0.83% to 3.15% relative improvement compared to TransNets model. Also, on two large categories, namely AZ-CSJ and AZ-Mov, our model achieves relative improvements of 8.08% and 7.56% compared to the DeepCoNN model, and relative improvements of 1.74% and 0.86% compared to the TransNets model, respectively.
['Saeedeh Momtazi', 'Parisa Abolfath Beygi Dezfouli', 'Mehdi Dehghan']
2020-03-16
null
null
null
null
['small-data']
['computer-vision']
[-1.36775091e-01 -2.42684826e-01 -3.12936187e-01 -4.53565001e-01 -3.20299149e-01 -2.40838468e-01 5.17001987e-01 2.07726613e-01 -4.25750643e-01 1.97821423e-01 3.66134703e-01 -1.84136301e-01 -3.02580774e-01 -1.10069537e+00 -4.02493507e-01 -3.45612884e-01 3.74722421e-01 1.45289034e-01 -8.45461860e-02 -5.00832498e-01 3.76696914e-01 3.69531550e-02 -1.64220262e+00 4.14789885e-01 1.13021910e+00 1.41933644e+00 2.23203808e-01 1.22114547e-01 -2.23062173e-01 5.21926284e-01 -2.62590617e-01 -6.62144363e-01 4.37128335e-01 -1.46183357e-01 -4.57750797e-01 -1.89416394e-01 3.56895357e-01 -2.85997391e-01 -4.95678008e-01 9.24067676e-01 3.37962389e-01 6.12662613e-01 6.28757179e-01 -9.43431497e-01 -1.13634109e+00 1.03951359e+00 -6.98205292e-01 5.47611620e-04 2.04198316e-01 -3.85185957e-01 1.52473974e+00 -1.15277112e+00 2.97502935e-01 8.95677924e-01 5.54456532e-01 2.50550151e-01 -7.37242222e-01 -7.69542456e-01 4.44899976e-01 2.14251429e-01 -1.21088278e+00 -1.06098063e-01 4.60170805e-01 -2.56718308e-01 9.30030406e-01 2.69957602e-01 7.18640268e-01 6.52530551e-01 1.66497558e-01 8.21934342e-01 6.49154425e-01 -6.22705184e-02 6.71990812e-02 1.28804728e-01 4.85038847e-01 4.12081331e-01 1.69517964e-01 -1.30446568e-01 -3.30458641e-01 -1.30767241e-01 4.30870622e-01 7.28164911e-01 -1.94810018e-01 4.83100712e-02 -8.94307017e-01 9.41524863e-01 8.45056176e-01 3.50881159e-01 -4.87905651e-01 -3.04778069e-01 2.78013051e-01 3.79995167e-01 5.89542508e-01 5.71361005e-01 -4.67042506e-01 2.09747881e-01 -6.68860793e-01 1.94724068e-01 7.93879807e-01 7.35367060e-01 6.42523348e-01 -8.55418295e-02 -2.35681236e-01 1.23674417e+00 6.03082895e-01 4.83846545e-01 8.80528152e-01 -5.04539549e-01 4.74554479e-01 8.54196727e-01 1.69121809e-02 -1.50582194e+00 -5.06365716e-01 -7.91291893e-01 -1.11874032e+00 -4.32309449e-01 1.34621918e-01 -8.51651803e-02 -7.04711854e-01 1.50682282e+00 1.64557695e-01 2.29645565e-01 8.77261981e-02 1.01603818e+00 1.25765634e+00 8.47505450e-01 -2.09257230e-01 -2.30773136e-01 1.21229124e+00 -1.26536095e+00 -4.36029702e-01 -3.01093888e-02 7.83252060e-01 -1.06012368e+00 1.23285449e+00 5.86474657e-01 -9.14466083e-01 -6.83330894e-01 -9.80334878e-01 7.68878460e-02 -3.97066504e-01 4.18873072e-01 4.42668319e-01 3.24317962e-01 -8.40291083e-01 8.50354850e-01 -1.92189351e-01 -4.19622928e-01 8.52971077e-02 5.00632644e-01 -6.95344508e-02 -2.27301136e-01 -1.38178337e+00 6.92281902e-01 -8.48817900e-02 2.10287854e-01 -4.14152384e-01 -5.84958494e-01 -4.45896864e-01 4.04346615e-01 4.00861472e-01 -5.16025543e-01 1.14874959e+00 -9.90988731e-01 -1.44368136e+00 2.39529520e-01 8.58454257e-02 -3.25590342e-01 -1.40670436e-02 -4.82250720e-01 -8.26220751e-01 -3.34180832e-01 -1.12923242e-01 2.74844527e-01 3.71156901e-01 -8.73982668e-01 -8.45372379e-01 -3.80089194e-01 4.36589599e-01 3.88356030e-01 -7.46436059e-01 2.17830818e-02 -7.70312190e-01 -7.01182663e-01 -3.07823215e-02 -1.00190651e+00 -5.13339341e-01 -4.24699008e-01 -2.31307998e-01 -4.42578226e-01 4.25764591e-01 -5.02743006e-01 1.51669669e+00 -1.83720636e+00 -2.64891726e-03 4.45981413e-01 3.13758552e-01 5.50488234e-01 -4.46845531e-01 5.17603219e-01 1.10628799e-01 5.95967844e-02 1.17380075e-01 -1.73818707e-01 4.12344672e-02 1.06894318e-02 -2.28160486e-01 2.51786262e-01 -3.38965744e-01 8.75555992e-01 -7.78883398e-01 1.32845789e-01 1.23851642e-01 5.37263989e-01 -6.09488487e-01 3.67171377e-01 -1.25431225e-01 -8.12842175e-02 -5.23060501e-01 4.31086451e-01 6.33432031e-01 -3.87800574e-01 1.97320491e-01 -3.42123240e-01 6.13137148e-03 3.82927388e-01 -1.10644984e+00 1.47946692e+00 -7.90868700e-01 2.87206978e-01 -2.83243001e-01 -9.11966026e-01 1.36641383e+00 1.89706221e-01 4.72908735e-01 -1.14469266e+00 4.17143136e-01 2.19345823e-01 1.52301610e-01 -3.83506805e-01 9.38559473e-01 2.27920175e-01 -9.77166113e-04 7.69849122e-01 -1.93938062e-01 5.32086074e-01 1.01145945e-01 4.65804845e-01 1.02194142e+00 -1.82422251e-01 1.36989787e-01 -1.91127464e-01 7.87163556e-01 -3.76710057e-01 4.76688057e-01 6.18046761e-01 4.24796164e-01 4.28567439e-01 1.51577577e-01 -5.92937768e-01 -7.37924993e-01 -4.90207553e-01 8.31333697e-02 1.17797935e+00 2.80409038e-01 -8.04921389e-01 -3.73646915e-01 -8.22728336e-01 3.93745229e-02 3.89641106e-01 -7.10653782e-01 -3.88454556e-01 -4.27286744e-01 -6.41895890e-01 -4.29432373e-03 4.93112296e-01 3.69575858e-01 -1.02224147e+00 -2.02485733e-03 1.71468914e-01 -7.67258108e-02 -7.16475189e-01 -6.86846852e-01 -1.37455687e-01 -8.18371356e-01 -1.04577374e+00 -7.32922614e-01 -6.65303826e-01 6.17132485e-01 9.83604670e-01 1.17895555e+00 5.77935636e-01 3.57614338e-01 -6.85567930e-02 -9.03802216e-01 1.79010574e-02 9.78761539e-02 3.97190124e-01 1.93367094e-01 3.65607142e-01 5.14385760e-01 -6.30235374e-01 -9.48989093e-01 6.52841151e-01 -9.27326560e-01 -1.64223999e-01 7.20698118e-01 8.03104639e-01 4.51128006e-01 -2.50761032e-01 8.60280573e-01 -1.19987571e+00 1.03366339e+00 -9.51114714e-01 -3.91164303e-01 2.62933463e-01 -1.18894875e+00 -2.08915100e-01 1.22846377e+00 -4.11702633e-01 -8.80311906e-01 -2.03542143e-01 -3.22157919e-01 -3.77042472e-01 1.39630780e-01 9.25989330e-01 -5.33184111e-02 1.47373140e-01 5.15548110e-01 4.55678888e-02 -3.16312730e-01 -7.91746616e-01 5.73920310e-01 1.00728703e+00 1.73784867e-01 -2.91319400e-01 4.99908984e-01 8.36599469e-02 -4.81497318e-01 -1.80322498e-01 -1.14797819e+00 -7.28682578e-01 -3.09049934e-01 -4.70362529e-02 4.52688366e-01 -8.87364388e-01 -7.97254086e-01 1.60559520e-01 -8.54128242e-01 1.82398841e-01 -8.49420801e-02 6.37515008e-01 3.96853276e-02 4.18918848e-01 -6.37493610e-01 -4.46603239e-01 -9.40702856e-01 -1.07260823e+00 4.77056950e-01 5.26315272e-01 -4.78565469e-02 -8.68208349e-01 8.94377902e-02 4.57272977e-01 7.18714654e-01 -2.77091980e-01 7.20888495e-01 -1.22369623e+00 -2.73792893e-01 -4.33312118e-01 -3.56366694e-01 3.68350804e-01 1.05004430e-01 -5.93461730e-02 -6.64851248e-01 -3.60885888e-01 -1.93246305e-01 -8.67992938e-02 7.71370947e-01 1.28055945e-01 1.07720232e+00 -3.21773410e-01 -2.02968314e-01 4.02557552e-01 1.35020018e+00 4.30989534e-01 5.47598779e-01 2.77374625e-01 8.85741830e-01 4.36077625e-01 9.06109154e-01 5.90752244e-01 6.01066828e-01 7.39152253e-01 4.89250004e-01 -5.25966436e-02 1.84087187e-01 -1.48221940e-01 2.24434733e-01 1.42066920e+00 9.14677531e-02 -4.79630321e-01 -5.13476133e-01 3.73566121e-01 -2.16653013e+00 -6.10155165e-01 -3.78377795e-01 2.27532578e+00 4.25851107e-01 -5.38353510e-02 8.06448832e-02 -3.40309218e-02 7.10026622e-01 1.37782386e-02 -6.94146156e-01 -4.95005876e-01 2.10289836e-01 2.84147054e-01 2.10001782e-01 1.43692851e-01 -7.68423617e-01 6.83706462e-01 5.03652716e+00 8.47660303e-01 -1.04337513e+00 -5.68560436e-02 4.98977661e-01 -2.21038729e-01 -4.30071890e-01 -1.78118840e-01 -8.06095898e-01 6.66995823e-01 9.95776594e-01 -3.12590808e-01 4.85026091e-01 6.87697768e-01 2.18802482e-01 2.05434859e-01 -8.65985394e-01 8.62783253e-01 2.51946598e-01 -1.19194627e+00 2.20591173e-01 1.51685312e-01 9.74042237e-01 1.63881928e-01 1.87951982e-01 5.65560818e-01 5.07494926e-01 -8.87290478e-01 2.41883323e-01 5.73434174e-01 2.27271065e-01 -1.14227462e+00 1.07428372e+00 3.97951305e-01 -1.35687304e+00 -1.87708929e-01 -8.87177110e-01 -9.95082557e-02 -1.01328567e-01 8.11088324e-01 -2.81006336e-01 8.38651955e-01 8.01179945e-01 1.31293547e+00 -3.67560714e-01 1.21363270e+00 -1.20307118e-01 5.60457945e-01 -8.61358121e-02 -2.32499212e-01 3.33276123e-01 -5.50708592e-01 -1.96104450e-03 9.45582569e-01 6.32509232e-01 1.60325661e-01 2.61047363e-01 4.45304424e-01 -5.09530008e-01 8.96636367e-01 -3.49183440e-01 2.01582223e-01 4.73478109e-01 1.78557098e+00 -3.88107091e-01 -4.10885334e-01 -5.98315120e-01 6.01608336e-01 4.54004616e-01 9.94703025e-02 -7.20937788e-01 -5.96398175e-01 6.53569758e-01 -1.87763587e-01 4.67638135e-01 3.04662615e-01 -2.01903462e-01 -1.27601743e+00 -4.63467315e-02 -9.64434564e-01 4.28872287e-01 -6.72625542e-01 -1.53509533e+00 8.87442470e-01 -5.96047103e-01 -1.55277145e+00 1.00049302e-01 -2.75761127e-01 -8.01607788e-01 9.54195321e-01 -1.38013804e+00 -8.99999559e-01 -4.33462411e-01 5.66951156e-01 5.10899127e-01 -3.42672348e-01 7.55600035e-01 6.88918829e-01 -7.35441685e-01 8.83857608e-01 5.57303965e-01 -1.13387518e-01 6.48969769e-01 -9.33850527e-01 4.91699010e-01 5.71713328e-01 2.24686787e-01 9.51712370e-01 2.10733712e-01 -4.31540877e-01 -1.39654386e+00 -1.25742841e+00 9.48047459e-01 -1.57679051e-01 6.70080841e-01 1.16635682e-02 -1.10251856e+00 2.47175276e-01 2.87730634e-01 -1.41653299e-01 1.02170277e+00 4.44727719e-01 -5.39921224e-01 -4.07396436e-01 -9.33427334e-01 6.34434700e-01 9.48312223e-01 -3.36913466e-01 -4.19679046e-01 1.82094440e-01 8.91953409e-01 -1.53941229e-01 -1.22699094e+00 2.62105197e-01 8.80697787e-01 -8.13028932e-01 7.42301822e-01 -5.80370426e-01 7.39809334e-01 -2.31018946e-01 -2.05581069e-01 -1.58197272e+00 -7.04918206e-01 -1.60652012e-01 -2.17605874e-01 1.30080175e+00 6.11608446e-01 -5.50236106e-01 6.66950583e-01 6.14075780e-01 -2.79975772e-01 -1.19744277e+00 -3.07219803e-01 -3.85445029e-01 -9.27996915e-03 -3.22659165e-01 9.21493948e-01 1.02626038e+00 8.99306163e-02 6.55816197e-01 -7.47956216e-01 -1.84252903e-01 8.77598971e-02 4.68025684e-01 7.50175357e-01 -1.41759539e+00 -3.10754359e-01 -5.39605558e-01 1.64084420e-01 -1.36296701e+00 -1.86142586e-02 -1.25012743e+00 -1.40384346e-01 -1.76655865e+00 2.23963812e-01 -6.65328383e-01 -9.98565853e-01 3.16319585e-01 -2.84479350e-01 3.50085586e-01 3.79994392e-01 4.21448022e-01 -7.48084426e-01 5.60170829e-01 1.29606700e+00 -1.02100372e-01 -1.44027725e-01 3.29297513e-01 -1.27100170e+00 5.24005711e-01 8.72058153e-01 -4.81865227e-01 -5.47395051e-01 -5.71745753e-01 6.78518295e-01 6.54328614e-02 -3.68108749e-01 -7.22989321e-01 4.03414607e-01 -3.28383483e-02 3.03031296e-01 -6.79384470e-01 1.67972460e-01 -8.68589282e-01 6.67490065e-02 2.17004657e-01 -6.66158915e-01 2.46035933e-01 -1.64771602e-01 6.32829070e-01 -1.88274190e-01 -2.83127129e-01 4.13047969e-01 3.02672964e-02 -4.29233432e-01 5.79651177e-01 -2.99471647e-01 -3.56767386e-01 5.74505031e-01 7.01675713e-02 -4.67352986e-01 -4.35948223e-01 -4.98958975e-01 4.38413680e-01 1.40255377e-01 8.01981568e-01 7.41029084e-01 -1.57708108e+00 -5.63976824e-01 -9.21622152e-04 2.72347391e-01 -3.13949436e-01 4.37519759e-01 7.45354116e-01 1.27900645e-01 3.91176075e-01 -4.88100983e-02 -1.08641423e-01 -1.19748175e+00 5.16362727e-01 1.74783856e-01 -4.91462350e-01 -3.83811295e-01 7.80746043e-01 1.56852543e-01 -6.57961428e-01 2.63076365e-01 -3.13008130e-01 -9.43246126e-01 1.88164487e-01 7.51629591e-01 4.56959099e-01 3.45470876e-01 -7.16200173e-01 -1.13746531e-01 6.89326048e-01 -6.20158195e-01 4.28595424e-01 1.48095667e+00 -2.22408846e-01 -9.99888256e-02 1.07462689e-01 1.30782020e+00 1.13662146e-02 -6.03113592e-01 -9.10898030e-01 -1.52246043e-01 -5.49738705e-01 1.99484497e-01 -9.77576792e-01 -1.81232619e+00 7.88311839e-01 4.47329044e-01 3.94553125e-01 1.10656416e+00 -2.34737352e-01 1.30422378e+00 6.04170322e-01 1.07871182e-01 -9.73115146e-01 6.62847469e-03 6.65749073e-01 6.65759742e-01 -1.05018306e+00 -2.05737874e-02 -6.97010458e-02 -6.76522851e-01 1.07749009e+00 8.64642799e-01 -4.06419784e-01 8.20148468e-01 -2.35098869e-01 2.59307846e-02 -2.54760832e-01 -1.01948524e+00 -2.65885800e-01 7.50451922e-01 1.21397868e-01 6.53587103e-01 1.01858251e-01 -6.94109857e-01 1.17807376e+00 -1.19813524e-01 9.99648962e-03 4.13496703e-01 5.49663126e-01 -6.38033450e-01 -1.24370790e+00 1.05093695e-01 1.04617631e+00 -4.72389758e-01 -3.01698864e-01 -2.90122807e-01 2.51805842e-01 9.28177685e-02 1.27977955e+00 5.05047734e-04 -1.20507419e+00 6.49914086e-01 -4.34705406e-01 -2.65461117e-01 -7.20865607e-01 -1.17990124e+00 1.01139829e-01 1.69700757e-02 -4.40935045e-01 -3.12002242e-01 -2.43868396e-01 -1.08285773e+00 -5.36286235e-01 -7.08993614e-01 4.79624838e-01 5.31733930e-01 1.03618646e+00 5.50364614e-01 4.98737216e-01 1.14142990e+00 -5.03791034e-01 -5.86965501e-01 -9.51667368e-01 -5.92597604e-01 4.28528011e-01 -2.45605737e-01 -4.91271347e-01 -2.55237609e-01 -5.53556144e-01]
[10.170783996582031, 5.659460067749023]
d7a948e4-6897-4202-b127-c99de153ee4d
plant-species-classification-using-transfer
2209.03076
null
https://arxiv.org/abs/2209.03076v1
https://arxiv.org/pdf/2209.03076v1.pdf
Plant Species Classification Using Transfer Learning by Pretrained Classifier VGG-19
Deep learning is currently the most important branch of machine learning, with applications in speech recognition, computer vision, image classification, and medical imaging analysis. Plant recognition is one of the areas where image classification can be used to identify plant species through their leaves. Botanists devote a significant amount of time to recognizing plant species by personally inspecting. This paper describes a method for dissecting color images of Swedish leaves and identifying plant species. To achieve higher accuracy, the task is completed using transfer learning with the help of pre-trained classifier VGG-19. The four primary processes of classification are image preprocessing, image augmentation, feature extraction, and recognition, which are performed as part of the overall model evaluation. The VGG-19 classifier grasps the characteristics of leaves by employing pre-defined hidden layers such as convolutional layers, max pooling layers, and fully connected layers, and finally uses the soft-max layer to generate a feature representation for all plant classes. The model obtains knowledge connected to aspects of the Swedish leaf dataset, which contains fifteen tree classes, and aids in predicting the proper class of an unknown plant with an accuracy of 99.70% which is higher than previous research works reported.
['Dheeraj Kumar Agrawal', 'Bhupendra Singh Kirar', 'Thiru Siddharth']
2022-09-07
null
null
null
null
['image-augmentation']
['computer-vision']
[ 3.91042233e-01 2.31174864e-02 -2.37007231e-01 -2.29642883e-01 -6.12074733e-02 -8.56982410e-01 3.35300803e-01 3.56351942e-01 -1.36926353e-01 2.80686110e-01 -4.32507664e-01 -6.42227113e-01 2.66116671e-02 -1.12826610e+00 -2.25151762e-01 -8.65743458e-01 -6.55569807e-02 2.95154095e-01 8.53936821e-02 1.41906455e-01 1.40184239e-01 1.24075568e+00 -1.51620030e+00 5.14088869e-01 6.62690282e-01 1.23718524e+00 4.12619859e-01 9.19432700e-01 -5.53493083e-01 6.90295279e-01 -7.40565658e-01 -5.47513776e-02 8.86234120e-02 -1.33810654e-01 -1.10301197e+00 6.29053891e-01 2.76000291e-01 -2.02858284e-01 1.60471037e-01 8.35502028e-01 3.00354838e-01 -2.45080277e-01 8.56926382e-01 -1.02447701e+00 -8.96707833e-01 6.30667686e-01 -5.09960592e-01 -6.39608502e-02 -2.05162898e-01 8.04466978e-02 7.27953672e-01 -7.48239517e-01 2.11560950e-01 1.05505657e+00 5.62703431e-01 2.98043936e-01 -1.32798398e+00 -3.43788147e-01 -1.45267934e-01 3.29477966e-01 -1.33914673e+00 7.78879747e-02 6.83054209e-01 -6.77943885e-01 5.67452788e-01 3.21543962e-01 6.77949250e-01 4.39909816e-01 7.10775405e-02 7.75508523e-01 9.14514780e-01 -4.99580830e-01 2.24198535e-01 2.62213320e-01 2.57860184e-01 8.63553703e-01 1.96400985e-01 -8.63514766e-02 2.83071458e-01 1.00022718e-01 7.49453545e-01 -4.58301743e-03 -1.72532976e-01 -2.48311192e-01 -7.25218356e-01 9.78960633e-01 9.99775648e-01 6.18237019e-01 -7.03286290e-01 -4.42705125e-01 3.74959290e-01 -1.45268649e-01 2.06173569e-01 5.78719020e-01 -8.22380364e-01 4.84919488e-01 -9.11623120e-01 -4.26005900e-01 1.01870728e+00 7.24084496e-01 6.08394027e-01 1.71642780e-01 -2.00616494e-01 9.03259277e-01 5.29831611e-02 3.54741126e-01 4.43923444e-01 -8.54719400e-01 -3.33417565e-01 1.13326383e+00 -3.50471109e-01 -7.02190757e-01 -2.61069447e-01 -3.78253549e-01 -1.03527689e+00 6.42752230e-01 5.46761215e-01 1.23110548e-01 -1.35017455e+00 1.18396294e+00 1.24618307e-01 -2.73037493e-01 2.22597390e-01 3.59885484e-01 1.22015035e+00 6.95566177e-01 4.53128457e-01 1.02775685e-01 1.63495994e+00 -7.01694250e-01 -3.11155289e-01 -3.08822513e-01 5.20660639e-01 -7.14984238e-01 7.44310439e-01 3.56337816e-01 -6.81247532e-01 -7.39653111e-01 -9.65639353e-01 -6.05855174e-02 -1.06787360e+00 9.99304533e-01 8.95499170e-01 6.92157209e-01 -8.10752034e-01 5.80018818e-01 -6.20941818e-01 -6.87018514e-01 9.97438848e-01 5.01586020e-01 -6.24660671e-01 -3.09862550e-02 -4.30372149e-01 9.91855264e-01 8.52136791e-01 5.36006272e-01 -8.88957858e-01 -4.56748605e-01 -8.66616368e-01 3.91493797e-01 1.89369962e-01 -1.30789429e-01 9.35733140e-01 -1.06285727e+00 -1.51741016e+00 1.29705477e+00 1.07469574e-01 -3.57242107e-01 -2.45493665e-01 3.94672751e-01 -2.16545552e-01 1.41097218e-01 -5.48306387e-03 9.42465901e-01 9.98256564e-01 -1.17850137e+00 -6.92704320e-01 -6.15405023e-01 -1.13828249e-01 -2.93867201e-01 -5.12556255e-01 -1.91823572e-01 5.18588722e-02 -3.08529437e-01 3.63365114e-01 -8.18503380e-01 -1.35902241e-01 2.72467554e-01 -3.44122648e-01 -2.39887565e-01 1.22637212e+00 -1.00659478e+00 6.43971860e-01 -2.17728376e+00 -3.87494005e-02 1.34583324e-01 1.72696069e-01 8.22634518e-01 -3.20507914e-01 3.02195810e-02 -3.19345713e-01 1.72071353e-01 -3.15347463e-01 3.36349607e-01 -4.53978926e-01 3.60849798e-01 3.86473499e-02 1.20078951e-01 6.50519252e-01 9.85077739e-01 -6.13667011e-01 -3.74840856e-01 6.40259147e-01 4.76509184e-01 1.15282856e-01 1.53675988e-01 -2.12276742e-01 2.41030186e-01 -3.11069965e-01 1.08838058e+00 9.81026828e-01 -7.10182190e-02 1.11248612e-01 -4.23500240e-01 -2.86804080e-01 -1.09268434e-01 -7.33982801e-01 1.04636240e+00 -3.24415982e-01 6.90510750e-01 2.62489408e-01 -1.36525333e+00 1.27054322e+00 1.53125137e-01 4.53721046e-01 4.64484319e-02 3.59999090e-01 8.19913745e-02 2.17626512e-01 -4.40633833e-01 8.90356079e-02 3.06239039e-01 4.23980176e-01 -1.93735436e-02 3.34645033e-01 -3.01888227e-01 3.72044384e-01 -8.78967941e-02 7.99124599e-01 9.11255628e-02 4.36191142e-01 -3.32893044e-01 8.07753265e-01 2.17032596e-01 2.61537105e-01 3.06995839e-01 -3.13989401e-01 2.06881493e-01 3.43054175e-01 -5.98553240e-01 -7.69796431e-01 -7.62373567e-01 -2.04873800e-01 9.71871972e-01 -4.24669117e-01 6.51459172e-02 -8.12412083e-01 -6.08990550e-01 7.78579712e-02 6.59781694e-01 -8.27984869e-01 -3.67727429e-01 -1.38137594e-01 -7.45665252e-01 5.19034863e-01 7.83402383e-01 9.06478286e-01 -1.83883977e+00 -9.10987020e-01 1.78030908e-01 6.79223835e-02 -1.03040290e+00 3.02562565e-01 9.91754055e-01 -9.18867171e-01 -1.36285663e+00 -7.34483123e-01 -1.24414992e+00 8.01925778e-01 2.07024977e-01 9.37046230e-01 2.23817751e-01 -1.04810154e+00 4.80857790e-02 -3.49823743e-01 -7.56163478e-01 -5.09701729e-01 5.14098644e-01 -7.25896299e-01 -1.53382063e-01 5.81002295e-01 -3.00734967e-01 -1.80181146e-01 -1.98709294e-01 -8.05000722e-01 -1.67209968e-01 9.54931855e-01 8.36286247e-01 4.69130933e-01 3.84071141e-01 1.75717309e-01 -7.27148890e-01 2.37693936e-01 -9.66624245e-02 -6.77999318e-01 5.49691617e-01 -2.07288474e-01 -1.71581894e-01 5.84446311e-01 -5.13566971e-01 -7.75934815e-01 7.21083879e-01 3.37388157e-03 1.52364280e-02 -6.84586525e-01 5.46184838e-01 -6.08370423e-01 -3.52974564e-01 7.38622725e-01 2.59769410e-01 2.06060842e-01 -3.44472528e-01 2.73445606e-01 8.25239062e-01 6.57399058e-01 -1.11954808e-01 7.19944954e-01 3.58686715e-01 2.59471297e-01 -1.36099601e+00 -9.21440721e-01 -4.11131173e-01 -1.35980511e+00 -1.68608457e-01 1.05511761e+00 -3.57370734e-01 -9.25815403e-01 8.02738547e-01 -9.33093727e-01 -3.84635478e-01 -3.64961594e-01 3.01324934e-01 -1.70099184e-01 1.41623139e-01 -5.21064520e-01 -7.99447536e-01 -4.91603255e-01 -8.85970175e-01 8.75220537e-01 6.92362249e-01 1.10728338e-01 -9.54887569e-01 -6.43977284e-01 1.67723909e-01 2.35307321e-01 2.78170198e-01 1.13784587e+00 -5.68088472e-01 -2.23349631e-01 -6.17793918e-01 -5.92473567e-01 7.32600689e-01 5.88272333e-01 4.86457139e-01 -1.23777032e+00 -1.12415709e-01 -2.87425101e-01 -4.09756243e-01 9.54446971e-01 6.79173231e-01 1.57104933e+00 6.17332831e-02 -3.69892985e-01 4.98635650e-01 1.33265126e+00 5.46234369e-01 6.69203341e-01 2.80123502e-01 7.24875271e-01 6.89710379e-01 3.07336032e-01 1.37351111e-01 -2.57332530e-02 -7.89695904e-02 8.84547651e-01 -5.79358935e-01 -3.90443765e-02 2.65211940e-01 -1.19460739e-01 -1.24919806e-02 -2.31450405e-02 4.48085973e-03 -8.45274508e-01 5.81974864e-01 -1.33726692e+00 -8.91546428e-01 -1.49058282e-01 2.07176018e+00 5.76052308e-01 -2.75065508e-02 6.03912910e-03 7.12480545e-01 6.47504032e-01 -2.40986615e-01 -5.87741733e-01 -5.61596692e-01 -1.34755880e-01 5.76164246e-01 7.06064045e-01 3.37621123e-01 -1.51467669e+00 1.34346664e+00 6.46764612e+00 3.70734274e-01 -1.40637207e+00 -5.28689027e-01 7.01585352e-01 7.63763309e-01 4.90723491e-01 -2.89185662e-02 -6.43043339e-01 -1.15553662e-02 6.28694773e-01 4.14571524e-01 5.04398167e-01 9.95828032e-01 -5.81346266e-02 -3.01234961e-01 -7.99416661e-01 4.72041547e-01 6.58350438e-03 -1.01120675e+00 1.84110537e-01 1.41488209e-01 2.13685632e-01 -5.13824224e-01 -2.34592438e-01 3.26197714e-01 4.13005739e-01 -1.11120129e+00 1.75787330e-01 2.04455703e-01 5.19672155e-01 -7.01670110e-01 6.56589091e-01 4.02756929e-01 -1.39669812e+00 -5.23285747e-01 -7.63683617e-01 1.12750351e-01 -5.53779006e-01 5.74046612e-01 -1.17474234e+00 4.18514341e-01 9.44503844e-01 6.19977772e-01 -9.78790283e-01 9.64605272e-01 -3.44637722e-01 6.96296453e-01 -3.20121706e-01 2.93590222e-02 2.87403196e-01 -1.89865455e-01 -4.57856692e-02 1.15018582e+00 1.49520382e-01 -4.36461647e-04 3.46296519e-01 9.49100435e-01 1.21614911e-01 8.94704014e-02 -4.61993307e-01 -6.13220334e-01 2.08070382e-01 1.84448063e+00 -1.32107019e+00 -4.86649632e-01 -3.24505791e-02 1.13709760e+00 7.99688101e-02 -1.16139362e-02 -2.54177779e-01 -5.84231973e-01 1.51353240e-01 -1.59157783e-01 6.52097166e-01 -1.56769305e-01 -4.56995457e-01 -4.03280020e-01 -3.91107261e-01 -6.22074187e-01 3.50766182e-01 -8.42335343e-01 -9.63984311e-01 6.06438875e-01 -2.95142621e-01 -6.24293387e-01 -2.61423411e-03 -1.19097865e+00 -6.28237128e-01 1.06343675e+00 -1.19659269e+00 -1.81504178e+00 -8.52865398e-01 3.35068703e-01 4.38251853e-01 -2.92806566e-01 1.47142625e+00 4.71016131e-02 -5.89580834e-01 5.01872301e-02 -1.46666437e-01 4.58876431e-01 2.61078179e-01 -1.34266269e+00 5.57268644e-03 6.29274130e-01 1.72363937e-01 -1.07206553e-02 8.84127989e-02 -5.05460203e-01 -9.44620073e-01 -1.26570272e+00 9.38587666e-01 4.14162036e-03 2.96607971e-01 -1.23201430e-01 -9.56614614e-01 4.83738929e-01 8.52820054e-02 1.04438879e-01 7.17953146e-01 -2.73040593e-01 -1.10991143e-01 -5.27649373e-02 -1.33726263e+00 3.62718433e-01 3.74319583e-01 -3.20302159e-01 -1.34637251e-01 3.64432633e-01 8.98324773e-02 4.27780347e-03 -8.75133812e-01 4.82600331e-01 5.40622175e-01 -4.13103729e-01 1.10541022e+00 -6.16891265e-01 3.07166189e-01 -2.80428618e-01 -7.33010694e-02 -1.25102675e+00 -8.86207998e-01 9.01384354e-02 2.84286946e-01 1.35016823e+00 3.70759130e-01 -3.37746054e-01 8.49413157e-01 2.07970992e-01 7.82955438e-02 -3.02146167e-01 -1.14241719e-01 -4.96638387e-01 8.23879614e-02 1.33008972e-01 4.57370162e-01 8.20709288e-01 -6.10432506e-01 2.24777430e-01 1.95304602e-01 2.79019505e-01 4.93865758e-01 2.59798676e-01 4.43355531e-01 -1.87948537e+00 2.46250764e-01 -6.72866702e-01 -5.43833733e-01 -3.37182492e-01 2.90814966e-01 -1.06671357e+00 5.42536899e-02 -1.79946721e+00 6.95114136e-02 -1.80639908e-01 -1.51356338e-02 9.97449398e-01 -1.50492236e-01 2.88003296e-01 9.45236236e-02 -7.90544152e-02 3.73132378e-01 7.59918839e-02 1.22688985e+00 -5.13590634e-01 -2.68988550e-01 4.05578166e-01 -7.74356306e-01 8.17274749e-01 1.25632012e+00 -1.57993987e-01 -1.94020510e-01 -2.57648081e-01 -6.35881424e-01 -6.58451796e-01 6.74549699e-01 -9.23873007e-01 -7.77283087e-02 -1.14062823e-01 1.08483160e+00 -8.75412762e-01 4.10195142e-01 -1.22035789e+00 -1.00343838e-01 9.38269019e-01 -2.16695815e-01 -4.40693378e-01 4.90067691e-01 -1.92784250e-01 -6.12156931e-03 -7.94837356e-01 1.06491649e+00 -4.91169512e-01 -1.07614625e+00 2.16182426e-01 -7.73087442e-01 -7.18012929e-01 1.21845448e+00 -4.01685655e-01 -1.79055892e-02 -8.43428448e-02 -9.47342575e-01 2.00727750e-02 -1.81574509e-01 2.42259324e-01 3.56273651e-01 -1.02291179e+00 -6.50613964e-01 4.25127298e-01 1.25631481e-01 -1.05617620e-01 -1.39996454e-01 3.09015751e-01 -8.41937721e-01 3.95499289e-01 -7.19733596e-01 -7.12638378e-01 -1.55754554e+00 7.10189819e-01 3.60408247e-01 -7.67462403e-02 -5.02905011e-01 8.12930286e-01 -7.43198693e-02 -5.23603320e-01 4.44947481e-01 -3.74270529e-01 -8.94616365e-01 1.87056556e-01 3.57166499e-01 2.13526428e-01 1.88833848e-01 -5.59172094e-01 -3.21208298e-01 3.96405339e-01 2.18204819e-02 4.26945955e-01 1.40787876e+00 4.69503999e-01 -2.75922596e-01 3.03881943e-01 8.51041436e-01 -5.78054249e-01 -8.67980301e-01 -6.58329204e-02 3.41324568e-01 -1.03961013e-01 3.31140250e-01 -1.14197648e+00 -1.30876601e+00 1.12021065e+00 9.25405562e-01 5.78204811e-01 1.33490491e+00 -1.28275797e-01 2.80141503e-01 5.52074075e-01 -1.78272888e-01 -8.96587074e-01 -2.87839681e-01 4.69428450e-01 8.64132702e-01 -1.35018647e+00 -2.41040349e-01 -7.55485833e-01 -2.82680213e-01 1.39506161e+00 7.41927087e-01 -4.38252948e-02 9.51579690e-01 5.48040092e-01 1.58388034e-01 -9.74415317e-02 -3.48058999e-01 -6.97485209e-01 3.26759189e-01 1.03812635e+00 7.27382004e-01 3.43496382e-01 -4.78278473e-02 4.84961301e-01 -6.90762550e-02 5.54739088e-02 1.52204409e-01 1.18730795e+00 -8.04443479e-01 -1.03172421e+00 -5.35923898e-01 5.57264209e-01 -3.80311012e-01 -1.15292795e-01 -1.12516570e+00 7.01698720e-01 2.91178167e-01 8.55120778e-01 2.49526519e-02 -2.65551925e-01 2.36912280e-01 1.86195090e-01 6.53519809e-01 -7.18121886e-01 -7.24433959e-01 -1.32413715e-01 -3.58097613e-01 -7.43034249e-03 -2.82459110e-01 -2.95993030e-01 -1.09615958e+00 -1.94600403e-01 -4.39592540e-01 -7.98962712e-02 1.01332438e+00 8.50932777e-01 2.49057580e-02 8.74562979e-01 6.60169125e-01 -9.87253904e-01 -3.73097122e-01 -1.26988208e+00 -5.79941034e-01 3.02255675e-02 -1.59485877e-01 -3.87690455e-01 -1.17106810e-01 4.46049780e-01]
[9.186015129089355, -1.5209834575653076]
185f468a-4f55-4dbf-93f9-ec32124abb63
exploiting-unlabeled-data-with-vision-and
2207.08954
null
https://arxiv.org/abs/2207.08954v1
https://arxiv.org/pdf/2207.08954v1.pdf
Exploiting Unlabeled Data with Vision and Language Models for Object Detection
Building robust and generic object detection frameworks requires scaling to larger label spaces and bigger training datasets. However, it is prohibitively costly to acquire annotations for thousands of categories at a large scale. We propose a novel method that leverages the rich semantics available in recent vision and language models to localize and classify objects in unlabeled images, effectively generating pseudo labels for object detection. Starting with a generic and class-agnostic region proposal mechanism, we use vision and language models to categorize each region of an image into any object category that is required for downstream tasks. We demonstrate the value of the generated pseudo labels in two specific tasks, open-vocabulary detection, where a model needs to generalize to unseen object categories, and semi-supervised object detection, where additional unlabeled images can be used to improve the model. Our empirical evaluation shows the effectiveness of the pseudo labels in both tasks, where we outperform competitive baselines and achieve a novel state-of-the-art for open-vocabulary object detection. Our code is available at https://github.com/xiaofeng94/VL-PLM.
['Dimitris Metaxas', 'Manmohan Chandraker', 'Anastasis Stathopoulos', 'Vijay Kumar B. G', 'Long Zhao', 'Samuel Schulter', 'Zhixing Zhang', 'Shiyu Zhao']
2022-07-18
null
null
null
null
['open-vocabulary-object-detection', 'semi-supervised-object-detection']
['computer-vision', 'computer-vision']
[ 1.34343386e-01 -3.49050108e-03 -2.07759514e-01 -4.93074417e-01 -1.10288203e+00 -9.72464979e-01 6.63429141e-01 1.53415695e-01 -4.99871224e-01 3.57400715e-01 -9.32619870e-02 -2.72657752e-01 6.16959751e-01 -5.02986014e-01 -8.05842578e-01 -4.53866035e-01 2.31769785e-01 5.90028763e-01 6.05294406e-01 2.03118265e-01 7.10404292e-02 3.54755670e-01 -1.67076445e+00 3.83249402e-01 3.45519006e-01 1.11503458e+00 4.44057912e-01 6.57637000e-01 -1.98948801e-01 5.52501678e-01 -2.59952456e-01 -3.32377195e-01 5.30760229e-01 -1.08342916e-02 -7.89178610e-01 3.72107863e-01 9.99812365e-01 -6.06760621e-01 -2.23704398e-01 1.08463395e+00 2.42877319e-01 -2.40930691e-02 8.48701119e-01 -1.19366181e+00 -9.28610325e-01 3.94131511e-01 -6.29198432e-01 2.63578266e-01 -1.33589312e-01 3.90103489e-01 1.21906888e+00 -1.28356314e+00 6.19004190e-01 1.22716510e+00 3.99354607e-01 7.63053358e-01 -1.34577668e+00 -9.59137261e-01 4.79023784e-01 -6.71605095e-02 -1.62645376e+00 -5.12226403e-01 2.82690525e-01 -6.49485290e-01 7.11647213e-01 -5.34059890e-02 2.75769085e-01 9.87436950e-01 -4.18472618e-01 9.43460166e-01 9.45146620e-01 -3.50755572e-01 1.79391176e-01 3.15939933e-01 3.77393752e-01 7.83047557e-01 4.87629116e-01 4.02582251e-02 -2.09219381e-01 -3.77175957e-02 6.29955530e-01 2.82196879e-01 -3.64937074e-02 -6.90990627e-01 -1.27845025e+00 9.53050017e-01 8.32821965e-01 -7.74818584e-02 -1.66481510e-01 4.24102098e-01 3.67796034e-01 -3.15102339e-02 5.80308437e-01 4.22364920e-01 -5.34194648e-01 5.67849815e-01 -9.13565278e-01 6.05499744e-02 5.35450697e-01 1.30935681e+00 8.63331735e-01 -3.86336327e-01 -3.94089937e-01 8.04968238e-01 4.57032055e-01 6.62982285e-01 2.19525397e-01 -1.01606274e+00 2.70708114e-01 6.40871167e-01 2.51909167e-01 -3.11105818e-01 -2.32573152e-01 -5.07939339e-01 -3.12641054e-01 6.89717606e-02 5.72970390e-01 1.09196492e-01 -1.40314150e+00 1.73019350e+00 6.30319536e-01 2.88009495e-01 -3.31669524e-02 9.66518462e-01 1.02358246e+00 5.01779199e-01 4.12693471e-01 2.43892282e-01 1.66603231e+00 -1.34320939e+00 -1.92227811e-01 -5.67345083e-01 6.74616456e-01 -6.72017157e-01 1.04394114e+00 -2.99875215e-02 -7.12807953e-01 -5.43271184e-01 -7.46384621e-01 -3.12113166e-01 -4.60934550e-01 4.59662378e-01 6.75545037e-01 3.21722358e-01 -1.01115322e+00 -5.70180044e-02 -7.56508470e-01 -5.59744895e-01 9.58947062e-01 1.92816705e-01 -2.84818053e-01 -3.24996263e-01 -6.42721236e-01 5.98104239e-01 5.50496697e-01 -2.45378792e-01 -1.43792260e+00 -5.19707918e-01 -8.59005928e-01 -1.09899968e-01 6.81709707e-01 -5.95717251e-01 1.54869485e+00 -9.98532712e-01 -6.61454260e-01 1.35385358e+00 -1.57867178e-01 -4.17841673e-01 4.46934819e-01 -1.10486776e-01 -1.15414280e-02 2.56212145e-01 5.01365244e-01 1.39714336e+00 9.29479122e-01 -1.28876781e+00 -1.07141626e+00 -4.09733355e-01 2.46590942e-01 1.00972272e-01 -2.94349194e-01 6.45767599e-02 -8.92646670e-01 -5.36283076e-01 1.31729037e-01 -1.06061029e+00 -2.71636158e-01 5.45346200e-01 -4.49818581e-01 -5.63307941e-01 7.32632399e-01 -4.19385582e-01 7.18277156e-01 -2.06664324e+00 -2.57443041e-01 -6.81221113e-02 4.57649440e-01 2.36401364e-01 -4.56045806e-01 -1.09704576e-01 2.91117162e-01 9.90863889e-02 -2.45388187e-02 -5.20231545e-01 -1.03157483e-01 7.25805685e-02 -5.96473932e-01 4.90522712e-01 4.40558821e-01 1.23814154e+00 -9.96188045e-01 -5.78019321e-01 5.09467497e-02 2.58147448e-01 -4.55506116e-01 1.33566916e-01 -5.45968890e-01 3.47055554e-01 -5.49351990e-01 9.71213520e-01 4.09605652e-01 -7.51508474e-01 -1.19139105e-01 -1.50658563e-01 1.24467619e-01 1.74945205e-01 -1.02285373e+00 1.55618072e+00 -3.37188751e-01 6.01142108e-01 4.83204648e-02 -8.28677416e-01 6.46455050e-01 4.71920595e-02 -6.68221265e-02 -2.13775232e-01 1.78065076e-01 2.43266165e-01 -1.93365917e-01 -3.06742907e-01 3.08358461e-01 7.86603242e-02 -1.35637686e-01 4.64422673e-01 2.05214664e-01 -1.47356600e-01 3.06398541e-01 4.35230166e-01 9.48183596e-01 1.37400344e-01 3.60658109e-01 -1.07612982e-01 3.54959488e-01 2.33108133e-01 3.87698650e-01 1.04245913e+00 -4.09848213e-01 6.23180687e-01 1.93459820e-02 -4.30724412e-01 -1.16349161e+00 -1.16713023e+00 -1.98299438e-01 1.57800841e+00 2.76118308e-01 -1.20687343e-01 -6.07790709e-01 -9.42691624e-01 2.92388022e-01 5.20351350e-01 -5.80335498e-01 6.66905344e-02 -1.95922375e-01 -3.33910942e-01 4.03320879e-01 7.60945380e-01 1.66725382e-01 -1.01305461e+00 -3.46455961e-01 -4.51962836e-02 -1.68622136e-01 -1.35785639e+00 -7.41755426e-01 1.76537931e-01 -6.73627377e-01 -1.06943548e+00 -7.09586442e-01 -1.12005615e+00 8.87847543e-01 7.81612098e-01 1.09601355e+00 1.77898914e-01 -7.37654328e-01 4.94352967e-01 -3.08771878e-01 -6.44502103e-01 -4.39536005e-01 2.27872301e-02 1.84091888e-02 1.14153116e-03 6.32173538e-01 8.93018767e-03 -7.59833336e-01 3.77811551e-01 -7.02429295e-01 7.15048611e-02 6.35893345e-01 6.93227172e-01 8.37571621e-01 -5.83955228e-01 6.23234212e-01 -8.36812317e-01 5.40209329e-03 -4.74570394e-01 -8.75301242e-01 2.42899746e-01 -3.38695556e-01 1.46109723e-02 2.65987873e-01 -7.60819614e-01 -6.74865365e-01 5.19339383e-01 2.00975522e-01 -6.83317900e-01 -3.06638449e-01 -9.39228833e-02 7.08960518e-02 -1.22378081e-01 8.18865299e-01 2.41188928e-01 -1.21164382e-01 -4.54599440e-01 9.90172923e-01 8.92835617e-01 5.77518702e-01 -3.56595129e-01 8.79015088e-01 8.55436385e-01 -4.46232200e-01 -4.73351091e-01 -1.39671552e+00 -1.07112598e+00 -6.58703685e-01 4.63263951e-02 9.38664079e-01 -1.50356007e+00 -2.51634747e-01 1.87283784e-01 -1.10619795e+00 -4.07830536e-01 -3.12330782e-01 2.71436334e-01 -3.88500154e-01 1.96470797e-01 -5.68508983e-01 -5.92987955e-01 -4.84244406e-01 -1.07650816e+00 1.59199929e+00 1.93794325e-01 1.06079631e-01 -6.15954340e-01 -3.39639157e-01 6.05518043e-01 2.34185487e-01 -2.75670499e-01 4.76896971e-01 -9.92939889e-01 -9.00683522e-01 -4.64642733e-01 -6.95273995e-01 3.47012699e-01 -4.08922546e-02 -2.92100161e-01 -1.20224953e+00 -4.04006481e-01 -4.02816623e-01 -8.64779115e-01 1.24143517e+00 2.08163187e-01 1.25713408e+00 -1.91245794e-01 -7.68391132e-01 5.00065565e-01 1.24770451e+00 -3.58055264e-01 3.48466821e-02 1.56571254e-01 8.20535183e-01 6.15321517e-01 7.63466835e-01 1.79654121e-01 5.19548595e-01 5.90652585e-01 4.04130727e-01 -1.63158834e-01 -4.29362833e-01 -2.91047603e-01 8.99533778e-02 1.50677487e-01 3.24225605e-01 -1.70062065e-01 -9.96776521e-01 9.03817952e-01 -1.74314666e+00 -6.93183661e-01 4.64414284e-02 2.12905025e+00 8.96690667e-01 1.36530399e-01 1.69793501e-01 -5.61903715e-01 1.00019169e+00 -1.67570353e-01 -8.64319980e-01 2.37139896e-01 1.77175626e-01 -6.38957098e-02 7.58073747e-01 2.33330995e-01 -1.57479823e+00 1.43469822e+00 5.91168070e+00 7.91570067e-01 -1.05367124e+00 4.32647228e-01 6.99359000e-01 -2.71379352e-01 7.55895004e-02 3.95788699e-02 -1.23498666e+00 1.98162541e-01 6.23913050e-01 1.48382366e-01 1.39417350e-01 1.22795415e+00 -6.56657200e-03 1.01317719e-01 -1.21352649e+00 9.87242281e-01 1.48325011e-01 -1.32343829e+00 2.71154076e-01 -2.40445491e-02 7.84471273e-01 6.57642007e-01 -1.37757078e-01 4.22120631e-01 5.55445731e-01 -7.32125044e-01 9.02677000e-01 1.27802208e-01 9.48073149e-01 -9.48655084e-02 4.71907288e-01 3.81962717e-01 -1.33599854e+00 -2.99096853e-01 -5.75770020e-01 1.15799986e-01 5.71380444e-02 1.98639423e-01 -9.82833385e-01 -1.27695858e-01 6.73041582e-01 5.94508469e-01 -8.40104818e-01 1.08869350e+00 -5.31974673e-01 6.77812517e-01 -4.43656534e-01 1.33342976e-02 3.22648793e-01 1.94168866e-01 3.08658183e-01 1.20588923e+00 1.00053689e-02 1.24787770e-01 7.19940543e-01 1.05231833e+00 -4.77154881e-01 5.43785691e-02 -5.82738876e-01 -4.70457301e-02 7.30590701e-01 1.67910349e+00 -9.63665783e-01 -7.20282435e-01 -6.02245569e-01 8.54431987e-01 5.51590860e-01 4.02325779e-01 -8.29412460e-01 -9.03425589e-02 5.30826628e-01 1.44483745e-01 7.11155355e-01 -9.85796303e-02 -8.16667601e-02 -1.24069667e+00 -8.19082782e-02 -5.13522089e-01 4.34277028e-01 -8.26141775e-01 -1.45077193e+00 3.23446780e-01 -9.10390988e-02 -1.09828770e+00 -8.20231736e-02 -7.87295580e-01 -3.05236965e-01 7.32753754e-01 -1.58674586e+00 -1.66449666e+00 -4.96878743e-01 4.10819054e-01 8.12350690e-01 -9.16043445e-02 5.95499456e-01 2.10922807e-01 -3.45379502e-01 5.62934101e-01 6.52767047e-02 4.79716182e-01 8.15213144e-01 -1.17719162e+00 6.23477399e-01 8.34047437e-01 4.83033478e-01 5.72457135e-01 3.45945776e-01 -5.48679769e-01 -1.14941883e+00 -1.66365576e+00 6.41702056e-01 -9.08134341e-01 8.04190159e-01 -7.88084149e-01 -7.02812791e-01 9.10007060e-01 -3.90536070e-01 7.21509218e-01 4.42591280e-01 2.87766587e-02 -9.07161057e-01 7.45080635e-02 -9.66602564e-01 4.68637139e-01 1.21478546e+00 -6.09803617e-01 -5.77062964e-01 7.96188056e-01 9.46830213e-01 -2.25358427e-01 -3.77307117e-01 2.90565461e-01 4.34920371e-01 -3.14447045e-01 1.09925377e+00 -6.80866420e-01 1.08117223e-01 -6.66622639e-01 -2.77706265e-01 -7.18621671e-01 -3.28278512e-01 7.14093726e-03 -7.02231675e-02 1.09046757e+00 5.17289817e-01 -5.84769428e-01 5.73040664e-01 4.99549210e-01 3.28338034e-02 -5.69299579e-01 -6.79429412e-01 -8.83754015e-01 -2.27701273e-02 -4.54715401e-01 2.57691383e-01 6.69591904e-01 -5.42247236e-01 6.14087582e-01 5.80457598e-02 4.96704608e-01 8.57674837e-01 4.09963310e-01 8.87304127e-01 -1.15862679e+00 -2.13203460e-01 -2.93485135e-01 -5.51907659e-01 -1.23695242e+00 2.40263090e-01 -1.18598998e+00 3.29354256e-01 -1.58528161e+00 6.86863959e-01 -7.37952769e-01 -3.83081943e-01 9.20965195e-01 -2.26964623e-01 9.58029807e-01 3.31275970e-01 6.21055841e-01 -1.18687475e+00 2.95858800e-01 9.56165373e-01 -4.39792335e-01 -9.30404011e-03 -3.19920219e-02 -9.19247091e-01 7.76085973e-01 5.94397724e-01 -5.10704458e-01 -3.10868561e-01 -4.41878110e-01 -3.18593442e-01 -5.48715115e-01 7.40719855e-01 -8.25953245e-01 1.60676569e-01 9.14613716e-03 3.01673681e-01 -4.80692357e-01 2.76347667e-01 -5.50452232e-01 -3.50524992e-01 3.51836383e-01 -4.57160950e-01 -4.85974133e-01 2.32167929e-01 8.60982120e-01 1.22652553e-01 -1.74074888e-01 9.23182130e-01 -1.89013869e-01 -1.18858647e+00 5.22559464e-01 5.25994375e-02 1.67966425e-01 1.28259993e+00 -7.39203542e-02 -4.19291317e-01 -1.03453115e-01 -6.85714126e-01 5.33259749e-01 6.88831568e-01 7.13318050e-01 4.36930746e-01 -1.02266371e+00 -7.16712415e-01 1.89870205e-02 7.93584228e-01 2.32945278e-01 -5.50980456e-02 5.18178165e-01 -3.27834010e-01 4.10255402e-01 2.28468776e-01 -9.73622620e-01 -1.36174464e+00 9.28518653e-01 3.45198214e-01 1.46505535e-01 -6.02839410e-01 1.21694100e+00 9.15849864e-01 -5.02382278e-01 3.83769274e-01 -3.49294692e-01 -6.56521246e-02 -6.15711734e-02 6.77878678e-01 -1.13301352e-01 -1.49617091e-01 -6.44596636e-01 -3.60414773e-01 5.61966062e-01 -4.33876693e-01 1.33799121e-01 1.03844845e+00 -2.85466641e-01 7.27156177e-02 3.21400285e-01 1.02882671e+00 -3.60152483e-01 -1.52812147e+00 -6.51883662e-01 -1.70448106e-02 -2.97072619e-01 9.26968455e-02 -7.91845441e-01 -8.09831560e-01 8.05476069e-01 8.43090594e-01 -2.24128470e-01 7.32990265e-01 7.55473495e-01 4.97688383e-01 5.33135593e-01 6.24771237e-01 -8.29756558e-01 2.02789500e-01 4.33968991e-01 5.94622076e-01 -1.78212976e+00 -1.31017566e-01 -5.86516023e-01 -5.94244063e-01 7.61432946e-01 7.07258046e-01 -8.33474174e-02 5.56887269e-01 4.07001860e-02 3.19766045e-01 -2.03500032e-01 -6.92393363e-01 -6.82595909e-01 5.28227389e-01 4.55144435e-01 1.94454104e-01 2.69737154e-01 6.41750023e-02 2.94529349e-01 2.81394005e-01 -8.59197453e-02 2.75716305e-01 8.30495894e-01 -8.09231520e-01 -8.90109479e-01 -3.56628090e-01 7.50016272e-01 -2.90545553e-01 -3.71555954e-01 -3.00201565e-01 5.90062916e-01 2.21251503e-01 8.67321372e-01 1.41520515e-01 9.39359292e-02 -7.51926675e-02 1.13020808e-01 2.70848453e-01 -1.31407118e+00 -9.61796418e-02 1.23532712e-01 -1.38841659e-01 -5.32704890e-01 -3.92847031e-01 -6.35940611e-01 -1.34184813e+00 3.94710809e-01 -7.08377779e-01 -2.17361614e-01 6.93633080e-01 8.70518386e-01 5.22940755e-01 1.71413094e-01 1.72312081e-01 -9.26526427e-01 -7.35957146e-01 -1.04616964e+00 -4.17330444e-01 5.59337199e-01 3.97503674e-01 -8.13068151e-01 -2.58569092e-01 3.60646009e-01]
[9.625821113586426, 1.469839334487915]
dce2b4e4-45e6-4f7d-9353-9172e44ba0f0
learning-to-stop-a-simple-yet-effective
2009.13112
null
https://arxiv.org/abs/2009.13112v3
https://arxiv.org/pdf/2009.13112v3.pdf
Learning to Stop: A Simple yet Effective Approach to Urban Vision-Language Navigation
Vision-and-Language Navigation (VLN) is a natural language grounding task where an agent learns to follow language instructions and navigate to specified destinations in real-world environments. A key challenge is to recognize and stop at the correct location, especially for complicated outdoor environments. Existing methods treat the STOP action equally as other actions, which results in undesirable behaviors that the agent often fails to stop at the destination even though it might be on the right path. Therefore, we propose Learning to Stop (L2Stop), a simple yet effective policy module that differentiates STOP and other actions. Our approach achieves the new state of the art on a challenging urban VLN dataset Touchdown, outperforming the baseline by 6.89% (absolute improvement) on Success weighted by Edit Distance (SED).
['William Yang Wang', 'Xin Eric Wang', 'Jiannan Xiang']
2020-09-28
null
https://aclanthology.org/2020.findings-emnlp.62
https://aclanthology.org/2020.findings-emnlp.62.pdf
findings-of-the-association-for-computational
['vision-language-navigation']
['computer-vision']
[ 1.82227567e-02 -2.72497535e-01 -1.37505203e-01 -3.81071478e-01 -5.63358426e-01 -9.40276146e-01 9.06152487e-01 4.68067918e-03 -1.02414417e+00 7.96620488e-01 2.31352210e-01 -6.96472704e-01 2.67798215e-01 -6.24069989e-01 -6.10885978e-01 -4.54694688e-01 -4.02385928e-02 5.53902447e-01 4.53218520e-01 -5.61579943e-01 4.12966400e-01 4.51941937e-01 -1.41562581e+00 -2.98191875e-01 9.46364403e-01 4.33114409e-01 5.49822986e-01 7.79701769e-01 -2.33379558e-01 9.91856158e-01 -2.30286211e-01 -2.82062013e-02 4.68461782e-01 -3.40205640e-01 -7.52713501e-01 -4.70550537e-01 7.06961095e-01 -6.04056358e-01 -3.71312052e-01 1.19719362e+00 2.64406174e-01 6.86324835e-01 5.88116467e-01 -1.45195365e+00 -4.23016489e-01 3.76430690e-01 -2.89841890e-01 3.00387423e-02 5.55701494e-01 5.91299295e-01 7.98705220e-01 -6.26164377e-01 6.71794415e-01 1.23074603e+00 5.04677474e-01 8.23474765e-01 -9.80802298e-01 -3.39515865e-01 7.05383897e-01 8.71656910e-02 -1.08377993e+00 -5.37697375e-01 6.23427853e-02 -3.94000947e-01 1.30622840e+00 -2.23416030e-01 3.67268145e-01 1.27166259e+00 1.80544466e-01 7.96597540e-01 7.18663692e-01 3.41593437e-02 3.67629588e-01 -5.55219650e-01 -1.25274166e-01 1.08915555e+00 3.66595089e-01 4.43554908e-01 -4.96145219e-01 1.54266044e-01 4.41805393e-01 -9.42544490e-02 -1.43532470e-01 -7.09378541e-01 -1.72303510e+00 7.04069436e-01 6.04663730e-01 -5.91190085e-02 -4.55470145e-01 5.57427406e-01 3.15579444e-01 2.43400484e-01 -3.92859161e-01 4.45202768e-01 -4.70443457e-01 -7.59835601e-01 -4.82436955e-01 5.04136086e-01 6.39250875e-01 1.15819740e+00 6.44510508e-01 1.86801225e-01 -3.02200168e-01 4.00933057e-01 3.26652259e-01 9.45540547e-01 1.46687537e-01 -1.40114486e+00 7.05979526e-01 4.52974588e-01 5.90223074e-01 -7.33877420e-01 -5.10282278e-01 -9.08258483e-02 -5.01398146e-01 8.93838227e-01 7.66967893e-01 -4.42002803e-01 -1.22612119e+00 1.96620989e+00 1.09344974e-01 1.26927465e-01 4.28711504e-01 8.70094597e-01 7.42133498e-01 5.26422739e-01 2.69990802e-01 3.65525335e-01 8.47434640e-01 -1.65245938e+00 -5.66479981e-01 -1.00999832e+00 1.00968683e+00 -4.54068869e-01 1.26416016e+00 1.95572555e-01 -5.07611573e-01 -3.26373875e-01 -8.66224706e-01 -1.35157272e-01 -4.71449107e-01 7.21188262e-02 6.68288887e-01 1.84017017e-01 -1.28301442e+00 2.65825659e-01 -1.11664248e+00 -5.81791461e-01 1.02747872e-01 1.31905302e-01 -6.73055589e-01 -1.93050802e-01 -8.89724076e-01 1.02760029e+00 1.59746572e-01 6.27086498e-04 -1.31629205e+00 -3.07396770e-01 -1.26023507e+00 -3.92566830e-01 6.07394338e-01 -6.23902500e-01 1.64307022e+00 -5.63100159e-01 -1.62126100e+00 6.97268903e-01 -4.16622549e-01 -7.34554648e-01 9.71546531e-01 -4.78884131e-01 -2.06167549e-01 -3.11308920e-01 7.24825561e-01 9.52219069e-01 3.35312873e-01 -1.00633895e+00 -1.30047059e+00 -5.65073006e-02 3.47360641e-01 4.86532897e-01 6.99483454e-01 -5.66510022e-01 -6.81745768e-01 -3.37501258e-01 1.71955243e-01 -1.09844720e+00 -6.05058312e-01 2.47495815e-01 -3.20493132e-01 -1.93271235e-01 7.18643367e-01 -6.11380041e-01 8.46236050e-01 -1.97338080e+00 -3.79046723e-02 1.55624282e-02 1.39696449e-01 1.94206998e-01 -3.56795967e-01 4.94334757e-01 5.82280219e-01 -1.53300673e-01 -3.07593465e-01 -4.68519717e-01 1.19819723e-01 3.36246312e-01 -3.91794384e-01 3.66315335e-01 -4.71329868e-01 9.72670197e-01 -1.45269299e+00 -9.01691709e-03 4.76805449e-01 1.55827329e-01 -5.62197626e-01 -6.08824268e-02 -3.95271122e-01 7.00583398e-01 -4.95087475e-01 5.87387264e-01 4.28542703e-01 2.50945866e-01 -1.03187799e-01 5.30366659e-01 -4.80153650e-01 4.70642656e-01 -1.06665850e+00 2.00817847e+00 -5.71256459e-01 9.98883307e-01 6.24548905e-02 -4.58402753e-01 6.96648598e-01 -1.77991748e-01 4.19935910e-03 -1.03840566e+00 -1.76391691e-01 4.43351358e-01 -1.16144225e-01 -3.43359530e-01 6.25175655e-01 4.34296548e-01 -2.91673154e-01 1.25176772e-01 -5.02223730e-01 -8.34921226e-02 2.07577959e-01 2.12849110e-01 1.38324106e+00 3.91335815e-01 4.44142222e-01 -7.66722336e-02 4.42699671e-01 4.46637571e-01 4.83527184e-01 1.48596120e+00 -7.66178668e-01 3.54400128e-01 2.21644104e-01 -4.85193223e-01 -5.46511829e-01 -1.22484469e+00 7.34690368e-01 1.28120244e+00 7.36917078e-01 -2.59652317e-01 -5.96515715e-01 -1.03213513e+00 -7.15303496e-02 1.22126079e+00 -4.70893472e-01 -1.35937050e-01 -7.46915221e-01 2.78597713e-01 6.20107234e-01 5.88757813e-01 9.13297296e-01 -1.23769057e+00 -1.05742168e+00 1.83565542e-01 -4.51373637e-01 -1.25268602e+00 -7.50102401e-01 2.88028300e-01 -4.36559975e-01 -1.04566276e+00 -4.60558146e-01 -1.04530382e+00 5.50178945e-01 6.12739444e-01 1.03571522e+00 9.47518274e-02 3.22370172e-01 4.65374649e-01 -2.98024207e-01 -6.34989515e-02 -3.55302185e-01 1.36991754e-01 2.57339895e-01 -4.31085527e-01 4.42021310e-01 -2.98489705e-02 -4.86685008e-01 2.61515170e-01 -1.48390159e-01 1.13127910e-01 3.24248552e-01 6.35358274e-01 5.73762119e-01 -1.77352637e-01 1.20660357e-01 -3.39074463e-01 6.68270469e-01 -9.08546001e-02 -1.11078441e+00 1.84118479e-01 -5.46747327e-01 3.31700027e-01 7.56479084e-01 -1.39227167e-01 -7.28004992e-01 2.20844358e-01 -1.36546105e-01 1.91266447e-01 -4.61977273e-01 2.52831250e-01 -1.24093801e-01 -2.07717761e-01 5.06800413e-01 4.55499530e-01 -1.35581613e-01 -1.66280538e-01 5.54330349e-01 2.50997096e-01 8.36872518e-01 -2.46339679e-01 7.94597864e-01 6.26864135e-01 -6.52420372e-02 -7.00273871e-01 -5.48506975e-01 -4.75356966e-01 -3.63485128e-01 -6.08008131e-02 9.52876508e-01 -1.05660689e+00 -1.01791394e+00 3.56796205e-01 -1.15049517e+00 -1.05955398e+00 -1.18324053e-04 4.52096641e-01 -7.07575202e-01 2.37797931e-01 -4.16899398e-02 -6.02840662e-01 6.58171065e-03 -1.31796396e+00 9.64617133e-01 4.74334806e-01 -1.89670712e-01 -8.38404715e-01 8.27378780e-02 4.53662984e-02 5.40975451e-01 1.38555482e-01 5.86238682e-01 -5.82374692e-01 -7.05276608e-01 1.01581477e-01 -8.89706686e-02 -1.93095475e-01 2.85442442e-01 -4.32004750e-01 -1.96710810e-01 -3.60831201e-01 -5.71943521e-01 -1.97058246e-01 1.06319404e+00 3.82907867e-01 4.90360022e-01 -8.65161940e-02 -5.55339098e-01 6.73703492e-01 1.32285404e+00 6.09659851e-01 5.19947588e-01 8.66119921e-01 6.50283515e-01 2.96596348e-01 8.55347216e-01 1.50103122e-01 7.85655200e-01 6.76448941e-01 8.08050156e-01 6.06694631e-02 -1.18660495e-01 -6.89996719e-01 7.78674960e-01 -6.63537458e-02 1.82825595e-01 -5.05017996e-01 -1.17799866e+00 7.59342611e-01 -2.17613029e+00 -1.04443836e+00 -3.16406973e-02 2.36235738e+00 3.24713707e-01 3.97917181e-01 -1.19319983e-01 -5.55972695e-01 3.66680235e-01 2.99755961e-01 -9.24661875e-01 -3.25200826e-01 -6.87227026e-02 -4.48448837e-01 8.58623743e-01 1.18353117e+00 -1.24900925e+00 1.73273540e+00 6.36624575e+00 3.81748050e-01 -1.18074751e+00 -1.84682220e-01 9.49876830e-02 -1.12926811e-02 -7.51791373e-02 -1.72948446e-02 -9.56587315e-01 2.74110585e-01 4.76449579e-01 3.55638191e-02 8.12548399e-01 1.09751236e+00 5.85373044e-01 -6.42646968e-01 -1.16092539e+00 1.08205926e+00 -1.17080122e-01 -1.15668714e+00 -2.40648650e-02 -2.18779057e-01 6.81736290e-01 7.18121707e-01 1.13039084e-01 6.70273364e-01 1.04756176e+00 -1.17952979e+00 9.48511600e-01 4.26640272e-01 4.89571899e-01 -6.71469510e-01 5.43682694e-01 5.94687939e-01 -1.29424942e+00 -6.20258227e-02 6.89338967e-02 -2.78899521e-01 5.06871343e-01 -3.05398047e-01 -9.73832667e-01 1.49047986e-01 7.02689171e-01 6.90779865e-01 -4.19876993e-01 1.18757021e+00 -7.68658578e-01 1.84857696e-01 -3.22490305e-01 -3.67634147e-01 1.00016594e+00 -4.21483546e-01 8.55056107e-01 9.33309793e-01 3.35478127e-01 -1.48548439e-01 5.81629872e-01 4.76163894e-01 2.50229031e-01 -3.11619103e-01 -1.06252074e+00 1.00128360e-01 3.70948255e-01 6.79220319e-01 -7.46789277e-01 -2.33737752e-01 -2.54667014e-01 1.40645289e+00 2.94065416e-01 7.04587281e-01 -9.95462656e-01 -6.05642140e-01 1.28100109e+00 -2.05821186e-01 4.12871361e-01 -9.21219170e-01 -2.94602308e-02 -7.59706676e-01 1.09631620e-01 -7.75134683e-01 6.65628910e-02 -7.81151593e-01 -7.96401799e-01 6.82753861e-01 -2.54503608e-01 -1.31894600e+00 -3.81141037e-01 -6.32120907e-01 -4.26325709e-01 6.08238816e-01 -1.69714057e+00 -8.40403974e-01 -5.20789146e-01 4.18561965e-01 8.89417470e-01 -2.41757870e-01 7.41862833e-01 7.90459067e-02 -1.55936241e-01 4.58674103e-01 1.37327090e-01 2.46544376e-01 6.61300778e-01 -1.27384925e+00 9.37714934e-01 1.19653285e+00 1.57647982e-01 6.56591177e-01 7.64032364e-01 -7.92012155e-01 -1.13574362e+00 -1.08952534e+00 1.13227785e+00 -5.92582047e-01 4.45681006e-01 -1.96449414e-01 -3.50126922e-01 9.33070600e-01 1.71441823e-01 -1.49266094e-01 8.89392048e-02 -4.42681313e-01 -3.32218707e-01 2.02912688e-01 -1.01135921e+00 1.56001222e+00 1.52096987e+00 -3.71278942e-01 -5.68746507e-01 2.33869866e-01 9.18039024e-01 -5.87673247e-01 5.07986724e-01 1.64074361e-01 4.39967543e-01 -1.09908748e+00 8.90237987e-01 -6.29357636e-01 -2.58422933e-05 -9.43430007e-01 -4.28346455e-01 -1.30230510e+00 -6.58016801e-02 -8.48885596e-01 1.72765687e-01 6.13821685e-01 6.37022913e-01 -6.66088223e-01 8.04880321e-01 5.23703575e-01 -1.05294205e-01 -4.25540507e-01 -1.11827564e+00 -9.01782930e-01 -3.48419279e-01 -6.32720530e-01 5.40939271e-01 4.87679571e-01 -3.82056743e-01 1.03422977e-01 -3.94823581e-01 4.49683666e-01 4.58532631e-01 -1.58518165e-01 9.95251298e-01 -8.10622334e-01 3.61149907e-01 -7.43596852e-01 -3.09986055e-01 -1.95253313e+00 3.41080934e-01 -8.46707642e-01 7.51685143e-01 -2.19094491e+00 -5.52841961e-01 -3.47843111e-01 -4.73620109e-02 6.11704350e-01 8.05668533e-02 -2.58655369e-01 2.01728493e-01 -1.56044900e-01 -1.06675363e+00 7.15884507e-01 1.01468313e+00 -4.65510100e-01 -6.44221723e-01 2.22495869e-01 -5.84571242e-01 9.18396533e-01 7.75001168e-01 -2.93152958e-01 -5.38819253e-01 -9.63839471e-01 1.06039427e-01 -7.78599977e-02 3.50997120e-01 -1.17810500e+00 6.73124790e-01 -4.72637892e-01 -2.20658973e-01 -7.63616920e-01 3.41401935e-01 -8.61802638e-01 -2.85541356e-01 8.12488317e-01 -3.96023422e-01 5.68700612e-01 2.97108531e-01 7.52299428e-01 6.76181838e-02 -2.98238415e-02 4.42215383e-01 -1.02330051e-01 -1.69516277e+00 2.42515251e-01 -8.80138814e-01 2.94720829e-01 1.09327042e+00 -3.84059519e-01 -3.94422144e-01 -7.69177020e-01 -3.91794682e-01 8.73517156e-01 6.13841653e-01 7.81713724e-01 7.70800948e-01 -1.29031944e+00 -4.66594219e-01 1.21729806e-01 2.67820954e-01 3.58598307e-02 -1.12655759e-01 7.38381803e-01 -1.03497005e+00 5.81154108e-01 -7.44784698e-02 -4.53921139e-01 -9.95800078e-01 3.52604181e-01 7.14806557e-01 -9.80793834e-02 -8.18267822e-01 9.72190559e-01 1.36246219e-01 -8.50022495e-01 6.57001734e-01 -5.41486144e-01 -4.13438290e-01 -3.37025970e-01 5.85781574e-01 3.23376656e-01 -2.57471025e-01 -7.77866006e-01 -6.60469949e-01 6.21296942e-01 2.04849206e-02 -4.22077447e-01 8.48045468e-01 -4.01009083e-01 1.24117419e-01 2.45344371e-01 6.78111017e-01 7.99874589e-02 -1.66761088e+00 -1.10518135e-01 2.87187278e-01 -3.62060279e-01 -9.05593187e-02 -1.00909448e+00 -5.87066531e-01 5.24017990e-01 7.64122605e-01 -2.90601671e-01 4.30851787e-01 -2.53735721e-01 8.13617349e-01 1.18610060e+00 8.95967066e-01 -1.08637524e+00 -7.43271085e-03 1.36565959e+00 7.34219611e-01 -1.65702462e+00 -4.82304007e-01 8.92994404e-02 -9.41890955e-01 8.25255871e-01 9.38898087e-01 1.07745782e-01 3.38961005e-01 4.26551513e-03 5.40654898e-01 4.85032797e-02 -5.43066025e-01 -5.68715036e-01 -1.54337659e-01 1.15125549e+00 -1.28012359e-01 1.73812762e-01 9.54221338e-02 9.56943445e-03 -2.92986989e-01 -3.12105954e-01 4.37209100e-01 1.03786361e+00 -6.68439209e-01 -7.96943545e-01 -4.64904495e-02 4.38280962e-02 1.47655770e-01 -1.84147656e-01 -3.29761505e-01 8.36893916e-01 6.91922801e-03 1.28775406e+00 5.06003350e-02 -4.32464868e-01 5.94929695e-01 -2.76574671e-01 1.42911542e-02 -5.89339197e-01 -3.40238452e-01 -4.30043012e-01 1.20824695e-01 -1.04429781e+00 -1.93507060e-01 -6.09481275e-01 -1.87214005e+00 -2.99130201e-01 2.18179494e-01 -1.05789021e-01 4.80025858e-01 9.90167499e-01 4.64777768e-01 4.01309907e-01 1.31375685e-01 -8.48759115e-01 -4.50330853e-01 -3.15560222e-01 1.21704238e-02 7.27327690e-02 9.79454517e-01 -7.69025207e-01 -3.52205843e-01 -3.01576674e-01]
[4.5236945152282715, 0.5720837116241455]
e1832e77-0488-4a6a-9e35-c6943237bcb0
transvpr-transformer-based-place-recognition
2201.02001
null
https://arxiv.org/abs/2201.02001v4
https://arxiv.org/pdf/2201.02001v4.pdf
TransVPR: Transformer-based place recognition with multi-level attention aggregation
Visual place recognition is a challenging task for applications such as autonomous driving navigation and mobile robot localization. Distracting elements presenting in complex scenes often lead to deviations in the perception of visual place. To address this problem, it is crucial to integrate information from only task-relevant regions into image representations. In this paper, we introduce a novel holistic place recognition model, TransVPR, based on vision Transformers. It benefits from the desirable property of the self-attention operation in Transformers which can naturally aggregate task-relevant features. Attentions from multiple levels of the Transformer, which focus on different regions of interest, are further combined to generate a global image representation. In addition, the output tokens from Transformer layers filtered by the fused attention mask are considered as key-patch descriptors, which are used to perform spatial matching to re-rank the candidates retrieved by the global image features. The whole model allows end-to-end training with a single objective and image-level supervision. TransVPR achieves state-of-the-art performance on several real-world benchmarks while maintaining low computational time and storage requirements.
['Nanning Zheng', 'Sanping Zhou', 'Weiliang Zuo', 'Yanqing Shen', 'Ruotong Wang']
2022-01-06
null
http://openaccess.thecvf.com//content/CVPR2022/html/Wang_TransVPR_Transformer-Based_Place_Recognition_With_Multi-Level_Attention_Aggregation_CVPR_2022_paper.html
http://openaccess.thecvf.com//content/CVPR2022/papers/Wang_TransVPR_Transformer-Based_Place_Recognition_With_Multi-Level_Attention_Aggregation_CVPR_2022_paper.pdf
cvpr-2022-1
['visual-place-recognition']
['computer-vision']
[ 2.21264154e-01 -2.51110822e-01 -1.47470713e-01 -4.08189565e-01 -7.91417122e-01 -1.31831810e-01 6.25826061e-01 3.56443286e-01 -6.15369558e-01 3.12975019e-01 1.87768843e-02 2.21049841e-02 -1.02722853e-01 -7.15134382e-01 -9.20026183e-01 -8.93085361e-01 2.79782712e-01 4.10097912e-02 5.55942297e-01 -1.93626717e-01 5.94345987e-01 5.41700482e-01 -1.98534250e+00 2.77215302e-01 1.07497454e+00 1.24253881e+00 8.88850749e-01 -4.79162075e-02 3.05574760e-03 8.40354145e-01 -2.83729851e-01 9.48044807e-02 3.29662412e-02 -7.85172358e-02 -4.96160984e-01 5.04974388e-02 6.77248001e-01 -6.58267736e-02 -3.03070456e-01 1.29075968e+00 4.15703356e-01 5.00142395e-01 4.81181383e-01 -1.08604908e+00 -8.66006196e-01 3.29710357e-02 -5.75967371e-01 3.67581397e-01 1.38673335e-01 1.79188833e-01 1.05014026e+00 -1.30988598e+00 5.16451955e-01 1.18198407e+00 8.04002583e-02 2.48981476e-01 -1.12904322e+00 -4.78900880e-01 5.96814156e-01 7.32758820e-01 -1.54475522e+00 -6.11206770e-01 1.07984006e+00 -2.47171253e-01 1.00697923e+00 1.62694946e-01 4.45152253e-01 7.05297530e-01 2.82030582e-01 8.98024738e-01 1.01263547e+00 -2.27254912e-01 1.27718270e-01 -9.58462246e-03 -7.18997568e-02 7.11185038e-01 1.82152130e-02 -1.16156362e-01 -8.97052228e-01 3.05102795e-01 5.91024995e-01 4.78085697e-01 -3.78573865e-01 -7.46213615e-01 -1.28779888e+00 5.75652838e-01 1.36776948e+00 3.32112849e-01 -7.42099345e-01 1.44609762e-02 7.25670010e-02 2.69217789e-02 2.82719731e-01 2.47884676e-01 -2.25893427e-02 3.04194957e-01 -7.21657157e-01 1.78025395e-01 -1.18687361e-01 8.14292431e-01 1.22906554e+00 -1.04451463e-01 -5.28215408e-01 1.08766270e+00 5.13640821e-01 5.63064873e-01 6.94795370e-01 -6.19361281e-01 5.98474264e-01 8.06526363e-01 9.53048095e-02 -1.33198524e+00 -1.36468083e-01 -6.39365315e-01 -7.67129242e-01 3.14960331e-01 4.55560982e-02 7.46481419e-01 -1.23984337e+00 1.60513413e+00 3.38910967e-01 6.97442815e-02 3.93699557e-02 1.24334085e+00 8.41719508e-01 6.64389670e-01 4.28319909e-02 2.06867099e-01 1.46129215e+00 -1.43173230e+00 -4.89592910e-01 -7.91977465e-01 4.24390823e-01 -5.74145257e-01 9.51158047e-01 3.53005677e-02 -9.38092470e-01 -8.21629226e-01 -1.20591700e+00 -5.27195573e-01 -6.30027413e-01 1.60074219e-01 2.41530344e-01 -8.63662437e-02 -1.02140641e+00 3.76489371e-01 -6.11857176e-01 -2.73676157e-01 5.31338453e-01 2.95713693e-01 -5.38382888e-01 -4.03080195e-01 -1.00668192e+00 8.70664120e-01 3.09473366e-01 4.77143079e-01 -1.00466931e+00 -2.61785984e-01 -1.27122736e+00 1.89742565e-01 1.07153371e-01 -3.67762655e-01 9.88197386e-01 -8.04915428e-01 -1.13049138e+00 9.87359107e-01 -6.14580393e-01 -3.32471311e-01 1.92588612e-01 -1.34350598e-01 -2.09508643e-01 1.45036399e-01 6.27933264e-01 8.20441246e-01 8.40011656e-01 -1.23554075e+00 -9.56983387e-01 -7.12842941e-01 -2.87076205e-01 6.17347240e-01 -1.22139744e-01 -2.54176766e-01 -7.20656157e-01 -3.76964658e-01 5.23719490e-01 -7.04826415e-01 -3.01585972e-01 -3.39166597e-02 -2.34070539e-01 -3.65520537e-01 7.43432999e-01 -4.22762632e-01 7.53187001e-01 -2.46945763e+00 2.58448660e-01 1.59838840e-01 2.77130127e-01 7.67223909e-02 -1.89111173e-01 1.69145852e-01 1.08841717e-01 -3.59591931e-01 -8.89175460e-02 -3.95551533e-01 -1.19309947e-01 2.93154269e-02 -3.65343869e-01 6.20850682e-01 3.29966366e-01 9.80348170e-01 -1.08090425e+00 -4.47219998e-01 6.20783687e-01 3.71669382e-01 -3.31593961e-01 1.38138503e-01 9.92978439e-02 5.34152389e-01 -5.77633142e-01 7.68375695e-01 6.68234766e-01 -1.80617362e-01 -4.05134261e-01 -9.40339789e-02 -3.16396207e-01 5.20108223e-01 -7.34276712e-01 1.81955016e+00 -5.04387319e-01 6.85796678e-01 -9.58746970e-02 -1.02921987e+00 1.13371718e+00 -2.75118202e-01 1.80615947e-01 -1.56925070e+00 1.38015300e-02 4.52742457e-01 -1.65603951e-01 -2.08800897e-01 6.19730175e-01 2.22501516e-01 -2.64558911e-01 -2.75544465e-01 1.42507136e-01 2.49127597e-01 -2.33314261e-02 -1.88557982e-01 7.32240558e-01 3.11054867e-02 2.55336374e-01 -2.45016351e-01 7.85320044e-01 -1.80897452e-02 5.67637324e-01 5.43420792e-01 -3.22789580e-01 6.22565448e-01 6.24336451e-02 -5.06086290e-01 -7.29161859e-01 -1.09176683e+00 -1.35793477e-01 1.19326508e+00 8.31410110e-01 -1.86532736e-01 -3.47732067e-01 -4.52558219e-01 6.47827834e-02 4.94105995e-01 -8.28037858e-01 -5.04690409e-01 -4.24519390e-01 -2.29651660e-01 4.61058579e-02 5.53458333e-01 6.96781993e-01 -1.33459449e+00 -9.47902977e-01 2.13465765e-01 -3.09094548e-01 -1.12267256e+00 -3.93591225e-01 4.41012532e-01 -5.82710803e-01 -7.67232597e-01 -9.77142334e-01 -1.14016080e+00 8.17119956e-01 9.25546050e-01 7.69642830e-01 -6.88412413e-02 -5.08706532e-02 -7.81398341e-02 -2.29348883e-01 -2.35088840e-01 2.03493387e-01 8.83911550e-02 -1.52795255e-01 5.95276237e-01 4.17574227e-01 -4.48416352e-01 -8.96717787e-01 3.92052233e-01 -6.28233969e-01 -4.03842889e-02 9.13015306e-01 1.04010355e+00 9.56596792e-01 -3.02826554e-01 3.13925713e-01 -3.00786763e-01 3.10773551e-01 -2.98873633e-01 -6.84898734e-01 3.04756910e-01 -3.23823541e-01 2.24575385e-01 6.83232605e-01 -1.70406774e-01 -7.69919813e-01 1.41527548e-01 9.77365673e-02 -7.73594737e-01 -8.65253434e-02 4.53816324e-01 -3.20204139e-01 -8.15315992e-02 5.07110000e-01 7.25492895e-01 -1.84070691e-01 -3.93438637e-01 2.22749040e-01 5.86178064e-01 6.89368606e-01 -1.28468513e-01 5.50331473e-01 4.91654128e-01 -6.17679358e-02 -9.01339948e-01 -7.46269643e-01 -7.74732292e-01 -5.52078784e-01 -2.27686256e-01 8.89148414e-01 -1.09914196e+00 -4.82938707e-01 5.18878758e-01 -1.09633505e+00 -4.49206941e-02 1.11046238e-02 3.69314253e-01 -4.45638567e-01 8.32369253e-02 -3.65691222e-02 -6.79102838e-01 -2.41788179e-01 -1.37532926e+00 1.47709060e+00 4.90324676e-01 2.81239539e-01 -5.08129179e-01 -1.12636074e-01 2.41624117e-01 2.93508232e-01 -2.06352351e-03 6.15881979e-01 -4.28353310e-01 -9.92351651e-01 -1.46038085e-01 -5.03285944e-01 9.10976678e-02 1.88929647e-01 -4.76638347e-01 -1.27069938e+00 -2.66545951e-01 -1.60146132e-01 -2.89525211e-01 1.33363795e+00 1.78911850e-01 1.14945602e+00 -9.09108594e-02 -6.22308671e-01 6.32371664e-01 1.43680501e+00 2.97268838e-01 7.36059248e-01 5.41451454e-01 9.78164613e-01 6.33897662e-01 9.36566889e-01 8.13055113e-02 5.09097338e-01 8.31863761e-01 6.75977826e-01 -2.47567877e-01 4.29361090e-02 -4.51915264e-01 3.40418905e-01 3.88267368e-01 2.85223514e-01 3.02493591e-02 -8.51264894e-01 9.93707180e-01 -2.09166837e+00 -8.80974293e-01 1.25872687e-01 2.47317696e+00 3.84654015e-01 1.42284319e-01 -3.00558835e-01 -3.97098027e-02 6.70994937e-01 5.65486610e-01 -5.59869170e-01 -7.96311647e-02 1.06025499e-03 -7.85977617e-02 5.10268271e-01 4.17541981e-01 -1.23366404e+00 1.03785026e+00 4.73019600e+00 8.58935654e-01 -1.51038396e+00 -6.56517744e-02 5.57536244e-01 7.92388394e-02 -1.44381776e-01 -1.04674742e-01 -6.95486903e-01 4.23410803e-01 3.13009739e-01 -6.43883497e-02 4.22574073e-01 1.00387561e+00 3.42807770e-02 -2.54322141e-01 -1.04466593e+00 1.31841409e+00 2.28274286e-01 -1.14227617e+00 1.14873178e-01 1.08107589e-01 6.17270648e-01 3.72031033e-01 4.73933458e-01 2.98605829e-01 -1.53004825e-01 -1.07860303e+00 1.00123000e+00 4.63467032e-01 5.85984170e-01 -8.30764890e-01 5.28103054e-01 3.16083342e-01 -1.48757303e+00 -3.55361581e-01 -6.56912506e-01 9.44420919e-02 -7.18912035e-02 3.77127767e-01 -6.39069915e-01 6.73682153e-01 8.85789633e-01 1.07971871e+00 -8.78661275e-01 1.30981231e+00 -1.31414846e-01 -1.43017888e-01 -1.98311105e-01 1.74390543e-02 6.33122683e-01 3.42858844e-02 3.61823440e-01 8.14162910e-01 3.88849139e-01 -3.24780315e-01 2.39341930e-01 9.24653172e-01 2.93052290e-02 7.08883703e-02 -9.38878179e-01 3.47645938e-01 5.41346073e-01 1.25087190e+00 -6.18676543e-01 -2.34210759e-01 -3.20164323e-01 1.08278716e+00 6.83136046e-01 3.86298239e-01 -6.18787050e-01 -4.76195574e-01 7.89527953e-01 5.93175627e-02 5.79219103e-01 -1.36630520e-01 -9.01731029e-02 -9.76426184e-01 3.77084225e-01 -5.27754724e-01 5.30493595e-02 -8.52763951e-01 -9.29176927e-01 7.40511298e-01 -3.25787961e-01 -1.60418236e+00 -9.26068872e-02 -6.90969408e-01 -5.08888304e-01 1.13144743e+00 -2.10034013e+00 -1.16082954e+00 -7.56173193e-01 7.59513497e-01 5.68530738e-01 6.35249093e-02 6.10334396e-01 3.34305644e-01 -4.72247183e-01 5.32993913e-01 1.01077355e-01 -7.21767661e-04 7.35645354e-01 -1.04553223e+00 3.48672241e-01 8.62351418e-01 1.94549382e-01 6.66864753e-01 3.63703936e-01 -3.43857229e-01 -1.17171311e+00 -1.32986605e+00 1.09490454e+00 -2.23002583e-01 1.45690992e-01 -3.96076858e-01 -9.91712630e-01 3.40263098e-01 2.01817691e-01 4.75599766e-01 1.34755954e-01 -9.74173695e-02 -4.93717402e-01 -5.56270480e-01 -8.66374254e-01 7.40302265e-01 9.91999447e-01 -7.71466434e-01 -7.09475458e-01 3.72947641e-02 4.57028031e-01 -4.24569011e-01 -2.39571989e-01 3.16385776e-01 2.93856233e-01 -1.00131512e+00 9.91029501e-01 -1.10382423e-01 2.61637390e-01 -6.92311406e-01 -1.60930604e-01 -1.27991283e+00 -7.46267617e-01 -1.69398591e-01 3.08246642e-01 1.03096581e+00 3.26438874e-01 -6.45081997e-01 3.82458180e-01 1.59394175e-01 -3.50929171e-01 -7.28541195e-01 -1.18852353e+00 -6.16986275e-01 -3.23347181e-01 -1.13460280e-01 4.35285360e-01 6.73377573e-01 -4.06169817e-02 6.09855056e-01 -9.57855210e-02 3.71645063e-01 6.54858351e-01 3.23995709e-01 5.13222456e-01 -1.29364204e+00 2.29784206e-01 -5.41837275e-01 -8.44083190e-01 -1.40629220e+00 1.60028100e-01 -9.76415813e-01 5.24207592e-01 -1.81400239e+00 1.36351719e-01 -3.71634275e-01 -8.32650542e-01 5.16246200e-01 -1.86812565e-01 3.90484989e-01 2.76825398e-01 3.65220279e-01 -9.71574664e-01 1.03948081e+00 1.32178974e+00 -4.69062746e-01 -1.37304202e-01 -2.48951197e-01 -6.30303442e-01 4.25606906e-01 7.16955841e-01 -2.95224994e-01 -5.13418078e-01 -6.33590698e-01 -1.90269783e-01 -1.92232236e-01 7.31394708e-01 -1.29706728e+00 5.14667749e-01 -1.47052079e-01 5.90634704e-01 -8.36436808e-01 5.70318282e-01 -8.94683957e-01 -3.59026670e-01 3.03683043e-01 -2.39331082e-01 2.10416615e-01 4.26594578e-02 6.26579225e-01 -6.16276920e-01 4.75726947e-02 8.22940052e-01 -4.03836481e-02 -1.28837788e+00 2.99656361e-01 -2.14940444e-01 -2.16286555e-01 1.15252197e+00 -3.53320688e-01 -3.50850254e-01 -1.69396885e-02 -3.66571456e-01 4.36127543e-01 6.78272665e-01 8.02478731e-01 1.04881370e+00 -1.61049914e+00 -4.34973150e-01 5.17345607e-01 7.33206451e-01 3.45457703e-01 5.53583324e-01 8.07460606e-01 -2.72408813e-01 6.72499239e-01 -5.72788894e-01 -1.04672348e+00 -1.09714758e+00 9.24457312e-01 5.12438715e-01 -5.94970174e-02 -6.41162574e-01 7.68298984e-01 7.93513834e-01 -1.74830362e-01 2.09207773e-01 -5.66045880e-01 -4.70569879e-01 9.22914967e-02 6.46820366e-01 -8.93570632e-02 2.34216198e-01 -1.07968867e+00 -8.19027722e-01 7.91492224e-01 -1.86129898e-01 -7.59194195e-02 1.13065624e+00 -3.06956261e-01 -3.09919287e-02 4.48112994e-01 1.30694580e+00 -1.69421896e-01 -1.35643923e+00 -5.27355790e-01 -7.17056468e-02 -6.17067516e-01 2.02917561e-01 -5.13157845e-01 -1.01642811e+00 1.03442550e+00 7.40648031e-01 -1.15277782e-01 1.08908772e+00 7.75555968e-02 4.83154535e-01 4.41930860e-01 5.80489695e-01 -9.30308759e-01 7.63262808e-02 5.84623873e-01 1.03494763e+00 -1.24807084e+00 -3.48297626e-01 -1.22460358e-01 -6.70182228e-01 6.00062609e-01 8.01951945e-01 -1.89413056e-01 3.39493454e-01 -3.87120992e-01 9.22972783e-02 -2.03434020e-01 -6.52121067e-01 -5.91155052e-01 6.07010305e-01 6.54762745e-01 7.66907781e-02 -9.58696082e-02 1.36686817e-01 3.32681477e-01 1.60756007e-01 -3.20625037e-01 -1.31658062e-01 9.23589468e-01 -7.43538797e-01 -6.24394536e-01 -3.53509426e-01 2.05579713e-01 -6.55387491e-02 -6.93972856e-02 -2.71527499e-01 3.33620697e-01 1.20329455e-01 8.43389988e-01 1.54706493e-01 -4.23354149e-01 4.87661153e-01 -9.11822245e-02 2.97729105e-01 -5.11647344e-01 -2.97519624e-01 -4.94399061e-03 -3.24124962e-01 -7.50031590e-01 -2.93541312e-01 -5.80350280e-01 -1.27309537e+00 2.13512212e-01 -2.49664802e-02 7.33192116e-02 3.77837986e-01 8.34670603e-01 6.91450655e-01 7.02445567e-01 7.67182767e-01 -1.32747316e+00 -3.49523872e-01 -8.08370650e-01 -4.39298928e-01 4.65800792e-01 8.33799601e-01 -1.00245738e+00 8.32135789e-03 -2.32611850e-01]
[9.813495635986328, -0.19518662989139557]
1f1feda4-74c4-4d4d-a894-1534c2bd3dc6
scamps-synthetics-for-camera-measurement-of
2206.04197
null
https://arxiv.org/abs/2206.04197v1
https://arxiv.org/pdf/2206.04197v1.pdf
SCAMPS: Synthetics for Camera Measurement of Physiological Signals
The use of cameras and computational algorithms for noninvasive, low-cost and scalable measurement of physiological (e.g., cardiac and pulmonary) vital signs is very attractive. However, diverse data representing a range of environments, body motions, illumination conditions and physiological states is laborious, time consuming and expensive to obtain. Synthetic data have proven a valuable tool in several areas of machine learning, yet are not widely available for camera measurement of physiological states. Synthetic data offer "perfect" labels (e.g., without noise and with precise synchronization), labels that may not be possible to obtain otherwise (e.g., precise pixel level segmentation maps) and provide a high degree of control over variation and diversity in the dataset. We present SCAMPS, a dataset of synthetics containing 2,800 videos (1.68M frames) with aligned cardiac and respiratory signals and facial action intensities. The RGB frames are provided alongside segmentation maps. We provide precise descriptive statistics about the underlying waveforms, including inter-beat interval, heart rate variability, and pulse arrival time. Finally, we present baseline results training on these synthetic data and testing on real-world datasets to illustrate generalizability.
['Tadas Baltrusaitis', 'Jonathan Lester', 'Javier Hernandez', 'Brian L. Hill', 'Xin Liu', 'Miah Wander', 'Daniel McDuff']
2022-06-08
null
null
null
null
['heart-rate-variability']
['medical']
[ 3.33134085e-01 -4.00564730e-01 4.20219004e-02 -5.07637560e-01 -4.81851816e-01 -6.84137106e-01 1.81240126e-01 1.42461546e-02 -1.35630578e-01 8.02884579e-01 1.74972899e-02 2.33081996e-01 2.46954896e-02 -1.58823878e-01 -2.44959176e-01 -1.06342196e+00 -3.17134112e-01 -5.80692366e-02 -2.83576220e-01 2.85314679e-01 -5.48877344e-02 6.35006726e-01 -1.43802297e+00 8.25935677e-02 3.74319792e-01 1.15070820e+00 -2.98429191e-01 9.65683579e-01 4.32683945e-01 5.69101393e-01 -8.37399542e-01 -7.63199776e-02 2.58950055e-01 -8.22146475e-01 -2.41631359e-01 3.14975560e-01 5.52103937e-01 -3.43753129e-01 -3.23787987e-01 6.45773411e-01 8.48545730e-01 -2.43223514e-02 3.60944927e-01 -1.37432909e+00 -5.79816513e-02 1.02765314e-01 -3.86838853e-01 4.92448330e-01 5.04438162e-01 6.93182230e-01 4.44178313e-01 -5.42342067e-01 5.77515185e-01 9.17078912e-01 7.65624881e-01 6.06754661e-01 -1.56854296e+00 -6.25442564e-01 -6.10253334e-01 4.55393977e-02 -1.38552940e+00 -8.38619828e-01 6.21652246e-01 -4.33808386e-01 4.85629737e-01 5.28951406e-01 9.60120201e-01 1.42414260e+00 2.43634179e-01 1.86240897e-01 1.43915284e+00 -3.17042060e-02 2.58882284e-01 1.82317793e-01 -3.36362451e-01 4.79543775e-01 1.87232777e-01 -7.14116246e-02 -8.29871356e-01 -9.87342894e-02 7.89415002e-01 9.69847478e-03 -7.95635402e-01 -2.43603528e-01 -1.68931651e+00 2.76320845e-01 -7.64481425e-02 1.46444831e-02 -4.93791908e-01 3.31387877e-01 4.30459738e-01 3.01661342e-01 1.35992482e-01 4.40650076e-01 -4.37008977e-01 -6.67931557e-01 -1.07731748e+00 -9.38950032e-02 8.13680828e-01 8.02959383e-01 3.91680658e-01 3.51372242e-01 -2.25392997e-01 6.50145471e-01 -1.11223482e-01 9.29847419e-01 4.46327746e-01 -1.69443774e+00 -1.15905859e-01 1.43894419e-01 2.17878506e-01 -1.14603317e+00 -7.29222357e-01 -8.08572322e-02 -1.01465964e+00 5.95316961e-02 6.81749642e-01 -3.34144950e-01 -6.18667662e-01 1.65035284e+00 3.26235056e-01 5.19722760e-01 2.31196191e-02 1.13856423e+00 8.18781853e-01 4.20685202e-01 4.82048765e-02 -7.33144701e-01 1.30964696e+00 -1.51819423e-01 -8.45176220e-01 -1.75144244e-02 1.68938592e-01 -6.43973887e-01 1.13593888e+00 5.10929525e-01 -1.24111855e+00 -5.16504705e-01 -7.36532748e-01 4.66450840e-01 1.03951380e-01 -1.30461901e-01 4.26818073e-01 8.37245226e-01 -1.09572232e+00 7.67492175e-01 -1.06069291e+00 -5.04120529e-01 3.58737439e-01 9.20298100e-02 -4.34737235e-01 -1.53413579e-01 -9.26548123e-01 5.11794448e-01 -6.80346265e-02 9.97527763e-02 -7.91890621e-01 -8.95465493e-01 -6.95510149e-01 -1.81888610e-01 1.20617501e-01 -5.27035475e-01 9.26724851e-01 -7.77911127e-01 -1.38793027e+00 9.19305265e-01 6.16889680e-03 -2.17143267e-01 4.43936199e-01 1.42605722e-01 -4.91413265e-01 9.21882689e-01 -2.49203995e-01 7.19869614e-01 9.02528763e-01 -8.30289066e-01 -3.33900680e-03 -6.89854696e-02 -3.32905740e-01 1.94821954e-01 -1.56283289e-01 2.73446620e-01 -3.67239565e-01 -3.63843977e-01 1.31536320e-01 -1.03317225e+00 7.36517385e-02 5.01908481e-01 -3.17539573e-01 7.25726068e-01 4.64631647e-01 -6.33697152e-01 9.32313263e-01 -2.27650619e+00 -1.16107918e-01 6.11409657e-02 3.32596004e-01 1.01186983e-01 2.35992149e-01 6.26736581e-02 -3.47486138e-02 2.01399222e-01 -3.24455768e-01 -7.22800195e-02 -2.73014724e-01 3.46759260e-01 1.71334565e-01 1.00396013e+00 8.84119794e-02 7.54525661e-01 -8.70575130e-01 -8.76735508e-01 5.81385553e-01 7.62109995e-01 -3.20328385e-01 2.48115242e-01 2.45976895e-01 1.15039086e+00 -5.38716614e-02 9.15483296e-01 3.02410424e-01 -2.22984567e-01 -8.31715092e-02 -4.94851053e-01 6.18001912e-03 -1.09989904e-01 -1.17982519e+00 1.52463138e+00 -1.69613257e-01 1.07271540e+00 1.77316591e-01 -7.08409250e-01 8.38545799e-01 7.31933475e-01 1.13859117e+00 -6.98752522e-01 1.88962460e-01 -1.14534341e-01 1.35064065e-01 -1.00571966e+00 -6.56045750e-02 -3.35591465e-01 1.57249421e-01 4.59415525e-01 -9.63166282e-02 -4.16222364e-01 2.35784799e-01 -8.82355943e-02 1.23753524e+00 8.51028189e-02 1.53882265e-01 -2.28381425e-01 2.09411234e-01 6.42104913e-03 6.68017566e-01 3.41634899e-01 -7.42634177e-01 1.06934857e+00 5.57706416e-01 -3.76552761e-01 -9.81584847e-01 -1.11417484e+00 -4.11683738e-01 2.57376254e-01 9.74875875e-03 -3.10210615e-01 -7.51605690e-01 2.16809422e-01 -1.28543824e-01 2.50854284e-01 -4.16769266e-01 -2.08578154e-01 -4.61261839e-01 -9.13568735e-01 8.49271476e-01 4.54470724e-01 3.68247390e-01 -1.19673908e+00 -1.46401477e+00 1.23672374e-01 -5.15869439e-01 -1.53141177e+00 -1.72542706e-01 2.35750973e-02 -9.65221524e-01 -1.19406962e+00 -6.22414291e-01 -1.20465174e-01 4.67641979e-01 4.25804406e-02 1.45422435e+00 -1.80569440e-01 -9.57264006e-01 8.55790019e-01 -2.43933871e-01 -3.01681131e-01 -2.16943100e-01 -7.04332054e-01 1.86927602e-01 2.13415846e-01 -1.06942502e-03 -6.91335917e-01 -1.09726679e+00 4.50903594e-01 -5.95889270e-01 -5.79049103e-02 2.34007865e-01 5.91630340e-01 7.85465121e-01 -3.37051928e-01 3.52424562e-01 -5.24553299e-01 4.18275476e-01 -3.23662370e-01 -2.27835968e-01 -6.90160468e-02 -3.61132145e-01 -5.15057862e-01 6.67593837e-01 -6.01116896e-01 -7.46675670e-01 -5.71217798e-02 3.50629866e-01 -8.08375716e-01 -4.78974015e-01 7.27434531e-02 2.04295024e-01 7.23431855e-02 9.49361742e-01 1.74803615e-01 1.37248248e-01 -8.20016488e-02 1.69951543e-01 5.35657883e-01 1.18819642e+00 -5.33231795e-01 4.30736780e-01 6.40797198e-01 2.39192083e-01 -1.11519885e+00 -4.44346786e-01 -3.92801821e-01 -9.32420969e-01 -6.00962996e-01 9.31795657e-01 -9.49463844e-01 -7.09900737e-01 5.56917071e-01 -7.07340121e-01 -4.78473157e-01 -5.45409262e-01 8.41595292e-01 -7.01541126e-01 2.46616289e-01 -6.09243333e-01 -8.10435236e-01 -2.88178146e-01 -8.26713264e-01 9.43826795e-01 4.19325233e-01 -6.55573487e-01 -9.92189288e-01 -1.22749083e-01 3.46072406e-01 6.98040545e-01 1.15952241e+00 4.69261438e-01 -4.91668172e-02 -3.02553058e-01 -1.46178663e-01 -3.82125564e-02 4.00280237e-01 3.76536548e-01 4.72423226e-01 -1.24826288e+00 -2.26250663e-01 2.11887524e-01 -4.87560183e-01 1.25427976e-01 6.36662722e-01 1.22417915e+00 -2.00728863e-01 6.37269542e-02 9.36850727e-01 1.14522302e+00 2.50805914e-01 6.97990775e-01 -2.68285066e-01 5.75139642e-01 7.71803200e-01 5.69910049e-01 8.03711176e-01 1.03389852e-01 4.97805953e-01 3.30018908e-01 -3.52261156e-01 -1.29825667e-01 3.83047462e-01 2.66310126e-01 7.63809383e-01 -9.47198272e-02 -1.56306192e-01 -8.57761383e-01 4.26968545e-01 -1.08867109e+00 -1.05439508e+00 -2.47934699e-01 2.32185555e+00 1.03981853e+00 -1.85121968e-01 2.26863652e-01 3.24570030e-01 7.48183489e-01 4.11933959e-02 -7.90053010e-01 -4.60610241e-01 -2.74832398e-01 3.71969432e-01 4.15524781e-01 -5.69811836e-03 -1.00050724e+00 2.63504893e-01 6.69579601e+00 -3.09685785e-02 -1.48764122e+00 -1.61019921e-01 7.78004110e-01 -5.51354289e-01 4.62141857e-02 -3.27129394e-01 -1.61883950e-01 7.27645457e-01 1.49357116e+00 -7.50991628e-02 6.39847636e-01 2.92912573e-01 6.69035912e-01 -2.68431693e-01 -1.14539313e+00 1.59578431e+00 1.23150066e-01 -9.44623828e-01 -7.16077328e-01 -3.10460418e-01 4.25431371e-01 2.97561195e-02 -1.32611573e-01 -3.49416554e-01 -4.10815388e-01 -1.15634656e+00 4.72467750e-01 6.17365301e-01 1.41030478e+00 -2.85394698e-01 4.20858234e-01 9.57525745e-02 -8.24707448e-01 2.09143385e-01 -1.40254825e-01 2.26650819e-01 2.15121120e-01 5.86794138e-01 -3.38737756e-01 8.25106576e-02 8.95890236e-01 7.77397752e-01 -5.01535714e-01 1.10872936e+00 -6.46726564e-02 7.64415145e-01 -5.41015863e-01 2.54851311e-01 -2.57487208e-01 -2.38855958e-01 4.49645817e-01 1.27871788e+00 3.89550954e-01 3.82929265e-01 -6.40532821e-02 8.62276375e-01 1.52462617e-01 -1.18369989e-01 -6.66988909e-01 -1.82395354e-01 5.43501616e-01 1.46416616e+00 -8.56358230e-01 -3.57502967e-01 -4.02433127e-01 7.29723573e-01 -5.59107363e-01 3.45076054e-01 -1.02875173e+00 -2.73801088e-01 7.98509300e-01 2.02284992e-01 -3.10668051e-01 -2.60338813e-01 -2.85739779e-01 -1.07699704e+00 -8.83005187e-02 -1.02691996e+00 2.08336994e-01 -1.17428517e+00 -8.57229710e-01 3.23734045e-01 2.01470509e-01 -1.27679658e+00 -3.97588998e-01 -3.01530540e-01 -6.15642190e-01 8.17677557e-01 -1.29606771e+00 -2.71391898e-01 -1.08397388e+00 9.07586515e-01 1.71471372e-01 3.50578338e-01 9.52749908e-01 5.56855679e-01 -7.65674174e-01 3.12783390e-01 -1.06935203e-01 3.53472270e-02 9.79173720e-01 -1.14386058e+00 3.96064930e-02 5.90658724e-01 3.10410094e-02 4.83978152e-01 7.68224359e-01 -1.49995506e-01 -1.67776012e+00 -9.75964308e-01 2.65132934e-01 -4.33335811e-01 3.28331411e-01 -2.15017185e-01 -7.70642400e-01 2.99654990e-01 -6.10097200e-02 6.07058346e-01 8.03828001e-01 -6.06279910e-01 9.76578370e-02 -3.87962341e-01 -1.38361371e+00 4.64606285e-01 6.05716527e-01 -4.62492943e-01 -9.65803042e-02 2.75783777e-01 2.36345321e-01 -6.97232664e-01 -1.29719067e+00 3.74586493e-01 8.20834637e-01 -1.20969224e+00 9.68252659e-01 -2.25982696e-01 1.80024981e-01 -3.24320197e-01 -1.11675501e-01 -1.17537558e+00 2.50079960e-01 -9.92731690e-01 -1.15639657e-01 1.15585339e+00 9.01282057e-02 -5.65385759e-01 6.53108120e-01 1.01998007e+00 -5.09679643e-03 -6.76206589e-01 -7.33129799e-01 -6.80917084e-01 -4.12761867e-01 -5.65272748e-01 2.22263247e-01 1.09998703e+00 -3.31366062e-02 -1.57291114e-01 -4.61384177e-01 -1.49193868e-01 7.73931921e-01 1.11996271e-01 7.11182654e-01 -1.09248626e+00 -1.50010139e-01 -5.48913367e-02 -5.88619471e-01 -3.12593222e-01 -3.16672683e-01 -4.50540036e-01 1.86980818e-03 -1.18456614e+00 4.80378196e-02 -4.67757165e-01 -3.66644084e-01 4.52014357e-01 3.56909670e-02 8.62827599e-01 3.68363112e-02 2.81201482e-01 -3.10202718e-01 3.48920785e-02 1.28210616e+00 1.18390568e-01 -1.47809520e-01 -2.33291954e-01 -2.87701666e-01 5.65533400e-01 9.74385381e-01 -4.47067618e-01 -4.87912029e-01 -6.41585886e-02 -1.44329667e-01 5.86779714e-01 5.00952184e-01 -1.28792667e+00 2.84870211e-02 -1.06602401e-01 6.86558306e-01 -1.01006225e-01 6.81479156e-01 -6.96874082e-01 5.39801061e-01 5.74124336e-01 -2.83810616e-01 2.78173864e-01 1.61151558e-01 3.12829673e-01 -2.03911141e-01 2.50790566e-01 1.14034462e+00 -3.76451820e-01 -4.31758344e-01 3.73008132e-01 -1.07085750e-01 6.58710539e-01 1.07100964e+00 -5.44290125e-01 -3.69397908e-01 -6.40492737e-01 -7.34688461e-01 1.03472106e-01 6.92601621e-01 1.45266086e-01 5.71227908e-01 -1.13387883e+00 -6.49904013e-01 3.63373876e-01 1.35182291e-01 -1.14281550e-01 2.27681309e-01 1.44410992e+00 -7.88381279e-01 -3.02801505e-02 -5.77310979e-01 -1.18867910e+00 -1.31689966e+00 2.57489502e-01 6.12924159e-01 5.20771861e-01 -8.53612542e-01 4.82565284e-01 -1.15379512e-01 2.40388334e-01 1.59788534e-01 -3.29707742e-01 2.20212271e-03 2.01035067e-01 4.01455551e-01 5.31106591e-01 -1.31496131e-01 -5.93876183e-01 -4.99790281e-01 7.57950664e-01 9.75066364e-01 3.46470289e-02 9.25462365e-01 -4.99844402e-01 2.05774345e-02 8.07702422e-01 1.07614136e+00 -3.01216692e-01 -1.51131523e+00 1.31574258e-01 -2.89978474e-01 -6.92832589e-01 -2.52731979e-01 -6.22755110e-01 -1.36994791e+00 1.31992531e+00 8.76547039e-01 1.91551447e-01 1.50576925e+00 -3.46813768e-01 6.43384874e-01 2.20192224e-02 2.04019293e-01 -8.83462310e-01 1.00586154e-01 -2.74456680e-01 4.61126596e-01 -1.12750041e+00 8.35460052e-02 -2.81243712e-01 -8.51514339e-01 1.18276143e+00 3.43138576e-01 3.39040965e-01 3.42419803e-01 4.74964857e-01 5.43009758e-01 -1.82650194e-01 -8.70789528e-01 1.26185283e-01 -9.24017653e-02 7.34199703e-01 5.82186520e-01 -3.67202088e-02 3.75614315e-02 -1.95569381e-01 -1.43081918e-01 -3.28041390e-02 8.35709870e-01 8.07663023e-01 -8.90781283e-02 -4.96083498e-01 -4.29875970e-01 5.59623957e-01 -9.58923340e-01 1.76435143e-01 -7.21217170e-02 7.86342502e-01 1.84479598e-02 1.08207798e+00 7.84047842e-02 -1.35204807e-01 3.30279648e-01 2.02719301e-01 6.81093514e-01 -3.44539791e-01 -4.15474534e-01 1.58143058e-01 -1.87258963e-02 -9.40007746e-01 -8.20535600e-01 -8.36481929e-01 -1.26908481e+00 -3.79126132e-01 2.24620789e-01 -2.58826762e-01 8.63403559e-01 4.65054363e-01 1.89192414e-01 5.18690944e-01 5.93434453e-01 -8.99957120e-01 -1.51917428e-01 -7.24078953e-01 -6.75154984e-01 8.31980169e-01 4.69120860e-01 -4.27954823e-01 -6.62886322e-01 7.04541385e-01]
[13.89770793914795, 2.829139232635498]
b9675a45-af14-441b-8777-47597db2714b
memen-multi-layer-embedding-with-memory
1707.09098
null
http://arxiv.org/abs/1707.09098v1
http://arxiv.org/pdf/1707.09098v1.pdf
MEMEN: Multi-layer Embedding with Memory Networks for Machine Comprehension
Machine comprehension(MC) style question answering is a representative problem in natural language processing. Previous methods rarely spend time on the improvement of encoding layer, especially the embedding of syntactic information and name entity of the words, which are very crucial to the quality of encoding. Moreover, existing attention methods represent each query word as a vector or use a single vector to represent the whole query sentence, neither of them can handle the proper weight of the key words in query sentence. In this paper, we introduce a novel neural network architecture called Multi-layer Embedding with Memory Network(MEMEN) for machine reading task. In the encoding layer, we employ classic skip-gram model to the syntactic and semantic information of the words to train a new kind of embedding layer. We also propose a memory network of full-orientation matching of the query and passage to catch more pivotal information. Experiments show that our model has competitive results both from the perspectives of precision and efficiency in Stanford Question Answering Dataset(SQuAD) among all published results and achieves the state-of-the-art results on TriviaQA dataset.
['Boyuan Pan', 'Zhou Zhao', 'Deng Cai', 'Bin Cao', 'Xiaofei He', 'Hao Li']
2017-07-28
null
null
null
null
['triviaqa']
['miscellaneous']
[ 2.62092669e-02 -4.36978154e-02 -2.81340722e-02 -4.97153640e-01 -8.36194038e-01 -3.57025862e-01 2.97247231e-01 5.23143768e-01 -8.54406178e-01 2.85599858e-01 6.19569182e-01 -5.06336331e-01 9.88904759e-03 -1.11242235e+00 -8.87651145e-01 -2.88381755e-01 3.88577193e-01 3.94363880e-01 5.75930417e-01 -6.35683417e-01 3.13416421e-01 -1.82978168e-01 -1.16550052e+00 6.15384817e-01 1.03046262e+00 9.74077404e-01 4.44874585e-01 6.97299600e-01 -7.28454709e-01 8.96064579e-01 -6.13172591e-01 -6.87677026e-01 -2.41472974e-01 -5.28254271e-01 -1.27674377e+00 -5.20082653e-01 5.24313092e-01 -4.70851988e-01 -6.14555061e-01 9.20909405e-01 6.52709186e-01 1.51153907e-01 5.44117093e-01 -6.62594199e-01 -1.12503326e+00 6.17604911e-01 -1.80402443e-01 6.56047463e-01 4.91101712e-01 2.19531711e-02 1.42969024e+00 -9.26079094e-01 3.44097197e-01 1.29345095e+00 1.85163781e-01 4.99542505e-01 -6.63836002e-01 -2.19583794e-01 2.56049454e-01 9.19285953e-01 -9.78957295e-01 -2.50796020e-01 7.12240040e-01 -8.19024816e-02 1.24660921e+00 2.97644734e-01 3.68959993e-01 8.81534874e-01 2.55966365e-01 1.08231270e+00 6.01235569e-01 -5.72744668e-01 -3.50285545e-02 7.12194219e-02 9.50674176e-01 8.46317589e-01 -8.96004774e-03 -3.54549468e-01 -1.57821357e-01 -2.95985807e-02 2.47067049e-01 1.37117300e-02 -4.65497732e-01 -4.60271649e-02 -1.12029850e+00 1.14551890e+00 6.92322731e-01 4.18455601e-01 -4.70128208e-01 1.74362645e-01 4.98890609e-01 5.48276186e-01 1.42538711e-01 4.80766982e-01 -5.84756792e-01 5.26479371e-02 -3.79972309e-01 2.63968855e-01 8.19109261e-01 7.98655152e-01 6.09779656e-01 -2.74772197e-01 -7.32408404e-01 7.60666847e-01 3.26199859e-01 5.78140140e-01 7.20309258e-01 -6.10027254e-01 9.08877313e-01 9.87749398e-01 -4.07891460e-02 -1.18133688e+00 -3.49182636e-01 -5.33004642e-01 -7.80336082e-01 -6.10480070e-01 2.30989099e-01 -1.14018664e-01 -6.17365420e-01 1.68153310e+00 2.14494914e-01 -2.32969537e-01 2.70051301e-01 9.29883063e-01 1.28469324e+00 1.12917387e+00 8.72613490e-02 1.65175158e-03 1.79354751e+00 -1.50629401e+00 -9.88174796e-01 -5.60377419e-01 7.48823404e-01 -5.65165222e-01 1.27671456e+00 -7.62283653e-02 -1.17091656e+00 -7.20631957e-01 -1.07953286e+00 -8.25649202e-01 -6.65474057e-01 7.64481649e-02 3.86615545e-01 1.91385090e-01 -7.26944566e-01 2.96223998e-01 -4.05844241e-01 -2.42750868e-01 1.98522046e-01 1.59723639e-01 -1.98751271e-01 -3.74181032e-01 -1.70547831e+00 1.09099436e+00 5.07100523e-01 2.82662660e-01 -6.24117970e-01 -4.91845578e-01 -8.72714221e-01 4.87589896e-01 2.76496917e-01 -9.32732761e-01 1.28983724e+00 -8.11253190e-01 -1.15396821e+00 5.94904423e-01 -5.38853407e-01 -3.77768695e-01 -2.47113556e-01 -5.45618117e-01 -2.99849808e-01 3.78403127e-01 -1.38678044e-01 6.36417329e-01 7.03773499e-01 -6.41972363e-01 -3.42127353e-01 -4.53849703e-01 4.48306143e-01 1.80246800e-01 -5.03480732e-01 -9.67709124e-02 -3.33622545e-01 -3.25841695e-01 6.00044653e-02 -2.75253236e-01 -1.52738877e-02 -2.58494139e-01 -8.94586593e-02 -8.42947662e-01 3.28329861e-01 -1.08741474e+00 1.70695233e+00 -2.08634496e+00 4.30657327e-01 -3.67310405e-01 1.28770471e-01 5.22299528e-01 -5.12408197e-01 7.53837287e-01 1.33989111e-01 1.08724702e-02 -1.60883605e-01 -8.46185163e-02 8.50203410e-02 3.26021969e-01 -6.00828528e-01 1.39310166e-01 4.45549458e-01 1.42965460e+00 -9.16861653e-01 -4.91986006e-01 -2.75884241e-01 2.49944434e-01 -6.23802125e-01 5.78893006e-01 -5.04867017e-01 -7.37563968e-02 -7.39807427e-01 2.56508231e-01 5.16032815e-01 -3.74919415e-01 -3.48364800e-01 -3.32384259e-01 3.74763697e-01 6.86505973e-01 -6.95088148e-01 1.97058535e+00 -4.74199712e-01 2.52633959e-01 -1.45431221e-01 -9.51504648e-01 7.97490776e-01 2.99692959e-01 -2.58845568e-01 -1.11481822e+00 1.58131316e-01 1.00252301e-01 1.69367224e-01 -1.05953014e+00 5.54672241e-01 4.25301827e-02 -5.44538796e-02 5.52080214e-01 2.97022790e-01 1.17729343e-01 1.55707568e-01 3.94902706e-01 9.96478975e-01 -3.28033626e-01 1.46809801e-01 -1.64873719e-01 9.67892289e-01 -7.71332011e-02 8.94382820e-02 6.79983199e-01 -3.77810337e-02 4.88557786e-01 6.64131641e-01 -4.53004599e-01 -9.73947644e-01 -7.79032052e-01 1.13378480e-01 1.52152085e+00 1.50825679e-01 -3.15581262e-01 -8.51779222e-01 -7.32058764e-01 -1.01056330e-01 9.19815481e-01 -7.65739024e-01 -5.58548987e-01 -9.61216509e-01 -5.36460817e-01 4.07452136e-01 5.95363438e-01 6.20150745e-01 -1.24961543e+00 -4.51040775e-01 3.34950000e-01 -4.40293163e-01 -9.73751545e-01 -6.58037961e-01 8.25764891e-03 -9.26205039e-01 -1.03980875e+00 -6.55895591e-01 -1.20786405e+00 4.66011971e-01 1.42696887e-01 1.37426817e+00 4.66099471e-01 4.22349945e-02 1.71869054e-01 -7.13221312e-01 -3.36549520e-01 3.77711817e-03 5.77861011e-01 -5.82981408e-01 5.35177700e-02 7.47250497e-01 -2.43999168e-01 -8.04647982e-01 3.00294440e-02 -1.02316570e+00 -1.34655997e-01 7.11173534e-01 1.04448199e+00 3.11741948e-01 -5.57571173e-01 8.24145138e-01 -6.97981834e-01 1.04858971e+00 -6.30876482e-01 -2.30186433e-01 6.00324333e-01 -3.36397827e-01 5.13144433e-01 6.28854096e-01 -2.98882991e-01 -8.31002653e-01 -4.56869155e-01 -7.02521980e-01 4.86754142e-02 1.24185860e-01 7.19047487e-01 -1.51848182e-01 3.83121401e-01 4.82112527e-01 6.31963134e-01 -9.85793695e-02 -7.34957039e-01 6.03584290e-01 6.84579968e-01 3.19619507e-01 -4.17342782e-01 4.42266196e-01 2.33543739e-02 -4.52674478e-01 -5.01139402e-01 -1.32650948e+00 -5.90529263e-01 -5.02274096e-01 3.38164151e-01 1.11780572e+00 -6.35641575e-01 -6.42974973e-01 1.91433445e-01 -1.64239514e+00 2.52264500e-01 -5.20565547e-02 3.50643992e-01 -1.91079795e-01 3.36558610e-01 -7.87078321e-01 -5.39015949e-01 -7.10076153e-01 -1.13300395e+00 1.07477260e+00 2.70864725e-01 1.99276671e-01 -8.93930614e-01 1.37007743e-01 5.25258958e-01 7.97149241e-01 -3.73591781e-01 1.58118510e+00 -9.56616342e-01 -6.01717472e-01 -2.01903194e-01 -4.27919537e-01 4.73678380e-01 -3.64768803e-01 -7.22587883e-01 -9.18930888e-01 -2.47797713e-01 3.62317026e-01 -5.12829840e-01 1.33378482e+00 7.17258900e-02 1.12140036e+00 -4.23527777e-01 1.22532323e-02 3.45089555e-01 1.40195918e+00 2.47483868e-02 6.91036046e-01 2.43966550e-01 5.19642770e-01 6.83050275e-01 2.85455972e-01 -7.59534687e-02 6.88736737e-01 3.34781647e-01 5.99482894e-01 2.29997724e-01 -4.27039526e-02 -5.69360614e-01 3.21103036e-01 1.45075870e+00 4.26846743e-01 -3.97233903e-01 -7.53571093e-01 7.27001667e-01 -1.67362642e+00 -7.65297472e-01 -1.28678977e-01 1.76798224e+00 8.24609518e-01 -1.75174307e-02 -3.63930970e-01 -8.17180425e-02 3.32323134e-01 4.97401267e-01 -5.13729274e-01 -5.61383128e-01 -7.32682720e-02 4.37380254e-01 5.86329103e-02 6.25024498e-01 -9.15333152e-01 9.58195686e-01 5.75376034e+00 7.84110308e-01 -6.93635166e-01 4.09870207e-01 3.57822895e-01 1.86790213e-01 -5.08871257e-01 -1.23616859e-01 -8.95693600e-01 2.05772430e-01 1.11160290e+00 -8.13176185e-02 1.53165624e-01 5.57337463e-01 -3.19932133e-01 -7.60560110e-02 -1.11889541e+00 8.59277546e-01 4.71954912e-01 -1.20561969e+00 5.23094714e-01 -4.91227508e-01 1.62070408e-01 -7.67042264e-02 -1.47930384e-01 8.77261341e-01 -4.27258849e-01 -1.20986521e+00 3.01598191e-01 8.28297019e-01 1.81840241e-01 -6.75168812e-01 1.27679074e+00 6.19748831e-01 -8.39452147e-01 -1.52137846e-01 -8.86199653e-01 -2.24654794e-01 3.18840832e-01 9.90235433e-02 -2.75083780e-01 5.77788889e-01 4.33791518e-01 3.98821861e-01 -9.60064530e-01 8.63466442e-01 -4.47437495e-01 8.06682944e-01 3.55904251e-02 -6.37790501e-01 4.95989233e-01 1.44026931e-02 2.84843624e-01 1.07928681e+00 -5.96638247e-02 4.01825786e-01 -2.24134758e-01 7.54107654e-01 -3.84253174e-01 5.31317949e-01 -3.02811950e-01 -2.18151659e-01 2.83046901e-01 8.86364877e-01 -3.47572379e-03 -4.86572057e-01 -5.97811460e-01 1.02150583e+00 7.27273226e-01 3.09773117e-01 -5.35959303e-01 -7.37094700e-01 3.76944065e-01 -2.08595321e-01 5.50573468e-01 -1.44394845e-01 6.49311332e-05 -1.27271473e+00 4.31941420e-01 -9.52491045e-01 5.22515595e-01 -6.87066734e-01 -1.36253607e+00 7.34079719e-01 -2.65268415e-01 -6.83988690e-01 -9.84367579e-02 -7.50924408e-01 -7.10199654e-01 1.05131924e+00 -1.83191168e+00 -9.65413749e-01 -1.23665132e-01 4.48248982e-01 7.51028180e-01 -7.70685226e-02 9.82711017e-01 4.27348495e-01 -4.84947979e-01 7.01039672e-01 2.91790031e-02 4.91958052e-01 5.42877793e-01 -1.08120441e+00 3.57075423e-01 5.83288968e-01 2.91660041e-01 8.30062985e-01 3.76439333e-01 -1.43731192e-01 -1.89961410e+00 -7.18393922e-01 1.53797185e+00 -6.13059580e-01 5.12894928e-01 -3.26605827e-01 -1.39893496e+00 5.81020176e-01 7.14685798e-01 -2.36634716e-01 6.17430806e-01 2.13994846e-01 -4.01957989e-01 -2.25460708e-01 -5.78142464e-01 2.91156977e-01 6.33575439e-01 -6.56053841e-01 -1.50902760e+00 3.81206363e-01 1.57462597e+00 -2.46611565e-01 -7.18221903e-01 3.23306352e-01 2.56114781e-01 -7.18926728e-01 9.20012474e-01 -1.16164470e+00 7.06410468e-01 -1.32489815e-01 -2.57299334e-01 -1.09193540e+00 -3.04376692e-01 1.12574503e-01 -3.47347945e-01 1.04817510e+00 6.28758311e-01 -4.09315407e-01 3.97513658e-01 2.77589113e-01 -2.02129230e-01 -1.16785586e+00 -9.10615146e-01 -2.75966704e-01 4.87840474e-01 -2.59769499e-01 7.38680780e-01 6.22960567e-01 3.62754948e-02 1.06837618e+00 -7.58590326e-02 8.89810398e-02 9.86590385e-02 2.80197799e-01 4.09414768e-01 -8.53819251e-01 -2.58035839e-01 -4.06343281e-01 -2.50415355e-01 -1.70946848e+00 2.55371988e-01 -1.00083768e+00 -1.77593887e-01 -1.93826735e+00 3.70848984e-01 2.69895077e-01 -4.39113885e-01 1.63699254e-01 -7.22633362e-01 -4.26448554e-01 8.90077055e-02 -1.12928286e-01 -9.07286644e-01 1.01222217e+00 1.43097687e+00 -3.84905279e-01 2.46314868e-01 -2.67235935e-01 -7.26479352e-01 3.80731672e-01 5.74524820e-01 -3.40475470e-01 -3.66156220e-01 -1.39508700e+00 4.46475595e-01 1.75936073e-01 2.43370026e-01 -6.28004432e-01 5.05057991e-01 1.79765314e-01 2.11486205e-01 -6.89404547e-01 2.42073402e-01 -7.54752994e-01 -8.17947268e-01 4.41791296e-01 -7.24736154e-01 6.12140298e-01 2.45974194e-02 5.95766783e-01 -6.12646878e-01 -6.68411374e-01 4.31479722e-01 -2.50410050e-01 -7.08840966e-01 2.85336584e-01 4.17204797e-02 5.56735218e-01 4.40570474e-01 4.58181828e-01 -5.48195243e-01 -4.10272092e-01 -3.67063373e-01 7.49988556e-01 -1.98583081e-01 7.62398660e-01 8.61066937e-01 -1.23940349e+00 -9.25041199e-01 1.73769698e-01 2.20110431e-01 -5.68381585e-02 3.83721501e-01 6.10315204e-01 -5.27539909e-01 8.81603122e-01 6.04124032e-02 -2.44592965e-01 -8.31522584e-01 8.12898755e-01 3.63448501e-01 -5.13618588e-01 -3.04720640e-01 1.04514992e+00 1.46205202e-01 -5.59712827e-01 3.87078732e-01 -5.51827431e-01 -7.87770629e-01 2.82625586e-01 9.51644063e-01 1.36320233e-01 1.42011210e-01 -5.00220060e-01 -1.82266414e-01 5.79545498e-01 -3.05745989e-01 1.18145682e-01 1.15200841e+00 -1.55864939e-01 -4.91719186e-01 4.99969840e-01 1.60274947e+00 -3.00072014e-01 -4.79960620e-01 -5.22015154e-01 1.16860017e-01 -2.64668074e-02 3.56916040e-02 -7.07919896e-01 -7.89129078e-01 1.47446811e+00 5.12259662e-01 -7.75951594e-02 9.14952636e-01 1.03495754e-01 1.33569860e+00 8.57456505e-01 -1.56788379e-02 -7.98530042e-01 1.65319905e-01 9.54135358e-01 1.02585709e+00 -1.33541608e+00 -4.08343911e-01 2.38165110e-02 -3.48604500e-01 1.18021429e+00 7.40185797e-01 -2.90717512e-01 6.00411832e-01 -4.75474745e-01 3.81706841e-02 -3.31993371e-01 -1.08281887e+00 -2.42175207e-01 6.08995736e-01 -1.91198271e-02 4.87950414e-01 -2.79234141e-01 -7.75974631e-01 8.94377410e-01 -2.11662576e-01 -3.06453705e-01 1.05742559e-01 8.94561350e-01 -9.46653128e-01 -9.59882796e-01 4.98810783e-02 5.24832308e-01 -4.90685284e-01 -5.18345237e-01 -1.77271828e-01 4.49018985e-01 -7.90764391e-02 9.57230031e-01 1.01114079e-01 -2.78203458e-01 6.62127614e-01 6.02349937e-01 3.04103702e-01 -5.69052279e-01 -8.52617621e-01 -5.82369208e-01 -4.35864031e-02 -4.85988826e-01 -1.72584876e-01 -1.55486718e-01 -1.03430521e+00 -8.84361193e-03 -4.42102700e-01 5.35794735e-01 4.85196173e-01 1.11418438e+00 4.70362008e-01 5.68934858e-01 4.24385905e-01 6.91315383e-02 -9.22662258e-01 -1.41441965e+00 -1.29798919e-01 5.58706105e-01 4.66780156e-01 -2.99874395e-01 -2.18213290e-01 -3.30352515e-01]
[11.134993553161621, 8.080790519714355]
171b887c-f79f-49b4-9cd8-1acab1c10c8a
learning-to-communicate-using-contrastive
2307.01403
null
https://arxiv.org/abs/2307.01403v1
https://arxiv.org/pdf/2307.01403v1.pdf
Learning to Communicate using Contrastive Learning
Communication is a powerful tool for coordination in multi-agent RL. But inducing an effective, common language is a difficult challenge, particularly in the decentralized setting. In this work, we introduce an alternative perspective where communicative messages sent between agents are considered as different incomplete views of the environment state. By examining the relationship between messages sent and received, we propose to learn to communicate using contrastive learning to maximize the mutual information between messages of a given trajectory. In communication-essential environments, our method outperforms previous work in both performance and learning speed. Using qualitative metrics and representation probing, we show that our method induces more symmetric communication and captures global state information from the environment. Overall, we show the power of contrastive learning and the importance of leveraging messages as encodings for effective communication.
['Michael Noukhovitch', 'Jakob Foerster', 'Biswa Sengupta', 'Yat Long Lo']
2023-07-03
null
null
null
null
['contrastive-learning', 'contrastive-learning']
['computer-vision', 'methodology']
[ 1.82456877e-02 2.15073302e-01 -2.34581023e-01 -2.52381593e-01 -1.07973158e+00 -9.07828808e-01 9.95364130e-01 4.51895356e-01 -5.68013072e-01 9.95766759e-01 7.19419062e-01 -2.55234796e-03 -2.33330712e-01 -6.88602746e-01 -7.04148352e-01 -7.43195891e-01 -6.75926268e-01 5.50764143e-01 -3.91920894e-01 -4.36163783e-01 2.87488438e-02 3.21974576e-01 -1.06596172e+00 7.01437667e-02 2.33441517e-01 4.16438818e-01 2.45349839e-01 1.02162015e+00 -1.27019025e-02 1.39557350e+00 -9.72576797e-01 1.12446964e-01 2.60965019e-01 -6.92838252e-01 -9.99546111e-01 2.66533971e-01 -1.58477113e-01 -3.42666864e-01 -2.67888457e-01 8.06176364e-01 5.50708652e-01 6.10238686e-02 2.95333803e-01 -1.48223877e+00 -2.09888294e-01 1.21941805e+00 -2.18595475e-01 7.57315680e-02 7.01608002e-01 3.38860631e-01 1.19039035e+00 -5.41452095e-02 7.66707420e-01 1.46215379e+00 2.95349658e-01 5.90446591e-01 -1.39331257e+00 -6.71611845e-01 5.15678763e-01 -1.62007809e-01 -9.25531387e-01 -7.03108013e-01 7.02887416e-01 -3.17187726e-01 7.38905013e-01 2.88814753e-01 5.86758435e-01 1.05889308e+00 2.63670534e-01 8.27186584e-01 1.34603143e+00 -2.31456921e-01 4.74251032e-01 3.74730267e-02 -2.36080989e-01 6.50474191e-01 -8.66050124e-02 2.84329683e-01 -8.69752705e-01 -3.14450264e-01 5.67128539e-01 -1.57386407e-01 -2.48021662e-01 -3.17378879e-01 -1.40306377e+00 8.00905824e-01 2.00471699e-01 1.97921962e-01 -3.26180845e-01 5.93573391e-01 3.12779039e-01 1.18598437e+00 3.96269828e-01 7.84952700e-01 -4.46234018e-01 -5.45236170e-01 -6.69670328e-02 3.35610390e-01 1.28819609e+00 8.56235921e-01 9.90363181e-01 -8.05166364e-02 1.34249017e-01 1.90118119e-01 4.11974013e-01 5.28568923e-01 -3.69219705e-02 -1.43786728e+00 6.44043326e-01 4.34422255e-01 4.03596699e-01 -8.98083448e-01 -6.41321898e-01 -3.26523900e-01 -5.94453156e-01 2.15076610e-01 4.13029432e-01 -8.59992504e-01 -4.41554412e-02 2.28668404e+00 2.28200689e-01 1.47554591e-01 7.04922795e-01 5.16578555e-01 2.64124125e-01 8.58774304e-01 -2.71676123e-01 -6.18655026e-01 5.80776215e-01 -7.71383524e-01 -7.82214463e-01 -3.52471888e-01 9.96861637e-01 -3.54175657e-01 6.98425829e-01 7.50999302e-02 -1.29412711e+00 1.26673877e-01 -8.03880453e-01 3.48401338e-01 -9.69395414e-03 -5.00314653e-01 7.76493788e-01 1.72621414e-01 -1.42869234e+00 4.77049053e-01 -9.38679159e-01 -2.87627488e-01 1.43717930e-01 5.58679938e-01 -2.58559406e-01 1.67312637e-01 -8.85736704e-01 9.10027802e-01 1.04685433e-01 -3.27657253e-01 -1.40698516e+00 -3.17827642e-01 -9.75229681e-01 -3.03220693e-02 5.25846601e-01 -6.40649557e-01 1.53910041e+00 -1.05240619e+00 -2.02381682e+00 3.37282747e-01 -6.76179826e-02 -5.12162685e-01 4.45512712e-01 1.34694517e-01 2.05273136e-01 5.72265126e-02 6.96179718e-02 4.27525431e-01 4.44804221e-01 -1.65858245e+00 -6.73606694e-01 -1.18038349e-01 7.86325991e-01 6.69746757e-01 9.16808844e-02 -7.93516934e-02 1.05515189e-01 -1.52936265e-01 -2.37327620e-01 -1.01846826e+00 -4.82259810e-01 -9.64065343e-02 -2.27237269e-01 -2.29668558e-01 6.49976850e-01 -4.57498319e-02 7.40352273e-01 -1.97279024e+00 7.06372559e-01 2.78541237e-01 5.05633116e-01 -4.09374386e-01 -5.48789501e-01 1.02544177e+00 5.04444242e-01 1.13507986e-01 -3.32320035e-02 -7.67234981e-01 2.89168358e-01 5.12780964e-01 -3.61897767e-01 8.07036579e-01 -7.70173594e-02 7.93530822e-01 -1.27859676e+00 -1.55839622e-01 5.27667142e-02 2.47062132e-01 -4.44994867e-01 5.09316444e-01 -4.05180156e-01 1.12825358e+00 -6.13859236e-01 -1.31925512e-02 5.65878823e-02 -2.58546323e-01 7.08254635e-01 6.08350873e-01 -1.56836018e-01 5.74140131e-01 -1.20587730e+00 1.86999047e+00 -1.01626611e+00 7.13338852e-01 6.72974944e-01 -7.84088790e-01 5.89650869e-01 4.36123043e-01 5.77507615e-01 -7.03432858e-01 1.35187611e-01 -1.61158577e-01 1.71763450e-01 -3.05538893e-01 7.33166188e-02 -1.09757304e-01 -5.08684278e-01 1.22297776e+00 -2.00537130e-01 -4.87582386e-01 1.64415300e-01 4.86831903e-01 1.19942605e+00 -2.40745783e-01 3.87378365e-01 -1.46777526e-01 2.19474599e-01 -3.31326365e-01 4.14956003e-01 1.09637237e+00 1.68441213e-03 -1.28466323e-01 9.86076474e-01 -2.24223316e-01 -7.93648601e-01 -8.81638944e-01 5.20031810e-01 1.24325824e+00 2.51085341e-01 -7.94181466e-01 -4.33431894e-01 -5.93749881e-01 -1.43852919e-01 5.09318888e-01 -4.71576869e-01 -1.15240686e-01 -7.67759085e-01 -5.18976927e-01 4.53555435e-01 1.27316177e-01 3.31604034e-01 -9.53218281e-01 -9.38847780e-01 2.92966038e-01 -1.45277634e-01 -9.91846800e-01 -5.48180759e-01 1.74955636e-01 -4.62158024e-01 -9.59531963e-01 -2.48191833e-01 -5.35165489e-01 5.44377208e-01 1.43997297e-01 1.11053848e+00 9.71447453e-02 1.71600923e-01 1.24201262e+00 -2.69845635e-01 -3.08459312e-01 -9.86425102e-01 4.44889255e-02 7.18731433e-02 2.03972142e-02 -4.30355906e-01 -7.11053491e-01 -2.64266908e-01 -1.05452295e-02 -8.05654883e-01 8.34801197e-02 4.36465383e-01 6.67331457e-01 5.08313533e-03 6.04558550e-02 6.39641285e-01 -6.52092159e-01 1.14831614e+00 -6.84407175e-01 -8.24234605e-01 4.65615653e-02 -3.25270355e-01 4.51757938e-01 8.21793377e-01 -3.17735791e-01 -8.56384277e-01 -1.13265499e-01 3.29538763e-01 3.53381574e-01 -1.45614401e-01 4.08544034e-01 9.97123718e-02 -2.10322753e-01 2.91257679e-01 3.58168930e-01 2.79881924e-01 -8.64074677e-02 5.90938807e-01 4.85262036e-01 -1.47647470e-01 -9.83232617e-01 5.57155192e-01 5.91647863e-01 1.00915127e-01 -7.58290231e-01 -5.10296762e-01 -1.41179040e-01 -3.23574066e-01 -2.88441509e-01 3.85409117e-01 -9.61242318e-01 -1.41530383e+00 3.77757967e-01 -1.36879683e+00 -7.49220073e-01 -2.86200523e-01 5.25013030e-01 -9.51148808e-01 2.14185789e-01 -6.11229360e-01 -1.02420378e+00 1.24457419e-01 -1.49682486e+00 1.01035035e+00 -1.61143199e-01 -1.47713795e-01 -1.30706346e+00 5.93235672e-01 -2.30601877e-01 6.72864377e-01 1.48935676e-01 6.18794501e-01 -6.66800380e-01 -9.29387331e-01 2.71761149e-01 2.31490418e-01 -1.73307464e-01 3.73409063e-01 -4.50929999e-01 -7.16616929e-01 -7.18461812e-01 -6.81465864e-02 -9.12629902e-01 4.71331626e-01 1.23460732e-01 5.32889664e-01 -1.02042282e+00 -1.52283370e-01 3.16414833e-01 1.21261120e+00 2.25682229e-01 1.04452699e-01 3.24208260e-01 2.61848927e-01 8.51861596e-01 9.98289436e-02 9.85855758e-01 1.05278504e+00 5.57018161e-01 4.74716723e-01 1.55474767e-01 1.25734702e-01 -1.88998833e-01 7.18949914e-01 9.53932166e-01 2.68195510e-01 -3.73422682e-01 -7.18881726e-01 2.83073336e-01 -2.02383924e+00 -9.87797022e-01 7.00703025e-01 1.99783385e+00 1.19431233e+00 -1.29131630e-01 1.35664448e-01 -3.36843222e-01 2.97899634e-01 5.13708651e-01 -6.25096262e-01 -3.19418281e-01 -1.68211102e-01 -4.41605821e-02 3.74238551e-01 1.07522023e+00 -8.11944485e-01 7.00244009e-01 7.06709003e+00 6.93801343e-02 -1.07291186e+00 1.97246552e-01 3.69783074e-01 -3.33740413e-01 -4.53599960e-01 1.08025454e-01 -5.06737292e-01 3.04212213e-01 1.02074814e+00 -2.70304024e-01 1.11974251e+00 2.61011124e-01 4.55902159e-01 -1.65940732e-01 -1.63091266e+00 8.29805195e-01 -1.15744509e-01 -1.55448890e+00 -1.22990787e-01 1.56140938e-01 8.01273227e-01 1.98251367e-01 7.66640902e-02 1.99802935e-01 1.16287374e+00 -1.08941710e+00 7.07003653e-01 3.90254021e-01 2.57197261e-01 -6.76446497e-01 4.66827244e-01 7.50792205e-01 -1.23156703e+00 -2.11000711e-01 1.64023727e-01 -6.98990643e-01 8.63594040e-02 -1.29049242e-01 -8.19702923e-01 4.17695642e-01 5.80846928e-02 8.75491977e-01 -4.77548242e-02 5.48327208e-01 -5.02752848e-02 3.73155773e-01 -3.68280739e-01 -2.65659392e-01 5.10919034e-01 -2.90967405e-01 6.71380162e-01 9.84841168e-01 -8.30244869e-02 -6.67331666e-02 8.48615825e-01 7.78353870e-01 -2.29275450e-01 -2.37904713e-01 -1.03082585e+00 -1.90998033e-01 6.99476779e-01 8.49707305e-01 -3.15483958e-01 -2.47369692e-01 -3.18308562e-01 6.10350192e-01 6.80449963e-01 5.08218765e-01 -3.64751607e-01 1.47627741e-01 9.09218311e-01 -6.52375221e-01 -1.62179828e-01 -4.95065898e-01 1.86671108e-01 -1.01563931e+00 -1.10366317e-02 -1.18708420e+00 1.98589757e-01 -1.01149887e-01 -9.97644544e-01 3.02609116e-01 1.17674358e-01 -9.74368155e-01 -7.34090209e-01 1.59276165e-02 -6.15807116e-01 3.61571193e-01 -1.58844113e+00 -9.46172714e-01 2.58786887e-01 5.19823015e-01 5.60489118e-01 -7.41168186e-02 9.12382662e-01 -1.33277729e-01 -3.97111952e-01 2.39386648e-01 1.56797841e-01 3.04431259e-03 3.81484389e-01 -1.37467647e+00 4.00825078e-03 4.02714908e-01 3.18113655e-01 6.53489947e-01 8.13888907e-01 -2.81281441e-01 -2.04251456e+00 -9.00187790e-01 5.68948805e-01 -4.49852288e-01 8.74702930e-01 -4.33045119e-01 -3.38138938e-01 1.10785007e+00 7.13558435e-01 -4.55878764e-01 5.65666020e-01 -1.96887683e-02 -2.01526701e-01 -9.43528265e-02 -9.98939633e-01 7.30885744e-01 8.16621125e-01 -6.91368043e-01 -3.26739401e-01 5.31386197e-01 9.06511605e-01 -4.13065165e-01 -8.14676583e-01 -2.65546292e-01 2.88301468e-01 -6.68219149e-01 5.51507831e-01 -6.38232350e-01 3.26774120e-02 -1.50249258e-01 -2.86020160e-01 -1.87066996e+00 2.32608244e-01 -1.39012647e+00 1.03274420e-01 1.01554120e+00 4.28786516e-01 -9.74911094e-01 4.65090603e-01 4.24168557e-01 2.27692828e-01 -6.31155431e-01 -1.03638697e+00 -5.37961364e-01 2.92522371e-01 -2.36127377e-01 6.14743531e-01 7.94851542e-01 4.60463405e-01 4.58632469e-01 -5.76068819e-01 4.15184885e-01 6.58641696e-01 9.57152918e-02 9.76891220e-01 -7.84228265e-01 -6.19442523e-01 -4.31194633e-01 -3.83831337e-02 -1.23572612e+00 7.22241640e-01 -8.33904386e-01 1.39904365e-01 -1.44304240e+00 2.52688527e-01 -5.92523575e-01 1.44844621e-01 3.10094535e-01 4.88880455e-01 -4.27174002e-01 6.29929304e-01 1.33871749e-01 -1.16473913e+00 6.47226453e-01 1.21931267e+00 -3.24819922e-01 -3.06063712e-01 1.31616648e-02 -6.78957343e-01 6.60993159e-01 9.97028589e-01 -4.67681408e-01 -5.87507486e-01 -6.81003511e-01 6.72369182e-01 5.35949051e-01 2.84975231e-01 -5.27941287e-01 6.16851807e-01 -4.33066070e-01 -4.10041273e-01 -7.25780055e-02 6.12926185e-01 -7.12636650e-01 1.54305920e-02 6.72616065e-01 -1.18494177e+00 4.39287454e-01 -7.95775875e-02 8.18277121e-01 -7.63264345e-03 4.60412540e-02 3.83071065e-01 -5.30023873e-01 -4.44142878e-01 1.12677380e-01 -7.64533997e-01 4.74484414e-01 1.01281142e+00 5.89779198e-01 -3.81166905e-01 -1.17042410e+00 -5.36451101e-01 7.37054169e-01 5.11067331e-01 8.12491998e-02 5.95956743e-01 -8.91117156e-01 -8.45061362e-01 -1.06074205e-02 -6.13390356e-02 2.61753537e-02 -3.12230885e-01 6.16224289e-01 2.09563132e-03 2.75040090e-01 1.00029379e-01 -4.00809944e-01 -1.01617146e+00 -3.74412648e-02 4.73505139e-01 -3.77195716e-01 -4.84940499e-01 5.25262713e-01 1.45776585e-01 -6.90143347e-01 6.29866540e-01 -3.75037193e-01 -4.72034924e-02 1.05453385e-02 5.70680916e-01 3.62131417e-01 -5.10494292e-01 -4.24729854e-01 -1.50033608e-01 3.80427539e-01 -3.33838649e-02 -6.30877376e-01 1.41258419e+00 -5.42971432e-01 -3.49503815e-01 6.63670182e-01 1.28093171e+00 8.81383419e-02 -1.59661436e+00 -3.81684065e-01 5.11231497e-02 -2.62457222e-01 -1.12589702e-01 -6.29230440e-01 -8.72820914e-01 3.35153788e-01 1.67609036e-01 4.66108680e-01 6.41143620e-01 6.28283858e-01 4.71324205e-01 9.57197666e-01 7.90795922e-01 -8.40014696e-01 3.50456238e-01 6.01978183e-01 8.69126022e-01 -1.16241932e+00 -2.35529736e-01 1.12497352e-01 -6.63682938e-01 8.72347891e-01 2.25427911e-01 -5.52299358e-02 4.21963215e-01 7.52542138e-01 1.78319201e-01 -1.79698214e-01 -1.42659569e+00 -2.72957116e-01 -4.94306535e-01 7.13524520e-01 1.39813334e-01 1.97042421e-01 8.36391822e-02 -4.23121899e-02 -2.61291981e-01 -4.20641541e-01 1.05228961e+00 1.28299296e+00 -3.43787223e-01 -1.25796461e+00 -6.30317479e-02 9.90721062e-02 -2.75723577e-01 1.10584550e-01 -5.81194580e-01 6.96897686e-01 -4.12943155e-01 1.37531328e+00 6.48073032e-02 -1.48879737e-01 -1.27578489e-02 -4.03632194e-01 5.56533813e-01 -7.52421021e-01 -7.27921486e-01 -2.14923307e-01 3.13565910e-01 -8.17955911e-01 -8.80152166e-01 -6.33053958e-01 -1.26540208e+00 -3.77068162e-01 5.92090152e-02 3.37261528e-01 5.43311417e-01 1.11123633e+00 5.08527935e-01 4.96097714e-01 1.16580796e+00 -8.93838823e-01 -1.02260351e+00 -5.40295184e-01 -3.41482103e-01 2.82584608e-01 1.13183105e+00 -4.72318232e-01 -5.73817790e-01 -2.97097981e-01]
[3.876816987991333, 1.9944801330566406]
41eeff84-a2b1-4d89-8edf-cb408f169e05
stableface-analyzing-and-improving-motion
2208.13717
null
https://arxiv.org/abs/2208.13717v1
https://arxiv.org/pdf/2208.13717v1.pdf
StableFace: Analyzing and Improving Motion Stability for Talking Face Generation
While previous speech-driven talking face generation methods have made significant progress in improving the visual quality and lip-sync quality of the synthesized videos, they pay less attention to lip motion jitters which greatly undermine the realness of talking face videos. What causes motion jitters, and how to mitigate the problem? In this paper, we conduct systematic analyses on the motion jittering problem based on a state-of-the-art pipeline that uses 3D face representations to bridge the input audio and output video, and improve the motion stability with a series of effective designs. We find that several issues can lead to jitters in synthesized talking face video: 1) jitters from the input 3D face representations; 2) training-inference mismatch; 3) lack of dependency modeling among video frames. Accordingly, we propose three effective solutions to address this issue: 1) we propose a gaussian-based adaptive smoothing module to smooth the 3D face representations to eliminate jitters in the input; 2) we add augmented erosions on the input data of the neural renderer in training to simulate the distortion in inference to reduce mismatch; 3) we develop an audio-fused transformer generator to model dependency among video frames. Besides, considering there is no off-the-shelf metric for measuring motion jitters in talking face video, we devise an objective metric (Motion Stability Index, MSI), to quantitatively measure the motion jitters by calculating the reciprocal of variance acceleration. Extensive experimental results show the superiority of our method on motion-stable face video generation, with better quality than previous systems.
['Li Song', 'Sheng Zhao', 'Yuchao Zhang', 'Runnan Li', 'Liyang Chen', 'Xu Tan', 'Jun Ling']
2022-08-29
null
null
null
null
['video-generation', 'talking-face-generation']
['computer-vision', 'computer-vision']
[-2.30402611e-02 -2.66249239e-01 8.36660936e-02 -2.49991640e-01 -6.75661683e-01 -4.03825909e-01 3.50718141e-01 -7.91062295e-01 4.35517021e-02 4.11071688e-01 3.32379788e-01 -1.25236601e-01 1.19443461e-02 -4.10032123e-01 -7.08926737e-01 -7.47528374e-01 2.34284326e-01 -2.86861658e-01 3.69899184e-01 -9.71477628e-02 2.49379978e-01 4.70658064e-01 -1.95887256e+00 3.47238421e-01 8.69493544e-01 1.06615186e+00 2.75734842e-01 5.86606979e-01 -3.40857744e-01 5.75776696e-01 -1.04847145e+00 -2.99256772e-01 9.35199559e-02 -7.54255593e-01 -3.14118356e-01 1.69016004e-01 2.95686632e-01 -6.11564636e-01 -2.88641483e-01 1.15476143e+00 8.12472165e-01 -1.59243152e-01 6.06173515e-01 -1.51801181e+00 -5.80121696e-01 3.40710402e-01 -5.94865680e-01 2.41516069e-01 4.28138912e-01 3.20180506e-01 1.16710700e-01 -9.20187294e-01 4.98322189e-01 1.69452262e+00 5.92292190e-01 8.08920383e-01 -9.18714285e-01 -8.85993719e-01 -4.96839359e-02 4.37229991e-01 -1.49726713e+00 -9.75105464e-01 8.25733423e-01 -3.93493980e-01 5.33379018e-01 2.46716976e-01 4.62991118e-01 1.09347975e+00 1.39259294e-01 4.29757863e-01 6.56053960e-01 -3.65356207e-01 6.85958415e-02 7.49450848e-02 -4.00569886e-01 4.63472635e-01 -2.86058515e-01 1.76913232e-01 -6.40434384e-01 1.21963002e-01 8.94568205e-01 -4.51970369e-01 -5.31934083e-01 2.90207088e-01 -7.46057689e-01 4.39295501e-01 -1.15522653e-01 2.38045901e-01 -1.78570643e-01 1.32688120e-01 4.38374966e-01 2.02380985e-01 5.25894046e-01 -2.52607524e-01 -2.22914726e-01 -3.04590970e-01 -1.18380415e+00 1.46812484e-01 3.79685968e-01 1.00531995e+00 4.22600627e-01 4.91927952e-01 -3.51520181e-01 9.04726803e-01 5.53836346e-01 6.94608629e-01 6.45033181e-01 -1.06471646e+00 4.74105895e-01 1.51930526e-01 -1.03485130e-01 -1.02477634e+00 5.94129879e-03 1.46864623e-01 -5.91243863e-01 3.43252778e-01 1.98806331e-01 -1.89539418e-01 -7.88886547e-01 1.92550683e+00 3.32470626e-01 4.42535788e-01 -5.37669025e-02 9.65232670e-01 9.89571273e-01 9.13355470e-01 -2.02941701e-01 -7.69892395e-01 1.17852545e+00 -7.50492513e-01 -1.34554005e+00 2.03579858e-01 1.73485689e-02 -1.31847453e+00 1.04375887e+00 2.09688514e-01 -1.41187322e+00 -1.02929652e+00 -1.09757149e+00 7.27220848e-02 1.38402835e-01 3.14821512e-01 2.71871984e-02 9.52673674e-01 -1.12588096e+00 6.06146336e-01 -7.11184144e-01 -1.86853670e-02 8.08208436e-02 2.90475726e-01 -6.05045035e-02 2.85356641e-01 -1.16429698e+00 7.13386714e-01 -6.39727190e-02 2.24445552e-01 -8.94402027e-01 -8.22530508e-01 -7.26795375e-01 3.65243778e-02 3.32811505e-01 -5.83976746e-01 1.31129801e+00 -1.13649213e+00 -2.13721204e+00 3.95449877e-01 -6.02850497e-01 1.00974135e-01 5.90244532e-01 -1.83595985e-01 -6.66377962e-01 2.13818729e-01 -1.68478772e-01 7.23041594e-01 1.36960840e+00 -1.14695787e+00 -5.29317200e-01 -1.03885166e-01 -3.11738759e-01 1.15560040e-01 -3.34613770e-01 2.77981520e-01 -8.69021952e-01 -9.89056826e-01 -2.25571003e-02 -7.93181419e-01 3.13111067e-01 9.42093655e-02 -8.30666050e-02 -1.37092933e-01 1.36537528e+00 -7.84698665e-01 1.48252416e+00 -2.46589065e+00 -9.06365141e-02 -2.39613712e-01 -2.61879683e-01 7.26223469e-01 -1.81568518e-01 8.98748934e-02 -9.98993739e-02 1.83174849e-01 8.30909535e-02 -4.19146478e-01 -2.36532807e-01 3.00461724e-02 -4.15474117e-01 2.60736525e-01 3.74482661e-01 4.06120360e-01 -4.79649991e-01 -5.92831135e-01 2.44593665e-01 9.39723432e-01 -6.51220143e-01 4.55344498e-01 -1.17289901e-01 3.58688772e-01 -2.11784244e-01 5.75860262e-01 1.05792677e+00 3.14769983e-01 -2.68654704e-01 -4.99055654e-01 -1.09084547e-01 2.22182661e-01 -1.39134312e+00 1.53440392e+00 -3.77286166e-01 6.76970124e-01 3.06602180e-01 -3.84449899e-01 1.05414486e+00 5.82023919e-01 3.43752295e-01 -5.08551836e-01 2.32074603e-01 1.29847556e-01 -1.64551362e-01 -9.59408760e-01 3.11818540e-01 -3.17811482e-02 5.14159858e-01 5.83848730e-02 -2.40279976e-02 -1.85989991e-01 1.01116011e-02 -1.19311571e-01 7.24877357e-01 3.43518168e-01 -4.30958778e-01 -1.85016111e-01 8.32299054e-01 -6.68834686e-01 8.23639572e-01 -4.48233932e-02 -2.90566951e-01 9.13851917e-01 6.04731143e-01 7.01289922e-02 -8.35553050e-01 -1.17140388e+00 1.35514081e-01 6.04449689e-01 1.41044125e-01 -5.92041135e-01 -1.26498532e+00 -2.83008695e-01 -2.96254307e-01 4.07482922e-01 -1.50537968e-01 -1.90118819e-01 -6.36476994e-01 -3.71230692e-01 7.01352537e-01 4.47956979e-01 5.29780388e-01 -8.68803859e-01 -3.22086900e-01 7.65313953e-02 -4.31326479e-01 -1.13979983e+00 -1.02216911e+00 -5.81476688e-01 -6.50279403e-01 -8.77727211e-01 -8.10478806e-01 -7.80541062e-01 6.00425959e-01 3.08717012e-01 5.88255227e-01 -3.60097028e-02 -3.19757998e-01 9.82640758e-02 -1.91242635e-01 -2.94337064e-01 -6.46894574e-01 -5.07847130e-01 3.51581931e-01 1.36126503e-01 -5.91488406e-02 -5.13973355e-01 -6.70545638e-01 7.28013456e-01 -9.66545522e-01 -9.03727412e-02 2.71881968e-01 4.82704997e-01 3.22970718e-01 3.16453367e-01 6.95581794e-01 -9.09161642e-02 7.84296453e-01 -5.99853173e-02 -7.36134052e-01 9.54870693e-03 -4.32451397e-01 -6.12792298e-02 6.45373583e-01 -7.73912787e-01 -1.38074052e+00 -1.45184189e-01 -4.20668185e-01 -9.35449898e-01 1.45869493e-01 3.87733094e-02 -6.50672913e-01 9.14068818e-02 3.85282844e-01 1.52967637e-02 4.92115676e-01 -3.91008288e-01 2.60321140e-01 9.67821896e-01 7.30667710e-01 -2.64281958e-01 6.63037360e-01 3.99137884e-01 -1.93815753e-01 -1.04391003e+00 -1.21662296e-01 -4.20547314e-02 -5.01549803e-02 -4.28982496e-01 8.06937695e-01 -9.70648170e-01 -1.13434553e+00 7.28425145e-01 -1.36797762e+00 -1.99098773e-02 1.82213426e-01 5.66914380e-01 -6.30862057e-01 5.60674906e-01 -7.24523783e-01 -9.10780489e-01 -4.36368942e-01 -1.64680684e+00 9.33764517e-01 5.29230118e-01 -9.97822285e-02 -4.06983465e-01 -2.59005696e-01 3.03027987e-01 5.76061785e-01 -1.30978122e-01 6.73318863e-01 2.64925092e-01 -5.57925403e-01 3.39227945e-01 -3.04755270e-02 7.03921080e-01 4.18397695e-01 6.67600393e-01 -1.16976941e+00 -2.35722840e-01 3.36858869e-01 2.88305938e-01 4.60928559e-01 7.35820949e-01 1.03741276e+00 -4.43740994e-01 -2.44837776e-01 7.61937737e-01 9.82334018e-01 5.60117602e-01 9.82377887e-01 -1.77812025e-01 3.52556169e-01 6.84291542e-01 5.89483142e-01 3.55919987e-01 1.00272084e-02 9.83492255e-01 2.76046246e-01 6.29067570e-02 -6.46397114e-01 -3.07200640e-01 9.18396294e-01 9.88531649e-01 6.53820531e-03 -3.23219955e-01 -2.85528868e-01 2.75017232e-01 -1.54011869e+00 -1.08079886e+00 -2.58752350e-02 2.21048689e+00 8.19037139e-01 6.82963654e-02 8.96210819e-02 2.83332884e-01 1.13785958e+00 1.07281230e-01 -2.91205615e-01 -3.81440967e-01 2.79261656e-02 2.26912685e-02 5.64662851e-02 5.68519354e-01 -6.88154697e-01 8.92290652e-01 6.21374416e+00 1.09296739e+00 -1.58141553e+00 -2.33196355e-02 4.71201241e-01 -3.84485006e-01 -3.14006805e-01 -1.05794802e-01 -1.06467271e+00 9.87863719e-01 9.26123977e-01 -2.31550723e-01 4.40967202e-01 6.90360665e-01 7.77735651e-01 5.38353249e-02 -8.89992952e-01 1.30696177e+00 2.21379116e-01 -1.21787441e+00 9.14966315e-02 -9.82979238e-02 4.35427845e-01 -6.56136870e-01 2.87333399e-01 -3.27657093e-03 -4.46568280e-01 -9.25616205e-01 9.60436463e-01 5.86450577e-01 1.01187181e+00 -9.60517406e-01 4.37005728e-01 1.23029314e-01 -1.36021733e+00 1.15455247e-01 -2.40921304e-01 3.49822968e-01 4.26691920e-01 5.12649417e-01 -7.24030793e-01 4.26254094e-01 7.97552109e-01 1.99803814e-01 -2.83757895e-01 9.71748233e-01 -2.66285896e-01 4.49073762e-01 -2.91825801e-01 1.34009004e-01 -2.20854521e-01 -9.12644565e-02 6.14143312e-01 1.02068830e+00 6.94486618e-01 6.70163939e-03 -3.55581611e-01 9.63334203e-01 5.83494678e-02 -2.81763431e-02 -3.72301340e-01 6.11158013e-02 7.02246189e-01 1.03200042e+00 -4.19581532e-01 -1.38171548e-02 -2.85411417e-01 9.28319633e-01 -5.01798272e-01 3.88171911e-01 -1.19883347e+00 -4.08606708e-01 1.01557159e+00 3.09248686e-01 3.14217031e-01 -1.52281880e-01 1.08642317e-01 -7.34785497e-01 3.10797870e-01 -9.37108099e-01 -9.34350118e-02 -9.23785150e-01 -8.99483263e-01 6.94881082e-01 -1.45625323e-01 -1.35730541e+00 -5.37934661e-01 -3.15749019e-01 -7.65131474e-01 9.34070408e-01 -1.28622639e+00 -8.05418015e-01 -2.51545906e-01 7.55266845e-01 8.11726153e-01 -1.68935657e-01 4.52947259e-01 7.28029609e-01 -5.61297238e-01 8.50366592e-01 -1.82312906e-01 -1.11940652e-01 1.05207741e+00 -3.84326696e-01 2.93261647e-01 9.15453076e-01 -3.53354305e-01 4.46921945e-01 7.31744885e-01 -5.51639795e-01 -1.48357105e+00 -8.67575169e-01 6.17605507e-01 -1.66271985e-01 2.36064166e-01 -2.23624870e-01 -9.24347699e-01 2.31824696e-01 6.51582852e-02 -4.57339846e-02 3.83948296e-01 -5.64633548e-01 -7.95230269e-02 -4.14656132e-01 -1.13940108e+00 5.81032097e-01 9.23151612e-01 -5.70094764e-01 -4.33988482e-01 -2.00147539e-01 9.21515822e-01 -4.44327652e-01 -5.73733449e-01 5.95610261e-01 5.90422451e-01 -1.19687426e+00 8.19736540e-01 1.03149988e-01 3.22233319e-01 -7.47723162e-01 3.52734327e-02 -1.17278290e+00 -1.83020830e-02 -1.11128366e+00 5.87352663e-02 2.02971053e+00 1.10728703e-01 -3.19363922e-01 6.61804676e-01 3.91099006e-01 -2.29782671e-01 -4.60156769e-01 -1.03214216e+00 -7.66279340e-01 -2.17854917e-01 -2.73156136e-01 6.38171315e-01 4.98833686e-01 -3.60683091e-02 1.80354461e-01 -4.38957810e-01 3.05024952e-01 3.68840456e-01 -3.59466851e-01 7.73519993e-01 -6.78215921e-01 -5.05738370e-02 -4.71287251e-01 -2.82372773e-01 -1.01195347e+00 8.60456154e-02 -2.10380286e-01 2.27896526e-01 -1.18680072e+00 -1.82412982e-01 -4.35427241e-02 2.90240079e-01 -7.43477568e-02 -2.04497293e-01 -1.11026786e-01 3.53103697e-01 1.41177326e-01 2.67833136e-02 7.63986647e-01 1.23828685e+00 5.53494804e-02 -3.14351767e-01 9.10787880e-02 -3.53812397e-01 8.20174098e-01 3.17451537e-01 -2.02390626e-01 -7.82281339e-01 -5.71598947e-01 -2.40126163e-01 5.37558794e-01 1.54498219e-01 -1.23753858e+00 2.77050376e-01 1.69124156e-02 3.62788320e-01 -7.02398777e-01 4.66751426e-01 -7.86521256e-01 4.23093408e-01 2.36817688e-01 -3.49225923e-02 8.61658826e-02 4.21473712e-01 1.37244195e-01 -3.56294274e-01 -1.77893728e-01 1.14055276e+00 2.73135424e-01 -3.61429125e-01 2.38797575e-01 -5.24435580e-01 -1.60865337e-01 9.39758480e-01 -1.67120367e-01 -3.89636755e-01 -5.87471962e-01 -4.27825063e-01 -1.40593484e-01 4.86424237e-01 6.56047881e-01 6.77523315e-01 -1.41015363e+00 -5.20442426e-01 6.05754554e-01 -4.88680840e-01 -2.28650674e-01 5.27796447e-01 4.68274921e-01 -4.79076684e-01 1.26784295e-01 -1.94049671e-01 -6.87772691e-01 -1.53112721e+00 5.85862398e-01 3.16097260e-01 4.73866731e-01 -3.36814314e-01 9.55740094e-01 1.80311918e-01 2.48557121e-01 5.12997568e-01 -3.49386930e-01 -6.14334755e-02 1.10386238e-01 8.32424462e-01 4.91459012e-01 1.49993718e-01 -6.89169049e-01 -4.01856095e-01 8.82600665e-01 1.34309590e-01 -2.49661073e-01 8.22989464e-01 -3.53729576e-01 6.15907609e-02 5.70287965e-02 1.30044210e+00 1.41312256e-01 -1.52554619e+00 3.13055813e-01 -4.72529382e-01 -6.17480040e-01 2.97135599e-02 -4.01075423e-01 -1.27811861e+00 9.14121628e-01 8.25512290e-01 3.56227793e-02 1.49720204e+00 -3.25403512e-01 9.59697783e-01 -3.94548953e-01 2.40342859e-02 -1.17008388e+00 3.08332443e-01 3.35814983e-01 8.45691323e-01 -7.07468688e-01 -2.93841600e-01 -7.18909919e-01 -5.51874340e-01 1.21166825e+00 5.81191361e-01 3.38313550e-01 5.99630833e-01 8.25663328e-01 2.93801010e-01 3.17983180e-01 -7.26131082e-01 2.42497921e-02 1.21234350e-01 6.44948184e-01 4.16592121e-01 -5.27105510e-01 -2.85975516e-01 6.94853306e-01 -3.58190864e-01 2.96168357e-01 3.92874509e-01 6.27373993e-01 -3.95965368e-01 -9.58012640e-01 -7.20694184e-01 -2.09466949e-01 -6.26552999e-01 2.34141219e-02 -5.16464897e-02 4.06233549e-01 2.45182827e-01 1.31410944e+00 1.66702539e-01 -5.49807608e-01 4.94657099e-01 1.35110304e-01 5.23734152e-01 -1.12790484e-02 -3.02215427e-01 5.80516875e-01 -1.92046210e-01 -5.36441028e-01 -3.19012016e-01 -3.62573028e-01 -1.24504578e+00 -5.92918873e-01 -4.05906081e-01 3.26655172e-02 9.24250305e-01 7.40981698e-01 6.20820642e-01 7.03221679e-01 8.70628476e-01 -9.09822524e-01 -4.28759336e-01 -9.62898791e-01 -3.80853802e-01 2.93316901e-01 3.84343237e-01 -6.60415530e-01 -7.31458962e-01 3.92322600e-01]
[13.275999069213867, -0.4048531651496887]
7366e7a5-3882-4536-85df-2c407230ba80
state-of-the-art-models-for-fake-news
null
null
https://ieeexplore.ieee.org/document/9089487
https://www.researchgate.net/publication/340478553_State_of_the_Art_Models_for_Fake_News_Detection_Tasks
State of the Art Models for Fake News Detection Tasks
This paper presents state of the art methods for addressing three important challenges in automated fake news detection: fake news detection, domain identification, and bot identification in tweets. The proposed solutions achieved first place in a recent international competition on fake news. For fake news detection, we present two models. The winning model in the competition combines similarity between the embedding of each article’s title and the embedding of the top five corresponding google search results. The new model relies on advances in Natural Language Understanding (NLU) end to end deep learning models to identify stylistic differences between legitimate and fake news articles. This second model was developed after the competition and outperforms the winning approach. For news domain detection, the winning model is a hybrid approach composed of named entity features concatenated with semantic embeddings derived from end to end models. For twitter bot detection, we propose to use the following features: duration between account creation and tweet date, presence of a tweet’s link, presence of user’s location, other tweet’s features, and the tweets’ metadata. Experiments include insights into the importance of the different features and the results indicate the superior performances of all proposed models.
['Wissam Antoun ; Fady Baly ; Rim Achour ; Amir Hussein ; Hazem Hajj']
2020-05-11
null
null
null
null
['twitter-bot-detection']
['miscellaneous']
[-2.11312458e-01 -6.51979372e-02 -5.42098463e-01 5.68137318e-02 -4.80493933e-01 -5.88480234e-01 1.22671866e+00 4.71832663e-01 -4.77338493e-01 5.13424933e-01 4.15437818e-01 -1.92952767e-01 1.79917619e-01 -6.70825422e-01 -3.73930484e-01 -2.20691338e-01 1.13059863e-01 4.70300555e-01 4.37478960e-01 -5.68398893e-01 9.80422199e-01 3.48666698e-01 -1.21540129e+00 5.72807312e-01 5.39339960e-01 9.84962285e-01 -1.86545998e-01 4.32785422e-01 -4.30058360e-01 1.03921449e+00 -1.03216064e+00 -6.34151220e-01 1.48300454e-01 -1.82203248e-01 -8.72717977e-01 -2.69317180e-01 3.11731130e-01 -5.59002101e-01 -8.59867454e-01 9.88091886e-01 3.48058522e-01 -5.14707386e-01 8.01801026e-01 -1.60849142e+00 -8.42027783e-01 6.00151777e-01 -5.44844449e-01 6.61233902e-01 2.95550168e-01 1.00314014e-01 9.82496262e-01 -7.08650708e-01 9.75535989e-01 1.03999257e+00 9.70467985e-01 4.17844504e-01 -6.84565842e-01 -5.92853606e-01 -3.34630013e-01 2.59136856e-01 -1.03425026e+00 -3.43414515e-01 7.36962974e-01 -1.04627681e+00 8.02864969e-01 -1.15343602e-02 4.35729384e-01 1.50525486e+00 4.23676856e-02 8.26158643e-01 1.09323430e+00 -2.47869655e-01 -1.34185642e-01 9.45980430e-01 4.36856091e-01 6.04754150e-01 3.36006850e-01 7.72937909e-02 -5.55322647e-01 -7.99769521e-01 2.49173582e-01 -4.99198586e-02 2.48005673e-01 1.51871070e-01 -1.03856647e+00 1.39522064e+00 3.56020480e-01 6.34429336e-01 -3.57153207e-01 5.49606569e-02 9.21183288e-01 3.29688251e-01 9.38800633e-01 8.04216266e-01 -2.31738880e-01 -1.63147554e-01 -1.13073123e+00 5.52671432e-01 9.64869499e-01 6.20700002e-01 4.68061060e-01 -1.84375241e-01 -4.10936624e-01 6.64991736e-01 1.70361280e-01 4.09830511e-01 8.85868609e-01 -1.01515658e-01 6.72831476e-01 7.29838967e-01 5.81978142e-01 -1.70367622e+00 -2.46521920e-01 -4.66592282e-01 -1.91884100e-01 -4.90353554e-01 5.33789694e-01 -1.40928775e-01 -5.49534082e-01 1.06305194e+00 1.16534583e-01 4.73463863e-01 -2.25702256e-01 6.40357733e-01 1.07927823e+00 6.25008762e-01 5.66966832e-03 1.34739593e-01 1.50199521e+00 -9.15608823e-01 -6.94480002e-01 -2.62714595e-01 8.13589633e-01 -9.25539374e-01 6.87930942e-01 -2.05476820e-01 -4.90215242e-01 -1.09868921e-01 -8.24218214e-01 -7.33485669e-02 -1.28462970e+00 2.98370451e-01 2.79156059e-01 7.51760483e-01 -5.51380515e-01 7.08666265e-01 -1.65288463e-01 -4.32393849e-01 4.77183312e-01 7.55364448e-02 -2.14769736e-01 3.05375606e-01 -1.69760132e+00 1.17073548e+00 3.59519929e-01 -4.00191993e-01 -6.23039484e-01 -3.45806450e-01 -3.61452013e-01 -1.78562224e-01 6.80179447e-02 -1.01620868e-01 1.11418986e+00 -9.19626296e-01 -1.15718877e+00 1.27966964e+00 -6.09057136e-02 -7.20077753e-01 6.31769359e-01 -1.66880004e-02 -7.45552421e-01 6.17606416e-02 6.12206757e-01 3.56613360e-02 1.30615973e+00 -9.95137572e-01 -6.46623909e-01 -4.82980728e-01 -1.61510065e-01 -2.76186109e-01 -6.55684531e-01 3.86708289e-01 5.77973127e-02 -7.00435758e-01 -1.94518507e-01 -7.69385457e-01 3.20725888e-01 -6.73073947e-01 -7.56621718e-01 -4.70254362e-01 1.36886358e+00 -1.17334116e+00 1.46051300e+00 -2.25377989e+00 -2.68704265e-01 -1.08251174e-03 6.08477652e-01 5.81351876e-01 9.82508510e-02 8.49189520e-01 9.96725783e-02 3.88821661e-01 2.47002423e-01 -4.21280086e-01 2.10046902e-01 -4.62970108e-01 -5.67052066e-01 8.79075408e-01 1.88694164e-01 8.41961563e-01 -9.74799573e-01 -2.43201360e-01 -1.55213013e-01 2.01062679e-01 -2.37036645e-01 -1.79859892e-01 -3.52197178e-02 2.71186709e-01 -5.73203683e-01 5.27776062e-01 5.41246235e-01 -1.16814680e-01 -2.51562536e-01 -1.99712478e-02 -2.43787065e-01 9.63871241e-01 -7.94404745e-01 6.13201976e-01 -3.00847620e-01 1.01020658e+00 -9.76739451e-02 -8.07336271e-01 9.42547262e-01 3.12806666e-01 3.60787570e-01 -5.62567294e-01 3.99640828e-01 4.81026441e-01 -2.79465646e-01 -5.06217599e-01 8.94009650e-01 1.18999146e-01 -3.21155846e-01 8.29132855e-01 -4.20186259e-02 5.99185564e-03 1.82265565e-02 4.30562794e-01 1.08887482e+00 -4.99669671e-01 3.48520964e-01 -7.53809065e-02 5.64146817e-01 3.75830978e-01 7.26312920e-02 1.03275931e+00 -6.56526566e-01 1.26870245e-01 9.23427701e-01 -6.84464514e-01 -1.19206071e+00 -5.52364111e-01 1.18024111e-01 1.24247718e+00 3.19381386e-01 -3.86573464e-01 -6.67107224e-01 -1.12405133e+00 3.89916152e-01 6.74087346e-01 -8.24151337e-01 -2.46033356e-01 -7.53791451e-01 -7.43547142e-01 1.07567751e+00 -1.16268478e-01 5.66212714e-01 -9.93561804e-01 -2.16607317e-01 2.02139214e-01 -5.54510713e-01 -1.18518567e+00 -4.22460735e-01 -2.41418675e-01 -4.87798363e-01 -9.67760324e-01 -5.15823841e-01 -7.61515498e-01 1.62504166e-01 3.01908165e-01 8.16950560e-01 1.17089793e-01 -1.47529125e-01 -1.08670130e-01 -5.37530124e-01 -4.45538431e-01 -7.78370857e-01 2.70028114e-01 1.06786184e-01 1.58754602e-01 6.57373250e-01 -1.10272922e-01 -1.48836747e-01 2.52755344e-01 -8.23136926e-01 -4.88613248e-01 3.05211455e-01 8.46259415e-01 -2.40492582e-01 5.24028689e-02 7.06138849e-01 -1.07977605e+00 1.13803911e+00 -1.17892015e+00 -2.64695942e-01 -2.74730414e-01 -3.77925366e-01 -1.42171562e-01 6.18405402e-01 -5.98026812e-01 -4.51672137e-01 -3.14170301e-01 -2.10819803e-02 -5.71148582e-02 -8.50703567e-02 3.78118604e-01 5.26223719e-01 1.98535118e-02 9.85898256e-01 4.80560035e-01 -1.59709733e-02 -5.47975540e-01 3.26693416e-01 1.35956848e+00 1.50350511e-01 9.02365074e-02 9.62571442e-01 6.50470197e-01 -7.17633069e-01 -1.04539669e+00 -7.38361239e-01 -1.13260198e+00 -4.02663499e-01 7.64887733e-03 6.25946105e-01 -6.92961097e-01 -5.79439938e-01 9.27347064e-01 -1.60597682e+00 3.70082319e-01 1.08428448e-01 2.95301735e-01 -1.79234996e-01 5.07277131e-01 -8.74516666e-01 -7.47898757e-01 -1.60310328e-01 -1.04027152e+00 1.01076186e+00 -3.02805811e-01 -2.67220110e-01 -9.96926248e-01 1.63676217e-01 5.82346201e-01 7.02809274e-01 3.82058501e-01 8.50591660e-01 -1.44636476e+00 7.21251518e-02 -8.46473157e-01 -5.76546431e-01 2.63366371e-01 1.13384046e-01 -8.26602951e-02 -7.23364532e-01 -9.21341330e-02 9.12568346e-02 -1.12027727e-01 9.82604623e-01 3.29475924e-02 4.25415963e-01 -8.87174487e-01 -6.14940286e-01 5.99517412e-02 1.22539103e+00 -2.31955335e-01 4.72590446e-01 7.76835620e-01 7.23493278e-01 5.98821282e-01 2.76043206e-01 6.03749990e-01 2.41535693e-01 9.04111624e-01 6.07417822e-01 2.43973181e-01 1.21207699e-01 -4.11237836e-01 6.27933204e-01 5.35404444e-01 2.68250585e-01 -2.03946263e-01 -9.10413921e-01 7.28237152e-01 -1.49717081e+00 -1.42613804e+00 -5.43414295e-01 1.99182510e+00 4.26934808e-01 2.56766677e-01 7.86566317e-01 2.94660274e-02 1.07877648e+00 3.87925267e-01 -2.78912157e-01 -6.24908686e-01 -1.73272267e-01 -2.19423324e-01 1.09505510e+00 5.69786251e-01 -1.48653829e+00 1.36295319e+00 5.97366810e+00 1.04127288e+00 -1.26755202e+00 7.31158853e-01 2.10238263e-01 2.30328515e-01 6.28235117e-02 -2.06514031e-01 -1.01263499e+00 9.75287735e-01 8.96849930e-01 -8.56692716e-02 3.06527406e-01 9.74886179e-01 3.93055350e-01 4.96951267e-02 -5.90251029e-01 7.52451718e-01 3.71779919e-01 -1.79135156e+00 -3.90719548e-02 3.46000940e-01 6.54185355e-01 3.17345113e-01 3.31181198e-01 3.35657239e-01 1.80827335e-01 -8.20859134e-01 9.31768179e-01 1.57912955e-01 4.75348145e-01 -5.16030610e-01 1.03300011e+00 6.61926031e-01 -5.20648062e-01 -4.85737681e-01 -2.08597720e-01 -9.72898975e-02 -3.13124843e-02 6.30597532e-01 -1.35905457e+00 1.45551831e-01 3.53938997e-01 1.00978816e+00 -6.02676272e-01 8.18147004e-01 -3.89867164e-02 6.31263196e-01 -1.95543200e-01 -5.32557309e-01 6.07446313e-01 1.72973022e-01 1.08707988e+00 1.59216237e+00 5.14771529e-02 -5.89717627e-01 -1.11183800e-01 8.16352308e-01 -1.65789917e-01 1.46087497e-01 -8.92695546e-01 -6.75913632e-01 6.16519272e-01 9.00933564e-01 -5.68152726e-01 -5.51664889e-01 -9.85726044e-02 1.12512445e+00 2.35329822e-01 -1.36742815e-01 -1.00318944e+00 -4.31856155e-01 5.90142310e-01 6.46311045e-01 8.09400380e-02 -2.21908614e-02 -3.81890267e-01 -1.03541791e+00 -5.78265972e-02 -8.35319281e-01 1.67834759e-01 -1.76908791e-01 -1.57087493e+00 5.99283338e-01 -1.05037875e-01 -1.19778705e+00 -2.09202155e-01 -4.95325506e-01 -5.86746752e-01 7.55312860e-01 -1.24974799e+00 -1.10498381e+00 1.15484476e-01 4.15656060e-01 3.48207533e-01 -5.75622022e-01 5.88453650e-01 5.02180994e-01 -4.72907275e-01 5.21061659e-01 5.19945979e-01 6.69869125e-01 6.55819178e-01 -9.77946699e-01 6.28823459e-01 4.34008330e-01 1.77812446e-02 3.48074794e-01 8.98453891e-01 -9.25861955e-01 -8.81618857e-01 -9.95710671e-01 1.57733095e+00 -8.44173968e-01 1.19283712e+00 -3.45431387e-01 -4.38187659e-01 5.35974920e-01 -7.54097179e-02 -2.56361187e-01 4.13545877e-01 -2.21851408e-01 -7.26615667e-01 4.89375532e-01 -1.47225618e+00 1.76441461e-01 4.43733782e-01 -7.40104914e-01 -6.36269629e-01 1.01149809e+00 4.29079324e-01 -9.23668668e-02 -1.86809808e-01 -2.59509146e-01 5.95418870e-01 -9.46106255e-01 8.05551648e-01 -1.01377785e+00 6.73022926e-01 1.88260190e-02 1.07434414e-01 -1.12289441e+00 -2.76527226e-01 -4.25415397e-01 -1.79368734e-01 9.64142859e-01 4.01048958e-01 -9.24616635e-01 7.73679197e-01 -2.44328693e-01 2.21905574e-01 -4.66030300e-01 -1.00754654e+00 -8.55414808e-01 1.26837224e-01 -3.79061811e-02 3.05067390e-01 1.43743980e+00 2.28019431e-01 3.75693500e-01 -8.03445935e-01 7.29896724e-02 3.67462963e-01 -1.90432161e-01 8.14049482e-01 -1.45087969e+00 -9.24257860e-02 -6.74583435e-01 -6.88961029e-01 -8.24853718e-01 2.20526397e-01 -8.01801920e-01 -4.68554497e-01 -1.29165220e+00 1.17510073e-01 -2.21098736e-01 1.26957759e-01 3.48191187e-02 1.03885494e-01 4.32272971e-01 1.77933723e-01 7.24883258e-01 -2.34629944e-01 1.00331143e-01 7.96374142e-01 -3.28271300e-01 -2.00748935e-01 2.78390616e-01 -5.56887209e-01 9.12698150e-01 8.65722775e-01 -1.07376468e+00 3.12954843e-01 -2.54423738e-01 1.13734566e-01 -1.87958658e-01 5.71785510e-01 -6.75450504e-01 5.50927781e-02 -1.77563895e-02 -1.15223691e-01 -4.85533506e-01 3.85434002e-01 -4.75846708e-01 -4.80158925e-01 7.38495767e-01 -3.06365192e-01 8.01872090e-02 -1.30017921e-01 8.64670455e-01 -2.36554071e-01 -4.22772676e-01 9.86767352e-01 -2.26998329e-01 -5.06091297e-01 3.90284471e-02 -7.97538340e-01 1.25406757e-01 9.27626193e-01 -3.25288892e-01 -6.49069846e-01 -5.59292614e-01 -6.64703965e-01 -4.57619548e-01 1.48366272e-01 5.96677244e-01 3.02512854e-01 -1.09249961e+00 -1.04179180e+00 -1.93826243e-01 2.14489430e-01 -1.18236351e+00 -2.99764514e-01 1.02294314e+00 -5.78488171e-01 5.85202575e-01 -1.29130304e-01 -7.46758506e-02 -1.14676964e+00 4.51919258e-01 7.07998723e-02 -7.44474530e-01 -1.28725290e-01 7.65320003e-01 -5.16508460e-01 -5.25354505e-01 -4.94400552e-03 1.57169864e-01 -4.69450444e-01 4.03033823e-01 5.98370016e-01 8.40206802e-01 8.26874971e-02 -1.45844245e+00 -4.23721164e-01 -3.15836892e-02 -3.23420972e-01 -9.61707607e-02 1.21368825e+00 8.25794935e-02 -3.30379099e-01 3.41388673e-01 1.54221523e+00 3.79631996e-01 -2.80029982e-01 -4.87063348e-01 4.48231667e-01 -6.94774985e-01 1.33328125e-01 -8.53337586e-01 -7.74650574e-01 6.31740153e-01 3.71791661e-01 8.27977359e-01 2.65272945e-01 8.79480019e-02 1.24075818e+00 1.25206023e-01 9.33986381e-02 -1.19074941e+00 3.36131960e-01 8.49548876e-01 5.47334790e-01 -1.37414241e+00 -2.18791291e-01 -3.57245266e-01 -5.55746257e-01 1.06691432e+00 3.05673778e-01 -3.40838552e-01 6.82577968e-01 -3.05972546e-01 -2.87969075e-02 -5.82464218e-01 -9.98187345e-03 -6.51524588e-02 1.63989678e-01 5.00381708e-01 1.78657368e-01 9.50752795e-02 -7.33674645e-01 4.89786118e-01 -2.18392923e-01 -3.70963693e-01 7.54045486e-01 7.59824991e-01 -6.21786356e-01 -8.13220739e-01 -3.33752692e-01 6.68183744e-01 -9.31853235e-01 -2.66064972e-01 -1.07019579e+00 7.79580414e-01 2.92149276e-01 1.06523442e+00 -1.96143329e-01 -6.76072538e-01 1.58733875e-01 1.12760663e-01 -1.44652292e-01 -7.19666421e-01 -9.41894531e-01 -5.57144046e-01 2.47826770e-01 -2.50840753e-01 -1.31079406e-01 -5.40131152e-01 -6.47912621e-01 -6.34805441e-01 -7.42122114e-01 3.86263460e-01 1.05700850e+00 1.14511359e+00 5.32241702e-01 -1.79042384e-01 8.20192873e-01 -6.18446708e-01 -9.05308723e-01 -1.30887437e+00 -6.25964105e-01 6.61215901e-01 4.94398296e-01 -7.84446180e-01 -8.72399032e-01 -1.84622064e-01]
[8.202781677246094, 10.235111236572266]
d8dbaf02-e664-4aba-99e7-89614f481339
reliable-and-efficient-image-cropping-a-grid
1904.04441
null
http://arxiv.org/abs/1904.04441v1
http://arxiv.org/pdf/1904.04441v1.pdf
Reliable and Efficient Image Cropping: A Grid Anchor based Approach
Image cropping aims to improve the composition as well as aesthetic quality of an image by removing extraneous content from it. Existing image cropping databases provide only one or several human-annotated bounding boxes as the groundtruth, which cannot reflect the non-uniqueness and flexibility of image cropping in practice. The employed evaluation metrics such as intersection-over-union cannot reliably reflect the real performance of cropping models, either. This work revisits the problem of image cropping, and presents a grid anchor based formulation by considering the special properties and requirements (e.g., local redundancy, content preservation, aspect ratio) of image cropping. Our formulation reduces the searching space of candidate crops from millions to less than one hundred. Consequently, a grid anchor based cropping benchmark is constructed, where all crops of each image are annotated and more reliable evaluation metrics are defined. We also design an effective and lightweight network module, which simultaneously considers the region of interest and region of discard for more accurate image cropping. Our model can stably output visually pleasing crops for images of different scenes and run at a speed of 125 FPS. Code and dataset are available at: https://github.com/HuiZeng/Grid-Anchor-based-Image-Cropping.
['Hui Zeng', 'Lida Li', 'Lei Zhang', 'Zisheng Cao']
2019-04-09
reliable-and-efficient-image-cropping-a-grid-1
http://openaccess.thecvf.com/content_CVPR_2019/html/Zeng_Reliable_and_Efficient_Image_Cropping_A_Grid_Anchor_Based_Approach_CVPR_2019_paper.html
http://openaccess.thecvf.com/content_CVPR_2019/papers/Zeng_Reliable_and_Efficient_Image_Cropping_A_Grid_Anchor_Based_Approach_CVPR_2019_paper.pdf
cvpr-2019-6
['image-cropping']
['computer-vision']
[ 2.17735901e-01 -9.54109579e-02 -1.54596820e-01 -1.17571220e-01 -5.02278566e-01 -7.80246198e-01 1.15211971e-01 3.98947477e-01 -1.13512628e-01 4.51002240e-01 -3.72779638e-01 -2.02987164e-01 1.98890746e-01 -1.14436662e+00 -9.23377752e-01 -6.54751480e-01 1.20766750e-02 -3.38171303e-01 3.73332053e-01 -2.70897299e-01 1.57845274e-01 4.29677159e-01 -1.91705883e+00 3.70225050e-02 1.15727627e+00 1.17552686e+00 7.67779648e-01 5.87794602e-01 -6.24317452e-02 1.06445208e-01 -5.50614178e-01 -2.01582298e-01 4.92391586e-01 -2.22218215e-01 -3.27647477e-01 4.17306870e-01 5.15815377e-01 -3.76805961e-01 1.97488189e-01 1.33670831e+00 3.55684221e-01 -3.41119289e-01 3.79620910e-01 -1.57292354e+00 -8.71484518e-01 4.45073217e-01 -9.55616772e-01 -2.99115300e-01 1.27653271e-01 2.12620422e-01 8.37050796e-01 -1.02952743e+00 5.90086937e-01 8.59040260e-01 6.87088847e-01 1.06750228e-01 -9.51543927e-01 -7.99179018e-01 2.30897516e-01 -1.64054483e-01 -1.55429816e+00 -1.04686007e-01 5.77641070e-01 -1.72391638e-01 2.73584008e-01 4.81244564e-01 7.39671171e-01 5.02133429e-01 1.96130216e-01 5.35071671e-01 9.67253327e-01 -5.06342649e-01 9.44432393e-02 2.21526846e-01 4.48378734e-03 6.54523730e-01 7.06669629e-01 -3.48895639e-02 -3.98013592e-01 1.13788724e-01 1.15109062e+00 1.06266057e-02 -4.44649518e-01 -7.55057752e-01 -1.57889760e+00 4.15326953e-01 8.07691574e-01 2.21248850e-01 -3.96562368e-01 1.88580424e-01 2.48592541e-01 1.43772904e-02 2.79751003e-01 3.39083135e-01 -3.84671271e-01 3.88651192e-01 -1.05200779e+00 2.90320188e-01 3.30069363e-01 1.54777861e+00 1.03217304e+00 -6.55307993e-02 2.71577835e-02 4.70963866e-01 -2.54437211e-03 8.37621868e-01 4.89010066e-02 -8.38553309e-01 3.73122960e-01 7.37953603e-01 3.95877987e-01 -1.45327699e+00 -3.47961605e-01 -2.33319908e-01 -1.13759446e+00 4.39229757e-01 5.27162790e-01 -1.53613910e-01 -9.08349097e-01 1.39836502e+00 4.62724626e-01 -6.04639314e-02 -9.06694382e-02 1.07976389e+00 9.26849425e-01 5.97678125e-01 1.39720753e-01 -1.91716626e-01 1.58062506e+00 -9.33888495e-01 -7.41866171e-01 -5.75353131e-02 4.58256125e-01 -1.07682180e+00 1.51178372e+00 5.14473140e-01 -1.08436227e+00 -5.92109859e-01 -1.28089237e+00 3.99262905e-02 -6.86287820e-01 3.83532524e-01 6.25716269e-01 6.37765586e-01 -1.09708965e+00 3.62921715e-01 -3.11218917e-01 -3.29215139e-01 2.14158490e-01 4.17975187e-02 -4.92564112e-01 -1.93715885e-01 -9.69997704e-01 5.52962005e-01 7.10601270e-01 2.21347272e-01 -5.39012432e-01 -6.80607498e-01 -8.77783418e-01 -5.48852375e-03 4.26666379e-01 -3.29093844e-01 8.62761259e-01 -1.15743899e+00 -8.86251688e-01 8.05415332e-01 1.50255188e-01 -2.70712584e-01 4.13421124e-01 -4.67937104e-02 -2.32160777e-01 1.89734861e-01 1.19878642e-01 9.82613862e-01 8.16460133e-01 -1.62927735e+00 -7.36408651e-01 -9.20227692e-02 1.76119894e-01 3.99349540e-01 -4.04513896e-01 -1.34682894e-01 -6.28967285e-01 -7.98456728e-01 1.56685278e-01 -6.89722598e-01 -2.76651204e-01 3.71183723e-01 -4.06235576e-01 2.83922911e-01 8.00314546e-01 -7.33686745e-01 1.39039981e+00 -2.13698721e+00 -4.12439585e-01 2.56427228e-01 1.62566826e-01 1.85676694e-01 -2.89310157e-01 3.65106136e-01 -6.45296574e-02 3.10054243e-01 -3.98477346e-01 2.97644865e-02 -2.22628072e-01 -2.21341439e-02 -9.22843590e-02 3.95123899e-01 4.78511542e-01 8.77773702e-01 -9.72975791e-01 -8.54632139e-01 4.94023055e-01 4.68073428e-01 -2.46809632e-01 -4.13182303e-02 -2.37873688e-01 1.34901032e-01 -2.90739030e-01 1.24834001e+00 1.33996868e+00 -3.01374882e-01 -1.63844263e-03 -5.44503033e-01 -3.54704648e-01 -5.86880028e-01 -1.31626749e+00 1.63734341e+00 -4.31443781e-01 3.82475048e-01 5.32214716e-02 -4.97436523e-01 1.21667266e+00 6.59152120e-02 4.89031255e-01 -7.10011423e-01 4.65544760e-02 2.75695026e-01 -4.95180666e-01 -2.38859996e-01 9.22120750e-01 5.67092121e-01 -1.55752808e-01 2.04760641e-01 -3.04613680e-01 -2.59561896e-01 4.65917349e-01 -3.53222160e-04 5.83807468e-01 3.87396961e-01 5.81194758e-01 -5.48960090e-01 2.62622297e-01 3.85894716e-01 4.41529542e-01 4.60110098e-01 -2.20682949e-01 1.05446482e+00 3.57765108e-01 -4.12864983e-01 -1.38100481e+00 -9.60187197e-01 -2.69959271e-01 9.61555779e-01 8.24142694e-01 -2.34473884e-01 -1.11171949e+00 -4.25020069e-01 -9.69109684e-02 4.35384065e-01 -5.53313017e-01 2.41681248e-01 -4.55411196e-01 -7.64547527e-01 2.77981818e-01 2.70445764e-01 7.79681027e-01 -1.17629457e+00 -9.77442861e-01 -4.32240479e-02 -5.01974225e-01 -1.08323789e+00 -6.06079161e-01 1.95034256e-03 -5.60900152e-01 -1.31464517e+00 -1.00296474e+00 -9.51998055e-01 9.39080954e-01 8.06018233e-01 1.36868298e+00 5.24755001e-01 -3.24199736e-01 3.34494421e-03 -7.79202878e-01 -5.60335040e-01 -2.68967390e-01 1.01520360e-01 -2.98484594e-01 -2.64527977e-01 1.50528058e-01 -3.02842170e-01 -8.97325695e-01 4.95077610e-01 -1.22874928e+00 3.65502983e-01 7.26351142e-01 7.30587006e-01 8.73077571e-01 2.62706369e-01 4.33743536e-01 -4.43580419e-01 4.66202497e-01 -3.18860978e-01 -8.47657681e-01 6.12614751e-01 -4.88604158e-01 -3.50420475e-01 4.84105080e-01 -5.23812354e-01 -7.51108587e-01 4.10866648e-01 4.14892435e-01 -3.35438818e-01 -9.03364792e-02 3.22272271e-01 -2.82039702e-01 -2.02381805e-01 5.97781837e-01 2.32359082e-01 -2.46034414e-02 -7.54394829e-02 5.27258337e-01 4.08382028e-01 6.64939761e-01 -3.46325994e-01 9.79769170e-01 2.70921975e-01 -4.78381142e-02 -6.95946693e-01 -3.43698710e-01 -3.72648448e-01 -6.81742251e-01 -5.22461891e-01 5.88818729e-01 -1.02106249e+00 -6.03366256e-01 6.17034912e-01 -1.12671542e+00 -2.44118884e-01 -2.24876791e-01 -8.92375188e-04 -4.62971479e-01 4.34005827e-01 -1.83450863e-01 -7.48493314e-01 -6.32985413e-01 -9.67931867e-01 1.28034890e+00 3.07088256e-01 2.88303077e-01 -3.92226309e-01 -5.37562013e-01 -1.31745577e-01 3.53169203e-01 8.69682550e-01 6.19302392e-01 2.86318183e-01 -8.04721534e-01 -1.97846040e-01 -7.69871116e-01 1.90215945e-01 1.30716398e-01 3.59800845e-01 -8.04915547e-01 -2.17880204e-01 -6.18061841e-01 1.64537765e-02 6.76770210e-01 4.97264862e-01 1.09705782e+00 -1.82235196e-01 -3.17023009e-01 4.70294178e-01 1.84553385e+00 4.88098478e-03 9.08328235e-01 3.89250934e-01 5.16119540e-01 6.98885262e-01 1.37726629e+00 5.22474587e-01 2.10854486e-01 4.42721725e-01 1.03213656e+00 -5.73376834e-01 -1.35479301e-01 -2.99959451e-01 2.43544698e-01 4.39559370e-01 8.80186856e-02 -5.25278091e-01 -8.47104013e-01 8.64187539e-01 -1.74323285e+00 -4.28162664e-01 -1.86581492e-01 2.45097351e+00 6.24451637e-01 -1.79843809e-02 7.11028501e-02 2.53413945e-01 1.10765147e+00 -6.19749688e-02 -5.53868294e-01 -9.18829963e-02 -3.90852600e-01 -1.41701594e-01 1.06017900e+00 1.20545119e-01 -1.19256473e+00 9.79263544e-01 5.82940435e+00 9.59227264e-01 -1.00805664e+00 -8.64032581e-02 8.82788420e-01 2.98803568e-01 -1.50632858e-01 -7.84329176e-02 -7.05519736e-01 4.39922154e-01 3.14832091e-01 -1.79849118e-01 2.56602913e-01 7.93671429e-01 2.26343617e-01 -3.39342803e-01 -5.20226598e-01 9.19208050e-01 -5.45620024e-02 -1.20987439e+00 8.73619467e-02 -6.01185523e-02 6.50560558e-01 -3.06515396e-01 1.00327022e-01 -3.23814481e-01 1.23212941e-01 -9.31439757e-01 1.01930654e+00 3.82618845e-01 1.20848858e+00 -7.52295494e-01 6.73902035e-01 1.42851204e-01 -1.67969394e+00 -8.68758839e-03 -4.36254144e-01 3.04765135e-01 2.16478184e-02 7.65693128e-01 -5.73464811e-01 7.71435976e-01 1.07350135e+00 4.28992301e-01 -7.09941506e-01 1.13533580e+00 1.35465786e-01 9.08621624e-02 -4.67077047e-01 -4.30365950e-02 2.33269528e-01 -2.68801153e-01 2.27781862e-01 1.16249108e+00 7.83822834e-01 -7.12093115e-02 1.69966564e-01 8.13622952e-01 7.66253695e-02 5.93037009e-01 -6.49489939e-01 3.48161347e-02 7.30743289e-01 1.51933014e+00 -1.19490016e+00 -1.53480887e-01 -2.42694288e-01 7.57557631e-01 -1.27430856e-01 1.32834449e-01 -9.28402543e-01 -4.91336852e-01 4.04837430e-01 3.24093491e-01 3.49906147e-01 -2.11538896e-01 -3.46476227e-01 -9.58744705e-01 3.24576259e-01 -8.82473230e-01 -2.96616387e-02 -1.09401655e+00 -9.26005781e-01 6.79008663e-01 5.24975592e-03 -1.75636590e+00 2.45638028e-01 -5.66897631e-01 -3.94331664e-01 7.93063283e-01 -1.55956244e+00 -1.54900968e+00 -1.11718285e+00 3.99077892e-01 4.41414952e-01 2.40075082e-01 7.47141063e-01 2.22768977e-01 -4.06404108e-01 5.41768014e-01 -8.35720226e-02 -3.18925492e-02 7.21612334e-01 -1.09644520e+00 5.72858572e-01 1.13265049e+00 -1.29518434e-01 1.74090266e-01 8.36515367e-01 -5.31466365e-01 -1.15009511e+00 -1.32280815e+00 5.99705696e-01 -1.29582405e-01 3.32153648e-01 -3.65728885e-01 -7.35178411e-01 1.31606936e-01 2.78346330e-01 1.31918833e-01 -7.56669343e-02 -5.93556583e-01 -1.99410453e-01 -2.07149163e-01 -1.19850242e+00 7.10329890e-01 1.00467825e+00 1.68824062e-01 2.71515578e-01 3.96397859e-01 1.19963527e+00 -4.95603770e-01 -9.09241319e-01 4.54140782e-01 5.42123318e-01 -1.05149591e+00 1.08704126e+00 1.64817750e-01 4.56791997e-01 -6.05130494e-01 -1.17886208e-01 -9.86867845e-01 -1.27452165e-01 -4.28700209e-01 2.68742412e-01 1.30704045e+00 2.93042779e-01 -4.27966416e-01 7.04585910e-01 1.96499527e-01 1.18413918e-01 -7.93826640e-01 -3.98640186e-01 -6.36028171e-01 -5.45002371e-02 -1.48822248e-01 1.18099594e+00 8.60489368e-01 -1.51539385e-01 -4.44201112e-01 -4.25363272e-01 4.41775680e-01 6.87936366e-01 3.03521991e-01 8.91597331e-01 -9.92780447e-01 3.07633221e-01 -3.07441533e-01 -3.18621159e-01 -7.11806715e-01 -4.33701217e-01 -3.85405064e-01 9.64822322e-02 -1.45068157e+00 -3.47209051e-02 -7.40738571e-01 -1.72704652e-01 5.98573744e-01 -1.71659186e-01 6.15900159e-01 3.56672972e-01 1.13544658e-01 -4.12758648e-01 1.66729569e-01 1.53455746e+00 5.42177446e-02 -1.18270338e-01 -2.27016151e-01 -7.52183855e-01 5.34368575e-01 1.17257917e+00 -1.63738847e-01 -2.93527991e-01 -3.07139844e-01 2.11431891e-01 -1.28056735e-01 4.35042143e-01 -1.08573306e+00 2.80322582e-02 -3.48162234e-01 4.17203665e-01 -8.10350120e-01 1.90676987e-01 -1.09558201e+00 2.55166650e-01 3.69431674e-01 -1.96635630e-02 3.06931704e-01 2.79523224e-01 3.73473614e-01 -2.08119541e-01 -2.57820100e-01 6.43849790e-01 -2.57963330e-01 -1.12264717e+00 3.28663170e-01 2.20108330e-01 -3.49088341e-01 1.36672091e+00 -4.52113181e-01 -3.93095344e-01 -1.94126084e-01 -2.86181629e-01 2.28131205e-01 8.67689610e-01 4.97376204e-01 5.31037986e-01 -1.25989556e+00 -7.76190996e-01 2.07726464e-01 5.48549950e-01 2.50539064e-01 3.54658872e-01 5.32521784e-01 -1.16170657e+00 1.95148408e-01 -4.94527429e-01 -5.95934689e-01 -1.45850646e+00 7.03656554e-01 1.77113846e-01 -1.86821893e-01 -5.98847508e-01 3.44809145e-01 4.16571498e-01 -3.09721112e-01 1.46333247e-01 -6.26654387e-01 -2.25631475e-01 -2.28379741e-02 5.22843957e-01 1.07982107e-01 1.36977825e-02 -6.50840878e-01 -1.12016268e-01 7.07386613e-01 2.46621847e-01 3.19810927e-01 9.07746792e-01 -4.19348508e-01 -7.34757707e-02 8.92316326e-02 6.93408430e-01 -2.64156222e-01 -1.30313611e+00 -1.23345397e-01 -3.25179249e-01 -7.17041790e-01 -6.19004341e-03 -7.03506529e-01 -1.20218050e+00 6.27760351e-01 7.79830039e-01 4.28367674e-01 1.63174033e+00 -4.82258230e-01 6.69266403e-01 1.23104192e-02 7.33329356e-01 -1.01129949e+00 -1.45971403e-01 -7.77587593e-02 1.14360297e+00 -1.36154771e+00 8.60831216e-02 -8.89630616e-01 -6.26233697e-01 1.05375051e+00 7.08160460e-01 -2.34297305e-01 5.10373175e-01 6.04898870e-01 2.07917541e-01 4.63296846e-02 -3.02483469e-01 -3.22506249e-01 8.73204172e-02 8.90364945e-01 2.30472431e-01 2.92086452e-01 -3.59216452e-01 3.02870780e-01 -1.94830179e-01 -2.20735166e-02 5.19747138e-01 7.63285339e-01 -5.87068617e-01 -7.77343094e-01 -7.80213475e-01 3.19162309e-01 -6.88147098e-02 -2.84474462e-01 -4.61161099e-02 1.08677816e+00 4.35795039e-01 8.62692714e-01 4.27999608e-02 -2.51002520e-01 3.33352745e-01 -5.43912768e-01 2.10882723e-01 -1.87196046e-01 -4.95833308e-01 1.31496936e-01 -1.82100683e-01 -5.70771217e-01 -5.61535716e-01 -3.59105319e-01 -9.40108240e-01 -5.48123658e-01 -7.37307012e-01 -2.40064889e-01 8.08142602e-01 1.47993207e-01 4.19541329e-01 3.71023804e-01 5.74113429e-01 -1.03207457e+00 -1.42169133e-01 -7.57628024e-01 -7.76853740e-01 1.99396834e-01 1.19600490e-01 -5.11760473e-01 -1.63875014e-01 4.38615143e-01]
[11.22376823425293, -1.1091156005859375]
b12f3386-581a-4b9a-a931-acda15dd53ae
parameters-or-privacy-a-provable-tradeoff
2202.01243
null
https://arxiv.org/abs/2202.01243v2
https://arxiv.org/pdf/2202.01243v2.pdf
Parameters or Privacy: A Provable Tradeoff Between Overparameterization and Membership Inference
A surprising phenomenon in modern machine learning is the ability of a highly overparameterized model to generalize well (small error on the test data) even when it is trained to memorize the training data (zero error on the training data). This has led to an arms race towards increasingly overparameterized models (c.f., deep learning). In this paper, we study an underexplored hidden cost of overparameterization: the fact that overparameterized models may be more vulnerable to privacy attacks, in particular the membership inference attack that predicts the (potentially sensitive) examples used to train a model. We significantly extend the relatively few empirical results on this problem by theoretically proving for an overparameterized linear regression model in the Gaussian data setting that membership inference vulnerability increases with the number of parameters. Moreover, a range of empirical studies indicates that more complex, nonlinear models exhibit the same behavior. Finally, we extend our analysis towards ridge-regularized linear regression and show in the Gaussian data setting that increased regularization also increases membership inference vulnerability in the overparameterized regime.
['Richard G. Baraniuk', 'Hamid Javadi', 'Blake Mason', 'Jasper Tan']
2022-02-02
null
null
null
null
['membership-inference-attack']
['computer-vision']
[ 1.25517100e-01 4.20598030e-01 -1.84034705e-01 -4.30621028e-01 -7.68726885e-01 -8.95275772e-01 3.80400360e-01 8.84264484e-02 -5.72507679e-01 9.60053921e-01 -2.92315781e-01 -6.58659637e-01 -1.25022069e-01 -7.09114969e-01 -9.57927227e-01 -9.46732461e-01 -2.48105273e-01 3.58738005e-01 -1.80272967e-01 4.14514206e-02 2.24130511e-01 6.51989698e-01 -9.79904830e-01 -1.51132315e-01 6.31191134e-01 7.16830552e-01 -8.61146212e-01 5.70091128e-01 6.82424366e-01 2.41910726e-01 -7.07136154e-01 -6.78970218e-01 5.01794159e-01 -5.51341176e-02 -7.41965413e-01 -1.55966118e-01 7.36081898e-01 -1.83649927e-01 -1.03721969e-01 1.26940119e+00 2.17770889e-01 1.40588224e-01 5.93188703e-01 -1.47805536e+00 -6.37974739e-01 6.76460385e-01 -3.33321005e-01 7.73064941e-02 -1.37713134e-01 1.02980681e-01 9.29211497e-01 -4.69871014e-01 3.18958700e-01 1.09720063e+00 9.09316242e-01 4.76808757e-01 -1.68032944e+00 -8.85057747e-01 8.36912356e-03 -3.92119378e-01 -1.56594360e+00 -3.72910768e-01 2.20569879e-01 -4.18832242e-01 4.95781600e-01 4.03207690e-01 1.92929581e-02 1.24363768e+00 1.57963023e-01 2.63718277e-01 1.13540900e+00 -1.80248916e-01 4.29539800e-01 6.98434412e-01 3.78364354e-01 5.12237430e-01 8.58921885e-01 2.50983864e-01 4.43871766e-02 -7.51702368e-01 5.47083497e-01 -2.14477614e-01 -3.32319826e-01 -4.41076458e-01 -5.32771468e-01 1.12762141e+00 5.31817563e-02 1.60336912e-01 -3.26339081e-02 1.43667877e-01 4.42124456e-01 6.93379343e-01 3.91758621e-01 9.53573465e-01 -6.38310015e-01 1.70306027e-01 -7.78807878e-01 3.11315626e-01 1.14815402e+00 7.58583367e-01 7.89226949e-01 1.06859483e-01 1.74019486e-01 4.60671633e-01 -9.67392176e-02 3.33221942e-01 4.18817341e-01 -7.93211758e-01 5.96032500e-01 1.84350640e-01 1.17779821e-01 -9.52256382e-01 -4.86886501e-01 -8.12166393e-01 -8.80133867e-01 2.93217957e-01 9.34031129e-01 -5.68998933e-01 -3.27996135e-01 2.13815594e+00 1.15610771e-01 -1.25252277e-01 2.20951989e-01 5.68765223e-01 -4.35258709e-02 5.15895247e-01 1.29549652e-01 -1.34522378e-01 1.01248729e+00 -4.14669394e-01 -2.87651390e-01 -1.24352887e-01 1.01318336e+00 -3.28281283e-01 1.24722421e+00 6.15552604e-01 -1.03024125e+00 -6.60635680e-02 -1.24239671e+00 1.03467070e-02 -5.47931373e-01 -3.87852192e-01 6.56571150e-01 1.12546587e+00 -8.65267932e-01 7.27591038e-01 -6.70041263e-01 -2.81712711e-01 6.76179528e-01 7.04155445e-01 -6.24370098e-01 3.12253069e-02 -1.40154183e+00 9.38462555e-01 3.94538760e-01 -3.67738269e-02 -5.00755489e-01 -8.72507691e-01 -6.58822596e-01 1.87049150e-01 4.14830446e-01 -5.02453089e-01 8.70891750e-01 -9.61607635e-01 -1.05792534e+00 8.44424605e-01 2.15335190e-01 -6.98001146e-01 9.09404337e-01 -1.74779192e-01 -1.54050365e-01 -2.34917298e-01 -4.27901596e-01 1.22532435e-01 9.22921896e-01 -1.09860909e+00 -1.62317157e-01 -7.15829790e-01 2.36726385e-02 -1.18709125e-01 -5.10509729e-01 -2.61204809e-01 1.09206274e-01 -5.56951761e-01 -1.68376178e-01 -1.29278255e+00 -3.56969595e-01 -2.00629476e-02 -9.24019694e-01 4.91319485e-02 6.05736494e-01 -3.95419270e-01 1.22860014e+00 -2.32667708e+00 -6.68370724e-02 7.59145856e-01 2.19089508e-01 2.93200612e-01 1.73084170e-01 2.88732827e-01 -3.89029533e-01 6.46538913e-01 -3.20133746e-01 -3.96665186e-01 2.09272370e-01 2.41656780e-01 -5.95996499e-01 9.78118539e-01 3.96558829e-02 7.29330838e-01 -2.43410274e-01 -1.70832872e-01 -3.98424834e-01 4.24917102e-01 -8.28500569e-01 -7.44350478e-02 3.28882411e-02 3.59487444e-01 -5.65900326e-01 2.73853362e-01 8.66507173e-01 -4.88104671e-01 1.26715586e-01 1.47548661e-01 1.49034753e-01 -1.49360374e-01 -9.07335997e-01 8.23898971e-01 -3.27074051e-01 7.24442303e-01 -1.07108010e-03 -1.13260663e+00 7.02240527e-01 2.39040077e-01 -8.96790326e-02 -1.11257844e-01 2.02480182e-01 1.28642723e-01 5.58872744e-02 -5.11802793e-01 4.12645131e-01 -2.86934406e-01 -3.25716048e-01 6.01742029e-01 -2.73478538e-01 2.31669784e-01 -4.46880311e-01 -1.11923451e-02 9.28373754e-01 -4.92186725e-01 2.13399082e-01 -4.04022455e-01 3.69294465e-01 -1.67591885e-01 4.23148155e-01 1.19538915e+00 -2.71539409e-02 4.56003368e-01 9.49347913e-01 -2.77018875e-01 -1.20620513e+00 -8.41253161e-01 -7.76977837e-01 1.02241313e+00 -3.16026330e-01 -1.36822239e-01 -9.17040944e-01 -7.88480759e-01 5.39377153e-01 8.79942417e-01 -9.54542398e-01 -5.28424382e-01 -4.99989569e-01 -1.12247872e+00 1.04865420e+00 3.30580711e-01 3.51191461e-01 -2.77765781e-01 -2.66388685e-01 -2.72559851e-01 3.19494843e-01 -8.35550070e-01 -3.11196089e-01 2.76426762e-01 -8.59073281e-01 -8.41649473e-01 -3.72127891e-01 -1.89074576e-01 9.13086653e-01 -5.20921290e-01 8.23361039e-01 2.24542752e-01 -1.29181817e-01 2.72303402e-01 6.23609200e-02 -4.11091775e-01 -6.60002410e-01 3.05424035e-01 1.01616837e-01 -2.07819487e-03 4.87559319e-01 -5.06129742e-01 -4.20351118e-01 4.11292225e-01 -1.06885946e+00 -6.54381275e-01 3.67219061e-01 7.86459446e-01 2.26479247e-01 -1.20741697e-02 6.87128663e-01 -1.53873479e+00 7.40513265e-01 -8.39965940e-01 -8.06403100e-01 3.57965857e-01 -9.41090643e-01 3.98327619e-01 9.31209862e-01 -9.17040229e-01 -6.21646285e-01 -1.33132994e-01 9.63849723e-02 -3.09657723e-01 3.42185646e-02 2.77270287e-01 -1.70199528e-01 -3.76308680e-01 8.10766101e-01 -2.17411364e-03 2.22212359e-01 -4.53271508e-01 1.29259646e-01 5.43673992e-01 3.03450108e-01 -6.87259257e-01 1.03388011e+00 4.50279772e-01 4.54679489e-01 -8.95362794e-01 -6.04063332e-01 3.52656782e-01 -2.60568827e-01 3.15538466e-01 4.38781619e-01 -6.14674985e-01 -9.44774270e-01 1.51515260e-01 -7.15864480e-01 -2.59675533e-01 -1.96152255e-01 5.23638010e-01 -5.26188731e-01 4.52313125e-01 -3.04701775e-01 -8.73877048e-01 8.62859841e-03 -1.01493156e+00 5.12528360e-01 -4.19389531e-02 -5.09479940e-01 -1.37745595e+00 -2.59485692e-02 2.22772285e-01 5.30724108e-01 4.55567062e-01 1.33547175e+00 -1.50779402e+00 -3.11899006e-01 -7.08693802e-01 7.04405829e-02 5.18240094e-01 -2.66287178e-01 1.34358583e-02 -1.14852333e+00 -5.39967775e-01 3.43474418e-01 -4.83736396e-01 8.13107133e-01 2.59083301e-01 1.41975200e+00 -7.71097720e-01 -2.08411187e-01 9.27775860e-01 1.33662975e+00 -1.43701956e-01 6.32209063e-01 3.18256199e-01 2.76981443e-01 6.54341877e-01 2.21189082e-01 3.21528286e-01 -9.20573547e-02 5.33876657e-01 3.16251516e-01 -3.86374705e-02 8.53154600e-01 -3.47799629e-01 2.03495741e-01 -2.40450367e-01 2.33161926e-01 -2.85366714e-01 -6.10193610e-01 2.82754935e-02 -1.54301631e+00 -7.64896393e-01 2.56695822e-02 2.72177315e+00 1.10826766e+00 1.70859620e-01 2.83936292e-01 -1.63203776e-02 8.01580191e-01 -1.44211918e-01 -8.45190823e-01 -7.14523613e-01 -2.18048900e-01 1.97785273e-01 1.00421381e+00 6.04276896e-01 -1.03555691e+00 7.85091817e-01 6.90214682e+00 7.09978640e-01 -1.19415522e+00 -7.69955441e-02 1.14022017e+00 -2.10304156e-01 -2.45539635e-01 -3.98512818e-02 -7.69469976e-01 3.74306500e-01 1.02591503e+00 -4.38327163e-01 5.13797522e-01 1.03188062e+00 -1.30637348e-01 -8.26056954e-03 -1.50977874e+00 5.35077989e-01 -1.16136029e-01 -1.10915720e+00 -2.59702235e-01 5.57698846e-01 5.11813462e-01 -3.02937388e-01 8.30144286e-01 3.79343539e-01 4.22076821e-01 -1.35369027e+00 1.54318631e-01 2.98529655e-01 7.35098183e-01 -8.55056703e-01 5.01785636e-01 4.10233378e-01 -1.35974303e-01 -2.69153237e-01 -5.56379080e-01 2.04788879e-01 -3.61649483e-01 3.91527683e-01 -8.48834038e-01 -6.16518557e-02 3.63630861e-01 1.89690124e-02 -8.27231765e-01 8.47815633e-01 2.27804124e-01 7.65940607e-01 -5.54919720e-01 1.62865385e-01 1.96194172e-01 -1.60268083e-01 4.60451126e-01 9.89966691e-01 1.16588868e-01 -1.57524720e-01 -3.20481211e-01 9.80502248e-01 -1.48613825e-01 5.58264144e-02 -8.49078000e-01 -1.09792121e-01 4.52702045e-01 8.07592213e-01 -1.31024539e-01 3.03361341e-02 -1.93870828e-01 5.56359708e-01 4.74497736e-01 6.46232545e-01 -6.10203505e-01 -1.76693618e-01 6.03416562e-01 1.23170182e-01 1.56371519e-01 -8.19535702e-02 -4.42284793e-01 -1.02721882e+00 1.00303236e-02 -8.25487673e-01 6.29484534e-01 -2.39057094e-01 -1.59962451e+00 4.54585344e-01 8.51002559e-02 -8.17776859e-01 -3.95522803e-01 -5.37765741e-01 -5.61029971e-01 1.01249099e+00 -1.13024521e+00 -6.54219091e-01 4.35016960e-01 6.15160704e-01 -2.46452227e-01 -1.63060054e-01 9.55141544e-01 3.64821292e-02 -7.97765791e-01 1.46707499e+00 4.62403268e-01 1.75528526e-01 7.27692544e-01 -1.09902000e+00 3.04749042e-01 6.86130345e-01 -2.43772522e-01 1.13264203e+00 8.76940548e-01 -5.26838541e-01 -1.32586026e+00 -1.10219896e+00 6.10718906e-01 -6.34407580e-01 6.95020378e-01 -5.29986262e-01 -1.33809090e+00 1.08233666e+00 -3.49125236e-01 1.12452105e-01 1.01032138e+00 3.34164083e-01 -7.83344865e-01 4.61389981e-02 -1.74534023e+00 7.23619044e-01 4.80732292e-01 -5.27196169e-01 -2.18951166e-01 2.45107278e-01 6.86701119e-01 -6.84695914e-02 -1.01121831e+00 1.17073320e-01 6.66797817e-01 -8.26396346e-01 8.29061508e-01 -1.20357251e+00 5.64911738e-02 2.49161288e-01 -2.35656530e-01 -1.04960740e+00 -1.48296664e-02 -1.00426722e+00 -1.10156171e-01 1.02337754e+00 8.14757407e-01 -1.24881613e+00 1.00707161e+00 1.45833158e+00 5.91124654e-01 -8.41902792e-01 -1.16099560e+00 -8.91354680e-01 8.49985361e-01 -2.27696151e-01 4.95406926e-01 1.19286323e+00 8.63211527e-02 -2.30325729e-01 -4.17921066e-01 5.54574966e-01 6.57825947e-01 -3.18356127e-01 8.49932075e-01 -1.27195871e+00 -4.57341552e-01 -2.53263265e-01 -5.41664839e-01 -7.19752431e-01 4.03826535e-01 -8.22080135e-01 -3.63159806e-01 -1.71063185e-01 -8.73135403e-02 -7.55092144e-01 -1.26337811e-01 3.26796323e-01 -5.08968346e-02 1.34717688e-01 -8.27675909e-02 8.91226679e-02 -1.65280819e-01 5.34056313e-02 8.80259693e-01 3.25938821e-01 -9.03320462e-02 5.78219295e-01 -1.06925642e+00 5.29605091e-01 8.94604802e-01 -7.44136691e-01 -4.26476181e-01 -2.15497781e-02 5.87157786e-01 -2.07619555e-02 5.56148112e-01 -6.41878843e-01 2.30332106e-01 -8.37813467e-02 3.59563589e-01 -2.05993652e-03 3.21898013e-01 -1.08554995e+00 1.17367864e-01 3.81287813e-01 -9.07097101e-01 -2.12131068e-02 2.18521103e-01 6.63488209e-01 3.95211697e-01 -3.88070911e-01 1.06277955e+00 1.46545902e-01 3.65321159e-01 4.07810181e-01 -3.34935725e-01 4.80733782e-01 1.02430952e+00 -2.75596738e-01 -5.19387603e-01 -5.45800865e-01 -8.64383280e-01 1.86496183e-01 6.91743910e-01 1.33093121e-02 3.60399604e-01 -8.54525089e-01 -6.12958491e-01 6.34970784e-01 2.42307726e-02 -1.24181218e-01 5.17760888e-02 8.64589453e-01 -2.09134340e-01 2.15089977e-01 1.43638641e-01 -2.95433789e-01 -1.21778035e+00 5.38566470e-01 6.36645496e-01 -9.87583101e-02 -4.57751811e-01 7.60068834e-01 3.14911842e-01 -2.10962385e-01 4.13163066e-01 -1.67243481e-01 3.44955415e-01 -1.40220359e-01 4.26840097e-01 4.53267187e-01 -9.40239057e-02 -3.64400417e-01 -1.06978841e-01 2.26331964e-01 -6.04832530e-01 -7.08218664e-02 1.20003343e+00 1.41779840e-01 -9.12387446e-02 4.26434666e-01 1.69991720e+00 1.29155308e-01 -1.23668754e+00 -2.28337839e-01 -1.29669420e-02 -4.85927850e-01 -1.40304610e-01 -5.66684723e-01 -8.08855176e-01 8.05456638e-01 3.42848599e-01 6.20819509e-01 6.31193042e-01 -1.49487406e-01 2.81154782e-01 7.77728617e-01 6.09577298e-02 -1.01172352e+00 -4.21190739e-01 2.07492366e-01 6.63841069e-01 -1.05520058e+00 1.53105676e-01 -9.60040465e-02 -7.19264567e-01 1.06337178e+00 4.16154712e-01 -2.54588217e-01 8.28987062e-01 2.46464863e-01 -2.40320131e-01 -4.48157154e-02 -6.64137900e-01 7.00827122e-01 3.06039061e-02 5.68083525e-01 -5.15735336e-02 -1.55586317e-01 1.02014251e-01 8.19796205e-01 -4.41220134e-01 -3.39298218e-01 7.87531674e-01 5.99840224e-01 -2.24173754e-01 -9.70443368e-01 -5.30771971e-01 4.55390483e-01 -8.55109096e-01 -9.55351815e-02 -5.35976827e-01 1.14242566e+00 -3.12447608e-01 7.98262358e-01 1.83936059e-01 -1.71317384e-01 1.08388700e-01 3.62920880e-01 1.94771439e-01 -4.67763990e-01 -7.92163193e-01 -4.78467882e-01 -7.17604309e-02 -4.35850084e-01 2.80023038e-01 -7.00784624e-01 -8.65173221e-01 -6.12748504e-01 -4.24175084e-01 3.09201002e-01 6.74000204e-01 8.32771420e-01 3.17097515e-01 -9.89046544e-02 8.12319696e-01 -1.47856653e-01 -1.46661520e+00 -6.56269312e-01 -9.18079615e-01 3.56212646e-01 7.36207128e-01 -2.69698918e-01 -1.11340308e+00 -4.76222336e-01]
[5.963038921356201, 6.969730377197266]
8b910fd1-ebc3-42f5-af64-da332a07a100
hirevae-an-online-and-adaptive-factor-model
2306.02848
null
https://arxiv.org/abs/2306.02848v1
https://arxiv.org/pdf/2306.02848v1.pdf
HireVAE: An Online and Adaptive Factor Model Based on Hierarchical and Regime-Switch VAE
Factor model is a fundamental investment tool in quantitative investment, which can be empowered by deep learning to become more flexible and efficient in practical complicated investing situations. However, it is still an open question to build a factor model that can conduct stock prediction in an online and adaptive setting, where the model can adapt itself to match the current market regime identified based on only point-in-time market information. To tackle this problem, we propose the first deep learning based online and adaptive factor model, HireVAE, at the core of which is a hierarchical latent space that embeds the underlying relationship between the market situation and stock-wise latent factors, so that HireVAE can effectively estimate useful latent factors given only historical market information and subsequently predict accurate stock returns. Across four commonly used real stock market benchmarks, the proposed HireVAE demonstrate superior performance in terms of active returns over previous methods, verifying the potential of such online and adaptive factor model.
['Dahua Lin', 'Bo Dai', 'Anyi Rao', 'Zikai Wei']
2023-06-05
null
null
null
null
['open-question', 'stock-prediction']
['natural-language-processing', 'time-series']
[-8.25968683e-01 -3.63191843e-01 -5.60710132e-01 -9.79225338e-02 -2.56045789e-01 -8.06571782e-01 6.38853848e-01 -3.47385764e-01 -1.37519464e-01 3.69898707e-01 4.67307180e-01 -6.45553589e-01 -4.93535370e-01 -1.03868771e+00 -3.11586887e-01 -3.91221017e-01 -1.87015221e-01 5.75563431e-01 1.67074695e-01 -2.88760304e-01 3.61064434e-01 4.99982029e-01 -1.20150328e+00 4.10534292e-02 3.73327583e-01 1.44546950e+00 -1.11281335e-01 3.56884092e-01 -2.70652354e-01 1.08133841e+00 -2.99611837e-01 -7.07487404e-01 1.10430539e+00 -4.83873487e-02 1.85659435e-02 -3.53175029e-02 -3.34569126e-01 -8.85460377e-01 -6.24128640e-01 7.05830812e-01 2.80226588e-01 -1.00819297e-01 6.30073845e-01 -9.74402130e-01 -4.93510395e-01 1.00784588e+00 -4.79130417e-01 7.10614681e-01 -3.26183885e-01 3.49616140e-01 1.44393384e+00 -8.80655110e-01 1.36068210e-01 9.63422120e-01 6.70514822e-01 -2.28967313e-02 -9.91472542e-01 -9.91005719e-01 3.19426507e-01 -4.31101732e-02 -5.98056138e-01 3.56112681e-02 8.52317631e-01 -8.22858989e-01 8.45853269e-01 -2.17446268e-01 1.20696831e+00 7.55454004e-01 7.25459874e-01 1.00785577e+00 1.13677490e+00 1.29210159e-01 2.77223915e-01 -1.42392084e-01 3.20791709e-03 1.38639584e-01 3.21373224e-01 5.02074480e-01 -7.46937811e-01 -1.28553227e-01 1.26522315e+00 7.85182953e-01 2.57381439e-01 -2.24747226e-01 -1.24278820e+00 1.10086930e+00 2.51683712e-01 3.34575564e-01 -9.54771280e-01 2.60407060e-01 1.62121311e-01 7.82982349e-01 6.47547901e-01 6.90486491e-01 -8.06414306e-01 -4.58856493e-01 -1.34429634e+00 6.16926253e-01 6.66662037e-01 5.56421518e-01 7.35979855e-01 6.66816711e-01 -3.01520735e-01 1.45817205e-01 4.72741246e-01 5.14250994e-01 9.23038006e-01 -8.32442760e-01 4.34020758e-01 9.07398999e-01 3.10155302e-01 -8.18333149e-01 -5.04319906e-01 -1.21550512e+00 -5.93394160e-01 2.33499750e-01 1.96022317e-01 -2.88663030e-01 -5.87018251e-01 1.22960150e+00 1.77754257e-02 3.70987892e-01 4.48698290e-02 4.03165251e-01 -5.75076938e-02 6.35497332e-01 -4.10855442e-01 -2.78006405e-01 1.10957265e+00 -1.04619110e+00 -9.95647967e-01 -3.25662822e-01 2.69340545e-01 -5.06147563e-01 5.16229331e-01 3.47821563e-01 -1.01631463e+00 -5.49790800e-01 -1.14494598e+00 5.36499143e-01 -3.21463734e-01 6.73140585e-03 1.30940974e+00 3.51561219e-01 -8.65625679e-01 8.26237619e-01 -9.50460732e-01 8.54525268e-01 3.51641625e-01 3.79753143e-01 1.78698674e-01 3.62536877e-01 -1.53387582e+00 5.02896130e-01 4.96790081e-01 3.56502205e-01 -9.46637690e-01 -1.16215980e+00 -2.13995382e-01 4.46694702e-01 5.36421716e-01 -6.41833007e-01 1.36019361e+00 -8.73565257e-01 -1.72790813e+00 -4.86829989e-02 5.25497794e-01 -9.86248434e-01 8.87595832e-01 -4.63226318e-01 -3.34373832e-01 -1.10848866e-01 -1.37711152e-01 -1.30383208e-01 1.25115299e+00 -4.34716314e-01 -6.96687877e-01 -2.54687399e-01 1.12043262e-01 -1.18288726e-01 -6.60119772e-01 -6.60829619e-02 -1.31300718e-01 -1.04947340e+00 2.45847995e-03 -6.67089641e-01 -5.14603436e-01 -4.71081167e-01 2.59166002e-01 -2.45928288e-01 3.15342844e-01 -7.80230045e-01 1.55442631e+00 -1.86757529e+00 -9.68217254e-02 3.11174393e-01 3.92156690e-01 8.76260623e-02 2.34903082e-01 6.80221558e-01 -2.87116885e-01 -1.21572912e-01 2.62095302e-01 -1.98305249e-01 5.76340914e-01 -1.45423964e-01 -9.00300920e-01 2.70736188e-01 1.79636583e-01 1.41955030e+00 -6.68953657e-01 3.00893217e-01 3.58867645e-03 -1.25519738e-01 -4.79488730e-01 5.07863760e-01 -1.25216469e-01 6.39336416e-03 -6.84206784e-01 6.24270737e-01 5.20873487e-01 -2.72146642e-01 -1.34799495e-01 1.84743062e-01 -3.21206957e-01 -6.48932979e-02 -1.32747543e+00 1.14455163e+00 -3.30544800e-01 4.79078919e-01 -3.92590523e-01 -9.70465481e-01 1.14511549e+00 3.13603729e-01 7.92213440e-01 -7.17413008e-01 -3.56498105e-03 5.05425155e-01 4.64601032e-02 2.15262488e-01 5.47908485e-01 -5.35807610e-01 -2.66418904e-01 1.08587694e+00 1.38703898e-01 2.88302511e-01 1.26991436e-01 -9.86104552e-03 1.09765506e+00 -1.01186968e-01 2.78724104e-01 -3.85291725e-01 2.26001162e-02 -4.03527826e-01 7.91496813e-01 6.35304511e-01 3.65798324e-02 2.46969610e-02 7.95756638e-01 -1.01483393e+00 -1.03034890e+00 -9.21806037e-01 4.93974239e-02 9.62673128e-01 -5.24655521e-01 -1.74067140e-01 -2.53149897e-01 -6.04193151e-01 5.46233296e-01 3.52540106e-01 -8.05189550e-01 -1.21233083e-01 -1.49382710e-01 -9.32718277e-01 1.92789529e-02 9.00407970e-01 5.95842063e-01 -8.21110904e-01 -7.17783451e-01 7.90782034e-01 5.46569288e-01 -7.19719112e-01 -3.21900725e-01 2.65000790e-01 -9.19676244e-01 -8.52015257e-01 -9.72401142e-01 9.91785154e-02 3.16524267e-01 1.22852825e-01 1.27751970e+00 -1.62979849e-02 2.41221711e-01 2.91820377e-01 -3.17396402e-01 -8.64555418e-01 -1.62978411e-01 2.09360883e-01 2.08597943e-01 4.22171026e-01 4.60108668e-01 -6.32749856e-01 -9.73106205e-01 3.58919472e-01 -1.18144846e+00 1.42546175e-02 8.91349375e-01 9.76996601e-01 1.62946507e-01 6.20454729e-01 8.67744148e-01 -7.99746335e-01 6.72941983e-01 -6.58255517e-01 -1.17017663e+00 2.52405286e-01 -1.13971639e+00 2.16604427e-01 4.41349864e-01 -3.71812969e-01 -1.04083991e+00 -3.74494046e-01 -2.96900328e-02 -6.44943476e-01 8.05668890e-01 1.07724237e+00 2.40945891e-01 3.24601531e-01 5.34063845e-04 2.82730073e-01 -1.74120203e-01 -5.82777441e-01 -5.09836944e-03 3.46252680e-01 3.40156764e-01 -2.88492709e-01 1.49168527e+00 3.06065530e-01 1.06356718e-01 1.85674950e-01 -1.09324205e+00 -4.56531048e-01 -9.14233267e-01 -2.09245816e-01 3.71663570e-01 -1.25650764e+00 -4.31673139e-01 7.90188968e-01 -4.85570818e-01 -4.54931259e-01 -4.54278022e-01 5.99115670e-01 -6.25897467e-01 -1.14950359e-01 -8.16121161e-01 -1.06213939e+00 -3.08055371e-01 -9.46550906e-01 7.87796319e-01 2.14463428e-01 1.64798424e-01 -1.38927639e+00 4.55866218e-01 2.77393907e-01 8.18065822e-01 1.87789559e-01 7.37394512e-01 -9.40339327e-01 -8.78340900e-01 -7.95539200e-01 9.55641910e-04 5.15435517e-01 1.71208486e-01 -9.57337841e-02 -6.43876493e-01 -6.43805861e-01 3.22038442e-01 -2.12675929e-02 9.95401561e-01 1.75135657e-01 5.15661657e-01 -3.69973749e-01 1.90050110e-01 7.76254714e-01 1.36097395e+00 3.54170442e-01 4.75187331e-01 7.77801812e-01 3.50127906e-01 4.15261149e-01 6.95887148e-01 9.06464398e-01 2.87324429e-01 4.29235786e-01 1.56854734e-01 1.77704229e-03 5.28140366e-01 -7.59322166e-01 5.27259529e-01 1.19004428e+00 -3.05110723e-01 1.56475350e-01 -8.21443498e-01 3.10806423e-01 -2.07962871e+00 -1.11625433e+00 6.59301817e-01 1.94558227e+00 4.48113829e-01 6.39453232e-01 3.78762811e-01 5.40332906e-02 1.45487249e-01 3.55265468e-01 -1.07328808e+00 2.14184120e-01 -2.06702590e-01 1.28754064e-01 7.96913624e-01 -4.54445966e-02 -9.73998070e-01 7.79495180e-01 6.79194403e+00 7.09787846e-01 -1.09647024e+00 -1.21976376e-01 8.72391224e-01 7.35676363e-02 -7.81064212e-01 4.67377156e-02 -1.09427559e+00 5.99102736e-01 1.21741593e+00 -5.62464833e-01 4.44289923e-01 9.96891260e-01 1.65661722e-01 5.26892483e-01 -9.61018145e-01 6.06061697e-01 -4.60536093e-01 -1.69448602e+00 -4.28168662e-02 4.71954137e-01 9.75980282e-01 -1.32779688e-01 5.41102946e-01 8.65559578e-01 4.26023602e-01 -8.22856843e-01 8.23612273e-01 1.08628106e+00 7.85160586e-02 -9.67670739e-01 1.08862090e+00 4.37890083e-01 -1.48093832e+00 -7.49059379e-01 -4.36075568e-01 -3.36931497e-01 2.07800977e-02 4.62096065e-01 -5.15843093e-01 6.29641235e-01 5.61403215e-01 1.04656255e+00 -5.93062878e-01 8.49920273e-01 -2.34828647e-02 7.11791933e-01 -4.01157141e-02 1.70522168e-01 5.42412937e-01 -4.34872478e-01 1.42797872e-01 5.71739614e-01 7.82527149e-01 -1.90957740e-01 4.52941991e-02 8.62624109e-01 1.02383278e-01 4.06045467e-02 -3.99046987e-01 -6.55594528e-01 1.42515197e-01 1.12306881e+00 -8.71091127e-01 -2.10846588e-01 -6.57350004e-01 3.34931612e-01 5.65452427e-02 3.87167782e-01 -5.72056592e-01 5.89171797e-02 5.76719761e-01 3.46199900e-01 6.78513944e-01 -4.67607409e-01 -6.00112937e-02 -1.52071095e+00 -1.75193567e-02 -9.23058033e-01 4.44624573e-01 -3.29398423e-01 -1.52241135e+00 3.55985433e-01 -5.74373752e-02 -1.53626168e+00 -7.15037346e-01 -1.01589096e+00 -8.55121076e-01 6.97661340e-01 -1.86252689e+00 -8.61279905e-01 4.07309026e-01 4.54045087e-01 4.98305231e-01 -1.01994956e+00 2.45475292e-01 -9.25463159e-03 -5.08797824e-01 3.99044186e-01 6.70124531e-01 2.38233685e-01 1.33059666e-01 -1.50786006e+00 7.85691500e-01 9.98988867e-01 3.66588324e-01 7.42536247e-01 4.27740544e-01 -8.04451287e-01 -1.61005151e+00 -9.76660252e-01 3.52170318e-01 -4.25860405e-01 1.61490428e+00 -3.10211480e-01 -8.86105597e-01 8.12254429e-01 1.47606537e-01 -5.21199889e-02 8.49224865e-01 -2.80112848e-02 -2.37914488e-01 -3.95095438e-01 -5.94672620e-01 3.23455662e-01 6.28586292e-01 -5.54497361e-01 -5.93346417e-01 1.52952373e-01 1.02006149e+00 -2.63765574e-01 -1.31158924e+00 1.64913252e-01 5.62305868e-01 -1.16016114e+00 9.67204332e-01 -6.41979396e-01 1.77685603e-01 -1.64499879e-02 5.12661934e-02 -1.39418638e+00 -7.93165386e-01 -1.26174986e+00 -5.53684175e-01 1.11036837e+00 3.94328207e-01 -1.05050921e+00 8.21529388e-01 8.11456263e-01 7.48082101e-02 -8.35804284e-01 -1.04488480e+00 -7.12605059e-01 1.90961614e-01 -4.00829166e-01 1.39840949e+00 7.78603375e-01 -3.23267430e-01 1.84522811e-02 -6.31698430e-01 -1.05076157e-01 5.52760899e-01 7.04517365e-01 8.63983154e-01 -1.41629350e+00 -7.41623402e-01 -7.61382520e-01 -2.69995898e-01 -1.11340559e+00 3.52669626e-01 -5.69672346e-01 -5.56306303e-01 -1.02629995e+00 5.94966449e-02 -1.86279818e-01 -1.08334446e+00 1.88776031e-01 -3.51681828e-01 -2.12489173e-01 3.97193044e-01 4.65000868e-01 -4.80913430e-01 8.68658364e-01 1.19964147e+00 -1.44552276e-01 -1.65128693e-01 6.35815561e-01 -9.29869473e-01 4.83061790e-01 6.40232623e-01 -1.52507588e-01 -3.21897477e-01 -2.38680262e-02 8.92396748e-01 4.41776782e-01 -4.28850390e-02 -6.64927006e-01 2.54742771e-01 -5.34446001e-01 5.53101242e-01 -7.50476182e-01 -1.00816011e-01 -8.63929212e-01 5.09168744e-01 6.62312448e-01 -1.88163564e-01 6.89639628e-01 3.30383070e-02 8.12431335e-01 -5.26255786e-01 -1.39195934e-01 1.23760127e-01 -1.45790234e-01 -6.16881430e-01 1.02579820e+00 -3.62046152e-01 -2.42995784e-01 9.26793396e-01 -1.13267787e-02 -1.40601337e-01 -5.08301318e-01 -2.44489655e-01 3.48887533e-01 9.70208719e-02 4.88699645e-01 4.49927717e-01 -1.70335448e+00 -6.83212757e-01 4.08709526e-01 -1.04179643e-01 -2.21633002e-01 3.82756203e-01 4.32680249e-01 -3.57211143e-01 7.74434566e-01 -2.35029876e-01 -5.35935462e-02 -6.23685680e-02 7.83028364e-01 4.26535875e-01 -9.78063285e-01 -5.44848323e-01 5.39989293e-01 2.22147763e-01 5.00763021e-02 -5.26294447e-02 -5.19718111e-01 -3.84445757e-01 6.32660449e-01 8.54878902e-01 2.24022493e-01 -2.06612665e-02 -3.48931879e-01 3.22233707e-01 4.88684326e-01 -5.25878966e-02 -5.84926009e-02 1.93227422e+00 5.84065430e-02 -6.10406697e-02 8.79840791e-01 7.20721245e-01 -1.28570378e-01 -2.03326988e+00 -6.38314068e-01 6.19736791e-01 -7.74205506e-01 3.70857567e-01 -3.31240535e-01 -1.53798997e+00 7.72437632e-01 5.28814554e-01 5.08229196e-01 9.96587574e-01 -4.77061361e-01 1.00300789e+00 3.39161009e-01 3.64953429e-01 -1.27618706e+00 4.21122670e-01 2.08921626e-01 6.95259869e-01 -9.94737804e-01 9.14787650e-02 6.20439231e-01 -6.40077949e-01 1.47801721e+00 2.21449912e-01 -4.45043087e-01 1.28922641e+00 2.45887995e-01 1.61594212e-01 -3.54173452e-01 -1.15462732e+00 1.29345313e-01 5.84825218e-01 -7.76872262e-02 -7.55805597e-02 1.55312821e-01 3.04904372e-01 1.21183860e+00 -5.09962082e-01 -4.48725410e-02 3.80150467e-01 8.19977105e-01 -3.63627583e-01 -1.10050666e+00 -6.40585199e-02 8.09119463e-01 -8.65923405e-01 1.12437807e-01 3.80264260e-02 6.72296584e-01 -3.70878220e-01 2.86534756e-01 2.74380803e-01 -2.91089416e-01 4.05782133e-01 3.94367538e-02 -2.31755957e-01 -5.28994799e-01 -7.05711365e-01 4.73839998e-01 -6.13031209e-01 -5.79866946e-01 -3.56048673e-01 -8.94942999e-01 -6.01389110e-01 -1.29145488e-01 -2.60946900e-01 1.78456411e-01 2.81679362e-01 1.12280738e+00 5.22367954e-02 6.61274612e-01 1.45189857e+00 -8.65539074e-01 -1.17531526e+00 -8.23699415e-01 -1.13372481e+00 1.75324529e-02 4.37289566e-01 -1.01184237e+00 -6.71828449e-01 -2.95226008e-01]
[4.442282676696777, 4.191076755523682]
76b80aaa-c9e8-4383-ac3f-2b33723343fa
deep-inverse-tone-mapping-using-ldr-based
1903.01277
null
http://arxiv.org/abs/1903.01277v1
http://arxiv.org/pdf/1903.01277v1.pdf
Deep Inverse Tone Mapping Using LDR Based Learning for Estimating HDR Images with Absolute Luminance
In this paper, a novel inverse tone mapping method using a convolutional neural network (CNN) with LDR based learning is proposed. In conventional inverse tone mapping with CNNs, generated HDR images cannot have absolute luminance, although relative luminance can. Moreover, loss functions suitable for learning HDR images are problematic, so it is difficult to train CNNs by directly using HDR images. In contrast, the proposed method enables us not only to estimate absolute luminance, but also to train a CNN by using LDR images. The CNN used in the proposed method learns a transformation from various input LDR images to LDR images mapped by Reinhard's global operator. Experimental results show that HDR images generated by the proposed method have higher-quality than HDR ones generated by conventional inverse tone mapping methods,in terms of HDR-VDP-2.2 and PU encoding + MS-SSIM.
[]
2019-02-28
null
null
null
null
['tone-mapping', 'inverse-tone-mapping']
['computer-vision', 'computer-vision']
[ 4.44219828e-01 -1.74707264e-01 -7.91815072e-02 -2.28994310e-01 -4.90372211e-01 -4.26479317e-02 3.56768489e-01 -5.33345938e-01 -3.30495059e-01 1.21522009e+00 -1.80985361e-01 -1.65186465e-01 2.13844076e-01 -1.28876460e+00 -8.14959228e-01 -7.31391490e-01 3.03756535e-01 -4.45974506e-02 1.04016736e-01 -5.54588854e-01 9.62480903e-02 3.19408506e-01 -1.77720571e+00 6.90430552e-02 1.01222992e+00 1.27613556e+00 4.84453112e-01 7.54807174e-01 1.46000102e-01 1.11395109e+00 -7.21331060e-01 -1.83227047e-01 3.76274109e-01 -9.05568898e-01 -5.63644826e-01 -1.30523637e-01 3.62350881e-01 -6.87319875e-01 -6.54842079e-01 1.10828233e+00 7.13470638e-01 2.16647144e-02 5.38619339e-01 -1.01705837e+00 -1.43901825e+00 3.35351557e-01 -5.62980831e-01 8.79819840e-02 2.48149067e-01 1.43252566e-01 6.00505173e-01 -5.83040655e-01 3.90661567e-01 1.02083945e+00 6.03650153e-01 6.01564407e-01 -1.43580401e+00 -7.95442581e-01 -5.99032998e-01 4.73335832e-01 -1.59343946e+00 -2.70629108e-01 8.72610152e-01 1.33293075e-02 9.08457696e-01 4.04010385e-01 6.72545373e-01 5.70011973e-01 1.41114563e-01 3.98002446e-01 1.62515247e+00 -5.73922276e-01 4.23672646e-02 1.88235000e-01 -5.66573739e-01 5.01663268e-01 -5.33495061e-02 4.83049721e-01 -3.67978722e-01 6.05695069e-01 1.31562674e+00 -3.86202097e-01 -6.05953932e-01 2.80054081e-02 -1.19440722e+00 7.26530313e-01 6.03827775e-01 2.90196657e-01 -4.59767953e-02 2.30467960e-01 2.48917297e-01 7.03325033e-01 5.69588482e-01 2.82205492e-01 -7.27034584e-02 -2.74686962e-02 -9.07218456e-01 -3.27844560e-01 3.46626103e-01 9.66411829e-01 1.01266897e+00 3.56992543e-01 2.44565476e-02 1.04046333e+00 1.06158569e-01 8.04715097e-01 2.95563996e-01 -1.15224338e+00 2.96334147e-01 9.42056812e-03 -2.76830457e-02 -8.50603163e-01 -2.82547157e-02 -2.50872574e-03 -1.40952837e+00 7.66081095e-01 2.48735577e-01 2.50705238e-02 -1.09386218e+00 1.54410946e+00 -1.64020538e-01 7.00408891e-02 2.28514835e-01 1.10721719e+00 7.31483936e-01 1.09596562e+00 -3.61073948e-02 -2.40340769e-01 8.87505889e-01 -5.92249930e-01 -1.14700472e+00 2.66732782e-01 2.57208914e-01 -8.97664428e-01 1.25771511e+00 5.90606511e-01 -1.33752453e+00 -1.11209118e+00 -1.40151882e+00 -5.12030840e-01 -3.07221800e-01 3.30211014e-01 2.51953214e-01 7.64774323e-01 -1.54812324e+00 6.51260614e-01 -1.14342660e-01 -1.54294923e-01 1.83366448e-01 4.14389104e-01 -1.44533202e-01 -1.11323856e-01 -1.79753888e+00 1.13432157e+00 5.68499386e-01 1.35283977e-01 -7.32135177e-01 -6.52684152e-01 -8.61484349e-01 1.15535952e-01 -1.99508578e-01 -3.71579140e-01 7.98416495e-01 -1.16302669e+00 -2.00092602e+00 7.60584235e-01 2.84779459e-01 -4.69039917e-01 6.11554325e-01 -4.64711562e-02 -7.00555384e-01 1.85438409e-01 -2.27944836e-01 8.41310859e-01 8.63490224e-01 -1.37714756e+00 -6.43682122e-01 1.32928863e-01 1.03569292e-01 1.93662256e-01 -1.84680149e-01 -3.44524264e-01 -2.66793221e-01 -6.19956553e-01 -7.66686127e-02 -4.73585635e-01 1.74496189e-01 3.90014648e-01 -3.33384752e-01 1.81191355e-01 1.30721724e+00 -7.95577526e-01 1.07240987e+00 -1.94969106e+00 -2.00687781e-01 1.21050812e-01 1.34026945e-01 4.97972310e-01 -2.10042313e-01 -1.06722787e-01 -4.05860782e-01 4.25820425e-02 -2.56759554e-01 -2.56518777e-02 2.20114011e-02 -5.97496964e-02 -2.95174181e-01 4.32464600e-01 2.44682983e-01 8.73705685e-01 -7.46638775e-01 -6.32538080e-01 8.17999899e-01 1.10641217e+00 -1.71190515e-01 4.23613548e-01 -5.47128404e-03 5.25583804e-01 2.38020614e-01 2.84536481e-01 1.04350650e+00 1.44005585e-02 5.72024621e-02 -7.04157233e-01 -3.40947270e-01 -1.86816320e-01 -8.31383526e-01 1.27787268e+00 -1.03729928e+00 1.14159298e+00 -4.69213665e-01 -8.50645304e-01 1.50602651e+00 3.85068864e-01 2.49178097e-01 -1.44900775e+00 1.92393139e-01 3.63882244e-01 -2.32158706e-01 -3.51949066e-01 5.50513387e-01 -3.04810286e-01 1.89727738e-01 4.97914851e-01 -8.74691978e-02 -3.79318774e-01 1.72337160e-01 -3.94177347e-01 4.98138100e-01 3.13115656e-01 2.03526124e-01 -4.13214900e-02 8.10934067e-01 -3.13241273e-01 2.84190238e-01 3.61764491e-01 3.44096720e-02 1.01367962e+00 3.32168370e-01 -6.72455668e-01 -1.69807661e+00 -1.32444036e+00 -3.51599574e-01 5.34836650e-01 3.25002104e-01 4.27330554e-01 -6.62564874e-01 -2.13419832e-02 -6.80405498e-01 7.61079550e-01 -6.06499791e-01 -3.08837593e-01 -7.88633585e-01 -6.77141964e-01 6.68669999e-01 3.99088770e-01 1.44585848e+00 -1.33451176e+00 -5.13675213e-01 1.36179656e-01 -5.75998485e-01 -1.06078136e+00 -5.94936728e-01 2.17851058e-01 -6.78703964e-01 -7.79404044e-01 -1.25888026e+00 -1.07756281e+00 5.36377609e-01 9.63106453e-02 1.21507061e+00 1.55191362e-01 -4.24639404e-01 -1.83556467e-01 -3.77676845e-01 1.82111412e-01 -7.00704217e-01 -9.62886885e-02 -2.67185628e-01 6.14523776e-02 1.44047797e-01 -4.27181333e-01 -7.43385911e-01 2.55612791e-01 -1.03122890e+00 3.05923164e-01 5.81142843e-01 8.81152928e-01 7.76523948e-01 4.11490411e-01 7.45983779e-01 -5.85874557e-01 4.14054364e-01 9.09338742e-02 -7.63106823e-01 3.19741547e-01 -5.67288637e-01 -2.69857854e-01 9.82125223e-01 -3.69031161e-01 -1.37775493e+00 -8.01484212e-02 -3.24268788e-01 -5.19936323e-01 -1.84362322e-01 -1.73683316e-01 -3.09324443e-01 -4.20919418e-01 6.16307855e-01 6.82980955e-01 -5.10434434e-02 3.19364108e-02 4.84592766e-01 7.53209174e-01 7.76460171e-01 -1.44927785e-01 1.02559471e+00 4.55660939e-01 -7.35646812e-03 -8.06293488e-01 -2.92902440e-01 3.73292506e-01 -4.97421026e-01 -3.03497702e-01 1.25924015e+00 -1.08077979e+00 -8.47236276e-01 8.13877463e-01 -9.64582920e-01 -8.37656796e-01 -4.39884752e-01 4.09965426e-01 -1.04357910e+00 1.67404816e-01 -9.56636012e-01 -4.70321327e-01 -1.49270251e-01 -1.15563464e+00 8.20765436e-01 2.84273863e-01 1.88078627e-01 -1.17765903e+00 -1.09659918e-01 4.45409454e-02 8.39447200e-01 1.34230345e-01 9.76335108e-01 3.68339121e-01 -8.61817479e-01 2.32664853e-01 -7.07611263e-01 6.82928741e-01 5.60585737e-01 -1.80489659e-01 -1.07794166e+00 -2.04533115e-01 5.70026152e-02 -4.82716948e-01 6.56447649e-01 6.73161924e-01 1.48266816e+00 -9.10189599e-02 1.94542095e-01 9.39302683e-01 1.97940159e+00 5.46123087e-01 1.71332848e+00 4.82613653e-01 5.70791841e-01 1.80459857e-01 6.52070999e-01 1.97866291e-01 6.23433329e-02 7.49491811e-01 1.36759818e-01 -8.84579539e-01 -7.48725235e-01 -1.18668020e-01 1.78951800e-01 6.99971080e-01 -1.22948587e-01 -3.45840275e-01 -3.60403240e-01 2.61169523e-01 -1.13363516e+00 -8.97400737e-01 -9.43450928e-02 2.21668053e+00 1.07922590e+00 -6.28681295e-03 -2.87967235e-01 4.41722542e-01 1.12162793e+00 1.55558124e-01 -4.26343262e-01 -4.57789749e-01 -4.89651918e-01 5.07300019e-01 6.09869778e-01 5.77021897e-01 -9.54056203e-01 8.36543560e-01 6.33373833e+00 8.71130049e-01 -1.54976833e+00 5.53314202e-02 1.11265337e+00 1.94146737e-01 -2.78796583e-01 -5.17768264e-01 -3.21897000e-01 5.83618939e-01 7.24053442e-01 -4.89632934e-02 7.64646053e-01 2.75081456e-01 3.69547457e-01 -2.51291066e-01 -7.52691031e-01 1.27299869e+00 1.98647499e-01 -1.17606556e+00 7.18717575e-02 -1.89231202e-01 1.05755627e+00 -5.63447654e-01 7.31437266e-01 1.48005143e-01 1.24812946e-01 -1.22231281e+00 4.84916687e-01 3.83048713e-01 1.78580296e+00 -1.02260613e+00 7.36778915e-01 -3.32577884e-01 -1.27652967e+00 3.26495737e-01 -6.61290646e-01 2.68878758e-01 1.53476804e-01 5.93711376e-01 -4.98881102e-01 3.64228696e-01 8.68477225e-01 7.01149821e-01 -3.55279624e-01 6.07286096e-01 -1.28591701e-01 2.24783525e-01 1.76380306e-01 3.21446776e-01 -2.43073553e-01 -2.84380794e-01 -8.41950402e-02 9.97345328e-01 6.89003646e-01 1.06913939e-01 -5.41482329e-01 1.27199614e+00 -1.38006881e-01 -9.85554531e-02 -8.33870947e-01 2.03460276e-01 2.87554085e-01 1.08926952e+00 -6.90972745e-01 -4.48832124e-01 -2.79541612e-01 1.22882426e+00 -2.28394359e-01 6.55858397e-01 -1.21197820e+00 -1.03449762e+00 3.94193411e-01 -1.02598490e-02 2.79114157e-01 1.97213814e-02 -2.38607362e-01 -8.53015125e-01 -2.94869095e-01 -8.09819758e-01 -3.18474531e-01 -1.32936466e+00 -1.08628416e+00 9.31288242e-01 -4.54276055e-01 -1.60894060e+00 1.83945894e-02 -4.96558905e-01 -4.30577368e-01 1.10977054e+00 -2.09831572e+00 -9.56907749e-01 -5.71348846e-01 7.33596981e-01 4.23593909e-01 -5.63006103e-02 5.51554680e-01 6.82778060e-01 -1.63361713e-01 5.71775496e-01 1.06753774e-01 1.82615370e-01 8.61869812e-01 -1.28897762e+00 3.75129491e-01 6.42581940e-01 -5.04894614e-01 9.04707462e-02 6.01927578e-01 -4.24001396e-01 -9.51667130e-01 -1.44145739e+00 5.98953187e-01 2.73248643e-01 2.11562738e-01 -1.22207008e-01 -9.13148940e-01 3.99354547e-01 5.78199983e-01 5.35372645e-02 2.75082350e-01 -6.62073314e-01 -2.30581537e-01 -4.84520614e-01 -1.43841684e+00 5.48146844e-01 6.79680109e-01 -8.17120075e-01 -5.66116795e-02 -2.38608066e-02 8.22764874e-01 -3.70122403e-01 -1.20926321e+00 3.73359501e-01 3.74542475e-01 -1.27436459e+00 1.13299751e+00 2.92087793e-01 7.86301732e-01 -6.84627175e-01 -2.17748642e-01 -1.43953204e+00 -1.90792292e-01 -2.01984212e-01 1.88831791e-01 1.14787900e+00 3.17881733e-01 -7.31562078e-01 5.72556138e-01 8.90071988e-02 -4.60092798e-02 -2.81138986e-01 -7.90047586e-01 -8.74140918e-01 4.70226724e-03 -1.55350389e-02 6.91711247e-01 9.89982784e-01 -5.08294582e-01 -2.29610205e-02 -7.73361683e-01 -9.41069704e-03 7.75889695e-01 -3.03637460e-02 4.12517756e-01 -8.11575711e-01 -7.87132457e-02 -2.67346621e-01 -3.73373955e-01 -8.26389253e-01 1.24510057e-01 -5.90047061e-01 4.78083044e-01 -1.48278821e+00 8.24850053e-02 -7.54835069e-01 -3.47948909e-01 2.68867910e-01 7.84961507e-02 1.10005355e+00 2.19560638e-01 -5.43153062e-02 -3.24847192e-01 7.08098590e-01 1.76627600e+00 -3.74548614e-01 -6.07771873e-02 -4.83783215e-01 -2.03151211e-01 2.90320814e-01 1.03830075e+00 -1.19389199e-01 -6.41569734e-01 -4.68214691e-01 1.36301368e-01 2.10912824e-01 4.40270752e-01 -1.51758111e+00 -5.92821278e-02 6.03444986e-02 8.87287796e-01 -4.68836159e-01 5.11129856e-01 -8.09970200e-01 4.50396538e-01 5.71713626e-01 -4.20567393e-01 -1.83642104e-01 -6.47199452e-02 9.80802923e-02 -4.82871085e-01 -3.89145277e-02 1.28217864e+00 -6.33804966e-03 -8.27368021e-01 2.35367566e-01 -2.39284426e-01 -1.88333198e-01 9.51192439e-01 -4.03217226e-01 -6.12999618e-01 -7.18528092e-01 -5.70029438e-01 -4.61416900e-01 6.50062680e-01 2.65588343e-01 1.06235111e+00 -1.72270703e+00 -5.80426574e-01 3.81936193e-01 -1.19552448e-01 -3.37387264e-01 5.37870526e-01 6.49894238e-01 -1.09308970e+00 2.55515158e-01 -8.92114699e-01 -3.94103914e-01 -8.30034256e-01 7.07222641e-01 4.91917402e-01 5.06529398e-02 -5.67989826e-01 3.26995879e-01 4.29427862e-01 -2.41413802e-01 -7.45170517e-03 -6.16585687e-02 -2.17293739e-01 -3.46005589e-01 5.58527529e-01 2.97825098e-01 -1.28496913e-02 -6.79589152e-01 2.15971366e-01 9.21615660e-01 3.00115407e-01 -8.52936953e-02 1.23975015e+00 -6.02770507e-01 -2.41547093e-01 2.92763501e-01 1.66008091e+00 -4.95637715e-01 -1.35806644e+00 2.53677294e-02 -6.14355147e-01 -6.91799760e-01 2.72853553e-01 -7.20856369e-01 -1.65909910e+00 1.14676702e+00 1.33346438e+00 2.12400645e-01 1.85591757e+00 -4.60414797e-01 1.09803867e+00 2.21832529e-01 4.59143966e-01 -1.19826865e+00 3.61920446e-01 2.38758415e-01 7.66563177e-01 -1.27314746e+00 -2.91554302e-01 -3.30539614e-01 -4.67154324e-01 1.38422143e+00 8.58945608e-01 -1.52670369e-01 4.27589923e-01 2.21989170e-01 4.89340335e-01 3.07790935e-01 -4.58060384e-01 -3.72585297e-01 -1.04695372e-01 9.92040455e-01 4.97840047e-01 -3.08072865e-01 -1.18818469e-01 -5.12097836e-01 -1.06761530e-01 3.96854222e-01 8.88970256e-01 4.64299232e-01 -4.72043663e-01 -9.26639915e-01 -5.03883481e-01 1.90408126e-01 -3.06694567e-01 -1.94518879e-01 2.88819194e-01 9.30650711e-01 3.74070406e-01 7.79847324e-01 4.74570483e-01 -6.57577455e-01 1.16194993e-01 -5.39709866e-01 6.49799585e-01 1.20268865e-02 2.54486185e-02 -1.02008589e-01 -2.04049438e-01 -3.59025508e-01 -5.87247908e-01 2.62775943e-02 -1.01488030e+00 -6.55131638e-01 -1.97311983e-01 -2.70709336e-01 6.77911818e-01 5.10405481e-01 -2.21625850e-01 8.05243790e-01 1.03093386e+00 -7.23388314e-01 9.77111906e-02 -6.82911277e-01 -9.13296819e-01 3.57591182e-01 5.99584639e-01 -4.18788671e-01 -2.17677400e-01 5.63843369e-01]
[10.97602367401123, -2.2164177894592285]
92775850-3293-40f0-b033-bc7761967b0e
an-analysis-of-svd-for-deep-rotation
2006.14616
null
https://arxiv.org/abs/2006.14616v1
https://arxiv.org/pdf/2006.14616v1.pdf
An Analysis of SVD for Deep Rotation Estimation
Symmetric orthogonalization via SVD, and closely related procedures, are well-known techniques for projecting matrices onto $O(n)$ or $SO(n)$. These tools have long been used for applications in computer vision, for example optimal 3D alignment problems solved by orthogonal Procrustes, rotation averaging, or Essential matrix decomposition. Despite its utility in different settings, SVD orthogonalization as a procedure for producing rotation matrices is typically overlooked in deep learning models, where the preferences tend toward classic representations like unit quaternions, Euler angles, and axis-angle, or more recently-introduced methods. Despite the importance of 3D rotations in computer vision and robotics, a single universally effective representation is still missing. Here, we explore the viability of SVD orthogonalization for 3D rotations in neural networks. We present a theoretical analysis that shows SVD is the natural choice for projecting onto the rotation group. Our extensive quantitative analysis shows simply replacing existing representations with the SVD orthogonalization procedure obtains state of the art performance in many deep learning applications covering both supervised and unsupervised training.
['Noah Snavely', 'Jake Levinson', 'Afshin Rostamizadeh', 'Angjoo Kanazawa', 'Ameesh Makadia', 'Kefan Chen', 'Carlos Esteves']
2020-06-25
null
http://proceedings.neurips.cc/paper/2020/hash/fec3392b0dc073244d38eba1feb8e6b7-Abstract.html
http://proceedings.neurips.cc/paper/2020/file/fec3392b0dc073244d38eba1feb8e6b7-Paper.pdf
neurips-2020-12
['3d-rotation-estimation']
['computer-vision']
[-2.23362640e-01 -1.00593030e-01 -3.81935328e-01 -1.31562456e-01 1.88621700e-01 -6.16812110e-01 6.57991230e-01 -2.48006344e-01 -7.55461872e-01 2.97375977e-01 3.91658634e-01 -5.11958301e-01 -3.13102715e-02 -2.84369677e-01 -7.27650285e-01 -8.02353323e-01 -2.63978213e-01 3.73495728e-01 -6.85208380e-01 -5.00292242e-01 3.35965514e-01 9.38891828e-01 -1.11163139e+00 -4.58100468e-01 1.38066784e-01 6.81545377e-01 -2.17607662e-01 4.95685160e-01 7.60791525e-02 3.87237221e-01 -1.64889500e-01 -5.43919444e-01 6.52746737e-01 -1.24584697e-01 -7.42142022e-01 -5.20455800e-02 5.91371417e-01 -3.23478699e-01 -7.41139948e-01 1.10587573e+00 5.25636971e-01 9.99743268e-02 7.03560233e-01 -1.15647018e+00 -8.29853833e-01 6.44888997e-01 -7.87861645e-01 6.92791566e-02 2.82444477e-01 -2.23593190e-01 1.45830560e+00 -1.00030434e+00 9.20338094e-01 1.36228418e+00 7.73395121e-01 3.91829789e-01 -1.33771551e+00 -3.27410966e-01 -1.15299009e-01 2.31295153e-01 -1.24584723e+00 -4.34624225e-01 8.59991431e-01 -4.46975201e-01 1.23993313e+00 2.44400218e-01 8.00723135e-01 1.07298183e+00 4.17260855e-01 9.44228411e-01 7.81970859e-01 -6.11684740e-01 -2.76249330e-02 -1.83197260e-01 1.94061637e-01 5.91653943e-01 5.15645087e-01 -1.55974701e-01 -6.19357824e-01 3.25097889e-01 1.09165132e+00 1.36328757e-01 -2.77018726e-01 -1.13904440e+00 -1.76715183e+00 9.26096201e-01 5.86470962e-01 7.20242634e-02 -3.88587743e-01 4.50991929e-01 4.56738591e-01 4.72373784e-01 2.10466728e-01 8.42160940e-01 -1.97751075e-01 2.93654632e-02 -6.34952843e-01 4.33389604e-01 6.60683334e-01 9.81398165e-01 6.31228447e-01 4.50005680e-01 3.06453198e-01 7.72930264e-01 3.75751734e-01 5.73774576e-01 6.03833497e-01 -1.21439373e+00 3.43024075e-01 3.39764178e-01 3.00987903e-02 -1.21068084e+00 -8.13515365e-01 -6.65727556e-01 -1.26772761e+00 -6.90868124e-02 2.83244699e-01 -7.22267628e-02 -6.93686962e-01 1.79818594e+00 1.94271449e-02 -2.56133884e-01 -5.55306626e-03 9.27730262e-01 2.91685939e-01 4.48141769e-02 -4.66177195e-01 -1.10541329e-01 1.23579645e+00 -7.35236049e-01 -6.73467577e-01 -3.66259038e-01 8.05461824e-01 -1.00256598e+00 7.18967199e-01 4.90386307e-01 -8.76448035e-01 -1.83345824e-01 -1.41197765e+00 -3.65641326e-01 -2.55797416e-01 2.31166899e-01 1.03788304e+00 5.55223823e-01 -1.18306065e+00 9.02895689e-01 -9.76282001e-01 -5.90595722e-01 1.99382961e-01 4.95652884e-01 -8.92482102e-01 -5.95813990e-02 -1.04571736e+00 1.23740947e+00 1.31902754e-01 2.36389250e-01 -5.53608954e-01 -2.07667455e-01 -1.17677414e+00 -2.44173989e-01 -1.73855215e-01 -9.33610797e-01 1.02432835e+00 -3.12957048e-01 -1.52691305e+00 9.45799351e-01 -1.56446368e-01 -7.48392522e-01 5.44565856e-01 -7.01172173e-01 1.91470817e-01 -1.48702800e-01 -4.35917415e-02 8.19535077e-01 1.14842892e+00 -6.42267227e-01 -7.23806024e-02 -6.23194098e-01 1.85971245e-01 5.35713673e-01 -4.93873954e-01 -6.99627176e-02 -3.23368102e-01 -7.07072020e-01 1.15574980e+00 -1.41757703e+00 -5.54248154e-01 -8.19431022e-02 -5.96294463e-01 -1.06876239e-01 3.94865692e-01 -6.73821270e-01 1.03419650e+00 -2.00011492e+00 8.18898141e-01 1.66615635e-01 3.40764046e-01 -3.36711928e-02 -3.67205893e-03 3.81163061e-01 -8.07906747e-01 -8.14805850e-02 3.84847703e-03 -3.60298514e-01 2.23894700e-01 2.35087708e-01 -5.29916525e-01 1.11713862e+00 1.19931750e-01 7.54538417e-01 -8.08487058e-01 -2.58325543e-02 3.21235806e-01 6.68153048e-01 -8.26274991e-01 -2.56440163e-01 4.11359698e-01 1.24862008e-01 -1.28200911e-02 2.41104379e-01 4.03577387e-01 1.13445241e-02 2.84975529e-01 -5.29341877e-01 -3.59942578e-02 4.47970599e-01 -1.66496670e+00 1.91289091e+00 -2.19542786e-01 9.96730566e-01 -9.69188884e-02 -1.33847940e+00 7.97093332e-01 2.43422031e-01 7.11118817e-01 -3.66477042e-01 3.91357124e-01 1.89017460e-01 1.42785370e-01 -2.61299908e-01 8.59258235e-01 -5.75484969e-02 1.11374453e-01 6.77771151e-01 3.86054754e-01 -2.04439998e-01 4.35202241e-01 2.08130866e-01 7.22465098e-01 4.15899754e-01 7.69853771e-01 -2.50766367e-01 2.65497088e-01 -3.00861925e-01 3.73779684e-01 2.90478081e-01 -2.10334852e-01 8.09773684e-01 6.74604893e-01 -7.87863612e-01 -1.25808656e+00 -7.61851668e-01 -2.67551035e-01 8.61318946e-01 -2.22101375e-01 -4.10930842e-01 -5.43136716e-01 -2.14276239e-01 -4.25956845e-02 3.54456604e-01 -5.78120947e-01 -1.74106106e-01 -8.72157931e-01 -8.18128645e-01 5.39732695e-01 5.88249862e-01 2.04915702e-01 -7.42197037e-01 -5.64014912e-01 1.23520456e-01 -1.65780947e-01 -1.06283939e+00 -2.17001870e-01 4.69305307e-01 -1.18792045e+00 -7.63833046e-01 -1.06216002e+00 -6.16153300e-01 8.09251845e-01 8.33619118e-01 8.12228560e-01 -5.35372436e-01 -8.30088463e-03 3.48229021e-01 -1.43373653e-01 -2.63894439e-01 8.53245426e-03 1.04433417e-01 1.00554693e+00 -1.62353158e-01 4.80153292e-01 -8.30530941e-01 -5.09486318e-01 2.79596210e-01 -6.57334566e-01 -1.54951975e-01 7.72498846e-01 8.47114205e-01 3.83195966e-01 -6.68190360e-01 5.66135813e-03 -5.24638116e-01 7.64737487e-01 -2.94189136e-02 -3.09596539e-01 -1.81955427e-01 -2.84727961e-01 6.15306973e-01 3.87096614e-01 -2.06142813e-01 -2.31109887e-01 1.20008230e-01 -7.40422495e-03 -8.73488784e-01 7.87209049e-02 7.07559645e-01 1.08285725e-01 -1.17582381e-01 1.05951428e+00 1.95155472e-01 2.00114116e-01 -4.80719090e-01 8.81455541e-01 3.10886145e-01 3.66054773e-01 -4.44153309e-01 8.49845529e-01 7.01994538e-01 3.07408303e-01 -1.21550584e+00 -5.61697185e-01 -4.97429252e-01 -1.13301778e+00 4.86765094e-02 7.83528686e-01 -1.01913571e+00 -6.18543863e-01 3.29971462e-01 -1.32134044e+00 3.77643019e-01 -1.88451678e-01 8.21906388e-01 -7.74142742e-01 7.37565398e-01 -4.02804941e-01 -3.97275984e-01 -2.76359886e-01 -1.46097195e+00 8.25309336e-01 -5.12867607e-02 -6.66716516e-01 -7.51960337e-01 2.08916694e-01 1.16341026e-03 1.33294696e-02 -5.00547700e-02 7.96476126e-01 -4.93150026e-01 -2.43343681e-01 -4.40060079e-01 -6.31591603e-02 5.94872117e-01 -1.07463144e-01 4.18510884e-02 -8.17512095e-01 -3.82329851e-01 -4.12451178e-02 -1.20837696e-01 9.89341795e-01 3.97548646e-01 7.72350073e-01 -9.31621343e-02 -9.39670652e-02 9.78330195e-01 1.10609126e+00 -1.02290235e-01 3.95475358e-01 6.49156570e-01 9.61379528e-01 6.11161590e-01 1.76013708e-01 3.38109881e-01 1.63676009e-01 5.96443951e-01 5.79599440e-01 1.63284913e-01 2.43946254e-01 6.42301664e-02 4.29574311e-01 1.33897495e+00 -6.28905475e-01 4.44731444e-01 -8.95854354e-01 5.02275527e-01 -1.63987923e+00 -9.06385183e-01 7.81341791e-02 2.39305782e+00 4.00601119e-01 1.21010840e-01 -1.09380722e-01 5.08204281e-01 4.50749695e-01 5.87468684e-01 -4.53889102e-01 -5.71240187e-01 -3.21105033e-01 3.13729316e-01 8.58363628e-01 2.92602479e-01 -1.25653636e+00 8.73876810e-01 6.97729301e+00 3.62667084e-01 -1.46051252e+00 -3.65339726e-01 8.92361850e-02 -3.38906758e-02 -2.09456116e-01 -8.90028551e-02 -8.17095697e-01 -4.03485954e-01 3.12045574e-01 6.01807274e-02 6.24497890e-01 1.09094739e+00 2.57770680e-02 4.31538373e-01 -1.56876981e+00 1.41895449e+00 2.59590954e-01 -1.34372115e+00 2.68358439e-01 2.71988124e-01 9.17888582e-01 3.34590316e-01 4.83048141e-01 1.66784167e-01 3.58785510e-01 -9.42091942e-01 7.52583325e-01 3.41769159e-02 6.59149289e-01 -6.31412804e-01 5.32775164e-01 6.76617026e-02 -8.39026034e-01 1.05936781e-01 -7.13309646e-01 -2.67435670e-01 1.00132013e-02 4.81893301e-01 -7.99137950e-01 5.26646793e-01 6.47706211e-01 1.20518875e+00 -2.75171578e-01 7.20758200e-01 -4.39382493e-01 8.31779912e-02 -5.21824479e-01 3.57819796e-02 4.48539525e-01 -7.93586612e-01 7.36375034e-01 9.05460656e-01 5.00031412e-01 -3.92155290e-01 -4.61474895e-01 3.84406388e-01 -2.04850703e-01 1.00979701e-01 -1.14265096e+00 -3.39905560e-01 2.01263785e-01 1.27942240e+00 -8.25245500e-01 1.84908099e-02 -3.48441213e-01 7.46176720e-01 4.26069677e-01 5.39346457e-01 -7.62717724e-01 -4.28003430e-01 1.26626050e+00 -2.99484104e-01 4.05478716e-01 -1.06864333e+00 -4.47955549e-01 -1.42373455e+00 4.79121506e-02 -1.00837553e+00 -1.15435990e-02 -5.74356437e-01 -7.56242275e-01 3.34243208e-01 -6.47289073e-03 -1.56380963e+00 -4.38809156e-01 -1.37911546e+00 -3.16540658e-01 6.29809141e-01 -1.12800574e+00 -7.05808818e-01 7.94831812e-02 4.98759478e-01 3.35040152e-01 -3.66570592e-01 7.44539678e-01 2.95316458e-01 -8.53171527e-01 4.94677454e-01 4.94901091e-01 2.98976630e-01 8.66246223e-01 -1.46661091e+00 7.02769637e-01 9.83707547e-01 8.19649696e-01 1.34843600e+00 1.15043950e+00 2.03206744e-02 -1.95118320e+00 -5.60298681e-01 7.14748621e-01 -4.51790154e-01 7.94677615e-01 -1.67796075e-01 -4.44227546e-01 1.23690414e+00 2.33997628e-01 -1.49534643e-01 5.36748409e-01 4.32727456e-01 -6.77996039e-01 -1.43069237e-01 -3.31819683e-01 1.25514781e+00 1.10143447e+00 -5.82148850e-01 -5.47548890e-01 3.05989176e-01 5.22735476e-01 -5.84078550e-01 -7.33599663e-01 2.95727015e-01 8.86640191e-01 -9.70586479e-01 1.30672646e+00 -9.31635499e-01 5.48153520e-01 -2.47495130e-01 -4.67033178e-01 -1.60680187e+00 -3.69829118e-01 -8.50651801e-01 -1.25405684e-01 4.21859592e-01 1.86721012e-01 -7.00181425e-01 8.27747524e-01 3.64222564e-02 -1.22536540e-01 -7.63553798e-01 -9.96928573e-01 -5.39558768e-01 1.60400853e-01 -6.05833828e-01 2.32538491e-01 1.24934554e+00 3.87006067e-02 5.38033903e-01 -5.99058986e-01 -1.78088266e-02 5.31260669e-01 -1.12256058e-01 1.15533507e+00 -9.20605600e-01 -8.78361166e-02 -8.39038432e-01 -8.07395279e-01 -1.46273494e+00 6.43631220e-02 -9.54127371e-01 -2.99896300e-01 -1.31961977e+00 -2.57556140e-01 7.80185908e-02 -2.77034163e-01 1.50749534e-01 1.70597225e-01 2.72683084e-01 3.07730347e-01 2.33621836e-01 -1.93623021e-01 6.11005127e-01 1.35878265e+00 -1.25755131e-01 -2.91331187e-02 -1.66156933e-01 -8.49318624e-01 1.12137413e+00 5.64689755e-01 -1.98198169e-01 -2.63853848e-01 -8.27829361e-01 8.69093835e-01 -3.79957229e-01 -1.14334166e-01 -1.02955246e+00 4.07948457e-02 6.35361299e-02 3.54676723e-01 -4.48666543e-01 4.41287667e-01 -5.63944221e-01 -1.35879681e-01 5.22008717e-01 -4.26126346e-02 5.24082243e-01 -1.90772429e-01 4.84623015e-01 -2.14251503e-01 -1.69283077e-01 6.10176861e-01 -2.09337503e-01 -6.96659386e-01 2.78665811e-01 -2.60033965e-01 -2.09535047e-01 6.13140702e-01 -3.47038835e-01 -9.11957249e-02 -3.94184083e-01 -8.75941217e-01 -1.21490672e-01 4.03799623e-01 4.70043302e-01 5.88321745e-01 -1.42495775e+00 -5.12242734e-01 3.25655282e-01 2.23846901e-02 8.92093331e-02 -8.56888220e-02 1.20629215e+00 -1.19972420e+00 8.02158535e-01 -7.10174859e-01 -8.99369717e-01 -1.09650600e+00 4.74316180e-01 1.59389675e-01 -2.35062107e-01 -4.59727913e-01 9.66008425e-01 1.74889043e-01 -7.08644629e-01 2.48080030e-01 -5.52873909e-01 -4.42204505e-01 4.00870323e-01 8.89017954e-02 5.08722365e-01 1.29102960e-01 -9.91393447e-01 -2.31239140e-01 7.99612939e-01 -1.77485406e-01 -2.06051052e-01 1.38599062e+00 4.34960462e-02 -3.11519682e-01 4.54085022e-01 1.33112800e+00 -1.99363515e-01 -8.14956725e-01 -2.61700183e-01 -1.42593235e-01 -2.01686453e-02 -6.78671226e-02 3.76156092e-01 -9.89241123e-01 1.42598200e+00 2.68196970e-01 8.27552229e-02 5.35616219e-01 -4.94868904e-01 4.19890642e-01 1.24496341e+00 3.66107762e-01 -8.07161510e-01 -9.92819946e-03 1.02907097e+00 1.07436180e+00 -1.07505560e+00 5.90402365e-01 7.32788965e-02 -5.18941164e-01 1.22473919e+00 2.69800782e-01 -5.95457733e-01 6.89266801e-01 -2.35616237e-01 1.89055115e-01 7.43597001e-02 -2.88446754e-01 2.39867810e-02 2.72746861e-01 5.12431741e-01 7.15649605e-01 1.52142599e-01 -4.40562308e-01 -8.53960440e-02 -8.43325615e-01 -6.93799913e-01 7.38505185e-01 8.25619996e-01 -2.89158374e-01 -9.86050487e-01 -5.78167081e-01 2.73607194e-01 -3.07560384e-01 4.00925800e-03 -3.04158747e-01 9.03871477e-01 -2.88842261e-01 2.00673640e-01 2.86745764e-02 -4.30875182e-01 1.25835404e-01 1.77099749e-01 8.47747505e-01 -3.84479195e-01 -9.61580873e-03 -7.46312225e-03 -1.33753240e-01 -7.01204538e-01 -5.18686891e-01 -7.73022592e-01 -8.17713678e-01 -3.38465810e-01 -1.50451511e-01 -1.57027751e-01 1.04736829e+00 1.06955922e+00 1.40871465e-01 2.59314716e-01 4.32522655e-01 -1.36386311e+00 -9.37716603e-01 -9.53258812e-01 -6.17731869e-01 4.31098789e-01 4.59317148e-01 -9.94315028e-01 -2.57492810e-01 -1.40626812e-02]
[8.8942289352417, 2.3297197818756104]
95d89543-161c-4228-8017-bcd132ac939d
tencent-avs-a-holistic-ads-video-dataset-for
2212.04700
null
https://arxiv.org/abs/2212.04700v1
https://arxiv.org/pdf/2212.04700v1.pdf
Tencent AVS: A Holistic Ads Video Dataset for Multi-modal Scene Segmentation
Temporal video segmentation and classification have been advanced greatly by public benchmarks in recent years. However, such research still mainly focuses on human actions, failing to describe videos in a holistic view. In addition, previous research tends to pay much attention to visual information yet ignores the multi-modal nature of videos. To fill this gap, we construct the Tencent `Ads Video Segmentation'~(TAVS) dataset in the ads domain to escalate multi-modal video analysis to a new level. TAVS describes videos from three independent perspectives as `presentation form', `place', and `style', and contains rich multi-modal information such as video, audio, and text. TAVS is organized hierarchically in semantic aspects for comprehensive temporal video segmentation with three levels of categories for multi-label classification, e.g., `place' - `working place' - `office'. Therefore, TAVS is distinguished from previous temporal segmentation datasets due to its multi-modal information, holistic view of categories, and hierarchical granularities. It includes 12,000 videos, 82 classes, 33,900 segments, 121,100 shots, and 168,500 labels. Accompanied with TAVS, we also present a strong multi-modal video segmentation baseline coupled with multi-label class prediction. Extensive experiments are conducted to evaluate our proposed method as well as existing representative methods to reveal key challenges of our dataset TAVS.
['Wei Liu', 'Qinglin Lu', 'Rongwei Quan', 'Jiangfeng Xiong', 'Zhimin Li', 'Jie Jiang']
2022-12-09
null
null
null
null
['scene-segmentation']
['computer-vision']
[ 7.96537474e-02 -4.93078828e-01 -8.20413530e-01 -4.44228321e-01 -1.00139832e+00 -9.47537422e-01 3.69407892e-01 -1.99444238e-02 9.55649931e-03 1.49004027e-01 4.28922713e-01 -2.24354789e-02 9.41829383e-02 -4.23042417e-01 -5.59978724e-01 -6.63316548e-01 -3.18065006e-03 9.12278816e-02 5.62283933e-01 -1.71663873e-02 5.64315617e-02 -2.36579846e-03 -1.65232503e+00 8.52715313e-01 5.44862807e-01 1.50271297e+00 -1.69159442e-01 3.95082802e-01 -2.52624422e-01 1.07484186e+00 -4.21410561e-01 -5.87433279e-01 3.22244056e-02 -4.42672372e-01 -1.06513321e+00 5.78662515e-01 7.38438010e-01 -4.83109742e-01 -4.52007860e-01 9.33373630e-01 1.46197632e-01 2.32453480e-01 6.82133615e-01 -1.60500550e+00 -7.18045533e-01 6.09076917e-01 -8.48667026e-01 3.10486466e-01 6.35929883e-01 4.70092669e-02 1.34567952e+00 -5.46237826e-01 9.40635562e-01 1.37405097e+00 7.07355261e-01 4.07402873e-01 -1.03560078e+00 -7.21557379e-01 6.26899242e-01 5.58281898e-01 -1.34632754e+00 -2.06312895e-01 8.02574873e-01 -8.70597839e-01 5.01190841e-01 5.28280139e-01 7.87456274e-01 1.56402028e+00 -2.70822734e-01 1.38466644e+00 1.07434583e+00 6.93939850e-02 -2.95042247e-02 -1.08312830e-01 4.56681639e-01 5.83795309e-01 -3.09934437e-01 -4.62317079e-01 -4.27087396e-01 1.83020562e-01 3.84788603e-01 2.59544462e-01 -1.98295951e-01 -3.90177190e-01 -1.41631615e+00 6.67075634e-01 7.06029236e-02 3.42239231e-01 -1.51526272e-01 -2.26409659e-02 1.11924052e+00 -4.85767983e-03 5.05805314e-01 -3.84898894e-02 -3.73162359e-01 -4.43268061e-01 -9.18550611e-01 2.02987611e-01 3.09189439e-01 1.42080939e+00 6.62397563e-01 5.83148636e-02 -6.64856136e-01 1.12166667e+00 1.84125945e-01 4.23579425e-01 4.46499079e-01 -1.41583598e+00 7.22134650e-01 5.67079365e-01 -1.07133783e-01 -1.10202074e+00 -3.07629108e-01 5.11407442e-02 -7.28346765e-01 -5.40315211e-01 5.28822690e-02 -4.33789082e-02 -1.10365260e+00 1.62617791e+00 2.64517695e-01 2.30170220e-01 -3.17322254e-01 8.95967364e-01 1.29764140e+00 9.20686543e-01 4.05179650e-01 -4.04820085e-01 1.57909453e+00 -1.33257151e+00 -8.95561814e-01 4.65753675e-02 7.00645089e-01 -5.71802735e-01 1.02022457e+00 5.93336582e-01 -8.96212935e-01 -6.90298557e-01 -5.65546572e-01 -4.11415696e-02 -4.29493994e-01 -2.72895233e-03 6.13287091e-01 5.15845537e-01 -8.94244790e-01 2.76976734e-01 -4.89746630e-01 -4.31696713e-01 6.41206205e-01 -1.31780252e-01 -2.37699762e-01 -3.40225726e-01 -1.13226938e+00 1.23790540e-01 5.06184220e-01 -2.84768462e-01 -1.16641200e+00 -6.43243074e-01 -9.42735434e-01 -2.96806604e-01 7.51559317e-01 -2.26975784e-01 1.36625278e+00 -1.14483690e+00 -1.13262403e+00 9.35057521e-01 -2.92855144e-01 -1.25202805e-01 4.22970444e-01 -6.02849051e-02 -8.28642249e-01 7.83290327e-01 4.47519124e-01 9.07895923e-01 7.60702014e-01 -1.34042943e+00 -1.24834764e+00 -1.89191476e-01 3.29423964e-01 1.44412324e-01 -4.88241255e-01 3.66952896e-01 -1.07350290e+00 -1.11399305e+00 1.25161530e-02 -1.02100968e+00 2.93562204e-01 -5.53654730e-01 -5.70177078e-01 -4.16718125e-01 1.15196383e+00 -7.81928182e-01 1.73665559e+00 -2.43978047e+00 3.39873314e-01 -2.28603676e-01 4.54867154e-01 -2.16027394e-01 -3.12319100e-02 3.44496816e-01 -1.49241731e-01 4.55602944e-01 1.32010654e-01 -3.10989141e-01 3.40315133e-01 5.37478738e-02 -1.79251343e-01 4.40112889e-01 -2.16464832e-01 8.57614338e-01 -8.84524345e-01 -9.76961255e-01 9.95915234e-02 1.30671278e-01 -5.94723225e-01 -3.82684432e-02 -2.82612503e-01 2.86787301e-01 -6.01430357e-01 1.55453503e+00 3.90230358e-01 -2.70662010e-01 -1.56427512e-03 -5.47188044e-01 -1.30911097e-01 -2.09079966e-01 -8.67128909e-01 1.72733676e+00 1.08126782e-01 7.85444379e-01 3.30103152e-02 -1.11205924e+00 2.14289084e-01 5.22889435e-01 1.28646266e+00 -7.26267517e-01 3.83566231e-01 -1.35590971e-01 -5.24189055e-01 -8.17652524e-01 5.30768335e-01 5.40963672e-02 -5.41479707e-01 1.05264880e-01 1.68734461e-01 3.67679477e-01 6.77433550e-01 3.73380095e-01 9.61228430e-01 4.02297944e-01 -2.30042323e-01 1.08309105e-01 2.78918862e-01 3.46968710e-01 9.38120544e-01 3.09819072e-01 -7.66057432e-01 7.88738549e-01 5.97523570e-01 -2.31258944e-01 -8.31564724e-01 -9.94542360e-01 -1.31913885e-01 1.59484327e+00 6.72601283e-01 -8.06842864e-01 -8.76722813e-01 -9.05141771e-01 -7.43611008e-02 3.74468535e-01 -7.29094744e-01 8.21798593e-02 -5.39258659e-01 -4.91796404e-01 5.49813330e-01 6.00063682e-01 6.69857144e-01 -8.95172238e-01 -7.26100802e-02 -3.81520241e-02 -9.24302995e-01 -1.43424892e+00 -7.15004742e-01 -2.83287734e-01 -5.47657728e-01 -1.15594351e+00 -8.06062937e-01 -9.22127783e-01 1.10126376e-01 7.98159182e-01 1.17388403e+00 -4.49152708e-01 -1.17709145e-01 6.17407382e-01 -9.96657073e-01 2.36906022e-01 1.00452639e-01 -1.42298386e-01 -4.02525440e-02 3.32051784e-01 4.12934095e-01 -2.26621374e-01 -6.62100315e-01 8.17267656e-01 -8.58897686e-01 1.34845614e-01 4.06733394e-01 3.37251842e-01 8.04229558e-01 2.43194684e-01 2.18296051e-01 -8.71290505e-01 6.25110418e-02 -8.65462542e-01 1.47194460e-01 3.57197464e-01 -1.11453369e-01 -8.34040582e-01 4.22984362e-01 -6.75074160e-01 -9.99188364e-01 -3.01488787e-01 -4.49241810e-02 -7.38989770e-01 -6.19282722e-01 4.39217955e-01 -3.28750342e-01 1.29983410e-01 1.29142672e-01 3.05780441e-01 -5.43663502e-01 -3.83369714e-01 3.18159521e-01 7.38700867e-01 4.43421572e-01 -7.60533273e-01 5.34193754e-01 4.29790616e-01 -4.81711477e-01 -7.26543248e-01 -9.95972097e-01 -9.78729367e-01 -6.29221857e-01 -7.70654440e-01 1.39753699e+00 -1.20523465e+00 -6.86747432e-01 6.20922387e-01 -6.07333422e-01 -1.37707338e-01 -3.07819359e-02 3.27316999e-01 -6.56064510e-01 7.76241481e-01 -9.58865702e-01 -3.91882300e-01 1.61353871e-01 -1.45859110e+00 1.18023336e+00 -3.61428820e-02 -1.44209862e-01 -7.32115865e-01 -4.31745499e-01 9.99559224e-01 -9.27828923e-02 5.56144774e-01 8.04357111e-01 -5.00333011e-01 -4.19501066e-01 -1.89012825e-01 -3.86800706e-01 4.08392191e-01 -3.04847974e-02 3.24650049e-01 -7.45873511e-01 -2.13227406e-01 -3.20675045e-01 -6.02158606e-01 8.83721650e-01 5.57449281e-01 1.44867730e+00 -2.81367451e-01 -5.43085098e-01 5.24004340e-01 1.27886903e+00 6.47326350e-01 5.02429426e-01 4.30713654e-01 1.21901143e+00 5.21692693e-01 1.17715156e+00 3.86369616e-01 7.20714331e-01 7.82733142e-01 5.23679614e-01 1.39992133e-01 8.68250504e-02 1.44374639e-01 6.21004939e-01 1.11407959e+00 -2.81305224e-01 -3.67615253e-01 -8.34190845e-01 6.11756682e-01 -1.78977871e+00 -1.42432046e+00 -1.95132375e-01 1.55587387e+00 5.64542770e-01 9.25461128e-02 7.80299842e-01 1.19625397e-01 9.71203983e-01 5.58479548e-01 -4.15167391e-01 -7.35306442e-02 -7.95764774e-02 -6.53251648e-01 3.74687612e-01 -2.06494167e-01 -1.76950705e+00 8.80766869e-01 5.78167963e+00 1.36232948e+00 -8.63159418e-01 1.99892506e-01 8.00359726e-01 -1.20192900e-01 -2.20058650e-01 -2.16519564e-01 -8.65148008e-01 9.51995313e-01 8.38152230e-01 3.27970564e-01 1.59376293e-01 9.37549472e-01 1.51648670e-01 1.12860225e-01 -8.97916496e-01 1.35510039e+00 3.75728846e-01 -1.31983531e+00 3.13113660e-01 1.92995649e-02 7.58150280e-01 -2.38661304e-01 1.19030304e-01 7.67441154e-01 -9.38435458e-03 -6.07481301e-01 1.25915492e+00 2.96058774e-01 8.80350173e-01 -7.97865093e-01 3.84452552e-01 4.32958044e-02 -1.73762202e+00 -4.14917469e-01 2.40628392e-01 4.53402847e-01 3.78924727e-01 7.72469789e-02 1.46693289e-01 6.96777880e-01 1.19511080e+00 1.43656766e+00 -7.21191525e-01 8.43682051e-01 4.78553593e-01 6.95266008e-01 1.74895301e-01 2.95966595e-01 7.49898374e-01 -5.09594560e-01 1.64273277e-01 1.45854032e+00 2.12095618e-01 4.05939668e-01 7.35504270e-01 1.31459862e-01 -5.03465757e-02 1.69302374e-01 -1.84086606e-01 -3.60819697e-01 4.38440353e-01 1.14061284e+00 -1.10846293e+00 -6.35407567e-01 -7.19699562e-01 8.63923788e-01 -2.27300525e-01 5.76909482e-01 -1.52710056e+00 -1.86344311e-01 6.32295310e-01 1.21554919e-01 4.27772343e-01 -2.28475668e-02 1.37661800e-01 -1.29446459e+00 -4.62083369e-02 -1.08332837e+00 6.51397049e-01 -7.71647513e-01 -1.05582523e+00 6.57316506e-01 2.03773782e-01 -1.64464784e+00 1.05104215e-01 -4.26956296e-01 5.14225513e-02 -1.73939634e-02 -1.15629363e+00 -1.53601062e+00 -4.11780238e-01 7.44453549e-01 1.23696911e+00 -1.04892775e-01 3.38981479e-01 9.23232555e-01 -1.06854260e+00 5.40049493e-01 1.31924272e-01 3.99852872e-01 8.63648832e-01 -1.19475269e+00 -1.58217207e-01 6.41441822e-01 -7.74739385e-02 9.07067880e-02 4.46306914e-01 -5.15876174e-01 -1.27380311e+00 -1.29751933e+00 3.27238113e-01 -5.03212631e-01 9.73511279e-01 -1.76313102e-01 -6.06338263e-01 9.01561022e-01 2.29796141e-01 -1.70926526e-01 9.41526711e-01 4.74480353e-02 -4.34902787e-01 -1.63502216e-01 -8.09403718e-01 4.93988246e-01 1.25271261e+00 -5.12410760e-01 -3.63435745e-01 5.23520887e-01 9.97831047e-01 -4.43276107e-01 -1.30017412e+00 4.78288591e-01 5.12627244e-01 -9.82149422e-01 1.21382236e+00 -4.48331833e-01 7.77780831e-01 -6.30475506e-02 -4.63796675e-01 -7.23128438e-01 -4.85645682e-01 -3.63805413e-01 -3.14451963e-01 1.77870977e+00 -9.56214443e-02 4.90376614e-02 7.68350601e-01 2.17718244e-01 -7.28977919e-01 -7.50198126e-01 -7.67111123e-01 -7.12203920e-01 -2.76385367e-01 -7.54481256e-01 2.87288874e-01 1.30718350e+00 -1.41830042e-01 2.81064421e-01 -6.81352258e-01 -1.08359411e-01 5.30585408e-01 4.42244381e-01 4.12593842e-01 -1.05995739e+00 1.18200341e-02 -8.01046431e-01 -4.75854963e-01 -1.23994386e+00 1.60255745e-01 -6.53777361e-01 -8.58128071e-02 -1.46420658e+00 6.11279786e-01 -2.28213832e-01 -3.50763619e-01 4.22036946e-01 2.01169237e-01 7.10154653e-01 1.20686322e-01 5.21903753e-01 -1.59349751e+00 2.27523431e-01 1.36814535e+00 -5.49118936e-01 1.73213631e-01 -1.93495631e-01 -5.97972333e-01 8.11019182e-01 4.43766534e-01 -8.42609033e-02 -4.17530239e-01 -4.71529186e-01 2.39190385e-02 1.96632177e-01 1.25859618e-01 -9.44372296e-01 -1.89947203e-01 -5.81338525e-01 1.41399100e-01 -9.58245277e-01 5.03797591e-01 -8.89001071e-01 1.78525940e-01 -2.01807246e-02 -2.85740674e-01 -7.77498186e-02 -8.99211913e-02 7.30876267e-01 -7.04863846e-01 1.54098332e-01 6.24834836e-01 -1.65971264e-01 -1.43418515e+00 7.03541100e-01 -4.59364623e-01 2.88854092e-01 1.46990132e+00 -6.21106267e-01 -3.50187451e-01 -1.10948347e-01 -1.10461175e+00 5.38814664e-01 5.70362926e-01 6.81980729e-01 4.78676289e-01 -1.69384408e+00 -2.77810246e-01 -4.02331561e-01 4.84446973e-01 -1.63251206e-01 9.05220926e-01 1.06549037e+00 -4.85561579e-01 4.30677235e-01 -1.78556934e-01 -8.47095191e-01 -1.37426257e+00 9.72127795e-01 -1.42435744e-01 1.77430347e-01 -7.87901282e-01 5.99333644e-01 3.53129387e-01 1.97573960e-01 5.69639862e-01 -3.33434194e-01 -6.71736360e-01 9.02588844e-01 3.76419693e-01 5.20247996e-01 -3.59656036e-01 -1.34242988e+00 -2.08628565e-01 8.27558517e-01 8.56885966e-03 2.66613752e-01 1.06868112e+00 -5.98681152e-01 2.70539690e-02 7.71361351e-01 1.41936815e+00 -1.98242262e-01 -1.15437114e+00 -1.36417717e-01 -8.85658264e-02 -5.06084621e-01 -1.25995651e-01 -5.98598480e-01 -1.38552368e+00 7.33660758e-01 4.03405160e-01 6.09176874e-01 1.18613994e+00 1.08216450e-01 1.25127983e+00 -1.49168521e-01 3.03473115e-01 -1.45471990e+00 5.05465627e-01 3.05304021e-01 4.90844250e-01 -1.33292830e+00 -1.14315577e-01 -6.96734190e-01 -1.12023032e+00 7.34050989e-01 7.29076028e-01 3.05224895e-01 5.51802337e-01 -2.62681425e-01 -7.55257457e-02 -2.06202313e-01 -4.41586435e-01 -2.64414549e-01 4.98600662e-01 3.43428910e-01 1.59136891e-01 8.91509429e-02 -4.63688225e-02 7.92858243e-01 2.26552099e-01 -2.17760205e-01 3.25492769e-01 9.33554411e-01 -5.11469960e-01 -7.26189733e-01 -2.70439655e-01 5.24582088e-01 -6.75159454e-01 1.76623628e-01 -1.03929512e-01 8.52397263e-01 5.02621114e-01 9.95191395e-01 2.31526382e-02 -8.21151555e-01 1.80622295e-01 -5.25507852e-02 1.19276941e-01 -4.75796252e-01 -3.44713211e-01 5.47296107e-01 1.29449427e-01 -7.86406815e-01 -8.88025582e-01 -1.02168846e+00 -9.76395965e-01 -4.78591114e-01 6.73528239e-02 1.63008556e-01 2.22614110e-01 9.70397294e-01 1.20279156e-01 7.57101715e-01 7.48120666e-01 -9.50795233e-01 1.83787063e-01 -6.89468324e-01 -6.82605326e-01 8.84370387e-01 9.20688733e-02 -9.19927418e-01 -3.07884306e-01 4.60483730e-01]
[9.539041519165039, 0.485289990901947]
9c5d7e27-8ccd-4cc1-b9aa-8e7b74ea56da
represent-compare-and-learn-a-similarity
2203.08354
null
https://arxiv.org/abs/2203.08354v1
https://arxiv.org/pdf/2203.08354v1.pdf
Represent, Compare, and Learn: A Similarity-Aware Framework for Class-Agnostic Counting
Class-agnostic counting (CAC) aims to count all instances in a query image given few exemplars. A standard pipeline is to extract visual features from exemplars and match them with query images to infer object counts. Two essential components in this pipeline are feature representation and similarity metric. Existing methods either adopt a pretrained network to represent features or learn a new one, while applying a naive similarity metric with fixed inner product. We find this paradigm leads to noisy similarity matching and hence harms counting performance. In this work, we propose a similarity-aware CAC framework that jointly learns representation and similarity metric. We first instantiate our framework with a naive baseline called Bilinear Matching Network (BMNet), whose key component is a learnable bilinear similarity metric. To further embody the core of our framework, we extend BMNet to BMNet+ that models similarity from three aspects: 1) representing the instances via their self-similarity to enhance feature robustness against intra-class variations; 2) comparing the similarity dynamically to focus on the key patterns of each exemplar; 3) learning from a supervision signal to impose explicit constraints on matching results. Extensive experiments on a recent CAC dataset FSC147 show that our models significantly outperform state-of-the-art CAC approaches. In addition, we also validate the cross-dataset generality of BMNet and BMNet+ on a car counting dataset CARPK. Code is at tiny.one/BMNet
['Zhiguo Cao', 'Chengxin Liu', 'Chen Feng', 'Hao Lu', 'Min Shi']
2022-03-16
null
http://openaccess.thecvf.com//content/CVPR2022/html/Shi_Represent_Compare_and_Learn_A_Similarity-Aware_Framework_for_Class-Agnostic_Counting_CVPR_2022_paper.html
http://openaccess.thecvf.com//content/CVPR2022/papers/Shi_Represent_Compare_and_Learn_A_Similarity-Aware_Framework_for_Class-Agnostic_Counting_CVPR_2022_paper.pdf
cvpr-2022-1
['object-counting']
['computer-vision']
[ 1.48543835e-01 -7.29114294e-01 -1.22975931e-01 -5.63478231e-01 -8.13908756e-01 -5.92448711e-01 8.11115921e-01 3.80237222e-01 -7.85272002e-01 1.63453788e-01 7.34512657e-02 2.91267280e-02 9.37263295e-02 -9.73262489e-01 -7.49995708e-01 -3.24850023e-01 -1.13537265e-02 4.70422059e-01 6.36814177e-01 7.28620514e-02 6.22868478e-01 6.55075133e-01 -1.77140379e+00 4.53165263e-01 6.07014239e-01 1.18591785e+00 6.80694655e-02 7.14216173e-01 -3.23811889e-01 1.06634283e+00 -6.32957458e-01 -5.37196398e-01 3.83451730e-01 -2.43906662e-01 -6.31520092e-01 -1.56921774e-01 8.96776974e-01 -3.42797101e-01 -4.72722113e-01 1.01092577e+00 2.68132687e-01 1.14942804e-01 7.79566705e-01 -1.43257284e+00 -7.86383688e-01 3.66798878e-01 -8.60431135e-01 6.04081929e-01 6.49462342e-02 3.98824006e-01 1.03762567e+00 -1.00524139e+00 2.58764625e-01 1.22198343e+00 8.22319806e-01 4.16142404e-01 -1.23812771e+00 -8.62203300e-01 1.03140786e-01 3.13717932e-01 -1.62954330e+00 -2.17556804e-01 4.97652024e-01 -3.30497026e-01 8.79223049e-01 2.34363154e-01 6.66114926e-01 7.55590856e-01 -3.02218527e-01 8.47881973e-01 9.94391084e-01 -3.09396029e-01 4.73805904e-01 -2.20464885e-01 3.29813749e-01 9.08377528e-01 4.28620815e-01 1.19817123e-01 -4.63285685e-01 -2.32560933e-01 7.04209387e-01 2.97252357e-01 2.19298065e-01 -4.58057314e-01 -1.19368887e+00 9.38183963e-01 8.84635031e-01 2.81431437e-01 -1.43681280e-03 6.96804702e-01 5.01920581e-01 1.12573355e-01 7.51497969e-02 3.93948585e-01 -1.13090672e-01 1.24630697e-01 -1.01890266e+00 4.27075058e-01 7.23490477e-01 1.12016642e+00 1.11087000e+00 -3.99864554e-01 -8.19326222e-01 9.28130269e-01 -2.55736820e-02 4.76277471e-01 4.52337116e-01 -1.09011352e+00 3.92900854e-01 7.78838873e-01 -1.77008823e-01 -1.12384200e+00 -1.95097312e-01 -1.83113307e-01 -5.46568215e-01 1.19829684e-01 5.11633694e-01 3.68240416e-01 -1.09204078e+00 1.65764880e+00 1.90354213e-01 6.08534575e-01 -3.86996835e-01 9.27803576e-01 7.89574802e-01 2.66036034e-01 2.73672193e-01 4.29969579e-01 1.42027295e+00 -8.95068288e-01 -7.97994584e-02 -4.25662130e-01 6.16465032e-01 -6.28317773e-01 1.31483901e+00 4.62815119e-03 -9.09822524e-01 -5.33842802e-01 -1.35406768e+00 -2.51658708e-01 -6.24154627e-01 2.33086601e-01 7.39827394e-01 6.55982733e-01 -7.33142138e-01 7.16971636e-01 -7.26557374e-01 -2.06553385e-01 6.90021992e-01 2.56715685e-01 -1.60755903e-01 -2.73195326e-01 -7.36535132e-01 7.94384003e-01 5.38843989e-01 -4.48010117e-01 -7.71570385e-01 -7.76375771e-01 -9.65760529e-01 2.49320850e-01 2.45065749e-01 -7.18765497e-01 1.20815647e+00 -6.78835094e-01 -8.52087617e-01 1.03760064e+00 -1.33793235e-01 -5.59570491e-01 4.38459277e-01 4.28447388e-02 -1.21541515e-01 2.35199600e-01 4.90486681e-01 1.02988291e+00 6.52916908e-01 -1.26069832e+00 -1.01781917e+00 -2.69864440e-01 6.70872107e-02 -2.53109336e-01 -4.12446529e-01 8.00641440e-03 -7.56501734e-01 -6.71519458e-01 -7.93963671e-02 -6.86733603e-01 -7.06064403e-02 3.85264695e-01 -1.48539588e-01 -3.92847717e-01 7.09959447e-01 -6.68200031e-02 1.12787580e+00 -2.04375601e+00 -3.12971383e-01 2.95392543e-01 3.82484853e-01 2.29181394e-01 -4.05563027e-01 1.32757843e-01 1.59034297e-01 -1.42176107e-01 -3.69809300e-01 -6.19439483e-01 4.84744646e-02 3.61129314e-01 -1.43885881e-01 5.51460743e-01 6.41463876e-01 1.05929470e+00 -1.14470410e+00 -8.44252944e-01 3.72614652e-01 3.94055307e-01 -6.55116260e-01 1.18338034e-01 -6.40795678e-02 2.72048218e-03 -1.07810825e-01 9.07339394e-01 8.50492597e-01 -4.81796682e-01 -2.07678676e-01 -4.12694454e-01 -1.98488459e-02 -1.71377435e-01 -1.28026056e+00 1.56590796e+00 -4.80249196e-01 3.25937271e-01 -4.93398249e-01 -9.00847554e-01 9.59935546e-01 -4.64491069e-01 2.45108113e-01 -8.98151636e-01 3.56377721e-01 2.28148624e-01 -1.62342161e-01 -7.28037357e-02 5.11277139e-01 2.26604030e-01 -2.05383569e-01 3.83378953e-01 2.32376277e-01 -2.41767436e-01 7.77580261e-01 4.32204157e-01 1.31122041e+00 -3.33948344e-01 4.19600427e-01 -9.45514217e-02 6.68780386e-01 -3.09336428e-02 6.14243329e-01 1.08784389e+00 -4.15821284e-01 9.84546959e-01 4.05409575e-01 -5.81782401e-01 -1.07335877e+00 -1.24169147e+00 -2.77329266e-01 1.19362843e+00 3.76976311e-01 -4.58129257e-01 -6.55286789e-01 -8.00002754e-01 3.36702377e-01 5.87374687e-01 -9.00421679e-01 -1.08959012e-01 -7.74078608e-01 -5.95937908e-01 6.25136614e-01 1.04464710e+00 7.51655817e-01 -8.60609353e-01 -6.71023071e-01 2.50870958e-02 6.33723512e-02 -1.30565500e+00 -9.15241539e-01 1.48301974e-01 -5.37367463e-01 -1.26025009e+00 -4.79090273e-01 -6.07920825e-01 5.92534781e-01 5.59675157e-01 1.37045419e+00 4.74108607e-01 -7.17305303e-01 4.98210907e-01 -2.52825409e-01 -4.15946156e-01 -2.79700696e-01 -1.14958882e-02 -2.49736696e-01 -4.78321761e-02 8.83473694e-01 -4.38443035e-01 -9.74277258e-01 3.45506907e-01 -8.79378617e-01 -3.18233430e-01 5.88860393e-01 7.37016141e-01 6.94457471e-01 -5.16040921e-01 2.79761046e-01 -6.96093440e-01 4.65608776e-01 -3.15498471e-01 -6.74795687e-01 2.75334060e-01 -4.48011786e-01 7.50370473e-02 5.25317311e-01 -5.48004448e-01 -5.52204967e-01 3.66159022e-01 2.57501751e-01 -5.93326867e-01 1.43166140e-01 -2.66399756e-02 1.14050798e-01 -1.84738100e-01 8.03645730e-01 7.54660070e-02 -2.09078655e-01 -1.17040917e-01 6.58644557e-01 3.83383274e-01 9.66630578e-01 -8.34058881e-01 8.79226446e-01 8.52672458e-01 -1.00214193e-02 -3.00499499e-01 -1.09784544e+00 -9.43504095e-01 -6.81867063e-01 -9.26255435e-02 7.75164247e-01 -9.18326259e-01 -8.87528241e-01 1.93606138e-01 -1.08390594e+00 -1.36468470e-01 -4.39805955e-01 4.10767347e-01 -5.73473752e-01 4.25381750e-01 -6.70443177e-01 -5.92413604e-01 -4.40333843e-01 -1.01006889e+00 1.33496845e+00 4.21159327e-01 -1.51492268e-01 -6.87096953e-01 1.66944742e-01 2.03543514e-01 3.83020669e-01 2.51152426e-01 6.98962152e-01 -5.59562564e-01 -6.92507684e-01 -2.79260486e-01 -9.08314943e-01 3.66170853e-01 -1.31407445e-02 1.31863460e-01 -1.05791056e+00 -3.04773808e-01 -4.62511033e-01 -5.03142178e-01 1.27193809e+00 1.53492674e-01 1.65577555e+00 1.33460075e-01 -3.34415436e-01 8.35329592e-01 1.57765913e+00 -2.82470763e-01 5.41785479e-01 3.62999618e-01 6.59782231e-01 2.01657668e-01 5.45645297e-01 3.93223017e-01 3.96699339e-01 5.18406868e-01 5.56737483e-01 -1.18572064e-01 -2.93092459e-01 -4.05980855e-01 -1.68691397e-01 5.55790663e-01 9.19070188e-03 1.28235787e-01 -7.77136564e-01 6.65809631e-01 -1.83968222e+00 -1.10996842e+00 1.19195674e-02 2.17285991e+00 6.67018592e-01 1.30114153e-01 3.46344173e-01 1.79791406e-01 8.58716726e-01 -2.53011137e-02 -5.17168581e-01 -2.23497629e-01 -9.82900783e-02 5.75135350e-01 8.01628768e-01 9.77199078e-02 -1.21310103e+00 8.27621222e-01 5.72208357e+00 9.49820280e-01 -8.09630036e-01 1.12699255e-01 6.55577242e-01 -1.81268036e-01 1.00911513e-01 2.20843460e-02 -8.34066272e-01 6.04506552e-01 4.53195363e-01 -1.27366245e-01 4.16965097e-01 9.74035203e-01 -3.93229127e-01 -1.29755795e-01 -1.35554683e+00 1.23238873e+00 2.46926039e-01 -1.46422696e+00 1.85580522e-01 -2.41743580e-01 5.60817361e-01 1.34385854e-01 -4.88900989e-02 6.21607304e-01 6.21489644e-01 -1.09072411e+00 8.33194375e-01 4.24209267e-01 8.57554436e-01 -7.95519531e-01 6.03302896e-01 9.62656960e-02 -1.63525772e+00 -2.50049949e-01 -7.09117711e-01 4.76148538e-02 -3.90565321e-02 4.78576660e-01 -6.35136724e-01 1.47049502e-01 8.20617557e-01 6.47448361e-01 -1.06620026e+00 1.38194382e+00 -1.19298078e-01 2.87162930e-01 -3.29352707e-01 -2.19873190e-02 3.15728158e-01 6.54068887e-02 -4.49057519e-02 1.70779252e+00 2.40587473e-01 -9.64568257e-02 2.68055916e-01 1.17527926e+00 -2.45139420e-01 -3.63901332e-02 -3.18005592e-01 3.43198270e-01 9.57523704e-01 1.38285053e+00 -1.02779460e+00 -5.05638123e-01 -5.64114928e-01 8.78278136e-01 7.47085690e-01 -8.38158876e-02 -9.84570146e-01 -3.95775378e-01 6.15242898e-01 3.90427536e-03 5.95418453e-01 -2.53033657e-02 -3.22570622e-01 -1.05266964e+00 4.13872907e-03 -6.56015933e-01 8.43770087e-01 -4.14422274e-01 -1.74013555e+00 1.55140519e-01 1.10810377e-01 -1.23076248e+00 -2.04973053e-02 -7.19184399e-01 -8.22271407e-01 6.05038285e-01 -1.52689552e+00 -1.32816005e+00 -7.80255079e-01 6.17878377e-01 4.01139289e-01 -2.43813768e-02 5.21028519e-01 3.68098855e-01 -3.87429565e-01 9.42448974e-01 -2.31487677e-01 4.93534267e-01 6.97124600e-01 -1.38153243e+00 8.85570407e-01 6.70309246e-01 4.62975353e-01 7.23100543e-01 1.74358308e-01 -4.11150545e-01 -1.05566359e+00 -1.29581690e+00 5.45937538e-01 -7.05539346e-01 6.91725373e-01 -5.92272758e-01 -9.67974365e-01 4.26480591e-01 -3.63809973e-01 8.18182647e-01 4.19080138e-01 -2.08148792e-01 -1.01958966e+00 -2.30759606e-01 -1.27141619e+00 4.82710600e-01 1.34133446e+00 -5.43641210e-01 -4.55613494e-01 1.41136542e-01 5.35776913e-01 -2.61413962e-01 -6.76612437e-01 4.05096769e-01 6.20528102e-01 -1.23014390e+00 1.25644577e+00 -3.25673431e-01 3.39466214e-01 -6.40938044e-01 -3.65916729e-01 -9.48260546e-01 -5.16161025e-01 9.92102642e-03 -1.30218506e-01 1.15660501e+00 6.92480803e-03 -4.10261989e-01 9.04324114e-01 3.84077132e-01 8.68181512e-02 -6.26613855e-01 -9.84310508e-01 -1.02664220e+00 2.61679679e-01 -5.12760460e-01 9.54893887e-01 8.43328297e-01 -3.14607471e-01 1.21339858e-01 5.76120950e-02 1.46552622e-01 7.32756853e-01 3.73938456e-02 1.02258611e+00 -1.09545648e+00 -1.87325194e-01 -5.40372849e-01 -9.26203132e-01 -8.65938365e-01 -5.67529770e-03 -1.02517509e+00 -1.21325165e-01 -1.13916445e+00 8.54329824e-01 -6.53809071e-01 -3.38632762e-01 3.10823381e-01 -3.79429966e-01 5.60105801e-01 5.27896523e-01 2.68720120e-01 -8.73897076e-01 2.83987612e-01 8.71135533e-01 -3.59233171e-01 1.25016063e-01 -2.40998104e-01 -5.08972824e-01 6.92979932e-01 7.35917985e-01 -6.15160048e-01 -2.42561057e-01 -4.48769808e-01 -2.16513295e-02 -4.19320792e-01 8.37539971e-01 -1.37577033e+00 5.63706219e-01 -1.13798060e-01 6.74220026e-01 -7.51722217e-01 1.78518087e-01 -6.59494936e-01 -2.72646844e-01 4.96760517e-01 -2.63898462e-01 1.15188628e-01 1.03532992e-01 6.68257475e-01 -1.39457479e-01 -5.26787937e-01 1.07817686e+00 -2.57631451e-01 -7.90725589e-01 3.99117649e-01 1.94518462e-01 4.11390156e-01 9.66856003e-01 -4.54013169e-01 -6.22086644e-01 1.19907940e-02 -1.61402747e-01 3.08445930e-01 6.39661014e-01 3.30840290e-01 5.07687390e-01 -1.53384614e+00 -6.69365525e-01 1.78415515e-02 7.19692647e-01 -3.70672867e-02 -1.09745100e-01 3.63906533e-01 -5.54108322e-01 1.28565058e-01 -1.09623745e-01 -9.31102514e-01 -1.10578012e+00 6.44320428e-01 3.83133113e-01 -2.19611526e-01 -3.71543437e-01 8.94637823e-01 3.50674987e-01 -4.19344217e-01 2.34099448e-01 -4.95900720e-01 -8.31996743e-03 -1.42967552e-01 7.97012985e-01 5.43162763e-01 1.39283761e-01 -5.44683158e-01 -4.73205000e-01 7.24513948e-01 -2.88184106e-01 1.13039702e-01 1.10040641e+00 2.18017116e-01 2.64351279e-01 2.89273322e-01 1.48321998e+00 -2.56403029e-01 -1.31123340e+00 -5.11345029e-01 1.99516088e-01 -8.50887001e-01 -8.04767311e-02 -5.19210160e-01 -1.18204272e+00 7.11376190e-01 7.55948186e-01 -2.50826091e-01 9.09401596e-01 1.78363472e-01 6.93263292e-01 5.01819849e-01 3.97854328e-01 -1.16130376e+00 5.75658739e-01 3.99050832e-01 5.99319041e-01 -1.46088207e+00 1.43302426e-01 -3.10351074e-01 -4.84252989e-01 1.08101654e+00 8.58184755e-01 -4.77374107e-01 3.48264605e-01 4.55385506e-01 -9.24689546e-02 -4.60610569e-01 -5.28327942e-01 -5.50675631e-01 3.31309021e-01 6.48613930e-01 3.72659802e-01 -8.91706124e-02 -1.72467411e-01 2.49746427e-01 -1.23430654e-01 6.64763674e-02 2.20254883e-01 9.99974072e-01 -4.72442210e-01 -7.02465475e-01 -4.05008912e-01 6.40654445e-01 -1.33582443e-01 -4.23723608e-02 -3.45090061e-01 9.34104025e-01 3.44033480e-01 7.42807150e-01 4.66446340e-01 -2.26215214e-01 5.99429071e-01 -3.59732300e-01 5.05044460e-01 -6.38292789e-01 -6.58517957e-01 -5.88370800e-01 -3.97292972e-01 -7.46565461e-01 -3.43762279e-01 -6.85034454e-01 -1.22889185e+00 -6.19724631e-01 -3.58410180e-01 -3.28334749e-01 5.60502946e-01 6.25530005e-01 7.20315352e-02 3.31734478e-01 5.95893979e-01 -7.79534400e-01 -7.89872229e-01 -7.31340051e-01 -4.04881209e-01 7.97490478e-01 -2.49797814e-02 -7.41390169e-01 -4.55650628e-01 -1.73224360e-01]
[8.99155044555664, 0.5052405595779419]
c376f60b-b308-4dbf-96c4-ad0137425fd0
cross-lingual-language-model-pretraining
1901.07291
null
http://arxiv.org/abs/1901.07291v1
http://arxiv.org/pdf/1901.07291v1.pdf
Cross-lingual Language Model Pretraining
Recent studies have demonstrated the efficiency of generative pretraining for English natural language understanding. In this work, we extend this approach to multiple languages and show the effectiveness of cross-lingual pretraining. We propose two methods to learn cross-lingual language models (XLMs): one unsupervised that only relies on monolingual data, and one supervised that leverages parallel data with a new cross-lingual language model objective. We obtain state-of-the-art results on cross-lingual classification, unsupervised and supervised machine translation. On XNLI, our approach pushes the state of the art by an absolute gain of 4.9% accuracy. On unsupervised machine translation, we obtain 34.3 BLEU on WMT'16 German-English, improving the previous state of the art by more than 9 BLEU. On supervised machine translation, we obtain a new state of the art of 38.5 BLEU on WMT'16 Romanian-English, outperforming the previous best approach by more than 4 BLEU. Our code and pretrained models will be made publicly available.
['Guillaume Lample', 'Alexis Conneau']
2019-01-22
cross-lingual-language-model-pretraining-1
http://papers.nips.cc/paper/8928-cross-lingual-language-model-pretraining
http://papers.nips.cc/paper/8928-cross-lingual-language-model-pretraining.pdf
neurips-2019-12
['unsupervised-machine-translation']
['natural-language-processing']
[ 3.76609527e-02 6.48458228e-02 -6.11039698e-01 -4.50400054e-01 -1.66124749e+00 -7.99472809e-01 8.24287474e-01 -7.04176202e-02 -6.06782913e-01 1.01690638e+00 1.76304340e-01 -8.33716691e-01 3.21290374e-01 -4.90068674e-01 -1.07333958e+00 -3.33654165e-01 2.79555976e-01 9.08767164e-01 -2.70413905e-01 -3.40069681e-01 -1.00691229e-01 -2.30704211e-02 -9.81963277e-01 5.89158893e-01 1.05632007e+00 4.59380776e-01 1.14998430e-01 5.06647229e-01 -3.19401145e-01 3.03196371e-01 -3.64218056e-01 -6.06017113e-01 1.96261212e-01 -6.28401458e-01 -1.02315068e+00 -2.36316219e-01 4.69221324e-01 3.46724354e-02 1.49724573e-01 8.97768438e-01 3.71472895e-01 -3.64678770e-01 6.18328571e-01 -7.81378686e-01 -8.49245965e-01 1.11643362e+00 -4.02463257e-01 -1.05944812e-01 2.40019783e-01 -2.21353937e-02 1.11463666e+00 -9.94088054e-01 8.10999036e-01 1.18078351e+00 6.63193643e-01 5.84176898e-01 -1.34651589e+00 -8.43330622e-01 -1.02864541e-01 -1.85094997e-01 -1.42831409e+00 -5.38773894e-01 3.16711932e-01 -3.26817334e-01 1.46365738e+00 -8.32593963e-02 1.96519166e-01 1.13899779e+00 2.92140126e-01 8.26483071e-01 1.56329691e+00 -1.00799620e+00 -1.91747293e-01 4.34864640e-01 -7.31493011e-02 6.52507246e-01 2.13701934e-01 1.91931590e-01 -4.78409439e-01 1.65135100e-01 5.26724637e-01 -4.09334332e-01 1.38390198e-01 -8.34617838e-02 -1.44877660e+00 9.73247290e-01 1.40591323e-01 6.64840102e-01 -9.50522125e-02 8.43487959e-03 4.49208796e-01 6.82708323e-01 8.18596244e-01 5.25336444e-01 -8.36672127e-01 -3.41057420e-01 -1.03941572e+00 -3.27196598e-01 7.93154895e-01 1.10032666e+00 9.48533118e-01 2.80219540e-02 2.84310967e-01 1.03271770e+00 1.81757852e-01 9.84070659e-01 6.31458282e-01 -4.91997421e-01 7.61649251e-01 5.22920191e-01 -1.86938047e-01 -1.33564085e-01 -1.15432054e-01 -6.85496926e-01 -6.56078160e-01 -2.60682166e-01 3.06951195e-01 -3.77885729e-01 -1.11084712e+00 1.82384622e+00 -3.02972972e-01 -3.38180304e-01 6.26855135e-01 3.19564492e-01 5.73268950e-01 9.38376367e-01 5.18265963e-02 -3.27747554e-01 1.22413921e+00 -1.19888151e+00 -5.38868845e-01 -4.68357027e-01 1.22551548e+00 -1.22160411e+00 1.31436217e+00 3.13456118e-01 -1.02509105e+00 -6.24679148e-01 -8.72361779e-01 -4.23235931e-02 -5.13990939e-01 5.50801694e-01 5.68635702e-01 8.29844952e-01 -1.21602368e+00 3.81508112e-01 -9.19796765e-01 -7.88019836e-01 -3.47896293e-02 4.96739805e-01 -5.19212544e-01 -2.69068211e-01 -1.25638151e+00 1.06673694e+00 6.44770384e-01 -3.77489984e-01 -8.14121008e-01 -5.03195882e-01 -8.19963396e-01 -2.13851243e-01 1.31915852e-01 -4.99215752e-01 1.31426156e+00 -9.90257442e-01 -1.59032822e+00 1.09810138e+00 -2.69914299e-01 -5.47872782e-01 5.06185293e-01 -3.82260203e-01 -6.04540825e-01 -2.70997226e-01 2.76775777e-01 8.58903170e-01 2.10032985e-01 -1.17976081e+00 -5.90263546e-01 -1.21743120e-01 -9.18392465e-02 1.06010780e-01 -3.50626796e-01 1.70985982e-01 -5.32401502e-01 -5.72524011e-01 -1.26421288e-01 -1.24256945e+00 -6.24350980e-02 -8.68124485e-01 -2.12853625e-01 -3.13886076e-01 3.78082901e-01 -8.10027599e-01 1.16810954e+00 -1.67637742e+00 1.29066795e-01 -1.30787626e-01 -3.40374082e-01 2.55919576e-01 -4.06110078e-01 7.07817018e-01 -1.01809073e-02 3.10156167e-01 -2.62119949e-01 -6.39338493e-01 2.57067010e-02 5.35918772e-01 -2.05679566e-01 1.33506447e-01 2.51279205e-01 1.20460272e+00 -8.78922105e-01 -2.42868289e-01 -5.69140352e-02 3.62100214e-01 -5.15815318e-01 9.23046917e-02 -9.70040560e-02 4.29874927e-01 -9.28352773e-02 4.92873609e-01 3.89373243e-01 -1.22241907e-01 4.91020977e-01 5.01663312e-02 -3.14618140e-01 8.26010227e-01 -5.90640366e-01 2.21397781e+00 -8.87758672e-01 5.08437216e-01 -3.93964499e-01 -8.00069213e-01 1.00279295e+00 4.53227490e-01 1.32555157e-01 -6.37367666e-01 3.48440148e-02 8.84041190e-01 2.01702505e-01 -4.52728607e-02 3.81733030e-01 -3.24728072e-01 -2.96132863e-01 6.95326030e-01 5.11469245e-01 -1.25704318e-01 3.48826170e-01 -2.35530119e-02 5.64510405e-01 4.63513404e-01 2.63674051e-01 -6.64098561e-01 4.34484631e-01 8.50264281e-02 2.54594445e-01 5.99710345e-01 2.93623030e-01 2.36729965e-01 1.95150882e-01 -3.18720877e-01 -1.23974252e+00 -1.03445220e+00 -8.44611079e-02 1.33258736e+00 -3.84519845e-01 -6.70769453e-01 -9.63650286e-01 -9.01265621e-01 -3.18733811e-01 9.63853478e-01 -3.97492766e-01 4.37099263e-02 -8.07879508e-01 -9.24178183e-01 7.95056522e-01 5.90279400e-01 4.45225358e-01 -8.44858825e-01 1.81067809e-01 7.65336975e-02 -5.44863224e-01 -1.33648872e+00 -4.88921672e-01 2.93177426e-01 -1.01001215e+00 -4.25651848e-01 -6.84677720e-01 -1.02866638e+00 5.18967927e-01 -1.72362119e-01 1.45502436e+00 -3.15614343e-01 3.01507711e-01 3.37492116e-02 -3.88427734e-01 -2.89493382e-01 -9.77901042e-01 8.67713094e-01 4.22969878e-01 -3.06470305e-01 5.19163311e-01 -3.56508523e-01 -4.70742695e-02 2.54674464e-01 -7.32335687e-01 2.53212154e-01 9.69841361e-01 8.96563292e-01 6.27041817e-01 -5.10564268e-01 5.67512810e-01 -1.04204202e+00 3.93902361e-01 -2.31744230e-01 -4.12025690e-01 4.43727672e-01 -1.11241972e+00 5.40767491e-01 7.19483435e-01 -3.24828267e-01 -9.68762934e-01 -8.32343698e-02 -3.33064288e-01 -1.39549837e-01 -2.71069705e-01 6.72737896e-01 -1.61104742e-02 1.18466981e-01 6.11129701e-01 3.17925334e-01 -4.28227365e-01 -7.19339252e-01 6.04562283e-01 8.29227984e-01 3.99214476e-01 -6.87808931e-01 6.98204935e-01 8.64462648e-03 -5.11892200e-01 -5.55299044e-01 -8.89828503e-01 -3.58614653e-01 -9.82925773e-01 2.72038400e-01 9.44269478e-01 -1.14194191e+00 5.10613173e-02 2.64344335e-01 -1.10584354e+00 -5.29054284e-01 6.22491026e-03 7.62610078e-01 -5.68929613e-01 1.41626865e-01 -8.79535198e-01 -4.28000957e-01 -5.74183524e-01 -1.19140065e+00 1.19736636e+00 -3.11283320e-01 -3.22403640e-01 -1.39424491e+00 3.94282699e-01 4.83361989e-01 3.57821971e-01 -2.57547259e-01 9.30740833e-01 -7.81945825e-01 -3.69259477e-01 -3.40369120e-02 -1.02915637e-01 4.59630549e-01 2.67864913e-01 -4.27153647e-01 -7.81070530e-01 -6.69063807e-01 -1.91939279e-01 -4.99291331e-01 7.93987691e-01 9.76546779e-02 3.05181473e-01 -1.83882639e-01 -3.50454241e-01 5.71213305e-01 1.52137339e+00 1.63998038e-01 5.49347579e-01 3.78251880e-01 6.45742893e-01 3.76920283e-01 4.62555021e-01 -3.50581348e-01 5.16762137e-01 6.83766186e-01 -1.49169445e-01 -3.62518519e-01 -2.75528222e-01 -4.95261073e-01 7.99201190e-01 1.69134736e+00 -1.84945419e-01 -2.59702623e-01 -1.20231342e+00 7.07813203e-01 -1.61629236e+00 -5.40886998e-01 -9.13127139e-03 2.19977045e+00 1.14210927e+00 2.60173082e-01 -6.72553554e-02 -3.28392744e-01 4.60293084e-01 -1.49090260e-01 -6.23566285e-02 -8.85149479e-01 -7.07540736e-02 5.38924873e-01 6.48517013e-01 9.35004950e-01 -9.72838104e-01 1.67822742e+00 6.61477232e+00 9.37044561e-01 -1.33218837e+00 3.33106995e-01 5.43783844e-01 2.03901216e-01 -4.59896058e-01 1.90185860e-01 -1.14975286e+00 2.00026438e-01 1.52536190e+00 -1.06002735e-02 4.71646637e-01 5.95207036e-01 1.87388062e-02 6.65098652e-02 -1.22815740e+00 9.00779188e-01 2.68161207e-01 -1.07904029e+00 2.60582685e-01 1.64622247e-01 1.25123990e+00 6.41821623e-01 -4.76772152e-02 6.94303274e-01 5.02697945e-01 -1.05204904e+00 3.50691736e-01 1.44505844e-01 1.20713568e+00 -8.29176664e-01 8.51869166e-01 5.23957968e-01 -1.07725167e+00 5.47867894e-01 -2.02307835e-01 -6.15109652e-02 2.04763636e-01 2.86380231e-01 -9.89558876e-01 8.67880225e-01 4.85777229e-01 8.65959108e-01 -5.18644273e-01 2.33490646e-01 -4.68531191e-01 9.88499224e-01 -3.72340739e-01 8.71384591e-02 6.50511682e-01 -2.19766513e-01 2.07088724e-01 1.68881810e+00 4.65530068e-01 -4.34266537e-01 4.14916813e-01 4.81602818e-01 -3.95348191e-01 6.38896227e-01 -6.72593474e-01 -3.18281949e-01 3.19289677e-02 8.73992801e-01 -4.48904037e-01 -6.55570745e-01 -5.12779355e-01 1.22879398e+00 4.09832060e-01 2.16343343e-01 -6.74665093e-01 -2.99204528e-01 3.36374253e-01 -7.50642270e-02 2.27276519e-01 -4.44938362e-01 -2.45969206e-01 -1.43160975e+00 -2.47256923e-02 -1.10886264e+00 8.46468657e-02 -5.06328762e-01 -1.25009084e+00 1.07158673e+00 1.86139438e-02 -1.20956814e+00 -7.64161825e-01 -7.13899910e-01 -1.49339736e-01 1.08813834e+00 -1.48392475e+00 -1.63793504e+00 3.65483910e-01 2.42863417e-01 7.99107671e-01 -4.69272017e-01 1.34452021e+00 4.61325049e-01 -2.73050785e-01 9.06372547e-01 4.66945440e-01 2.56644368e-01 1.14924026e+00 -1.09766626e+00 8.69869351e-01 9.65345144e-01 5.94095945e-01 7.92200446e-01 3.81546527e-01 -6.52288198e-01 -1.28597939e+00 -1.04033637e+00 1.73893583e+00 -7.36782730e-01 9.14103150e-01 -5.58106780e-01 -7.48769581e-01 1.16181183e+00 7.78291047e-01 -4.87292588e-01 8.40903640e-01 5.41993260e-01 -5.69003940e-01 -2.04258785e-02 -7.08155215e-01 5.93398631e-01 9.19057488e-01 -7.03027427e-01 -7.09795058e-01 4.11607355e-01 7.34923124e-01 -2.97761530e-01 -1.09104013e+00 5.50225496e-01 6.01664543e-01 -6.34066164e-01 6.02544963e-01 -7.21401632e-01 5.23277938e-01 -8.13991129e-02 -2.71630049e-01 -1.59310818e+00 -1.11928917e-01 -6.24579370e-01 3.35399330e-01 1.16393936e+00 1.07848334e+00 -8.25104296e-01 4.44991142e-01 -1.69943467e-01 -3.31865251e-01 -7.29540706e-01 -7.26798952e-01 -1.04912794e+00 7.68069983e-01 -5.20449877e-01 2.05281660e-01 1.18830824e+00 2.68837456e-02 9.01398122e-01 -5.97837925e-01 -1.18266992e-01 3.17241728e-01 2.00430766e-01 8.07736993e-01 -8.70100200e-01 -4.59495455e-01 -3.59339118e-01 3.46870674e-03 -1.33157802e+00 3.50042105e-01 -1.41124916e+00 5.19272573e-02 -1.56288397e+00 4.96145785e-01 -2.82944053e-01 -2.43732184e-01 7.87479162e-01 -4.06526215e-02 5.71905553e-01 1.14213668e-01 3.24829787e-01 -4.32922184e-01 2.63019264e-01 9.04408693e-01 -5.64407893e-02 -2.00178996e-01 -3.49358797e-01 -5.99204957e-01 6.24852180e-01 9.70344961e-01 -5.98452330e-01 -2.40499765e-01 -1.02023113e+00 6.79613128e-02 -3.40743482e-01 -3.12457025e-01 -8.89832377e-01 -1.24652460e-01 1.39518380e-01 1.50186032e-01 -4.69614565e-01 1.22661166e-01 -4.64966029e-01 3.22424471e-02 5.70637703e-01 -3.91374111e-01 3.99920553e-01 5.63517392e-01 4.12495025e-02 -3.31521153e-01 -4.45020013e-02 6.48538589e-01 -1.41138792e-01 -4.47178036e-01 5.17875478e-02 -3.56585801e-01 1.22248374e-01 5.20628333e-01 2.06656560e-01 -2.82556057e-01 -2.96334535e-01 -5.76622248e-01 -5.88697270e-02 5.39396822e-01 7.05180466e-01 6.02556877e-02 -1.35510039e+00 -1.08186078e+00 4.26910728e-01 2.72342861e-01 -7.29395568e-01 -3.17930609e-01 9.16291356e-01 -3.35673928e-01 9.70297217e-01 -8.15066621e-02 -7.60893464e-01 -1.02661014e+00 3.17375213e-01 1.40307352e-01 -7.32722700e-01 -4.62310202e-02 5.47132075e-01 3.24645415e-02 -9.73202527e-01 -2.75029719e-01 -3.53419065e-01 2.10758537e-01 -2.57892102e-01 7.74498135e-02 -2.26822887e-02 3.17700177e-01 -7.88535535e-01 -3.15723151e-01 8.36039245e-01 -3.11181515e-01 -4.44523960e-01 1.12483191e+00 -6.23446479e-02 -2.53662050e-01 7.76384592e-01 1.41494489e+00 4.82801348e-01 -6.07662499e-01 -3.51047724e-01 1.26315460e-01 2.12975442e-02 -8.52378532e-02 -1.16938150e+00 -5.29113650e-01 9.04501617e-01 5.58633685e-01 -2.05121845e-01 9.80387807e-01 2.20282003e-01 9.50732231e-01 5.13547122e-01 6.38417184e-01 -9.15471673e-01 -2.68871069e-01 1.03831100e+00 5.90337515e-01 -1.53919637e+00 -2.36692712e-01 -2.07920462e-01 -5.66874921e-01 9.19644475e-01 2.24807501e-01 1.71753392e-02 4.42778379e-01 3.32813174e-01 4.65681553e-01 2.37204134e-01 -7.40335464e-01 -2.24550679e-01 6.01695538e-01 1.26408130e-01 1.09097302e+00 4.14430797e-01 -5.79681396e-01 3.92158747e-01 -5.00656426e-01 -4.52841818e-02 4.87673143e-03 7.09875345e-01 -2.16701746e-01 -1.89882207e+00 -7.68689290e-02 7.30508119e-02 -6.58157647e-01 -7.11226761e-01 -4.38028246e-01 9.42752302e-01 7.43619576e-02 8.66088808e-01 -1.09704928e-02 -3.46964002e-01 2.06133753e-01 5.50251245e-01 6.30709052e-01 -8.39314699e-01 -5.19997418e-01 4.43386197e-01 3.59583169e-01 -1.70694634e-01 -5.04488945e-01 -5.86220384e-01 -1.01606441e+00 -2.09492505e-01 -2.93628901e-01 4.39993262e-01 7.26253271e-01 1.15631664e+00 2.73176640e-01 1.73310593e-01 4.11926687e-01 -4.18350756e-01 -3.52144510e-01 -1.30259645e+00 2.55722851e-02 1.33338675e-01 -1.38429236e-02 -2.06643000e-01 -2.15067759e-01 3.48947227e-01]
[11.430704116821289, 10.212129592895508]
f0194cdf-efd7-471f-8c41-301fea8ea400
prediction-of-good-reaction-coordinates-and
2208.10962
null
https://arxiv.org/abs/2208.10962v1
https://arxiv.org/pdf/2208.10962v1.pdf
Prediction of good reaction coordinates and future evolution of MD trajectories using Regularized Sparse Autoencoders: A novel deep learning approach
Identifying reaction coordinates(RCs) is an active area of research, given the crucial role RCs play in determining the progress of a chemical reaction. The choice of the reaction coordinate is often based on heuristic knowledge. However, an essential criterion for the choice is that the coordinate should capture both the reactant and product states unequivocally. Also, the coordinate should be the slowest one so that all the other degrees of freedom can easily equilibrate along the reaction coordinate. Also, the coordinate should be the slowest one so that all the other degrees of freedom can easily equilibrate along the reaction coordinate. We used a regularised sparse autoencoder, an energy-based model, to discover a crucial set of reaction coordinates. Along with discovering reaction coordinates, our model also predicts the evolution of a molecular dynamics(MD) trajectory. We showcased that including sparsity enforcing regularisation helps in choosing a small but important set of reaction coordinates. We used two model systems to demonstrate our approach: alanine dipeptide system and proflavine and DNA system, which exhibited intercalation of proflavine into DNA minor groove in an aqueous environment. We model MD trajectory as a multivariate time series, and our latent variable model performs the task of multi-step time series prediction. This idea is inspired by the popular sparse coding approach - to represent each input sample as a linear combination of few elements taken from a set of representative patterns.
['Arnab Mukherjee', 'Abhijit Gupta']
2022-08-22
null
null
null
null
['time-series-prediction']
['time-series']
[ 2.12751493e-01 -3.74449730e-01 -2.23987609e-01 6.63554519e-02 -2.77741909e-01 -6.03652537e-01 7.51796961e-01 4.19050485e-01 -4.52659488e-01 9.56595600e-01 1.84488580e-01 -4.03314620e-01 -1.70332342e-01 -8.03574562e-01 -8.55723977e-01 -1.51007187e+00 -1.95978492e-01 6.39029443e-01 -4.25772294e-02 -2.97175229e-01 4.35944021e-01 9.64287937e-01 -1.37814021e+00 1.59928441e-01 6.60741568e-01 5.16991317e-01 1.68602869e-01 6.88937604e-01 3.22943889e-02 7.80798435e-01 -2.77552605e-01 1.74580783e-01 2.69322008e-01 -6.55243158e-01 -7.37374842e-01 -2.89214492e-01 -2.50942886e-01 3.64414245e-01 -2.59352744e-01 8.65957022e-01 4.96727049e-01 5.80642939e-01 1.17192030e+00 -7.84416556e-01 -2.61542052e-01 8.81034881e-02 -2.41477683e-01 2.26208627e-01 4.15589988e-01 2.65997708e-01 1.08963954e+00 -7.50059843e-01 1.06007457e+00 7.26532280e-01 3.95733148e-01 3.23514968e-01 -1.49982929e+00 -2.04278514e-01 -4.27426072e-03 2.22449809e-01 -1.30374527e+00 -2.61984348e-01 7.99429119e-01 -5.74721158e-01 1.31947482e+00 3.95834565e-01 9.48204279e-01 1.04229248e+00 7.79364407e-01 3.27522278e-01 8.12166691e-01 -4.26219374e-01 6.66167378e-01 -2.35640064e-01 4.17814963e-02 6.65324450e-01 -2.08590571e-02 2.43683040e-01 -3.16519856e-01 -4.42622066e-01 6.12579763e-01 4.57695007e-01 -3.22884083e-01 -5.09497404e-01 -1.26140320e+00 1.10326481e+00 2.23331735e-01 4.55744505e-01 -7.91194320e-01 -1.41794711e-01 1.06396250e-01 2.15474799e-01 -7.00745583e-02 7.65500605e-01 -4.66382265e-01 -3.34136367e-01 -7.66652405e-01 2.54617006e-01 9.23018575e-01 1.50724247e-01 8.56797338e-01 -5.20105027e-02 2.43633196e-01 3.89386982e-01 2.17516735e-01 1.04084499e-01 6.07837439e-01 -7.39247918e-01 5.20696566e-02 7.18802929e-01 2.64059067e-01 -8.67868960e-01 -4.93990451e-01 -1.18047751e-01 -1.09255755e+00 4.20020163e-01 6.06070459e-01 -2.08446503e-01 -6.90920413e-01 1.60877264e+00 4.00775731e-01 -2.75241017e-01 2.82637089e-01 9.41067100e-01 2.58406132e-01 1.31317580e+00 3.08766328e-02 -9.17857945e-01 1.10327959e+00 -5.25873423e-01 -5.27554870e-01 3.40829104e-01 6.41153932e-01 -6.97291434e-01 6.12069130e-01 5.07974088e-01 -9.36892986e-01 -3.64162326e-01 -1.08783484e+00 1.06374599e-01 -3.72629404e-01 -1.52043819e-01 6.64557517e-01 1.00526839e-01 -4.49827105e-01 1.03469300e+00 -1.07649255e+00 -2.25975752e-01 -4.55981344e-01 3.97560894e-01 -7.11875439e-01 1.64256901e-01 -1.04217446e+00 8.67289603e-01 4.06995744e-01 1.64115012e-01 -8.36380541e-01 -4.61913019e-01 -6.49058878e-01 -2.81676166e-02 1.78856969e-01 -5.14292419e-01 7.92192101e-01 -9.96160030e-01 -1.58809900e+00 3.92323405e-01 -4.12384450e-01 -3.24940592e-01 3.47009212e-01 2.68194467e-01 -3.60723972e-01 8.58012214e-02 -2.43067458e-01 3.01544130e-01 8.79800022e-01 -8.98676276e-01 -1.57669000e-02 -2.01292500e-01 -2.37847343e-01 1.72095954e-01 2.49854252e-01 -1.08235002e-01 5.81441559e-02 -3.95428777e-01 3.24171364e-01 -1.06748569e+00 -4.52654988e-01 -3.77242774e-01 -4.15749282e-01 -2.60746568e-01 5.17980218e-01 -4.76716191e-01 1.25797093e+00 -1.93022335e+00 8.58445704e-01 4.27403629e-01 3.75880182e-01 -1.76018849e-02 3.00790787e-01 1.04564667e+00 -8.33933055e-01 -1.19401470e-01 -3.16250861e-01 2.17055097e-01 -2.41144374e-01 1.18956782e-01 -2.30007306e-01 7.39569426e-01 1.86410338e-01 5.00362396e-01 -8.20090473e-01 6.77055493e-03 8.70596841e-02 6.56958163e-01 -4.40531194e-01 3.28518331e-01 -3.21004719e-01 5.38043797e-01 -4.37012553e-01 3.09410214e-01 2.56661236e-01 -4.60354418e-01 3.03434491e-01 -4.75401789e-01 -7.23755538e-01 1.33679196e-01 -1.40322876e+00 1.30064571e+00 -3.63923684e-02 2.47404188e-01 -5.00657141e-01 -9.96739268e-01 9.22626376e-01 4.19648647e-01 9.92554247e-01 -5.71803987e-01 1.12730198e-01 5.29221492e-03 3.62711757e-01 -3.91271174e-01 1.62004232e-01 -3.37764978e-01 1.41842484e-01 8.22386324e-01 -2.06074491e-01 4.82513085e-02 2.53279895e-01 2.42901236e-01 8.97020876e-01 -6.34077415e-02 7.50499308e-01 -8.33824873e-02 6.61518097e-01 1.42867073e-01 5.72565436e-01 3.70702326e-01 -8.42729211e-02 5.14818788e-01 8.47278059e-01 -9.67701674e-01 -1.60181808e+00 -5.18391788e-01 -2.65466105e-02 6.80633366e-01 -1.40800565e-01 -4.21416730e-01 -6.17247522e-01 -3.66778761e-01 -1.11249976e-01 3.84988517e-01 -8.02623272e-01 -2.24718899e-01 -6.50892019e-01 -9.50374842e-01 4.28313091e-02 3.63533534e-02 -1.74703404e-01 -1.27712798e+00 -5.35496294e-01 5.73753119e-01 1.28003195e-01 -2.43033528e-01 -4.64182466e-01 9.10427332e-01 -6.48324370e-01 -1.35888696e+00 -5.10014474e-01 -4.77271557e-01 7.73079634e-01 2.58440543e-02 6.86780810e-01 -6.16469271e-02 -4.64741677e-01 -1.05636187e-01 -1.15696296e-01 4.07080464e-02 -6.93918347e-01 -1.19831890e-01 2.24339932e-01 2.02853754e-01 1.65228248e-01 -5.27785122e-01 -7.55643547e-01 1.18582025e-01 -7.88751006e-01 1.03678882e-01 1.87699705e-01 8.90861690e-01 9.47130382e-01 1.67502701e-01 1.28419476e-03 -5.93051970e-01 4.35718596e-01 -6.56591058e-01 -5.45168519e-01 1.11717932e-01 -4.96339113e-01 6.99663222e-01 1.01874912e+00 -5.90697944e-01 -6.17956161e-01 4.25076246e-01 -2.97431350e-01 -3.18511605e-01 -1.19806722e-01 6.26973569e-01 -1.27788410e-01 -3.73271899e-03 6.42533302e-01 6.98403418e-01 1.42425418e-01 -3.40997696e-01 2.62076646e-01 1.04972832e-01 1.99636355e-01 -6.55227065e-01 4.41467613e-01 4.68876481e-01 3.47609222e-01 -8.52404237e-01 1.09751401e-02 -5.20641148e-01 -7.09500313e-01 -1.20310210e-01 8.56098354e-01 -4.82265800e-01 -1.23365200e+00 3.00160557e-01 -1.08747232e+00 -7.57711977e-02 -1.22921422e-01 4.87533510e-01 -6.44328177e-01 5.14096677e-01 -4.93966103e-01 -6.95550919e-01 -1.89192355e-01 -1.42312479e+00 5.63864410e-01 4.51352680e-04 -4.62077320e-01 -8.17280233e-01 6.87292576e-01 -2.31027007e-01 1.52470708e-01 5.40953219e-01 1.23707414e+00 -6.85662150e-01 -4.57524061e-01 -4.21410263e-01 4.33203548e-01 -1.13250047e-01 9.24313366e-02 5.59674501e-01 -3.55335683e-01 -4.41968203e-01 1.24083787e-01 6.58913748e-03 7.74524629e-01 5.33022344e-01 6.39015973e-01 -3.33734065e-01 -3.51778418e-01 5.75664222e-01 1.31732011e+00 7.07820535e-01 3.88834357e-01 1.75557494e-01 7.30147183e-01 4.25501019e-01 1.44572377e-01 4.35262233e-01 1.30738333e-01 7.78323650e-01 3.88029844e-01 -2.59758290e-02 3.66333336e-01 -1.78115830e-01 4.52787876e-01 8.21199238e-01 -4.79601622e-01 -2.01940715e-01 -8.57699990e-01 1.01349294e-01 -1.66771066e+00 -1.24133527e+00 -1.68232068e-01 2.25509906e+00 8.73488724e-01 -2.65723616e-02 4.50195640e-01 4.55024987e-01 4.23913568e-01 1.96693048e-01 -8.28248501e-01 -6.05063796e-01 -1.43596515e-01 1.74883693e-01 3.65916967e-01 6.99710846e-01 -7.93811798e-01 6.04655325e-01 6.43003988e+00 4.11109239e-01 -1.62813592e+00 -1.65756971e-01 5.06768048e-01 -2.84767412e-02 -2.28658110e-01 6.88710958e-02 -7.73020566e-01 6.35797799e-01 1.15680611e+00 -3.91551219e-02 5.30575693e-01 6.56850457e-01 4.74023432e-01 -1.33238345e-01 -1.10902929e+00 6.89708531e-01 -4.06655818e-01 -1.63977873e+00 -5.35195693e-02 5.66267334e-02 6.48814678e-01 -1.27328962e-01 -2.09138364e-01 -2.42579862e-01 3.57746840e-01 -9.91470993e-01 6.86386585e-01 5.77046692e-01 5.01185536e-01 -7.67440379e-01 3.48483145e-01 5.13393998e-01 -1.06279337e+00 -5.46557643e-03 -1.78809986e-01 -2.08199650e-01 8.15507174e-02 4.97364342e-01 -7.40146697e-01 2.35534117e-01 6.41307756e-02 7.63311684e-01 1.60531044e-01 6.80426538e-01 -8.94889683e-02 5.21687567e-01 -3.51613373e-01 -2.54320830e-01 3.73717487e-01 -7.33813286e-01 4.14268345e-01 9.04949725e-01 1.40625343e-01 3.19758356e-01 9.14125144e-02 6.91491723e-01 3.61157954e-01 1.99601114e-01 -6.49729311e-01 -3.25334460e-01 2.41985455e-01 8.42186928e-01 -8.16831589e-01 -1.22671900e-02 -3.45226765e-01 8.86888504e-01 2.65959322e-01 6.30324781e-01 -5.94740689e-01 -2.63805777e-01 9.55267549e-01 -3.80615778e-02 3.97227257e-01 -4.52511072e-01 2.32395694e-01 -1.14980865e+00 -3.32709730e-01 -1.06502843e+00 3.36016238e-01 -4.04607266e-01 -7.80913651e-01 5.40236294e-01 -4.08344835e-01 -9.35519099e-01 -2.93759316e-01 -3.20800781e-01 -4.46072906e-01 1.16487646e+00 -1.26324856e+00 -4.56800789e-01 1.32731110e-01 5.02328098e-01 4.31839973e-01 -1.66685879e-01 1.00636029e+00 1.04882866e-02 -9.89009440e-01 1.21762559e-01 6.13627851e-01 -2.01061621e-01 5.18243849e-01 -1.10062611e+00 1.84722453e-01 6.49810553e-01 2.10905924e-01 1.04660261e+00 9.68432426e-01 -6.74028993e-01 -1.76067340e+00 -6.10506356e-01 8.29836130e-01 -3.11782539e-01 5.31738281e-01 -3.14834028e-01 -9.63626385e-01 4.22832161e-01 -1.88370734e-01 -1.12820379e-02 7.62922168e-01 -9.27241072e-02 -1.94011614e-01 1.95255488e-01 -7.67955542e-01 5.61364651e-01 5.19951642e-01 -5.19316971e-01 -4.15066034e-01 4.54617739e-01 4.46480274e-01 -3.17934752e-01 -1.03335428e+00 -1.88233197e-01 6.75490201e-01 -9.64010119e-01 1.07291341e+00 -9.63738382e-01 3.13832313e-01 -6.11563206e-01 5.69066741e-02 -1.21630919e+00 -6.37022257e-01 -9.03330743e-01 -5.50234973e-01 3.63599151e-01 2.69507945e-01 -5.93555450e-01 8.80442679e-01 5.80826938e-01 9.64456201e-02 -9.21734750e-01 -8.43321681e-01 -4.68464553e-01 -5.45603670e-02 -2.30798647e-02 4.79778171e-01 1.04057562e+00 1.45664839e-02 3.03094298e-01 -4.89077955e-01 8.25502872e-02 2.66772449e-01 1.81490660e-01 3.70714486e-01 -1.12887168e+00 -3.86609674e-01 -3.86780679e-01 -2.71323949e-01 -7.88643658e-01 5.49633503e-02 -8.64837408e-01 -1.75371319e-01 -1.01161432e+00 1.32744193e-01 -2.94008851e-01 -4.84104872e-01 3.58685464e-01 -2.73574870e-02 -3.30841005e-01 -5.64224496e-02 4.75778311e-01 -2.39108175e-01 4.61289406e-01 1.01963520e+00 -7.84473047e-02 -4.87524420e-01 -4.89019975e-02 -5.01932383e-01 3.00883114e-01 5.45437574e-01 -7.48202384e-01 -1.98241606e-01 2.35010341e-01 6.36070907e-01 4.97614324e-01 -9.97968763e-02 -6.62134767e-01 3.68914157e-01 -5.40919363e-01 3.81912172e-01 -4.31488156e-01 3.56650054e-01 -8.23180914e-01 7.41432667e-01 8.14533472e-01 -2.65859157e-01 3.21709484e-01 -2.03851134e-01 6.09866619e-01 -1.47559687e-01 -1.15417562e-01 7.58857489e-01 -1.08271502e-01 -2.81522721e-01 3.99629593e-01 -7.74826884e-01 -5.73773086e-01 1.04022276e+00 -3.95044148e-01 1.92981526e-01 -8.04181471e-02 -8.75158548e-01 -2.15253219e-01 7.02660322e-01 9.33280811e-02 2.62953669e-01 -1.11715245e+00 -4.62161034e-01 5.93817770e-01 -9.50298309e-02 -1.05091557e-01 1.02945231e-01 8.31549168e-01 -7.52553701e-01 6.08244896e-01 -2.04590648e-01 -4.78426695e-01 -1.11690974e+00 7.78666317e-01 4.47753757e-01 -2.45094776e-01 -5.96739948e-01 5.59648633e-01 -1.84571445e-01 -1.29835024e-01 -1.46233365e-02 -3.41619402e-01 -2.98195779e-01 2.91583329e-01 5.00641525e-01 3.15302402e-01 2.39274442e-01 -6.46538556e-01 -2.48672426e-01 7.08579838e-01 -2.47632593e-01 3.28428239e-01 1.75717592e+00 3.52448732e-01 -2.49360338e-01 5.94510019e-01 1.26380765e+00 -1.32103458e-01 -1.45318079e+00 7.00720549e-02 -8.70829597e-02 -2.26289295e-02 1.02119155e-01 -5.11028230e-01 -6.03067219e-01 6.30240083e-01 3.63139331e-01 2.58268088e-01 7.47491241e-01 -4.58116412e-01 4.16819870e-01 6.32801235e-01 1.81337520e-01 -7.67730713e-01 -5.59223443e-03 5.54332197e-01 5.56261241e-01 -1.13650215e+00 5.97931482e-02 1.10255517e-01 -5.53321004e-01 1.50756919e+00 -3.68060358e-02 -1.78819254e-01 4.55445111e-01 1.40590295e-01 -2.58120865e-01 -3.30781698e-01 -8.87875080e-01 2.61005163e-01 1.35121733e-01 1.08553499e-01 6.88734949e-01 -4.79436032e-02 -4.16971445e-01 2.48607427e-01 1.23467967e-01 -2.73075700e-01 3.53585184e-01 8.81339490e-01 -3.41698855e-01 -1.31481493e+00 -3.58044088e-01 5.59984259e-02 -1.03392318e-01 -5.18310256e-02 -2.52721548e-01 5.52508295e-01 -8.27441514e-02 4.22374398e-01 9.79261380e-03 1.36490881e-01 1.34956256e-01 4.63351399e-01 4.43300962e-01 -3.22858363e-01 -5.54328442e-01 2.08193492e-02 -2.33736187e-01 -6.22302473e-01 -1.70764133e-01 -9.34104145e-01 -1.44467962e+00 -4.88333762e-01 -2.22569376e-01 6.64829433e-01 7.72313476e-01 9.15600598e-01 3.25302869e-01 3.35726708e-01 8.99037659e-01 -7.29839146e-01 -3.75007987e-01 -6.22739494e-01 -4.68434185e-01 4.96481419e-01 6.51238918e-01 -6.82331920e-01 -4.26595658e-01 1.57352626e-01]
[5.050365447998047, 5.2684831619262695]
25f97cf0-5d79-40bc-a87c-08a886cff2b4
assessing-the-severity-of-health-states-based
2009.09600
null
https://arxiv.org/abs/2009.09600v1
https://arxiv.org/pdf/2009.09600v1.pdf
Assessing the Severity of Health States based on Social Media Posts
The unprecedented growth of Internet users has resulted in an abundance of unstructured information on social media including health forums, where patients request health-related information or opinions from other users. Previous studies have shown that online peer support has limited effectiveness without expert intervention. Therefore, a system capable of assessing the severity of health state from the patients' social media posts can help health professionals (HP) in prioritizing the user's post. In this study, we inspect the efficacy of different aspects of Natural Language Understanding (NLU) to identify the severity of the user's health state in relation to two perspectives(tasks) (a) Medical Condition (i.e., Recover, Exist, Deteriorate, Other) and (b) Medication (i.e., Effective, Ineffective, Serious Adverse Effect, Other) in online health communities. We propose a multiview learning framework that models both the textual content as well as contextual-information to assess the severity of the user's health state. Specifically, our model utilizes the NLU views such as sentiment, emotions, personality, and use of figurative language to extract the contextual information. The diverse NLU views demonstrate its effectiveness on both the tasks and as well as on the individual disease to assess a user's health.
['Joy Prakash Sain', 'Sriparna Saha', 'Amit Sheth', 'Asif Ekbal', 'Shweta Yadav', 'Pushpak Bhattacharyya']
2020-09-21
null
null
null
null
['multiview-learning']
['computer-vision']
[ 5.22535928e-02 3.39629710e-01 -7.73287475e-01 -3.67008150e-01 -4.62994635e-01 -2.80891001e-01 2.98634857e-01 1.11583447e+00 1.05413154e-01 6.02973819e-01 9.48922932e-01 -2.02602804e-01 -3.25850368e-01 -4.93489474e-01 2.80688763e-01 -4.79673535e-01 -4.59364615e-02 3.03669989e-01 -2.95958459e-01 -2.60843843e-01 4.54610765e-01 -1.01549782e-01 -1.49009550e+00 8.10784042e-01 9.54013646e-01 9.20618176e-01 4.65326793e-02 4.87772346e-01 -2.02789679e-01 1.19541717e+00 -5.83671510e-01 -9.36830193e-02 -3.41133863e-01 -3.32348406e-01 -7.94928551e-01 2.00484380e-01 -9.84574482e-02 -5.46665132e-01 4.22397941e-01 8.10718060e-01 6.98879242e-01 -4.79955047e-01 7.08331764e-01 -1.14851081e+00 -3.71357113e-01 2.83479273e-01 -1.82138890e-01 8.48268252e-03 1.12261546e+00 4.03393526e-03 9.67485547e-01 -6.35138988e-01 8.61955762e-01 1.29586673e+00 5.60546577e-01 3.91451269e-01 -6.73708320e-01 -3.12226534e-01 1.30339995e-01 2.32227314e-02 -7.77066827e-01 -1.83958650e-01 7.23362744e-01 -8.77875865e-01 4.15547460e-01 5.27147353e-01 6.93363369e-01 1.16162801e+00 5.50090194e-01 6.39086187e-01 1.39669526e+00 -1.38670811e-02 3.45628649e-01 4.77244049e-01 3.91659498e-01 5.08659363e-01 -6.88999221e-02 -6.06272519e-01 -3.91888618e-01 -9.00263488e-01 2.12629288e-01 6.37128949e-01 -1.67621031e-01 4.06847417e-01 -1.13590503e+00 7.88565338e-01 1.14698812e-01 2.61642486e-01 -7.06541061e-01 -7.69933224e-01 6.41807020e-01 2.90625781e-01 9.20839727e-01 4.01900083e-01 -6.90788746e-01 9.72498432e-02 -4.45566297e-01 3.39711620e-03 1.14936614e+00 2.86763966e-01 5.93980253e-01 -6.51150286e-01 -4.99676108e-01 7.42357492e-01 4.71537858e-01 4.93362546e-01 5.56591868e-01 -8.07466269e-01 2.33142167e-01 1.27638352e+00 2.26330370e-01 -1.41236806e+00 -5.81667006e-01 -3.51760387e-01 -9.84255135e-01 -3.48546296e-01 -3.03898484e-01 -6.44818068e-01 -5.19228995e-01 1.48325610e+00 8.01532447e-01 -4.70005348e-02 2.28823990e-01 6.04738176e-01 1.35159266e+00 6.80762768e-01 4.41549540e-01 -6.55709743e-01 1.81010723e+00 -4.76480365e-01 -1.25057292e+00 -1.14286102e-01 7.78650403e-01 -8.83102000e-01 7.67680526e-01 4.32930857e-01 -1.00026035e+00 -1.15451872e-01 -6.27079904e-01 1.79529086e-01 -4.95707601e-01 1.70213029e-01 2.87007421e-01 1.17029227e-01 -9.53247726e-01 6.98695958e-01 -2.88958013e-01 -6.43465400e-01 1.39581189e-01 1.28714278e-01 -2.91800380e-01 1.59814149e-01 -1.62051868e+00 8.03845286e-01 2.88342208e-01 -2.72297889e-01 -4.36718464e-01 -8.36680591e-01 -6.50612295e-01 -1.79393843e-01 3.97752792e-01 -1.10684824e+00 9.08253193e-01 -9.01469588e-01 -1.10568917e+00 8.38183284e-01 -9.75134149e-02 6.15502857e-02 2.55893648e-01 -1.06498748e-01 -5.39597631e-01 3.60501766e-01 1.35883585e-01 3.09520036e-01 5.82161129e-01 -8.28789294e-01 -9.27408993e-01 -7.89292276e-01 1.23648413e-01 5.02609909e-01 -7.37498879e-01 2.65676469e-01 -5.24812564e-02 -4.55481648e-01 -2.36812770e-01 -4.61081237e-01 -2.77731121e-01 2.93974206e-02 -4.64322835e-01 -6.85150981e-01 9.29050446e-01 -1.09353411e+00 1.78255343e+00 -2.00659943e+00 1.19733289e-01 8.55168998e-02 6.10383391e-01 3.01207751e-01 2.38617554e-01 9.65453506e-01 1.45298392e-01 5.98876715e-01 2.32706722e-02 1.21863648e-01 -4.40941632e-01 1.51231915e-01 3.18093359e-01 1.29817039e-01 2.30681971e-01 3.95654589e-01 -1.26373017e+00 -8.16442788e-01 -3.77724767e-02 6.68555140e-01 -6.64019585e-01 4.30755138e-01 -1.19004540e-01 6.11439645e-01 -1.01374567e+00 6.75803721e-01 7.66148865e-02 -7.46744812e-01 4.76964653e-01 -3.12787920e-01 -6.67114109e-02 4.22662258e-01 -9.21375811e-01 8.72831583e-01 -3.90588075e-01 -1.20158076e-01 4.17297930e-01 -5.28910160e-01 6.11971676e-01 9.19805467e-01 9.65590894e-01 -2.18794182e-01 1.76260900e-02 2.33670026e-02 -2.78643101e-01 -1.20196319e+00 -2.46509895e-01 -4.98075075e-02 -3.37336548e-02 7.33353078e-01 -4.54855710e-01 4.17233288e-01 -1.21795245e-01 4.84385759e-01 1.03937376e+00 -3.32634270e-01 1.13622129e+00 -2.72237808e-01 6.98535323e-01 -1.75504029e-01 5.11109948e-01 1.53929248e-01 -3.75155389e-01 5.94726726e-02 9.87384677e-01 -3.79775226e-01 -3.80540460e-01 -3.42346877e-01 -1.10948585e-01 1.16667628e+00 5.06038517e-02 -7.29737997e-01 -5.07883608e-01 -5.01644313e-01 -1.51489243e-01 3.72027874e-01 -7.69210100e-01 8.84747803e-02 -5.44637144e-02 -6.77849233e-01 -9.46184173e-02 2.16806203e-01 2.50079215e-01 -1.21276486e+00 -6.34193420e-01 1.51126578e-01 -6.65758133e-01 -6.78987145e-01 -4.26089287e-01 -5.37871361e-01 -8.38636518e-01 -1.31687105e+00 -5.75902820e-01 -7.47640312e-01 6.44676208e-01 1.90541372e-02 1.19991374e+00 2.13419542e-01 -1.29264563e-01 8.04162860e-01 -4.04232740e-01 -4.85460877e-01 -5.38399816e-01 -3.04059386e-01 5.19341230e-03 2.10030332e-01 1.52753755e-01 -2.97466010e-01 -1.14025128e+00 1.89319134e-01 -1.02324176e+00 2.51413226e-01 3.06571484e-01 4.55588251e-01 3.98638278e-01 -7.68678859e-02 8.59442949e-01 -1.41988254e+00 1.27269685e+00 -1.43894172e+00 5.45889378e-01 2.91001290e-01 -7.59590566e-01 -3.68072301e-01 4.15601462e-01 -3.37437004e-01 -1.08617795e+00 -2.82541901e-01 -8.25481713e-02 3.05868924e-01 -3.40517759e-01 1.21924341e+00 1.93209723e-01 5.45577586e-01 5.38582027e-01 -2.09346578e-01 1.69839412e-01 -3.30395758e-01 -8.66972953e-02 1.39178622e+00 -1.79262206e-01 -7.48747215e-02 -2.28976440e-02 3.66095603e-01 -3.26929450e-01 -8.81355703e-01 -1.23479784e+00 -1.08784950e+00 8.76945443e-03 -5.86517751e-01 1.03125703e+00 -1.06193388e+00 -1.05868542e+00 2.29263604e-01 -1.20762599e+00 2.22962946e-01 3.77601206e-01 1.36759639e-01 4.58272472e-02 3.58869523e-01 -7.81521678e-01 -9.17361438e-01 -1.01380360e+00 -8.68493080e-01 9.08151567e-01 1.25190392e-01 -8.70927215e-01 -1.30410790e+00 1.17989562e-01 6.32269800e-01 4.39526081e-01 4.29038435e-01 1.19101965e+00 -8.43867958e-01 2.44086623e-01 -3.35662365e-01 -1.15291759e-01 -7.05530569e-02 7.24372983e-01 -3.83079648e-02 -5.87793350e-01 -7.28543848e-02 4.86339569e-01 -4.28287148e-01 1.51100665e-01 3.71292919e-01 8.83972049e-01 -1.02182531e+00 -3.17644536e-01 -2.46591702e-01 1.03650832e+00 2.29782313e-01 5.55940807e-01 -3.28442156e-01 6.47850215e-01 1.17242622e+00 6.42913580e-01 1.12696981e+00 7.34882712e-01 1.69452474e-01 5.25333941e-01 -3.33599865e-01 3.05140227e-01 -1.93824530e-01 6.74057484e-01 1.02642822e+00 -3.47536206e-02 3.16998437e-02 -9.35525358e-01 2.15574831e-01 -1.87246370e+00 -5.86486936e-01 -6.45529628e-02 1.94793582e+00 9.69103456e-01 -2.32967943e-01 9.53250602e-02 -1.32875443e-01 7.98508883e-01 1.67563751e-01 -5.94744325e-01 -3.89315039e-01 4.03884113e-01 -3.74191791e-01 -1.85381964e-01 3.42378110e-01 -8.42553258e-01 2.40082651e-01 5.54966068e+00 3.61706018e-01 -1.12290823e+00 1.26242146e-01 1.19986331e+00 3.27711970e-01 -5.76864064e-01 -3.52838606e-01 -6.44112647e-01 5.60330749e-01 6.61176980e-01 -7.21159950e-02 7.33465552e-02 5.54842770e-01 7.76964426e-01 -9.54008698e-02 -1.03584099e+00 7.73869634e-01 2.83833951e-01 -1.05057621e+00 9.52106640e-02 -1.77736178e-01 7.20291436e-01 -3.06317449e-01 -3.55249681e-02 1.12720758e-01 -1.58913970e-01 -5.56006074e-01 5.35760783e-02 7.15620279e-01 7.28430748e-01 -7.52287135e-02 8.97202253e-01 6.51416183e-01 -1.07877231e+00 2.66008265e-03 2.32731029e-01 -1.19022839e-01 5.56966215e-02 5.51711977e-01 -1.18681228e+00 5.94071984e-01 4.52407241e-01 1.25692058e+00 -4.74557221e-01 5.06488323e-01 -9.31717008e-02 4.91843790e-01 1.53469563e-01 -1.79341614e-01 -2.31132284e-02 -2.97615588e-01 3.93034577e-01 9.52862799e-01 3.06216657e-01 5.65840483e-01 4.52879936e-01 3.19376796e-01 1.96251780e-01 9.25945878e-01 -8.13382924e-01 -4.40693289e-01 1.96390346e-01 1.22100687e+00 -4.99422610e-01 -7.68020034e-01 -3.73957425e-01 3.80833834e-01 1.07142860e-02 2.83363968e-01 -2.31398314e-01 1.67452529e-01 3.27882022e-01 5.84742486e-01 -3.62200797e-01 6.04896545e-01 -1.95993334e-01 -1.01411569e+00 -2.37886772e-01 -1.03351367e+00 8.00301969e-01 -6.61216319e-01 -1.51549745e+00 2.92338252e-01 -2.77704716e-01 -1.39112151e+00 -2.34014928e-01 -3.06704402e-01 -7.84508705e-01 5.95715046e-01 -1.26221180e+00 -8.65594923e-01 -2.21344340e-03 5.21081924e-01 5.10696948e-01 1.84858605e-01 9.52228844e-01 2.94432908e-01 -5.76798618e-01 -1.43608764e-01 -4.67860669e-01 -1.91786215e-01 1.08759463e+00 -1.01987290e+00 -6.44714236e-01 -2.15981275e-01 -9.91369128e-01 6.53279424e-01 6.03319168e-01 -1.14332223e+00 -1.29832745e+00 -8.22115004e-01 1.61477065e+00 -3.23849648e-01 5.93137562e-01 3.66372466e-01 -7.95904577e-01 1.76776960e-01 3.26901585e-01 -5.57660222e-01 1.41151774e+00 3.92693579e-01 -7.80183775e-03 1.38462141e-01 -1.47525012e+00 7.88209617e-01 4.29505587e-01 -4.56383049e-01 -6.26541972e-01 8.10660422e-01 5.75993598e-01 1.41995000e-02 -1.27878869e+00 3.94437104e-01 5.60546875e-01 -7.10613787e-01 9.51500535e-01 -1.08686399e+00 1.10559285e+00 6.62545562e-02 1.06396079e-01 -1.18350494e+00 -1.16064690e-01 -5.21112621e-01 -1.00746110e-01 8.62550676e-01 4.20528054e-01 -8.08702707e-01 2.11744890e-01 8.43653083e-01 1.79594979e-01 -1.26306367e+00 -1.38272628e-01 5.84496498e-01 -6.12089097e-01 8.55760723e-02 1.64458171e-01 1.43177700e+00 7.10122943e-01 6.13769591e-01 -4.60147858e-01 1.20707192e-01 1.97674274e-01 1.12903245e-01 1.96983218e-01 -1.78474283e+00 -2.76569892e-02 -2.10651994e-01 1.24266893e-01 -2.74910301e-01 -3.29171360e-01 -7.19720662e-01 -5.04485786e-01 -2.15667367e+00 6.43477023e-01 -1.50298893e-01 -5.18147767e-01 4.37429041e-01 -4.78189856e-01 -4.45790261e-01 -3.34003866e-02 4.31788206e-01 -8.69373143e-01 8.39350075e-02 1.56048465e+00 -1.44235596e-01 -4.33855981e-01 2.01085210e-01 -9.21629250e-01 9.66005683e-01 8.10419977e-01 -3.70095462e-01 -2.70868003e-01 1.36707455e-01 1.11953485e+00 8.42387676e-01 -1.82315130e-02 -2.69121438e-01 2.00674102e-01 -4.25770104e-01 -2.20345538e-02 -4.92068470e-01 1.65839508e-01 -6.67718232e-01 2.38121711e-02 7.80796409e-01 -7.50704944e-01 -2.63493005e-02 -3.17725003e-01 6.17309034e-01 -4.39586639e-01 -9.25630853e-02 2.72379279e-01 -3.89002621e-01 -2.18027800e-01 3.38443428e-01 -5.51292360e-01 1.00356415e-02 8.32528651e-01 2.50885069e-01 -4.59582180e-01 -8.72913539e-01 -1.09593511e+00 6.92264140e-01 8.99841338e-02 2.01751754e-01 7.05778062e-01 -9.41624463e-01 -9.38550949e-01 -1.58734515e-01 2.19198585e-01 -3.33955914e-01 7.49590874e-01 1.23006177e+00 -3.10686648e-01 4.20439281e-02 -4.12383303e-02 -3.46700609e-01 -1.58629239e+00 5.78615725e-01 -1.52852699e-01 -6.63731694e-01 -1.22381479e-01 1.40078306e-01 2.06092879e-01 -3.51001769e-01 4.33814764e-01 -6.64712116e-02 -1.13716114e+00 8.85279536e-01 7.99465358e-01 5.42728841e-01 -1.38694257e-01 -5.69060922e-01 -2.34770685e-01 3.35295856e-01 -1.17881216e-01 2.42652863e-01 1.29483259e+00 -2.63166368e-01 -6.34532332e-01 6.82224154e-01 1.18882513e+00 -1.82224810e-01 -2.79804707e-01 -2.56365865e-01 -4.43413109e-02 1.01885326e-01 -5.97851863e-03 -1.00303090e+00 -7.33748972e-01 7.33644426e-01 4.99824673e-01 6.80853248e-01 1.04480660e+00 1.27150528e-02 7.11642921e-01 2.07620561e-01 -8.64502490e-02 -1.24650407e+00 2.43711650e-01 1.55287907e-01 1.01188052e+00 -1.56477582e+00 1.59488678e-01 -4.85615194e-01 -1.15221179e+00 1.04487908e+00 3.15105170e-01 4.54014212e-01 1.19987559e+00 -1.95350185e-01 4.46651876e-01 -6.52128279e-01 -1.21783495e+00 -1.79108081e-03 3.69588912e-01 6.61373362e-02 7.30178654e-01 4.99571711e-01 -5.99725723e-01 5.52600861e-01 4.60402906e-01 4.61382493e-02 3.62585127e-01 1.04990292e+00 -4.33046371e-01 -9.42315519e-01 -3.59035105e-01 7.87007809e-01 -8.01380634e-01 -5.24930470e-03 -7.81303406e-01 9.41677913e-02 2.40565643e-01 1.19665432e+00 -3.48426282e-01 -2.50398308e-01 1.93908110e-01 2.15008885e-01 -4.07396913e-01 -9.84658420e-01 -9.54828620e-01 4.10499513e-01 3.53670657e-01 -5.16918421e-01 -7.43817627e-01 -4.46292311e-01 -9.93767262e-01 1.67514592e-01 1.16985641e-01 1.88989937e-03 6.43108606e-01 1.23420393e+00 6.60771489e-01 3.82903993e-01 9.34012353e-01 -2.03161433e-01 -3.17172140e-01 -8.46007168e-01 -1.75876632e-01 7.53027081e-01 4.46090877e-01 -2.57955730e-01 -4.97405410e-01 -4.74708080e-02]
[8.657608032226562, 8.86437702178955]
41fabb8e-b51e-4053-a147-c39a65d9b282
open-domain-question-answering-via-chain-of
2210.12338
null
https://arxiv.org/abs/2210.12338v1
https://arxiv.org/pdf/2210.12338v1.pdf
Open-domain Question Answering via Chain of Reasoning over Heterogeneous Knowledge
We propose a novel open-domain question answering (ODQA) framework for answering single/multi-hop questions across heterogeneous knowledge sources. The key novelty of our method is the introduction of the intermediary modules into the current retriever-reader pipeline. Unlike previous methods that solely rely on the retriever for gathering all evidence in isolation, our intermediary performs a chain of reasoning over the retrieved set. Specifically, our method links the retrieved evidence with its related global context into graphs and organizes them into a candidate list of evidence chains. Built upon pretrained language models, our system achieves competitive performance on two ODQA datasets, OTT-QA and NQ, against tables and passages from Wikipedia. In particular, our model substantially outperforms the previous state-of-the-art on OTT-QA with an exact match score of 47.3 (45 % relative gain).
['Jianfeng Gao', 'Eric Nyberg', 'Xiaodong Liu', 'Hao Cheng', 'Kaixin Ma']
2022-10-22
null
null
null
null
['open-domain-question-answering']
['natural-language-processing']
[-4.16445166e-01 4.90163773e-01 -2.75960058e-01 7.63892159e-02 -1.83310902e+00 -1.07133222e+00 6.00655317e-01 8.31929505e-01 -4.28013146e-01 8.44783306e-01 6.49214208e-01 -5.36190271e-01 -3.99042070e-01 -1.15209174e+00 -9.62402165e-01 1.26316756e-01 5.45976497e-03 1.03094327e+00 1.27440894e+00 -7.34861255e-01 1.78844020e-01 -3.37688019e-03 -1.11072040e+00 5.76335073e-01 1.28167617e+00 1.02251399e+00 -1.52496412e-01 7.44628012e-01 -5.01060903e-01 1.31816757e+00 -5.27383029e-01 -1.23612225e+00 -1.77634820e-01 -1.21406630e-01 -1.61954975e+00 -6.74484849e-01 6.74820840e-01 -3.21247190e-01 -5.16359568e-01 8.40086341e-01 4.58208501e-01 9.42053050e-02 4.49112028e-01 -7.58499563e-01 -1.05887485e+00 8.36710334e-01 -1.72011822e-01 4.78330940e-01 1.04860318e+00 -1.83213905e-01 1.70133567e+00 -1.00848329e+00 9.80581105e-01 1.03049028e+00 4.19597596e-01 1.32397994e-01 -8.30388308e-01 -1.05817080e-01 9.11896676e-02 6.88175201e-01 -1.26822317e+00 -1.63027823e-01 2.94211537e-01 2.93104425e-02 1.42734098e+00 2.42391229e-01 2.68929541e-01 5.95809340e-01 -1.26306549e-01 7.52502024e-01 7.51522005e-01 -6.44209266e-01 6.49835765e-02 -1.39106661e-01 5.76435208e-01 1.06763148e+00 4.08067405e-01 -5.78270078e-01 -7.66043663e-01 -4.09919918e-01 1.52596310e-01 -7.66434252e-01 -1.38686866e-01 -1.72467098e-01 -1.17670238e+00 5.44148326e-01 6.12501800e-01 7.72067085e-02 -4.54006642e-01 -1.61321033e-02 1.95989296e-01 5.61775625e-01 3.88475448e-01 6.41742945e-01 -5.61211526e-01 1.32293269e-01 -4.94453609e-01 4.75204647e-01 1.38036335e+00 9.96583283e-01 6.04373097e-01 -9.61512625e-01 -5.60018003e-01 8.87977600e-01 5.66697359e-01 6.47388339e-01 -3.09726279e-02 -9.38541651e-01 9.59613562e-01 8.94313693e-01 2.82916665e-01 -8.96500409e-01 -1.21169493e-01 -5.05811334e-01 1.14231326e-01 -7.74606109e-01 5.85647345e-01 2.70296261e-02 -6.89451993e-01 1.37456191e+00 7.41210938e-01 -1.61837548e-01 4.22760874e-01 8.35882425e-01 1.41775596e+00 8.06893110e-01 1.24022603e-01 1.68687791e-01 1.75993228e+00 -1.28396308e+00 -7.17056215e-01 -2.67443299e-01 6.04447365e-01 -7.54088998e-01 1.04782987e+00 2.09057912e-01 -1.26917279e+00 -4.60435562e-02 -1.00302470e+00 -7.91686952e-01 -5.95275462e-01 -2.60878682e-01 2.18222916e-01 1.72986299e-01 -1.28060794e+00 1.19909219e-01 -2.87304819e-01 -4.75034267e-01 2.72757057e-02 -5.46328491e-03 -2.43685827e-01 -6.72143579e-01 -1.69037879e+00 1.06928885e+00 2.96841711e-01 -1.50768176e-01 -9.93513703e-01 -6.62233651e-01 -6.62528276e-01 1.85563669e-01 9.53870952e-01 -1.22724104e+00 1.53572559e+00 1.06206961e-01 -1.34525061e+00 8.28305662e-01 -3.27074289e-01 -4.52981353e-01 1.25145957e-01 -5.33769011e-01 -7.33142912e-01 5.13623178e-01 4.38557804e-01 2.52445936e-01 2.81158417e-01 -1.14520741e+00 -5.31582713e-01 -1.88283041e-01 7.79656589e-01 3.08699340e-01 -6.63882270e-02 1.77669734e-01 -1.22771466e+00 -2.18456358e-01 8.97343159e-02 -4.98008460e-01 3.56268510e-02 -2.11674005e-01 -4.41934973e-01 -7.73635268e-01 1.83795452e-01 -9.26120281e-01 1.55134618e+00 -1.57588494e+00 1.59632206e-01 2.70962328e-01 3.43150645e-01 1.20853975e-01 -2.59398252e-01 9.86053467e-01 6.62621498e-01 4.26525716e-03 -8.99423957e-02 6.28756359e-02 3.55186999e-01 1.98388949e-01 -6.52254105e-01 -1.36240989e-01 3.22919369e-01 1.33409798e+00 -1.54001081e+00 -1.03570318e+00 -5.32331884e-01 8.93401541e-03 -5.68398476e-01 1.99653938e-01 -9.42823410e-01 -5.21245264e-02 -8.02050173e-01 9.24836457e-01 4.03520346e-01 -6.32538974e-01 2.57717043e-01 -1.14934832e-01 3.51489246e-01 1.16166878e+00 -8.87426257e-01 1.95100629e+00 -2.87889212e-01 9.37807113e-02 -2.09482342e-01 -1.51170835e-01 7.37543583e-01 2.42106751e-01 -4.07466143e-02 -1.06629789e+00 -3.67905140e-01 6.32045269e-01 -2.14237869e-01 -6.44864321e-01 8.52091968e-01 2.86073565e-01 -2.10806578e-01 4.81765956e-01 5.01615345e-01 -1.46666333e-01 5.89214444e-01 1.03973997e+00 1.59918213e+00 1.93375692e-01 1.29405543e-01 -1.80476025e-01 7.62053132e-01 4.21810180e-01 7.62848882e-03 9.67541039e-01 1.38751239e-01 1.65062010e-01 5.56102931e-01 -8.80356804e-02 -5.73492408e-01 -1.51574361e+00 1.16805173e-01 1.27684724e+00 2.26349235e-01 -7.35019088e-01 -4.38554078e-01 -1.12860465e+00 1.47887707e-01 7.94428885e-01 -4.44635332e-01 4.03541736e-02 -6.37382686e-01 -1.04331180e-01 7.17505395e-01 4.58952159e-01 4.22886848e-01 -9.82316971e-01 2.69763488e-02 3.64082307e-01 -6.80145502e-01 -1.33926010e+00 -2.17251182e-01 -2.40327775e-01 -6.83517039e-01 -1.25081515e+00 -4.28987175e-01 -7.79113054e-01 2.46686384e-01 2.78921574e-01 1.92835391e+00 3.44058543e-01 2.93049395e-01 7.10787177e-01 -6.36700988e-01 -2.67446250e-01 -3.48154277e-01 3.89202207e-01 -5.98952532e-01 -4.42876428e-01 3.56439441e-01 -2.68441588e-01 -8.30668986e-01 2.35372931e-01 -8.21845770e-01 -3.18835884e-01 5.36822021e-01 4.58419532e-01 6.88815951e-01 -4.33430821e-01 9.03435171e-01 -7.91777849e-01 9.21172559e-01 -9.99697566e-01 -6.16910577e-01 9.90159154e-01 -5.83631039e-01 3.28427553e-01 3.61270517e-01 2.72875398e-01 -1.36702609e+00 -7.36134350e-01 -2.77155727e-01 2.77460039e-01 3.62351000e-01 1.08446252e+00 -1.83707085e-02 1.79566726e-01 7.95182824e-01 -1.68160081e-01 -5.99520922e-01 -5.27347088e-01 1.04107189e+00 4.33765769e-01 7.50848711e-01 -1.08283198e+00 8.04559827e-01 3.55098784e-01 -2.54540950e-01 -1.66675910e-01 -1.41611445e+00 -8.91597807e-01 -4.92831975e-01 -2.94813097e-01 7.61726975e-01 -9.52345371e-01 -4.96318638e-01 -5.00591919e-02 -1.24042583e+00 -1.76908284e-01 -3.37629825e-01 7.53255785e-02 -1.58020452e-01 3.79706204e-01 -9.01984632e-01 -4.91129458e-01 -5.18038929e-01 -6.65009916e-01 1.14853513e+00 9.48326960e-02 -1.44631686e-02 -1.00243700e+00 6.34203136e-01 1.01720560e+00 2.65346140e-01 -1.87144011e-01 1.15538847e+00 -9.70456421e-01 -1.03004146e+00 -2.54689723e-01 -4.53275651e-01 -2.01890215e-01 -1.59688681e-01 -2.40714982e-01 -7.96486795e-01 1.82826102e-01 -7.01771379e-01 -9.38560724e-01 8.81458700e-01 -4.23736751e-01 6.68092906e-01 -3.30413252e-01 -2.98762321e-01 -1.70479923e-01 1.43686712e+00 -2.21778035e-01 5.90120137e-01 7.46237755e-01 3.74574423e-01 6.05926335e-01 6.79041624e-01 -6.30683973e-02 1.15691388e+00 4.07212168e-01 5.19465208e-01 3.29945773e-01 -4.19516981e-01 -7.05821455e-01 2.34254137e-01 1.33736324e+00 6.96034580e-02 -6.18214905e-01 -1.07631361e+00 1.02590227e+00 -1.80000353e+00 -7.70511508e-01 -2.44110540e-01 1.88982677e+00 1.11994421e+00 1.41593099e-01 -9.56805795e-03 -3.40270609e-01 2.20651031e-01 1.71889514e-01 -4.57794398e-01 -2.53989875e-01 -2.76023090e-01 5.77426314e-01 1.69423938e-01 6.75431371e-01 -6.46993816e-01 1.03369188e+00 6.41782475e+00 7.76138842e-01 -5.63578829e-02 1.12533808e-01 1.08251832e-01 1.15495920e-01 -7.73018956e-01 2.90578961e-01 -8.64248157e-01 -1.03601895e-01 1.15893304e+00 -3.40116143e-01 2.48111501e-01 5.00723064e-01 -6.20169163e-01 -1.52916178e-01 -8.70339930e-01 5.71224019e-02 2.12998137e-01 -1.53008294e+00 4.92805451e-01 -3.76536667e-01 7.11846948e-01 4.53225493e-01 -2.46519133e-01 6.66182876e-01 1.01164436e+00 -7.97230005e-01 7.19263732e-01 6.70320988e-01 4.29655224e-01 -4.90016222e-01 9.08995867e-01 1.94645628e-01 -1.22880042e+00 2.48589460e-02 -2.15453207e-01 2.92417854e-01 3.07087690e-01 4.36070412e-01 -9.10687506e-01 1.44900846e+00 7.93790102e-01 3.91384155e-01 -7.79174030e-01 1.03346121e+00 -7.60661185e-01 7.80238211e-01 -3.78288716e-01 -3.57857019e-01 3.93825829e-01 1.13628626e-01 6.09603643e-01 1.26429045e+00 1.34708270e-01 2.80014336e-01 -1.21273252e-03 5.68708420e-01 -6.72621131e-01 3.53184521e-01 -3.45645189e-01 9.43929553e-02 7.86378205e-01 1.07997215e+00 -2.25737140e-01 -6.31825566e-01 -7.07260907e-01 6.89126074e-01 1.02751875e+00 2.51722932e-01 -6.34764314e-01 -5.81154823e-01 1.59856789e-02 -2.24516466e-01 5.23466587e-01 -1.73641313e-02 3.12048644e-01 -1.24242711e+00 5.13085902e-01 -1.12905121e+00 1.32831860e+00 -1.01711106e+00 -1.61476314e+00 6.16917431e-01 -5.83308861e-02 -7.83627033e-01 -1.69023499e-01 -3.58752817e-01 -1.94367111e-01 8.09570849e-01 -2.11978078e+00 -1.22350991e+00 -1.57485768e-01 7.80140817e-01 3.09101433e-01 2.25107610e-01 8.36346149e-01 4.61550713e-01 -1.50490895e-01 3.28907877e-01 1.93879064e-02 2.72866458e-01 8.93555462e-01 -1.51208532e+00 5.05243242e-01 9.34408426e-01 6.24063313e-01 9.10758972e-01 4.34748501e-01 -7.47227788e-01 -1.64407718e+00 -8.26577723e-01 1.33214235e+00 -1.12266254e+00 1.41763902e+00 -1.15297578e-01 -1.25794005e+00 6.52751029e-01 6.51832104e-01 -2.30495110e-02 5.41736960e-01 4.57010895e-01 -1.05775726e+00 -1.32840797e-01 -7.67561615e-01 4.98089284e-01 1.11944067e+00 -8.92528594e-01 -1.43586254e+00 4.51408833e-01 1.30945194e+00 -7.18015671e-01 -1.23031926e+00 3.56798679e-01 3.83008152e-01 -5.38627684e-01 1.15259242e+00 -8.58096063e-01 4.33944702e-01 -5.51027298e-01 -2.52544135e-01 -9.81682599e-01 -7.94966593e-02 -5.59504569e-01 -6.66162491e-01 1.16518688e+00 1.10740221e+00 -6.28805757e-01 3.39225858e-01 3.26129705e-01 -2.61136830e-01 -7.31697261e-01 -1.03416097e+00 -6.17933273e-01 1.13518268e-01 -4.05272216e-01 5.92951357e-01 7.26448476e-01 3.53662103e-01 5.63279867e-01 2.96483964e-01 5.29884875e-01 4.33847576e-01 2.66723186e-01 6.41458631e-01 -1.12674916e+00 -4.18961436e-01 -1.01819389e-01 2.55051881e-01 -1.39794445e+00 -1.88689828e-02 -1.13380194e+00 5.52020781e-02 -2.40289235e+00 2.24553555e-01 -5.06256044e-01 -6.33269668e-01 3.92743677e-01 -5.10040343e-01 -1.26844821e-02 1.86095312e-02 2.16424629e-01 -1.63136899e+00 5.00396013e-01 1.21880245e+00 -2.14157909e-01 1.79034501e-01 -5.27407944e-01 -9.02317107e-01 3.79725933e-01 3.65664542e-01 -5.81216574e-01 -4.38054651e-01 -9.07582104e-01 1.09291494e+00 1.82000428e-01 3.94266576e-01 -7.56521821e-01 7.47406065e-01 1.92958236e-01 -3.56572717e-01 -7.04575002e-01 2.23452210e-01 -4.38161135e-01 -4.27326202e-01 1.86072707e-01 -4.37342674e-01 2.80197889e-01 3.23468268e-01 8.16548824e-01 -4.54830825e-01 -1.97737202e-01 -1.09445661e-01 -8.08777735e-02 -7.09888458e-01 1.22591257e-01 -1.53212979e-01 9.07843292e-01 3.19163233e-01 5.80606163e-01 -1.35899043e+00 -4.50655967e-01 -3.44416082e-01 7.30448067e-01 8.17580968e-02 3.54503155e-01 5.08913398e-01 -1.10619533e+00 -8.43518078e-01 -7.00265884e-01 6.15577698e-01 2.04243228e-01 7.89747313e-02 7.70567298e-01 -5.43291390e-01 9.96130586e-01 3.45622808e-01 -1.52833611e-01 -8.52051377e-01 4.81385678e-01 1.05475418e-01 -1.00983834e+00 -4.42429543e-01 8.08091462e-01 -3.98114443e-01 -5.87340415e-01 -2.57615708e-02 -3.84599715e-01 -4.12904948e-01 1.14646358e-02 5.20948231e-01 4.16605532e-01 4.82546747e-01 -1.35715127e-01 -4.51032311e-01 4.32815462e-01 -1.77259803e-01 -3.44244599e-01 1.18154955e+00 -1.92224562e-01 -5.59870958e-01 2.32714996e-01 8.34956706e-01 6.83340728e-01 -5.29246330e-01 -8.15795004e-01 6.09956563e-01 -4.97141108e-02 -1.62731081e-01 -1.40746689e+00 -4.10578668e-01 3.28058124e-01 -2.08305091e-01 2.92922109e-01 8.59161019e-01 7.51286387e-01 9.86616313e-01 1.06809175e+00 5.91311812e-01 -7.30039418e-01 3.36460739e-01 8.99626255e-01 1.08263814e+00 -1.17544127e+00 6.95914868e-03 -5.17378807e-01 -4.37194556e-01 8.81790102e-01 5.06297827e-01 7.32439831e-02 2.83307552e-01 -2.04058200e-01 7.74279609e-02 -9.14950073e-01 -1.22388077e+00 -6.25065267e-01 8.72022629e-01 2.55637795e-01 3.16191137e-01 -3.51050287e-01 -3.17678183e-01 5.50549150e-01 -1.36049286e-01 -7.70304445e-03 2.00345233e-01 1.08505452e+00 -6.53614104e-01 -1.04512370e+00 -1.33432895e-01 2.26277381e-01 -5.19757330e-01 -4.56229299e-01 -6.10702932e-01 8.67280960e-01 -3.67362857e-01 1.27313685e+00 -2.12442219e-01 -6.17388263e-03 7.06580460e-01 3.53678614e-01 4.35102969e-01 -5.72837293e-01 -8.12464416e-01 -5.53303361e-01 9.30174351e-01 -6.39512837e-01 -3.90760750e-01 -3.16260576e-01 -1.20060849e+00 1.44251790e-02 -3.21360826e-01 6.63757503e-01 3.06678325e-01 9.37484026e-01 7.22101390e-01 4.57866669e-01 -4.06334102e-02 5.47667146e-01 -4.23905790e-01 -8.74688745e-01 3.11906524e-02 2.74328947e-01 3.08416307e-01 -4.97299820e-01 -1.36064455e-01 -2.32442170e-01]
[10.757257461547852, 7.910121917724609]
587deed2-62e3-4678-843f-579030e9337d
aio-p-expanding-neural-performance-predictors
2211.17228
null
https://arxiv.org/abs/2211.17228v2
https://arxiv.org/pdf/2211.17228v2.pdf
AIO-P: Expanding Neural Performance Predictors Beyond Image Classification
Evaluating neural network performance is critical to deep neural network design but a costly procedure. Neural predictors provide an efficient solution by treating architectures as samples and learning to estimate their performance on a given task. However, existing predictors are task-dependent, predominantly estimating neural network performance on image classification benchmarks. They are also search-space dependent; each predictor is designed to make predictions for a specific architecture search space with predefined topologies and set of operations. In this paper, we propose a novel All-in-One Predictor (AIO-P), which aims to pretrain neural predictors on architecture examples from multiple, separate computer vision (CV) task domains and multiple architecture spaces, and then transfer to unseen downstream CV tasks or neural architectures. We describe our proposed techniques for general graph representation, efficient predictor pretraining and knowledge infusion techniques, as well as methods to transfer to downstream tasks/spaces. Extensive experimental results show that AIO-P can achieve Mean Absolute Error (MAE) and Spearman's Rank Correlation (SRCC) below 1% and above 0.5, respectively, on a breadth of target downstream CV tasks with or without fine-tuning, outperforming a number of baselines. Moreover, AIO-P can directly transfer to new architectures not seen during training, accurately rank them and serve as an effective performance estimator when paired with an algorithm designed to preserve performance while reducing FLOPs.
['Shangling Jui', 'Wei Lu', 'Jialin Zhang', 'Puyuan Liu', 'Fred X. Han', 'Weichen Qiu', 'Mohammad Salameh', 'Di Niu', 'Keith G. Mills']
2022-11-30
aio-p-expanding-neural-performance-predictors
https://arxiv.org/abs/2211.17228
https://arxiv.org/pdf/2211.17228.pdf
null
['panoptic-segmentation', '2d-human-pose-estimation']
['computer-vision', 'computer-vision']
[ 2.91028917e-01 -6.70279488e-02 -2.65149176e-01 -5.55532753e-01 -6.91420376e-01 -6.46900594e-01 4.10377651e-01 -2.26591099e-02 -3.79633904e-01 4.75866646e-01 -2.91833967e-01 -4.92134839e-01 -2.54134208e-01 -4.74937171e-01 -1.07954109e+00 -4.59494948e-01 -2.02535093e-01 7.16513693e-01 4.23387885e-01 -3.10881902e-02 2.29820147e-01 6.59802139e-01 -1.36457729e+00 4.66195166e-01 5.88271022e-01 1.46384728e+00 3.54222327e-01 8.81295741e-01 1.19360603e-01 8.92072797e-01 -6.22586787e-01 -4.55897659e-01 3.22692484e-01 -6.98943883e-02 -9.01934803e-01 -4.21116054e-01 8.90184999e-01 1.09433882e-01 -2.04642192e-01 7.43966520e-01 2.47225121e-01 1.62809920e-02 9.20491636e-01 -1.32928252e+00 -7.65995204e-01 7.29118288e-01 -3.60581219e-01 3.63277525e-01 -4.45343316e-01 3.52237254e-01 1.21051848e+00 -1.00808537e+00 4.88352507e-01 9.99357164e-01 9.86673117e-01 6.15882576e-01 -1.54069340e+00 -8.31977785e-01 2.52328843e-01 3.08641464e-01 -1.21450293e+00 -6.51034474e-01 7.14195490e-01 -5.38522065e-01 1.38059258e+00 1.26346216e-01 4.59714681e-01 1.29872096e+00 3.03648889e-01 8.47464681e-01 6.43612921e-01 -1.17725559e-01 2.58101016e-01 4.07212898e-02 4.61614311e-01 9.13591325e-01 2.45752886e-01 2.13132039e-01 -5.71795285e-01 1.33389831e-01 5.23139298e-01 -2.05688983e-01 -3.94233793e-01 -5.43714464e-01 -1.06191146e+00 6.72017038e-01 9.63343382e-01 -1.65873058e-02 -3.78082007e-01 5.83481491e-01 6.52753472e-01 6.02605820e-01 2.04342350e-01 9.10798192e-01 -9.74403977e-01 5.62010482e-02 -8.02694976e-01 3.86331826e-02 8.66857409e-01 1.20000243e+00 7.56344616e-01 3.63270372e-01 -3.90771925e-01 8.67605746e-01 1.25376090e-01 1.22222885e-01 6.73516154e-01 -6.70728147e-01 6.60050213e-01 6.97417915e-01 -6.46779299e-01 -8.46166849e-01 -4.79269147e-01 -1.02601159e+00 -9.72730041e-01 1.33099392e-01 5.63481972e-02 -1.73110634e-01 -1.24634051e+00 1.75941288e+00 -1.61682636e-01 3.17522168e-01 1.03128059e-02 7.32468188e-01 9.45752740e-01 6.21503115e-01 1.59726277e-01 2.11976394e-02 1.02724445e+00 -1.43046570e+00 1.02825560e-01 -8.84503603e-01 8.66399765e-01 -3.71568292e-01 1.27279818e+00 4.77099210e-01 -9.48518336e-01 -8.35808635e-01 -1.22335136e+00 -8.41843933e-02 -2.96188295e-01 2.35498339e-01 4.15131092e-01 5.17852366e-01 -1.55075347e+00 8.19791317e-01 -7.68916428e-01 1.36585860e-02 6.83583677e-01 7.80200362e-01 -2.95672923e-01 -1.77010037e-02 -6.92570150e-01 9.22587097e-01 5.53042650e-01 -2.78895460e-02 -1.38693643e+00 -9.93567824e-01 -5.43743849e-01 4.67270166e-01 1.21587202e-01 -8.27725410e-01 1.16004777e+00 -1.04184186e+00 -1.43114567e+00 5.98610938e-01 1.50153413e-01 -8.23087513e-01 2.28831574e-01 -1.93733707e-01 -2.45962963e-01 -3.38019133e-01 -1.41756058e-01 8.81132364e-01 9.53431070e-01 -1.14189827e+00 -5.21044672e-01 -3.51164758e-01 -3.81850660e-01 6.09518141e-02 -6.99959457e-01 -3.82812023e-01 -6.58394754e-01 -4.59586710e-01 -1.00080326e-01 -9.82592404e-01 -3.19778830e-01 -3.06428671e-01 -5.34858525e-01 -2.93361276e-01 7.17939138e-01 -3.14781964e-01 1.09621000e+00 -2.09067321e+00 2.46901333e-01 3.87820452e-01 5.60937226e-01 4.70493495e-01 -7.88703382e-01 -7.31941760e-02 -6.34903908e-02 1.18061319e-01 -8.76428559e-02 -4.28080082e-01 -2.77294129e-01 1.33200198e-01 -1.30154058e-01 3.59326482e-01 4.92385745e-01 1.06606567e+00 -5.95318735e-01 -2.41715863e-01 -1.47310436e-01 2.49076784e-01 -5.85011899e-01 1.89076960e-01 -1.08744517e-01 4.26588692e-02 -2.85454452e-01 5.27649581e-01 2.20971689e-01 -6.50725424e-01 2.40949914e-01 -3.21707666e-01 3.43012035e-01 2.40201637e-01 -7.02047765e-01 1.42390752e+00 -6.04930818e-01 8.53123248e-01 -8.90112519e-02 -1.27248847e+00 1.34991193e+00 -1.51713610e-01 6.39743432e-02 -6.75879955e-01 2.59764761e-01 2.17318892e-01 2.32801586e-01 6.16007484e-03 2.81788915e-01 4.42692488e-01 2.25808993e-01 5.25082387e-02 3.54918778e-01 1.82534158e-01 1.11892112e-01 -1.89430296e-01 1.62962699e+00 -2.77223200e-01 1.42783791e-01 -3.01515132e-01 4.56856489e-01 3.02028835e-01 4.79881823e-01 7.53408253e-01 -7.53564909e-02 4.02622640e-01 4.94051576e-01 -6.30755007e-01 -9.77515697e-01 -9.61425364e-01 6.20713038e-03 1.56629002e+00 -1.84499085e-01 -3.61742854e-01 -5.08789718e-01 -9.39250231e-01 1.61077172e-01 7.87601650e-01 -7.15945244e-01 -6.21011496e-01 -8.10883641e-01 -4.13487643e-01 5.64533412e-01 9.60671961e-01 3.05338800e-01 -1.07639432e+00 -4.62581664e-01 1.75974160e-01 4.56815898e-01 -1.01858389e+00 -4.60012674e-01 7.53963292e-01 -1.21064901e+00 -1.12531209e+00 -5.24694800e-01 -1.25166476e+00 6.01838768e-01 1.29966080e-01 1.55486822e+00 9.69290212e-02 -1.11950897e-01 3.32303047e-02 -6.44537527e-03 -3.78076702e-01 -4.89114434e-01 6.97216332e-01 -6.95633367e-02 -1.51294827e-01 2.96196997e-01 -4.99258280e-01 -6.65490091e-01 4.63250875e-01 -2.61345446e-01 4.72353138e-02 1.14034760e+00 1.04346108e+00 7.68940032e-01 -3.91707927e-01 5.20611465e-01 -1.10419250e+00 8.73401105e-01 -4.93914008e-01 -7.73223460e-01 4.18900400e-01 -1.33422399e+00 3.54797989e-01 1.06270003e+00 -5.84754527e-01 -4.58583862e-01 3.55834812e-01 1.91750318e-01 -1.10086083e+00 5.92568610e-03 5.93880117e-01 1.75110057e-01 -2.14364886e-01 1.21781266e+00 2.51474380e-01 5.61176278e-02 -4.43706214e-01 3.17150444e-01 2.00710461e-01 7.07212746e-01 -3.49709690e-01 5.79125822e-01 -2.81828165e-01 2.54937381e-01 -4.12664890e-01 -7.45696068e-01 -4.17143524e-01 -5.17366827e-01 -1.97143666e-02 4.80546981e-01 -8.03481281e-01 -7.59316981e-01 1.01393588e-01 -9.78656411e-01 -8.15607250e-01 -2.69773025e-02 2.55707800e-01 -2.55637258e-01 -2.51218766e-01 -6.42954528e-01 -2.94788808e-01 -8.11712265e-01 -1.41882956e+00 7.46865094e-01 -4.08336753e-03 -1.64111391e-01 -1.12324035e+00 -7.59768486e-03 1.85153902e-01 5.75706363e-01 8.50369874e-03 1.07788610e+00 -1.01849127e+00 -5.39339840e-01 -2.83643126e-01 -5.32805622e-01 6.36231005e-01 -3.70454431e-01 -1.22472346e-01 -9.56066608e-01 -5.11840820e-01 -3.76339138e-01 -5.84958434e-01 1.14461505e+00 6.05192780e-01 1.53482854e+00 -5.19571424e-01 -5.91119111e-01 9.95385706e-01 1.39575756e+00 5.00780344e-02 3.19443852e-01 5.27455449e-01 8.94475818e-01 2.88312018e-01 2.28391767e-01 -3.57731953e-02 1.75833121e-01 6.15477026e-01 6.96683824e-01 1.08893074e-01 -3.01424950e-01 -1.09367676e-01 4.60920334e-01 8.33426535e-01 8.04974139e-02 -2.28410348e-01 -1.32776308e+00 4.81015861e-01 -1.75403643e+00 -2.88824022e-01 -3.33526492e-01 1.98488510e+00 6.83502614e-01 4.06352818e-01 -1.03923187e-01 -2.18121707e-01 4.63904083e-01 -7.25407153e-02 -9.36622143e-01 -5.25164664e-01 1.01838700e-01 3.92859817e-01 8.66094291e-01 4.47724238e-02 -9.71820176e-01 1.08502257e+00 6.64897013e+00 7.82638967e-01 -1.24840605e+00 9.66056883e-02 9.49866235e-01 -8.75084102e-02 -1.10054100e-02 -2.43778661e-01 -9.49904621e-01 8.05190504e-02 1.35246253e+00 5.31570539e-02 4.46309656e-01 1.54598844e+00 -4.68273938e-01 5.81847370e-01 -1.66249406e+00 1.10530794e+00 -1.34632466e-02 -1.72029448e+00 9.64760631e-02 -6.38738871e-02 8.72440100e-01 6.09984934e-01 3.06195855e-01 8.06245923e-01 3.76826346e-01 -1.34460926e+00 5.98881364e-01 2.35598728e-01 9.81848121e-01 -6.66031301e-01 7.92901874e-01 3.34828854e-01 -1.19207227e+00 -3.79554838e-01 -6.30797803e-01 1.41494676e-01 -5.36178172e-01 2.21849889e-01 -1.39024794e+00 -9.73690376e-02 8.59309793e-01 8.17381203e-01 -9.83392179e-01 9.86637294e-01 -1.67332590e-01 7.47212946e-01 -5.74449375e-02 -5.13131917e-01 3.58037233e-01 1.55654997e-01 1.88286290e-01 1.32488418e+00 3.29672515e-01 -3.26677918e-01 2.41830666e-03 7.85109520e-01 -5.23026824e-01 1.23060346e-01 -4.80247617e-01 1.44437075e-01 6.59810066e-01 1.18652427e+00 -3.73643309e-01 -2.21144736e-01 -2.14141443e-01 7.62705386e-01 9.42766130e-01 3.56771648e-01 -7.47762680e-01 -2.10268438e-01 6.32716417e-01 -1.77608915e-02 3.81558776e-01 5.14802337e-02 -7.18143761e-01 -7.25695431e-01 -3.79687659e-02 -8.38477731e-01 4.51405317e-01 -3.79643530e-01 -1.35056901e+00 9.67538059e-01 -3.25385779e-01 -1.06910372e+00 -1.74317330e-01 -1.03788555e+00 -7.50797868e-01 6.90774500e-01 -1.47791874e+00 -9.95728850e-01 -5.41592538e-01 5.93413711e-01 7.14212656e-01 -9.21622872e-01 6.93731725e-01 1.06289536e-01 -8.06259751e-01 1.10143828e+00 1.20960683e-01 1.65842429e-01 5.32997549e-01 -1.25775719e+00 7.30491281e-01 6.45080149e-01 4.17077571e-01 4.96963233e-01 4.62018102e-01 -3.83198828e-01 -1.66077983e+00 -1.52192485e+00 3.87488067e-01 -4.46214229e-01 6.40454829e-01 -3.11532259e-01 -1.08692300e+00 8.57006073e-01 -9.02926773e-02 2.53469169e-01 4.92618263e-01 7.09017217e-01 -5.07531106e-01 -5.67604482e-01 -5.77938735e-01 4.93023604e-01 1.31741834e+00 -1.67838544e-01 -3.13969374e-01 4.86409724e-01 7.65694439e-01 -4.16792989e-01 -9.31019008e-01 6.37050986e-01 3.16561908e-01 -8.49102199e-01 1.03955209e+00 -8.94266784e-01 6.58504963e-01 1.40225723e-01 1.37464758e-02 -1.51993561e+00 -7.54808486e-01 -2.78476894e-01 -3.85932922e-01 8.54616821e-01 1.11169183e+00 -5.21582007e-01 1.29882514e+00 4.96754974e-01 -7.21401215e-01 -1.09903622e+00 -7.55387306e-01 -9.64781284e-01 7.86496401e-02 -5.53545356e-01 3.60380530e-01 8.46737623e-01 -6.77576900e-01 8.62332523e-01 -2.07696259e-01 1.04226388e-01 4.84969169e-01 -6.35551214e-02 9.42190945e-01 -1.37088430e+00 -5.28923035e-01 -1.04719079e+00 -5.07242620e-01 -1.07504094e+00 1.89458832e-01 -1.20969343e+00 2.10161224e-01 -1.35644186e+00 -4.12145220e-02 -7.12188065e-01 -6.43052518e-01 6.61132991e-01 -3.92011218e-02 4.50379476e-02 9.69169382e-03 5.68100393e-01 -4.89174217e-01 4.71304834e-01 9.91601586e-01 -3.27695131e-01 -2.88044930e-01 1.69999421e-01 -5.48243523e-01 5.40357769e-01 8.31902564e-01 -5.26852012e-01 -7.59377658e-01 -7.60493457e-01 4.93460335e-02 -1.25927314e-01 2.45888621e-01 -1.36565161e+00 4.77213442e-01 4.16404344e-02 6.91327929e-01 -2.62115479e-01 2.93554127e-01 -7.79219627e-01 9.69572738e-03 7.12449193e-01 -6.58704519e-01 4.53769028e-01 4.13927764e-01 8.60160112e-01 -1.12533450e-01 -2.21971303e-01 7.84912407e-01 1.44301340e-01 -1.15016234e+00 4.92951155e-01 1.49247974e-01 6.89969584e-02 1.04707348e+00 -2.28557825e-01 -6.12344503e-01 6.85918182e-02 -5.88691413e-01 4.40962076e-01 1.10687554e-01 4.59334850e-01 1.00869203e+00 -1.15507734e+00 -7.92572260e-01 1.96465492e-01 3.97706598e-01 -5.91370054e-02 -4.88419719e-02 7.04284012e-01 -7.80252576e-01 5.45754850e-01 -3.59077632e-01 -9.17285681e-01 -1.29455471e+00 6.75309300e-01 2.81081468e-01 -5.12009501e-01 -4.58448380e-01 1.46211064e+00 3.87138039e-01 -5.16669571e-01 4.79713768e-01 -2.62398422e-01 -2.49175951e-01 -2.67218590e-01 2.14837626e-01 9.46165025e-02 2.98217028e-01 -1.22750387e-01 -4.43522632e-01 2.83347279e-01 -2.77901173e-01 5.54360569e-01 1.40904534e+00 3.40403974e-01 2.59282529e-01 2.38702461e-01 1.54419661e+00 -7.20580757e-01 -1.29651546e+00 -3.56665105e-01 3.89766693e-01 -2.90963799e-02 2.71529347e-01 -8.01468968e-01 -1.50619566e+00 8.54549229e-01 7.07469523e-01 4.07040305e-02 1.23803961e+00 6.59760684e-02 6.48349643e-01 7.87540853e-01 9.71551836e-02 -9.67929244e-01 3.77309859e-01 6.80588484e-01 1.12497628e+00 -1.21766210e+00 -1.68484136e-01 -2.33064830e-01 -7.87823558e-01 1.30208051e+00 1.23929870e+00 -3.20488900e-01 6.55688345e-01 1.17221348e-01 -1.29011080e-01 -4.01363224e-01 -1.40395534e+00 1.84030712e-01 8.71091783e-01 6.25882208e-01 3.72938156e-01 3.38427126e-02 4.82633442e-01 5.05370438e-01 -3.29706073e-01 -3.50203604e-01 9.82129797e-02 4.86574113e-01 -4.70354557e-01 -7.49437451e-01 1.22202002e-01 1.04073834e+00 1.95439951e-03 -3.21065068e-01 -4.97143358e-01 8.07069361e-01 -3.22017729e-01 4.50949371e-01 1.59175575e-01 -9.70150590e-01 4.46586937e-01 3.21939737e-02 1.42774358e-01 -7.02698052e-01 -8.94507527e-01 -2.75591165e-01 3.47752690e-01 -6.85046256e-01 2.84885317e-01 -3.07235926e-01 -8.45744729e-01 -2.60736763e-01 -2.60362387e-01 -9.16679576e-02 7.72746205e-01 6.65727437e-01 6.67817593e-01 9.18449461e-01 4.49587166e-01 -7.39937305e-01 -8.80279541e-01 -9.73396480e-01 -2.14423403e-01 1.76533401e-01 1.97785735e-01 -5.32409489e-01 -1.63975403e-01 -1.38726667e-01]
[8.736870765686035, 3.313368320465088]
f809cf12-de23-4a0a-a911-a467e0d73b66
late-multimodal-fusion-for-image-and-audio
2204.03063
null
https://arxiv.org/abs/2204.03063v3
https://arxiv.org/pdf/2204.03063v3.pdf
Late multimodal fusion for image and audio music transcription
Music transcription, which deals with the conversion of music sources into a structured digital format, is a key problem for Music Information Retrieval (MIR). When addressing this challenge in computational terms, the MIR community follows two lines of research: music documents, which is the case of Optical Music Recognition (OMR), or audio recordings, which is the case of Automatic Music Transcription (AMT). The different nature of the aforementioned input data has conditioned these fields to develop modality-specific frameworks. However, their recent definition in terms of sequence labeling tasks leads to a common output representation, which enables research on a combined paradigm. In this respect, multimodal image and audio music transcription comprises the challenge of effectively combining the information conveyed by image and audio modalities. In this work, we explore this question at a late-fusion level: we study four combination approaches in order to merge, for the first time, the hypotheses regarding end-to-end OMR and AMT systems in a lattice-based search space. The results obtained for a series of performance scenarios -- in which the corresponding single-modality models yield different error rates -- showed interesting benefits of these approaches. In addition, two of the four strategies considered significantly improve the corresponding unimodal standard recognition frameworks.
['Jorge Calvo-Zaragoza', 'José M. Iñesta', 'Jose J. Valero-Mas', 'María Alfaro-Contreras']
2022-04-06
null
null
null
null
['music-transcription', 'music-information-retrieval']
['music', 'music']
[ 8.53213906e-01 -2.98067510e-01 -7.99115002e-02 1.00524187e-01 -1.22063875e+00 -8.36998701e-01 8.42544854e-01 7.08337501e-02 -3.52235496e-01 4.40403849e-01 3.20622295e-01 6.77593872e-02 -6.91218257e-01 -3.79163772e-01 -2.40450844e-01 -8.38649154e-01 3.29990983e-01 4.27404106e-01 -1.70627534e-01 -1.98872268e-01 3.82404357e-01 4.67228442e-01 -2.07852125e+00 5.66502333e-01 4.68459010e-01 9.85707819e-01 2.87630677e-01 6.97694302e-01 -3.07904482e-01 4.11845297e-01 -4.54306901e-01 -4.95230138e-01 1.75082818e-01 -5.53639233e-01 -6.97036147e-01 1.46541879e-01 3.10836256e-01 4.04727399e-01 -5.52329831e-02 1.10223675e+00 7.99936056e-01 4.06626388e-02 6.66701615e-01 -1.02329373e+00 -2.25591958e-01 7.75187016e-01 -3.21951956e-01 -2.63956130e-01 8.94334614e-01 -2.69913316e-01 1.20464909e+00 -8.72041285e-01 7.92492092e-01 1.07657194e+00 3.61775398e-01 3.74695092e-01 -1.53776991e+00 -2.89323479e-01 -2.52640605e-01 3.88823867e-01 -1.59164059e+00 -7.46598721e-01 7.98380911e-01 -6.37125254e-01 5.18219113e-01 6.60666108e-01 4.53657508e-01 1.08429480e+00 -3.24245572e-01 8.50385427e-01 1.28660941e+00 -8.15093517e-01 1.80974364e-01 9.08332914e-02 -1.03483712e-02 -1.53486714e-01 1.05394185e-01 6.60638958e-02 -1.07916737e+00 -3.34281400e-02 3.82198185e-01 -4.36144054e-01 -3.67133200e-01 -2.73987383e-01 -1.42438471e+00 3.23652804e-01 -1.85917124e-01 9.22932088e-01 -2.39476815e-01 -1.06864505e-01 4.68987316e-01 4.32333261e-01 9.68740135e-02 4.50703561e-01 1.15408540e-01 -4.20792282e-01 -1.32604146e+00 2.74104357e-01 9.18037176e-01 6.28495276e-01 2.87253767e-01 5.92217967e-02 -2.31353104e-01 9.98347819e-01 4.06460285e-01 3.44416648e-01 4.04034287e-01 -6.97177768e-01 5.68962812e-01 2.77150154e-01 3.55408750e-02 -8.99617255e-01 -2.76161849e-01 -6.68539286e-01 -9.58967149e-01 4.00201976e-02 4.56301033e-01 4.15471762e-01 -5.01148641e-01 1.71366608e+00 1.58461794e-01 6.11851700e-02 2.77648330e-01 1.11362445e+00 5.93947411e-01 5.43413699e-01 -1.72498837e-01 -4.88702893e-01 1.50619042e+00 -5.33340871e-01 -8.45372379e-01 2.72875041e-01 2.80929863e-01 -1.28020954e+00 8.51566315e-01 8.63287628e-01 -1.16865623e+00 -5.90819299e-01 -1.06496537e+00 4.17137444e-02 -3.11622351e-01 4.97555435e-01 5.02738394e-02 6.44926846e-01 -1.08190918e+00 5.57064354e-01 -1.82411835e-01 -4.19573784e-01 -1.75299734e-01 1.63030773e-01 -2.98343688e-01 -1.18250422e-01 -9.95188653e-01 6.70523584e-01 4.45821106e-01 2.02192083e-01 -5.07155001e-01 -2.99490124e-01 -3.46272737e-01 8.35119337e-02 4.19591159e-01 -6.69889808e-01 1.10749125e+00 -1.03532517e+00 -1.51042819e+00 1.03769982e+00 -1.22835010e-01 -3.40883017e-01 6.09845757e-01 -1.45169690e-01 -4.73043561e-01 1.01647370e-01 -1.89473465e-01 4.39826488e-01 1.07867944e+00 -1.33229268e+00 -4.31644797e-01 -5.19949794e-01 -2.20936403e-01 5.29320359e-01 -2.25823462e-01 2.09107757e-01 -5.70360005e-01 -9.37222421e-01 2.15254933e-01 -1.06259596e+00 1.80970356e-01 -5.45223892e-01 -4.73911047e-01 1.38885836e-04 2.74661243e-01 -4.80794519e-01 1.44509423e+00 -2.43962407e+00 8.37133110e-01 3.87779653e-01 -1.54169932e-01 1.54658303e-01 -2.85773486e-01 8.16028416e-01 -3.43216479e-01 -2.10256636e-01 -3.12073171e-01 -5.06725550e-01 4.07927126e-01 -5.78537062e-02 -5.68658710e-01 2.28799790e-01 -1.90142229e-01 5.78267455e-01 -5.95017970e-01 -5.25068343e-01 1.72214121e-01 6.21373177e-01 -2.01403186e-01 -3.49154277e-03 -1.53717935e-01 7.95824349e-01 -1.77863121e-01 5.72633564e-01 2.73263752e-01 1.49780840e-01 3.50973934e-01 -3.17810357e-01 -2.44856924e-01 7.30047002e-02 -1.58608711e+00 2.16594434e+00 -3.70996743e-01 6.56122923e-01 8.34756270e-02 -9.09838617e-01 9.59168673e-01 7.51779139e-01 7.66646981e-01 -7.05358207e-01 8.26820210e-02 5.29578745e-01 -2.13808045e-02 -3.48808229e-01 8.97767067e-01 -2.23437086e-01 -1.63861781e-01 4.05067086e-01 5.19775674e-02 -4.72273044e-02 3.94111097e-01 -2.34657690e-01 7.41353691e-01 1.94226250e-01 3.30630869e-01 2.12183729e-01 7.27167487e-01 -9.33400691e-02 -1.12204425e-01 7.52087057e-01 2.33983681e-01 9.55343187e-01 1.56081647e-01 9.98609960e-02 -8.23586285e-01 -1.04646957e+00 -2.10316643e-01 1.01979744e+00 1.95301659e-02 -6.70690238e-01 -7.44693577e-01 5.11714444e-02 -3.24282348e-01 4.93187129e-01 -9.50440839e-02 7.10130408e-02 -3.66018504e-01 -6.31999254e-01 7.66530216e-01 -5.39935865e-02 1.36105597e-01 -1.05133677e+00 -5.91020286e-01 2.40033537e-01 -5.14581919e-01 -1.03923881e+00 -1.46718070e-01 1.60394639e-01 -7.74461865e-01 -8.77309859e-01 -1.11022890e+00 -4.16988432e-01 4.34942171e-03 9.04652029e-02 1.04088449e+00 -3.97259653e-01 -3.29779774e-01 8.84941041e-01 -6.47662699e-01 -1.28491759e-01 -5.30928552e-01 1.36178583e-01 7.30684176e-02 5.81185281e-01 1.37229022e-02 -8.03397000e-01 -3.20140272e-01 1.12171263e-01 -1.46445537e+00 8.95339027e-02 7.87194371e-01 4.71982121e-01 6.70814574e-01 -1.30264059e-01 4.32217240e-01 -3.74925166e-01 8.91332984e-01 -1.63187340e-01 -3.94176364e-01 5.01020968e-01 -4.70770836e-01 1.07640743e-01 3.00301164e-01 -5.24269640e-01 -8.21638942e-01 2.24003285e-01 -1.74430281e-01 -4.97465193e-01 -3.54081899e-01 7.35172570e-01 -2.93082029e-01 8.62899944e-02 5.44117153e-01 5.43170273e-01 -2.93689609e-01 -7.80307293e-01 5.15014052e-01 9.21115875e-01 8.08512151e-01 -7.43223548e-01 5.76270580e-01 1.67555660e-01 2.90374309e-01 -1.03437686e+00 -4.16026711e-01 -7.56298184e-01 -5.22421122e-01 -5.33127546e-01 8.27906013e-01 -7.37374485e-01 -5.97636223e-01 3.21420461e-01 -1.15694058e+00 3.00643802e-01 -5.49567521e-01 5.10234177e-01 -9.12423313e-01 5.25649607e-01 -4.26196337e-01 -1.09716749e+00 -2.22360864e-01 -1.34132612e+00 1.39326560e+00 -1.69118956e-01 -4.41973239e-01 -5.14477193e-01 3.60334277e-01 4.50363457e-01 2.49084517e-01 -8.68612677e-02 7.81178176e-01 -5.61144173e-01 -6.78364635e-01 -2.09238216e-01 4.57939645e-03 2.21376866e-01 -1.70657501e-01 -3.26956004e-01 -1.19432569e+00 -1.36879206e-01 -7.93996379e-02 -1.10070482e-01 9.75154221e-01 9.26365629e-02 7.97558188e-01 1.05927266e-01 7.86863714e-02 2.33704656e-01 1.34127069e+00 1.61079884e-01 8.05436194e-01 2.54044384e-01 3.01495582e-01 7.20430255e-01 6.01786554e-01 5.39778054e-01 -5.87914400e-02 1.49673951e+00 2.40543470e-01 1.90795615e-01 -3.25733066e-01 -1.45787492e-01 5.35681665e-01 1.14174366e+00 -2.21394509e-01 -3.08462292e-01 -7.24975467e-01 1.49716482e-01 -1.93301213e+00 -1.06458819e+00 4.36252207e-02 2.43812370e+00 6.65724337e-01 -2.86237627e-01 2.18633920e-01 7.81832695e-01 6.47127926e-01 2.22174719e-01 -2.81220898e-02 -2.32483149e-01 -3.86803448e-01 2.32173771e-01 -3.59311476e-02 2.50017375e-01 -1.01673007e+00 3.79167140e-01 5.50183725e+00 1.29241276e+00 -1.03099370e+00 1.09835863e-01 -1.01946041e-01 -8.14228281e-02 -3.01469207e-01 -2.61838548e-02 -4.83330250e-01 2.96030909e-01 1.05020595e+00 3.82490568e-02 7.07898438e-01 2.69161403e-01 1.62477911e-01 -8.66773874e-02 -1.25598550e+00 1.68067026e+00 2.31890991e-01 -1.06306744e+00 3.71919334e-01 1.05400346e-01 3.82561475e-01 -3.42841834e-01 2.96080649e-01 -6.77709430e-02 -5.35482526e-01 -9.95090902e-01 1.14439332e+00 9.99685526e-01 9.16163206e-01 -4.30797011e-01 5.14182985e-01 3.19534779e-01 -1.30846536e+00 4.04556207e-02 1.92137629e-01 1.34429246e-01 3.84150624e-01 5.72956860e-01 -5.21146178e-01 1.25552273e+00 3.29716742e-01 6.34162962e-01 -3.70595336e-01 1.32167268e+00 2.03656390e-01 4.68597382e-01 -2.63607204e-01 2.48049006e-01 -1.50933206e-01 -4.50903326e-01 1.08872688e+00 1.30656576e+00 7.22401500e-01 -2.77872950e-01 -5.16784051e-03 7.97383249e-01 1.10017784e-01 4.39513743e-01 -6.09653771e-01 -2.49058917e-01 1.44137785e-01 1.20807528e+00 -7.64352918e-01 -1.75009184e-02 -4.03595716e-02 8.50704074e-01 -2.68401623e-01 3.12705994e-01 -4.97353226e-01 -1.04426846e-01 2.84302235e-01 2.44205650e-02 5.53719699e-02 -2.16482073e-01 -1.70196176e-01 -1.19989288e+00 1.87617496e-01 -1.07269168e+00 3.05730402e-01 -8.91689479e-01 -1.10312057e+00 6.18539989e-01 9.56577659e-02 -1.78476918e+00 -4.62540239e-01 -4.24071699e-01 -1.50224275e-03 7.23534048e-01 -1.24956357e+00 -9.71410155e-01 1.67584587e-02 6.45105958e-01 3.98933321e-01 -1.75982133e-01 9.51689780e-01 8.52854609e-01 -1.82179525e-01 4.15449709e-01 2.98844188e-01 -3.27887028e-01 7.29127169e-01 -1.00006747e+00 -2.93596029e-01 5.93398392e-01 9.56331253e-01 3.32749814e-01 7.24724710e-01 -2.73618698e-01 -1.64148903e+00 -5.77924609e-01 9.63279188e-01 -2.28712812e-01 3.85979444e-01 -7.39299953e-02 -6.30695879e-01 4.86044660e-02 1.27601728e-01 -5.82641065e-01 8.36903870e-01 5.86783178e-02 -3.58797610e-01 -7.82915503e-02 -5.63592851e-01 5.83636463e-01 9.87736702e-01 -9.18657899e-01 -5.75128078e-01 -5.94719686e-02 3.29128742e-01 -2.16085911e-01 -9.50901389e-01 4.47610110e-01 6.66613579e-01 -9.84134853e-01 1.11467338e+00 -2.83176690e-01 2.54794955e-01 -6.87773705e-01 -5.68985999e-01 -9.56198752e-01 2.25078270e-01 -6.77848756e-01 3.72354276e-02 1.41619062e+00 2.42454574e-01 -2.21784890e-01 2.90984273e-01 -5.14883436e-02 -6.06403723e-02 -3.13073665e-01 -1.26089799e+00 -6.53317451e-01 -4.40972716e-01 -9.83635843e-01 3.46560329e-01 6.99233472e-01 1.17663622e-01 5.70590496e-01 -5.11624336e-01 -1.38559192e-01 4.22565758e-01 3.53334665e-01 7.07904279e-01 -1.25571585e+00 -5.60999990e-01 -7.00905800e-01 -6.53744340e-01 -8.73818219e-01 -2.58418061e-02 -1.23665988e+00 1.91175146e-03 -1.22241116e+00 7.03436062e-02 -1.59672275e-01 -3.39058727e-01 1.01758651e-01 4.12294447e-01 4.58895624e-01 7.65029967e-01 5.75053275e-01 -7.03550577e-01 4.48836058e-01 8.66593659e-01 -2.19748527e-01 -2.14607522e-01 4.84468266e-02 -3.11550051e-01 4.84783590e-01 4.28801715e-01 -3.14843386e-01 -3.22002292e-01 -2.36995503e-01 6.64825439e-01 5.25935233e-01 4.87726510e-01 -1.24935758e+00 4.30297852e-01 2.78185219e-01 -5.54252714e-02 -6.96559787e-01 7.29945004e-01 -9.03084815e-01 7.77107775e-01 -4.57006656e-02 -6.11868858e-01 -6.69345781e-02 1.82825909e-03 5.54472566e-01 -6.82123423e-01 -4.26726341e-01 4.25059289e-01 1.73796505e-01 -4.33033735e-01 -2.26494357e-01 -3.96609664e-01 -1.19600244e-01 6.05798900e-01 -4.32857066e-01 1.85355201e-01 -4.66631591e-01 -1.11917412e+00 -4.35364664e-01 1.68040946e-01 4.44940060e-01 5.69434941e-01 -1.28471780e+00 -7.22924769e-01 -8.68883878e-02 2.54104227e-01 -4.89922881e-01 3.92311722e-01 1.27381074e+00 -5.37541993e-02 5.67259073e-01 -4.79968190e-02 -8.23082209e-01 -1.46954441e+00 3.97825837e-01 1.81199908e-01 -3.92130852e-01 -2.73025423e-01 2.31291443e-01 -4.34771739e-02 -1.37517795e-01 5.09279191e-01 6.75165802e-02 -3.99118274e-01 5.54195166e-01 3.95006955e-01 3.54485333e-01 2.82565266e-01 -9.91555393e-01 -9.19846743e-02 7.98540056e-01 4.76306379e-01 -7.99937308e-01 1.09011638e+00 -2.37637684e-01 -2.42378920e-01 8.86804700e-01 9.14256692e-01 2.11893782e-01 -3.45787734e-01 -1.92085862e-01 3.88001561e-01 -4.21270132e-01 -1.49009168e-01 -8.86569142e-01 -6.63735330e-01 9.15031075e-01 8.13273966e-01 3.33399415e-01 1.43677330e+00 -6.28633127e-02 3.52789968e-01 3.38128060e-01 4.91796762e-01 -8.98287058e-01 -1.20997243e-01 3.78524959e-01 1.03667521e+00 -6.79733753e-01 -2.70024985e-01 -2.79604435e-01 -3.64865392e-01 1.27811003e+00 -2.52111614e-01 4.33148503e-01 2.83335119e-01 1.42342955e-01 -7.22737312e-02 -1.01206533e-03 -4.53236699e-01 -6.98641241e-01 6.82971060e-01 2.78727204e-01 5.66050053e-01 9.50546563e-02 -5.10834217e-01 7.37317026e-01 -3.20333362e-01 2.47448370e-01 1.38492361e-01 7.12740958e-01 -1.23033293e-01 -1.52277935e+00 -9.34554279e-01 -9.79932677e-03 -4.04955208e-01 -3.81401367e-02 -6.69504642e-01 4.70636934e-01 1.07892960e-01 9.68859136e-01 -4.74264115e-01 -4.88806635e-01 4.28859085e-01 4.18566614e-01 7.10564613e-01 -3.05906206e-01 -7.07903087e-01 5.70202708e-01 2.06981730e-02 -3.28062862e-01 -9.40358102e-01 -7.68217087e-01 -7.40367353e-01 1.27267599e-01 -3.11529487e-01 2.31035605e-01 8.29576790e-01 9.26561415e-01 1.98636398e-01 5.43532073e-01 3.23335886e-01 -1.16005242e+00 -5.85729837e-01 -8.35761607e-01 -7.50387132e-01 5.66821337e-01 1.31945908e-01 -5.17526090e-01 -1.11859255e-01 2.54325598e-01]
[15.78703498840332, 5.310118675231934]
b7825602-ccd2-4114-80bd-e9864ac19fb7
reinforcement-learning-with-reward-machines
2305.17372
null
https://arxiv.org/abs/2305.17372v1
https://arxiv.org/pdf/2305.17372v1.pdf
Reinforcement Learning With Reward Machines in Stochastic Games
We investigate multi-agent reinforcement learning for stochastic games with complex tasks, where the reward functions are non-Markovian. We utilize reward machines to incorporate high-level knowledge of complex tasks. We develop an algorithm called Q-learning with reward machines for stochastic games (QRM-SG), to learn the best-response strategy at Nash equilibrium for each agent. In QRM-SG, we define the Q-function at a Nash equilibrium in augmented state space. The augmented state space integrates the state of the stochastic game and the state of reward machines. Each agent learns the Q-functions of all agents in the system. We prove that Q-functions learned in QRM-SG converge to the Q-functions at a Nash equilibrium if the stage game at each time step during learning has a global optimum point or a saddle point, and the agents update Q-functions based on the best-response strategy at this point. We use the Lemke-Howson method to derive the best-response strategy given current Q-functions. The three case studies show that QRM-SG can learn the best-response strategies effectively. QRM-SG learns the best-response strategies after around 7500 episodes in Case Study I, 1000 episodes in Case Study II, and 1500 episodes in Case Study III, while baseline methods such as Nash Q-learning and MADDPG fail to converge to the Nash equilibrium in all three case studies.
['Yongming Liu', 'Ufuk Topcu', 'Zhe Xu', 'Yanze Wang', 'Jean-Raphaël Gaglione', 'Jueming Hu']
2023-05-27
null
null
null
null
['q-learning', 'multi-agent-reinforcement-learning']
['methodology', 'methodology']
[-4.53484207e-01 1.70708403e-01 -2.00355023e-01 3.93271625e-01 -1.07474303e+00 -5.85628629e-01 8.31561387e-02 -1.63919225e-01 -8.62181127e-01 1.25125706e+00 -3.55531499e-02 -3.51198584e-01 -7.13929713e-01 -5.61893106e-01 -5.38296759e-01 -9.26627517e-01 -4.68708843e-01 7.33598232e-01 1.83539286e-01 -4.34869826e-01 2.52031624e-01 -8.74171034e-02 -9.38441396e-01 -1.50976598e-01 8.40172410e-01 7.29516923e-01 5.07375658e-01 1.33225989e+00 1.29042342e-01 1.05223560e+00 -7.82380521e-01 -1.09997034e-01 6.39132917e-01 -7.33359575e-01 -7.95452058e-01 1.97216533e-02 -6.44201100e-01 -6.03758454e-01 -4.05157119e-01 1.17704046e+00 5.78594387e-01 5.65003037e-01 6.08304322e-01 -1.56468630e+00 9.45263207e-02 5.73473811e-01 -7.81046271e-01 2.67024469e-02 1.98045105e-01 6.86561227e-01 1.03361225e+00 -1.85323760e-01 3.46987396e-01 1.42193055e+00 4.37470704e-01 9.95948017e-01 -1.11487782e+00 -4.37778801e-01 2.62968123e-01 -6.76156208e-02 -8.48783791e-01 1.57224953e-01 1.85090959e-01 -2.24020392e-01 1.07404149e+00 -3.28112006e-01 1.03696501e+00 5.54929614e-01 8.89535546e-01 8.95096660e-01 1.22676826e+00 -1.72538817e-01 7.65485942e-01 -4.01714027e-01 -3.97570997e-01 7.33773887e-01 -2.82128691e-03 6.49371386e-01 -3.93409520e-01 -3.26810420e-01 1.10900080e+00 2.38974124e-01 3.16449910e-01 -4.01897728e-01 -9.18693721e-01 1.07112360e+00 9.61524323e-02 -2.36296162e-01 -1.13686693e+00 7.56833613e-01 2.83483654e-01 9.68745887e-01 1.21731542e-01 7.80061185e-01 -6.56566858e-01 -5.89700997e-01 -3.88652027e-01 7.69700408e-01 9.52192664e-01 6.57155335e-01 6.72474980e-01 3.74226898e-01 -3.67458105e-01 5.04383385e-01 2.22536519e-01 7.10670531e-01 3.35568935e-01 -1.80860782e+00 6.74065232e-01 1.06003836e-01 9.26844299e-01 -2.75688529e-01 -6.70934439e-01 -4.31266934e-01 -4.81142342e-01 6.33159757e-01 7.38789439e-01 -1.19995475e+00 -5.16806424e-01 1.92689168e+00 1.97553098e-01 3.58674340e-02 6.44776940e-01 8.05625677e-01 -3.54035720e-02 7.68877566e-01 -1.34900451e-01 -7.25093484e-01 8.13463330e-01 -7.94145048e-01 -4.34785277e-01 -3.23570579e-01 5.48694134e-01 -2.67437398e-01 9.42373335e-01 3.18670243e-01 -1.51076007e+00 -1.00617416e-01 -6.63918495e-01 1.11682129e+00 4.09916043e-01 -5.90441346e-01 1.81044549e-01 2.65939742e-01 -1.12685716e+00 8.72174919e-01 -8.80492628e-01 3.01963277e-02 3.23258132e-01 5.42834878e-01 3.50810289e-01 1.60879537e-01 -1.24624550e+00 9.10546362e-01 3.42172921e-01 -1.85121164e-01 -1.72218096e+00 -4.02664751e-01 -3.19525361e-01 2.51217242e-02 1.21448696e+00 -7.48079002e-01 2.14650440e+00 -1.01068389e+00 -2.18671989e+00 1.16290912e-01 3.00099760e-01 -4.36209619e-01 7.15520382e-01 3.41626368e-02 3.73478204e-01 6.25897571e-02 2.88614631e-01 4.44512308e-01 8.77968311e-01 -1.02167380e+00 -9.91446853e-01 -1.50902867e-01 3.97387534e-01 9.50813472e-01 1.52344674e-01 -6.50805905e-02 1.98215216e-01 -1.47427425e-01 -5.49001932e-01 -1.05921352e+00 -8.99640620e-01 -6.40141249e-01 -4.21626717e-02 -5.04141867e-01 -5.74221909e-02 -7.98795521e-02 8.41480434e-01 -1.62714005e+00 1.79729387e-01 3.06974053e-02 2.30407581e-01 -2.22519357e-02 -5.01448452e-01 7.33804345e-01 4.00669664e-01 -8.79646000e-03 1.69405222e-01 -1.14453807e-01 1.82239205e-01 3.77556235e-01 -1.16761979e-02 3.78386468e-01 -1.31975695e-01 1.00345969e+00 -1.23181367e+00 -9.11181271e-02 -1.17580108e-01 -5.86253345e-01 -7.66523957e-01 4.75600302e-01 -4.93683189e-01 1.63797945e-01 -8.36918890e-01 -5.52162714e-02 1.70428082e-01 -1.93281874e-01 3.48807275e-01 9.47810531e-01 -7.08434358e-02 -1.01426750e-01 -1.48443687e+00 1.05522716e+00 -4.88909274e-01 -9.10457503e-03 2.65712172e-01 -9.81094480e-01 6.56938851e-01 4.31150436e-01 8.65571856e-01 -7.38058448e-01 1.48123186e-02 1.26599073e-01 3.07670951e-01 -2.98518211e-01 4.53432173e-01 -6.93741083e-01 -4.13615674e-01 9.51307714e-01 9.38670486e-02 -5.08604884e-01 3.49458218e-01 3.49047422e-01 1.18277550e+00 7.09793717e-02 4.34508860e-01 -1.41446054e-01 1.25880167e-01 1.17937177e-01 7.55119145e-01 1.52030766e+00 -4.68814075e-01 -1.78676888e-01 1.25903523e+00 -1.34097576e-01 -9.43406940e-01 -1.14994240e+00 8.86404753e-01 1.41074097e+00 1.71558931e-01 -1.12799294e-01 -6.64094985e-01 -5.83132744e-01 9.84345004e-02 5.18459320e-01 -5.77411652e-01 -3.31190705e-01 -3.57248843e-01 -5.89742839e-01 -7.77487755e-02 2.26896778e-01 6.59739256e-01 -1.54402280e+00 -1.02137709e+00 7.10052252e-01 2.86176410e-02 -4.91155267e-01 -9.47636902e-01 2.47661769e-01 -8.99421036e-01 -1.23303306e+00 -9.14107978e-01 -4.47990209e-01 5.25356174e-01 6.46637604e-02 8.75069976e-01 -2.24912345e-01 3.52087617e-01 7.37966001e-01 9.84479487e-03 -4.77302700e-01 -6.16113484e-01 1.06396094e-01 2.08094150e-01 -2.44510651e-01 -2.81891137e-01 -1.84767276e-01 -7.57265151e-01 4.33268219e-01 -4.76095885e-01 -1.75650448e-01 3.05619389e-01 9.60058928e-01 4.94223118e-01 2.37230629e-01 1.20675969e+00 -5.72758257e-01 1.30910850e+00 -5.11812329e-01 -1.11044323e+00 1.90391034e-01 -5.97307801e-01 3.31078589e-01 9.57504570e-01 -6.68109894e-01 -9.25628006e-01 2.38478035e-02 2.27566376e-01 -2.27544382e-01 5.27112126e-01 5.56590617e-01 1.56299055e-01 4.14988279e-01 6.11699581e-01 3.00254256e-01 5.39186478e-01 6.54830411e-02 2.86918133e-01 4.29786354e-01 8.43098909e-02 -8.20950449e-01 5.87371945e-01 -2.08762199e-01 1.52181119e-01 -1.69304416e-01 -6.67265475e-01 -2.12199092e-01 -8.22233558e-02 -5.72464049e-01 4.37067747e-01 -8.79755497e-01 -1.64701557e+00 8.51138592e-01 -6.89813316e-01 -1.02739298e+00 -7.90830374e-01 4.40895826e-01 -1.31962252e+00 -9.26837623e-02 -6.84620440e-01 -1.42543399e+00 -2.08984166e-01 -8.56550992e-01 3.40532392e-01 8.12027693e-01 2.64440507e-01 -9.35915530e-01 3.67921233e-01 -1.19899511e-01 3.27874720e-01 -1.92631945e-01 5.84980249e-01 -3.58646005e-01 -3.76955450e-01 1.44964054e-01 3.92856926e-01 1.82480607e-02 -7.34799653e-02 -4.54274565e-01 -7.73071423e-02 -7.79381752e-01 1.49093037e-02 -6.34968281e-01 3.71864855e-01 9.97121274e-01 3.80593508e-01 -7.36125648e-01 6.67339042e-02 4.16062213e-02 1.62115502e+00 9.61885750e-01 3.38530034e-01 5.43912590e-01 -4.77844812e-02 2.49431223e-01 8.08850944e-01 1.01840281e+00 4.62886900e-01 2.25663990e-01 6.49267137e-01 5.47469974e-01 7.21020639e-01 -3.61023575e-01 9.90318537e-01 3.16421777e-01 -6.96782246e-02 -6.73804879e-02 -7.49925375e-01 4.24106926e-01 -2.42542648e+00 -1.19549239e+00 4.78120565e-01 2.46881890e+00 9.29844081e-01 4.25701141e-01 9.05932844e-01 -4.99008954e-01 4.57798094e-01 -1.15636811e-01 -1.40877092e+00 -7.42870629e-01 2.87672997e-01 4.86735962e-02 6.70959592e-01 7.40251422e-01 -6.00569427e-01 1.03609562e+00 6.90469456e+00 6.63125098e-01 -6.08319700e-01 1.63164958e-01 4.68130767e-01 -5.13393641e-01 5.35032414e-02 -1.19131722e-01 -5.66640615e-01 3.29983324e-01 1.07365704e+00 -8.09945285e-01 1.13078463e+00 7.80711532e-01 7.82394171e-01 -4.95925963e-01 -6.10645711e-01 7.53601313e-01 -7.43119240e-01 -1.16890454e+00 -7.33664572e-01 1.97244033e-01 1.14226830e+00 1.07486397e-01 4.08539623e-02 9.36572075e-01 1.63623905e+00 -7.48822212e-01 5.48509240e-01 5.81290960e-01 4.70881760e-01 -1.36810005e+00 8.12243581e-01 7.69537210e-01 -9.69149053e-01 -6.70185268e-01 -3.09344739e-01 -6.04842484e-01 7.32530430e-02 -8.91032889e-02 -8.25060725e-01 1.95812687e-01 2.38995045e-01 2.64742583e-01 1.10666469e-01 9.47854817e-01 -2.84593046e-01 4.50203300e-01 -1.47794634e-01 -7.45220304e-01 6.27277911e-01 -4.91545886e-01 4.59079027e-01 3.93994927e-01 1.74453892e-02 2.41825238e-01 7.88316905e-01 6.73426688e-01 1.76492721e-01 -2.14748736e-02 -3.96102160e-01 -2.07488194e-01 2.71559805e-01 9.97561932e-01 -6.92880929e-01 -1.11006312e-01 -8.35305676e-02 6.90510631e-01 2.11609513e-01 6.34924173e-01 -5.44776976e-01 -4.36881542e-01 1.04226577e+00 -3.27161789e-01 8.12506080e-02 -2.97548532e-01 1.92681432e-01 -8.04729939e-01 -4.40484136e-01 -9.81572986e-01 5.52503049e-01 -7.22830296e-01 -1.24606776e+00 1.19034007e-01 -1.12791233e-01 -1.04744196e+00 -1.26722515e+00 -1.42763555e-01 -7.89937794e-01 7.56591141e-01 -8.77454340e-01 -1.55352250e-01 5.80379307e-01 7.86982238e-01 5.46331465e-01 -6.00826263e-01 4.31634903e-01 -5.65215588e-01 -6.31573915e-01 1.91914916e-01 8.41826439e-01 -8.60277005e-03 1.13021642e-01 -1.58264434e+00 2.04914346e-01 4.47770625e-01 -3.88520390e-01 -8.69826376e-02 6.16772950e-01 -6.10934556e-01 -1.57779574e+00 -6.32797241e-01 -1.50750801e-01 2.20562052e-02 8.94577980e-01 -4.80412766e-02 -3.20869297e-01 4.13339615e-01 1.47240445e-01 -2.30279624e-01 -2.46273372e-02 -4.10411060e-02 4.84107822e-01 -2.13439479e-01 -1.06942701e+00 9.15384412e-01 6.72521770e-01 -3.21863115e-01 -2.65183985e-01 2.19334155e-01 7.76428461e-01 -4.53098387e-01 -5.48276484e-01 -3.46069962e-01 3.10975015e-01 -4.99161720e-01 4.85413969e-01 -1.08903658e+00 3.37198794e-01 7.64389038e-02 1.38216168e-01 -2.18402004e+00 -3.76073003e-01 -1.53121650e+00 -1.39125526e-01 4.53730077e-01 3.43057454e-01 -8.74453425e-01 1.06773746e+00 4.81131881e-01 1.17793918e-01 -8.20277810e-01 -1.29508269e+00 -1.00380599e+00 6.69089973e-01 -5.99075630e-02 2.28505313e-01 1.02930747e-01 1.74632072e-01 4.24041420e-01 -5.27782202e-01 -2.16800749e-01 9.70870674e-01 1.89472884e-01 7.48186648e-01 -6.14939690e-01 -7.57330835e-01 -4.82732832e-01 4.51173335e-01 -1.08826125e+00 3.17739189e-01 -4.34025824e-01 3.87326300e-01 -1.79597437e+00 5.19861758e-01 -2.83994436e-01 -3.76281053e-01 4.15467769e-01 -3.00006092e-01 -5.88524044e-01 7.84317911e-01 2.33887620e-02 -1.19601882e+00 7.22495675e-01 1.77603531e+00 -1.04297325e-02 -8.32294941e-01 6.87117457e-01 -8.15698504e-01 4.57577556e-01 1.11953676e+00 -7.28656769e-01 -6.12686753e-01 1.16205454e-01 3.90524626e-01 1.20075440e+00 -1.66348591e-02 -5.74831545e-01 1.78840086e-01 -1.09630322e+00 -5.26063330e-02 -3.53663117e-01 7.90045634e-02 -3.30998123e-01 -9.23875049e-02 1.25078213e+00 -4.42639172e-01 3.69819015e-01 -3.40449251e-02 7.75956213e-01 3.12658131e-01 -5.92742503e-01 8.03314924e-01 -4.29736555e-01 -4.80143160e-01 4.60351288e-01 -1.18572581e+00 7.15786695e-01 1.14037383e+00 -4.83319089e-02 -6.10983223e-02 -1.12082756e+00 -7.74299204e-01 1.06473184e+00 -4.19701822e-02 3.52964737e-02 7.04372883e-01 -1.10238421e+00 -7.86932170e-01 -2.01667026e-01 -6.52851403e-01 -2.46231630e-01 3.11933517e-01 3.65913630e-01 2.52990294e-02 -5.67402206e-02 -4.06439781e-01 -1.69321850e-01 -7.84907401e-01 2.75873184e-01 1.16950679e+00 -1.00853050e+00 -2.57252753e-01 3.00462604e-01 -6.92132786e-02 -4.23924297e-01 3.67984362e-02 3.07923928e-03 -1.52866133e-02 1.44657820e-01 3.47480595e-01 6.61472917e-01 -6.14402950e-01 6.06675521e-02 -5.97275384e-02 2.41098017e-01 7.59587809e-02 -9.47310865e-01 1.36887443e+00 -2.13585988e-01 3.33759874e-01 3.98080736e-01 8.05147827e-01 -6.95436001e-01 -1.99515760e+00 -3.10760081e-01 -5.85102476e-02 -2.05746770e-01 -1.66138098e-01 -8.19315493e-01 -1.09380877e+00 4.21124518e-01 4.94207025e-01 3.11902761e-01 8.36323023e-01 -3.02797228e-01 3.93835366e-01 6.50084078e-01 8.89127851e-01 -1.61905909e+00 8.19372535e-01 1.05296016e+00 6.24302089e-01 -9.31396723e-01 -2.39867792e-01 7.09963441e-01 -1.31754422e+00 8.27950120e-01 8.54552925e-01 -4.66751635e-01 6.04532599e-01 1.83265790e-01 1.01744188e-02 5.62123805e-02 -1.41071868e+00 -4.63369638e-01 -5.03352404e-01 6.91781640e-01 -3.74333680e-01 3.76007259e-01 -1.89328641e-01 6.29671931e-01 7.21060019e-03 1.53393567e-01 9.95838046e-01 1.08173954e+00 -7.51854241e-01 -1.05410874e+00 -3.69633019e-01 7.22482979e-01 -3.54642570e-01 3.58015537e-01 4.86306958e-02 5.10114372e-01 -5.81028700e-01 1.12915564e+00 1.34772524e-01 -1.21349305e-01 4.52375531e-01 -1.25383049e-01 7.98952222e-01 -6.59943521e-01 -7.41948485e-01 2.70958960e-01 -2.66865343e-01 -6.14350736e-01 -9.72064659e-02 -7.50417590e-01 -1.69017375e+00 -4.73910302e-01 -5.92104085e-02 3.56028497e-01 1.25256538e-01 1.03350770e+00 1.53873667e-01 4.51728493e-01 1.16182017e+00 -4.29843813e-01 -1.53105903e+00 -6.67934239e-01 -1.01591861e+00 3.74802575e-02 4.15872872e-01 -5.89581013e-01 -1.50375679e-01 -6.64581597e-01]
[4.167448043823242, 2.534403085708618]
ae95d78c-3f2f-46b7-9c4f-31a2f13f79e2
multisum-a-dataset-for-multimodal
2306.04216
null
https://arxiv.org/abs/2306.04216v1
https://arxiv.org/pdf/2306.04216v1.pdf
MultiSum: A Dataset for Multimodal Summarization and Thumbnail Generation of Videos
Multimodal summarization with multimodal output (MSMO) has emerged as a promising research direction. Nonetheless, numerous limitations exist within existing public MSMO datasets, including insufficient upkeep, data inaccessibility, limited size, and the absence of proper categorization, which pose significant challenges to effective research. To address these challenges and provide a comprehensive dataset for this new direction, we have meticulously curated the MultiSum dataset. Our new dataset features (1) Human-validated summaries for both video and textual content, providing superior human instruction and labels for multimodal learning. (2) Comprehensively and meticulously arranged categorization, spanning 17 principal categories and 170 subcategories to encapsulate a diverse array of real-world scenarios. (3) Benchmark tests performed on the proposed dataset to assess varied tasks and methods, including video temporal segmentation, video summarization, text summarization, and multimodal summarization. To champion accessibility and collaboration, we release the MultiSum dataset and the data collection tool as fully open-source resources, fostering transparency and accelerating future developments. Our project website can be found at https://multisum-dataset.github.io/.
['Lijuan Wang', 'Ding Zhao', 'Bo Li', 'JianFeng Wang', 'Linjie Li', 'Zhengyuan Yang', 'Claire Jin', 'Karthik Mittal', 'Aditesh Kumar', 'William Han', 'Jiacheng Zhu', 'JieLin Qiu']
2023-06-07
null
null
null
null
['text-summarization']
['natural-language-processing']
[ 2.80061990e-01 -1.07383892e-01 -4.51930910e-01 -8.79531503e-02 -1.07937002e+00 -7.23880768e-01 5.53146541e-01 2.26553977e-01 -2.85355330e-01 6.58696830e-01 5.89579284e-01 -1.48240358e-01 -2.01489069e-02 -2.17506468e-01 -4.31679815e-01 -4.54367965e-01 1.30555615e-01 1.21679768e-01 1.27545267e-01 -1.06764689e-01 6.74835503e-01 1.20229099e-03 -1.77413356e+00 5.46235144e-01 1.34964597e+00 8.08007002e-01 2.62050837e-01 8.21294546e-01 -5.86866885e-02 6.20051920e-01 -8.39033842e-01 -5.32976925e-01 -2.32400954e-01 -5.66999555e-01 -8.94535601e-01 1.85348257e-01 8.06495547e-01 -3.96394610e-01 -4.36899215e-01 8.88399065e-01 8.45555305e-01 3.22513849e-01 7.78542280e-01 -1.42620087e+00 -8.37003469e-01 7.74125874e-01 -4.70623374e-01 1.09592572e-01 8.16436887e-01 2.64793038e-01 1.04560649e+00 -7.08682239e-01 8.33543479e-01 9.38697219e-01 4.39684987e-01 5.86579561e-01 -6.71727538e-01 -6.19749188e-01 8.34891796e-02 3.26116204e-01 -1.11992288e+00 -6.23302162e-01 6.09759450e-01 -3.63429815e-01 7.67386198e-01 8.00274014e-01 5.60140073e-01 1.47700155e+00 -8.75855610e-02 1.24985957e+00 8.72670710e-01 -1.88643545e-01 3.69053558e-02 1.87716171e-01 4.20203805e-01 4.92017120e-01 1.72076538e-01 -5.79715788e-01 -7.13015676e-01 1.55520692e-01 1.19059443e-01 1.04806880e-02 -5.41149259e-01 -6.76388815e-02 -1.54247129e+00 5.03027797e-01 -5.22268042e-02 1.74168110e-01 -5.76206110e-02 -3.03537816e-01 7.51691878e-01 1.13827735e-01 2.65643060e-01 3.01025659e-01 -1.00491367e-01 -5.26757538e-01 -1.10408497e+00 2.57204533e-01 7.80089796e-01 1.20863986e+00 4.05118883e-01 -2.29875743e-02 -3.59862804e-01 1.02540982e+00 -3.92743982e-02 5.98778725e-01 8.27101052e-01 -1.23758996e+00 9.82324660e-01 7.85164177e-01 -6.76241368e-02 -1.20271695e+00 -3.80862653e-01 -3.77679765e-02 -9.46164310e-01 -5.68811119e-01 8.67290720e-02 -1.65126637e-01 -6.48676932e-01 1.51325929e+00 1.11375429e-01 -1.32723406e-01 2.39652097e-01 8.79147947e-01 1.81148160e+00 9.12159562e-01 -1.11008503e-01 -3.49633127e-01 1.26570261e+00 -1.17589176e+00 -9.52377915e-01 4.55341861e-03 6.57634258e-01 -9.32052791e-01 1.21432197e+00 3.41218442e-01 -1.13927519e+00 -4.15730238e-01 -8.63195181e-01 -2.00365528e-01 -5.98986864e-01 3.22719842e-01 4.33813065e-01 4.84179258e-01 -1.09940934e+00 2.83412516e-01 -4.97274518e-01 -8.26140583e-01 3.55107725e-01 3.00061926e-02 -5.19623458e-01 -9.65020731e-02 -1.06014001e+00 5.18600464e-01 6.42359376e-01 1.14416815e-01 -6.14078701e-01 -4.76159185e-01 -9.58713591e-01 -1.93000600e-01 5.92107713e-01 -6.33482635e-01 1.25284791e+00 -6.27327442e-01 -1.17489457e+00 8.31091762e-01 -1.09831735e-01 -1.42855763e-01 5.85058987e-01 -1.46894127e-01 -3.37257802e-01 5.50631464e-01 1.94650814e-01 7.92336941e-01 5.67535460e-01 -1.28716755e+00 -7.20582247e-01 -2.01008141e-01 -1.50844529e-01 6.46742046e-01 -7.15358317e-01 -6.82590827e-02 -7.66683102e-01 -8.40973556e-01 -1.29313335e-01 -7.53778756e-01 1.18943945e-01 -7.54917145e-01 -7.10706234e-01 -2.05551401e-01 8.80520642e-01 -1.05568504e+00 1.77020216e+00 -2.20556283e+00 3.84609789e-01 -4.94056612e-01 3.68679136e-01 2.49839187e-01 -2.23025084e-01 8.24308157e-01 1.94375589e-01 3.54249328e-01 -3.45157027e-01 -4.57109004e-01 2.55714029e-01 -2.43796781e-01 -5.47315143e-02 1.76971138e-01 -8.47470611e-02 9.58946109e-01 -1.02592456e+00 -9.23338175e-01 1.87438935e-01 1.22169137e-01 -1.83829755e-01 7.85530284e-02 -3.84724177e-02 4.03011739e-01 -4.40255702e-01 1.05431604e+00 6.03376389e-01 -1.35030717e-01 -5.71828969e-02 -3.37594509e-01 -1.77498817e-01 -1.07427403e-01 -1.12359750e+00 1.61655772e+00 -3.03094983e-02 1.03016281e+00 -4.12507877e-02 -8.60102296e-01 5.86735189e-01 3.40127081e-01 4.83944714e-01 -6.09245479e-01 1.43844694e-01 1.51306316e-01 -5.54565787e-01 -9.65525746e-01 1.16747725e+00 5.45213938e-01 -3.57973069e-01 3.89209330e-01 1.55840531e-01 -1.40210405e-01 7.33210325e-01 7.35728741e-01 9.05447423e-01 -6.06854400e-03 2.83278137e-01 9.80400518e-02 3.99153411e-01 2.94713050e-01 3.39829147e-01 7.17258751e-01 -4.87080812e-01 8.00691009e-01 5.55602014e-01 1.44141659e-01 -8.74611676e-01 -7.97424912e-01 -3.20728645e-02 1.15250254e+00 2.06491902e-01 -6.56872809e-01 -8.35527956e-01 -6.18507266e-01 -1.62818462e-01 5.50823390e-01 -4.04125333e-01 4.40027826e-02 -3.07920396e-01 -6.60548568e-01 7.36705720e-01 3.76365840e-01 6.65626287e-01 -1.11541796e+00 -4.10928339e-01 -2.62795955e-01 -9.46075320e-01 -1.17786133e+00 -7.70662010e-01 -3.96853268e-01 -8.13364863e-01 -1.19322765e+00 -9.98365462e-01 -8.86140466e-01 5.20149112e-01 6.89842820e-01 9.23324287e-01 -8.85876119e-02 -1.49560601e-01 9.36230659e-01 -6.39126062e-01 -1.63301602e-01 -4.38739866e-01 2.58879304e-01 -1.24097049e-01 -2.04379857e-01 1.22678399e-01 -8.48476589e-02 -6.31703377e-01 3.23442012e-01 -1.14547873e+00 2.68692970e-01 5.79052389e-01 5.95463336e-01 3.20990890e-01 -9.68593210e-02 8.48215401e-01 -6.91998482e-01 9.34716880e-01 -5.99267662e-01 -1.10370614e-01 3.79426301e-01 -4.42706704e-01 -5.98992348e-01 2.62370348e-01 -2.53349811e-01 -1.11807859e+00 -2.94808149e-01 8.71583447e-02 -8.14431831e-02 -3.20284605e-01 7.17762470e-01 -1.49741188e-01 3.35068703e-01 2.91640073e-01 4.26532328e-01 -1.01828784e-01 -2.85028845e-01 5.23082674e-01 1.11434877e+00 7.72734165e-01 -4.32553440e-01 2.99099177e-01 1.24556243e-01 -5.05997300e-01 -1.16886938e+00 -5.94044328e-01 -6.11780882e-01 -6.51008666e-01 -6.36519313e-01 8.43318820e-01 -1.02072501e+00 -5.59862137e-01 6.63512528e-01 -1.01903784e+00 -2.08045706e-01 1.37471408e-01 4.05389339e-01 -2.93314397e-01 9.21303511e-01 -5.85376024e-01 -6.24128759e-01 -5.88141263e-01 -1.12351751e+00 9.42747474e-01 5.90597808e-01 -4.66599107e-01 -8.94901693e-01 -1.71261832e-01 1.15837371e+00 2.18658954e-01 5.12310147e-01 5.88996410e-01 -7.69746661e-01 -4.78601992e-01 -3.37942481e-01 -2.37210736e-01 2.48412013e-01 -7.97356889e-02 6.50299609e-01 -7.76172161e-01 -2.52092391e-01 -3.78928721e-01 -6.34252071e-01 1.02452099e+00 3.22182477e-01 1.07233799e+00 -3.66185635e-01 -1.92813724e-01 2.21098214e-01 9.14488494e-01 5.45732342e-02 3.69824022e-01 4.51146603e-01 8.22368979e-01 7.53190279e-01 7.84954548e-01 5.88567555e-01 8.71263325e-01 2.91152596e-01 3.23515296e-01 2.36279577e-01 -2.01531604e-01 -7.47464448e-02 4.58499968e-01 1.56322837e+00 -1.05164424e-01 -5.66767752e-01 -9.98224080e-01 6.66652560e-01 -2.01927471e+00 -1.01886702e+00 -4.81210411e-01 1.96107638e+00 7.64834166e-01 -2.44246274e-01 2.81318814e-01 -1.28053529e-02 8.49935353e-01 5.13558507e-01 -3.75465095e-01 -3.18583041e-01 -3.38726878e-01 -6.31926477e-01 9.71060395e-02 2.25865066e-01 -1.09316599e+00 7.80387044e-01 6.23199558e+00 1.04074502e+00 -9.87026572e-01 4.58868444e-02 5.40579677e-01 -2.51893848e-01 -2.77748227e-01 -4.18604553e-01 -6.50860965e-01 7.10730255e-01 9.36988711e-01 -5.72188616e-01 2.36543387e-01 5.30341029e-01 3.38135153e-01 -2.77714133e-01 -9.32660758e-01 1.15364087e+00 5.81412792e-01 -1.30932307e+00 2.05575749e-01 -1.97904348e-01 9.86445189e-01 -1.03320144e-01 2.77113825e-01 5.76988816e-01 -1.72189623e-01 -8.63423705e-01 7.26838827e-01 4.57043111e-01 7.83158779e-01 -6.06975138e-01 8.25540721e-01 2.81952649e-01 -1.06722617e+00 -1.00400776e-01 7.67657952e-03 1.86128274e-01 5.23375757e-02 2.02926576e-01 -3.95930588e-01 9.29098427e-01 7.31619358e-01 1.01968670e+00 -9.59154427e-01 1.10996079e+00 1.28207773e-01 5.39865792e-01 8.89664814e-02 -1.98759660e-01 2.14835316e-01 -2.92221338e-01 6.77047729e-01 1.61594307e+00 2.98327714e-01 4.92645726e-02 1.77406609e-01 1.97419584e-01 -2.70275235e-01 4.16674703e-01 -6.62715495e-01 -4.54753786e-01 6.21844947e-01 1.33680606e+00 -8.71584475e-01 -5.07209837e-01 -4.96484607e-01 8.72019529e-01 3.27923894e-02 5.90904891e-01 -8.68448317e-01 -5.76509595e-01 1.64893657e-01 -3.00419092e-01 -6.57702312e-02 -2.00328782e-01 -4.19317633e-01 -1.38636947e+00 3.92512279e-03 -1.18904305e+00 7.72144020e-01 -7.21923113e-01 -9.80828583e-01 4.12910372e-01 3.48929435e-01 -1.34737253e+00 -4.18565162e-02 -2.52817750e-01 -5.57683706e-01 3.20151001e-01 -1.22513306e+00 -1.16035306e+00 -7.36953020e-01 3.14039469e-01 9.59614992e-01 -3.28992695e-01 4.32776332e-01 5.32443106e-01 -1.18239105e+00 6.00912452e-01 2.63204962e-01 -1.09066933e-01 1.15205038e+00 -1.07841575e+00 -5.86113296e-02 7.96443045e-01 -2.34292865e-01 5.46203911e-01 6.71093762e-01 -6.24146938e-01 -1.54670596e+00 -1.01931572e+00 7.72995949e-01 -7.06793368e-01 7.84590185e-01 -6.26623556e-02 -8.19794238e-01 6.98471248e-01 6.07209444e-01 -7.71234751e-01 8.72779250e-01 -2.45656163e-01 1.25758946e-01 1.04366817e-01 -8.86277318e-01 8.26765537e-01 9.20678139e-01 -3.31010699e-01 -6.58420682e-01 4.43354368e-01 7.88615704e-01 -6.31212473e-01 -9.39707994e-01 4.01924431e-01 5.07710397e-01 -9.44917917e-01 6.22134924e-01 -3.50615263e-01 9.42282677e-01 4.43418743e-03 -1.11957446e-01 -1.06856501e+00 2.09001496e-01 -7.15547144e-01 -3.60381693e-01 1.81845224e+00 4.11829591e-01 -5.15261292e-01 5.63386858e-01 6.18911564e-01 -5.18966019e-01 -8.23142350e-01 -6.73863173e-01 -4.40377146e-01 -8.19863677e-02 -4.13561195e-01 3.36499482e-01 1.04834950e+00 3.00630122e-01 3.94544542e-01 -4.73758072e-01 -6.24603368e-02 4.68170375e-01 1.82763845e-01 9.54638064e-01 -7.81092823e-01 2.93117225e-01 -9.03319657e-01 -1.66532412e-01 -1.01255035e+00 9.65991020e-02 -1.03705716e+00 -2.44262263e-01 -1.95225465e+00 7.21395552e-01 2.41108671e-01 1.10826388e-01 3.18227082e-01 -4.42571789e-01 4.49443012e-01 2.28508934e-01 4.23202723e-01 -1.30685508e+00 6.86447263e-01 1.27593958e+00 -2.26452425e-01 -1.76280931e-01 -1.71685606e-01 -9.19687390e-01 5.91622055e-01 8.36379051e-01 -4.49021533e-02 -3.74418020e-01 -4.62470800e-01 4.22756709e-02 1.75694391e-01 1.97976291e-01 -8.67101014e-01 3.05384308e-01 -2.84190625e-01 1.00233354e-01 -1.00893974e+00 2.30421871e-01 -3.69939983e-01 -9.87585858e-02 9.80731025e-02 -5.13849497e-01 1.87835187e-01 2.07316324e-01 4.11423117e-01 -4.98214781e-01 -4.24195409e-01 3.23057383e-01 -5.71384619e-04 -1.11194205e+00 1.71654448e-01 -4.12095100e-01 2.71652609e-01 1.14385295e+00 -4.43738908e-01 -7.62491524e-01 -5.98470390e-01 -5.19449890e-01 7.62398064e-01 6.63563967e-01 5.63794494e-01 7.04288185e-01 -1.20408726e+00 -8.15987587e-01 -3.35493594e-01 2.74805695e-01 -1.54659441e-02 7.29773223e-01 1.03657913e+00 -6.36738777e-01 5.75686157e-01 -2.43977889e-01 -6.20012045e-01 -1.58745873e+00 2.91651815e-01 -2.37129018e-01 1.17667660e-01 -6.60820603e-01 3.76548648e-01 -2.24037841e-01 -6.37559593e-01 5.14410734e-01 -1.67823195e-01 -5.92213631e-01 6.01186574e-01 5.05193412e-01 8.70874822e-01 -1.60105556e-01 -8.27516079e-01 -7.43403286e-02 3.45971674e-01 8.08002204e-02 2.76887249e-02 1.12713647e+00 -6.61417007e-01 -1.45307526e-01 7.40657806e-01 1.09578729e+00 2.93609034e-02 -9.24914896e-01 6.48861751e-02 -1.24449357e-02 -3.19588155e-01 -3.26890647e-01 -7.67799139e-01 -7.08964705e-01 6.69102371e-01 2.35331953e-01 4.91715819e-01 1.17353189e+00 -2.80524623e-02 9.61848497e-01 3.69708210e-01 -8.71008933e-02 -1.30750608e+00 3.02450806e-01 5.08331060e-01 9.46856499e-01 -1.50374687e+00 1.35406896e-01 -2.52522677e-01 -1.20270205e+00 1.02806246e+00 6.18362367e-01 5.91029048e-01 3.15239429e-02 -2.33060718e-01 1.72319904e-01 -6.70360178e-02 -7.70371139e-01 -3.53652202e-02 4.88208413e-01 5.03586531e-01 6.38481855e-01 5.73889725e-02 -6.05127811e-01 8.34608138e-01 -2.67639577e-01 -1.69838846e-01 7.94897914e-01 9.39548612e-01 -3.84604871e-01 -6.34581506e-01 -4.42246467e-01 6.03982866e-01 -3.72018367e-01 1.31573349e-01 -5.83157837e-01 8.93560231e-01 -3.24748605e-01 1.20019460e+00 -2.24214867e-01 -3.58675480e-01 3.54336858e-01 5.17064743e-02 1.35987759e-01 -4.26353931e-01 -3.17358255e-01 -7.44605586e-02 2.59893209e-01 -3.68382037e-01 -4.83379275e-01 -7.65004039e-01 -1.00683308e+00 -5.08683026e-01 6.67093992e-02 1.60584450e-01 7.15181887e-01 7.17242658e-01 4.99682456e-01 5.58867335e-01 4.61005151e-01 -1.11795819e+00 -3.57394159e-01 -1.05478144e+00 -2.51079500e-01 5.30709088e-01 7.63452649e-02 -2.67521530e-01 -3.94429028e-01 3.24956357e-01]
[10.685943603515625, 0.6552039980888367]
881e8487-5736-489b-80f9-798dc4d3ce96
evaluation-of-a-canonical-image
2304.09243
null
https://arxiv.org/abs/2304.09243v1
https://arxiv.org/pdf/2304.09243v1.pdf
Evaluation of a Canonical Image Representation for Sidescan Sonar
Acoustic sensors play an important role in autonomous underwater vehicles (AUVs). Sidescan sonar (SSS) detects a wide range and provides photo-realistic images in high resolution. However, SSS projects the 3D seafloor to 2D images, which are distorted by the AUV's altitude, target's range and sensor's resolution. As a result, the same physical area can show significant visual differences in SSS images from different survey lines, causing difficulties in tasks such as pixel correspondence and template matching. In this paper, a canonical transformation method consisting of intensity correction and slant range correction is proposed to decrease the above distortion. The intensity correction includes beam pattern correction and incident angle correction using three different Lambertian laws (cos, cos2, cot), whereas the slant range correction removes the nadir zone and projects the position of SSS elements into equally horizontally spaced, view-point independent bins. The proposed method is evaluated on real data collected by a HUGIN AUV, with manually-annotated pixel correspondence as ground truth reference. Experimental results on patch pairs compare similarity measures and keypoint descriptor matching. The results show that the canonical transformation can improve the patch similarity, as well as SIFT descriptor matching accuracy in different images where the same physical area was ensonified.
['John Folkesson', 'Jun Zhang', 'Yiping Xie', 'Li Ling', 'Weiqi Xu']
2023-04-18
null
null
null
null
['template-matching']
['computer-vision']
[ 1.73192739e-01 -3.63075256e-01 6.56558514e-01 -5.86084366e-01 -3.83634001e-01 -5.51073611e-01 3.64720047e-01 -3.26435976e-02 -9.03265774e-01 3.47593963e-01 -2.84248102e-03 2.58816838e-01 -3.26286778e-02 -1.15428197e+00 -7.53724158e-01 -6.84757590e-01 5.31551391e-02 2.73477495e-01 7.04129875e-01 -6.93088531e-01 4.34696019e-01 6.72351360e-01 -1.65847933e+00 -3.96998793e-01 7.38489568e-01 9.67767477e-01 3.82709086e-01 4.28213865e-01 -1.22666411e-01 -1.63175717e-01 -5.88401973e-01 -3.52669239e-01 8.69548798e-01 -1.69971034e-01 2.43441880e-01 -1.21865883e-01 9.03922975e-01 -5.41702271e-01 -3.36426288e-01 1.55336845e+00 5.13469100e-01 3.02399606e-01 4.81698632e-01 -1.05078018e+00 -7.07264468e-02 -5.65889897e-03 -7.99689591e-01 -2.55607903e-01 2.80518055e-01 -2.22663343e-01 6.27608180e-01 -7.99702823e-01 4.75013196e-01 1.18323183e+00 1.14512622e+00 4.90575731e-02 -7.05889583e-01 -7.48407364e-01 -5.69638073e-01 3.65557045e-01 -1.45980036e+00 -2.15214804e-01 6.92754626e-01 -2.44302675e-01 5.88014841e-01 3.10161114e-01 9.44347382e-01 2.20633909e-01 5.96022785e-01 2.03921851e-02 1.07477665e+00 -4.88577008e-01 1.18672207e-01 -1.02767371e-01 -2.37983674e-01 4.87297684e-01 6.89682364e-01 1.79887041e-01 -4.06797081e-01 -8.64934251e-02 7.70102024e-01 4.21478271e-01 -8.33558738e-01 -7.18715370e-01 -8.33591878e-01 7.91912735e-01 4.37642157e-01 -1.50793940e-02 -1.89096838e-01 -3.15458737e-02 2.12359801e-01 3.76167297e-01 -1.60316721e-01 4.36823279e-01 -1.39651462e-01 2.05818620e-02 -4.25120860e-01 1.57786265e-01 5.74060798e-01 1.21727180e+00 1.17754364e+00 1.47268683e-01 7.04556525e-01 8.88831556e-01 7.32783377e-01 1.34107041e+00 5.08397043e-01 -6.63705289e-01 2.48482555e-01 3.82194668e-01 2.15025797e-01 -1.33598256e+00 -5.60741127e-01 -1.78118691e-01 -7.15786457e-01 5.14767587e-01 1.96775720e-01 -4.01146114e-02 -8.57601643e-01 1.03360784e+00 3.07312548e-01 5.32702953e-02 5.47402143e-01 1.21614254e+00 9.42563534e-01 7.46435523e-01 -6.95104837e-01 -2.63097048e-01 1.29242194e+00 -4.08824086e-01 -9.20289338e-01 -4.84509349e-01 3.40411305e-01 -1.03041935e+00 6.40511692e-01 2.27548152e-01 -6.40992105e-01 -3.90007585e-01 -1.38471115e+00 2.68190205e-01 -1.36651367e-01 -1.57489434e-01 1.48861974e-01 3.96337032e-01 -7.75047362e-01 2.55381316e-01 -7.20899820e-01 -4.71466810e-01 -2.67951101e-01 -1.54228359e-01 -5.96934259e-01 -2.88616598e-01 -1.32308388e+00 9.81670201e-01 -3.55911404e-02 3.13775212e-01 -3.20184469e-01 -4.89409268e-01 -1.32382619e+00 -1.56236604e-01 -3.95261310e-02 -1.10811710e-01 8.79169583e-01 -5.34680903e-01 -1.32181609e+00 6.20524883e-01 1.80269346e-01 -4.15343374e-01 4.15660024e-01 -9.64151397e-02 -5.83517790e-01 1.05755441e-01 1.62632048e-01 1.34627432e-01 5.62473297e-01 -1.34745038e+00 -8.74563277e-01 -6.83443129e-01 -3.61310959e-01 6.84059322e-01 1.64110377e-01 -4.12583828e-01 -3.93725276e-01 -1.33773655e-01 1.09920204e+00 -8.16052079e-01 -1.25227913e-01 2.89100587e-01 1.11985944e-01 3.83439690e-01 8.05440307e-01 -3.16971600e-01 5.69617808e-01 -2.32713580e+00 -2.62166411e-01 3.30810189e-01 -4.56370592e-01 2.33812872e-02 -2.18410213e-02 8.03704262e-01 5.17446518e-01 -4.29369748e-01 -2.40355358e-01 -1.03007495e-01 -2.43018642e-01 6.18254900e-01 -4.53309864e-02 9.60348845e-01 -6.30800188e-01 1.62469611e-01 -7.30157316e-01 -2.51657546e-01 3.44374478e-01 3.84552449e-01 -2.75506616e-01 3.31166208e-01 4.63766217e-01 -9.36007202e-02 -4.30791155e-02 5.69795012e-01 1.45047784e+00 9.24317777e-01 -1.48016572e-01 -7.34548926e-01 -7.37909198e-01 -2.55999476e-01 -1.56493306e+00 1.57423294e+00 -3.74458492e-01 8.78475964e-01 4.65850741e-01 -5.48926890e-01 1.62097323e+00 -2.51966119e-01 9.18291360e-02 -9.82211232e-01 4.35205968e-03 4.40708935e-01 -1.79740414e-01 -8.19276333e-01 7.41267264e-01 -3.36204141e-01 -4.87782098e-02 -1.76471561e-01 -3.78575951e-01 -8.16204667e-01 -2.98051596e-01 -1.27358168e-01 5.44636905e-01 -1.35578722e-01 4.84637558e-01 -4.47226226e-01 2.53145814e-01 2.28636801e-01 8.78628612e-01 6.82097971e-01 1.05617754e-01 8.98357987e-01 1.11340573e-02 -5.99865139e-01 -1.06557274e+00 -9.12070930e-01 -5.16572475e-01 1.97109610e-01 1.26602149e+00 5.19386344e-02 -2.83886582e-01 -6.47994056e-02 2.76170582e-01 2.26799324e-01 -5.97537816e-01 -1.09282970e-01 -5.03693759e-01 -4.23588276e-01 3.87260377e-01 2.30854556e-01 8.83253276e-01 -5.77172101e-01 -1.11079514e+00 1.63142726e-01 5.12466617e-02 -8.83039296e-01 -2.68272847e-01 -1.97966844e-01 -6.34750605e-01 -1.15735257e+00 -5.36715627e-01 -7.79115379e-01 7.41690040e-01 1.04727018e+00 5.22670209e-01 -9.28513557e-02 -2.31565490e-01 2.39358336e-01 -7.52849460e-01 -5.30943811e-01 -1.62710547e-01 -9.90422785e-01 1.58812135e-01 -1.01004653e-02 3.52108955e-01 -3.43882173e-01 -7.15587854e-01 7.43452132e-01 -7.55359650e-01 -1.66018873e-01 5.56281865e-01 1.10519826e+00 5.23648739e-01 -1.83660269e-01 -2.05931574e-01 -2.52793819e-01 -3.41904089e-02 -1.11298040e-01 -1.09107900e+00 -8.55442286e-02 -4.85325515e-01 -3.55118334e-01 2.77750075e-01 -3.68925673e-03 -8.48324895e-01 1.24051478e-02 -9.32638198e-02 -2.17088401e-01 -1.22061022e-01 4.86363232e-01 -1.11173831e-01 -6.67413294e-01 6.49898648e-01 7.26174116e-01 5.08136511e-01 -3.47115070e-01 -3.61736029e-01 8.51564765e-01 6.27545416e-01 2.13186800e-01 1.09246409e+00 9.24031138e-01 2.89234817e-01 -1.41495848e+00 -1.05169594e-01 -7.37516046e-01 -4.63312805e-01 -3.25200051e-01 6.46664858e-01 -1.15357423e+00 -5.14129460e-01 7.72922397e-01 -1.04814136e+00 1.13587730e-01 2.05935538e-02 1.09213090e+00 6.83641881e-02 8.19270790e-01 -3.12157035e-01 -6.27022266e-01 -3.31271648e-01 -1.03074062e+00 9.11025286e-01 7.74160147e-01 4.09662008e-01 -6.73034489e-01 3.22387189e-01 4.28120680e-02 3.74251723e-01 1.74091727e-01 -2.57242955e-02 -9.46249887e-02 -4.81113553e-01 -4.24850851e-01 -3.08413982e-01 2.48469532e-01 1.92883119e-01 -6.98849233e-03 -6.77718103e-01 -4.59591806e-01 8.79455358e-02 1.34953946e-01 7.44264722e-01 2.97130406e-01 -1.92810923e-01 -5.26338667e-02 -2.20262855e-01 1.16197360e+00 1.84986651e+00 6.78771317e-01 7.84006357e-01 7.69813955e-01 5.78489721e-01 5.16697705e-01 1.29959166e+00 5.25537372e-01 4.54854965e-01 7.58468628e-01 9.30193961e-01 -1.64723843e-01 2.00235143e-01 -1.57641813e-01 1.57074511e-01 4.48877782e-01 -2.73533612e-01 3.72142866e-02 -6.18539631e-01 5.17240703e-01 -1.58549929e+00 -8.10977340e-01 -6.45366967e-01 2.54659724e+00 1.58725083e-01 -1.81481525e-01 -7.43219733e-01 -1.05391197e-01 6.12796426e-01 6.86415657e-02 -1.80662259e-01 -3.00374448e-01 -3.68582428e-01 -3.57072502e-01 1.34185016e+00 8.70618761e-01 -7.45949090e-01 5.62050700e-01 4.81094360e+00 4.36826378e-01 -1.24414444e+00 -3.56899738e-01 -5.39583027e-01 6.53809249e-01 -4.95175302e-01 4.29877788e-02 -9.79283273e-01 3.48333538e-01 1.97038963e-03 3.89492139e-02 -3.87969203e-02 8.00207019e-01 2.02604413e-01 -4.11529183e-01 -2.47459233e-01 1.19644213e+00 4.29235518e-01 -9.06353176e-01 -1.46377757e-01 9.27099609e-04 6.76858842e-01 1.15916155e-01 -5.69877923e-01 -2.04843685e-01 -2.90817618e-01 -2.49700844e-01 8.45741689e-01 4.11315680e-01 8.31803262e-01 -5.93880892e-01 1.31190383e+00 1.07695982e-01 -1.29289079e+00 2.48281091e-01 -9.25553381e-01 -3.48317087e-01 2.78146446e-01 1.20061681e-01 -1.04403245e+00 4.99650538e-01 9.80997205e-01 6.15776539e-01 -8.94557536e-02 1.34805095e+00 -1.83952630e-01 -6.75788745e-02 -7.33759284e-01 -2.94037223e-01 2.95547783e-01 -9.09441292e-01 8.18256736e-01 9.62371051e-01 1.00439954e+00 5.44587672e-01 -1.71042591e-01 2.81359613e-01 5.20672798e-01 3.06387365e-01 -8.48818719e-01 8.53969276e-01 8.56708825e-01 1.03417158e+00 -4.35137868e-01 -5.12793250e-02 -6.00014925e-01 6.67077422e-01 -6.46039367e-01 1.80048600e-01 -4.67434466e-01 -9.10824478e-01 8.62698853e-01 2.43684128e-01 1.65591240e-01 -3.32247883e-01 -6.92320243e-02 -7.94509292e-01 -9.36824605e-02 -2.26575285e-01 6.81984127e-02 -9.69377100e-01 -7.13236511e-01 4.10927773e-01 -1.67216212e-02 -2.01519060e+00 2.49108225e-01 -5.11923134e-01 -6.91569746e-01 8.62188160e-01 -1.60175157e+00 -8.69456410e-01 -1.08331752e+00 3.28689992e-01 3.91916275e-01 -6.55039772e-02 6.42903745e-01 2.07342759e-01 1.32125631e-01 4.22829986e-01 7.35381544e-01 3.67862321e-02 9.12024140e-01 -1.12609982e+00 1.77563131e-01 8.71714950e-01 -1.76499307e-01 2.94349968e-01 1.26835585e+00 -7.64798641e-01 -1.80796421e+00 -5.98607659e-01 4.49028105e-01 4.03646708e-01 4.46412921e-01 -7.30336234e-02 -1.05469203e+00 4.58216637e-01 -2.39161570e-02 6.23380132e-02 2.82904893e-01 -5.80142021e-01 -2.97595151e-02 -6.59280002e-01 -1.23709309e+00 2.36505836e-01 7.34994709e-01 7.83751067e-03 -6.06218874e-01 7.71312714e-02 1.74385533e-01 -9.67101276e-01 -7.36287296e-01 6.61709309e-01 1.01221824e+00 -1.38546479e+00 8.69656324e-01 3.24238956e-01 -8.02321825e-03 -7.17240512e-01 -4.58572716e-01 -1.58904219e+00 -5.57438470e-02 -8.22200477e-02 1.06788790e+00 8.91404986e-01 4.35418300e-02 -1.07165587e+00 5.62901258e-01 -3.63922422e-03 -6.86042726e-01 -1.06224388e-01 -1.19999361e+00 -6.69541121e-01 -5.63176572e-01 -5.38478382e-02 3.80358011e-01 8.28952968e-01 -2.50853039e-02 9.66226086e-02 -4.00112003e-01 1.03733265e+00 1.06966519e+00 4.63995904e-01 1.21505916e+00 -1.43064344e+00 7.62660522e-03 -9.18414667e-02 -1.00545692e+00 -1.21802628e+00 -4.95867848e-01 -3.91103253e-02 4.96642560e-01 -1.58613920e+00 -3.89120638e-01 -4.68547732e-01 3.95082891e-01 9.94237214e-02 2.84621507e-01 6.57366157e-01 9.26978234e-03 4.08416808e-01 7.88438618e-02 6.42314911e-01 1.28202963e+00 -9.42013320e-03 -1.16794743e-01 1.34787485e-01 7.78550804e-02 8.66834998e-01 4.04213876e-01 -2.24027872e-01 -1.27923608e-01 -7.25297987e-01 2.51344562e-01 3.44356000e-01 9.54274833e-02 -1.09881473e+00 6.30464554e-01 -2.11746484e-01 1.33587718e-01 -6.80259824e-01 5.83908617e-01 -1.15551460e+00 3.99266541e-01 7.70706296e-01 4.51183051e-01 -1.61712110e-01 8.50149617e-02 5.85139632e-01 -6.20526195e-01 -6.77652359e-01 1.12841105e+00 -1.40283421e-01 -1.30925643e+00 6.33279160e-02 -9.52746198e-02 -3.93336028e-01 9.42212105e-01 -7.71461546e-01 -4.74519819e-01 -5.10237873e-01 3.45708383e-03 2.19044447e-01 9.82528090e-01 1.81156725e-01 1.19444180e+00 -1.08083713e+00 -8.12683880e-01 7.62277961e-01 4.31110770e-01 3.44788015e-01 7.02931106e-01 6.27034247e-01 -1.48337233e+00 -2.16685444e-01 -4.36313421e-01 -8.50228906e-01 -1.58195066e+00 -1.20058060e-01 5.33772290e-01 8.21745098e-01 -8.48999381e-01 9.43373799e-01 2.61250377e-01 -4.19630706e-01 -1.05248660e-01 -1.68394834e-01 -3.25000554e-01 1.48995325e-01 5.72540522e-01 3.50387871e-01 -2.66346205e-02 -9.85154212e-01 -3.00518900e-01 1.68049610e+00 2.29930654e-01 5.50555624e-02 1.12036324e+00 -4.84541237e-01 1.08552866e-01 1.08347796e-01 1.19898748e+00 6.19635820e-01 -1.34861553e+00 -2.54286200e-01 -7.48824179e-01 -1.07822728e+00 1.45005798e-02 -2.53231436e-01 -8.32553625e-01 8.16941082e-01 9.02967632e-01 3.30448747e-01 9.77242589e-01 -2.36514270e-01 6.72502697e-01 3.83488536e-01 6.41508400e-01 -8.75179350e-01 -5.99230647e-01 6.36709452e-01 9.38515425e-01 -1.40332830e+00 2.85886079e-01 -3.66080880e-01 -6.60647213e-01 1.51910806e+00 5.60885251e-01 -3.39956760e-01 3.49526584e-01 3.09807539e-01 7.32484400e-01 -1.94221623e-02 1.27505839e-01 2.23969594e-02 -1.70855314e-01 6.08040392e-01 -3.65462631e-01 -1.33642063e-01 -5.71508527e-01 1.49254665e-01 -3.99275839e-01 -7.72583067e-01 1.12586486e+00 9.64588046e-01 -1.11432636e+00 -4.64765638e-01 -9.03288543e-01 4.28158194e-02 1.96983367e-01 4.33544070e-02 4.44206119e-01 9.99394476e-01 1.86836720e-01 4.91979331e-01 5.34402728e-01 -5.94287753e-01 9.55114365e-01 -6.51554346e-01 2.36200705e-01 -1.74543083e-01 2.37516358e-01 2.56437898e-01 -6.95596412e-02 -2.93481350e-01 -3.13040644e-01 -7.76009202e-01 -1.51322508e+00 9.59737450e-02 -4.02878106e-01 7.11889863e-01 1.18467283e+00 5.13751030e-01 -5.02652191e-02 -1.85573310e-01 1.00953734e+00 -1.07171476e+00 -5.01054704e-01 -1.20750237e+00 -1.05340064e+00 2.99167097e-01 3.82699728e-01 -8.09651494e-01 -1.05095077e+00 -1.88050702e-01]
[7.540106773376465, -1.8375636339187622]
e2bcf2a5-c2ba-4579-86ac-e544b4e985e3
a-survey-on-csi-based-human-behavior
null
null
https://doi.org/10.1109/ACCESS.2019.2922244
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8735849
A Survey on CSI-Based Human Behavior Recognition in Through-the-Wall Scenario
Recent years have witnessed increasing research interest in human behavior recognition as it provides attractive applications in various sensing scenarios. Among these encouraging implementations, device-free behavior recognition based on WiFi channel state information (CSI) has attracted significant attention due to the popularity of WiFi devices and abundant channel characteristics from CSI. Meanwhile, the CSI signal provides us with additional benefits because it can propagate through a wall. This through-the-wall and device-free scheme not only enables us to identify specific human actions but also to infer person activities by collecting the data from different rooms. This paper presents a survey on the state-of-art progresses in device-free through-the-wall human behavior recognition based on CSI. Specifically, this paper first introduces the basic concept of CSI and describes the signal variation caused by human behavior. Then, it illustrates that different human behaviors can cause signal transformation. Therefore, the unique map relationship between action and signal variation can be leveraged to recognize human behavior. Next, it provides the general architecture of through-the-wall behavior recognition and highlights its core characteristic. It investigates the state-of-art applications in various scenarios and analyzes specific design schemes and implementations. Afterward, it discusses the various across wall applications and makes a detailed comparison between non-through-the-wall and through-the-wall applications. Meanwhile, it analyzes many factors that affect recognition accuracy and emphasizes performance differences under across wall scenarios. Finally, this paper concludes by summarizing the issues and challenges faced and providing insights into the possible solution and future research trend.
['Yinjing Guo', 'Yushan Hou', 'Wenwen Dou', 'Chengming Zhang', 'Kangkang Jiang', 'Zhengjie Wang', 'Zehua Huang']
2019-06-12
null
null
null
ieee-access-volume-7-2019-6
['rf-based-pose-estimation']
['computer-vision']
[ 4.36037034e-01 -5.29013872e-01 -3.19318444e-01 -2.58391738e-01 -5.61273694e-01 -3.10840219e-01 1.05677515e-01 -2.76191652e-01 -5.83576076e-02 6.55071259e-01 5.03277183e-01 -1.26135200e-01 -2.32666478e-01 -6.23141885e-01 -4.03101683e-01 -8.34031940e-01 -4.56005752e-01 -3.37685019e-01 -2.00276226e-01 2.65837321e-03 -4.85774368e-01 1.87345386e-01 -1.23689437e+00 4.98101953e-03 3.08995098e-01 1.34730303e+00 3.81804118e-03 8.21805179e-01 2.68336058e-01 8.11113656e-01 -1.00959671e+00 3.30759250e-02 1.84755340e-01 -3.99138868e-01 -7.03746304e-02 -6.82002679e-02 6.82527199e-02 -2.82710284e-01 -9.41507816e-01 4.44878548e-01 8.80444646e-01 8.03778172e-02 1.58285692e-01 -1.27104330e+00 -1.74489141e-01 5.03875911e-01 -4.49231654e-01 3.17637026e-01 1.37147152e+00 -1.43093333e-01 3.84080321e-01 -2.86669463e-01 -2.16459506e-03 7.70570219e-01 1.15227342e+00 3.96137804e-01 -8.27570260e-01 -7.75452912e-01 1.64220154e-01 9.83941406e-02 -1.68876362e+00 -5.26842773e-01 5.38410723e-01 -2.47013852e-01 7.63923287e-01 7.43154287e-01 8.49505603e-01 1.51469219e+00 -6.95791170e-02 9.24431205e-01 7.46059477e-01 -3.05498391e-01 2.45376825e-02 -1.93534091e-01 3.06051880e-01 3.96203399e-01 3.93224657e-01 1.33679867e-01 -9.31889117e-01 -2.95547724e-01 5.85390627e-01 4.09745842e-01 -5.69495380e-01 -7.61259124e-02 -1.31673253e+00 -1.83441579e-01 3.25307190e-01 4.10899341e-01 -4.58006531e-01 4.50046331e-01 2.10380808e-01 3.94647270e-02 -2.46615946e-01 7.80972317e-02 -1.83309391e-01 -9.06558931e-01 -9.60055053e-01 2.54772678e-02 9.38415408e-01 1.07234287e+00 2.48464569e-01 3.64888579e-01 -5.66924632e-01 7.08716631e-01 3.18660080e-01 1.38488412e+00 7.81710818e-02 -6.15717173e-01 5.08243024e-01 -2.30598282e-02 3.38789165e-01 -1.12880421e+00 -6.33837938e-01 -7.74309099e-01 -9.17627335e-01 -7.65450597e-01 4.05120105e-01 -7.12237716e-01 -3.92731339e-01 1.62776554e+00 6.64539859e-02 7.39985287e-01 -2.14003086e-01 7.31978476e-01 8.18831921e-01 2.74731696e-01 -1.36045590e-01 -2.56626546e-01 1.46018577e+00 -7.15951025e-01 -1.04987514e+00 -3.07465762e-01 3.38322341e-01 -4.92479771e-01 8.05751860e-01 4.29591864e-01 -5.30490696e-01 -6.08058631e-01 -1.07521963e+00 9.46824789e-01 -5.88620901e-02 3.50116521e-01 8.08969557e-01 1.55749905e+00 -6.28765762e-01 -1.30868122e-01 -1.27648509e+00 -6.72391653e-01 1.78628862e-01 5.01213253e-01 1.22060971e-02 -2.03940660e-01 -1.14400589e+00 2.78273582e-01 -4.63037461e-01 3.30853552e-01 -4.87282485e-01 -5.81620932e-01 -8.29664290e-01 1.10269058e-02 3.87235314e-01 -5.97847641e-01 1.43694532e+00 -4.78169888e-01 -1.57296097e+00 7.77846649e-02 -7.52770007e-01 -3.35632861e-01 3.43609840e-01 -2.54482210e-01 -1.43841696e+00 -1.19742647e-01 1.57278344e-01 -4.86362040e-01 3.04164410e-01 -9.54582334e-01 -6.62584364e-01 -2.80857712e-01 -7.83736110e-02 -1.55882478e-01 -4.63037610e-01 -2.52073616e-01 -8.14700723e-01 -6.39816403e-01 1.25839338e-01 -1.01347601e+00 -6.81686550e-02 -5.27083278e-01 -6.86540186e-01 5.89149535e-01 7.05600858e-01 -3.91958535e-01 1.88805449e+00 -2.15180063e+00 -8.34271848e-01 7.74357617e-01 8.99627656e-02 -1.86850522e-02 3.01534802e-01 7.04513729e-01 5.07554352e-01 -3.97000283e-01 2.33687371e-01 -2.94532180e-01 -2.25063041e-02 4.63159047e-02 -4.95695621e-02 7.97947526e-01 -7.93649435e-01 7.88098514e-01 -9.20152724e-01 1.04321435e-01 5.44044614e-01 6.89073384e-01 -3.50741535e-01 8.92928243e-02 9.31291461e-01 7.81570315e-01 -8.47796619e-01 9.10787880e-01 7.62938380e-01 -2.03997195e-01 3.75493169e-01 -2.70757496e-01 -2.06051975e-01 1.60337955e-01 -1.28455639e+00 1.32358599e+00 -6.63477182e-01 6.28366709e-01 1.16950363e-01 -1.03871620e+00 8.54897201e-01 4.35202122e-01 9.28470135e-01 -8.52843523e-01 1.68512389e-01 8.05610195e-02 -2.37083703e-01 -6.11881316e-01 1.81647092e-01 3.84234607e-01 -6.96099997e-01 3.80485326e-01 -5.20475626e-01 6.73708141e-01 -9.67711955e-02 -9.82011333e-02 1.86155438e+00 -6.66631386e-02 6.59024239e-01 -7.35728908e-03 7.49386489e-01 -4.61171895e-01 5.21714449e-01 1.41932702e+00 -5.39552748e-01 2.49198694e-02 -5.10821283e-01 -1.34846941e-01 1.30773872e-01 -1.38126147e+00 -1.39797330e-01 9.42527711e-01 7.34968841e-01 -6.32812023e-01 -4.27175313e-01 -2.08169580e-01 3.03819031e-01 1.89519703e-01 -6.98183715e-01 -9.75360274e-02 -4.35098261e-01 -6.97052121e-01 1.28062677e+00 6.63077891e-01 1.36094761e+00 -5.03633142e-01 -7.30979502e-01 2.99742281e-01 -6.99886799e-01 -1.55813205e+00 -1.28583238e-01 -9.01378393e-02 -3.75328273e-01 -7.23842144e-01 -6.89082980e-01 -2.83445776e-01 4.08685207e-01 1.05877030e+00 6.72356367e-01 -1.00777845e-03 -3.71800780e-01 1.18611979e+00 -4.01825219e-01 -2.50859410e-01 5.18594027e-01 -8.06464925e-02 5.01119137e-01 4.17936951e-01 6.79932177e-01 -5.38712800e-01 -7.02641368e-01 7.43440688e-01 1.10201038e-01 -6.17044389e-01 3.66966367e-01 3.03822339e-01 1.77978843e-01 2.20670491e-01 3.11508507e-01 -4.27607536e-01 3.09437722e-01 -6.54466212e-01 -2.01213229e-02 2.94353753e-01 -1.85192749e-01 -4.29949313e-01 5.04044890e-01 -1.57139868e-01 -1.34597683e+00 1.54452533e-01 -1.08226188e-01 1.22871168e-01 -2.99498826e-01 3.00858527e-01 -4.55586880e-01 -3.03628057e-01 6.38835549e-01 4.00261611e-01 -5.00457048e-01 -3.92287552e-01 -1.42893065e-02 1.10436976e+00 7.90182710e-01 -4.96175647e-01 7.27072954e-01 9.42848444e-01 -1.25814632e-01 -1.45978570e+00 -8.97401094e-01 -1.23008895e+00 -2.72111297e-01 -6.70159638e-01 6.33471787e-01 -1.14586079e+00 -1.32367933e+00 6.95671082e-01 -6.17915988e-01 -4.94112611e-01 2.07530022e-01 6.60008490e-01 -3.36239070e-01 2.48254046e-01 -5.97284079e-01 -1.43217838e+00 4.39799055e-02 -8.85757923e-01 1.06015217e+00 3.40264857e-01 -6.42518938e-01 -9.95043159e-01 -6.76460415e-02 5.27253568e-01 6.59479916e-01 3.18492025e-01 -2.09265649e-01 1.58373099e-02 -2.98223644e-01 -5.05390882e-01 2.37325415e-01 -4.43267316e-01 5.12715399e-01 -7.40447938e-01 -1.22594345e+00 -3.76640469e-01 -1.60369471e-01 3.79772395e-01 3.03079367e-01 8.64214003e-01 1.14744222e+00 -1.41309902e-01 -1.11705220e+00 9.97494102e-01 1.05540252e+00 2.73974121e-01 8.11360419e-01 3.84420395e-01 6.90670252e-01 -5.54985926e-02 6.08460367e-01 8.43594551e-01 4.00783658e-01 1.25464225e+00 -1.34764254e-01 -2.93966830e-01 -2.66237110e-01 -2.93107122e-01 5.94420135e-01 4.65577602e-01 -3.30742002e-01 -6.05693817e-01 -6.56953156e-01 -2.62469575e-02 -1.83821058e+00 -1.25434625e+00 -4.69546407e-01 2.34459543e+00 -1.37533158e-01 -1.45338222e-01 4.36596692e-01 5.23833930e-01 6.78103089e-01 4.84604053e-02 -4.54118401e-01 6.51592836e-02 -1.25879645e-01 -6.25561103e-02 1.08329749e+00 4.64642912e-01 -1.27963924e+00 5.11511147e-01 7.02242136e+00 4.62364644e-01 -1.00591171e+00 1.72249779e-01 -7.34640704e-03 -1.07136853e-01 2.96732724e-01 -6.56731009e-01 -9.50600088e-01 5.14401853e-01 7.04826117e-01 2.27834463e-01 3.19900453e-01 8.05847168e-01 6.49954498e-01 -4.71945673e-01 -8.38678360e-01 1.73495507e+00 2.30257213e-01 -9.42396522e-01 -5.53249061e-01 3.52465510e-01 5.06334662e-01 2.68077143e-02 -3.62230539e-02 2.22421810e-01 -6.79053292e-02 -6.86196506e-01 6.95382059e-01 4.56406832e-01 8.49615693e-01 -3.93627971e-01 7.38761842e-01 1.38136268e-01 -2.11699486e+00 -5.18306434e-01 2.15082198e-01 -7.81345010e-01 5.47787309e-01 8.79180729e-01 -4.00715590e-01 6.69433177e-01 1.01309812e+00 9.08938587e-01 -2.46741235e-01 1.13251638e+00 -2.07692638e-01 1.02109790e+00 -4.72241223e-01 -2.13630572e-01 -2.00316295e-01 9.45453253e-03 4.17474270e-01 1.61952353e+00 7.54282892e-01 2.94923753e-01 5.41953981e-01 2.42640927e-01 4.66563314e-01 -3.48908126e-01 -6.70478225e-01 3.73552293e-01 1.05355656e+00 8.87658477e-01 -5.65680325e-01 -1.98294565e-01 -7.24644423e-01 9.91381049e-01 -5.45002520e-01 7.27455974e-01 -8.29805613e-01 -3.56898248e-01 8.98244381e-01 4.02953148e-01 1.55664682e-01 -5.56859493e-01 -2.35262915e-01 -1.17848349e+00 1.42706810e-02 -5.99456251e-01 3.19002569e-01 -2.52801478e-01 -8.29329729e-01 1.05559297e-01 -9.03120935e-02 -1.53131711e+00 -2.57265009e-02 -3.05099040e-01 -4.68080878e-01 4.23713148e-01 -1.02980995e+00 -1.06492662e+00 -9.61130619e-01 1.02551472e+00 2.50173599e-01 -1.84325464e-02 9.42042589e-01 8.70356381e-01 -8.14538598e-01 1.24142766e+00 3.57565850e-01 4.97345567e-01 2.17638865e-01 -5.83430111e-01 3.44417512e-01 1.01707625e+00 2.50378111e-03 1.12848043e+00 7.93448389e-01 -6.04354739e-01 -1.81753600e+00 -7.98992932e-01 6.55627370e-01 -3.16381544e-01 3.37516040e-01 -6.42604709e-01 -1.51234165e-01 5.82539201e-01 -4.06914622e-01 1.34315073e-01 1.19325566e+00 3.88276488e-01 -4.28063720e-02 -5.09612143e-01 -9.99075234e-01 7.06958950e-01 1.58014965e+00 -3.90895665e-01 -2.40132362e-02 -2.06674591e-01 -2.43427306e-01 -2.59011149e-01 -6.68950140e-01 6.81774244e-02 1.31579673e+00 -9.80478585e-01 1.20728159e+00 2.28011981e-01 -7.83099651e-01 -4.25856411e-01 -5.05953372e-01 -9.33838069e-01 -5.08975148e-01 -9.63440359e-01 -5.63873351e-01 1.06978977e+00 8.19275454e-02 -7.54982650e-01 8.27465117e-01 5.44310629e-01 7.63619170e-02 -2.03596696e-01 -8.07490468e-01 -1.06069660e+00 -1.07967448e+00 -1.02393389e+00 8.59895468e-01 6.26892805e-01 5.64080179e-01 1.52986785e-02 -9.00545895e-01 2.17973232e-01 7.11757243e-01 -2.33990550e-01 1.27691770e+00 -9.59330618e-01 -7.09291935e-01 -5.57731185e-03 -6.65209174e-01 -1.98378706e+00 -3.42128038e-01 -4.06986505e-01 8.68420750e-02 -1.31683373e+00 -4.23836671e-02 -5.40180743e-01 -4.36914176e-01 4.20115799e-01 1.93692207e-01 5.01518011e-01 -1.79591421e-02 1.43797278e-01 -7.94598639e-01 2.07137153e-01 8.12760174e-01 -1.80630893e-01 -4.15838122e-01 6.21677995e-01 -6.77720666e-01 6.24646783e-01 9.23359931e-01 -2.33891308e-02 -3.77037555e-01 -1.86207697e-01 1.58427551e-01 7.72005022e-02 3.80477071e-01 -1.76343048e+00 2.15957046e-01 -2.08523758e-02 6.75807893e-01 -2.08138078e-01 7.32463002e-01 -1.01939142e+00 2.73077577e-01 7.11114824e-01 1.12963989e-01 -3.04092526e-01 1.16664223e-01 8.47482204e-01 1.78508818e-01 5.85447311e-01 1.97989255e-01 1.86313942e-01 -9.20213580e-01 7.62935877e-02 -8.87920439e-01 -2.73849994e-01 8.76769245e-01 -7.47446537e-01 -5.42945750e-02 -9.60208356e-01 -5.97456694e-01 2.05754992e-02 -6.57293424e-02 3.63211244e-01 2.29885072e-01 -1.40139461e+00 2.15113815e-02 4.97460753e-01 2.95044303e-01 -9.47501123e-01 5.10573864e-01 1.21289027e+00 1.07863136e-02 6.83192492e-01 5.58538437e-02 -8.52079749e-01 -1.47535324e+00 -9.37277526e-02 4.51252103e-01 -4.34556557e-03 -5.50225139e-01 7.51277030e-01 1.76103041e-03 -3.02522242e-01 5.44376493e-01 -2.24701390e-01 -1.28909960e-01 -6.06325269e-01 8.68521392e-01 8.56947780e-01 8.78688172e-02 -9.37600791e-01 -8.77322853e-01 8.14505577e-01 7.54060388e-01 -1.24903627e-01 7.00782895e-01 -7.99319863e-01 7.11210549e-01 3.48104984e-01 8.39181840e-01 3.77069086e-01 -1.11911952e+00 -2.23506853e-01 -2.78232455e-01 -6.85807347e-01 -6.12756722e-02 -8.76535714e-01 -9.91584837e-01 4.89645839e-01 8.63090396e-01 2.12043807e-01 9.70743120e-01 -1.51220873e-01 1.04931974e+00 5.46619236e-01 1.17163181e+00 -9.39737558e-01 -1.44335076e-01 3.24824274e-01 1.80110022e-01 -8.18710208e-01 -6.96996343e-04 -8.14126492e-01 -3.03899914e-01 7.33186066e-01 2.48265415e-01 3.13617289e-01 1.06749523e+00 7.48697400e-01 3.41483861e-01 -1.33708447e-01 1.01389274e-01 -4.59367603e-01 -4.79390100e-02 1.38510382e+00 6.17923558e-01 5.76250851e-01 9.02989209e-02 9.65424776e-01 -5.80842912e-01 1.19361915e-01 1.91924527e-01 1.09579289e+00 -2.19490841e-01 -1.05470872e+00 -8.97188842e-01 8.01117122e-01 -4.74005818e-01 4.45128560e-01 -1.94210723e-01 5.38033485e-01 1.52748138e-01 1.80555916e+00 -3.15450251e-01 -8.86795044e-01 7.00579464e-01 -4.62049633e-01 4.40991074e-01 -1.89627245e-01 -5.16302466e-01 6.22915067e-02 2.96986043e-01 -9.83197212e-01 -4.83903885e-01 -8.09422255e-01 -1.15507936e+00 -4.97625440e-01 -8.61644745e-02 2.06739753e-01 4.29657519e-01 1.02855730e+00 4.86852527e-01 8.90869915e-01 5.20717680e-01 -9.02701914e-01 7.72987455e-02 -5.95523536e-01 -9.04399633e-01 3.30654949e-01 3.93499374e-01 -1.05469179e+00 -2.50885904e-01 -9.83181447e-02]
[6.6919779777526855, 0.722235381603241]
274194af-8dcc-4103-a267-558e46a29ca7
escnet-an-end-to-end-superpixel-enhanced
null
null
https://ieeexplore.ieee.org/document/9474911
https://ieeexplore.ieee.org/document/9474911
ESCNet: An End-to-End Superpixel-Enhanced Change Detection Network for Very-High-Resolution Remote Sensing Images
Change detection (CD), as one of the central problems in Earth observation, has attracted a lot of research interest over recent decades. Due to the rapid development of satellite sensors in recent years, we have witnessed an enrichment of the CD source data with the availability of very-high-resolution (VHR) multispectral imagery, which provides abundant change clues. However, precisely locating real changed areas still remains a challenge. In this article, we propose an end-to-end superpixel-enhanced CD network (ESCNet) for VHR images, which combines differentiable superpixel segmentation and a deep convolutional neural network (DCNN). Two weight-sharing superpixel sampling networks (SSNs) are tailored for the feature extraction and superpixel segmentation of bitemporal image pairs. A UNet-based Siamese neural network is then employed to mine the different information. The superpixels are then leveraged to reduce the latent noise in the pixel-level feature maps while preserving the edges, where a novel superpixelation module is used to serve this purpose. Furthermore, to compensate for the dependence on the number of superpixels, we propose an innovative adaptive superpixel merging (ASM) module, which has a concise form and is fully differentiable. A pixel-level refinement module making use of the multilevel decoded features is also appended to the end of the framework. Experiments on two public datasets confirmed the superiority of ESCNet compared to the traditional and state-of-the-art (SOTA) deep learning-based CD (DLCD) methods.
['Liangpei Zhang', 'Guangyi Yang', 'Manhui Lin', 'Hongyan zhang']
2021-07-05
null
null
null
ieee-transactions-on-neural-networks-and-9
['change-detection', 'superpixels', 'change-detection-for-remote-sensing-images']
['computer-vision', 'computer-vision', 'miscellaneous']
[ 3.33676964e-01 -3.24418008e-01 -8.61779694e-03 -4.21238661e-01 -7.23385930e-01 -1.73884258e-01 5.73197842e-01 -2.54975319e-01 -6.64993942e-01 6.18005753e-01 6.86725378e-02 -2.74687782e-02 -1.15596540e-01 -1.04588437e+00 -6.18187904e-01 -9.18211937e-01 -1.07729912e-01 -6.74808994e-02 3.41470957e-01 -3.73705089e-01 1.25539437e-01 5.05138099e-01 -1.44209325e+00 -3.91929969e-03 1.36586773e+00 1.15278876e+00 4.50840443e-01 2.79344618e-01 2.20822934e-02 3.41448009e-01 -5.08551951e-03 -1.82596579e-01 6.33983552e-01 -3.50703120e-01 -5.22230804e-01 2.21056968e-01 7.39895880e-01 -5.26259959e-01 -3.34437490e-01 1.50171328e+00 4.74322885e-01 9.65117812e-02 3.18774283e-01 -6.61060095e-01 -5.89276016e-01 4.31269526e-01 -1.14455330e+00 4.33792889e-01 -1.95470244e-01 2.32130706e-01 1.04937696e+00 -8.11731160e-01 6.42134964e-01 1.10479629e+00 6.99817896e-01 6.59630895e-02 -1.08637285e+00 -6.66130245e-01 2.08947435e-01 3.15890938e-01 -1.47656631e+00 -1.46344915e-01 1.00871432e+00 -3.39393437e-01 5.60609579e-01 9.73990709e-02 8.47869575e-01 6.19615555e-01 8.59643966e-02 8.61935556e-01 1.30111206e+00 -1.06288135e-01 9.55345333e-02 -3.43797743e-01 1.15088835e-01 7.27374494e-01 2.79319704e-01 1.87027007e-01 -2.06581458e-01 1.99675202e-01 8.74339223e-01 3.06115925e-01 -4.99068916e-01 -2.15065286e-01 -1.05066812e+00 8.97916794e-01 1.06684124e+00 5.37752390e-01 -6.16236806e-01 -1.04824966e-02 1.05949357e-01 1.01537317e-01 9.67876792e-01 2.94042259e-01 -3.74330729e-01 2.61491746e-01 -1.47179735e+00 2.63922721e-01 3.44064921e-01 2.05135301e-01 1.07032144e+00 1.97293572e-02 -1.24902628e-01 1.07900286e+00 4.20520633e-01 5.66220224e-01 3.95389736e-01 -9.28703189e-01 4.48352963e-01 8.26020420e-01 -1.70140900e-02 -1.34027982e+00 -4.34356779e-01 -8.19143951e-01 -1.26176047e+00 2.59886116e-01 9.41308290e-02 3.88836749e-02 -1.08648431e+00 1.41508734e+00 6.32366240e-01 3.15417677e-01 -1.91902250e-01 1.20677638e+00 4.84223574e-01 7.24476874e-01 -2.26570591e-01 5.49816042e-02 1.28991270e+00 -8.75640512e-01 -4.07464653e-01 -2.78807640e-01 2.54312396e-01 -2.58581221e-01 7.24475980e-01 1.62867248e-01 -7.50570178e-01 -5.16133964e-01 -1.18929613e+00 -1.84336409e-01 -4.92011726e-01 1.00811735e-01 5.68140268e-01 5.13422489e-01 -9.48855937e-01 7.46159256e-01 -1.04002655e+00 -1.93322524e-01 8.50409985e-01 1.06270574e-01 -2.10815787e-01 -2.67033726e-01 -1.28425109e+00 5.62336564e-01 5.28855622e-01 6.87118411e-01 -7.88938105e-01 -6.71815217e-01 -8.25104713e-01 1.06215753e-01 2.74692029e-01 -5.90871811e-01 6.11751437e-01 -1.36488676e+00 -1.41315353e+00 9.11199331e-01 1.85272634e-01 -5.49734652e-01 7.48076916e-01 -1.57306567e-01 -3.26990455e-01 4.69881654e-01 2.64496386e-01 8.47936332e-01 9.32318747e-01 -1.10649168e+00 -1.07261133e+00 -6.67751968e-01 -4.28104773e-02 2.75483727e-01 -1.84729978e-01 -1.09815873e-01 -5.59957445e-01 -9.26160991e-01 4.09311712e-01 -7.10658312e-01 -3.35725158e-01 1.68816075e-01 -2.64587611e-01 -4.05484326e-02 8.26837063e-01 -1.11077571e+00 1.20426881e+00 -2.28929758e+00 2.46990398e-01 2.65972912e-01 2.96697408e-01 3.23246211e-01 -2.62368411e-01 -1.28659919e-01 -5.74339479e-02 -3.73494141e-02 -1.09080470e+00 -3.00979018e-01 -1.35851324e-01 1.63060799e-01 -6.61271065e-02 8.18637311e-01 2.98519850e-01 7.86658704e-01 -9.80108142e-01 -4.55117047e-01 3.25266302e-01 3.73708427e-01 -5.45172393e-01 -2.87200622e-02 -2.98620045e-01 4.89883721e-01 -3.89877617e-01 8.80287647e-01 1.26755857e+00 -8.50298442e-03 -1.96888074e-01 -3.40522915e-01 -6.95503354e-01 -1.72083050e-01 -1.28827441e+00 1.70245135e+00 -1.51888967e-01 3.94024760e-01 3.88584852e-01 -1.02900445e+00 7.59774506e-01 -1.59617841e-01 5.19455850e-01 -9.73285615e-01 -7.33212307e-02 4.50958490e-01 -2.25410849e-01 -5.00724971e-01 6.02785051e-01 -6.55759797e-02 2.28319287e-01 1.23805083e-01 -3.20747554e-01 -1.15789078e-01 2.34489486e-01 -8.55331197e-02 6.03193402e-01 3.43673408e-01 2.60524213e-01 -2.91528434e-01 6.07345879e-01 2.86426157e-01 8.31116796e-01 3.56365383e-01 -3.54208827e-01 7.27840781e-01 -6.73502334e-04 -4.70777482e-01 -9.06353533e-01 -7.06350684e-01 -3.23403865e-01 7.76939988e-01 4.06389683e-01 3.66067529e-01 -7.08330214e-01 -4.80701447e-01 6.47602975e-03 3.94024223e-01 -7.45192766e-01 1.66612819e-01 -7.38337815e-01 -1.28002310e+00 4.83228713e-01 2.76895344e-01 1.34648097e+00 -7.97756314e-01 -6.34320617e-01 3.00809503e-01 -4.27349627e-01 -8.71585906e-01 -5.07430732e-01 -2.13004556e-02 -8.87219131e-01 -9.15782809e-01 -9.43211436e-01 -6.99104726e-01 5.54473102e-01 6.81398094e-01 6.17806077e-01 -2.07695395e-01 -3.20845634e-01 -6.83284849e-02 -3.54561388e-01 1.62613317e-01 -5.75354137e-03 1.61813170e-01 -2.94317901e-01 4.35790092e-01 1.15835652e-01 -5.36121607e-01 -9.68693495e-01 -2.28966475e-02 -1.35773432e+00 2.22542599e-01 8.76163244e-01 9.62952554e-01 7.23497987e-01 1.36346534e-01 3.21684837e-01 -6.81377411e-01 1.08747289e-01 -5.20081460e-01 -8.27280760e-01 2.30272010e-01 -5.34439683e-01 -7.31098354e-02 2.81002998e-01 -4.63670976e-02 -1.42747653e+00 9.62142795e-02 -3.44525069e-01 -9.60929394e-02 -1.46330297e-01 8.63030136e-01 -1.09186754e-01 -1.70735464e-01 3.62608373e-01 4.43431020e-01 -1.44830868e-01 -5.71552336e-01 4.40495640e-01 9.00212169e-01 7.12014914e-01 -2.17482932e-02 9.02512968e-01 9.70641315e-01 -2.13712275e-01 -9.11943197e-01 -9.69448745e-01 -7.41940200e-01 -6.42757833e-01 -8.99878144e-02 1.06910336e+00 -1.15485251e+00 -1.44160941e-01 1.00458038e+00 -7.91996419e-01 -3.64204407e-01 -1.26252130e-01 4.26186293e-01 -1.17239989e-01 6.28437698e-01 -5.25781095e-01 -4.89877731e-01 -6.47214472e-01 -9.85720456e-01 1.17302859e+00 4.44413006e-01 4.82566357e-01 -8.29062760e-01 2.79576808e-01 5.08526981e-01 4.73566383e-01 4.90807056e-01 6.40052021e-01 -1.21639937e-01 -8.71319056e-01 1.57984849e-02 -6.70471370e-01 6.59411669e-01 1.79735258e-01 -2.19847392e-02 -8.29972684e-01 -3.87137294e-01 4.67831269e-02 -1.54292528e-02 1.47271836e+00 7.62286961e-01 9.74063337e-01 -1.98504671e-01 -2.93080598e-01 1.12108183e+00 1.68101442e+00 -1.27906203e-01 7.57615209e-01 6.10076427e-01 8.73190463e-01 3.75005424e-01 5.72998405e-01 4.08921510e-01 5.12717664e-01 6.42231524e-01 7.04360843e-01 -5.35197794e-01 -2.87794411e-01 1.03213310e-01 3.87405604e-01 4.73613501e-01 -1.02414370e-01 1.95458323e-01 -6.81235611e-01 7.35745370e-01 -1.57050705e+00 -1.11303174e+00 -2.62180716e-01 1.95629096e+00 9.29083109e-01 -2.48904720e-01 -2.41361305e-01 -5.63878827e-02 8.09035361e-01 6.60928965e-01 -8.39912117e-01 3.26589525e-01 -4.97164369e-01 5.53394035e-02 7.81604350e-01 3.40765327e-01 -1.57895863e+00 9.45532680e-01 4.40253305e+00 1.05657935e+00 -1.30411577e+00 2.11148709e-01 6.67759120e-01 2.92019248e-01 -1.28175423e-01 -1.96464077e-01 -7.11113870e-01 5.25912046e-01 2.03724369e-01 3.37118477e-01 4.82153505e-01 6.55238330e-01 5.43021441e-01 -3.16348344e-01 -2.86483645e-01 8.30778360e-01 1.15842119e-01 -1.25510013e+00 1.65751800e-02 -2.56778989e-02 1.20092535e+00 6.20818973e-01 1.02457263e-01 9.88785326e-02 3.04797918e-01 -5.77606916e-01 6.49962902e-01 5.37744820e-01 7.03069568e-01 -8.37996304e-01 7.36496449e-01 2.10116014e-01 -1.30115926e+00 -2.91290849e-01 -4.75966930e-01 2.31991485e-01 8.88584256e-02 9.91648436e-01 -2.72984892e-01 9.20893788e-01 9.48619604e-01 1.06461012e+00 -7.39066899e-01 1.29516256e+00 -2.61055678e-01 6.15372121e-01 -4.43822980e-01 5.09109378e-01 6.67147040e-01 -6.63979828e-01 6.92501903e-01 1.24330926e+00 2.36421108e-01 8.93839374e-02 9.87933427e-02 9.92611051e-01 -2.69673795e-01 -1.57467322e-03 -1.36793047e-01 1.51349157e-01 1.79609209e-01 1.69684517e+00 -7.78958738e-01 -3.16165775e-01 -3.71489495e-01 1.28025973e+00 2.31116086e-01 2.78802544e-01 -6.30644500e-01 -3.58839422e-01 6.41657352e-01 -2.11198926e-01 6.09723389e-01 -2.49204904e-01 -2.61062324e-01 -1.36260486e+00 -1.18558161e-01 -7.87724376e-01 3.88470471e-01 -4.66287404e-01 -1.32611287e+00 3.26927960e-01 -2.87942678e-01 -1.38141119e+00 3.70669544e-01 -1.26515448e-01 -5.55385113e-01 8.17075312e-01 -2.33410645e+00 -1.31994593e+00 -6.73623979e-01 3.72683227e-01 5.12817085e-01 2.59090424e-01 1.81417733e-01 5.55769503e-01 -8.67792904e-01 1.40307218e-01 5.37609041e-01 3.65676880e-01 4.40491587e-01 -1.07135570e+00 2.76780397e-01 1.37085414e+00 -1.39658272e-01 2.74052888e-01 2.62988657e-01 -7.69777834e-01 -1.05799437e+00 -1.77638924e+00 5.66458583e-01 3.86271268e-01 6.10849142e-01 5.06375544e-02 -1.03439260e+00 3.98763537e-01 -2.89233252e-02 4.19012234e-02 3.13027292e-01 -5.36900401e-01 -1.53379306e-01 -4.70807463e-01 -1.17264414e+00 5.32197177e-01 8.75852346e-01 -4.09217983e-01 -5.84433913e-01 1.80200681e-01 7.38830268e-01 -3.02964926e-01 -6.95813119e-01 5.49517930e-01 3.11366588e-01 -9.69900787e-01 9.90103424e-01 7.92203024e-02 5.31963587e-01 -6.36762202e-01 -1.77668199e-01 -1.35719383e+00 -4.50235784e-01 -1.30143508e-01 2.12577403e-01 1.10506821e+00 -3.07034776e-02 -6.91858947e-01 4.72439587e-01 5.13911545e-02 -4.26231563e-01 -4.70313251e-01 -1.07415056e+00 -6.54519260e-01 -6.94039762e-02 -1.50609031e-01 4.54938829e-01 1.08803737e+00 -6.10008955e-01 3.34763899e-02 -2.73553014e-01 5.65342665e-01 8.35809469e-01 5.50812960e-01 3.79250020e-01 -1.29088783e+00 -1.70288622e-01 -7.00054586e-01 -2.06748426e-01 -1.30903876e+00 -4.00870070e-02 -1.01124203e+00 2.15556562e-01 -1.72841811e+00 2.52168149e-01 -3.20958257e-01 -2.45115593e-01 4.48055804e-01 -5.36237299e-01 6.11298501e-01 -2.30059661e-02 3.88218403e-01 -3.94587666e-01 9.01478469e-01 1.29054236e+00 -3.89517635e-01 -4.14769530e-01 -4.39262278e-02 -4.57394272e-01 7.02061057e-01 6.70210421e-01 -4.28633392e-01 2.36620769e-01 -5.78290522e-01 9.67039615e-02 -2.81223118e-01 6.64618850e-01 -1.01717389e+00 2.46342570e-01 -6.26094863e-02 3.51384401e-01 -8.70954514e-01 -3.76498252e-02 -7.98052192e-01 8.86768028e-02 7.23944008e-01 -7.51803890e-02 -4.59050924e-01 -1.56939134e-01 6.30108893e-01 -4.47429717e-01 -5.30513339e-02 1.25702953e+00 -1.79966524e-01 -9.72042620e-01 6.77814841e-01 -1.43257231e-01 -2.62855589e-01 9.63506877e-01 -2.69073576e-01 -1.55159727e-01 2.46638715e-01 -1.95633188e-01 4.12083387e-01 3.50347102e-01 1.25945613e-01 5.55572510e-01 -1.03501582e+00 -9.76675212e-01 1.89536124e-01 4.46253270e-02 3.28921169e-01 6.53798461e-01 1.11319709e+00 -7.86494613e-01 1.89620242e-01 -2.76478618e-01 -5.26633739e-01 -9.81556594e-01 2.09332913e-01 6.45230651e-01 -3.76493096e-01 -7.33490944e-01 6.20518923e-01 8.75213817e-02 -5.02869666e-01 -1.89598143e-01 -4.98995006e-01 -3.34175736e-01 4.76638407e-01 5.22475302e-01 5.84916890e-01 1.06942831e-02 -7.28222549e-01 -2.02830255e-01 6.85219407e-01 3.69853415e-02 7.98691884e-02 1.91572750e+00 -5.06814480e-01 -4.10073102e-01 3.50767709e-02 1.17624342e+00 -2.34482542e-01 -1.71972990e+00 -4.77668196e-01 -1.48332149e-01 -5.16873240e-01 5.84938943e-01 -5.75937808e-01 -1.53430140e+00 9.47225451e-01 9.25261021e-01 -8.21319073e-02 1.25046802e+00 -3.65643770e-01 1.04063702e+00 2.13012457e-01 2.46608645e-01 -1.21265364e+00 -8.20431262e-02 3.77478004e-01 6.64227784e-01 -1.43511987e+00 4.94153909e-02 -2.60128617e-01 -3.26153338e-01 9.73163962e-01 3.21205854e-01 -1.99592233e-01 5.69125295e-01 -1.15574054e-01 -3.13881002e-02 -2.66172141e-01 8.78899172e-02 -5.40289879e-01 1.27585322e-01 4.29782331e-01 -1.17854543e-01 8.37599412e-02 -3.53218377e-01 3.37400973e-01 2.11270943e-01 1.55757621e-01 2.44742721e-01 7.29967475e-01 -7.48868167e-01 -5.87230146e-01 -4.54663157e-01 4.28337663e-01 -3.64056468e-01 -3.37553203e-01 9.78318080e-02 5.18376589e-01 4.94659096e-01 7.48663306e-01 1.12745747e-01 -2.45818179e-02 6.30976781e-02 -4.11293864e-01 8.74495059e-02 -3.42870027e-01 -6.43311501e-01 5.92203364e-02 -3.81092787e-01 -5.95063984e-01 -8.55599701e-01 -7.48030245e-01 -1.13519025e+00 -4.24320288e-02 -3.28174770e-01 -1.95802107e-01 4.92053270e-01 9.46755111e-01 3.09868366e-01 4.78822052e-01 9.70075786e-01 -1.13792765e+00 -5.16504407e-01 -1.01435912e+00 -9.20110583e-01 3.02829057e-01 3.99664313e-01 -4.61917430e-01 -3.87441635e-01 1.30530015e-01]
[9.697587966918945, -1.3918957710266113]
da800672-f4b1-43ff-bf79-9eb0cf3a1f8b
intelligent-home-3d-automatic-3d-house-design
2003.00397
null
https://arxiv.org/abs/2003.00397v1
https://arxiv.org/pdf/2003.00397v1.pdf
Intelligent Home 3D: Automatic 3D-House Design from Linguistic Descriptions Only
Home design is a complex task that normally requires architects to finish with their professional skills and tools. It will be fascinating that if one can produce a house plan intuitively without knowing much knowledge about home design and experience of using complex designing tools, for example, via natural language. In this paper, we formulate it as a language conditioned visual content generation problem that is further divided into a floor plan generation and an interior texture (such as floor and wall) synthesis task. The only control signal of the generation process is the linguistic expression given by users that describe the house details. To this end, we propose a House Plan Generative Model (HPGM) that first translates the language input to a structural graph representation and then predicts the layout of rooms with a Graph Conditioned Layout Prediction Network (GC LPN) and generates the interior texture with a Language Conditioned Texture GAN (LCT-GAN). With some post-processing, the final product of this task is a 3D house model. To train and evaluate our model, we build the first Text-to-3D House Model dataset.
['Yu-Han Wang', 'Shuai Wang', 'Rui Tang', 'Qi Wu', 'Qi Chen', 'Mingkui Tan']
2020-03-01
intelligent-home-3d-automatic-3d-house-design-1
http://openaccess.thecvf.com/content_CVPR_2020/html/Chen_Intelligent_Home_3D_Automatic_3D-House_Design_From_Linguistic_Descriptions_Only_CVPR_2020_paper.html
http://openaccess.thecvf.com/content_CVPR_2020/papers/Chen_Intelligent_Home_3D_Automatic_3D-House_Design_From_Linguistic_Descriptions_Only_CVPR_2020_paper.pdf
cvpr-2020-6
['text-to-3d']
['computer-vision']
[ 4.24984068e-01 5.13609469e-01 6.64894819e-01 -5.54496586e-01 -2.37261266e-01 -4.45100576e-01 7.84083366e-01 -3.01560223e-01 5.06804049e-01 5.78879714e-01 5.74357092e-01 -3.64997238e-01 3.61763656e-01 -1.58378851e+00 -8.00761700e-01 -4.21395123e-01 3.02688777e-01 6.43910885e-01 -4.21175033e-01 -3.78544480e-01 -4.12416160e-01 2.82663882e-01 -1.44836283e+00 4.13053423e-01 7.98729360e-01 9.13818657e-01 5.75525224e-01 6.88341677e-01 -1.77903533e-01 1.24511826e+00 -3.12366486e-01 -3.08976680e-01 1.74264103e-01 -8.62843275e-01 -8.32259357e-01 6.64473236e-01 -5.00803478e-02 -3.36795390e-01 1.04749091e-02 7.46018469e-01 4.80586916e-01 1.61357209e-01 9.30968523e-01 -1.06624007e+00 -1.09258425e+00 7.78175533e-01 -1.60630345e-01 -8.71498883e-01 7.17333138e-01 2.99037248e-01 1.12483621e+00 -6.55904174e-01 7.77580678e-01 1.32632017e+00 3.39427114e-01 3.59576344e-01 -1.42258716e+00 -6.18068837e-02 3.71154010e-01 -2.39227936e-01 -1.51117206e+00 -2.64884025e-01 1.08951116e+00 -6.95201933e-01 7.68240511e-01 3.06073129e-01 1.27238655e+00 1.22165596e+00 -8.66240263e-03 9.68931675e-01 1.02778447e+00 -5.09129703e-01 4.91897166e-01 -9.29195900e-03 -6.87945008e-01 1.02655768e+00 -2.57530510e-01 -1.36257097e-01 -9.88580734e-02 3.58691484e-01 9.67117190e-01 5.17234067e-03 -5.45663714e-01 -3.86316270e-01 -8.88537586e-01 8.82188141e-01 7.69314289e-01 1.42956227e-01 -4.77901399e-01 1.14146069e-01 -1.09158568e-01 1.62942499e-01 3.38497609e-01 3.33512276e-01 -9.31236297e-02 2.47050241e-01 -9.35110271e-01 5.07621169e-01 9.71394598e-01 1.60265970e+00 7.20818341e-01 -4.13027555e-02 -4.21522945e-01 7.22013652e-01 6.49816036e-01 7.24275947e-01 -3.46267410e-02 -4.58933026e-01 4.71270472e-01 7.20488191e-01 6.01324551e-02 -9.43941832e-01 -2.54964441e-01 -2.87437916e-01 -1.15239775e+00 1.02757908e-01 1.44609034e-01 -1.76533580e-01 -9.60227609e-01 1.70636153e+00 1.97027966e-01 -4.18941349e-01 -2.07753733e-01 9.05668855e-01 8.81226838e-01 9.55419242e-01 1.60715818e-01 1.53539613e-01 1.42305005e+00 -1.18890905e+00 -8.00270796e-01 -4.52717274e-01 4.44068432e-01 -7.69395053e-01 1.48205733e+00 1.14444517e-01 -9.94634151e-01 -6.78568363e-01 -1.12311113e+00 -3.96919876e-01 -4.14896876e-01 3.29760194e-01 5.49550653e-01 4.05063182e-01 -1.02901340e+00 3.85410041e-01 -5.60428441e-01 -7.81629324e-01 4.06267822e-01 -7.23220780e-03 -1.16835631e-01 -2.46781498e-01 -9.19382989e-01 7.20520377e-01 2.02211812e-01 4.15555358e-01 -6.85585737e-01 -4.25144464e-01 -1.21105039e+00 1.15145005e-01 2.54955143e-01 -1.49180186e+00 1.29528880e+00 -8.07923198e-01 -1.56467843e+00 8.90308321e-01 5.87096512e-02 1.64536089e-02 6.45143330e-01 -4.43683900e-02 -1.48637131e-01 -5.06811976e-01 2.23822474e-01 6.28465176e-01 7.94554710e-01 -1.64032865e+00 -4.32493299e-01 -2.58267373e-01 1.23492673e-01 2.06616253e-01 4.19991821e-01 -5.36957562e-01 -4.57895458e-01 -8.46906066e-01 1.16370292e-02 -8.77466381e-01 -2.22608417e-01 -2.76100766e-02 -9.69794631e-01 9.51562375e-02 4.42077458e-01 -1.00452018e+00 1.17155993e+00 -2.07234311e+00 3.00873160e-01 5.01912832e-01 9.77469012e-02 -5.13434112e-01 -1.00705095e-01 6.25786245e-01 7.07081631e-02 1.08972237e-01 -3.49767208e-01 -7.25330293e-01 6.55469954e-01 3.83707732e-02 -3.55019659e-01 -8.51224363e-03 1.76261410e-01 1.31149995e+00 -7.86097050e-01 -2.90483683e-01 4.17448938e-01 6.75134718e-01 -7.50851214e-01 7.09206581e-01 -7.37804592e-01 6.21694386e-01 -6.40233874e-01 5.27845740e-01 4.28930253e-01 -5.04949868e-01 4.32160050e-01 -2.84675688e-01 4.20233943e-02 2.20008641e-01 -9.67767656e-01 1.85020947e+00 -8.96256924e-01 2.64121294e-01 -2.24868342e-01 -4.67622459e-01 1.18595803e+00 1.72850564e-01 1.06389366e-01 -7.89876580e-01 2.84088582e-01 -1.06232120e-02 -3.89768392e-01 -5.18005908e-01 3.08640569e-01 -3.05841148e-01 -3.84808779e-01 5.23801744e-01 -1.87428683e-01 -6.85299098e-01 -1.31319195e-01 -2.01095253e-01 1.04067457e+00 4.86862808e-01 2.46858105e-01 -8.90684575e-02 2.47825295e-01 -2.17561156e-01 -1.09485937e-02 3.61423194e-01 6.78281486e-01 7.67651379e-01 4.09655899e-01 -5.79980552e-01 -1.34801829e+00 -1.10964453e+00 4.87332284e-01 7.27991998e-01 -3.67655188e-01 -5.54429293e-01 -1.09872723e+00 -3.72760624e-01 -2.40576014e-01 9.71525550e-01 -7.48297870e-01 -5.11336885e-02 -4.13209677e-01 -2.79252738e-01 -1.60127714e-01 5.28507531e-01 8.20242465e-01 -1.32566428e+00 -8.08935285e-01 2.78613925e-01 -1.90823227e-01 -1.19256067e+00 -5.75640559e-01 3.16045314e-01 -2.84574628e-01 -7.23534822e-01 -7.19310403e-01 -1.07250392e+00 8.93092632e-01 -3.86810660e-01 1.44648349e+00 7.71184191e-02 -2.39136398e-01 1.86018646e-01 -5.99808037e-01 -2.80514717e-01 -3.70657861e-01 4.05099481e-01 -5.17402232e-01 1.05775177e-01 -1.39768496e-02 -1.01591015e+00 -4.70413834e-01 -1.03654020e-01 -7.56459892e-01 1.12415135e+00 5.58740318e-01 6.12254322e-01 7.24033177e-01 2.83256114e-01 -1.11338124e-01 -1.03374672e+00 6.41010880e-01 -1.22922532e-01 -4.26246434e-01 4.14704382e-01 -7.95520246e-02 2.35110402e-01 9.12158489e-01 6.11264221e-02 -1.14762771e+00 4.35108960e-01 -3.04027915e-01 -7.72936493e-02 -4.13198173e-01 4.46958572e-01 -1.00069308e+00 4.86706555e-01 2.85856664e-01 2.56156802e-01 -5.28495789e-01 -4.51404929e-01 7.32257664e-01 3.93609703e-01 2.55917847e-01 -8.35586488e-01 1.19205439e+00 1.57705203e-01 2.65036221e-03 -9.02923763e-01 -8.94803524e-01 3.52204621e-01 -8.47067416e-01 -3.00189286e-01 1.35643005e+00 -7.46659756e-01 -4.72094387e-01 3.75576705e-01 -1.26916814e+00 -1.09717882e+00 -3.96251976e-01 -2.20326617e-01 -1.01176381e+00 -2.95312911e-01 -3.19776922e-01 -8.28172326e-01 -2.03931957e-01 -1.04772913e+00 1.44037914e+00 -2.64498144e-02 -5.27419269e-01 -9.01376784e-01 -4.13028821e-02 1.90699488e-01 3.41675311e-01 9.42554832e-01 1.47102666e+00 3.48692656e-01 -9.75039661e-01 1.06624402e-01 -1.22620963e-01 1.51276946e-01 5.12214601e-01 -2.53871441e-01 -1.11286700e+00 1.75828636e-01 -1.25256017e-01 -4.47857492e-02 2.25864395e-01 4.10436004e-01 1.05594540e+00 -7.18584359e-01 -1.52419344e-01 9.09790456e-01 1.59793353e+00 3.97399545e-01 7.05481529e-01 2.52367526e-01 9.70131755e-01 8.11346412e-01 1.19586773e-01 5.45196056e-01 6.88290954e-01 6.69366181e-01 1.07663438e-01 -3.64773840e-01 -2.80026406e-01 -1.26771176e+00 4.37973849e-02 7.13804185e-01 -2.74745226e-01 -5.54260373e-01 -7.62516439e-01 2.40781695e-01 -1.79627991e+00 -7.91002572e-01 -6.03737123e-03 1.74230993e+00 5.64739406e-01 -5.34244813e-02 -5.54296151e-02 1.64370075e-01 3.19004923e-01 2.06635430e-01 -2.31812328e-01 -3.51050079e-01 -1.47889972e-01 3.42657059e-01 -9.19662565e-02 6.10294878e-01 -1.13218284e+00 1.15470672e+00 5.02844477e+00 3.47028106e-01 -8.07108641e-01 -3.19589525e-01 9.47697997e-01 1.98047295e-01 -4.94003505e-01 -8.20215046e-02 -3.41510862e-01 2.82847106e-01 3.69989634e-01 7.86488429e-02 9.63855803e-01 9.63846147e-01 5.14418006e-01 5.67605868e-02 -1.46623397e+00 1.13370240e+00 -7.75456056e-02 -1.14092863e+00 3.89414340e-01 2.27585331e-01 7.95189321e-01 -6.74310029e-01 -3.32631990e-02 2.91749239e-01 6.04362667e-01 -1.35861766e+00 1.34523678e+00 8.11702609e-01 8.49073291e-01 -6.47336662e-01 3.69904280e-01 2.60866344e-01 -1.61307180e+00 2.40971446e-01 -5.14843129e-03 -2.10125938e-01 4.84944820e-01 4.06158030e-01 -8.48812163e-01 5.30655086e-01 4.44374323e-01 4.54952449e-01 -5.78916788e-01 6.69310987e-01 -8.63038480e-01 2.33239219e-01 -2.60437816e-01 -1.06056839e-01 9.95546356e-02 -4.32358146e-01 -1.16647013e-01 9.63110626e-01 7.71293521e-01 2.49941319e-01 4.27372634e-01 1.51106083e+00 -1.84903573e-03 1.57018587e-01 -9.96579945e-01 -1.09550253e-01 1.56922132e-01 1.08523297e+00 -8.18527281e-01 -7.25003406e-02 -1.65427297e-01 1.56197810e+00 2.86651284e-01 5.74902713e-01 -7.64026999e-01 -1.96010828e-01 2.32032225e-01 5.26122987e-01 3.46045673e-01 -2.32895449e-01 -4.39934462e-01 -1.01237595e+00 7.66009092e-02 -8.05743158e-01 -3.25752974e-01 -1.40760195e+00 -1.30700624e+00 8.33402276e-01 -3.42880160e-01 -8.40688229e-01 -2.79316485e-01 -4.87886041e-01 -5.34314036e-01 1.03616047e+00 -1.17851317e+00 -2.00369954e+00 -6.40585244e-01 4.38476771e-01 5.45081258e-01 1.26736119e-01 1.00652742e+00 2.73966312e-01 -2.74352968e-01 2.86100209e-01 -6.84504092e-01 3.98706347e-01 5.48335630e-03 -1.36825216e+00 7.09346056e-01 5.52955449e-01 9.88558456e-02 4.26967829e-01 7.30886221e-01 -7.39574373e-01 -1.32187033e+00 -1.62068450e+00 1.14582646e+00 -3.75299543e-01 2.86579788e-01 -9.30975616e-01 -2.72556931e-01 8.80690634e-01 2.69426733e-01 -5.75440824e-01 5.66115201e-01 -1.53988585e-01 1.03181697e-01 2.61200994e-01 -8.98064017e-01 9.17325377e-01 1.55138159e+00 -5.33976734e-01 -3.18442613e-01 3.02381754e-01 6.17667317e-01 -3.88168752e-01 -8.03448737e-01 1.25475645e-01 4.21623498e-01 -8.25905561e-01 6.36877000e-01 -1.07854359e-01 7.84576118e-01 -5.54762185e-01 -5.34403026e-01 -1.43356764e+00 -5.87311566e-01 -6.49786890e-01 1.88972548e-01 1.33929729e+00 5.54150522e-01 6.54619411e-02 6.68951392e-01 6.90519273e-01 -1.66280955e-01 -6.26426697e-01 -2.68324584e-01 -3.55431855e-01 -2.93808550e-01 -5.47352552e-01 1.30725098e+00 7.70696104e-01 -2.34121248e-01 8.02840471e-01 -4.23973054e-01 1.26939043e-01 5.11803031e-01 4.17209655e-01 1.01264691e+00 -1.13774145e+00 -5.75631082e-01 -3.57770950e-01 5.17947562e-02 -1.27823758e+00 3.64427902e-02 -9.89121020e-01 2.75365919e-01 -2.25101495e+00 -2.96532419e-02 -3.78126293e-01 5.84310651e-01 5.12165785e-01 2.41724253e-01 -1.53827339e-01 3.43931288e-01 -3.40413481e-01 -2.29450315e-01 1.00388479e+00 1.69643664e+00 -4.08298284e-01 -4.33006912e-01 -1.20628618e-01 -7.55088270e-01 5.71736813e-01 5.17404854e-01 3.09223551e-02 -7.32479990e-01 -5.48876762e-01 5.43037951e-01 1.63303658e-01 6.21640861e-01 -1.04068851e+00 -1.70347959e-01 -1.19495124e-01 6.84126675e-01 -4.06226754e-01 2.74157375e-01 -9.90126371e-01 7.36945689e-01 2.00371698e-01 -9.76250544e-02 6.21987469e-02 -1.46138817e-01 1.44966900e-01 1.71549395e-02 3.14352423e-01 4.69895095e-01 -4.90635693e-01 -4.28355485e-01 3.98377806e-01 -3.34390670e-01 -3.38516831e-01 8.16635430e-01 -1.37647286e-01 5.08870631e-02 -6.65193319e-01 -1.16835046e+00 9.39040408e-02 7.37960517e-01 5.01027822e-01 7.12145805e-01 -1.80006635e+00 -4.70299542e-01 6.07700050e-01 8.40356275e-02 4.15071726e-01 3.95516485e-01 4.28051092e-02 -7.07054198e-01 3.44002455e-01 -5.32632973e-03 -2.05145329e-01 -7.04400778e-01 6.36531770e-01 3.58442903e-01 -3.75284553e-01 -9.24174786e-01 6.39755666e-01 5.51330984e-01 -6.06133461e-01 -5.81847578e-02 -8.09696674e-01 -9.00512114e-02 -1.42252952e-01 1.92665830e-01 -1.85155943e-01 -2.43299901e-01 -5.80335021e-01 1.01447202e-01 6.87574089e-01 6.37808561e-01 -2.15148538e-01 1.59749496e+00 2.01533325e-02 -2.47896269e-01 4.88323361e-01 1.03567135e+00 -4.12021428e-01 -1.18970811e+00 2.10341781e-01 -1.68274581e-01 -1.64895013e-01 -5.27359806e-02 -8.98976982e-01 -9.64349985e-01 7.95141459e-01 4.57687795e-01 2.09560245e-01 1.17989242e+00 1.90644249e-01 8.55160952e-01 1.83365047e-01 8.71752918e-01 -1.00612390e+00 -4.31489758e-02 3.64985436e-01 1.47831392e+00 -7.25308359e-01 -3.64782572e-01 -6.44274294e-01 -5.31131148e-01 9.10943329e-01 4.66284513e-01 -8.69188905e-02 7.42808819e-01 2.99419671e-01 -2.61919886e-01 -4.27235931e-01 -5.53655028e-01 -3.09546202e-01 5.07688999e-01 7.84248233e-01 5.50458610e-01 4.80964303e-01 3.66458178e-01 8.85189831e-01 -1.13291693e+00 1.07812338e-01 7.07174465e-03 9.27968502e-01 -1.93639994e-01 -1.16930759e+00 -1.73559889e-01 1.84775695e-01 3.56967151e-01 -1.62638947e-01 -5.17510474e-01 5.79777360e-01 5.11813283e-01 8.51871133e-01 -4.77085076e-02 -6.91821098e-01 8.08672249e-01 1.28107414e-01 6.55100584e-01 -8.65796447e-01 -3.03799301e-01 1.42477632e-01 1.41121000e-01 -4.24695045e-01 -2.44871691e-01 -4.74800110e-01 -9.06077087e-01 -4.08449084e-01 3.87260914e-01 -7.35652223e-02 2.72683591e-01 9.36256230e-01 -1.27413322e-03 1.01100647e+00 4.69867855e-01 -9.34534848e-01 2.38057464e-01 -8.82325888e-01 -8.78042161e-01 4.37015802e-01 1.16100274e-01 -4.63047832e-01 9.72286910e-02 5.69993973e-01]
[9.378976821899414, -2.948323965072632]
2a0187dd-a7cf-4289-8cd1-b27230f29a87
advancing-biomedicine-with-graph
2306.10456
null
https://arxiv.org/abs/2306.10456v2
https://arxiv.org/pdf/2306.10456v2.pdf
Advancing Biomedicine with Graph Representation Learning: Recent Progress, Challenges, and Future Directions
Graph representation learning (GRL) has emerged as a pivotal field that has contributed significantly to breakthroughs in various fields, including biomedicine. The objective of this survey is to review the latest advancements in GRL methods and their applications in the biomedical field. We also highlight key challenges currently faced by GRL and outline potential directions for future research.
['Cui Tao', 'Zenan Sun', 'Yi Nian', 'Fang Li']
2023-06-18
null
null
null
null
['graph-representation-learning']
['methodology']
[ 5.02355576e-01 1.64444819e-01 -3.03750932e-01 -1.56534463e-01 -6.38591051e-01 -1.54126987e-01 2.26149216e-01 8.18545341e-01 2.65691094e-02 7.50715911e-01 1.52430698e-01 -4.54290181e-01 -3.02784085e-01 -6.69948280e-01 -1.48406878e-01 -9.89244461e-01 -5.13623476e-01 2.63047487e-01 -2.60555241e-02 -1.62080377e-01 2.20331773e-01 8.25814545e-01 -1.08905423e+00 3.07369858e-01 5.14171183e-01 4.74179894e-01 1.08242959e-01 5.96795380e-01 -1.35284603e-01 9.17480469e-01 -6.24902964e-01 -2.82287925e-01 -2.75639474e-01 -7.55058587e-01 -6.17388904e-01 -2.40978420e-01 1.66899398e-01 5.25411665e-01 -7.17187166e-01 9.97378349e-01 7.78624535e-01 3.84272635e-01 8.68125796e-01 -8.85986924e-01 -7.02180505e-01 3.25830102e-01 -8.01754415e-01 5.38776517e-01 4.10447121e-01 -5.60033143e-01 8.28960657e-01 -8.35795045e-01 6.34481907e-01 1.23742700e+00 5.89539349e-01 6.47267759e-01 -1.04281044e+00 -3.68210614e-01 3.01018119e-01 5.08882225e-01 -1.52887058e+00 -1.96586400e-01 1.01773047e+00 -3.42395127e-01 9.11075175e-01 1.09530956e-01 5.04743695e-01 6.84094667e-01 9.64569628e-01 8.31786335e-01 1.02124012e+00 -6.20142221e-01 1.64227366e-01 -4.17651266e-01 3.49939436e-01 9.57165837e-01 6.74062192e-01 -3.31787437e-01 -4.78982389e-01 -3.83490831e-01 8.40964079e-01 6.71165437e-02 -2.56441802e-01 -3.09003472e-01 -9.75118756e-01 1.23720765e+00 4.66926426e-01 8.49562526e-01 -1.78830609e-01 2.80719221e-01 5.06908417e-01 1.50108665e-01 5.50053596e-01 4.85852450e-01 -2.89237127e-02 2.86492348e-01 -6.65680051e-01 -3.15207988e-02 4.57220554e-01 7.04587638e-01 1.63735330e-01 3.24644476e-01 5.78388646e-02 9.78670955e-01 3.07934850e-01 -2.79592350e-02 3.04431796e-01 -4.11629736e-01 9.20292437e-02 6.52238488e-01 -6.03212118e-01 -1.36707067e+00 -7.30525970e-01 -6.53538942e-01 -1.23464477e+00 -3.40630740e-01 -1.21882893e-02 1.19100586e-01 -5.31482518e-01 1.28100812e+00 1.48960650e-01 4.57556933e-01 -1.39785245e-01 4.36790645e-01 1.48561752e+00 4.49863851e-01 2.56368041e-01 -3.79541039e-01 1.32419062e+00 -7.05724418e-01 -1.00253868e+00 -9.74819511e-02 7.46350527e-01 -5.46619534e-01 2.93301314e-01 2.92248040e-01 -6.60130203e-01 -2.69524515e-01 -8.30873251e-01 -1.38261154e-01 -5.38128614e-01 8.45403373e-02 1.02127349e+00 7.08756030e-01 -1.24221396e+00 2.90970355e-01 -9.80600119e-01 -7.71097839e-01 6.05876207e-01 4.54086244e-01 -4.36824501e-01 -4.60285991e-01 -1.37213135e+00 7.72974372e-01 2.06310645e-01 3.91477913e-01 -5.19028425e-01 -3.71560633e-01 -1.17570734e+00 -3.43887299e-01 1.72501534e-01 -6.49822056e-01 7.18803048e-01 4.68347013e-01 -1.15182543e+00 1.01705921e+00 -1.83607876e-01 -3.87904346e-01 6.58431835e-03 1.07531853e-01 -5.62494874e-01 1.57911047e-01 -1.03766933e-01 3.59816194e-01 1.69000085e-02 -8.95280838e-01 -2.92470098e-01 -5.02363503e-01 -4.68641698e-01 1.53904974e-01 -1.10862568e-01 6.43355623e-02 -5.57428122e-01 -8.10295343e-01 2.08484650e-01 -7.86795318e-01 -7.82816052e-01 -1.29528731e-01 -2.42193639e-01 -4.97024596e-01 4.72709179e-01 -4.69748497e-01 1.39216340e+00 -1.77539110e+00 4.29115891e-01 2.54754543e-01 5.25815725e-01 2.95922160e-01 -1.96593389e-01 9.90092933e-01 -1.10968657e-01 1.21816158e-01 -1.42250612e-01 -1.05323739e-01 -6.75414324e-01 1.91197425e-01 -1.75341770e-01 9.21753228e-01 -1.00914463e-01 1.23585618e+00 -1.39109910e+00 -3.92227918e-01 5.34357309e-01 7.10583389e-01 7.41707236e-02 -2.13617310e-02 8.97243842e-02 7.75865614e-01 -8.14691544e-01 8.19615126e-01 1.58557042e-01 -5.61931610e-01 7.75707006e-01 -5.05558439e-02 1.37636364e-01 1.40213668e-01 -6.20082259e-01 1.60071504e+00 -3.29462183e-03 6.02112532e-01 -2.95413822e-01 -1.53099990e+00 1.22484529e+00 2.53244013e-01 8.22052598e-01 -2.75909901e-01 2.20906660e-01 3.30891348e-02 -5.55277942e-03 -3.51918250e-01 -1.47086740e-01 -2.30184481e-01 6.55350536e-02 1.25208914e-01 -6.45543560e-02 -4.30511124e-02 1.01705767e-01 2.79457897e-01 1.10327101e+00 -2.07898885e-01 1.24159384e+00 -1.96374416e-01 9.20094609e-01 -2.85775125e-01 3.80335003e-01 6.40612483e-01 -3.42546940e-01 2.97897935e-01 6.21539235e-01 -6.49848521e-01 -2.19277546e-01 -6.73481703e-01 -1.16251484e-01 8.05764735e-01 4.11489829e-02 -6.91861272e-01 -2.55358756e-01 -7.60797739e-01 -1.31443992e-01 8.78197402e-02 -9.10129249e-01 -1.18854731e-01 -6.82009399e-01 -1.22330391e+00 4.93442863e-01 5.03515542e-01 7.36714341e-03 -1.08220911e+00 3.57930809e-02 4.22031403e-01 -5.34553900e-02 -1.22859085e+00 -5.67747615e-02 -7.73526058e-02 -1.35817564e+00 -1.48206949e+00 -8.28788042e-01 -1.12064791e+00 1.09240532e+00 6.10538065e-01 9.01781499e-01 2.90572047e-01 -7.93393135e-01 4.20098037e-01 -4.84080374e-01 -3.71044844e-01 -2.68094063e-01 1.51992962e-01 -1.38943344e-01 -4.29313034e-01 2.92955011e-01 -2.02643529e-01 -5.41507185e-01 -8.69129598e-03 -9.07035589e-01 -2.85726905e-01 3.95472169e-01 8.06017756e-01 1.06446862e+00 2.83414572e-01 9.11356926e-01 -1.62506831e+00 9.68013883e-01 -5.46969414e-01 -4.83094543e-01 5.16079187e-01 -5.59348941e-01 -2.41733000e-01 4.95276242e-01 1.11524820e-01 -5.27469873e-01 -3.43542427e-01 -4.48461592e-01 1.52826309e-01 2.52058983e-01 1.03816783e+00 -1.93165411e-04 -5.93791187e-01 4.76512164e-01 2.22966820e-03 1.23161435e-01 -6.19282842e-01 2.99050003e-01 4.46063161e-01 1.59205511e-01 -4.50588167e-01 1.89394847e-01 4.49361056e-01 7.09307253e-01 -1.30815065e+00 -8.84015799e-01 -7.26716936e-01 -3.16295505e-01 -2.47833148e-01 5.86780429e-01 -5.11638939e-01 -5.72560549e-01 2.71520406e-01 -9.71277237e-01 -1.01348907e-01 1.02607518e-01 4.73362595e-01 -2.79503495e-01 8.20398748e-01 -8.94558191e-01 -6.68183804e-01 -6.60594523e-01 -1.15071404e+00 9.74964559e-01 9.13693607e-02 -1.10230163e-01 -1.58633101e+00 4.51343179e-01 3.53998899e-01 2.51370877e-01 5.87527812e-01 1.24893093e+00 -7.16461539e-01 1.59624554e-02 -5.80522239e-01 -3.46638635e-02 6.68086708e-02 5.14733255e-01 -1.89905807e-01 -8.56506824e-01 -6.71249628e-01 -3.06011532e-02 -1.71489552e-01 8.69085312e-01 5.99682212e-01 1.38701117e+00 2.56098390e-01 -1.02904558e+00 3.18413556e-01 1.41917253e+00 3.41856182e-01 5.23238599e-01 -2.58311361e-01 7.98407316e-01 5.52270114e-01 4.03540343e-01 -4.57048751e-02 2.89944232e-01 5.05742073e-01 1.27369881e-01 -4.51383591e-01 -4.95082855e-01 -2.38699406e-01 -2.01177094e-02 1.46898150e+00 -2.09484413e-01 -3.97060126e-01 -8.83208752e-01 2.29891986e-01 -1.96043289e+00 -5.97638369e-01 -2.66355664e-01 1.90184510e+00 3.67339492e-01 -2.04314739e-01 -5.37520796e-02 1.80985481e-01 8.20665300e-01 2.83137292e-01 -3.56436849e-01 -2.59647697e-01 -1.89836472e-01 3.66333038e-01 2.49826059e-01 5.05565226e-01 -9.21522558e-01 1.15948737e+00 8.20580196e+00 6.90392792e-01 -1.06476355e+00 -1.76268294e-01 6.08553827e-01 5.94444394e-01 -1.15091562e-01 -3.18974525e-01 -6.95245326e-01 -1.26004145e-01 6.90878451e-01 -2.58133948e-01 2.21608207e-01 7.23518550e-01 2.94733167e-01 1.80969797e-02 -8.42789114e-01 1.01915967e+00 4.60491985e-01 -1.55363572e+00 3.61130685e-01 6.61297739e-02 6.23971641e-01 2.46049240e-01 -1.44614667e-01 8.97366703e-02 6.39748052e-02 -1.11236882e+00 -6.48523211e-01 3.42952728e-01 1.07926679e+00 -5.18338978e-01 1.13199294e+00 -3.33343223e-02 -1.55055380e+00 4.26501989e-01 -6.70989335e-01 -2.04626322e-01 8.94232094e-02 8.43857229e-01 -1.11251688e+00 1.07356024e+00 2.23733008e-01 1.31972909e+00 -3.74944478e-01 1.33868229e+00 -5.28656244e-01 5.70784867e-01 2.53637999e-01 -1.26823530e-01 -3.01571824e-02 -3.16326976e-01 3.28487515e-01 1.33656144e+00 1.89838428e-02 1.89415202e-01 4.19993103e-01 3.08749795e-01 -1.01229191e-01 5.37004054e-01 -9.44273651e-01 -6.54369771e-01 4.41396445e-01 1.01505435e+00 -1.09708297e+00 -4.71509844e-02 -6.03115201e-01 4.75914598e-01 5.63522339e-01 3.62583637e-01 -4.03811246e-01 -5.19019067e-01 4.57728386e-01 -1.83190450e-01 -2.26926163e-01 -3.85902464e-01 8.41086283e-02 -1.09067023e+00 -6.17449999e-01 -6.91239238e-01 1.01758301e+00 -2.99106866e-01 -1.43932462e+00 6.38315439e-01 1.19471848e-02 -9.94322479e-01 5.64292260e-02 -7.99942970e-01 -2.60190368e-01 7.18927622e-01 -1.81201994e+00 -9.76173699e-01 4.14211378e-02 1.32542849e-01 5.90628922e-01 -1.63809612e-01 1.18741238e+00 2.21540779e-01 -7.22414076e-01 4.77918208e-01 2.15369791e-01 1.21532641e-01 5.27895689e-01 -8.61202121e-01 6.11480594e-01 4.72671509e-01 4.00094032e-01 9.77860332e-01 4.52466130e-01 -7.73160696e-01 -1.42331266e+00 -1.26758659e+00 1.16599929e+00 -1.21122971e-01 6.14460409e-01 -1.71480581e-01 -9.12015498e-01 6.43228412e-01 -2.78673768e-01 2.22273931e-01 1.28687894e+00 -8.22078530e-03 -1.73479691e-02 5.92197441e-02 -1.08352304e+00 5.66421628e-01 8.25817823e-01 -5.57580411e-01 -2.44185135e-01 6.24972463e-01 2.19676837e-01 -3.58552575e-01 -1.07858407e+00 4.82713491e-01 4.00141865e-01 -3.53054851e-01 1.00106704e+00 -8.10624361e-01 -3.49427134e-01 -4.01686251e-01 1.93354607e-01 -1.23667288e+00 -7.03570366e-01 -5.79726219e-01 -3.07900280e-01 5.74821174e-01 -1.67805061e-01 -7.72476196e-01 8.71086776e-01 1.96037620e-01 -2.70151883e-01 -1.44300318e+00 -9.11969900e-01 -5.28132796e-01 1.89323276e-01 -1.86706200e-01 1.77489191e-01 7.63725877e-01 4.65473294e-01 4.64845777e-01 -2.74723560e-01 -2.91865498e-01 5.78088820e-01 -6.97830841e-02 4.34882939e-01 -1.36150181e+00 7.94676021e-02 -5.00246167e-01 -8.67505908e-01 -8.15490484e-01 3.87137383e-01 -1.44820547e+00 -2.63198256e-01 -2.26384330e+00 2.80819982e-01 -1.20444693e-01 -9.59279180e-01 6.33237064e-01 -3.00119072e-01 5.61453640e-01 -2.07227260e-01 3.41732316e-02 -6.39228582e-01 1.52129486e-01 1.26829004e+00 -2.57474571e-01 -1.14480324e-01 1.30487001e-02 -1.12327218e+00 3.90633196e-01 9.62357223e-01 -3.40120912e-01 -6.23885870e-01 -1.32278383e-01 3.25211942e-01 5.11402003e-02 -3.50425184e-01 -5.00135303e-01 7.79294074e-02 -1.59111679e-01 1.95229843e-01 -4.45820808e-01 7.72991031e-02 -4.57303792e-01 1.17366754e-01 5.98400176e-01 -2.51324505e-01 1.65023401e-01 1.43862933e-01 8.20341229e-01 -1.59695223e-01 -3.17494422e-02 7.24555731e-01 -2.30417088e-01 -6.95041955e-01 3.35065693e-01 -5.74977517e-01 -1.17288217e-01 1.08081400e+00 -7.32087344e-02 -3.88245106e-01 -1.92290440e-01 -5.90368509e-01 2.75521576e-01 1.55888677e-01 3.87132019e-01 8.57686996e-01 -9.85424578e-01 -5.98886907e-01 4.03040349e-02 3.64751309e-01 -2.61029184e-01 4.24696118e-01 8.19354117e-01 -3.93176734e-01 8.18673253e-01 1.69462524e-02 -2.71716177e-01 -1.41742241e+00 5.39947808e-01 1.23941764e-01 -4.84214902e-01 -9.75474894e-01 7.63148427e-01 2.44902670e-01 -5.14857471e-02 3.07222724e-01 5.26522137e-02 -8.60988438e-01 -3.60607505e-01 6.31460071e-01 7.05846369e-01 4.19289261e-01 -4.24496293e-01 -8.07075322e-01 6.24708116e-01 -2.80274123e-01 5.50093889e-01 1.48962069e+00 5.99594191e-02 -6.33493841e-01 7.92329907e-01 1.21472108e+00 -1.16184697e-01 -1.65214419e-01 8.14115396e-04 1.12596169e-01 -1.12750441e-01 1.08699434e-01 -4.41924363e-01 -1.13984132e+00 8.36665273e-01 3.11338276e-01 -1.74425468e-02 7.41628289e-01 3.89552623e-01 5.81891835e-01 3.36151123e-01 9.17178452e-01 -5.21658719e-01 2.98511591e-02 3.56916368e-01 9.40358698e-01 -9.32499349e-01 5.50274849e-01 -9.20105636e-01 -2.92328477e-01 1.15352607e+00 2.57839765e-02 -1.12665504e-01 9.41144109e-01 -1.71380371e-01 1.07536793e-01 -6.73847854e-01 -4.11928296e-01 -8.53867903e-02 6.26118779e-01 8.03672671e-01 1.30347002e+00 3.35519016e-02 -8.73458326e-01 2.03512296e-01 3.77388328e-01 2.28377860e-02 3.16873908e-01 1.13787699e+00 -2.27595150e-01 -1.76212728e+00 -5.66222072e-02 8.72130930e-01 -3.61890912e-01 -5.59023349e-03 -4.84921098e-01 5.49391329e-01 -3.86669844e-01 9.94244277e-01 -5.25769711e-01 -1.65315554e-01 7.84783065e-02 -1.60303786e-01 8.16570759e-01 -1.01317775e+00 -2.32585594e-01 2.85696238e-01 -2.43650917e-02 -3.97615314e-01 -3.41474533e-01 -5.90895414e-01 -1.36092949e+00 1.29476234e-01 -2.75267452e-01 4.52426970e-01 5.32807887e-01 6.13997698e-01 5.60779452e-01 8.02152574e-01 2.78274804e-01 -4.59737033e-01 1.31721303e-01 -6.40347183e-01 -8.73596668e-01 -2.77025729e-01 1.75469532e-03 -9.80534673e-01 -2.53652215e-01 -1.33058861e-01]
[7.12399959564209, 6.395130634307861]
577565aa-9ad8-4c72-b8f7-981c00626be1
automatic-keyphrase-generation-by
null
null
https://aclanthology.org/2022.coling-1.204
https://aclanthology.org/2022.coling-1.204.pdf
Automatic Keyphrase Generation by Incorporating Dual Copy Mechanisms in Sequence-to-Sequence Learning
The keyphrase generation task is a challenging work that aims to generate a set of keyphrases for a piece of text. Many previous studies based on the sequence-to-sequence model were used to generate keyphrases, and they introduce a copy mechanism to achieve good results. However, we observed that most of the keyphrases are composed of some important words (seed words) in the source text, and if these words can be identified accurately and copied to create more keyphrases, the performance of the model might be improved. To address this challenge, we propose a DualCopyNet model, which introduces an additional sequence labeling layer for identifying seed words, and further copies the words for generating new keyphrases by dual copy mechanisms. Experimental results demonstrate that our model outperforms the baseline models and achieves an obvious performance improvement.
['Yin Wang', 'Yao Huang', 'Jianhui Jiang', 'Siyu Wang']
null
null
null
null
coling-2022-10
['keyphrase-generation']
['natural-language-processing']
[ 2.30636299e-01 -1.47961944e-01 -4.30287659e-01 2.36102924e-01 -8.66838396e-01 -9.01049972e-01 9.94700909e-01 3.38539124e-01 -4.14406478e-01 8.61947179e-01 8.11515212e-01 -4.14440662e-01 5.02375722e-01 -8.97962928e-01 -6.26944304e-01 -4.37641501e-01 2.64340043e-01 4.21653353e-02 6.13671720e-01 -6.36666059e-01 7.18788445e-01 1.62527382e-01 -1.32256997e+00 4.81695741e-01 8.72474074e-01 3.23471159e-01 5.28925717e-01 7.63509154e-01 -8.80507648e-01 8.75808954e-01 -1.05175376e+00 -3.40499401e-01 3.10297519e-01 -5.55728078e-01 -8.29660118e-01 -2.79249936e-01 2.71032393e-01 -7.29546309e-01 -5.51565051e-01 1.08453715e+00 3.60245913e-01 2.47121695e-02 4.90466475e-01 -1.34349859e+00 -7.82405317e-01 1.45962298e+00 -4.76874173e-01 8.30404013e-02 5.42508960e-01 5.86463585e-02 1.26389897e+00 -1.09144807e+00 5.53719282e-01 1.17661309e+00 3.38729560e-01 3.07882786e-01 -5.96417606e-01 -1.09369719e+00 1.86630726e-01 1.64639622e-01 -1.44637620e+00 -1.42168790e-01 6.02420866e-01 -1.50937159e-02 7.11812675e-01 2.42446512e-01 6.59936368e-01 1.07844114e+00 1.81326836e-01 1.10951924e+00 8.76871049e-01 -5.54434001e-01 -1.80150047e-01 -1.36000350e-01 2.45765075e-01 5.46813071e-01 4.58204180e-01 -4.59458232e-02 -5.74416697e-01 -4.63635951e-01 7.77310967e-01 1.19824067e-01 -2.47561038e-01 3.71644825e-01 -1.67445469e+00 7.42006302e-01 1.49547905e-01 4.61613119e-01 -5.15928984e-01 1.40576318e-01 4.72805381e-01 1.86678082e-01 3.40830594e-01 8.07700217e-01 -3.78514707e-01 -2.16664553e-01 -1.12963843e+00 7.19290674e-01 8.53643477e-01 1.14971936e+00 8.88425648e-01 -2.80549139e-01 -8.27951252e-01 5.60117245e-01 -1.72906350e-02 4.89280403e-01 6.98716760e-01 -4.17763621e-01 7.18719602e-01 8.33391607e-01 3.64459068e-01 -1.07667780e+00 2.92964354e-02 -2.16351122e-01 -7.94143081e-01 -4.92726177e-01 -1.37776166e-01 -3.31505597e-01 -1.26965809e+00 1.30885661e+00 7.82568529e-02 3.34913075e-01 -1.51045844e-01 4.21133786e-01 8.79361153e-01 1.24453783e+00 -1.14431709e-01 -1.87940344e-01 1.50530362e+00 -1.29163551e+00 -1.03389716e+00 -2.64259249e-01 4.56849724e-01 -1.26037681e+00 1.20590830e+00 1.25981033e-01 -7.89463758e-01 -6.50216877e-01 -1.07089579e+00 -5.51885776e-02 -4.45482105e-01 1.74599111e-01 4.26703453e-01 3.45544189e-01 -9.09685075e-01 4.12143290e-01 -3.26861978e-01 -1.39362961e-01 5.04931547e-02 -3.46015155e-01 8.72683898e-03 4.71370406e-02 -1.55927348e+00 4.56343263e-01 1.06107163e+00 -3.52390975e-01 -9.79658544e-01 -8.88093948e-01 -7.00512409e-01 3.73555198e-02 8.70236456e-01 -5.84282577e-01 1.54264104e+00 -4.20303255e-01 -1.07696390e+00 2.93388844e-01 -3.95756394e-01 -2.71161884e-01 3.42554897e-01 -3.62058848e-01 -4.33174014e-01 2.89370596e-01 4.05101061e-01 6.48309827e-01 1.12969756e+00 -1.23894882e+00 -1.09950244e+00 3.61338705e-01 1.86363891e-01 1.78632140e-01 -5.78292608e-01 3.01183730e-01 -7.82881558e-01 -1.54764318e+00 -2.20513985e-01 -7.57881641e-01 -3.76574129e-01 -5.07138789e-01 -7.23330140e-01 -3.55606556e-01 6.81844592e-01 -8.52052808e-01 1.96751845e+00 -1.85748363e+00 -2.97437429e-01 2.80510217e-01 3.40312839e-01 4.57661211e-01 -4.20770079e-01 1.15274131e+00 3.14903781e-02 6.59440517e-01 -2.18387797e-01 8.23162124e-02 -1.53719172e-01 -8.68905410e-02 -1.20589197e+00 -4.73924905e-01 -5.65785402e-03 1.25575387e+00 -1.10312164e+00 -4.88742322e-01 -3.03219378e-01 -4.83910032e-02 -7.49832839e-02 5.07262468e-01 -6.31579518e-01 -2.86207587e-01 -6.25785351e-01 3.33573341e-01 5.48946559e-01 -3.26352835e-01 -1.00334495e-01 -2.56450742e-01 1.91967562e-02 5.61358571e-01 -1.28471088e+00 1.33114052e+00 -3.31751227e-01 2.25693464e-01 -4.53529567e-01 -2.50914872e-01 8.71863484e-01 4.17587101e-01 4.09837037e-01 -1.97853521e-01 2.02239528e-02 1.52980402e-01 -2.86721408e-01 -1.79773480e-01 1.25367069e+00 2.28468865e-01 -1.60872146e-01 1.11643136e+00 -3.09639126e-01 -4.23348516e-01 5.94726861e-01 9.40099776e-01 1.10697305e+00 -1.00984521e-01 4.59921032e-01 6.77284226e-03 5.53730965e-01 2.31163025e-01 1.89756170e-01 1.03231633e+00 3.02447677e-01 6.27644122e-01 4.45740581e-01 -2.82100528e-01 -1.23410606e+00 -5.77978075e-01 5.85742116e-01 1.03114235e+00 2.13478312e-01 -1.14007723e+00 -7.68292844e-01 -1.00838029e+00 -2.09150612e-01 7.70501196e-01 -4.98381227e-01 -3.39643240e-01 -6.51698232e-01 -5.44056654e-01 8.75331879e-01 5.94187796e-01 5.30002415e-01 -1.12146914e+00 -7.88160320e-03 6.31683171e-01 -6.47316217e-01 -1.13654971e+00 -1.25792050e+00 -3.15606385e-01 -3.87266576e-01 -9.79433417e-01 -1.20672154e+00 -8.30561757e-01 8.36051881e-01 8.17873120e-01 1.18287671e+00 5.37574351e-01 -5.15779778e-02 9.05726478e-02 -1.01978934e+00 -5.62344551e-01 -7.44972885e-01 4.35543209e-01 -1.90303519e-01 -1.30812004e-01 2.39155188e-01 -2.27456793e-01 -4.15822327e-01 -3.19713205e-02 -1.44414556e+00 3.02721024e-01 8.01899254e-01 7.18245149e-01 3.25804830e-01 3.66519690e-01 3.85398746e-01 -8.27123344e-01 1.45623207e+00 -2.04712063e-01 -3.43178302e-01 4.73398954e-01 -8.96695912e-01 3.62684667e-01 9.08920288e-01 -6.94164455e-01 -8.79235148e-01 -4.38431919e-01 -1.53128460e-01 -3.71407643e-02 1.76144153e-01 5.89198411e-01 1.99571580e-01 1.70545548e-01 4.59428281e-01 9.86964345e-01 -2.60754436e-01 -5.27555645e-01 7.77231812e-01 6.99214816e-01 3.32782984e-01 -7.31292307e-01 1.35005152e+00 2.45084375e-01 -4.32024360e-01 -4.74305868e-01 -9.39040363e-01 -5.44963598e-01 -3.45578223e-01 1.22890450e-01 4.07665312e-01 -1.07384884e+00 -6.32710010e-02 7.12110937e-01 -1.41366112e+00 3.42945382e-02 -2.79392481e-01 7.04589486e-02 -3.62145305e-02 8.82080257e-01 -6.20223284e-01 -4.39842522e-01 -9.60160792e-01 -9.70997572e-01 1.17581820e+00 3.65375429e-01 -4.57198173e-01 -5.93658566e-01 4.24745027e-03 -1.52671248e-01 2.73018420e-01 -3.41495395e-01 1.11568606e+00 -8.02973270e-01 -6.07172132e-01 -2.86360562e-01 -2.90972352e-01 3.07350695e-01 5.39950132e-01 1.00265548e-01 -3.00398916e-01 -2.21434757e-01 -3.34982008e-01 -2.11605877e-01 9.60437596e-01 -3.06396127e-01 1.23177552e+00 -7.86886513e-01 -4.40690398e-01 2.36238658e-01 1.02725852e+00 2.68910229e-01 6.96562171e-01 3.25235814e-01 8.19305420e-01 3.21110487e-01 5.88097334e-01 4.15187627e-01 7.27792680e-01 2.24151880e-01 -2.18687067e-03 -1.55852899e-01 -2.18182594e-01 -1.13222253e+00 4.88611698e-01 1.11577880e+00 3.33149314e-01 -3.49106401e-01 -6.82618797e-01 8.31572950e-01 -1.66520858e+00 -9.36149716e-01 -1.73953772e-01 1.76080406e+00 1.34721947e+00 3.15241665e-01 4.97545935e-02 2.30204254e-01 7.26782143e-01 6.45263076e-01 -1.65524885e-01 7.35133067e-02 -1.54870227e-01 4.48002160e-01 5.56125224e-01 3.54251206e-01 -6.89753890e-01 1.36724079e+00 7.25396395e+00 1.29343069e+00 -9.18000638e-01 -2.85675019e-01 3.57563764e-01 4.47614789e-01 -8.46372306e-01 3.22593391e-01 -1.29417169e+00 6.69262886e-01 5.07172406e-01 -7.45344043e-01 2.87662536e-01 6.82633340e-01 5.64542226e-02 8.72786492e-02 -5.25169194e-01 8.14910352e-01 2.36594155e-01 -1.58923864e+00 8.22672129e-01 -2.54840344e-01 9.57447410e-01 -4.11746204e-01 -2.90125310e-01 2.93756604e-01 8.05406511e-01 -7.24583209e-01 5.72384238e-01 3.41090024e-01 5.92606306e-01 -7.04951346e-01 5.77716827e-01 3.92754436e-01 -1.48430347e+00 1.10494308e-01 -3.38435143e-01 1.21632583e-01 2.61241585e-01 7.96286941e-01 -1.13779974e+00 5.80574512e-01 3.06949168e-01 4.62092459e-01 -8.25653970e-01 8.47527742e-01 -8.93472373e-01 8.17943990e-01 -1.17812686e-01 -6.32682502e-01 4.84865040e-01 1.94667317e-02 4.10454094e-01 1.30604863e+00 4.72957075e-01 2.94039156e-02 5.50244629e-01 7.60066092e-01 -4.12059933e-01 3.97297025e-01 -3.16476911e-01 -6.86926365e-01 8.34011257e-01 1.62381887e+00 -9.02472973e-01 -9.60810363e-01 -2.07876325e-01 1.19245970e+00 -2.07263395e-01 3.78345191e-01 -6.18013680e-01 -9.99813735e-01 5.62321901e-01 1.51292130e-01 3.34904134e-01 -4.36315060e-01 -5.23835644e-02 -1.28507066e+00 1.81578144e-01 -1.33404171e+00 2.61443734e-01 -9.80364919e-01 -1.17451048e+00 4.66934651e-01 1.58485234e-01 -1.12316442e+00 -2.82977223e-01 -1.02918290e-01 -7.68188655e-01 9.62411702e-01 -1.56884813e+00 -1.20962596e+00 -2.50616193e-01 3.75452012e-01 7.74123549e-01 -1.62799120e-01 4.86565560e-01 -1.44193158e-01 -5.10006174e-02 5.54328918e-01 -7.71181136e-02 5.04791141e-01 8.75396311e-01 -1.25166142e+00 1.41042602e+00 1.30686557e+00 3.70158434e-01 1.00653696e+00 5.39687335e-01 -1.17854297e+00 -1.26613855e+00 -1.10439241e+00 1.19672227e+00 -3.50703865e-01 8.96184623e-01 -4.11505044e-01 -9.57623482e-01 4.37608242e-01 5.56970179e-01 -5.62853277e-01 4.76122051e-01 -4.32887614e-01 -5.71569622e-01 2.02262580e-01 -4.50224340e-01 1.18856084e+00 1.07239807e+00 -6.11741066e-01 -1.00015378e+00 3.43684286e-01 1.51392281e+00 -3.74242783e-01 -4.30726618e-01 2.83210307e-01 3.37506264e-01 -4.30658132e-01 8.51431549e-01 -5.43353319e-01 6.65487766e-01 -5.95334053e-01 3.06783408e-01 -1.60516810e+00 -3.66293997e-01 -1.13066185e+00 -4.74975407e-01 1.44043016e+00 4.70784217e-01 -4.70396906e-01 5.11447728e-01 4.64462787e-02 1.20987497e-01 -4.57321256e-01 -1.96945891e-01 -7.44277775e-01 -2.58911923e-02 -2.85878599e-01 1.34760606e+00 7.56557941e-01 -1.52704507e-01 6.11514986e-01 -5.28903127e-01 -1.30258664e-01 3.23508114e-01 1.23680890e-01 1.00515044e+00 -8.00790906e-01 -7.35746144e-05 -5.28734505e-01 4.21092451e-01 -1.58346999e+00 2.94297258e-03 -9.12359536e-01 1.44917339e-01 -1.78504348e+00 2.40161002e-01 -2.65679777e-01 -9.67138037e-02 7.22710490e-01 -1.18405139e+00 -9.56673324e-02 2.99008965e-01 4.22079384e-01 -1.70679554e-01 5.28262556e-01 1.42470312e+00 -3.30173105e-01 -1.69653103e-01 -1.53771237e-01 -1.30026853e+00 2.95612097e-01 6.26904905e-01 -6.20833397e-01 -4.95782733e-01 -1.36649281e-01 4.96218950e-01 -2.47406304e-01 2.29775836e-03 -5.71097672e-01 4.06911105e-01 -5.58941305e-01 1.63470387e-01 -8.83660913e-01 -1.53173149e-01 -5.62263906e-01 -1.90800905e-01 5.40431976e-01 -4.67969984e-01 4.51567739e-01 6.34299815e-02 3.23684424e-01 -2.62260079e-01 -4.24055040e-01 2.50917196e-01 -4.71147299e-01 -8.37848604e-01 6.13045335e-01 -4.76299226e-01 1.33938864e-01 8.76186192e-01 2.25771721e-02 -3.85951281e-01 -6.30816579e-01 2.84515828e-01 2.53835589e-01 2.99739480e-01 8.55620265e-01 7.53807902e-01 -1.52625668e+00 -1.03156459e+00 1.58480659e-01 3.81856084e-01 -4.66082878e-02 -2.72349596e-01 4.89124507e-02 -4.94553655e-01 4.13150579e-01 1.81071013e-01 2.12835491e-01 -1.04901278e+00 6.13006890e-01 -2.07431331e-01 -5.64625084e-01 -5.69070995e-01 7.83739805e-01 -1.28959343e-01 -3.42134893e-01 7.71454573e-02 -5.20560026e-01 -3.26588780e-01 2.74762064e-01 1.26471972e+00 1.00532576e-01 -4.10576537e-02 -2.69100636e-01 1.75993070e-01 4.35093641e-01 -7.32543409e-01 -2.73359776e-01 9.75439370e-01 -1.97342150e-02 -4.26544964e-01 6.19727634e-02 9.37293887e-01 3.77237111e-01 -7.28885710e-01 -6.43702507e-01 9.53960195e-02 -3.18658441e-01 -2.08488479e-01 -7.45488107e-01 -6.81567311e-01 4.86648470e-01 -1.40341371e-01 1.65187761e-01 1.00906551e+00 -2.00063884e-01 1.61269081e+00 5.36055207e-01 3.63467008e-01 -1.12019134e+00 4.76215422e-01 9.52239633e-01 8.38125885e-01 -9.01132107e-01 8.32226053e-02 -5.17419755e-01 -5.75223327e-01 1.22553611e+00 5.97074687e-01 1.53106183e-01 5.60061753e-01 4.33285296e-01 2.68692225e-01 1.37342751e-01 -7.73011744e-01 -2.83238471e-01 4.22629148e-01 3.50968659e-01 1.84426442e-01 -1.83314905e-01 -6.84659183e-01 6.77241564e-01 -5.30519128e-01 -1.57708582e-02 6.91534758e-01 1.11621952e+00 -7.67809629e-01 -1.55732000e+00 -3.69669586e-01 6.07837915e-01 -5.78218639e-01 -7.79110253e-01 -8.52276444e-01 2.82500297e-01 -1.77580759e-01 8.91579747e-01 -2.63109535e-01 -7.25760520e-01 1.94237918e-01 1.69269040e-01 -1.08941607e-01 -9.35024261e-01 -7.72633672e-01 4.25987784e-03 -1.28422156e-01 -2.27851391e-01 4.46650141e-04 6.37202635e-02 -1.34886932e+00 -3.93520981e-01 -3.43918085e-01 5.36677659e-01 3.78377587e-01 8.70256841e-01 4.26562577e-01 4.96533364e-01 9.81594563e-01 -2.89465785e-01 -6.36858582e-01 -1.12878966e+00 -3.06563675e-01 3.78500640e-01 1.77814499e-01 9.74724889e-02 -1.26033604e-01 3.56677949e-01]
[12.308065414428711, 8.889082908630371]
59680308-574d-401c-80a9-d7dfd9180e5d
bayesian-optimization-meets-self-distillation
2304.12666
null
https://arxiv.org/abs/2304.12666v1
https://arxiv.org/pdf/2304.12666v1.pdf
Bayesian Optimization Meets Self-Distillation
Bayesian optimization (BO) has contributed greatly to improving model performance by suggesting promising hyperparameter configurations iteratively based on observations from multiple training trials. However, only partial knowledge (i.e., the measured performances of trained models and their hyperparameter configurations) from previous trials is transferred. On the other hand, Self-Distillation (SD) only transfers partial knowledge learned by the task model itself. To fully leverage the various knowledge gained from all training trials, we propose the BOSS framework, which combines BO and SD. BOSS suggests promising hyperparameter configurations through BO and carefully selects pre-trained models from previous trials for SD, which are otherwise abandoned in the conventional BO process. BOSS achieves significantly better performance than both BO and SD in a wide range of tasks including general image classification, learning with noisy labels, semi-supervised learning, and medical image analysis tasks.
['Donggeun Yoo', 'Suyeong Park', 'Gi-hyeon Lee', 'Hyeonsoo Lee', 'Heon Song', 'Hyunjae Lee']
2023-04-25
null
null
null
null
['learning-with-noisy-labels', 'learning-with-noisy-labels']
['computer-vision', 'natural-language-processing']
[ 2.01202407e-01 -9.68312398e-02 -4.54578251e-01 -5.39350688e-01 -1.08212566e+00 -2.11996436e-01 5.35165906e-01 6.80893734e-02 -6.09699070e-01 8.62615883e-01 1.14783451e-01 3.17230076e-02 -2.57256359e-01 -9.12585706e-02 -5.47050714e-01 -1.01716220e+00 3.36152554e-01 5.20435333e-01 9.64025706e-02 2.51112789e-01 2.26429179e-01 1.26204088e-01 -9.17337358e-01 1.80025980e-01 8.42769444e-01 9.60615754e-01 2.66866922e-01 3.72334063e-01 1.02092542e-01 4.82447475e-01 -5.71294904e-01 -2.09510937e-01 -5.94611131e-02 -3.07700366e-01 -6.80034339e-01 1.77897111e-01 2.00438961e-01 -2.21709043e-01 -7.05279186e-02 1.04784596e+00 5.64936638e-01 3.03942204e-01 4.99448240e-01 -7.86958098e-01 -3.92798036e-01 8.56843829e-01 -4.05156583e-01 1.98564947e-01 3.26535553e-02 3.55749816e-01 9.02971506e-01 -7.93970764e-01 4.10222381e-01 1.13733721e+00 5.92641234e-01 4.47382599e-01 -1.53217125e+00 -7.77145803e-01 4.02422965e-01 1.61435172e-01 -1.36978209e+00 -4.54203069e-01 6.73033655e-01 -4.39727962e-01 7.80102372e-01 -1.24034368e-01 4.70669627e-01 1.21402657e+00 -3.81547026e-02 9.97139812e-01 1.12390089e+00 -3.65544349e-01 5.06780267e-01 4.36514556e-01 4.98717308e-01 6.22591376e-01 2.01713383e-01 1.10350788e-01 -7.45642304e-01 -3.20481479e-01 3.10778230e-01 -2.46659443e-01 -4.50790197e-01 -3.50803643e-01 -1.24686897e+00 5.70607662e-01 3.59182358e-01 2.20487267e-01 -4.86887276e-01 1.20379604e-01 2.44854942e-01 4.59415019e-02 3.31501305e-01 9.24843132e-01 -7.75058031e-01 -8.13552514e-02 -9.01787460e-01 5.49764149e-02 5.98457038e-01 7.22520888e-01 1.01919949e+00 8.87398571e-02 -4.50505733e-01 1.22004247e+00 4.59561795e-01 5.49266040e-01 6.41726375e-01 -1.01454616e+00 4.86267060e-01 4.67345446e-01 1.38009369e-01 -6.13864005e-01 -4.77906644e-01 -7.88244545e-01 -6.63613260e-01 -2.76887268e-01 2.57520825e-01 -2.37057075e-01 -1.19721627e+00 1.62820232e+00 4.03098494e-01 1.73023939e-01 -4.35935669e-02 8.84833038e-01 8.81618857e-01 4.56666321e-01 1.40102789e-01 -3.01311553e-01 1.00207591e+00 -8.88878405e-01 -5.42663991e-01 -4.67060834e-01 5.75397968e-01 -4.47093993e-01 1.04717910e+00 8.28804672e-01 -7.11375594e-01 -4.56496060e-01 -1.10218203e+00 4.19352114e-01 6.55954257e-02 3.84225935e-01 4.98972237e-01 5.75991333e-01 -7.34847248e-01 7.50785112e-01 -8.87208581e-01 -1.91426780e-02 7.84876943e-01 6.13325715e-01 -1.26610100e-01 -3.51763248e-01 -9.49197173e-01 6.97743058e-01 4.12587494e-01 4.13783699e-01 -1.05437469e+00 -7.18952835e-01 -5.91734231e-01 -4.52874005e-02 8.82445097e-01 -6.56688511e-01 1.34280813e+00 -6.02492750e-01 -1.85440636e+00 4.37991291e-01 -2.77054489e-01 -3.29882562e-01 3.41943085e-01 -5.29602587e-01 4.68404293e-02 -1.35903478e-01 -2.65896589e-01 8.62348735e-01 9.85554338e-01 -1.26770175e+00 -4.08881724e-01 -3.05637360e-01 -3.34341198e-01 1.94096729e-01 -4.53444660e-01 -1.75832927e-01 -7.56766856e-01 -4.24656123e-01 3.73623610e-01 -1.16975856e+00 -4.24532145e-01 -3.16413164e-01 -7.01396465e-01 -3.00425023e-01 3.17570359e-01 -3.44328344e-01 1.19800973e+00 -1.98280013e+00 1.61128581e-01 3.92625183e-01 2.16991276e-01 5.67375243e-01 -2.33720765e-01 -5.67058362e-02 2.02202097e-01 2.52225637e-01 -1.14432544e-01 -3.35696608e-01 -1.62661791e-01 3.65002126e-01 -1.40381232e-01 2.13485286e-01 7.54073113e-02 7.91767716e-01 -9.71531689e-01 -4.57873583e-01 1.24769695e-01 2.78027564e-01 -4.70982581e-01 1.29435852e-01 -4.32641923e-01 8.26365948e-01 -7.97100663e-01 5.52277565e-01 4.56331074e-01 -7.62891710e-01 2.42426768e-01 -3.18695098e-01 5.41809797e-01 3.56102079e-01 -1.06703138e+00 1.51607978e+00 -3.78185540e-01 4.17356104e-01 -3.34840596e-01 -9.71244097e-01 9.33949769e-01 1.85565695e-01 4.55258310e-01 -4.85054791e-01 1.64257497e-01 3.07882279e-02 7.43276104e-02 -6.16134167e-01 -1.05227008e-02 6.56536296e-02 2.14393958e-01 4.09109414e-01 1.87489614e-01 -1.66832939e-01 8.57783854e-02 -1.63654029e-01 1.00441146e+00 3.08718830e-01 3.98459911e-01 5.17518856e-02 3.45630974e-01 -2.22837478e-02 9.07417536e-01 1.18124986e+00 -3.20255905e-01 5.52629828e-01 3.82030070e-01 -2.82461941e-01 -7.07710326e-01 -9.11032438e-01 -4.49172020e-01 1.09598672e+00 1.05789550e-01 -4.65144604e-01 -5.49686849e-01 -9.99271810e-01 -1.11845590e-01 6.95580721e-01 -5.65029800e-01 -4.04936284e-01 -4.98659611e-01 -1.31307137e+00 2.88431466e-01 4.15765852e-01 5.22508442e-01 -9.77664411e-01 -1.98609591e-01 1.94904432e-01 -1.07190438e-01 -1.13520122e+00 -3.12228858e-01 3.58597815e-01 -1.23951745e+00 -8.69339049e-01 -6.10194325e-01 -3.24427783e-01 6.56833172e-01 -6.65462539e-02 8.95080328e-01 -3.45951729e-02 -2.81788427e-02 -7.19112009e-02 -3.06890279e-01 -3.14106256e-01 -4.06089514e-01 2.72378743e-01 -2.25674901e-02 -8.82639363e-02 2.28794575e-01 -1.78242803e-01 -4.62198555e-01 7.11419642e-01 -3.84111553e-01 -4.00158800e-02 8.74638617e-01 1.03253984e+00 7.87957847e-01 5.88698648e-02 5.27931154e-01 -1.24780428e+00 4.48102266e-01 -4.08231497e-01 -4.75403607e-01 4.97074842e-01 -1.12583721e+00 3.95827889e-01 3.93650800e-01 -7.39128530e-01 -1.22602701e+00 -7.76558891e-02 1.13235384e-01 -4.72201884e-01 -2.64031082e-01 5.68207502e-01 2.57435590e-02 2.19051495e-01 8.19052219e-01 -1.01859197e-01 -1.70094535e-01 -7.70265758e-01 -4.64768062e-04 5.88077664e-01 3.19366574e-01 -7.24303782e-01 2.76500195e-01 3.46303284e-02 -1.86406627e-01 -4.91388947e-01 -1.30921364e+00 -3.28362554e-01 -4.98566270e-01 -9.57539678e-02 4.54946399e-01 -7.90105283e-01 -4.65609699e-01 6.06856585e-01 -8.28937948e-01 -6.05029583e-01 -5.44132618e-03 8.44152272e-01 -1.34056807e-01 1.16532706e-01 -2.26965323e-01 -7.56871819e-01 -1.27754197e-01 -1.35224438e+00 8.63799095e-01 3.47224146e-01 -4.14315075e-01 -9.60205674e-01 -5.62147647e-02 6.81443870e-01 2.31434241e-01 -2.83703268e-01 9.18208063e-01 -7.94361293e-01 -5.37987590e-01 -2.36778393e-01 -1.56403169e-01 5.44331193e-01 1.19097650e-01 -1.86757594e-01 -1.16926670e+00 -3.70201498e-01 2.24726889e-02 -5.41787207e-01 1.19690597e+00 7.11793840e-01 1.32817185e+00 -1.09530970e-01 -4.95171636e-01 6.26232624e-01 9.68947053e-01 2.31293395e-01 2.15028554e-01 3.76719445e-01 6.20899677e-01 4.42926943e-01 4.92019683e-01 3.61148149e-01 5.79011776e-02 7.63845146e-01 8.95067975e-02 1.84702709e-01 -9.50319096e-02 -1.00653291e-01 3.62005949e-01 5.22405922e-01 1.40053287e-01 -8.80440101e-02 -9.35194314e-01 6.79967627e-02 -1.74728501e+00 -3.66702408e-01 1.46366566e-01 2.25060654e+00 1.11959612e+00 3.99688214e-01 -1.81122944e-01 -8.79910812e-02 6.99131668e-01 1.13768995e-01 -1.14025116e+00 1.85744800e-02 -6.38396367e-02 -4.99579273e-02 3.85031283e-01 2.70633817e-01 -1.00499225e+00 8.82728696e-01 6.90542221e+00 9.80205476e-01 -1.11694622e+00 7.87539482e-02 7.74458349e-01 -3.56099844e-01 -1.40635476e-01 1.45660013e-01 -1.04610145e+00 4.11952853e-01 6.60740674e-01 2.31810793e-01 5.67683935e-01 8.08497548e-01 2.79062986e-01 -6.39401793e-01 -1.09024060e+00 8.77287090e-01 1.14107355e-01 -1.19040215e+00 -1.68616802e-01 -9.94680077e-02 1.06284893e+00 2.41583616e-01 1.95011541e-01 4.66254443e-01 6.26825333e-01 -8.42400670e-01 4.00337607e-01 5.35137653e-01 3.83978933e-01 -2.37852395e-01 1.07545924e+00 4.29692119e-01 -4.84533370e-01 -3.69059771e-01 -2.59917438e-01 4.78244126e-01 -6.62696958e-02 1.02336907e+00 -1.30912793e+00 3.86498839e-01 6.85268879e-01 7.46157169e-01 -7.67096043e-01 1.21273458e+00 -5.06956697e-01 1.33686411e+00 -4.86794800e-01 -3.41453925e-02 1.41614646e-01 -1.61380112e-01 5.20368516e-01 1.04469538e+00 6.04485814e-03 -1.53559923e-01 2.57273823e-01 8.23667705e-01 -8.83953199e-02 -1.79528110e-02 9.31576639e-03 1.26475453e-01 6.49917424e-01 1.01969540e+00 -5.77821732e-01 -3.25839579e-01 -2.86913943e-02 2.11588472e-01 2.70161688e-01 6.11390531e-01 -5.37135303e-01 1.02523662e-01 2.11066619e-01 -2.98058987e-01 4.34593588e-01 2.29256172e-02 -5.53484857e-01 -9.22783971e-01 -2.59658009e-01 -7.19354510e-01 5.85747480e-01 -8.54433537e-01 -1.19781470e+00 5.13613522e-01 2.89083660e-01 -1.03133917e+00 -2.28010550e-01 -4.49693024e-01 -4.55179304e-01 6.48174465e-01 -1.53113413e+00 -6.68467522e-01 -2.86146045e-01 1.00673549e-01 4.72419828e-01 -3.09234500e-01 3.64432991e-01 -1.38433129e-01 -1.26140428e+00 6.44819856e-01 4.36841249e-01 -1.03085466e-01 1.07864618e+00 -1.21711338e+00 -1.47741958e-01 4.96461481e-01 2.65470415e-01 7.32251525e-01 6.49388909e-01 -5.93127310e-01 -1.01463270e+00 -8.33916128e-01 4.80328977e-01 -1.80992082e-01 3.57047290e-01 -9.91174579e-03 -1.12610412e+00 2.68454045e-01 -3.01833451e-01 -5.69958985e-02 6.78433061e-01 5.42804718e-01 -2.44696990e-01 -3.96916360e-01 -7.83205211e-01 5.10195255e-01 6.38820350e-01 -2.18866959e-01 -3.38521481e-01 4.06948358e-01 3.80923033e-01 -4.01099652e-01 -8.36066961e-01 5.94068348e-01 4.39514190e-01 -5.72666645e-01 9.54449832e-01 -6.08731568e-01 1.07128553e-01 -1.58176944e-01 -1.73367858e-01 -1.54143369e+00 -1.59516528e-01 -6.20341539e-01 9.08664428e-03 1.14861691e+00 7.57581830e-01 -5.08208692e-01 8.63775134e-01 7.66126156e-01 -1.11239634e-01 -1.13070548e+00 -6.28611207e-01 -6.79653406e-01 -1.82033107e-01 -6.81884527e-01 5.74448466e-01 6.41670823e-01 -2.86929190e-01 4.87894446e-01 -3.46482128e-01 1.08884588e-01 5.89805841e-01 7.58505613e-02 8.91008973e-01 -1.15651822e+00 -5.96189618e-01 -3.75849187e-01 9.84148383e-02 -1.33186579e+00 1.74812526e-01 -8.50306451e-01 5.21266639e-01 -1.16662180e+00 5.09095788e-01 -7.38151073e-01 -5.96662819e-01 8.41802537e-01 -6.86610162e-01 7.94431865e-02 5.52639924e-02 6.51448667e-01 -8.30375016e-01 5.52934825e-01 1.34085584e+00 9.83753521e-03 -5.41214824e-01 3.31752747e-01 -7.36066461e-01 8.45555782e-01 7.04134047e-01 -9.45291400e-01 -2.91221440e-01 -4.80563372e-01 2.17626363e-01 -8.30349922e-02 2.80138224e-01 -7.48438358e-01 2.44398192e-01 -4.58843261e-01 5.09837031e-01 -2.88270056e-01 2.93786168e-01 -4.29928243e-01 -8.26075114e-03 2.42099658e-01 -7.49783039e-01 -7.07970500e-01 -4.13483824e-04 8.13053548e-01 1.35613391e-02 -5.38348675e-01 7.99359620e-01 1.03008099e-01 -4.39898252e-01 2.17956170e-01 -1.39985234e-01 1.48214355e-01 4.41905648e-01 -1.59781873e-01 -3.01554263e-01 -1.34631470e-01 -1.15199947e+00 5.64952016e-01 8.76177847e-02 1.80950150e-01 3.29746962e-01 -8.69958401e-01 -5.78208089e-01 2.63052970e-01 9.50032696e-02 3.66381288e-01 1.99353844e-01 1.13556647e+00 2.04274878e-01 3.82475734e-01 2.21374899e-01 -1.07918549e+00 -1.14989042e+00 1.44518197e-01 2.82106906e-01 -4.61080998e-01 -4.64396238e-01 8.20194423e-01 5.73922209e-02 -3.51632208e-01 4.31513548e-01 -3.27131987e-01 -1.55400395e-01 -2.65642554e-02 3.06113154e-01 4.50951487e-01 3.93548399e-01 -2.01136265e-02 -1.94599435e-01 4.88427877e-01 -5.12980402e-01 -8.73610750e-02 1.35618865e+00 -5.92311919e-02 3.10696095e-01 6.34103656e-01 1.03945613e+00 -3.83329153e-01 -1.67311084e+00 -8.35015178e-01 1.24190897e-01 -4.51215744e-01 5.45225918e-01 -1.17529655e+00 -9.52159584e-01 7.85295844e-01 3.58146250e-01 -4.15700585e-01 9.71343458e-01 1.31428063e-01 5.79176247e-01 1.01637006e+00 2.87366360e-01 -1.25905097e+00 4.39443082e-01 3.25083852e-01 7.14009285e-01 -1.43854523e+00 3.32432240e-01 -1.08520702e-01 -9.87707794e-01 9.76507604e-01 6.40349150e-01 2.59876460e-01 6.64420068e-01 -5.94315343e-02 -3.56359221e-02 -8.26413780e-02 -9.41045225e-01 6.31843600e-03 4.91261035e-01 4.11156982e-01 5.55096641e-02 -2.49405473e-01 1.53885037e-01 7.14714646e-01 3.03442795e-02 1.93650499e-01 5.06315194e-02 7.07702637e-01 -5.09038448e-01 -9.87135530e-01 -4.55100596e-01 7.96416700e-01 -2.17950985e-01 6.46053860e-03 -2.10847110e-01 4.46791410e-01 7.32866600e-02 9.63858664e-01 -3.40427428e-01 -4.12262738e-01 1.31796569e-01 1.78877205e-01 3.67649585e-01 -6.91036344e-01 -5.69631934e-01 3.35493654e-01 1.46851406e-01 -4.64258283e-01 -2.83144921e-01 -7.66985714e-01 -1.03379154e+00 1.79373533e-01 -8.80194426e-01 1.38357371e-01 8.18829179e-01 1.23800480e+00 3.60817850e-01 5.78650296e-01 6.50747716e-01 -7.59703100e-01 -1.06237364e+00 -1.11039639e+00 -2.76984990e-01 1.45106554e-01 2.85417736e-01 -9.15814579e-01 -3.82028520e-01 -4.37924974e-02]
[9.337357521057129, 3.571859359741211]
19ad9ea8-af7c-4fd1-a747-45bcb4ed6a28
rgb-t-tracking-based-on-mixed-attention
2304.04264
null
https://arxiv.org/abs/2304.04264v4
https://arxiv.org/pdf/2304.04264v4.pdf
RGB-T Tracking Based on Mixed Attention
RGB-T tracking involves the use of images from both visible and thermal modalities. The primary objective is to adaptively leverage the relatively dominant modality in varying conditions to achieve more robust tracking compared to single-modality tracking. An RGB-T tracker based on mixed attention mechanism to achieve complementary fusion of modalities (referred to as MACFT) is proposed in this paper. In the feature extraction stage, we utilize different transformer backbone branches to extract specific and shared information from different modalities. By performing mixed attention operations in the backbone to enable information interaction and self-enhancement between the template and search images, it constructs a robust feature representation that better understands the high-level semantic features of the target. Then, in the feature fusion stage, a modality-adaptive fusion is achieved through a mixed attention-based modality fusion network, which suppresses the low-quality modality noise while enhancing the information of the dominant modality. Evaluation on multiple RGB-T public datasets demonstrates that our proposed tracker outperforms other RGB-T trackers on general evaluation metrics while also being able to adapt to longterm tracking scenarios.
['Jin Yu', 'Xiqing Guo', 'Mingtao Dong', 'Yang Luo']
2023-04-09
null
null
null
null
['rgb-t-tracking']
['computer-vision']
[ 1.11224249e-01 -1.89653620e-01 -8.98256674e-02 -1.72098801e-01 -9.21197712e-01 -5.44091821e-01 5.64058721e-01 -2.65141368e-01 -3.36471915e-01 2.41210938e-01 3.38079482e-01 3.03656280e-01 -2.37502679e-02 -2.04880714e-01 -5.67685902e-01 -1.10602462e+00 4.16225344e-01 -1.83001444e-01 3.54730725e-01 -9.00589004e-02 3.38326171e-02 2.63589621e-01 -1.67804837e+00 -8.10466334e-03 8.11987877e-01 1.56337667e+00 3.12939554e-01 4.13541257e-01 -1.22514434e-01 4.91836250e-01 -3.50006044e-01 -2.84898847e-01 4.73331183e-01 -1.20861515e-01 -5.76220304e-02 5.66704059e-03 7.68780887e-01 -1.32869110e-01 -4.06136811e-01 1.01691580e+00 7.15884566e-01 1.18079811e-01 2.12733403e-01 -1.40027916e+00 -5.63125610e-01 1.55272499e-01 -7.98351347e-01 3.40290308e-01 5.35353720e-01 6.74996495e-01 6.00086272e-01 -9.47790384e-01 3.19089592e-01 1.11133289e+00 6.77065730e-01 6.16289973e-01 -8.20417881e-01 -8.19904149e-01 2.46323809e-01 1.76359579e-01 -1.05079520e+00 -4.80872750e-01 9.05832291e-01 -1.61045000e-01 6.27304912e-01 2.91048020e-01 8.99857938e-01 9.85737801e-01 5.46848059e-01 7.66670823e-01 1.14150906e+00 -2.17576399e-01 -1.65838271e-01 9.65889767e-02 -5.16930073e-02 6.00347877e-01 1.62207916e-01 3.13191116e-01 -1.10938990e+00 -3.03580444e-02 5.27390420e-01 1.48728132e-01 -4.00222570e-01 -2.74400920e-01 -1.47223020e+00 1.71316847e-01 8.36966574e-01 5.20648956e-01 -6.08826101e-01 3.04669648e-01 8.18180144e-02 -8.23770314e-02 1.10990122e-01 -4.70076166e-02 -1.79221436e-01 -1.12876050e-01 -7.42637098e-01 -2.46446878e-01 1.77230105e-01 9.37774241e-01 6.78271890e-01 1.24555506e-01 -7.32572496e-01 2.86027938e-01 9.29697096e-01 1.08217430e+00 4.44877952e-01 -6.97603106e-01 4.65017051e-01 8.11363637e-01 1.60706803e-01 -6.08833611e-01 -4.02164966e-01 -8.71356249e-01 -5.70028961e-01 3.09879720e-01 3.53830248e-01 -1.07811518e-01 -1.22070849e+00 1.85264349e+00 7.81638324e-01 2.27618217e-01 -6.13751309e-03 1.36829042e+00 1.10610628e+00 1.96176380e-01 3.96300077e-01 -5.14736697e-02 1.79442286e+00 -7.52624929e-01 -9.18162286e-01 -2.71214396e-01 1.32226106e-02 -8.35899115e-01 6.43027067e-01 -1.42286137e-01 -8.58264148e-01 -9.29122686e-01 -1.10150909e+00 -5.24191298e-02 -4.58566487e-01 9.61432904e-02 4.71214831e-01 8.65967095e-01 -1.01396525e+00 -1.25796786e-02 -8.63272190e-01 -5.19152880e-01 3.11370909e-01 5.78133106e-01 -4.76385534e-01 9.34368223e-02 -1.05681658e+00 1.00676978e+00 3.12417001e-01 5.48993707e-01 -8.10017943e-01 -8.58574092e-01 -6.91011071e-01 -1.71803907e-01 4.62497234e-01 -9.38824892e-01 8.38278174e-01 -8.81496191e-01 -1.50942183e+00 4.28135931e-01 -1.48918375e-01 -2.42881794e-02 2.06694990e-01 -1.63499624e-01 -3.81722450e-01 1.19321398e-01 1.81827303e-02 7.11535990e-01 1.24823451e+00 -1.16835201e+00 -8.31589460e-01 -5.89205563e-01 -2.03484580e-01 6.09578550e-01 -4.09083873e-01 -4.52450961e-02 -8.59379113e-01 -4.57214862e-01 4.40098047e-01 -1.03274477e+00 2.31359884e-01 1.69659197e-01 -3.12608272e-01 -1.22190416e-01 1.11912405e+00 -6.59196317e-01 8.35058451e-01 -2.10970473e+00 3.45750034e-01 1.42176360e-01 4.85026330e-01 1.07005470e-01 3.40752751e-02 4.74911965e-02 1.55785039e-01 -4.14418995e-01 2.12641016e-01 -4.41731215e-01 -1.40820146e-02 -9.85761806e-02 1.18331864e-01 7.11309373e-01 6.13790229e-02 1.26878893e+00 -8.59338284e-01 -6.37076497e-01 5.25339842e-01 6.94087684e-01 -2.38391347e-02 1.15681775e-01 -1.16719082e-01 1.12272859e+00 -7.92499900e-01 1.16229022e+00 7.25023866e-01 -1.94867015e-01 -1.37968749e-01 -9.41089511e-01 -2.99746990e-01 -3.18750381e-01 -8.71425867e-01 2.01702523e+00 -3.20396185e-01 6.34674311e-01 2.96262592e-01 -2.96328694e-01 6.89290941e-01 2.88568020e-01 8.31054211e-01 -1.00443041e+00 5.64162612e-01 1.39302671e-01 -3.52223292e-02 -5.05357444e-01 6.76036537e-01 8.49822089e-02 -1.49419442e-01 2.76939154e-01 2.05614045e-01 3.80670309e-01 -3.05672079e-01 1.91427529e-01 9.15522099e-01 5.08638561e-01 -2.62341529e-01 1.06822170e-01 5.65490067e-01 -5.92226125e-02 5.94997704e-01 8.26618552e-01 -5.87140679e-01 2.07711577e-01 -5.35114110e-01 -1.84219927e-01 -6.05668962e-01 -1.04934192e+00 1.18486807e-01 1.19596243e+00 7.99525678e-01 -1.32059529e-01 -3.42043847e-01 -6.15154326e-01 1.89509287e-01 2.39363730e-01 -7.77477860e-01 -3.80673438e-01 -2.63650388e-01 -4.56026286e-01 3.47314745e-01 5.25547028e-01 9.15758848e-01 -7.64357746e-01 -9.70373392e-01 -7.29402378e-02 -4.42573518e-01 -1.08324158e+00 -7.63204336e-01 1.88021958e-01 -7.83293545e-01 -8.80491257e-01 -6.69604480e-01 -2.12625384e-01 3.94758284e-01 6.17690563e-01 4.87354755e-01 -8.74532461e-02 -1.20348446e-01 1.05810177e+00 -3.24046254e-01 -3.25433522e-01 1.37154832e-01 -1.02005251e-01 7.99425021e-02 5.96645892e-01 2.30669886e-01 -1.25468299e-01 -7.19717264e-01 2.47201726e-01 -8.27407777e-01 -1.04647376e-01 9.03385341e-01 6.06389940e-01 4.54335779e-01 -1.97792158e-01 -7.66880289e-02 2.18683690e-01 -1.02729816e-02 -2.65531361e-01 -5.35437286e-01 7.33591139e-01 -2.49432936e-01 2.00881585e-01 1.10920809e-01 -6.53181493e-01 -1.45988142e+00 2.69572049e-01 3.41459125e-01 -7.67117083e-01 1.75908566e-01 2.05868453e-01 -3.45915616e-01 -7.52191603e-01 3.48808676e-01 5.14713705e-01 -8.97689909e-03 -4.10808474e-01 4.70286816e-01 5.11447728e-01 7.02277899e-01 -3.59125257e-01 1.25326264e+00 7.00314760e-01 3.11308866e-03 -6.06309950e-01 -8.22666049e-01 -4.89167750e-01 -5.92417121e-01 -9.95180726e-01 1.07779384e+00 -1.06955898e+00 -1.09689653e+00 6.30432189e-01 -6.90993190e-01 1.85127750e-01 -1.44397989e-01 7.73635805e-01 -2.00988844e-01 1.34862021e-01 -2.56863236e-01 -9.29640353e-01 -4.82274473e-01 -1.23979080e+00 1.61728597e+00 1.02817726e+00 5.80969751e-01 -6.50460303e-01 3.39284763e-02 5.83573401e-01 8.47582221e-01 1.10589348e-01 4.77393828e-02 -2.51106590e-01 -1.01620281e+00 -1.89103067e-01 -1.94759578e-01 -2.03823850e-01 2.26820201e-01 -3.12869310e-01 -1.26425004e+00 -3.94754529e-01 -1.59546826e-02 -1.43406317e-01 8.34317923e-01 4.36966240e-01 5.35392523e-01 3.78436714e-01 -5.43409705e-01 8.68063271e-01 1.31093490e+00 8.67528319e-02 4.46022630e-01 3.69173825e-01 9.56279337e-01 1.71956718e-01 8.51371169e-01 1.46577403e-01 5.11062384e-01 8.74339819e-01 5.24362087e-01 -2.19096363e-01 -3.76994431e-01 -4.67503853e-02 5.86607337e-01 4.37131733e-01 -3.18269916e-02 1.11839011e-01 -5.83249629e-01 8.66333023e-02 -1.84242809e+00 -8.47225130e-01 1.78540766e-01 2.20737243e+00 3.74386966e-01 1.70747619e-02 1.58892184e-01 -3.32243681e-01 8.70642781e-01 -5.40324971e-02 -7.52969623e-01 4.41332728e-01 -2.58327246e-01 -1.72005594e-01 7.86292374e-01 -5.57811595e-02 -1.06763792e+00 7.06372201e-01 5.50122690e+00 7.07290411e-01 -1.32801008e+00 4.13205653e-01 -1.73731297e-02 -3.38460535e-01 -1.48750097e-01 6.90616891e-02 -7.26791322e-01 5.08893609e-01 7.08339274e-01 -2.43436508e-02 1.12286665e-01 3.20897490e-01 1.14293121e-01 -3.13062876e-01 -7.06759453e-01 1.14035833e+00 6.65811226e-02 -7.77833581e-01 -4.87065315e-01 1.77985430e-01 3.65737975e-01 1.38862208e-01 3.06842864e-01 2.47320592e-01 -1.36143893e-01 -5.10179043e-01 1.11161339e+00 1.08802009e+00 5.71015358e-01 -2.91876644e-01 5.89189231e-01 -3.14599536e-02 -1.86546576e+00 -2.95378447e-01 1.44164160e-01 5.85263431e-01 9.69025716e-02 3.00242782e-01 -3.54109854e-01 1.08888996e+00 9.48420107e-01 7.32552528e-01 -9.03392971e-01 1.30852604e+00 2.79939510e-02 1.92769587e-01 -6.22380793e-01 1.38402745e-01 1.34091038e-04 4.05940227e-02 8.39133441e-01 7.53690064e-01 3.78726840e-01 1.55116161e-02 3.51775259e-01 7.46919453e-01 2.46515229e-01 -2.70002782e-01 -2.41475090e-01 1.63173288e-01 6.17673159e-01 1.64298892e+00 -6.61429584e-01 -1.80369273e-01 -6.15858197e-01 8.32849979e-01 -3.68665345e-02 3.81766826e-01 -1.26893079e+00 1.19370125e-01 3.29950035e-01 -2.79791117e-01 4.13075864e-01 -2.58888990e-01 -7.57828308e-03 -1.09087598e+00 8.74512419e-02 -5.35988629e-01 4.50687289e-01 -1.21549380e+00 -9.84767854e-01 6.65929377e-01 -9.10813510e-02 -1.63491213e+00 2.53383130e-01 -3.43549073e-01 -3.03740710e-01 1.13048482e+00 -1.56217897e+00 -1.89947224e+00 -8.13791811e-01 8.85417163e-01 6.53307885e-02 -8.83949175e-03 1.01362981e-01 4.39137578e-01 -6.90644145e-01 6.77763879e-01 -1.99786946e-01 -2.47145146e-01 8.76563847e-01 -9.72406089e-01 -2.72468328e-01 1.02979922e+00 -1.64344937e-01 6.69926941e-01 5.82862079e-01 -1.02458441e+00 -2.31512022e+00 -8.65316629e-01 1.17264524e-01 -4.81864184e-01 4.82882738e-01 -9.21708420e-02 -6.14439905e-01 4.46119577e-01 3.75106424e-01 2.14130789e-01 3.56686175e-01 -2.19514281e-01 -3.18450034e-01 -3.57633561e-01 -1.08522928e+00 2.16244474e-01 8.54453981e-01 -5.71143508e-01 -5.24147868e-01 -1.34535655e-01 5.79025030e-01 -7.20170259e-01 -1.09902179e+00 3.93619686e-01 1.05709124e+00 -7.42450774e-01 9.95708525e-01 -4.14414983e-03 -3.49427432e-01 -8.44465554e-01 -1.18997343e-01 -9.11948621e-01 -3.14265698e-01 -5.91780722e-01 -2.71566987e-01 1.34356332e+00 -2.50722934e-02 -6.09897912e-01 6.36678696e-01 7.20576227e-01 -2.51218796e-01 -1.35160848e-01 -1.25177217e+00 -5.72663128e-01 -7.54459620e-01 -3.28830242e-01 4.91619170e-01 4.87237900e-01 -2.98630536e-01 1.16088539e-01 -5.66435754e-01 4.19999808e-01 1.22610474e+00 1.92440867e-01 6.32461548e-01 -1.04584396e+00 -8.39881897e-02 -2.47672945e-01 -4.02810246e-01 -8.13939452e-01 -2.11443946e-01 -8.22288811e-01 3.06745797e-01 -1.16988039e+00 3.36040050e-01 -3.42017710e-01 -6.47286296e-01 5.02308965e-01 -4.34079915e-01 4.80758369e-01 5.27315915e-01 3.24322283e-01 -1.03536344e+00 8.88796508e-01 1.48968565e+00 -2.46814549e-01 -1.97922781e-01 -1.93041429e-01 -7.68372715e-01 9.45668295e-02 3.32397789e-01 -3.66242349e-01 -1.49472477e-02 -3.89793158e-01 -2.26217046e-01 8.23077187e-02 6.64563894e-01 -1.38952553e+00 8.24000597e-01 9.96801704e-02 7.86268413e-01 -9.43356395e-01 5.90991437e-01 -1.29964292e+00 5.92869937e-01 4.25397575e-01 -3.96060571e-02 -5.73740378e-02 3.59789610e-01 8.16755831e-01 6.44848868e-02 3.80222470e-01 5.87554514e-01 9.99362320e-02 -9.01963532e-01 4.17173296e-01 2.38814093e-02 -4.20239329e-01 1.03484356e+00 -5.20084202e-01 -4.13508326e-01 -2.40244702e-01 -7.19859719e-01 5.49768686e-01 4.99478430e-01 7.35777676e-01 4.66937780e-01 -1.64993012e+00 -3.49308461e-01 2.11705476e-01 4.64856923e-01 -6.18740201e-01 5.65470338e-01 1.50712466e+00 2.32663244e-01 3.80907148e-01 -3.74839246e-01 -1.07168496e+00 -1.30641675e+00 4.07829285e-01 6.52548552e-01 -1.61832627e-02 -5.29389799e-01 6.32437468e-01 -1.03357233e-01 -5.14507666e-02 4.16908801e-01 -2.34350786e-01 -1.74852014e-01 8.12250655e-03 5.83345950e-01 1.93795100e-01 4.85916771e-02 -1.20714593e+00 -8.21236849e-01 9.29106653e-01 2.66338468e-01 -2.41547272e-01 9.53406036e-01 -6.48321807e-01 1.09367035e-01 1.98781669e-01 7.98557580e-01 -8.99708569e-02 -1.52555978e+00 -5.69642603e-01 -2.35677645e-01 -6.04683399e-01 4.36264902e-01 -9.75151360e-01 -1.63657081e+00 4.01506871e-01 1.34239352e+00 2.15140581e-02 1.58739018e+00 2.20474284e-02 6.61897480e-01 -8.26076940e-02 3.46212357e-01 -8.01357627e-01 9.11105126e-02 1.05266705e-01 5.28432250e-01 -1.38061893e+00 5.80340764e-03 -1.53658584e-01 -5.52043319e-01 8.99240851e-01 7.50877321e-01 3.36283445e-01 3.84577751e-01 2.97376156e-01 1.49036422e-01 -4.25610423e-01 -6.30177557e-01 -7.64511883e-01 7.70651877e-01 6.48117363e-01 1.47928178e-01 -3.47685277e-01 1.91287652e-01 4.07575727e-01 3.90800506e-01 -1.15444779e-01 -2.42204174e-01 1.02174425e+00 -3.87954831e-01 -9.12571609e-01 -1.01418090e+00 4.06378418e-01 -2.37651229e-01 1.57249659e-01 -3.87640864e-01 7.00045705e-01 3.71530712e-01 1.11113250e+00 -2.99403548e-01 -7.68180966e-01 2.66808450e-01 3.00220307e-02 8.26506972e-01 -1.91572495e-02 -1.17438853e+00 5.34978449e-01 -1.99595720e-01 -9.74314332e-01 -1.08597565e+00 -7.39248097e-01 -8.42586160e-01 -6.14456609e-02 -6.35836363e-01 -7.37897009e-02 9.66756821e-01 1.14664876e+00 4.76558745e-01 8.45662832e-01 4.88302857e-01 -1.29098046e+00 -2.36285716e-01 -8.74717653e-01 -3.23782176e-01 3.67088169e-01 5.62137425e-01 -1.15958393e+00 -1.82852238e-01 -1.49825275e-01]
[6.339169502258301, -2.199431896209717]
bae35e43-af4c-4d83-bb99-4b61c525b3dd
mm-diffusion-learning-multi-modal-diffusion
2212.09478
null
https://arxiv.org/abs/2212.09478v2
https://arxiv.org/pdf/2212.09478v2.pdf
MM-Diffusion: Learning Multi-Modal Diffusion Models for Joint Audio and Video Generation
We propose the first joint audio-video generation framework that brings engaging watching and listening experiences simultaneously, towards high-quality realistic videos. To generate joint audio-video pairs, we propose a novel Multi-Modal Diffusion model (i.e., MM-Diffusion), with two-coupled denoising autoencoders. In contrast to existing single-modal diffusion models, MM-Diffusion consists of a sequential multi-modal U-Net for a joint denoising process by design. Two subnets for audio and video learn to gradually generate aligned audio-video pairs from Gaussian noises. To ensure semantic consistency across modalities, we propose a novel random-shift based attention block bridging over the two subnets, which enables efficient cross-modal alignment, and thus reinforces the audio-video fidelity for each other. Extensive experiments show superior results in unconditional audio-video generation, and zero-shot conditional tasks (e.g., video-to-audio). In particular, we achieve the best FVD and FAD on Landscape and AIST++ dancing datasets. Turing tests of 10k votes further demonstrate dominant preferences for our model. The code and pre-trained models can be downloaded at https://github.com/researchmm/MM-Diffusion.
['Baining Guo', 'Qin Jin', 'Nicholas Jing Yuan', 'Jianlong Fu', 'Bei Liu', 'Huiguo He', 'Huan Yang', 'Yiyang Ma', 'Ludan Ruan']
2022-12-19
null
http://openaccess.thecvf.com//content/CVPR2023/html/Ruan_MM-Diffusion_Learning_Multi-Modal_Diffusion_Models_for_Joint_Audio_and_Video_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Ruan_MM-Diffusion_Learning_Multi-Modal_Diffusion_Models_for_Joint_Audio_and_Video_CVPR_2023_paper.pdf
cvpr-2023-1
['video-generation']
['computer-vision']
[ 3.10704522e-02 1.23745963e-01 1.23932809e-01 -2.07409680e-01 -1.57346749e+00 -5.88823915e-01 6.32710218e-01 -6.28143966e-01 -3.61627489e-02 5.19324899e-01 6.42618060e-01 9.87000912e-02 1.12405956e-01 -6.75106645e-01 -1.10893297e+00 -7.68075407e-01 2.32380062e-01 5.80643676e-02 3.76028195e-02 -1.87974453e-01 -1.77357495e-01 -2.04868436e-01 -1.54089880e+00 8.32291365e-01 7.28285015e-01 9.10220385e-01 4.00176138e-01 1.19441760e+00 1.89121485e-01 1.22357655e+00 -6.68531299e-01 -7.89634287e-01 5.30614890e-02 -8.75135779e-01 -5.86918652e-01 1.11442469e-01 4.89645749e-01 -8.67964804e-01 -7.78138041e-01 7.92014599e-01 9.68366027e-01 2.70137548e-01 5.84562063e-01 -1.36540663e+00 -1.26148140e+00 1.04573250e+00 -5.21180630e-01 -1.91347629e-01 6.69123650e-01 5.57667136e-01 1.13538587e+00 -9.66033936e-01 7.04558134e-01 1.41842389e+00 6.08813107e-01 8.93773496e-01 -1.30896950e+00 -7.35931575e-01 1.74085237e-02 3.17542374e-01 -1.39953780e+00 -7.11731553e-01 8.31780076e-01 -1.86699226e-01 7.01032162e-01 2.64779866e-01 6.80883765e-01 1.81177783e+00 -9.16327983e-02 1.16148043e+00 5.34586310e-01 -1.34663135e-01 6.15763515e-02 -1.89603895e-01 -7.00570583e-01 2.74511129e-01 -5.17667174e-01 1.77309215e-02 -9.71852183e-01 -2.39723083e-02 8.65348041e-01 -1.30079910e-01 -3.32672060e-01 1.03621729e-01 -1.20597720e+00 6.57387614e-01 1.46526843e-01 3.07797790e-01 -4.66138244e-01 6.29184186e-01 3.47948134e-01 3.12854260e-01 3.71701449e-01 2.08395366e-02 1.06679825e-02 -4.78888571e-01 -1.13174975e+00 3.89278412e-01 4.07261997e-01 1.11990249e+00 3.04636121e-01 4.49440360e-01 -6.37201607e-01 9.81939912e-01 2.65949786e-01 6.50793374e-01 4.84503806e-01 -1.61819661e+00 3.73757392e-01 -2.46671930e-01 -6.53825700e-05 -8.00861299e-01 2.33829677e-01 -2.32713670e-01 -9.97309268e-01 -1.85921401e-01 5.67132868e-02 -4.01467979e-01 -8.04566860e-01 2.16040850e+00 2.34722137e-01 6.23436213e-01 6.85047656e-02 1.00845480e+00 9.84498739e-01 1.11656570e+00 5.19771166e-02 -1.72697231e-01 9.69866276e-01 -1.16181111e+00 -8.44744921e-01 1.65925086e-01 2.71979243e-01 -9.09612179e-01 1.13085079e+00 4.33474958e-01 -1.60621250e+00 -8.40520144e-01 -7.35664964e-01 -2.64115512e-01 3.77459705e-01 3.29170637e-02 2.29228705e-01 2.08005428e-01 -1.35163760e+00 4.09204870e-01 -7.28146672e-01 -2.15096679e-02 2.44769484e-01 -5.07260263e-02 -3.17329228e-01 -1.90969273e-01 -1.35407257e+00 2.62853771e-01 1.13706797e-01 1.62633192e-02 -1.65751934e+00 -6.79606557e-01 -8.79291534e-01 -1.67205445e-02 1.58677995e-01 -9.32779789e-01 1.51459718e+00 -1.29627013e+00 -1.70031881e+00 5.52902341e-01 -1.54004514e-01 -4.23121631e-01 6.84712648e-01 -5.00002384e-01 -3.89499038e-01 4.26795334e-01 8.30715448e-02 1.39390540e+00 1.22202563e+00 -1.44803584e+00 -5.82892597e-01 1.89929366e-01 1.13019183e-01 3.68906289e-01 -6.14692926e-01 -9.44230109e-02 -7.60802627e-01 -1.07903552e+00 -3.21372688e-01 -7.50356674e-01 1.01541296e-01 -1.77060902e-01 -4.93970275e-01 -6.13851324e-02 7.55505621e-01 -9.62070823e-01 1.40419066e+00 -2.37408996e+00 5.42914391e-01 -1.75906479e-01 2.00547665e-01 -1.88195556e-01 -6.57588899e-01 4.42884207e-01 -1.33328319e-01 5.30431718e-02 -1.41779542e-01 -8.06732893e-01 2.05012485e-01 6.17084727e-02 -4.17443991e-01 9.16894972e-02 2.52079844e-01 1.00447166e+00 -8.44110727e-01 -3.80253434e-01 -8.64810050e-02 8.61463785e-01 -9.94760871e-01 4.68233615e-01 -2.36772180e-01 5.00815630e-01 8.52386188e-03 5.73062479e-01 5.86989820e-01 -3.66709009e-02 7.92976618e-02 -1.71386838e-01 2.75650263e-01 -1.76102575e-02 -1.09941697e+00 2.23266840e+00 -5.96689939e-01 7.02949524e-01 2.63523817e-01 -5.46125114e-01 5.33468068e-01 7.42439866e-01 3.23365211e-01 -7.85275936e-01 4.99958545e-02 1.44874483e-01 -3.79853368e-01 -5.84336579e-01 7.88951576e-01 -2.43348390e-01 -6.72841668e-02 3.97747785e-01 5.43679595e-01 -3.04004848e-01 3.33983362e-01 6.73797846e-01 1.01989961e+00 4.87706393e-01 -5.66349268e-01 2.07758218e-01 3.48725282e-02 -4.90538895e-01 4.40717250e-01 6.80268526e-01 5.29256463e-03 1.24238813e+00 4.27029371e-01 1.53689086e-01 -1.26481009e+00 -1.31717587e+00 4.13646817e-01 1.21225190e+00 1.45901263e-01 -6.27548575e-01 -1.09715247e+00 -2.65065074e-01 -2.44103923e-01 8.68069530e-01 -6.02058768e-01 -2.65322834e-01 -4.18176353e-01 -2.69585907e-01 8.85929942e-01 5.51348627e-01 2.25633502e-01 -1.08170950e+00 -4.65864968e-03 2.29874894e-01 -8.97277117e-01 -8.59748602e-01 -9.50595677e-01 -1.89570397e-01 -5.42719364e-01 -5.09244025e-01 -1.30162764e+00 -9.08527076e-01 2.57298619e-01 2.16613501e-01 1.17850530e+00 -1.29981488e-01 1.84031390e-02 6.29634380e-01 -4.09227371e-01 1.75166994e-01 -5.31740904e-01 -1.00391835e-01 1.04171768e-01 1.28674865e-01 -7.08144307e-02 -9.36332762e-01 -7.45664716e-01 3.02138984e-01 -1.33447647e+00 2.62390256e-01 4.48107779e-01 1.03619671e+00 6.68933392e-01 -1.64286971e-01 6.47539616e-01 -1.60555333e-01 8.06010604e-01 -6.84465349e-01 -9.71701927e-03 2.22584382e-02 1.09147862e-01 -2.75792211e-01 4.96349037e-01 -8.24219346e-01 -1.27836704e+00 -2.67368019e-01 -4.88058895e-01 -1.01060259e+00 -1.21440463e-01 3.40991110e-01 -2.54571438e-01 5.62559247e-01 6.85698330e-01 3.78728747e-01 -2.15776846e-01 -3.13464761e-01 7.71902800e-01 7.27019250e-01 8.09167862e-01 -7.02634335e-01 5.67120790e-01 2.95034289e-01 -6.97445035e-01 -7.07036853e-01 -5.53342760e-01 4.58758324e-02 -4.00278941e-02 -6.58345580e-01 1.03765917e+00 -1.42209136e+00 -6.46457672e-01 8.40301335e-01 -1.32263756e+00 -8.44061673e-01 -4.90905881e-01 3.84662151e-01 -8.51543367e-01 3.44315052e-01 -1.07982421e+00 -7.42558002e-01 -1.73185199e-01 -1.16630983e+00 1.36175418e+00 1.71009347e-01 -4.06612098e-01 -6.66749358e-01 8.32410306e-02 5.85976720e-01 3.84226143e-01 -1.58914961e-02 3.41001809e-01 -3.96526642e-02 -7.11469531e-01 1.32335469e-01 7.07704350e-02 6.37461901e-01 -2.02066526e-01 2.90085584e-01 -9.74175870e-01 -2.76466966e-01 -7.78534114e-02 -8.14908504e-01 1.02201879e+00 6.57533526e-01 1.18185258e+00 -3.68515909e-01 1.91453815e-01 6.12519979e-01 9.90600765e-01 2.69815382e-02 9.53652322e-01 -1.34795532e-01 6.71635389e-01 3.47063333e-01 3.87666285e-01 7.63054967e-01 5.63270628e-01 5.60519874e-01 3.47722411e-01 4.16956544e-02 -5.56917429e-01 -6.75661266e-01 8.55798364e-01 1.31157362e+00 -1.64100677e-02 -7.19884753e-01 -4.69909668e-01 8.12432885e-01 -1.76854944e+00 -1.32751918e+00 -2.79280543e-03 1.81109548e+00 9.87257481e-01 -1.10024009e-02 2.30375856e-01 -7.93992635e-03 8.33833694e-01 1.36457980e-01 -4.49395776e-01 4.82909307e-02 -4.52640533e-01 6.20991886e-02 -2.02026576e-01 6.37186408e-01 -8.14068437e-01 8.23909879e-01 5.80998325e+00 1.56341755e+00 -9.43368554e-01 4.54285979e-01 9.56569910e-01 -7.02288628e-01 -9.20108795e-01 -3.03458989e-01 -3.84761751e-01 6.64499462e-01 8.94429386e-01 1.36772245e-01 6.38164997e-01 5.52263975e-01 4.30658519e-01 1.01205468e-01 -9.10452783e-01 1.16584635e+00 2.32441634e-01 -1.36565435e+00 3.34894150e-01 -2.63110697e-01 1.03658581e+00 -1.67426914e-01 4.35931504e-01 3.79583538e-01 5.12483954e-01 -8.37970495e-01 1.47749639e+00 5.65025926e-01 1.08228064e+00 -8.03625584e-01 2.71531999e-01 1.49746656e-01 -1.20281196e+00 -1.59825400e-01 -4.89607565e-02 2.85497487e-01 7.11267054e-01 6.05225086e-01 7.25030452e-02 5.09931386e-01 9.51605499e-01 6.29081964e-01 -2.14550644e-03 6.72909617e-01 -4.36231196e-01 7.40457177e-01 -2.16389775e-01 4.05868828e-01 1.96950912e-01 1.62269417e-02 7.09878564e-01 1.25204039e+00 8.39244366e-01 2.65293010e-02 -2.11731195e-01 9.68328595e-01 -4.43139553e-01 -1.98172301e-01 -4.93182808e-01 -5.06580658e-02 4.31075513e-01 1.18562484e+00 -3.28673810e-01 -5.18234015e-01 -1.22638434e-01 1.36293364e+00 8.78167152e-02 5.88466048e-01 -1.42257071e+00 -3.31461161e-01 6.28537476e-01 1.31617114e-02 5.18630862e-01 -2.00946376e-01 1.11751489e-01 -1.32206726e+00 1.35828495e-01 -1.16762257e+00 2.39900097e-01 -1.32033491e+00 -1.32876706e+00 7.58951068e-01 -1.86021268e-01 -1.20592189e+00 -3.77442092e-01 1.23802505e-01 -5.78671813e-01 5.60521603e-01 -1.08465803e+00 -1.36509991e+00 -2.90689766e-01 8.51967037e-01 8.08918118e-01 7.12436065e-02 4.48318303e-01 6.67897046e-01 -2.94345230e-01 7.69903302e-01 1.19084947e-01 -7.16381967e-02 1.03487563e+00 -9.57404912e-01 3.53632957e-01 8.43388557e-01 2.00700998e-01 3.24389994e-01 6.66829884e-01 -6.35460019e-01 -1.23860884e+00 -1.12273419e+00 6.08589947e-01 -3.36001515e-01 6.12267852e-01 -4.24654096e-01 -7.17156351e-01 5.60189307e-01 8.46593022e-01 -4.65073884e-01 7.04504371e-01 -2.73030758e-01 -3.27550501e-01 -6.72344044e-02 -7.70331025e-01 8.70051384e-01 1.25518465e+00 -7.56798267e-01 -1.58282548e-01 1.95873067e-01 1.16708708e+00 -4.05360639e-01 -9.13431525e-01 1.62290454e-01 6.20685101e-01 -1.18627858e+00 9.02809262e-01 -3.37651640e-01 1.10644341e+00 -1.93243653e-01 -4.63455796e-01 -1.42896461e+00 -3.38689238e-01 -1.15391910e+00 -4.05773878e-01 1.58507657e+00 3.95655513e-01 3.84040847e-02 5.94824553e-01 1.97094083e-01 -3.01240921e-01 -5.61808348e-01 -9.31814790e-01 -6.40701413e-01 -4.02853228e-02 -8.08049977e-01 4.25091445e-01 8.94117117e-01 -5.52857481e-02 4.15537477e-01 -1.06297708e+00 -7.37483725e-02 5.64021230e-01 -1.80805832e-01 7.85224557e-01 -3.57976913e-01 -9.21548784e-01 -5.01341939e-01 1.00067168e-01 -1.53999937e+00 6.36763722e-02 -6.16785645e-01 2.54349560e-01 -1.48755670e+00 3.07752013e-01 1.40691549e-01 -3.71357501e-02 3.11784804e-01 -2.30351403e-01 5.99660575e-01 3.22615862e-01 1.38242736e-01 -8.18788290e-01 1.15424573e+00 1.42680645e+00 -2.86577880e-01 -2.27815658e-01 -3.97391617e-01 -7.22671628e-01 3.84702086e-01 5.08756697e-01 -4.53866422e-01 -5.63583553e-01 -9.18031335e-01 2.11583734e-01 4.66729760e-01 3.70505750e-01 -1.04983282e+00 2.05384701e-01 3.86857390e-02 2.58355200e-01 -3.19911093e-01 7.74391353e-01 -3.72697055e-01 5.37594438e-01 1.35418698e-01 -6.18369818e-01 9.46470723e-03 8.66767392e-02 6.59705460e-01 -5.66071153e-01 1.13711439e-01 6.40081644e-01 -2.63494477e-02 -3.91622722e-01 3.27170581e-01 -5.45361161e-01 8.69250894e-02 8.99203658e-01 -8.24033618e-02 -1.73957109e-01 -1.23106325e+00 -1.06960380e+00 3.01118970e-01 3.33948612e-01 6.25495434e-01 7.78486967e-01 -1.89957070e+00 -1.04204774e+00 -1.74990684e-01 -1.79322869e-01 -4.87384945e-02 1.12892628e+00 8.38485301e-01 -2.82868356e-01 -9.98434275e-02 -6.98473398e-03 -5.71229219e-01 -1.16423070e+00 3.46267045e-01 1.41510993e-01 -4.27719578e-02 -2.55940318e-01 1.24086869e+00 3.82856905e-01 -2.18388736e-01 3.52591515e-01 1.43570462e-02 3.23277444e-01 1.89646602e-01 5.74760318e-01 3.51134539e-01 -3.13162059e-01 -5.76179445e-01 1.08498797e-01 1.55651137e-01 1.26992539e-01 -6.53056860e-01 1.39505744e+00 -3.09869230e-01 6.17836714e-02 3.96825790e-01 1.12335122e+00 1.70825571e-02 -1.68551564e+00 -3.08207273e-02 -8.83659303e-01 -3.88080657e-01 -8.09261948e-02 -6.00677252e-01 -1.31817353e+00 9.79279459e-01 2.79543757e-01 1.03586972e-01 1.41593480e+00 1.54579021e-02 1.19806981e+00 8.41662195e-03 -5.26147615e-03 -1.16356730e+00 7.37420797e-01 3.21468771e-01 1.20646274e+00 -7.33924150e-01 -5.93279243e-01 7.66160190e-02 -9.73424554e-01 7.67800093e-01 7.46757448e-01 6.46207780e-02 3.59132409e-01 3.88347089e-01 8.85460526e-02 2.10074440e-01 -1.18666482e+00 3.83644067e-02 -4.72532585e-03 6.22033060e-01 4.79027152e-01 -1.00822791e-01 9.34804976e-02 8.65590930e-01 -9.78536308e-02 1.27838403e-01 4.85858142e-01 7.14696348e-01 -1.55095696e-01 -9.18711185e-01 -4.48127240e-01 5.98627925e-02 -4.27725494e-01 -2.90333450e-01 -2.35386491e-01 3.03989470e-01 1.09535143e-01 1.05669641e+00 1.07903630e-01 -8.27001512e-01 1.73161954e-01 -6.51285723e-02 3.92245471e-01 -1.27732120e-02 -5.88997126e-01 5.78743637e-01 4.56281640e-02 -5.88505745e-01 -3.78691673e-01 -5.67742467e-01 -7.71320462e-01 -6.36086762e-01 -2.83058256e-01 1.29038766e-01 3.72228622e-01 4.46105212e-01 6.30981266e-01 8.72570395e-01 6.59735382e-01 -1.22984922e+00 -3.82091582e-01 -1.08183646e+00 -5.46908915e-01 6.61662042e-01 1.78398848e-01 -2.44310737e-01 -3.25730324e-01 5.05642951e-01]
[15.365753173828125, 5.240102767944336]
8516432d-8237-4b43-a2ee-8b4050e349d5
scopeit-scoping-task-relevant-sentences-in
2003.04988
null
https://arxiv.org/abs/2003.04988v2
https://arxiv.org/pdf/2003.04988v2.pdf
ScopeIt: Scoping Task Relevant Sentences in Documents
Intelligent assistants like Cortana, Siri, Alexa, and Google Assistant are trained to parse information when the conversation is synchronous and short; however, for email-based conversational agents, the communication is asynchronous, and often contains information irrelevant to the assistant. This makes it harder for the system to accurately detect intents, extract entities relevant to those intents and thereby perform the desired action. We present a neural model for scoping relevant information for the agent from a large query. We show that when used as a preprocessing step, the model improves performance of both intent detection and entity extraction tasks. We demonstrate the model's impact on Scheduler (Cortana is the persona of the agent, while Scheduler is the name of the service. We use them interchangeably in the context of this paper.) - a virtual conversational meeting scheduling assistant that interacts asynchronously with users through email. The model helps the entity extraction and intent detection tasks requisite by Scheduler achieve an average gain of 35% in precision without any drop in recall. Additionally, we demonstrate that the same approach can be used for component level analysis in large documents, such as signature block identification.
['Charles Lee', 'Pamela Bhattacharya', 'Chala Fufa', 'Barun Patra', 'Vishwas Suryanarayanan']
2020-02-23
null
https://aclanthology.org/2020.coling-industry.20
https://aclanthology.org/2020.coling-industry.20.pdf
coling-2020-8
['entity-extraction']
['natural-language-processing']
[ 3.72438878e-01 6.19189918e-01 -9.55555961e-03 -3.93009216e-01 -8.19639146e-01 -6.08623743e-01 6.36628628e-01 4.35921639e-01 -6.26241565e-01 6.49996042e-01 5.06360233e-01 -5.75932026e-01 -1.55750245e-01 -5.45238376e-01 -2.76652813e-01 -3.18551272e-01 -1.50973007e-01 1.00114620e+00 1.41146734e-01 -2.18070984e-01 2.20558047e-01 4.41548198e-01 -1.42279792e+00 2.49317318e-01 5.30609369e-01 5.14352739e-01 2.94282049e-01 8.95897746e-01 -2.84253448e-01 9.67252672e-01 -1.08906400e+00 9.43301171e-02 -3.03798243e-02 -1.63506180e-01 -1.16451478e+00 -3.58334482e-02 1.11377455e-01 -5.67803860e-01 -5.09591520e-01 5.58513165e-01 4.35888559e-01 2.11984560e-01 6.81214035e-01 -1.68897593e+00 1.51250497e-01 6.98090136e-01 -3.56227279e-01 1.85339481e-01 7.32081234e-01 -2.17214823e-01 1.07868147e+00 -4.64929104e-01 6.62812948e-01 1.17588651e+00 4.82260019e-01 4.45052505e-01 -8.74693930e-01 -4.19235766e-01 3.99067670e-01 -1.04477093e-01 -1.28363264e+00 -9.03918386e-01 3.36833596e-01 -1.01484977e-01 1.29167449e+00 5.44174969e-01 -3.09010912e-02 7.47856438e-01 5.26699573e-02 1.04475093e+00 4.13852364e-01 -5.63308060e-01 2.19721764e-01 3.32318693e-01 8.71525288e-01 7.95297623e-01 7.31142461e-02 -3.90337199e-01 -7.73332357e-01 -7.46749699e-01 8.04982930e-02 1.39945326e-02 -3.27923506e-01 4.85349558e-02 -9.72421825e-01 5.95594108e-01 -1.61451846e-01 2.86256015e-01 -6.50518239e-01 -2.52069421e-02 3.85680377e-01 5.67401111e-01 2.91312963e-01 6.67833328e-01 -6.32134318e-01 -5.58176041e-01 -4.53138202e-01 5.28954446e-01 1.85260296e+00 1.45494461e+00 5.28706908e-01 -4.45461929e-01 -2.53034830e-01 7.29195476e-01 4.81946729e-02 3.68004769e-01 3.54698569e-01 -1.06391680e+00 3.41319025e-01 7.95262933e-01 6.33105218e-01 -7.68587708e-01 -8.75443816e-01 -6.89022690e-02 -1.15418702e-01 -1.73467427e-01 4.96652514e-01 -6.48554027e-01 -6.82269871e-01 1.58728004e+00 2.53294677e-01 -1.95729926e-01 1.23874225e-01 6.04673624e-01 8.41240883e-01 4.34346765e-01 2.33913958e-02 -4.57725167e-01 1.91374350e+00 -8.64100218e-01 -9.73656595e-01 -5.27772963e-01 7.70311296e-01 -7.77808845e-01 5.71196616e-01 1.10511795e-01 -1.06127548e+00 3.31875607e-02 -8.41359735e-01 -2.99693365e-02 -4.30744857e-01 -7.54173771e-02 7.84978390e-01 2.75608957e-01 -1.14066327e+00 1.98401675e-01 -8.71652901e-01 -8.44795167e-01 -2.12914690e-01 6.75611734e-01 -1.53992519e-01 2.59891730e-02 -1.06350493e+00 9.45277154e-01 -5.32663474e-03 -3.29048693e-01 -3.32768917e-01 -3.60043854e-01 -7.81240106e-01 3.32438618e-01 5.74506402e-01 -7.68245459e-01 2.08049059e+00 -5.80979586e-01 -1.20333898e+00 6.47343218e-01 -8.61410439e-01 -3.86571497e-01 2.33823434e-01 -1.99720591e-01 -3.11045527e-01 1.32249687e-02 3.68179530e-01 3.87802333e-01 5.31428695e-01 -9.17887032e-01 -1.23929727e+00 -5.90390801e-01 4.54727948e-01 5.55464506e-01 -2.99842179e-01 3.21214348e-01 -8.13956916e-01 4.97458689e-02 5.27669750e-02 -1.08534575e+00 8.16061050e-02 -6.25233591e-01 -3.42435449e-01 -6.84120417e-01 9.67447758e-01 -6.31295025e-01 1.31310880e+00 -2.06360054e+00 -3.13227475e-01 3.79205614e-01 4.25171435e-01 -8.75889212e-02 -9.52998735e-03 7.19936728e-01 2.06852779e-01 -6.18126728e-02 1.38034627e-01 -2.55261511e-01 1.89065173e-01 5.11614233e-03 -2.80778995e-03 1.78055629e-01 -1.72655493e-01 7.13709354e-01 -6.99607134e-01 -3.90695781e-01 -3.71342659e-01 1.19161427e-01 -2.92547971e-01 4.48579043e-01 -2.10362479e-01 -1.80197805e-01 -6.43391192e-01 4.29123729e-01 1.25741422e-01 -4.63551581e-01 5.31679630e-01 1.38852999e-01 -1.15865581e-02 8.80902171e-01 -1.07166827e+00 1.33868361e+00 -6.44700408e-01 8.05284858e-01 9.13357019e-01 -4.29497600e-01 5.75439513e-01 6.22689426e-01 4.32332098e-01 -3.00096035e-01 -2.48461384e-02 2.15789862e-03 -9.34454147e-03 -4.40208972e-01 7.57333636e-01 3.67779016e-01 -1.74270138e-01 1.19384265e+00 -4.36031103e-01 2.63553470e-01 3.86102915e-01 7.65519381e-01 1.74650347e+00 -4.95983481e-01 4.06824231e-01 -3.71304378e-02 3.61517817e-01 2.36242250e-01 2.90371031e-01 1.10914791e+00 -1.91644818e-01 -1.31305493e-03 5.10038137e-01 -4.37431484e-01 -5.97209990e-01 -4.48969841e-01 2.70210564e-01 1.83034420e+00 1.77188188e-01 -7.17871130e-01 -7.16661453e-01 -7.32986391e-01 2.78325438e-01 8.97609055e-01 -6.20318279e-02 -7.68671781e-02 -6.60348892e-01 -4.97119188e-01 3.90969992e-01 5.06069243e-01 2.24239811e-01 -1.07902360e+00 -5.33693910e-01 4.58469033e-01 -5.85987866e-01 -1.29378581e+00 -7.21808255e-01 4.39603746e-01 -4.60599154e-01 -1.08762443e+00 -5.84198795e-02 -8.92489612e-01 6.94055140e-01 5.55640280e-01 1.24601340e+00 3.96526724e-01 -1.50257558e-01 8.21903527e-01 -1.10304862e-01 -7.40294337e-01 -5.86081088e-01 4.21734810e-01 1.17690668e-01 -3.95251393e-01 1.10289824e+00 -4.33101803e-01 -1.23609103e-01 4.41581011e-01 -5.53379357e-01 -1.28310606e-01 4.38615501e-01 5.94297647e-01 -4.46447045e-01 6.04729056e-02 6.92966044e-01 -1.20630288e+00 1.21280825e+00 -4.00413066e-01 -3.69785011e-01 7.97475800e-02 -6.99197352e-01 9.99070108e-02 3.21518838e-01 -8.30896869e-02 -9.79252875e-01 3.04089319e-02 1.99911475e-01 3.95052016e-01 -4.99757707e-01 4.58805501e-01 -1.38317630e-01 3.37424397e-01 6.15241528e-01 4.63885032e-02 8.43698084e-02 -3.57871205e-01 -2.07342356e-01 1.43918240e+00 3.35744619e-01 -3.88063878e-01 5.07219672e-01 2.59947032e-01 -5.83055794e-01 -9.40435708e-01 -4.17000830e-01 -9.69018817e-01 -1.44104108e-01 1.59138218e-01 4.13539141e-01 -9.12078440e-01 -1.35285759e+00 2.80644655e-01 -1.44849861e+00 -2.40451232e-01 2.39575446e-01 2.83964694e-01 -4.47680116e-01 4.79397736e-02 -8.38959694e-01 -1.04900134e+00 -5.37700474e-01 -9.72854793e-01 1.10023201e+00 2.96724617e-01 -9.38603997e-01 -6.76191926e-01 -4.19364095e-01 5.15955806e-01 4.25315648e-01 -5.91891468e-01 9.77313459e-01 -1.46995699e+00 -6.01445794e-01 -5.91067910e-01 -3.17577273e-01 -4.41337198e-01 3.19606721e-01 -3.93776536e-01 -1.08841252e+00 -3.13087136e-01 -2.38517091e-01 -1.44052217e-02 3.99097025e-01 2.49026030e-01 5.56857526e-01 -4.92985874e-01 -9.54951227e-01 -5.86124063e-02 5.15491962e-01 5.68763077e-01 6.74556419e-02 3.73653531e-01 2.46067911e-01 7.30571508e-01 5.31389892e-01 4.81439829e-01 6.39827311e-01 7.46593952e-01 -4.90838476e-02 2.55440995e-02 3.21104199e-01 1.74920335e-02 3.71558368e-01 4.50282693e-01 2.08957151e-01 -4.01359618e-01 -9.25060868e-01 4.96490151e-01 -2.16971183e+00 -7.14677095e-01 1.48666590e-01 1.95454419e+00 7.71204710e-01 2.27419436e-01 2.03047201e-01 -1.94534793e-01 6.97184384e-01 6.59323763e-03 -4.97832596e-01 -4.13640410e-01 5.53897977e-01 -2.10239395e-01 5.88760257e-01 9.31784868e-01 -1.03791714e+00 7.58566737e-01 6.34628439e+00 1.08004011e-01 -5.73839247e-01 -1.21422105e-01 3.62310559e-01 -1.97388530e-01 -4.35341895e-02 -1.03613131e-01 -1.11932909e+00 1.62646547e-01 1.20903885e+00 -4.31645572e-01 5.87958694e-01 9.39880967e-01 3.19045484e-01 -3.14134419e-01 -1.66332924e+00 7.06541657e-01 6.67377263e-02 -9.67414916e-01 -4.86623615e-01 1.43392220e-01 1.29420161e-01 -1.00733731e-02 -5.73226333e-01 6.47669733e-01 7.03380942e-01 -7.61633217e-01 6.06185235e-02 2.33919680e-01 2.29055196e-01 -7.94387639e-01 8.68166029e-01 6.56158209e-01 -8.77116501e-01 -1.72739401e-01 6.76891133e-02 -2.23725051e-01 1.46538645e-01 2.62136728e-01 -1.68144107e+00 1.41702279e-01 3.15717936e-01 -5.32566868e-02 -2.13351265e-01 6.75660074e-01 4.43317108e-02 4.00812775e-01 -5.28653204e-01 -4.51987565e-01 5.65678589e-02 1.12016700e-01 8.29952121e-01 1.21803117e+00 -2.55804718e-01 2.69026995e-01 4.66142356e-01 3.72731924e-01 -2.83538193e-01 -1.45824691e-02 -8.00230682e-01 -2.58586496e-01 8.63526106e-01 1.27963424e+00 -6.59983933e-01 -5.62801659e-01 -4.15228456e-01 9.93703902e-01 2.51756251e-01 7.01089442e-01 -2.99431592e-01 -8.39571059e-01 8.13625038e-01 2.22700268e-01 -4.26775450e-03 -1.19493291e-01 9.20122936e-02 -6.17358208e-01 1.28109813e-01 -1.25586987e+00 4.09686387e-01 -5.98350346e-01 -8.28700066e-01 4.81031299e-01 -1.19463056e-02 -4.73861635e-01 -1.02087021e+00 -4.83148128e-01 -6.20570898e-01 9.08504903e-01 -1.09077728e+00 -6.91804469e-01 -3.57991606e-01 3.23165655e-01 6.08383775e-01 -1.96220443e-01 1.15283859e+00 1.76727921e-01 -3.43877345e-01 4.11647260e-01 -2.00278610e-01 3.69873166e-01 9.08044040e-01 -1.26946390e+00 5.34318507e-01 4.21751291e-01 -1.13035254e-01 1.14786768e+00 9.60332453e-01 -5.55799305e-01 -1.75811923e+00 -6.53262496e-01 1.41474724e+00 -5.91332197e-01 4.80968177e-01 -3.51055503e-01 -6.79274082e-01 1.01066911e+00 4.50202107e-01 -8.52396011e-01 5.89106560e-01 8.20098877e-01 -2.32941229e-02 -4.09061350e-02 -9.94580388e-01 7.89510250e-01 9.83577609e-01 -6.43916607e-01 -5.47762752e-01 7.82885671e-01 9.60267186e-01 -4.10233319e-01 -3.63436908e-01 -5.69531396e-02 5.68513036e-01 -5.63598216e-01 5.96594274e-01 -8.41344833e-01 -2.55608618e-01 -3.57329510e-02 1.52421579e-01 -1.09473860e+00 -1.37289077e-01 -1.09725583e+00 -2.99992979e-01 1.05662739e+00 6.53516531e-01 -7.36702561e-01 7.13558376e-01 1.48303223e+00 3.43232066e-03 -2.04765379e-01 -5.12211621e-01 -3.05897206e-01 -8.13245177e-01 -2.90903240e-01 7.41530836e-01 6.19929910e-01 6.46776795e-01 9.29690778e-01 1.63298845e-01 2.92741537e-01 1.52699158e-01 1.12812892e-01 1.03234017e+00 -1.44548070e+00 -1.86135605e-01 -2.89055586e-01 3.03255897e-02 -1.38631654e+00 2.93711931e-01 -5.60168922e-01 3.49918932e-01 -1.62458241e+00 -2.60303821e-02 -3.35374027e-01 1.40489475e-03 6.75987720e-01 -1.29026398e-01 -6.03554368e-01 -3.02316621e-02 1.81597948e-01 -7.66317844e-01 -1.38045788e-01 6.75138295e-01 -2.97105521e-01 -6.89024031e-01 6.06314182e-01 -1.08847523e+00 8.02706420e-01 5.23950160e-01 -3.87251437e-01 -3.70571613e-01 -1.73157617e-01 4.46351804e-02 3.55201006e-01 -8.76062214e-02 -6.15990996e-01 8.71371686e-01 4.74218316e-02 -5.36956899e-02 -3.61517489e-01 3.57491791e-01 -9.82732356e-01 -3.54518592e-01 2.43953988e-01 -6.99746668e-01 2.50465989e-01 2.07760483e-01 5.22263885e-01 -4.40216660e-02 -4.06429023e-01 7.64579102e-02 -1.54229760e-01 -4.25164878e-01 -5.32387756e-02 -1.02752841e+00 -1.38986960e-01 6.46580517e-01 1.48593888e-01 -4.00070459e-01 -1.16751015e+00 -4.96751308e-01 5.09373784e-01 3.70699108e-01 4.90747511e-01 2.02629790e-01 -6.67216837e-01 -5.47445774e-01 2.21264705e-01 2.05403835e-01 -5.22538200e-02 -2.99001396e-01 8.24498475e-01 -1.03996396e-01 6.91548347e-01 3.78958195e-01 -3.54358822e-01 -1.69964731e+00 1.47406697e-01 1.46987155e-01 -3.13261122e-01 -4.52636331e-01 6.86121941e-01 2.73860693e-01 -5.50694644e-01 9.84857798e-01 -2.99405515e-01 -1.92087054e-01 2.06673935e-01 9.66188669e-01 2.48480827e-01 4.54628527e-01 -1.31336451e-01 -5.99163711e-01 -5.65700173e-01 -7.27737188e-01 -3.38041604e-01 1.22261941e+00 -2.98645914e-01 -2.39282563e-01 3.03239435e-01 8.71488154e-01 1.23959899e-01 -6.01203442e-01 -3.85273457e-01 2.70713687e-01 7.81275891e-03 -2.90196743e-02 -7.27403939e-01 -4.91856337e-01 2.18661413e-01 1.54523656e-01 6.45272613e-01 6.81017220e-01 3.97117771e-02 8.04417253e-01 1.14012206e+00 3.90295953e-01 -1.21362114e+00 -3.03817749e-01 8.07184994e-01 6.44213021e-01 -1.18255353e+00 -7.34928995e-02 -2.12816820e-01 -5.95850170e-01 8.92883420e-01 6.86602294e-01 5.71543396e-01 5.12998581e-01 7.23867357e-01 2.03274935e-01 -4.73737180e-01 -1.20305312e+00 2.30141580e-02 -1.20639004e-01 5.40445209e-01 6.85648322e-01 6.76695257e-02 -1.99327156e-01 4.93910581e-01 -9.07952264e-02 -1.75552666e-01 7.07880497e-01 1.50302541e+00 -6.38110936e-01 -8.76643240e-01 -3.50416929e-01 7.63953447e-01 -4.04740274e-01 -2.70343214e-01 -7.99327731e-01 7.23265409e-01 -7.00105011e-01 1.47724509e+00 2.47371182e-01 -1.98963717e-01 4.97075826e-01 6.75936401e-01 -1.61713585e-01 -7.73382723e-01 -8.18711281e-01 3.26674916e-02 8.76577079e-01 -4.56551313e-01 -1.75242499e-01 -6.03155375e-01 -1.53701234e+00 -4.84180748e-01 -5.51955283e-01 6.46431863e-01 7.31758773e-01 1.07044506e+00 7.22384453e-01 4.81207043e-01 5.45859694e-01 -3.48941177e-01 -5.19117653e-01 -1.13492692e+00 -5.18689454e-01 2.05563739e-01 4.52446252e-01 -4.49804842e-01 -3.39499176e-01 -2.03096941e-02]
[12.497527122497559, 7.8071208000183105]
824c7faf-1d6a-440d-9c74-c1026e090f65
novelty-detection-in-network-traffic-using
2301.06229
null
https://arxiv.org/abs/2301.06229v1
https://arxiv.org/pdf/2301.06229v1.pdf
Novelty Detection in Network Traffic: Using Survival Analysis for Feature Identification
Intrusion Detection Systems are an important component of many organizations' cyber defense and resiliency strategies. However, one downside of these systems is their reliance on known attack signatures for detection of malicious network events. When it comes to unknown attack types and zero-day exploits, modern Intrusion Detection Systems often fall short. In this paper, we introduce an unconventional approach to identifying network traffic features that influence novelty detection based on survival analysis techniques. Specifically, we combine several Cox proportional hazards models and implement Kaplan-Meier estimates to predict the probability that a classifier identifies novelty after the injection of an unknown network attack at any given time. The proposed model is successful at pinpointing PSH Flag Count, ACK Flag Count, URG Flag Count, and Down/Up Ratio as the main features to impact novelty detection via Random Forest, Bayesian Ridge, and Linear Support Vector Regression classifiers.
['Nathaniel Bastian', 'Elie Alhajjar', 'Taylor Bradley']
2023-01-16
null
null
null
null
['survival-analysis']
['miscellaneous']
[ 1.31597653e-01 -4.14593697e-01 -6.77191734e-01 -5.74104451e-02 -1.57957405e-01 -6.74112380e-01 7.38073587e-01 7.80401647e-01 -4.23327446e-01 7.81038642e-01 -3.92644018e-01 -1.24396515e+00 -3.81761074e-01 -8.63916159e-01 7.63159096e-02 -4.94410306e-01 -4.93319988e-01 3.51332188e-01 4.36802685e-01 6.72672363e-03 9.27800238e-01 1.06944668e+00 -6.20942354e-01 -1.24650955e-01 3.49370688e-02 1.06899166e+00 -8.42989743e-01 1.01906788e+00 -1.11903124e-01 7.18425512e-01 -7.92224348e-01 -8.42682421e-02 2.19684348e-01 -2.30161622e-01 -3.54051322e-01 -6.66517973e-01 -4.36807036e-01 -4.16483611e-01 -4.80958641e-01 6.92285717e-01 1.65626049e-01 -1.83845371e-01 6.72644019e-01 -1.64092267e+00 1.37503639e-01 4.87024665e-01 -6.53175771e-01 9.18715239e-01 2.60539889e-01 2.09029064e-01 4.52199459e-01 -3.81576329e-01 4.71366286e-01 1.01708210e+00 7.35867441e-01 1.29272610e-01 -1.28051186e+00 -8.42160583e-01 -1.94812432e-01 2.32063066e-02 -1.20672965e+00 -3.00899863e-01 5.09513795e-01 -5.11250854e-01 6.64311349e-01 2.49186516e-01 2.42877439e-01 1.10861313e+00 1.26785815e+00 -1.14561319e-01 1.26387715e+00 -3.16385716e-01 4.52989548e-01 2.32048184e-01 5.79592347e-01 6.12700701e-01 7.33473778e-01 7.71635234e-01 -1.60182044e-01 -9.68445063e-01 6.21546686e-01 5.27198195e-01 1.05847843e-01 1.46199510e-01 -7.42887437e-01 1.01940930e+00 1.64984062e-01 4.10683036e-01 -4.17561352e-01 -1.15596168e-01 6.84303045e-01 3.40158671e-01 9.59799737e-02 3.72507364e-01 -5.03200710e-01 -2.79286772e-01 -6.24091864e-01 -1.34766564e-01 1.09462726e+00 1.81105226e-01 4.23309296e-01 1.98191419e-01 -9.54886153e-02 -9.16351378e-02 2.13385209e-01 2.36069962e-01 2.89426148e-01 -3.39084327e-01 -3.92028332e-01 5.73115170e-01 -2.45671600e-01 -9.48360384e-01 -5.32004416e-01 -7.71385789e-01 -5.47918558e-01 3.14967752e-01 3.00455362e-01 -1.56118259e-01 -7.62107074e-01 1.10244811e+00 2.46088117e-01 3.42071265e-01 -1.78122029e-01 -3.01627144e-02 -7.39056151e-03 3.94279391e-01 3.38551402e-01 -5.44355929e-01 1.00146723e+00 -1.58849820e-01 -5.87418854e-01 7.64614120e-02 5.49806237e-01 -5.21217287e-01 3.13336253e-01 4.19791609e-01 -2.90098697e-01 -2.69310791e-02 -8.23010981e-01 1.00364220e+00 -4.04842734e-01 -7.31628597e-01 7.66456246e-01 1.27280581e+00 -1.78991571e-01 6.85674548e-01 -8.50572050e-01 -3.60514969e-01 2.97265023e-01 1.31254703e-01 -1.48376808e-01 -1.78190157e-01 -9.66468096e-01 7.89457858e-01 4.53056902e-01 -3.11158210e-01 -1.07891881e+00 -3.58970344e-01 -3.62592101e-01 5.21494150e-02 5.97905874e-01 -3.76363873e-01 8.56526613e-01 -1.94939137e-01 -1.08987844e+00 2.37839952e-01 1.33804426e-01 -6.13738298e-01 1.56862468e-01 3.31983894e-01 -7.69912541e-01 -4.52652434e-03 -1.24334641e-01 -5.75029254e-01 1.02542341e+00 -8.95730257e-01 -6.73797190e-01 -7.74843752e-01 -6.90316781e-02 -5.41078091e-01 9.47473291e-03 2.54287362e-01 5.52069068e-01 -4.11501408e-01 6.47804141e-02 -8.27854097e-01 -5.95539629e-01 -6.04960442e-01 -5.66935897e-01 2.08325580e-01 1.12062764e+00 -3.75502497e-01 1.39346266e+00 -2.11582971e+00 -6.05363727e-01 7.21569777e-01 1.42668664e-01 1.24897353e-01 3.73043925e-01 5.45313120e-01 -3.33243385e-02 4.94379967e-01 -6.18867762e-02 4.31730151e-01 -2.98979878e-01 2.08990276e-03 -5.14591277e-01 5.82955718e-01 4.22911951e-03 3.09910506e-01 -6.79438055e-01 -8.40906650e-02 2.95997024e-01 1.50778323e-01 -2.07017988e-01 1.58643663e-01 4.18980241e-01 5.14032602e-01 -4.48512286e-01 1.11172223e+00 2.13340193e-01 1.88678071e-01 2.28079796e-01 2.60537416e-01 -2.22006857e-01 1.28284678e-01 -8.51725221e-01 4.70611304e-01 -1.11415975e-01 2.00777426e-01 -2.35221863e-01 -9.75666940e-01 1.04600000e+00 1.77167043e-01 1.37915075e-01 -2.65897036e-01 5.53501368e-01 3.49148303e-01 2.71546870e-01 1.16993837e-01 -6.98632747e-02 -5.93130827e-01 -3.12280595e-01 3.88993263e-01 -3.85438316e-02 4.17803675e-01 -7.66765401e-02 1.36930093e-01 1.79267037e+00 -6.84816420e-01 9.66675699e-01 3.70486528e-02 4.25412923e-01 7.00654462e-02 7.90379345e-01 1.19066072e+00 -6.27282083e-01 -2.30633959e-01 1.00986588e+00 -5.64191818e-01 -5.04094064e-01 -1.28321159e+00 -4.41542715e-01 5.76010704e-01 -3.58054280e-01 -5.96831590e-02 7.35302642e-02 -1.14709330e+00 2.35058844e-01 9.71425772e-01 -4.87379342e-01 -5.98217905e-01 -1.35766566e-01 -8.82122934e-01 7.48121977e-01 4.22055684e-02 2.17240900e-01 -6.33781195e-01 -3.71762872e-01 3.99354070e-01 5.32069206e-01 -8.30694437e-01 1.92678452e-01 5.41958272e-01 -1.15491796e+00 -1.51225770e+00 4.67316136e-02 1.74546108e-01 4.62870687e-01 7.52632916e-02 3.29083443e-01 3.90087627e-02 -9.08592224e-01 3.48033071e-01 -3.21468174e-01 -6.17667973e-01 -6.08808756e-01 -7.16851205e-02 3.44860911e-01 -4.93254438e-02 4.75703597e-01 -5.45463264e-01 -3.88244510e-01 2.97995597e-01 -6.45844460e-01 -1.19425285e+00 7.78348386e-01 5.87417066e-01 1.07920028e-01 5.00204504e-01 8.68111610e-01 -1.22592759e+00 6.70770764e-01 -1.02519476e+00 -3.77588093e-01 1.28486205e-03 -1.17587280e+00 -4.68668252e-01 4.07951951e-01 -5.51009893e-01 -6.52010024e-01 -1.94661081e-01 2.90324569e-01 -2.63642490e-01 -4.23744291e-01 6.63904607e-01 2.17654854e-01 -2.79203117e-01 8.78843665e-01 1.79560304e-01 1.60950944e-01 3.40570472e-02 -3.25693399e-01 5.29403746e-01 5.72461933e-02 -1.71408355e-01 1.07098770e+00 5.67992747e-01 4.46820915e-01 -8.94074440e-01 -4.37411517e-01 -6.64931476e-01 -3.19426090e-01 -3.03786844e-01 3.00246894e-01 -3.94546419e-01 -9.70569253e-01 3.79141718e-01 -6.16115391e-01 3.42084795e-01 7.09765404e-02 6.53202116e-01 1.25957876e-02 3.61334860e-01 -4.74990249e-01 -1.23069811e+00 -2.27873757e-01 -7.34402299e-01 1.37534708e-01 2.95452565e-01 -4.35047925e-01 -1.04275930e+00 3.51185232e-01 7.22303316e-02 7.36238122e-01 7.25794971e-01 1.02075136e+00 -1.50689173e+00 -3.15288275e-01 -1.26041281e+00 -8.20730329e-02 2.64603496e-01 2.70198047e-01 3.82346243e-01 -5.65644026e-01 -4.91917878e-01 3.96516293e-01 3.74771029e-01 3.69286746e-01 5.15189409e-01 7.46904314e-01 -4.49138761e-01 -7.12387502e-01 3.55423629e-01 1.51538348e+00 9.41179812e-01 4.60153192e-01 6.51865721e-01 1.13222666e-01 5.07265329e-01 5.71421683e-01 8.87278497e-01 -2.58683950e-01 2.37628832e-01 3.50762844e-01 3.48621637e-01 7.33549297e-01 -1.31170258e-01 3.25466126e-01 4.53742892e-02 2.01082334e-01 3.72138917e-02 -1.21671188e+00 1.45104632e-01 -1.20661974e+00 -9.28612947e-01 -1.15547568e-01 2.55161810e+00 1.50311694e-01 9.09207761e-01 3.64019275e-01 3.86115253e-01 7.18458593e-01 -7.57055581e-02 -4.19076085e-01 -7.09839106e-01 1.99361935e-01 1.93996727e-01 8.25556815e-01 2.12795869e-01 -8.13993514e-01 4.13572192e-01 6.57274342e+00 4.36927587e-01 -9.53755617e-01 -1.56052873e-01 8.07875216e-01 2.39040375e-01 2.67965585e-01 6.79919660e-01 -9.79589283e-01 3.73913169e-01 1.62427211e+00 -4.11186218e-01 3.72664491e-03 9.38116252e-01 1.75107569e-01 -4.78316039e-01 -6.87521219e-01 1.92666337e-01 2.54691131e-02 -1.25349474e+00 -3.07394918e-02 3.40636939e-01 1.46681875e-01 -5.05485713e-01 1.47542253e-01 5.68526089e-01 2.16843739e-01 -7.54005969e-01 -1.60630450e-01 4.98640925e-01 4.03259158e-01 -1.10496867e+00 9.69189882e-01 2.78428555e-01 -6.24366403e-01 -5.97120643e-01 1.18118696e-01 -2.87732571e-01 2.73940116e-01 9.42205429e-01 -1.45227051e+00 3.14157784e-01 3.06334972e-01 1.13550508e-02 -3.11164200e-01 1.05168104e+00 -6.30939454e-02 8.65792334e-01 -1.16923451e-01 2.81117130e-02 2.40075104e-02 3.76639843e-01 9.01923418e-01 8.76841962e-01 1.15371577e-01 1.33033171e-01 2.17372850e-01 3.58752012e-01 5.40923536e-01 -1.14171714e-01 -9.18823779e-01 -4.25503135e-01 8.03561568e-01 1.20643163e+00 -1.25170791e+00 9.35467184e-02 -2.72114962e-01 3.28946233e-01 -3.92473817e-01 2.68680975e-02 -5.44155002e-01 -8.39188814e-01 4.83128726e-01 4.32862550e-01 -2.81628430e-01 -4.81713295e-01 -6.23901963e-01 -6.40010834e-01 -7.78869033e-01 -6.25482142e-01 7.44889855e-01 9.94533375e-02 -1.26558530e+00 4.03521240e-01 2.19455846e-02 -8.99481058e-01 -2.79685944e-01 -6.12856627e-01 -1.09954214e+00 6.95376933e-01 -1.21404445e+00 -7.39887655e-01 6.19711280e-02 4.04618859e-01 1.17765576e-01 -5.18378019e-01 8.51557553e-01 -1.15468621e-01 -9.02437627e-01 3.35504174e-01 -5.75548410e-02 3.10206227e-02 5.05548656e-01 -8.52348864e-01 2.39355221e-01 8.94305766e-01 -4.31283593e-01 8.88126194e-01 8.25686991e-01 -1.18542159e+00 -1.01741612e+00 -7.83566833e-01 5.78263819e-01 -4.36828911e-01 1.01846159e+00 -7.76242116e-04 -6.70122445e-01 7.98314691e-01 -5.94138324e-01 -1.83024585e-01 1.01019561e+00 3.53992164e-01 -2.39577994e-01 1.49709567e-01 -1.41310453e+00 5.14352560e-01 1.12468623e-01 -2.83529669e-01 -5.36458731e-01 8.03854167e-02 5.07141531e-01 2.27205187e-01 -8.76979113e-01 5.38237751e-01 5.12951910e-01 -9.82835233e-01 8.32247317e-01 -8.55678558e-01 -1.68786705e-01 -3.82648334e-02 -1.55020103e-01 -6.12907469e-01 -4.01581317e-01 -7.60418653e-01 -5.79479448e-02 1.02983141e+00 3.75291556e-01 -1.07690692e+00 6.96888685e-01 3.53716820e-01 3.54672551e-01 -6.29562795e-01 -1.13996696e+00 -9.26691055e-01 -4.88619149e-01 -2.12900594e-01 1.80986956e-01 1.18094587e+00 1.22449853e-01 2.55135328e-01 -2.15960026e-01 3.42451781e-01 1.01335704e+00 -3.15549165e-01 7.52726376e-01 -1.56733429e+00 -2.27602005e-01 -3.58724475e-01 -8.29647064e-01 3.74931842e-01 3.23627032e-02 -4.63717669e-01 -5.53305566e-01 -7.58289099e-01 2.55272716e-01 -4.09072429e-01 -6.60592258e-01 4.43132579e-01 -1.62230711e-02 -1.48716256e-01 -2.36699536e-01 2.29195938e-01 2.11068153e-01 5.23994602e-02 2.11058393e-01 2.36558422e-01 -3.97678822e-01 7.14925349e-01 -4.06834155e-01 7.09587574e-01 1.14507532e+00 -9.30057943e-01 -1.67805448e-01 9.02673125e-01 -8.25846195e-02 6.16144240e-01 5.33593893e-01 -8.86137068e-01 2.20186383e-01 -5.09759009e-01 3.82336438e-01 -4.01035786e-01 -2.03059584e-01 -6.85707152e-01 1.90776050e-01 1.15583873e+00 6.36780187e-02 3.66630197e-01 3.34165484e-01 1.00074661e+00 3.65896560e-02 -3.70991826e-01 9.34393823e-01 2.35657673e-02 -3.33738923e-01 2.52854228e-01 -7.58930385e-01 -3.20427418e-01 1.39059055e+00 -4.34832841e-01 -4.52294946e-01 -2.18411878e-01 -8.57450068e-01 -2.05175698e-01 2.62659669e-01 3.05277050e-01 7.25702107e-01 -7.65359938e-01 -4.16300088e-01 2.56073475e-01 3.52993384e-02 -9.85033989e-01 3.58849168e-01 9.28716660e-01 -4.25871640e-01 6.28657579e-01 -4.56183165e-01 -9.38061401e-02 -1.17245817e+00 6.43119395e-01 1.27815112e-01 -4.74817276e-01 -2.41312206e-01 5.38722277e-01 -4.33516145e-01 -2.41265781e-02 1.01895288e-01 4.57200646e-01 -9.74982306e-02 -1.32044330e-01 5.50110281e-01 9.64479268e-01 -1.16226651e-01 -8.66653174e-02 -6.18777514e-01 -1.88093886e-01 -5.80663145e-01 -2.49507297e-02 8.74982893e-01 1.58332452e-01 -3.40555727e-01 8.06313455e-01 8.08849156e-01 -1.00688480e-01 -5.05614698e-01 -1.30214870e-01 4.74567235e-01 -7.90524244e-01 1.98829427e-01 -8.07520807e-01 -4.71150309e-01 4.09025580e-01 7.49811113e-01 3.51464540e-01 7.53898442e-01 -3.11954111e-01 4.06042486e-01 1.61504447e-01 3.71809483e-01 -5.01543939e-01 -4.87785898e-02 4.06360239e-01 1.48875684e-01 -1.01356387e+00 1.77025229e-01 -2.16820940e-01 -2.63425142e-01 1.22411966e+00 3.47806484e-01 -1.43989503e-01 1.01361918e+00 2.93964028e-01 -3.01048428e-01 -1.84316456e-01 -7.87002385e-01 3.71774316e-01 -2.73355365e-01 6.95526302e-01 1.68293342e-01 1.87170401e-01 -5.61528802e-01 4.21929091e-01 2.88048953e-01 -2.27680296e-01 9.12855089e-01 1.12732017e+00 -7.80019403e-01 -9.41026688e-01 -4.75101292e-01 8.93379748e-01 -8.91368508e-01 -1.10543631e-02 -3.65396231e-01 8.65045547e-01 -4.59474742e-01 8.75299692e-01 -9.18548629e-02 -5.06459951e-01 3.58646482e-01 3.37513149e-01 -1.41714811e-01 -6.70816243e-01 -4.88611698e-01 -3.48462254e-01 -1.22152083e-02 -4.00559247e-01 3.36073756e-01 -7.34230936e-01 -1.10540605e+00 -5.03111303e-01 -5.05868971e-01 1.66546211e-01 9.86670077e-01 8.30395281e-01 2.13232294e-01 5.72532594e-01 1.02055585e+00 -3.41610163e-01 -9.24470723e-01 -8.35381269e-01 -8.52026880e-01 -4.33970422e-01 2.96311051e-01 -9.13517594e-01 -1.03744841e+00 -6.87024057e-01]
[5.277563571929932, 7.2189531326293945]
89fbf220-bd99-42a9-bc0e-7d401906bda2
on-hallucinating-context-and-background
1811.07104
null
https://arxiv.org/abs/1811.07104v3
https://arxiv.org/pdf/1811.07104v3.pdf
On Hallucinating Context and Background Pixels from a Face Mask using Multi-scale GANs
We propose a multi-scale GAN model to hallucinate realistic context (forehead, hair, neck, clothes) and background pixels automatically from a single input face mask. Instead of swapping a face on to an existing picture, our model directly generates realistic context and background pixels based on the features of the provided face mask. Unlike face inpainting algorithms, it can generate realistic hallucinations even for a large number of missing pixels. Our model is composed of a cascaded network of GAN blocks, each tasked with hallucination of missing pixels at a particular resolution while guiding the synthesis process of the next GAN block. The hallucinated full face image is made photo-realistic by using a combination of reconstruction, perceptual, adversarial and identity preserving losses at each block of the network. With a set of extensive experiments, we demonstrate the effectiveness of our model in hallucinating context and background pixels from face masks varying in facial pose, expression and lighting, collected from multiple datasets subject disjoint with our training data. We also compare our method with two popular face swapping and face completion methods in terms of visual quality and recognition performance. Additionally, we analyze our cascaded pipeline and compare it with the recently proposed progressive growing of GANs.
['Patrick J. Flynn', 'Kevin W. Bowyer', 'Walter J. Scheirer', 'Sandipan Banerjee']
2018-11-17
null
null
null
null
['facial-inpainting']
['computer-vision']
[ 7.91757464e-01 5.21411598e-01 4.90736455e-01 -4.19552207e-01 -7.77697802e-01 -5.32161117e-01 5.13394773e-01 -1.01160991e+00 1.33254513e-01 9.48570609e-01 4.16917562e-01 3.27018350e-01 5.96857548e-01 -5.74334085e-01 -8.90112460e-01 -6.47300422e-01 5.39870322e-01 3.35639179e-01 -3.41519654e-01 -2.38244608e-02 -3.74201685e-01 7.91228414e-01 -1.56585670e+00 6.85633898e-01 3.53312582e-01 9.50192511e-01 1.30944941e-02 6.60623550e-01 2.34206438e-01 8.70896280e-01 -7.68925905e-01 -7.09401071e-01 6.18918061e-01 -8.43034148e-01 -4.39444691e-01 7.00728834e-01 1.10488021e+00 -8.64722371e-01 -4.62887853e-01 7.38206685e-01 5.34488797e-01 -9.81123745e-02 2.61435390e-01 -1.33648193e+00 -7.67242074e-01 1.42731100e-01 -9.02430177e-01 -4.01105791e-01 5.78463256e-01 4.99009669e-01 4.60381120e-01 -1.29163289e+00 9.23942626e-01 1.65348899e+00 6.51801527e-01 1.24114668e+00 -1.57376063e+00 -9.20871615e-01 -4.18997929e-02 -3.94556791e-01 -1.26167917e+00 -9.86956418e-01 7.48117387e-01 -4.27943282e-02 3.35056335e-01 2.51765311e-01 5.27878046e-01 1.52922630e+00 -5.54638132e-02 4.10077393e-01 1.28090310e+00 -3.11861813e-01 1.44577533e-01 6.61076456e-02 -8.49457622e-01 6.53450966e-01 -2.00638324e-01 1.17986135e-01 -6.34743989e-01 -3.58864486e-01 1.11275673e+00 7.99805000e-02 -5.01953542e-01 -1.58108875e-01 -7.55964518e-01 4.73755002e-01 1.51190728e-01 -2.23440096e-01 -5.80096483e-01 3.95150125e-01 -2.43765101e-01 2.30954096e-01 5.01716673e-01 -4.01197746e-03 -8.06759447e-02 5.18517196e-01 -1.40699291e+00 3.08167368e-01 6.16204083e-01 9.75103796e-01 9.08544362e-01 4.46047544e-01 -5.33461392e-01 9.40352976e-01 9.92228240e-02 4.84246910e-01 -9.08192098e-02 -1.54116750e+00 4.09154333e-02 1.24402009e-01 2.72062421e-01 -6.30432725e-01 2.65238434e-01 -1.33401588e-01 -7.49862313e-01 6.15207314e-01 1.38733655e-01 -3.95433843e-01 -1.22272480e+00 2.10339856e+00 4.63377327e-01 4.45567608e-01 -1.08244933e-01 8.84565532e-01 8.20322931e-01 4.95327920e-01 -3.98062468e-02 -3.02888989e-01 1.21582186e+00 -1.10017192e+00 -8.94086480e-01 -3.73840392e-01 -4.82675135e-01 -9.38924670e-01 8.99227798e-01 4.12455559e-01 -1.72508395e+00 -6.32896245e-01 -6.41073585e-01 -3.93577307e-01 2.27243289e-01 1.25329539e-01 2.59614110e-01 5.75757325e-01 -1.73714745e+00 3.96777034e-01 -3.84658664e-01 -1.99060053e-01 7.83540249e-01 3.40711981e-01 -6.77838564e-01 -4.81870353e-01 -7.34946489e-01 6.58131063e-01 -1.75004795e-01 1.75765812e-01 -1.58340025e+00 -7.89065182e-01 -9.12007451e-01 -5.42234294e-02 1.62230190e-02 -9.42283750e-01 1.09355664e+00 -1.65416098e+00 -1.44205630e+00 1.01009429e+00 -4.65290248e-01 -1.18441425e-01 8.60111833e-01 -1.38584360e-01 -3.22391391e-01 2.16381088e-01 6.67282566e-02 1.29231882e+00 1.54513752e+00 -1.77711034e+00 -1.32905528e-01 -3.15116584e-01 -1.61962003e-01 1.86503068e-01 4.09354381e-02 1.93454742e-01 -7.18790591e-01 -8.18438172e-01 -3.32588851e-01 -8.37312281e-01 -1.04359584e-02 5.54495573e-01 -3.64418358e-01 6.29706800e-01 1.26060808e+00 -1.37847054e+00 5.46185255e-01 -2.26286697e+00 2.56222427e-01 -7.75354169e-03 1.84186503e-01 1.06218837e-01 -4.78312552e-01 3.67462456e-01 -3.47336978e-01 -1.19042695e-01 -6.58261478e-01 -1.15903568e+00 -3.88102025e-01 3.67608786e-01 -5.17110169e-01 4.62373257e-01 5.32786191e-01 9.77368474e-01 -6.41829133e-01 -2.10127130e-01 1.56629696e-01 1.11915195e+00 -6.43525422e-01 4.75182712e-01 -3.63057822e-01 9.92935598e-01 3.49510998e-01 8.68554890e-01 1.14343119e+00 3.46103348e-02 3.59559506e-01 -9.54366848e-02 4.81337577e-01 -2.84277380e-01 -9.35753644e-01 1.92393005e+00 -5.71507812e-01 6.00761652e-01 6.90556586e-01 -2.45891765e-01 7.64224350e-01 5.44945061e-01 1.98603526e-01 -2.86999315e-01 -3.12690511e-02 3.98450857e-03 -3.55996877e-01 -2.39943847e-01 2.70187169e-01 -4.01928097e-01 3.47353697e-01 4.75127548e-01 1.92986846e-01 -3.11741471e-01 -2.56183058e-01 5.83406761e-02 1.03177285e+00 4.15459186e-01 -1.38461322e-01 2.70395190e-01 3.91978562e-01 -6.16202116e-01 3.41699004e-01 2.16497898e-01 1.92262065e-02 1.29286981e+00 4.57191139e-01 -2.86866695e-01 -1.28097773e+00 -1.25465643e+00 7.11031407e-02 7.89414406e-01 -2.49894857e-01 -8.68825614e-02 -1.17969704e+00 -6.66124701e-01 -1.03190411e-02 6.92560554e-01 -1.03724289e+00 -5.70331849e-02 -5.55572152e-01 -3.13774824e-01 5.56911051e-01 2.57554412e-01 7.15849578e-01 -1.43869185e+00 -2.38198131e-01 -1.62960410e-01 -2.17394471e-01 -1.27445006e+00 -7.92847574e-01 -4.77436185e-01 -5.41506767e-01 -9.10910249e-01 -8.85333300e-01 -8.73002887e-01 1.02966654e+00 1.59612060e-01 1.29112232e+00 1.51570752e-01 -4.27669317e-01 5.25113046e-01 6.65136799e-02 -5.18980101e-02 -7.04770386e-01 -6.52261376e-01 -1.12400956e-01 6.79218471e-01 -2.96418518e-01 -8.88851643e-01 -8.79259825e-01 1.44194365e-01 -1.29325688e+00 2.43366465e-01 4.72845703e-01 8.92231047e-01 5.25701404e-01 -1.75890625e-01 3.00570965e-01 -9.19344842e-01 4.78125572e-01 -5.64766884e-01 -3.19584548e-01 1.92633584e-01 -1.47713562e-02 -3.25086415e-01 4.87945497e-01 -4.26560104e-01 -1.34799278e+00 2.63449013e-01 -7.62189999e-02 -1.05050921e+00 -2.23010078e-01 -4.25130248e-01 -7.22336292e-01 -2.20258921e-01 4.80118990e-01 2.48167783e-01 2.25462630e-01 -4.21960175e-01 6.65095985e-01 4.12941307e-01 8.43308389e-01 -4.10998434e-01 8.35539818e-01 8.98787022e-01 -1.30154982e-01 -5.93929410e-01 -5.39482176e-01 3.25022697e-01 -4.87159491e-01 -2.58432090e-01 7.74967253e-01 -1.20744860e+00 -3.92794877e-01 5.62746704e-01 -1.30300558e+00 -4.66031849e-01 -6.46424055e-01 -2.36083046e-01 -6.35224700e-01 1.12513721e-01 -6.33842885e-01 -8.26615274e-01 -3.38325471e-01 -1.02368534e+00 1.62705600e+00 1.21035114e-01 -1.91844985e-01 -7.28335083e-01 -1.53846040e-01 5.74787438e-01 4.13491994e-01 7.69230843e-01 4.33398753e-01 2.25469872e-01 -5.41697800e-01 4.52292934e-02 -2.18000561e-01 7.29715288e-01 4.06807154e-01 -8.98806378e-02 -1.52313042e+00 -5.53780913e-01 2.28228346e-01 -5.24016082e-01 8.38863552e-01 2.16406733e-01 1.22504675e+00 -5.96370578e-01 -5.55402525e-02 8.39192033e-01 1.38919759e+00 -1.04318894e-02 1.01909792e+00 -4.89021122e-01 7.24775016e-01 8.60107780e-01 9.44980308e-02 4.08661455e-01 -9.15553272e-02 5.73873520e-01 7.06417441e-01 -4.70697552e-01 -8.10925484e-01 -5.49782038e-01 6.55635417e-01 -5.40345125e-02 7.26121217e-02 -1.06298491e-01 -3.47512305e-01 5.28140008e-01 -1.33264744e+00 -1.05097568e+00 3.94275099e-01 2.10394645e+00 9.71030235e-01 -5.17975807e-01 6.94438815e-02 -2.60050833e-01 8.49201739e-01 1.52085917e-02 -7.68561900e-01 -3.07137847e-01 -4.10503149e-01 6.71622276e-01 -1.60494391e-02 7.61188567e-01 -5.95090210e-01 1.07024610e+00 6.71398878e+00 6.20813310e-01 -1.09822845e+00 4.24016356e-01 1.19995737e+00 -6.31249309e-01 -5.40865362e-01 -8.83403271e-02 -4.35301334e-01 4.06400412e-01 6.16486728e-01 4.78644192e-01 9.86908317e-01 4.78736877e-01 1.13177851e-01 4.83084433e-02 -1.22817218e+00 1.08709991e+00 5.23434043e-01 -1.32985139e+00 3.14826012e-01 7.36689121e-02 1.18998992e+00 -3.90775681e-01 3.90629709e-01 -1.71479315e-01 4.68123943e-01 -1.57968032e+00 8.77810478e-01 5.57944179e-01 1.53217328e+00 -5.79532146e-01 1.21138558e-01 -1.35871544e-01 -6.82416081e-01 -2.56198514e-02 -5.67071587e-02 1.36183247e-01 2.22575262e-01 3.82319629e-01 -7.89838254e-01 2.80590773e-01 5.98921061e-01 3.95030260e-01 -3.79780114e-01 3.02141786e-01 -4.09355015e-01 2.94634104e-01 -1.24144770e-01 1.02751815e+00 -2.78857559e-01 -1.93351254e-01 4.44921404e-01 8.79882574e-01 3.56873870e-01 3.82373147e-02 -2.57225186e-01 1.40014601e+00 -5.66538274e-01 -2.63453633e-01 -7.16080487e-01 3.68732661e-01 4.68665242e-01 1.45722294e+00 -2.39701971e-01 -2.47539058e-01 -3.98931533e-01 1.65144968e+00 2.46686175e-01 7.15536892e-01 -8.97880137e-01 3.10834557e-01 8.90618622e-01 4.23045993e-01 3.30424041e-01 1.39631093e-01 -1.47524282e-01 -1.02487409e+00 2.13211238e-01 -1.23154140e+00 5.50778536e-03 -1.23649442e+00 -1.26021230e+00 8.59884977e-01 -2.30687261e-01 -8.77824962e-01 -1.57920197e-01 -1.85657635e-01 -6.96329594e-01 1.32397497e+00 -1.39077842e+00 -1.67000365e+00 -6.56204760e-01 9.34961975e-01 6.62254930e-01 -1.73259020e-01 8.81533325e-01 2.09515989e-01 -5.03662586e-01 6.85088873e-01 -2.87141919e-01 -2.25655697e-02 9.22598660e-01 -7.58966029e-01 6.33822203e-01 8.52091610e-01 -9.86853763e-02 4.06186700e-01 6.99524164e-01 -7.67507851e-01 -1.18222439e+00 -1.45435286e+00 3.48915488e-01 -4.29387778e-01 -1.50139686e-02 -7.07082808e-01 -6.64944768e-01 1.00347054e+00 7.24557042e-01 3.09454769e-01 5.03266573e-01 -6.91197813e-01 -4.12840307e-01 -1.72205314e-01 -1.76534915e+00 7.67800212e-01 1.19818974e+00 -5.15757620e-01 -1.31021217e-01 3.06204498e-01 5.58755398e-01 -4.42426443e-01 -5.90027332e-01 2.61414200e-01 7.01265156e-01 -1.25446296e+00 9.80581880e-01 -4.42305773e-01 7.57515430e-01 -3.52917522e-01 -2.46055573e-01 -1.27479422e+00 -5.29311560e-02 -1.12586677e+00 -6.08343370e-02 1.38368821e+00 6.20749593e-02 -3.79886597e-01 8.85105610e-01 7.15535939e-01 1.16914861e-01 -4.92865533e-01 -7.49750316e-01 -3.84750485e-01 -8.55734572e-02 -6.22859038e-02 8.09573650e-01 7.99821496e-01 -8.11720252e-01 1.09263703e-01 -9.35347080e-01 9.53233764e-02 7.65687287e-01 -2.71725450e-02 1.04370761e+00 -6.24604702e-01 -5.44514596e-01 8.88574421e-02 5.48181310e-02 -5.58358729e-01 3.26737672e-01 -5.60540617e-01 -6.07452951e-02 -1.24476230e+00 3.02434295e-01 -9.63605382e-03 2.67126590e-01 7.62451708e-01 -4.87241782e-02 9.10907626e-01 2.89288193e-01 1.36832997e-01 3.89089920e-02 5.73222280e-01 1.59848702e+00 1.36379087e-02 7.69995898e-02 -3.50273401e-01 -8.73869598e-01 6.15398467e-01 4.39945698e-01 -2.42365628e-01 -5.34745574e-01 -5.32217979e-01 -1.15152247e-01 3.62805575e-01 7.61744738e-01 -8.43730032e-01 -2.26461083e-01 -1.09183930e-01 1.07497704e+00 3.42792459e-02 9.54501808e-01 -8.46407413e-01 8.82861435e-01 2.04792768e-01 -2.81548053e-01 -7.75328325e-03 3.96274000e-01 5.00882030e-01 -1.26752093e-01 3.11537594e-01 1.07501495e+00 -3.10466230e-01 -3.09144974e-01 4.14866805e-01 6.96483627e-02 -4.39360663e-02 1.05452323e+00 -3.59426051e-01 -7.41464719e-02 -8.83935034e-01 -9.69005525e-01 -1.66204587e-01 1.02957213e+00 4.14308220e-01 9.25682306e-01 -1.43040156e+00 -1.03850949e+00 7.92635381e-01 -3.49490732e-01 1.31768107e-01 4.61160749e-01 4.37527627e-01 -5.57344139e-01 -3.00255090e-01 -5.58333576e-01 -3.08070391e-01 -1.32188630e+00 5.10094106e-01 5.00930429e-01 6.21706396e-02 -3.36362451e-01 9.25045252e-01 8.61696422e-01 -2.89977133e-01 9.08722803e-02 2.21139044e-01 3.44611675e-01 -2.28808969e-01 7.17111826e-01 1.52722254e-01 -5.24016283e-02 -8.19065630e-01 -1.12637103e-01 4.83883828e-01 7.27781728e-02 -5.16313672e-01 1.26031339e+00 -1.93238318e-01 -4.74971652e-01 -5.24879061e-02 1.01751876e+00 3.90554458e-01 -1.94608510e+00 1.57689154e-02 -8.81167531e-01 -7.92566240e-01 -2.84506440e-01 -1.04579723e+00 -1.49572897e+00 6.26462519e-01 4.96607393e-01 -4.71810430e-01 1.68749213e+00 -3.61639261e-02 8.34867001e-01 -5.57640612e-01 2.94179887e-01 -5.53815901e-01 3.27343911e-01 8.95221066e-03 1.49240196e+00 -1.01265848e+00 -1.73256129e-01 -5.91655731e-01 -6.90476835e-01 7.97395349e-01 7.01261640e-01 -2.91190058e-01 4.66714919e-01 6.01809680e-01 1.00760154e-01 3.27204578e-02 -7.56803274e-01 2.02949092e-01 1.80465221e-01 8.18216324e-01 2.54529178e-01 -1.86696202e-01 2.81623513e-01 2.65963346e-01 -2.77373552e-01 -2.72490680e-02 4.92800802e-01 5.72344840e-01 9.17540938e-02 -1.20007324e+00 -7.16244817e-01 3.95945981e-02 -4.45801347e-01 -2.46557251e-01 -7.07487583e-01 6.16266429e-01 4.33443576e-01 8.77881706e-01 6.86737373e-02 -9.81695205e-02 -2.61963614e-05 1.28666386e-01 7.94933021e-01 -7.98004389e-01 -6.00003183e-01 1.45051196e-01 -7.82517567e-02 -7.88701534e-01 -2.63299227e-01 -7.67102242e-01 -7.67644405e-01 -3.21459115e-01 1.90517172e-01 -5.13237774e-01 5.12048364e-01 5.88351190e-01 5.50928175e-01 4.06781256e-01 7.33966351e-01 -1.16543984e+00 -1.90580860e-01 -1.00044060e+00 -6.55458033e-01 6.75232589e-01 5.23442030e-01 -3.66362065e-01 -2.79595435e-01 4.58522260e-01]
[12.566381454467773, -0.27086618542671204]
7d52f6d5-9d93-4525-b8db-bf1ff269a0a8
icon-interactive-conversational-memory
null
null
https://aclanthology.org/D18-1280
https://aclanthology.org/D18-1280.pdf
ICON: Interactive Conversational Memory Network for Multimodal Emotion Detection
Emotion recognition in conversations is crucial for building empathetic machines. Present works in this domain do not explicitly consider the inter-personal influences that thrive in the emotional dynamics of dialogues. To this end, we propose Interactive COnversational memory Network (ICON), a multimodal emotion detection framework that extracts multimodal features from conversational videos and hierarchically models the self- and inter-speaker emotional influences into global memories. Such memories generate contextual summaries which aid in predicting the emotional orientation of utterance-videos. Our model outperforms state-of-the-art networks on multiple classification and regression tasks in two benchmark datasets.
['Devamanyu Hazarika', 'Soujanya Poria', 'Roger Zimmermann', 'Erik Cambria', 'Rada Mihalcea']
2018-10-01
null
null
null
emnlp-2018-10
['multimodal-emotion-recognition', 'emotion-recognition-in-conversation', 'multimodal-emotion-recognition']
['computer-vision', 'natural-language-processing', 'speech']
[-3.23652983e-01 -7.73652866e-02 -3.42920154e-01 -6.39128268e-01 -3.15611988e-01 -3.68402898e-01 6.96755171e-01 -5.69413938e-02 -7.52659440e-02 6.18228972e-01 1.02144814e+00 4.59813714e-01 2.91165859e-01 -5.30893266e-01 -2.58038193e-01 -5.04058659e-01 -1.46083370e-01 -5.84467612e-02 -4.13912266e-01 -4.70537156e-01 4.07169193e-01 2.34087303e-01 -1.45935202e+00 1.25775981e+00 2.96272904e-01 1.09032488e+00 -3.23579252e-01 1.02258050e+00 -4.84438688e-01 1.70012379e+00 -6.30335391e-01 -5.92896879e-01 -7.89360881e-01 -7.91327059e-01 -8.62144291e-01 1.33520663e-01 -1.17981754e-01 -5.65515086e-02 -4.21751022e-01 7.41330385e-01 5.16489089e-01 3.44875723e-01 7.89310396e-01 -1.27714145e+00 -4.53636587e-01 6.83445990e-01 -1.99573457e-01 1.64214388e-01 6.80848062e-01 -1.68016791e-01 9.09358621e-01 -1.01487255e+00 1.01666081e+00 1.47482467e+00 5.87251365e-01 6.53009593e-01 -8.47065806e-01 -4.83952224e-01 1.60998851e-01 5.69925368e-01 -6.97726846e-01 -6.06591165e-01 1.29725683e+00 -4.39400643e-01 1.18734992e+00 2.99346149e-01 8.53206754e-01 1.91288519e+00 4.37589943e-01 1.02460623e+00 9.18370128e-01 -2.11285263e-01 -9.94765311e-02 3.89568150e-01 2.79723674e-01 3.61216992e-01 -9.92104053e-01 -2.72858679e-01 -1.28755307e+00 -2.04477787e-01 2.48649701e-01 -6.34362474e-02 -2.12031990e-01 1.09752364e-01 -8.93322766e-01 1.08580148e+00 2.86716968e-01 5.00700712e-01 -6.22134149e-01 -1.31925419e-02 8.70420933e-01 5.72252870e-01 6.32260680e-01 2.86013514e-01 -1.44904420e-01 -7.47422576e-01 -4.43742812e-01 -1.97647467e-01 1.10571337e+00 4.01192069e-01 4.19376016e-01 -1.40879452e-01 -4.37673889e-02 1.34321713e+00 -2.80666798e-02 5.34548573e-02 5.03706336e-01 -1.04821527e+00 6.55150712e-02 8.33952844e-01 -1.15255892e-01 -1.49080908e+00 -7.26337254e-01 -2.93422621e-02 -6.77489460e-01 -3.20791841e-01 -1.66775063e-01 -4.75881964e-01 1.78394392e-02 1.79355550e+00 2.97501832e-01 4.13140170e-02 4.03361291e-01 7.84238040e-01 1.47282553e+00 9.80115712e-01 2.92022765e-01 -4.90290254e-01 1.30447614e+00 -1.25026214e+00 -1.17700100e+00 -2.64986545e-01 4.27663207e-01 -7.44693696e-01 7.52966464e-01 4.48089272e-01 -1.16256785e+00 -4.29933906e-01 -8.97908866e-01 7.38347396e-02 -5.13547063e-01 -5.50949648e-02 5.87170243e-01 1.40593648e-01 -6.65379703e-01 2.99414366e-01 -3.80655408e-01 -3.57940316e-01 1.50076464e-01 3.61808836e-02 -5.54887474e-01 3.85542721e-01 -1.31948209e+00 1.16733718e+00 4.68844026e-02 2.23931268e-01 -6.31988049e-01 -2.91396290e-01 -7.87133694e-01 -1.19908042e-01 1.06460892e-01 -3.70070487e-01 1.26601899e+00 -1.62561834e+00 -1.94958866e+00 7.74682343e-01 -2.44306684e-01 -1.35699689e-01 -4.09275182e-02 -3.58459115e-01 -6.65816545e-01 4.95479703e-01 -5.79972088e-01 8.11998785e-01 7.46916890e-01 -1.38204682e+00 -2.35414639e-01 -3.13684851e-01 -1.68198183e-01 3.99087846e-01 -7.52224207e-01 4.22053486e-01 -2.40609571e-01 -4.02630717e-01 -2.99509853e-01 -8.25730741e-01 -4.20818441e-02 -6.03139579e-01 -2.27455422e-01 -3.42979014e-01 1.07455742e+00 -5.58353186e-01 1.34069037e+00 -2.25310802e+00 5.91639161e-01 -7.73170218e-02 1.38970822e-01 -2.44333118e-01 -1.81710064e-01 7.98374891e-01 -8.78912769e-03 -2.88721085e-01 4.05277640e-01 -6.13855720e-01 1.68533131e-01 1.31117001e-01 -5.02742589e-01 1.91981211e-01 4.19769734e-02 8.68429720e-01 -7.76498437e-01 -7.20363736e-01 1.08562857e-01 7.58634865e-01 -4.47131068e-01 6.41940475e-01 -1.38622254e-01 6.53377652e-01 -2.62463301e-01 5.57451248e-01 1.31875798e-01 1.40955925e-01 4.29526508e-01 -2.26463437e-01 -9.72281396e-02 2.13877335e-01 -5.22206187e-01 1.59463573e+00 -6.76224768e-01 1.07574892e+00 3.64414096e-01 -7.31171012e-01 1.03622270e+00 5.54612339e-01 4.91118878e-01 -6.91936433e-01 5.97473621e-01 -2.60547876e-01 -3.24341476e-01 -1.08494687e+00 7.00631022e-01 -3.15771848e-01 -6.15604639e-01 5.65660477e-01 3.53082061e-01 2.12057725e-01 -2.03383356e-01 2.52376974e-01 7.36513317e-01 -1.80291623e-01 4.23425794e-01 1.62013650e-01 5.22334397e-01 -3.92380863e-01 5.51188290e-01 3.64275992e-01 -4.00333822e-01 1.81098714e-01 9.17401671e-01 -6.07476175e-01 -3.51140350e-01 -7.02021778e-01 6.57583326e-02 1.74989545e+00 1.40334263e-01 -5.38765490e-01 -6.14255011e-01 -4.99768645e-01 -4.46462750e-01 7.29309440e-01 -1.01825559e+00 -2.66586721e-01 -1.71485841e-01 -4.61125106e-01 4.83862728e-01 5.31959414e-01 2.40309328e-01 -1.83403897e+00 -5.52464247e-01 2.81954974e-01 -7.59213626e-01 -1.11933017e+00 -3.42548311e-01 1.77382037e-01 -4.16890800e-01 -8.53750706e-01 -1.76094696e-02 -6.62569344e-01 1.85754478e-01 -1.55996174e-01 1.22311890e+00 -2.46542260e-01 -3.77891719e-01 8.45479071e-01 -5.31884491e-01 -3.20535153e-01 -6.49135649e-01 -1.29622281e-01 -1.38708845e-01 3.22685659e-01 5.41274786e-01 -8.54124844e-01 -5.06307364e-01 3.55596393e-01 -6.40195012e-01 5.85970320e-02 1.37699708e-01 8.24181259e-01 4.73364778e-02 -3.68315369e-01 8.92533004e-01 -5.83439589e-01 1.08728147e+00 -8.66944849e-01 4.30014431e-01 2.45870531e-01 2.14968637e-01 -3.19896936e-01 3.87375742e-01 -6.93182528e-01 -1.43952799e+00 -1.28984764e-01 -1.12407148e-01 -3.50808859e-01 -2.98186898e-01 6.02749586e-01 -3.00682560e-02 2.48961285e-01 3.86894584e-01 -3.64502601e-04 -2.25249343e-02 1.15012275e-02 5.42004645e-01 1.01804650e+00 7.35928178e-01 -5.00585020e-01 -4.04229432e-01 4.18565482e-01 -3.57516438e-01 -1.09253645e+00 -8.31069589e-01 -5.71577191e-01 -5.68911672e-01 -1.13628042e+00 1.14981246e+00 -8.86496484e-01 -1.11119413e+00 4.09549296e-01 -1.43671596e+00 -7.91382641e-02 2.43521258e-01 3.30425650e-01 -7.68070042e-01 1.13736570e-01 -1.05822229e+00 -1.16592801e+00 -5.35843730e-01 -7.41447568e-01 6.20869815e-01 3.52670431e-01 -1.07582855e+00 -1.12674463e+00 5.18702686e-01 6.32209718e-01 4.00291711e-01 4.78000641e-01 5.72937131e-01 -5.99214017e-01 1.66092679e-01 -1.16106614e-01 2.30767816e-01 1.93316281e-01 -3.46660197e-01 2.04593927e-01 -1.11766720e+00 3.89003754e-01 1.28736809e-01 -1.03320813e+00 8.53291392e-01 1.12285532e-01 7.52546072e-01 -5.98715365e-01 -7.84640461e-02 9.94592011e-02 7.65311718e-01 3.29087406e-01 7.97898054e-01 1.84007525e-01 3.24241459e-01 1.25863016e+00 6.41406000e-01 8.58479917e-01 5.52774429e-01 4.67143774e-01 5.68396032e-01 1.05732366e-01 5.29190779e-01 -1.00759074e-01 9.87023413e-01 1.19460392e+00 3.04894336e-02 -2.27970123e-01 -5.97064137e-01 4.65359896e-01 -2.30891681e+00 -1.48309445e+00 -1.12381980e-01 1.22560787e+00 8.01286876e-01 -3.46468031e-01 8.21991637e-02 -5.44513226e-01 6.26951039e-01 5.60457647e-01 -5.05227625e-01 -1.10905671e+00 -4.01307851e-01 -1.80836573e-01 -5.59086978e-01 5.42742252e-01 -1.18123472e+00 9.22722876e-01 6.01547003e+00 3.53989810e-01 -1.11571133e+00 1.02295555e-01 7.95863211e-01 -4.57707226e-01 3.49587537e-02 -4.60817665e-01 -3.55591446e-01 9.18616578e-02 1.20267928e+00 1.60456076e-01 4.61982667e-01 1.10215974e+00 1.92072496e-01 -1.58847407e-01 -1.06833541e+00 1.16314530e+00 5.85520566e-01 -1.12991726e+00 -1.87101230e-01 -2.80478090e-01 5.86224973e-01 -7.80224502e-02 6.01974465e-02 5.76895654e-01 2.52607954e-03 -1.10832834e+00 4.15916890e-01 1.10053134e+00 6.84971735e-02 -1.12342525e+00 7.85034716e-01 3.10103565e-01 -8.62604201e-01 -1.65431976e-01 -1.80195779e-01 -6.70482442e-02 4.01340127e-01 9.45520997e-02 -5.37515342e-01 2.01528869e-03 8.13765764e-01 7.95517027e-01 -2.06775874e-01 1.98466912e-01 -1.08723350e-01 4.47589010e-01 1.29996613e-01 -2.58540422e-01 2.58288413e-01 -2.25500777e-01 6.19917512e-01 1.89497316e+00 1.44557253e-01 3.78281683e-01 -1.92161277e-01 6.38183653e-01 -2.75949329e-01 4.55625474e-01 -6.71230078e-01 -5.18285334e-01 3.62120420e-01 1.72507751e+00 -3.38117242e-01 -2.26287737e-01 -2.63487965e-01 1.13208354e+00 5.26586235e-01 7.06185400e-02 -9.44990158e-01 -1.37849107e-01 8.03047299e-01 -6.14051580e-01 1.04516722e-01 8.67522582e-02 -3.19375955e-02 -1.15082896e+00 -3.62468779e-01 -9.59799230e-01 5.67554355e-01 -8.94567847e-01 -1.53344142e+00 8.83442521e-01 -4.89618659e-01 -9.08412635e-01 -5.37216365e-01 -3.47866982e-01 -1.06871104e+00 1.18469700e-01 -9.39564466e-01 -1.17597294e+00 -4.89791304e-01 7.60051727e-01 6.59827828e-01 -1.93179503e-01 1.16604650e+00 6.98723048e-02 -5.91401935e-01 2.92256087e-01 -3.73918444e-01 1.10061415e-01 1.11527848e+00 -7.51914263e-01 -5.74361026e-01 5.96878454e-02 1.03560947e-02 5.87185144e-01 7.87528992e-01 -2.52502024e-01 -1.33802390e+00 -6.06080413e-01 1.05313969e+00 -5.31573117e-01 9.86357450e-01 -3.58604312e-01 -7.43907928e-01 4.80327308e-01 1.10466444e+00 -5.14948606e-01 1.54065776e+00 4.85607892e-01 -3.97881716e-01 2.05676779e-01 -8.66559923e-01 5.80113053e-01 7.62327015e-01 -7.78943121e-01 -7.30136633e-01 7.57978996e-03 4.77550924e-01 -9.75705907e-02 -1.18191779e+00 2.74886072e-01 9.62295949e-01 -1.41283357e+00 8.67994368e-01 -1.03741825e+00 1.10306275e+00 4.88613486e-01 -3.70859772e-01 -1.44924557e+00 1.50814146e-01 -6.74138844e-01 -2.50394821e-01 1.49654090e+00 2.26129159e-01 -1.79598466e-01 4.15658057e-01 6.67266071e-01 -1.75499663e-01 -8.51958513e-01 -8.36207509e-01 1.31814376e-01 -1.88363343e-01 -5.08198917e-01 1.18785009e-01 1.41564310e+00 1.08534372e+00 8.56541216e-01 -8.64099324e-01 -2.77509391e-01 8.12516510e-02 2.49347687e-01 9.05565202e-01 -9.94658887e-01 -6.84340000e-02 -7.39980340e-01 -2.46496737e-01 -4.77661431e-01 7.51671970e-01 -4.72010583e-01 -3.76612395e-02 -1.15224695e+00 5.33222079e-01 4.70175773e-01 -1.38889760e-01 2.91846603e-01 1.44090980e-01 1.60297856e-01 1.85519740e-01 -1.17030472e-01 -1.27428532e+00 1.04217887e+00 9.26527441e-01 1.94585100e-02 -5.35029233e-01 -5.06351650e-01 -5.66051781e-01 9.41719055e-01 9.37371910e-01 -1.95481390e-01 -1.07315339e-01 2.61925250e-01 4.33315516e-01 5.51261604e-01 2.32977435e-01 -6.45094931e-01 3.30759853e-01 -2.18778253e-01 2.51315534e-01 -7.74712980e-01 9.53762710e-01 -6.16877854e-01 2.23589633e-02 -1.29615337e-01 -7.87148774e-01 1.03648692e-01 1.37923986e-01 4.47231680e-01 -8.21101546e-01 9.85667333e-02 6.79967105e-01 -7.83626214e-02 -4.71640319e-01 -1.20252237e-01 -8.76998186e-01 -1.37580395e-01 1.01956809e+00 5.33606261e-02 -6.13159657e-01 -1.11267507e+00 -1.22345734e+00 3.12372744e-01 6.03920023e-04 8.35499942e-01 7.23649859e-01 -1.50495827e+00 -4.52263445e-01 -2.71634489e-01 2.09162638e-01 -7.79111803e-01 8.71620059e-01 1.07813823e+00 1.15832657e-01 2.32817501e-01 -4.63406235e-01 -3.47839296e-01 -1.84861994e+00 3.91012013e-01 3.65585208e-01 -2.04639867e-01 -1.32358059e-01 8.49717319e-01 2.40551531e-01 -5.52628696e-01 4.19924796e-01 3.05280983e-01 -8.06163371e-01 8.52121294e-01 6.42193556e-01 4.19877321e-01 -5.14031708e-01 -1.02950621e+00 -1.21989481e-01 2.33424038e-01 4.91419844e-02 -2.02972561e-01 1.68974721e+00 -3.35169256e-01 -5.46743333e-01 1.13623023e+00 1.52113366e+00 -5.99670745e-02 -9.26533222e-01 3.02099753e-02 -1.31501123e-01 6.85050562e-02 -6.95440099e-02 -8.64357889e-01 -8.02042425e-01 1.08518171e+00 9.61792916e-02 1.99486747e-01 1.09574676e+00 1.89973071e-01 9.88179743e-01 6.52785003e-01 -9.36502367e-02 -1.75547385e+00 8.86740029e-01 7.62043595e-01 1.28572452e+00 -1.21253526e+00 -3.94246250e-01 -5.57061881e-02 -1.43531382e+00 1.56113541e+00 7.67562032e-01 9.93416458e-02 6.77688062e-01 1.73184291e-01 4.16037679e-01 -4.72455800e-01 -1.40111196e+00 2.77222842e-01 2.45817199e-01 1.34023383e-01 5.46377242e-01 9.87733752e-02 -1.58063993e-01 1.30256248e+00 -2.25188375e-01 -2.67942607e-01 4.06057596e-01 7.65480340e-01 -4.78470474e-01 -7.86909997e-01 -4.21720356e-01 -9.23247412e-02 -4.76952165e-01 1.88754201e-01 -1.31174111e+00 6.58253431e-01 -4.69465405e-02 1.10072947e+00 6.89372346e-02 -8.32893491e-01 1.45536482e-01 6.08831823e-01 2.27772981e-01 -2.07840465e-02 -1.00566161e+00 2.10188344e-01 7.48969018e-01 -7.96065032e-01 -8.69972706e-01 -8.44867051e-01 -1.24765599e+00 -2.98794448e-01 7.64596909e-02 2.01042607e-01 8.03039789e-01 8.18092465e-01 5.23449302e-01 4.28249419e-01 8.35623622e-01 -1.10609460e+00 1.27085388e-01 -1.23270857e+00 -3.58720839e-01 5.52366793e-01 7.04767779e-02 -5.53464234e-01 -6.41229808e-01 1.70094706e-02]
[13.0523681640625, 5.950223922729492]
073e600a-1a4a-44a0-9581-daf0170d5209
security-of-distributed-parameter-cyber
2107.14159
null
https://arxiv.org/abs/2107.14159v2
https://arxiv.org/pdf/2107.14159v2.pdf
Security of Distributed Parameter Cyber-Physical Systems: Cyber-Attack Detection in Linear Parabolic PDEs
Security of Distributed Parameter Cyber-Physical Systems (DPCPSs) is of critical importance in the face of cyber-attack threats. Although security aspects of Cyber-Physical Systems (CPSs) modelled by Ordinary differential Equations (ODEs) have been extensively explored during the past decade, security of DPCPSs has not received its due attention despite its safety-critical nature. In this work, we explore the security aspects of DPCPSs from a system theoretic viewpoint. Specifically, we focus on DPCPSs modelled by linear parabolic Partial Differential Equations (PDEs) subject to cyber-attacks in actuation channel. First, we explore the detectability of such attacks and derive conditions for stealthy attacks. Next, we develop a design framework for cyber-attack detection algorithms based on output injection observers. Such attack detection algorithms explicitly consider stability, robustness and attack sensitivity in their design. Finally, theoretical analysis and simulation studies are performed to illustrate the effectiveness of the proposed approach.
['Satadru Dey', 'Tanushree Roy']
2021-07-29
null
null
null
null
['cyber-attack-detection']
['miscellaneous']
[ 3.34913135e-01 2.57186234e-01 1.50218934e-01 4.79362369e-01 -3.59151065e-02 -1.02974319e+00 6.71627283e-01 2.50046402e-01 -7.41151273e-02 7.13874102e-01 -5.36542237e-01 -6.50512278e-01 -6.35521352e-01 -4.47080940e-01 -4.29152787e-01 -9.44815159e-01 -3.51546705e-01 -3.87487471e-01 3.03383112e-01 -1.80121541e-01 6.18929826e-02 7.87790477e-01 -8.27330351e-01 -7.91276813e-01 6.69843793e-01 9.98337805e-01 -4.18814719e-01 6.55520558e-01 8.03963006e-01 2.84710288e-01 -7.36792445e-01 1.85696691e-01 3.97799760e-01 -1.53667897e-01 -5.56534767e-01 -2.95010302e-02 -4.81541902e-01 -2.72665173e-01 -3.42344552e-01 1.34244919e+00 2.75331229e-01 2.91141599e-01 6.93189204e-01 -1.98231483e+00 -4.18505222e-02 1.76500767e-01 -3.52783293e-01 6.65298058e-03 2.86941469e-01 4.26772833e-01 5.00991046e-01 -4.37315255e-01 2.94956326e-01 1.14992237e+00 3.03544879e-01 6.12938404e-01 -1.07851291e+00 -4.90801573e-01 3.18054348e-01 -2.46144265e-01 -1.27139962e+00 2.08232939e-01 8.21340263e-01 -4.55089152e-01 4.53222066e-01 6.44015014e-01 5.20650566e-01 9.96927261e-01 8.83904397e-01 4.17758375e-01 9.48814034e-01 -3.43791880e-02 6.40884817e-01 8.25431272e-02 3.24492812e-01 5.48732057e-02 8.27040613e-01 4.05719936e-01 4.99853581e-01 -6.57059908e-01 6.89486921e-01 3.61672118e-02 -4.09580946e-01 -4.22834963e-01 -9.16776061e-01 5.75654805e-01 2.08052099e-01 2.81783640e-01 -5.49525797e-01 -4.85359281e-02 3.85311484e-01 2.34455168e-01 -8.16975832e-02 5.28193772e-01 -3.69093686e-01 1.71632811e-01 3.61089259e-01 4.58213717e-01 9.58459496e-01 9.16965127e-01 -2.95536876e-01 4.90088135e-01 4.74005759e-01 1.58358887e-02 5.01458228e-01 7.52004147e-01 -2.26330191e-01 -6.20219529e-01 1.41250938e-01 4.14255589e-01 5.25009692e-01 -1.14550924e+00 -3.45201075e-01 -8.81444067e-02 -9.27021682e-01 3.84141415e-01 6.13270374e-03 -6.76846445e-01 -2.58054316e-01 1.50499177e+00 6.91925824e-01 1.64284229e-01 4.45569038e-01 8.92032743e-01 -3.60210001e-01 8.98167253e-01 1.59327790e-01 -4.90308434e-01 1.17263854e+00 -2.93281022e-02 -7.15746462e-01 3.57388437e-01 1.58712566e-01 -3.62737983e-01 4.68035430e-01 3.91413480e-01 -7.48762727e-01 3.03001772e-03 -1.10226107e+00 9.74409759e-01 -2.59263307e-01 -3.37943941e-01 -2.13814333e-01 8.37398589e-01 -6.25185907e-01 3.19072872e-01 -8.79292309e-01 -3.57796997e-01 -4.21113729e-01 4.25206512e-01 7.21632838e-02 8.10264409e-01 -1.41286170e+00 9.81164992e-01 3.88996273e-01 1.46225333e-01 -9.15123522e-01 -8.54925394e-01 -5.42617321e-01 -1.74724475e-01 6.40821874e-01 -2.90068477e-01 1.20532143e+00 -3.61515850e-01 -1.85982192e+00 -8.77405778e-02 7.62903869e-01 -4.27924454e-01 3.71912181e-01 -7.01117367e-02 -7.17804492e-01 3.43702942e-01 -5.23304522e-01 -4.15494561e-01 8.60343039e-01 -1.20611119e+00 -3.05198282e-01 -6.29343465e-02 4.56047654e-01 -1.81893602e-01 -3.55996788e-01 2.25763872e-01 9.17859674e-01 -4.11360115e-01 -9.57145393e-02 -1.34234929e+00 -5.27123213e-01 -1.77339539e-02 -7.81125724e-01 2.59333581e-01 1.56341362e+00 -2.51912504e-01 1.02058673e+00 -2.14845920e+00 2.46681765e-01 5.17082036e-01 -1.60232365e-01 1.00505483e+00 3.73224556e-01 1.11091352e+00 -8.61622319e-02 3.33236754e-02 -4.06594157e-01 3.73078018e-01 2.80812263e-01 3.48017327e-02 -9.70025480e-01 9.49852645e-01 4.50002223e-01 1.38211459e-01 -5.00307620e-01 -3.93515043e-02 4.30015445e-01 5.42655110e-01 -4.16766435e-01 1.66155338e-01 -9.22926962e-02 7.88708806e-01 -1.28075588e+00 5.30185163e-01 8.18494141e-01 3.53051275e-01 2.77596176e-01 -6.47894666e-02 -6.39131308e-01 -4.42873508e-01 -1.40101159e+00 3.03706914e-01 -4.00950491e-01 1.75892618e-02 8.42055261e-01 -1.01962245e+00 6.08710885e-01 7.67201900e-01 5.61604559e-01 -1.91011205e-01 7.28513300e-01 1.37330115e-01 2.44097471e-01 -2.24594563e-01 3.37962061e-01 -1.00175478e-01 -4.62867647e-01 4.55748260e-01 -4.92077976e-01 -5.71123898e-01 -3.13056707e-01 9.26400125e-02 1.25782609e+00 -2.91976571e-01 6.89786732e-01 -5.81716001e-01 1.17945611e+00 3.85247953e-02 6.55309975e-01 -3.00600957e-02 -5.25373936e-01 -4.10799623e-01 6.92488432e-01 6.07605912e-02 -1.03633726e+00 -8.97814333e-01 -2.92995691e-01 8.39505643e-02 5.11856616e-01 6.07000403e-02 -7.74812043e-01 -3.28754097e-01 1.56917095e-01 6.49900019e-01 -1.93835795e-01 -7.01182663e-01 -5.74640810e-01 -4.83821392e-01 7.47015357e-01 2.29492560e-01 2.87567377e-01 -3.43390048e-01 -1.11128974e+00 4.21441078e-01 5.49347699e-01 -1.28822362e+00 4.71983701e-02 -6.10071123e-02 -5.61144769e-01 -1.38260949e+00 -4.75286514e-01 -4.57428753e-01 7.37271249e-01 -6.60996735e-02 3.61446291e-02 -1.79608464e-01 -5.32949567e-01 1.09389067e+00 -9.35087129e-02 -6.58009350e-01 -7.84533381e-01 -4.91504222e-01 6.79731965e-01 2.57526487e-01 -4.96897370e-01 -2.93553203e-01 -5.40069997e-01 7.78089881e-01 -1.26968527e+00 -5.47739208e-01 1.73701599e-01 3.77515495e-01 2.05400750e-01 3.41030151e-01 8.09310973e-01 -3.53949010e-01 9.06987429e-01 -4.96070743e-01 -1.40917957e+00 9.68769845e-03 -2.05377668e-01 -2.62324452e-01 1.17474294e+00 -6.33794367e-01 -9.61132348e-01 5.31957373e-02 7.89879784e-02 -3.58368248e-01 -2.52645314e-01 2.45020390e-01 -3.36761981e-01 -6.13110781e-01 3.63106191e-01 6.55332580e-02 2.40173265e-01 -1.87435418e-01 1.33057786e-02 5.59165776e-01 1.50587007e-01 -7.86334336e-01 1.13966858e+00 3.78909618e-01 7.54578590e-01 -1.24452460e+00 4.39844243e-02 -1.22275807e-01 -2.32968673e-01 -3.11080724e-01 4.23160881e-01 -2.98557937e-01 -1.55757034e+00 6.30732119e-01 -1.05598378e+00 -5.76248355e-02 -1.24870121e-01 3.61200899e-01 -5.84592164e-01 4.57496762e-01 -7.28427768e-01 -1.62945175e+00 -1.70340598e-01 -1.11701632e+00 6.98068917e-01 1.02022514e-01 -2.68544495e-01 -1.06431103e+00 3.58098507e-01 -4.99345839e-01 5.01608014e-01 1.07335532e+00 7.10297823e-01 -5.81642449e-01 -5.52565575e-01 -6.00719213e-01 3.05211067e-01 3.06576580e-01 -6.09092116e-02 3.21009427e-01 -6.21514916e-01 -6.78322792e-01 6.02248430e-01 1.61287054e-01 -2.74779499e-01 1.50265038e-01 5.17545104e-01 -5.45563817e-01 -5.50074756e-01 -2.30987594e-02 1.71063840e+00 6.15260005e-01 1.87300310e-01 2.68677231e-02 3.84338170e-01 8.31405580e-01 8.63134444e-01 6.40978873e-01 -1.80767581e-01 4.42708403e-01 1.03268468e+00 3.41586471e-01 9.34348702e-01 2.76737750e-01 4.50691551e-01 4.01433319e-01 1.17392607e-01 -3.47272575e-01 -7.49418855e-01 2.94097662e-01 -1.50745666e+00 -2.41536066e-01 -4.07751381e-01 2.19969916e+00 2.70363867e-01 -5.38539439e-02 1.79430500e-01 7.31192112e-01 1.07958221e+00 -8.35748836e-02 -4.61756468e-01 -7.23509192e-01 1.28141955e-01 -1.75977424e-01 7.54087150e-01 3.99937272e-01 -1.08349752e+00 1.37662306e-01 5.45365667e+00 2.53976136e-01 -1.17198503e+00 -2.76092976e-01 -1.83793344e-02 5.01668870e-01 3.19497198e-01 2.57194996e-01 -5.87818086e-01 3.86304885e-01 9.99923468e-01 -5.16882181e-01 -2.82357484e-02 8.86779010e-01 4.68321025e-01 -1.61626652e-01 -5.56949019e-01 2.05032513e-01 -5.19301713e-01 -4.09957558e-01 -4.02219772e-01 2.77628124e-01 6.23700976e-01 -8.18540990e-01 4.85934228e-01 -1.16106428e-01 1.51959747e-01 -3.04951370e-01 3.82520795e-01 2.03552946e-01 1.24163277e-01 -9.53555822e-01 3.99156630e-01 5.88721097e-01 -1.22376227e+00 -4.92386758e-01 -3.95401455e-02 -2.26702660e-01 6.93803966e-01 2.75068104e-01 -3.83659780e-01 7.79412508e-01 -1.15413412e-01 1.85850099e-01 9.81320813e-02 1.06461155e+00 -1.14111766e-01 5.03984809e-01 -4.98774588e-01 -4.48276311e-01 4.34670597e-01 -1.04271583e-01 1.13350284e+00 6.35072827e-01 2.05071330e-01 7.01593161e-01 1.41106904e-01 7.74014831e-01 8.28567088e-01 -2.39655226e-01 -7.32230186e-01 -2.47129485e-01 2.39375308e-01 1.12681007e+00 -5.11935949e-01 -1.31347338e-02 -4.80458885e-02 3.35214078e-01 -6.33092403e-01 4.07442391e-01 -1.15389168e+00 -8.49000096e-01 1.19928169e+00 1.30404010e-01 -1.53098881e-01 -6.77917004e-01 -3.18840481e-02 -8.07408631e-01 -1.18036926e-01 -6.44773960e-01 2.63658911e-01 -2.05427751e-01 -1.25271761e+00 4.00226444e-01 2.99209148e-01 -1.89702022e+00 -3.86821419e-01 -5.70474148e-01 -8.76195431e-01 6.41301632e-01 -1.03270602e+00 -7.25725293e-01 2.81777948e-01 7.00375915e-01 -5.94797321e-02 3.64448130e-01 7.02684462e-01 -3.16651464e-02 -7.77247965e-01 -5.20664118e-02 2.01568559e-01 -2.76915491e-01 1.02772467e-01 -8.21928620e-01 2.65027553e-01 1.13813603e+00 -1.13815725e+00 7.74868846e-01 1.31723559e+00 -6.60355389e-01 -2.03809500e+00 -9.90139663e-01 2.14783609e-01 8.80138502e-02 1.25806963e+00 -1.05832078e-01 -9.04585302e-01 5.49735188e-01 3.01950604e-01 3.50462198e-02 6.62892032e-03 -9.71468389e-01 9.08757895e-02 1.75700039e-01 -1.57635713e+00 6.95670664e-01 3.71841550e-01 -2.66308188e-01 -4.16093022e-01 4.12543453e-02 7.22847342e-01 -3.11089367e-01 -9.35221851e-01 5.21767139e-01 2.04607368e-01 -1.17978081e-01 1.02097046e+00 -4.65857923e-01 -3.81073773e-01 -5.46041489e-01 1.70843527e-02 -1.44439793e+00 1.10898972e-01 -1.08686388e+00 1.71439294e-02 9.92542446e-01 -2.53434107e-02 -1.22468472e+00 1.27269268e-01 7.55050421e-01 3.72013194e-03 -4.46680427e-01 -9.98379827e-01 -1.29799390e+00 1.45626798e-01 -6.90009221e-02 3.35537314e-01 9.78667676e-01 7.13355303e-01 -2.36640975e-01 -3.28740209e-01 1.03371286e+00 7.46739089e-01 -2.83517778e-01 5.54396093e-01 -1.01025152e+00 -2.61602968e-01 -1.91313371e-01 -4.40577269e-01 -1.12223133e-01 5.73067041e-03 -7.17747211e-02 1.45026118e-01 -1.07809126e+00 -6.87824190e-01 -8.69678110e-02 -1.22647524e-01 1.54077724e-01 2.50566542e-01 -2.70703465e-01 2.60997623e-01 -9.32467952e-02 -1.41708106e-01 6.65656745e-01 9.04065251e-01 7.25422874e-02 -2.44375467e-01 4.14285839e-01 2.41486073e-01 6.99182451e-01 9.98017013e-01 -3.22154969e-01 -7.57973611e-01 3.00776154e-01 -1.81198880e-01 6.42284036e-01 8.17672551e-01 -1.20464063e+00 2.33602703e-01 -6.31781697e-01 -4.47845995e-01 -3.80499810e-01 4.46553797e-01 -1.31701005e+00 5.20299256e-01 1.24537015e+00 -6.41515404e-02 1.39818013e-01 3.32197845e-01 8.62411261e-01 -2.42260188e-01 -8.63143988e-03 1.07740915e+00 5.16342342e-01 -2.17709541e-01 1.08374931e-01 -1.04741323e+00 -3.83729517e-01 1.74672472e+00 2.61030451e-04 -4.90762591e-01 -9.31262076e-02 -8.28634083e-01 1.49571136e-01 2.31905431e-01 2.89521605e-01 4.50735837e-01 -8.38744462e-01 -1.72833167e-02 1.06459379e-01 -1.60726324e-01 -3.92474145e-01 4.76004452e-01 1.02674603e+00 -3.05302948e-01 7.81587005e-01 -3.68217796e-01 -2.57101089e-01 -1.20859718e+00 8.01303029e-01 3.03328812e-01 1.08683653e-01 -4.42639828e-01 8.62766355e-02 2.24171877e-01 -2.33219102e-01 3.67400795e-02 -5.33401906e-01 -1.17603214e-02 -5.02180219e-01 2.54988521e-01 6.10229909e-01 -2.82823443e-01 -5.74757576e-01 -5.74228287e-01 5.08414388e-01 4.08708006e-01 -2.33852714e-01 9.63472128e-01 -1.07423909e-01 5.91480397e-02 1.24225721e-01 8.85993302e-01 -3.61409515e-01 -1.35478878e+00 1.79051712e-01 -2.45673936e-02 -3.37577164e-01 -2.41711348e-01 -3.81336480e-01 -7.75043964e-01 5.38004637e-01 1.90302759e-01 1.04235804e+00 9.31110144e-01 -6.16121173e-01 7.48953104e-01 2.46045664e-01 6.69482589e-01 -9.02302921e-01 7.90732075e-03 6.21736526e-01 8.18485975e-01 -2.22225353e-01 -1.47533730e-01 -1.03922725e+00 -4.90106314e-01 9.02589083e-01 5.36494434e-01 -1.04602432e+00 1.11600888e+00 6.46024466e-01 -4.45831299e-01 2.32325241e-01 -5.72076857e-01 3.96929234e-01 -1.24092922e-01 6.99341834e-01 -2.00893775e-01 1.01741359e-01 -6.58250153e-01 6.82973564e-01 5.22678792e-01 -3.33598673e-01 9.86784160e-01 1.52436996e+00 -2.68669277e-01 -1.00761163e+00 -8.11947525e-01 -4.13883686e-01 -4.58877325e-01 6.73572242e-01 -4.25797224e-01 1.05584335e+00 -6.31220281e-01 9.27296340e-01 -3.87646884e-01 -2.95547843e-01 7.58997202e-01 -2.30002344e-01 1.35807525e-02 -1.87433586e-01 -6.86967373e-01 -2.58486085e-02 -5.78837134e-02 -4.08822268e-01 -1.52024701e-01 -5.82620025e-01 -1.22685778e+00 -2.25111470e-01 -3.60413879e-01 2.60230541e-01 9.89867330e-01 7.68494427e-01 3.86418968e-01 7.33404875e-01 8.83803606e-01 -5.60692966e-01 -1.34591877e+00 -4.58538353e-01 -9.35919642e-01 -3.77032906e-01 3.63696128e-01 -7.89530098e-01 -6.77609146e-01 -4.85368490e-01]
[5.327722072601318, 2.6192643642425537]
3a0a7409-509b-43b2-9d4f-02e486f35ac1
contrastive-learning-with-stronger-1
2104.07713
null
https://arxiv.org/abs/2104.07713v2
https://arxiv.org/pdf/2104.07713v2.pdf
Contrastive Learning with Stronger Augmentations
Representation learning has significantly been developed with the advance of contrastive learning methods. Most of those methods have benefited from various data augmentations that are carefully designated to maintain their identities so that the images transformed from the same instance can still be retrieved. However, those carefully designed transformations limited us to further explore the novel patterns exposed by other transformations. Meanwhile, as found in our experiments, the strong augmentations distorted the images' structures, resulting in difficult retrieval. Thus, we propose a general framework called Contrastive Learning with Stronger Augmentations~(CLSA) to complement current contrastive learning approaches. Here, the distribution divergence between the weakly and strongly augmented images over the representation bank is adopted to supervise the retrieval of strongly augmented queries from a pool of instances. Experiments on the ImageNet dataset and downstream datasets showed the information from the strongly augmented images can significantly boost the performance. For example, CLSA achieves top-1 accuracy of 76.2% on ImageNet with a standard ResNet-50 architecture with a single-layer classifier fine-tuned, which is almost the same level as 76.5% of supervised results. The code and pre-trained models are available in https://github.com/maple-research-lab/CLSA.
['Guo-Jun Qi', 'Xiao Wang']
2021-04-15
contrastive-learning-with-stronger
https://openreview.net/forum?id=KJSC_AsN14
https://openreview.net/pdf?id=KJSC_AsN14
null
['self-supervised-image-classification']
['computer-vision']
[ 1.58004135e-01 -1.73976898e-01 -2.67713398e-01 -4.34626609e-01 -7.98848808e-01 -4.58171248e-01 9.25189853e-01 -2.94067740e-01 -6.03831351e-01 5.21028996e-01 1.72454566e-01 -4.32752930e-02 -2.26663742e-02 -6.83147252e-01 -8.38894129e-01 -8.65656912e-01 2.05569431e-01 3.69881183e-01 1.53942496e-01 -5.32459438e-01 4.05080542e-02 4.37382787e-01 -1.70855761e+00 5.03267884e-01 6.71903253e-01 1.02328324e+00 3.61326069e-01 3.19755644e-01 -8.39412883e-02 6.70864105e-01 -5.99560499e-01 -4.03126150e-01 5.68036437e-01 -1.82878122e-01 -7.85279691e-01 -2.38976404e-01 7.04062343e-01 -5.19609511e-01 -7.94311047e-01 1.10152256e+00 5.34063518e-01 1.77003011e-01 6.05096519e-01 -1.21511316e+00 -1.21584356e+00 4.54565585e-01 -7.16244102e-01 4.90131229e-01 9.08555742e-03 1.62627891e-01 1.05232322e+00 -1.40577161e+00 5.19194365e-01 1.31902826e+00 1.94402829e-01 7.10476637e-01 -1.10071146e+00 -1.13024211e+00 3.69098514e-01 3.94403130e-01 -1.56155169e+00 -4.93296653e-01 6.79475367e-01 -1.17338248e-01 7.52201438e-01 1.73050165e-01 4.59234595e-01 1.11956620e+00 -2.58063108e-01 1.02637553e+00 1.10206926e+00 -4.92605448e-01 -2.99841106e-01 3.61168146e-01 1.23755090e-01 5.23065746e-01 8.83414075e-02 2.06967324e-01 -4.34373885e-01 3.13549191e-02 6.41773224e-01 2.57136256e-01 -3.90926123e-01 -3.69858444e-01 -1.08017683e+00 7.17935920e-01 1.00774539e+00 4.39866573e-01 -3.19405794e-01 3.86417732e-02 1.70334786e-01 4.23160672e-01 7.08637834e-01 4.76589650e-01 -5.80827594e-01 4.71186429e-01 -7.45083809e-01 1.75914317e-01 2.10997179e-01 1.08561671e+00 9.40629601e-01 1.02330148e-01 -2.50428110e-01 9.60514963e-01 2.82769322e-01 5.84997356e-01 7.16693223e-01 -5.75513244e-01 4.55364078e-01 7.06059456e-01 -2.03668684e-01 -8.14779818e-01 5.56609780e-02 -8.77921343e-01 -9.52008903e-01 7.30765285e-03 2.79559731e-01 2.31685072e-01 -1.29087460e+00 1.98751521e+00 -4.46959361e-02 2.90573090e-01 2.70834751e-02 8.59841347e-01 9.27421927e-01 8.32022130e-01 1.33769125e-01 1.32277519e-01 1.31602466e+00 -1.08095145e+00 -6.02322578e-01 -3.18317682e-01 5.69912136e-01 -8.62700343e-01 1.45108187e+00 1.32359028e-01 -1.10039246e+00 -8.27064693e-01 -1.14910150e+00 -1.36194661e-01 -5.71019411e-01 5.24924882e-02 5.05757511e-01 2.66696244e-01 -1.27753186e+00 1.86429843e-01 -5.08841455e-01 -2.10385397e-01 7.23820686e-01 3.48180085e-01 -4.62573409e-01 -4.56293315e-01 -1.34474730e+00 9.44215059e-01 3.54705572e-01 4.18194234e-02 -9.95368779e-01 -7.90614188e-01 -5.28999329e-01 1.68407902e-01 2.34871656e-01 -4.91454363e-01 1.24112344e+00 -1.03886616e+00 -9.53461885e-01 1.15243840e+00 1.20979156e-02 -2.89278865e-01 3.29435706e-01 -5.21613777e-01 -3.90681505e-01 1.77812517e-01 2.13067204e-01 1.17005408e+00 8.70734513e-01 -1.41314578e+00 -5.12249231e-01 -3.90883982e-01 5.25881387e-02 3.78219455e-01 -7.17529476e-01 -4.61793281e-02 -8.22304070e-01 -7.91101158e-01 4.66428660e-02 -1.17473245e+00 -1.32798791e-01 4.34669107e-02 -7.74637461e-02 -2.35509172e-01 7.07611918e-01 -4.48260576e-01 9.69329476e-01 -2.40530610e+00 -9.61877964e-03 1.71030656e-01 1.57690123e-01 6.95464015e-01 -6.81698143e-01 2.43446797e-01 -3.79800379e-01 8.67381915e-02 -4.04616110e-02 -2.09610403e-01 -1.33985654e-01 1.23906337e-01 -7.53393590e-01 2.28768051e-01 3.65819067e-01 1.01838839e+00 -8.32607806e-01 -2.43499249e-01 4.94487286e-02 5.11083305e-01 -5.47894359e-01 4.45224494e-01 -8.64830315e-02 4.30665612e-01 -3.57965291e-01 4.84640419e-01 8.54577839e-01 -4.79656547e-01 -7.43732005e-02 -3.52403343e-01 3.16349417e-01 7.54732862e-02 -7.32958734e-01 1.72244835e+00 -2.79823065e-01 4.51400280e-01 -1.97456002e-01 -9.41006899e-01 8.82774413e-01 6.76705986e-02 2.82493532e-01 -1.18982434e+00 -4.78873886e-02 2.00489819e-01 1.20487101e-01 -3.16578001e-01 4.86459732e-01 1.52537882e-01 1.97034866e-01 2.96715587e-01 2.28466555e-01 6.31829947e-02 4.07744497e-02 4.36137170e-01 8.00055444e-01 4.82890755e-02 3.78132723e-02 -1.68259755e-01 4.60641861e-01 -9.33517814e-02 4.31584805e-01 9.42032516e-01 -1.39021501e-01 7.26940632e-01 -3.56917121e-02 -2.88043648e-01 -9.09170508e-01 -1.18874645e+00 -2.93941468e-01 1.65559852e+00 2.69812971e-01 -2.87968963e-01 -5.02302825e-01 -6.72146440e-01 -8.50041434e-02 5.00163436e-01 -6.36921048e-01 -5.97712100e-01 -6.40203297e-01 -8.17867637e-01 6.26825511e-01 4.58301216e-01 9.72356141e-01 -1.13766885e+00 1.03672385e-01 -2.13033170e-01 -1.43475175e-01 -9.28203344e-01 -3.34684730e-01 1.01306669e-01 -7.07204223e-01 -9.91124809e-01 -9.15703058e-01 -1.08011317e+00 9.12243843e-01 6.32745862e-01 1.12646914e+00 3.12688947e-01 -1.66658804e-01 3.83708417e-01 -5.00434220e-01 -3.90476555e-01 -1.67448416e-01 2.44550154e-01 7.53154512e-03 -4.40593474e-02 4.92876023e-01 -5.04372239e-01 -8.28422368e-01 3.89703393e-01 -1.23347604e+00 -1.06131449e-01 8.50007057e-01 1.06168365e+00 5.97879529e-01 -2.03378811e-01 7.57782817e-01 -8.38924766e-01 5.16825199e-01 -6.35160983e-01 -4.37110603e-01 3.53265345e-01 -7.86602199e-01 6.06001467e-02 4.28037643e-01 -6.42690480e-01 -1.23076820e+00 -1.85391262e-01 -9.50814337e-02 -6.20126188e-01 -4.82972935e-02 4.65777129e-01 -1.30207852e-01 4.42121178e-03 8.51185024e-01 4.41583395e-01 1.18727516e-02 -5.10466218e-01 5.65055311e-01 5.69488287e-01 5.07242262e-01 -5.40800929e-01 1.10660505e+00 3.90155524e-01 -4.35670137e-01 -5.05302846e-01 -1.15426576e+00 -4.40293401e-01 -4.41082358e-01 -3.97920348e-02 4.16666299e-01 -1.18889260e+00 -2.42979273e-01 5.50555944e-01 -9.16715980e-01 -2.74532408e-01 -3.82413328e-01 5.22912025e-01 -1.47582933e-01 1.03115834e-01 -6.48732126e-01 -3.11814904e-01 -3.15856844e-01 -1.21190214e+00 8.21520209e-01 1.81150451e-01 1.67848319e-02 -6.69900894e-01 6.08717278e-02 4.70432252e-01 7.35330403e-01 -4.23484653e-01 1.06019795e+00 -9.70595121e-01 -7.52442896e-01 -2.49254212e-01 -4.04268593e-01 6.23366833e-01 1.74955174e-01 -1.52060777e-01 -1.35574603e+00 -6.41060174e-01 -7.24908486e-02 -6.45734191e-01 1.15621674e+00 1.14945538e-01 1.38817787e+00 -2.20751330e-01 -2.50181675e-01 6.73305750e-01 1.12674177e+00 1.79180592e-01 9.13750470e-01 5.40634274e-01 6.17367625e-01 5.09574592e-01 7.37436116e-01 3.98769751e-02 7.24641979e-02 5.97968817e-01 5.03988564e-01 -3.82646829e-01 -4.13180500e-01 -3.51837814e-01 2.60141015e-01 8.39071214e-01 9.11438465e-03 -3.04227561e-01 -7.82037079e-01 5.33513725e-01 -1.70722091e+00 -7.76658475e-01 2.72134632e-01 2.01012993e+00 1.02184224e+00 4.51476537e-02 -3.62480640e-01 -1.47083461e-01 8.16890001e-01 3.57279152e-01 -6.38757288e-01 2.18880951e-01 -2.94467896e-01 4.28073913e-01 3.94296885e-01 2.76706487e-01 -1.06972647e+00 1.19015789e+00 6.14087486e+00 1.04977608e+00 -1.12655449e+00 -3.02685965e-02 7.21375406e-01 -2.90266991e-01 -3.66795003e-01 -1.40145943e-01 -8.97584558e-01 4.11959320e-01 6.97363198e-01 -2.45693773e-01 3.19899946e-01 7.78097391e-01 -3.35885018e-01 2.59168208e-01 -1.04491496e+00 1.05521965e+00 7.39197731e-02 -1.15471041e+00 6.44713879e-01 4.94677387e-02 9.21985626e-01 4.23305571e-01 7.79154241e-01 8.00262690e-01 2.36211672e-01 -1.02218568e+00 4.17260617e-01 4.57327366e-01 8.32773864e-01 -5.46023667e-01 7.30860829e-01 2.47335121e-01 -8.30237389e-01 -1.33467643e-02 -6.07955098e-01 1.91025913e-01 -3.30624074e-01 3.40102345e-01 -8.93426299e-01 3.07773858e-01 8.91034603e-01 6.48887753e-01 -9.03770566e-01 8.90441358e-01 -2.53358245e-01 6.43375516e-01 -1.10427991e-01 3.18260789e-01 2.72189319e-01 -7.97579736e-02 2.64926106e-01 1.05094922e+00 2.01329619e-01 1.47313112e-02 -5.96589074e-02 7.04248846e-01 -6.51302338e-01 1.32332250e-01 -7.13160574e-01 1.84315756e-01 5.75750053e-01 1.31462514e+00 -2.79171228e-01 -4.28595811e-01 -5.08820534e-01 7.50245035e-01 4.99235064e-01 6.90669000e-01 -7.09039032e-01 -1.49261251e-01 5.89335382e-01 9.54387859e-02 1.76900670e-01 1.96470127e-01 2.63234377e-01 -1.22491646e+00 7.52094015e-02 -1.07200313e+00 5.24266899e-01 -8.31624389e-01 -1.58037317e+00 8.33033502e-01 1.69031546e-01 -1.33267617e+00 -1.80792809e-01 -5.36897540e-01 -3.73168886e-01 9.16936994e-01 -1.81087482e+00 -1.14751172e+00 -4.24248934e-01 8.67792249e-01 4.89485800e-01 -5.58070362e-01 8.63402665e-01 6.23893082e-01 -4.75194275e-01 9.56065655e-01 1.44289300e-01 3.14838231e-01 1.19193637e+00 -9.33836818e-01 2.30995417e-01 7.35753775e-01 2.92025149e-01 8.71888459e-01 3.49298388e-01 -4.19640124e-01 -1.09830260e+00 -1.16120756e+00 5.21749139e-01 -3.11443120e-01 5.41587353e-01 -3.01868528e-01 -1.44018674e+00 9.10635829e-01 4.78831291e-01 3.63592148e-01 6.74534559e-01 6.20567426e-02 -7.47045100e-01 -4.14712667e-01 -9.70601618e-01 6.64315045e-01 1.07562399e+00 -7.44757891e-01 -7.06973135e-01 4.37812716e-01 6.98911309e-01 -1.91215128e-01 -6.16201878e-01 7.19377518e-01 3.40201795e-01 -6.79980636e-01 1.10387039e+00 -8.31468761e-01 3.93731415e-01 -2.24896148e-01 -2.73625404e-01 -1.22426260e+00 -5.79748452e-01 -7.71781132e-02 -4.82901111e-02 1.23325646e+00 3.92416745e-01 -7.74175107e-01 7.16480792e-01 4.84218389e-01 -1.50308713e-01 -5.67260742e-01 -5.79683244e-01 -8.33485425e-01 2.37701759e-01 -2.58383229e-02 5.20412207e-01 1.14959621e+00 -5.02744794e-01 4.86241788e-01 -2.67546475e-01 9.88461226e-02 4.91783679e-01 -9.11096483e-02 7.79332697e-01 -1.00386381e+00 -1.13025650e-01 -3.27813089e-01 -2.08736449e-01 -1.34707427e+00 3.51555109e-01 -1.27021086e+00 -9.25475284e-02 -1.16021669e+00 4.86893713e-01 -6.21241271e-01 -9.02238846e-01 6.15478754e-01 -4.45604473e-01 5.86427689e-01 1.87060058e-01 6.60793781e-01 -5.99498332e-01 8.61699283e-01 1.44071770e+00 -3.70083928e-01 1.27510782e-02 -1.03506781e-01 -9.92940724e-01 6.02958024e-01 1.01005375e+00 -3.97925973e-01 -8.33066881e-01 -6.76514149e-01 1.32388934e-01 -4.53236163e-01 3.50714862e-01 -7.24230409e-01 2.75645226e-01 2.32032418e-01 6.71255410e-01 -7.11789489e-01 5.12197912e-01 -8.13746750e-01 -5.11280447e-02 3.14787984e-01 -7.38786638e-01 1.07136250e-01 3.52574885e-01 5.39055407e-01 -5.65559328e-01 -2.24986285e-01 7.97352493e-01 -2.08130121e-01 -8.54409397e-01 5.57931483e-01 -7.46337846e-02 4.96146753e-02 8.88591826e-01 2.11102515e-02 -6.87722981e-01 -4.18919116e-01 -6.93793118e-01 1.87307164e-01 2.26626992e-01 5.22506237e-01 7.69418955e-01 -1.44176602e+00 -8.37131679e-01 3.40945691e-01 4.49801207e-01 1.00922830e-01 2.85203218e-01 4.83414829e-01 -2.23339394e-01 3.69028151e-01 -3.70955229e-01 -5.71497560e-01 -1.23831975e+00 5.63099802e-01 1.72320381e-01 -3.44613254e-01 -3.81459236e-01 9.62377250e-01 8.05540502e-01 -4.47031051e-01 2.44172260e-01 9.13504437e-02 -2.46320784e-01 9.46665332e-02 6.27021968e-01 3.37152109e-02 1.05725132e-01 -6.20448887e-01 -2.98644125e-01 3.51823032e-01 -9.31798637e-01 3.84389199e-02 1.48519897e+00 -5.38088083e-02 -7.13869259e-02 8.67663845e-02 1.41708946e+00 -1.45138472e-01 -1.20212460e+00 -8.23923528e-01 -1.76391140e-01 -5.08122027e-01 5.96978180e-02 -8.67801011e-01 -1.44472003e+00 8.04423034e-01 8.44735563e-01 -9.90640521e-02 1.20185173e+00 1.42420620e-01 5.07723629e-01 6.63740933e-01 2.54085600e-01 -7.44751632e-01 4.22177404e-01 4.63474333e-01 1.13933969e+00 -1.27978325e+00 -1.23815062e-02 -2.47820064e-01 -4.81806159e-01 6.55560970e-01 1.00162387e+00 -2.73796320e-01 6.15522087e-01 -1.22419886e-01 1.46338388e-01 -2.98724145e-01 -8.10263097e-01 -2.69398361e-01 4.75187004e-01 4.91521508e-01 3.92063290e-01 -2.30632946e-01 1.86933130e-01 3.02440792e-01 -6.88969046e-02 -2.12529913e-01 1.67828560e-01 9.24215078e-01 -2.24022746e-01 -1.04669046e+00 -2.17637450e-01 5.35223424e-01 -4.32850897e-01 -4.04862344e-01 -3.69468629e-01 1.03806186e+00 -1.79478079e-01 6.87979639e-01 1.22858815e-01 -3.75761747e-01 4.14751261e-01 8.33257511e-02 5.56547761e-01 -7.26559818e-01 -4.89008635e-01 -8.13941658e-03 -3.03245723e-01 -4.16882604e-01 -4.66071844e-01 -2.44068936e-01 -1.01886272e+00 -1.79678544e-01 -3.46462756e-01 1.08588442e-01 3.11825901e-01 5.85790515e-01 3.93111944e-01 4.46580350e-01 7.63516188e-01 -7.04273641e-01 -6.83471918e-01 -1.18220258e+00 -4.77314204e-01 6.97817326e-01 2.30882972e-01 -6.30177438e-01 -4.96265739e-01 -7.31871724e-02]
[9.864394187927246, 2.1261606216430664]
38820c74-f30c-436a-a569-23a227ad8e33
red-psm-regularization-by-denoising-of
2304.03483
null
https://arxiv.org/abs/2304.03483v1
https://arxiv.org/pdf/2304.03483v1.pdf
RED-PSM: Regularization by Denoising of Partially Separable Models for Dynamic Imaging
Dynamic imaging addresses the recovery of a time-varying 2D or 3D object at each time instant using its undersampled measurements. In particular, in the case of dynamic tomography, only a single projection at a single view angle may be available at a time, making the problem severely ill-posed. In this work, we propose an approach, RED-PSM, which combines for the first time two powerful techniques to address this challenging imaging problem. The first, are partially separable models, which have been used to efficiently introduce a low-rank prior for the spatio-temporal object. The second is the recent Regularization by Denoising (RED), which provides a flexible framework to exploit the impressive performance of state-of-the-art image denoising algorithms, for various inverse problems. We propose a partially separable objective with RED and an optimization scheme with variable splitting and ADMM, and prove convergence of our objective to a value corresponding to a stationary point satisfying the first order optimality conditions. Convergence is accelerated by a particular projection-domain-based initialization. We demonstrate the performance and computational improvements of our proposed RED-PSM with a learned image denoiser by comparing it to a recent deep-prior-based method TD-DIP.
['Yoram Bresler', 'Marc L. Klasky', 'Berk Iskender']
2023-04-07
null
null
null
null
['video-reconstruction']
['computer-vision']
[ 4.45618510e-01 -1.92486286e-01 5.31339943e-01 -1.57017112e-01 -9.82157052e-01 -1.90399960e-01 4.76721346e-01 -3.78267735e-01 -5.38217306e-01 6.99256897e-01 1.26468599e-01 1.31134048e-01 -6.38672709e-01 -4.79602724e-01 -4.78212357e-01 -1.45451379e+00 1.84068620e-01 6.97076499e-01 7.88365025e-03 -9.31717977e-02 -9.15485027e-04 5.84015012e-01 -1.15743661e+00 -3.00680231e-02 9.97546911e-01 8.86443853e-01 5.50218761e-01 3.88213843e-01 2.02217743e-01 5.36628902e-01 -4.32857163e-02 -6.48519173e-02 4.59090322e-01 -4.56444204e-01 -5.17116427e-01 4.75859791e-01 2.50871062e-01 -4.20243472e-01 -5.56216955e-01 1.07534277e+00 5.51480412e-01 4.02910531e-01 4.42022264e-01 -7.91542888e-01 -2.47437879e-01 8.22625011e-02 -7.85308242e-01 1.84254676e-01 1.60421088e-01 1.19541325e-01 5.83320737e-01 -1.03372550e+00 8.81466806e-01 8.98943305e-01 5.33663869e-01 3.61728728e-01 -1.56591463e+00 -1.86298862e-01 -8.18900317e-02 2.90250093e-01 -1.24467897e+00 -5.21069467e-01 1.03428328e+00 -6.15457356e-01 4.12484497e-01 2.28042722e-01 5.82171857e-01 8.85480881e-01 1.20617576e-01 3.82824570e-01 1.52473962e+00 -4.44059372e-01 2.51422524e-01 -2.98982322e-01 8.87004286e-02 5.87690771e-01 8.62895399e-02 -4.55530621e-02 -5.48847020e-01 -2.06810966e-01 8.97420406e-01 2.53335536e-01 -7.94136703e-01 -5.13191760e-01 -1.30575967e+00 6.66050315e-01 2.69008040e-01 4.31392789e-01 -7.87338436e-01 -1.92796588e-01 7.22381100e-02 2.18357340e-01 7.79956937e-01 2.58135706e-01 -1.23956792e-01 1.17342122e-01 -1.11154723e+00 1.46497190e-01 5.96922398e-01 4.33699101e-01 6.73391461e-01 1.76286235e-01 -1.64030492e-01 7.65847623e-01 3.09749156e-01 6.65190816e-01 1.66064471e-01 -1.26765370e+00 2.68155754e-01 -1.81970410e-02 3.19077402e-01 -8.45242143e-01 -3.62868220e-01 -7.09564269e-01 -1.24948347e+00 3.78184050e-01 5.31855464e-01 6.87039569e-02 -7.54000247e-01 1.88568926e+00 6.50464714e-01 5.01505077e-01 4.19943705e-02 1.12243009e+00 4.54881012e-01 7.52735913e-01 -4.61350083e-01 -8.98841143e-01 1.14012098e+00 -7.13239431e-01 -8.95574808e-01 -1.04961611e-01 -8.75645969e-03 -8.07628930e-01 5.96072018e-01 7.69478559e-01 -1.22416461e+00 -2.35544026e-01 -9.34081495e-01 1.43175796e-02 4.25132990e-01 -1.17757797e-01 2.29878768e-01 2.92649120e-01 -9.40673828e-01 8.15521240e-01 -1.05383480e+00 -1.23415552e-01 1.55347943e-01 1.45986632e-01 -5.51475883e-01 -5.78040719e-01 -6.78859532e-01 8.25226665e-01 -9.80818495e-02 4.67424989e-01 -1.06358027e+00 -8.11463892e-01 -4.95110065e-01 -1.70016959e-01 5.92093050e-01 -9.78794694e-01 7.08378375e-01 -5.49326718e-01 -1.69334948e+00 8.54578733e-01 -4.01476771e-01 -2.72174448e-01 8.59211266e-01 -2.15536028e-01 -1.35661528e-01 6.73592746e-01 1.55570731e-01 -1.31162271e-01 1.22539198e+00 -1.49284267e+00 8.28947723e-02 -6.22362554e-01 -1.17765114e-01 2.03323692e-01 -1.04865633e-01 -1.78663030e-01 -5.09755552e-01 -6.30446255e-01 6.70076311e-01 -1.02757275e+00 -5.84149122e-01 6.71080276e-02 -2.99068809e-01 5.69340348e-01 6.68678999e-01 -1.10381281e+00 7.45431781e-01 -2.08505964e+00 9.88472700e-01 7.85200819e-02 3.99629951e-01 1.54474691e-01 5.59140854e-02 4.11205590e-01 -2.06003934e-01 -5.05070031e-01 -7.64409602e-01 -7.38131344e-01 -3.45451564e-01 5.17095864e-01 -1.99529469e-01 9.57539320e-01 -3.00437033e-01 4.43530589e-01 -8.26241374e-01 -2.33024597e-01 3.92430484e-01 8.35620642e-01 -5.59017479e-01 3.44064087e-01 1.12082511e-01 1.38945305e+00 -4.39410180e-01 1.72048196e-01 1.16004848e+00 -2.41133407e-01 1.90927356e-01 -4.41830307e-01 -4.98020679e-01 -1.21478818e-01 -1.43294358e+00 2.01650906e+00 -6.50689781e-01 2.95187116e-01 7.98204958e-01 -1.34884512e+00 4.48082209e-01 4.43044871e-01 1.02385342e+00 -5.83356082e-01 -1.50729507e-01 4.51220900e-01 -2.50663906e-01 -6.35173202e-01 1.36639655e-01 -7.04546332e-01 4.18744683e-01 2.71950960e-01 -5.30997990e-03 -2.24582955e-01 9.98961553e-02 1.49660781e-01 1.33434844e+00 5.54597564e-02 8.97128806e-02 -3.46926600e-01 7.16460466e-01 -2.65559733e-01 7.42029607e-01 8.25355589e-01 1.43445820e-01 1.12492895e+00 2.54446000e-01 -1.70904949e-01 -9.90037799e-01 -9.72211421e-01 -4.37470824e-01 3.18131089e-01 -2.72858497e-02 1.04651973e-02 -6.17185652e-01 -2.09309667e-01 -2.93163240e-01 5.27802825e-01 -4.55923289e-01 1.31762296e-01 -9.05965745e-01 -1.06163073e+00 -2.68947899e-01 -1.57904267e-01 6.06734872e-01 -5.22099912e-01 -5.55330396e-01 3.81353557e-01 -5.55625916e-01 -1.29913056e+00 -1.51631892e-01 2.53924370e-01 -1.00338864e+00 -9.17123020e-01 -1.09528887e+00 -3.66448551e-01 7.33225226e-01 4.98947322e-01 8.77473414e-01 -1.86893687e-01 1.87709183e-02 6.10116899e-01 -2.18521908e-01 4.15297925e-01 -2.48862565e-01 -5.21766543e-01 8.92285779e-02 6.29963934e-01 -5.83528459e-01 -1.15843594e+00 -7.65591919e-01 3.05655569e-01 -1.16134143e+00 2.22401708e-01 5.21433115e-01 1.14152324e+00 7.88631499e-01 1.78622425e-01 2.62025706e-02 -6.97362840e-01 3.82389016e-02 -4.59529519e-01 -6.78875685e-01 8.47916380e-02 -3.41612071e-01 1.58828154e-01 6.23352826e-01 -3.41514409e-01 -1.27430010e+00 1.68428913e-01 -3.28790575e-01 -6.99830949e-01 2.61339426e-01 5.95889986e-01 -2.19572559e-01 -4.79715526e-01 2.46151283e-01 5.72427928e-01 1.24771774e-01 -8.39501560e-01 1.97085410e-01 1.19450025e-01 6.53755903e-01 -6.39482319e-01 1.07733119e+00 1.15185761e+00 5.01588225e-01 -1.17417443e+00 -1.01173818e+00 -7.48928607e-01 -7.41801083e-01 -3.37553769e-01 8.24230194e-01 -7.88798809e-01 -5.56896687e-01 7.66227961e-01 -1.08226061e+00 -3.16276342e-01 -4.78688031e-01 7.53831267e-01 -6.96168542e-01 8.79542291e-01 -6.74503088e-01 -6.56435311e-01 -1.87769771e-01 -1.23464572e+00 1.01209211e+00 -2.13698864e-01 4.55480158e-01 -9.11359012e-01 2.03314751e-01 5.04352570e-01 4.15746450e-01 4.61478472e-01 6.81567252e-01 3.62548009e-02 -7.68963099e-01 2.08244130e-01 3.84193729e-03 5.00741959e-01 -2.76043355e-01 -4.65175807e-01 -7.45223761e-01 -4.86054808e-01 1.08483636e+00 6.26587719e-02 9.46637452e-01 7.80002713e-01 8.49649012e-01 -1.96454480e-01 -1.58898935e-01 1.12365556e+00 1.76086295e+00 -1.45004177e-02 6.12102449e-01 1.17065199e-01 6.08397305e-01 4.47825015e-01 4.53720987e-01 6.88028097e-01 4.56627123e-02 1.07843566e+00 4.98306245e-01 -4.99945953e-02 -1.36737436e-01 2.89995819e-01 1.83094852e-02 1.17712307e+00 -4.83282328e-01 -1.44216329e-01 -8.01473439e-01 4.59458262e-01 -1.88917267e+00 -8.54891121e-01 -5.29080272e-01 2.50555134e+00 5.63247681e-01 -3.06953520e-01 -3.99715841e-01 2.91983455e-01 4.10111010e-01 3.37502450e-01 -4.93434399e-01 4.56751794e-01 -1.87504247e-01 3.29351544e-01 3.92086089e-01 7.46250451e-01 -8.70749176e-01 2.53152728e-01 5.32639647e+00 7.02830255e-01 -1.08779013e+00 7.32127130e-01 2.15450510e-01 -4.44055982e-02 -2.97088414e-01 1.00370221e-01 -3.18472147e-01 3.28925103e-01 6.37671590e-01 -8.59282911e-02 5.73175907e-01 2.94769138e-01 4.91220832e-01 -2.05528155e-01 -8.16305101e-01 1.15784383e+00 6.80499375e-02 -1.07338345e+00 -2.29558572e-01 2.44676679e-01 7.91831911e-01 -2.50851195e-02 1.25216872e-01 -2.13280827e-01 -5.53029850e-02 -5.52970648e-01 5.99144697e-01 8.01691771e-01 5.93788385e-01 -4.00850028e-01 3.65042835e-01 3.97216886e-01 -8.16212952e-01 -5.38732717e-03 -2.30098262e-01 1.23546913e-01 7.68988371e-01 1.36407697e+00 -3.44196446e-02 1.01533258e+00 5.58474481e-01 9.14047480e-01 1.14600532e-01 1.03844047e+00 -2.92761505e-01 4.69492078e-01 -4.90260959e-01 8.60359073e-01 1.81065843e-01 -9.65985119e-01 1.23855734e+00 7.98891127e-01 6.97582364e-01 5.98447084e-01 1.64071515e-01 7.42410004e-01 1.98959351e-01 -3.06659728e-01 -3.70449007e-01 5.30734420e-01 -1.32718638e-01 1.37832153e+00 -5.37929416e-01 -9.47971418e-02 -3.72222781e-01 1.19683909e+00 -5.55645339e-02 7.07139075e-01 -6.70126736e-01 3.54627281e-01 4.74754035e-01 2.44562089e-01 3.81846756e-01 -6.78373754e-01 -4.54283804e-02 -1.61365068e+00 3.62775832e-01 -7.27603555e-01 2.71635622e-01 -7.72949636e-01 -1.16745770e+00 4.96511042e-01 -5.37613519e-02 -1.27447283e+00 -8.86924118e-02 -4.02476519e-01 -3.64868581e-01 8.49197507e-01 -1.67525244e+00 -1.05991220e+00 -4.04641509e-01 8.16688836e-01 3.13211083e-01 2.21986577e-01 5.72763979e-01 6.65674031e-01 -4.38317209e-01 -1.86972827e-01 7.62208879e-01 -2.82076806e-01 6.17879450e-01 -1.17199826e+00 -2.46781752e-01 1.12833261e+00 -1.02967799e-01 5.46585560e-01 1.07400370e+00 -3.75710934e-01 -1.84789324e+00 -5.92318356e-01 3.63321334e-01 -7.87317380e-02 6.40104651e-01 -2.06197247e-01 -1.10645223e+00 7.21840501e-01 7.75446817e-02 3.58248115e-01 2.77738065e-01 -2.30513752e-01 6.19847476e-02 -2.79151797e-01 -1.22018504e+00 2.63680398e-01 9.58839238e-01 -3.62058610e-01 -3.78579795e-01 6.64945781e-01 3.82473528e-01 -6.33684874e-01 -8.82738292e-01 3.78736764e-01 2.75995851e-01 -1.20241201e+00 1.34651828e+00 -6.83404356e-02 4.54775661e-01 -4.23249274e-01 -2.07152054e-01 -1.44227219e+00 -3.31438303e-01 -9.82530177e-01 -2.25768238e-01 1.07802701e+00 -1.34312645e-01 -8.27496171e-01 5.31752408e-01 3.95838916e-01 -4.18519586e-01 -5.13949335e-01 -1.43585038e+00 -7.76064694e-01 -1.48452386e-01 -4.07822281e-01 -2.03837276e-01 9.45206106e-01 -6.13606989e-01 4.98293601e-02 -7.73508132e-01 5.73142529e-01 1.25410759e+00 1.76216200e-01 5.12901366e-01 -1.05129266e+00 -8.10763180e-01 7.00796768e-02 -2.07176492e-01 -1.33153462e+00 7.81518035e-03 -6.51646197e-01 7.13750497e-02 -1.40639150e+00 3.87460709e-01 -4.41502035e-01 -1.60910085e-01 -1.14653006e-01 1.54205307e-01 1.68039337e-01 1.92419454e-01 4.43093270e-01 -1.99718297e-01 7.16919065e-01 1.44117832e+00 -5.67882359e-02 -1.01136841e-01 1.25167906e-01 -2.81057447e-01 7.37453699e-01 8.23651105e-02 -6.51875257e-01 -2.10799366e-01 -6.33545280e-01 2.09388033e-01 6.77519917e-01 6.69631720e-01 -1.05372679e+00 3.24638158e-01 -1.02232829e-01 -1.38554517e-02 -4.63640630e-01 8.96781802e-01 -1.00030375e+00 6.68838322e-01 4.22212958e-01 1.04796542e-02 -3.81800622e-01 -1.37738481e-01 7.54206061e-01 -3.60227525e-01 -5.00435948e-01 1.12071776e+00 -2.83570349e-01 -3.66860181e-01 3.23342472e-01 -1.52376264e-01 2.49584056e-02 6.13832355e-01 2.19577312e-01 2.30673235e-02 -3.25659454e-01 -1.16882050e+00 -1.52925164e-01 4.74068612e-01 -3.67205769e-01 6.50014520e-01 -1.08620405e+00 -9.37989354e-01 6.87505752e-02 -3.29157233e-01 9.03255269e-02 7.97180772e-01 1.53546309e+00 -3.99633646e-01 -1.06566481e-01 -9.36504155e-02 -8.19022775e-01 -1.16199613e+00 5.28131187e-01 4.22674090e-01 -6.07333064e-01 -1.14759731e+00 6.54202104e-01 3.87076050e-01 -2.82481700e-01 -2.91025966e-01 1.95497535e-02 4.31164131e-02 -5.10038957e-02 5.56906164e-01 5.04978657e-01 2.24887833e-01 -7.57517517e-01 -1.34247169e-01 9.77950513e-01 2.80664831e-01 -4.89023030e-01 1.86479700e+00 -3.77836853e-01 -5.06171346e-01 3.94429207e-01 1.25250196e+00 1.38021722e-01 -1.50518429e+00 -5.02631307e-01 -4.25604284e-01 -6.67579532e-01 4.67000037e-01 -4.30169880e-01 -1.34095562e+00 8.12411368e-01 5.74912906e-01 4.49720435e-02 1.39070356e+00 -2.03688294e-01 8.87963474e-01 1.52767882e-01 5.01492083e-01 -6.90925241e-01 5.55386022e-02 3.69121760e-01 1.09712076e+00 -1.02349281e+00 3.72575760e-01 -6.28261805e-01 -2.31143266e-01 1.04403627e+00 -1.19778208e-01 -1.40356705e-01 6.71240032e-01 1.01448156e-01 -2.66031504e-01 -1.82516515e-01 -3.29535544e-01 -1.38642386e-01 1.93136021e-01 4.08452481e-01 -5.19251488e-02 -2.48093545e-01 -4.81214941e-01 1.77868847e-02 4.28696871e-01 1.11012841e-02 5.85566282e-01 7.96866357e-01 -7.55494013e-02 -1.11703014e+00 -7.01412797e-01 7.74908662e-02 -5.38109541e-01 -1.70210679e-03 3.18139732e-01 6.67764425e-01 -1.21550143e-01 8.47896695e-01 -3.87305290e-01 1.84233591e-01 2.59936333e-01 -1.01214856e-01 7.81218886e-01 -4.64240104e-01 -1.94534123e-01 4.64592069e-01 -2.13787064e-01 -6.44589603e-01 -8.21782649e-01 -1.00989807e+00 -9.72190797e-01 -8.94856825e-02 -3.67505737e-02 2.16261353e-02 6.25160217e-01 1.01254427e+00 7.13691413e-02 4.52527374e-01 7.41909146e-01 -1.28006780e+00 -6.13390207e-01 -6.72290027e-01 -8.86900961e-01 4.93373126e-01 5.35233796e-01 -7.06309676e-01 -7.49988616e-01 1.21976752e-02]
[11.78738021850586, -2.4414846897125244]