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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
29e47874-f6ec-43db-b2f5-c62ab0a52022 | sdc-stacked-dilated-convolution-a-unified | 1904.03076 | null | http://arxiv.org/abs/1904.03076v1 | http://arxiv.org/pdf/1904.03076v1.pdf | SDC - Stacked Dilated Convolution: A Unified Descriptor Network for Dense Matching Tasks | Dense pixel matching is important for many computer vision tasks such as
disparity and flow estimation. We present a robust, unified descriptor network
that considers a large context region with high spatial variance. Our network
has a very large receptive field and avoids striding layers to maintain spatial
resolution. These properties are achieved by creating a novel neural network
layer that consists of multiple, parallel, stacked dilated convolutions (SDC).
Several of these layers are combined to form our SDC descriptor network. In our
experiments, we show that our SDC features outperform state-of-the-art feature
descriptors in terms of accuracy and robustness. In addition, we demonstrate
the superior performance of SDC in state-of-the-art stereo matching, optical
flow and scene flow algorithms on several famous public benchmarks. | ['Oliver Wasenmüller', 'René Schuster', 'Didier Stricker', 'Christian Unger'] | 2019-04-05 | null | null | null | null | ['stereo-matching'] | ['computer-vision'] | [ 7.13646784e-02 -9.52755153e-01 -2.26288944e-01 -3.55856925e-01
-2.07829848e-01 -2.20770687e-01 7.18629837e-01 -2.71004528e-01
-5.12851357e-01 5.25602520e-01 3.92194629e-01 3.23423184e-02
-2.30729277e-03 -9.32318628e-01 -5.47396302e-01 -6.80544674e-01
-1.30398944e-01 -1.73772112e-01 6.10981584e-01 -2.71156609e-01
6.34834051e-01 7.93228388e-01 -1.86253464e+00 2.66863793e-01
8.96075189e-01 1.17184424e+00 4.59963530e-02 6.26489520e-01
-6.70374259e-02 8.88306260e-01 -2.46734664e-01 -2.46355787e-01
6.08992457e-01 -1.11104980e-01 -8.31604362e-01 -3.89516532e-01
1.27107406e+00 -5.27783155e-01 -7.96424985e-01 1.02985775e+00
6.19464934e-01 4.04754162e-01 5.50187588e-01 -1.26375961e+00
-7.77301431e-01 -2.28790790e-02 -7.03706443e-01 4.87374038e-01
4.29797471e-01 2.68622547e-01 9.38595653e-01 -9.17998791e-01
6.89890265e-01 1.53132617e+00 7.92751789e-01 4.27730352e-01
-1.03501153e+00 -8.18437219e-01 -3.45705897e-02 3.06734920e-01
-1.32247615e+00 -5.68894029e-01 7.51048028e-01 -5.23711860e-01
1.27127266e+00 -5.44227436e-02 6.18184030e-01 8.60196650e-01
3.14679652e-01 6.11876845e-01 8.55789065e-01 -1.48274124e-01
-7.36213997e-02 -4.70017970e-01 1.67898461e-01 8.91983330e-01
2.90934324e-01 5.62893867e-01 -5.43002248e-01 7.99414143e-02
1.18091822e+00 1.35543123e-01 -3.57291371e-01 -2.90123910e-01
-1.30493915e+00 7.41379619e-01 9.40264642e-01 1.47752002e-01
-1.90902755e-01 6.42451525e-01 5.34505308e-01 1.73061803e-01
1.55059218e-01 2.57005811e-01 -9.56577957e-02 -1.90807819e-01
-6.66423798e-01 3.20428401e-01 4.99859035e-01 9.03083861e-01
1.11438501e+00 1.12451538e-01 -2.71849126e-01 9.58036005e-01
3.34066190e-02 7.40924954e-01 4.41499144e-01 -1.45253372e+00
5.00617266e-01 4.53481436e-01 -1.82357550e-01 -1.49169803e+00
-2.95486122e-01 -1.68642610e-01 -1.25092614e+00 4.35795993e-01
1.94556743e-01 2.51526088e-01 -6.92314208e-01 1.69310141e+00
2.18576696e-02 8.13102126e-01 -6.88093081e-02 1.07591546e+00
1.07342803e+00 5.67479730e-01 4.17353436e-02 3.91490251e-01
9.70739484e-01 -1.18925571e+00 -4.00118381e-01 -1.90086842e-01
4.09999579e-01 -8.52772534e-01 6.09175980e-01 -1.53040931e-01
-9.77212131e-01 -1.04266024e+00 -9.94008601e-01 -4.93518651e-01
-4.47315514e-01 -3.11732471e-01 1.08274829e+00 4.00373250e-01
-1.20569003e+00 8.93297315e-01 -5.09159029e-01 -3.03215414e-01
6.63903117e-01 4.36934024e-01 -5.69848418e-01 -1.84763923e-01
-1.05006456e+00 7.41790414e-01 4.76715751e-02 2.41533238e-02
-4.91457224e-01 -8.36001217e-01 -1.24747753e+00 1.12255648e-01
-3.42578679e-01 -1.02742279e+00 6.53044701e-01 -6.07584834e-01
-1.37964249e+00 1.05259979e+00 -5.41125298e-01 -2.88273782e-01
4.00825232e-01 -1.42378911e-01 -4.27856058e-01 3.26747149e-01
3.06326330e-01 1.19748783e+00 8.32639277e-01 -7.02212811e-01
-9.82593298e-01 -1.29469797e-01 -5.69259338e-02 -1.23391636e-01
-4.40911502e-02 6.34656101e-02 -4.95750487e-01 -6.79966807e-01
1.13548517e-01 -6.70318961e-01 -1.98156998e-01 5.02310216e-01
1.05261296e-01 -2.69520104e-01 9.17145848e-01 8.72965753e-02
1.14570355e+00 -2.12098098e+00 -6.79518357e-02 1.70287862e-01
3.78290147e-01 5.49919546e-01 -4.52386916e-01 -3.47883739e-02
-5.74685149e-02 -1.60908148e-01 1.52240833e-03 -3.13363224e-01
-1.15761593e-01 9.43971202e-02 -3.69473249e-01 7.62297869e-01
3.52946252e-01 8.99378479e-01 -9.52604175e-01 -7.04652131e-01
7.43556559e-01 6.14056766e-01 -7.74617255e-01 -7.55405053e-03
4.16983277e-01 3.38526756e-01 -3.66271049e-01 5.66916764e-01
1.17658830e+00 -2.67650902e-01 -4.24268961e-01 -1.96806043e-01
-3.73721957e-01 1.91990107e-01 -1.21341538e+00 1.95928574e+00
-4.91315454e-01 1.14580035e+00 -2.38116622e-01 -8.14660490e-01
1.05280924e+00 -1.62101701e-01 6.22159839e-01 -1.01673245e+00
1.21416762e-01 5.36821783e-01 -1.58361152e-01 -2.36388534e-01
5.04667461e-01 3.99200112e-01 2.91810811e-01 2.42923856e-01
1.58518672e-01 -1.11418143e-01 3.68480414e-01 8.14661267e-04
9.93043602e-01 -1.19516842e-01 2.28422970e-01 -6.75107479e-01
1.12358630e+00 -3.03979933e-01 8.93591225e-01 8.20311427e-01
-7.74691761e-01 8.60554934e-01 2.51176059e-01 -1.00333750e+00
-7.66415358e-01 -9.32162106e-01 -4.45953280e-01 7.65489042e-01
6.08029127e-01 -3.07096243e-01 -3.18514794e-01 -4.02364939e-01
5.84555686e-01 -2.77801275e-01 -7.00678885e-01 -1.03069693e-01
-1.00634003e+00 9.35467426e-03 6.60809875e-01 7.94698417e-01
1.16952932e+00 -1.07570839e+00 -7.48814344e-01 1.99531764e-01
-8.04696977e-02 -1.31430936e+00 -9.46158469e-01 -2.92238533e-01
-6.79861188e-01 -1.38651073e+00 -7.57452250e-01 -1.16785693e+00
4.51915443e-01 8.50013614e-01 1.29763484e+00 2.30221108e-01
-5.54747283e-01 2.36205712e-01 -4.68424894e-02 4.69521768e-02
1.71145320e-01 -1.17349915e-01 -1.34296760e-01 -6.51997104e-02
5.80093682e-01 -7.35288382e-01 -1.14846611e+00 4.79339510e-01
-7.42381334e-01 -9.99750495e-02 1.28339544e-01 9.11475539e-01
4.04681325e-01 -5.09635329e-01 1.23630911e-01 -3.67844343e-01
4.37205583e-01 9.26524848e-02 -9.43898320e-01 5.31458035e-02
-3.63281697e-01 1.36105016e-01 5.73869169e-01 -3.23852628e-01
-8.35172474e-01 8.20871964e-02 -1.28473312e-01 -5.34689963e-01
-1.44127488e-01 -2.74315953e-01 4.01966751e-01 -8.99642110e-01
6.29267931e-01 1.21736899e-01 8.21380392e-02 -2.30068043e-01
2.23046333e-01 5.89097917e-01 8.75228047e-01 -6.93686724e-01
6.50025010e-01 6.95371926e-01 5.26961923e-01 -6.56706572e-01
-6.03097856e-01 -8.43768239e-01 -7.91473448e-01 -2.31058016e-01
7.06252217e-01 -1.06950009e+00 -1.20109200e+00 1.02933931e+00
-1.46328998e+00 -8.42417628e-02 -9.47173759e-02 6.18235469e-01
-6.87131107e-01 2.91090101e-01 -7.99683452e-01 -2.62198716e-01
-4.60210741e-01 -1.41559386e+00 1.24189973e+00 4.81759310e-01
6.58266023e-02 -1.27022827e+00 3.38993639e-01 1.69252453e-03
8.58231187e-01 2.52901912e-01 4.71092373e-01 9.24048796e-02
-8.27541709e-01 1.46756113e-01 -8.51154149e-01 4.51885045e-01
1.96124792e-01 2.81951994e-01 -1.09681630e+00 -2.69040883e-01
-6.78750932e-01 -2.49693960e-01 1.51755834e+00 6.33888602e-01
1.35614872e+00 3.10251981e-01 -3.54738533e-01 1.49394333e+00
1.69619656e+00 1.06166489e-01 9.87175524e-01 4.46405649e-01
7.75226831e-01 3.92416716e-01 1.85080692e-01 4.22828496e-01
3.13463360e-01 5.43755949e-01 3.64084840e-01 -2.72205055e-01
-5.88038623e-01 -3.63619961e-02 1.90492999e-03 4.52844203e-01
-3.70887935e-01 9.35217887e-02 -7.97187030e-01 4.83159661e-01
-1.89400840e+00 -1.22595930e+00 -2.50851899e-01 1.99149966e+00
4.67964500e-01 -5.35940900e-02 -2.77138472e-01 -1.38873845e-01
8.47986937e-01 6.19226336e-01 -2.71934032e-01 -4.97859925e-01
-5.37473261e-01 6.48037195e-01 7.00835943e-01 6.55476034e-01
-1.55989027e+00 1.12036562e+00 7.19051027e+00 6.16769552e-01
-1.24849534e+00 -2.89327323e-01 4.33396816e-01 2.66554803e-01
-7.40741119e-02 -1.34193778e-01 -8.53346646e-01 4.05503273e-01
2.10818067e-01 -3.80755439e-02 3.42147440e-01 6.28243387e-01
8.81064907e-02 -9.75814983e-02 -1.12506795e+00 1.60612583e+00
1.53049566e-02 -1.98484373e+00 5.11793755e-02 -3.44538689e-03
1.05320811e+00 3.03333580e-01 6.10338859e-02 1.54343560e-01
2.35102892e-01 -1.06305277e+00 3.15859795e-01 4.12965417e-01
9.17528450e-01 -4.57329184e-01 7.29838073e-01 -4.05257881e-01
-1.59366858e+00 -1.68294720e-02 -6.69172704e-01 -2.17784181e-01
-4.58073840e-02 4.08678293e-01 3.41172457e-01 2.68828809e-01
7.84094930e-01 1.55375504e+00 -6.06633604e-01 1.41994357e+00
7.08960220e-02 -8.90256464e-02 -2.07206085e-01 1.65527344e-01
5.81262350e-01 -3.66743356e-01 2.39497781e-01 1.44227207e+00
1.87649846e-01 -1.81667805e-01 2.60229222e-02 9.16735768e-01
-1.85723901e-01 -7.93134868e-02 -8.47182572e-01 6.77846968e-01
5.04786372e-01 1.03265440e+00 -2.82861561e-01 -3.90387535e-01
-7.22642422e-01 9.30248022e-01 4.23800677e-01 3.49482358e-01
-6.00760400e-01 -8.36468220e-01 1.46953523e+00 -3.07215691e-01
3.18767309e-01 -1.49013713e-01 -2.03202829e-01 -1.29752398e+00
-1.94329787e-02 -3.86002511e-01 3.17133188e-01 -5.46046972e-01
-1.44451487e+00 4.67505842e-01 -3.96507829e-01 -1.57816911e+00
-1.93513855e-01 -9.70351219e-01 -9.44884598e-01 1.10284960e+00
-2.24653029e+00 -7.53812671e-01 -8.59310627e-01 1.09611619e+00
2.53722459e-01 -2.91990370e-01 6.90086663e-01 6.07670009e-01
-5.81741035e-01 4.59742039e-01 3.53364088e-02 5.38435102e-01
1.05084503e+00 -9.92078662e-01 7.54591584e-01 9.14849639e-01
-1.67977303e-01 6.34809196e-01 5.15143201e-02 -1.62820607e-01
-1.09196794e+00 -1.09580815e+00 9.08775091e-01 2.74025388e-02
5.42061388e-01 -3.96841951e-02 -8.09791148e-01 3.29566568e-01
1.49815947e-01 9.53795314e-01 2.54706383e-01 -2.09372398e-02
-8.58262181e-01 -5.29513001e-01 -1.04205048e+00 3.15590560e-01
1.48846209e+00 -7.12820947e-01 -4.19432342e-01 -6.39777957e-03
4.48494583e-01 -4.41164374e-01 -6.45243585e-01 4.90172863e-01
7.95862615e-01 -1.45253372e+00 1.42576492e+00 -5.66734552e-01
5.35424352e-01 -2.06266731e-01 -2.30960652e-01 -1.01981926e+00
-7.67119348e-01 -6.40107453e-01 1.06984429e-01 9.73151147e-01
-1.56956762e-01 -9.57068861e-01 7.31084943e-01 2.88386643e-01
-2.39685029e-01 -3.57686222e-01 -9.13319528e-01 -9.07386661e-01
2.05067396e-01 -2.62521744e-01 6.89757168e-01 8.45088363e-01
-2.53627986e-01 -7.79441325e-04 -2.09083945e-01 -1.77269891e-01
7.84533083e-01 3.69029105e-01 7.60155499e-01 -1.37114513e+00
2.76812941e-01 -8.69367003e-01 -1.02388310e+00 -1.47945416e+00
5.06994426e-01 -7.18631208e-01 -1.28040075e-01 -1.30473876e+00
1.09952554e-01 -4.73163038e-01 -4.66154188e-01 2.17348903e-01
-1.59613505e-01 4.17675912e-01 2.88944453e-01 2.76971489e-01
-6.94477737e-01 4.88118589e-01 1.65287805e+00 -2.73070574e-01
-1.58714920e-01 -2.16380715e-01 -2.89098620e-01 5.96573710e-01
6.45429671e-01 -3.27979624e-02 -1.06397137e-01 -7.32392609e-01
-3.27600449e-01 -2.57872432e-01 5.78734159e-01 -1.22483063e+00
4.10570771e-01 -1.87271699e-01 4.50860262e-01 -5.26930153e-01
1.86574876e-01 -6.37053251e-01 -2.88653046e-01 6.37563944e-01
-1.87792838e-01 9.76326019e-02 1.69511899e-01 1.41729653e-01
-6.94931448e-01 6.92488551e-02 1.08162010e+00 -2.42271647e-02
-1.35015166e+00 6.85993195e-01 1.18872598e-01 2.14808345e-01
8.54062080e-01 -4.31633413e-01 -8.37366819e-01 -1.83834359e-01
1.38809070e-01 3.47688735e-01 3.29929620e-01 5.98894775e-01
8.41676414e-01 -1.71022999e+00 -6.40616238e-01 6.59001052e-01
2.76462734e-01 -3.81191671e-02 1.95008472e-01 6.61523223e-01
-1.18118346e+00 7.90965557e-01 -7.53166080e-01 -8.02361369e-01
-9.73514199e-01 4.11796153e-01 5.12708545e-01 6.42133579e-02
-7.41450846e-01 7.59625733e-01 3.97466689e-01 -1.50441721e-01
3.28464955e-01 -4.00502056e-01 -1.28552899e-01 -2.57275432e-01
6.64010882e-01 6.55314147e-01 -2.13831037e-01 -8.24729085e-01
-6.40990078e-01 1.34472632e+00 2.63637662e-01 4.85825390e-01
1.17073274e+00 -9.23127979e-02 -2.30959713e-01 -3.31156075e-01
1.76959574e+00 -3.19967896e-01 -1.50462675e+00 -4.02078122e-01
-4.31637406e-01 -1.08468640e+00 1.53194219e-01 -1.47385910e-01
-1.48897326e+00 1.04119742e+00 6.97835863e-01 -3.46320570e-01
1.25252283e+00 -3.18035871e-01 9.62976158e-01 4.12761867e-01
2.88944185e-01 -1.05480421e+00 -2.42909510e-02 9.08940077e-01
6.76272452e-01 -1.63726842e+00 -1.55958310e-01 -5.98864257e-01
-1.56069264e-01 1.46046770e+00 7.95294344e-01 -5.87732852e-01
9.55988646e-01 3.28366399e-01 1.35197639e-01 -3.10095306e-02
-6.22448146e-01 -6.41688108e-01 5.20893276e-01 8.13812077e-01
3.78171533e-01 -2.16592565e-01 -1.46586195e-01 -3.84439677e-01
8.49634185e-02 7.60252625e-02 1.11887150e-01 7.62560368e-01
-4.70365852e-01 -9.21303630e-01 -6.34571761e-02 4.51765023e-02
-3.20764393e-01 -2.27786005e-01 8.70071352e-02 7.94085741e-01
2.20136434e-01 6.87766671e-01 5.22795975e-01 -3.90435219e-01
3.36766511e-01 -5.28554082e-01 4.73866493e-01 -2.87769129e-03
-5.63266516e-01 -2.60373533e-01 -3.03224623e-01 -1.23554099e+00
-8.24277937e-01 -3.64208698e-01 -9.82722521e-01 -8.67380381e-01
1.70501888e-01 -3.96869630e-01 2.28707656e-01 5.48283219e-01
4.77736890e-01 2.70462811e-01 8.47919226e-01 -9.20955420e-01
-1.13607228e-01 -6.05461955e-01 -5.47019243e-01 9.36258376e-01
7.63040602e-01 -8.39932024e-01 -3.83791417e-01 -6.20820560e-02] | [8.904060363769531, -2.0800135135650635] |
6a512bae-c0cc-449b-9aec-9fe0029f7c2d | generating-token-level-explanations-for | 1904.10717 | null | http://arxiv.org/abs/1904.10717v1 | http://arxiv.org/pdf/1904.10717v1.pdf | Generating Token-Level Explanations for Natural Language Inference | The task of Natural Language Inference (NLI) is widely modeled as supervised
sentence pair classification. While there has been a lot of work recently on
generating explanations of the predictions of classifiers on a single piece of
text, there have been no attempts to generate explanations of classifiers
operating on pairs of sentences. In this paper, we show that it is possible to
generate token-level explanations for NLI without the need for training data
explicitly annotated for this purpose. We use a simple LSTM architecture and
evaluate both LIME and Anchor explanations for this task. We compare these to a
Multiple Instance Learning (MIL) method that uses thresholded attention make
token-level predictions. The approach we present in this paper is a novel
extension of zero-shot single-sentence tagging to sentence pairs for NLI. We
conduct our experiments on the well-studied SNLI dataset that was recently
augmented with manually annotation of the tokens that explain the entailment
relation. We find that our white-box MIL-based method, while orders of
magnitude faster, does not reach the same accuracy as the black-box methods. | ['Christos Christodoulopoulos', 'James Thorne', 'Arpit Mittal', 'Andreas Vlachos'] | 2019-04-24 | generating-token-level-explanations-for-1 | https://aclanthology.org/N19-1101 | https://aclanthology.org/N19-1101.pdf | naacl-2019-6 | ['sentence-pair-classification'] | ['natural-language-processing'] | [ 7.05706656e-01 8.84779751e-01 -2.51832545e-01 -9.44466054e-01
-1.21177161e+00 -2.97548354e-01 9.22640026e-01 4.17871535e-01
-1.75424710e-01 1.26285827e+00 2.82061070e-01 -8.72063100e-01
5.03149554e-02 -6.28994942e-01 -9.66496527e-01 -2.38628134e-01
1.24159209e-01 6.62290931e-01 5.14389314e-02 -1.50094226e-01
3.73858780e-01 -1.35862991e-01 -1.57385600e+00 9.75298643e-01
8.80970895e-01 7.51972020e-01 -1.40300179e-02 9.96335685e-01
-6.27879918e-01 9.63861525e-01 -7.45612144e-01 -6.54445291e-01
-2.65033752e-01 -6.41281962e-01 -1.34768677e+00 -1.52759135e-01
7.82851040e-01 1.30762801e-01 1.21556163e-01 5.63287199e-01
8.24384019e-02 -2.09182594e-02 7.87166178e-01 -1.42459631e+00
-7.07884610e-01 1.21255457e+00 -1.64132535e-01 1.95736781e-01
5.04534006e-01 -3.35558861e-01 1.42869568e+00 -9.55202579e-01
4.14644837e-01 1.23729873e+00 7.14168847e-01 7.98701465e-01
-1.28644681e+00 -4.45222288e-01 1.74688041e-01 2.98956364e-01
-7.11872697e-01 -2.26154566e-01 3.79303008e-01 -2.05714226e-01
1.57756782e+00 2.47379407e-01 1.44207925e-01 1.11453998e+00
3.84280801e-01 8.98570061e-01 8.91809106e-01 -1.03686869e+00
1.30789846e-01 3.47425282e-01 5.64452112e-01 6.62430525e-01
1.06094889e-01 -9.90245584e-03 -5.43618321e-01 -7.18032345e-02
2.21497566e-01 -1.51094645e-01 -5.24779260e-02 9.76439118e-02
-1.46463013e+00 1.15884745e+00 3.86624426e-01 7.20728934e-01
2.65998184e-03 3.13222647e-01 4.96074617e-01 2.77101636e-01
8.84951353e-01 6.96653605e-01 -9.42718148e-01 2.81264521e-02
-9.79557514e-01 3.18098038e-01 1.15075803e+00 9.45620596e-01
5.77238262e-01 -1.11104153e-01 -3.72273952e-01 4.31310415e-01
1.89678445e-01 -2.17611134e-01 5.77862263e-01 -7.18408763e-01
6.96463645e-01 3.72798026e-01 1.61273032e-01 -5.26037395e-01
-2.97393501e-01 -2.31804729e-01 -6.15763426e-01 1.66514367e-02
5.63111663e-01 -2.83984810e-01 -6.48665965e-01 1.68794084e+00
2.30060704e-02 4.02810216e-01 4.82462347e-01 5.32317579e-01
1.07936287e+00 6.87877476e-01 1.69103235e-01 2.73674745e-02
1.17534447e+00 -1.05414295e+00 -5.36805809e-01 -4.59791839e-01
1.06247795e+00 -6.87871873e-01 8.59203517e-01 2.32243463e-01
-1.07838023e+00 -6.83306754e-01 -1.10786688e+00 -1.84793234e-01
-5.04033744e-01 -1.55805042e-02 8.96989882e-01 2.57124752e-01
-1.01722646e+00 7.69226372e-01 -3.66301566e-01 -3.39735419e-01
1.83829263e-01 4.59796876e-01 -4.07176197e-01 1.53741390e-01
-1.47997272e+00 1.19602370e+00 4.67442900e-01 -2.75189709e-02
-3.20131481e-01 -5.68736434e-01 -1.09004533e+00 2.53163397e-01
1.96821034e-01 -5.80798447e-01 1.62936687e+00 -1.19014919e+00
-1.09532356e+00 9.24894989e-01 -5.63897908e-01 -9.41107452e-01
1.81382030e-01 -1.24919958e-01 -1.41043544e-01 -2.47409597e-01
4.16609079e-01 1.06339538e+00 4.28507239e-01 -1.20741498e+00
-5.82676888e-01 -1.46550328e-01 3.48143518e-01 -1.79847717e-01
1.41328704e-02 -3.81506346e-02 6.25174522e-01 -2.84700602e-01
-5.31102531e-02 -6.54333472e-01 -3.02069604e-01 -4.64577287e-01
-7.22350359e-01 -7.45043814e-01 4.18674529e-01 -5.04789889e-01
7.04407275e-01 -1.65638220e+00 -1.91378132e-01 -2.52116799e-01
1.23322591e-01 2.84895122e-01 -2.13197455e-01 3.79320979e-01
-4.45536047e-01 4.89835292e-01 -3.45140547e-01 -6.80837452e-01
2.25476280e-01 4.89491701e-01 -7.54614651e-01 -7.63632506e-02
6.07189953e-01 9.54111636e-01 -1.12748110e+00 -7.00075746e-01
5.74358813e-02 2.09898323e-01 -6.41481161e-01 3.21977913e-01
-6.55282617e-01 1.99141309e-01 -1.41234502e-01 5.45941815e-02
2.40895420e-01 -2.58862197e-01 9.29503962e-02 1.27376109e-01
-8.03811848e-02 1.08303130e+00 -6.00802064e-01 1.70813429e+00
-7.03241229e-01 9.33349192e-01 -4.71065253e-01 -1.36441350e+00
9.62794781e-01 8.13218772e-01 -9.26875547e-02 -1.56252772e-01
-4.66267988e-02 4.82771099e-01 1.45034343e-01 -4.85334992e-01
5.09295821e-01 -6.06460869e-01 -3.48884284e-01 5.03378272e-01
2.82136977e-01 -2.74486057e-02 2.32387125e-01 3.02425832e-01
1.17122066e+00 3.04124892e-01 5.05519509e-01 -2.21871193e-02
4.94124353e-01 2.91060001e-01 3.01525682e-01 1.08610094e+00
1.34003147e-01 6.37971282e-01 5.08797526e-01 -6.30667686e-01
-1.13971186e+00 -7.77615488e-01 -1.41582802e-01 8.89818132e-01
-2.64151543e-01 -4.05199736e-01 -8.00994456e-01 -9.63292778e-01
-9.06793624e-02 1.33264160e+00 -5.61527967e-01 1.44798785e-01
-3.85421336e-01 -3.37283820e-01 4.94339645e-01 7.43392408e-01
1.89995855e-01 -1.37787855e+00 -4.26473707e-01 3.39261204e-01
-3.70441139e-01 -1.00703919e+00 -1.08727001e-01 7.46084452e-01
-7.63894856e-01 -7.72114694e-01 -1.24155350e-01 -9.68528867e-01
5.61099410e-01 2.77275685e-02 1.56507421e+00 3.23314577e-01
-2.05279306e-01 9.10021365e-02 -4.45085764e-01 -6.62189424e-01
-8.01751614e-01 2.40419388e-01 -5.96379526e-02 -4.16515201e-01
7.92066395e-01 -3.23945910e-01 5.60298041e-02 -1.68086454e-01
-7.38232255e-01 3.63798589e-01 5.52629173e-01 1.25009251e+00
3.34810048e-01 -4.44981396e-01 8.09874535e-01 -1.28703284e+00
5.44907749e-01 -5.67564964e-01 -4.21563871e-02 3.82303476e-01
-3.77075136e-01 5.35137057e-01 7.77447939e-01 -1.82384104e-01
-1.08718705e+00 9.30224452e-03 -3.84900540e-01 1.08894184e-01
-7.17075825e-01 5.87463796e-01 -2.56353114e-02 3.31169486e-01
5.94089150e-01 -1.27135769e-01 -1.62334889e-01 -2.61828989e-01
4.26171809e-01 8.96455407e-01 4.60307807e-01 -4.40494478e-01
5.12629867e-01 -1.08053558e-01 -3.61303873e-02 -4.60983574e-01
-1.77460194e+00 -1.78767115e-01 -8.58589590e-01 5.61300367e-02
9.49660659e-01 -4.62020636e-01 -5.94993651e-01 -3.69121611e-01
-1.84700119e+00 -4.05526906e-01 -2.52213180e-01 3.53878647e-01
-7.77889729e-01 2.62624055e-01 -5.02594471e-01 -8.51002932e-01
-3.70685607e-01 -7.01836348e-01 1.01204789e+00 -9.99625102e-02
-8.33103895e-01 -1.19740391e+00 1.05913945e-01 6.27322614e-01
2.50327438e-01 6.34790659e-02 1.04990375e+00 -1.42275441e+00
-4.27687556e-01 -2.68716425e-01 -1.09527186e-01 3.30075532e-01
-9.57192704e-02 -1.29257917e-01 -1.14215517e+00 2.12240115e-01
7.06836805e-02 -5.56626379e-01 9.55766857e-01 3.14495891e-01
1.23495305e+00 -5.59812844e-01 -2.97857940e-01 -5.91605194e-02
1.26750338e+00 -8.78374279e-02 5.76863527e-01 1.63171321e-01
3.82227182e-01 7.41656780e-01 8.01303685e-01 8.92057642e-03
5.16565442e-01 5.43219984e-01 4.57060367e-01 -1.76761732e-01
6.24480769e-02 -2.60751873e-01 3.12886357e-01 4.31265742e-01
1.93335041e-01 -4.97144014e-01 -7.85339236e-01 7.70805657e-01
-2.12429118e+00 -1.33263385e+00 -4.84236419e-01 2.01397920e+00
8.91141832e-01 4.47360933e-01 -4.05296326e-01 3.70999992e-01
6.30806327e-01 -1.58431694e-01 -6.30775765e-02 -1.12460506e+00
2.17554435e-01 3.56340438e-01 1.95002228e-01 1.04753685e+00
-1.11135817e+00 8.44063938e-01 6.36492014e+00 5.06237149e-01
-6.76520824e-01 2.01408908e-01 8.12927604e-01 1.04561545e-01
-5.91878295e-01 3.10974300e-01 -9.96308327e-01 1.74158230e-01
1.38873827e+00 1.19276531e-01 5.88599732e-03 8.24286401e-01
3.16403322e-02 -1.02785267e-01 -1.59012258e+00 3.68759215e-01
2.60277033e-01 -1.50995517e+00 3.09717115e-02 -2.12267309e-01
6.37328982e-01 -1.81626946e-01 -3.31341177e-01 4.34961349e-01
2.17140749e-01 -1.18553472e+00 6.47046983e-01 4.17555630e-01
4.42567378e-01 -6.29840016e-01 1.19325173e+00 5.44322848e-01
-7.35755801e-01 3.93842161e-02 -4.97292578e-01 -6.99628472e-01
9.66794416e-02 7.05757558e-01 -1.35763705e+00 3.58549356e-01
1.26440987e-01 7.07829833e-01 -4.48687732e-01 7.30383575e-01
-6.56717837e-01 8.35002124e-01 7.49347657e-02 -4.34773952e-01
4.79404032e-01 2.78587788e-01 2.92082429e-01 1.36327684e+00
2.48895019e-01 -4.17792685e-02 1.91893145e-01 1.06593680e+00
3.22315767e-02 -9.12164673e-02 -9.35201585e-01 7.28969127e-02
2.09274158e-01 1.14675176e+00 -4.32177931e-01 -7.81651735e-01
-6.04849398e-01 1.05587184e+00 5.94735026e-01 -8.38450938e-02
-8.26304197e-01 -4.91428167e-01 3.55907917e-01 -9.47741568e-02
2.35491782e-01 9.37789753e-02 -6.35781109e-01 -1.12312353e+00
-2.53004371e-03 -4.44608659e-01 2.49169260e-01 -1.06936145e+00
-1.47733963e+00 7.74624050e-01 6.27500052e-03 -8.45858395e-01
-1.04680359e+00 -7.28556395e-01 -1.16132867e+00 9.73971069e-01
-1.53621638e+00 -1.03337193e+00 6.02641441e-02 -4.37111215e-04
8.36812377e-01 -2.09342651e-02 1.20982277e+00 -1.22957692e-01
-1.79816231e-01 5.14817238e-01 -1.23588197e-01 1.78525880e-01
5.99433482e-01 -1.62400961e+00 6.79447472e-01 7.20474362e-01
6.94384217e-01 6.27502799e-01 1.21622586e+00 -4.64299440e-01
-6.65492237e-01 -9.67702091e-01 2.22562885e+00 -6.06769204e-01
6.37638867e-01 -3.82146388e-01 -9.55209970e-01 1.04765618e+00
5.47753096e-01 -2.26858631e-01 9.03410494e-01 4.65194762e-01
-2.97799081e-01 1.35606155e-01 -1.04812801e+00 3.62563252e-01
7.98300028e-01 -4.46585536e-01 -1.11727679e+00 7.37217307e-01
9.70896959e-01 -1.60438731e-01 -6.26618147e-01 2.31036440e-01
2.90151238e-01 -1.01064336e+00 8.47565413e-01 -1.13635898e+00
1.14030635e+00 -7.21410066e-02 1.63930133e-02 -1.39004600e+00
-5.63695468e-02 -1.08506307e-01 2.60515183e-01 1.50178492e+00
1.07904065e+00 -5.80793917e-01 6.53729200e-01 8.64940107e-01
-4.82276440e-01 -8.85696352e-01 -9.69927847e-01 -6.18436515e-01
2.60453373e-01 -6.33678019e-01 4.87260997e-01 5.28412163e-01
7.36360312e-01 7.59749591e-01 -1.69929042e-01 -1.01612344e-01
5.25582135e-01 4.15759265e-01 6.30670846e-01 -1.27256274e+00
-3.69146496e-01 -1.69952840e-01 -3.01876932e-01 -7.59371459e-01
8.00179780e-01 -1.29058182e+00 5.95937550e-01 -1.88446236e+00
3.90088171e-01 -2.15570971e-01 1.60047766e-02 8.00865531e-01
-2.80234486e-01 2.23760396e-01 -8.46373588e-02 -1.80983827e-01
-6.68023527e-01 1.98886544e-01 7.56368577e-01 -1.70067593e-01
3.13175470e-01 -7.80619383e-02 -8.22775245e-01 7.70779252e-01
9.74660277e-01 -6.34087861e-01 -2.52646208e-01 -4.37228948e-01
2.06144869e-01 1.89841077e-01 5.81728578e-01 -8.34001422e-01
4.91739735e-02 8.44853744e-02 2.81935960e-01 -5.75848222e-01
2.08647087e-01 -3.53926331e-01 -2.99457788e-01 4.15816844e-01
-1.12199211e+00 -7.16656521e-02 3.97492461e-02 4.63214844e-01
-3.58803004e-01 -8.31293166e-01 4.32244629e-01 -4.31425124e-01
-4.05958891e-01 -1.85242698e-01 -5.89497566e-01 -1.35537507e-02
6.59832120e-01 -3.09510659e-02 -2.69660443e-01 -4.81972873e-01
-6.19827092e-01 5.21847978e-02 9.71101746e-02 3.08905125e-01
5.85023940e-01 -1.05918133e+00 -8.48437786e-01 -4.84714694e-02
1.04135528e-01 -9.56377238e-02 -2.37912506e-01 4.74251419e-01
8.79420191e-02 9.56557095e-01 9.80400071e-02 -3.68759632e-01
-1.31726766e+00 5.86961210e-01 1.12602130e-01 -6.08718038e-01
-5.03590763e-01 9.62182045e-01 3.19071114e-02 -7.15220094e-01
3.58224958e-02 -5.29637396e-01 -2.39397287e-01 -2.62867481e-01
5.15534878e-01 -2.37745315e-01 -5.23537491e-03 -3.74374747e-01
-1.78350300e-01 -4.95780371e-02 -2.77161915e-02 -1.38862297e-01
1.35592508e+00 1.30673483e-01 -2.01678023e-01 9.15219963e-01
9.55840111e-01 -3.61048847e-01 -7.71132469e-01 -1.22286165e-02
2.66904324e-01 -2.17511058e-01 -3.86031657e-01 -8.91639650e-01
-3.90254676e-01 1.32426441e+00 -1.40485540e-01 5.60887098e-01
3.86316687e-01 2.25282669e-01 8.77809286e-01 4.68539536e-01
3.41191202e-01 -6.10134065e-01 -2.37494032e-03 7.25473583e-01
8.34937930e-01 -1.49210429e+00 -3.88258427e-01 -3.05897743e-01
-5.30646801e-01 1.14765298e+00 6.99791670e-01 -1.54844046e-01
1.26536399e-01 2.51563936e-01 -1.46235034e-01 -6.27971441e-02
-1.15266192e+00 -1.65347774e-02 2.12019563e-01 3.56851012e-01
1.15266287e+00 8.81099179e-02 -3.94537866e-01 8.22174609e-01
-2.79154927e-01 -6.42225295e-02 6.73759162e-01 5.25901258e-01
-6.40065074e-01 -1.31688571e+00 -2.64126599e-01 7.17388511e-01
-5.00426769e-01 -5.55643678e-01 -6.66299105e-01 7.04257488e-01
9.27533060e-02 1.26383007e+00 1.35064051e-01 -2.93926597e-01
-2.57086098e-01 7.83607781e-01 4.00385141e-01 -1.02813601e+00
-8.19478929e-01 -8.56473684e-01 5.79062700e-01 -1.71816781e-01
-4.97002423e-01 -6.42789543e-01 -1.51139307e+00 -7.07853436e-02
-4.92987812e-01 5.18252850e-01 7.03988373e-01 1.76503599e+00
1.20990731e-01 6.06194735e-01 2.93389678e-01 -1.10954404e+00
-5.85454345e-01 -1.15631533e+00 -2.74872720e-01 4.77701217e-01
2.82713234e-01 -4.28172022e-01 -5.75699508e-01 -1.63757354e-02] | [10.34440803527832, 8.541993141174316] |
3478acb1-7ae4-4a81-a255-b6e4144ad9e6 | fast-full-resolution-target-adaptive-cnn | null | null | https://www.mdpi.com/2072-4292/15/2/319 | https://www.mdpi.com/2072-4292/15/2/319/pdf?version=1673404470 | Fast Full-Resolution Target-Adaptive CNN-Based Pansharpening Framework | In the last few years, there has been a renewed interest in data fusion techniques, and, in particular, in pansharpening due to a paradigm shift from model-based to data-driven approaches, supported by the recent advances in deep learning. Although a plethora of convolutional neural networks (CNN) for pansharpening have been devised, some fundamental issues still wait for answers. Among these, cross-scale and cross-datasets generalization capabilities are probably the most urgent ones since most of the current networks are trained at a different scale (reduced-resolution), and, in general, they are well-fitted on some datasets but fail on others. A recent attempt to address both these issues leverages on a target-adaptive inference scheme operating with a suitable full-resolution loss. On the downside, such an approach pays an additional computational overhead due to the adaptation phase. In this work, we propose a variant of this method with an effective target-adaptation scheme that allows for the reduction in inference time by a factor of ten, on average, without accuracy loss. A wide set of experiments carried out on three different datasets, GeoEye-1, WorldView-2 and WorldView-3, prove the computational gain obtained while keeping top accuracy scores compared to state-of-the-art methods, both model-based and deep-learning ones. The generality of the proposed solution has also been validated, applying the new adaptation framework to different CNN models. | ['Giuseppe Scarpa', 'Matteo Ciotola'] | 2023-01-05 | null | null | null | mdpi-remote-sensing-2023-1 | ['image-super-resolution', 'satellite-image-super-resolution', 'pansharpening'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 3.06884140e-01 -1.31939501e-01 -1.72363773e-01 -3.16427112e-01
-6.00989342e-01 -2.99040884e-01 6.97193086e-01 2.48816863e-01
-5.20443261e-01 7.40877926e-01 -1.41080141e-01 7.56833702e-02
-6.84099019e-01 -1.05382836e+00 -5.45445204e-01 -7.93259561e-01
1.14449263e-02 1.03383228e-01 4.96091008e-01 -4.60824579e-01
1.06038461e-02 7.51027167e-01 -2.02706385e+00 -3.72365788e-02
1.02406681e+00 1.45723450e+00 2.65071131e-02 4.49580818e-01
7.69549385e-02 3.28157753e-01 -3.60298663e-01 -5.29906332e-01
4.75303829e-01 6.39360398e-02 -2.70054489e-01 -1.17384054e-01
6.09538019e-01 7.53330532e-03 -1.44272521e-01 9.75492656e-01
4.60475951e-01 -1.46702826e-02 1.89286217e-01 -1.04902434e+00
-1.08221173e-01 2.84445792e-01 -5.36772430e-01 1.51676796e-02
-7.17911199e-02 -6.55530468e-02 8.68270218e-01 -7.24861443e-01
3.21731657e-01 9.49459016e-01 9.38566387e-01 9.67585742e-02
-1.35846353e+00 -6.78332865e-01 -7.78340995e-02 2.27520227e-01
-1.40029967e+00 -1.95645839e-01 8.20942223e-01 -3.08724284e-01
7.54081309e-01 2.85551101e-01 6.84806883e-01 9.93662238e-01
3.35737050e-01 2.48074517e-01 1.30188072e+00 -3.85697573e-01
2.65346766e-01 1.47767365e-01 -1.39471680e-01 3.79400700e-01
6.46651685e-01 3.11724722e-01 -4.42068458e-01 1.77903548e-01
5.53799808e-01 2.42077000e-02 -2.79238075e-01 -5.50545275e-01
-8.63966405e-01 7.50791907e-01 5.85123360e-01 5.49757719e-01
-5.41355729e-01 -1.58764452e-01 4.46657836e-01 8.60698894e-02
6.17622018e-01 4.14314985e-01 -5.58507204e-01 4.57051918e-02
-1.38455880e+00 5.10166764e-01 7.89176702e-01 4.07345921e-01
8.99046421e-01 7.09474981e-02 1.65119037e-01 6.79593742e-01
9.56065208e-02 3.86726141e-01 3.44762325e-01 -5.52200854e-01
4.22964126e-01 8.66999269e-01 -1.31546631e-01 -1.48584878e+00
-7.69351721e-01 -8.38566184e-01 -1.34948754e+00 5.24335802e-01
3.29727799e-01 -1.48316219e-01 -8.92876983e-01 1.85400224e+00
4.33272868e-01 7.14347735e-02 9.31806490e-02 7.22999513e-01
3.78427982e-01 3.68239760e-01 -2.68132519e-02 6.22381270e-02
1.35729086e+00 -6.72213912e-01 -5.64888537e-01 -8.46612453e-02
1.80468947e-01 -7.76539266e-01 5.67067385e-01 8.52555692e-01
-6.84236825e-01 -9.14709091e-01 -1.42305231e+00 1.40501201e-01
-9.35881674e-01 2.15132743e-01 5.68672836e-01 7.23902166e-01
-8.82416904e-01 8.40023160e-01 -7.30627537e-01 -7.64399409e-01
2.59426385e-01 3.75072390e-01 -5.33066690e-01 1.47195756e-01
-1.28989756e+00 1.02065504e+00 7.89135396e-01 3.20676804e-01
-3.45237434e-01 -6.87025726e-01 -4.85998452e-01 2.46373102e-01
6.32776856e-01 -5.72256863e-01 7.24636734e-01 -9.32313859e-01
-1.43615687e+00 5.39169490e-01 3.76920462e-01 -9.80919778e-01
7.50086427e-01 -4.88132745e-01 -8.52360845e-01 -5.92892617e-02
-3.17192703e-01 6.11628234e-01 8.93158078e-01 -1.23711765e+00
-9.20719385e-01 -5.09614468e-01 2.01654255e-01 -1.06000721e-01
-5.09504557e-01 -1.89917311e-01 -2.71094799e-01 -5.97991049e-01
1.20218873e-01 -8.79912794e-01 -8.66194516e-02 4.95176055e-02
-1.70518294e-01 -3.57681774e-02 9.04972315e-01 -5.94554961e-01
1.32171774e+00 -1.91611564e+00 1.34299904e-01 2.10108742e-01
1.51474848e-02 7.47177780e-01 2.17204809e-01 5.09194136e-01
-7.47078508e-02 -1.57611042e-01 -5.14887810e-01 -3.54140431e-01
-1.90837801e-01 4.37395215e-01 -2.43474036e-01 3.75439107e-01
3.55056763e-01 3.92491519e-01 -5.38724720e-01 -3.73171091e-01
6.81950688e-01 7.98383415e-01 -2.98551261e-01 1.71095338e-02
-1.44112200e-01 3.12045544e-01 -1.30438536e-01 5.33879876e-01
1.00538433e+00 7.06802309e-02 1.26572654e-01 -5.91097534e-01
-5.38506866e-01 -2.79987901e-01 -1.53101313e+00 1.69909096e+00
-4.69345033e-01 4.23010796e-01 2.78706849e-01 -1.22503603e+00
1.13136458e+00 2.95158058e-01 4.92397666e-01 -7.63336003e-01
2.13875204e-01 5.05962670e-01 5.32171242e-02 -2.56509036e-01
7.40276456e-01 -3.31744961e-02 5.89413866e-02 -1.12828694e-01
1.67384237e-01 2.62673479e-02 3.15361828e-01 -3.32455188e-01
6.16156042e-01 4.86001879e-01 5.12813389e-01 -3.36162239e-01
8.97982061e-01 5.46177756e-03 6.39320195e-01 6.54796362e-01
-9.38709676e-02 6.73857510e-01 7.02602714e-02 -6.29304588e-01
-1.02754176e+00 -5.75341702e-01 -2.58503526e-01 7.20050037e-01
6.92153051e-02 -3.69442850e-01 -8.57881784e-01 -4.17629480e-01
4.18624319e-02 4.67714250e-01 -6.50252759e-01 -9.60233286e-02
-6.06144845e-01 -9.34703290e-01 6.54247701e-01 4.79400784e-01
9.86492038e-01 -7.79835045e-01 -1.16037321e+00 3.68128359e-01
2.53840536e-01 -1.22191811e+00 5.53281069e-01 3.17178607e-01
-1.14153373e+00 -1.14090848e+00 -5.23603737e-01 1.04794605e-02
1.31575420e-01 9.46048722e-02 1.07463217e+00 -2.23244503e-01
-2.42574587e-01 4.52226922e-02 -3.59175384e-01 -6.09612942e-01
-2.84689695e-01 4.70510989e-01 4.77049202e-02 2.56863624e-01
3.40470642e-01 -7.09869206e-01 -5.51342726e-01 2.04371348e-01
-1.10744905e+00 1.56883225e-01 9.34985340e-01 6.44152999e-01
4.54332322e-01 2.74529308e-01 5.26198685e-01 -7.17944026e-01
3.24952871e-01 -3.17834914e-01 -1.05693901e+00 3.00618112e-01
-1.01608980e+00 3.25530767e-02 8.45626473e-01 -6.84841871e-02
-1.19604826e+00 8.68520737e-02 -2.82259643e-01 -3.24005038e-01
-4.29029524e-01 6.43178582e-01 -1.71121150e-01 -1.68983847e-01
7.25632668e-01 -2.78078299e-02 3.81508581e-02 -7.14461505e-01
3.44737351e-01 5.13276398e-01 8.34594250e-01 -3.71434689e-01
1.08827066e+00 6.59390092e-01 5.04590154e-01 -9.15992558e-01
-8.31567049e-01 -2.68102169e-01 -7.04508722e-01 -2.59019822e-01
8.21659148e-01 -8.90708268e-01 -4.31386679e-01 6.91602647e-01
-8.76332164e-01 1.63426638e-01 -2.02203900e-01 4.14201081e-01
-1.39351040e-01 2.90036142e-01 7.27339834e-02 -7.61964500e-01
-5.87048709e-01 -1.08230174e+00 8.51511180e-01 6.25268102e-01
1.39295936e-01 -8.93756092e-01 1.51370481e-01 1.72174335e-01
9.12591517e-01 6.34359598e-01 6.88237131e-01 -6.54292703e-01
-3.82416189e-01 -5.58196664e-01 -4.96412605e-01 5.87325990e-01
1.68040916e-01 2.15147734e-01 -1.16416645e+00 -2.65372992e-01
-3.62730026e-02 -1.23653680e-01 9.03223455e-01 2.67921120e-01
1.04819536e+00 1.93581522e-01 -2.70847201e-01 6.01099670e-01
1.78569961e+00 -4.52324972e-02 5.90061069e-01 4.87149864e-01
5.38049996e-01 5.00040174e-01 6.33147717e-01 3.58552605e-01
2.84549534e-01 9.65835512e-01 7.96315134e-01 -4.04367477e-01
-1.42994925e-01 -3.64700034e-02 -1.99003980e-01 3.50148737e-01
-4.73287463e-01 -1.10023655e-01 -6.80944562e-01 4.29094404e-01
-1.94924653e+00 -8.52837861e-01 -2.98618097e-02 2.54350090e+00
4.08790082e-01 3.11506718e-01 6.10146858e-02 5.60573995e-01
5.20556986e-01 3.68854374e-01 -4.36217338e-01 -2.61648595e-01
-3.32719058e-01 4.30854321e-01 8.04869652e-01 1.89133570e-01
-1.35570288e+00 5.22404492e-01 5.46084785e+00 7.73507118e-01
-1.61346185e+00 -1.07083008e-01 3.15333694e-01 2.65811682e-01
3.21758628e-01 6.68004379e-02 -8.42792630e-01 3.03926259e-01
1.08349693e+00 1.18898624e-03 2.03433573e-01 9.97335255e-01
1.45093232e-01 -2.92415291e-01 -7.19998777e-01 8.50354910e-01
1.53583437e-01 -1.33595061e+00 -2.50950962e-01 1.44155219e-01
5.40614545e-01 9.56785828e-02 -1.75380901e-01 2.90902436e-01
-2.54395962e-01 -7.33008265e-01 5.81558228e-01 6.14128411e-01
3.91722560e-01 -8.57059956e-01 1.01518929e+00 2.64438957e-01
-1.26037204e+00 -2.81018734e-01 -2.94447154e-01 -7.15210810e-02
7.83868581e-02 9.03703749e-01 -4.05642450e-01 1.31632805e+00
9.45659697e-01 4.68450844e-01 -8.34630549e-01 1.09280932e+00
-4.13644165e-02 4.37942147e-01 -6.05539143e-01 2.76705056e-01
2.09450409e-01 -2.44082958e-01 5.73846102e-01 1.23964465e+00
5.04452527e-01 -3.06322843e-01 -1.33832008e-01 6.94159448e-01
1.50628701e-01 -6.48856536e-02 -5.34239471e-01 2.62081474e-01
2.21738681e-01 1.75309622e+00 -5.84114432e-01 -3.21142703e-01
-4.42776799e-01 4.96842563e-01 9.05347168e-02 4.91874591e-02
-8.78579259e-01 -4.10761714e-01 5.23249865e-01 2.41129488e-01
5.05789578e-01 -1.79752782e-01 -2.02019840e-01 -9.70219553e-01
-9.14024562e-02 -8.24190259e-01 3.59885544e-01 -3.74971002e-01
-1.09640002e+00 6.69506788e-01 3.16916704e-01 -1.26600361e+00
-2.20023423e-01 -7.77267694e-01 -3.07444423e-01 7.74425089e-01
-1.76414669e+00 -1.28747654e+00 -8.51177633e-01 3.66576463e-01
2.64121890e-01 -1.28402680e-01 8.62326026e-01 6.51189983e-01
-5.64086795e-01 3.69221210e-01 6.39023259e-02 -1.87636733e-01
6.21532142e-01 -1.06770003e+00 1.20654151e-01 1.20780432e+00
9.70202014e-02 4.30310667e-01 9.11179781e-01 -2.65351832e-01
-1.27524519e+00 -9.68044937e-01 7.06065893e-01 1.27477013e-02
4.94323283e-01 -3.54480654e-01 -1.07621026e+00 2.75986582e-01
3.02286237e-01 1.14808634e-01 3.13266098e-01 1.56472057e-01
-3.49293470e-01 -7.18971550e-01 -1.31316531e+00 3.42853397e-01
6.98056340e-01 -8.95738676e-02 -4.75143790e-01 -1.28366023e-01
3.11744064e-01 -4.23484057e-01 -1.10716176e+00 9.28740203e-01
7.78855503e-01 -1.60687482e+00 8.76236200e-01 -7.74468035e-02
1.88698828e-01 -5.11391819e-01 -3.39855701e-01 -1.14472353e+00
-8.70258212e-02 -2.73627967e-01 -2.17041522e-01 1.24889588e+00
1.05025917e-01 -7.49532998e-01 7.52356052e-01 4.39684212e-01
-2.42772279e-03 -7.66916215e-01 -1.13183427e+00 -6.66815281e-01
-4.98908982e-02 -4.74332929e-01 6.52839065e-01 6.35132015e-01
-6.75160468e-01 1.88293397e-01 -5.56669831e-01 1.49585843e-01
6.38219714e-01 1.24528930e-01 9.78499115e-01 -1.85278773e+00
-2.64232099e-01 -4.74886954e-01 -6.15554035e-01 -5.52897274e-01
-3.02632302e-01 -3.08857381e-01 -2.68409014e-01 -1.32253349e+00
-2.67412663e-01 -3.90187860e-01 -4.21422750e-01 4.26587313e-01
2.40134797e-03 3.66593838e-01 2.14068919e-01 -9.22939857e-04
-1.79058746e-01 3.56595278e-01 6.44098699e-01 1.93777367e-01
-4.26145419e-02 -2.37141158e-02 -4.88560200e-01 7.42748320e-01
8.07856500e-01 -2.94507027e-01 -2.09727749e-01 -2.81621754e-01
3.26884985e-01 -2.66753852e-01 5.52867949e-01 -1.78999698e+00
2.57803977e-01 1.88404173e-01 4.27911371e-01 -7.68306732e-01
4.93148893e-01 -1.26175773e+00 5.32444715e-01 4.70845520e-01
1.16679519e-01 5.36752865e-02 4.31064963e-01 5.25893927e-01
-4.10363644e-01 5.80990948e-02 1.08457410e+00 1.03722438e-01
-7.96724498e-01 1.52924255e-01 6.80692568e-02 -4.98557717e-01
9.77497816e-01 -3.86709780e-01 -2.22331583e-01 3.78834689e-03
-4.29682851e-01 -1.43918291e-01 2.90936768e-01 5.91637671e-01
5.86687475e-02 -1.00319409e+00 -5.81689298e-01 2.12746397e-01
1.94623321e-01 1.87930658e-01 3.82756352e-01 1.10503399e+00
-3.70470524e-01 5.75395763e-01 -3.89824539e-01 -6.90824986e-01
-1.17456603e+00 3.96048188e-01 4.71991450e-01 -6.14430428e-01
-5.56871831e-01 4.31113660e-01 -2.78324783e-01 -3.60036731e-01
7.09443241e-02 -3.90443981e-01 -2.56650448e-01 3.40449363e-01
3.10490578e-01 5.51710546e-01 5.00724792e-01 -7.29537487e-01
-3.15254122e-01 7.40523338e-01 1.15703009e-01 1.45462170e-01
1.42894280e+00 4.91164327e-02 6.92622811e-02 2.68361926e-01
7.58265734e-01 -3.27025354e-02 -1.19285500e+00 -2.68911153e-01
1.64637551e-01 -4.44401145e-01 2.48000130e-01 -8.88114989e-01
-1.25609410e+00 9.15741086e-01 1.05429494e+00 5.08758128e-01
1.40172195e+00 -4.77819532e-01 5.68387270e-01 2.60064483e-01
3.98374528e-01 -1.20560586e+00 -4.18391287e-01 1.00761890e-01
7.23491728e-01 -1.28493357e+00 2.80214876e-01 -2.14401856e-01
-1.68059126e-01 1.25645196e+00 5.34465849e-01 -2.05568567e-01
6.61268890e-01 2.36298010e-01 -1.14397984e-02 -1.24174543e-01
-4.33425993e-01 -2.45709121e-01 2.39106685e-01 4.08200830e-01
4.04985577e-01 1.76821835e-02 -3.38010371e-01 3.12272996e-01
-5.09225503e-02 4.82289284e-01 1.00534037e-01 7.47524619e-01
-3.97260547e-01 -1.16146553e+00 -5.54161847e-01 3.42126340e-01
-4.21827137e-01 2.11840346e-01 -7.04296306e-02 1.24661791e+00
5.07724941e-01 9.59563494e-01 5.02094887e-02 -3.67415905e-01
6.53065443e-01 -8.48343596e-02 3.38772625e-01 -2.82708872e-02
-7.21414626e-01 -2.25882288e-02 -2.10420266e-02 -6.48717821e-01
-8.32876146e-01 -5.86458683e-01 -7.58427441e-01 -3.04249942e-01
-4.76816535e-01 -8.59864131e-02 1.00851393e+00 9.35614824e-01
4.94536012e-01 5.85561395e-01 4.08992648e-01 -1.02406037e+00
-6.02128863e-01 -1.05664921e+00 -4.60313588e-01 2.45369673e-01
1.35435581e-01 -9.62601185e-01 -2.76885599e-01 -3.75871003e-01] | [9.713468551635742, -1.6862986087799072] |
8dcba349-81a0-4448-9a85-22342b43183d | pain-or-anxiety-the-health-consequences-of | 2301.10675 | null | https://arxiv.org/abs/2301.10675v1 | https://arxiv.org/pdf/2301.10675v1.pdf | Pain or Anxiety? The Health Consequences of Rising Robot Adoption in China | The rising adoption of industrial robots is radically changing the role of workers in the production process. Robots can be used for some of the more physically demanding and dangerous production work, thus reducing the possibility of worker injury. On the other hand, robots may replace workers, potentially increasing worker anxiety about their job safety. In this paper, we investigate how individual physical health and mental health outcomes vary with local exposure to robots for manufacturing workers in China. We find a link between robot exposure and better physical health of workers, particularly for younger workers and those with less education. However, we also find that robot exposure is associated with more mental stress for Chinese workers, particularly older and less educated workers. | ['Robert Seamans', 'Sen Luo', 'Qiren Liu'] | 2023-01-25 | null | null | null | null | ['industrial-robots'] | ['robots'] | [-1.01694711e-01 5.31125247e-01 -3.64258081e-01 2.44225755e-01
3.55971009e-01 -1.41281068e-01 -1.30433083e-01 4.02950913e-01
-3.57993603e-01 3.51954222e-01 2.55057931e-01 -1.41283348e-01
-4.75048304e-01 -7.51769125e-01 -4.56085116e-01 -5.18022954e-01
3.27556193e-01 -1.84534252e-01 -3.90711933e-01 -3.10953140e-01
1.61607504e-01 8.33974004e-01 -1.18311179e+00 -5.14013469e-01
7.80491531e-01 4.22424167e-01 7.53206551e-01 -1.63570672e-01
7.17553318e-01 5.41350245e-01 -4.50153053e-01 -3.95452261e-01
2.00700797e-02 -2.38550141e-01 -7.03911245e-01 3.05314630e-01
-2.26970270e-01 -4.03443873e-01 -5.44662178e-01 8.54570687e-01
4.37498897e-01 4.23861116e-01 6.48784816e-01 -1.22375989e+00
-1.18582261e+00 3.72753292e-01 -4.46862608e-01 -4.08896983e-01
2.24788025e-01 4.14170682e-01 9.73380059e-02 -6.88724279e-01
4.43538159e-01 1.60413253e+00 7.89364338e-01 6.49374574e-02
-1.14175284e+00 -8.73967648e-01 -1.72265768e-01 -2.87622362e-02
-1.15219319e+00 -3.06586117e-01 5.34760058e-01 -4.56363589e-01
8.21498513e-01 -1.96935132e-01 5.68433940e-01 9.27380562e-01
1.05953288e+00 -4.37517941e-01 9.09207106e-01 -1.91158906e-01
-2.97405809e-01 2.72264600e-01 2.97419995e-01 4.55083162e-01
9.82785463e-01 -1.99036330e-01 -5.14101470e-03 -1.99864224e-01
7.18857825e-01 6.12188339e-01 -2.48604603e-02 -4.46168073e-02
-1.18380737e+00 4.75933939e-01 3.85290384e-01 3.71695787e-01
-8.35433006e-01 4.19007599e-01 4.45835352e-01 3.71751457e-01
2.80973881e-01 6.65014446e-01 -4.98152316e-01 1.33881625e-02
6.15602612e-01 4.52649593e-02 4.67078835e-01 6.41107678e-01
8.80945802e-01 -3.88784617e-01 -2.30153665e-01 1.01252270e+00
1.30457491e-01 6.54073358e-01 1.07648022e-01 -1.49346852e+00
1.93923011e-01 5.08069336e-01 2.81215519e-01 -1.22796047e+00
-7.25051880e-01 -2.61290580e-01 -7.01129377e-01 4.94577795e-01
2.84733117e-01 -5.64272761e-01 -3.32782388e-01 1.60731244e+00
-2.08756030e-02 -5.53648233e-01 7.50311613e-02 6.70249104e-01
1.37464143e-02 4.40942824e-01 4.65489000e-01 -2.74204999e-01
1.78215587e+00 -7.74203241e-01 -1.13634527e+00 -6.59237087e-01
5.77805758e-01 -5.04474521e-01 1.29218638e+00 1.96047664e-01
-7.61153519e-01 -1.07347775e+00 -6.54466569e-01 -4.27949727e-02
-3.70502889e-01 1.16475455e-01 5.93071580e-01 1.15103900e+00
-6.01796925e-01 9.44097817e-01 -6.62403286e-01 -7.54876256e-01
2.31561124e-01 4.21016097e-01 -4.66928273e-01 -5.94599903e-01
-9.29848433e-01 1.42168868e+00 4.63603362e-02 8.24451968e-02
-3.22752655e-01 -5.81845820e-01 -1.00224113e+00 2.92589404e-02
5.47972441e-01 -6.17765248e-01 1.17576408e+00 -2.12594315e-01
-1.20875204e+00 1.62415192e-01 2.95710951e-01 1.70895964e-01
-6.40782118e-02 -6.25873148e-01 -4.22904402e-01 -1.70956746e-01
6.00702286e-01 5.61460435e-01 5.07946253e-01 -1.14865267e+00
-2.66914368e-02 -6.92316055e-01 1.82289705e-01 2.70123005e-01
-2.28744671e-01 1.65612116e-01 4.24461156e-01 -2.80188799e-01
-3.62549797e-02 -1.58187532e+00 -5.52636802e-01 -1.36333600e-01
-2.96487123e-01 -7.26259410e-01 7.48964369e-01 -6.40943229e-01
2.66932368e-01 -2.67428517e+00 -3.56028646e-01 -1.08877324e-01
-5.44011518e-02 -4.58411992e-01 2.65880406e-01 5.71184754e-01
9.81352106e-03 2.61687249e-01 4.91780072e-01 -4.33557779e-02
1.28159031e-01 4.47832018e-01 3.89610857e-01 9.07062650e-01
4.81465787e-01 5.28890073e-01 -1.00907135e+00 -3.26355875e-01
1.79856852e-01 2.62251735e-01 -1.27893373e-01 -1.81827828e-01
8.08551073e-01 4.70375776e-01 -2.99748778e-01 4.58157390e-01
6.70311153e-01 3.12722951e-01 7.49346763e-02 5.31810820e-01
-3.48919392e-01 1.92295685e-01 -5.43586910e-01 1.28395665e+00
-7.05220938e-01 4.23449665e-01 4.57284838e-01 -3.33285034e-01
9.06927526e-01 5.93429625e-01 2.67821312e-01 -7.68592000e-01
2.64270276e-01 -8.37622657e-02 5.23326874e-01 -9.34378088e-01
5.19710362e-01 -4.00355726e-01 -3.91995788e-01 4.27050292e-01
-5.23580849e-01 -4.67943490e-01 -9.20687988e-02 -1.57663286e-01
1.00389421e+00 -2.91332215e-01 6.77846596e-02 -4.07087356e-01
-2.13678144e-02 -1.15332060e-01 7.83975840e-01 5.71822107e-01
-4.85875785e-01 -3.30367565e-01 6.73294127e-01 3.20736170e-01
-4.30778056e-01 -1.16867507e+00 -1.92765370e-01 1.07800031e+00
4.21249330e-01 1.36951342e-01 -3.33290815e-01 -3.80739778e-01
3.85484666e-01 1.10564506e+00 -2.03443825e-01 -8.92202914e-01
-3.60562891e-01 -2.35859022e-01 3.38560909e-01 5.47845781e-01
5.17899632e-01 -1.29847562e+00 -8.77350569e-01 -1.29399449e-01
-7.07927207e-03 -6.08729899e-01 -3.70573670e-01 3.00722659e-01
-7.68359423e-01 -6.79479718e-01 -8.07674110e-01 -8.56703281e-01
9.40300822e-01 7.03351021e-01 5.83607614e-01 4.04736012e-01
-1.29193172e-01 3.35122257e-01 -2.26517335e-01 -1.19626868e+00
-5.06096423e-01 8.63982737e-02 6.26585960e-01 -8.47843170e-01
-2.81157017e-01 -1.61559179e-01 -4.42682266e-01 5.35454988e-01
-4.10198569e-01 -3.61831218e-01 1.03638458e+00 1.05496109e-01
-2.69582748e-01 8.73858929e-01 1.27997708e+00 -7.26869762e-01
8.67573619e-01 -6.43736422e-01 4.44295198e-01 -1.13704026e-01
-9.22071397e-01 -5.20578206e-01 3.24782163e-01 -5.33567488e-01
-1.51713848e+00 2.86698178e-03 2.97464013e-01 3.01512390e-01
-5.52071333e-01 3.58867079e-01 -3.55835408e-01 2.77723521e-01
5.25696993e-01 -9.12034512e-01 3.48462045e-01 -4.22202736e-01
-1.29366666e-01 6.38904095e-01 3.32104623e-01 -3.56622815e-01
7.37691939e-01 7.26272687e-02 1.96105495e-01 -1.12691653e+00
-3.15838993e-01 -2.60132343e-01 -6.32232785e-01 -5.77752769e-01
1.26206434e+00 -8.14395905e-01 -1.09110117e+00 3.13892573e-01
-8.26947749e-01 -8.07722032e-01 -5.69947213e-02 1.03829038e+00
-2.44722575e-01 2.94649810e-01 -7.27884829e-01 -1.06079566e+00
2.86316276e-01 -1.05183971e+00 8.64465296e-01 6.70253858e-02
-6.97815597e-01 -3.44040871e-01 -2.93970227e-01 5.16821027e-01
-1.42838631e-03 3.07150483e-01 1.31656790e+00 3.27843934e-01
1.44100964e-01 -1.77338257e-01 -1.46905974e-01 4.77097660e-01
6.51264966e-01 -5.26710331e-01 -5.36671400e-01 -9.74162221e-02
4.30258363e-01 -3.37020010e-01 3.18026930e-01 2.97033012e-01
8.53549659e-01 6.18203133e-02 -7.13110268e-01 -3.76756072e-01
7.04513967e-01 7.39122152e-01 8.83025050e-01 1.04186786e-02
5.30004382e-01 1.43873942e+00 1.26890755e+00 -1.56603888e-01
-1.30236462e-01 1.92354009e-01 1.89570144e-01 -3.40333134e-01
2.64808416e-01 1.42773286e-01 5.48469901e-01 2.05282807e-01
-3.21062714e-01 -2.19233394e-01 -7.91104734e-01 3.24018508e-01
-1.54145885e+00 -5.98842263e-01 -7.14522660e-01 1.62524760e+00
3.67147535e-01 5.41223623e-02 6.29838780e-02 3.48831058e-01
6.90290213e-01 -4.78767574e-01 -4.33721989e-01 -6.30176485e-01
6.43319845e-01 8.76660272e-02 9.03481364e-01 -1.84608534e-01
-4.07735467e-01 1.74572840e-01 6.53706264e+00 -1.60114542e-01
-4.06388283e-01 3.43461782e-01 7.42889881e-01 -3.63627560e-02
3.88944626e-01 8.85978192e-02 9.92173702e-02 4.46151346e-02
9.76582289e-01 -4.96279076e-03 2.84685850e-01 9.11058903e-01
5.03416598e-01 -7.20078945e-01 -8.62903595e-01 4.21865106e-01
-2.08822265e-01 -2.30306342e-01 -9.61854100e-01 4.96661514e-01
3.79172206e-01 -7.50277698e-01 2.28827089e-01 5.70676684e-01
-1.21203773e-01 -8.63023877e-01 1.01624334e+00 3.89030606e-01
4.72384453e-01 -1.13945520e+00 1.00587344e+00 3.79229993e-01
-6.99366271e-01 -5.30963182e-01 -5.14431655e-01 -1.14603698e+00
2.69135743e-01 8.28189194e-01 -9.03005123e-01 1.77887663e-01
9.99809623e-01 1.15070399e-02 -5.73745608e-01 1.65977702e-01
6.12720177e-02 1.06374443e-01 2.56481528e-01 -5.90264089e-02
-3.76202911e-01 -3.02926779e-01 -2.27104351e-01 5.57840645e-01
4.55104142e-01 1.06830545e-01 -2.09771786e-02 7.61928082e-01
1.23884350e-01 -3.48679811e-01 -1.17058313e+00 -1.77901521e-01
3.90181512e-01 1.13341081e+00 -1.32938743e+00 1.05832946e-02
-3.81775826e-01 1.09746087e+00 -8.17400441e-02 1.63580716e-01
-7.37469971e-01 -8.88214350e-01 9.20384228e-01 1.47010177e-01
-6.24582350e-01 -7.66983569e-01 -6.41204000e-01 2.09044620e-01
-2.82264888e-01 -3.56066853e-01 -1.73051104e-01 -8.13312769e-01
-1.10516310e+00 -4.52785254e-01 1.47553712e-01 -3.76825482e-01
2.08468318e-01 -7.17081785e-01 -4.37928557e-01 8.69856358e-01
-5.52263319e-01 -3.79465282e-01 -1.90178409e-01 -1.03026271e-01
4.14440274e-01 4.26855743e-01 6.02712154e-01 3.91588062e-02
-8.82901430e-01 -4.79928032e-02 -2.70722985e-01 -3.56602341e-01
9.93802965e-01 -7.71077216e-01 1.77758947e-01 2.23090932e-01
-1.04221594e+00 1.06971943e+00 3.58653009e-01 -1.41270268e+00
-1.57745898e+00 -1.07304633e+00 7.63160646e-01 -6.56476140e-01
2.33346298e-01 -5.24983704e-01 -4.79979366e-01 6.04327440e-01
3.77131104e-01 -9.87494826e-01 5.62344432e-01 4.66499835e-01
5.87262809e-01 1.95311442e-01 -1.35874426e+00 9.81367588e-01
1.47965407e+00 -6.78448915e-01 -4.68422860e-01 5.94387472e-01
1.05126977e+00 3.01250309e-01 -1.18322027e+00 7.29224905e-02
3.95043999e-01 -6.77607954e-01 1.16077459e+00 2.77757615e-01
3.24907213e-01 9.90360752e-02 7.12866366e-01 -1.24789894e+00
-1.12174869e+00 -2.60728568e-01 8.22702229e-01 1.35528696e+00
4.62922961e-01 -9.82220531e-01 1.64389759e-01 9.26463366e-01
-7.66177416e-01 -3.91029328e-01 -7.49338269e-01 -1.24174488e+00
1.50773361e-01 -2.90337116e-01 4.73419696e-01 9.61788058e-01
5.36659598e-01 5.09542882e-01 1.42704010e-01 2.49511451e-01
-9.19930339e-02 -8.44438195e-01 4.80913013e-01 -1.39528811e+00
1.98226329e-02 -1.33542538e-01 -1.26639307e-01 1.63250137e-02
5.35138667e-01 -6.23999119e-01 4.72995341e-01 -2.09488058e+00
3.03189844e-01 -3.73816669e-01 2.49126017e-01 5.65162957e-01
-9.51120779e-02 3.68756354e-02 1.23016879e-01 -1.68041244e-01
1.27221167e-01 3.38895619e-01 1.51804388e+00 9.69505608e-02
-5.38232207e-01 9.31187496e-02 -9.18817341e-01 6.37042284e-01
1.43684137e+00 -5.73405087e-01 -5.17643929e-01 -1.50861502e-01
4.38279033e-01 7.85689577e-02 4.26556647e-01 -1.03021574e+00
-4.47894186e-01 -5.85721910e-01 6.05372906e-01 -2.82476068e-01
3.63552630e-01 -1.16428888e+00 5.90497017e-01 1.46721780e+00
1.13199614e-02 4.77870584e-01 1.00731030e-01 4.34782296e-01
8.67519140e-01 -5.59406459e-01 4.32180762e-01 1.48775220e-01
2.05637068e-01 -7.03783929e-01 -1.34042358e+00 -8.21052790e-01
1.24471521e+00 -9.35797114e-03 -4.15923238e-01 -1.45372152e-01
-5.08703411e-01 2.14322403e-01 7.85054743e-01 5.97000659e-01
1.89598471e-01 -1.14159477e+00 -1.09416112e-01 -1.31552413e-01
-1.97830480e-02 -2.44026959e-01 3.61094058e-01 1.24617016e+00
-3.56233507e-01 3.71994883e-01 -4.28063422e-01 3.27619493e-01
-7.71107376e-01 1.10709107e+00 -2.22147852e-01 4.68599230e-01
-6.13256037e-01 3.61510724e-01 3.21227819e-01 -2.41644874e-01
-3.90359461e-02 -8.82273257e-01 2.05156967e-01 -1.14656724e-01
-3.42743285e-02 1.73365390e+00 -2.38327131e-01 -5.91500401e-01
-3.87720227e-01 -3.84488888e-02 4.73230481e-01 7.59901255e-02
7.77950168e-01 -3.67966205e-01 -4.88613307e-01 8.37936640e-01
7.22550094e-01 2.04203978e-01 -1.18319774e+00 7.31989980e-01
-3.49657498e-02 -4.23421293e-01 -3.78436565e-01 -4.87142175e-01
-6.01554155e-01 2.93846965e-01 7.99855530e-01 4.56434727e-01
7.86416233e-01 5.08127332e-01 9.61166501e-01 9.33333576e-01
6.58813715e-01 -1.35371518e+00 5.69412231e-01 4.59711432e-01
9.53061998e-01 -7.66303420e-01 -2.15542629e-01 -9.35224771e-01
-7.04582393e-01 4.67372268e-01 6.47673607e-01 5.54894656e-02
6.50566578e-01 5.80313504e-02 -1.11310624e-01 -4.69006449e-01
-1.43706901e-02 7.36168101e-02 -2.82292575e-01 1.06120145e+00
7.33171880e-01 3.32408100e-01 -5.21206260e-01 6.56220376e-01
-4.14074033e-01 -2.55125046e-01 6.70194447e-01 9.49768603e-01
-5.38495481e-01 -9.22782063e-01 -8.74474525e-01 7.55175591e-01
-3.48352641e-01 5.62090993e-01 -7.57677376e-01 7.84112275e-01
8.15745413e-01 1.77707493e+00 2.03808680e-01 -2.55765557e-01
8.17683280e-01 3.61745149e-01 3.16124082e-01 -1.00032520e+00
-6.32187009e-01 -3.31101537e-01 6.58424735e-01 -2.04493403e-01
-1.61339805e-01 -1.08702326e+00 -1.53528690e+00 -4.45148021e-01
-4.39006984e-01 -2.68215984e-01 5.52166522e-01 8.51785362e-01
2.92204648e-01 9.52655613e-01 3.13618302e-01 -1.13384283e+00
-3.17258656e-01 -1.41725898e+00 -1.28347826e+00 -6.52214931e-03
-9.09102038e-02 -1.09547496e+00 6.04490601e-02 -1.63155511e-01] | [4.9331374168396, 0.9661966562271118] |
93a0657b-f1e0-4afa-9dfc-8d0b9ac794bf | prioritycut-occlusion-aware-regularization | null | null | https://openreview.net/forum?id=wVYtfckXU0T | https://openreview.net/pdf?id=wVYtfckXU0T | PriorityCut: Occlusion-aware Regularization for Image Animation | Image animation generates a video of a source image following the motion of a driving video. Self-supervised image animation approaches do not require explicit pose references as inputs, thus offering large flexibility in learning. State-of-the-art self-supervised image animation approaches mostly warp the source image according to the motion of the driving video, and recover the warping artifacts by inpainting. When the source and the driving images have large pose differences, heavy inpainting is necessary. Without guidance, heavily inpainted regions usually suffer from loss of details. While previous data augmentation techniques such as CutMix are effective in regularizing non-warp-based image generation, directly applying them to image animation ignores the difficulty of inpainting on the warped image. We propose PriorityCut, a novel augmentation approach that uses the top-$k$ percent occluded pixels of the foreground to regularize image animation. By taking into account the difficulty of inpainting, PriorityCut preserves better identity than vanilla CutMix and outperforms state-of-the-art image animation models in terms of the pixel-wise difference, low-level similarity, keypoint distance, and feature embedding distance. | ['Gyeongsu Chae', 'Wai Ting Cheung'] | 2021-01-01 | null | null | null | null | ['image-animation'] | ['computer-vision'] | [ 3.49160522e-01 4.88183536e-02 -3.90457153e-01 -1.57986253e-01
-6.08600855e-01 -4.77758646e-01 6.23969316e-01 -4.07973945e-01
-3.48824471e-01 4.97877002e-01 2.11541116e-01 1.23272955e-01
2.80604869e-01 -7.20208764e-01 -9.99597192e-01 -8.90510499e-01
8.25902596e-02 2.42109567e-01 1.98298618e-01 -2.65379101e-01
1.89559370e-01 3.79392147e-01 -1.55679178e+00 2.41473585e-01
7.82407999e-01 5.42365313e-01 1.35040507e-01 6.15823805e-01
-3.69433820e-01 7.52782524e-01 -6.28145695e-01 -9.68116447e-02
5.58893442e-01 -9.73883033e-01 -7.04081535e-01 4.02880281e-01
9.06583369e-01 -4.14025098e-01 -7.35652983e-01 1.07359684e+00
2.57340699e-01 2.29349151e-01 5.15683413e-01 -1.47093892e+00
-4.93637472e-01 4.10580307e-01 -1.11853409e+00 9.92982388e-02
4.08872873e-01 3.97628456e-01 6.89065516e-01 -8.00051093e-01
1.10380030e+00 1.23656416e+00 3.96149606e-01 6.61230862e-01
-1.56026745e+00 -8.91465962e-01 1.37860700e-01 1.68138966e-01
-1.17962039e+00 -1.97560877e-01 1.46784914e+00 -2.76906878e-01
3.89338344e-01 1.98103175e-01 9.92319882e-01 9.69581306e-01
1.17089137e-01 7.97961414e-01 1.03639877e+00 -4.60408121e-01
8.92464221e-02 -9.16750263e-03 -3.16527039e-01 9.18951869e-01
-1.07403509e-01 3.53038430e-01 -6.67006969e-01 -8.78271740e-03
1.08459747e+00 -3.15081291e-02 -3.21438193e-01 -6.02120340e-01
-1.39520562e+00 8.71793985e-01 4.47005153e-01 1.05305031e-01
-2.19371289e-01 3.68548512e-01 3.11623693e-01 2.53115207e-01
5.29703617e-01 5.00251770e-01 -1.11106306e-01 -7.74871111e-02
-1.34869516e+00 2.14796662e-01 3.33681226e-01 8.75953555e-01
1.02817869e+00 5.69748640e-01 2.30548158e-01 3.94285560e-01
-1.12205997e-01 3.70053053e-01 5.10762930e-01 -1.42188859e+00
2.85547376e-01 4.25754130e-01 -1.63632259e-01 -1.14514065e+00
1.31936088e-01 7.99710304e-02 -8.16361606e-01 7.03097939e-01
3.89983386e-01 -1.21973120e-01 -1.06644464e+00 1.76846814e+00
4.42744851e-01 4.66110289e-01 2.44064797e-02 9.05901968e-01
6.58032000e-01 1.02967918e+00 -8.48886371e-03 -4.88116205e-01
9.29326296e-01 -1.21849024e+00 -8.05321336e-01 -2.50665963e-01
3.89566571e-01 -8.42759609e-01 1.30779088e+00 1.97916701e-01
-1.26697874e+00 -8.45865548e-01 -1.04372382e+00 -3.60758118e-02
4.81685139e-02 -2.88876653e-01 3.85000050e-01 1.73988193e-01
-8.50879848e-01 8.90483320e-01 -7.72943079e-01 -5.71344085e-02
2.65800178e-01 1.35952115e-01 -7.01765180e-01 -5.92264161e-02
-1.00520623e+00 7.67750263e-01 1.20224096e-01 -4.13996428e-01
-9.15449917e-01 -1.11701787e+00 -1.14892828e+00 -2.29444966e-01
3.04930627e-01 -4.33023363e-01 7.26433098e-01 -1.52421761e+00
-1.56023955e+00 7.51964450e-01 -9.92596224e-02 -2.15429544e-01
7.92935967e-01 -3.20184052e-01 -1.97413206e-01 4.28169549e-01
1.59722924e-01 1.34560263e+00 1.50474191e+00 -1.47515500e+00
-3.57727200e-01 1.15583137e-01 -1.82859808e-01 3.90083522e-01
-2.21626818e-01 -2.01540172e-01 -5.94635010e-01 -1.11672366e+00
-2.24361181e-01 -1.01685882e+00 -3.32459390e-01 5.19571781e-01
-3.04245204e-01 1.56396911e-01 1.67577636e+00 -6.17217064e-01
9.79332566e-01 -2.40179086e+00 4.41503763e-01 1.73437856e-02
-9.77983698e-03 3.65431726e-01 -5.61363280e-01 2.49271750e-01
-4.43024069e-01 -1.08842611e-01 -5.12851000e-01 -3.75114381e-01
-5.49628198e-01 4.97692585e-01 -4.82059419e-01 5.94499409e-01
3.47783864e-01 7.03571141e-01 -1.01406467e+00 -8.87594223e-01
4.68441188e-01 7.32298017e-01 -6.07124627e-01 4.13557559e-01
-4.05871361e-01 6.73258066e-01 -3.65849733e-02 3.53336692e-01
4.05080050e-01 1.44264111e-02 -1.35589793e-01 -3.95166457e-01
-1.66241199e-01 -2.08961114e-01 -1.11926627e+00 2.11767840e+00
-2.56864518e-01 9.87895012e-01 -5.07301018e-02 -7.93513000e-01
8.55818152e-01 1.30498279e-02 6.23811722e-01 -4.08605248e-01
1.22079710e-02 -1.34984389e-01 -5.51653616e-02 -4.44613069e-01
3.91671866e-01 -7.48424279e-03 9.37642306e-02 5.31596303e-01
-6.26990646e-02 -5.99117875e-01 3.21972549e-01 3.23141694e-01
6.88094378e-01 6.45474434e-01 -9.41527355e-03 -1.81517541e-01
2.72999972e-01 1.41582444e-01 4.86773014e-01 4.95150015e-02
3.44779752e-02 1.05267048e+00 4.69588399e-01 -5.66861212e-01
-1.30219030e+00 -7.50505984e-01 2.81985492e-01 8.57766867e-01
2.57116377e-01 -3.07494491e-01 -1.08361340e+00 -6.68110728e-01
-4.70746495e-02 6.06390834e-01 -7.49453485e-01 -2.26296231e-01
-9.69636798e-01 -3.37172419e-01 3.52600783e-01 5.25984108e-01
6.08043194e-01 -1.12705076e+00 -5.82005858e-01 7.69700184e-02
-2.25001186e-01 -1.00724375e+00 -1.12422919e+00 -1.61507249e-01
-8.87602806e-01 -1.14769292e+00 -9.51763868e-01 -1.04214454e+00
1.06145132e+00 4.71516430e-01 7.54148245e-01 1.27163365e-01
-5.04029214e-01 1.30627096e-01 -8.83097425e-02 8.53388682e-02
-6.59429491e-01 -2.83908159e-01 -2.54130224e-03 1.46544911e-02
-3.19162577e-01 -6.98249757e-01 -6.37378991e-01 2.47179955e-01
-1.00209415e+00 3.06141108e-01 4.56528544e-01 8.83725226e-01
7.54773378e-01 -3.32224742e-02 -1.91186275e-02 -8.57965291e-01
1.97309822e-01 -4.24174452e-03 -4.44292635e-01 -1.87687837e-02
-5.69178164e-01 2.46517643e-01 7.73587584e-01 -9.38279152e-01
-7.70637870e-01 4.34447497e-01 1.30206943e-01 -1.09041691e+00
5.74645512e-02 -5.22903576e-02 -1.43592238e-01 -2.88984120e-01
6.41966522e-01 3.43547344e-01 6.18145227e-01 -1.68103710e-01
7.06455469e-01 -5.70903271e-02 8.41870129e-01 -3.68265986e-01
1.18100572e+00 6.58597469e-01 8.57790038e-02 -9.52283740e-01
-5.30964375e-01 -1.15763523e-01 -6.28295958e-01 -2.73764819e-01
9.72798049e-01 -7.30815291e-01 -2.40702555e-01 4.95835602e-01
-1.04565167e+00 -5.22756994e-01 -7.05088437e-01 2.75950909e-01
-7.35374570e-01 7.06928372e-01 -5.23689449e-01 -3.08082968e-01
-2.94110268e-01 -1.23837507e+00 8.50721776e-01 1.20120734e-01
-4.80513543e-01 -8.54978621e-01 1.66179314e-01 1.14214346e-01
1.88280970e-01 6.86584949e-01 9.51686025e-01 -4.00733761e-02
-3.91521335e-01 -5.90380169e-02 2.62363181e-02 4.44272608e-01
2.83547074e-01 3.92491907e-01 -4.72903609e-01 -2.80593038e-01
-2.09820002e-01 -1.86081201e-01 7.55585372e-01 2.57845551e-01
1.02345467e+00 -4.13126230e-01 -6.96903393e-02 9.54505503e-01
1.27559996e+00 1.27704665e-01 8.38278651e-01 2.12579295e-01
1.10901260e+00 7.69473553e-01 5.92267632e-01 4.45074290e-02
9.35694575e-03 6.28562629e-01 4.51107383e-01 -5.59626579e-01
-5.40912330e-01 -4.99369711e-01 5.61862648e-01 6.71055317e-01
-9.45551991e-02 1.49997622e-01 -5.77131450e-01 4.90792632e-01
-1.62818408e+00 -1.17352974e+00 7.34989047e-02 2.08611369e+00
1.23176432e+00 6.43908903e-02 2.35481486e-01 2.35189691e-01
7.78968513e-01 5.73509157e-01 -6.16588771e-01 -1.61665350e-01
-1.22681111e-01 2.44590923e-01 3.99955302e-01 8.27847004e-01
-9.93213832e-01 1.07080531e+00 6.08958769e+00 8.21599483e-01
-1.42268133e+00 -1.00020461e-01 6.98714256e-01 -1.97604507e-01
-3.30471337e-01 1.37360543e-01 -4.08561617e-01 5.67611277e-01
4.93306905e-01 8.55552405e-02 4.06882882e-01 9.45047915e-01
2.25276187e-01 -3.73717435e-02 -1.05598974e+00 9.38908875e-01
3.93096864e-01 -1.56226087e+00 3.94588768e-01 -2.33887330e-01
7.51789331e-01 -5.37961602e-01 1.67576119e-01 -9.75933373e-02
4.44583595e-02 -7.56730020e-01 7.15662181e-01 1.99404046e-01
1.07939363e+00 -9.52534318e-01 1.66361913e-01 1.32348493e-01
-1.05811012e+00 1.99974865e-01 -1.09322220e-01 1.56720147e-01
3.35481048e-01 1.42802283e-01 -2.56259829e-01 8.24134424e-02
5.46657860e-01 7.51346052e-01 -2.99688935e-01 4.84683424e-01
-4.43600565e-01 5.33170760e-01 -1.04382873e-01 5.33037663e-01
3.22421670e-01 -4.78752255e-01 5.90724349e-01 1.18999207e+00
2.30748266e-01 1.89814284e-01 1.57114059e-01 6.62663877e-01
3.89591940e-02 1.46681871e-02 -6.44156694e-01 -1.64892431e-02
2.56649703e-01 1.04739976e+00 -6.93790913e-01 -4.48116273e-01
-3.55397254e-01 1.32131732e+00 -7.78376833e-02 3.49090159e-01
-8.94670665e-01 -4.03064191e-01 8.04478526e-01 4.36756283e-01
3.01855922e-01 -2.88626969e-01 -7.16720894e-02 -9.39462066e-01
-3.19938540e-01 -1.05716348e+00 1.32296413e-01 -7.89705634e-01
-8.73481989e-01 6.39186263e-01 1.94228336e-01 -1.59317136e+00
-4.53168243e-01 -1.31508008e-01 -8.91484797e-01 5.01164019e-01
-1.45121455e+00 -1.16275585e+00 -5.09415507e-01 8.15103829e-01
9.92806792e-01 -8.24414715e-02 6.85691059e-01 1.81657612e-01
-4.79642540e-01 5.92095137e-01 -1.72671109e-01 2.58834153e-01
1.00317335e+00 -9.74237978e-01 4.72850621e-01 9.26953197e-01
2.63193011e-01 3.42858076e-01 9.58806574e-01 -6.65536046e-01
-1.44763553e+00 -1.17543960e+00 3.77755463e-01 -3.61039229e-02
5.06719410e-01 -1.37260228e-01 -1.08741319e+00 4.05837953e-01
5.96919417e-01 4.77525234e-01 3.55246902e-01 -7.41137564e-01
-4.99768823e-01 -1.68394968e-01 -1.10350513e+00 8.56932700e-01
9.46618497e-01 -3.51650208e-01 -4.22292769e-01 2.29291171e-01
7.66701698e-01 -4.26088899e-01 -4.66383427e-01 -1.21751679e-02
3.82788777e-01 -6.83179140e-01 1.08548963e+00 -6.16488457e-01
7.93487906e-01 -4.64787871e-01 2.79189318e-01 -1.35677671e+00
-2.79604226e-01 -1.11529362e+00 -1.84340984e-01 1.08974612e+00
1.11324385e-01 -2.10487526e-02 1.13562918e+00 3.83309305e-01
4.89582717e-02 -5.09775162e-01 -7.22255826e-01 -6.69864774e-01
7.37615377e-02 -1.13265496e-02 1.72595471e-01 1.26986170e+00
1.92555673e-02 1.12404220e-01 -8.07183623e-01 -1.28398240e-01
8.65412831e-01 1.38530970e-01 9.95710254e-01 -7.53260434e-01
-2.63921261e-01 -2.81993955e-01 -3.79913092e-01 -8.03220630e-01
4.14205372e-01 -5.06899536e-01 1.71284601e-01 -1.11925638e+00
-3.58334370e-02 -1.15790129e-01 3.05490524e-01 5.62980890e-01
-3.35262597e-01 5.83370745e-01 4.13621753e-01 4.90035564e-01
-1.13601666e-02 5.21165252e-01 1.58847773e+00 -1.98882684e-01
-3.19818974e-01 -4.62362498e-01 -2.71679848e-01 9.10037100e-01
6.85077608e-01 -5.08483946e-01 -6.24550223e-01 -4.37209070e-01
-3.18179101e-01 1.62895486e-01 1.99651361e-01 -9.10805225e-01
7.12701082e-02 -4.61604178e-01 4.03663188e-01 -2.39490032e-01
5.09191334e-01 -7.34621167e-01 4.03164834e-01 6.64667070e-01
-5.04639983e-01 3.76357436e-01 9.31573287e-02 5.44664145e-01
-2.65654206e-01 -8.84005949e-02 1.16927457e+00 -2.33026117e-01
-7.59329796e-01 4.30621654e-01 -3.41558844e-01 8.57142955e-02
1.18119419e+00 -4.16339159e-01 -3.13670412e-02 -5.73808253e-01
-4.00134861e-01 -1.53033704e-01 8.38201225e-01 3.83787632e-01
7.98994958e-01 -1.29184866e+00 -6.12485886e-01 4.05491799e-01
-2.35886902e-01 2.14147598e-01 8.16569701e-02 5.10025501e-01
-8.05618942e-01 -3.48189831e-01 -5.83437562e-01 -4.73624676e-01
-1.46249211e+00 7.17043877e-01 3.14220674e-02 1.82983896e-03
-9.47273016e-01 7.08977461e-01 4.36953872e-01 2.22729668e-01
2.84156233e-01 2.21769195e-02 1.13077708e-01 1.27341121e-01
5.18711209e-01 2.81524956e-01 -4.24981922e-01 -7.35279441e-01
-2.79538948e-02 9.21773136e-01 -2.32148215e-01 -2.89860010e-01
1.23325694e+00 1.75013080e-01 6.07223287e-02 2.16356426e-01
1.44508839e+00 8.28322321e-02 -1.97516787e+00 -1.11451000e-01
-5.52000165e-01 -7.69879818e-01 1.13631234e-01 -1.51146948e-01
-1.57815754e+00 8.61336589e-01 4.13055092e-01 -2.87722826e-01
1.09581363e+00 -2.67446905e-01 1.07155216e+00 -1.00868173e-01
1.08020484e-01 -1.02917218e+00 6.97439671e-01 1.26497895e-01
1.04211903e+00 -1.15634310e+00 2.99976856e-01 -5.97221255e-01
-8.70817602e-01 1.01627982e+00 7.47063398e-01 -5.68844795e-01
3.96418482e-01 5.29002309e-01 3.81023258e-01 1.34011954e-01
-3.57817292e-01 1.24772638e-01 1.69213533e-01 8.47973228e-01
1.20705426e-01 -4.02806371e-01 -2.78099597e-01 -1.69833332e-01
-1.21734761e-01 -4.06947583e-01 5.94238400e-01 1.11437488e+00
-1.92799658e-01 -1.10084033e+00 -4.25019056e-01 2.84967162e-02
-4.88548055e-02 -5.37862554e-02 -4.03617978e-01 1.01771319e+00
6.66363388e-02 3.42967063e-01 3.50124240e-01 -4.17849720e-01
2.21606880e-01 -6.08908832e-02 5.32400370e-01 -4.54018772e-01
-4.84840453e-01 2.48029545e-01 -2.49620005e-01 -6.40490174e-01
-6.19334221e-01 -6.83432400e-01 -1.43936610e+00 -4.16358858e-01
-1.80525377e-01 9.64693427e-02 4.68080461e-01 5.50213635e-01
3.34357888e-01 2.84314394e-01 7.34419227e-01 -1.36296117e+00
-3.20135266e-01 -6.51413739e-01 -2.69048393e-01 7.42469370e-01
4.63298827e-01 -5.58441460e-01 -4.59460109e-01 5.84635079e-01] | [10.948349952697754, -0.8534174561500549] |
ab3ca633-bc89-47f2-a71b-eb4d7dc52b27 | to-balance-or-not-to-balance-an | 1912.04486 | null | https://arxiv.org/abs/1912.04486v2 | https://arxiv.org/pdf/1912.04486v2.pdf | To Balance or Not to Balance: A Simple-yet-Effective Approach for Learning with Long-Tailed Distributions | Real-world visual data often exhibits a long-tailed distribution, where some ''head'' classes have a large number of samples, yet only a few samples are available for ''tail'' classes. Such imbalanced distribution causes a great challenge for learning a deep neural network, which can be boiled down into a dilemma: on the one hand, we prefer to increase the exposure of tail class samples to avoid the excessive dominance of head classes in the classifier training. On the other hand, oversampling tail classes makes the network prone to over-fitting, since head class samples are often consequently under-represented. To resolve this dilemma, in this paper, we propose a simple-yet-effective auxiliary learning approach. The key idea is to split a network into a classifier part and a feature extractor part, and then employ different training strategies for each part. Specifically, to promote the awareness of tail-classes, a class-balanced sampling scheme is utilised for training both the classifier and the feature extractor. For the feature extractor, we also introduce an auxiliary training task, which is to train a classifier under the regular random sampling scheme. In this way, the feature extractor is jointly trained from both sampling strategies and thus can take advantage of all training data and avoid the over-fitting issue. Apart from this basic auxiliary task, we further explore the benefit of using self-supervised learning as the auxiliary task. Without using any bells and whistles, our model achieves superior performance over the state-of-the-art solutions. | ['Jun-Jie Zhang', 'Chunhua Shen', 'Peng Wang', 'Lingqiao Liu'] | 2019-12-10 | null | null | null | null | ['auxiliary-learning'] | ['methodology'] | [ 2.40124226e-01 4.73932028e-02 -1.87204778e-01 -4.56870019e-01
-5.29167116e-01 -1.64233416e-01 4.46509868e-01 1.73620448e-01
-4.69835728e-01 6.17719889e-01 -2.81957060e-01 -1.61588326e-01
2.78746150e-02 -9.70484495e-01 -5.77054977e-01 -1.17386854e+00
3.35737795e-01 2.43501514e-01 3.45507443e-01 1.51793852e-01
9.97298434e-02 3.94375533e-01 -1.83978581e+00 1.28703699e-01
1.07051110e+00 1.34722686e+00 3.00996065e-01 9.03104767e-02
-4.03920293e-01 8.05777609e-01 -7.52310634e-01 -2.34737277e-01
7.38264620e-02 -3.76816362e-01 -3.38200361e-01 3.88285786e-01
2.30851471e-01 -2.29404181e-01 -4.10458520e-02 1.16920829e+00
6.70702696e-01 1.44844025e-01 6.68684244e-01 -1.39663672e+00
-1.76122159e-01 3.33158344e-01 -9.49307859e-01 5.55596873e-03
-9.63665172e-02 2.28748754e-01 8.95583987e-01 -7.48657703e-01
2.10548371e-01 9.07446384e-01 4.49705452e-01 5.05621850e-01
-1.17028189e+00 -9.27108586e-01 3.61559600e-01 2.02042133e-01
-1.32126570e+00 -4.37275171e-01 1.15711594e+00 -3.29747736e-01
2.22319767e-01 2.71314353e-01 8.22120190e-01 9.78313684e-01
-7.60345086e-02 1.22467971e+00 1.11163116e+00 -2.31069639e-01
3.83613050e-01 5.04724145e-01 2.31130674e-01 4.87364948e-01
2.92998433e-01 2.43077874e-02 -1.25275135e-01 -3.92160788e-02
5.17963946e-01 2.81922877e-01 -3.65185410e-01 -5.66438317e-01
-6.60029948e-01 8.36561978e-01 5.27699411e-01 1.51269153e-01
-3.70262593e-01 -3.61211240e-01 4.24887300e-01 7.57356361e-02
4.25564140e-01 1.45747602e-01 -4.27589238e-01 1.65813416e-01
-1.06923807e+00 2.22418174e-01 6.71919286e-01 6.74925685e-01
1.05317497e+00 -1.73737556e-01 -2.01683387e-01 1.02743649e+00
3.25158715e-01 1.26177102e-01 5.29589355e-01 -3.17496210e-01
5.44548690e-01 9.74782348e-01 -9.94702876e-02 -1.12090230e+00
-3.07229251e-01 -9.73326862e-01 -1.26188231e+00 3.78987342e-01
6.53159618e-01 -3.80242579e-02 -9.19499338e-01 1.73292100e+00
4.79532927e-01 -1.35648221e-01 -1.73035994e-01 8.09333920e-01
6.08637273e-01 5.99311471e-01 -7.44077861e-02 -4.94820178e-01
1.40091980e+00 -1.04727304e+00 -6.04197264e-01 -4.02246386e-01
3.61264288e-01 -5.35727322e-01 1.10213506e+00 5.32597661e-01
-8.58752549e-01 -6.48447812e-01 -1.11107361e+00 2.00285003e-01
-3.33490938e-01 3.79153103e-01 5.16386211e-01 3.51482451e-01
-3.77188891e-01 5.28911948e-01 -6.70894980e-01 8.22586939e-02
8.66855741e-01 2.78307945e-01 -2.89014846e-01 -2.60173410e-01
-8.97989988e-01 5.28295398e-01 5.17831683e-01 3.25083613e-01
-6.39782310e-01 -4.23639745e-01 -7.35775530e-01 1.38779029e-01
6.87750995e-01 -3.87319952e-01 1.00450361e+00 -1.38576400e+00
-1.34420967e+00 6.21663451e-01 2.09237225e-02 -3.33335437e-02
6.93429172e-01 -2.78278310e-02 -2.72878021e-01 -1.49557665e-01
1.18843594e-03 2.86159068e-01 1.22723460e+00 -1.28431273e+00
-7.75297165e-01 -5.03531754e-01 -1.87919214e-01 9.12675858e-02
-6.03155136e-01 -2.62554348e-01 -5.34283221e-01 -6.46527469e-01
8.61336812e-02 -6.17734671e-01 -1.25772700e-01 1.78396687e-01
-5.19052923e-01 -2.89633811e-01 8.94352615e-01 -2.92905778e-01
1.49157000e+00 -2.41615963e+00 -9.12171230e-02 2.58394361e-01
4.46627557e-01 5.90802073e-01 -2.72065178e-02 9.69119519e-02
-2.27267951e-01 -2.36570433e-01 -2.97007322e-01 -3.99046630e-01
-1.08847827e-01 2.69162774e-01 -2.26149082e-01 4.30315852e-01
4.38962370e-01 4.91596937e-01 -8.49732578e-01 -4.51703697e-01
1.44858196e-01 3.76475811e-01 -5.18979788e-01 5.39437473e-01
-1.64712802e-01 2.33502954e-01 -5.51938236e-01 4.40197051e-01
1.07507551e+00 -3.34250927e-01 4.71693501e-02 -3.36778343e-01
-1.66461155e-01 1.81168854e-01 -1.45065868e+00 1.15426946e+00
-2.34011859e-01 2.68078297e-01 1.02618359e-01 -1.37350714e+00
1.09829092e+00 3.28586437e-02 2.09311470e-01 -6.74663186e-01
2.90739596e-01 3.41381699e-01 2.23825112e-01 -5.00291824e-01
6.06954321e-02 -2.16095284e-01 1.70275271e-01 3.48604918e-01
-4.80200499e-02 2.52690881e-01 4.68559191e-02 -1.10247433e-01
7.56214917e-01 4.06320021e-02 4.32724774e-01 1.13144070e-02
7.05736041e-01 -3.78095716e-01 7.63476431e-01 4.79107440e-01
-1.55387089e-01 4.15286392e-01 7.45763898e-01 -5.85301280e-01
-7.77114272e-01 -6.97868586e-01 -1.42019257e-01 1.02135205e+00
5.67313135e-02 -2.29721621e-01 -7.26656973e-01 -9.86257434e-01
-2.58639734e-02 4.89538938e-01 -7.62381852e-01 -4.24314469e-01
-3.38410288e-01 -1.07776165e+00 2.34075859e-01 5.64939499e-01
6.89047098e-01 -1.10044622e+00 -8.06492746e-01 2.00874597e-01
6.70678690e-02 -7.65414596e-01 -2.89106429e-01 5.75669229e-01
-7.68818736e-01 -1.16479874e+00 -8.00592601e-01 -5.73661149e-01
8.23947966e-01 3.41510952e-01 9.16495919e-01 2.54895836e-01
-1.45277724e-01 -4.62412924e-01 -2.97436982e-01 -5.11531174e-01
-2.22765431e-01 2.64008015e-01 -2.09510580e-01 4.85055089e-01
5.48122287e-01 -6.05421841e-01 -7.58441329e-01 4.05348390e-01
-9.60851431e-01 1.11772612e-01 9.42594886e-01 1.04776001e+00
4.57913160e-01 2.62551039e-01 7.83234239e-01 -7.58234501e-01
2.94966787e-01 -4.08909470e-01 -5.93791127e-01 -1.68621223e-02
-5.06584525e-01 -4.20550406e-02 1.07061887e+00 -6.34050667e-01
-7.90860772e-01 1.32707760e-01 -2.37054750e-01 -4.07874107e-01
-1.56650290e-01 4.44412708e-01 -6.80832863e-01 6.54798970e-02
4.29016650e-01 3.61570895e-01 3.25176924e-01 -5.71150303e-01
-1.40064508e-01 1.03923905e+00 2.72848845e-01 -3.84315908e-01
8.12423706e-01 4.00922179e-01 -2.20581725e-01 -7.30174601e-01
-1.11907065e+00 -4.27012384e-01 -4.63309288e-01 -1.38679937e-01
4.71971542e-01 -7.17963994e-01 -5.54668546e-01 8.32828164e-01
-7.76578307e-01 -2.58711457e-01 -3.88372004e-01 3.02387416e-01
-2.51583099e-01 2.68826783e-01 -1.11121103e-01 -9.74751711e-01
-1.66010126e-01 -1.19483709e+00 9.41272974e-01 5.72937906e-01
1.35996357e-01 -6.92202151e-01 -1.80899337e-01 2.29995459e-01
3.77822101e-01 1.85920447e-01 9.34066951e-01 -1.00725377e+00
-4.92248833e-01 -4.71571386e-01 -5.37684023e-01 6.73873663e-01
3.70813280e-01 -2.93253232e-02 -1.22221935e+00 -5.87787092e-01
1.18013717e-01 -5.20662308e-01 8.73792410e-01 3.92678343e-02
1.46725023e+00 -2.48433515e-01 -3.20683837e-01 5.91555357e-01
1.30907774e+00 6.79260343e-02 5.87174058e-01 4.46882308e-01
6.19122744e-01 7.26176381e-01 6.01580560e-01 5.07877052e-01
3.64179581e-01 5.60276747e-01 5.09645402e-01 -2.51323402e-01
9.18996930e-02 -2.43667752e-01 9.01752934e-02 6.04768634e-01
5.59811518e-02 -1.11116469e-01 -6.42735541e-01 3.14740330e-01
-1.72540975e+00 -7.70836532e-01 2.06790447e-01 2.45970893e+00
1.08966219e+00 2.56073147e-01 3.32884908e-01 5.93137980e-01
7.55474746e-01 3.13258201e-01 -6.59782767e-01 3.98648642e-02
7.64392987e-02 -4.24498618e-02 8.03589746e-02 -9.66945663e-03
-1.14222574e+00 5.32070458e-01 4.57182550e+00 1.25006473e+00
-1.52305973e+00 -2.70115733e-01 7.37193942e-01 1.13717742e-01
-1.14043765e-01 -2.52839606e-02 -9.05163527e-01 8.23859274e-01
4.01255697e-01 2.83873171e-01 3.16701531e-01 9.07756746e-01
-4.37140875e-02 -2.06674695e-01 -1.01214588e+00 9.70117807e-01
-1.18317222e-02 -8.87664616e-01 -9.99099389e-02 2.07105726e-01
3.12938571e-01 -2.72930533e-01 -1.11337781e-01 4.63263214e-01
-1.70663565e-01 -8.96475077e-01 6.80846274e-01 2.90918231e-01
7.09202945e-01 -8.81975591e-01 8.25782955e-01 9.52389300e-01
-1.02956223e+00 -2.95811176e-01 -5.42194784e-01 -1.37129510e-02
-2.08882779e-01 1.04205954e+00 -2.40736634e-01 6.24850035e-01
5.76260269e-01 5.72296500e-01 -6.19519293e-01 1.30493820e+00
-1.38673365e-01 4.26435977e-01 -2.62101561e-01 3.74058299e-02
1.83523968e-01 -2.79703975e-01 3.15483272e-01 9.69246387e-01
7.34059885e-02 -2.03366622e-01 4.40092832e-01 6.49748385e-01
4.17499021e-02 2.07206771e-01 -4.03683186e-01 1.03611872e-01
4.31677908e-01 1.38785732e+00 -7.59249806e-01 -4.09721553e-01
-4.86871600e-01 7.19312012e-01 6.21705830e-01 2.72502512e-01
-7.49230564e-01 -5.99803269e-01 2.85600483e-01 1.17927924e-01
4.46233481e-01 2.34942541e-01 -1.54206529e-01 -1.21783423e+00
2.53614545e-01 -1.05625916e+00 3.79097819e-01 -3.13428402e-01
-1.37047124e+00 5.73536277e-01 -1.94645971e-01 -1.46894705e+00
-1.48756266e-01 -6.51084125e-01 -6.80885792e-01 8.70955825e-01
-1.74383819e+00 -9.87181246e-01 -6.80940390e-01 6.00252151e-01
3.13214868e-01 -9.35070440e-02 5.10727167e-01 5.76915860e-01
-9.51120377e-01 8.97624314e-01 8.68725777e-02 1.99718311e-01
6.92790449e-01 -1.12398052e+00 -1.41183630e-01 5.95177650e-01
-2.36206725e-01 5.17451882e-01 4.97086138e-01 -3.29333305e-01
-1.13354552e+00 -1.16956413e+00 5.58332801e-01 7.65352845e-02
4.82797444e-01 -5.27289271e-01 -1.27611148e+00 2.90312648e-01
-1.79061636e-01 3.58790666e-01 7.07123876e-01 2.60732211e-02
-3.29026729e-01 -5.45434177e-01 -1.10253441e+00 4.78283256e-01
5.74924111e-01 -3.32541704e-01 -5.09008884e-01 5.28657027e-02
2.57254779e-01 -1.93762392e-01 -4.97834265e-01 5.00779331e-01
6.37764573e-01 -1.16857600e+00 6.95449054e-01 -3.18651259e-01
4.15077686e-01 -3.89833778e-01 5.84280826e-02 -1.38360882e+00
-2.36077175e-01 -4.67932343e-01 -3.25252384e-01 1.36849821e+00
9.94230434e-02 -8.05549204e-01 9.33458447e-01 2.12877631e-01
4.00990248e-02 -1.25877857e+00 -7.87089944e-01 -6.75936162e-01
-8.22848305e-02 -2.67155021e-01 6.09461129e-01 8.79534185e-01
-3.59323055e-01 4.24152911e-01 -2.79625803e-01 -1.17349371e-01
5.79899728e-01 3.29189122e-01 8.47922981e-01 -1.40703726e+00
-3.32806468e-01 -4.55640852e-01 -2.25878999e-01 -1.11103344e+00
-1.51277274e-01 -5.65428793e-01 2.22229287e-01 -1.11071289e+00
3.76829863e-01 -7.32939899e-01 -2.80504674e-01 4.86184955e-01
-4.70847368e-01 2.47767478e-01 2.26520583e-01 2.20420957e-01
-4.30987298e-01 7.84019828e-01 1.41747022e+00 -7.16485875e-03
-1.73643887e-01 5.31862497e-01 -9.82769132e-01 9.31434512e-01
6.50020719e-01 -4.36613411e-01 -5.71896136e-01 -7.86873922e-02
1.72698155e-01 -1.99380308e-01 5.33884108e-01 -8.94867718e-01
2.51815736e-01 1.33195827e-02 5.78980863e-01 -5.71601868e-01
5.27117774e-02 -9.91645157e-01 -2.12813720e-01 3.37931186e-01
-5.59931397e-02 -2.82922506e-01 4.33218554e-02 5.01950681e-01
-3.09578627e-01 -3.51750672e-01 1.06707728e+00 -1.27441317e-01
-2.37353548e-01 4.91568953e-01 -2.08017424e-01 2.56635211e-02
9.70623732e-01 -4.26741838e-01 -3.19337279e-01 -9.11778510e-02
-3.83487672e-01 4.08745289e-01 5.09018123e-01 1.41085654e-01
4.73100603e-01 -1.28633213e+00 -4.25728142e-01 6.31309986e-01
1.58879846e-01 6.40007198e-01 3.38743865e-01 9.75665808e-01
-5.72865345e-02 -1.67304769e-01 -1.78239509e-01 -5.93843102e-01
-8.82988453e-01 8.06623459e-01 2.19974160e-01 -4.58024919e-01
-5.48688889e-01 6.64997756e-01 3.44061971e-01 -5.21222591e-01
5.12961388e-01 -8.57442170e-02 -4.27783251e-01 6.28481805e-01
7.20223129e-01 2.44130239e-01 4.32786718e-02 -3.42506766e-01
-3.34524900e-01 4.54286307e-01 -3.81037980e-01 3.95399988e-01
1.32511568e+00 1.99941695e-02 -1.37845263e-01 5.16077876e-01
1.36689687e+00 -3.83791630e-03 -1.38193536e+00 -2.47618854e-01
-2.11923510e-01 -5.71237028e-01 1.80881936e-02 -7.01005101e-01
-1.24735057e+00 1.14176869e+00 4.92712975e-01 5.34926295e-01
1.51220012e+00 -2.55439699e-01 6.86205864e-01 1.29017159e-01
1.14313044e-01 -1.11940408e+00 5.00015430e-02 3.43517810e-01
5.71986258e-01 -1.27112019e+00 -5.63745573e-02 -3.44349533e-01
-4.78005081e-01 1.06924617e+00 7.89940953e-01 -2.20962569e-01
4.98804688e-01 2.29270786e-01 1.41729444e-01 -1.78427786e-01
-6.53347313e-01 -4.65768240e-02 2.62631297e-01 3.72304082e-01
2.08028316e-01 -1.60177141e-01 -1.82811648e-01 7.57573783e-01
1.14756888e-02 8.46116617e-02 2.42310628e-01 7.69487321e-01
-5.25148928e-01 -1.02856886e+00 -2.24398077e-01 7.07014263e-01
-3.61186445e-01 6.53997436e-02 -4.72106785e-02 7.88626552e-01
3.94993663e-01 5.82625687e-01 1.80210263e-01 -3.45992088e-01
4.08537596e-01 3.37402076e-02 1.32671595e-01 -4.69833374e-01
-4.73912686e-01 2.88766056e-01 -2.30781823e-01 -5.35112798e-01
-4.22090471e-01 -4.67573106e-01 -8.14625144e-01 -4.85101789e-02
-6.59939706e-01 2.10440978e-01 3.66794050e-01 1.06346309e+00
1.46690503e-01 7.44596720e-01 9.78730321e-01 -9.17494833e-01
-1.02660418e+00 -9.62966323e-01 -6.89607620e-01 3.12350780e-01
5.58242738e-01 -7.61698186e-01 -5.60731828e-01 -3.77052277e-01] | [9.380597114562988, 3.4129552841186523] |
6b7864fe-6d62-40a4-b728-c740c3cfaa5d | fully-dnn-based-multi-label-regression-for | 1606.07695 | null | http://arxiv.org/abs/1606.07695v2 | http://arxiv.org/pdf/1606.07695v2.pdf | Fully DNN-based Multi-label regression for audio tagging | Acoustic event detection for content analysis in most cases relies on lots of
labeled data. However, manually annotating data is a time-consuming task, which
thus makes few annotated resources available so far. Unlike audio event
detection, automatic audio tagging, a multi-label acoustic event classification
task, only relies on weakly labeled data. This is highly desirable to some
practical applications using audio analysis. In this paper we propose to use a
fully deep neural network (DNN) framework to handle the multi-label
classification task in a regression way. Considering that only chunk-level
rather than frame-level labels are available, the whole or almost whole frames
of the chunk were fed into the DNN to perform a multi-label regression for the
expected tags. The fully DNN, which is regarded as an encoding function, can
well map the audio features sequence to a multi-tag vector. A deep pyramid
structure was also designed to extract more robust high-level features related
to the target tags. Further improved methods were adopted, such as the Dropout
and background noise aware training, to enhance its generalization capability
for new audio recordings in mismatched environments. Compared with the
conventional Gaussian Mixture Model (GMM) and support vector machine (SVM)
methods, the proposed fully DNN-based method could well utilize the long-term
temporal information with the whole chunk as the input. The results show that
our approach obtained a 15% relative improvement compared with the official
GMM-based method of DCASE 2016 challenge. | ['Philip J. B. Jackson', 'Wenwu Wang', 'Yong Xu', 'Qiang Huang', 'Mark D. Plumbley'] | 2016-06-24 | null | null | null | null | ['audio-tagging'] | ['audio'] | [ 2.46903747e-01 -2.33682796e-01 2.26391330e-01 -3.48878860e-01
-1.06823850e+00 -2.33889073e-01 1.61577344e-01 2.94497967e-01
-7.12235153e-01 5.99497139e-01 1.71220273e-01 1.31905854e-01
1.66495532e-01 -6.00217164e-01 -5.13556778e-01 -9.59783792e-01
9.80028324e-03 2.19775201e-03 5.53459346e-01 1.35868862e-01
-1.11324921e-01 2.31106490e-01 -1.90815604e+00 4.85179782e-01
5.45548379e-01 1.38169062e+00 4.43330079e-01 4.82654452e-01
-2.42186278e-01 7.97467768e-01 -5.40899813e-01 -1.18689008e-01
-1.46025285e-01 -2.41378456e-01 -4.95255738e-01 -5.22472896e-02
8.64996165e-02 -2.15695575e-01 -2.04836801e-01 1.10754907e+00
8.55355918e-01 4.51877326e-01 5.12162507e-01 -1.07347429e+00
7.24095404e-02 6.50689781e-01 -3.78603697e-01 2.06650600e-01
2.37412959e-01 -2.82900214e-01 8.45290601e-01 -1.01713717e+00
2.62398690e-01 9.43675935e-01 8.20725501e-01 3.25125813e-01
-5.22829592e-01 -9.99695420e-01 2.43102804e-01 7.93483734e-01
-1.56681406e+00 -4.41337228e-01 9.95941341e-01 -4.11290407e-01
6.67905748e-01 2.22286269e-01 5.95511675e-01 1.14317310e+00
-1.75473496e-01 7.65630066e-01 9.18406725e-01 -4.03497517e-01
2.03232497e-01 1.55216366e-01 -9.90815368e-03 3.26566160e-01
-4.82876986e-01 -1.59787461e-01 -5.48425496e-01 -1.43756434e-01
3.86187911e-01 -4.39443029e-02 -3.30595285e-01 1.22210026e-01
-1.14260387e+00 5.58321476e-01 -1.90524210e-03 5.49248219e-01
-4.24008787e-01 -5.64049697e-03 8.78287733e-01 2.64031943e-02
7.24863887e-01 -7.79490992e-02 -6.13301039e-01 -5.05054474e-01
-1.05320060e+00 -1.10597432e-01 6.35217428e-01 6.88124239e-01
6.93397701e-01 3.49178106e-01 -2.08991870e-01 1.14092290e+00
2.90923566e-01 2.53319800e-01 6.92784429e-01 -7.20437706e-01
4.35677856e-01 2.98750937e-01 -3.07589415e-02 -9.18204784e-01
-6.75880134e-01 -7.74820030e-01 -8.73431444e-01 -1.09335832e-01
1.69743136e-01 -3.61863673e-01 -7.21012294e-01 1.77836967e+00
5.35578966e-01 7.03473568e-01 -1.01666294e-01 9.05152261e-01
8.78824532e-01 1.01427877e+00 2.89056391e-01 -5.15674949e-01
1.38471711e+00 -8.77689540e-01 -9.59688663e-01 -2.15787478e-02
6.78856611e-01 -1.03534043e+00 8.49341273e-01 5.70470810e-01
-7.86001027e-01 -9.22173798e-01 -9.55249965e-01 2.84037352e-01
-2.65185118e-01 3.63915563e-01 2.47581452e-01 5.85583448e-01
-7.70482481e-01 4.75931466e-01 -7.97246635e-01 -2.19459996e-01
2.09409326e-01 3.62536192e-01 -3.33113760e-01 1.08289219e-01
-1.53716028e+00 6.42867863e-01 7.56060421e-01 4.74331141e-01
-1.00899148e+00 -3.67457151e-01 -8.40564728e-01 2.21313685e-01
4.42296445e-01 1.21506713e-02 1.19915736e+00 -8.17215025e-01
-1.55709589e+00 4.51279342e-01 -9.26465169e-02 -3.94542664e-01
2.45721832e-01 -2.42034197e-01 -7.83191442e-01 2.22023264e-01
-1.88836426e-01 4.84097332e-01 8.22326958e-01 -9.28560317e-01
-1.02634430e+00 -1.61338709e-02 -2.82800466e-01 3.54840308e-01
-7.80887187e-01 2.91542292e-01 -3.63463283e-01 -8.78368795e-01
9.28412303e-02 -7.84066617e-01 9.12136585e-02 -4.56777692e-01
-1.92261338e-01 -3.74741614e-01 9.14829254e-01 -9.60935235e-01
1.43720973e+00 -2.56946039e+00 -4.86902259e-02 -1.14968844e-01
-1.21701531e-01 3.79613638e-01 -3.32775223e-03 3.15076441e-01
-1.03633650e-01 -3.61808002e-01 6.25330135e-02 -5.24823487e-01
3.31712253e-02 -4.63913865e-02 -2.41835371e-01 5.03903508e-01
8.18675980e-02 2.60847956e-01 -8.33821774e-01 -8.09124351e-01
3.13079864e-01 5.09955347e-01 -4.15931284e-01 4.12059069e-01
2.38919836e-02 5.92628241e-01 -1.69316143e-01 5.85377336e-01
6.93499327e-01 2.86167234e-01 -2.14065045e-01 -5.35498142e-01
-1.84529856e-01 1.48920611e-01 -1.53906333e+00 1.83800864e+00
-6.37829244e-01 4.39410537e-01 2.03073725e-01 -1.30602849e+00
1.07430422e+00 9.10622001e-01 7.14576721e-01 -3.72038275e-01
2.33478650e-01 2.66071230e-01 -9.15904567e-02 -7.07509577e-01
3.32516313e-01 -2.92739481e-01 -2.17881039e-01 2.78171692e-02
3.75586480e-01 3.31727207e-01 -7.79466331e-02 -2.45914131e-01
1.11427832e+00 1.70507148e-01 5.80705740e-02 6.70415908e-02
7.03523695e-01 -2.51339972e-01 1.09729707e+00 5.25022626e-01
-2.73895919e-01 5.72707295e-01 2.56823730e-02 -2.94199288e-01
-7.79299319e-01 -7.41540790e-01 -2.47671425e-01 1.38094544e+00
-4.16141469e-03 -4.67138261e-01 -6.52139783e-01 -6.08102143e-01
-4.76901203e-01 5.45397520e-01 -1.84870452e-01 -3.01791877e-01
-5.72987616e-01 -8.43776405e-01 7.88385272e-01 4.52587098e-01
6.75532103e-01 -1.31165671e+00 -3.18011701e-01 7.48387337e-01
-5.74858963e-01 -1.30440283e+00 -3.72297049e-01 7.07392454e-01
-5.16562223e-01 -6.79122150e-01 -8.04342926e-01 -9.74973857e-01
3.29182073e-02 -1.70438260e-01 4.49770570e-01 -3.61952007e-01
-1.17783388e-02 -1.67351849e-02 -7.86962926e-01 -3.06177795e-01
-3.61136377e-01 1.13659225e-01 2.23190233e-01 4.64692712e-01
4.40893233e-01 -8.17904770e-01 -4.72304016e-01 4.10926372e-01
-8.00768435e-01 -5.70497699e-02 5.58123589e-01 8.97895336e-01
5.49791157e-01 3.31261575e-01 1.05584335e+00 -4.56774831e-01
3.10017854e-01 -3.99629682e-01 -2.94483155e-01 4.66735065e-02
-3.27120721e-02 -3.45073223e-01 8.08122993e-01 -8.76138389e-01
-1.20281303e+00 3.35621625e-01 -7.14881361e-01 -4.59494531e-01
-4.73253757e-01 3.95757079e-01 -5.40790319e-01 1.86967626e-01
3.47378731e-01 3.62692624e-01 -3.89008075e-01 -7.80467749e-01
5.33484221e-02 1.17918146e+00 5.23040950e-01 -3.98413807e-01
3.85167003e-01 7.22236782e-02 -1.38986051e-01 -7.04743266e-01
-9.52992499e-01 -7.50388563e-01 -6.30832732e-01 -4.63061392e-01
1.02938700e+00 -1.21143866e+00 -6.78663373e-01 7.22397387e-01
-1.21827316e+00 -7.67463371e-02 -1.01812422e-01 9.65157092e-01
-4.13885951e-01 4.04086798e-01 -8.37598503e-01 -1.02714503e+00
-1.64146349e-01 -1.13568747e+00 1.06133199e+00 9.78336707e-02
-8.36747438e-02 -6.70444489e-01 3.10520325e-02 2.75972724e-01
2.26374120e-01 1.14657171e-02 7.19203770e-01 -1.17220664e+00
-4.63281237e-02 -2.59691417e-01 -4.67935205e-02 7.78641164e-01
6.43197522e-02 -3.35535467e-01 -1.41696370e+00 -1.78167000e-01
3.64541471e-01 -4.33823556e-01 7.93009937e-01 2.50219524e-01
1.30654979e+00 -5.95653467e-02 -1.93165272e-01 4.22997534e-01
1.07883310e+00 5.14715612e-01 3.86985302e-01 1.74810067e-01
9.26478863e-01 4.58785862e-01 1.04375672e+00 6.83146536e-01
2.19121233e-01 7.94626713e-01 4.55853343e-01 -5.85302003e-02
-3.51968817e-02 -1.45392101e-02 5.41774035e-01 1.50207031e+00
7.92642906e-02 -3.41553092e-01 -7.22834051e-01 4.75114644e-01
-1.85150838e+00 -8.49247575e-01 -8.59901831e-02 2.10496163e+00
9.97829378e-01 1.82175636e-01 -1.25946179e-02 5.95001101e-01
1.16645169e+00 6.09069020e-02 -3.34742576e-01 -1.00641116e-01
-4.18182984e-02 2.74987519e-01 1.79341555e-01 4.95693609e-02
-1.62349725e+00 6.25309706e-01 4.85028219e+00 1.60170889e+00
-1.03080368e+00 6.07139468e-01 2.13874951e-01 -1.03921086e-01
2.56344676e-01 -1.98736042e-01 -9.65435863e-01 7.42715001e-01
1.25773919e+00 3.69349808e-01 7.11216256e-02 7.46460855e-01
4.47312772e-01 -2.23975684e-02 -9.01706994e-01 1.17806339e+00
-7.79211447e-02 -8.64689469e-01 -3.36078346e-01 -6.82963058e-02
3.76159042e-01 -1.80621848e-01 -1.22530051e-01 5.98303139e-01
-1.66411400e-01 -5.81210852e-01 8.16748798e-01 4.29947197e-01
6.86258078e-01 -9.15390730e-01 1.13697481e+00 4.89534825e-01
-1.53821003e+00 -1.47311419e-01 -4.98090953e-01 2.77396813e-02
3.86003286e-01 9.64619040e-01 -6.95915520e-01 6.87996566e-01
8.09916675e-01 6.92434907e-01 -3.12397182e-01 1.27306116e+00
4.49291989e-02 1.09056592e+00 -5.08749485e-01 1.40711248e-01
2.01947600e-01 1.79017514e-01 4.38944310e-01 1.51021516e+00
7.38357663e-01 -1.84262067e-01 4.36631680e-01 2.30307579e-01
3.17708328e-02 5.36492586e-01 -2.24042162e-01 2.95258790e-01
5.56966484e-01 1.51089704e+00 -7.29755402e-01 -2.83587605e-01
-5.31556487e-01 9.27257061e-01 1.47387818e-01 9.88576710e-02
-1.15796959e+00 -5.58130026e-01 2.32737392e-01 -1.64156079e-01
4.17230666e-01 6.46121474e-03 2.19277069e-01 -9.92556453e-01
4.26286645e-03 -7.33213544e-01 4.41081733e-01 -7.18811750e-01
-1.02787364e+00 7.65185654e-01 -1.93451658e-01 -1.70150411e+00
-1.33064225e-01 -1.82504058e-01 -4.81835783e-01 7.31578350e-01
-1.35117114e+00 -1.10967672e+00 -3.26242834e-01 6.55251503e-01
5.46104610e-01 -1.73728898e-01 1.00773847e+00 1.08166325e+00
-6.51189506e-01 6.64667487e-01 1.74254835e-01 1.15697429e-01
1.09373510e+00 -9.90696669e-01 -4.08994287e-01 6.82927251e-01
2.62186885e-01 -6.55409787e-03 5.10094404e-01 -5.49422681e-01
-7.34664023e-01 -1.35066497e+00 7.52326250e-01 2.09679872e-01
6.92935407e-01 -3.77793014e-01 -1.10215735e+00 3.84931475e-01
1.71207577e-01 1.09356411e-01 8.64703476e-01 -8.71576592e-02
1.42186165e-01 -2.98219770e-01 -8.18200529e-01 7.57870898e-02
8.34022522e-01 -6.13104701e-01 -5.59744120e-01 2.74700761e-01
8.57279539e-01 -3.47560793e-01 -9.75570142e-01 6.73574388e-01
2.78718740e-01 -7.73965716e-01 7.65710890e-01 -1.02311820e-01
7.89317861e-02 -4.98657644e-01 -4.15097237e-01 -1.28819454e+00
-2.22854719e-01 -3.38797092e-01 -6.84114397e-02 1.85127866e+00
1.23104431e-01 -1.68699354e-01 5.57356000e-01 -9.21861231e-02
-5.76071918e-01 -6.08271658e-01 -1.36048436e+00 -7.49047995e-01
-3.89459044e-01 -8.11384559e-01 3.84106487e-01 7.84448147e-01
-3.79903466e-02 2.39225149e-01 -7.47810423e-01 3.77264202e-01
3.52251917e-01 -1.92727759e-01 1.76164344e-01 -1.29846931e+00
-2.09071249e-01 -1.00288264e-01 -7.06721663e-01 -8.05694520e-01
4.34564173e-01 -8.36481214e-01 3.20818454e-01 -1.18197596e+00
-1.42733917e-01 -4.74175364e-01 -8.30547869e-01 4.54549283e-01
-1.63152456e-01 4.36186522e-01 2.95807049e-03 1.07485704e-01
-7.53219128e-01 7.26142824e-01 8.58848274e-01 -1.85095102e-01
-1.89669594e-01 4.18339312e-01 -1.34706095e-01 8.79189491e-01
7.56085873e-01 -7.41595387e-01 -2.38167778e-01 -9.76666212e-02
-1.70866568e-02 2.72147715e-01 2.18262792e-01 -1.44848979e+00
4.49903041e-01 1.20139472e-01 3.31428379e-01 -8.18822086e-01
7.22960174e-01 -9.98629987e-01 2.85941929e-01 2.16409370e-01
-3.10221225e-01 -2.88678616e-01 2.69603252e-01 5.70358872e-01
-8.03930104e-01 -5.44916272e-01 6.22318745e-01 1.08185457e-02
-8.67820084e-01 2.29483098e-01 -6.14456177e-01 -1.51205510e-01
9.66712296e-01 -1.47273140e-02 9.21477377e-02 -2.98650742e-01
-1.17509925e+00 -2.50277482e-02 -1.89311698e-01 2.46342525e-01
4.61902887e-01 -1.64947188e+00 -6.25972033e-01 -1.16091795e-01
1.10790916e-01 3.08454428e-02 7.78532326e-01 8.56162965e-01
-1.94339957e-02 7.35142231e-02 -3.33359689e-02 -6.23057008e-01
-1.18089449e+00 5.74906647e-01 1.04056232e-01 -2.94808000e-01
-4.59963202e-01 8.63248348e-01 1.51315466e-01 -2.40316704e-01
6.27317488e-01 -2.68157870e-01 -6.04033232e-01 5.95137417e-01
4.88919258e-01 3.71857345e-01 2.92669028e-01 -7.77911782e-01
-4.63786811e-01 5.38011074e-01 2.14206100e-01 -1.49029925e-01
1.32812440e+00 -2.85031557e-01 -4.12729010e-02 7.66692400e-01
1.39948940e+00 2.48967372e-02 -1.06757164e+00 -3.01398486e-01
-8.53503719e-02 2.98668072e-02 2.85625517e-01 -4.74958479e-01
-1.02424610e+00 1.23076451e+00 8.49886775e-01 1.78569004e-01
1.38325047e+00 -9.82623920e-02 9.49951470e-01 2.13767573e-01
2.12538272e-01 -1.30093324e+00 -6.95752427e-02 4.13901180e-01
5.12380838e-01 -9.80571747e-01 -3.42404336e-01 -3.15834641e-01
-5.50000250e-01 1.26934516e+00 6.26820445e-01 1.58250287e-01
8.67534041e-01 3.77602190e-01 7.68600404e-02 2.24198982e-01
-5.32455146e-01 -3.44364285e-01 2.30719373e-01 5.25204062e-01
1.70521110e-01 -1.67268544e-01 -9.70985219e-02 1.05569577e+00
3.91160771e-02 -4.03093547e-02 2.65673369e-01 7.59939849e-01
-6.49436355e-01 -1.14381182e+00 -4.98184055e-01 4.20249492e-01
-6.80161715e-01 -7.34134763e-02 4.38119560e-01 3.99926096e-01
6.60888851e-01 1.13759434e+00 1.21450452e-02 -7.80220270e-01
3.15806121e-01 4.84447092e-01 3.71533409e-02 -6.57837868e-01
-7.55585432e-01 5.94821393e-01 6.47605062e-02 -3.40786576e-01
-7.20207691e-01 -6.89965248e-01 -1.21317673e+00 2.06818730e-01
-6.08021557e-01 2.34957799e-01 5.53351879e-01 1.04682910e+00
8.33386704e-02 8.34848106e-01 7.63528824e-01 -9.39989924e-01
-3.80547583e-01 -1.31367648e+00 -7.70003796e-01 3.15506607e-01
1.89440042e-01 -7.05848873e-01 -4.49815780e-01 2.42263258e-01] | [15.212989807128906, 5.165895462036133] |
621628dd-abe1-4926-878f-0950580207e7 | deep-graph-kernel-point-processes | 2306.11313 | null | https://arxiv.org/abs/2306.11313v1 | https://arxiv.org/pdf/2306.11313v1.pdf | Deep graph kernel point processes | Point process models are widely used to analyze asynchronous events occurring within a graph that reflect how different types of events influence one another. Predicting future events' times and types is a crucial task, and the size and topology of the graph add to the challenge of the problem. Recent neural point process models unveil the possibility of capturing intricate inter-event-category dependencies. However, such methods utilize an unfiltered history of events, including all event categories in the intensity computation for each target event type. In this work, we propose a graph point process method where event interactions occur based on a latent graph topology. The corresponding undirected graph has nodes representing event categories and edges indicating potential contribution relationships. We then develop a novel deep graph kernel to characterize the triggering and inhibiting effects between events. The intrinsic influence structures are incorporated via the graph neural network (GNN) model used to represent the learnable kernel. The computational efficiency of the GNN approach allows our model to scale to large graphs. Comprehensive experiments on synthetic and real-world data show the superior performance of our approach against the state-of-the-art methods in predicting future events and uncovering the relational structure among data. | ['Yao Xie', 'Xiuyuan Cheng', 'Matthew Repasky', 'Zheng Dong'] | 2023-06-20 | null | null | null | null | ['point-processes'] | ['methodology'] | [ 2.04074327e-02 -2.97948569e-02 -9.24707428e-02 -3.46489608e-01
1.47307500e-01 -5.57125211e-01 1.07608902e+00 8.81741941e-01
-1.99497174e-02 4.00387734e-01 4.85449433e-01 -3.06032121e-01
-4.75582093e-01 -1.43679750e+00 -7.62591422e-01 -5.40313363e-01
-8.93314302e-01 6.01255894e-01 4.24919456e-01 5.05239628e-02
6.29666150e-02 5.24342418e-01 -9.80815709e-01 2.79094994e-01
5.20141542e-01 6.30860686e-01 -2.44306147e-01 6.89729512e-01
-1.94171384e-01 1.39416862e+00 -6.42490745e-01 -2.95428187e-01
-2.71004029e-02 -2.48816103e-01 -4.87060964e-01 -1.26114950e-01
-2.78896004e-01 2.50859205e-02 -1.04248238e+00 5.97324908e-01
2.77244240e-01 1.84151202e-01 8.16637039e-01 -1.42417359e+00
-4.44495976e-01 1.19192696e+00 -6.48330331e-01 7.10311830e-01
1.46332428e-01 4.84388620e-02 1.16118705e+00 -6.51012421e-01
5.65760553e-01 1.23268187e+00 7.05423772e-01 -6.83986470e-02
-1.23006725e+00 -6.15022302e-01 5.54880738e-01 3.17181200e-01
-1.25263405e+00 9.54182595e-02 1.07414901e+00 -4.06139702e-01
1.07865179e+00 3.49283405e-02 9.38266277e-01 1.28172076e+00
7.10626543e-01 6.81245446e-01 5.47208548e-01 3.87744643e-02
3.48125875e-01 -7.05112040e-01 4.38623369e-01 6.78531528e-01
2.95982182e-01 2.58179177e-02 -7.92526007e-01 -4.62158293e-01
9.56054926e-01 7.50506580e-01 -2.46684235e-02 -1.48760170e-01
-1.23722982e+00 7.40887761e-01 4.84263152e-01 1.69443041e-01
-5.97756982e-01 5.16051888e-01 4.13418889e-01 2.84889817e-01
8.42875361e-01 3.85159515e-02 -4.47330356e-01 -1.17858596e-01
-6.09848082e-01 1.57912821e-02 1.18439698e+00 6.72945678e-01
7.41564691e-01 -1.44322857e-01 -4.33580011e-01 4.42044616e-01
4.34442610e-01 1.74139977e-01 -1.38798401e-01 -2.46401474e-01
5.65028369e-01 1.23170984e+00 -2.70544857e-01 -1.51158535e+00
-6.46809340e-01 -4.99292016e-01 -1.25407517e+00 -3.80196512e-01
5.41598678e-01 -1.34129107e-01 -8.18078458e-01 1.74790621e+00
3.85050416e-01 9.80457008e-01 -2.79567629e-01 3.34630728e-01
5.19156575e-01 1.07921493e+00 3.76727372e-01 -2.39598483e-01
1.27503443e+00 -4.45588052e-01 -7.04389513e-01 -1.98516488e-01
3.08115661e-01 -1.55959651e-01 5.20942450e-01 3.41155343e-02
-9.00224686e-01 -2.38519192e-01 -5.79582930e-01 3.12958598e-01
-3.84653181e-01 -3.30277264e-01 1.35826874e+00 9.08443183e-02
-8.63572896e-01 6.75056100e-01 -1.22835040e+00 -2.75853515e-01
4.89429086e-01 4.77969944e-01 -1.21905223e-01 2.11659521e-01
-1.60695505e+00 3.82248223e-01 4.35619414e-01 3.03412139e-01
-1.25569618e+00 -7.91593432e-01 -6.22896492e-01 6.85549676e-01
6.14045680e-01 -6.85423136e-01 8.06600749e-01 -5.93902707e-01
-1.06347275e+00 3.41095746e-01 -1.75382927e-01 -4.62499827e-01
1.29277945e-01 1.69765681e-01 -6.20295048e-01 1.88265875e-01
-6.14512973e-02 -1.27557769e-01 7.76098013e-01 -7.53148913e-01
-6.88567758e-01 -2.76054353e-01 1.00544449e-02 1.30478352e-01
-3.11300546e-01 9.87876905e-04 -5.30782521e-01 -6.70462132e-01
1.15966974e-02 -6.94697022e-01 -3.31009299e-01 -4.06896740e-01
-3.43565106e-01 -7.77440429e-01 6.44674718e-01 -2.99825937e-01
1.65095139e+00 -1.98614705e+00 1.36135787e-01 4.25805628e-01
7.93504894e-01 -2.82168865e-01 3.48771140e-02 1.13754010e+00
-8.40846300e-02 -4.73450422e-02 4.14125063e-02 -1.33162275e-01
-4.41716239e-02 2.30654702e-01 -6.03459895e-01 4.22404110e-01
3.83457005e-01 1.05602765e+00 -1.14287257e+00 -2.69869179e-01
6.64837584e-02 6.93009257e-01 -1.85752422e-01 2.51258731e-01
-2.48513460e-01 3.31101418e-01 -6.50431752e-01 3.80377740e-01
2.52684712e-01 -7.65487313e-01 3.89418840e-01 3.93120348e-02
1.94401622e-01 3.06129277e-01 -1.24359775e+00 1.32914042e+00
-4.36486416e-02 4.03699011e-01 -4.67829794e-01 -1.09597576e+00
8.71912003e-01 4.66401994e-01 6.08958244e-01 -2.54276872e-01
-7.91701227e-02 -2.67538846e-01 5.14798500e-02 -5.37768416e-02
1.04542412e-01 2.32781041e-02 -2.64407963e-01 6.71344042e-01
1.32565051e-01 3.90194505e-01 4.41024333e-01 6.48137867e-01
1.83886409e+00 -3.87369990e-01 2.63936460e-01 -1.57886863e-01
1.67903751e-01 -5.21186709e-01 7.16985226e-01 1.01789367e+00
-1.16072312e-01 1.31770939e-01 1.32490385e+00 -8.77174497e-01
-7.43456602e-01 -1.52656281e+00 2.81130940e-01 1.12425041e+00
-4.72526625e-02 -5.41708827e-01 -2.94338077e-01 -7.71781802e-01
-1.49849067e-02 4.34531868e-01 -8.80641818e-01 -3.56565595e-01
-7.54759669e-01 -1.20644283e+00 5.17234564e-01 6.88759625e-01
1.54243633e-01 -1.34720361e+00 6.70401305e-02 5.80308020e-01
-1.71322361e-01 -9.49434817e-01 -4.48458821e-01 2.26995364e-01
-9.18125808e-01 -1.13056004e+00 7.90089462e-03 -6.47796810e-01
5.78168154e-01 -1.90474510e-01 1.34644961e+00 -1.60277724e-01
-1.27079144e-01 5.05964756e-01 -9.20479968e-02 -5.08333683e-01
-2.06109509e-01 6.50303662e-02 -1.06372140e-01 5.25132537e-01
5.08308649e-01 -1.11913145e+00 -8.49535406e-01 1.45785913e-01
-1.11871541e+00 7.29845138e-03 4.40072000e-01 4.85462010e-01
4.30984378e-01 5.79208910e-01 6.86499715e-01 -1.09274483e+00
7.37483382e-01 -1.01584637e+00 -5.81551909e-01 2.48744771e-01
-5.25895953e-01 7.75187537e-02 7.84004390e-01 -7.42425919e-01
-1.07222986e+00 -1.50815442e-01 5.69673479e-01 -2.93797582e-01
2.84219272e-02 7.78236032e-01 -1.46436006e-01 5.17301917e-01
3.07885498e-01 2.66167074e-01 -4.73724842e-01 1.67457029e-01
3.41752708e-01 -6.21800125e-02 1.99659273e-01 -5.13430357e-01
5.62735438e-01 9.19820786e-01 4.32329714e-01 -5.22717714e-01
-6.81175947e-01 -4.66522843e-01 -4.36692506e-01 -3.89916301e-01
5.42104959e-01 -8.17026496e-01 -9.94762719e-01 6.30055368e-01
-1.30420613e+00 -3.44366729e-01 -2.68677980e-01 4.03370440e-01
-3.36271405e-01 1.37509257e-01 -1.35412407e+00 -7.25699008e-01
-3.26295137e-01 -6.28494680e-01 9.73982692e-01 1.50800109e-01
-1.27331063e-01 -1.71993506e+00 4.98651534e-01 -3.25512975e-01
1.43726051e-01 5.16751051e-01 9.76792574e-01 -7.65676200e-01
-8.12780261e-01 -4.27593946e-01 -2.09389567e-01 -5.00988781e-01
2.44239450e-01 5.45637310e-02 -4.48867351e-01 -1.20859616e-01
-5.66202849e-02 4.45351511e-01 9.28967118e-01 6.41160131e-01
1.06549132e+00 -2.67340451e-01 -7.41312146e-01 5.22812426e-01
1.16126764e+00 6.51143193e-02 5.44063628e-01 -1.00875154e-01
1.16714978e+00 5.71532607e-01 -8.52826834e-02 6.49302363e-01
6.77595198e-01 1.21365011e-01 4.23989117e-01 -1.35283947e-01
2.37669870e-01 -7.49857128e-01 3.41667593e-01 9.12101865e-01
-2.27290198e-01 -7.77469158e-01 -1.23227787e+00 4.24207091e-01
-2.27703786e+00 -1.31494892e+00 -5.90120256e-01 2.00221014e+00
6.90673113e-01 4.13277090e-01 -1.32414252e-01 1.26410518e-02
9.75357294e-01 5.80610752e-01 -4.17321503e-01 -4.40417342e-02
-1.48862362e-01 1.80961967e-01 3.10276061e-01 3.97666186e-01
-1.24001884e+00 8.25938225e-01 6.02291870e+00 6.30985320e-01
-8.61592233e-01 -1.64884552e-01 6.63392544e-01 7.98512250e-02
-2.82842994e-01 2.07869500e-01 -7.55680501e-01 4.90011185e-01
1.34781265e+00 -2.43018225e-01 3.98414999e-01 1.95379362e-01
4.67222273e-01 2.21188799e-01 -1.30705714e+00 6.22994423e-01
-2.10168943e-01 -1.42874730e+00 2.94307470e-01 4.47739996e-02
6.23778880e-01 1.03179790e-01 -2.39840046e-01 3.71280104e-01
8.41036260e-01 -8.87308180e-01 2.96829998e-01 9.65732515e-01
2.56896853e-01 -8.93446386e-01 2.87265956e-01 3.33820492e-01
-1.72266507e+00 -7.79887736e-02 -2.60735214e-01 -4.72630292e-01
3.92610341e-01 1.06277263e+00 -1.00513124e+00 5.55536985e-01
5.28261244e-01 1.33057380e+00 -4.94867951e-01 6.59431398e-01
-4.78609145e-01 1.27104020e+00 -4.53167528e-01 -1.22737519e-01
8.48579109e-02 -1.85536504e-01 6.99215531e-01 1.27509642e+00
6.65106773e-02 3.34988013e-02 2.55792320e-01 8.29646707e-01
-2.32124507e-01 -2.10978299e-01 -7.76695848e-01 -3.90208423e-01
4.18269306e-01 1.36187792e+00 -1.28019392e+00 -3.65910619e-01
-5.44531167e-01 7.97037959e-01 7.25942671e-01 7.12688088e-01
-8.21107626e-01 -1.77203700e-01 4.65367466e-01 2.72569060e-01
2.63341367e-01 -3.43220741e-01 -1.38392717e-01 -1.37293756e+00
-1.54592052e-01 -4.49878305e-01 9.10268247e-01 -4.42305654e-01
-1.84871662e+00 2.80489832e-01 -3.51754837e-02 -9.73383784e-01
-3.72223742e-02 -2.75762379e-01 -1.12157106e+00 7.38858938e-01
-1.15956879e+00 -1.25962949e+00 -1.14950612e-01 8.00117791e-01
2.33989745e-01 7.51686245e-02 5.42897046e-01 1.68175921e-01
-6.53234363e-01 -1.51605951e-02 -1.63453966e-01 4.50788140e-01
3.65445733e-01 -1.49520445e+00 6.82895780e-01 9.19321418e-01
4.49222863e-01 6.44322038e-01 4.27900106e-01 -1.18262255e+00
-1.61806464e+00 -1.23845649e+00 9.12715614e-01 -8.23149979e-01
1.35288382e+00 -7.45718479e-01 -1.10841811e+00 1.07873476e+00
2.24773437e-02 2.14727744e-01 6.91455007e-01 5.08549213e-01
-2.96699166e-01 -2.09933385e-01 -3.97612810e-01 7.48110592e-01
1.23388493e+00 -5.45310438e-01 -4.66569155e-01 2.81390637e-01
7.23279119e-01 -1.01945456e-02 -8.23072374e-01 3.39745462e-01
1.56658262e-01 -5.09679556e-01 9.30925071e-01 -7.08247483e-01
5.87947428e-01 -3.21397036e-01 5.56516707e-01 -1.31189966e+00
-7.12611377e-01 -7.68215179e-01 -8.21206808e-01 1.30294180e+00
3.31034511e-01 -6.83284044e-01 7.29425013e-01 4.22790885e-01
2.38356590e-01 -5.80161691e-01 -6.85656667e-01 -1.68905422e-01
-2.74981618e-01 -6.18725538e-01 5.91186404e-01 1.19883955e+00
1.80595845e-01 8.31238806e-01 -3.53244007e-01 6.21492565e-01
7.22292840e-01 1.58066332e-01 4.94930297e-01 -1.59814453e+00
-5.04412532e-01 -3.83220583e-01 -5.61472833e-01 -6.05564952e-01
8.26575533e-02 -1.18018162e+00 -4.38126266e-01 -1.66769254e+00
4.44345891e-01 -2.35082924e-01 -6.47185922e-01 4.18957829e-01
-6.78731084e-01 -4.15698618e-01 -2.00051218e-01 3.42281520e-01
-8.31721306e-01 5.56917489e-01 9.39657509e-01 -1.29523948e-01
-3.69881272e-01 1.39117911e-01 -2.78857499e-01 7.63966799e-01
7.72226751e-01 -6.02845669e-01 -6.44204259e-01 -3.83684695e-01
7.88652420e-01 4.59870607e-01 4.44388300e-01 -7.09394872e-01
5.75187087e-01 -1.35414019e-01 4.66591001e-01 -5.82618773e-01
8.86915848e-02 -7.03719139e-01 4.65089351e-01 2.02431202e-01
-3.42146605e-01 2.48260528e-01 -3.51092555e-02 1.40674031e+00
-2.94180572e-01 5.27433932e-01 1.84103306e-02 1.10249140e-03
-5.17305970e-01 1.00374186e+00 -4.65479970e-01 7.24823922e-02
1.08709514e+00 2.00170338e-01 -2.89563745e-01 -6.28459215e-01
-8.25538516e-01 3.23730528e-01 8.94007797e-04 3.87710005e-01
6.44426823e-01 -1.22358656e+00 -8.05439830e-01 8.09653848e-02
4.37378362e-02 -9.14587546e-03 3.20133746e-01 8.00701678e-01
-3.56832772e-01 -1.12275958e-01 1.31493643e-01 -5.23538888e-01
-8.91293585e-01 8.09341848e-01 1.70174599e-01 -1.07298505e+00
-8.85468125e-01 5.51268578e-01 6.14026964e-01 -3.08590919e-01
-4.31268699e-02 -3.44963998e-01 -4.32331175e-01 1.62106112e-01
4.71972346e-01 4.44000185e-01 -3.00176173e-01 -5.43381087e-02
-1.72401071e-01 4.28054370e-02 -5.09892963e-02 -1.46072106e-02
1.53161776e+00 -3.64175625e-02 -4.54363257e-01 9.25551414e-01
7.56433964e-01 -2.01466292e-01 -1.41197455e+00 -3.94612193e-01
2.80820221e-01 4.95633148e-02 -1.74596176e-01 -4.42347795e-01
-9.95378196e-01 9.03522432e-01 -7.68487528e-02 7.49110341e-01
1.11539578e+00 2.54339486e-01 6.43846571e-01 1.90531015e-01
3.11243385e-01 -7.07425177e-01 2.10884959e-02 7.23614514e-01
5.09972155e-01 -8.59160304e-01 -1.59296095e-01 -6.29665375e-01
-3.36680532e-01 1.19968176e+00 2.50087559e-01 -4.04977709e-01
1.15234745e+00 3.31878841e-01 -4.16137338e-01 -6.01042569e-01
-1.22107756e+00 1.46633402e-01 2.27211967e-01 2.14766279e-01
2.72205442e-01 3.13641220e-01 -2.88073361e-01 5.93177855e-01
2.81981677e-01 -1.07998163e-01 4.81540263e-01 8.16033602e-01
-3.37477885e-02 -8.61668110e-01 -5.63406684e-02 6.88611209e-01
-5.04701078e-01 -2.37044871e-01 -3.08280826e-01 4.64378953e-01
-3.00443202e-01 8.35303903e-01 3.56984228e-01 -1.98105335e-01
2.98588783e-01 7.18803182e-02 1.32980421e-01 -6.32309854e-01
-8.02603483e-01 -3.36304642e-02 -1.12499416e-01 -5.38328528e-01
-1.19029872e-01 -5.90041518e-01 -1.38352156e+00 -3.84625077e-01
-5.02471812e-02 5.58193736e-02 1.18838526e-01 7.46846318e-01
5.56079924e-01 8.55203927e-01 6.26097858e-01 -4.55784947e-01
-3.05536576e-02 -8.94184709e-01 -7.85192788e-01 7.28237629e-01
1.14759570e-02 -5.33156931e-01 -4.23284233e-01 1.38508692e-01] | [7.245387077331543, 5.788229942321777] |
78d283a5-8b44-44da-a210-5c259ac2b3b9 | radiopathomics-multimodal-learning-in-non | 2204.12423 | null | https://arxiv.org/abs/2204.12423v1 | https://arxiv.org/pdf/2204.12423v1.pdf | RadioPathomics: Multimodal Learning in Non-Small Cell Lung Cancer for Adaptive Radiotherapy | The current cancer treatment practice collects multimodal data, such as radiology images, histopathology slides, genomics and clinical data. The importance of these data sources taken individually has fostered the recent raise of radiomics and pathomics, i.e. the extraction of quantitative features from radiology and histopathology images routinely collected to predict clinical outcomes or to guide clinical decisions using artificial intelligence algorithms. Nevertheless, how to combine them into a single multimodal framework is still an open issue. In this work we therefore develop a multimodal late fusion approach that combines hand-crafted features computed from radiomics, pathomics and clinical data to predict radiation therapy treatment outcomes for non-small-cell lung cancer patients. Within this context, we investigate eight different late fusion rules (i.e. product, maximum, minimum, mean, decision template, Dempster-Shafer, majority voting, and confidence rule) and two patient-wise aggregation rules leveraging the richness of information given by computer tomography images and whole-slide scans. The experiments in leave-one-patient-out cross-validation on an in-house cohort of 33 patients show that the proposed multimodal paradigm with an AUC equal to $90.9\%$ outperforms each unimodal approach, suggesting that data integration can advance precision medicine. As a further contribution, we also compare the hand-crafted representations with features automatically computed by deep networks, and the late fusion paradigm with early fusion, another popular multimodal approach. In both cases, the experiments show that the proposed multimodal approach provides the best results. | ['Paolo Soda', 'Sara Ramella', 'Giuseppe Perrone', 'Edy Ippolito', 'Lorenzo Nibid', 'Rosa Sicilia', 'Ermanno Cordelli', 'Matteo Tortora'] | 2022-04-26 | null | null | null | null | ['data-integration'] | ['knowledge-base'] | [ 3.20961326e-01 3.30361282e-03 -3.61585617e-01 -3.48153889e-01
-1.15831017e+00 -4.42993790e-01 6.67332053e-01 8.50850165e-01
-6.73959136e-01 9.44493711e-01 2.14054629e-01 -4.26294774e-01
-6.88379526e-01 -7.52934635e-01 -2.06718981e-01 -1.07501817e+00
7.40151554e-02 6.71665728e-01 -6.79986030e-02 -6.13598861e-02
-1.86617952e-02 8.73385191e-01 -1.34561408e+00 4.47827578e-01
7.20744967e-01 1.29344571e+00 3.13126519e-02 7.31095731e-01
-1.20070755e-01 6.47275448e-01 -6.87914863e-02 -2.76762605e-01
5.32979704e-02 -1.71140209e-01 -5.02374351e-01 -1.60803869e-01
1.25004932e-01 -9.85063836e-02 3.66326720e-02 6.96428120e-01
7.24370718e-01 -3.56083393e-01 8.78455818e-01 -9.49405491e-01
-1.86422750e-01 3.57916147e-01 -5.32979965e-01 1.19033225e-01
2.13091031e-01 3.70986938e-01 9.26166654e-01 -7.23345339e-01
7.64343739e-01 5.49813271e-01 7.50723302e-01 1.10139474e-01
-1.05325878e+00 -2.37087220e-01 -3.04874688e-01 -1.26840562e-01
-1.36924899e+00 -1.55234143e-01 2.62101591e-01 -6.15845919e-01
7.24390805e-01 5.66278756e-01 5.32664418e-01 8.89168739e-01
6.97365344e-01 2.89402395e-01 1.40813756e+00 -4.75002289e-01
1.84462667e-01 3.19025844e-01 1.57057360e-01 7.95066595e-01
5.31485558e-01 1.57365918e-01 -2.51117557e-01 -4.41908926e-01
3.14252019e-01 5.23620903e-01 -2.29603022e-01 -1.31454811e-01
-1.14723277e+00 6.88819885e-01 4.40441191e-01 7.53902197e-01
-5.88783741e-01 -5.94374686e-02 3.22368205e-01 5.26484512e-02
2.25955322e-01 1.25498816e-01 -3.56793314e-01 3.08395684e-01
-1.01097906e+00 -6.44835904e-02 5.92400491e-01 2.14952469e-01
6.77401602e-01 -5.86548805e-01 -3.01278979e-01 6.68404341e-01
5.15293598e-01 4.68946993e-01 9.04086649e-01 -3.03265691e-01
3.24050635e-02 9.60014880e-01 -1.85225874e-01 -8.19672108e-01
-1.04316282e+00 -4.59924459e-01 -1.12474298e+00 -4.39259969e-02
4.76052582e-01 -2.28121504e-02 -9.73928452e-01 1.50452542e+00
3.90145779e-01 -1.41618580e-01 8.78860503e-02 8.49902749e-01
8.99140060e-01 -1.73145123e-02 4.86235708e-01 -3.42307836e-01
1.68921030e+00 -4.83310640e-01 -5.83000779e-01 3.89269322e-01
9.49741364e-01 -6.53070569e-01 3.52255046e-01 4.09098655e-01
-7.23035216e-01 -8.54843929e-02 -8.88527751e-01 2.63796121e-01
-7.18450606e-01 4.47627366e-01 6.79922521e-01 6.49329901e-01
-1.10047674e+00 4.57295179e-01 -7.16473758e-01 -6.45818770e-01
5.82120240e-01 7.08094299e-01 -7.78028131e-01 -1.91557985e-02
-1.01011395e+00 9.18680191e-01 3.65219355e-01 5.04627973e-02
-5.92185140e-01 -7.86808968e-01 -4.62143689e-01 -1.47670746e-01
3.37351829e-01 -1.11449671e+00 9.78761494e-01 -6.63832903e-01
-1.31628108e+00 8.30605686e-01 1.13831190e-02 -4.85098362e-01
4.34195220e-01 3.41302574e-01 -2.87007302e-01 3.23850065e-01
-1.41706064e-01 5.08421004e-01 2.97725469e-01 -1.00680196e+00
-8.58588755e-01 -8.56440127e-01 -3.31669092e-01 1.09463990e-01
-4.69984263e-01 -3.45115453e-01 -5.86754568e-02 -3.60610455e-01
4.49846126e-02 -9.03159618e-01 -5.05424917e-01 -7.36609176e-02
-3.99197429e-01 -1.93902269e-01 4.07257169e-01 -5.61228573e-01
1.19997978e+00 -1.91360450e+00 9.47560966e-02 5.82341909e-01
3.86269510e-01 6.05145767e-02 1.52593717e-01 4.26562905e-01
-1.51592791e-01 3.33418101e-01 -1.20319404e-01 -1.43591732e-01
-1.01403482e-01 -5.00658527e-02 3.73658061e-01 5.54855108e-01
1.26181960e-01 1.03376555e+00 -6.55604601e-01 -7.51435399e-01
3.90020311e-01 5.71184993e-01 -1.58872992e-01 -3.07628680e-02
1.30567150e-02 3.75689119e-01 -5.99992156e-01 9.60285008e-01
4.09546435e-01 -4.00202066e-01 3.61115634e-01 -3.61518979e-01
5.56228459e-02 -3.81772816e-01 -7.32776761e-01 1.64416039e+00
-2.94217438e-01 1.05125234e-01 -7.12584630e-02 -7.25641370e-01
7.42951512e-01 5.89816928e-01 9.34296310e-01 -7.02998519e-01
6.88836455e-01 4.02444094e-01 -5.63069247e-02 -6.01098180e-01
1.09858066e-01 -5.42542756e-01 5.56686223e-02 3.37859482e-01
7.54346177e-02 -2.85557788e-02 1.12982251e-01 -6.00412264e-02
1.37637103e+00 -3.10923040e-01 7.23386109e-01 -6.30509183e-02
9.07845497e-01 1.52688503e-01 1.98911637e-01 5.93666971e-01
-2.44153351e-01 7.00050056e-01 5.17744124e-01 -2.62156755e-01
-6.31924391e-01 -8.06183755e-01 -2.41997331e-01 7.20543683e-01
-3.50756615e-01 -2.88137421e-02 -4.55975026e-01 -8.18175375e-01
1.79872751e-01 2.54212111e-01 -9.63913023e-01 7.02702701e-02
-2.71505415e-01 -1.09347796e+00 5.88095844e-01 3.79573792e-01
-2.66935322e-02 -6.16814196e-01 -6.95291758e-01 2.22708851e-01
1.58971503e-01 -8.48348498e-01 1.69838876e-01 5.19119322e-01
-1.00921571e+00 -1.33112645e+00 -8.40833068e-01 -2.10740700e-01
5.85937500e-01 -5.43907322e-02 7.38871515e-01 1.66678563e-01
-5.87617576e-01 3.79493982e-01 -4.12639141e-01 -3.59091192e-01
-4.18436229e-01 1.49015024e-01 -2.10994899e-01 2.01112524e-01
3.36966455e-01 -5.32975018e-01 -7.17357516e-01 5.99787533e-02
-1.14915395e+00 -1.29661322e-01 1.26941907e+00 8.05968344e-01
8.85245502e-01 -1.65017515e-01 4.79382366e-01 -9.53580439e-01
5.17368793e-01 -7.48138666e-01 -2.25117743e-01 4.99061763e-01
-5.39549589e-01 -5.36830677e-03 5.39249778e-01 -7.28273531e-03
-8.67004275e-01 3.22241634e-01 -9.93404314e-02 -2.24779367e-01
-4.72946763e-01 8.44044268e-01 1.25564620e-01 -1.40740544e-01
7.71570086e-01 -5.62305525e-02 2.84426391e-01 -1.38186634e-01
1.62623316e-01 7.16209769e-01 3.76879066e-01 -4.42122877e-01
5.11162102e-01 7.16427922e-01 5.18341422e-01 -5.13044775e-01
-5.27472377e-01 -5.16460419e-01 -6.76693976e-01 -2.16475293e-01
9.76180017e-01 -7.23898053e-01 -8.72957349e-01 3.03914189e-01
-8.45758915e-01 1.42546028e-01 -3.08509618e-01 7.10918069e-01
-4.70875591e-01 1.26410678e-01 -4.38572407e-01 -7.02020049e-01
-6.59212410e-01 -1.23701167e+00 1.25596189e+00 4.55824673e-01
-1.91985071e-01 -9.39838648e-01 3.03375959e-01 3.02603662e-01
5.57853699e-01 7.64777660e-01 9.77102160e-01 -1.09229755e+00
-3.77562016e-01 -6.25226617e-01 -2.77922422e-01 -4.80584092e-02
2.20176622e-01 2.18673244e-01 -1.09145772e+00 -1.01899151e-02
-2.28643790e-01 -2.55037755e-01 8.93857121e-01 4.65416759e-01
9.68690932e-01 1.82937384e-01 -5.40588021e-01 3.50884646e-01
1.74161708e+00 2.44580492e-01 3.62855673e-01 2.69302398e-01
3.13367605e-01 6.57398283e-01 3.87626380e-01 6.19635522e-01
3.62024039e-01 3.57825756e-01 5.58910191e-01 6.81229692e-04
-3.37031996e-03 1.62194222e-01 -1.64637148e-01 3.41009319e-01
-2.42838800e-01 -4.66568232e-01 -1.18975556e+00 2.42993385e-01
-1.63468218e+00 -6.23955846e-01 -8.03223476e-02 2.02286720e+00
5.14439523e-01 1.49167571e-02 -1.32205918e-01 2.03860566e-01
5.80818415e-01 -3.14213514e-01 -2.57488847e-01 -1.71811327e-01
-1.25826433e-01 4.05092537e-01 4.71283436e-01 1.42287120e-01
-9.80297267e-01 2.23430663e-01 5.43227625e+00 8.51372063e-01
-1.38682508e+00 1.03684179e-01 9.54923093e-01 3.98607068e-02
-2.34273061e-01 -2.68148482e-01 -5.28916419e-01 1.44520625e-01
1.28737986e+00 -1.08935509e-03 -1.47446021e-01 4.57102627e-01
3.74726593e-01 -5.80083311e-01 -1.11098003e+00 7.64991879e-01
8.54456872e-02 -1.38936794e+00 -2.60170344e-02 5.42192519e-01
4.67051655e-01 1.10931106e-01 7.24786296e-02 -3.92990559e-02
1.71882063e-01 -1.12966275e+00 1.49077669e-01 1.12901568e+00
7.15565860e-01 -4.61175650e-01 1.31346393e+00 3.17041367e-01
-8.77110183e-01 -1.65614396e-01 2.69284278e-01 4.58314359e-01
2.40733866e-02 8.10663164e-01 -1.20354152e+00 1.23156822e+00
2.64822423e-01 4.60436881e-01 -8.62736344e-01 1.01741683e+00
3.40110093e-01 2.74112046e-01 -4.10684854e-01 -1.30678117e-01
2.80610323e-01 2.34882563e-01 1.40914574e-01 1.15394056e+00
6.24070287e-01 2.38473520e-01 -1.05928905e-01 4.16944712e-01
2.07230642e-01 4.39573735e-01 -4.18086052e-01 -2.13346243e-01
1.61718354e-01 1.84639668e+00 -1.03706443e+00 -2.92731583e-01
-3.52238089e-01 2.67198861e-01 1.76266376e-02 -3.01927980e-02
-6.93182945e-01 3.92376706e-02 -4.57681939e-02 2.05015913e-01
1.30047545e-01 2.82585561e-01 -4.94146049e-01 -7.88563371e-01
-3.80992860e-01 -6.83160365e-01 7.82758892e-01 -6.00504458e-01
-1.53300428e+00 7.45049596e-01 6.68770075e-03 -1.29864907e+00
-2.23158017e-01 -8.47257793e-01 -5.66994190e-01 6.92157388e-01
-1.41443908e+00 -1.21694171e+00 -5.43418229e-01 4.10596013e-01
-1.77211314e-01 -1.48687840e-01 8.76095712e-01 2.53320366e-01
-6.03641212e-01 4.90412474e-01 9.08286572e-02 -1.73383281e-01
6.57868326e-01 -1.12940252e+00 -8.14737678e-01 4.60012406e-02
-3.12826812e-01 3.87523413e-01 3.74048293e-01 -4.74106580e-01
-1.44871175e+00 -8.68258655e-01 6.45873785e-01 -2.97950566e-01
5.80014348e-01 3.43777925e-01 -6.60807312e-01 1.46350682e-01
1.67081237e-01 3.04410547e-01 1.36090767e+00 -2.98209667e-01
9.31894556e-02 -2.42786244e-01 -1.47144473e+00 2.31580645e-01
3.40501130e-01 -2.05494449e-01 -2.94727921e-01 2.28305951e-01
2.43530303e-01 -2.00707316e-01 -1.47360957e+00 8.53557527e-01
7.41719186e-01 -1.12356222e+00 7.66394973e-01 -4.48391318e-01
3.85115534e-01 -3.02585989e-01 -3.25330257e-01 -1.06902444e+00
-1.11306794e-01 5.08613735e-02 3.04820031e-01 1.00258183e+00
6.18057489e-01 -6.40109003e-01 8.40084970e-01 5.42178094e-01
-1.16638064e-01 -1.30169463e+00 -1.04245317e+00 -1.90236852e-01
9.48387906e-02 -3.57496440e-01 4.90471482e-01 6.56524479e-01
-1.09561153e-01 7.20142499e-02 2.29781792e-01 -7.57896751e-02
2.77912557e-01 -1.28412440e-01 5.70417881e-01 -1.40455782e+00
-1.93164781e-01 -6.29732251e-01 -7.24610209e-01 1.55765731e-02
-1.45235330e-01 -1.06848276e+00 -2.91562140e-01 -1.54163015e+00
5.19383788e-01 -4.24344003e-01 -7.65292346e-01 4.64846492e-01
-4.51283567e-02 3.24837029e-01 7.91751146e-02 8.12917128e-02
-3.52995306e-01 2.69034207e-01 1.11670780e+00 -8.20254162e-02
-4.30990718e-02 -1.73229232e-01 -7.97905266e-01 7.20854223e-01
6.71647847e-01 -4.98890579e-01 -6.23782091e-02 1.39083371e-01
1.25591561e-01 5.08490980e-01 4.28120226e-01 -1.14150310e+00
3.88562918e-01 -2.53686547e-01 7.12564468e-01 -7.68909812e-01
2.44404882e-01 -1.08209908e+00 3.49767268e-01 6.70583606e-01
-1.11990497e-01 5.43510402e-03 5.61226308e-02 7.08755553e-01
-3.33067983e-01 -1.26965076e-01 5.67735195e-01 -1.69235647e-01
-1.97370365e-01 2.35859320e-01 -3.09274316e-01 -4.69416380e-01
1.41112244e+00 -3.54115069e-01 -4.73406315e-01 5.37418462e-02
-9.70799088e-01 1.87103584e-01 7.14445412e-02 -1.06953876e-02
3.10981303e-01 -1.18565130e+00 -8.19254994e-01 -4.52115051e-02
2.08037749e-01 -9.02216733e-02 6.21934950e-01 1.60961485e+00
-4.84976232e-01 6.46009922e-01 -1.74664214e-01 -8.82960141e-01
-1.18548167e+00 3.14229965e-01 3.78637165e-01 -8.72581720e-01
1.00237936e-01 5.94838977e-01 -4.63483706e-02 -4.00115728e-01
-1.24608293e-01 -3.46939057e-01 -4.49393958e-01 4.40059572e-01
2.54717708e-01 2.01859429e-01 5.23790956e-01 -6.62940860e-01
-5.14197588e-01 5.73361754e-01 -2.10642338e-01 -1.63087785e-01
1.41838133e+00 1.47463202e-01 -2.74024457e-01 3.89616668e-01
1.20450675e+00 -1.16462693e-01 -5.11313021e-01 -1.81380212e-01
9.18252692e-02 -1.26173198e-01 2.11359859e-02 -1.20350540e+00
-1.04395449e+00 7.22634792e-01 9.29327428e-01 1.85294837e-01
1.39524710e+00 9.50216595e-03 3.91058981e-01 2.98342973e-01
1.56787977e-01 -7.62796104e-01 -1.04085855e-01 1.33005127e-01
6.99478090e-01 -1.23587883e+00 2.67082155e-01 -2.95709789e-01
-5.74188590e-01 1.45831764e+00 3.42296958e-01 1.68092728e-01
8.07227314e-01 3.33310813e-01 7.59644434e-02 -2.78041571e-01
-1.00494552e+00 -3.96090716e-01 3.29477429e-01 2.28883415e-01
6.72753930e-01 1.92852125e-01 -6.39034271e-01 9.19480681e-01
-2.13088281e-02 4.95531946e-01 1.92499250e-01 1.01977420e+00
-5.27820468e-01 -1.10827076e+00 -5.63172340e-01 7.81509578e-01
-6.06329501e-01 1.52503923e-01 -5.89498580e-01 1.06388760e+00
4.32486445e-01 6.43192530e-01 -2.07650304e-01 -4.77831751e-01
2.19385356e-01 2.29868338e-01 3.15469623e-01 -3.03940177e-01
-1.03550458e+00 2.10023433e-01 -1.06815241e-01 -1.82481095e-01
-5.77715278e-01 -6.17469728e-01 -1.23271775e+00 2.77217943e-02
-4.44548696e-01 -3.06528131e-03 9.29562807e-01 1.09091938e+00
2.69885510e-01 6.47583842e-01 6.34634435e-01 -5.90029657e-01
-4.98786509e-01 -1.00730419e+00 -3.59617591e-01 1.02270275e-01
2.44459018e-01 -5.57657719e-01 -3.38412553e-01 -1.63130254e-01] | [15.081212043762207, -2.843771457672119] |
08f33bc3-6089-4cd1-b090-1628dc1eed76 | weakly-supervised-optical-flow-estimation-for | 2210.05298 | null | https://arxiv.org/abs/2210.05298v2 | https://arxiv.org/pdf/2210.05298v2.pdf | Weakly-Supervised Optical Flow Estimation for Time-of-Flight | Indirect Time-of-Flight (iToF) cameras are a widespread type of 3D sensor, which perform multiple captures to obtain depth values of the captured scene. While recent approaches to correct iToF depths achieve high performance when removing multi-path-interference and sensor noise, little research has been done to tackle motion artifacts. In this work we propose a training algorithm, which allows to supervise Optical Flow (OF) networks directly on the reconstructed depth, without the need of having ground truth flows. We demonstrate that this approach enables the training of OF networks to align raw iToF measurements and compensate motion artifacts in the iToF depth images. The approach is evaluated for both single- and multi-frequency sensors as well as multi-tap sensors, and is able to outperform other motion compensation techniques. | ['Timo Ropinski', 'Pedro Hermosilla', 'Michael Schelling'] | 2022-10-11 | null | null | null | null | ['motion-compensation'] | ['computer-vision'] | [ 5.58326066e-01 -1.11799724e-01 1.36456311e-01 -2.50391066e-01
-2.45987833e-01 -4.35832620e-01 3.34290981e-01 -6.13140225e-01
-6.40461206e-01 6.61783338e-01 2.08531395e-01 5.70811592e-02
-8.13761652e-02 -5.11776447e-01 -5.63278794e-01 -4.88220721e-01
1.50061354e-01 2.27719601e-02 3.60131204e-01 2.89602578e-01
3.99916381e-01 6.27937555e-01 -1.50590789e+00 -1.01492472e-01
5.70573151e-01 9.57460761e-01 2.80935049e-01 1.09019613e+00
2.03457773e-01 9.19078767e-01 -5.22982597e-01 -3.58583778e-02
4.97628242e-01 -4.90829051e-01 -4.95614380e-01 1.54527172e-01
1.00400293e+00 -1.06561744e+00 -5.68764746e-01 9.52244937e-01
5.40597379e-01 2.12803148e-02 3.95624667e-01 -9.37858820e-01
2.23515108e-02 4.93825711e-02 -4.22186524e-01 3.19287777e-01
5.26262820e-01 3.65423739e-01 4.44310904e-01 -8.08377445e-01
7.19024956e-01 1.13280296e+00 7.85761833e-01 7.35460520e-01
-1.05935740e+00 -2.90051430e-01 -3.45918626e-01 1.28447756e-01
-8.33005369e-01 -5.69766045e-01 8.35192740e-01 -5.08432567e-01
9.77657914e-01 -1.97939336e-01 8.96826684e-01 1.03030014e+00
3.20439398e-01 4.03314143e-01 8.20176303e-01 -3.18822682e-01
2.87035927e-02 -4.53316450e-01 -6.98660016e-02 4.61475968e-01
3.74461204e-01 3.19168657e-01 -9.84819889e-01 2.49198481e-01
1.41715026e+00 -5.41071128e-03 -8.66351962e-01 -4.15262789e-01
-1.54611337e+00 2.84853667e-01 2.59900481e-01 3.51432979e-01
-4.47878361e-01 7.65453219e-01 1.50057256e-01 1.19479045e-01
3.66482735e-01 3.37636501e-01 -3.40500772e-01 -5.66299796e-01
-1.04695952e+00 -7.67614469e-02 5.87119460e-01 7.22883463e-01
9.89121914e-01 3.16796899e-01 2.24475861e-01 2.75592864e-01
4.96827036e-01 8.25234532e-01 4.66293037e-01 -1.84950888e+00
4.93342251e-01 2.28027388e-01 2.66965210e-01 -1.01769066e+00
-4.44058686e-01 -7.32192993e-02 -8.70993078e-01 6.26638949e-01
7.31200516e-01 -2.05070823e-01 -7.01487839e-01 1.50876617e+00
2.16326430e-01 7.61546969e-01 3.06857750e-02 1.25135589e+00
5.84345460e-01 4.14469510e-01 -7.52368808e-01 -3.39419007e-01
7.36595273e-01 -8.24872017e-01 -8.81993532e-01 -2.36969605e-01
2.81580865e-01 -6.59316421e-01 7.10588992e-01 7.14289069e-01
-1.25164258e+00 -6.14263296e-01 -1.07955921e+00 -2.58208275e-01
1.83458418e-01 -1.50494859e-01 3.10596794e-01 4.04503435e-01
-1.11645305e+00 1.03164291e+00 -1.18478703e+00 -1.80090636e-01
2.18348101e-01 3.46217006e-01 -4.76599842e-01 -1.83094189e-01
-7.96257913e-01 7.72193015e-01 8.24996978e-02 3.70307922e-01
-9.23256457e-01 -9.97420490e-01 -6.75931394e-01 -2.76670367e-01
1.69683874e-01 -1.05945313e+00 1.18709648e+00 -9.51589525e-01
-1.98988640e+00 5.40403485e-01 -3.36147189e-01 -5.03352463e-01
6.81328833e-01 -6.58258557e-01 1.68962665e-02 6.78750336e-01
-1.43835485e-01 7.10507870e-01 1.09986889e+00 -9.45132256e-01
-4.55839723e-01 -2.41647854e-01 -4.95838039e-02 3.20114233e-02
-2.88647801e-01 -4.35107768e-01 -1.45801052e-01 -2.40442500e-01
3.49664330e-01 -6.63290203e-01 -1.26672640e-01 5.77448606e-01
-3.74395519e-01 5.59577525e-01 8.78578722e-01 -3.33933920e-01
8.60102415e-01 -1.84082067e+00 7.08446056e-02 -9.77459997e-02
2.21124351e-01 3.54610503e-01 -7.11176246e-02 6.49100840e-02
8.28146413e-02 -4.07430321e-01 -2.87841618e-01 -6.33319616e-01
-6.04183674e-01 3.92497748e-01 -1.32382259e-01 5.88354647e-01
-2.66579334e-02 7.15885818e-01 -1.07010734e+00 -1.39575034e-01
8.88121068e-01 8.11221123e-01 -6.39478445e-01 2.90458888e-01
9.02849957e-02 9.71137285e-01 -7.35263079e-02 3.82685035e-01
7.18571544e-01 -1.88121185e-01 -6.21873736e-02 -5.18633187e-01
-4.14446533e-01 2.35656992e-01 -1.24195874e+00 2.04843903e+00
-4.39553052e-01 1.03806591e+00 3.21208298e-01 -6.09712958e-01
7.19072700e-01 4.18335617e-01 9.53367233e-01 -6.45521820e-01
3.75700355e-01 4.64217275e-01 -2.25510076e-01 -6.31702363e-01
4.58125561e-01 -9.72743556e-02 5.02265692e-01 2.63713717e-01
3.12932581e-02 -2.94788659e-01 -9.74213518e-03 -2.29878023e-01
1.30024230e+00 4.08001572e-01 9.11329761e-02 -3.03246360e-02
7.80504882e-01 -2.18671381e-01 6.36166334e-01 6.31419957e-01
-3.19020718e-01 8.94718707e-01 -2.23521926e-02 -5.60349822e-01
-1.05269897e+00 -1.07131994e+00 7.89934620e-02 -5.85769638e-02
3.84570777e-01 -3.75837609e-02 -7.30095506e-01 -2.57307053e-01
-8.78740549e-02 1.35005191e-01 -4.55552936e-01 1.49439305e-01
-7.39999831e-01 -2.02106878e-01 4.60253596e-01 3.98959190e-01
7.23238528e-01 -6.86165452e-01 -1.44901645e+00 3.02919894e-01
-3.91045481e-01 -1.56209290e+00 -5.60419977e-01 -7.91095272e-02
-1.34871411e+00 -1.36211574e+00 -7.68042445e-01 -1.91794559e-01
4.97550398e-01 6.48446739e-01 8.84982288e-01 -3.04308593e-01
-3.79359014e-02 8.10685635e-01 -1.24466801e-02 -9.67663676e-02
-1.95687622e-01 -3.07440549e-01 3.44102010e-02 2.31714562e-01
1.75016612e-01 -1.00974131e+00 -8.48747194e-01 3.07828069e-01
-1.03672862e+00 3.23045850e-02 2.58569390e-01 6.67909205e-01
4.21552896e-01 -3.82243484e-01 -8.38969722e-02 -4.45629060e-01
2.15289757e-01 9.95316133e-02 -9.88867164e-01 -4.48481292e-01
-4.92760509e-01 -8.88517797e-02 5.90189397e-01 -3.78889322e-01
-8.58712316e-01 2.62322932e-01 2.66898051e-03 -1.11739659e+00
-1.71846449e-01 6.15900271e-02 1.29398361e-01 -3.18738848e-01
5.83006322e-01 -1.81368776e-02 3.06811929e-01 -2.94572443e-01
3.35948281e-02 2.76203930e-01 1.15552306e+00 -1.23541296e-01
7.78769076e-01 1.06945550e+00 5.55211306e-01 -1.04884350e+00
-8.37851584e-01 -7.00016379e-01 -1.01364136e+00 -6.49371564e-01
1.13302767e+00 -9.39406872e-01 -8.16195309e-01 9.13894355e-01
-1.43952453e+00 -4.25763428e-01 -4.85644311e-01 9.56531227e-01
-7.47056782e-01 6.21350884e-01 -6.84248865e-01 -5.95994174e-01
-1.37452170e-01 -1.11383057e+00 1.04052258e+00 3.41193199e-01
-5.47322817e-02 -1.20954907e+00 3.03506970e-01 2.05251426e-01
3.58687609e-01 2.99006909e-01 -3.19977328e-02 5.58848441e-01
-1.07653809e+00 1.63816646e-01 -2.80663799e-02 7.28992701e-01
2.25141063e-01 1.95712700e-01 -1.30930865e+00 -2.51487970e-01
3.09955865e-01 3.21151018e-02 7.34555423e-01 8.84179175e-01
6.23865962e-01 5.80341741e-02 1.55770360e-02 1.05992746e+00
1.65470028e+00 1.33230954e-01 1.01932633e+00 4.59305853e-01
9.38378751e-01 6.17633998e-01 4.24844116e-01 5.51380575e-01
1.57875732e-01 7.58292139e-01 8.60435545e-01 1.13455780e-01
-3.46730292e-01 -1.28494561e-01 5.83807945e-01 7.60930419e-01
-4.05249923e-01 -1.69011399e-01 -6.38046622e-01 5.84640861e-01
-1.60120690e+00 -9.15353358e-01 -7.34118104e-01 2.21489882e+00
2.78571337e-01 -2.25843936e-02 -1.85315832e-01 4.79023337e-01
4.67383057e-01 1.92323744e-01 -6.47594929e-01 -1.16392158e-01
-1.66576833e-01 1.77025735e-01 7.38763154e-01 8.20109606e-01
-8.42745185e-01 4.69918668e-01 6.65344763e+00 -1.42165735e-01
-1.32439291e+00 6.29705936e-02 4.96473126e-02 -2.96515316e-01
-2.28206515e-01 -1.25062987e-02 -4.55338746e-01 4.40115780e-01
9.18558836e-01 2.32416764e-01 3.59015971e-01 4.04208958e-01
5.89488924e-01 -4.97657120e-01 -1.09174252e+00 1.30551887e+00
1.90741178e-02 -1.18220317e+00 -1.31953895e-01 1.39478937e-01
7.82995582e-01 1.39960527e-01 -2.49548301e-01 -6.76433504e-01
-2.00641319e-01 -6.29610121e-01 6.65936232e-01 8.52340877e-01
7.22237945e-01 -4.30905551e-01 7.75349200e-01 3.73506486e-01
-9.94240701e-01 5.56775443e-02 -3.65269572e-01 -3.19284678e-01
6.26959443e-01 1.08351767e+00 -5.21329522e-01 4.84543890e-01
6.81390941e-01 1.39906538e+00 -1.14237897e-01 1.07647669e+00
-2.46837661e-01 2.18474343e-01 -4.18449104e-01 2.93955564e-01
5.61436713e-02 -3.66274536e-01 7.69385099e-01 7.91088402e-01
8.29238892e-01 -7.48100281e-02 -2.41526201e-01 7.76212990e-01
1.84451804e-01 -5.79540193e-01 -6.63868427e-01 3.15156430e-01
2.15368047e-01 1.02638769e+00 -3.44530195e-01 -2.57167101e-01
-5.81768990e-01 1.06525302e+00 -2.22209290e-01 3.26972067e-01
-5.45329869e-01 -1.81727499e-01 1.08817661e+00 3.10138404e-01
3.01588178e-01 -6.28653347e-01 -1.49005204e-01 -1.42036772e+00
1.43179476e-01 -2.58002490e-01 -1.81938279e-02 -1.29810667e+00
-9.73394573e-01 2.30443329e-01 -1.75542295e-01 -1.65518212e+00
-4.68916178e-01 -8.02575171e-01 -4.39763904e-01 6.51916683e-01
-1.96912110e+00 -5.38260043e-01 -8.96912932e-01 9.33697045e-01
2.08507583e-01 3.05343747e-01 4.44076538e-01 5.56564569e-01
-3.41828257e-01 -1.74347088e-02 3.28886509e-02 4.96787131e-02
8.12144101e-01 -9.13195491e-01 8.94933417e-02 1.27078867e+00
-1.95778310e-01 2.96395659e-01 7.43290126e-01 -3.78570378e-01
-1.51894784e+00 -7.95895576e-01 6.84135079e-01 -3.11952204e-01
4.55604970e-01 1.69221967e-01 -6.88997746e-01 6.07281864e-01
1.35971919e-01 4.40450072e-01 1.19146649e-02 -7.61133969e-01
-9.06857327e-02 -3.74232382e-01 -1.05471551e+00 6.21055029e-02
1.16076684e+00 -6.37845814e-01 -4.15684164e-01 -2.74007767e-01
3.65217417e-01 -6.80002630e-01 -7.86945820e-01 3.22956443e-01
7.03960359e-01 -1.74408698e+00 1.06300986e+00 3.37642848e-01
7.40437627e-01 -5.27150154e-01 -3.32292705e-03 -1.19796562e+00
1.21887140e-02 -7.76186466e-01 -4.70953524e-01 8.74863923e-01
-2.19183639e-01 -8.40697765e-01 1.15633142e+00 2.69686848e-01
-3.22964221e-01 -9.42800120e-02 -1.23169851e+00 -6.77187979e-01
-4.21196491e-01 -6.79007232e-01 8.31076205e-02 7.11012781e-01
-4.38105553e-01 1.47678256e-01 -5.90691149e-01 2.44647771e-01
1.05339265e+00 -1.61459863e-01 1.02195573e+00 -1.25999057e+00
-3.18727076e-01 -2.97069073e-01 -6.82330191e-01 -1.70436561e+00
-9.37994123e-02 -2.72177428e-01 1.57236129e-01 -1.49991941e+00
-5.70227504e-01 5.21647409e-02 1.04900643e-01 -2.04550643e-02
1.56374738e-01 5.02388597e-01 1.31712377e-01 3.42010260e-01
-8.33918750e-02 3.91177863e-01 1.63866830e+00 1.96780920e-01
-1.60767809e-01 -7.21867010e-02 6.01399243e-02 8.16648126e-01
3.99840891e-01 -3.51497561e-01 -6.18073642e-01 -8.08136761e-01
1.32294133e-01 3.92792612e-01 6.91819847e-01 -1.48856890e+00
3.73492926e-01 4.59236763e-02 4.83879983e-01 -8.06769133e-01
5.28468907e-01 -1.12610722e+00 4.90126044e-01 5.13474405e-01
-6.19290629e-03 1.07192628e-01 -7.61724710e-02 4.60682511e-01
-4.53970253e-01 -2.55154133e-01 9.09016490e-01 -2.32989177e-01
-7.20308900e-01 4.29315984e-01 -4.36817527e-01 5.34693673e-02
6.38700724e-01 -6.82438254e-01 -2.11085871e-01 -5.38016260e-01
-3.33214849e-01 -1.97737053e-01 6.73555315e-01 6.62618279e-02
9.07481611e-01 -1.22200036e+00 -2.76127458e-01 4.37144965e-01
-2.26069689e-01 4.09074157e-01 2.74236143e-01 9.79575038e-01
-9.87448752e-01 3.61434430e-01 -3.55262935e-01 -1.08391464e+00
-9.75689054e-01 1.51116312e-01 8.56948078e-01 -1.22861482e-01
-8.90099168e-01 6.52993500e-01 -4.70634438e-02 -1.46491691e-01
1.39472514e-01 -6.84875190e-01 -3.72710638e-02 -6.82386234e-02
7.02604771e-01 7.15387583e-01 -7.56713934e-03 -6.61998689e-01
-2.11713180e-01 1.24519527e+00 6.17483079e-01 -3.11665207e-01
1.17836821e+00 -4.81456250e-01 -8.36357009e-03 5.65368354e-01
1.28323066e+00 -8.31076428e-02 -1.90984428e+00 6.75592348e-02
-2.76105881e-01 -8.40748966e-01 1.47429109e-01 -1.88660040e-01
-1.43475258e+00 1.25849748e+00 6.06378973e-01 8.61945078e-02
1.29744160e+00 -6.28079772e-01 9.51476872e-01 2.99071968e-01
5.59077680e-01 -7.17114627e-01 3.21184844e-01 7.19488978e-01
2.70165712e-01 -1.01051176e+00 -7.94027597e-02 -4.41893488e-01
1.83886528e-01 1.57858396e+00 4.59263951e-01 -3.53473723e-01
5.78722417e-01 3.63456398e-01 3.53890449e-01 5.19653447e-02
-4.83915538e-01 -7.43039623e-02 -2.58734040e-02 7.94880271e-01
2.18862236e-01 -5.99549115e-01 1.04745127e-01 -4.76966143e-01
6.94568232e-02 5.23039520e-01 1.01334810e+00 9.16871548e-01
-2.66961932e-01 -9.62420344e-01 -4.17085826e-01 1.25301734e-01
-3.49653006e-01 1.49933025e-01 4.36460190e-02 6.08513892e-01
5.02922349e-02 7.64193535e-01 1.08520843e-01 -2.94723690e-01
6.56437516e-01 -2.79951662e-01 8.17557216e-01 -2.21792117e-01
-4.81968850e-01 6.63100779e-02 5.81430346e-02 -1.26098466e+00
-1.31404507e+00 -6.93073750e-01 -1.18465328e+00 -2.88860828e-01
-8.00937712e-02 -1.97555423e-01 5.34559608e-01 7.71934152e-01
2.74736524e-01 5.28902352e-01 6.30925417e-01 -1.37226522e+00
3.70185040e-02 -7.64903843e-01 -6.63599968e-01 5.11203468e-01
1.04785955e+00 -5.66513538e-01 -8.49491656e-01 1.68830886e-01] | [8.794475555419922, -2.1651573181152344] |
cd3f15e0-1ce0-4007-897c-d1a490b1a345 | lane-detection-with-position-embedding | 2203.12301 | null | https://arxiv.org/abs/2203.12301v1 | https://arxiv.org/pdf/2203.12301v1.pdf | Lane detection with Position Embedding | Recently, lane detection has made great progress in autonomous driving. RESA (REcurrent Feature-Shift Aggregator) is based on image segmentation. It presents a novel module to enrich lane feature after preliminary feature extraction with an ordinary CNN. For Tusimple dataset, there is not too complicated scene and lane has more prominent spatial features. On the basis of RESA, we introduce the method of position embedding to enhance the spatial features. The experimental results show that this method has achieved the best accuracy 96.93% on Tusimple dataset. | ['Jianwei Shuai', 'Kaer Huang', 'Feng Chen', 'Dezhen Qi', 'Jiacheng Han', 'Jun Xie'] | 2022-03-23 | null | null | null | null | ['lane-detection'] | ['computer-vision'] | [-3.27053696e-01 -1.00344621e-01 -2.27503717e-01 -4.76440072e-01
-3.18867326e-01 -1.00480206e-01 6.47588491e-01 -4.51492310e-01
-7.65047371e-01 6.23907566e-01 2.47087970e-01 -5.10775924e-01
1.07421733e-01 -8.93996894e-01 -4.72732991e-01 -7.01289713e-01
7.51425847e-02 -4.94568765e-01 6.03779256e-01 -6.48614168e-01
4.59489226e-01 7.83003747e-01 -1.76176012e+00 -6.62280694e-02
8.93693447e-01 9.05905426e-01 2.96310484e-01 4.77112085e-01
-3.27595741e-01 6.69859171e-01 -2.49426678e-01 1.41459048e-01
5.81064522e-01 -1.91364974e-01 -3.88706416e-01 9.98587459e-02
4.66035485e-01 -3.82923394e-01 -8.22772980e-01 1.05544341e+00
4.62541461e-01 1.57035813e-01 3.64420772e-01 -1.57077849e+00
-2.11910561e-01 5.10776699e-01 -5.96319497e-01 4.73307788e-01
-1.88743800e-01 2.15020120e-01 5.25242507e-01 -1.02469325e+00
7.06321299e-01 1.24854076e+00 6.84408426e-01 5.27212806e-02
-4.82601792e-01 -5.52548528e-01 -3.02265808e-02 7.36082435e-01
-1.44946742e+00 -2.10766718e-01 1.00557387e+00 -2.39379317e-01
8.00080061e-01 1.32053614e-01 6.41070187e-01 3.20103586e-01
7.75999308e-01 1.02670848e+00 1.13534641e+00 2.51377597e-02
-2.20089853e-01 -1.50420200e-02 6.89726651e-01 9.34764862e-01
3.66037309e-01 1.62031174e-01 -1.46483257e-01 6.08741701e-01
5.19248128e-01 9.23973694e-02 1.74413938e-02 -4.76884484e-01
-1.49015343e+00 9.62421417e-01 7.12615848e-01 -2.52598021e-02
-5.12722507e-02 7.34232739e-02 5.04380167e-01 5.12913950e-02
-1.60158858e-01 9.60450023e-02 -2.40614992e-02 -1.31961659e-01
-5.22185862e-01 3.50563765e-01 1.88815579e-01 1.17221785e+00
1.22373593e+00 4.21636313e-01 1.09872380e-02 5.93342543e-01
1.50017098e-01 6.89814687e-01 6.77309871e-01 -1.10586226e+00
3.36215556e-01 6.63341880e-01 -1.73407868e-01 -1.15054524e+00
-7.02349067e-01 -4.07988161e-01 -8.72403324e-01 5.27144730e-01
4.59474744e-03 -1.65464386e-01 -1.15718865e+00 1.10545969e+00
-4.83814031e-02 1.66182503e-01 4.02949005e-01 8.73884141e-01
1.08012831e+00 8.99859011e-01 -9.16924551e-02 1.05664521e-01
1.10507476e+00 -1.40900528e+00 -1.05651200e+00 -1.92703351e-01
9.12247360e-01 -5.91774404e-01 7.87597120e-01 9.09684524e-02
-4.72109586e-01 -1.11657298e+00 -1.52511144e+00 -3.48701328e-01
-9.73367453e-01 4.06880260e-01 8.05967093e-01 4.07890469e-01
-9.77252662e-01 1.82560861e-01 -1.41370490e-01 -3.37596178e-01
3.42549473e-01 1.96751624e-01 -7.47142673e-01 6.15678243e-02
-1.31843889e+00 9.25875485e-01 7.67922461e-01 6.62524343e-01
-2.86502570e-01 -1.43556237e-01 -1.31292784e+00 -2.99797416e-01
3.56810004e-01 -2.12297097e-01 8.37275445e-01 -2.08571404e-01
-1.48595953e+00 4.89616364e-01 -3.16279292e-01 -7.06241310e-01
6.12416863e-01 -2.02223033e-01 -6.53132677e-01 -2.06950635e-01
2.57880002e-01 1.09907925e+00 4.79019672e-01 -1.01759160e+00
-1.22199714e+00 -1.63910761e-02 -2.00309575e-01 2.17875615e-01
3.07755291e-01 -4.02883708e-01 -2.84663737e-01 -2.02495351e-01
3.88157099e-01 -7.83761501e-01 -5.31391859e-01 -2.30652317e-01
-2.97432452e-01 -3.97393286e-01 1.46857250e+00 -3.08800101e-01
1.15478611e+00 -2.46809101e+00 -4.52324897e-01 1.89914584e-01
3.44996095e-01 3.73035163e-01 -2.49944609e-02 1.79750696e-01
7.83229247e-02 -2.15207022e-02 -1.18240990e-01 1.79727837e-01
8.56644474e-03 2.45594516e-01 -3.41401368e-01 3.63077909e-01
2.39052743e-01 1.28951895e+00 -7.94667006e-01 -7.00835884e-01
7.48236597e-01 5.51744364e-02 -2.75271863e-01 -2.52138853e-01
3.90028983e-01 6.24806434e-02 -3.83346349e-01 4.25453037e-01
1.14524996e+00 3.59164566e-01 -5.86832047e-01 -4.08665419e-01
-9.73526716e-01 -3.05398762e-01 -1.11193228e+00 1.41029108e+00
-1.69551358e-01 1.30344236e+00 -4.21129495e-01 -6.62300944e-01
1.32318020e+00 -6.29911780e-01 2.17738643e-01 -9.60047662e-01
5.44443190e-01 1.82417676e-01 2.04650462e-01 -8.36480439e-01
1.04953468e+00 5.49707592e-01 -3.79670113e-01 -3.46333176e-01
-2.43129343e-01 -1.22274794e-01 1.95309922e-01 9.50252712e-02
9.09675956e-01 5.68390861e-02 7.75126517e-02 -4.30908382e-01
8.46873105e-01 3.87816280e-01 7.27892160e-01 5.64877689e-01
-8.53251398e-01 5.13984621e-01 2.79540747e-01 -5.15440106e-01
-1.08757758e+00 -9.50871527e-01 -2.04729870e-01 3.91136616e-01
6.71138287e-01 -7.95078650e-02 -4.31246161e-01 -6.44418240e-01
3.25490117e-01 4.96592283e-01 -6.18210137e-01 -2.43061468e-01
-8.23863983e-01 -2.99907416e-01 3.97397101e-01 7.15564609e-01
1.34252751e+00 -1.06034994e+00 -5.84369302e-01 1.75159812e-01
8.82491544e-02 -1.27283323e+00 -4.13002521e-01 7.87904710e-02
-5.23251712e-01 -7.93879628e-01 -4.68079895e-01 -1.12943363e+00
5.66711366e-01 8.54041159e-01 2.89659947e-01 -2.91307658e-01
-3.40752780e-01 -1.64136395e-01 -2.20670596e-01 -4.12442565e-01
1.97258055e-01 1.34520829e-01 -2.63490438e-01 -2.40241885e-02
4.88022476e-01 -2.93356390e-03 -5.27869344e-01 2.33654946e-01
-4.53299552e-01 1.95988059e-01 8.87542188e-01 7.03943312e-01
4.36334610e-01 7.20015587e-03 6.53905213e-01 -6.11768484e-01
7.36299276e-01 -2.43962675e-01 -6.25867486e-01 -2.73291141e-01
-5.89835405e-01 -7.39756227e-02 5.40605426e-01 8.18899125e-02
-8.43293428e-01 1.99135885e-01 -2.55519450e-01 -3.31356198e-01
-4.08309698e-01 4.66641605e-01 -9.86784473e-02 -2.34051213e-01
2.56517738e-01 4.20593411e-01 4.44122821e-01 -1.87761664e-01
5.30432582e-01 9.02777910e-01 9.36889529e-01 2.99976707e-01
7.39865124e-01 5.70681691e-01 2.37322196e-01 -9.88746881e-01
-3.23206097e-01 -6.40777588e-01 -1.00349665e+00 -5.45859635e-01
9.50542927e-01 -8.86711836e-01 -8.90593708e-01 8.46479833e-01
-8.40698361e-01 -1.90925956e-01 -2.85593301e-01 7.48850763e-01
-5.45730054e-01 4.04531479e-01 -2.43150458e-01 -3.61882031e-01
6.94075823e-02 -1.28060126e+00 6.41338646e-01 6.39363348e-01
4.24401879e-01 -5.91615915e-01 -2.55579054e-01 1.49471641e-01
4.98517215e-01 3.57937276e-01 4.15176183e-01 -2.67356277e-01
-8.12869906e-01 -3.71480733e-01 -6.00794196e-01 3.81211460e-01
4.79972139e-02 3.05565834e-01 -8.52498293e-01 2.46313795e-01
-5.87470949e-01 2.50007749e-01 1.39363956e+00 3.56681883e-01
1.14762449e+00 3.73630792e-01 -3.03554893e-01 7.41640985e-01
1.31996584e+00 5.67118227e-01 9.72468257e-01 8.32069695e-01
9.36477780e-01 4.19822872e-01 1.07558048e+00 7.18741724e-03
5.66038787e-01 1.62694439e-01 3.99159878e-01 -3.55371684e-01
-2.65907049e-01 -2.27126047e-01 3.70913297e-01 9.84608531e-01
2.02259600e-01 7.95867145e-02 -7.90429354e-01 6.52925432e-01
-2.02301574e+00 -1.00774586e+00 -7.17322171e-01 1.53755426e+00
2.24300586e-02 5.97811282e-01 6.22777492e-02 2.00765699e-01
6.74912333e-01 5.15429556e-01 -3.28739941e-01 -6.97198033e-01
-5.36324978e-01 -4.39800262e-01 1.23333943e+00 5.85983336e-01
-1.41198218e+00 1.29559052e+00 6.68511629e+00 7.78115213e-01
-1.47003996e+00 -1.94690317e-01 3.00981879e-01 7.06840932e-01
-8.74309391e-02 -2.31264517e-01 -1.14656067e+00 5.28129160e-01
6.51166975e-01 -3.83820087e-01 -2.00343773e-01 1.05541289e+00
3.93754214e-01 -4.17079777e-01 -3.09472889e-01 1.01327491e+00
1.51208311e-01 -1.38614821e+00 8.45333040e-02 7.68813398e-03
5.65694332e-01 1.19827002e-01 2.87991494e-01 6.23698056e-01
-6.82695433e-02 -8.34481001e-01 6.13408148e-01 8.04892361e-01
6.07795358e-01 -1.21404314e+00 1.34737885e+00 3.37943792e-01
-1.35873199e+00 -2.86579669e-01 -6.61459744e-01 -1.84375063e-01
2.88790822e-01 2.58768708e-01 -8.36683571e-01 5.74186742e-01
4.28243816e-01 1.13200533e+00 -8.45067501e-01 1.36671340e+00
2.66963411e-02 3.31013143e-01 -1.89507484e-01 -2.45038584e-01
8.33475769e-01 -3.48521054e-01 3.97413969e-01 1.38046968e+00
3.03638846e-01 -6.73286393e-02 1.45949200e-01 4.26679283e-01
2.88874209e-01 2.41387352e-01 -1.15090430e+00 4.01317328e-01
2.56823570e-01 1.37548089e+00 -7.05284953e-01 -3.26455235e-01
-4.45282042e-01 5.30861318e-01 -7.92359859e-02 1.90575868e-01
-9.50788200e-01 -1.14840984e+00 5.06507039e-01 -3.48026566e-02
3.75894397e-01 -5.92911482e-01 -2.49915376e-01 -6.10896647e-01
-2.20300496e-01 -1.50458023e-01 -9.35517326e-02 -1.00348413e+00
-6.69010282e-01 4.46502179e-01 9.82145965e-02 -1.45368123e+00
-4.30297069e-02 -9.27289605e-01 -7.69906342e-01 6.57716990e-01
-1.89372683e+00 -1.09627700e+00 -6.60242081e-01 4.86139923e-01
8.23689222e-01 -3.44762594e-01 5.15954047e-02 3.03326637e-01
-9.58563745e-01 4.61060345e-01 2.75361568e-01 2.66524673e-01
4.96969581e-01 -9.81566548e-01 4.28506106e-01 9.61192548e-01
-6.75899148e-01 2.99381763e-01 4.92776483e-01 -5.44168234e-01
-1.31308579e+00 -1.39143384e+00 8.85171235e-01 9.10196046e-04
5.70993662e-01 -2.01036990e-01 -5.89510977e-01 7.14657068e-01
4.33324665e-01 1.51810572e-01 3.34379897e-02 -5.01357496e-01
1.44172922e-01 -5.22908270e-01 -1.10190535e+00 6.82374716e-01
1.04626477e+00 -2.03265741e-01 -5.36159813e-01 -2.90883482e-01
6.08282804e-01 -3.97428721e-01 -3.99897426e-01 8.58628154e-01
3.96024317e-01 -9.32054758e-01 6.98664427e-01 -1.59123410e-02
9.12584811e-02 -8.86984229e-01 3.86467241e-02 -9.98095989e-01
-6.68324530e-01 -2.98803866e-01 3.38354111e-01 8.43355477e-01
3.25738192e-01 -1.04607022e+00 7.03087270e-01 -4.39644493e-02
-8.13921750e-01 -6.71725750e-01 -8.09971988e-01 -7.96403408e-01
-1.67319328e-01 -3.96633089e-01 6.91107094e-01 7.24311113e-01
-2.70598531e-01 2.25268841e-01 -1.93266615e-01 1.97237372e-01
4.45328146e-01 -4.36496288e-02 1.12148237e+00 -1.15182066e+00
9.04471815e-01 -5.85258901e-01 -1.09058821e+00 -1.25027156e+00
2.99351454e-01 -8.83114755e-01 2.99397081e-01 -1.69925559e+00
-4.39597368e-02 -2.27136880e-01 -4.84862685e-01 3.22450846e-01
-4.11998518e-02 3.30852568e-01 1.38295248e-01 -1.79228514e-01
-7.26150513e-01 6.11464620e-01 1.66947103e+00 -3.42045754e-01
-1.39300168e-01 -2.39623576e-01 -3.11589241e-01 7.44809568e-01
1.27878511e+00 3.12221944e-01 -4.22986388e-01 -2.21967008e-02
-2.24812612e-01 -3.45232993e-01 3.47577840e-01 -1.57098222e+00
4.17007595e-01 -1.36162579e-01 2.98465222e-01 -1.65244234e+00
1.40487880e-01 -7.71307349e-01 -1.62151054e-01 6.21019125e-01
4.90951631e-03 3.92036676e-01 5.25808752e-01 4.36116099e-01
-5.23963034e-01 -7.24716187e-02 6.03724539e-01 2.95345873e-01
-2.05701542e+00 1.11708231e-01 -7.98415363e-01 -4.76255983e-01
1.33852911e+00 -8.44426692e-01 -5.41916132e-01 -1.52208224e-01
-5.02977371e-01 4.66470093e-01 5.91345392e-02 7.60695159e-01
9.31662381e-01 -1.65867591e+00 -4.84983861e-01 5.78222156e-01
3.18183333e-01 -1.63006186e-02 3.83409232e-01 7.71588087e-01
-9.96394992e-01 7.54749537e-01 -7.90943027e-01 -6.28168166e-01
-1.18705046e+00 4.41985577e-01 3.47616464e-01 4.18654770e-01
-8.73850644e-01 4.16634232e-01 7.92688727e-02 -3.76991957e-01
-1.15951598e-01 -5.46502352e-01 -8.18347871e-01 6.14849143e-02
2.55421430e-01 6.80668652e-01 -1.42644569e-01 -1.22515643e+00
-1.24568641e-01 8.03675354e-01 -1.01610556e-01 1.39830276e-01
9.35459673e-01 -4.45113540e-01 -1.32218683e-02 5.35638928e-01
1.35604310e+00 1.87027261e-01 -1.31732666e+00 1.71417609e-01
-3.00810933e-01 -5.09288907e-01 2.34184161e-01 -2.33695507e-01
-1.04682720e+00 1.05407858e+00 1.01024556e+00 1.27441332e-01
6.13134801e-01 -5.18660426e-01 1.02913189e+00 5.88222444e-01
5.05460322e-01 -1.22330272e+00 -3.39371651e-01 1.14512646e+00
8.59397829e-01 -1.46791816e+00 -1.85292378e-01 -5.58508158e-01
-5.63155472e-01 1.38776219e+00 8.97612035e-01 -4.56808418e-01
9.59932387e-01 1.86530605e-01 3.91979009e-01 -5.70949353e-02
-2.43868560e-01 -6.79121673e-01 5.48439212e-02 4.83448297e-01
-9.97149348e-02 1.27368197e-01 -6.64233863e-01 1.81931704e-01
-5.16885340e-01 -2.28751957e-01 7.22064078e-01 9.16646063e-01
-1.05996644e+00 -5.43407917e-01 -1.30603522e-01 4.32293832e-01
1.64310172e-01 2.41922498e-01 7.51561001e-02 1.27463472e+00
4.80563611e-01 6.37698591e-01 3.69296461e-01 -8.66893351e-01
5.64080119e-01 -1.12261049e-01 -1.24646150e-01 -5.71924299e-02
1.55087069e-01 -1.84363574e-01 6.05846895e-03 -7.27201104e-01
-1.12932310e-01 -6.17940068e-01 -1.65661931e+00 -4.84074831e-01
-2.20181853e-01 3.15563045e-02 8.54615152e-01 5.92652082e-01
4.84608859e-01 7.02870131e-01 8.77617717e-01 -8.38523507e-01
9.02082324e-02 -7.16895401e-01 -6.52366102e-01 -4.42253575e-02
5.21688581e-01 -7.67465234e-01 -3.33603889e-01 -1.79661110e-01] | [8.06304931640625, -1.3873035907745361] |
2f4845c0-8c4c-4d35-beeb-0c80db23144b | exploiting-wordnet-synset-and-hypernym | null | null | https://aclanthology.org/2020.aacl-main.14 | https://aclanthology.org/2020.aacl-main.14.pdf | Exploiting WordNet Synset and Hypernym Representations for Answer Selection | Answer selection (AS) is an important subtask of document-based question answering (DQA). In this task, the candidate answers come from the same document, and each answer sentence is semantically related to the given question, which makes it more challenging to select the true answer. WordNet provides powerful knowledge about concepts and their semantic relations so we employ WordNet to enrich the abilities of paraphrasing and reasoning of the network-based question answering model. Specifically, we exploit the synset and hypernym concepts to enrich the word representation and incorporate the similarity scores of two concepts that share the synset or hypernym relations into the attention mechanism. The proposed WordNet-enhanced hierarchical model (WEHM) consists of four modules, including WordNet-enhanced word representation, sentence encoding, WordNet-enhanced attention mechanism, and hierarchical document encoding. Extensive experiments on the public WikiQA and SelQA datasets demonstrate that our proposed model significantly improves the baseline system and outperforms all existing state-of-the-art methods by a large margin. | ['Yunfang Wu', 'Weikang Li'] | 2020-12-01 | null | null | null | asian-chapter-of-the-association-for | ['answer-selection'] | ['natural-language-processing'] | [ 3.32306847e-02 1.42048344e-01 -8.98622125e-02 -4.36270595e-01
-7.69213676e-01 -4.55852747e-01 3.31245482e-01 5.77467203e-01
-6.53500140e-01 5.50498128e-01 7.58621156e-01 -4.06214476e-01
-4.57009017e-01 -1.18759072e+00 -2.52792746e-01 -1.17004141e-01
4.04973119e-01 8.67494464e-01 8.03461492e-01 -8.64626586e-01
5.62178552e-01 -1.49396122e-01 -1.53594184e+00 6.61199868e-01
1.29794312e+00 1.07966197e+00 3.39434475e-01 5.17903566e-01
-1.03576112e+00 1.25364602e+00 -7.11074352e-01 -7.46862411e-01
-3.06535661e-01 -6.74766779e-01 -1.62438798e+00 -4.57605362e-01
6.14874482e-01 -2.54767478e-01 -6.08374417e-01 9.79313016e-01
4.47687507e-01 5.74424505e-01 3.33531350e-01 -8.77630293e-01
-1.09114885e+00 5.72022617e-01 -5.21439798e-02 5.03251493e-01
7.95201480e-01 -1.39387563e-01 1.74730396e+00 -1.00515699e+00
4.34872329e-01 1.63581026e+00 1.64035603e-01 5.43531120e-01
-5.48223078e-01 -5.16524494e-01 1.08604640e-01 7.93796182e-01
-1.19091845e+00 -9.20950100e-02 6.56149149e-01 -7.49808410e-03
1.38640249e+00 2.33682409e-01 3.38724196e-01 5.16939521e-01
-8.77640471e-02 8.86340380e-01 5.35250068e-01 -3.97972643e-01
2.32060209e-01 -1.63835317e-01 8.72458994e-01 7.93705165e-01
1.48051888e-01 -6.90831721e-01 -4.52488869e-01 -5.00548899e-01
2.42109671e-01 8.43209848e-02 -4.12456572e-01 8.72437842e-03
-1.01195514e+00 1.16554964e+00 6.17159426e-01 2.36050248e-01
-3.48814845e-01 2.42706582e-01 5.12805045e-01 6.14083350e-01
7.71696940e-02 1.01444256e+00 -3.61934751e-01 2.67145663e-01
-3.68170440e-01 5.69543362e-01 8.57472599e-01 9.99795139e-01
8.74724746e-01 -5.25151491e-01 -8.10279071e-01 1.12795901e+00
4.45890844e-01 3.62386793e-01 8.93529058e-01 -9.87876356e-01
8.67897868e-01 1.13552260e+00 -9.50069502e-02 -1.36427963e+00
-2.62124211e-01 -4.44386095e-01 -5.18642128e-01 -8.09718430e-01
-5.31320088e-02 1.25903353e-01 -6.88594401e-01 1.54238057e+00
3.38519245e-01 1.50297746e-01 3.78973126e-01 1.03125226e+00
1.65282857e+00 9.02909398e-01 3.06017727e-01 9.60711241e-02
1.75737858e+00 -1.40670133e+00 -8.91595423e-01 -5.65452278e-01
7.22652078e-01 -6.03617907e-01 1.30422843e+00 -3.07545334e-01
-9.55785275e-01 -6.79320693e-01 -9.85363901e-01 -6.28511727e-01
-4.76487875e-01 -2.84331352e-01 5.37529767e-01 7.98699856e-02
-7.32634723e-01 2.53312975e-01 1.24518447e-01 -5.05423546e-01
3.27312320e-01 1.42380655e-01 1.73047313e-03 -7.53870130e-01
-2.01992583e+00 1.05085742e+00 5.75827599e-01 -1.70208782e-01
-5.03063917e-01 -5.69186509e-01 -1.01930439e+00 4.08984214e-01
5.37424147e-01 -1.22159863e+00 1.23305118e+00 -3.23534548e-01
-9.43798542e-01 8.10865521e-01 -5.36387503e-01 -3.53712499e-01
-4.94168639e-01 -3.75820249e-02 -5.81541181e-01 5.31073987e-01
4.02171165e-01 6.40315890e-01 5.07404447e-01 -8.41605961e-01
-5.44871449e-01 -3.25826973e-01 6.49896026e-01 5.38660049e-01
-4.93165106e-01 1.13975391e-01 -5.72712600e-01 -4.82149869e-01
2.67332941e-01 -1.32445961e-01 -1.07252643e-01 -3.08945000e-01
-7.10064694e-02 -9.65729713e-01 4.98369366e-01 -6.75569892e-01
1.56961322e+00 -1.75588322e+00 3.72830816e-02 8.32648724e-02
4.07554865e-01 2.48018682e-01 -6.41826928e-01 6.33470416e-01
1.41453966e-01 -7.28541911e-02 -2.78996736e-01 1.87098190e-01
4.87640016e-02 4.66231972e-01 -6.46283507e-01 -4.18962121e-01
2.45978341e-01 1.18680155e+00 -1.38079858e+00 -6.62567556e-01
-2.87455857e-01 -2.06777409e-01 -5.62599897e-01 4.96115178e-01
-5.13159394e-01 -2.32929155e-01 -8.61446679e-01 5.35775363e-01
4.39834952e-01 -4.60615069e-01 -8.18989053e-02 -2.40476236e-01
8.37207437e-01 9.26398635e-01 -8.74339342e-01 1.88900089e+00
-5.78049183e-01 1.37386292e-01 -4.90948588e-01 -8.65182340e-01
1.14564335e+00 2.44236693e-01 -1.20461695e-01 -1.12249887e+00
1.84562951e-02 3.94894034e-01 2.12232888e-01 -1.07960665e+00
7.52984941e-01 -9.59780812e-02 5.05197048e-02 6.24527633e-01
5.42526364e-01 -3.70436311e-01 4.11674201e-01 8.17799449e-01
1.36366379e+00 -2.60238916e-01 2.87850916e-01 -3.84560108e-01
1.05256736e+00 6.37725443e-02 2.15319633e-01 7.56193876e-01
3.85344289e-02 4.60830629e-01 4.10976201e-01 -1.86188310e-01
-4.52705979e-01 -9.58949864e-01 1.99135154e-01 1.31656802e+00
5.04589558e-01 -6.69323683e-01 -5.15909910e-01 -9.97510016e-01
1.10286817e-01 9.83771503e-01 -5.56143522e-01 -7.10453153e-01
-6.58598006e-01 -3.18996936e-01 4.15245622e-01 6.63509727e-01
5.41030765e-01 -1.24363863e+00 -1.05441377e-01 3.05487543e-01
-6.31731331e-01 -1.03631186e+00 -6.43463552e-01 -2.15725094e-01
-8.00358713e-01 -1.23513305e+00 -5.61057150e-01 -1.02828956e+00
5.24637938e-01 4.52981859e-01 1.80569458e+00 6.41557872e-01
9.45830271e-02 4.62368637e-01 -6.47575855e-01 -1.14920631e-01
9.75664631e-02 1.12468630e-01 -4.08229917e-01 -1.49623320e-01
9.11498189e-01 -2.53350198e-01 -8.48099589e-01 2.09001318e-01
-1.01270640e+00 -4.72223908e-01 1.99205890e-01 8.60643148e-01
3.23465258e-01 -1.92128032e-01 1.24742186e+00 -9.50230837e-01
1.40885532e+00 -8.44305396e-01 -1.21548930e-02 7.15561986e-01
-4.90851611e-01 1.62048176e-01 6.36144221e-01 -6.26324639e-02
-1.04753137e+00 -6.74919903e-01 -5.51029682e-01 1.00753553e-01
2.60401577e-01 9.09905672e-01 -3.06305438e-01 1.16412818e-01
4.79820132e-01 1.59409314e-01 -3.16068470e-01 -2.83372313e-01
6.64843619e-01 6.09685361e-01 4.89799887e-01 -6.85073614e-01
4.98613358e-01 1.13740318e-01 -2.79085696e-01 -3.82931858e-01
-1.63828874e+00 -1.04292607e+00 -2.28277415e-01 3.40361781e-02
8.42607319e-01 -6.72580421e-01 -5.77774882e-01 -9.93265137e-02
-1.58342361e+00 4.93737727e-01 -4.06023830e-01 2.69398510e-01
-1.87762380e-01 3.81531328e-01 -4.99105960e-01 -4.90074754e-01
-7.55102754e-01 -8.85462940e-01 8.61856282e-01 4.61770922e-01
-4.31404978e-01 -1.08994567e+00 1.26889855e-01 9.00665164e-01
5.23514569e-01 -5.17101288e-01 1.65962815e+00 -1.26829541e+00
-3.93633425e-01 -2.07086951e-01 -4.30795401e-01 3.08551908e-01
4.29128446e-02 -7.30394721e-01 -6.76125228e-01 -3.87113690e-02
1.70391023e-01 -7.52248287e-01 1.37065780e+00 -1.94036707e-01
1.30384147e+00 -2.95445979e-01 9.04951170e-02 -8.83351639e-02
1.22595358e+00 -1.17161579e-03 6.44670665e-01 1.45297408e-01
6.45611644e-01 1.02011967e+00 6.56731606e-01 1.81855727e-02
8.39060247e-01 2.60219723e-01 5.14118969e-01 3.51131022e-01
-3.61258626e-01 -4.95791286e-01 -1.04439616e-01 1.42025137e+00
5.29018998e-01 -4.69424963e-01 -8.94527793e-01 9.27704453e-01
-1.83840275e+00 -8.28760862e-01 -3.70346129e-01 1.93069530e+00
1.05443335e+00 -1.83294192e-01 -2.99555004e-01 -7.16645569e-02
6.34261727e-01 2.79468596e-01 -4.58137274e-01 -3.30409527e-01
-1.98285833e-01 7.10310221e-01 -1.80086702e-01 8.03550839e-01
-5.41771173e-01 9.60984945e-01 5.59746981e+00 1.10072792e+00
-3.80004905e-02 1.83854043e-01 2.97552556e-01 3.06204617e-01
-9.45596457e-01 9.60002616e-02 -7.29428172e-01 2.43753105e-01
6.26268208e-01 -4.00426626e-01 5.79276942e-02 5.95654905e-01
-4.68195230e-01 8.76388624e-02 -1.03936100e+00 6.57246828e-01
7.13244915e-01 -1.56383193e+00 7.59418190e-01 -7.75962710e-01
6.25172794e-01 -1.66257843e-01 -3.23244452e-01 7.71415472e-01
2.48284549e-01 -9.85645115e-01 3.38939652e-02 6.45878494e-01
2.61765212e-01 -8.58862340e-01 1.28425181e+00 9.54867080e-02
-1.22000575e+00 -2.59897768e-01 -8.30298126e-01 -2.46562473e-02
1.56256661e-01 2.91807979e-01 -3.14417392e-01 6.23625398e-01
5.73621631e-01 5.96102178e-01 -7.98061669e-01 1.02427840e+00
-6.60714746e-01 7.07782447e-01 8.89654383e-02 -4.87977833e-01
6.37256622e-01 2.80774012e-02 2.62829572e-01 9.68636096e-01
1.89108223e-01 5.59506714e-01 -5.16339019e-02 8.48174810e-01
-5.82459688e-01 4.24263895e-01 -1.27206698e-01 -4.60902117e-02
6.84205234e-01 1.12059510e+00 -1.73470810e-01 -5.42533278e-01
-5.91575861e-01 1.01945972e+00 5.29890299e-01 3.55702102e-01
-3.08155209e-01 -1.15111387e+00 6.14830792e-01 -1.94763333e-01
1.35849789e-01 2.44674489e-01 4.34020720e-02 -1.22419930e+00
1.47735894e-01 -9.46174443e-01 1.08658993e+00 -9.99810040e-01
-1.83383965e+00 5.51972628e-01 -3.54742885e-01 -7.51130521e-01
-1.05929740e-01 -4.67060387e-01 -9.97811675e-01 1.26303172e+00
-1.88278782e+00 -8.64570737e-01 -4.26679581e-01 6.15328193e-01
5.85167170e-01 -2.83676982e-01 9.61589098e-01 3.43314946e-01
-6.87727183e-02 3.66416872e-01 -1.72576532e-01 1.81805387e-01
5.88969350e-01 -1.15389681e+00 4.70897764e-01 5.74995100e-01
1.64273918e-01 9.48828757e-01 2.41691887e-01 -5.56705654e-01
-1.24296165e+00 -1.13277054e+00 1.54133081e+00 -4.90674883e-01
7.72460282e-01 -1.50662404e-03 -1.40268874e+00 3.17455888e-01
5.70584655e-01 -2.42059723e-01 9.56638396e-01 1.14327230e-01
-5.94284832e-01 -9.95039567e-02 -8.73258710e-01 5.27037978e-01
8.42867434e-01 -8.46637070e-01 -1.56681716e+00 5.93702018e-01
1.49823058e+00 6.04694895e-02 -6.05886281e-01 3.79693121e-01
-4.65793870e-02 -5.48683286e-01 1.11122739e+00 -1.22594845e+00
6.92127228e-01 -4.84182447e-01 -3.43232960e-01 -1.17651188e+00
-3.58245164e-01 -5.84714748e-02 -6.08169079e-01 1.31909907e+00
6.35766804e-01 -5.88860035e-01 5.14347553e-01 5.30824602e-01
-1.85994327e-01 -1.05108082e+00 -9.39491808e-01 -6.17578566e-01
1.68363512e-01 -1.30671382e-01 9.94718432e-01 9.75532413e-01
2.23875165e-01 1.05316520e+00 2.00110346e-01 6.85762323e-04
1.91919282e-01 4.13764834e-01 1.15543544e-01 -1.23837984e+00
-1.30209967e-01 -4.65532511e-01 -1.91224143e-01 -1.60336947e+00
4.64396030e-01 -1.24040902e+00 -2.69137342e-02 -2.14333367e+00
3.95151347e-01 -1.85260326e-01 -5.23038924e-01 1.99437857e-01
-9.04739141e-01 -1.36515811e-01 -4.32134829e-02 -4.22588503e-03
-1.23453259e+00 9.85575199e-01 1.51902890e+00 -3.68652344e-01
4.05118614e-01 -3.59603524e-01 -1.00638258e+00 5.71315229e-01
5.95142841e-01 -4.93998259e-01 -8.52521896e-01 -9.75398362e-01
7.85654068e-01 -5.27055338e-02 2.43081138e-01 -4.90588099e-01
6.62000179e-01 4.59972816e-03 -8.74293521e-02 -4.39991236e-01
2.12091401e-01 -7.11541831e-01 -1.08057463e+00 4.17988509e-01
-1.00166416e+00 2.00645953e-01 -1.17922790e-01 5.93670011e-01
-7.11964905e-01 -9.61652160e-01 3.47252816e-01 -3.10311973e-01
-8.71997833e-01 3.82323563e-01 -1.60338894e-01 9.42518651e-01
3.18320692e-01 1.34925500e-01 -6.85791612e-01 -5.48514068e-01
-1.11359827e-01 9.21016395e-01 -3.58190775e-01 7.48537779e-01
1.21468449e+00 -1.58138633e+00 -8.24276328e-01 -1.82553828e-01
7.16412127e-01 -2.01536436e-02 4.68258739e-01 2.26321384e-01
-4.75799799e-01 5.96957624e-01 2.09878087e-01 5.93941323e-02
-9.43619728e-01 4.88026291e-01 3.74031782e-01 -3.51486713e-01
-2.12654397e-01 1.30105138e+00 1.10014923e-01 -9.18127239e-01
1.16276704e-01 -2.10333303e-01 -9.22458827e-01 -2.27213800e-02
8.82884145e-01 3.16606909e-01 1.03246383e-01 -3.20285171e-01
-3.32582295e-01 5.35772502e-01 -2.13173106e-01 2.02539250e-01
8.42669070e-01 -2.06599191e-01 -7.85174787e-01 1.09458596e-01
1.25170338e+00 -3.12977821e-01 -4.10185568e-02 -7.65505373e-01
2.21981555e-01 -4.17079449e-01 -8.78722519e-02 -8.06893766e-01
-7.15506554e-01 1.08577466e+00 -4.81754392e-02 7.43140057e-02
9.65052009e-01 3.70485753e-01 1.42576635e+00 9.77256894e-01
4.03725803e-02 -9.75215793e-01 4.90882933e-01 1.03917384e+00
1.08681464e+00 -1.05866992e+00 -2.95382440e-01 -4.37645495e-01
-5.78570664e-01 1.02423656e+00 1.20328093e+00 -1.21601105e-01
3.12699437e-01 -6.08912766e-01 -1.79876573e-02 -5.89780450e-01
-9.09260690e-01 -4.60205585e-01 7.07348049e-01 3.25494170e-01
4.75460351e-01 -4.84192312e-01 -6.43109560e-01 1.01806426e+00
-9.55746174e-02 -4.91792560e-01 7.29709417e-02 7.12712526e-01
-9.02654588e-01 -8.50792825e-01 1.25333384e-01 6.84044182e-01
-1.63326800e-01 -6.00699782e-01 -4.28240001e-01 7.97219351e-02
-1.60798989e-02 1.36291265e+00 3.96853350e-02 -3.40249538e-01
4.64872479e-01 2.98741668e-01 1.90156326e-01 -9.91920531e-01
-8.93904567e-01 -9.01050329e-01 2.75360405e-01 -4.76700664e-01
-2.26465464e-01 -1.58508494e-02 -1.56179070e+00 1.47498713e-03
-4.49819565e-01 7.82088161e-01 3.60285165e-04 1.36058104e+00
4.56697136e-01 6.82823479e-01 3.92154098e-01 5.26518703e-01
-5.24821281e-01 -1.01008272e+00 -3.19470912e-01 6.12451196e-01
1.11012138e-01 -3.27388644e-01 -1.48523837e-01 -4.69420850e-01] | [11.048158645629883, 8.047056198120117] |
6fcc24da-be7c-4ab5-960b-f6b86f542a10 | on-the-utility-of-self-supervised-models-for | 2210.07185 | null | https://arxiv.org/abs/2210.07185v2 | https://arxiv.org/pdf/2210.07185v2.pdf | On the Utility of Self-supervised Models for Prosody-related Tasks | Self-Supervised Learning (SSL) from speech data has produced models that have achieved remarkable performance in many tasks, and that are known to implicitly represent many aspects of information latently present in speech signals. However, relatively little is known about the suitability of such models for prosody-related tasks or the extent to which they encode prosodic information. We present a new evaluation framework, SUPERB-prosody, consisting of three prosody-related downstream tasks and two pseudo tasks. We find that 13 of the 15 SSL models outperformed the baseline on all the prosody-related tasks. We also show good performance on two pseudo tasks: prosody reconstruction and future prosody prediction. We further analyze the layerwise contributions of the SSL models. Overall we conclude that SSL speech models are highly effective for prosody-related tasks. | ['Nigel G. Ward', 'Hung-Yi Lee', 'Chen-An Li', 'Tzu-Han Lin', 'Yuan Tseng', 'Wei-Ping Huang', 'Chi-Luen Feng', 'Guan-Ting Lin'] | 2022-10-13 | null | null | null | null | ['prosody-prediction'] | ['natural-language-processing'] | [ 1.09389566e-01 3.76436234e-01 -7.49748647e-01 -6.22922421e-01
-1.00794327e+00 -5.17765760e-01 5.33523619e-01 -4.99452837e-02
-2.22267166e-01 6.95429504e-01 1.17652416e+00 1.65285934e-02
2.24130422e-01 -9.22869220e-02 -4.91267115e-01 -3.92585963e-01
-3.65042947e-02 2.24490345e-01 2.60400057e-01 -3.35377574e-01
-6.51297942e-02 1.41123328e-02 -1.30685663e+00 8.62126231e-01
9.13768634e-02 8.75085831e-01 3.23363632e-01 4.88738745e-01
-3.35578680e-01 9.94964063e-01 -6.01597309e-01 -1.99780956e-01
-1.70755893e-01 -4.54137892e-01 -8.59824538e-01 -9.15437117e-02
1.70238853e-01 5.22958003e-02 4.48937342e-02 7.67451942e-01
4.62014139e-01 6.47680461e-02 2.78508335e-01 -9.46077466e-01
-3.82098109e-01 1.64213371e+00 -3.96633893e-02 4.38766479e-01
2.52365381e-01 -1.52574815e-02 1.50913990e+00 -1.05129683e+00
4.73580956e-01 1.50216663e+00 7.65122771e-01 9.34837997e-01
-1.62783420e+00 -8.09420764e-01 2.87061453e-01 -1.13552362e-01
-8.11319828e-01 -1.27219737e+00 1.29138768e+00 -3.50827307e-01
1.23972023e+00 3.14681947e-01 3.82411212e-01 1.42012525e+00
-6.20842092e-02 1.28509736e+00 1.12837386e+00 -3.85482252e-01
1.33676782e-01 2.59633124e-01 5.13836086e-01 -9.46603995e-03
-6.51614010e-01 3.90350997e-01 -1.29273343e+00 -2.69483496e-02
3.20535302e-01 -8.59954596e-01 -1.23701900e-01 -2.46808454e-02
-1.09851873e+00 6.94231868e-01 3.13140750e-02 4.23655182e-01
-3.50656867e-01 3.19687426e-02 8.25306475e-01 3.71890366e-01
9.63232040e-01 4.57444221e-01 -8.80840838e-01 -3.92261624e-01
-1.10534513e+00 2.34988872e-02 7.69483447e-01 8.15546930e-01
2.91804671e-01 4.81476307e-01 -1.37752384e-01 1.67609179e+00
4.33087438e-01 2.34549008e-02 6.69597805e-01 -1.13860273e+00
6.08685255e-01 -2.19103232e-01 -1.78594649e-01 -1.19720273e-01
-5.15648782e-01 -1.66186407e-01 -2.81858206e-01 -4.19207752e-01
1.83057398e-01 -2.30235547e-01 -4.64051276e-01 2.34163499e+00
-1.65830538e-01 2.77635306e-02 2.50903934e-01 6.00722134e-01
8.14024806e-01 1.20942497e+00 6.43960714e-01 -8.86606038e-01
1.14095032e+00 -1.20028985e+00 -1.12457776e+00 -6.00417435e-01
1.56755477e-01 -9.11877036e-01 1.29483509e+00 2.79829621e-01
-1.54369223e+00 -8.53149831e-01 -9.87948120e-01 -1.14753753e-01
4.28312346e-02 -2.33937986e-02 6.92184269e-01 8.34908843e-01
-1.06241214e+00 3.39259267e-01 -7.34939873e-01 1.88304946e-01
7.18411878e-02 1.55942053e-01 -8.99988413e-02 6.77065551e-01
-1.36569071e+00 6.75760567e-01 5.88146806e-01 -1.79455444e-01
-7.50331938e-01 -1.00532675e+00 -1.06489718e+00 1.89807698e-01
1.11429535e-01 -6.70423210e-02 1.77684140e+00 -1.03046131e+00
-1.86773753e+00 8.95871878e-01 -3.67920280e-01 -8.36693764e-01
-1.14812575e-01 -3.49464774e-01 -5.80955982e-01 -1.93709247e-02
-1.49223864e-01 1.01564860e+00 7.38336384e-01 -1.06352830e+00
-4.84628230e-01 -3.78463417e-02 -6.00528717e-01 4.48711365e-01
-3.07301223e-01 5.81184864e-01 -1.52089342e-01 -9.07765687e-01
-6.72998279e-02 -7.28433609e-01 1.59447223e-01 -4.57984000e-01
-2.46320307e-01 -4.71885234e-01 4.62520748e-01 -7.21694946e-01
1.45729625e+00 -2.35785580e+00 8.10358152e-02 -4.20056581e-01
-3.89006406e-01 2.26434723e-01 -1.74749658e-01 3.92110854e-01
-4.57567096e-01 1.44996420e-01 -2.48088941e-01 -9.38561499e-01
2.12736785e-01 2.56459028e-01 -1.00389469e+00 -1.99704930e-01
3.88217270e-01 9.33429837e-01 -6.57320321e-01 -2.55773097e-01
1.55451179e-01 3.18400681e-01 -4.21011150e-01 1.14013463e-01
-6.17822707e-01 4.67817068e-01 1.50972173e-01 3.38049173e-01
2.01167122e-01 4.49670523e-01 3.40971977e-01 8.07896629e-02
-3.63750190e-01 1.50263107e+00 -4.90826249e-01 1.62249541e+00
-6.08832061e-01 7.79205918e-01 4.57443118e-01 -5.99436104e-01
8.54092598e-01 9.54417109e-01 4.26855087e-01 -4.67706680e-01
-3.00090462e-01 2.30227429e-02 3.21665823e-01 -1.73452273e-01
3.98780704e-01 -9.16187406e-01 -2.45654285e-01 3.34721357e-01
3.50289822e-01 -3.57604414e-01 -1.42211139e-01 -2.26767465e-01
7.33275950e-01 5.57590388e-02 4.99389440e-01 -3.51229578e-01
5.53657413e-01 -6.44503161e-02 8.42108190e-01 4.59650189e-01
-5.68372130e-01 4.17049259e-01 4.26943332e-01 -9.01571214e-02
-9.80480433e-01 -1.27387559e+00 -3.35049868e-01 1.64789319e+00
-3.87085408e-01 -5.72104871e-01 -6.94885850e-01 -3.06782246e-01
-3.43993217e-01 1.19363189e+00 -9.01905894e-02 -3.98225375e-02
-8.54758084e-01 -5.92224360e-01 8.17670405e-01 7.24109769e-01
7.14663938e-02 -1.93247104e+00 8.10101479e-02 5.13258755e-01
-3.64731729e-01 -1.30841899e+00 -7.10009754e-01 4.82495129e-01
-9.56228137e-01 -4.34778720e-01 -4.63948250e-01 -1.09444368e+00
-2.74615586e-01 -4.28034887e-02 1.09839046e+00 -5.89948237e-01
4.58365202e-01 2.03707248e-01 -2.47506440e-01 -3.78828794e-01
-1.01970565e+00 -1.24052325e-02 2.33071193e-01 -1.17222108e-01
4.63697344e-01 -6.74486458e-01 1.40816821e-02 1.96649477e-01
-3.82421941e-01 1.33656934e-01 2.86782742e-01 7.31011927e-01
5.77922881e-01 -2.24446412e-02 1.17411435e+00 -8.60545218e-01
9.32016373e-01 -3.96489292e-01 -1.27450213e-01 -9.64373872e-02
-1.78897262e-01 5.10829221e-03 4.80577707e-01 -5.26801825e-01
-1.56825161e+00 1.53774440e-01 -8.91341925e-01 -1.79003671e-01
-4.73108381e-01 5.28144598e-01 -3.47611815e-01 7.74911344e-01
5.04572868e-01 1.70210838e-01 6.97518066e-02 -8.52261901e-01
4.52573806e-01 7.90583909e-01 5.31629264e-01 -6.74367070e-01
2.10786119e-01 2.44757049e-02 -6.40584171e-01 -1.40512300e+00
-1.25744474e+00 -7.28593946e-01 -3.92049253e-01 1.30711421e-01
9.52744842e-01 -1.14499986e+00 -3.27036023e-01 4.30507094e-01
-1.32659900e+00 -7.63221323e-01 -5.75952649e-01 5.24195194e-01
-1.06076980e+00 3.57763827e-01 -1.17464948e+00 -1.23031795e+00
-4.50714290e-01 -9.83517230e-01 8.70772719e-01 -2.74887890e-01
-8.69853437e-01 -1.05778027e+00 2.90208042e-01 4.75562215e-01
2.59506583e-01 -4.25319642e-01 1.17437136e+00 -8.04563820e-01
7.47245504e-03 4.43935812e-01 2.82769531e-01 8.13825846e-01
2.89089084e-01 -5.30908883e-01 -1.43704772e+00 8.29575956e-02
5.34779131e-01 -6.68945372e-01 1.17615962e+00 7.96083689e-01
8.20218086e-01 -2.82603145e-01 -2.28530876e-02 2.97678083e-01
5.68582177e-01 3.63672733e-01 2.34328061e-01 -3.00939023e-01
2.11681992e-01 1.01924753e+00 4.99890685e-01 -1.23076342e-01
2.35945717e-01 8.55321348e-01 -1.38455480e-01 1.59775138e-01
-6.74812198e-01 -6.69873416e-01 1.07330513e+00 1.46885896e+00
4.33357716e-01 -5.88533618e-02 -6.99154615e-01 7.95989215e-01
-1.73445797e+00 -1.06734395e+00 3.79002206e-02 1.82853758e+00
1.50343049e+00 6.39595807e-01 3.96886826e-01 2.15208754e-01
6.03991568e-01 6.62847400e-01 -3.72525007e-01 -6.27602279e-01
-2.99337059e-01 3.96861643e-01 -1.53546885e-01 8.47035170e-01
-1.24969804e+00 1.53274763e+00 7.68365526e+00 7.16924727e-01
-8.14814627e-01 3.40164125e-01 4.13359433e-01 6.64665550e-02
-2.67392099e-01 -8.59410763e-02 -1.07500255e+00 3.85425597e-01
1.31957853e+00 -5.78795327e-03 3.18975925e-01 9.94841933e-01
4.81932521e-01 1.92639351e-01 -1.41311836e+00 7.33334839e-01
-1.94029044e-02 -1.28324211e+00 -7.30285123e-02 -3.45880330e-01
4.40181524e-01 1.96189120e-01 1.76194564e-01 6.21856153e-01
5.21344990e-02 -7.35034764e-01 1.04127026e+00 4.70028706e-02
5.68356097e-01 -5.07736921e-01 2.50176042e-01 6.79414511e-01
-1.15401852e+00 2.98306393e-03 -2.09295288e-01 -2.47146383e-01
7.15568662e-01 3.14976349e-02 -1.09333110e+00 -1.33004799e-01
4.12993580e-01 6.96842551e-01 -1.92579273e-02 5.94355166e-01
-6.43979430e-01 1.49543810e+00 -1.98275670e-01 1.63975433e-01
1.69887483e-01 2.62232751e-01 8.65821242e-01 1.53297937e+00
-2.97939986e-01 -8.92863572e-02 8.17838591e-03 1.01829875e+00
7.77143985e-02 1.55077040e-01 -4.44469303e-01 -3.10122699e-01
6.00109458e-01 7.14211345e-01 -4.31999922e-01 -4.90347818e-02
-4.25111294e-01 4.78323430e-01 2.62701184e-01 1.51612073e-01
-4.50639635e-01 3.88877898e-01 1.03710580e+00 -5.09959226e-03
1.64062545e-01 -3.95899236e-01 -3.49559665e-01 -9.91825402e-01
-1.32717639e-01 -7.90889978e-01 4.12802219e-01 -8.10183764e-01
-1.47585475e+00 6.59340918e-01 1.33052647e-01 -6.74663186e-01
-7.36825645e-01 -6.49237990e-01 -6.53095782e-01 8.32088709e-01
-1.57130408e+00 -1.08634055e+00 6.33041322e-01 3.18926573e-01
1.54479039e+00 -4.18163806e-01 1.16103864e+00 -1.15432538e-01
-4.57562238e-01 1.97788253e-01 -5.04471898e-01 2.08929166e-01
6.92537189e-01 -1.33067513e+00 8.04135323e-01 5.86424887e-01
5.97266793e-01 5.95877111e-01 6.58726633e-01 -5.86130798e-01
-5.50852358e-01 -5.75379729e-01 1.13781714e+00 -5.12566924e-01
8.54521215e-01 -6.79897189e-01 -9.44733262e-01 9.40652609e-01
5.28697014e-01 -4.21847016e-01 1.13927186e+00 5.96654713e-01
-2.06188485e-01 -2.28606886e-03 -5.16637683e-01 4.55916822e-01
9.91181493e-01 -9.05116677e-01 -1.31936765e+00 -4.65897508e-02
1.33337224e+00 -4.98247631e-02 -6.01751924e-01 3.74121219e-01
4.06009465e-01 -9.86199021e-01 1.03456354e+00 -7.07171381e-01
3.82024944e-01 2.24661931e-01 -3.84626061e-01 -1.40943646e+00
-1.98857576e-01 -8.18314731e-01 -2.31468022e-01 1.57698715e+00
8.03268194e-01 -1.94226295e-01 6.17776096e-01 3.63150835e-01
-6.83179438e-01 -2.53850758e-01 -1.04284060e+00 -9.14374709e-01
3.15240085e-01 -1.00061679e+00 2.80013472e-01 8.58201504e-01
5.78113139e-01 1.05087376e+00 -4.69185352e-01 -7.08191544e-02
3.79618227e-01 -5.53691983e-02 2.44663283e-01 -1.00943613e+00
-5.07221758e-01 -6.24097168e-01 1.22805066e-01 -1.24216735e+00
9.55200911e-01 -1.02991939e+00 2.65374869e-01 -1.04883456e+00
-8.16621706e-02 -3.47414464e-01 -5.85899413e-01 7.41175532e-01
1.16593324e-01 -4.47578803e-02 3.63891542e-01 2.60541290e-01
-1.86661631e-01 6.70168877e-01 7.35052884e-01 7.63763934e-02
-7.62346923e-01 5.13153076e-01 -6.23339534e-01 9.94851649e-01
1.05631995e+00 -2.56950855e-01 -5.79930663e-01 -1.63223773e-01
-1.18745349e-01 5.50321341e-01 -7.56672770e-02 -6.77269757e-01
8.59330222e-02 -1.35528296e-01 9.35998000e-03 -7.54539251e-01
9.79846418e-01 -4.83288229e-01 -1.81009755e-01 5.97292855e-02
-8.70973110e-01 -5.86793959e-01 5.82631350e-01 1.35450438e-01
-6.77461326e-01 -3.20988297e-01 7.51184583e-01 -8.01759362e-02
-8.33935678e-01 -3.00007164e-01 -6.41994834e-01 2.04618126e-01
5.84388971e-01 1.13579072e-02 -1.69088796e-01 -5.56507468e-01
-1.19166291e+00 6.71265349e-02 -2.97555089e-01 8.79596651e-01
5.29135883e-01 -1.28480041e+00 -5.20384490e-01 3.47860008e-01
-2.09268741e-02 -4.28392023e-01 -2.05156058e-02 5.30598640e-01
1.50942668e-01 7.32214510e-01 -4.18976471e-02 -3.59966546e-01
-1.26866114e+00 5.95456123e-01 1.90375924e-01 -1.50494084e-01
-5.78372121e-01 9.86292899e-01 3.83105218e-01 -4.29138273e-01
5.67009389e-01 -4.06686246e-01 -1.95040703e-01 3.24101329e-01
4.98308331e-01 9.01862532e-02 -2.96446949e-01 -7.85643995e-01
-3.31113547e-01 2.72991899e-02 -7.54654258e-02 -6.66565895e-01
1.55643129e+00 -2.31285870e-01 7.04213977e-02 1.11430597e+00
7.75034249e-01 4.06738192e-01 -1.58317769e+00 -2.74131924e-01
6.27480507e-01 4.20030415e-01 1.26015261e-01 -9.44859922e-01
-4.46238995e-01 1.10966611e+00 -1.29608184e-01 3.81882429e-01
1.01697290e+00 3.39898050e-01 8.74898255e-01 4.49253172e-02
8.89581740e-02 -1.32952690e+00 7.05375969e-02 8.58320713e-01
1.00677228e+00 -9.38651800e-01 -4.46561664e-01 -7.72035837e-01
-1.12463021e+00 9.07123387e-01 4.74410087e-01 1.24253683e-01
9.56525505e-01 5.53037643e-01 2.20881879e-01 1.56477839e-01
-1.23956954e+00 -2.64008224e-01 4.26125675e-01 4.85493451e-01
8.89189065e-01 1.94941789e-01 -2.80143619e-01 1.12654841e+00
-7.69492328e-01 -4.89844829e-01 5.84844463e-02 3.67360055e-01
-6.52093709e-01 -1.39522874e+00 -1.23897165e-01 -2.14828894e-01
-5.35078704e-01 -4.15225863e-01 -7.83770800e-01 5.87384164e-01
-1.41699210e-01 1.10909522e+00 1.24224313e-01 -2.17626289e-01
1.98272720e-01 1.04207015e+00 1.51524946e-01 -1.17849958e+00
-7.00742006e-01 8.05640399e-01 6.69202626e-01 -4.85801816e-01
-6.93406761e-01 -7.64914334e-01 -1.18019187e+00 5.70961058e-01
-1.18427321e-01 2.15879515e-01 4.29796994e-01 9.22878981e-01
-1.17710002e-01 3.90740007e-01 5.37123740e-01 -7.47938752e-01
-7.26847708e-01 -1.15410089e+00 -5.09915471e-01 1.47576228e-01
4.13825244e-01 -3.73115391e-01 -3.25831324e-01 2.31840104e-01] | [14.4093656539917, 6.923627853393555] |
0ba31d2c-06ff-4cc8-b3a2-fa794a37107b | towards-federated-multivariate-statistical | 2211.01645 | null | https://arxiv.org/abs/2211.01645v2 | https://arxiv.org/pdf/2211.01645v2.pdf | Towards federated multivariate statistical process control (FedMSPC) | The ongoing transition from a linear (produce-use-dispose) to a circular economy poses significant challenges to current state-of-the-art information and communication technologies. In particular, the derivation of integrated, high-level views on material, process, and product streams from (real-time) data produced along value chains is challenging for several reasons. Most importantly, sufficiently rich data is often available yet not shared across company borders because of privacy concerns which make it impossible to build integrated process models that capture the interrelations between input materials, process parameters, and key performance indicators along value chains. In the current contribution, we propose a privacy-preserving, federated multivariate statistical process control (FedMSPC) framework based on Federated Principal Component Analysis (PCA) and Secure Multiparty Computation to foster the incentive for closer collaboration of stakeholders along value chains. We tested our approach on two industrial benchmark data sets - SECOM and ST-AWFD. Our empirical results demonstrate the superior fault detection capability of the proposed approach compared to standard, single-party (multiway) PCA. Furthermore, we showcase the possibility of our framework to provide privacy-preserving fault diagnosis to each data holder in the value chain to underpin the benefits of secure data sharing and federated process modeling. | ['Ramin Nikzad-Langerodi', 'David Gabauer', 'Du Nguyen Duy'] | 2022-11-03 | null | null | null | null | ['fault-detection'] | ['miscellaneous'] | [ 3.28462809e-01 -1.21262796e-01 -4.37195078e-02 -1.44651935e-01
-5.71305633e-01 -1.10404205e+00 8.06489527e-01 5.56246996e-01
2.46112525e-01 1.64254099e-01 8.19752067e-02 -3.38951588e-01
-4.98478264e-01 -1.03476727e+00 -4.77963269e-01 -7.62849331e-01
3.73683125e-02 4.01700169e-01 -4.83425915e-01 2.81726182e-01
1.85537770e-01 6.26020908e-01 -1.09261632e+00 5.25774360e-01
3.32853138e-01 1.45236444e+00 -4.11833972e-01 3.46858174e-01
-3.88707072e-02 6.95991337e-01 -6.56477287e-02 -1.06100535e+00
8.45840514e-01 5.94474114e-02 -5.93226850e-01 2.34714597e-01
-3.70932609e-01 -2.19717190e-01 9.23942402e-03 1.08708251e+00
-9.03713629e-02 -2.74067670e-01 3.90761793e-01 -1.87635756e+00
-5.41525543e-01 4.92330045e-01 -4.68995392e-01 -7.16022491e-01
3.31091881e-01 5.02949357e-01 1.06877065e+00 -5.06747246e-01
8.76239896e-01 8.30638409e-01 4.51327860e-01 -8.27338621e-02
-1.50985396e+00 -4.85671759e-01 3.51701640e-02 -7.57938623e-02
-7.82424927e-01 -1.20147496e-01 8.61199200e-01 -5.61454237e-01
5.53332865e-01 6.32167578e-01 7.22663999e-01 1.10528755e+00
6.15399599e-01 5.32623410e-01 1.50088942e+00 -1.07174195e-01
5.08612216e-01 2.08822057e-01 1.15851618e-01 7.40241492e-03
8.39525461e-01 5.38617745e-02 -6.15004003e-01 -8.79406810e-01
4.74121273e-01 7.49573648e-01 -1.15099646e-01 -8.67236435e-01
-1.27982020e+00 5.01908422e-01 -1.25039041e-01 1.55465737e-01
-7.49110281e-01 5.69162332e-02 5.23581326e-01 6.64223850e-01
2.50010431e-01 4.21962261e-01 -7.28151619e-01 -7.62721151e-02
-4.77645338e-01 2.11125612e-01 1.29621124e+00 1.23549747e+00
5.35637736e-01 -5.19858837e-01 -1.49586782e-01 -1.31553143e-01
5.42499125e-01 4.21807528e-01 -2.42168933e-01 -1.20246458e+00
7.13231266e-01 6.30538642e-01 3.40826035e-01 -1.03979254e+00
1.57580882e-01 -1.51353568e-01 -6.88124120e-01 1.50261700e-01
3.97437096e-01 -1.01949885e-01 -1.06891274e-01 9.30092514e-01
1.09103061e-01 -4.46817726e-01 2.75907427e-01 6.02154553e-01
-2.81981915e-01 4.86491650e-01 1.20763391e-01 -2.56142795e-01
1.51388931e+00 -6.32229388e-01 -9.79227006e-01 3.72966677e-01
1.09800816e-01 -5.98738074e-01 3.19771141e-01 8.00352991e-01
-8.76825035e-01 9.09428596e-02 -6.86011553e-01 2.85502404e-01
-3.71974289e-01 -3.39338720e-01 9.03195083e-01 8.15996110e-01
-4.56796348e-01 8.93028200e-01 -8.61966848e-01 -6.93457574e-02
7.33298004e-01 3.06973726e-01 -8.86472940e-01 -3.82891089e-01
-8.03423047e-01 4.35762554e-01 5.49383881e-03 3.83812003e-02
-1.00366044e+00 -9.14841890e-01 -3.80737185e-01 3.89580131e-01
7.03219771e-01 -8.10378551e-01 1.12855530e+00 -4.70371217e-01
-1.32628024e+00 2.58577675e-01 4.06439811e-01 -5.32200813e-01
8.46267819e-01 -1.33656099e-01 -5.03472090e-01 1.84778348e-01
-3.93700004e-01 -2.58911312e-01 8.48566830e-01 -1.45958006e+00
-8.36759329e-01 -8.83993983e-01 -3.78295869e-01 -2.48037294e-01
-9.29367021e-02 2.00360298e-01 -5.13221894e-04 -1.99806184e-01
2.91782349e-01 -8.61376584e-01 -3.56604874e-01 2.47638106e-01
-4.50295478e-01 4.24361169e-01 9.77706552e-01 -9.27204370e-01
7.50923812e-01 -2.11186194e+00 -2.15010464e-01 6.40657961e-01
1.35941759e-01 -1.94953561e-01 4.09919053e-01 1.06996918e+00
-9.96935926e-03 3.53120118e-01 -2.61291564e-01 -2.56798267e-01
4.23464775e-01 2.30536267e-01 -3.27667296e-01 7.51158178e-01
1.43303603e-01 7.95169294e-01 -7.03387678e-01 4.45913263e-02
2.68873066e-01 3.09392661e-01 -8.83056149e-02 1.14292828e-02
-1.03117183e-01 5.04458666e-01 -5.30536473e-01 1.13815725e+00
9.15996671e-01 7.59384111e-02 5.75853944e-01 -1.77761093e-01
-1.48156852e-01 -2.94375867e-01 -1.42467487e+00 1.56828642e+00
-3.82401347e-01 -1.21157773e-01 7.06800461e-01 -2.82956630e-01
8.65110099e-01 7.48819411e-01 9.90453601e-01 -7.34931827e-02
3.58492225e-01 2.23878454e-02 -4.28050727e-01 -2.30242684e-01
5.45953333e-01 -1.12039313e-01 -3.32799703e-01 7.42915273e-01
-5.71403503e-02 -9.57872421e-02 -2.58620858e-01 -2.50100553e-01
1.46305335e+00 9.31194723e-02 2.48506412e-01 -9.95870531e-02
3.99034709e-01 -2.80111909e-01 7.32594669e-01 9.52412188e-02
-3.49534452e-01 3.25899899e-01 8.78952384e-01 -1.71714544e-01
-9.43723321e-01 -1.00877702e+00 1.76378608e-01 2.80866534e-01
-1.24310218e-01 -4.59702969e-01 -3.80975425e-01 -8.31774473e-01
6.12469673e-01 6.71802700e-01 -1.92825869e-01 1.78656384e-01
9.15605724e-02 -4.73336279e-01 3.84344399e-01 3.24922562e-01
2.34152034e-01 -4.45600241e-01 -4.03810889e-01 3.45462829e-01
4.06634569e-01 -1.05994880e+00 -5.19405454e-02 1.22893430e-01
-7.21908391e-01 -1.26903033e+00 -7.39286691e-02 -1.76264346e-02
6.26764357e-01 1.34493724e-01 4.74232793e-01 -7.50416815e-01
-4.27740924e-02 4.95410442e-01 -5.89875244e-02 -5.41392148e-01
-5.82033277e-01 -3.91963422e-01 -1.25360310e-01 5.96315682e-01
1.23226091e-01 -5.62427104e-01 -4.14602995e-01 1.90498799e-01
-8.61163259e-01 -3.15488666e-01 5.79312503e-01 5.79320729e-01
6.56518638e-01 7.25542128e-01 2.89221406e-01 -1.15851724e+00
7.45958805e-01 -7.31199920e-01 -8.42539787e-01 4.56181675e-01
-1.39502180e+00 -3.43444310e-02 4.64458793e-01 -1.10954799e-01
-1.39766169e+00 3.07440758e-01 5.32325387e-01 -4.88019526e-01
-4.53340501e-01 4.37653184e-01 -8.55162621e-01 1.01101324e-01
-1.42072394e-01 -1.60267223e-02 3.22739899e-01 -6.04415119e-01
6.49214804e-01 6.49775684e-01 5.74927390e-01 -5.94930768e-01
7.84566700e-01 9.17196810e-01 4.90549237e-01 -5.94133101e-02
3.95660639e-01 -5.41179538e-01 -2.49307141e-01 9.25731733e-02
6.70366645e-01 -9.01287377e-01 -1.26847303e+00 5.86249471e-01
-1.16096568e+00 4.85111684e-01 -7.04975903e-01 4.22147185e-01
-5.38833320e-01 3.95922035e-01 -8.01855385e-01 -1.00996673e+00
-6.92594886e-01 -9.30159748e-01 6.73390210e-01 -2.88206607e-01
-2.40654618e-01 -6.20143712e-01 -5.24162464e-02 6.63863838e-01
5.88981569e-01 6.11704946e-01 8.79786909e-01 -9.85837579e-01
-1.03173637e+00 -8.07269931e-01 5.15680872e-02 6.63822234e-01
4.33957905e-01 1.78197056e-01 -9.64867890e-01 -1.90045312e-01
5.99634945e-01 4.64267522e-01 -8.90456215e-02 -3.65373194e-01
9.05331254e-01 -5.36767125e-01 -4.14719805e-02 3.87836367e-01
1.62509108e+00 1.90349549e-01 5.23304164e-01 2.62536287e-01
5.90883851e-01 1.14065135e+00 7.08993971e-01 8.24669302e-01
3.42426777e-01 1.64988995e-01 5.10182679e-01 5.54304779e-01
5.83398640e-01 -3.13624561e-01 2.28853896e-01 5.40581882e-01
-2.21540019e-01 -1.52753843e-02 -6.43312454e-01 4.08373564e-01
-1.96450436e+00 -9.52964127e-01 -6.65359676e-01 2.11388469e+00
3.06602389e-01 -1.78575009e-01 -2.75745660e-01 3.62866551e-01
6.22849941e-01 -1.99254707e-01 -2.56669492e-01 -6.00395858e-01
1.43883646e-01 -4.93953787e-02 1.17450583e+00 -2.99205035e-01
-9.07197833e-01 -5.54015040e-02 4.74823475e+00 2.38658458e-01
-6.57056808e-01 3.92894745e-01 4.35575217e-01 -1.34679140e-03
-7.51604915e-01 4.66357023e-01 -2.51216829e-01 3.86222899e-01
1.21314728e+00 -4.32115853e-01 7.10124135e-01 1.14084411e+00
1.70850947e-01 2.35043645e-01 -1.38273501e+00 9.08909261e-01
-4.77179378e-01 -1.28527439e+00 -3.91315043e-01 7.19438791e-01
9.42908406e-01 -2.13600636e-01 -1.01181334e-02 -5.13646424e-01
5.93247354e-01 -5.90801477e-01 8.55119348e-01 6.53973579e-01
2.67259479e-01 -9.73150909e-01 6.93289220e-01 5.04590571e-02
-1.12701142e+00 -7.02329755e-01 1.36526868e-01 1.98278680e-01
2.79244035e-01 8.20156395e-01 -6.57719433e-01 1.53856146e+00
5.44620216e-01 2.43981332e-01 2.63849739e-02 7.26783693e-01
-1.89644903e-01 4.05847251e-01 -7.37197278e-03 5.39069235e-01
-3.26300561e-01 -6.37461662e-01 3.86160940e-01 7.22167969e-01
6.38340056e-01 -4.48814631e-01 -2.30107799e-01 1.01469862e+00
-1.71309471e-01 7.68293217e-02 -8.20024014e-01 -4.89974678e-01
4.42239225e-01 1.55474031e+00 -4.68229085e-01 2.72739619e-01
-6.48631632e-01 8.02258790e-01 -5.40713549e-01 2.21373647e-01
-2.00269148e-01 1.44857883e-01 9.25462008e-01 2.83938676e-01
3.52908403e-01 -3.19392353e-01 -8.14496279e-01 -9.70193267e-01
2.07535744e-01 -1.01416588e+00 4.15686965e-01 -2.53980488e-01
-1.82650042e+00 -1.42730206e-01 -2.03802541e-01 -1.16929018e+00
-1.73059732e-01 -3.78403902e-01 -3.12455684e-01 9.98444796e-01
-1.28298283e+00 -1.58078253e+00 1.65723398e-01 6.54189587e-01
-3.08305979e-01 -1.57551050e-01 9.82745230e-01 5.00958487e-02
-2.95339078e-01 -1.52357116e-01 4.62211251e-01 -1.99905112e-01
4.91464287e-01 -1.19363177e+00 4.19475257e-01 9.86863196e-01
-3.58029120e-02 7.14774489e-01 3.19980115e-01 -9.86126482e-01
-2.53884912e+00 -1.19322348e+00 6.90676093e-01 -5.09639502e-01
1.02865732e+00 -5.14880657e-01 -7.31786132e-01 6.76649988e-01
2.64125943e-01 2.88904935e-01 9.07641172e-01 -1.51343867e-01
-3.83824557e-01 -6.40797913e-01 -1.89091361e+00 1.27710737e-02
6.40408993e-01 -8.58206928e-01 -2.21546844e-01 3.18387263e-02
6.68971717e-01 2.78883487e-01 -1.72146702e+00 -4.87811211e-03
5.09691358e-01 -7.21718967e-01 5.18636286e-01 -4.04804319e-01
2.34689757e-01 -3.68611187e-01 -4.53304321e-01 -9.30176556e-01
-1.81061178e-01 -1.33857763e+00 -4.93956029e-01 1.77990746e+00
1.52234109e-02 -1.09373844e+00 7.61436760e-01 1.52414560e+00
2.26765603e-01 -3.59223843e-01 -9.57760453e-01 -7.33314097e-01
-2.29648232e-01 -7.42806494e-01 1.52096367e+00 1.14662468e+00
1.53686225e-01 -5.24338961e-01 -4.19305675e-02 6.23919845e-01
9.97207761e-01 4.41985399e-01 7.85513222e-01 -1.39686167e+00
-3.81082684e-01 5.00345677e-02 -4.10204142e-01 2.41809696e-01
-2.47322038e-01 -7.25659430e-01 -3.85053724e-01 -1.38978016e+00
5.72581254e-02 -5.23380458e-01 -5.35642028e-01 3.51224869e-01
5.24999559e-01 -5.78859985e-01 6.55151248e-01 2.48833701e-01
-1.10113293e-01 3.52260917e-01 1.20495033e+00 -3.21837105e-02
1.57584235e-01 1.65622354e-01 -1.08645272e+00 2.66309559e-01
5.95238268e-01 -6.10462546e-01 -3.79521579e-01 6.09131455e-02
1.74381182e-01 4.34957415e-01 5.03998518e-01 -6.59576595e-01
3.75265688e-01 -4.03065890e-01 -9.86272320e-02 -3.91394228e-01
7.63684809e-02 -1.82521546e+00 1.36077380e+00 5.99123001e-01
-7.79041415e-03 1.23970360e-01 -4.31308419e-01 9.73935485e-01
-1.29625738e-01 -7.67202303e-02 -1.67628169e-01 -1.97536200e-01
-1.83823064e-01 3.64644080e-01 -1.85792789e-01 -8.65075529e-01
1.42169833e+00 7.45665580e-02 -6.38823509e-01 -8.40433091e-02
-4.43696320e-01 9.10354406e-02 9.61539805e-01 2.32595667e-01
1.54223233e-01 -1.01867044e+00 -4.62101310e-01 5.36411762e-01
1.76950619e-01 1.34373158e-01 1.92456126e-01 7.46712625e-01
-2.37194985e-01 4.19926882e-01 -1.93159863e-01 -1.66198120e-01
-1.04122472e+00 1.08619308e+00 -3.87235552e-01 -5.08190155e-01
-3.67249787e-01 6.32543042e-02 -4.05881405e-01 -4.08924013e-01
-1.25436768e-01 -6.99662745e-01 5.63805759e-01 1.36305496e-01
2.95189977e-01 9.55448925e-01 3.89807671e-01 -2.57569224e-01
-1.27750948e-01 -1.52247727e-01 2.72134274e-01 -2.89558470e-01
1.65068281e+00 -3.27581823e-01 -6.14420295e-01 2.46710315e-01
9.30887640e-01 1.67055011e-01 -1.33443558e+00 4.13539074e-02
3.16759318e-01 -7.81700075e-01 1.38822440e-02 -7.95099139e-01
-1.21926355e+00 4.47445065e-01 4.40068871e-01 4.77394819e-01
9.80381072e-01 -3.33595365e-01 6.55996382e-01 2.51673877e-01
8.32705855e-01 -1.08773112e+00 -7.22477853e-01 -3.73847246e-01
5.91563463e-01 -7.21168578e-01 1.71873480e-01 -9.77048099e-01
-8.61424029e-01 9.64100897e-01 -2.22028404e-01 3.19577128e-01
8.05865288e-01 5.70582926e-01 -1.61903650e-01 -2.51941532e-01
-8.16454768e-01 4.94624078e-01 -3.38865131e-01 7.65719056e-01
-3.28949273e-01 5.07851541e-01 1.00489654e-01 1.18370116e+00
2.43808925e-01 3.82866323e-01 6.51257157e-01 1.57215464e+00
5.08324504e-01 -1.76340353e+00 -7.01652050e-01 3.95359427e-01
-8.36138844e-01 4.41860139e-01 -2.15307757e-01 5.01040459e-01
2.78077245e-01 1.07904673e+00 -2.28771105e-01 -3.64497632e-01
5.30482590e-01 3.51767123e-01 1.64219379e-01 -2.21223593e-01
-1.03086340e+00 8.26809555e-02 2.63768882e-01 -7.43159056e-01
-1.99228093e-01 -1.44359553e+00 -1.01583147e+00 -5.70972502e-01
-2.19221205e-01 -1.81289285e-01 1.42497039e+00 4.71680701e-01
6.94353819e-01 4.08170015e-01 9.35108185e-01 -2.58808434e-01
-1.12999821e+00 -2.72923738e-01 -1.13030064e+00 6.85573041e-01
-1.74749196e-01 -1.08909592e-01 -1.26725122e-01 2.57689595e-01] | [5.986578941345215, 6.604734420776367] |
9312b165-3ff2-4cf5-9972-803161a2d0c1 | hyperparameters-in-reinforcement-learning-and | 2306.01324 | null | https://arxiv.org/abs/2306.01324v1 | https://arxiv.org/pdf/2306.01324v1.pdf | Hyperparameters in Reinforcement Learning and How To Tune Them | In order to improve reproducibility, deep reinforcement learning (RL) has been adopting better scientific practices such as standardized evaluation metrics and reporting. However, the process of hyperparameter optimization still varies widely across papers, which makes it challenging to compare RL algorithms fairly. In this paper, we show that hyperparameter choices in RL can significantly affect the agent's final performance and sample efficiency, and that the hyperparameter landscape can strongly depend on the tuning seed which may lead to overfitting. We therefore propose adopting established best practices from AutoML, such as the separation of tuning and testing seeds, as well as principled hyperparameter optimization (HPO) across a broad search space. We support this by comparing multiple state-of-the-art HPO tools on a range of RL algorithms and environments to their hand-tuned counterparts, demonstrating that HPO approaches often have higher performance and lower compute overhead. As a result of our findings, we recommend a set of best practices for the RL community, which should result in stronger empirical results with fewer computational costs, better reproducibility, and thus faster progress. In order to encourage the adoption of these practices, we provide plug-and-play implementations of the tuning algorithms used in this paper at https://github.com/facebookresearch/how-to-autorl. | ['Roberta Raileanu', 'Marius Lindauer', 'Theresa Eimer'] | 2023-06-02 | null | null | null | null | ['automl', 'hyperparameter-optimization'] | ['methodology', 'methodology'] | [-2.98390657e-01 -3.17167133e-01 -4.04726982e-01 -1.91327766e-01
-7.70616174e-01 -8.70170474e-01 2.66357213e-01 4.89781424e-02
-6.50485218e-01 9.72092748e-01 5.40797450e-02 -4.23447162e-01
-4.21437800e-01 -6.43753231e-01 -7.11592317e-01 -7.95632482e-01
-1.03440031e-01 3.99475455e-01 -2.37748623e-02 -1.42748132e-01
3.94913167e-01 4.02186185e-01 -1.63487625e+00 -3.10267508e-01
6.44272804e-01 4.92557704e-01 -7.03876615e-02 5.21393657e-01
3.04203659e-01 5.02620995e-01 -8.10664475e-01 -3.21224004e-01
3.67911875e-01 -5.03282726e-01 -6.04371011e-01 -3.58407944e-01
2.95178266e-04 -4.55709279e-01 2.60263197e-02 7.49817193e-01
1.02855361e+00 1.82702199e-01 2.70304650e-01 -1.40768421e+00
-4.08499032e-01 9.08150911e-01 -4.78356451e-01 1.26447991e-01
1.26814216e-01 8.28845441e-01 1.09364057e+00 -1.41841009e-01
4.36449260e-01 1.04455113e+00 7.14365184e-01 4.29740250e-01
-1.39181113e+00 -1.13449705e+00 7.31489658e-02 3.98667753e-02
-1.30804324e+00 -5.87211251e-01 5.78350484e-01 -1.94196686e-01
9.48910952e-01 1.04342625e-01 9.60180640e-01 1.37677097e+00
9.32397321e-02 6.99078381e-01 1.19209290e+00 -4.05166984e-01
6.57641947e-01 1.03849351e-01 -1.53256178e-01 5.37841082e-01
6.53566420e-01 4.26980019e-01 -3.41275990e-01 -2.59097427e-01
9.01525259e-01 -5.64060271e-01 -1.49534553e-01 -5.97553909e-01
-1.05384791e+00 1.11074197e+00 1.49200141e-01 1.16201311e-01
-2.39302412e-01 5.59905231e-01 3.94589037e-01 2.43626907e-01
1.52681628e-02 1.01017821e+00 -5.13091147e-01 -6.39586806e-01
-6.60470009e-01 6.73014939e-01 8.75699818e-01 6.00037217e-01
5.06075382e-01 -1.03389293e-01 -5.53138219e-02 1.03540719e+00
4.83731270e-01 2.54360348e-01 4.63084430e-01 -1.48621535e+00
2.41377532e-01 3.18303764e-01 2.27453649e-01 -7.41582096e-01
-6.59165502e-01 -7.44470954e-01 -2.53396153e-01 6.10029817e-01
4.81701732e-01 -4.82096702e-01 -4.45462674e-01 2.04855180e+00
4.22839969e-01 -1.37260437e-01 5.87632470e-02 9.81290877e-01
3.85619760e-01 2.76440889e-01 1.67814955e-01 -1.00132190e-01
1.16573393e+00 -7.32057035e-01 -4.04833227e-01 -3.33023190e-01
7.79152989e-01 -6.81938469e-01 1.39160430e+00 4.12592173e-01
-1.16536546e+00 -9.39358473e-02 -1.26217067e+00 2.89374411e-01
-1.47908702e-01 -3.28159183e-02 9.13763225e-01 7.88024127e-01
-8.40360582e-01 6.78748190e-01 -1.00009656e+00 -5.81988752e-01
4.93123829e-01 4.76792455e-01 1.47114381e-01 1.92323536e-01
-1.06743085e+00 8.37354541e-01 4.47987258e-01 -6.21617064e-02
-7.57634401e-01 -7.95732856e-01 -4.16187644e-01 -3.41235064e-02
6.33990467e-01 -7.68992543e-01 1.60518706e+00 -7.70938516e-01
-1.90656304e+00 4.06565994e-01 4.13859665e-01 -3.63932401e-01
6.29246712e-01 -1.96418464e-01 -5.30798733e-02 -1.83152303e-01
-1.21926598e-01 7.43992269e-01 4.07576799e-01 -1.14573133e+00
-5.46406925e-01 -2.03341216e-01 2.63680458e-01 4.31165814e-01
-5.09347171e-02 -1.16768286e-01 -4.60905939e-01 -3.38138133e-01
-3.67031723e-01 -1.07653952e+00 -1.96906567e-01 -3.97227079e-01
-1.30513653e-01 -8.95001590e-02 2.34973401e-01 -1.79284699e-02
1.34457898e+00 -2.03409004e+00 -5.41728362e-02 7.99520090e-02
2.12364625e-02 7.82090649e-02 -2.33534962e-01 5.03856301e-01
2.10292771e-01 3.80726218e-01 6.52737692e-02 -6.50914237e-02
3.54099244e-01 1.53192237e-01 -8.70874617e-03 7.03560948e-01
-1.24294013e-01 7.76245832e-01 -9.90446389e-01 -4.09426510e-01
1.58314362e-01 3.81089091e-01 -9.80835736e-01 1.13548368e-01
-3.97053748e-01 3.51522028e-01 -5.87859690e-01 4.96901840e-01
2.80787826e-01 -3.94150108e-01 2.93510646e-01 8.48399997e-02
-2.47259021e-01 5.22102475e-01 -1.28674459e+00 1.43530476e+00
-4.26930040e-01 5.31327128e-01 1.43866334e-02 -5.41554749e-01
6.95018589e-01 4.01445664e-02 5.43830454e-01 -8.39774728e-01
3.68120968e-01 1.65030435e-01 4.11132544e-01 -2.46476442e-01
2.39066273e-01 3.05293351e-01 2.13393673e-01 8.35387170e-01
-1.59678966e-01 -1.57358274e-01 5.67110181e-01 -1.50409997e-01
1.15238023e+00 4.59049016e-01 3.07863802e-01 -3.13330561e-01
-6.68491721e-02 7.19770640e-02 5.34143925e-01 1.00552917e+00
-2.55417556e-01 1.89932212e-01 7.53276825e-01 -4.68644202e-02
-1.08899510e+00 -7.83604264e-01 -3.77876550e-01 1.31929946e+00
-2.20691711e-01 -4.51902300e-01 -8.12736213e-01 -3.09433699e-01
1.48414031e-01 9.33749616e-01 -7.21033335e-01 -1.65738195e-01
-4.77190375e-01 -1.09812462e+00 7.71889448e-01 2.94225574e-01
4.15894747e-01 -1.26126385e+00 -1.19633913e+00 3.20125788e-01
1.76607668e-01 -6.81702435e-01 -2.10392937e-01 4.30166960e-01
-7.32478321e-01 -9.76186335e-01 -5.10790050e-01 -3.33256841e-01
3.44280154e-01 2.98051983e-02 1.22704649e+00 2.80246377e-01
-1.49722829e-01 4.81071800e-01 -4.59900767e-01 -4.32721019e-01
-5.22912025e-01 4.09696609e-01 -5.97721487e-02 -6.25857949e-01
8.53031203e-02 -6.61643565e-01 -8.19083989e-01 5.03867924e-01
-7.77960360e-01 -1.22548781e-01 7.86102176e-01 6.07025743e-01
4.17704999e-01 -3.88907082e-02 5.63694477e-01 -6.82025015e-01
1.14913809e+00 -5.24276078e-01 -1.07050776e+00 7.88688362e-02
-1.22299433e+00 3.76508623e-01 5.19484103e-01 -6.01078689e-01
-5.90299547e-01 -3.60047847e-01 8.03655162e-02 -2.21405834e-01
1.40847769e-02 3.77936900e-01 1.44494891e-01 -2.12156802e-01
9.32712376e-01 -7.44445175e-02 2.43194014e-01 -1.47129402e-01
2.04095870e-01 3.84413600e-01 -4.55628596e-02 -8.17793369e-01
5.53872287e-01 -5.75707108e-02 -1.84175402e-01 -4.88776386e-01
-7.41118491e-01 1.66665968e-02 1.45119932e-02 -2.66002476e-01
4.17737156e-01 -6.79620028e-01 -1.07256162e+00 1.34154558e-01
-4.73332107e-01 -1.08565354e+00 -2.25198850e-01 6.01976693e-01
-6.98955774e-01 -1.28527367e-02 -3.72278839e-01 -5.83579898e-01
-2.54990250e-01 -1.50877380e+00 7.35359013e-01 6.01733506e-01
-4.54071939e-01 -1.00236213e+00 5.10900497e-01 3.44945610e-01
6.10622108e-01 3.06329839e-02 8.69036555e-01 -5.69034338e-01
-4.14461434e-01 2.86651164e-01 1.86287358e-01 8.31384435e-02
-1.25399888e-01 3.08914691e-01 -7.74810791e-01 -5.49621820e-01
-3.55572641e-01 -6.61506593e-01 2.98660845e-01 6.78970158e-01
1.14067698e+00 -2.26621658e-01 -5.97557612e-02 8.61011386e-01
1.28037882e+00 1.30243644e-01 4.56781417e-01 1.16509664e+00
1.97868541e-01 2.67930806e-01 6.91481531e-01 7.52108276e-01
3.28873754e-01 6.72804832e-01 3.78597885e-01 8.40087906e-02
1.84914365e-01 -1.86039165e-01 3.83522332e-01 3.89889449e-01
1.27061218e-01 -3.06640059e-01 -8.68727863e-01 1.96241632e-01
-1.84278464e+00 -7.65850604e-01 4.27034318e-01 2.40621614e+00
1.11168849e+00 2.80120611e-01 5.95719814e-01 -1.04592882e-01
3.67516071e-01 1.06064439e-01 -8.81235719e-01 -4.89812046e-01
9.55971926e-02 6.99610785e-02 7.50935197e-01 3.90139669e-01
-5.82487524e-01 9.33145225e-01 6.62676191e+00 7.02064931e-01
-1.25441241e+00 -6.87305704e-02 5.31212747e-01 -6.14244342e-01
-1.95190772e-01 7.51750767e-02 -8.03977251e-01 3.79800797e-01
1.03139603e+00 -4.06203538e-01 1.16604304e+00 8.23950946e-01
5.31366825e-01 -1.92595720e-01 -1.07453465e+00 8.19201052e-01
-4.09278870e-01 -1.08271837e+00 -4.35317695e-01 2.08445117e-01
5.12013257e-01 2.28259653e-01 1.37854412e-01 5.50510526e-01
9.06064272e-01 -1.16227245e+00 7.03866005e-01 5.72175011e-02
4.04388815e-01 -8.42061698e-01 5.52488089e-01 3.21430191e-02
-7.17089534e-01 -2.12482616e-01 -3.97093743e-01 1.85264319e-01
-3.95648301e-01 2.99164802e-01 -9.64536965e-01 1.46640524e-01
8.68869424e-01 4.27307010e-01 -5.95374048e-01 1.32477129e+00
-2.72719979e-01 9.27906394e-01 -4.92093593e-01 -5.43478131e-01
2.27657691e-01 -2.40238950e-01 2.69431442e-01 9.45621550e-01
6.86718896e-02 -8.97208154e-02 -1.10891975e-01 9.86320674e-01
3.49447057e-02 2.69528627e-01 -2.87148863e-01 -2.15491191e-01
9.17700350e-01 1.34294748e+00 -7.48427629e-01 2.22307235e-01
-8.84793997e-02 3.01885843e-01 5.15726686e-01 4.46963996e-01
-1.19993961e+00 -1.54868707e-01 8.41076314e-01 -4.41043302e-02
2.33677357e-01 -2.25789681e-01 -3.80311966e-01 -6.51569128e-01
-1.41027689e-01 -1.41581559e+00 4.45906967e-01 -5.50962031e-01
-9.90550041e-01 2.51457959e-01 2.01732025e-01 -1.01101232e+00
-2.41623610e-01 -3.25025052e-01 -3.86688560e-01 4.43261981e-01
-1.25282955e+00 -5.49025714e-01 -1.93400770e-01 2.41349235e-01
2.93254495e-01 -4.14956249e-02 4.90741044e-01 5.99334463e-02
-9.49010253e-01 8.55353057e-01 2.86422670e-01 -2.46108726e-01
9.06412065e-01 -1.00532472e+00 1.16240010e-01 3.90538186e-01
-2.17520483e-02 7.53859997e-01 1.02774978e+00 -4.21783864e-01
-1.71053040e+00 -7.27185309e-01 -1.58505782e-01 -3.54591727e-01
8.98199320e-01 -2.31230915e-01 -5.69556832e-01 6.02656960e-01
6.65658042e-02 -5.07625163e-01 6.92835808e-01 5.35759985e-01
-2.41807662e-02 -1.68574721e-01 -9.20897841e-01 1.13655066e+00
9.70984161e-01 3.36374789e-02 -1.14079513e-01 3.18900377e-01
4.30463314e-01 -4.17534351e-01 -1.00014126e+00 2.88073242e-01
6.50516152e-01 -1.15347004e+00 8.52021396e-01 -2.82941431e-01
1.73391253e-01 -1.88168287e-01 2.20687734e-03 -1.50831449e+00
-4.01628345e-01 -9.08141911e-01 1.41252860e-01 1.24841249e+00
6.49256647e-01 -9.48776066e-01 7.19949663e-01 4.63018030e-01
1.56435996e-01 -1.06831861e+00 -6.81965351e-01 -7.16226757e-01
3.32376987e-01 -5.31470478e-01 7.12781727e-01 6.69284761e-01
-1.37652338e-01 2.58404881e-01 -7.13924915e-02 9.15548801e-02
6.37338638e-01 -4.04552631e-02 9.80531931e-01 -9.68534648e-01
-6.91051602e-01 -8.60094070e-01 9.76962224e-02 -6.97238326e-01
9.76039693e-02 -6.44287586e-01 6.20780094e-03 -1.18609202e+00
1.88883319e-01 -9.05034721e-01 -1.93893284e-01 6.07796967e-01
-1.02421418e-01 -7.17479736e-03 1.81318507e-01 3.12230796e-01
-8.28076303e-01 6.72224939e-01 1.24055195e+00 3.62182796e-01
-6.41086102e-01 -1.69999555e-01 -8.97775412e-01 5.47258139e-01
1.27446806e+00 -7.03397572e-01 -5.61114490e-01 -4.24172044e-01
4.77309316e-01 -1.85223579e-01 1.67957067e-01 -1.15177691e+00
3.29126753e-02 -4.01020139e-01 3.43826175e-01 1.15219973e-01
9.18421447e-02 -5.32435477e-01 1.80642530e-01 4.96339023e-01
-5.62579453e-01 1.64304301e-01 4.89935666e-01 9.54057202e-02
4.42989111e-01 -3.79458010e-01 8.55280161e-01 2.49172561e-02
-2.33280912e-01 3.91208194e-02 -3.36180389e-01 3.84519756e-01
1.00316441e+00 -1.19128697e-01 -5.92899740e-01 -4.84167427e-01
-1.78196907e-01 5.25722384e-01 8.34239244e-01 2.97649771e-01
1.88642852e-02 -8.55277658e-01 -5.62388003e-01 -2.69166119e-02
-5.75833917e-02 -2.50332028e-01 -9.71346498e-02 9.14898932e-01
-5.48841596e-01 3.57392401e-01 -2.94224322e-01 -4.57540721e-01
-1.03066790e+00 2.13854015e-01 5.93572259e-01 -1.70398489e-01
-6.20544910e-01 5.24936974e-01 -3.12157840e-01 -4.96740222e-01
6.25951052e-01 -3.29926193e-01 -1.49986250e-02 -1.45588189e-01
2.54037350e-01 4.24463987e-01 -9.55905393e-02 1.00515038e-01
-3.60374063e-01 4.04355526e-01 -6.29418865e-02 -4.00793850e-01
1.34082365e+00 3.66139300e-02 3.14890325e-01 3.80777776e-01
7.04219401e-01 1.22489899e-01 -1.64865589e+00 2.54029363e-01
-1.49628550e-01 -2.95302629e-01 3.21245223e-01 -1.06636608e+00
-1.01232147e+00 2.53254622e-01 6.86085165e-01 2.90129811e-01
1.06617188e+00 -5.30160964e-02 2.64065504e-01 4.50236261e-01
5.87727547e-01 -1.25160706e+00 -5.51777668e-02 2.79205084e-01
7.51636863e-01 -1.07157326e+00 3.75680298e-01 1.48466334e-01
-7.99551964e-01 1.01292253e+00 5.76405346e-01 -8.80176388e-03
2.28975147e-01 5.08237422e-01 1.88483030e-01 -1.15033224e-01
-1.19238508e+00 -1.94613203e-01 -3.27079058e-01 3.97505522e-01
4.74733323e-01 -1.74586158e-02 -3.97550523e-01 3.39359134e-01
-6.48796976e-01 1.72057241e-01 4.93593365e-01 9.99098897e-01
-2.73728848e-01 -1.29537439e+00 -4.26328510e-01 3.72740835e-01
-5.19113421e-01 9.30107161e-02 -2.33685151e-01 1.19932687e+00
-2.87330925e-01 1.03911340e+00 -3.07443172e-01 -2.22667962e-01
2.67370105e-01 -1.60435021e-01 6.37010753e-01 -2.38405183e-01
-8.36206079e-01 -1.35860769e-02 2.10693181e-01 -8.72799516e-01
-1.11445121e-01 -8.10358882e-01 -1.29955506e+00 -5.76020360e-01
-1.07010022e-01 2.51025081e-01 8.80667329e-01 7.85662353e-01
5.52125096e-01 5.51703274e-01 3.70447904e-01 -6.80325031e-01
-9.31643605e-01 -6.92082107e-01 -3.94748420e-01 7.02206343e-02
3.57707441e-02 -9.69198525e-01 -5.65430820e-01 -6.24536395e-01] | [4.327173709869385, 1.9294966459274292] |
7c191cc0-559c-4963-9a44-ec6d536ee13c | resources-and-evaluations-for-danish-entity | null | null | https://aclanthology.org/2021.crac-1.7 | https://aclanthology.org/2021.crac-1.7.pdf | Resources and Evaluations for Danish Entity Resolution | Automatic coreference resolution is understudied in Danish even though most of the Danish Dependency Treebank (Buch-Kromann, 2003) is annotated with coreference relations. This paper describes a conversion of its partial, yet well-documented, coreference relations into coreference clusters and the training and evaluation of coreference models on this data. To the best of our knowledge, these are the first publicly available, neural coreference models for Danish. We also present a new entity linking annotation on the dataset using WikiData identifiers, a named entity disambiguation (NED) dataset, and a larger automatically created NED dataset enabling wikily supervised NED models. The entity linking annotation is benchmarked using a state-of-the-art neural entity disambiguation model. | ['Anders Søgaard', 'Barbara Plank', 'Ophélie Lacroix', 'Martin Wu', 'Hieu Lam', 'Maria Barrett'] | null | null | null | null | crac-acl-2021-11 | ['entity-disambiguation', 'entity-resolution'] | ['natural-language-processing', 'natural-language-processing'] | [-2.18368337e-01 1.00673449e+00 -4.63405132e-01 -4.13958192e-01
-8.03011775e-01 -7.62613714e-01 6.22042000e-01 4.30825114e-01
-8.52721691e-01 1.09634948e+00 9.35279131e-01 -9.50698853e-02
-3.85158509e-01 -6.02146268e-01 -3.53694260e-01 -2.22038642e-01
-3.21052782e-02 1.64935017e+00 3.69458795e-01 -4.84577507e-01
-3.39634687e-01 3.20601374e-01 -1.34052181e+00 3.20767850e-01
7.44202375e-01 2.57546097e-01 2.63472736e-01 4.08342272e-01
-1.48617029e-01 6.69776320e-01 -5.88328481e-01 -9.87208724e-01
-2.58099407e-01 4.43258882e-02 -1.53114331e+00 -1.03705525e+00
6.55453682e-01 1.22825816e-01 -3.82256716e-01 1.03788137e+00
7.93465316e-01 2.36300856e-01 2.65898317e-01 -8.58638167e-01
-6.27758503e-01 1.66113162e+00 2.08225679e-02 3.48066509e-01
4.82726753e-01 -6.11969709e-01 1.61694515e+00 -3.57795328e-01
1.53845572e+00 1.35274935e+00 8.05115163e-01 8.54298830e-01
-1.14540279e+00 -7.26944864e-01 -2.07984060e-01 2.84266949e-01
-1.08848310e+00 -5.30744314e-01 4.92013365e-01 -1.73799679e-01
1.75898814e+00 5.71361557e-02 2.26843834e-01 1.08710659e+00
-5.42566240e-01 6.71792328e-01 5.70867538e-01 -6.47700429e-01
-3.17287654e-01 -1.07425109e-01 5.10646880e-01 2.76310027e-01
6.43565834e-01 6.19646132e-01 -3.25857490e-01 -1.38042018e-01
4.37367737e-01 -9.64137137e-01 -4.77449387e-01 -5.01171589e-01
-9.42083120e-01 6.13271475e-01 4.74311680e-01 6.92476928e-01
-2.33036712e-01 -4.00799453e-01 7.82183826e-01 1.85275078e-01
1.80312961e-01 7.15644300e-01 -1.13052058e+00 -8.82475972e-02
-5.29547334e-01 2.70234793e-01 1.40394616e+00 1.09657001e+00
3.56608182e-01 -3.01230550e-01 2.13329539e-01 9.96138036e-01
1.66302457e-01 3.05966526e-01 4.20337170e-01 -1.22600198e+00
6.71171248e-01 6.05507672e-01 1.92358568e-01 -6.30022585e-01
-9.91668105e-01 -2.64079776e-02 -3.16747308e-01 -1.60768718e-01
5.59889674e-01 -3.34760070e-01 -3.31678480e-01 2.09387827e+00
2.92610973e-01 1.86334968e-01 8.92189980e-01 7.59081841e-01
1.76260126e+00 2.59105004e-02 6.74222410e-01 -4.06633973e-01
1.64972830e+00 -5.52142978e-01 -9.63606715e-01 -1.71864517e-02
9.03052568e-01 -5.26592255e-01 2.25288436e-01 1.67711508e-02
-1.12624693e+00 -3.11443388e-01 -6.93172157e-01 -3.53816897e-01
-5.66946685e-01 -8.50366503e-02 1.01651514e+00 4.49288458e-01
-7.70175159e-01 6.30364776e-01 -8.63751233e-01 -8.11073184e-01
2.43777800e-02 5.95556498e-01 -9.74483848e-01 3.22560757e-01
-1.88141227e+00 1.48171818e+00 1.36162889e+00 -2.09435880e-01
-3.29541415e-01 -8.47435117e-01 -1.39106143e+00 -5.10111041e-02
3.38018507e-01 -4.73778516e-01 1.57486320e+00 -5.53747654e-01
-1.02246690e+00 1.73231077e+00 1.44337654e-01 -7.00292230e-01
-4.76742797e-02 -5.80338955e-01 -1.15944469e+00 -8.25727731e-02
3.38546246e-01 8.57402563e-01 -4.07028645e-01 -1.28526103e+00
-1.09831119e+00 -3.84260505e-01 7.85415247e-02 1.90472558e-01
1.13759659e-01 5.72439849e-01 -3.69067878e-01 -5.27013004e-01
-2.69985855e-01 -8.35254669e-01 -3.63262035e-02 -9.52667952e-01
-3.25471640e-01 -6.25342607e-01 5.89141965e-01 -7.59510159e-01
1.47945082e+00 -1.90749550e+00 1.47479728e-01 -3.02289546e-01
5.21923937e-02 6.50492430e-01 -6.06170297e-02 5.04248738e-01
-7.51551747e-01 -7.27469325e-02 -5.49920909e-02 -9.35180932e-02
1.27942875e-01 7.88377941e-01 -9.90781263e-02 1.94882348e-01
-1.11248111e-02 8.48485947e-01 -1.27611756e+00 -7.55513251e-01
5.61507307e-02 3.66949499e-01 -4.80929613e-01 4.48215492e-02
-2.47995287e-01 2.03698739e-01 -1.97776914e-01 3.08519512e-01
5.05269945e-01 3.41347963e-01 1.03915465e+00 -6.53548777e-01
-4.51513171e-01 8.59446466e-01 -1.18806469e+00 1.95404172e+00
-1.37991086e-01 4.78817552e-01 1.40675977e-01 -7.65855014e-01
9.66259658e-01 9.28159356e-01 4.39148247e-01 -5.44052243e-01
1.99823946e-01 4.14487690e-01 1.31766245e-01 -4.54198033e-01
1.01187503e+00 -3.35809588e-02 -6.45925820e-01 1.57354295e-01
5.95572591e-01 3.05790514e-01 5.43447852e-01 2.25727692e-01
9.07454967e-01 4.51275945e-01 7.77908325e-01 -3.91834825e-01
5.23001611e-01 5.96526861e-01 1.07615161e+00 2.18689993e-01
-2.23038241e-01 1.53019994e-01 2.42484778e-01 -5.05020857e-01
-7.59233713e-01 -1.11853576e+00 -7.00300395e-01 1.16053391e+00
-1.48187608e-01 -6.69326246e-01 -6.89580321e-01 -8.15449774e-01
-3.41008976e-02 1.00315833e+00 -4.84909952e-01 2.81525552e-01
-1.22004282e+00 -4.06831712e-01 1.16528285e+00 6.45641029e-01
3.74781907e-01 -1.71583688e+00 -5.71325064e-01 3.79199117e-01
-3.45567703e-01 -1.35971642e+00 -1.13934427e-01 5.65379262e-01
-5.22673428e-01 -1.70079625e+00 4.45443653e-02 -1.43894458e+00
6.95654228e-02 -6.76422000e-01 1.86613989e+00 -1.53527185e-01
5.43456413e-02 1.18738301e-01 -2.85397828e-01 -3.46626133e-01
-6.08678818e-01 2.81121820e-01 1.21602029e-01 -1.19364488e+00
1.27427137e+00 -3.24178010e-01 2.30833292e-01 -1.37825891e-01
-3.83174956e-01 -6.26435101e-01 2.06805915e-01 1.01455200e+00
4.43173110e-01 -3.08586657e-01 3.32326710e-01 -1.48948681e+00
3.62214476e-01 -3.61009330e-01 -6.12841189e-01 1.74933895e-01
-3.68663818e-01 1.84811875e-01 1.20196149e-01 -3.77065092e-02
-1.74020243e+00 1.40098929e-01 -4.95462030e-01 -7.39740999e-03
-7.94046104e-01 5.59471250e-01 -6.54511511e-01 5.43905854e-01
6.83483779e-01 -6.71169817e-01 -6.26502633e-01 -8.43061507e-01
9.01145518e-01 7.01757193e-01 1.55204666e+00 -9.91779566e-01
2.68710047e-01 9.60737169e-02 -1.77957669e-01 -5.50608158e-01
-1.05587888e+00 -6.60276353e-01 -1.46511924e+00 4.45053101e-01
1.08820701e+00 -1.00767589e+00 -7.65723169e-01 -6.32192418e-02
-1.52191460e+00 -2.14635372e-01 -4.02515471e-01 5.01258016e-01
-5.33180237e-01 2.97919720e-01 -8.98549736e-01 -3.36238652e-01
-5.64352512e-01 -4.82333332e-01 6.48915231e-01 2.02448085e-01
-1.04450274e+00 -1.29318631e+00 9.05551434e-01 2.47985080e-01
-3.44766885e-01 4.04599875e-01 7.38791168e-01 -1.38340724e+00
2.10358158e-01 7.30089545e-02 -2.21800283e-01 -3.16285640e-01
-7.39557967e-02 -1.57637164e-01 -7.91651487e-01 -7.47132450e-02
-7.43347585e-01 -1.95150495e-01 8.31483901e-01 1.82858959e-01
-3.02029997e-01 -1.77158594e-01 -1.01590145e+00 4.43813831e-01
1.33695114e+00 1.81141570e-01 4.52908039e-01 9.28394139e-01
4.79921341e-01 6.42565012e-01 8.35523665e-01 3.16254571e-02
7.91513085e-01 7.63404369e-01 2.05237970e-01 2.28148922e-01
-4.16185379e-01 -3.18910033e-01 -7.26665780e-02 7.27423608e-01
-2.74426728e-01 -7.25303136e-04 -1.17563415e+00 1.20629334e+00
-2.00592709e+00 -1.21442223e+00 -4.79739130e-01 1.81929576e+00
1.34851098e+00 -4.03139666e-02 -1.59286961e-01 -1.95670813e-01
1.04922593e+00 -1.21421866e-01 3.12664621e-02 -4.82010782e-01
-5.22914350e-01 7.24198937e-01 3.73531044e-01 7.65282035e-01
-1.65730369e+00 1.60133290e+00 6.25062466e+00 1.53134525e-01
-2.28548110e-01 2.24412933e-01 -4.39669788e-01 2.44102776e-01
-3.31959650e-02 2.19425350e-01 -1.25130963e+00 9.45257768e-02
1.16979563e+00 6.69256747e-02 -1.35547936e-01 7.92492390e-01
-3.65002602e-01 2.74157107e-01 -1.17415273e+00 6.45808876e-01
-3.53010409e-02 -1.28971863e+00 -2.12142080e-01 -2.62996137e-01
5.85061848e-01 5.34173906e-01 -8.26936603e-01 5.26813447e-01
1.50015712e+00 -7.35174179e-01 4.45399731e-01 2.06458211e-01
8.49262834e-01 -8.87306094e-01 1.19516397e+00 -1.64106369e-01
-1.21284497e+00 -8.89060423e-02 -3.17100883e-01 2.24075958e-01
5.97983003e-01 1.21931873e-01 -4.04613167e-01 8.11373413e-01
1.02065647e+00 8.23597252e-01 -1.51386023e-01 7.74529576e-01
-7.68923581e-01 2.61419594e-01 -2.31136546e-01 4.64248568e-01
1.66259203e-02 3.19228768e-01 8.31564605e-01 1.58762169e+00
-1.01590538e-02 4.91626769e-01 1.35537058e-01 5.36493301e-01
-2.02623919e-01 6.51041931e-03 -4.16516662e-01 9.03920904e-02
1.20626163e+00 1.27203691e+00 -4.53841574e-02 -2.06599206e-01
-5.21431983e-01 6.17046833e-01 8.87399197e-01 1.27416179e-02
-2.81593531e-01 -4.29225117e-01 9.88547146e-01 -3.48543316e-01
4.35994297e-01 2.22056463e-01 1.77728355e-01 -9.02500689e-01
-6.52854145e-01 -6.25680387e-01 1.47582471e+00 -5.57703435e-01
-1.35296786e+00 5.97006917e-01 2.91499376e-01 -6.22095823e-01
-5.75461626e-01 -7.00600922e-01 -3.75904918e-01 6.84574544e-01
-1.54770076e+00 -1.54879868e+00 1.07321553e-02 4.92679715e-01
-6.94578514e-02 -3.28565568e-01 1.57904553e+00 5.20722032e-01
-5.34898400e-01 7.02812076e-01 -2.17952684e-01 9.57393408e-01
1.22249842e+00 -1.56953108e+00 5.84965467e-01 3.83831531e-01
2.22403005e-01 6.29612565e-01 8.38767350e-01 -9.16525960e-01
-5.86952507e-01 -8.72640789e-01 1.76867223e+00 -8.94392371e-01
1.01472366e+00 3.29296052e-01 -9.78377998e-01 1.35715675e+00
6.66017592e-01 -3.24439943e-01 8.76535714e-01 6.64095402e-01
-5.50740600e-01 3.18067938e-01 -1.14199793e+00 2.25386694e-01
1.36047161e+00 -3.33262891e-01 -1.71601367e+00 -1.36746883e-01
7.45096028e-01 -8.88341963e-01 -1.55230379e+00 6.15324080e-01
2.16204658e-01 -5.75863183e-01 1.05272138e+00 -1.14150012e+00
1.38754398e-01 -7.45244026e-02 -2.83650070e-01 -1.36448419e+00
-3.97792757e-01 -6.18264198e-01 -2.46772572e-01 2.01106787e+00
6.08194590e-01 -3.17844748e-01 7.15890408e-01 4.89052355e-01
-5.21002412e-01 3.20006669e-01 -1.16576040e+00 -5.17930865e-01
5.09504974e-01 -2.16848060e-01 6.37714386e-01 1.72506809e+00
9.67059314e-01 7.39513636e-01 1.84962839e-01 3.91705483e-01
6.45177722e-01 1.74912140e-01 3.91754776e-01 -2.01919174e+00
-1.24526389e-01 -4.38729078e-01 -2.89267570e-01 -3.47572118e-01
9.49428439e-01 -1.08745551e+00 -1.65174268e-02 -1.49799407e+00
8.96047894e-03 -6.04723573e-01 -1.27898380e-01 8.78464639e-01
-1.08799137e-01 -1.43762127e-01 2.13868916e-02 1.61287889e-01
-5.89336753e-01 -1.34214386e-01 5.86279154e-01 7.41422269e-03
-1.87531129e-01 -5.63161135e-01 -6.62550628e-01 8.71299505e-01
7.20621943e-01 -5.81084490e-01 2.14963451e-01 -5.44693828e-01
2.39907101e-01 1.65137649e-01 -6.90782741e-02 -6.31831944e-01
4.35536832e-01 2.23335564e-01 -1.13769665e-01 -6.57423079e-01
4.00143601e-02 -6.82931244e-01 2.24946722e-01 1.64061219e-01
-3.09957206e-01 -4.97312434e-02 5.42623281e-01 8.45743567e-02
-4.06640977e-01 -4.69910264e-01 5.79528511e-01 -4.62511182e-01
-1.32775247e+00 -1.37848616e-01 1.57514453e-01 8.17971647e-01
8.07219028e-01 2.11485237e-01 -8.16460907e-01 2.89719135e-01
-1.27473187e+00 3.88369739e-01 3.31608802e-01 6.29491866e-01
-7.38077238e-02 -1.36033165e+00 -8.76686931e-01 -3.53600621e-01
2.41964772e-01 4.98052239e-02 -2.39332253e-03 3.93766671e-01
-1.03025809e-01 8.05843294e-01 -3.61112744e-01 -5.18798130e-03
-1.57230842e+00 8.52852404e-01 2.71989822e-01 -6.78458571e-01
-6.18751109e-01 7.23707557e-01 -2.99902827e-01 -1.10415530e+00
1.66960761e-01 1.38250262e-01 -1.03287506e+00 4.22394782e-01
3.68086398e-01 3.09135079e-01 9.17695404e-04 -1.36872923e+00
-6.86362505e-01 3.36568236e-01 -9.13542435e-02 1.36859581e-01
1.48937643e+00 -1.53025519e-02 -4.13406521e-01 -2.53708754e-03
7.17165947e-01 1.12636417e-01 -5.21266341e-01 -3.89255494e-01
7.23460138e-01 1.80512831e-01 -2.09962949e-01 -9.85911191e-01
-8.97756279e-01 4.19341505e-01 2.22409472e-01 -1.01838708e-01
6.06406510e-01 6.09410465e-01 7.91314602e-01 6.10245168e-01
3.24663877e-01 -1.12940931e+00 -8.12265337e-01 1.03874874e+00
6.50071383e-01 -1.07069635e+00 -1.21598102e-01 -5.79000711e-01
-7.16336727e-01 8.51696789e-01 8.50451469e-01 -5.09618409e-03
7.30605900e-01 7.99587011e-01 4.40013468e-01 -4.69332397e-01
-6.52607501e-01 -7.34407246e-01 -3.30345184e-02 1.28294110e+00
9.05889213e-01 1.71261132e-01 -6.90022707e-01 1.23444033e+00
-6.76703990e-01 -1.43097237e-01 3.33195150e-01 3.77883554e-01
-2.53502041e-01 -1.52659774e+00 -7.94211552e-02 -9.62132215e-02
-7.26379514e-01 -5.19650757e-01 -5.68897545e-01 1.36394835e+00
3.12551737e-01 8.15337360e-01 4.50610936e-01 2.90031396e-02
8.52413476e-01 1.19680330e-01 3.94537717e-01 -8.25569093e-01
-8.07429254e-01 -2.88215339e-01 1.20960581e+00 -3.75660717e-01
-1.06320560e+00 -9.27189410e-01 -1.81699061e+00 -2.17275605e-01
-3.92843276e-01 6.58333898e-01 -4.30001952e-02 9.76400673e-01
7.82019347e-02 3.35500002e-01 -4.76532280e-01 -6.55454040e-01
5.50199561e-02 -1.01000512e+00 -4.71310318e-01 8.40831161e-01
-2.49402046e-01 -8.16875994e-01 -6.16128631e-02 -6.97005028e-03] | [9.318680763244629, 9.554218292236328] |
b9feea1a-fe70-4e37-a6a2-5a100ea84c72 | dynamic-loss-for-robust-learning | 2211.12506 | null | https://arxiv.org/abs/2211.12506v1 | https://arxiv.org/pdf/2211.12506v1.pdf | Dynamic Loss For Robust Learning | Label noise and class imbalance commonly coexist in real-world data. Previous works for robust learning, however, usually address either one type of the data biases and underperform when facing them both. To mitigate this gap, this work presents a novel meta-learning based dynamic loss that automatically adjusts the objective functions with the training process to robustly learn a classifier from long-tailed noisy data. Concretely, our dynamic loss comprises a label corrector and a margin generator, which respectively correct noisy labels and generate additive per-class classification margins by perceiving the underlying data distribution as well as the learning state of the classifier. Equipped with a new hierarchical sampling strategy that enriches a small amount of unbiased metadata with diverse and hard samples, the two components in the dynamic loss are optimized jointly through meta-learning and cultivate the classifier to well adapt to clean and balanced test data. Extensive experiments show our method achieves state-of-the-art accuracy on multiple real-world and synthetic datasets with various types of data biases, including CIFAR-10/100, Animal-10N, ImageNet-LT, and Webvision. Code will soon be publicly available. | ['Tingfa Xu', 'Ying Wang', 'Jizhou Zhang', 'Jianan Li', 'Shenwang Jiang'] | 2022-11-22 | null | null | null | null | ['learning-with-noisy-labels', 'learning-with-noisy-labels'] | ['computer-vision', 'natural-language-processing'] | [ 2.51022369e-01 -1.76432073e-01 -2.17768908e-01 -8.38651836e-01
-1.19688845e+00 -4.11386907e-01 3.92012328e-01 1.50026813e-01
-6.06222808e-01 9.10601377e-01 -1.67939618e-01 -4.25252430e-02
-4.06149775e-02 -5.59962034e-01 -8.22756350e-01 -1.01631892e+00
1.13399543e-01 4.04157251e-01 -1.27416700e-02 5.29111214e-02
1.62249967e-01 -5.22489510e-02 -1.65324950e+00 3.56093466e-01
1.05997109e+00 1.43683815e+00 -7.89102614e-02 5.46503365e-01
5.69199957e-02 8.71084571e-01 -6.21043563e-01 -5.11127412e-01
1.92214444e-01 -1.46067798e-01 -4.91134167e-01 6.89693093e-02
6.08335853e-01 7.90910497e-02 7.17252865e-03 1.42281997e+00
8.47838700e-01 -1.40676856e-01 6.28821433e-01 -1.41868567e+00
-4.35288996e-01 7.88215220e-01 -6.04636192e-01 1.50298849e-01
-2.59870887e-01 4.85742152e-01 8.86318922e-01 -7.71330237e-01
2.75409967e-01 1.30817008e+00 8.99632454e-01 6.30866945e-01
-1.28317475e+00 -9.94390726e-01 5.01880169e-01 1.89056650e-01
-1.12377250e+00 -4.25824821e-01 7.44239151e-01 -3.84734064e-01
7.21208379e-02 1.58368647e-01 3.05369467e-01 1.58428097e+00
2.74638474e-01 7.60297418e-01 1.40717614e+00 -1.70502886e-01
5.27847171e-01 3.55114192e-01 4.16802138e-01 3.63796502e-01
2.91499645e-01 2.11980179e-01 -4.76903915e-01 -1.78540602e-01
2.70216782e-02 -3.01203653e-02 -2.71744668e-01 -4.07040805e-01
-1.09175682e+00 6.13883674e-01 3.37230235e-01 -1.99073359e-01
-1.70852765e-01 1.79925859e-01 5.96687138e-01 3.53198022e-01
7.76121855e-01 2.55906612e-01 -7.19630718e-01 3.89421359e-02
-6.95236981e-01 4.36558425e-01 6.43081129e-01 7.82342017e-01
7.06884444e-01 -6.57725111e-02 -6.65874958e-01 1.05992091e+00
3.25698346e-01 7.21959412e-01 6.37298048e-01 -7.07648039e-01
6.00153387e-01 5.35218298e-01 9.45969969e-02 -8.68747890e-01
-3.13216269e-01 -1.12132287e+00 -1.03225327e+00 3.68220627e-01
5.05393624e-01 -2.16935143e-01 -1.12177515e+00 2.10696793e+00
5.91502547e-01 1.78737134e-01 -1.42607093e-01 8.07272315e-01
5.95459878e-01 3.84041607e-01 2.88490623e-01 -2.48015821e-02
1.20631158e+00 -9.89340246e-01 -8.18247497e-01 -3.73643190e-01
4.87964392e-01 -4.40099955e-01 1.38099670e+00 6.37794375e-01
-8.83882403e-01 -5.29636502e-01 -1.20925486e+00 2.54319608e-01
-3.35639507e-01 1.91310048e-01 2.20379949e-01 6.16248190e-01
-4.12322193e-01 5.31581640e-01 -5.27148187e-01 2.42641792e-01
8.18941474e-01 7.30642900e-02 -1.18288681e-01 -1.74272552e-01
-1.12907231e+00 6.60628736e-01 2.88374841e-01 1.73852816e-01
-1.23281777e+00 -8.81860435e-01 -5.32728434e-01 -1.42456964e-01
4.33924049e-01 -5.94902158e-01 1.40832365e+00 -1.29440391e+00
-1.21932352e+00 9.87852216e-01 4.25164819e-01 -4.15546685e-01
1.07504559e+00 -2.06081495e-01 -3.68527353e-01 -4.03193921e-01
6.76423684e-02 4.13658202e-01 1.06735325e+00 -1.49298978e+00
-8.69207442e-01 -6.28818810e-01 -3.23151410e-01 4.94775921e-02
-3.85666758e-01 -2.97202289e-01 -4.99475561e-02 -9.01779413e-01
5.88206202e-02 -6.86895192e-01 -1.79610595e-01 1.01617232e-01
-4.79030877e-01 -1.47507399e-01 7.02480137e-01 -2.17899427e-01
1.17723870e+00 -2.01785374e+00 -3.02782003e-02 1.13602720e-01
2.01752067e-01 4.21800166e-01 -2.69911051e-01 -1.30799457e-01
-2.11307496e-01 1.14405409e-01 -3.85453045e-01 -5.56727886e-01
-2.25620680e-02 1.17478572e-01 -4.15534526e-01 6.50713921e-01
1.14064375e-02 6.53961420e-01 -1.07277119e+00 -2.77758509e-01
-1.09509520e-01 3.78000587e-01 -3.34789544e-01 4.00555372e-01
-2.85408527e-01 3.60252112e-01 -2.01658934e-01 6.84658229e-01
9.76323426e-01 -1.61218971e-01 -3.38767804e-02 -1.95164502e-01
2.34397158e-01 -2.66136564e-02 -1.49613154e+00 1.38721406e+00
-4.38769877e-01 1.53603122e-01 2.90725857e-01 -9.79309201e-01
8.89155746e-01 8.56328662e-03 1.56082138e-02 -8.01269770e-01
2.84040362e-01 3.15049142e-01 -3.32581699e-01 -7.09361970e-01
1.74045146e-01 -2.08084404e-01 -1.08090699e-01 2.54561275e-01
1.87620744e-01 -3.28035913e-02 1.40846763e-02 -1.43057242e-01
8.92960846e-01 3.79256606e-02 -1.72720607e-02 -2.21018344e-01
4.86103952e-01 -3.31534922e-01 1.01574433e+00 1.05444837e+00
-5.32681644e-01 6.36992097e-01 5.25392175e-01 -5.27394772e-01
-9.13520157e-01 -1.02791333e+00 -3.29921633e-01 1.33158326e+00
7.52281025e-02 7.06463829e-02 -8.38402987e-01 -1.07775164e+00
1.57716691e-01 6.42695725e-01 -8.32995653e-01 -4.91318762e-01
-2.45074391e-01 -1.40185535e+00 4.90473032e-01 3.55191380e-01
4.71772343e-01 -8.68582904e-01 -2.38862991e-01 4.63495925e-02
-1.57355428e-01 -7.16506898e-01 -3.84040833e-01 6.69691980e-01
-7.47764647e-01 -1.20750165e+00 -3.16470385e-01 -6.31830275e-01
7.12842226e-01 -1.99392792e-02 1.30551434e+00 -3.76942605e-02
-2.88640678e-01 -1.94696426e-01 -3.31382036e-01 -5.72359741e-01
-3.54361981e-01 1.94698915e-01 2.71997862e-02 3.63924444e-01
3.70815210e-02 -5.59203982e-01 -5.88199794e-01 4.16040957e-01
-1.00231135e+00 -2.09465861e-01 4.85947162e-01 1.27149677e+00
7.30388522e-01 9.03639868e-02 9.61577475e-01 -1.25699973e+00
3.31043631e-01 -8.25053334e-01 -7.03874230e-01 2.54255414e-01
-9.01372671e-01 1.13019124e-01 6.91846550e-01 -7.72920310e-01
-1.05347681e+00 -1.03474520e-01 -1.92805722e-01 -2.32836723e-01
-8.76057968e-02 2.71286100e-01 -5.67188323e-01 1.72170252e-01
8.76333058e-01 -5.11832070e-03 -2.08450958e-01 -5.90400875e-01
3.56430769e-01 9.61093187e-01 7.35931396e-01 -8.02878082e-01
8.59967232e-01 5.04531443e-01 -2.05229223e-01 -8.52357745e-02
-1.50441933e+00 -2.30065092e-01 -3.75686526e-01 -1.46738291e-01
3.34057868e-01 -1.11879468e+00 -6.34516895e-01 1.16189730e+00
-7.02106297e-01 -5.04411280e-01 -4.67192054e-01 2.22541511e-01
-4.23835099e-01 -1.01142250e-01 -4.38437909e-01 -7.73473859e-01
-4.58027363e-01 -1.13822639e+00 1.05320680e+00 2.02758610e-01
2.17767552e-01 -7.96967804e-01 1.81287929e-01 3.63960683e-01
5.29795229e-01 4.03979659e-01 8.86418998e-01 -7.14375556e-01
-3.50581646e-01 -3.86200666e-01 -2.32131436e-01 8.95371437e-01
-4.56849225e-02 -5.96170016e-02 -1.39392984e+00 -5.17545640e-01
1.40445903e-01 -8.23905528e-01 1.10833609e+00 2.09086388e-01
1.70449793e+00 -5.47004700e-01 -1.17572956e-01 9.30355608e-01
1.41821432e+00 -1.25239313e-01 3.64139318e-01 4.60159093e-01
6.08471990e-01 5.38303316e-01 7.30758607e-01 4.65886652e-01
4.96668607e-01 3.87843966e-01 8.18438649e-01 4.90643121e-02
-1.26511410e-01 -2.48386636e-01 1.64971083e-01 6.37033045e-01
6.29205287e-01 -2.83679754e-01 -7.89187849e-01 4.43385094e-01
-1.83023190e+00 -7.57196188e-01 1.03468992e-01 2.28220272e+00
1.38092279e+00 3.51573169e-01 -4.50402834e-02 3.68073791e-01
7.15979099e-01 2.25062072e-01 -9.66782689e-01 2.45827124e-01
-3.53040159e-01 -2.09856525e-01 5.90703487e-01 3.58236969e-01
-1.47897100e+00 5.80358982e-01 5.77643538e+00 1.07375920e+00
-1.37746286e+00 2.80447423e-01 1.30125737e+00 -3.11386049e-01
-3.49944621e-01 -3.79667282e-01 -8.54306757e-01 8.29528332e-01
8.96965802e-01 1.66773528e-01 2.80278683e-01 9.99150932e-01
-5.12193830e-04 5.76177388e-02 -1.07976246e+00 1.02352166e+00
-6.07025204e-03 -9.97455776e-01 -1.78509116e-01 -2.99160600e-01
9.39459205e-01 1.98361799e-01 4.74998236e-01 4.33094531e-01
4.47237641e-01 -1.04422581e+00 1.25930536e+00 8.13651502e-01
6.17834926e-01 -7.19607115e-01 8.99134934e-01 6.42974019e-01
-6.26750231e-01 -5.33222258e-01 -3.14789712e-01 1.85686052e-01
-2.75475860e-01 1.27844977e+00 -3.53007674e-01 1.95634946e-01
7.95198500e-01 6.06856525e-01 -8.56324255e-01 1.27431893e+00
-2.49733508e-01 8.29819262e-01 -1.40207455e-01 1.45618305e-01
-8.00585151e-02 1.30048439e-01 5.07589996e-01 1.09892917e+00
-1.09470440e-02 -4.17893529e-01 2.80977219e-01 7.47353137e-01
-4.29812998e-01 -1.12494484e-01 -1.33488089e-01 3.52922380e-01
6.71769559e-01 1.44753683e+00 -4.29160684e-01 -2.72645086e-01
1.21279873e-01 5.01141906e-01 5.62273562e-01 3.59731168e-01
-8.64327490e-01 -3.83964837e-01 7.17774212e-01 -1.87106710e-02
2.18870472e-02 2.92877614e-01 -7.63665617e-01 -1.29093874e+00
2.09009811e-01 -1.19547772e+00 4.33822393e-01 -4.39156801e-01
-1.77905691e+00 6.21153951e-01 -2.72650242e-01 -1.20100343e+00
1.64651930e-01 -4.76943582e-01 -3.69487226e-01 7.12596297e-01
-1.90548563e+00 -1.16005874e+00 -6.16065681e-01 2.31510818e-01
3.40156436e-01 -1.15307406e-01 5.03946066e-01 5.61401367e-01
-9.53753054e-01 9.69282746e-01 3.77586693e-01 7.16809705e-02
1.18658280e+00 -1.39505136e+00 1.34549737e-01 6.84482336e-01
-2.33990088e-01 2.35225141e-01 7.80713320e-01 -4.83135462e-01
-1.02334106e+00 -1.56610310e+00 4.87780452e-01 -5.10259092e-01
4.55053419e-01 -6.08748734e-01 -9.67794836e-01 4.48783875e-01
-1.04121156e-01 5.71882725e-01 4.04779881e-01 -2.63137296e-02
-7.80262589e-01 -8.35652113e-01 -1.44733822e+00 2.06106886e-01
1.01426101e+00 -2.69368708e-01 -2.98506707e-01 5.21991432e-01
7.12626874e-01 -6.79096758e-01 -6.29107475e-01 6.56083167e-01
4.02337909e-01 -9.44908023e-01 7.69909739e-01 -7.25539148e-01
3.26917648e-01 -3.23113650e-01 -3.39535177e-01 -1.57655334e+00
-1.34651393e-01 -3.47874135e-01 -6.76746517e-02 1.58165157e+00
3.58346641e-01 -6.78909957e-01 6.56527400e-01 1.65536672e-01
-1.15956560e-01 -1.01236558e+00 -9.80332851e-01 -6.20872736e-01
1.15116216e-01 -6.14230812e-01 7.19120622e-01 8.56019676e-01
-6.06142402e-01 1.41431838e-01 -4.21160251e-01 8.94387960e-02
1.14522123e+00 -2.03421444e-01 6.86533391e-01 -1.35011888e+00
-5.20282760e-02 -3.97881389e-01 -7.30361268e-02 -6.23043060e-01
3.22659701e-01 -8.37670147e-01 4.25635397e-01 -8.01354527e-01
2.78048664e-01 -8.69388640e-01 -5.58038116e-01 4.82586145e-01
-6.26050293e-01 2.45531857e-01 -1.27633914e-01 7.20764101e-02
-9.04403150e-01 8.53636384e-01 9.61957872e-01 -2.98782438e-01
1.12835191e-01 3.17929924e-01 -9.84499276e-01 7.94649065e-01
6.37410283e-01 -8.54835331e-01 -3.95114928e-01 -3.98034424e-01
3.75242323e-01 -5.16384602e-01 4.75733936e-01 -1.06662858e+00
1.75730169e-01 -2.51639992e-01 4.90076542e-01 -3.93009782e-01
-7.59470165e-02 -8.19303870e-01 -1.87955797e-01 2.98666447e-01
-8.12878847e-01 -1.00890756e-01 -1.85750380e-01 9.32010949e-01
-1.62045449e-01 -1.50720254e-01 1.40176904e+00 -7.17543811e-02
-3.30807716e-01 4.43143040e-01 1.75586313e-01 6.52764142e-01
9.40219760e-01 1.96734458e-01 -6.40665412e-01 -1.03684962e-01
-4.98325735e-01 6.17906988e-01 3.25922132e-01 5.11499643e-01
2.44018555e-01 -1.38320661e+00 -8.24704945e-01 3.86349946e-01
3.37429285e-01 2.59110183e-01 4.20580745e-01 5.91974318e-01
-5.25001734e-02 -1.75299048e-01 -8.03940818e-02 -7.60365784e-01
-9.06214058e-01 3.90204400e-01 7.22610891e-01 -3.33100528e-01
-1.38623059e-01 1.15102959e+00 -1.22725688e-01 -8.90311897e-01
8.61660480e-01 -3.08554441e-01 -1.65289175e-02 4.16397244e-01
7.87283838e-01 5.10493338e-01 2.65734881e-01 -3.34581524e-01
-2.62079179e-01 3.70831817e-01 -8.10291246e-02 2.13138193e-01
1.38665164e+00 -3.68447065e-01 -1.06567197e-01 7.95263648e-01
1.21121216e+00 -3.39981914e-01 -1.72155273e+00 -4.58813876e-01
4.49445620e-02 -5.02121210e-01 3.76996063e-02 -1.23067391e+00
-1.22598994e+00 8.15966070e-01 1.04047585e+00 -9.85743199e-03
1.21237600e+00 -3.53407949e-01 7.24031150e-01 1.65880203e-01
2.30267540e-01 -1.39587712e+00 2.94857353e-01 5.28060973e-01
6.41526639e-01 -1.58996606e+00 -2.48176634e-01 1.25348577e-02
-5.38960338e-01 7.59243071e-01 7.63591051e-01 7.91922733e-02
7.48805404e-01 3.83794487e-01 4.41690028e-01 1.69019923e-01
-1.08128858e+00 1.15276121e-01 1.06882833e-01 6.45235598e-01
1.06543459e-01 2.30744742e-02 -8.17756876e-02 9.63187754e-01
-1.89924743e-02 2.87494913e-04 2.54335940e-01 6.70390129e-01
-4.86268461e-01 -9.10596013e-01 -4.19346690e-01 3.92354488e-01
-7.08940983e-01 9.79790837e-02 1.49825260e-01 2.88714826e-01
5.97943246e-01 8.95538747e-01 -5.88050000e-02 -3.92111361e-01
5.27554631e-01 1.06442228e-01 4.51758578e-02 -4.62850928e-01
-7.02920377e-01 -2.27664262e-01 -1.63164362e-01 -5.98307312e-01
-2.35733882e-01 -5.79443574e-01 -7.99271822e-01 -8.13625008e-02
-3.01397592e-01 4.46434394e-02 7.98606753e-01 8.27326000e-01
2.17154548e-01 6.94390774e-01 9.76526558e-01 -6.72186792e-01
-1.47140837e+00 -1.10319328e+00 -5.99506617e-01 6.60858810e-01
7.16854811e-01 -7.19204783e-01 -8.15023363e-01 -1.07600808e-01] | [9.300859451293945, 3.8833847045898438] |
cbe6cc9b-0710-4ba0-bfd7-f928f5339ab2 | learning-data-driven-reflectance-priors-for | 1510.02413 | null | http://arxiv.org/abs/1510.02413v1 | http://arxiv.org/pdf/1510.02413v1.pdf | Learning Data-driven Reflectance Priors for Intrinsic Image Decomposition | We propose a data-driven approach for intrinsic image decomposition, which is
the process of inferring the confounding factors of reflectance and shading in
an image. We pose this as a two-stage learning problem. First, we train a model
to predict relative reflectance ordering between image patches (`brighter',
`darker', `same') from large-scale human annotations, producing a data-driven
reflectance prior. Second, we show how to naturally integrate this learned
prior into existing energy minimization frameworks for intrinsic image
decomposition. We compare our method to the state-of-the-art approach of Bell
et al. on both decomposition and image relighting tasks, demonstrating the
benefits of the simple relative reflectance prior, especially for scenes under
challenging lighting conditions. | ['Philipp Krähenbühl', 'Tinghui Zhou', 'Alexei A. Efros'] | 2015-10-08 | learning-data-driven-reflectance-priors-for-1 | http://openaccess.thecvf.com/content_iccv_2015/html/Zhou_Learning_Data-Driven_Reflectance_ICCV_2015_paper.html | http://openaccess.thecvf.com/content_iccv_2015/papers/Zhou_Learning_Data-Driven_Reflectance_ICCV_2015_paper.pdf | iccv-2015-12 | ['intrinsic-image-decomposition', 'image-relighting'] | ['computer-vision', 'computer-vision'] | [ 8.90121877e-01 1.12805009e-01 2.35129938e-01 -5.11988819e-01
-7.36125112e-01 -5.12742817e-01 4.06627983e-01 -2.29860649e-01
-1.29507899e-01 2.34980255e-01 3.90627831e-01 -1.13583170e-01
2.90116519e-01 -6.56414688e-01 -8.08653414e-01 -6.70175612e-01
3.92044216e-01 5.41263297e-02 -5.45901954e-02 -1.57320812e-01
1.44165531e-01 4.56623435e-01 -1.58158958e+00 4.94328588e-01
7.02161491e-01 8.94717038e-01 6.98791221e-02 7.86546886e-01
1.54318362e-01 6.52512312e-01 -1.20509483e-01 -2.85056621e-01
4.70412016e-01 -3.98329407e-01 -8.42718124e-01 4.91435170e-01
5.98345160e-01 -5.62372386e-01 8.26235563e-02 8.40612113e-01
1.80200070e-01 4.01882529e-02 7.00427473e-01 -8.84852111e-01
-7.20269263e-01 -1.60837427e-01 -1.05868471e+00 -2.82827407e-01
2.86473393e-01 3.26706499e-01 1.14172614e+00 -1.04101121e+00
6.11671090e-01 1.08027470e+00 6.10715389e-01 4.59410429e-01
-1.95266247e+00 -9.49796513e-02 1.48581192e-01 -1.90896749e-01
-1.21535218e+00 -7.79754758e-01 9.68134999e-01 -8.63162160e-01
9.30050313e-01 3.81155938e-01 5.80017567e-01 7.78003812e-01
-2.93212444e-01 7.00609386e-01 1.37360644e+00 -5.64457893e-01
6.00143149e-02 -2.67950334e-02 3.97167392e-02 6.54521286e-01
9.20782238e-02 2.04114959e-01 -5.67765296e-01 -3.47682908e-02
7.95994461e-01 -8.92177299e-02 -3.52598667e-01 -5.43899715e-01
-1.05795670e+00 4.37063426e-01 3.11818570e-01 -2.06121027e-01
-4.77912217e-01 3.73495311e-01 -8.79806057e-02 -4.22053374e-02
7.96572685e-01 2.39450648e-01 -5.29237151e-01 2.87603170e-01
-8.67512405e-01 7.82086775e-02 5.21917582e-01 6.24435306e-01
1.19536972e+00 -1.85650423e-01 -1.40375411e-02 9.81583536e-01
7.30382025e-01 3.59467030e-01 -3.33746672e-01 -1.35014367e+00
1.34941697e-01 4.81132418e-01 4.58921075e-01 -8.93634379e-01
-2.35694885e-01 -1.35751307e-01 -5.12774289e-01 6.04070485e-01
2.96976566e-01 7.17471689e-02 -9.59702849e-01 1.82249808e+00
4.54335153e-01 -6.77192360e-02 -3.54990140e-02 1.01393664e+00
5.34246504e-01 5.01105428e-01 -9.85007361e-02 -9.42308828e-02
1.42615199e+00 -1.10190392e+00 -3.91753942e-01 -3.85593921e-01
2.21502677e-01 -9.12558198e-01 1.27867472e+00 5.76524436e-01
-1.21318769e+00 -4.92776304e-01 -8.30622554e-01 -6.06119931e-01
-1.12846131e-02 4.49474484e-01 5.80210924e-01 6.37372434e-01
-1.20696187e+00 3.78904700e-01 -9.61061656e-01 -3.28063607e-01
4.32618082e-01 9.63742062e-02 -1.91378772e-01 -8.77089724e-02
-4.41437513e-01 6.79308295e-01 -1.60398364e-01 2.43538424e-01
-1.16331196e+00 -8.71993005e-01 -8.33025455e-01 -1.74702972e-01
1.75061911e-01 -8.43621373e-01 1.11130631e+00 -1.31507850e+00
-1.58474112e+00 1.50647795e+00 -6.19375944e-01 -5.25230728e-03
3.60145628e-01 -3.93704891e-01 1.47995995e-02 1.87679604e-01
-1.61527306e-01 7.18612373e-01 1.03380334e+00 -2.09428382e+00
-2.73952842e-01 -4.80845153e-01 9.58434045e-02 3.32376808e-01
-3.07310373e-02 2.49701254e-02 -4.85352069e-01 -4.99610990e-01
1.20773338e-01 -6.86719418e-01 -2.05959156e-01 5.32923341e-01
-3.76849532e-01 2.06925854e-01 4.50948834e-01 -8.99830699e-01
8.37273717e-01 -2.28351784e+00 4.29565430e-01 4.59933542e-02
3.26021850e-01 -1.72736377e-01 -2.50287831e-01 3.59218150e-01
-1.88564390e-01 -4.19994779e-02 -5.54188848e-01 -6.85663044e-01
5.77527881e-02 2.12532043e-01 -3.42409104e-01 5.06285131e-01
3.29412133e-01 6.17221832e-01 -7.34426379e-01 -1.78656399e-01
1.63191468e-01 8.80853832e-01 -6.54877126e-01 5.09028733e-01
-3.40588331e-01 7.77211368e-01 3.88277359e-02 5.84830523e-01
8.27870905e-01 -1.32229418e-01 2.12745667e-01 -7.29157627e-01
-3.37829918e-01 2.72607535e-01 -9.76691306e-01 1.82845354e+00
-5.69052279e-01 5.46782374e-01 6.16138518e-01 -7.02924669e-01
8.30774486e-01 5.37420735e-02 6.33266509e-01 -4.73632514e-01
2.76991841e-03 5.36125712e-02 -3.62887949e-01 -5.44780016e-01
3.04638833e-01 -3.02734017e-01 4.94632214e-01 6.35425448e-01
-2.38410681e-01 -2.78176457e-01 -5.06248251e-02 3.07230726e-02
9.18853760e-01 8.02481294e-01 -1.39831379e-02 -2.69367009e-01
4.04133350e-01 -2.00050801e-01 4.94206220e-01 3.11651021e-01
3.66228223e-02 1.01821506e+00 5.39514005e-01 -4.38629985e-01
-1.15800571e+00 -1.10134172e+00 -6.56588078e-02 1.23272943e+00
3.03180497e-02 -3.54240268e-01 -8.77497315e-01 -2.77514249e-01
-7.23856390e-02 6.15121245e-01 -7.47416615e-01 1.28034934e-01
-4.24224943e-01 -8.21645379e-01 6.01824597e-02 4.74381238e-01
3.10041159e-01 -6.75025523e-01 -8.34989250e-01 -2.46946663e-01
-3.41765612e-01 -1.18963718e+00 -4.46782112e-01 7.50226229e-02
-6.82644427e-01 -9.89554286e-01 -3.28776360e-01 -4.51807588e-01
1.12735629e+00 5.45908034e-01 1.67156708e+00 2.99362272e-01
-5.89688957e-01 7.38584936e-01 -2.51171619e-01 -7.53064975e-02
-1.76113144e-01 -4.69516993e-01 -4.31219608e-01 3.35976750e-01
6.26362488e-02 -7.14549661e-01 -1.11982143e+00 3.75001162e-01
-1.11007082e+00 6.05243921e-01 3.75891984e-01 7.01981783e-01
9.06686544e-01 -7.07007647e-02 -1.67152330e-01 -9.59120333e-01
2.62614697e-01 -1.08751424e-01 -6.96771383e-01 3.00929368e-01
-5.26630759e-01 -4.30983640e-02 2.37692699e-01 6.04432891e-04
-1.24402392e+00 4.21053708e-01 -6.73199594e-02 -1.89031258e-01
-3.37781250e-01 7.49412626e-02 -4.05048609e-01 -1.15316667e-01
5.01243532e-01 9.73320156e-02 -7.68552348e-02 -5.89759648e-01
6.90249979e-01 3.32109779e-01 6.97593391e-01 -9.78695929e-01
8.29588175e-01 1.03432846e+00 -3.26042175e-02 -7.51233399e-01
-1.05560243e+00 -4.73863274e-01 -8.70366871e-01 -1.11273050e-01
1.25377214e+00 -1.07763040e+00 -7.29748726e-01 5.05297601e-01
-1.19547248e+00 -8.76502395e-01 -4.50336993e-01 -2.67023947e-02
-6.95604622e-01 4.51425433e-01 -4.14477170e-01 -8.74224007e-01
-2.65158623e-01 -1.11364293e+00 1.58868432e+00 -3.30084972e-02
1.47402734e-02 -1.01225090e+00 3.46650004e-01 7.66481102e-01
3.12628388e-01 5.02558112e-01 8.96129966e-01 4.17516500e-01
-7.44576454e-01 3.51961434e-01 -6.17595255e-01 4.98291105e-01
2.87091941e-01 3.41190457e-01 -1.29578090e+00 -3.11710447e-01
1.70345083e-01 -4.36432272e-01 9.59021211e-01 2.90656716e-01
1.13756752e+00 -3.09925824e-01 -2.39107776e-02 1.09738302e+00
1.74145508e+00 -3.21941763e-01 8.46959472e-01 1.06910028e-01
9.19907033e-01 8.61253321e-01 4.60719705e-01 3.78397197e-01
6.94340527e-01 6.49227262e-01 6.35241449e-01 -6.92354500e-01
-3.82122725e-01 -1.75268784e-01 3.19442183e-01 3.36410314e-01
-1.94469720e-01 -3.80491838e-02 -8.60942900e-01 5.46264231e-01
-1.74237561e+00 -6.19177878e-01 -2.49848202e-01 2.30249929e+00
1.00526881e+00 -3.25532556e-01 7.42681175e-02 -2.31645733e-01
3.34350675e-01 3.21705669e-01 -4.54810441e-01 -2.70909458e-01
-1.00528806e-01 9.21089277e-02 3.05134028e-01 8.97524774e-01
-8.77676427e-01 8.93032134e-01 7.14454412e+00 1.57433838e-01
-9.02166665e-01 5.27505390e-02 9.23236907e-01 -4.51436965e-03
-8.99492562e-01 4.29153681e-01 -4.10227478e-01 9.39435810e-02
4.50168669e-01 5.53777099e-01 9.73988414e-01 3.51608902e-01
4.01238948e-01 -4.24596906e-01 -1.25446975e+00 8.28433633e-01
2.09272504e-01 -8.88998449e-01 -1.20482013e-01 7.64461458e-02
8.28410566e-01 2.65508443e-02 9.88654885e-03 -4.15326267e-01
6.37417555e-01 -1.12125266e+00 6.17119730e-01 5.33920109e-01
9.47143734e-01 -2.41564497e-01 5.23300143e-03 -2.61246134e-02
-1.08700001e+00 2.04540148e-01 -2.81946599e-01 6.81264699e-02
2.18955696e-01 9.01313663e-01 -4.08108205e-01 3.83654237e-01
7.94490099e-01 8.17066312e-01 -4.75944310e-01 3.93235356e-01
-5.58836281e-01 4.38013881e-01 -3.05164635e-01 8.85166109e-01
-3.22163582e-01 -7.02905118e-01 3.32038283e-01 1.08685291e+00
-8.40069503e-02 3.26119602e-01 9.95064229e-02 1.34981406e+00
-1.07613504e-01 -1.17154457e-01 -4.77187306e-01 2.22214937e-01
-5.89698814e-02 1.61326218e+00 -5.99212408e-01 -6.21211566e-02
-4.84304249e-01 1.29874706e+00 4.92145479e-01 7.32422233e-01
-6.41416669e-01 1.95135459e-01 7.60529160e-01 2.26238728e-01
3.21367443e-01 -3.82978469e-01 -5.95595717e-01 -1.27363658e+00
6.16047829e-02 -8.11805606e-01 1.15973823e-01 -1.19759226e+00
-1.16636896e+00 2.45492563e-01 -2.03513533e-01 -9.52636600e-01
3.33650947e-01 -8.43936205e-01 -5.22101462e-01 9.58115637e-01
-1.91527486e+00 -1.52559781e+00 -6.72768772e-01 5.45718610e-01
4.32027936e-01 4.66869891e-01 8.24129760e-01 1.70070618e-01
-5.43981850e-01 2.39896048e-02 8.15447718e-02 1.03353895e-02
7.09449172e-01 -1.37128484e+00 3.04490685e-01 1.06206131e+00
1.08861357e-01 6.86261952e-01 7.70218849e-01 -3.13270658e-01
-1.66208673e+00 -8.07631791e-01 3.57624203e-01 -6.49273634e-01
4.03256387e-01 -5.30382752e-01 -7.29246557e-01 6.79712653e-01
3.77132684e-01 1.60237789e-01 9.04111087e-01 2.24879175e-01
-7.80047417e-01 -2.73707092e-01 -1.04037738e+00 5.90974092e-01
1.04639482e+00 -6.82516694e-01 -1.52288809e-01 3.61105621e-01
4.38748956e-01 -3.29175293e-01 -9.60263550e-01 3.08268726e-01
7.39545166e-01 -1.36264431e+00 1.27700603e+00 -2.57819384e-01
7.33822882e-01 -5.90633571e-01 -3.57724190e-01 -1.15410733e+00
-3.48761261e-01 -6.56101108e-01 -3.51142255e-03 1.14310253e+00
3.12497675e-01 -5.10074675e-01 5.27293324e-01 7.74885416e-01
2.88037322e-02 -5.47388494e-01 -4.19052184e-01 -2.99306899e-01
-1.44063547e-01 -2.03643098e-01 3.34507763e-01 7.90849805e-01
-4.83929515e-01 3.14777136e-01 -4.01708037e-01 2.29736552e-01
8.98231804e-01 3.59179497e-01 1.01589823e+00 -1.14312065e+00
-5.85304379e-01 -3.18690717e-01 1.21837825e-01 -1.25834370e+00
-2.15701666e-02 -6.48118377e-01 3.62451494e-01 -1.68671739e+00
5.61157405e-01 -3.88098270e-01 -1.77900791e-01 5.56937277e-01
-3.43318284e-01 6.67853951e-01 -5.35009317e-02 3.84211540e-01
-3.63817364e-01 4.65245187e-01 1.16492319e+00 -4.72113863e-02
-1.14190884e-01 -4.34008121e-01 -1.00937450e+00 6.94320023e-01
5.19619286e-01 -2.36263543e-01 -4.52290833e-01 -1.09957802e+00
6.21888995e-01 -2.43979916e-01 6.10567153e-01 -3.96512896e-01
-1.36466786e-01 -3.18963885e-01 3.06827843e-01 -3.83609027e-01
4.20884669e-01 -9.29183483e-01 2.46623293e-01 1.21068187e-01
-3.02607059e-01 -2.31471300e-01 2.27466170e-02 5.21014929e-01
2.45451387e-02 1.67395398e-02 8.84179235e-01 -1.68620721e-01
-4.29083794e-01 1.51377052e-01 9.15532634e-02 -4.58099879e-02
6.85169756e-01 -3.23288471e-01 -3.81979942e-01 -2.23052233e-01
-5.55128694e-01 2.70151626e-02 1.00035334e+00 9.56917927e-02
6.41125977e-01 -1.02576470e+00 -7.95822740e-01 3.02331746e-01
1.11691639e-01 3.00126314e-01 3.60837542e-02 1.04401076e+00
-6.26785159e-01 -3.52612942e-01 -1.96815804e-02 -8.22508454e-01
-1.29382694e+00 4.20533121e-01 4.59530026e-01 -1.11961313e-01
-6.85932040e-01 8.07427526e-01 7.00880349e-01 -4.05643404e-01
-4.50124070e-02 -2.59757131e-01 2.46877104e-01 -2.55254358e-01
3.03114653e-01 2.66022235e-01 -2.70611402e-02 -6.32608891e-01
-2.69429505e-01 9.92458880e-01 1.91086873e-01 -3.19316268e-01
1.64008844e+00 -4.27153140e-01 -6.51129067e-01 3.31135541e-01
1.09600854e+00 2.26954103e-01 -1.82070112e+00 -1.22321442e-01
-3.50063473e-01 -7.95258880e-01 2.37308845e-01 -8.84726167e-01
-1.08355355e+00 1.05568600e+00 5.57221115e-01 4.51280437e-02
1.51346159e+00 -1.13541879e-01 4.87850994e-01 -2.35971157e-03
7.63667375e-02 -1.06189024e+00 2.21077830e-01 4.24727350e-02
9.67015862e-01 -1.30978513e+00 3.77125740e-01 -7.60118067e-01
-6.03448391e-01 1.03117549e+00 3.60502124e-01 -2.99275257e-02
4.65968937e-01 4.35485482e-01 2.72115350e-01 -4.10159379e-01
-6.79968953e-01 -2.68921435e-01 5.20948112e-01 6.11332595e-01
6.62169933e-01 -1.46866962e-01 -3.19989882e-02 6.42132759e-03
1.44440085e-01 -9.22330543e-02 2.40964770e-01 7.59850323e-01
-2.97752887e-01 -1.14505148e+00 -3.69989127e-01 -7.94781372e-03
-2.01804399e-01 -2.08055571e-01 -5.14833927e-01 1.96746856e-01
5.48679791e-02 9.09178793e-01 -5.05644269e-02 -6.36997223e-02
2.13232741e-01 -1.13137849e-01 7.00185597e-01 -6.96397662e-01
-3.69618118e-01 5.36054909e-01 6.60316050e-02 -8.28452051e-01
-8.17091286e-01 -6.63798749e-01 -9.31967676e-01 1.97760314e-01
-1.28345594e-01 -4.66224968e-01 7.31004298e-01 7.86385596e-01
3.68918389e-01 5.19092441e-01 8.02133918e-01 -1.04076803e+00
-1.28445834e-01 -4.60944980e-01 -3.70026886e-01 8.36330295e-01
5.37397683e-01 -3.24699670e-01 -3.75951618e-01 6.70432985e-01] | [9.895909309387207, -2.9223687648773193] |
fd4a0c31-6f85-4d3e-b7cf-0ebb9a2a85de | where-to-focus-deep-attention-based-spatially | 1709.05769 | null | http://arxiv.org/abs/1709.05769v1 | http://arxiv.org/pdf/1709.05769v1.pdf | Where to Focus: Deep Attention-based Spatially Recurrent Bilinear Networks for Fine-Grained Visual Recognition | Fine-grained visual recognition typically depends on modeling subtle
difference from object parts. However, these parts often exhibit dramatic
visual variations such as occlusions, viewpoints, and spatial transformations,
making it hard to detect. In this paper, we present a novel attention-based
model to automatically, selectively and accurately focus on critical object
regions with higher importance against appearance variations. Given an image,
two different Convolutional Neural Networks (CNNs) are constructed, where the
outputs of two CNNs are correlated through bilinear pooling to simultaneously
focus on discriminative regions and extract relevant features. To capture
spatial distributions among the local regions with visual attention, soft
attention based spatial Long-Short Term Memory units (LSTMs) are incorporated
to realize spatially recurrent yet visually selective over local input
patterns. All the above intuitions equip our network with the following novel
model: two-stream CNN layers, bilinear pooling layer, spatial recurrent layer
with location attention are jointly trained via an end-to-end fashion to serve
as the part detector and feature extractor, whereby relevant features are
localized and extracted attentively. We show the significance of our network
against two well-known visual recognition tasks: fine-grained image
classification and person re-identification. | ['Lin Wu', 'Yang Wang'] | 2017-09-18 | null | null | null | null | ['deep-attention', 'fine-grained-visual-recognition', 'deep-attention'] | ['computer-vision', 'computer-vision', 'natural-language-processing'] | [ 1.73164129e-01 -3.37969691e-01 -1.09316923e-01 -5.19376099e-01
-4.20267463e-01 -4.78878945e-01 5.97660244e-01 -2.61653829e-02
-3.49213898e-01 4.32257980e-01 6.01089597e-01 3.05402011e-01
7.00772926e-02 -6.12606883e-01 -8.82983863e-01 -5.31099677e-01
6.12234510e-02 -1.54858604e-01 2.34024659e-01 1.31024227e-01
4.15003359e-01 8.89525831e-01 -1.46211386e+00 6.82089865e-01
4.45509046e-01 1.21709001e+00 1.08427830e-01 6.96402609e-01
1.80731744e-01 7.47276604e-01 -5.01920402e-01 -1.07904412e-01
2.34680220e-01 -7.33491331e-02 -7.73383498e-01 3.94473910e-01
7.20903099e-01 -3.23698878e-01 -6.05259061e-01 1.04058576e+00
4.98016179e-01 2.90133297e-01 6.86947405e-01 -8.21928680e-01
-1.46824574e+00 1.71144679e-01 -8.40934217e-01 6.38433337e-01
3.21232855e-01 5.42753756e-01 8.28666210e-01 -1.19605517e+00
1.32838607e-01 1.43691182e+00 6.42281294e-01 5.17028570e-01
-1.13260984e+00 -4.45051163e-01 7.27806330e-01 2.41599441e-01
-1.51497054e+00 -4.36064750e-01 7.48694837e-01 -4.21970934e-01
1.14008725e+00 3.07431936e-01 6.21500611e-01 1.10122740e+00
2.61043489e-01 8.28114510e-01 1.00248837e+00 -2.30290115e-01
-8.36828351e-02 2.13353008e-01 2.88724482e-01 6.74334884e-01
6.65780678e-02 2.06633449e-01 -5.90263486e-01 2.12055504e-01
9.99564707e-01 9.16823506e-01 -2.32319981e-01 -1.42302111e-01
-1.10668898e+00 3.81280243e-01 1.15023959e+00 4.17644441e-01
-5.70259988e-01 1.76369905e-01 3.93034592e-02 7.94765428e-02
2.88186550e-01 1.75618380e-01 -4.13027644e-01 4.77239609e-01
-7.35330939e-01 1.29228234e-01 -1.00383013e-01 7.81769872e-01
8.45440388e-01 -1.02282181e-01 -9.70904708e-01 8.43100369e-01
4.20032978e-01 2.60140687e-01 6.80015385e-01 -3.32470953e-01
4.19222742e-01 9.58085835e-01 -2.75158249e-02 -1.04863954e+00
-2.44799599e-01 -7.35927999e-01 -1.21304083e+00 1.92876592e-01
1.41525865e-01 1.61597177e-01 -1.30778289e+00 1.63125932e+00
2.94379964e-02 1.59998760e-01 -2.42720842e-01 1.13624680e+00
7.64570534e-01 3.98312151e-01 4.28370059e-01 3.69126767e-01
1.67804193e+00 -1.13216734e+00 -1.94156975e-01 -4.80575383e-01
9.05335546e-02 -5.79681873e-01 1.13093090e+00 -6.89384490e-02
-1.15267503e+00 -1.08438468e+00 -8.09150398e-01 -4.21444803e-01
-7.77740300e-01 2.83306062e-01 2.72797495e-01 2.86177963e-01
-1.04467463e+00 4.65396583e-01 -4.40866023e-01 -4.71725017e-01
8.92871916e-01 3.21170926e-01 -5.06065428e-01 -1.40644476e-01
-8.89668941e-01 8.29710186e-01 -9.90382396e-03 5.79208612e-01
-9.73741651e-01 -5.16305327e-01 -8.43595147e-01 2.60908276e-01
-3.18625480e-01 -9.99352992e-01 8.17356110e-01 -1.30298364e+00
-1.13245058e+00 9.79603231e-01 -5.53506434e-01 -2.82606512e-01
1.34005919e-01 -1.50025770e-01 -2.65034705e-01 -3.50030437e-02
2.33637869e-01 6.89444900e-01 1.16878951e+00 -1.05624270e+00
-6.08008325e-01 -7.06418157e-01 -9.42954868e-02 1.82489976e-01
-3.00595134e-01 3.93376410e-01 -4.84092474e-01 -9.42791402e-01
2.01468449e-02 -4.74015117e-01 -3.09937865e-01 9.21460763e-02
-6.84500754e-01 -3.73773962e-01 8.16543579e-01 -7.27572203e-01
9.31220412e-01 -2.24028111e+00 1.51892647e-01 2.03039065e-01
3.62226278e-01 2.03304276e-01 -2.89545625e-01 -6.33084076e-03
-2.37120032e-01 1.64420500e-01 8.58430490e-02 -3.28535825e-01
1.63595397e-02 -2.40988180e-01 -4.92229819e-01 5.01498640e-01
6.45593226e-01 1.57660389e+00 -6.11410081e-01 -1.91533267e-01
4.86991107e-01 6.90829396e-01 -3.76213878e-01 3.38241756e-01
2.13731259e-01 3.78219098e-01 -4.31307763e-01 8.88295650e-01
5.94897926e-01 -3.38596404e-01 -3.98546249e-01 -5.16906023e-01
-5.22265658e-02 -4.60525490e-02 -8.60341311e-01 1.40228844e+00
-2.65395254e-01 6.83098853e-01 -2.48691320e-01 -9.12267148e-01
8.11296582e-01 -7.99844936e-02 -9.69461650e-02 -1.23655045e+00
1.36116996e-01 -3.19281608e-01 -2.24505693e-01 -4.77182031e-01
4.08309668e-01 4.78325263e-02 -1.90177873e-01 2.62857884e-01
9.52243358e-02 6.59170926e-01 -4.34232920e-01 -9.94691625e-02
7.09585428e-01 -6.00352548e-02 2.79961705e-01 -8.05473030e-02
4.29001838e-01 -5.46729863e-01 3.43468726e-01 8.19524288e-01
-5.47672391e-01 9.88788188e-01 4.04307898e-03 -8.12437356e-01
-9.24345672e-01 -1.09884095e+00 9.04665440e-02 1.54305661e+00
3.75665873e-01 -3.11300792e-02 -4.86652672e-01 -6.34651363e-01
8.53408128e-02 6.16034009e-02 -1.19206893e+00 -3.39786023e-01
-6.28766894e-01 -4.14878279e-01 5.08645535e-01 1.05878913e+00
8.54589701e-01 -1.35706902e+00 -6.74045384e-01 -2.79454030e-02
2.24427313e-01 -8.48675609e-01 -9.45274770e-01 9.31417346e-02
-5.22574902e-01 -9.35095370e-01 -1.11703408e+00 -1.04061794e+00
8.20817232e-01 6.97611094e-01 9.45104480e-01 -8.47167615e-03
-6.06783211e-01 3.08491737e-01 -7.77450949e-02 -1.12413332e-01
6.04793012e-01 1.87627450e-02 -1.16610182e-02 5.87678790e-01
5.54846704e-01 -5.10312676e-01 -1.10009801e+00 3.08429062e-01
-6.10044658e-01 -9.42813978e-02 9.73049581e-01 9.61769104e-01
7.25742400e-01 -2.99391001e-01 3.50558698e-01 -4.31881875e-01
5.63508451e-01 -2.37052768e-01 -1.58088028e-01 5.65852046e-01
1.76183075e-01 2.25867257e-01 6.44512236e-01 -5.26856780e-01
-9.76760805e-01 1.49939939e-01 1.26486093e-01 -7.10602164e-01
-5.06976366e-01 -3.08307656e-03 -3.08090955e-01 -2.03151956e-01
4.60381180e-01 5.89167476e-01 -5.96887529e-01 -5.91751277e-01
5.58652759e-01 5.61669886e-01 6.32091761e-01 -2.55586982e-01
7.41704822e-01 4.91259396e-01 -4.16757703e-01 -6.62690580e-01
-9.38643813e-01 -3.90391111e-01 -9.63634849e-01 -8.17298517e-03
1.02012396e+00 -9.17991936e-01 -8.69164884e-01 5.03046274e-01
-1.14180076e+00 -3.12849283e-01 -4.52922940e-01 2.01726958e-01
-5.92013188e-02 9.85588655e-02 -4.70075488e-01 -6.24307990e-01
-4.35356587e-01 -9.93097961e-01 1.40838778e+00 7.68065512e-01
-8.93852189e-02 -7.96235442e-01 -2.35215515e-01 2.08245385e-02
5.42063653e-01 1.16988346e-01 5.71562529e-01 -2.48909414e-01
-8.13140750e-01 -1.91467792e-01 -9.96153295e-01 1.04002006e-01
2.28473708e-01 -2.45679542e-01 -1.29160309e+00 -4.63707060e-01
-3.24818343e-01 -2.54880846e-01 1.36583602e+00 5.55475473e-01
1.58016193e+00 -5.52214682e-01 -5.83953977e-01 9.15650427e-01
1.19014800e+00 -1.30441561e-01 6.52437091e-01 1.58063039e-01
1.06278539e+00 6.14967108e-01 3.17743979e-02 2.84392893e-01
2.92757571e-01 6.82753384e-01 2.22091585e-01 -4.89886940e-01
-3.82423013e-01 -2.98251480e-01 2.00813100e-01 7.60495886e-02
-1.49114266e-01 1.56630024e-01 -4.71383780e-01 7.02988923e-01
-1.66710722e+00 -1.09304357e+00 3.93309236e-01 2.06609559e+00
5.39290607e-01 -7.90357292e-02 1.29268393e-01 -2.78600216e-01
8.31527114e-01 2.99821675e-01 -7.27837563e-01 -3.28407258e-01
-3.66938949e-01 1.71693683e-01 4.25558627e-01 4.19225216e-01
-1.29817247e+00 1.00681293e+00 5.65327501e+00 5.45316458e-01
-1.37000573e+00 6.51159063e-02 9.56432521e-01 -2.53892690e-01
-1.32207766e-01 -4.10682470e-01 -7.57597268e-01 4.45542693e-01
3.53814721e-01 2.83692956e-01 3.10531169e-01 6.42529309e-01
7.64460638e-02 3.01423371e-01 -1.01583278e+00 1.25148201e+00
2.81063855e-01 -1.30986607e+00 3.71058285e-01 -2.47679465e-02
7.63135195e-01 -2.20058281e-02 5.77591419e-01 2.45532542e-01
3.09318811e-01 -1.64358532e+00 7.71966040e-01 9.49932814e-01
8.30689609e-01 -6.54613733e-01 4.82575625e-01 -5.20118587e-02
-1.61319172e+00 -3.66952360e-01 -5.35659790e-01 -1.43393785e-01
-9.25129727e-02 2.33417526e-01 -5.05788088e-01 1.00934841e-01
1.01358461e+00 8.40485334e-01 -9.47685838e-01 1.11667538e+00
-1.24766737e-01 7.04348981e-02 4.17589247e-02 6.40155673e-02
2.64273435e-01 4.70568240e-01 2.55280614e-01 1.51367807e+00
-2.99057476e-02 8.45010951e-02 5.41578606e-02 1.22920835e+00
-2.00661436e-01 -3.00838232e-01 -5.28159797e-01 3.39928031e-01
1.81453571e-01 1.28382659e+00 -4.77705896e-01 -3.03533852e-01
-5.56392133e-01 1.54972947e+00 6.49766624e-01 7.82014966e-01
-6.62509918e-01 -5.32829165e-01 8.41224730e-01 1.23920955e-01
6.43501043e-01 -4.74668667e-02 -4.76200610e-01 -1.30695105e+00
1.35736033e-01 -5.23187459e-01 5.00126421e-01 -7.12778330e-01
-1.70515239e+00 7.80990779e-01 -4.40979511e-01 -1.04640973e+00
4.59812619e-02 -7.53896415e-01 -8.15053344e-01 1.40357161e+00
-1.66181552e+00 -1.63278627e+00 -5.65617681e-01 1.06205511e+00
6.80591524e-01 -1.23093113e-01 6.04937673e-01 2.24445790e-01
-7.40433514e-01 9.93203700e-01 -2.60253966e-01 5.38929701e-01
6.18400991e-01 -1.11179292e+00 6.17167294e-01 1.06263542e+00
9.06016976e-02 9.72909749e-01 3.72732654e-02 -4.77478325e-01
-1.02766728e+00 -1.64483893e+00 9.25821900e-01 -5.20455956e-01
1.21169604e-01 -5.57001293e-01 -9.62640047e-01 8.59505475e-01
7.95631707e-02 6.15042448e-01 3.78589600e-01 1.53416723e-01
-8.35412681e-01 -1.77862227e-01 -9.81245875e-01 5.53209484e-01
1.08889174e+00 -9.85555768e-01 -7.13929892e-01 1.20293848e-01
5.04621863e-01 -6.28124923e-02 -3.96871299e-01 1.83561280e-01
7.88438916e-01 -8.98007393e-01 1.36052120e+00 -9.35289264e-01
1.25365555e-01 -3.57669681e-01 -2.01653779e-01 -1.08957839e+00
-1.05699658e+00 -1.78202987e-01 3.88397314e-02 1.04486179e+00
2.60363936e-01 -6.41546011e-01 6.22576594e-01 5.88668942e-01
5.19997180e-02 -6.98591053e-01 -8.06123435e-01 -3.90561908e-01
-1.15886629e-01 -6.13474362e-02 5.93648195e-01 6.86701655e-01
-1.73730195e-01 4.69597310e-01 -3.87655705e-01 3.87527853e-01
4.82592225e-01 3.07286918e-01 4.06016678e-01 -1.01807606e+00
-3.05203497e-01 -7.12228179e-01 -5.52366316e-01 -1.38303769e+00
-1.45322204e-01 -5.43058813e-01 -7.24102184e-02 -1.42430973e+00
6.24022067e-01 -1.60825774e-01 -8.04429471e-01 7.09356844e-01
-3.82714003e-01 8.32508862e-01 2.70917773e-01 3.03712457e-01
-8.90921652e-01 5.24911225e-01 1.16683042e+00 -4.84077573e-01
-2.06678391e-01 -2.27573309e-02 -1.07981062e+00 6.98591173e-01
5.40979028e-01 7.27048814e-02 -1.30904540e-02 -6.49434566e-01
-2.76079237e-01 -4.38453704e-01 1.11802459e+00 -8.90642524e-01
4.41792160e-01 -9.99003425e-02 1.41910565e+00 -5.75285614e-01
2.09454209e-01 -5.63005745e-01 -3.50515097e-01 2.69722670e-01
-5.84357083e-01 6.51017874e-02 2.64010370e-01 5.08472025e-01
-2.33651578e-01 3.24801117e-01 8.09996545e-01 -2.19153225e-01
-1.03791308e+00 8.07751179e-01 -1.79488391e-01 -2.17345744e-01
9.91707444e-01 -5.00435114e-01 -1.98182970e-01 -1.52359784e-01
-9.74145234e-01 -6.50877655e-02 1.64736003e-01 5.70070863e-01
1.01099241e+00 -1.53713286e+00 -6.63852990e-01 7.42272437e-01
2.34565452e-01 -1.20381430e-01 8.01724792e-01 4.39550191e-01
8.37365985e-02 7.23987162e-01 -5.71860373e-01 -5.55527806e-01
-1.06070375e+00 9.49894667e-01 7.49118030e-01 -4.97727953e-02
-5.85101545e-01 1.24820113e+00 8.09814990e-01 -1.30847991e-01
3.32341105e-01 -6.05182886e-01 -3.70639563e-01 -2.18893975e-01
9.22515869e-01 -1.32631725e-02 -2.39143521e-01 -9.10579443e-01
-5.74480236e-01 9.82573211e-01 -2.81689882e-01 1.90970778e-01
1.19160390e+00 -4.51065123e-01 1.55127972e-01 1.69819787e-01
1.35265183e+00 -1.40628427e-01 -1.66922486e+00 -6.59209430e-01
-3.67397815e-01 -5.59750021e-01 -1.48603633e-01 -8.42952371e-01
-1.11775315e+00 1.30447316e+00 8.60496759e-01 1.48968592e-01
1.24226415e+00 2.71971405e-01 5.85834026e-01 -1.16944842e-01
-1.30938366e-01 -7.33181775e-01 3.47520232e-01 4.14394289e-01
1.13614571e+00 -1.21217620e+00 -3.73933822e-01 3.12780589e-02
-5.74447870e-01 9.61142361e-01 8.95864010e-01 -4.57449257e-01
6.00144207e-01 -3.37165827e-03 -2.48321012e-01 -2.00991899e-01
-5.76852143e-01 -5.68969965e-01 8.28804255e-01 8.68200839e-01
3.09319794e-01 7.94847384e-02 6.99093163e-01 9.07248020e-01
9.59419310e-02 -2.79747516e-01 -2.98045665e-01 4.63279217e-01
-5.18612683e-01 -5.53882897e-01 -3.62547904e-01 2.57903248e-01
-3.03730041e-01 -2.18550235e-01 -6.10893309e-01 3.84049624e-01
5.00167787e-01 5.34327984e-01 4.48745281e-01 -4.59342211e-01
4.37690049e-01 -2.85096347e-01 4.92084742e-01 -6.29285455e-01
-7.35596478e-01 -2.41321743e-01 -5.40275037e-01 -5.73462546e-01
-1.12632141e-01 -4.18813109e-01 -7.91970313e-01 -1.04777686e-01
1.15368515e-01 -3.42244893e-01 3.81126557e-03 1.06683385e+00
6.32372439e-01 8.15048993e-01 5.95298827e-01 -1.20096874e+00
-2.34114155e-01 -1.02622807e+00 -4.24690694e-01 6.05106950e-01
7.48339653e-01 -6.08668745e-01 1.70498654e-01 1.01731889e-01] | [9.581080436706543, 1.974305272102356] |
6b25249b-2d9d-4086-9215-883d1c2c1314 | multi-message-shuffled-privacy-in-federated | 2302.11152 | null | https://arxiv.org/abs/2302.11152v1 | https://arxiv.org/pdf/2302.11152v1.pdf | Multi-Message Shuffled Privacy in Federated Learning | We study differentially private distributed optimization under communication constraints. A server using SGD for optimization aggregates the client-side local gradients for model updates using distributed mean estimation (DME). We develop a communication-efficient private DME, using the recently developed multi-message shuffled (MMS) privacy framework. We analyze our proposed DME scheme to show that it achieves the order-optimal privacy-communication-performance tradeoff resolving an open question in [1], whether the shuffled models can improve the tradeoff obtained in Secure Aggregation. This also resolves an open question on the optimal trade-off for private vector sum in the MMS model. We achieve it through a novel privacy mechanism that non-uniformly allocates privacy at different resolutions of the local gradient vectors. These results are directly applied to give guarantees on private distributed learning algorithms using this for private gradient aggregation iteratively. We also numerically evaluate the private DME algorithms. | ['Suhas Diggavi', 'Antonious M. Girgis'] | 2023-02-22 | null | null | null | null | ['distributed-optimization', 'open-question'] | ['methodology', 'natural-language-processing'] | [-4.98283505e-02 -4.19458449e-02 5.44523448e-02 -6.41712248e-01
-1.55368650e+00 -1.03032601e+00 8.76198709e-02 1.95774138e-01
-5.59410930e-01 7.27955043e-01 3.44176173e-01 -2.89682478e-01
-2.99567908e-01 -7.31081903e-01 -8.75591576e-01 -1.40219820e+00
-5.41619241e-01 2.60525703e-01 -3.52268577e-01 1.21417023e-01
5.01672588e-02 7.28401363e-01 -1.02383721e+00 3.59017164e-01
5.00748158e-01 1.12679493e+00 -1.76475212e-01 1.12066329e+00
1.96672961e-01 6.20279849e-01 -5.27460158e-01 -7.14209914e-01
1.04364324e+00 -2.66940892e-01 -8.42878103e-01 -1.72681898e-01
5.19462466e-01 -8.48035336e-01 -2.47020543e-01 1.29140186e+00
8.32195938e-01 -1.48536950e-01 3.38539898e-01 -1.49135208e+00
-2.67842293e-01 8.09413075e-01 -5.76765716e-01 -1.01253860e-01
1.17349997e-02 -2.40055144e-01 1.11465240e+00 -2.27418706e-01
5.52665055e-01 1.18404484e+00 6.47300184e-01 7.26357341e-01
-1.58698523e+00 -5.84489524e-01 -1.90901279e-01 3.54397148e-02
-1.33818817e+00 -4.35189933e-01 5.20057380e-01 -4.90410663e-02
4.74080831e-01 1.24704421e+00 2.06817061e-01 6.07982635e-01
1.30355552e-01 9.24506664e-01 1.48306465e+00 -2.36095205e-01
6.82903707e-01 5.32760799e-01 1.64332360e-01 4.59578246e-01
5.64239264e-01 -8.87608528e-02 -7.81681597e-01 -1.21540844e+00
3.31317753e-01 8.72432664e-02 -6.39026523e-01 -6.97422624e-01
-6.46444440e-01 9.49344218e-01 7.32955486e-02 -1.69529989e-01
-2.64320850e-01 4.96679574e-01 4.66926783e-01 9.59658384e-01
8.70859861e-01 -5.79714589e-02 -8.93985212e-01 -7.35378563e-02
-6.81197405e-01 7.53355861e-01 1.53765607e+00 8.77338946e-01
9.20971215e-01 -5.77081919e-01 -2.82339990e-01 4.70639557e-01
2.62935191e-01 5.19533217e-01 5.02873063e-02 -1.46928823e+00
8.04042637e-01 -2.37735078e-01 4.43490863e-01 -1.06678128e+00
9.35256481e-02 -2.93328404e-01 -1.00934112e+00 4.77875412e-01
4.33301806e-01 -6.08622313e-01 1.92167535e-01 2.08471751e+00
7.74266481e-01 -1.54013053e-01 3.99643809e-01 7.83365488e-01
-1.86090160e-03 6.72260940e-01 -3.32237810e-01 -4.12392884e-01
8.32008719e-01 -8.04677725e-01 -5.60177803e-01 3.99299145e-01
1.14566696e+00 -4.92325306e-01 7.21044958e-01 2.88748682e-01
-1.22278225e+00 6.12930477e-01 -7.99525619e-01 -2.53975570e-01
-6.61298484e-02 -2.44921193e-01 7.60474682e-01 1.16439247e+00
-1.54211521e+00 9.77288127e-01 -9.51356471e-01 -1.73413441e-01
5.65987170e-01 7.73848414e-01 -5.55969715e-01 1.81901336e-01
-7.32250988e-01 1.47321388e-01 -1.95785969e-01 -2.76860029e-01
-9.06843722e-01 -9.13753390e-01 -3.91037852e-01 2.29234785e-01
1.19803324e-02 -9.74296689e-01 1.38065636e+00 -7.15890646e-01
-1.53522670e+00 1.08620393e+00 -2.51290321e-01 -8.34884226e-01
1.12439632e+00 1.27766266e-01 4.73139882e-01 -1.00123440e-03
-3.21342796e-01 -1.09537765e-01 4.88504529e-01 -1.29747927e+00
-8.73719811e-01 -1.12236965e+00 8.84755775e-02 3.58506948e-01
-7.36990869e-01 -3.94512974e-02 1.46657482e-01 -5.29005706e-01
8.15964788e-02 -9.99754906e-01 -4.59245980e-01 5.04334986e-01
-2.87307948e-01 1.28967196e-01 9.89525616e-01 -7.56738663e-01
1.00755465e+00 -2.20436978e+00 1.36047319e-01 5.02488256e-01
6.89177752e-01 -3.16808611e-01 7.47850314e-02 7.09879339e-01
4.17011619e-01 9.37184319e-02 -1.84224576e-01 -1.17559516e+00
4.71675456e-01 3.70009154e-01 -5.08092165e-01 1.06138778e+00
-9.63734448e-01 5.59532464e-01 -5.36198020e-01 -2.08942726e-01
-2.94611812e-01 2.44345993e-01 -8.51349115e-01 3.35677981e-01
5.19357733e-02 3.03904980e-01 -6.90742373e-01 9.54230726e-02
1.47462487e+00 -3.22691113e-01 4.75387275e-01 -5.01554413e-03
1.93316817e-01 -3.33097391e-03 -1.47503364e+00 1.54206753e+00
-3.98390591e-01 2.87875086e-01 1.21406269e+00 -8.16849649e-01
2.55872399e-01 2.37767428e-01 5.24018288e-01 2.26015434e-01
-6.85121343e-02 4.31721598e-01 -8.40106249e-01 -1.32255182e-01
2.80804306e-01 -9.08815395e-03 -2.09493667e-01 9.52226758e-01
-2.30437785e-01 1.39034465e-01 -7.76809216e-01 4.56438273e-01
1.24963212e+00 -5.67262590e-01 -5.71830198e-02 -5.61459124e-01
4.60301846e-01 -5.40820301e-01 4.88205880e-01 1.17630315e+00
-2.14033499e-01 4.74280179e-01 6.08229339e-01 -2.61647522e-01
-7.32422709e-01 -9.11301732e-01 -1.47734405e-02 1.28371739e+00
7.81156402e-03 -6.68155015e-01 -1.14893699e+00 -7.51267731e-01
4.93926018e-01 4.08187389e-01 -6.03796899e-01 1.63940102e-01
-8.30346197e-02 -9.86630678e-01 4.53694552e-01 -6.10758411e-03
6.11899137e-01 -6.86989108e-04 -2.42831081e-01 1.09750116e-02
-2.44650412e-02 -6.85123801e-01 -8.85712802e-01 2.28319585e-01
-8.82497668e-01 -4.55071181e-01 -8.57804298e-01 -3.94752055e-01
6.36019945e-01 4.05380994e-01 5.46419084e-01 -2.42400751e-01
9.99780744e-02 9.82809961e-01 2.94033825e-01 -2.15325996e-01
-4.87293392e-01 2.07363501e-01 2.00271830e-02 4.51825052e-01
2.74934262e-01 -9.83216405e-01 -7.74300635e-01 -4.03967686e-02
-8.54398012e-01 -6.16911910e-02 -4.28445004e-02 6.34554029e-01
5.09273112e-01 -1.79607734e-01 1.92807466e-02 -1.06066942e+00
1.03457820e+00 -4.65141416e-01 -9.85122085e-01 4.04508471e-01
-9.61308837e-01 3.92771959e-01 6.09458804e-01 -8.92252177e-02
-9.06404376e-01 1.58724651e-01 -6.56164214e-02 -2.29464009e-01
2.40322322e-01 -1.56713948e-01 -4.13487434e-01 -8.59538913e-01
3.45642000e-01 4.50635850e-01 3.95386517e-01 -8.14132094e-01
5.20143390e-01 1.14865518e+00 1.54479057e-01 -1.04707968e+00
5.34112930e-01 8.26171219e-01 4.08545256e-01 -4.63660896e-01
-4.00080949e-01 -3.56647015e-01 -3.14127700e-03 5.12880921e-01
3.78355712e-01 -9.56725180e-01 -1.50266528e+00 7.31182098e-01
-1.14854729e+00 -3.03349048e-01 -5.85271657e-01 2.35629365e-01
-9.89176869e-01 8.01311910e-01 -9.84062135e-01 -1.10599649e+00
-8.12863350e-01 -9.71771896e-01 1.01902890e+00 -1.88033879e-01
1.95149913e-01 -1.08670008e+00 2.19515026e-01 2.27959290e-01
6.25792623e-01 2.63082743e-01 4.78687435e-01 -7.60446489e-01
-8.55495036e-01 -1.27967760e-01 -1.02426760e-01 4.68793899e-01
-2.13201046e-01 -7.90679514e-01 -1.15242648e+00 -8.65260482e-01
5.76452076e-01 -2.81618327e-01 5.94780445e-01 2.07518101e-01
1.68646300e+00 -1.40946198e+00 -1.84213266e-01 1.19219446e+00
1.69017756e+00 -6.00744665e-01 2.43180871e-01 6.16755569e-03
4.95242327e-01 4.90358412e-01 7.23335221e-02 1.30892575e+00
4.98707771e-01 5.50851524e-01 2.70303905e-01 1.86913490e-01
5.49031973e-01 -1.26040161e-01 5.27801812e-01 7.00905859e-01
2.17942953e-01 -1.68623120e-01 -2.78468281e-01 3.00687671e-01
-2.11866403e+00 -7.64491141e-01 -7.01937154e-02 2.59997940e+00
1.34311008e+00 -7.64664114e-01 5.54399081e-02 -1.37829378e-01
4.39865023e-01 3.54561955e-01 -5.16917527e-01 -5.71953654e-01
-4.24995720e-01 2.66114604e-02 1.48197436e+00 9.14998114e-01
-9.23312604e-01 5.07753193e-01 6.41482973e+00 1.16577089e+00
-7.18040407e-01 7.55076826e-01 8.14414859e-01 -4.34237480e-01
-6.87674105e-01 -3.93193252e-02 -9.16372120e-01 4.94654536e-01
9.74228799e-01 -7.51802146e-01 1.04442132e+00 1.07844841e+00
2.52027586e-02 1.46696076e-01 -1.15685666e+00 1.33222866e+00
-2.14117631e-01 -1.41407979e+00 -2.76160061e-01 7.33507931e-01
9.30093229e-01 4.42977734e-02 3.02593261e-01 -4.18964922e-01
7.16268301e-01 -5.57348311e-01 1.77211046e-01 1.78647012e-01
4.95884985e-01 -7.93783486e-01 3.32227260e-01 4.42355067e-01
-7.37580836e-01 -2.44863808e-01 -5.41260004e-01 9.66316164e-02
-5.07572293e-02 6.00054920e-01 -4.32552606e-01 2.86833495e-01
8.49199891e-01 1.29924119e-01 -2.32041612e-01 6.99369550e-01
3.04521322e-01 6.93063974e-01 -9.08959925e-01 -1.08144365e-01
1.37174428e-01 -5.23900628e-01 7.89839089e-01 1.08584201e+00
2.44327560e-01 8.21027979e-02 -1.96392775e-01 6.07124627e-01
-3.33367944e-01 3.75779510e-01 -5.24510324e-01 3.49189460e-01
3.33200067e-01 1.07100344e+00 1.49151742e-01 -3.54220152e-01
-3.13509032e-02 1.67692268e+00 4.12410349e-01 4.64973569e-01
-1.91510499e-01 -2.49396622e-01 1.42444241e+00 -4.81441319e-02
2.44947061e-01 -2.32222959e-01 -3.58795404e-01 -1.33703899e+00
3.73071641e-01 -8.04450393e-01 8.41201484e-01 6.90377280e-02
-1.65528011e+00 -8.88672099e-02 -8.37978050e-02 -6.13357127e-01
-1.07578136e-01 -1.66925609e-01 -2.40452588e-01 9.67763782e-01
-1.19697750e+00 -1.05256343e+00 9.81671736e-02 9.44462597e-01
-2.01249182e-01 -1.09549351e-02 1.13261092e+00 3.07156146e-01
-3.64178792e-02 1.26465642e+00 1.33760130e+00 -2.47519553e-01
6.49615109e-01 -1.46558297e+00 9.71096158e-02 5.80532193e-01
-5.46231493e-02 5.32568872e-01 7.49381423e-01 -2.88399279e-01
-1.96094513e+00 -9.56353664e-01 1.03235626e+00 -2.48135075e-01
3.83816004e-01 -8.15741777e-01 -6.33818746e-01 9.27964211e-01
3.31312120e-02 3.07635933e-01 1.08077919e+00 -1.00763902e-01
-3.40586215e-01 -7.27366507e-01 -1.97287357e+00 2.95445532e-01
1.01686656e+00 -7.16456294e-01 2.83362299e-01 9.30054009e-01
8.56923044e-01 -4.67712492e-01 -9.83368754e-01 -2.41183668e-01
5.81525743e-01 -9.34894562e-01 8.58142555e-01 -7.18325615e-01
-2.40018591e-01 1.66517254e-02 -8.57074261e-01 -9.76413190e-01
1.88282743e-01 -1.58232498e+00 -4.44591433e-01 1.11061239e+00
8.02029744e-02 -1.29737258e+00 1.24633884e+00 1.37818038e+00
7.78076172e-01 -6.82426929e-01 -1.45743144e+00 -6.92639828e-01
3.35953534e-01 -1.68858826e-01 8.80866230e-01 8.39108229e-01
-2.01199397e-01 -4.30071265e-01 -7.36596107e-01 5.20736933e-01
1.44542432e+00 1.32371783e-01 1.07963669e+00 -6.24330103e-01
-8.50294232e-01 3.57021997e-03 -4.63204563e-01 -1.27616537e+00
2.54159749e-01 -1.02263010e+00 -5.68870068e-01 -8.36612642e-01
3.60920340e-01 -4.90944445e-01 -2.04045832e-01 3.37319821e-01
2.75756091e-01 -2.02510938e-01 2.85180688e-01 1.88007340e-01
-6.99051797e-01 5.06194770e-01 7.81193614e-01 1.22951968e-02
-8.50477368e-02 4.89899606e-01 -8.30111563e-01 1.41654626e-01
6.46180570e-01 -7.74088323e-01 -2.61421621e-01 -3.94438952e-01
1.53677464e-01 6.34434521e-02 4.38999534e-01 -5.96273065e-01
5.72663367e-01 -1.66795433e-01 -5.48700020e-02 -3.39817762e-01
1.95605114e-01 -1.14909303e+00 3.73025447e-01 5.39012015e-01
-7.20259368e-01 -9.55650210e-02 -2.55019367e-01 7.96478033e-01
3.68287534e-01 -7.00087398e-02 7.99776316e-01 -9.32210833e-02
8.18884149e-02 5.88152468e-01 -1.96599916e-01 6.11691922e-03
1.12806499e+00 2.05082342e-01 1.13585450e-01 -9.72614944e-01
-8.60913455e-01 3.46598506e-01 7.13414013e-01 -4.22449291e-01
2.15613499e-01 -1.13931715e+00 -8.66621435e-01 2.31304780e-01
-3.26842189e-01 -2.95030534e-01 3.29621941e-01 6.55077100e-01
-4.88886714e-01 -8.43937173e-02 4.45463002e-01 -3.99069399e-01
-1.77455783e+00 3.39663386e-01 5.32645643e-01 -3.90664995e-01
-4.62934554e-01 1.02380812e+00 4.66941595e-02 -6.05239630e-01
6.51085377e-01 7.97393695e-02 9.78151619e-01 -3.26828718e-01
7.44038284e-01 7.27061391e-01 6.26176819e-02 -1.03643313e-01
-8.48335922e-02 2.80187994e-01 -1.78866550e-01 -5.57842314e-01
1.47365439e+00 -7.55493343e-01 -6.15584671e-01 2.01826021e-01
2.16376591e+00 3.47874612e-01 -1.27776468e+00 -4.10421878e-01
-4.09380615e-01 -9.49875057e-01 1.28469035e-01 -4.66310382e-01
-1.20475781e+00 6.37877822e-01 1.02672517e+00 5.24031073e-02
1.03024149e+00 -9.14041549e-02 9.74647999e-01 4.87493306e-01
8.20279300e-01 -9.53115761e-01 -8.89129460e-01 1.26514994e-02
7.02127218e-01 -9.92909670e-01 1.13079540e-01 -3.23703349e-01
-2.57999718e-01 8.47316206e-01 -1.63900495e-01 3.73538174e-02
1.06181931e+00 5.69380939e-01 -5.45758232e-02 1.43376306e-01
-8.09152067e-01 5.30638695e-01 -3.36187035e-01 3.79928350e-01
-2.17585087e-01 3.18466157e-01 -7.21602142e-01 7.48427093e-01
-1.19267948e-01 -1.09676860e-01 2.33829319e-01 1.09789848e+00
-1.37079746e-01 -1.50316358e+00 -4.07919496e-01 3.75817984e-01
-8.90724063e-01 1.14735395e-01 -4.79030997e-01 -2.50002556e-02
-4.29972768e-01 7.27498770e-01 -2.05700576e-01 -2.23659411e-01
-3.50439072e-01 -7.70189539e-02 3.95055354e-01 1.85355954e-02
-9.84315932e-01 -2.08537385e-01 -1.78033471e-01 -7.86059618e-01
2.23860722e-02 -6.04637444e-01 -7.88348854e-01 -1.04140806e+00
-1.43377438e-01 5.28942168e-01 9.51129436e-01 3.24979872e-01
1.07652116e+00 -6.51299000e-01 1.43588173e+00 -3.86463851e-01
-1.84248078e+00 -2.60396451e-01 -1.19719231e+00 4.72389847e-01
6.30979657e-01 5.44810832e-01 -1.05341601e+00 -3.60209942e-01] | [5.949897766113281, 6.566749572753906] |
14eb6050-746f-4302-8115-01c1f62ae9ce | oxkbc-outcome-explanation-for-factorization | null | null | https://openreview.net/forum?id=nqYhFwaUj | https://openreview.net/pdf?id=nqYhFwaUj | OxKBC: Outcome Explanation for Factorization Based Knowledge Base Completion | State-of-the-art models for Knowledge Base Completion (KBC) are based on tensor factorization (TF), e.g, DistMult, ComplEx. While they produce good results, they cannot expose any rationale behind their predictions, potentially reducing the trust of a user in the model. Previous works have explored creating an inherently explainable model, e.g. Neural Theorem Proving (NTP), DeepPath, MINERVA, but explainability comes at the cost of performance. Others have tried to create an auxiliary explainable model having high fidelity with the underlying TF model, but unfortunately, they do not scale on large KBs such as FB15k and YAGO. In this work, we propose OxKBC -- an Outcome eXplanation engine for KBC, which provides a post-hoc explanation for every triple inferred by an (uninterpretable) factorization based model. It first augments the underlying Knowledge Graph by introducing weighted edges between entities based on their similarity given by the underlying model. In the augmented graph, it defines a notion of human-understandable explanation paths along with a language to generate them. Depending on the edges, the paths are aggregated into second-order templates for further selection. The best template with its grounding is then selected by a neural selection module that is trained with minimal supervision by a novel loss function. Experiments over Mechanical Turk demonstrate that users find our explanations more trustworthy compared to rule mining. | ['Mausam', 'Parag Singla', 'Mayank Singh Chauhan', 'Aman Agrawal', 'Ankesh Gupta', 'Yatin Nandwani'] | 2020-02-14 | null | null | null | akbc-2020-6 | ['knowledge-base-completion', 'knowledge-base-completion', 'automated-theorem-proving', 'automated-theorem-proving'] | ['graphs', 'knowledge-base', 'miscellaneous', 'reasoning'] | [-1.18256472e-01 1.08667099e+00 -4.53558475e-01 -2.91611552e-01
-3.33058774e-01 -6.02494836e-01 4.22030449e-01 1.17914580e-01
2.73414969e-01 8.99497390e-01 3.94815058e-01 -8.32127631e-01
-4.31311250e-01 -9.43201482e-01 -1.12791431e+00 1.70732681e-02
-2.13006243e-01 8.69432569e-01 -8.36710855e-02 -2.79871911e-01
-3.12645435e-02 -3.88177298e-02 -1.36098373e+00 8.88827384e-01
1.09745908e+00 5.67474961e-01 -2.67595142e-01 5.40460467e-01
-3.66871245e-02 8.78879070e-01 -8.55476782e-02 -1.26518965e+00
1.41467586e-01 -2.14036047e-01 -1.39240539e+00 -4.41374183e-01
2.59622306e-01 -2.29335606e-01 -1.83824390e-01 6.60804570e-01
-4.21712637e-01 1.82899646e-02 5.49403191e-01 -1.70342958e+00
-1.13075352e+00 1.37259519e+00 9.75100473e-02 -2.55736142e-01
7.19061196e-01 5.22007830e-02 1.46832883e+00 -1.15563869e+00
9.42506135e-01 1.27855849e+00 7.64670968e-01 7.29493380e-01
-1.44157481e+00 -4.97189462e-01 1.48808643e-01 6.53773785e-01
-1.19558275e+00 -6.18074127e-02 3.46632123e-01 -1.28917068e-01
1.45317996e+00 6.31494641e-01 7.70672619e-01 1.25580537e+00
9.57310572e-02 5.84393024e-01 7.84174383e-01 -5.76101005e-01
1.32255003e-01 4.48646069e-01 4.80373144e-01 1.22670472e+00
6.90863073e-01 5.81662320e-02 -8.88416409e-01 -4.26917464e-01
3.25847864e-01 -1.60336912e-01 -4.62676853e-01 -3.70486856e-01
-1.20137203e+00 9.08950508e-01 6.65684521e-01 -1.01396136e-01
-4.69725639e-01 3.39284956e-01 2.43859924e-03 3.28906089e-01
2.51617730e-01 1.02498078e+00 -8.84216130e-01 1.43877774e-01
-6.70087099e-01 5.93020499e-01 1.16903079e+00 1.00458741e+00
1.04452360e+00 -4.31027949e-01 -1.25450343e-01 1.20602567e-02
5.53865731e-01 2.66264677e-01 4.24477458e-02 -1.00367594e+00
5.81285655e-01 1.03901434e+00 9.65451896e-02 -1.31454659e+00
-2.34932408e-01 -3.39104474e-01 -5.69310486e-01 -1.18993230e-01
1.18892930e-01 8.89834985e-02 -8.79850745e-01 1.61541915e+00
3.92270029e-01 3.11416715e-01 2.69796520e-01 9.60872114e-01
8.25206637e-01 3.30089778e-01 1.90843612e-01 3.05434078e-01
1.36544740e+00 -9.87376332e-01 -6.01751089e-01 -2.22250640e-01
8.42589855e-01 -2.39949986e-01 9.70624030e-01 5.79360962e-01
-8.11025679e-01 -1.63152635e-01 -1.08048296e+00 -2.65931755e-01
-6.22466147e-01 -1.59994900e-01 1.36622500e+00 5.43226421e-01
-9.24327612e-01 8.16536009e-01 -7.31957197e-01 -2.37602130e-01
3.14980865e-01 4.34049964e-01 -7.55288064e-01 -3.22223634e-01
-1.54841399e+00 1.14157641e+00 6.85375929e-01 4.63951714e-02
-8.19053054e-01 -9.83404458e-01 -1.02737868e+00 3.33575547e-01
7.05309272e-01 -1.37281692e+00 9.62671161e-01 -5.82383394e-01
-9.60045636e-01 2.33648956e-01 -2.23110586e-01 -8.14479887e-01
8.03790316e-02 -2.67738104e-01 -5.14210880e-01 -1.01969600e-01
3.02681863e-01 8.19614887e-01 6.47854209e-01 -1.37569857e+00
-4.04522419e-01 -1.74554691e-01 7.23128438e-01 -1.64877698e-01
-6.79113194e-02 -4.64410961e-01 -3.90172482e-01 -3.17262620e-01
1.15848139e-01 -1.08387923e+00 -3.97215098e-01 -3.99224728e-01
-9.65709031e-01 -2.36433074e-01 5.97184300e-01 -7.84539819e-01
1.38547933e+00 -1.54460371e+00 3.72927248e-01 6.50625467e-01
6.26245141e-01 -1.42221451e-02 -1.31125227e-01 4.63862091e-01
-1.66913137e-01 8.11081707e-01 1.65770188e-01 -4.99644011e-01
2.23532483e-01 4.96606588e-01 -4.55059141e-01 -5.96726090e-02
3.46467972e-01 1.06969452e+00 -9.14686143e-01 -3.94590944e-01
-1.93104580e-01 3.37624282e-01 -1.10793912e+00 -1.46039978e-01
-7.14489043e-01 -4.21837978e-02 -4.80730683e-01 5.16563177e-01
4.92083400e-01 -5.69467545e-01 4.27000761e-01 -3.70348722e-01
4.02666479e-01 5.11593759e-01 -1.15106642e+00 1.61176705e+00
-3.75499487e-01 1.89912200e-01 -5.40804625e-01 -5.88358045e-01
5.84259331e-01 4.54274923e-01 -1.65008485e-01 2.69181319e-02
-3.68543208e-01 2.51675904e-01 -2.55350154e-02 -4.24970806e-01
7.91798472e-01 -1.45067036e-01 1.68127492e-01 6.28335714e-01
2.78307855e-01 1.25431642e-01 1.82951182e-01 9.26746726e-01
1.28074718e+00 2.64278948e-01 3.40789407e-01 1.01422081e-02
3.58354777e-01 4.21290696e-01 4.45605934e-01 8.67923141e-01
4.90230858e-01 4.95529085e-01 6.22129261e-01 -8.85788679e-01
-8.45219493e-01 -7.31350720e-01 3.46661776e-01 5.65270007e-01
-2.63747051e-02 -1.25559449e+00 -6.70767903e-01 -1.28283811e+00
3.14028591e-01 1.14805317e+00 -9.25955653e-01 -2.87323684e-01
-1.02603287e-01 -3.00084036e-02 7.32146919e-01 4.03742641e-01
5.90883642e-02 -1.03247213e+00 -1.81685254e-01 1.22602269e-01
-6.41369879e-01 -1.04762793e+00 -1.27822921e-01 1.02317743e-01
-7.73867607e-01 -1.32642078e+00 3.99515361e-01 -2.70590242e-02
9.58435476e-01 8.57141763e-02 1.37050688e+00 7.21421421e-01
6.26369268e-02 3.39879602e-01 -5.67981660e-01 -5.02002656e-01
-3.83186251e-01 2.36073703e-01 2.29405358e-01 -1.06643505e-01
5.06584942e-01 -4.97334898e-01 -3.68806452e-01 1.66094393e-01
-8.25937986e-01 5.39585531e-01 3.93157333e-01 6.99919403e-01
5.12432158e-01 2.71082431e-01 2.67736167e-01 -1.26212788e+00
6.00038826e-01 -6.49904549e-01 -1.49811313e-01 5.39988816e-01
-1.16963637e+00 5.48617423e-01 6.75826192e-01 -1.81702390e-01
-9.86530721e-01 -3.25616039e-02 4.18847740e-01 -5.12877345e-01
2.14961335e-01 1.12096739e+00 -1.19232193e-01 1.54148921e-01
1.10539520e+00 -2.43211836e-01 -1.70160741e-01 -3.03796411e-01
8.75461102e-01 3.87721471e-02 4.38040644e-01 -7.73128271e-01
1.26995051e+00 2.62378156e-01 3.24775204e-02 1.17101140e-01
-1.24461555e+00 -3.72547507e-02 -4.26855236e-01 9.29140523e-02
6.96555853e-01 -6.50922179e-01 -1.01270783e+00 -6.12459481e-01
-1.46646678e+00 -7.71559700e-02 -1.35515168e-01 5.36511064e-01
-1.73340365e-01 1.63730577e-01 -3.28020781e-01 -7.37663448e-01
-4.81693238e-01 -8.60742450e-01 7.13711679e-01 1.68677121e-02
-7.49845982e-01 -9.31857586e-01 -1.08946182e-01 8.15341294e-01
2.75415212e-01 2.37347499e-01 1.42292631e+00 -6.94445252e-01
-1.01857257e+00 -3.02271694e-01 -7.22474754e-02 7.23792091e-02
-2.67215908e-01 2.71562301e-02 -7.97720492e-01 1.93837926e-01
-7.97197044e-01 -2.70546794e-01 7.80728519e-01 -1.03009023e-01
1.10924160e+00 -8.40010166e-01 -5.93725085e-01 4.64026302e-01
1.15192568e+00 -4.39693004e-01 7.37557888e-01 5.07909596e-01
8.05693567e-01 3.40618312e-01 4.98934060e-01 4.91399840e-02
7.93656349e-01 4.07540560e-01 8.19227755e-01 2.42800098e-02
1.79862589e-01 -8.05611730e-01 2.41106406e-01 9.67089683e-02
-6.93541467e-01 -1.97060481e-01 -8.55643988e-01 3.33987474e-01
-2.21056056e+00 -1.01282418e+00 -5.03855944e-01 1.90192771e+00
7.27250993e-01 2.25225300e-01 -3.71332169e-01 1.60335541e-01
2.10126847e-01 -4.58843976e-01 -2.91295022e-01 -5.90988040e-01
-1.02453738e-01 3.80674928e-01 1.93452865e-01 7.79555678e-01
-6.06918931e-01 1.06437361e+00 5.53714323e+00 4.07338619e-01
-5.99358916e-01 -9.83868912e-02 4.01992708e-01 3.07104085e-02
-1.04894245e+00 7.85030544e-01 -7.13886499e-01 -1.22968845e-01
7.85532415e-01 -1.64378211e-01 8.34974229e-01 9.99897242e-01
-7.31968507e-02 1.38907894e-01 -1.42095351e+00 3.88327420e-01
-7.10561872e-02 -1.85361445e+00 5.04403651e-01 2.46372566e-01
5.68306565e-01 -4.18548256e-01 -1.49422541e-01 8.50440741e-01
5.41187167e-01 -1.37378740e+00 8.13264489e-01 8.00107062e-01
5.68149269e-01 -7.61791766e-01 9.08633053e-01 3.57429653e-01
-8.32359672e-01 -9.95545313e-02 -4.07476634e-01 -2.58519709e-01
-8.60222951e-02 6.63864553e-01 -1.38178301e+00 1.05818045e+00
5.48280239e-01 3.90620917e-01 -7.86227465e-01 3.52272213e-01
-8.46385539e-01 8.34018230e-01 -1.73543155e-01 3.31380032e-02
1.39267862e-01 1.44750597e-02 4.96862352e-01 9.43108618e-01
4.03585374e-01 3.11517686e-01 2.44999584e-03 1.36417472e+00
-3.20004016e-01 -1.35047361e-01 -7.09129989e-01 -1.09299988e-01
4.62336004e-01 1.61430001e+00 -3.30666304e-01 -5.45352817e-01
-1.76936850e-01 9.08900142e-01 7.30044901e-01 4.10665691e-01
-9.71057475e-01 7.73096159e-02 5.15161812e-01 2.99548805e-01
1.09409593e-01 1.22125156e-01 -3.74470562e-01 -1.30384612e+00
1.90563485e-01 -1.07200313e+00 6.04919016e-01 -1.10696399e+00
-1.09077346e+00 6.78714454e-01 5.24442084e-02 -6.70496345e-01
-4.34788704e-01 -5.29896200e-01 -5.89507461e-01 1.03533208e+00
-1.28214312e+00 -1.56457686e+00 -2.25219831e-01 6.10537350e-01
8.96258093e-03 -8.01396966e-02 1.21592963e+00 -1.99711040e-01
-3.53161752e-01 5.50304711e-01 -8.64985228e-01 -1.83344498e-01
3.54390293e-01 -1.51135731e+00 5.49640179e-01 8.79521370e-01
4.89313543e-01 1.52006185e+00 9.69775140e-01 -1.01386738e+00
-1.68044400e+00 -1.26559806e+00 1.40546548e+00 -9.98957396e-01
6.69820368e-01 -2.04999372e-01 -1.04051435e+00 1.09987736e+00
1.09838329e-01 1.12434916e-01 7.24434078e-01 9.39374208e-01
-8.64471555e-01 1.89210698e-01 -1.07469678e+00 7.53053248e-01
1.22971582e+00 -5.68868756e-01 -5.74958861e-01 3.66944879e-01
1.05690050e+00 -3.19184184e-01 -8.62747550e-01 2.78658122e-01
6.49296463e-01 -8.82098138e-01 7.88205624e-01 -1.35144186e+00
9.13928807e-01 -5.90670705e-01 2.64953468e-02 -1.41260481e+00
-7.58179665e-01 -6.77516162e-01 -6.78866148e-01 1.05548894e+00
1.32040238e+00 -3.89245808e-01 1.07072854e+00 1.15316749e+00
-1.43473402e-01 -8.92717838e-01 -5.75682580e-01 -5.62718630e-01
-4.69255298e-01 -8.35966885e-01 1.18686461e+00 1.05377245e+00
6.21243179e-01 6.33213639e-01 -3.60170692e-01 6.11441731e-01
6.24030352e-01 1.06828816e-01 8.97955835e-01 -1.36172652e+00
-4.98949796e-01 1.43017337e-01 -1.28647685e-01 -2.94792652e-01
3.93962979e-01 -1.37618542e+00 -2.61479229e-01 -1.84998262e+00
3.82073283e-01 -4.99734223e-01 -6.71230853e-02 1.16268754e+00
-3.69550437e-01 2.73479316e-02 1.21861100e-01 7.46231750e-02
-6.90643787e-01 2.57651955e-01 1.07756257e+00 -9.27794948e-02
-1.30410120e-01 -3.15321624e-01 -1.21114683e+00 6.73933566e-01
6.07397318e-01 -5.54026365e-01 -5.97244084e-01 -3.98667246e-01
8.32959116e-01 9.90421996e-02 7.90065587e-01 -5.91319203e-01
2.92789549e-01 -1.31366357e-01 3.19421381e-01 -2.22730368e-01
1.35698676e-01 -7.67933846e-01 7.07525134e-01 2.40485236e-01
-3.80803406e-01 8.36445987e-02 4.59595360e-02 7.31959045e-01
6.28269417e-03 -1.77114278e-01 -1.37754709e-01 -1.11650921e-01
-4.19527829e-01 3.41633499e-01 -7.10907504e-02 -4.32404280e-01
4.73093241e-01 -6.72952086e-02 -4.77808535e-01 -5.72024167e-01
-8.31618726e-01 2.87824303e-01 1.99959114e-01 3.69096935e-01
7.56972432e-01 -1.45038319e+00 -6.62469089e-01 -3.99058573e-02
3.91028672e-01 -6.51530027e-02 3.69042158e-02 7.86738873e-01
-9.20325592e-02 5.88309467e-01 1.70498207e-01 -6.16160296e-02
-1.21632242e+00 7.61224627e-01 9.85280126e-02 -6.92520380e-01
-5.83919525e-01 6.77132249e-01 -9.35909748e-02 -6.76459014e-01
-9.03705657e-02 -5.36104739e-01 -7.50436680e-04 -5.16887665e-01
4.50577080e-01 9.69675109e-02 1.98614568e-01 -6.57481477e-02
-4.63516027e-01 -1.52117640e-01 -2.01128259e-01 -1.93281919e-02
1.38700020e+00 2.84713537e-01 -2.49845177e-01 -2.33772904e-01
5.92202842e-01 -1.91373248e-02 -5.05031347e-01 -7.30564669e-02
1.71488106e-01 -4.37836826e-01 -2.07848370e-01 -1.18902481e+00
-7.95321286e-01 2.98468947e-01 -3.00041616e-01 3.00358832e-01
6.86840236e-01 9.62067917e-02 4.81765062e-01 7.77702153e-01
5.31217158e-01 -5.83082974e-01 -1.01278014e-01 5.46952963e-01
1.14994776e+00 -1.07708538e+00 5.83184287e-02 -7.43076265e-01
-7.87424803e-01 1.15816498e+00 6.57997131e-01 2.24162564e-01
4.76628780e-01 -2.56446097e-02 -2.79762238e-01 -5.93095958e-01
-1.25044012e+00 3.47586125e-02 6.11983597e-01 7.45656252e-01
4.32022005e-01 3.86258543e-01 -1.03064984e-01 1.24330544e+00
-6.51153564e-01 1.39339894e-01 5.75399816e-01 2.15089291e-01
-6.67489134e-03 -1.16457546e+00 -3.20726871e-01 4.43845987e-01
-2.38911331e-01 -4.56778288e-01 -7.22199738e-01 9.73428726e-01
4.51291233e-01 1.08610511e+00 -6.67604625e-01 -8.33712816e-01
3.11862290e-01 2.06451878e-01 3.11509162e-01 -6.69804156e-01
-6.46950543e-01 -7.64683545e-01 8.27595294e-01 -7.46545792e-01
-2.72005666e-02 -3.66868734e-01 -1.52127671e+00 -6.85106635e-01
-4.76880759e-01 5.48736393e-01 4.84797686e-01 1.13208878e+00
5.57309926e-01 2.34794155e-01 1.08610325e-01 -3.46393406e-01
-2.31684580e-01 -7.87280619e-01 -3.12124938e-01 5.89239359e-01
-1.12769052e-01 -5.93885481e-01 -3.24818999e-01 1.70060188e-01] | [8.891743659973145, 7.58951473236084] |
8be70e97-09d2-4453-8a05-19d07154d7c5 | a-factored-generalized-additive-model-for | 1907.12596 | null | https://arxiv.org/abs/1907.12596v1 | https://arxiv.org/pdf/1907.12596v1.pdf | A Factored Generalized Additive Model for Clinical Decision Support in the Operating Room | Logistic regression (LR) is widely used in clinical prediction because it is simple to deploy and easy to interpret. Nevertheless, being a linear model, LR has limited expressive capability and often has unsatisfactory performance. Generalized additive models (GAMs) extend the linear model with transformations of input features, though feature interaction is not allowed for all GAM variants. In this paper, we propose a factored generalized additive model (F-GAM) to preserve the model interpretability for targeted features while allowing a rich model for interaction with features fixed within the individual. We evaluate F-GAM on prediction of two targets, postoperative acute kidney injury and acute respiratory failure, from a single-center database. We find superior model performance of F-GAM in terms of AUPRC and AUROC compared to several other GAM implementations, random forests, support vector machine, and a deep neural network. We find that the model interpretability is good with results with high face validity. | ['Bradley A Fritz', 'Zhicheng Cui', 'Michael S Avidan', 'Yixin Chen', 'Christopher R King'] | 2019-07-29 | null | null | null | null | ['respiratory-failure'] | ['medical'] | [ 1.09948970e-01 2.59703845e-01 -6.27348900e-01 -6.60424054e-01
-4.52468187e-01 -2.29801342e-01 2.33052552e-01 1.04895927e-01
-2.78948158e-01 8.70530307e-01 1.93251684e-01 -8.81231606e-01
-5.72281122e-01 -7.64599562e-01 -4.21440393e-01 -6.21464312e-01
-2.84363806e-01 5.00808001e-01 -2.99181223e-01 -8.65827799e-02
-1.13741763e-01 4.91678268e-01 -9.50146496e-01 4.31490690e-01
1.14133179e+00 9.06359255e-01 -2.22457603e-01 4.58721250e-01
2.50962526e-01 1.21655023e+00 -1.17266834e-01 -4.44911718e-01
1.98845699e-01 -4.47787046e-01 -3.48507315e-01 -4.51396078e-01
6.87374547e-02 -4.91427571e-01 -1.07551806e-01 1.84682086e-01
5.49115181e-01 -1.53085500e-01 1.00567853e+00 -1.51774263e+00
-8.53317022e-01 4.29408997e-01 -3.49710315e-01 -1.60062924e-01
2.90583283e-01 1.26988113e-01 9.87106264e-01 -9.76585686e-01
2.24647716e-01 1.26642990e+00 1.27936387e+00 5.48330545e-01
-1.49768484e+00 -7.82716930e-01 1.05938740e-01 -1.10678695e-01
-1.14689827e+00 5.71095198e-02 2.44066253e-01 -5.96258402e-01
1.07711732e+00 6.08619511e-01 6.04002953e-01 7.29948580e-01
8.83080184e-01 5.14180779e-01 1.29173279e+00 -5.60531914e-01
3.28489691e-02 4.13057089e-01 5.76616585e-01 7.47717321e-01
1.72831699e-01 4.30954903e-01 -4.58314896e-01 -8.33082855e-01
7.95325458e-01 5.73132277e-01 -1.51933478e-02 -2.67453194e-01
-9.55029190e-01 1.18755972e+00 5.16868830e-01 -1.35795221e-01
-6.02872074e-01 -2.52952371e-02 1.04312100e-01 3.89008850e-01
5.13149321e-01 3.41990709e-01 -7.04628885e-01 1.65227339e-01
-6.59413457e-01 2.76640445e-01 8.87621224e-01 5.76731682e-01
3.20204347e-01 -5.09348921e-02 -4.27015305e-01 1.12710333e+00
2.08169177e-01 3.13366383e-01 5.69150746e-01 -8.26507270e-01
-2.30166614e-02 7.52738476e-01 -2.03924432e-01 -8.38975847e-01
-1.00295484e+00 -7.07567990e-01 -1.29114354e+00 4.21687514e-01
1.95549980e-01 -6.15537427e-02 -9.15612996e-01 1.79836714e+00
4.18929309e-02 -2.33077645e-01 5.32206111e-02 4.47334886e-01
1.01507282e+00 1.13916576e-01 5.35743356e-01 -2.63155818e-01
1.32551324e+00 -1.04797029e+00 -5.72539151e-01 -2.99981415e-01
8.89320314e-01 -2.26200178e-01 1.06230915e+00 5.12004912e-01
-6.83525860e-01 -1.92406371e-01 -5.84171057e-01 -9.52784196e-02
-7.02945217e-02 1.28659412e-01 9.25175846e-01 9.62841034e-01
-9.59666073e-01 5.63775957e-01 -8.73505771e-01 -5.76012842e-02
4.67441320e-01 7.73574412e-01 -7.20456421e-01 -2.74266213e-01
-1.12292624e+00 1.21187961e+00 -5.25178900e-03 -1.44647107e-01
-2.60668486e-01 -1.03164721e+00 -6.73880160e-01 2.79908832e-02
-2.10340302e-02 -1.36107528e+00 1.09507358e+00 -9.43488359e-01
-1.26224136e+00 4.69196558e-01 -2.66141623e-01 -6.02085173e-01
6.97371960e-01 -2.13485599e-01 -2.07645353e-02 -3.70255709e-01
-1.33445516e-01 5.57860672e-01 4.05532479e-01 -7.25997210e-01
-3.97582203e-01 -5.05314052e-01 -2.04441607e-01 1.84286788e-01
-7.78718293e-02 2.30500653e-01 3.84388536e-01 -6.41775906e-01
-1.30346662e-03 -1.12514746e+00 -7.46429384e-01 1.23366974e-01
-2.48030484e-01 1.94407254e-02 3.23798686e-01 -9.29921329e-01
1.54030478e+00 -1.96234143e+00 -3.28524977e-01 1.61897808e-01
4.51094002e-01 -1.33222327e-01 2.17027113e-01 2.29844451e-01
-4.99886274e-01 4.50801015e-01 -3.99408579e-01 -2.16643259e-01
-2.91145056e-01 3.02084535e-01 5.43926954e-02 2.08376646e-01
3.36123228e-01 1.12754214e+00 -6.04281008e-01 -3.95913124e-01
1.00249954e-01 4.91089255e-01 -1.02295423e+00 6.33237138e-02
4.86197144e-01 2.92920113e-01 -3.04768980e-01 6.71840966e-01
5.48901260e-01 -4.32600409e-01 2.04631090e-01 3.39566946e-01
1.97681472e-01 1.06787682e-01 -6.54969275e-01 7.25445986e-01
-4.69141811e-01 1.04594946e-01 -3.68489891e-01 -6.93882644e-01
1.02774322e+00 2.48137340e-01 7.02508152e-01 -3.31093878e-01
-1.51182443e-01 2.64100909e-01 4.24845219e-01 -2.55067647e-01
-8.26857015e-02 -6.84938610e-01 8.04754421e-02 4.33403105e-01
-1.14638686e-01 9.80439186e-02 -4.68343705e-01 -9.79872346e-02
1.12427342e+00 -1.28687739e-01 1.20805931e+00 -3.04185212e-01
6.39869869e-02 2.34751049e-02 6.83120191e-01 1.02409375e+00
5.53291589e-02 6.37394845e-01 5.39286315e-01 -8.27861786e-01
-5.78174531e-01 -1.01103294e+00 -5.49943686e-01 1.10427427e+00
-5.77625513e-01 -2.95794964e-01 -2.46339753e-01 -6.47666991e-01
3.64192814e-01 9.92642879e-01 -1.00759971e+00 -4.49780345e-01
-3.38181585e-01 -1.45480692e+00 5.57685792e-01 7.43539631e-01
8.63754451e-02 -8.85087907e-01 -4.81789440e-01 2.62881994e-01
6.96475580e-02 -5.34692228e-01 -2.26249591e-01 6.89857721e-01
-1.07222235e+00 -1.08455360e+00 -4.87761468e-01 -3.95614117e-01
5.82300484e-01 -5.50930910e-02 1.22889543e+00 -6.39971867e-02
-2.71437645e-01 6.58895588e-03 -1.78869799e-01 -9.45011437e-01
-3.99664193e-01 -5.37704937e-02 9.23266187e-02 -2.77624488e-01
4.58835959e-01 -3.16090375e-01 -6.78452075e-01 4.55638319e-01
-6.21137142e-01 2.68825144e-01 8.01543534e-01 1.37852323e+00
4.68862206e-01 -3.50989819e-01 6.31831229e-01 -1.42106330e+00
6.54820025e-01 -7.89199531e-01 2.32317641e-01 1.90010201e-02
-1.34271717e+00 -2.62458801e-01 4.18651909e-01 -5.44935644e-01
-7.73371279e-01 1.34667873e-01 -4.48508859e-02 -2.07071379e-01
9.93191078e-02 8.11834753e-01 2.46614113e-01 -5.39006479e-02
1.09010267e+00 -6.97879866e-02 4.09966379e-01 -5.01330137e-01
1.87942777e-02 6.91903412e-01 8.83325711e-02 -2.39202321e-01
3.09871256e-01 6.27918020e-02 1.75470233e-01 -3.33378911e-01
-5.46505094e-01 -1.62368894e-01 -4.64436293e-01 2.22641096e-01
5.07922232e-01 -9.42742407e-01 -4.41749811e-01 2.57865280e-01
-5.99398434e-01 -4.40269738e-01 -1.72168911e-01 6.55555010e-01
-6.41293049e-01 -8.68523940e-02 -5.88867307e-01 -8.85515153e-01
-6.56363785e-01 -1.11308289e+00 7.78155029e-01 -1.94273829e-01
-7.39940464e-01 -1.09490693e+00 -9.07597914e-02 1.61784336e-01
6.14005446e-01 4.88531142e-01 1.45222771e+00 -1.08465421e+00
9.67544224e-03 -4.69342113e-01 -2.03944266e-01 4.25201327e-01
4.34446812e-01 8.36250260e-02 -8.00949097e-01 -4.32757922e-02
-8.23533610e-02 6.06112415e-03 8.63071918e-01 9.21763420e-01
1.30897772e+00 -7.61700809e-01 -4.29399759e-01 6.82736099e-01
1.13172865e+00 5.67707479e-01 6.31517708e-01 4.16743785e-01
4.12886888e-01 5.66494405e-01 5.35220802e-01 3.14092308e-01
4.96205330e-01 5.48321247e-01 2.22332552e-01 -5.25461257e-01
3.67358297e-01 -1.57737553e-01 1.67381123e-01 9.03360844e-02
-4.33317661e-01 -9.45492461e-03 -1.15251553e+00 2.68740263e-02
-1.90405858e+00 -7.64442742e-01 -3.46450537e-01 2.23283052e+00
6.78489208e-01 1.87479146e-02 3.41410726e-01 7.13533983e-02
3.23051870e-01 -5.38081229e-01 -4.52038646e-01 -9.21376228e-01
-8.87753963e-02 2.25255802e-01 5.63247979e-01 2.93326259e-01
-1.02798641e+00 3.77833217e-01 7.99050522e+00 5.70830941e-01
-1.00179243e+00 2.18301788e-01 1.36904943e+00 -2.41873220e-01
-1.07050605e-01 -1.55688241e-01 -5.20801008e-01 3.32822293e-01
1.00442517e+00 -2.94110626e-01 2.96076506e-01 1.15415227e+00
3.89490694e-01 1.43904701e-01 -1.24058163e+00 6.48087204e-01
-2.23810747e-01 -1.02610922e+00 -1.43252671e-01 1.93516552e-01
4.43031073e-01 -1.14791386e-01 3.92139047e-01 5.95071554e-01
5.93570292e-01 -1.69155788e+00 3.88819218e-01 5.89495122e-01
1.06582725e+00 -5.12632787e-01 1.03495419e+00 5.28575361e-01
-3.71410757e-01 -4.02674347e-01 -1.50960505e-01 -3.11169833e-01
-1.97524503e-01 5.14041662e-01 -1.29147267e+00 3.52713466e-01
6.89890802e-01 4.13599133e-01 -5.04575908e-01 9.61718857e-01
1.42122298e-01 7.17539012e-01 -2.99078047e-01 8.74316394e-02
1.44814625e-02 7.95645081e-03 1.64780155e-01 1.28995812e+00
4.17766243e-01 9.60653350e-02 3.64601947e-02 5.05118191e-01
3.80169660e-01 6.14985466e-01 -7.44791746e-01 3.84497017e-01
5.83911575e-02 1.00518763e+00 -1.68769792e-01 -1.18582770e-02
-6.69342995e-01 4.37438279e-01 1.36457220e-01 1.84074223e-01
-7.16540337e-01 1.55454263e-01 5.00154078e-01 4.40196186e-01
-2.93833524e-01 3.36607069e-01 -8.96165609e-01 -7.22359955e-01
-2.18529701e-01 -1.03414726e+00 7.74215996e-01 -7.49847710e-01
-1.46916294e+00 7.44797528e-01 2.94098612e-02 -1.53867447e+00
-5.77116191e-01 -8.13944519e-01 -3.35749835e-01 1.15818441e+00
-1.02493763e+00 -1.47732103e+00 -3.10603172e-01 4.01583523e-01
1.01640858e-01 -2.25754797e-01 1.35307562e+00 -4.31153774e-02
-4.45458740e-01 9.37973678e-01 1.19287886e-01 -5.95482849e-02
8.09100807e-01 -1.26246178e+00 3.94357741e-02 1.02007233e-01
-5.16687691e-01 1.06035030e+00 5.76450884e-01 -6.07957780e-01
-8.84512901e-01 -1.19643199e+00 9.70250189e-01 -4.70491886e-01
3.95521134e-01 1.08965538e-01 -7.83234298e-01 9.86407936e-01
-2.38030866e-01 7.87498578e-02 1.47234595e+00 5.39304197e-01
-4.28770572e-01 -2.50415951e-02 -1.39799321e+00 4.22690004e-01
7.09717274e-01 -1.80802032e-01 -3.97729307e-01 3.43908787e-01
6.68915331e-01 -3.42911392e-01 -1.21582210e+00 7.65923321e-01
1.11880302e+00 -9.80456650e-01 1.07202423e+00 -1.15951145e+00
6.69098437e-01 2.30779126e-01 -1.84689954e-01 -1.45166814e+00
-9.30844486e-01 -2.58792609e-01 5.27874269e-02 6.38770640e-01
9.30606544e-01 -1.23717868e+00 5.65191090e-01 1.35490417e+00
3.53986062e-02 -1.31205988e+00 -8.52708340e-01 -6.26616538e-01
4.11016703e-01 -3.50364685e-01 6.41447842e-01 9.41755295e-01
2.17483565e-01 2.98770338e-01 -6.41741216e-01 -2.86251068e-01
2.40818739e-01 -1.68004259e-01 6.56890333e-01 -1.47264707e+00
-6.88782215e-01 -4.47500467e-01 -4.39984947e-01 -1.25165805e-01
-1.00024641e-01 -9.95937884e-01 -4.69931602e-01 -1.45114803e+00
6.67630255e-01 -7.46696174e-01 -8.36449862e-01 1.06554270e+00
-6.10904753e-01 1.64843082e-01 2.97690406e-02 2.40718022e-01
9.61432457e-02 1.30278274e-01 1.01005220e+00 6.17379807e-02
-5.48290670e-01 5.56334674e-01 -1.22581923e+00 7.19923913e-01
9.17754889e-01 -6.75339043e-01 -3.07951659e-01 4.26865136e-03
-1.06785493e-02 3.63395303e-01 3.21573853e-01 -2.88030267e-01
-1.40897527e-01 -4.39349920e-01 6.46258652e-01 -1.81066036e-01
1.60965547e-01 -7.39880621e-01 7.46860206e-01 7.06667483e-01
-4.86692876e-01 2.54352838e-01 1.78514779e-01 3.62406492e-01
-7.16450214e-02 -1.80677287e-02 6.98778391e-01 -3.65246385e-02
-6.77953959e-02 4.01147991e-01 -4.29644793e-01 -4.54298824e-01
9.40138340e-01 -2.76628464e-01 -3.61514002e-01 -4.85199541e-01
-9.11330223e-01 -1.49008846e-02 3.15917134e-01 2.99041390e-01
5.82597136e-01 -1.19255412e+00 -9.04623508e-01 2.87295192e-01
2.94003427e-01 -2.24429920e-01 2.28152752e-01 1.36427402e+00
-5.85632741e-01 4.97157902e-01 -1.87448427e-01 -3.73844236e-01
-1.51717913e+00 6.41418874e-01 4.94526714e-01 -6.94688320e-01
-5.41553974e-01 5.72632134e-01 6.57753110e-01 -6.14959061e-01
-1.10011436e-01 -3.21602821e-01 -2.37699077e-01 -2.06428722e-01
5.92383623e-01 4.53536004e-01 1.33394182e-01 -2.56402284e-01
-5.83064973e-01 1.58894479e-01 -2.88910419e-02 3.59095424e-01
1.48102129e+00 2.93555796e-01 -3.42078567e-01 4.19433564e-01
9.08073127e-01 -2.42482260e-01 -7.31136501e-01 -1.00214027e-01
-1.06213383e-01 -2.45538279e-01 4.23245877e-02 -1.20449388e+00
-7.03820884e-01 6.34243488e-01 6.05744839e-01 -6.46920279e-02
1.37178767e+00 -3.87894988e-01 -6.72912076e-02 2.11876318e-01
3.23185414e-01 -4.16601688e-01 -4.89515096e-01 1.13463521e-01
1.00700366e+00 -1.28261864e+00 1.56984136e-01 -6.73763096e-01
-1.06318665e+00 1.05390060e+00 4.53216672e-01 -6.50267303e-02
9.56974685e-01 4.34914440e-01 1.55984908e-01 2.43698359e-01
-1.12010419e+00 3.53611797e-01 4.97365385e-01 6.49174452e-01
6.65751040e-01 5.88683426e-01 -5.95940709e-01 1.43774092e+00
-3.86390656e-01 4.25123543e-01 1.81900144e-01 6.34251356e-01
-1.02630042e-01 -1.06256533e+00 -2.91539937e-01 1.37556565e+00
-7.61220753e-01 -4.19007093e-01 -3.69941354e-01 1.00598145e+00
-1.16150863e-01 7.31545985e-01 -1.19189061e-01 -5.94531059e-01
3.47486258e-01 3.69168758e-01 5.27704991e-02 -5.77356398e-01
-9.16261256e-01 -2.49909069e-02 2.55688936e-01 -4.81408477e-01
3.00026312e-02 -8.14414740e-01 -8.61920357e-01 -3.41194540e-01
-2.23748803e-01 -5.04325442e-02 3.76133442e-01 6.61975563e-01
4.37385350e-01 5.31027198e-01 6.29501462e-01 -1.91784099e-01
-7.09398746e-01 -1.07323194e+00 -6.90991700e-01 -6.20287508e-02
4.64970112e-01 -7.93887794e-01 -3.96971911e-01 -1.70839146e-01] | [8.160487174987793, 5.662962436676025] |
a0016147-a235-4d3e-b71f-87990060f4fb | smurff-a-high-performance-framework-for | 1904.02514 | null | https://arxiv.org/abs/1904.02514v3 | https://arxiv.org/pdf/1904.02514v3.pdf | SMURFF: a High-Performance Framework for Matrix Factorization | Bayesian Matrix Factorization (BMF) is a powerful technique for recommender systems because it produces good results and is relatively robust against overfitting. Yet BMF is more computationally intensive and thus more challenging to implement for large datasets. In this work we present SMURFF a high-performance feature-rich framework to compose and construct different Bayesian matrix-factorization methods. The framework has been successfully used in to do large scale runs of compound-activity prediction. SMURFF is available as open-source and can be used both on a supercomputer and on a desktop or laptop machine. Documentation and several examples are provided as Jupyter notebooks using SMURFF's high-level Python API. | ['Jörg Wegner', 'José Felipe Golib Dzib', 'Thomas J. Ashby', 'Wilfried Verachtert', 'Vladimir Chupakhin', 'Imen Chakroun', 'Hugo Ceulemans', 'Adam Arany', 'Tom Vander Aa', 'Thanh Le Van', 'Roel Wuyts', 'Yves Moreau', 'Jaak Simm'] | 2019-04-04 | null | null | null | null | ['activity-prediction', 'activity-prediction'] | ['computer-vision', 'time-series'] | [-2.85384536e-01 -4.76040989e-01 -1.73743501e-01 -2.95941293e-01
-5.12576878e-01 -5.52510977e-01 3.50020140e-01 5.69398254e-02
-2.19016254e-01 8.94288898e-01 1.55217603e-01 -7.23660052e-01
-3.81924927e-01 -8.21450412e-01 -4.31832880e-01 -6.04892731e-01
1.54501072e-03 7.67644048e-01 3.97968769e-01 -2.07972467e-01
1.65393636e-01 4.30689245e-01 -1.75014067e+00 4.49075341e-01
6.35971487e-01 4.22752023e-01 4.65898633e-01 7.79726386e-01
5.01645841e-02 7.59832799e-01 -2.70360827e-01 -4.43598062e-01
2.65079051e-01 -1.29130781e-01 -6.48336589e-01 -4.54415172e-01
1.90502882e-01 -1.14420159e-02 -4.10963707e-02 6.32280171e-01
6.34333372e-01 3.88638854e-01 5.08369625e-01 -1.01857245e+00
-2.55359352e-01 4.40767080e-01 -3.27760786e-01 3.06860089e-01
6.99256420e-01 -2.15878755e-01 9.79402006e-01 -1.08893430e+00
3.77874851e-01 1.09912980e+00 8.09746563e-01 1.40314683e-01
-1.46387279e+00 -6.46409333e-01 -4.53461148e-02 1.37404621e-01
-1.27365732e+00 -3.17389935e-01 1.15090698e-01 -6.54315174e-01
1.12978661e+00 6.30768657e-01 6.72329366e-01 1.12937200e+00
5.58201075e-02 7.61786461e-01 1.22149324e+00 -2.27738470e-01
2.72725254e-01 2.02985242e-01 4.34939951e-01 4.49064076e-01
2.81104296e-01 6.63283393e-02 -6.03701890e-01 -8.71142805e-01
7.37904251e-01 2.73049295e-01 -5.44080846e-02 -2.20369250e-02
-1.11248815e+00 1.10693419e+00 -4.18530926e-02 3.66714299e-01
-4.57299411e-01 2.28688747e-01 9.18301567e-02 2.84298211e-01
5.18119216e-01 2.20859468e-01 -7.81228602e-01 -5.63036263e-01
-8.33561063e-01 7.39306688e-01 9.53707397e-01 4.95788842e-01
5.58170080e-01 -1.57975838e-01 -9.68059078e-02 1.12756860e+00
4.95584875e-01 4.29392874e-01 3.61854106e-01 -8.70315254e-01
-2.08544657e-01 1.82434529e-01 4.32134897e-01 -8.83151650e-01
-6.88261271e-01 -3.10408533e-01 -5.62301397e-01 4.79243547e-02
4.51834619e-01 -6.68205395e-02 -3.61940563e-01 1.28183544e+00
6.50109887e-01 4.73690867e-01 -3.30454201e-01 8.12013388e-01
1.08995640e+00 7.77443171e-01 1.48971364e-01 -3.28708917e-01
1.48171806e+00 -8.82535577e-01 -4.99427527e-01 3.51642370e-02
5.69016278e-01 -1.08188426e+00 8.09280634e-01 9.74002838e-01
-8.74446750e-01 -4.66724902e-01 -6.99088812e-01 2.05737948e-02
-4.72064584e-01 2.35361964e-01 1.37158048e+00 1.23957050e+00
-9.14866924e-01 9.43292379e-01 -1.10660350e+00 -3.46121043e-01
2.06424862e-01 4.14612442e-01 -4.12312716e-01 -9.40770656e-02
-1.08527255e+00 9.00952399e-01 3.77716660e-01 -2.62838125e-01
-6.43802404e-01 -9.02490973e-01 -5.78491509e-01 2.90618036e-02
3.76434505e-01 -9.22045827e-01 1.26029718e+00 -4.85808790e-01
-1.76526356e+00 2.67389446e-01 -3.70427929e-02 -2.45921046e-01
3.66883069e-01 -2.88703591e-01 -4.13892597e-01 -1.97985068e-01
-1.88481569e-01 1.88730627e-01 6.09233260e-01 -5.08261502e-01
-3.26627582e-01 -4.98881191e-01 3.80881839e-02 -8.40215236e-02
-1.30017415e-01 4.43219870e-01 -3.15592796e-01 -7.05021918e-01
-1.00252576e-01 -9.84544337e-01 -4.20411795e-01 -5.67979395e-01
-1.47516083e-03 -3.91616136e-01 4.50705677e-01 -5.61607122e-01
1.49431944e+00 -1.68640101e+00 5.03512844e-02 3.28864068e-01
-1.11535475e-01 4.70584899e-01 6.53786138e-02 7.12141573e-01
-3.75413388e-01 -1.69961900e-01 1.30378529e-01 1.06178030e-01
-8.70577246e-03 2.36446947e-01 -1.36884395e-02 3.47016662e-01
-3.51584733e-01 4.10829514e-01 -9.75970924e-01 9.04308259e-02
4.87139255e-01 6.55673683e-01 -7.03815937e-01 1.01624923e-02
-1.24536566e-01 4.73843992e-01 -2.65354931e-01 3.99972886e-01
6.99736476e-01 -3.33030760e-01 4.80229884e-01 7.56294280e-02
-3.34073961e-01 3.93641591e-01 -1.69291162e+00 1.62439477e+00
-3.34685743e-01 1.62255645e-01 -2.47591406e-01 -6.33179128e-01
8.27430904e-01 2.09849074e-01 5.77242672e-01 1.26478476e-02
7.70335421e-02 3.59772472e-03 -5.80237396e-02 -2.95839131e-01
5.70218325e-01 -1.15191683e-01 2.85859764e-01 9.54652071e-01
2.54613370e-01 -9.16770473e-03 7.91890085e-01 5.40751517e-01
1.11734140e+00 3.07433605e-01 5.34339786e-01 -6.39590681e-01
5.97765923e-01 -2.90838033e-01 4.70320433e-01 7.28573561e-01
4.13015574e-01 1.07805558e-01 3.06230187e-01 -6.03327870e-01
-6.96743548e-01 -9.55643654e-01 -4.46005642e-01 1.63960230e+00
-6.02364719e-01 -1.31993246e+00 -3.84274185e-01 -2.96058536e-01
2.38810152e-01 6.18902564e-01 -3.69886160e-01 2.74905771e-01
1.15303144e-01 -1.16061759e+00 1.42605320e-01 4.53793883e-01
-9.03208107e-02 -6.87346756e-01 -1.24390945e-01 3.00173312e-01
2.64222287e-02 -5.57804704e-01 -5.43513633e-02 2.00816438e-01
-8.52385879e-01 -1.05667162e+00 -4.98226047e-01 -6.83465824e-02
1.94081619e-01 6.12379491e-01 1.27847135e+00 1.19604999e-02
-2.06313476e-01 2.01743916e-01 -4.91877794e-01 -6.53692961e-01
-1.57395899e-01 4.66548055e-02 4.23606545e-01 -3.65438014e-01
6.21813416e-01 -7.53086865e-01 -5.28538167e-01 5.89384496e-01
-6.68284833e-01 -1.32469207e-01 1.87276825e-01 7.45192111e-01
4.57538992e-01 7.89089054e-02 4.72029686e-01 -1.30825531e+00
8.25091064e-01 -7.06110299e-01 -9.52259004e-01 -1.01873260e-02
-8.25847149e-01 -1.70872584e-01 3.33752066e-01 -1.64594740e-01
-8.67678404e-01 2.22466454e-01 -6.45945430e-01 -8.78136680e-02
-7.05107898e-02 7.42112339e-01 7.96828642e-02 7.61695951e-02
8.87510419e-01 -7.88588077e-02 -2.15373591e-01 -1.01969838e+00
2.27945641e-01 7.74247408e-01 -8.79604369e-03 -7.07666993e-01
4.92013156e-01 1.39108315e-01 -1.22356467e-01 -1.01704097e+00
-7.87801027e-01 -8.33661139e-01 -4.47668344e-01 -2.09466964e-01
4.59875405e-01 -1.03759098e+00 -1.08375835e+00 2.51778990e-01
-5.55249512e-01 -2.31276885e-01 4.09609318e-01 5.41033447e-01
-2.45666072e-01 4.09396887e-01 -4.01944607e-01 -7.62295902e-01
-3.65050077e-01 -8.25608492e-01 6.61819041e-01 3.10986936e-01
-3.04597169e-01 -9.03989971e-01 5.76762319e-01 7.28158712e-01
2.76763231e-01 -1.42161995e-01 4.72094655e-01 -4.90889609e-01
-1.48383081e-01 -1.59627303e-01 3.19529139e-02 2.26000249e-01
-1.62562579e-01 5.52291870e-01 -9.78530109e-01 -3.13807458e-01
-4.23255444e-01 -1.39938414e-01 8.51875365e-01 4.44143742e-01
1.17762840e+00 3.41570005e-02 -4.36706662e-01 3.25127035e-01
1.10513651e+00 -4.49889228e-02 6.61447823e-01 4.65144590e-02
3.90235007e-01 3.11800599e-01 6.52309716e-01 8.51743519e-01
2.65261978e-01 7.75854468e-01 -8.20364654e-02 2.69850254e-01
4.44769412e-01 -7.47298151e-02 4.01925534e-01 6.56302750e-01
-5.83244681e-01 1.13473069e-02 -8.27649236e-01 -1.05336674e-01
-2.26343060e+00 -1.25834858e+00 -9.24185276e-01 2.31765437e+00
7.03275681e-01 -1.43545270e-01 7.23124146e-01 1.93663850e-01
2.75970429e-01 -2.92861611e-01 -9.98521447e-02 -5.63403547e-01
2.01020256e-01 4.84296203e-01 3.11451465e-01 4.01198030e-01
-1.12641811e+00 7.95440435e-01 7.66260719e+00 9.41617072e-01
-6.42977357e-01 5.35600185e-01 2.01226354e-01 -2.99637705e-01
-1.07370056e-01 1.38749212e-01 -7.96616793e-01 6.25018477e-01
1.36810112e+00 -1.34190377e-02 6.42055571e-01 1.15992296e+00
4.06288356e-01 -6.85231686e-01 -7.57970870e-01 1.08937573e+00
-4.10037428e-01 -1.54789400e+00 -4.98452038e-01 1.18288159e-01
5.42948961e-01 4.38152552e-01 -1.45443141e-01 3.62754345e-01
7.46908307e-01 -8.39003563e-01 2.48067737e-01 4.29214329e-01
3.12025845e-01 -9.17991817e-01 7.47563779e-01 4.66586888e-01
-9.38531518e-01 -1.25128612e-01 -7.09913552e-01 -5.06284595e-01
-2.57717986e-02 1.15953314e+00 -7.34712064e-01 6.57338738e-01
9.13371801e-01 7.50830889e-01 -6.25104010e-01 1.06774986e+00
-1.52317330e-01 7.46514142e-01 -4.81352687e-01 -6.84169456e-02
-8.51850584e-02 -5.89062154e-01 1.48925319e-01 1.19976509e+00
2.73398966e-01 7.80739635e-02 2.66849339e-01 3.49316776e-01
3.74857455e-01 4.27414238e-01 -2.81431466e-01 -1.96964249e-01
2.35840216e-01 1.54642940e+00 -8.25738490e-01 -3.79102767e-01
-5.80438375e-01 4.36802149e-01 3.83605987e-01 -1.29312947e-01
-9.08333719e-01 -1.84301436e-01 9.45390582e-01 6.74999729e-02
3.53560895e-01 -3.32927227e-01 1.28617957e-01 -1.43788075e+00
-6.73763275e-01 -9.59863424e-01 7.14304030e-01 -9.88744557e-01
-1.14972782e+00 4.33338791e-01 8.67641047e-02 -7.99595356e-01
-8.38534161e-02 -9.37284052e-01 -3.10051054e-01 8.10949147e-01
-8.41761768e-01 -7.14893281e-01 -2.93286473e-01 7.82381415e-01
5.09165302e-02 -2.16085911e-01 1.35642374e+00 6.41068220e-01
-8.23393822e-01 1.69348940e-01 6.49166703e-01 -4.27779824e-01
8.88660848e-01 -1.33833325e+00 -1.90769006e-02 5.69293499e-01
4.88445371e-01 1.25485063e+00 9.38679218e-01 -6.41824901e-01
-1.45788193e+00 -7.47797251e-01 6.49508119e-01 -7.79569209e-01
7.33674049e-01 -4.58720177e-01 -6.55637980e-01 4.93224561e-01
4.80575822e-02 -1.82148889e-01 1.73702335e+00 8.56274545e-01
-4.68982190e-01 -1.03622034e-01 -9.31444705e-01 2.28693292e-01
6.44230545e-01 -3.00862044e-01 -4.24309045e-01 9.53127801e-01
-4.35197055e-02 -2.56171763e-01 -1.30849969e+00 4.05910090e-02
8.35804522e-01 -1.29081726e+00 1.03199208e+00 -5.97419500e-01
1.16246551e-01 -5.67658126e-01 -1.57021895e-01 -1.13395238e+00
-8.00272882e-01 -9.48224664e-01 -3.55603904e-01 8.59327972e-01
3.75194877e-01 -6.69192016e-01 6.86417818e-01 5.29269218e-01
1.57885328e-01 -5.48495293e-01 -5.67482591e-01 -7.58195937e-01
-1.03502780e-01 -7.33779430e-01 5.64700246e-01 7.48154759e-01
4.28314269e-01 5.94097972e-01 -6.88468456e-01 -1.28397048e-01
2.21148446e-01 3.08472544e-01 1.13276291e+00 -1.55969238e+00
-8.87101889e-01 -9.49554816e-02 -1.60823479e-01 -6.53655052e-01
-5.39417118e-02 -8.95540595e-01 -5.26799262e-01 -1.22525597e+00
3.87044162e-01 -3.05064112e-01 -3.30107123e-01 6.37260199e-01
-2.03360036e-01 6.21391475e-01 4.28412184e-02 1.36969462e-01
-4.42335963e-01 -7.19266245e-03 8.61340344e-01 1.44507289e-01
-2.11616829e-01 5.16066253e-01 -9.08676147e-01 6.06177628e-01
7.03632593e-01 -6.96673334e-01 -3.66566867e-01 1.31789058e-01
7.45647430e-01 -9.42776129e-02 -8.51355866e-02 -7.34634578e-01
-2.72991776e-01 -1.67025641e-01 5.31199217e-01 -8.83295894e-01
4.10535216e-01 -6.74622536e-01 8.71003807e-01 4.00453001e-01
2.20225275e-01 1.85676552e-02 1.09874137e-01 5.33920527e-01
1.99466452e-01 -4.39456820e-01 7.06622601e-01 -3.45864117e-01
-1.80237561e-01 2.40728319e-01 -7.93233454e-01 -5.80992520e-01
1.00089180e+00 1.02931306e-01 -1.97012916e-01 -2.78056711e-01
-9.89667773e-01 -4.85135503e-02 4.07227784e-01 2.77520120e-01
1.96898773e-01 -1.10613096e+00 -4.33863133e-01 1.69898823e-01
-1.64212808e-02 -7.70193040e-01 4.35619026e-01 1.17827165e+00
-6.64744973e-01 5.63768744e-01 -3.10068399e-01 -3.74332994e-01
-1.51147485e+00 6.71692669e-01 -1.41793385e-01 -2.99780965e-01
-4.98851746e-01 7.83618748e-01 -2.64258921e-01 -5.55717230e-01
-7.54389763e-02 -1.53219089e-01 -2.79201895e-01 2.49831989e-01
1.25056767e+00 7.51322508e-01 2.70389378e-01 -3.98570925e-01
-4.34833407e-01 1.29554689e-01 -5.89328520e-02 1.42812416e-01
1.75459766e+00 2.26765767e-01 -4.73012656e-01 3.73734772e-01
6.68161094e-01 1.24078393e-01 -7.27181494e-01 1.88665837e-01
-1.04359552e-01 -6.70400739e-01 4.12424862e-01 -7.12546945e-01
-7.00082898e-01 4.52971339e-01 4.52297390e-01 2.45203242e-01
7.74173081e-01 -2.34026477e-01 4.38286901e-01 5.69774389e-01
4.69495237e-01 -1.08368325e+00 -2.55994946e-01 5.88949919e-01
5.69235325e-01 -1.03696084e+00 6.91514313e-01 -3.96741658e-01
-5.35251737e-01 1.09546864e+00 2.11171925e-01 -1.70315608e-01
1.18989265e+00 1.67132631e-01 -1.82788819e-01 -1.35775656e-01
-9.58751976e-01 -1.69228852e-01 2.95124412e-01 6.61357939e-01
9.35280442e-01 3.63421738e-01 -7.14884937e-01 8.09325576e-01
-2.22631201e-01 2.33190760e-01 6.51539385e-01 8.47675204e-01
-2.52942443e-01 -1.68290293e+00 -6.03737295e-01 8.13320160e-01
-7.26234436e-01 -1.75638676e-01 -2.05963492e-01 1.47810504e-01
7.02457726e-02 1.06455290e+00 -4.28192139e-01 -4.61009741e-01
-1.09793946e-01 5.63330874e-02 6.52503908e-01 -7.90715337e-01
-9.32444513e-01 3.81239027e-01 3.95836473e-01 -8.21739674e-01
-5.98628521e-01 -1.02092612e+00 -7.29962230e-01 -7.34488606e-01
-5.12989163e-01 5.67053199e-01 8.59058380e-01 6.56970143e-01
4.75488663e-01 2.60164589e-01 3.02636474e-01 -9.20143366e-01
-1.26384929e-01 -1.22063053e+00 -8.26073468e-01 -5.97022250e-02
-4.56384033e-01 -1.02024937e+00 -1.33334100e-01 -2.85895705e-01] | [7.4619035720825195, 4.438168525695801] |
80395a59-74ec-4795-ae44-5dacbe38cfd2 | emotion-recognition-with-pre-trained | 2212.13885 | null | https://arxiv.org/abs/2212.13885v1 | https://arxiv.org/pdf/2212.13885v1.pdf | Emotion Recognition with Pre-Trained Transformers Using Multimodal Signals | In this paper, we address the problem of multimodal emotion recognition from multiple physiological signals. We demonstrate that a Transformer-based approach is suitable for this task. In addition, we present how such models may be pretrained in a multimodal scenario to improve emotion recognition performances. We evaluate the benefits of using multimodal inputs and pre-training with our approach on a state-ofthe-art dataset. | ['James L Crowley', 'Julien Cumin', 'Grégoire Lefebvre', 'Juan Vazquez-Rodriguez'] | 2022-12-22 | null | null | null | null | ['multimodal-emotion-recognition', 'multimodal-emotion-recognition'] | ['computer-vision', 'speech'] | [ 4.52372819e-01 -9.78714675e-02 3.36658061e-01 -7.96051025e-01
-6.96643949e-01 -5.38794816e-01 3.37288499e-01 -3.26209627e-02
-6.44374251e-01 7.22675383e-01 2.87059173e-02 -6.27895398e-03
-4.30105291e-02 -3.27427626e-01 -5.06308794e-01 -4.23778713e-01
-7.36187920e-02 1.78543717e-01 -2.92698473e-01 -4.94135022e-01
5.36474958e-02 6.04425251e-01 -1.61999238e+00 9.06404734e-01
6.38420999e-01 1.37938213e+00 -2.61074603e-01 6.97790384e-01
4.28973734e-02 3.91583592e-01 -7.58396685e-01 -6.49610043e-01
-4.98315655e-02 -4.12201405e-01 -8.71421039e-01 -1.85996797e-02
5.64354174e-02 2.19802968e-02 3.36372592e-02 6.98674560e-01
7.14988887e-01 2.55064040e-01 7.39468932e-01 -1.34443820e+00
-4.99721318e-02 3.78139734e-01 -1.03117727e-01 6.92424874e-05
7.41008937e-01 -2.29755476e-01 8.21272433e-01 -9.94347274e-01
4.69246775e-01 1.09557283e+00 5.43436646e-01 8.27841341e-01
-1.29961371e+00 -3.54289621e-01 1.61545333e-02 3.05202097e-01
-1.10378492e+00 -7.40428805e-01 1.08379948e+00 -4.67138514e-02
1.25397348e+00 3.44340682e-01 5.77552319e-01 1.56000960e+00
6.16021082e-02 1.02699566e+00 1.41152847e+00 -5.00802755e-01
2.34510764e-01 3.62581700e-01 2.02220455e-01 4.44815427e-01
-5.29658079e-01 -6.40026554e-02 -8.37822616e-01 -3.33419591e-01
3.62772584e-01 -3.41356307e-01 -2.63618588e-01 -7.62854442e-02
-1.10767853e+00 5.96115768e-01 1.79826140e-01 5.22448897e-01
-7.05632150e-01 1.07167922e-01 7.91530609e-01 6.16146505e-01
3.71606499e-01 6.38553262e-01 -5.72757304e-01 -3.61811012e-01
-8.51726890e-01 -1.77951902e-01 9.98597324e-01 6.17479861e-01
4.87103969e-01 6.85700551e-02 1.82949435e-02 1.02860463e+00
1.83493838e-01 3.41430724e-01 5.36318541e-01 -8.12021673e-01
1.90707073e-01 3.53686243e-01 1.82577074e-02 -7.12016881e-01
-6.50846303e-01 -3.49224024e-02 -5.36912560e-01 -8.80520344e-02
2.89923221e-01 -4.26583350e-01 -8.58264804e-01 1.68559384e+00
-8.29104856e-02 1.95831895e-01 5.61590075e-01 8.60742688e-01
7.93993056e-01 6.49925113e-01 3.09378028e-01 -2.65366137e-01
1.29329360e+00 -7.02570498e-01 -1.01852846e+00 -3.50643881e-02
3.38747084e-01 -5.09761751e-01 8.76680911e-01 9.10721958e-01
-1.13956988e+00 -3.36078703e-01 -6.84049606e-01 2.56205738e-01
-7.81936109e-01 3.95201683e-01 7.00605929e-01 9.76298630e-01
-1.03590620e+00 3.98663253e-01 -7.08110929e-01 -5.61259687e-01
-3.23332548e-02 7.38341212e-01 -6.07887268e-01 3.31791997e-01
-1.38195765e+00 1.18055940e+00 4.48406667e-01 4.12303835e-01
-6.38475895e-01 -7.40490109e-02 -9.31855023e-01 1.48232028e-01
-8.51321965e-02 -3.75970095e-01 1.08545649e+00 -1.11175227e+00
-1.91841936e+00 7.29937613e-01 -4.57701296e-01 -1.36170357e-01
-6.15956523e-02 -4.66990732e-02 -7.54820704e-01 5.20777941e-01
-7.67616153e-01 7.07356215e-01 9.11295474e-01 -1.34278083e+00
-3.72989587e-02 -3.82280856e-01 -1.21363364e-01 1.14099994e-01
-7.76548147e-01 3.83553952e-01 -7.97996148e-02 -3.63730848e-01
-1.62821203e-01 -7.23681867e-01 -1.87386259e-01 -5.18279731e-01
-5.01131006e-02 -9.09050833e-03 5.71639121e-01 -5.57132781e-01
8.30638170e-01 -2.24185991e+00 5.45996964e-01 8.13619196e-01
-2.42796600e-01 1.98952332e-02 -4.58942354e-01 6.62028372e-01
-1.53903350e-01 1.66461244e-01 -2.29623064e-01 -5.92899561e-01
2.71834254e-01 4.12360072e-01 -3.67838711e-01 5.13233468e-02
6.28266752e-01 9.98204291e-01 -4.15048420e-01 -3.33619267e-01
2.34404311e-01 7.88394570e-01 -2.72366881e-01 5.69013834e-01
2.08209634e-01 5.24463952e-01 -1.52894661e-01 8.05455744e-01
4.57687467e-01 2.24558964e-01 3.77489477e-01 -4.56672847e-01
1.04216926e-01 -1.48613034e-02 -9.30208981e-01 1.74836504e+00
-6.71041727e-01 6.88637733e-01 2.43974656e-01 -1.38446128e+00
1.05765235e+00 8.31008434e-01 4.88441527e-01 -7.30523944e-01
5.84985852e-01 1.25153825e-01 -1.42064076e-02 -7.55034029e-01
4.65859920e-01 -5.54992795e-01 -2.34642372e-01 3.31261188e-01
7.08332598e-01 -9.36113372e-02 4.95181940e-02 -2.17491761e-01
7.68098652e-01 3.27148512e-02 2.88791656e-01 1.69337895e-02
7.33511388e-01 -2.17749000e-01 9.41752270e-02 4.30697709e-01
-2.18839064e-01 3.80858928e-01 4.86784726e-01 -1.13787107e-01
-5.35076022e-01 -8.12968969e-01 -2.70170212e-01 1.32039106e+00
-1.91632032e-01 -4.18060869e-01 -6.99477732e-01 -5.61650217e-01
-3.60526770e-01 6.00097537e-01 -7.02941537e-01 -1.75295681e-01
-9.25613865e-02 -8.40233922e-01 8.50232661e-01 6.87685311e-01
2.23510545e-02 -1.16135907e+00 -8.78596544e-01 1.89254031e-01
-4.33531314e-01 -1.29384792e+00 2.65233070e-01 7.91118383e-01
-8.80447984e-01 -6.74236357e-01 -7.22688198e-01 -4.75564390e-01
4.83321071e-01 -4.07905400e-01 9.01446939e-01 -2.14553595e-01
-6.09373264e-02 1.04686475e+00 -4.38774943e-01 -5.82866311e-01
-8.43088031e-02 -4.10665944e-02 -9.65510122e-03 2.91042268e-01
3.93386841e-01 -6.08268917e-01 -2.02633917e-01 2.90492684e-01
-1.16817200e+00 -4.06951755e-01 4.82434422e-01 8.82423580e-01
1.76761597e-01 -1.31729409e-01 7.54539192e-01 -4.52510357e-01
1.08007789e+00 -3.61862570e-01 -2.44380999e-02 5.44177771e-01
-1.80945218e-01 -8.22346006e-03 5.32082915e-01 -7.89575338e-01
-1.08071828e+00 3.30365866e-01 -5.55379868e-01 -4.44236308e-01
-6.35483146e-01 9.04651701e-01 -1.94312736e-01 -6.59780085e-01
4.47994530e-01 1.27015054e-01 -2.67615523e-02 -3.09745163e-01
2.89633244e-01 7.59326100e-01 3.60497832e-01 -7.40773976e-01
1.31390721e-01 2.36343220e-01 1.26461210e-02 -7.26384521e-01
-1.42955214e-01 -3.85814011e-01 -6.92985415e-01 -5.44842482e-01
6.19062185e-01 -7.80057430e-01 -9.17582452e-01 3.63696486e-01
-1.22345400e+00 -1.37201980e-01 7.05947587e-03 5.16487718e-01
-7.95060694e-01 1.77886084e-01 -6.52273178e-01 -1.21573007e+00
-3.36691469e-01 -1.03547537e+00 1.13750029e+00 2.55335450e-01
-3.33887964e-01 -1.19006014e+00 1.68902814e-01 1.16931736e-01
5.69552958e-01 1.34874821e-01 5.61105490e-01 -1.03929889e+00
2.77529091e-01 -1.92893684e-01 1.09250240e-01 3.64334136e-01
-1.42815650e-01 -1.37370721e-01 -1.36325943e+00 1.04346000e-01
9.96718556e-03 -8.90360653e-01 6.71936154e-01 2.20904443e-02
1.02035391e+00 3.36778313e-01 -1.33803397e-01 2.76573658e-01
1.23159754e+00 1.69718832e-01 7.58269429e-01 7.30459690e-02
2.44487569e-01 9.24281061e-01 4.43373859e-01 3.82505774e-01
3.35958660e-01 5.76080501e-01 3.80225718e-01 -2.54153132e-01
6.36924863e-01 2.66582817e-01 4.74207133e-01 6.86768413e-01
-1.38691753e-01 -4.24853444e-01 -1.03782701e+00 3.70868534e-01
-1.88504803e+00 -9.30058897e-01 3.89529318e-02 1.53408074e+00
6.91474855e-01 -3.67306679e-01 2.01687783e-01 2.52283096e-01
3.07429433e-01 -2.36595005e-01 -1.72187746e-01 -1.09657252e+00
-4.53429461e-01 7.27909148e-01 -1.97413340e-01 3.29384983e-01
-1.02972305e+00 7.65486300e-01 8.13000298e+00 3.08308482e-01
-1.41100228e+00 5.03378138e-02 3.57675135e-01 -1.59587320e-02
-2.53639758e-01 -4.54955816e-01 -2.37999111e-01 5.30937761e-02
1.40839541e+00 2.52188236e-01 5.48046887e-01 4.37506557e-01
4.48218286e-02 -2.23499879e-01 -1.24611557e+00 1.27787209e+00
4.91345078e-01 -6.22650325e-01 -1.53060898e-01 -1.97629765e-01
2.01849386e-01 -1.68603674e-01 -3.76691073e-02 3.57192516e-01
-4.43180323e-01 -1.15977609e+00 4.91816789e-01 8.07793558e-01
3.93561691e-01 -7.49038398e-01 1.02441025e+00 8.32548887e-02
-8.65367472e-01 1.69146899e-02 3.91987897e-03 -3.29914875e-02
2.67706811e-01 2.62189686e-01 -5.15227675e-01 8.69590461e-01
5.63282132e-01 5.65029740e-01 -5.90583682e-01 9.21897411e-01
-6.60944730e-02 4.57009226e-01 -4.83730078e-01 -1.86497658e-01
8.07496831e-02 -8.36358219e-02 1.75570905e-01 1.66843843e+00
4.51510131e-01 2.52821326e-01 -2.35810056e-01 6.59539938e-01
1.40477389e-01 3.49058896e-01 -8.13168705e-01 -3.17041606e-01
-2.46005535e-01 1.62591112e+00 -4.42587167e-01 -4.30676997e-01
-3.51461411e-01 1.24368870e+00 2.70578206e-01 5.82433581e-01
-7.59524107e-01 -5.23765624e-01 3.51483703e-01 -7.84087360e-01
2.44926989e-01 -1.12921856e-01 -1.71633199e-01 -1.51462853e+00
-8.78065303e-02 -9.27835882e-01 3.55829716e-01 -1.23090589e+00
-1.60608530e+00 9.94546056e-01 7.21003264e-02 -9.87126589e-01
-4.79783654e-01 -1.17489803e+00 -6.25133395e-01 9.35226440e-01
-1.42987430e+00 -1.05733013e+00 -1.08079098e-01 9.23086405e-01
-1.65333711e-02 -3.25934379e-03 1.28439498e+00 4.63486105e-01
-4.04685110e-01 5.01729429e-01 -3.22082460e-01 -1.84046105e-01
8.20895433e-01 -1.15223682e+00 -3.59313637e-01 4.29378748e-01
1.71250120e-01 6.44847751e-01 6.95892155e-01 -4.46254686e-02
-1.61136603e+00 -4.30057198e-01 7.99951196e-01 -3.20901543e-01
6.15728319e-01 -5.00091493e-01 -9.11665022e-01 6.30770862e-01
8.24010730e-01 -2.47737572e-01 1.32470453e+00 4.61108238e-01
-2.08335564e-01 5.95731251e-02 -1.30808020e+00 3.50505292e-01
5.33806801e-01 -8.72349799e-01 -8.31917167e-01 -1.92060009e-01
3.61363813e-02 -1.95768639e-01 -1.29646075e+00 6.99323475e-01
8.13839853e-01 -7.11933672e-01 8.46365333e-01 -8.95164788e-01
2.21126467e-01 -6.44617677e-02 -3.41606319e-01 -1.78163552e+00
2.85082370e-01 -5.17013848e-01 -6.30131960e-02 1.09240639e+00
6.30436838e-01 -5.62154770e-01 3.60333741e-01 8.61780107e-01
1.08820744e-01 -6.48755670e-01 -9.77175057e-01 -4.44880128e-01
-4.22101989e-02 -6.93421960e-01 3.12693208e-01 8.88951659e-01
8.87111425e-01 3.92782867e-01 -5.56021273e-01 -5.52478479e-03
1.82223395e-01 -3.00568454e-02 4.74743605e-01 -9.96537924e-01
1.29244765e-02 -3.98598999e-01 -5.13508797e-01 -6.19594753e-01
5.80347776e-01 -6.70043170e-01 2.23284274e-01 -1.08327889e+00
9.49850157e-02 2.50091791e-01 -8.18721533e-01 7.03216791e-01
2.34151278e-02 3.61785382e-01 2.58324057e-01 -4.39901680e-01
-6.50992990e-01 7.34241843e-01 7.14119673e-01 -1.11566350e-01
-1.59768000e-01 -4.22383577e-01 -3.83346498e-01 3.79309028e-01
1.18465507e+00 -2.32783839e-01 -2.03616366e-01 -1.64308533e-01
3.12794268e-01 1.93290696e-01 4.66747224e-01 -8.22698474e-01
2.80879080e-01 2.27958515e-01 5.88996768e-01 -3.99910063e-01
9.69441473e-01 -1.20313489e+00 1.20489988e-02 -2.52785664e-02
-4.37453061e-01 3.18589568e-01 8.24310124e-01 1.37950659e-01
-6.58205569e-01 -2.62328535e-01 3.48638356e-01 1.52889326e-01
-6.11276567e-01 -3.96653980e-01 -8.77161145e-01 -4.61357594e-01
7.11165130e-01 2.95102596e-02 -9.82333422e-02 -5.40154636e-01
-1.23458636e+00 1.94661617e-01 5.91301247e-02 3.08985084e-01
1.03804636e+00 -1.31072283e+00 -1.84220523e-01 2.45663896e-01
3.62152249e-01 -1.14266384e+00 9.36488658e-02 1.26417577e+00
1.89582214e-01 3.51196706e-01 -8.27854037e-01 -5.73290646e-01
-1.48912883e+00 4.14460689e-01 5.67009330e-01 1.53151795e-03
1.83598399e-01 3.50992143e-01 -2.28948280e-01 -7.75771737e-01
2.22509027e-01 -1.95372164e-01 -5.43577194e-01 2.85815597e-01
3.23770016e-01 -4.55979258e-02 1.16808563e-01 -5.82558095e-01
-6.24517977e-01 4.67099011e-01 3.34361434e-01 -8.58474910e-01
1.27558148e+00 -1.06449142e-01 -3.01145911e-01 8.57145309e-01
1.18694317e+00 -2.18042016e-01 -3.59497309e-01 1.48971871e-01
4.22248431e-02 -8.63644853e-03 -4.89326678e-02 -1.15869009e+00
-9.27382469e-01 1.20420277e+00 7.04328418e-01 2.08684132e-01
1.73423183e+00 -1.00165568e-01 3.28785181e-01 7.85605609e-01
2.95681179e-01 -1.18957865e+00 5.54521158e-02 5.50443590e-01
8.95220041e-01 -1.26240087e+00 -4.29091930e-01 -2.43350565e-01
-1.05438292e+00 1.52221107e+00 3.91369164e-01 1.55027166e-01
6.28260851e-01 2.76060611e-01 3.09979916e-01 -3.74238670e-01
-1.03189433e+00 -4.70613748e-01 6.01362765e-01 5.35241544e-01
6.33711815e-01 -1.28947854e-01 -3.06921333e-01 8.47985446e-01
1.02234684e-01 1.19563162e-01 2.70702064e-01 1.09129393e+00
6.19803928e-02 -1.37998533e+00 -4.91468877e-01 1.66327178e-01
-5.60507238e-01 -1.24114171e-01 -9.19995666e-01 6.22480810e-01
-1.27195358e-01 1.15064931e+00 -2.97724515e-01 -7.14223325e-01
5.36352158e-01 8.71510863e-01 8.05442452e-01 -2.94529021e-01
-1.08076525e+00 3.47552091e-01 4.62414116e-01 -6.84407711e-01
-1.02141142e+00 -7.09712923e-01 -1.06329191e+00 2.12431267e-01
-1.53467447e-01 2.38168791e-01 9.49067831e-01 1.00283599e+00
4.81582344e-01 4.74137217e-01 5.08918583e-01 -9.28865969e-01
-8.93174261e-02 -1.07585633e+00 -4.15195912e-01 4.99405771e-01
1.07430458e-01 -5.35771191e-01 -2.88208723e-01 1.97628066e-01] | [13.316887855529785, 5.4105682373046875] |
ad871dbb-4faf-4c6f-b607-adfab0ac8bbb | understanding-the-stochastic-dynamics-of | 2208.06245 | null | https://arxiv.org/abs/2208.06245v2 | https://arxiv.org/pdf/2208.06245v2.pdf | Understanding the stochastic dynamics of sequential decision-making processes: A path-integral analysis of multi-armed bandits | The multi-armed bandit (MAB) model is one of the most classical models to study decision-making in an uncertain environment. In this model, a player chooses one of $K$ possible arms of a bandit machine to play at each time step, where the corresponding arm returns a random reward to the player, potentially from a specific unknown distribution. The target of the player is to collect as many rewards as possible during the process. Despite its simplicity, the MAB model offers an excellent playground for studying the trade-off between exploration versus exploitation and designing effective algorithms for sequential decision-making under uncertainty. Although many asymptotically optimal algorithms have been established, the finite-time behaviors of the stochastic dynamics of the MAB model appear much more challenging to analyze, due to the intertwine between the decision-making and the rewards being collected. In this paper, we employ techniques in statistical physics to analyze the MAB model, which facilitates the characterization of the distribution of cumulative regrets at a finite short time, the central quantity of interest in an MAB algorithm, as well as the intricate dynamical behaviors of the model. Our analytical results, in good agreement with simulations, point to the emergence of an interesting multimodal regret distribution, with large regrets resulting from excess exploitation of sub-optimal arms due to an initial unlucky output from the optimal one. | ['Chi Ho Yeung', 'Bo Li'] | 2022-08-11 | null | null | null | null | ['decision-making-under-uncertainty', 'decision-making-under-uncertainty'] | ['medical', 'reasoning'] | [-7.98678547e-02 8.16845223e-02 -2.87454545e-01 -5.60191367e-03
-5.90746045e-01 -8.98618340e-01 2.97317743e-01 2.20423788e-01
-4.99968380e-01 9.79732454e-01 -1.68201715e-01 -4.88364905e-01
-8.36005032e-01 -7.27739573e-01 -7.70396113e-01 -1.18180609e+00
-2.05128416e-01 9.83451664e-01 -2.88481832e-01 -2.14955524e-01
3.16076726e-01 4.81425256e-01 -1.27262223e+00 -7.51645267e-02
6.08703196e-01 1.52313769e+00 4.13161330e-02 6.89115226e-01
-3.73995364e-01 6.45650446e-01 -4.03638572e-01 -7.42345452e-01
4.81153220e-01 -5.37806153e-01 -7.46171594e-01 -1.14631720e-01
-6.45334840e-01 -1.25066712e-01 -1.39439330e-01 1.15772712e+00
1.48682326e-01 3.49392354e-01 6.68371260e-01 -8.78549278e-01
-9.25004110e-02 9.09807384e-01 -6.78676426e-01 3.71301144e-01
9.99623686e-02 -2.97594834e-02 1.04581022e+00 9.68486294e-02
4.61826295e-01 1.08358169e+00 5.50917052e-02 4.33713436e-01
-1.23776579e+00 -2.29237571e-01 6.92941621e-02 5.92659451e-02
-9.00332749e-01 -8.37815255e-02 7.33863652e-01 -2.64015675e-01
7.14825913e-02 6.41837120e-01 8.47660840e-01 6.71031535e-01
3.53399992e-01 1.06650937e+00 1.30689991e+00 -5.46984017e-01
8.27794850e-01 2.29575410e-01 2.34876364e-01 1.51361823e-01
4.85458255e-01 5.19157588e-01 -4.19432491e-01 -3.33867311e-01
5.29739916e-01 1.33473337e-01 -1.55810669e-01 -7.71389127e-01
-5.94487786e-01 7.59912193e-01 3.82224023e-01 1.02195367e-01
-8.03047359e-01 3.72453332e-01 1.75126866e-02 5.87338746e-01
3.33018959e-01 3.11549544e-01 -1.89216748e-01 -4.16780919e-01
-6.99614942e-01 6.05327785e-01 8.64540935e-01 4.97750223e-01
4.42108929e-01 -5.53954005e-01 -3.18872541e-01 2.46033162e-01
5.89227527e-02 4.37039405e-01 8.57520755e-03 -9.09507632e-01
8.71817529e-01 9.39292759e-02 1.04542625e+00 -5.74928582e-01
-3.85804951e-01 -8.14395070e-01 -6.90782189e-01 5.59538126e-01
1.11266601e+00 -3.07163060e-01 -3.58220637e-01 1.50204289e+00
1.55062854e-01 -6.84034824e-01 1.43126119e-02 1.09688783e+00
-3.51634234e-01 6.49764359e-01 -1.83235168e-01 -6.65293872e-01
9.37630177e-01 -4.65315998e-01 -4.95450586e-01 -2.56098956e-01
2.69456029e-01 -4.25473899e-01 5.64242780e-01 6.45837069e-01
-1.36697078e+00 3.01327050e-01 -6.71351969e-01 5.53045213e-01
-6.03095293e-02 -4.14396703e-01 6.87659383e-01 8.71828377e-01
-5.31495452e-01 8.97346973e-01 -7.63147593e-01 -1.20071424e-02
3.81539792e-01 2.47035339e-01 2.12127209e-01 -2.28329733e-01
-7.47387111e-01 8.16222370e-01 2.69024700e-01 5.75984180e-01
-5.96169770e-01 -1.75720319e-01 2.50135548e-02 4.08653527e-01
7.66819000e-01 -5.92121124e-01 1.56206203e+00 -1.30797613e+00
-1.64382172e+00 3.69777471e-01 -1.07289501e-01 -5.42143703e-01
1.10384548e+00 7.97026157e-02 2.60522395e-01 -6.60250336e-02
-2.23064467e-01 -1.78747088e-01 7.20520794e-01 -1.02259815e+00
-5.41555703e-01 -6.89667523e-01 2.78154075e-01 4.01463032e-01
2.78604269e-01 -1.63530886e-01 1.19047262e-01 -2.90679723e-01
2.06622571e-01 -1.07654107e+00 -5.38027823e-01 -4.46614385e-01
-3.57289165e-01 6.35381639e-02 -2.45503783e-01 -1.70963481e-01
8.47541332e-01 -1.97655153e+00 3.69703293e-01 3.39019746e-01
-8.23312327e-02 -2.98311055e-01 1.95310712e-01 6.99029207e-01
1.51310280e-01 1.52181834e-01 5.26491255e-02 -1.20789327e-01
2.46435046e-01 7.44555295e-02 -4.49179202e-01 6.85954273e-01
-3.58125776e-01 8.21527600e-01 -8.27749968e-01 1.90272719e-01
1.69046931e-02 -4.07222509e-01 -2.74991661e-01 1.25825197e-01
-4.12330955e-01 2.87741065e-01 -1.07898879e+00 3.72709721e-01
6.87591374e-01 6.26581430e-04 1.59801096e-01 5.18074751e-01
-3.93740177e-01 -5.05473502e-02 -1.29287803e+00 1.01455045e+00
-3.08505595e-01 2.42133364e-01 4.91587013e-01 -1.32530642e+00
3.79593134e-01 -7.37195974e-03 4.79203820e-01 -6.38889313e-01
4.68547285e-01 2.71438122e-01 1.12357117e-01 -3.32631022e-01
5.46846390e-01 -7.43685305e-01 -2.31777981e-01 8.30921352e-01
-3.96778464e-01 6.93379417e-02 1.06745772e-01 -2.25650761e-02
1.08437812e+00 -4.53952372e-01 2.34362587e-01 -4.31038558e-01
5.38440458e-02 5.22691049e-02 2.79885232e-01 1.25375116e+00
-2.89397895e-01 2.96912670e-01 1.00145721e+00 -5.78876555e-01
-7.39157259e-01 -1.05389464e+00 8.06886405e-02 9.37165856e-01
5.62741339e-01 3.78636986e-01 -4.94559735e-01 -3.24988604e-01
4.30330694e-01 8.14824820e-01 -9.26832974e-01 -4.67449240e-02
-1.53786704e-01 -1.07022476e+00 -2.40300000e-01 -8.68802890e-02
2.65826374e-01 -8.43571603e-01 -8.55837643e-01 3.42146724e-01
-3.03514712e-02 -5.14292002e-01 -1.76382273e-01 3.72710526e-01
-9.33629572e-01 -1.02420354e+00 -7.62190521e-01 2.94816822e-01
4.40680772e-01 1.25408024e-01 9.15242136e-01 -4.59921926e-01
-6.88677356e-02 3.58965337e-01 -2.65509665e-01 -7.07655728e-01
-3.75109971e-01 -1.94740325e-01 -4.70495224e-02 3.72877657e-01
-6.50489628e-02 -1.05622746e-01 -7.63448834e-01 2.94293612e-01
-8.63023520e-01 -1.15506880e-01 6.06083691e-01 6.05903327e-01
4.45068806e-01 2.53560930e-01 3.34349006e-01 -5.38365364e-01
7.38184929e-01 -6.29993498e-01 -1.11318576e+00 3.89765888e-01
-3.85092080e-01 4.21969414e-01 6.47239208e-01 -3.28561485e-01
-1.07378757e+00 -8.03522840e-02 3.77320230e-01 -1.91208161e-02
2.16287717e-01 4.53158319e-01 -2.44935765e-03 -3.20454724e-02
3.43515784e-01 1.72372431e-01 1.69102862e-01 -5.62051237e-01
3.61438334e-01 4.80471522e-01 7.03909546e-02 -9.57194090e-01
4.33095515e-01 5.64112902e-01 4.50042069e-01 -4.96650517e-01
-9.04183686e-01 -1.93642229e-01 -7.73151740e-02 -3.60463619e-01
2.95322299e-01 -2.08244637e-01 -1.41687596e+00 4.55498427e-01
-9.88842010e-01 -2.01943949e-01 -7.44653761e-01 5.58008373e-01
-1.05167627e+00 -3.15329134e-02 -3.54992837e-01 -1.75377488e+00
-1.86073463e-02 -9.41216707e-01 2.90170759e-01 5.45070648e-01
2.70022810e-01 -6.23657346e-01 3.78615186e-02 4.62378591e-01
5.25419354e-01 1.91650122e-01 9.17014599e-01 -4.18162495e-01
-8.85949433e-01 -4.66781884e-01 6.60902560e-02 -1.14265151e-01
-3.31409276e-01 -2.18466520e-01 -3.74533504e-01 -3.27095240e-01
3.14641297e-01 -1.91659257e-01 7.00304329e-01 8.61993074e-01
9.58227336e-01 -4.15588319e-01 -2.71440715e-01 1.89363956e-01
1.48223948e+00 4.74302858e-01 3.00090671e-01 5.72339714e-01
-2.94298202e-01 6.33068979e-01 6.57796144e-01 7.41679728e-01
-6.96485788e-02 6.56085730e-01 8.96456718e-01 5.33839107e-01
9.56014395e-01 4.90966104e-02 1.41923696e-01 -1.56150803e-01
-3.01447481e-01 -3.87177914e-01 -6.51316226e-01 2.67155379e-01
-2.10110807e+00 -1.02249217e+00 1.46659851e-01 2.84637141e+00
4.98572528e-01 2.32197180e-01 5.59826195e-01 -8.12041685e-02
6.73587799e-01 -1.24643438e-01 -8.70517492e-01 -7.22607017e-01
-1.36930615e-01 -5.66554330e-02 9.85301435e-01 5.00441253e-01
-6.69452846e-01 3.39888871e-01 6.47401142e+00 9.47727203e-01
-9.86842275e-01 -3.17330509e-02 1.02857506e+00 -8.13394427e-01
-3.08027953e-01 5.72937764e-02 -5.45517981e-01 7.39501953e-01
7.64942467e-01 -5.98399103e-01 9.20285940e-01 7.02897310e-01
4.69768047e-01 -7.04147339e-01 -8.11222255e-01 7.78858185e-01
-7.64733970e-01 -1.46048629e+00 -3.50721568e-01 4.54651147e-01
6.52667522e-01 -1.04233108e-01 1.37383670e-01 9.33567900e-03
6.25691295e-01 -8.69137287e-01 1.09081292e+00 8.65996957e-01
2.24185407e-01 -1.16092789e+00 8.13502789e-01 8.25234413e-01
-4.94573176e-01 -6.20627403e-01 -2.71894276e-01 -5.14130652e-01
1.61976099e-01 8.72932732e-01 -2.27240682e-01 5.95701456e-01
4.11014855e-01 -1.87926158e-01 1.67998061e-01 1.26769698e+00
2.58456141e-01 4.16042060e-01 -7.47432947e-01 -6.65338755e-01
4.63737130e-01 -8.49958479e-01 7.05903351e-01 5.40990114e-01
4.11766469e-01 3.70307326e-01 -3.21768343e-01 9.87253249e-01
5.63755929e-02 1.16185464e-01 -3.09943378e-01 -3.53600889e-01
1.36937916e-01 8.86300981e-01 -9.31308806e-01 6.58329576e-02
7.05055147e-02 7.82583892e-01 3.94095033e-01 3.71193320e-01
-5.90721428e-01 -1.17948264e-01 7.44569778e-01 1.08102679e-01
2.98211724e-01 -1.52496114e-01 -5.33287644e-01 -9.36714351e-01
3.93000066e-01 -3.18791538e-01 3.92814666e-01 -2.37806216e-01
-1.20357943e+00 1.15189821e-01 -1.24907419e-01 -7.83119023e-01
-2.83383489e-01 -3.65168840e-01 -5.97585082e-01 9.31150913e-01
-1.21164012e+00 -3.58314127e-01 4.48573053e-01 2.05818444e-01
1.26491383e-01 1.78596646e-01 3.23623776e-01 -3.53850096e-01
-5.69581211e-01 2.16570407e-01 1.24173379e+00 -3.54784042e-01
-2.07792357e-01 -1.18571866e+00 -1.83346659e-01 5.00744998e-01
-7.77845979e-02 3.86360466e-01 1.17412937e+00 -3.39126199e-01
-1.92072272e+00 -3.39517325e-01 2.46397763e-01 -3.50403994e-01
9.54251707e-01 -1.91229969e-01 -2.94341713e-01 2.67148107e-01
-2.21990734e-01 -1.15709506e-01 3.86328429e-01 2.57227421e-01
3.55154902e-01 -2.33173057e-01 -1.23618746e+00 4.32814360e-01
6.85402632e-01 4.78904396e-02 -2.84453124e-01 1.06780909e-01
1.33924454e-01 -3.50140542e-01 -2.33293489e-01 -9.14382860e-02
8.24526072e-01 -1.31377733e+00 5.66958547e-01 -7.62148798e-01
2.77280360e-01 1.84810907e-01 -2.23615319e-02 -1.45368552e+00
-1.50500506e-01 -1.05603278e+00 1.90284681e-02 8.20461690e-01
4.28080261e-01 -6.66163564e-01 1.19631934e+00 9.91351485e-01
6.11376047e-01 -9.14735675e-01 -1.56908572e+00 -9.03126657e-01
3.80316943e-01 -4.71352279e-01 4.64733481e-01 2.00678021e-01
2.28404120e-01 -1.53806597e-01 -3.10415775e-01 -3.34467962e-02
8.98313105e-01 6.57005250e-01 4.11650538e-01 -9.24854279e-01
-6.59389198e-01 -8.08043182e-01 7.41015673e-02 -1.13993740e+00
-2.91817218e-01 -3.81991088e-01 1.11616097e-01 -1.17743003e+00
2.72156477e-01 -8.15230548e-01 -4.88777339e-01 -2.61342645e-01
8.62119198e-02 -3.49159688e-01 5.53123653e-01 2.75727719e-01
-6.97295904e-01 4.20257688e-01 1.18689895e+00 -6.72368482e-02
-3.79569173e-01 9.55599070e-01 -9.12830293e-01 4.94927734e-01
6.45488799e-01 -5.89247525e-01 -6.24957606e-02 -2.05443949e-01
7.16619372e-01 7.80739546e-01 9.54875648e-02 -4.65351164e-01
1.29878983e-01 -6.69139981e-01 1.47507349e-02 -5.11180103e-01
2.99022317e-01 -8.60934854e-01 3.58929992e-01 5.93182147e-01
-5.04247963e-01 -2.29225591e-01 -1.87943533e-01 9.69092131e-01
1.45130083e-01 -7.64237761e-01 6.84867442e-01 -2.92529225e-01
1.12119213e-01 1.42084226e-01 -7.69420445e-01 -8.47076103e-02
1.13623512e+00 1.91573665e-01 -3.92065980e-02 -7.75623620e-01
-9.73465621e-01 3.48263741e-01 2.76250422e-01 -1.08955763e-01
-7.29070604e-02 -9.42661703e-01 -3.67855757e-01 -1.62059307e-01
-3.91249329e-01 -1.10831492e-01 4.76451188e-01 7.16461003e-01
-3.25110316e-01 4.39857841e-01 -7.04514682e-02 -1.13236099e-01
-7.04458773e-01 6.05892122e-01 5.31111240e-01 -6.92991912e-01
-6.74893856e-02 6.82166100e-01 -2.70525478e-02 2.66989201e-01
3.80593501e-02 -3.30336869e-01 2.24605396e-01 2.61552423e-01
5.46281040e-01 6.48165345e-01 -5.03602102e-02 -1.16911076e-03
5.83229139e-02 2.20164284e-01 -1.77928694e-02 -4.49671417e-01
1.19476986e+00 -3.36624265e-01 -9.77703631e-02 5.37578404e-01
6.13424361e-01 1.53363394e-02 -1.33155394e+00 -2.18552381e-01
2.08320260e-01 -6.23737872e-01 2.22317539e-02 -9.44061160e-01
-8.25395882e-01 6.00831985e-01 2.41903201e-01 9.67301190e-01
8.98066938e-01 1.33335426e-01 3.76193523e-01 3.41789454e-01
7.94594407e-01 -1.07452703e+00 -3.59880477e-01 3.83615434e-01
7.97065258e-01 -9.92605388e-01 -2.50993907e-01 1.45048136e-02
-5.11621535e-01 1.09791148e+00 -1.28321037e-01 -1.59476265e-01
5.30157745e-01 1.72344431e-01 -2.42280215e-01 1.61921471e-01
-6.42309725e-01 -2.10067883e-01 -1.62861362e-01 7.44192749e-02
-2.34776661e-01 6.10113025e-01 -5.83276451e-01 9.32807326e-01
-2.45502442e-01 1.15911871e-01 6.43662751e-01 8.53682458e-01
-4.85745847e-01 -1.04845214e+00 -5.78510702e-01 5.66983640e-01
-7.50581145e-01 2.90567338e-01 -4.02596653e-01 4.85018700e-01
-3.57355326e-01 8.93327355e-01 1.29486978e-01 3.00484210e-01
2.12461784e-01 5.40193953e-02 6.55099690e-01 -1.05796397e-01
-3.56710672e-01 -4.68254723e-02 -1.38304578e-02 -5.53891122e-01
-3.71396653e-02 -1.01480544e+00 -7.51558840e-01 -5.78194857e-01
-3.01618576e-01 7.87079513e-01 8.89097810e-01 1.27982032e+00
1.45220220e-01 -2.51162127e-02 9.52492058e-01 -8.10842991e-01
-1.21031284e+00 -6.27746582e-01 -1.17739832e+00 1.38529062e-01
5.55766523e-01 -4.88524884e-01 -5.71020782e-01 -7.94354856e-01] | [4.494750499725342, 3.2777392864227295] |
79da5201-f92f-4270-aa76-c803f51c1cc3 | celebv-text-a-large-scale-facial-text-video | 2303.14717 | null | https://arxiv.org/abs/2303.14717v1 | https://arxiv.org/pdf/2303.14717v1.pdf | CelebV-Text: A Large-Scale Facial Text-Video Dataset | Text-driven generation models are flourishing in video generation and editing. However, face-centric text-to-video generation remains a challenge due to the lack of a suitable dataset containing high-quality videos and highly relevant texts. This paper presents CelebV-Text, a large-scale, diverse, and high-quality dataset of facial text-video pairs, to facilitate research on facial text-to-video generation tasks. CelebV-Text comprises 70,000 in-the-wild face video clips with diverse visual content, each paired with 20 texts generated using the proposed semi-automatic text generation strategy. The provided texts are of high quality, describing both static and dynamic attributes precisely. The superiority of CelebV-Text over other datasets is demonstrated via comprehensive statistical analysis of the videos, texts, and text-video relevance. The effectiveness and potential of CelebV-Text are further shown through extensive self-evaluation. A benchmark is constructed with representative methods to standardize the evaluation of the facial text-to-video generation task. All data and models are publicly available. | ['Wayne Wu', 'Weidong Cai', 'Chen Change Loy', 'Liming Jiang', 'Hao Zhu', 'Jianhui Yu'] | 2023-03-26 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Yu_CelebV-Text_A_Large-Scale_Facial_Text-Video_Dataset_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Yu_CelebV-Text_A_Large-Scale_Facial_Text-Video_Dataset_CVPR_2023_paper.pdf | cvpr-2023-1 | ['video-generation', 'text-to-video-generation'] | ['computer-vision', 'natural-language-processing'] | [ 3.55334520e-01 -1.20790906e-01 -4.39481018e-03 -5.80767214e-01
-9.84879911e-01 -1.48821786e-01 1.05858278e+00 -6.92364275e-01
1.07537024e-01 8.16540956e-01 5.56421995e-01 6.33310974e-01
9.64752957e-02 -4.56738055e-01 -5.71437836e-01 -8.71934712e-01
1.83063582e-01 4.06307548e-01 -2.69254267e-01 -2.92862296e-01
-5.14469072e-02 1.42845809e-01 -2.19564128e+00 5.47378600e-01
6.44593000e-01 1.37416089e+00 8.10860321e-02 6.31772101e-01
4.32852432e-02 6.01426303e-01 -6.49594426e-01 -1.08924425e+00
1.45612061e-01 -8.27324390e-01 -3.95557433e-01 3.30839485e-01
1.19605744e+00 -5.96995294e-01 -4.39774364e-01 6.59308434e-01
9.30271089e-01 5.20980246e-02 8.70032609e-01 -1.74228084e+00
-8.10123861e-01 2.70126700e-01 -4.26393479e-01 -1.82965890e-01
7.29342937e-01 2.72376209e-01 8.31375360e-01 -1.45290816e+00
1.32711542e+00 1.39482975e+00 2.87643790e-01 9.95954692e-01
-9.17150021e-01 -9.20153320e-01 -1.93903461e-01 1.19643345e-01
-1.41046917e+00 -1.20872188e+00 6.80348575e-01 -5.97129285e-01
5.55687487e-01 3.09081912e-01 9.16762233e-01 1.80765891e+00
1.20423704e-01 5.57601213e-01 6.25455856e-01 -3.18756044e-01
-5.39915189e-02 -5.50068356e-02 -8.37283611e-01 6.06801927e-01
7.31318370e-02 2.30727732e-01 -1.28033268e+00 -5.87598793e-03
5.03241301e-01 -4.26809847e-01 -2.93628722e-01 -1.32258162e-01
-1.17320383e+00 6.70350194e-01 -2.75402933e-01 1.32697940e-01
-2.86024988e-01 1.64942116e-01 5.91345191e-01 3.00854504e-01
7.76401758e-01 1.75662115e-02 8.62653553e-02 -3.49344671e-01
-1.38553512e+00 7.44838774e-01 4.30465370e-01 1.56584191e+00
3.96663547e-01 5.78050792e-01 -8.88061166e-01 9.78775680e-01
2.18371421e-01 1.06215286e+00 4.15693372e-01 -1.02955806e+00
4.38697368e-01 8.67839307e-02 -4.75616120e-02 -1.25989687e+00
1.83097228e-01 3.53271633e-01 -8.11299562e-01 -1.31758600e-01
4.22528498e-02 -2.23166078e-01 -7.63914227e-01 1.80266273e+00
3.32202107e-01 1.60252124e-01 -1.39480472e-01 8.79360437e-01
1.35895550e+00 7.96279252e-01 -1.97329298e-02 -6.19762957e-01
1.20385361e+00 -7.38736987e-01 -9.95677054e-01 2.65033871e-01
8.93396437e-02 -1.15302372e+00 8.34967494e-01 2.27188945e-01
-1.36045051e+00 -7.30474412e-01 -4.82651830e-01 -1.25691980e-01
8.13602954e-02 6.38363123e-01 2.28079647e-01 7.72837460e-01
-1.42690444e+00 1.10043325e-01 2.45207362e-02 -5.11947930e-01
7.45082498e-01 3.16485614e-02 -7.94013321e-01 -1.79755330e-01
-1.15554404e+00 3.87831986e-01 1.77995309e-01 4.44962531e-02
-1.29256690e+00 -6.56425953e-01 -9.93707120e-01 -3.77282768e-01
3.26946586e-01 -8.71974885e-01 1.34508073e+00 -1.35196674e+00
-1.58875334e+00 1.07312489e+00 -5.05480528e-01 -7.63141736e-02
8.07302356e-01 -1.52723014e-01 -4.92696404e-01 5.23648977e-01
2.46737078e-01 1.09390914e+00 1.66764283e+00 -1.05610061e+00
-6.12802505e-01 -1.40324906e-01 -3.53853464e-01 2.05268964e-01
-6.79266930e-01 2.10002556e-01 -7.74696410e-01 -1.12279940e+00
-6.18690968e-01 -7.87762940e-01 3.26499313e-01 2.05634445e-01
-2.49891296e-01 -2.87182182e-01 9.52179790e-01 -6.28533542e-01
1.24719763e+00 -2.05132580e+00 1.26970842e-01 -2.13509858e-01
2.42058396e-01 1.94157138e-01 -5.25969863e-01 7.11320877e-01
-1.58258036e-01 1.11434288e-01 3.24502945e-01 -6.65429652e-01
-3.76588032e-02 -2.44677246e-01 -1.32328928e-01 2.30638474e-01
4.25016642e-01 9.38701868e-01 -8.09501469e-01 -1.05429161e+00
2.04565987e-01 5.66620171e-01 -5.69105387e-01 3.00751925e-01
-3.34627539e-01 2.02143773e-01 -3.63164872e-01 9.86366630e-01
6.51271343e-01 2.02379093e-01 -1.17933027e-01 -4.61372703e-01
3.10279459e-01 -4.57598895e-01 -6.04957461e-01 1.71102726e+00
-1.16767205e-01 1.16108310e+00 5.98464906e-02 -4.69046175e-01
8.97662103e-01 6.16221249e-01 7.60088801e-01 -7.82386005e-01
3.18198770e-01 1.36835739e-01 -6.37656033e-01 -8.51077020e-01
9.62747693e-01 -2.54124343e-01 1.85063645e-01 3.13148737e-01
4.75199491e-01 -2.76674509e-01 9.44725811e-01 5.25728881e-01
5.41307569e-01 4.83155549e-01 -1.05986834e-01 4.11959477e-02
4.81357843e-01 -2.71947950e-01 3.19974214e-01 1.93387598e-01
-2.21397832e-01 9.68913019e-01 5.67480505e-01 -9.92387682e-02
-1.21568704e+00 -9.16001141e-01 -6.74945042e-02 1.12507284e+00
-1.68957412e-01 -8.95165980e-01 -1.10020661e+00 -4.83852148e-01
-9.58512723e-02 3.32581401e-01 -8.84835422e-01 -1.30240232e-01
-2.01398492e-01 -4.29899305e-01 5.71221530e-01 1.50752589e-01
3.37893099e-01 -1.27632678e+00 -8.87399018e-02 -1.50733009e-01
-7.30912566e-01 -1.44125056e+00 -1.07273924e+00 -1.05445826e+00
-4.49931592e-01 -1.13744640e+00 -1.11032712e+00 -9.48495269e-01
7.60924220e-01 2.86580533e-01 1.36231434e+00 -4.72782850e-02
-5.10899961e-01 5.99334717e-01 -6.83682501e-01 -2.79635310e-01
-5.74669302e-01 -3.80374938e-01 9.54027772e-02 5.63898802e-01
-3.04400511e-02 -2.39123061e-01 -4.89070535e-01 5.31927109e-01
-8.73614848e-01 4.43087727e-01 3.56871873e-01 1.00959623e+00
5.42017877e-01 -1.62723064e-01 6.42730415e-01 -5.23537636e-01
6.77232087e-01 -2.99179018e-01 -3.15975308e-01 1.74875066e-01
-3.39006722e-01 -4.56156403e-01 3.86141419e-01 -5.27496874e-01
-1.47181070e+00 -1.44700125e-01 2.51282491e-02 -8.08292031e-01
5.94907291e-02 1.48203090e-01 -3.11484218e-01 1.31316304e-01
5.89249849e-01 4.13888067e-01 2.13202700e-01 9.34400335e-02
4.41203505e-01 7.04457998e-01 6.04421139e-01 -6.75362229e-01
8.82355034e-01 3.22473258e-01 -8.98357201e-03 -1.13430643e+00
-4.21186209e-01 -3.29965651e-02 -3.87498260e-01 -9.21970665e-01
7.64326394e-01 -1.20366168e+00 -4.55929726e-01 8.26561689e-01
-1.08937013e+00 -1.38860762e-01 -3.25106084e-01 1.01735622e-01
-8.99057448e-01 1.84420630e-01 -4.40911770e-01 -7.13159263e-01
-5.51663637e-01 -1.12018216e+00 1.72699797e+00 2.91258153e-02
-5.74218221e-02 -5.49174786e-01 -8.63847807e-02 7.48389602e-01
2.22180784e-01 4.71993566e-01 2.96191603e-01 -5.35410596e-03
-3.92112851e-01 -1.32612348e-01 -2.01166302e-01 3.16503733e-01
1.94262743e-01 7.58330345e-01 -1.02692080e+00 -3.85148704e-01
-4.23054308e-01 -8.21118176e-01 5.83479285e-01 4.76071358e-01
1.03559852e+00 -4.56567258e-01 -8.50879475e-02 6.11552894e-01
8.92528355e-01 -1.95076577e-02 8.06935608e-01 -2.57000118e-01
6.72936440e-01 8.59004259e-01 1.05556393e+00 7.38803506e-01
2.39956170e-01 9.63595152e-01 2.15048343e-01 1.37927502e-01
-4.49465334e-01 -4.57576513e-01 6.92736208e-01 6.81003034e-01
-2.91245311e-01 -5.86408615e-01 -3.92415226e-01 5.65462947e-01
-1.58137453e+00 -1.55762661e+00 -1.34668157e-01 1.82030416e+00
9.86317039e-01 -5.14476180e-01 3.25928301e-01 -1.18394740e-01
9.27057564e-01 1.90206543e-01 -2.23598674e-01 3.17733288e-02
-4.22838449e-01 -5.29265553e-02 -3.07268053e-01 -8.31940025e-02
-1.03646517e+00 8.83789480e-01 6.46857929e+00 1.44450581e+00
-1.01436353e+00 5.03794244e-03 8.76794636e-01 -5.96224546e-01
-1.08028777e-01 -4.18854862e-01 -8.51376772e-01 7.42043495e-01
8.85921001e-01 -5.85528314e-01 1.28880695e-01 7.45643437e-01
6.64774954e-01 -6.63264608e-03 -1.06429827e+00 1.50601363e+00
7.37789631e-01 -1.48087513e+00 7.25730717e-01 5.72458375e-03
1.19261277e+00 -5.25134504e-01 2.31272534e-01 9.51392278e-02
-3.15189242e-01 -1.01538169e+00 1.23376191e+00 6.10698700e-01
1.68402076e+00 -6.47904932e-01 4.94986534e-01 -2.83144951e-01
-1.34711778e+00 5.99115118e-02 -2.82141358e-01 6.43879235e-01
3.44591469e-01 5.58743894e-01 -5.74073315e-01 7.35202730e-01
8.42279732e-01 1.15963483e+00 -7.38499880e-01 6.37984037e-01
8.41569230e-02 2.21836403e-01 8.71961042e-02 -1.59963593e-02
-2.18676910e-01 -1.19708948e-01 5.20693958e-01 1.13940406e+00
8.24599922e-01 -1.40554402e-02 -1.59349591e-01 6.95587575e-01
-3.36482882e-01 4.53445971e-01 -7.35456645e-01 -5.25063097e-01
4.62608665e-01 1.49757445e+00 -2.51693934e-01 -4.31429088e-01
-2.73208857e-01 8.81225944e-01 -1.58502489e-01 3.13071877e-01
-9.75093007e-01 -1.68655947e-01 7.10988879e-01 3.78469676e-01
1.80607110e-01 2.63064444e-01 4.48348254e-01 -1.13427246e+00
1.50732920e-01 -1.19851029e+00 4.24116477e-02 -1.33549368e+00
-1.39349926e+00 9.56205904e-01 2.64740020e-01 -1.49034476e+00
-6.85401261e-01 -4.51285928e-01 -3.07020158e-01 4.65834409e-01
-1.07953191e+00 -1.60296237e+00 -9.27683234e-01 1.07221758e+00
9.38106477e-01 -8.12358439e-01 4.78569686e-01 4.86437172e-01
-5.29645205e-01 1.01026237e+00 -9.67385024e-02 -2.33377162e-02
1.32210922e+00 -4.91234571e-01 3.71584982e-01 7.20404387e-01
-5.05987443e-02 1.86850294e-01 5.19132555e-01 -7.70798385e-01
-1.51866007e+00 -1.42976749e+00 7.54384220e-01 -4.83206630e-01
4.14774984e-01 -5.09127736e-01 -2.35278711e-01 3.05895358e-01
3.59942377e-01 -4.66605313e-02 4.51461434e-01 -6.44955277e-01
-9.66843218e-02 -3.61872494e-01 -1.00421047e+00 8.20692778e-01
1.39592588e+00 -4.43749040e-01 -4.96380217e-02 2.94289887e-01
4.85906094e-01 -3.77184808e-01 -8.04218888e-01 2.86918968e-01
7.85999537e-01 -1.22777987e+00 7.12453723e-01 -2.28698388e-01
9.38407302e-01 -1.62732191e-02 -9.34207514e-02 -1.18762267e+00
1.49212599e-01 -1.25742924e+00 6.65108487e-02 1.78876770e+00
8.03309232e-02 -9.89874229e-02 8.59371364e-01 2.10406512e-01
-3.50205265e-02 -4.31894273e-01 -8.43285322e-01 -5.82479239e-01
-4.10969883e-01 -3.34666014e-01 6.44172490e-01 7.03996837e-01
-2.20041886e-01 2.42156357e-01 -9.90298331e-01 -8.32812190e-01
5.30557454e-01 -1.44435972e-01 1.26205277e+00 -9.86300170e-01
3.11542034e-01 -4.38834667e-01 -5.02938926e-01 -6.08198762e-01
3.16573620e-01 -7.38977432e-01 1.66117787e-01 -1.18586469e+00
3.79261404e-01 5.46461605e-02 6.13649666e-01 3.24214071e-01
-1.21042706e-01 6.65113628e-01 3.18120867e-01 1.60389796e-01
-5.65893173e-01 1.12073839e+00 1.74185979e+00 -9.09999162e-02
2.64794350e-01 -3.07759196e-01 -4.85483587e-01 3.20445925e-01
4.18602645e-01 -2.04782337e-01 -7.66449273e-01 -2.11706609e-01
1.18891276e-01 3.05476367e-01 2.39507586e-01 -8.21268916e-01
-1.41686574e-01 -3.17413211e-01 5.15113950e-01 -4.97104555e-01
7.66544163e-01 -4.50082093e-01 4.64258969e-01 -1.27866298e-01
-3.94370705e-01 6.77657947e-02 1.08110979e-01 5.22821903e-01
-4.25736576e-01 6.18043765e-02 7.16344953e-01 1.79699630e-01
-6.04439676e-01 8.92329395e-01 -2.07702950e-01 3.70979071e-01
1.26926756e+00 -5.29735088e-01 -4.27772582e-01 -1.03119123e+00
-4.64903086e-01 1.37878675e-02 4.81709450e-01 8.12001407e-01
1.11748576e+00 -1.85127389e+00 -1.27485526e+00 5.40099382e-01
5.20096421e-01 -4.82371926e-01 5.20104408e-01 7.11288095e-01
-4.50497776e-01 3.05622965e-01 -5.34526348e-01 -5.55530727e-01
-1.69819999e+00 1.82856932e-01 1.09780475e-01 3.12684566e-01
-3.13154668e-01 6.94815576e-01 4.41261321e-01 1.29520193e-01
4.52998728e-02 2.93348432e-01 -1.60325736e-01 4.37217861e-01
7.81583905e-01 3.56995255e-01 -4.59053703e-02 -1.41000009e+00
7.08107948e-02 7.20014334e-01 1.86159596e-01 -2.39587218e-01
1.06424272e+00 -3.03132713e-01 -1.29701635e-02 -1.24005735e-01
9.35567498e-01 -1.26172468e-01 -1.18955946e+00 1.08834587e-01
-5.37501693e-01 -9.05707538e-01 -2.38294140e-01 -4.62010354e-01
-1.55280459e+00 6.77760839e-01 4.54135984e-01 -2.82039851e-01
1.39974761e+00 -2.84853399e-01 8.59566867e-01 7.10084336e-03
3.84413898e-01 -1.15587604e+00 6.32500470e-01 1.98637307e-01
1.41421556e+00 -1.13298738e+00 -2.76418999e-02 -4.32866871e-01
-8.74764025e-01 1.03875959e+00 8.81824791e-01 4.48277026e-01
4.32601750e-01 9.40715745e-02 -3.52095789e-03 -1.92199666e-02
-1.12363791e+00 1.19588792e-01 6.09110534e-01 9.96822000e-01
5.24351120e-01 -3.12507212e-01 -2.48697460e-01 3.17320466e-01
-5.59888482e-01 2.95964658e-01 4.66537237e-01 5.43785155e-01
9.85854305e-03 -1.00910616e+00 -4.09708053e-01 5.72726369e-01
-4.42909271e-01 -7.28399977e-02 -5.21787405e-01 7.43762493e-01
3.49240266e-02 1.21522236e+00 -1.04852758e-01 -4.82019305e-01
1.50611386e-01 -8.24213773e-02 6.60372674e-01 -2.98679650e-01
-2.77791679e-01 3.46173555e-01 3.04551572e-01 -7.36090839e-01
-6.67022407e-01 -8.07797492e-01 -6.74034834e-01 -6.90992236e-01
-2.27543041e-01 -4.18150704e-03 4.87754941e-01 4.65044826e-01
5.61767936e-01 2.06185505e-01 7.77967036e-01 -1.39836276e+00
-5.33430018e-02 -9.62947130e-01 -6.16444409e-01 7.55576015e-01
1.27388820e-01 -8.07386160e-01 -2.31083438e-01 7.56469965e-01] | [12.77625846862793, -0.08386053889989853] |
53fadcff-54a1-4260-8551-9c8a3d5846ac | smarnet-teaching-machines-to-read-and | 1710.02772 | null | http://arxiv.org/abs/1710.02772v1 | http://arxiv.org/pdf/1710.02772v1.pdf | Smarnet: Teaching Machines to Read and Comprehend Like Human | Machine Comprehension (MC) is a challenging task in Natural Language
Processing field, which aims to guide the machine to comprehend a passage and
answer the given question. Many existing approaches on MC task are suffering
the inefficiency in some bottlenecks, such as insufficient lexical
understanding, complex question-passage interaction, incorrect answer
extraction and so on. In this paper, we address these problems from the
viewpoint of how humans deal with reading tests in a scientific way.
Specifically, we first propose a novel lexical gating mechanism to dynamically
combine the words and characters representations. We then guide the machines to
read in an interactive way with attention mechanism and memory network. Finally
we add a checking layer to refine the answer for insurance. The extensive
experiments on two popular datasets SQuAD and TriviaQA show that our method
exceeds considerable performance than most state-of-the-art solutions at the
time of submission. | ['Rongqin Yang', 'Zhou Zhao', 'Zheqian Chen', 'Deng Cai', 'Bin Cao', 'Xiaofei He'] | 2017-10-08 | null | null | null | null | ['triviaqa'] | ['miscellaneous'] | [ 3.12365025e-01 -2.25527529e-02 2.80833811e-01 -4.99575227e-01
-7.60831773e-01 -7.57682204e-01 3.48106861e-01 6.85901761e-01
-6.21230781e-01 8.87868285e-01 3.28071028e-01 -7.26399601e-01
7.63717741e-02 -8.34608436e-01 -6.25507414e-01 -7.07508549e-02
4.99068350e-01 4.85383242e-01 6.13607228e-01 -3.88328582e-01
9.76319313e-01 -2.25370258e-01 -1.38265824e+00 4.65751648e-01
1.13670826e+00 5.93900979e-01 3.63937557e-01 1.03448439e+00
-7.43899584e-01 9.25234675e-01 -6.38125241e-01 -4.23560590e-01
-2.66367793e-01 -9.33232665e-01 -1.55804956e+00 -3.35000366e-01
2.22395569e-01 -1.69195265e-01 1.49012417e-01 1.12722707e+00
6.07351065e-01 3.16766500e-01 5.38414240e-01 -6.25649333e-01
-7.66566753e-01 4.45488334e-01 -4.31138277e-01 9.08675969e-01
1.03189743e+00 2.34271407e-01 8.44191134e-01 -7.24532545e-01
3.69910508e-01 1.37358809e+00 1.38847440e-01 6.86689138e-01
-6.93622470e-01 -2.50046909e-01 3.88604701e-01 7.88829684e-01
-7.29841948e-01 -2.17156187e-01 5.05355239e-01 -3.69905293e-01
1.09602284e+00 4.22633469e-01 3.33412886e-01 7.21934974e-01
3.48306119e-01 6.58500135e-01 1.04065561e+00 -6.03531241e-01
2.51335502e-01 -9.22588706e-02 8.93765926e-01 8.20138156e-01
-7.58533785e-03 -5.30038357e-01 -5.29990315e-01 -2.41375342e-02
3.68936598e-01 -4.54715610e-01 -4.44869965e-01 2.34829724e-01
-1.05944526e+00 9.72105920e-01 1.46395087e-01 1.72941446e-01
-2.35869348e-01 -2.75874674e-01 3.87610108e-01 5.77935815e-01
1.09442966e-02 7.00016737e-01 -5.60313702e-01 -2.89257914e-01
-4.70163167e-01 3.28402430e-01 1.17277694e+00 7.16808677e-01
4.12926823e-01 -4.87645268e-01 -6.15197659e-01 5.92782617e-01
2.43524536e-01 1.32535756e-01 6.23000324e-01 -6.01899266e-01
9.11994994e-01 8.06530118e-01 1.41530216e-01 -9.91484225e-01
-4.37316298e-01 -2.32280254e-01 -6.93171799e-01 -1.14358455e-01
4.39718395e-01 -2.39878923e-01 -8.57905626e-01 1.50207567e+00
3.68652701e-01 -7.86332637e-02 9.74928029e-03 1.03250015e+00
1.37180948e+00 8.35623741e-01 4.01012123e-01 -2.44311810e-01
1.96865189e+00 -1.31694472e+00 -9.93683219e-01 -4.18617368e-01
3.32700372e-01 -9.24721181e-01 1.42496586e+00 4.71807033e-01
-1.31694281e+00 -8.21713030e-01 -1.03033793e+00 -5.31847477e-01
-3.54753762e-01 -1.97624341e-01 2.61448562e-01 2.45354563e-01
-6.51561618e-01 3.18806201e-01 -3.08869213e-01 -4.51880544e-01
2.45018259e-01 2.08070233e-01 -5.82548901e-02 -1.06359378e-01
-1.38933122e+00 1.14651668e+00 4.58015859e-01 1.89347059e-01
-5.59472263e-01 -3.73827994e-01 -6.79862618e-01 2.11322680e-01
5.50451517e-01 -1.12620437e+00 1.49743664e+00 -8.25774789e-01
-1.58809829e+00 1.01187682e+00 -4.73257005e-01 -3.52310002e-01
3.11632067e-01 -6.57009363e-01 -2.90501267e-01 2.46161669e-01
6.43978566e-02 4.87832755e-01 5.20713627e-01 -5.78716636e-01
-4.67037678e-01 -3.16732377e-01 2.44694427e-01 3.96831214e-01
2.08439425e-01 1.39191315e-01 -4.75829482e-01 -2.63993293e-01
-2.32153293e-02 -2.21787393e-01 -2.10373566e-01 -3.84100229e-01
-2.67464280e-01 -7.22294152e-01 3.57253462e-01 -9.35187578e-01
1.56230247e+00 -1.79456663e+00 4.45443779e-01 -2.49775454e-01
2.87273616e-01 5.20704210e-01 -1.71228394e-01 4.85831648e-01
5.63875251e-02 3.32107693e-01 -2.10928172e-01 -9.03263409e-03
-1.26977578e-01 5.08714840e-02 -2.29181632e-01 -4.89417426e-02
4.47137773e-01 9.97402847e-01 -8.94257903e-01 -6.93835676e-01
-1.59882292e-01 -2.27115422e-01 -5.64029515e-01 8.51278305e-01
-7.11932659e-01 5.80583513e-01 -6.92893386e-01 4.08962458e-01
6.52660131e-01 -4.09850121e-01 -1.58704743e-01 2.17962891e-01
-1.17115587e-01 5.95905364e-01 -8.82700622e-01 1.76749206e+00
-1.29491985e-01 3.08870435e-01 2.90662684e-02 -1.06973302e+00
7.94164479e-01 1.93873629e-01 -6.63101017e-01 -8.99863064e-01
3.52150738e-01 5.83210215e-02 2.88332343e-01 -1.30487788e+00
4.67305630e-01 -1.13291992e-02 -7.40044266e-02 6.06984317e-01
-2.62060110e-02 5.38415760e-02 4.08657551e-01 2.31735274e-01
9.96445894e-01 1.66757435e-01 5.58869243e-01 -3.84026766e-01
1.18801773e+00 1.03341468e-01 2.20103011e-01 8.99534404e-01
-2.95127213e-01 6.36353076e-01 7.25189745e-01 -4.18309212e-01
-7.56859839e-01 -9.53362942e-01 3.21268737e-01 1.33610141e+00
1.52702272e-01 -1.18528970e-01 -1.07073617e+00 -6.33925021e-01
-3.98732483e-01 9.03824031e-01 -5.54660857e-01 -1.34694874e-01
-7.60885000e-01 -5.31731784e-01 2.74132878e-01 3.88786495e-01
8.42113435e-01 -1.58630013e+00 -8.13112259e-01 3.60641211e-01
-5.66304982e-01 -8.07196558e-01 -4.24423903e-01 -7.93971680e-03
-7.62363315e-01 -1.16761088e+00 -3.53846312e-01 -1.16914058e+00
5.58919132e-01 3.92008349e-02 1.55141592e+00 6.71151161e-01
-3.27027649e-01 -4.57199514e-02 -5.86819053e-01 -4.70595956e-01
-2.28135392e-01 3.51941198e-01 -5.96145570e-01 -2.72099197e-01
7.73691416e-01 -2.79972076e-01 -8.53979588e-01 1.00590447e-02
-8.65691423e-01 2.24570200e-01 4.34846222e-01 7.12517500e-01
3.46981823e-01 -4.11263794e-01 9.71411705e-01 -1.06394184e+00
1.24874210e+00 -7.86039710e-01 -3.04968059e-01 5.50519705e-01
-2.34096572e-01 2.57488072e-01 6.89330161e-01 -1.89552844e-01
-1.14622056e+00 -3.80758852e-01 -5.58043420e-01 4.92335588e-01
-4.50602442e-01 5.83567679e-01 -2.03370124e-01 3.54290932e-01
5.59147239e-01 3.89278382e-01 -7.82555342e-02 -5.96499562e-01
2.35943332e-01 5.71811974e-01 6.48186743e-01 -6.04845345e-01
4.48147684e-01 -2.24057958e-01 -3.99127066e-01 -3.80517691e-01
-1.20736098e+00 -5.03325582e-01 -5.78901708e-01 -4.03845161e-02
1.27075350e+00 -4.41722840e-01 -1.02571046e+00 2.89140433e-01
-1.47601557e+00 8.46521650e-03 9.15193260e-02 1.40913665e-01
-4.17603731e-01 5.32802165e-01 -6.90450490e-01 -5.71170568e-01
-7.43329525e-01 -8.75038505e-01 6.36364281e-01 7.78395414e-01
-4.19413120e-01 -7.89521992e-01 1.70112252e-01 7.55430579e-01
4.68840092e-01 -6.32610098e-02 1.53643715e+00 -8.49559128e-01
-4.19673711e-01 1.20845884e-01 -2.83412576e-01 1.87664658e-01
-1.71843439e-01 -3.88568282e-01 -6.63328171e-01 3.61542553e-02
3.78729820e-01 -5.34685135e-01 9.45296049e-01 -8.60375091e-02
1.27664816e+00 -1.86041474e-01 7.63437897e-02 1.81286067e-01
1.25824046e+00 2.05707297e-01 7.53720462e-01 2.61856794e-01
2.94946402e-01 7.50466704e-01 5.86384535e-01 9.95417908e-02
5.30607462e-01 7.29766581e-03 3.43722641e-01 2.48476699e-01
7.34144077e-02 -3.88381779e-01 -6.55381903e-02 1.26269734e+00
2.80325592e-01 -6.27884686e-01 -9.25974965e-01 5.11844575e-01
-2.02508235e+00 -8.20788085e-01 -5.11711717e-01 1.95326519e+00
9.96288240e-01 2.31610239e-01 -2.14624941e-01 1.74756031e-02
6.41353428e-01 1.31889861e-02 -4.26412463e-01 -9.10185397e-01
6.44824132e-02 4.33429301e-01 -2.56417811e-01 7.34719157e-01
-8.85020554e-01 9.83170331e-01 6.32988930e+00 5.74870825e-01
-6.20064139e-01 1.91073984e-01 6.99757636e-01 2.94073552e-01
-2.59449184e-01 -4.74369749e-02 -7.23066330e-01 2.99303561e-01
8.20346177e-01 -1.37623951e-01 2.95074880e-01 2.96419859e-01
1.52478829e-01 -7.07172036e-01 -1.09626222e+00 7.81194508e-01
4.25818801e-01 -1.15968311e+00 3.08192372e-01 -6.90637827e-01
3.17096442e-01 -3.63144100e-01 -3.13717186e-01 5.22229314e-01
2.84067933e-02 -1.40412152e+00 3.71839404e-01 8.47934663e-01
1.81546345e-01 -6.50516450e-01 9.87868488e-01 8.26262593e-01
-7.59181619e-01 -5.63043403e-03 -5.27701318e-01 -6.92109168e-01
3.08568388e-01 9.83010978e-02 -4.53284711e-01 5.01851737e-01
4.86437947e-01 2.54531279e-02 -8.89742792e-01 1.24970353e+00
-6.10968947e-01 8.06659341e-01 9.91147608e-02 -7.08271444e-01
3.03317428e-01 -1.01521030e-01 2.29427680e-01 1.15525186e+00
3.07733715e-02 7.21936166e-01 5.47296554e-02 8.23552847e-01
-1.95916697e-01 5.27615666e-01 -2.34399945e-01 3.99993286e-02
1.62106946e-01 9.76825178e-01 -5.75724900e-01 -5.37141621e-01
-3.84475797e-01 1.02877629e+00 6.53583229e-01 2.41461232e-01
-7.09004462e-01 -5.21472573e-01 -1.31739285e-02 1.72209670e-03
1.88098848e-01 -1.48883194e-01 -4.47482586e-01 -1.20315313e+00
2.43909732e-01 -1.12460732e+00 6.19097650e-01 -9.09697652e-01
-1.21050429e+00 6.91620171e-01 -2.89434582e-01 -7.63339639e-01
-3.09334579e-03 -5.76033115e-01 -1.08448088e+00 1.14537191e+00
-1.58742976e+00 -4.61116284e-01 -3.73179287e-01 4.38974231e-01
1.18579173e+00 1.19339682e-01 7.96505332e-01 2.59977221e-01
-5.11295617e-01 2.30941251e-01 -4.20219839e-01 1.16266966e-01
5.53546846e-01 -1.18766451e+00 4.79187369e-01 7.52866685e-01
-1.88836083e-01 7.65827835e-01 6.70716882e-01 -6.05528593e-01
-1.34991443e+00 -4.12187129e-01 1.25926328e+00 -5.10460436e-01
4.41388845e-01 -2.00358704e-01 -1.30459630e+00 2.69774795e-01
9.10536110e-01 -5.26836336e-01 6.83979273e-01 4.42378856e-02
-6.00308143e-02 3.34696472e-01 -9.51291382e-01 5.69869637e-01
6.77362919e-01 -3.79947782e-01 -1.49341643e+00 3.17484319e-01
8.93254638e-01 -3.93594593e-01 -1.71374917e-01 1.57368436e-01
2.17928603e-01 -9.21537995e-01 4.93625611e-01 -1.20774293e+00
6.90276742e-01 -4.85107005e-01 1.84707657e-01 -1.03905821e+00
-4.34499562e-01 -5.59186995e-01 6.10012896e-02 1.20856369e+00
5.56142330e-01 -3.28441799e-01 3.66974354e-01 4.73245293e-01
-4.41847816e-02 -7.79272199e-01 -7.28787482e-01 8.74905214e-02
2.34754264e-01 -1.78705812e-01 5.79382837e-01 5.04824519e-01
1.80845663e-01 1.01891124e+00 -7.85963610e-02 -1.89963561e-02
1.57174006e-01 1.62894726e-01 5.63642323e-01 -1.06782508e+00
-2.40709752e-01 -4.07696545e-01 1.08541556e-01 -1.46084046e+00
-1.26994804e-01 -6.05597138e-01 2.71850914e-01 -1.74448419e+00
5.47539890e-01 3.46022189e-01 -1.53533295e-01 -1.14279568e-01
-8.84317636e-01 -2.61446834e-01 1.10094517e-01 -1.98597252e-01
-1.10139573e+00 3.60987663e-01 1.49054849e+00 -2.84350663e-02
1.11481678e-02 -1.00121833e-01 -8.18435490e-01 7.49799907e-01
9.38694656e-01 -3.53263140e-01 -4.04790133e-01 -9.23663080e-01
6.08200073e-01 3.58879060e-01 3.05630624e-01 -7.73173451e-01
5.88858545e-01 -1.22599751e-01 3.27834338e-01 -7.17813134e-01
-2.02834800e-01 -2.17977419e-01 -5.83804905e-01 4.27065939e-01
-5.49952388e-01 6.47444844e-01 2.41853625e-01 5.31724513e-01
-3.53624582e-01 -7.51521170e-01 5.45211554e-01 -4.79533583e-01
-6.51209354e-01 -4.63427864e-02 -5.98427176e-01 6.91954970e-01
8.44426274e-01 1.66240364e-01 -6.13376915e-01 -4.12398309e-01
-7.03628480e-01 9.19671416e-01 -1.79075420e-01 5.81185579e-01
7.24076092e-01 -8.53899181e-01 -9.33663011e-01 1.94569796e-01
-5.11793196e-02 1.53351929e-02 4.20550823e-01 5.87240875e-01
-1.04201758e+00 5.04189730e-01 -2.59488106e-01 -1.71687633e-01
-1.27838731e+00 6.40889168e-01 2.52290457e-01 -4.95094180e-01
-2.02727094e-01 1.08290207e+00 1.51028425e-01 -3.59342247e-01
3.24510396e-01 -4.07561779e-01 -8.62967849e-01 6.90355748e-02
8.52170944e-01 1.74500600e-01 -3.44562740e-03 -1.67072624e-01
-2.07101718e-01 6.97887182e-01 -2.19271064e-01 5.42928949e-02
8.82849872e-01 -3.12272489e-01 -4.22139198e-01 4.45632428e-01
7.88645208e-01 -2.34839365e-01 -5.92456102e-01 -1.17716670e-01
2.36924767e-01 -2.56686479e-01 -5.54094970e-01 -1.12731874e+00
-3.02155733e-01 1.31225324e+00 3.16555589e-01 3.60986471e-01
8.91648173e-01 -1.59567539e-02 9.83402550e-01 6.13429070e-01
-5.89001029e-05 -1.15960133e+00 6.81166798e-02 8.37510705e-01
1.01280761e+00 -1.33847678e+00 -1.11622646e-01 -3.53386641e-01
-4.75454897e-01 1.23988760e+00 1.09129691e+00 -1.82826936e-01
3.81204456e-01 -2.48608321e-01 1.94707975e-01 -2.92797118e-01
-1.16217601e+00 -3.82017829e-02 2.53354281e-01 1.24738395e-01
7.10376680e-01 -2.29699165e-01 -9.23373640e-01 8.63660514e-01
-2.21667469e-01 -2.11692527e-02 4.30011034e-01 9.47425425e-01
-9.92788613e-01 -1.03809643e+00 -4.57357913e-01 4.59049344e-01
-6.66645706e-01 -3.19011837e-01 -5.99428952e-01 4.48993802e-01
1.28968567e-01 1.29362750e+00 -4.14112061e-02 2.32180208e-02
4.08833444e-01 4.07924652e-01 3.37749660e-01 -8.06411207e-01
-1.08592081e+00 -3.23310882e-01 1.34938225e-01 -2.27340490e-01
-1.12958118e-01 -3.46850127e-01 -1.35131550e+00 -2.06278384e-01
-1.83109760e-01 4.89234328e-01 1.75711513e-01 1.21170390e+00
2.46964186e-01 6.99710488e-01 6.95310459e-02 -2.60336744e-03
-7.31548846e-01 -1.16726756e+00 7.91284069e-02 4.48510796e-01
3.44646245e-01 -1.56605512e-01 -1.24169424e-01 -3.28311487e-03] | [11.293660163879395, 8.143415451049805] |
17955790-369f-4ce4-9bff-d6247dd29265 | music-source-separation-with-generative-flow | 2204.09079 | null | https://arxiv.org/abs/2204.09079v4 | https://arxiv.org/pdf/2204.09079v4.pdf | Music Source Separation with Generative Flow | Fully-supervised models for source separation are trained on parallel mixture-source data and are currently state-of-the-art. However, such parallel data is often difficult to obtain, and it is cumbersome to adapt trained models to mixtures with new sources. Source-only supervised models, in contrast, only require individual source data for training. In this paper, we first leverage flow-based generators to train individual music source priors and then use these models, along with likelihood-based objectives, to separate music mixtures. We show that in singing voice separation and music separation tasks, our proposed method is competitive with a fully-supervised approach. We also demonstrate that we can flexibly add new types of sources, whereas fully-supervised approaches would require retraining of the entire model. | ['Zhiyao Duan', 'Anton Selitskiy', 'Fei Jiang', 'Jordan Darefsky', 'Ge Zhu'] | 2022-04-19 | null | null | null | null | ['music-source-separation'] | ['music'] | [ 3.74202073e-01 -7.68626183e-02 -8.95864815e-02 -1.92743987e-02
-1.19158399e+00 -9.23632264e-01 5.13794661e-01 -2.80541182e-01
-5.98960230e-03 6.41348600e-01 3.20096552e-01 -7.51553178e-02
-1.94020301e-01 -2.66192585e-01 -5.39726853e-01 -6.67611241e-01
-3.38401571e-02 4.97888476e-01 6.48902496e-03 3.40832509e-02
-2.02277035e-01 2.93960571e-01 -1.49126554e+00 1.20804846e-01
1.09101748e+00 4.71405596e-01 2.87464619e-01 1.09965038e+00
-2.20327780e-01 6.76059067e-01 -6.67374015e-01 -1.11557022e-01
3.84865969e-01 -1.19644952e+00 -4.44646657e-01 3.86113107e-01
4.23208565e-01 -1.08371161e-01 -3.41376036e-01 9.97520924e-01
5.45864224e-01 3.33126783e-01 7.80463040e-01 -1.05192506e+00
-2.52084315e-01 1.00538778e+00 -4.20276344e-01 8.78084376e-02
7.39828274e-02 3.19693275e-02 1.02497649e+00 -9.24242377e-01
1.66965991e-01 9.90312219e-01 5.53175092e-01 5.66828549e-01
-1.39902329e+00 -7.15804040e-01 5.41642793e-02 -8.85268077e-02
-1.06878126e+00 -1.14135301e+00 1.22148490e+00 -3.18512261e-01
5.82958221e-01 2.05343083e-01 6.36607170e-01 9.09836471e-01
-6.38888836e-01 1.04531741e+00 1.00988996e+00 -6.70866847e-01
2.05561355e-01 8.86198431e-02 -2.70641088e-01 4.94341671e-01
6.33021742e-02 2.99935639e-02 -8.55555534e-01 -1.71118706e-01
8.04321289e-01 -2.78931528e-01 -2.78454691e-01 -3.80539238e-01
-1.34038043e+00 5.88309407e-01 1.88721512e-02 2.11896658e-01
-2.83257931e-01 2.13577837e-01 3.93441767e-02 5.15658371e-02
4.43009764e-01 5.63544154e-01 -2.72517502e-01 -2.90576190e-01
-1.85360873e+00 2.13440344e-01 1.05349636e+00 1.07363296e+00
6.06788993e-01 6.11678779e-01 5.70183881e-02 1.18904018e+00
3.51884335e-01 6.87674522e-01 6.34468853e-01 -1.30275059e+00
1.90533012e-01 -1.38395071e-01 8.75234380e-02 -2.58034527e-01
-9.41917449e-02 -1.01137578e+00 -7.66422808e-01 1.73530653e-01
6.34024620e-01 -4.11842912e-01 -9.73496079e-01 1.79698884e+00
1.44815788e-01 7.24975765e-01 2.50110209e-01 7.60453999e-01
3.19839299e-01 5.82940042e-01 -4.69839722e-01 -4.53167051e-01
7.95607090e-01 -1.39474869e+00 -8.59271884e-01 -4.18897480e-01
5.85813448e-02 -1.06117773e+00 8.52812946e-01 5.64020514e-01
-1.44017959e+00 -6.77081168e-01 -1.21824944e+00 2.64873028e-01
1.69301748e-01 2.42465764e-01 5.91568053e-01 9.07334626e-01
-7.52656639e-01 8.79238009e-01 -1.12621093e+00 1.96376629e-02
2.47332171e-01 3.86264920e-01 1.30799443e-01 2.55474627e-01
-8.60666096e-01 4.99746233e-01 1.63894385e-01 8.94098356e-02
-1.23336434e+00 -6.36637688e-01 -8.29289138e-01 1.44663900e-01
4.03296649e-01 -6.77464306e-01 1.61778986e+00 -9.52968955e-01
-2.11624455e+00 1.99880704e-01 -4.80296433e-01 -4.34772551e-01
1.54892966e-01 -3.14440787e-01 -4.85410541e-01 2.80779451e-01
-3.16016972e-02 5.13111413e-01 1.30686247e+00 -1.41983986e+00
-5.56062579e-01 2.03215361e-01 -2.01724336e-01 3.11015606e-01
-3.68950188e-01 -7.03171715e-02 -3.18501502e-01 -8.70378494e-01
1.28304094e-01 -9.78192985e-01 -3.04039627e-01 -4.01143461e-01
-5.42163014e-01 2.48951524e-01 5.94237030e-01 -6.51053131e-01
1.10337710e+00 -2.19599795e+00 3.07966292e-01 9.84349474e-02
-1.01783715e-01 2.38812834e-01 -2.56746233e-01 2.12598085e-01
4.94264103e-02 -8.68087336e-02 -5.63795388e-01 -8.30628037e-01
1.87708855e-01 8.49440023e-02 -6.02920055e-01 1.94862559e-01
3.06774795e-01 5.69562137e-01 -1.05119193e+00 -3.58711243e-01
2.32110023e-01 3.31366450e-01 -8.73120666e-01 3.19908589e-01
-3.77585083e-01 8.23449612e-01 1.93389207e-01 4.38333988e-01
4.52998549e-01 -6.61430135e-02 2.61460125e-01 4.77655344e-02
2.46339235e-02 8.09507847e-01 -1.58373415e+00 2.15012479e+00
-6.25084698e-01 5.27250290e-01 3.93066317e-01 -9.07856822e-01
7.32594609e-01 5.93226492e-01 5.07498264e-01 2.98744023e-01
-1.08671926e-01 5.36876798e-01 3.51807415e-01 -1.76471442e-01
3.29872847e-01 -5.83629966e-01 7.40351826e-02 6.69103980e-01
6.10851228e-01 -5.83228469e-01 3.81243050e-01 6.39194995e-02
7.95214772e-01 4.59846884e-01 2.01539189e-01 7.87747428e-02
3.40763986e-01 -1.33397505e-01 6.87592030e-01 7.34601617e-01
1.75135016e-01 8.98736119e-01 1.94829315e-01 5.62327087e-01
-7.62021840e-01 -1.36639678e+00 8.70765597e-02 9.69563305e-01
-1.67663977e-01 -5.88829279e-01 -7.94678032e-01 -5.72152436e-01
-2.12632373e-01 8.01116765e-01 2.10222170e-01 3.66634540e-02
-7.00270295e-01 -7.23717809e-01 8.15982163e-01 5.06219089e-01
4.82289195e-02 -6.83512986e-01 -1.82240337e-01 4.60747480e-01
-1.74323201e-01 -9.24146056e-01 -8.70335758e-01 4.63109106e-01
-1.15236104e+00 -7.56535769e-01 -1.03042269e+00 -6.30948067e-01
4.01879549e-01 2.87812382e-01 8.18433940e-01 -3.38041306e-01
1.31585315e-01 3.42308819e-01 -1.16116084e-01 -5.09148002e-01
-7.75897026e-01 2.49407053e-01 3.37938249e-01 1.76880822e-01
-7.19235390e-02 -1.13575363e+00 -2.83615291e-01 1.47647083e-01
-8.57445121e-01 -8.56942032e-03 5.92843175e-01 8.96513164e-01
5.00771523e-01 2.48212010e-01 1.05294967e+00 -7.92398095e-01
5.56416810e-01 -3.28975201e-01 -4.80883032e-01 -1.38308648e-02
-5.05469561e-01 1.71679318e-01 7.99628854e-01 -7.31740117e-01
-1.24742913e+00 3.18537712e-01 -1.35521293e-01 -8.94376636e-01
-3.12694281e-01 4.27943707e-01 -3.73308003e-01 3.12964439e-01
6.72589660e-01 1.79928392e-01 -1.20555803e-01 -8.93154263e-01
7.52634823e-01 7.42098451e-01 8.77912164e-01 -5.66492438e-01
1.12569666e+00 2.85202414e-01 -1.82518020e-01 -9.28932309e-01
-9.34851646e-01 -6.32517457e-01 -8.30342114e-01 9.45014954e-02
5.27764142e-01 -1.01691127e+00 -8.75660777e-02 5.68962693e-01
-1.05939960e+00 -5.16271949e-01 -5.98149896e-01 9.12123501e-01
-7.04015017e-01 4.24537271e-01 -5.98266840e-01 -1.09557939e+00
-3.91370505e-02 -8.35055709e-01 7.76273012e-01 2.51977533e-01
-2.56232440e-01 -1.19584870e+00 2.42034838e-01 3.69024813e-01
4.01616871e-01 -3.35583657e-01 5.70089638e-01 -8.64705682e-01
-4.76894945e-01 -8.95955339e-02 4.72990990e-01 5.63386559e-01
6.81834102e-01 -1.89694017e-01 -1.38330507e+00 -1.21128485e-01
2.68433005e-01 -3.14440489e-01 9.93987262e-01 3.96159172e-01
8.54085147e-01 -2.59240538e-01 -1.79563254e-01 7.53092885e-01
7.63346374e-01 -5.19753024e-02 2.31201679e-01 -3.42474550e-01
7.99126804e-01 5.16195059e-01 1.78109676e-01 2.92071998e-01
6.73758015e-02 5.93408167e-01 -1.05574809e-01 -6.74667060e-02
-5.91370285e-01 -6.10696673e-01 7.02673018e-01 1.36332953e+00
-9.02434140e-02 -1.58063740e-01 -4.36587423e-01 6.65240407e-01
-1.60683560e+00 -1.04036355e+00 8.36513489e-02 2.29101133e+00
1.25297809e+00 9.32585448e-02 4.10152525e-01 4.68198240e-01
7.02601790e-01 2.44312733e-01 -6.16752505e-01 2.16191083e-01
-1.04329094e-01 5.86035192e-01 2.18915433e-01 6.71905160e-01
-1.12486708e+00 9.24631417e-01 7.05138493e+00 8.87862563e-01
-1.11081517e+00 3.16382319e-01 -4.03510146e-02 -6.10129416e-01
-5.54406285e-01 3.51021588e-01 -6.11666799e-01 3.02459002e-01
1.03556192e+00 -9.93302017e-02 9.50650215e-01 5.36540568e-01
1.36670917e-01 9.48566571e-02 -1.34568346e+00 9.42895055e-01
2.37674594e-01 -9.28468883e-01 -1.92765921e-01 -1.33980647e-01
8.58304143e-01 -2.08543926e-01 -7.56476372e-02 2.49686077e-01
3.77738774e-01 -9.26272750e-01 8.91853034e-01 2.60527402e-01
6.19158566e-01 -5.82751989e-01 1.77531242e-01 6.33474708e-01
-1.09957290e+00 4.28387038e-02 -1.94122329e-01 1.67245567e-02
7.18662560e-01 7.81707823e-01 -7.72517383e-01 6.27665877e-01
1.33250669e-01 7.31435895e-01 -1.36759177e-01 1.20221686e+00
-6.53411508e-01 1.15361249e+00 -4.92357880e-01 4.11612093e-01
-1.96391195e-01 -1.06362537e-01 1.00684857e+00 1.29199636e+00
5.69647193e-01 -3.18551481e-01 2.09926456e-01 1.01220381e+00
-5.93004711e-02 -1.96346611e-01 -3.66058499e-01 -3.52117389e-01
4.39615428e-01 1.34523070e+00 -5.67291498e-01 -4.78233337e-01
-3.32377136e-01 1.06754935e+00 1.50334314e-01 5.13695717e-01
-6.22528791e-01 -3.39819491e-01 5.29168069e-01 6.70064706e-03
5.44785500e-01 -4.72353846e-01 -1.91851467e-01 -1.57490051e+00
-3.97190630e-01 -1.01727271e+00 2.03842685e-01 -5.96300304e-01
-1.46883631e+00 5.01756310e-01 8.59328061e-02 -1.37249064e+00
-8.40111613e-01 -3.71000916e-01 -7.85032809e-01 9.98269856e-01
-1.53298259e+00 -9.68989193e-01 1.41288027e-01 4.95805740e-01
7.79390514e-01 -3.41607124e-01 8.88550937e-01 2.63644308e-01
-4.37309653e-01 5.51743388e-01 7.67592490e-02 1.38126537e-02
9.77054894e-01 -1.44041657e+00 2.11470410e-01 9.99264359e-01
9.72249806e-01 7.30076015e-01 6.26618266e-01 -5.32091141e-01
-1.20626557e+00 -9.59576011e-01 5.64425707e-01 -1.86304182e-01
6.42259955e-01 -5.02658844e-01 -8.63869786e-01 5.93064785e-01
5.22423446e-01 -8.96731317e-02 1.21257687e+00 2.36188725e-01
-2.32418105e-01 -1.15813658e-01 -5.29435515e-01 5.59276760e-01
1.19508827e+00 -6.67953193e-01 -7.87967265e-01 1.65400580e-01
4.83463824e-01 -3.75545979e-01 -4.83965307e-01 1.52129799e-01
4.09123361e-01 -8.37789536e-01 9.72488582e-01 -4.95029509e-01
1.04170881e-01 -5.98668456e-01 -1.82231609e-02 -1.72210729e+00
-2.98948735e-01 -1.15093148e+00 -5.31066418e-01 1.50157857e+00
7.37575173e-01 -4.67106491e-01 5.65594375e-01 9.63491946e-02
-3.78131300e-01 -1.77024364e-01 -6.86337292e-01 -1.09797931e+00
1.24649070e-01 -6.27908826e-01 5.42343020e-01 7.92113662e-01
2.76882321e-01 6.12531066e-01 -5.72343707e-01 3.46148968e-01
8.04425061e-01 4.72803503e-01 8.91275525e-01 -1.03970301e+00
-1.26818717e+00 -6.47504032e-01 3.16775441e-01 -1.48138738e+00
3.68700147e-01 -1.18930507e+00 3.94006312e-01 -1.40411627e+00
-6.05145097e-02 -6.42659783e-01 -4.41050619e-01 3.73640239e-01
-4.09618199e-01 2.07943365e-01 4.89261478e-01 4.65916157e-01
-2.82021195e-01 6.83241010e-01 1.01356411e+00 -2.05789730e-01
-6.86367035e-01 4.29566920e-01 -7.75098681e-01 1.06245828e+00
7.88104117e-01 -7.06459820e-01 -6.98375881e-01 -3.80813897e-01
-1.57889396e-01 2.38536686e-01 2.40997642e-01 -1.24237072e+00
3.64437819e-01 -1.58793971e-01 3.81469250e-01 -3.61123145e-01
6.61042809e-01 -4.07217205e-01 9.01184604e-02 7.11517781e-02
-3.51596057e-01 -7.62957156e-01 2.22670272e-01 5.94336510e-01
-4.07331645e-01 -6.61348701e-01 5.33339977e-01 -3.85223851e-02
8.20772871e-02 1.16799831e-01 -4.70044702e-01 1.86413065e-01
3.16169083e-01 1.81104422e-01 2.62155384e-01 -7.42696643e-01
-8.69005024e-01 -2.18325227e-01 3.34303439e-01 1.14783846e-01
3.68094057e-01 -1.35699785e+00 -7.44416714e-01 2.43176460e-01
-3.27577829e-01 2.47352958e-01 7.84292072e-02 8.83605361e-01
8.65137875e-02 2.75969207e-02 2.83563644e-01 -5.05614936e-01
-8.20385575e-01 5.70260346e-01 3.57794374e-01 -2.48492911e-01
-3.25089276e-01 7.05421448e-01 1.52492195e-01 -5.66860557e-01
1.58921391e-01 -1.20107964e-01 4.88588661e-02 -2.78988294e-02
4.23102558e-01 4.14287269e-01 -2.32300088e-01 -3.57401103e-01
-1.67084545e-01 5.08536041e-01 4.00370121e-01 -9.56162274e-01
9.76568878e-01 -6.81135152e-03 1.03772447e-01 8.35106492e-01
5.73701203e-01 9.08497512e-01 -1.39890110e+00 -2.74575561e-01
-7.07038864e-02 -3.10152292e-01 -1.31419808e-01 -6.35259986e-01
-8.66114736e-01 1.24588776e+00 3.06092910e-02 5.78514710e-02
1.25461829e+00 -3.82500216e-02 8.14985752e-01 3.11918199e-01
4.41990383e-02 -8.71155143e-01 9.57496315e-02 4.01860893e-01
7.10796952e-01 -8.13483834e-01 -2.12932631e-01 -5.57055056e-01
-4.63093668e-01 1.13821721e+00 3.28634828e-01 -8.55181664e-02
7.86010563e-01 5.66896319e-01 1.74655050e-01 3.38612378e-01
-6.48158789e-01 -4.67745692e-01 6.06009781e-01 5.67314327e-01
5.61333895e-01 -6.66870922e-03 2.01486498e-01 1.04167271e+00
-3.97731960e-01 6.23919070e-03 4.91175473e-01 7.43296206e-01
-3.92515570e-01 -1.65864408e+00 -6.74815178e-01 3.80569220e-01
-4.37921375e-01 -4.31755722e-01 -2.42002770e-01 1.10259913e-01
1.20818570e-01 1.45470083e+00 -5.61551563e-02 -2.23732799e-01
-2.36316323e-02 5.98772168e-01 7.61687219e-01 -1.06079304e+00
-2.45922089e-01 1.02441239e+00 1.39164120e-01 1.49233252e-01
-6.88849628e-01 -8.33642006e-01 -1.27357209e+00 1.79489687e-01
-5.55149198e-01 2.93465525e-01 4.55484092e-01 9.54404652e-01
3.17605913e-01 7.28589892e-01 5.74150443e-01 -1.12967432e+00
-6.29133105e-01 -1.32422662e+00 -6.43741727e-01 5.11535406e-02
4.74673271e-01 -5.09367764e-01 -6.48476541e-01 5.63666999e-01] | [15.479768753051758, 5.589348793029785] |
2a105ee0-8c9c-4cc5-8e59-c10f61dfc68a | videochat-chat-centric-video-understanding | 2305.06355 | null | https://arxiv.org/abs/2305.06355v1 | https://arxiv.org/pdf/2305.06355v1.pdf | VideoChat: Chat-Centric Video Understanding | In this study, we initiate an exploration into video understanding by introducing VideoChat, an end-to-end chat-centric video understanding system. It integrates video foundation models and large language models via a learnable neural interface, excelling in spatiotemporal reasoning, event localization, and causal relationship inference. To instructively tune this system, we propose a video-centric instruction dataset, composed of thousands of videos matched with detailed descriptions and conversations. This dataset emphasizes spatiotemporal reasoning and causal relationships, providing a valuable asset for training chat-centric video understanding systems. Preliminary qualitative experiments reveal our system's potential across a broad spectrum of video applications and set the standard for future research. Access our code and data at https://github.com/OpenGVLab/Ask-Anything | ['Yu Qiao', 'LiMin Wang', 'Yali Wang', 'Ping Luo', 'Wenhai Wang', 'Yizhuo Li', 'Yi Wang', 'Yinan He', 'Kunchang Li'] | 2023-05-10 | null | null | null | null | ['video-understanding'] | ['computer-vision'] | [-4.94097441e-01 2.89240628e-02 -4.39340383e-01 -4.79868859e-01
-4.48330671e-01 -5.04155278e-01 4.51625198e-01 -1.59828603e-01
2.29553673e-02 5.78394771e-01 7.65931189e-01 -5.67673564e-01
9.49545279e-02 -3.61064285e-01 -9.73149061e-01 -1.53013747e-02
-2.06485942e-01 2.27460265e-01 2.74059772e-01 -3.00058097e-01
4.17949170e-01 -2.63842881e-01 -1.55197430e+00 9.12215054e-01
5.15586555e-01 4.35471386e-01 2.93910950e-01 1.14151299e+00
3.38160098e-02 2.01693535e+00 -5.87967634e-01 -3.72801930e-01
-3.77569884e-01 -5.64181924e-01 -1.32140779e+00 -2.79742032e-01
5.26278913e-01 -8.47220242e-01 -9.10045028e-01 4.68909264e-01
7.54017979e-02 3.81853491e-01 1.34361744e-01 -1.82215166e+00
-1.02436674e+00 9.44519818e-01 7.10804313e-02 7.43664324e-01
1.00472498e+00 7.40793705e-01 8.75461519e-01 -7.26875961e-01
9.05322671e-01 1.24310851e+00 5.07053435e-01 7.19782948e-01
-5.69921017e-01 -7.67337203e-01 3.62381160e-01 8.86235535e-01
-1.02991390e+00 -5.11345625e-01 4.17468965e-01 -5.24037778e-01
1.29731536e+00 9.65616107e-02 9.20790315e-01 1.75998080e+00
1.64941698e-01 1.25145936e+00 3.89029950e-01 -8.60305503e-02
2.05179267e-02 -1.67459652e-01 2.30835319e-01 9.85946476e-01
-5.08690178e-01 -2.52676547e-01 -1.32945752e+00 3.56634408e-01
9.47631061e-01 2.06006274e-01 -4.96522307e-01 -5.57968132e-02
-1.54794288e+00 5.49406528e-01 1.85750023e-01 2.07927346e-01
-9.64801162e-02 7.03284979e-01 5.31829476e-01 4.89196777e-01
2.15375245e-01 3.48765343e-01 -3.15761000e-01 -1.22507274e+00
-4.73035693e-01 1.11844182e-01 1.06256282e+00 1.24714410e+00
3.63259107e-01 -1.03093563e-02 -7.30860755e-02 3.97571564e-01
2.33889922e-01 -1.52661011e-01 2.13353992e-01 -1.72528231e+00
4.08385575e-01 5.72332621e-01 -2.04105899e-01 -7.43197858e-01
-5.12072258e-02 3.53265256e-01 -2.71587260e-02 -4.50404942e-01
4.19396400e-01 -2.18570590e-01 -3.19719046e-01 1.67785323e+00
-4.53035571e-02 1.05009747e+00 -1.32245198e-03 8.83013964e-01
1.31736958e+00 7.71238089e-01 2.79695302e-01 5.96699491e-02
1.26399553e+00 -1.41553485e+00 -9.17089581e-01 1.02467857e-01
1.07404709e+00 -4.49876398e-01 1.49064529e+00 4.01850194e-01
-1.19950533e+00 -3.68396699e-01 -6.30596578e-01 -3.89190942e-01
-4.13552880e-01 -1.97942331e-01 8.30547929e-01 1.62886698e-02
-1.31262171e+00 4.06211942e-01 -1.16269863e+00 -7.71184027e-01
3.44457388e-01 -5.68101741e-02 -2.75869429e-01 -7.94285387e-02
-1.21218467e+00 6.54584229e-01 2.92277277e-01 -1.67804912e-01
-1.38126338e+00 -1.04540133e+00 -8.64172101e-01 1.26784220e-01
7.39029288e-01 -6.56567633e-01 1.84064305e+00 -8.38578939e-01
-1.55195975e+00 5.54427147e-01 -1.79709166e-01 -2.91418254e-01
2.69224524e-01 -6.64904654e-01 -2.06250206e-01 6.54204071e-01
-8.02735910e-02 7.81835616e-01 2.72000492e-01 -8.24487209e-01
-5.35344660e-01 1.65505275e-01 6.53155923e-01 2.02344701e-01
-5.13715982e-01 2.94856548e-01 -8.27910602e-01 -2.89172530e-01
-6.53407514e-01 -6.23318255e-01 2.52140850e-01 -1.81807816e-01
-7.14554116e-02 -4.12321538e-01 1.22480118e+00 -7.90714622e-01
1.41310859e+00 -1.99142838e+00 4.18888658e-01 -5.73764384e-01
5.19689322e-01 -1.37334689e-01 -2.72343099e-01 1.06696427e+00
-1.87472120e-01 4.36598033e-01 5.19605398e-01 -1.30443648e-01
-6.12467006e-02 8.75585154e-02 -2.02978164e-01 -6.60406519e-03
2.46961474e-01 9.77941453e-01 -1.28604913e+00 -6.01759911e-01
3.87936413e-01 6.10490978e-01 -9.25305188e-01 5.46748281e-01
-6.31753743e-01 5.49131513e-01 -7.00935185e-01 7.29692459e-01
-1.65017381e-01 -5.42787552e-01 1.35025561e-01 -7.36321956e-02
-1.99715067e-02 4.07871962e-01 -6.05946779e-01 2.07870722e+00
-4.92261261e-01 1.42226160e+00 6.28149777e-04 -8.09339762e-01
3.41058761e-01 7.43094146e-01 5.85351229e-01 -4.35435593e-01
-4.36379202e-02 -4.19103265e-01 -3.22339743e-01 -1.23059475e+00
4.84634817e-01 4.81505036e-01 -6.31707534e-02 7.03875721e-01
5.66970944e-01 1.63443625e-01 2.38028944e-01 1.01232803e+00
1.50946105e+00 5.11667728e-01 -8.32674876e-02 1.70259792e-02
1.06417894e-01 3.41145515e-01 8.14264640e-02 5.79745233e-01
-4.23665732e-01 2.95773059e-01 8.44805658e-01 -5.81745267e-01
-6.29288137e-01 -1.01636708e+00 3.84197950e-01 1.44851089e+00
3.56536567e-01 -1.02676916e+00 -8.71059835e-01 -6.20530486e-01
-3.17919314e-01 8.65291536e-01 -5.79675674e-01 -2.00521141e-01
-6.47453427e-01 1.68604642e-01 6.26712203e-01 8.02782238e-01
4.11092043e-01 -1.23859203e+00 -4.75957334e-01 -1.18290126e-01
-8.89041543e-01 -1.26183283e+00 -5.57290375e-01 -1.69129550e-01
-6.25909507e-01 -1.38323689e+00 -2.09747665e-02 -6.50051475e-01
2.96785742e-01 5.53696632e-01 1.56340230e+00 6.72853231e-01
-1.86793432e-01 1.21381974e+00 -6.58805490e-01 -1.00191772e-01
-4.19078737e-01 -5.71859442e-02 -7.17361346e-02 -5.47482431e-01
5.93204021e-01 -6.18853092e-01 -6.04088962e-01 5.64394414e-01
-7.07476139e-01 3.84765565e-01 -4.45461161e-02 6.49788916e-01
-3.25082541e-02 -3.65439862e-01 4.36865807e-01 -5.29508591e-01
6.76865995e-01 -9.57494497e-01 -1.43076286e-01 1.75818607e-01
4.20421958e-02 -3.34503233e-01 5.31769037e-01 -4.54792976e-01
-1.24623752e+00 -3.47094655e-01 5.17345704e-02 -8.17734778e-01
-4.34364498e-01 5.13355613e-01 8.87273028e-02 2.54347235e-01
4.22177255e-01 1.57540947e-01 3.61180604e-02 -5.90503551e-02
5.10050237e-01 5.22375047e-01 8.40297461e-01 -8.20279181e-01
3.11891049e-01 1.88581020e-01 -8.82116437e-01 -7.77414858e-01
-5.20832002e-01 -5.03861010e-01 -6.43818498e-01 -1.00668049e+00
1.17760360e+00 -1.39398444e+00 -1.54378366e+00 9.03079212e-02
-1.22428608e+00 -1.08289504e+00 1.40950352e-01 5.83366871e-01
-1.00573158e+00 1.51145801e-01 -1.20992231e+00 -4.30025041e-01
1.75161660e-01 -1.16356277e+00 9.52383637e-01 3.47388297e-01
-6.22160137e-01 -1.30870903e+00 -5.86210527e-02 1.02130997e+00
2.11674273e-01 -8.99228081e-02 6.40369534e-01 -6.69506371e-01
-1.22778940e+00 1.12441078e-01 -8.48218799e-02 -1.54286563e-01
-2.92342514e-01 5.35315394e-01 -7.32041359e-01 -2.01609097e-02
-2.64257818e-01 -8.93046260e-01 6.13588274e-01 1.18416965e-01
1.35747421e+00 -2.49535874e-01 -3.87492150e-01 5.64166307e-01
1.08815527e+00 3.32319200e-01 6.14838421e-01 3.39656264e-01
9.34138715e-01 4.16781187e-01 6.18687272e-01 5.48671424e-01
9.84222054e-01 3.85165662e-01 5.69106042e-01 3.43378842e-01
-9.64159891e-02 -6.27877116e-01 7.90385306e-01 1.17031276e+00
-3.20767164e-01 -6.98253095e-01 -1.06842864e+00 6.38924003e-01
-2.25154781e+00 -1.46104991e+00 -1.03838652e-01 1.37342846e+00
7.29506910e-01 -1.08731620e-01 1.80452451e-01 -5.33190906e-01
4.10352737e-01 1.46984428e-01 -3.85297507e-01 -4.05584782e-01
4.52910990e-01 -3.45664829e-01 -1.73046753e-01 4.77084219e-01
-6.15281463e-01 1.19096482e+00 6.58364677e+00 6.30362034e-01
-9.61961746e-01 1.03468716e-01 5.25868237e-01 -4.06959444e-01
-3.61378133e-01 4.59063388e-02 -5.16984463e-01 3.57020497e-01
1.35325456e+00 -3.06014836e-01 5.57899296e-01 7.69176841e-01
6.48280680e-01 -2.77253479e-01 -1.45236993e+00 6.54121697e-01
5.95760942e-02 -1.94036150e+00 -1.24563405e-03 -3.06984037e-01
2.75456786e-01 2.04713102e-02 -1.98049411e-01 7.33352125e-01
5.50644755e-01 -1.01233685e+00 7.66116381e-01 3.59996527e-01
5.37360430e-01 -2.76627630e-01 3.35251689e-01 3.60939920e-01
-1.20873404e+00 -7.71524012e-03 1.63402468e-01 -5.97681701e-01
3.01926613e-01 -4.04740125e-01 -7.56634116e-01 7.17848465e-02
1.02424395e+00 1.43197465e+00 -3.11158419e-01 6.53973877e-01
-2.52874076e-01 9.47630405e-01 -7.97848590e-03 -1.70240760e-01
2.09266871e-01 5.43995425e-02 3.44064534e-01 1.35083604e+00
6.71683848e-02 6.42398417e-01 2.08278492e-01 8.89535308e-01
-2.49594420e-01 -3.17963600e-01 -8.50978255e-01 -4.12155151e-01
7.12557614e-01 9.59946632e-01 -7.57805884e-01 -5.25054753e-01
-7.80546606e-01 8.22370350e-01 3.69806468e-01 5.95338881e-01
-1.41351652e+00 9.92164835e-02 9.68031526e-01 -5.06856814e-02
9.40463096e-02 -3.72394741e-01 1.12021059e-01 -1.53562772e+00
-1.81921214e-01 -1.01176047e+00 4.66564119e-01 -1.40916598e+00
-8.24216604e-01 2.81542540e-01 3.41194659e-01 -1.01549733e+00
-4.68265206e-01 -4.42671359e-01 -1.07668734e+00 1.12049609e-01
-1.05108869e+00 -1.01569784e+00 -7.04267561e-01 9.10642147e-01
1.21026170e+00 6.72699586e-02 7.46841311e-01 3.15699905e-01
-9.12373841e-01 2.98369557e-01 -3.77290994e-01 3.42418432e-01
8.89629602e-01 -1.01638758e+00 2.44631678e-01 6.04745269e-01
6.85484111e-02 7.44645417e-01 8.11480880e-01 -7.09299624e-01
-1.88264108e+00 -7.20324457e-01 4.80183274e-01 -1.04260945e+00
1.17477667e+00 -3.53476614e-01 -9.39540267e-01 1.47172165e+00
9.14835155e-01 -3.01369548e-01 8.29013586e-01 1.88682839e-01
-4.17006701e-01 4.53950018e-01 -6.01925135e-01 9.27547693e-01
1.46375036e+00 -9.01262045e-01 -5.38689256e-01 5.56257129e-01
1.16066730e+00 -6.10280752e-01 -1.31060910e+00 -6.92756101e-02
6.80362582e-01 -1.23780179e+00 1.01788068e+00 -8.93473506e-01
1.20941305e+00 6.08424246e-02 1.43210545e-01 -1.14709413e+00
-1.71687279e-03 -9.26417768e-01 -3.50284666e-01 1.25249255e+00
2.57393956e-01 -1.51647003e-02 8.29129815e-01 7.18508244e-01
-4.64120358e-01 -6.77899182e-01 -3.62188220e-01 -5.25077879e-01
-2.46830389e-01 -7.65737593e-01 1.75700888e-01 1.25616097e+00
8.40294361e-01 2.64037371e-01 -3.41512382e-01 1.02517270e-01
1.28823504e-01 -1.38885126e-01 9.31952417e-01 -5.55811822e-01
-3.01437944e-01 -5.10148525e-01 -2.04003930e-01 -1.13562000e+00
4.09756899e-01 -5.43324113e-01 -1.22563064e-01 -1.45125699e+00
4.16832358e-01 3.28129590e-01 4.68479730e-02 6.00162148e-01
1.68962721e-02 -3.49541306e-02 3.57221335e-01 1.50207490e-01
-1.44176769e+00 4.18351531e-01 1.27994752e+00 5.26194647e-02
-1.34543329e-01 -3.42582077e-01 -4.12034124e-01 8.36024046e-01
6.95331156e-01 -1.15396120e-01 -7.53210008e-01 -7.97817469e-01
1.07227564e-01 7.30842710e-01 5.02407968e-01 -8.48792195e-01
4.91357684e-01 -4.56532240e-01 -7.90626109e-02 -3.89828473e-01
4.88673091e-01 -5.41424572e-01 8.17206725e-02 1.66798681e-01
-8.28257620e-01 3.46514046e-01 5.29007256e-01 4.58418757e-01
-5.61889350e-01 2.46340528e-01 4.11926247e-02 -2.94069439e-01
-1.47148263e+00 1.02982730e-01 -8.67716491e-01 2.98127800e-01
1.27862799e+00 -3.18302661e-01 -8.71876955e-01 -1.22795928e+00
-7.17311740e-01 7.91122556e-01 4.29488301e-01 8.59873295e-01
8.24702024e-01 -1.09196115e+00 -5.25658309e-01 -1.21426277e-01
2.45839715e-01 -2.61790186e-01 4.71135497e-01 9.46237803e-01
-6.55352473e-01 3.84065360e-01 -3.54353547e-01 -6.96349084e-01
-1.38293076e+00 4.35167283e-01 2.13522494e-01 2.15298429e-01
-7.53632843e-01 9.65434492e-01 2.66881108e-01 -1.69440582e-01
5.37677884e-01 -5.85866034e-01 -1.20085225e-01 -2.88114130e-01
7.01581180e-01 3.16636771e-01 -4.72047418e-01 -1.98490262e-01
-2.06489518e-01 8.94755647e-02 1.05089761e-01 1.83173656e-01
1.35068917e+00 -4.10125971e-01 1.42813623e-01 7.08093107e-01
1.25425363e+00 -4.25760627e-01 -1.54546952e+00 5.94427660e-02
-1.11616679e-01 -3.18777144e-01 -1.08134590e-01 -6.75025940e-01
-9.68063235e-01 9.82726693e-01 -2.04052608e-02 2.15947896e-01
9.23853457e-01 2.61800498e-01 7.85403430e-01 5.12812495e-01
2.65055984e-01 -7.07655728e-01 8.61047089e-01 7.21281290e-01
7.59737015e-01 -1.40963328e+00 -2.68743217e-01 -4.07006711e-01
-8.15418005e-01 1.31649292e+00 1.19345093e+00 8.18937346e-02
5.29588401e-01 5.11483848e-01 2.29230270e-01 -5.04927635e-01
-1.79583049e+00 9.96307880e-02 -2.36275047e-01 5.74986696e-01
7.53769577e-01 -2.34139144e-01 2.58727074e-01 5.18911242e-01
-6.81370720e-02 6.45293355e-01 9.92961526e-01 8.46140921e-01
-9.57140476e-02 -5.54567158e-01 -1.33474097e-01 -6.10962808e-02
-2.27038726e-01 -1.25944659e-01 -3.01621467e-01 1.14184821e+00
-3.84621143e-01 1.01868284e+00 3.89727145e-01 -7.07984269e-01
4.40153480e-02 2.18051881e-01 2.51072317e-01 -4.63398963e-01
-7.02084303e-01 -3.24842513e-01 3.90268147e-01 -1.27162826e+00
-4.31267440e-01 -4.93891031e-01 -1.53588438e+00 -8.02897036e-01
1.09273709e-01 9.67240483e-02 3.00338119e-01 9.07481611e-01
5.63278854e-01 7.81548023e-01 1.79422662e-01 -8.16287220e-01
1.75679684e-01 -8.08829486e-01 8.53509232e-02 3.47831100e-01
1.04108095e-01 -5.17012060e-01 -2.66758382e-01 6.25945270e-01] | [10.45586109161377, 0.9872857928276062] |
84351885-d44b-4873-befe-27798e515191 | on-the-lifting-and-reconstruction-of | 2304.11860 | null | https://arxiv.org/abs/2304.11860v1 | https://arxiv.org/pdf/2304.11860v1.pdf | On the lifting and reconstruction of dynamical systems with multiple attractors | The Koopman operator provides a linear perspective on non-linear dynamics by focusing on the evolution of observables in an invariant subspace. Observables of interest are typically linearly reconstructed from the Koopman eigenfunctions. Despite the broad use of Koopman operators over the past few years, there exist some misconceptions about the applicability of Koopman operators to dynamical systems with more than one fixed point. In this work, an explanation is provided for the mechanism of lifting for the Koopman operator of a dynamical system with multiple attractors. Considering the example of the Duffing oscillator, we show that by exploiting the inherent symmetry between the basins of attraction, a linear reconstruction with three degrees of freedom in the Koopman observable space is sufficient to globally linearize the system. | ['Karthik Duraisamy', 'Shaowu Pan'] | 2023-04-24 | null | null | null | null | ['misconceptions'] | ['miscellaneous'] | [-9.84552279e-02 2.90853351e-01 1.25280526e-02 2.05137819e-01
1.97276235e-01 -6.59758806e-01 7.52409160e-01 -1.33251861e-01
-2.22511724e-01 8.28904271e-01 1.85177371e-01 -1.88573897e-01
-6.24046206e-01 -4.69275564e-01 -3.05702358e-01 -1.24485540e+00
-5.12253284e-01 9.02027637e-02 -1.44425277e-02 -5.93807578e-01
1.06965765e-01 7.66368866e-01 -1.20483673e+00 -1.74123943e-01
6.10842168e-01 5.39145768e-01 -2.01493680e-01 8.83439720e-01
4.82673585e-01 3.84678751e-01 -1.77972019e-01 3.60511482e-01
6.65540397e-01 -8.91891181e-01 -8.42758298e-01 -1.38169542e-01
2.02167422e-01 2.16814294e-01 -8.24806809e-01 1.30046356e+00
-1.58303604e-02 2.87029922e-01 6.91011071e-01 -1.02393270e+00
-6.24214768e-01 5.05822778e-01 1.35908976e-01 4.27968740e-01
2.84447253e-01 2.10453302e-01 1.35905635e+00 -6.38189256e-01
7.57337272e-01 1.08581269e+00 5.03003657e-01 5.23281515e-01
-1.73267663e+00 -8.77245143e-02 -4.63837057e-01 -5.67697063e-02
-1.75564694e+00 -1.30255461e-01 6.28759205e-01 -5.88931799e-01
8.68941009e-01 6.98741198e-01 1.00013471e+00 5.03875613e-01
1.12841380e+00 -8.68120492e-02 1.19690847e+00 -6.88847780e-01
2.12162912e-01 2.25122124e-01 5.64741492e-01 6.47109807e-01
4.85381931e-01 6.89287841e-01 -3.49864393e-01 -2.99673736e-01
8.85061383e-01 -7.90000409e-02 -4.00868565e-01 -9.38606679e-01
-1.15708542e+00 7.87520468e-01 4.30502355e-01 7.01368093e-01
-1.09063335e-01 9.64131355e-02 3.56235616e-02 5.62882304e-01
-6.58673123e-02 1.09334469e+00 -1.17051877e-01 6.23817928e-02
-5.04662156e-01 3.36015671e-01 8.89512956e-01 5.17550766e-01
9.41592813e-01 -5.58630042e-02 2.82617152e-01 -2.92803049e-01
8.23884755e-02 6.90613985e-01 3.41182560e-01 -7.52361059e-01
-1.07213721e-01 7.81298637e-01 1.97641566e-01 -7.86614537e-01
-7.28141367e-01 -2.67137796e-01 -9.91458178e-01 3.07011306e-01
5.47409832e-01 -1.39550507e-01 -4.16988432e-01 1.75496054e+00
5.78090968e-03 -2.96055824e-01 3.50511849e-01 9.48635280e-01
5.36473058e-02 8.45237970e-01 -5.70116222e-01 -3.63967836e-01
8.42390239e-01 -4.02674735e-01 -7.69531012e-01 1.63308948e-01
8.17570210e-01 -2.16783196e-01 7.74313807e-01 1.54023757e-02
-9.44171071e-01 -2.04198003e-01 -1.08312941e+00 2.20638976e-01
-3.14929307e-01 -1.46218911e-01 4.08264071e-01 4.48007554e-01
-9.91904795e-01 9.78686571e-01 -1.01344562e+00 -7.42173851e-01
-6.19952500e-01 5.63096881e-01 -6.72646403e-01 5.54005861e-01
-1.16730130e+00 1.33326674e+00 6.45367503e-01 2.62908548e-01
-4.96915042e-01 -4.88243103e-01 -5.50520599e-01 2.48741820e-01
6.71697110e-02 -3.89458567e-01 5.14331281e-01 -8.61940563e-01
-1.49088407e+00 4.12284315e-01 -7.14053884e-02 -4.39019412e-01
1.60659462e-01 2.37522244e-01 -3.70091170e-01 3.34414579e-02
-2.08113045e-01 -4.96234233e-03 5.41725874e-01 -6.93186283e-01
3.29942018e-01 -2.05893949e-01 1.48609906e-01 -7.77504519e-02
-2.40161419e-01 -1.74549848e-01 6.86248302e-01 -6.55450374e-02
5.04899681e-01 -1.52835560e+00 -2.82213092e-01 -6.50450408e-01
-2.90702015e-01 -2.25518160e-02 7.21749485e-01 -1.53212741e-01
1.47364891e+00 -2.24627733e+00 8.76233399e-01 4.49824154e-01
2.71420926e-01 -4.57709357e-02 1.33417591e-01 1.09652805e+00
-5.64369678e-01 1.60225123e-01 -3.88774127e-01 4.71834928e-01
-2.48575248e-02 1.44869372e-01 -6.89370751e-01 9.86207187e-01
1.72353104e-01 8.37546468e-01 -7.54142225e-01 -3.11695132e-02
2.16582298e-01 2.61283427e-01 -6.38822258e-01 -2.61752397e-01
3.74092370e-01 7.86156058e-01 -4.00450796e-01 9.85541642e-02
4.05910134e-01 -2.00653940e-01 2.70040363e-01 5.72317876e-02
-6.81237400e-01 2.08495453e-01 -1.33796322e+00 1.10691142e+00
2.69881804e-02 9.47952449e-01 -9.34155285e-02 -1.01302087e+00
5.91645837e-01 4.84311551e-01 7.88422823e-01 -3.51893395e-01
8.39738455e-03 3.59508932e-01 8.48976016e-01 -3.83648783e-01
5.14147758e-01 -5.64090192e-01 -4.35949951e-01 4.06995535e-01
3.41755837e-01 -1.51340336e-01 3.04204792e-01 1.49620324e-01
1.21459746e+00 -2.09604472e-01 7.53493786e-01 -1.19979668e+00
8.09678495e-01 1.01504438e-01 1.94347426e-01 6.75430477e-01
-2.43735164e-01 3.44533205e-01 7.58269727e-01 -6.23599529e-01
-1.25558960e+00 -1.11886537e+00 -8.17645073e-01 2.30758950e-01
3.73226941e-01 -7.72655010e-01 -6.91950440e-01 8.22314173e-02
1.95236020e-02 3.91082406e-01 -8.80940855e-01 -5.87857246e-01
-6.56268239e-01 -8.53612661e-01 3.87314945e-01 -1.60642684e-01
2.84717053e-01 -9.25775766e-01 -9.39896464e-01 8.20878670e-02
2.30018273e-01 -7.73235023e-01 -2.71657139e-01 2.40752101e-01
-9.61559534e-01 -9.42977548e-01 -3.36558133e-01 -7.00814202e-02
7.35068262e-01 -6.51884004e-02 5.35436392e-01 -1.30369216e-01
-4.55078900e-01 5.16052067e-01 1.14394769e-01 1.03365377e-01
-6.74168289e-01 2.19021123e-02 6.78418875e-01 6.95663765e-02
-1.67508125e-02 -5.52536011e-01 -2.85688192e-01 3.16652685e-01
-8.70161712e-01 -3.17099482e-01 2.03383416e-01 7.99850702e-01
1.18520364e-01 9.18132141e-02 1.61120385e-01 -2.29659170e-01
4.75808829e-01 -2.98034340e-01 -9.50410485e-01 -6.73650801e-02
-5.78567505e-01 6.94011450e-01 8.65934372e-01 -4.52063590e-01
-6.87508464e-01 8.33087862e-02 5.61926723e-01 -6.03835136e-02
3.40981245e-01 5.04523456e-01 2.03566208e-01 -5.76787293e-01
8.42096031e-01 3.33611101e-01 4.71514910e-02 -1.33970723e-01
2.19149709e-01 1.11915372e-01 1.47789642e-01 -3.74551415e-01
8.92210305e-01 6.22518837e-01 9.02656436e-01 -1.26630294e+00
-4.75232393e-01 -4.26953763e-01 -1.05109942e+00 -3.85396391e-01
8.34992588e-01 -3.86661589e-01 -1.03257024e+00 1.50124028e-01
-8.97321701e-01 2.08426099e-02 -9.05777454e-01 7.34148502e-01
-6.80725038e-01 2.16019154e-01 -3.70696694e-01 -1.06699049e+00
2.74157584e-01 -1.07778287e+00 4.82217908e-01 1.99398041e-01
-3.98024470e-01 -1.32934344e+00 7.40630746e-01 -3.81979614e-01
4.97959822e-01 1.17637552e-01 8.14076841e-01 -4.34057504e-01
-6.83153987e-01 -5.07695794e-01 3.62170368e-01 4.81234267e-02
-1.19526364e-01 8.73154327e-02 -6.29565656e-01 -5.46871066e-01
3.52283984e-01 1.59882411e-01 7.44656861e-01 4.03036743e-01
2.25979403e-01 -3.54505867e-01 -4.46228713e-01 2.43431717e-01
1.47710681e+00 2.41281420e-01 4.75729495e-01 1.44545779e-01
4.38888490e-01 6.10603690e-01 -1.25075862e-01 1.37025967e-01
-1.32390141e-01 6.56543732e-01 1.82742685e-01 3.27554941e-01
4.24395472e-01 -2.48132050e-02 7.95355856e-01 1.07394814e+00
-4.36731696e-01 9.50687453e-02 -8.20644915e-01 4.05960023e-01
-1.77293730e+00 -1.13985169e+00 -3.76508802e-01 2.52512074e+00
3.76368403e-01 -8.89427289e-02 -1.26648191e-02 -1.36514813e-01
5.38252950e-01 1.97023153e-01 -3.71089488e-01 -4.59075063e-01
-4.06997859e-01 -4.45930362e-02 7.53646970e-01 9.14132178e-01
-7.63869107e-01 6.08573794e-01 7.62638044e+00 1.94142491e-01
-1.17326343e+00 -4.58592130e-03 -3.53526771e-01 1.10302188e-01
-2.51345281e-02 8.81831706e-01 -6.42553329e-01 4.90251303e-01
1.05752409e+00 -6.94413066e-01 4.94722992e-01 4.28522229e-01
3.64550352e-01 -3.78038228e-01 -9.93664443e-01 5.09812176e-01
-1.96063221e-01 -1.16124654e+00 -2.65744686e-01 6.93648279e-01
9.15572345e-01 -1.43967301e-01 2.64591835e-02 -1.33168876e-01
7.99544379e-02 -9.73114014e-01 2.02632442e-01 9.13004577e-01
2.86651134e-01 -4.73474652e-01 4.78151858e-01 5.19324064e-01
-9.97392476e-01 -1.68859899e-01 -3.99198085e-01 -6.48835540e-01
1.92942247e-01 4.73803729e-01 -3.15615058e-01 2.46919096e-01
2.10989460e-01 6.31078839e-01 -3.26932788e-01 8.27956021e-01
2.19694629e-01 3.96821856e-01 -6.10220969e-01 3.75884883e-02
3.20272088e-01 -1.00904047e+00 1.04664218e+00 6.46026433e-01
4.63609338e-01 4.65576261e-01 -3.27301741e-01 1.16649723e+00
5.05838811e-01 -1.28006041e-01 -9.33967531e-01 -4.17074531e-01
-2.49738917e-01 1.15333200e+00 -7.19837725e-01 -5.17378375e-02
-1.44545794e-01 6.18575752e-01 4.54440191e-02 3.26832980e-01
-4.86393780e-01 -3.83703500e-01 7.46073663e-01 1.67825222e-01
1.85241122e-02 -6.71117544e-01 6.04103245e-02 -1.62243032e+00
-1.60511091e-01 -3.72540087e-01 -9.71805677e-02 -5.15285909e-01
-6.52033627e-01 3.87898207e-01 3.31345618e-01 -1.12840497e+00
-2.52968758e-01 -6.80006087e-01 -6.96714938e-01 8.49706948e-01
-4.75193858e-01 -4.61614788e-01 2.02565253e-01 2.73409128e-01
-3.42134118e-01 -6.32569864e-02 9.53661621e-01 -1.76730469e-01
-6.74552381e-01 -2.03581750e-01 5.88920176e-01 -3.35866362e-01
2.39989519e-01 -1.46566701e+00 1.46555096e-01 8.62572253e-01
1.43265575e-01 9.82940495e-01 1.36599219e+00 -4.05585229e-01
-1.68939590e+00 -3.66031051e-01 8.65297914e-01 -6.38798892e-01
1.09619081e+00 -4.15866286e-01 -7.94200182e-01 9.72790778e-01
3.28776568e-01 4.45622541e-02 2.27667972e-01 -1.58719406e-01
2.28422731e-01 1.66766383e-02 -7.36501753e-01 7.15368509e-01
8.44633102e-01 -6.32970870e-01 -4.64188486e-01 4.54788595e-01
2.80505061e-01 -1.71231568e-01 -9.59806442e-01 1.67469919e-01
6.07512176e-01 -1.02763379e+00 7.43893921e-01 -8.19267035e-01
-3.73490714e-02 -3.52096975e-01 3.69675010e-02 -1.38869750e+00
-4.46853101e-01 -1.10788155e+00 2.26251066e-01 4.16822046e-01
1.87514365e-01 -1.07956648e+00 1.92906603e-01 5.10910392e-01
1.36079147e-01 -3.03473204e-01 -1.37607372e+00 -9.62867498e-01
3.22041243e-01 9.99261141e-02 -2.61053536e-02 8.23789954e-01
9.24183190e-01 2.87674576e-01 -3.73017043e-01 3.81734222e-01
2.47739494e-01 1.45524278e-01 4.93329257e-01 -1.29076862e+00
-1.99052840e-01 -2.99693197e-01 -9.33013558e-01 -8.10482383e-01
3.33385140e-01 -1.11685944e+00 -1.65971264e-01 -7.91305542e-01
1.83059901e-01 -1.35318980e-01 -3.41298729e-01 -5.19899167e-02
4.97965783e-01 7.21873567e-02 3.08427870e-01 5.85495949e-01
-1.75142139e-01 4.70138311e-01 1.17331278e+00 3.13188791e-01
-5.12655556e-01 -8.71214122e-02 -1.49208859e-01 6.62136793e-01
6.26447678e-01 -6.39188111e-01 -6.92985579e-02 3.52234215e-01
6.09923482e-01 1.68240592e-01 5.88214874e-01 -1.08126736e+00
2.72746861e-01 -3.06330055e-01 -2.25926086e-01 -1.64938629e-01
5.39928615e-01 -6.94381356e-01 5.81891716e-01 8.66838276e-01
-4.43393052e-01 3.20403039e-01 6.19796477e-03 5.37064254e-01
-3.23317558e-01 -5.20878315e-01 8.75363648e-01 -3.44264656e-02
-2.48154938e-01 -1.52870208e-01 -9.21403825e-01 -2.05647737e-01
1.10202694e+00 5.37602901e-02 -2.45705366e-01 -2.45920822e-01
-1.01499462e+00 -8.91313255e-02 6.58885360e-01 -1.39837787e-01
9.96662378e-02 -1.34865952e+00 -1.52252272e-01 5.32527864e-01
-4.15276699e-02 -6.60681903e-01 2.27025062e-01 1.20287919e+00
-5.59510112e-01 1.06460917e+00 -5.62172472e-01 -6.13048434e-01
-1.06234109e+00 5.05904019e-01 9.49792624e-01 -3.18887144e-01
-6.06070876e-01 4.78114076e-02 3.58238280e-01 -3.79807919e-01
-7.49976873e-01 -6.18528068e-01 2.55417228e-01 -1.10431515e-01
1.41850755e-01 6.23011053e-01 -3.18791807e-01 -1.07558072e+00
-4.06552017e-01 7.83632755e-01 4.33853924e-01 -4.50268924e-01
9.15407181e-01 -1.10754550e-01 -6.55179560e-01 8.81174147e-01
1.16651595e+00 3.57647806e-01 -8.72716069e-01 1.27765194e-01
-5.44756278e-02 -1.63671166e-01 -1.01158477e-01 -1.28008842e-01
-4.01320398e-01 8.28393817e-01 2.90384680e-01 9.87747192e-01
6.71035290e-01 8.36088285e-02 1.45203650e-01 7.47649968e-01
2.06445321e-01 -6.97417021e-01 -3.36939842e-01 7.74367332e-01
1.07034767e+00 -8.01153481e-01 9.02508572e-02 -1.06085576e-01
-3.53334963e-01 1.48225105e+00 1.11449078e-01 -7.76951194e-01
7.53901422e-01 1.88263748e-02 -4.08881664e-01 -2.17450082e-01
-7.39779949e-01 -4.58938293e-02 4.30292875e-01 -2.38846894e-02
3.11990976e-01 2.96879530e-01 -6.47044659e-01 -3.45198549e-02
-3.15839767e-01 -4.18255925e-01 1.34104967e+00 6.45987868e-01
-4.68736708e-01 -9.76912618e-01 -5.04763424e-01 4.42963839e-03
-1.46915819e-02 2.01069340e-01 -5.89229167e-01 1.20759773e+00
-1.92698583e-01 5.09364665e-01 9.36889648e-02 -1.13227233e-01
1.97570294e-01 3.52368414e-01 6.42136872e-01 -4.96640384e-01
-4.33942020e-01 -1.64339334e-01 -3.92817229e-01 -4.60030466e-01
-4.96718794e-01 -9.15185153e-01 -1.22354198e+00 -3.06370825e-01
-4.66832012e-01 4.27926391e-01 3.39868039e-01 1.20845175e+00
3.93824279e-01 2.87095517e-01 4.50325340e-01 -6.22780085e-01
-7.55764246e-01 -9.04359639e-01 -1.05224705e+00 2.11377636e-01
8.01491141e-01 -6.81789100e-01 -8.92928243e-01 -2.50798106e-01] | [6.052515506744385, 4.417259693145752] |
55a6733c-8fea-4266-bed4-a199c79b2217 | a-circular-window-based-cascade-transformer | 2208.14209 | null | https://arxiv.org/abs/2208.14209v1 | https://arxiv.org/pdf/2208.14209v1.pdf | A Circular Window-based Cascade Transformer for Online Action Detection | Online action detection aims at the accurate action prediction of the current frame based on long historical observations. Meanwhile, it demands real-time inference on online streaming videos. In this paper, we advocate a novel and efficient principle for online action detection. It merely updates the latest and oldest historical representations in one window but reuses the intermediate ones, which have been already computed. Based on this principle, we introduce a window-based cascade Transformer with a circular historical queue, where it conducts multi-stage attentions and cascade refinement on each window. We also explore the association between online action detection and its counterpart offline action segmentation as an auxiliary task. We find that such an extra supervision helps discriminative history clustering and acts as feature augmentation for better training the classifier and cascade refinement. Our proposed method achieves the state-of-the-art performances on three challenging datasets THUMOS'14, TVSeries, and HDD. Codes will be available after acceptance. | ['Lin Ma', 'Wei zhang', 'Bairui Wang', 'Weixin Luo', 'Shuqiang Cao'] | 2022-08-30 | null | null | null | null | ['online-action-detection', 'action-segmentation'] | ['computer-vision', 'computer-vision'] | [ 4.09964174e-01 -1.50432304e-01 -5.67432582e-01 -2.16774389e-01
-7.34622478e-01 -2.05997989e-01 4.61209148e-01 1.75246537e-01
-6.07374251e-01 4.81524378e-01 3.63699198e-01 -1.01063244e-01
1.50459930e-01 -4.46334749e-01 -5.60161233e-01 -7.33100712e-01
-4.19150591e-01 1.02876358e-01 8.78812015e-01 1.22638233e-01
2.29676858e-01 3.38709593e-01 -1.62660503e+00 5.95532417e-01
5.84672928e-01 1.38141382e+00 2.00670078e-01 8.46424639e-01
1.02218695e-01 1.35204732e+00 -4.08117086e-01 -3.87879610e-01
2.19722793e-01 -7.06565082e-01 -7.03363240e-01 4.68611956e-01
8.36444423e-02 -7.22790062e-01 -6.12972796e-01 5.95531940e-01
4.21434253e-01 3.07299942e-01 2.61901706e-01 -1.16835058e+00
-2.51759619e-01 6.29633486e-01 -6.49820626e-01 7.94600487e-01
3.41619521e-01 1.67972729e-01 1.14138997e+00 -6.36962354e-01
5.09063184e-01 9.06558990e-01 4.95125145e-01 5.07327378e-01
-6.34402215e-01 -5.59409916e-01 6.47286475e-01 8.33256245e-01
-1.01962221e+00 -5.16854465e-01 7.94313908e-01 -2.06219375e-01
8.62635076e-01 8.29304233e-02 1.15334034e+00 9.79599297e-01
1.27197087e-01 1.52342355e+00 4.86128002e-01 -2.84080237e-01
3.72887760e-01 -5.86968839e-01 -1.78804800e-01 7.47785568e-01
-4.25893307e-01 -1.86116114e-01 -7.24742353e-01 -6.20294027e-02
7.85528004e-01 2.55874395e-01 -1.15456112e-01 4.26027961e-02
-1.03593028e+00 5.75369060e-01 9.42673385e-02 2.48748973e-01
-5.75048506e-01 1.27610162e-01 5.12649477e-01 1.70336798e-01
6.60483301e-01 -1.57847509e-01 -5.04218817e-01 -6.08689129e-01
-1.02669132e+00 -3.49874981e-02 5.05414307e-01 7.69657135e-01
4.43007559e-01 -1.83427453e-01 -4.25933838e-01 5.49995363e-01
4.74372469e-02 -1.11221075e-01 5.44434845e-01 -1.23878539e+00
4.75793093e-01 6.92106962e-01 3.35560411e-01 -8.16973925e-01
-2.26749063e-01 -2.55681545e-01 -5.24391115e-01 -1.72808338e-02
3.97341460e-01 -6.01121001e-02 -7.21416533e-01 1.36657655e+00
7.17383146e-01 1.08204985e+00 -2.58694917e-01 8.18658829e-01
2.40794301e-01 5.97089171e-01 3.06511253e-01 -6.15528345e-01
1.04578686e+00 -1.40435123e+00 -7.37092912e-01 -8.04085210e-02
6.13700211e-01 -4.74727660e-01 7.08811581e-01 4.94275093e-01
-1.10885632e+00 -7.30455637e-01 -9.04165745e-01 -3.66892852e-02
1.04015552e-01 4.73956674e-01 8.15592110e-01 3.14258754e-01
-7.69676685e-01 8.91572475e-01 -1.45314693e+00 -2.37873241e-01
6.79104567e-01 2.08210662e-01 -9.88340098e-03 7.78106824e-02
-9.91726339e-01 3.56263608e-01 2.75605083e-01 3.53300780e-01
-1.12576056e+00 -3.71735692e-01 -6.52805746e-01 -1.87178090e-01
8.58449638e-01 -3.12109351e-01 1.50353062e+00 -1.18207943e+00
-1.79267645e+00 6.10110819e-01 -3.85369390e-01 -9.28555667e-01
8.54036570e-01 -5.66660583e-01 -3.50959688e-01 6.35640264e-01
-6.04651086e-02 4.21897531e-01 1.09497893e+00 -4.06233072e-01
-1.27015591e+00 -3.71720582e-01 3.51367474e-01 4.42260385e-01
-5.06519794e-01 1.80197403e-01 -9.03697670e-01 -6.61869347e-01
9.96676460e-02 -7.52717555e-01 -3.38026643e-01 1.61445692e-01
3.09540448e-03 -5.28919637e-01 7.18148470e-01 -7.80958652e-01
1.64610565e+00 -2.12429738e+00 1.15379103e-01 -1.71609759e-01
7.48765990e-02 2.49127343e-01 1.81902006e-01 3.65200609e-01
1.36638388e-01 -2.98799127e-01 5.76125123e-02 -5.56526959e-01
-1.74270809e-01 3.03597957e-01 -1.88119411e-01 6.92049205e-01
1.75820947e-01 6.49406433e-01 -9.37625110e-01 -8.93421948e-01
2.22516820e-01 1.93579838e-01 -5.69578409e-01 2.63725430e-01
-3.12133819e-01 6.78626120e-01 -6.92159832e-01 5.11579454e-01
2.56843954e-01 -4.75951523e-01 2.86138654e-01 -6.36775568e-02
-3.33949089e-01 1.45313010e-01 -1.22729099e+00 1.99754870e+00
-1.45957217e-01 3.69328499e-01 -1.37072191e-01 -1.18864775e+00
4.19045448e-01 3.98815721e-01 9.76750433e-01 -5.84023058e-01
1.27127990e-01 -1.49712250e-01 -2.62039244e-01 -5.47943354e-01
3.47654670e-01 2.35215291e-01 1.17448904e-01 4.42032933e-01
2.66326517e-02 7.02714443e-01 4.59842294e-01 2.53600955e-01
1.33663499e+00 5.18056691e-01 3.14820379e-01 3.02879661e-01
8.12199593e-01 -1.99422121e-01 8.66626263e-01 7.36835957e-01
-4.35768455e-01 3.28030080e-01 4.80997682e-01 -5.33671141e-01
-7.67737389e-01 -6.44910157e-01 2.64528662e-01 1.59579027e+00
2.78787762e-01 -5.66719949e-01 -6.53443396e-01 -9.79276597e-01
-3.10933858e-01 3.31155032e-01 -7.07038403e-01 -9.37772766e-02
-7.68660069e-01 -5.70486724e-01 2.53706604e-01 9.55636501e-01
5.89057803e-01 -1.08306372e+00 -8.65522563e-01 4.77533907e-01
-2.55297989e-01 -1.16326618e+00 -5.84378719e-01 -2.12282777e-01
-1.03769302e+00 -1.24639785e+00 -7.90794194e-01 -5.95467985e-01
4.01480794e-01 2.69941002e-01 9.04067278e-01 3.31076384e-01
7.84389675e-04 3.73576283e-01 -7.91457891e-01 -1.39804423e-01
-1.46487519e-01 2.55012568e-02 -1.56462476e-01 5.82468569e-01
1.55604318e-01 -6.29296064e-01 -1.11111176e+00 3.96777153e-01
-6.32775009e-01 1.75520003e-01 6.47968769e-01 5.65272987e-01
7.65572608e-01 1.62352070e-01 4.34668630e-01 -6.57077968e-01
-5.02787642e-02 -4.11747128e-01 -5.00301301e-01 2.28226498e-01
-2.79403836e-01 -2.59356409e-01 6.28902972e-01 -4.77530241e-01
-1.25982463e+00 2.94515103e-01 -3.18359315e-01 -5.85322082e-01
-6.72395974e-02 1.78203896e-01 -8.82788301e-02 2.81214386e-01
2.11459860e-01 3.18575621e-01 -3.81465197e-01 -7.51761258e-01
2.58633524e-01 5.50115883e-01 6.17248535e-01 -2.40363717e-01
3.97208005e-01 9.28648651e-01 -4.15307850e-01 -6.19168818e-01
-1.12841368e+00 -6.92337751e-01 -8.26694071e-01 -5.25104582e-01
1.02654207e+00 -1.12789142e+00 -8.83738041e-01 8.19486320e-01
-9.82423306e-01 -5.41555345e-01 -3.55802447e-01 5.12893140e-01
-6.03897095e-01 4.66892123e-01 -9.27611172e-01 -1.06189799e+00
-1.21958800e-01 -5.88679016e-01 1.09651852e+00 3.45764399e-01
-7.56584713e-03 -7.45768249e-01 -2.39790771e-02 3.44249964e-01
-2.05576539e-01 8.87933075e-02 3.13795000e-01 -6.66171372e-01
-6.48831606e-01 -3.06063503e-01 1.41838059e-01 3.49657089e-01
-9.13243666e-02 6.46266490e-02 -7.64210701e-01 -3.03576678e-01
6.19171448e-02 -1.56284079e-01 1.10134852e+00 4.08560634e-01
1.47807169e+00 -1.88331723e-01 -2.74339825e-01 3.62354130e-01
1.14818752e+00 4.22938377e-01 7.43853569e-01 1.42800853e-01
5.51325738e-01 2.05040798e-01 1.08820701e+00 9.03760433e-01
2.75610000e-01 5.61756313e-01 3.12229514e-01 1.22265778e-01
1.29161701e-01 -4.43010807e-01 5.48275888e-01 8.13528717e-01
-4.76996094e-01 -2.60944217e-01 -4.33556676e-01 5.44289768e-01
-2.21431994e+00 -1.27708995e+00 8.44445825e-02 2.16218209e+00
7.82275677e-01 5.18204153e-01 5.05964577e-01 4.04633135e-01
7.88468182e-01 5.08119941e-01 -7.52685606e-01 6.40664771e-02
2.04757348e-01 2.58074015e-01 2.55275667e-01 2.37426043e-01
-1.44892561e+00 8.70373726e-01 5.82336760e+00 1.03700995e+00
-8.06093514e-01 3.70377213e-01 8.35160732e-01 -2.64942855e-01
3.67378950e-01 1.17277883e-01 -8.63029957e-01 6.43529236e-01
7.48931885e-01 1.13276757e-01 2.00942829e-01 8.97075891e-01
5.07701397e-01 -5.47553599e-01 -1.04682279e+00 8.38959515e-01
1.29360640e-02 -1.35749757e+00 -1.84661508e-01 -2.17166722e-01
5.25737762e-01 -1.48037612e-01 -4.23923373e-01 3.05231541e-01
-4.95172329e-02 -3.34297746e-01 7.65758693e-01 5.56998670e-01
3.78934711e-01 -7.88877070e-01 4.14645582e-01 5.10648429e-01
-1.58678329e+00 -5.96662521e-01 -4.24010456e-02 -3.37036133e-01
3.48161966e-01 3.77135366e-01 -1.41755149e-01 5.32339990e-01
6.30080521e-01 1.20347714e+00 -4.48297590e-01 1.11680138e+00
-3.76507193e-01 7.81615555e-01 -1.90208435e-01 4.04838659e-02
1.46331042e-01 -5.58605455e-02 2.77384222e-01 9.90287244e-01
9.12424698e-02 4.78967756e-01 4.93671030e-01 3.32107916e-02
6.86522648e-02 5.79756452e-03 1.65151746e-03 -1.57441327e-03
2.09095001e-01 1.14879572e+00 -1.00274622e+00 -7.41079569e-01
-6.98079765e-01 1.28187430e+00 2.08306789e-01 1.41778961e-01
-1.28096712e+00 1.78339034e-02 4.25390273e-01 4.18808647e-02
7.88736105e-01 -3.15727204e-01 1.79500222e-01 -1.32317543e+00
1.17618255e-01 -5.77568352e-01 7.92834103e-01 -4.69905466e-01
-9.66267765e-01 1.64813682e-01 -2.19311595e-01 -1.52674103e+00
-1.47695795e-01 -2.54432470e-01 -8.09769809e-01 6.72065541e-02
-1.32098734e+00 -9.88846004e-01 -1.62522733e-01 8.27554286e-01
9.86395359e-01 2.91540712e-01 4.57189947e-01 5.51619351e-01
-8.69064152e-01 3.77008319e-01 -1.61934093e-01 3.75961453e-01
2.87969738e-01 -9.57151890e-01 2.37107158e-01 9.79041278e-01
2.89230376e-01 -9.91133451e-02 5.41490376e-01 -7.36093044e-01
-1.22049487e+00 -1.04177570e+00 6.79444373e-01 -1.14523590e-01
8.78653288e-01 -8.46605897e-02 -8.04045260e-01 7.55414009e-01
2.29079295e-02 3.45285311e-02 5.84478915e-01 -2.47505531e-01
2.13394389e-01 -2.59212226e-01 -6.32789314e-01 4.25316960e-01
1.52271044e+00 -3.60597998e-01 -3.89193892e-01 4.97685194e-01
6.44024134e-01 -5.29428780e-01 -8.07091773e-01 1.51941597e-01
6.25502110e-01 -1.13994300e+00 7.17440903e-01 -7.79974401e-01
4.33644056e-01 -1.37488797e-01 2.08975494e-01 -5.91243923e-01
-1.79465592e-01 -1.05717409e+00 -6.93892658e-01 1.08431768e+00
1.24293767e-01 -1.88358620e-01 1.08251595e+00 4.98940229e-01
-2.16264740e-01 -1.11213517e+00 -9.20262158e-01 -6.73780024e-01
-7.29321778e-01 -6.91505194e-01 2.50732630e-01 4.91213948e-01
-1.07042097e-01 1.30516410e-01 -6.12536967e-01 1.94882870e-01
5.16115010e-01 2.58538216e-01 6.59787536e-01 -8.51739883e-01
-6.24327481e-01 -2.24772561e-02 -4.68407393e-01 -1.64271927e+00
7.20970035e-02 -2.71864057e-01 -8.42548832e-02 -1.23785877e+00
1.66407406e-01 -7.58417770e-02 -6.39299393e-01 4.40731406e-01
-3.18943501e-01 2.83670276e-01 1.79132268e-01 2.82578260e-01
-1.44102240e+00 7.26576388e-01 1.27104902e+00 2.05264613e-01
-3.52084845e-01 5.66293240e-01 -3.31158280e-01 9.33196247e-01
6.84526622e-01 -5.08825958e-01 -4.43743199e-01 -2.40499690e-01
-4.80045490e-02 3.75549704e-01 3.20056915e-01 -1.31409490e+00
4.63805705e-01 -1.71781823e-01 4.59051520e-01 -7.40432203e-01
4.31424618e-01 -6.64071977e-01 -1.67589590e-01 5.26353538e-01
-5.02677023e-01 -3.30402330e-03 -2.51391202e-01 1.12144709e+00
-1.48787990e-01 -3.46897612e-03 6.13021374e-01 -1.50005534e-01
-1.17539871e+00 6.68889046e-01 -4.77754593e-01 1.67729124e-01
1.17931330e+00 -2.90693492e-01 1.41887173e-01 -4.04108047e-01
-1.09603155e+00 3.76876622e-01 1.45436183e-01 3.53628993e-01
5.15662730e-01 -1.25358331e+00 -5.42846620e-01 -3.05721108e-02
-2.38899544e-01 -1.79486915e-01 5.35954118e-01 1.28945541e+00
-1.99800625e-01 -6.49807006e-02 -5.59461154e-02 -5.65969229e-01
-1.37131560e+00 5.62381089e-01 7.24646077e-02 -4.34255630e-01
-8.00672352e-01 8.81360292e-01 -8.83807167e-02 6.57860219e-01
4.77545500e-01 -2.42213592e-01 -5.28905690e-01 3.22613806e-01
8.67250264e-01 6.75606966e-01 -2.10688636e-01 -4.78290558e-01
-2.82591641e-01 3.34274769e-01 6.17750315e-03 -8.26600716e-02
1.23966372e+00 -3.73767525e-01 2.03485966e-01 6.45430505e-01
1.13549078e+00 -2.40678087e-01 -1.97957170e+00 -3.14634800e-01
1.47123471e-01 -6.56312943e-01 -1.36427104e-01 -2.63074070e-01
-1.24591243e+00 6.68929279e-01 5.35534561e-01 1.82602838e-01
1.55566084e+00 9.67149436e-02 1.22847521e+00 1.46307752e-01
3.35145205e-01 -1.56396329e+00 2.82896519e-01 3.22897464e-01
5.22257268e-01 -1.02945316e+00 1.46288514e-01 -3.29106838e-01
-6.04495347e-01 9.55443144e-01 5.52048862e-01 -2.33435825e-01
6.55787587e-01 6.15575770e-03 -2.83252835e-01 7.57947192e-02
-9.30094242e-01 -4.49396282e-01 1.11197978e-01 2.54395097e-01
1.55043937e-02 -2.11456463e-01 -4.69356209e-01 5.99465847e-01
5.10246873e-01 4.21456963e-01 1.12982154e-01 1.18479204e+00
-5.41584194e-01 -9.11623299e-01 1.06774576e-01 4.44170952e-01
-6.82832062e-01 1.90035105e-01 -1.43237472e-01 5.10380924e-01
2.62236685e-01 9.72462654e-01 2.52320379e-01 -4.13944721e-01
2.18635991e-01 1.12337075e-01 4.78077173e-01 -4.23607349e-01
-4.18289065e-01 3.17740887e-01 2.09961697e-01 -1.13119686e+00
-8.02288830e-01 -9.86315072e-01 -1.33508956e+00 -7.49463513e-02
-3.52080435e-01 1.23828106e-01 1.76860616e-01 1.16502869e+00
5.18109262e-01 4.89805788e-01 9.27583218e-01 -1.04243374e+00
-4.82200593e-01 -8.08769882e-01 -4.80410576e-01 3.92700940e-01
1.88917801e-01 -6.54402435e-01 -2.41486713e-01 3.88580769e-01] | [8.335536003112793, 0.46648791432380676] |
6e381124-2948-41d9-b3d2-773e52788a1d | graph-based-active-learning-for-surface-water | 2306.10440 | null | https://arxiv.org/abs/2306.10440v1 | https://arxiv.org/pdf/2306.10440v1.pdf | Graph-based Active Learning for Surface Water and Sediment Detection in Multispectral Images | We develop a graph active learning pipeline (GAP) to detect surface water and in-river sediment pixels in satellite images. The active learning approach is applied within the training process to optimally select specific pixels to generate a hand-labeled training set. Our method obtains higher accuracy with far fewer training pixels than both standard and deep learning models. According to our experiments, our GAP trained on a set of 3270 pixels reaches a better accuracy than the neural network method trained on 2.1 million pixels. | ['Jon Schwenk', 'Andrea L. Bertozzi', 'Kevin Miller', 'Bohan Chen'] | 2023-06-17 | null | null | null | null | ['active-learning', 'active-learning'] | ['methodology', 'natural-language-processing'] | [ 3.79520059e-01 7.95399606e-01 -9.34733376e-02 -4.82978553e-01
-1.06963015e+00 -4.17518079e-01 5.28687239e-01 1.28607184e-01
-6.25275612e-01 5.63888490e-01 1.87621266e-01 -2.44797215e-01
2.82179862e-01 -1.61156034e+00 -7.45633721e-01 -8.14012289e-01
-6.44292474e-01 3.93617928e-01 5.24191380e-01 7.50124827e-02
-1.25917792e-01 4.54809844e-01 -1.13696861e+00 1.51203811e-01
1.15318727e+00 5.38989186e-01 3.01812083e-01 8.00987780e-01
-4.70745385e-01 9.24581170e-01 -3.48687083e-01 -1.81083024e-01
5.59006810e-01 -2.21616894e-01 -7.86047816e-01 3.08846831e-01
7.82523036e-01 -4.43258554e-01 -2.61043519e-01 1.11396563e+00
5.03683329e-01 7.77637735e-02 4.64002967e-01 -6.90386176e-01
-4.27558333e-01 1.26356614e+00 -6.02572620e-01 3.04117411e-01
-3.80567640e-01 6.26533151e-01 1.27964246e+00 -1.22683156e+00
3.30075145e-01 1.17523968e+00 7.87249744e-01 2.94677913e-01
-1.17147350e+00 -4.98136818e-01 3.02365780e-01 -9.19042379e-02
-1.45137036e+00 -4.51774597e-01 6.56874716e-01 -2.69679457e-01
9.63451862e-01 1.81997284e-01 1.26894116e+00 5.78440189e-01
-2.65398733e-02 1.07848620e+00 7.98315942e-01 -4.44351882e-01
6.33561492e-01 -2.67602473e-01 3.28932405e-01 8.41745198e-01
4.77957875e-01 1.80632681e-01 -4.24752414e-01 -2.94754893e-01
8.88154149e-01 -1.86834261e-01 -1.06543958e-01 2.59588510e-01
-6.32044852e-01 1.13511288e+00 9.36875284e-01 -1.85418054e-01
-5.24787724e-01 3.44171911e-01 -1.81164652e-01 -1.12716295e-01
1.05380023e+00 3.96614224e-01 -3.42468351e-01 6.01730525e-01
-1.21224821e+00 -1.96138158e-01 7.00634241e-01 5.01503885e-01
1.39202237e+00 4.73657608e-01 1.03509516e-01 7.12796628e-01
1.08078957e+00 8.15436721e-01 -2.96090126e-01 -8.13733220e-01
4.48953778e-01 1.27800810e+00 -1.24509372e-01 -7.16620207e-01
-4.53122109e-01 -4.97929990e-01 -5.50178528e-01 8.54263783e-01
2.55918354e-01 -6.51644468e-01 -1.39396429e+00 7.73415089e-01
1.45697922e-01 4.65264171e-01 1.55011922e-01 6.50319993e-01
1.29029644e+00 1.06641388e+00 4.00083691e-01 8.21982324e-02
6.82399750e-01 -1.02184474e+00 -2.98037708e-01 -1.23162508e+00
4.55729872e-01 -1.93603951e-02 9.19684291e-01 7.77355880e-02
-6.95597291e-01 -2.44524911e-01 -9.70796347e-01 4.16151643e-01
-3.88620138e-01 -8.41595158e-02 8.44107032e-01 5.72360158e-01
-1.24806798e+00 2.74949461e-01 -1.18945646e+00 -1.36907220e-01
1.03632092e+00 2.87724674e-01 -1.06856242e-01 2.28416860e-01
-1.05756879e+00 5.64797699e-01 4.76686746e-01 5.71589112e-01
-1.15710354e+00 -5.84504187e-01 -1.11246550e+00 1.53373569e-01
1.24794357e-01 -3.17251861e-01 9.56617415e-01 -1.28247821e+00
-1.27311099e+00 1.08389378e+00 3.40651155e-01 -8.16163659e-01
5.13106525e-01 -2.69447893e-01 -7.62804821e-02 9.50491726e-02
-4.19424921e-02 9.41350698e-01 6.17243290e-01 -1.38180780e+00
-6.11422062e-01 -3.02847415e-01 -2.35078409e-02 3.32640916e-01
-2.21983239e-01 -3.70926112e-02 -3.94697905e-01 -3.89139086e-01
3.20282012e-01 -8.71638179e-01 -1.01806462e+00 2.93589830e-01
-4.28567439e-01 -1.18273728e-01 6.72044992e-01 -3.50845575e-01
8.06591749e-01 -1.98784983e+00 -2.40095824e-01 4.06738788e-01
4.12164837e-01 3.28553051e-01 -2.81586885e-01 2.13400930e-01
7.75893629e-02 1.59966886e-01 -8.34131181e-01 -1.51355639e-01
-2.30553403e-01 3.58783960e-01 -1.82086140e-01 5.08786976e-01
6.18317187e-01 1.11279368e+00 -1.20096493e+00 -8.27411175e-01
1.19945355e-01 1.69573158e-01 -1.61575198e-01 3.98353994e-01
-4.34763253e-01 1.43309042e-01 -3.89644533e-01 8.64388108e-01
8.62222672e-01 -4.62446332e-01 2.42437989e-01 2.48114541e-01
-1.20934375e-01 -1.29132625e-02 -1.05284238e+00 1.33162534e+00
-4.21885811e-02 1.04767394e+00 -7.07521336e-03 -6.57492042e-01
1.31328130e+00 -3.22473869e-02 4.98666376e-01 -6.19850159e-01
-2.30885357e-01 3.96814346e-02 -8.93826932e-02 -6.28838718e-01
4.12576824e-01 3.75490338e-01 2.33966768e-01 3.06327254e-01
-6.68501034e-02 8.64881873e-02 3.20001692e-02 2.06243172e-01
1.24958456e+00 2.06019014e-01 9.41298455e-02 -4.84746933e-01
-3.42967622e-02 6.25017881e-01 6.11640394e-01 1.00489545e+00
-4.12759095e-01 4.98127639e-01 3.22439343e-01 -6.86735511e-01
-7.21736908e-01 -1.09027815e+00 -5.49265593e-02 1.30016971e+00
1.07143573e-01 -2.12293379e-02 -5.84420502e-01 -8.99893522e-01
1.01430207e-01 5.28208017e-01 -6.54308319e-01 4.40592259e-01
-8.27888846e-01 -1.61090362e+00 6.37822986e-01 5.86465597e-01
7.96337187e-01 -1.28313994e+00 -3.88245821e-01 2.19836012e-01
4.14413452e-01 -6.98980868e-01 2.56626099e-01 3.04930091e-01
-1.23869753e+00 -1.11080122e+00 -5.36277235e-01 -8.57195735e-01
8.68505836e-01 1.88217536e-02 1.62804317e+00 2.21607506e-01
-1.16840899e-01 -1.98992580e-01 -4.50531155e-01 -5.53963840e-01
-3.47066909e-01 2.30786324e-01 -9.23469245e-01 -7.03344047e-02
4.81992036e-01 -1.50258303e-01 -9.62326586e-01 -1.46626696e-01
-8.01935434e-01 -8.18892494e-02 6.95705891e-01 4.05665576e-01
4.67378765e-01 -1.08699299e-01 1.92763209e-01 -1.22372615e+00
4.23435979e-02 -6.90517247e-01 -1.03137350e+00 1.59641877e-01
-7.24056542e-01 -1.03610188e-01 -6.11132234e-02 -7.80760348e-02
-9.41953480e-01 1.18537319e+00 -1.45742118e-01 4.19230044e-01
1.15477905e-01 8.71792436e-01 3.97979021e-02 -1.26269311e-01
1.22950506e+00 -6.74582943e-02 -2.19844822e-02 -4.00441349e-01
3.05224776e-01 7.08466411e-01 3.89655471e-01 1.70329392e-01
7.86808789e-01 6.19565725e-01 -2.71755636e-01 -7.82516003e-01
-8.89902174e-01 -3.35765690e-01 -8.64199996e-01 -6.82402134e-01
5.81964135e-01 -1.41369939e+00 1.93467751e-01 7.23164797e-01
-7.99689770e-01 -1.17260730e+00 -3.22208524e-01 4.22195524e-01
1.67708099e-01 1.60093158e-01 -5.08401275e-01 -9.53850985e-01
-8.91768038e-01 -7.96567321e-01 1.23981929e+00 5.05518198e-01
1.25678390e-01 -1.33688772e+00 3.59338313e-01 3.57877463e-02
3.57765347e-01 3.78047407e-01 1.48997277e-01 -5.51064610e-01
-8.45090508e-01 -1.78293828e-02 -1.81229174e-01 3.34965326e-02
1.11490108e-01 6.24334157e-01 -1.14506352e+00 -1.49469674e-01
-7.54630744e-01 -1.57916695e-01 1.58617365e+00 5.35082698e-01
6.20067775e-01 -2.35506520e-01 -2.96204329e-01 5.72931051e-01
1.67098570e+00 -5.37455827e-02 1.17105687e+00 5.32852352e-01
5.90757787e-01 5.10451078e-01 4.72053081e-01 2.22489819e-01
2.13016316e-01 -1.48225367e-01 1.03424346e+00 -9.75153506e-01
-9.12805721e-02 1.14354365e-01 4.87802416e-01 2.38832474e-01
5.67030050e-02 -4.05217290e-01 -1.85774624e+00 6.38983130e-01
-1.88241470e+00 -8.88026416e-01 -8.52909565e-01 1.93258572e+00
8.30430090e-01 3.34511340e-01 -1.91909239e-01 -2.12609142e-01
8.38646412e-01 5.52694798e-01 -6.92212999e-01 3.60316783e-01
-5.84094703e-01 3.33592415e-01 1.21679091e+00 8.55869114e-01
-1.72334170e+00 1.40201342e+00 7.83136845e+00 1.95844114e-01
-1.21966946e+00 -1.72827840e-01 7.32890725e-01 1.68627471e-01
-2.97017068e-01 1.63564876e-01 -8.88311625e-01 2.16931045e-01
7.65703261e-01 2.26862952e-01 -2.95046926e-01 9.52746153e-01
5.44335425e-01 -3.03029239e-01 -4.64738041e-01 3.84555370e-01
-2.20594540e-01 -1.59734738e+00 -1.30089909e-01 1.40504718e-01
9.48401213e-01 1.00488901e+00 -4.39096898e-01 5.81520759e-02
1.17561579e+00 -1.06436992e+00 5.13189852e-01 7.35055566e-01
4.51670140e-01 -2.72339940e-01 7.98321962e-01 3.27711284e-01
-1.12925053e+00 -3.59365018e-03 -4.94813621e-01 -2.39039332e-01
-7.82674029e-02 7.77885735e-01 -1.25449705e+00 -1.34860083e-01
7.24633932e-01 9.27490830e-01 -1.10018361e+00 1.54287720e+00
-5.92819035e-01 1.42621410e+00 -5.36997676e-01 -2.95375865e-02
4.69052970e-01 -2.91937262e-01 3.99526209e-01 1.39957273e+00
-1.51320606e-01 1.36837155e-01 3.01781148e-01 7.92813420e-01
-3.43126766e-02 1.28827587e-01 -5.10305882e-01 9.63937957e-03
6.91747725e-01 1.37197661e+00 -1.15489841e+00 -5.00541925e-01
-3.07100892e-01 5.52700639e-01 4.31361973e-01 3.44041377e-01
-5.66638052e-01 -2.21940592e-01 2.63526648e-01 3.11891418e-02
9.11881924e-02 -3.27026516e-01 -5.57074666e-01 -7.80472159e-01
-5.73109746e-01 -2.43591934e-01 4.15931582e-01 -7.95006275e-01
-1.33080387e+00 6.50042295e-01 -2.70117640e-01 -1.33752203e+00
3.34463231e-02 -5.33385098e-01 -1.04035568e+00 8.46703768e-01
-1.74788141e+00 -1.29780102e+00 -9.06427681e-01 4.73514833e-02
5.31530023e-01 -3.21627736e-01 8.22222471e-01 1.35380208e-01
-7.49772191e-01 -7.30710477e-02 1.21970780e-01 1.02218425e+00
1.32544130e-01 -1.53268290e+00 1.10413635e+00 1.31421387e+00
2.05063969e-01 1.46946672e-03 3.57953638e-01 -8.11286747e-01
-1.21103275e+00 -1.59393644e+00 7.08323002e-01 -1.86253816e-01
6.66071713e-01 -1.97670326e-01 -1.02325130e+00 9.80822384e-01
3.79416913e-01 5.63116610e-01 7.70191669e-01 -1.84329689e-01
-2.64849188e-03 -1.48096442e-01 -7.91498721e-01 2.03409478e-01
7.56556571e-01 -1.19761050e-01 -4.57804263e-01 5.42248249e-01
4.52086866e-01 -3.26107860e-01 -6.46886051e-01 4.13468808e-01
9.11503583e-02 -6.18463039e-01 8.10963213e-01 -5.44007123e-01
3.71726602e-01 -3.00246924e-01 1.30258814e-01 -1.19403589e+00
-6.47063732e-01 -1.33043990e-01 1.01221867e-01 9.29657280e-01
1.12841105e+00 -5.67096710e-01 1.47815859e+00 4.66567636e-01
-3.67007941e-01 -2.98890412e-01 -7.50411808e-01 -3.32846344e-01
1.11335546e-01 -2.54947811e-01 5.42469621e-01 9.65692818e-01
-4.90404814e-01 3.46212029e-01 7.69697353e-02 7.00094223e-01
1.04480529e+00 9.32656303e-02 5.94799876e-01 -1.61753392e+00
9.52075645e-02 -1.53931707e-01 -3.79017591e-01 -9.23150957e-01
1.31640077e-01 -7.73727775e-01 4.86348271e-01 -2.20739269e+00
1.38811365e-01 -7.93271184e-01 -1.88265115e-01 1.26298916e+00
-4.85136181e-01 6.85245156e-01 -2.89751142e-01 3.75550926e-01
-4.71945494e-01 4.81013715e-01 9.26618814e-01 -7.64505565e-01
-5.11966467e-01 7.45879859e-02 -3.75268817e-01 7.56683588e-01
9.12335753e-01 -6.95075810e-01 -8.48991144e-03 -7.74950027e-01
7.37543941e-01 -3.58279765e-01 3.52030098e-01 -6.95996881e-01
3.13604802e-01 -3.20051581e-01 4.03617859e-01 -5.97747922e-01
-6.84460849e-02 -7.38783658e-01 2.13058352e-01 6.23348236e-01
-3.27436805e-01 -4.38259125e-01 3.23100947e-02 5.79468548e-01
-9.58123431e-02 -3.01536649e-01 8.99191916e-01 -5.53354084e-01
-1.41344786e+00 3.57043535e-01 -6.22309268e-01 -7.52351657e-02
8.18440676e-01 -3.79787296e-01 -6.19078755e-01 -2.24452894e-02
-9.23852086e-01 4.91037548e-01 1.89300925e-01 4.52202559e-03
6.25193775e-01 -9.47020769e-01 -1.39505553e+00 1.01594508e-01
2.59668767e-01 4.91616249e-01 -2.61120975e-01 2.77372271e-01
-1.35411680e+00 -4.57104981e-01 7.61885662e-03 -9.92243171e-01
-1.10906339e+00 -5.10165036e-01 9.10791457e-01 -1.12184761e-02
-9.52055037e-01 1.05570221e+00 -4.78036433e-01 -4.89371896e-01
9.32481214e-02 5.34379371e-02 -6.55330062e-01 4.24964190e-01
2.58558631e-01 3.61309201e-01 1.45093381e-01 -5.37399292e-01
-3.13721359e-01 2.31341749e-01 4.47414994e-01 1.15334149e-02
1.84788477e+00 1.87268585e-01 -2.45240167e-01 2.58882016e-01
8.59476030e-01 8.28802511e-02 -1.77337432e+00 -6.83063507e-01
3.20112914e-01 -4.03028786e-01 7.78982520e-01 -5.07830381e-01
-1.64196348e+00 3.60004932e-01 1.03848910e+00 4.26471859e-01
8.84800017e-01 1.02146484e-01 1.68945745e-01 5.57930350e-01
6.32869974e-02 -1.11354184e+00 -8.74360185e-03 5.69546342e-01
6.16806567e-01 -1.66641378e+00 4.39712405e-01 -6.07016444e-01
-4.94212389e-01 1.37828863e+00 6.65922105e-01 -4.13069487e-01
8.42802942e-01 7.52542436e-01 6.19329810e-01 -6.13462090e-01
-5.59433341e-01 -7.01631486e-01 4.81872335e-02 4.26368326e-01
4.68886405e-01 1.64685935e-01 1.39681906e-01 -2.13225514e-01
6.69753999e-02 -2.99221426e-01 4.90475535e-01 9.71655250e-01
-1.14225864e+00 -7.14459360e-01 -3.64763767e-01 6.03361189e-01
-2.97792535e-02 -4.68444526e-01 -7.54246354e-01 6.78935409e-01
-4.04627658e-02 8.87719274e-01 7.70421803e-01 -1.95169538e-01
1.14597909e-01 -1.58117592e-01 -9.00275260e-02 -1.05374920e+00
-8.99843931e-01 -7.79528916e-02 2.68889844e-01 -3.09685290e-01
-8.90284419e-01 -5.42745769e-01 -1.51174474e+00 2.64255516e-02
-5.08404672e-01 9.74926353e-02 4.45244253e-01 6.66038811e-01
1.89602315e-01 1.27113059e-01 9.00958717e-01 -7.70404518e-01
-1.14981100e-01 -1.19904947e+00 -7.36257374e-01 -9.00825486e-02
1.39899433e-01 6.28936291e-02 -5.54874182e-01 2.70899683e-01] | [9.513100624084473, -1.4287621974945068] |
78355e1c-62b0-403a-a199-4e9c84dbd728 | hierarchical-internal-representation-of | 1711.07792 | null | http://arxiv.org/abs/1711.07792v3 | http://arxiv.org/pdf/1711.07792v3.pdf | Hierarchical internal representation of spectral features in deep convolutional networks trained for EEG decoding | Recently, there is increasing interest and research on the interpretability
of machine learning models, for example how they transform and internally
represent EEG signals in Brain-Computer Interface (BCI) applications. This can
help to understand the limits of the model and how it may be improved, in
addition to possibly provide insight about the data itself. Schirrmeister et
al. (2017) have recently reported promising results for EEG decoding with deep
convolutional neural networks (ConvNets) trained in an end-to-end manner and,
with a causal visualization approach, showed that they learn to use spectral
amplitude changes in the input. In this study, we investigate how ConvNets
represent spectral features through the sequence of intermediate stages of the
network. We show higher sensitivity to EEG phase features at earlier stages and
higher sensitivity to EEG amplitude features at later stages. Intriguingly, we
observed a specialization of individual stages of the network to the classical
EEG frequency bands alpha, beta, and high gamma. Furthermore, we find first
evidence that particularly in the last convolutional layer, the network learns
to detect more complex oscillatory patterns beyond spectral phase and
amplitude, reminiscent of the representation of complex visual features in
later layers of ConvNets in computer vision tasks. Our findings thus provide
insights into how ConvNets hierarchically represent spectral EEG features in
their intermediate layers and suggest that ConvNets can exploit and might help
to better understand the compositional structure of EEG time series. | ['Robin Tibor Schirrmeister', 'Kay Gregor Hartmann', 'Tonio Ball'] | 2017-11-21 | null | null | null | null | ['eeg-decoding', 'eeg-decoding'] | ['medical', 'time-series'] | [ 2.82422513e-01 2.38108203e-01 5.15334129e-01 -3.31215352e-01
2.63590276e-01 -6.68344200e-01 8.03410888e-01 2.76248276e-01
-3.63442838e-01 6.08232379e-01 3.34264636e-01 -4.37846780e-01
-4.72667813e-01 -5.68650603e-01 -5.68652570e-01 -7.23233283e-01
-8.95522773e-01 -8.62577856e-02 -3.50894295e-02 -4.24225718e-01
1.16007328e-01 5.57159841e-01 -1.26252103e+00 7.97121108e-01
5.66241205e-01 1.04326284e+00 5.77021465e-02 5.55343509e-01
3.98593366e-01 4.71726149e-01 -9.05860424e-01 -1.13067287e-03
-1.91550195e-01 -5.80558956e-01 -6.66552246e-01 -2.84912586e-01
-4.70129065e-02 1.50798053e-01 -4.53733861e-01 9.39061224e-01
3.03589553e-01 -1.85459659e-01 6.83684647e-01 -1.05242252e+00
-4.88144845e-01 7.49441028e-01 -3.95682395e-01 1.00866783e+00
2.25551486e-01 4.70388800e-01 7.45691240e-01 -5.75889230e-01
4.28929448e-01 9.03077245e-01 7.49170661e-01 2.07912251e-01
-1.58324921e+00 -7.50875890e-01 2.69001007e-01 6.10182047e-01
-1.09747946e+00 -2.30371088e-01 8.01140070e-01 -7.42235243e-01
1.13297212e+00 1.82380304e-01 1.40304983e+00 1.34065652e+00
6.00008786e-01 4.16593760e-01 1.43703413e+00 -1.18292555e-01
1.03924103e-01 -4.94200327e-02 5.29870391e-01 4.84553277e-01
1.15685992e-01 2.16459662e-01 -8.76353681e-01 -6.74001053e-02
7.18972325e-01 -6.51334152e-02 -6.65601969e-01 2.16265634e-01
-1.40438831e+00 5.71904480e-01 9.47710633e-01 5.57172835e-01
-6.38448417e-01 4.67281729e-01 4.33817327e-01 2.78734535e-01
3.28934103e-01 8.34284067e-01 -5.86279631e-01 -1.57191381e-01
-9.66403365e-01 -1.88976437e-01 4.32493985e-01 3.36734027e-01
5.80455661e-01 1.32498413e-01 -4.45935130e-02 3.97857487e-01
1.08265877e-01 -1.00921594e-01 4.86783147e-01 -6.27281725e-01
1.32645354e-01 4.49380428e-01 -4.47647780e-01 -6.74406230e-01
-9.81802523e-01 -7.99777985e-01 -9.97945905e-01 3.79643738e-01
3.78471076e-01 -2.12515697e-01 -5.80990672e-01 1.70056963e+00
-5.29746354e-01 2.53536671e-01 -2.27694541e-01 9.02843595e-01
4.79939878e-01 3.72980088e-01 1.75169379e-01 -4.40929867e-02
1.62319362e+00 -5.53195700e-02 -4.79265988e-01 -3.55067939e-01
5.02284348e-01 -1.61898807e-01 9.80211318e-01 7.03613281e-01
-1.06668448e+00 -6.98638082e-01 -1.26794565e+00 1.82449773e-01
-5.61744750e-01 3.76124382e-02 7.93214440e-01 3.43507499e-01
-1.31534576e+00 1.13160181e+00 -9.43606317e-01 -2.44437262e-01
8.86947989e-01 6.60770357e-01 -2.56500632e-01 5.71315765e-01
-1.27703512e+00 9.68044758e-01 5.48752069e-01 3.69003087e-01
-6.58699334e-01 -1.09082055e+00 -4.14719939e-01 3.82185698e-01
-6.16471350e-01 -6.60156369e-01 7.37437844e-01 -1.25572360e+00
-1.00676465e+00 6.19171262e-01 -6.50139153e-02 -5.35958588e-01
5.86017384e-04 -6.30469248e-03 -3.71439546e-01 3.35023135e-01
-5.33466816e-01 6.36151254e-01 6.67353451e-01 -8.95291746e-01
-1.90969691e-01 -5.22660017e-01 -3.04890908e-02 -3.75890553e-01
-2.57239729e-01 -9.36383978e-02 4.22863781e-01 -6.82294250e-01
1.06362090e-01 -6.24276340e-01 1.05065770e-01 -1.06441244e-01
-9.68953818e-02 -7.36565962e-02 5.69936574e-01 -7.54196167e-01
9.46158946e-01 -2.28851032e+00 2.32856512e-01 5.22896767e-01
7.38043189e-01 -3.27490014e-03 -1.15762996e-02 3.99129719e-01
-8.40345442e-01 2.09409714e-01 -2.03554258e-01 3.62411201e-01
-9.24984962e-02 -8.78317803e-02 -5.12860000e-01 5.68327129e-01
7.80940115e-01 1.17585337e+00 -8.82955611e-01 3.75237584e-01
1.74985096e-01 7.39395380e-01 -4.98758316e-01 -1.15542494e-01
-1.84640719e-03 8.12271237e-01 1.77103072e-01 2.32239105e-02
3.83404791e-01 -3.26146543e-01 1.11495100e-01 -6.18654728e-01
-2.14149445e-01 7.10149169e-01 -4.97954249e-01 1.45280099e+00
-3.76728594e-01 1.83028805e+00 -3.11819524e-01 -1.47630572e+00
4.41115379e-01 5.09073257e-01 3.45570564e-01 -9.68912840e-01
3.67097437e-01 -4.23955359e-02 8.76072466e-01 -5.26516080e-01
-2.07062066e-01 -1.98574308e-02 4.15759861e-01 4.41477507e-01
4.23062205e-01 1.06578074e-01 -8.59204307e-02 -2.81503517e-02
1.11247277e+00 -1.03018731e-01 2.15458319e-01 -5.38555324e-01
1.20362937e-01 -2.95485556e-01 -1.68714657e-01 5.67914784e-01
1.06089450e-01 5.70487857e-01 1.10095441e+00 -4.34317231e-01
-7.77948499e-01 -1.09612381e+00 -5.20597935e-01 9.55133796e-01
-3.00955087e-01 -4.96218592e-01 -6.90378487e-01 -1.11494656e-03
-1.77007854e-01 5.13390779e-01 -1.11275446e+00 -6.51578665e-01
-5.27489960e-01 -9.76463974e-01 6.75528347e-01 6.56416833e-01
4.62463051e-01 -1.17461872e+00 -1.13914704e+00 3.39009374e-01
1.19032837e-01 -9.79099631e-01 3.48434120e-01 8.30540836e-01
-1.10842466e+00 -1.16591954e+00 -5.56157470e-01 -5.01918137e-01
6.10086203e-01 -2.11806417e-01 9.09228504e-01 1.20215751e-01
-5.82812726e-01 3.28456163e-01 -2.13919550e-01 -7.13331163e-01
-9.94915050e-03 1.54674257e-04 -1.19878203e-01 -2.06904113e-01
3.67938191e-01 -1.24308145e+00 -9.74776506e-01 1.16434344e-03
-7.65355289e-01 2.02350736e-01 5.28652251e-01 6.89630687e-01
1.57864261e-02 -5.59735782e-02 5.52219331e-01 -5.67321539e-01
9.68505859e-01 -5.76776087e-01 -2.05414698e-01 5.72425723e-02
-3.49967360e-01 2.68153995e-01 7.45560288e-01 -5.98819256e-01
-6.87599778e-01 -4.04553264e-01 -7.13651925e-02 6.64736331e-02
-2.10442185e-01 6.50604129e-01 3.36566478e-01 -3.07217896e-01
9.88197625e-01 2.52319187e-01 -3.84721965e-01 -1.61509797e-01
-1.05782726e-03 3.31363887e-01 5.44823647e-01 -4.07886833e-01
3.89989972e-01 4.31543291e-01 1.17393389e-01 -9.24839258e-01
-3.93458784e-01 -4.74095047e-02 -7.48439431e-01 -3.11485797e-01
8.11520517e-01 -8.28285754e-01 -1.12732875e+00 3.02556217e-01
-1.34055305e+00 -6.40903413e-01 -2.08562747e-01 5.18058300e-01
-4.04571593e-01 -1.43598557e-01 -6.29548609e-01 -6.19899154e-01
-9.96277332e-02 -9.47645962e-01 7.19169617e-01 1.38011038e-01
-6.75881088e-01 -1.07434535e+00 -2.21189111e-01 -4.74769235e-01
4.19072330e-01 4.25067365e-01 1.46611249e+00 -5.09570420e-01
-1.81162044e-01 2.52005219e-01 -2.95284182e-01 -2.53046528e-02
-1.36621997e-01 -5.94870523e-02 -1.45936859e+00 -1.55966207e-01
7.62972981e-02 2.95016766e-02 1.15898430e+00 4.65288550e-01
1.66060615e+00 -3.44260693e-01 -2.71834046e-01 7.95293868e-01
1.06200397e+00 2.61289537e-01 7.64667630e-01 2.39361838e-01
3.88454646e-01 8.19374681e-01 -4.28172112e-01 2.49571234e-01
-8.03685710e-02 3.55996490e-01 2.92377740e-01 -2.56253690e-01
-1.35779038e-01 1.54967785e-01 2.84074366e-01 6.06820703e-01
-5.63829839e-01 2.85730928e-01 -1.02562642e+00 3.35628420e-01
-1.56775498e+00 -8.15294981e-01 -3.12101305e-01 1.95177448e+00
8.76065195e-01 3.21566492e-01 1.29333967e-02 5.08665383e-01
1.88216418e-01 -4.87322807e-02 -4.37243611e-01 -4.93536592e-01
-1.65484190e-01 8.65864754e-01 7.03555197e-02 -6.70534745e-02
-6.01392269e-01 4.93159562e-01 6.66544247e+00 1.39123604e-01
-1.46999633e+00 1.21137157e-01 6.40342712e-01 -1.01251632e-01
-1.25992060e-01 -2.16769516e-01 9.20330882e-02 3.85864794e-01
1.33771074e+00 -5.96159957e-02 9.28142428e-01 7.21897632e-02
3.61650527e-01 -2.50625134e-01 -1.52854407e+00 1.01677024e+00
-4.57915604e-01 -1.56518602e+00 -2.54762352e-01 1.48591161e-01
3.50677997e-01 2.00328559e-01 2.96158344e-01 4.12394069e-02
-1.10496640e-01 -1.53331733e+00 8.44241500e-01 7.88448334e-01
6.66290879e-01 -6.51961565e-01 5.36582589e-01 1.22961134e-01
-1.11786664e+00 -2.96072274e-01 -3.24407488e-01 -4.01983649e-01
-1.67256564e-01 3.81493896e-01 -6.18892193e-01 -3.47399935e-02
9.94407058e-01 9.78525162e-01 -9.28381324e-01 1.24751580e+00
-3.04143101e-01 9.00777161e-01 -4.74880151e-02 3.33200805e-02
1.36323646e-01 -8.01575407e-02 2.60452420e-01 1.45782995e+00
3.32429975e-01 3.95116597e-01 -7.81210482e-01 1.25935495e+00
2.51008451e-01 -5.69282234e-01 -5.25719643e-01 -2.46437475e-01
7.60679021e-02 1.09538889e+00 -8.92171860e-01 -1.50599033e-01
-1.37022510e-01 7.88303018e-01 3.49686593e-01 8.42465460e-01
-5.64908445e-01 -3.97529304e-01 6.49329841e-01 6.54756650e-02
3.71690802e-02 -5.21922946e-01 -7.30580509e-01 -1.00934696e+00
-9.05927792e-02 -5.60156107e-01 3.38924937e-02 -1.30952382e+00
-9.47404802e-01 7.68006802e-01 3.49585861e-02 -7.83008158e-01
-1.74663454e-01 -1.00857532e+00 -1.00212085e+00 9.81611729e-01
-1.18682182e+00 -5.53595960e-01 -3.48615110e-01 8.14420044e-01
2.43146554e-01 1.53437972e-01 8.50267470e-01 -5.12784235e-02
-2.92641342e-01 2.23595947e-01 2.32328125e-03 1.60758674e-01
1.76789537e-01 -1.23074043e+00 4.79509741e-01 6.32284939e-01
5.73215604e-01 9.33723092e-01 8.09574604e-01 -1.72758758e-01
-1.08963645e+00 -8.11214864e-01 3.54792953e-01 -1.72604576e-01
8.67618322e-01 -7.62333453e-01 -1.17313898e+00 5.90433300e-01
6.03200257e-01 -1.71110615e-01 7.60543883e-01 3.49281490e-01
-3.24819952e-01 -1.25768214e-01 -6.77591503e-01 6.37426853e-01
1.02349102e+00 -7.45279610e-01 -8.43476057e-01 2.27708861e-01
3.50192726e-01 -1.11488877e-02 -7.76310861e-01 2.24380493e-01
6.27626002e-01 -1.21717942e+00 9.07636881e-01 -6.03525281e-01
3.95731509e-01 3.17956172e-02 4.51052487e-01 -1.99604583e+00
-5.46631455e-01 -4.93547410e-01 -1.49090067e-01 5.64517498e-01
5.95619202e-01 -8.07846785e-01 1.32119015e-01 3.26520354e-01
-2.65774727e-01 -7.64583111e-01 -7.88857758e-01 -4.28510010e-01
1.11269601e-01 -6.41088307e-01 4.29525495e-01 6.86133087e-01
5.53693116e-01 1.90668404e-01 3.18414629e-01 1.24314561e-01
1.59258038e-01 -3.11907917e-01 -2.38687351e-01 -1.59467769e+00
-1.46228164e-01 -8.87015760e-01 -6.04305744e-01 -4.90385562e-01
1.18048996e-01 -1.11989737e+00 -3.68445367e-01 -1.43447363e+00
1.40424781e-02 -9.75297093e-02 -6.84094667e-01 7.15648353e-01
6.93429112e-02 5.01425087e-01 1.93711370e-01 1.36075661e-01
5.12729995e-02 1.47161052e-01 8.82807612e-01 -1.87314332e-01
-2.13739038e-01 -2.14428902e-01 -6.91453993e-01 7.08048224e-01
9.77042198e-01 -3.51230890e-01 -4.31417406e-01 -4.47116345e-01
6.89685583e-01 -1.33285537e-01 9.84003961e-01 -1.23554003e+00
2.09897280e-01 2.94512123e-01 1.34333098e+00 -5.75624816e-02
1.61182404e-01 -7.45192170e-01 1.48519173e-01 6.35512710e-01
-5.24408281e-01 2.02003017e-01 6.11684322e-01 4.63278681e-01
-1.23327129e-01 1.77249327e-01 5.23108482e-01 -4.30505686e-02
-5.50508499e-01 1.63937919e-02 -9.57147658e-01 -1.18166178e-01
6.05688810e-01 -3.80401194e-01 -5.56698143e-01 -2.61475205e-01
-1.03121459e+00 -6.75634146e-02 -2.04668403e-01 3.20283443e-01
6.14797413e-01 -1.18635643e+00 -5.13642490e-01 5.24429500e-01
-7.54639357e-02 -6.12447202e-01 1.45446127e-02 1.34960735e+00
-3.59563500e-01 3.72055441e-01 -7.13254035e-01 -8.33837628e-01
-8.62194717e-01 2.37785146e-01 6.59287632e-01 2.95354426e-01
-7.42526114e-01 8.56420875e-01 4.77837712e-01 4.53524768e-01
1.06459662e-01 -6.48437977e-01 -5.01120567e-01 3.28409404e-01
4.46515799e-01 -3.27386484e-02 2.67997772e-01 -2.87439585e-01
-4.66995776e-01 2.70380378e-01 2.33857721e-01 -2.73574948e-01
1.66221881e+00 2.75109172e-01 -3.85539711e-01 6.68396771e-01
1.22864008e+00 -5.14306843e-01 -1.38149512e+00 1.99261054e-01
2.26858303e-01 1.08218614e-04 -1.75807644e-02 -1.03684783e+00
-1.23933518e+00 1.50168633e+00 7.21978605e-01 5.03869116e-01
1.35986423e+00 5.47418334e-02 1.80774301e-01 4.34222966e-01
2.18785226e-01 -8.34224045e-01 1.28266037e-01 4.67738420e-01
1.11389005e+00 -7.26906121e-01 -4.80353162e-02 1.10975496e-01
-2.81847179e-01 1.92471731e+00 2.42501348e-01 -4.39713448e-01
8.29573929e-01 2.52184570e-01 -4.53629732e-01 -5.70488155e-01
-9.13674116e-01 -2.61080619e-02 7.40236640e-01 7.90217757e-01
6.76906884e-01 8.09722170e-02 -1.34239033e-01 8.42797697e-01
-6.71409190e-01 -2.67542392e-01 5.34286737e-01 3.84964794e-01
-4.07882959e-01 -5.57243168e-01 -2.06723124e-01 7.74507463e-01
-2.47306481e-01 -3.35701913e-01 -3.70271474e-01 7.94343591e-01
3.11055154e-01 7.64867008e-01 3.16665977e-01 -5.09709060e-01
2.42586076e-01 4.58197027e-01 8.75284314e-01 -5.37835181e-01
-1.01742899e+00 -5.92904277e-02 -1.44320294e-01 -5.80677211e-01
-3.29766780e-01 -5.34384429e-01 -1.30481410e+00 1.19517654e-01
6.04992323e-02 -7.89032280e-02 6.75865650e-01 1.08623457e+00
3.18665177e-01 1.21222889e+00 2.80183434e-01 -1.12243378e+00
2.26131499e-01 -1.07725918e+00 -4.74044323e-01 7.73380995e-02
7.70120323e-01 -5.54329693e-01 -3.75200778e-01 2.49699220e-01] | [12.975566864013672, 3.4309611320495605] |
67525cbf-bd14-4175-b8a8-59bea1a11e71 | foundation-model-for-endoscopy-video-analysis | 2306.16741 | null | https://arxiv.org/abs/2306.16741v1 | https://arxiv.org/pdf/2306.16741v1.pdf | Foundation Model for Endoscopy Video Analysis via Large-scale Self-supervised Pre-train | Foundation models have exhibited remarkable success in various applications, such as disease diagnosis and text report generation. To date, a foundation model for endoscopic video analysis is still lacking. In this paper, we propose Endo-FM, a foundation model specifically developed using massive endoscopic video data. First, we build a video transformer, which captures both local and global long-range dependencies across spatial and temporal dimensions. Second, we pre-train our transformer model using global and local views via a self-supervised manner, aiming to make it robust to spatial-temporal variations and discriminative across different scenes. To develop the foundation model, we construct a large-scale endoscopy video dataset by combining 9 publicly available datasets and a privately collected dataset from Baoshan Branch of Renji Hospital in Shanghai, China. Our dataset overall consists of over 33K video clips with up to 5 million frames, encompassing various protocols, target organs, and disease types. Our pre-trained Endo-FM can be easily adopted for a given downtream task via fine-tuning by serving as the backbone. With experiments on 3 different types of downstream tasks, including classification, segmentation, and detection, our Endo-FM surpasses the current state-of-the-art self-supervised pre-training and adapter-based transfer learning methods by a significant margin, such as VCL (3.1% F1 for classification, 4.8% Dice for segmentation, and 5.5% F1 for detection) and ST-Adapter (5.9% F1 for classification, 9.6% Dice for segmentation, and 9.9% F1 for detection). Code, datasets, and models are released at https://github.com/med-air/Endo-FM. | ['Qi Dou', 'Shaoting Zhang', 'Chang Liu', 'Zhao Wang'] | 2023-06-29 | null | null | null | null | ['transfer-learning'] | ['miscellaneous'] | [ 4.19612266e-02 -2.93894887e-01 -3.76884967e-01 -1.65254861e-01
-1.21879923e+00 -7.44598627e-01 1.00914821e-01 1.22830942e-01
-4.23245609e-01 3.68015915e-01 2.57950902e-01 -2.70474792e-01
4.40774485e-02 -6.50375605e-01 -8.62699389e-01 -7.30634749e-01
-1.66893974e-01 -6.82861432e-02 3.40492398e-01 1.82648122e-01
-2.16068625e-01 -1.42058134e-02 -9.84639466e-01 7.97888756e-01
7.87583649e-01 1.10296273e+00 2.24833503e-01 9.25439715e-01
4.34481472e-01 7.55952299e-01 -3.24932069e-01 -2.05912054e-01
-2.02489607e-02 -4.24676001e-01 -6.39992714e-01 9.23979580e-02
2.37406373e-01 -5.07783651e-01 -5.31991839e-01 8.48176718e-01
5.72630346e-01 -2.51923233e-01 4.84195292e-01 -6.98484957e-01
-4.95163471e-01 3.53495836e-01 -3.05528671e-01 3.74974012e-01
2.82204181e-01 4.66405958e-01 8.06861341e-01 -6.26747906e-01
7.30659425e-01 6.64520621e-01 7.53769934e-01 6.76455021e-01
-1.01187468e+00 -7.54950881e-01 1.22817636e-01 -1.43198311e-01
-1.07433033e+00 -7.79888779e-02 3.57456297e-01 -6.07012630e-01
5.51069736e-01 2.13115349e-01 7.90879309e-01 1.28838778e+00
4.42551494e-01 1.03417540e+00 9.24635649e-01 4.47025299e-02
-6.47552386e-02 1.63660888e-02 -3.13642591e-01 9.76315618e-01
-6.25537559e-02 1.39588907e-01 -3.66852582e-01 -1.12499893e-01
9.83319402e-01 1.62229270e-01 -5.20678222e-01 -1.22747675e-01
-1.49680018e+00 8.46909404e-01 5.49742997e-01 1.11759588e-01
-9.39774811e-02 -1.20756954e-01 6.00137770e-01 3.12182546e-01
3.10873240e-01 2.27671459e-01 -5.26502430e-01 -8.28313977e-02
-5.80793023e-01 -9.85243395e-02 6.36885822e-01 1.15150940e+00
1.38635606e-01 -3.16027880e-01 -1.63300991e-01 8.36746573e-01
1.27977103e-01 2.64955789e-01 9.50549901e-01 -7.19734967e-01
2.93947130e-01 5.34247160e-01 -2.56651103e-01 -6.79910839e-01
-6.26657903e-01 -6.17421389e-01 -9.63187635e-01 -3.84584814e-01
3.24928761e-01 -3.17058235e-01 -9.20162976e-01 1.68255723e+00
3.30270469e-01 3.08770239e-01 -7.06639141e-02 8.35761786e-01
1.19328964e+00 4.59311515e-01 -4.61480394e-02 -9.31783691e-02
1.48769271e+00 -1.29792106e+00 -3.37021798e-01 1.70440786e-02
1.08073545e+00 -8.20698619e-01 1.05771041e+00 6.53957903e-01
-9.76045370e-01 -4.93370950e-01 -8.20459604e-01 -2.95967255e-02
4.71537597e-02 5.17854393e-01 4.94435966e-01 2.96060830e-01
-8.37513924e-01 3.02774638e-01 -1.10664606e+00 -2.81871349e-01
4.97657359e-01 4.02079105e-01 -6.11756146e-01 -3.45378399e-01
-8.38042676e-01 2.88440734e-01 1.47299707e-01 -9.57276672e-02
-1.05345106e+00 -8.74154210e-01 -8.17348659e-01 -2.81687707e-01
3.65922302e-01 -9.69450057e-01 1.35138524e+00 -6.48274302e-01
-1.46795726e+00 1.10400605e+00 6.28198087e-02 -2.86995709e-01
6.23525381e-01 -9.06867310e-02 -4.16699886e-01 4.70135659e-01
1.04287907e-01 5.37204087e-01 5.15105188e-01 -7.22723126e-01
-7.94001102e-01 -2.52610266e-01 2.85834670e-01 -1.44729363e-02
-2.63095975e-01 -1.75036520e-01 -8.79805982e-01 -9.90814447e-01
-1.64096415e-01 -1.32887721e+00 -3.09431404e-01 3.62280488e-01
-3.83081824e-01 1.33516878e-01 5.06193399e-01 -7.37551689e-01
1.20473909e+00 -2.51649189e+00 9.77861285e-02 -9.81342942e-02
3.92415613e-01 2.56820202e-01 -2.22498745e-01 5.93452789e-02
3.51261459e-02 9.21253338e-02 -9.87413228e-02 -3.01468521e-01
-4.99996871e-01 -1.23259947e-01 1.46783024e-01 5.14705896e-01
7.52898725e-03 8.38568807e-01 -1.18997025e+00 -6.76198483e-01
3.92933309e-01 3.87868196e-01 -9.41849172e-01 3.71335685e-01
5.65722622e-02 7.95737565e-01 -5.43617606e-01 8.02223742e-01
3.53634536e-01 -6.42167330e-01 2.20846877e-01 -5.20727634e-01
7.06616193e-02 1.91723287e-01 -6.94235682e-01 2.19638562e+00
-5.36163807e-01 4.49331760e-01 1.52740166e-01 -9.37754810e-01
6.03446722e-01 3.95208091e-01 9.11288381e-01 -6.51197672e-01
2.14278668e-01 3.45657825e-01 -1.74783058e-02 -7.98912585e-01
-1.88158136e-02 -4.32095006e-02 -1.69300124e-01 1.66050857e-03
3.41957778e-01 1.59597561e-01 3.98922980e-01 1.40703753e-01
1.29191482e+00 -8.57726857e-02 1.66540608e-01 -1.78281397e-01
5.03754020e-01 1.65074751e-01 6.44521177e-01 5.00678658e-01
-3.29636931e-01 9.32844043e-01 5.52699983e-01 -3.39092940e-01
-6.23010576e-01 -1.08108401e+00 -3.90990525e-01 7.31520951e-01
2.86107928e-01 -5.78995824e-01 -6.47510350e-01 -9.19209301e-01
-9.96475518e-02 2.08500296e-01 -6.62563682e-01 -2.57589549e-01
-4.54382837e-01 -8.98538709e-01 4.20158982e-01 5.55712283e-01
3.74925792e-01 -7.90756047e-01 -4.31781679e-01 3.44054908e-01
-3.69228452e-01 -1.24619663e+00 -8.96066129e-01 -1.48009837e-01
-1.02461219e+00 -1.41120303e+00 -8.53398442e-01 -9.90298808e-01
6.91686690e-01 4.79836196e-01 9.57869649e-01 1.77156609e-02
-5.97180665e-01 3.52719575e-01 -4.78421032e-01 -1.52537853e-01
-2.75716573e-01 1.54414117e-01 -2.38301426e-01 -3.57881910e-03
4.02004011e-02 -2.52282888e-01 -1.10194409e+00 6.36930048e-01
-9.93384123e-01 1.85235843e-01 5.71886957e-01 1.21330810e+00
8.93567085e-01 -3.39416981e-01 2.55851835e-01 -9.65867698e-01
1.65930286e-01 -6.70615971e-01 -5.18540263e-01 1.28717735e-01
-2.48096958e-01 -4.00978923e-01 6.52505040e-01 -5.25003791e-01
-7.20385551e-01 -3.89379933e-02 -2.28286684e-01 -5.39263785e-01
1.18997099e-03 6.63906276e-01 3.44971538e-01 1.00090541e-02
5.68221629e-01 2.19739109e-01 2.76561648e-01 -3.70578945e-01
2.44202726e-02 7.42877603e-01 4.96965706e-01 -3.36339682e-01
3.70563596e-01 6.06590927e-01 -3.67251426e-01 -4.37329173e-01
-1.05301297e+00 -7.46494830e-01 -3.05707127e-01 2.99282596e-02
1.01888371e+00 -1.39755893e+00 -6.01429641e-01 5.61678052e-01
-7.42166162e-01 -5.19750893e-01 -2.94468030e-02 8.20171118e-01
-4.84522134e-01 2.67463475e-01 -1.22780037e+00 -1.40591413e-02
-6.17253184e-01 -1.58939433e+00 1.02305877e+00 1.87583655e-01
-1.23315811e-01 -1.05729425e+00 4.94603962e-02 4.90145206e-01
3.62499774e-01 4.33525413e-01 7.17447042e-01 -4.87744212e-01
-6.11310065e-01 -2.84914702e-01 -2.34065875e-01 4.09103066e-01
4.17943478e-01 2.22459575e-03 -4.87057030e-01 -5.65254569e-01
-1.14155650e-01 -3.92340809e-01 8.39297414e-01 6.41932130e-01
1.55092978e+00 -1.29768640e-01 -5.35874903e-01 1.12018371e+00
1.22655761e+00 2.45380193e-01 3.83034676e-01 3.84968191e-01
5.76558173e-01 2.75579989e-01 7.11684525e-01 3.94701302e-01
4.75604802e-01 6.68873191e-01 2.87315428e-01 -2.98316777e-01
-1.99972391e-01 -3.69273931e-01 3.77180576e-01 1.00283241e+00
6.70157298e-02 -2.42061317e-01 -6.98133945e-01 5.61495006e-01
-1.75293171e+00 -6.86054587e-01 1.75940380e-01 2.16234779e+00
7.65415847e-01 8.73746574e-02 2.95452029e-01 -4.65213865e-01
4.90606517e-01 -4.45397161e-02 -6.46510661e-01 1.89795330e-01
1.97725847e-01 -1.38063252e-01 4.50704366e-01 1.85342655e-01
-1.46368730e+00 6.24387681e-01 5.57927465e+00 8.26446772e-01
-1.56552196e+00 6.60247803e-02 7.93926179e-01 -3.46192181e-01
-1.11378074e-01 -3.53186369e-01 -5.82424998e-01 7.34700501e-01
7.85869658e-01 -1.41066223e-01 1.18955821e-01 8.65034997e-01
1.56981051e-01 1.58284515e-01 -9.98435795e-01 1.05741107e+00
9.38379616e-02 -1.53054154e+00 -3.28336544e-02 1.83776110e-01
6.77346647e-01 4.10615295e-01 1.48202792e-01 4.07921940e-01
3.29466872e-02 -8.72571588e-01 3.16910952e-01 1.45674959e-01
1.21950793e+00 -2.69538462e-01 9.09630358e-01 1.52500346e-01
-1.15772080e+00 4.52073850e-02 3.01448796e-02 5.56811929e-01
2.10498095e-01 6.24215543e-01 -5.55896580e-01 5.56518912e-01
8.43692839e-01 1.09624040e+00 -3.23964953e-01 1.18722463e+00
-1.06115110e-01 5.20955384e-01 -3.59117687e-01 1.84290871e-01
3.29356015e-01 -6.08261637e-02 3.88324648e-01 1.41787970e+00
5.41519523e-01 6.22691885e-02 3.13361347e-01 3.84124637e-01
-2.05233008e-01 1.70222580e-01 -2.62710273e-01 7.90154561e-03
1.18189305e-01 1.41906190e+00 -5.50994933e-01 -4.34289455e-01
-7.07092583e-01 9.57193017e-01 7.83046484e-02 1.63166910e-01
-1.23108804e+00 -2.22930223e-01 4.63273823e-01 9.10329074e-02
3.25891465e-01 2.75588240e-02 1.99777305e-01 -1.66292024e+00
5.51536232e-02 -1.11441147e+00 9.13767517e-01 -4.67109680e-01
-1.31386173e+00 6.65346920e-01 -2.16266528e-01 -1.82884061e+00
-1.02043174e-01 -1.01494169e+00 -5.28787315e-01 2.64301091e-01
-1.46692348e+00 -1.11713648e+00 -6.59615755e-01 7.81263471e-01
6.08538508e-01 -1.45623177e-01 8.92300725e-01 4.72330481e-01
-7.14753211e-01 7.97567844e-01 2.14910910e-01 4.03530806e-01
1.12795353e+00 -1.13160110e+00 -2.74517015e-02 5.04377484e-01
-9.14661586e-02 6.12915993e-01 1.24234222e-01 -3.08046639e-01
-1.50011241e+00 -1.30769753e+00 2.98091620e-01 -3.99013609e-01
7.68313169e-01 -1.25888437e-01 -6.36004150e-01 8.04476619e-01
-2.32535020e-01 5.33695877e-01 1.09465873e+00 -5.29810041e-02
-4.23811495e-01 -2.25036755e-01 -1.15641391e+00 5.43962598e-01
1.02683771e+00 -4.76069510e-01 -1.26246557e-01 7.63845384e-01
7.27745235e-01 -1.05840862e+00 -1.33159363e+00 6.09407365e-01
7.04987407e-01 -1.00810540e+00 9.59001184e-01 -4.32227880e-01
7.58962333e-01 -1.22372396e-01 -1.90605670e-02 -1.05686343e+00
-3.11143786e-01 -6.27358139e-01 -5.33618294e-02 7.28448987e-01
5.36877811e-01 -5.46052635e-01 7.57418990e-01 3.12449485e-01
-5.19222856e-01 -1.12952948e+00 -8.48021686e-01 -5.02505720e-01
4.10399362e-02 -3.41827661e-01 -2.86354013e-02 8.49838316e-01
2.14094758e-01 6.07756712e-02 -2.33932048e-01 2.85911933e-02
4.32054311e-01 3.39471698e-01 7.32151031e-01 -5.57853758e-01
-5.85376859e-01 -3.30014586e-01 -2.82868594e-01 -1.24296200e+00
-4.16859478e-01 -1.02642024e+00 -1.74245521e-01 -1.44769561e+00
4.48532432e-01 -3.87468725e-01 -7.12809563e-01 3.95671338e-01
-1.67587832e-01 2.73411989e-01 5.24385720e-02 3.67078900e-01
-6.83133543e-01 2.87269920e-01 1.49120986e+00 -2.52147198e-01
-2.71508098e-01 1.36413947e-01 -6.53591752e-01 6.19268060e-01
7.37556100e-01 -4.56072330e-01 -2.52211720e-01 -4.82557088e-01
-3.11566830e-01 2.63331383e-01 3.06889862e-01 -9.40346777e-01
1.86568573e-01 8.08414295e-02 3.49197656e-01 -2.74191409e-01
1.68430328e-01 -4.98603910e-01 1.81216776e-01 8.65273535e-01
-2.64908820e-01 -2.30430160e-02 3.85113865e-01 5.62150478e-01
-6.28961980e-01 1.77368984e-01 7.63214946e-01 -3.72462541e-01
-5.40310085e-01 7.27812767e-01 -1.91846803e-01 2.41849706e-01
1.12240529e+00 8.85216296e-02 -4.66358453e-01 -2.06146106e-01
-8.22238863e-01 4.51737881e-01 3.79804641e-01 5.06000996e-01
6.46818817e-01 -1.04567885e+00 -7.54626036e-01 2.29934454e-01
4.60111737e-01 2.89329916e-01 8.44938815e-01 1.36378109e+00
-6.30803823e-01 3.93258274e-01 6.04052693e-02 -9.56786454e-01
-1.17930663e+00 5.62160313e-01 3.57239962e-01 -4.94883716e-01
-9.04557526e-01 8.58238876e-01 6.93498790e-01 -4.26593870e-01
6.43663952e-06 -6.46646678e-01 1.29690960e-01 -1.65282518e-01
5.93801796e-01 -2.24317480e-02 1.23968627e-02 -2.72928387e-01
-3.45110416e-01 7.07840800e-01 -4.10261959e-01 3.91989231e-01
1.13566673e+00 5.69115207e-03 2.10617408e-01 1.54628709e-01
1.59243643e+00 -8.00578818e-02 -1.25100136e+00 -2.11299345e-01
-5.36199510e-01 -2.96316922e-01 2.05707595e-01 -7.61313617e-01
-1.37948346e+00 8.30442667e-01 5.42768061e-01 -1.45186886e-01
1.27356708e+00 7.88613111e-02 1.13930082e+00 -1.16318658e-01
3.82086515e-01 -6.57806814e-01 2.62756050e-01 2.21201152e-01
7.48101115e-01 -1.34643769e+00 -1.28198028e-01 -5.45989990e-01
-6.55429780e-01 1.07538974e+00 6.17195487e-01 -7.08790943e-02
6.34042799e-01 3.19591999e-01 2.32918143e-01 -3.83659005e-02
-7.24057913e-01 1.61187991e-01 3.92347217e-01 2.96927810e-01
6.04136944e-01 9.71988589e-02 -1.32481471e-01 6.83766901e-01
-4.91198571e-03 4.44467574e-01 2.98020482e-01 8.41831267e-01
6.72615767e-02 -8.95487130e-01 1.45633087e-01 5.85903406e-01
-7.20478654e-01 -8.91650748e-03 2.88718641e-01 7.95073509e-01
-2.17893738e-02 6.43851101e-01 -1.77795321e-01 -5.94270825e-01
3.44238102e-01 -6.31310463e-01 3.30185115e-01 -5.59417188e-01
-5.83887696e-01 4.85367417e-01 8.93891901e-02 -8.22566390e-01
-4.64302868e-01 -5.79202414e-01 -1.07715917e+00 -4.78663966e-02
-2.20101625e-01 -6.29308894e-02 3.14953804e-01 5.78908384e-01
5.00549972e-01 7.09824204e-01 6.04012370e-01 -5.48701406e-01
-3.78809094e-01 -7.08662152e-01 -3.34576130e-01 4.92628902e-01
5.39131999e-01 -4.77214903e-01 -5.41565001e-01 2.31041968e-01] | [14.337888717651367, -3.0341906547546387] |
ac2a6487-2f54-4166-8e05-934eec42521f | face-synthesis-for-eyeglass-robust-face | 1806.01196 | null | https://arxiv.org/abs/1806.01196v2 | https://arxiv.org/pdf/1806.01196v2.pdf | Face Synthesis for Eyeglass-Robust Face Recognition | In the application of face recognition, eyeglasses could significantly degrade the recognition accuracy. A feasible method is to collect large-scale face images with eyeglasses for training deep learning methods. However, it is difficult to collect the images with and without glasses of the same identity, so that it is difficult to optimize the intra-variations caused by eyeglasses. In this paper, we propose to address this problem in a virtual synthesis manner. The high-fidelity face images with eyeglasses are synthesized based on 3D face model and 3D eyeglasses. Models based on deep learning methods are then trained on the synthesized eyeglass face dataset, achieving better performance than previous ones. Experiments on the real face database validate the effectiveness of our synthesized data for improving eyeglass face recognition performance. | ['Jianzhu Guo', 'Zhen Lei', 'Stan Z. Li', 'Xiangyu Zhu'] | 2018-06-04 | null | null | null | null | ['robust-face-recognition'] | ['computer-vision'] | [-2.71801889e-01 7.03622475e-02 2.69733012e-01 -4.24666971e-01
-1.37261033e-01 -1.60953570e-02 7.41807893e-02 -1.14897501e+00
2.76845992e-01 7.21956074e-01 -1.91168524e-02 1.76857203e-01
3.13726783e-01 -7.75315523e-01 -1.00798488e+00 -9.82531130e-01
3.37146789e-01 -1.18129283e-01 -4.71157014e-01 3.45974267e-02
1.16989212e-02 8.30827057e-01 -2.02602363e+00 7.91804418e-02
5.90600252e-01 1.15674663e+00 9.55400243e-02 8.33993852e-02
1.49473818e-02 2.91873723e-01 -1.00263727e+00 -3.69100332e-01
6.90092742e-01 -6.09717906e-01 -7.73817003e-02 9.59755659e-01
9.37071323e-01 -8.00086439e-01 -5.00894964e-01 1.08380127e+00
4.48622137e-01 1.87577377e-03 3.57025206e-01 -1.16129899e+00
-1.26808417e+00 -2.64710672e-02 -7.48877645e-01 -3.79313648e-01
3.58682394e-01 3.89541298e-01 2.31516719e-01 -1.04287112e+00
4.35132146e-01 1.56858015e+00 4.71103847e-01 1.22050011e+00
-1.06868386e+00 -1.01448524e+00 -8.54469761e-02 1.82298794e-02
-1.62169838e+00 -8.83902013e-01 7.06611216e-01 -3.06958437e-01
4.77291316e-01 -2.92412541e-03 6.53695226e-01 1.07607222e+00
2.30485335e-01 3.34809572e-01 1.18031967e+00 -4.77246314e-01
-1.72866344e-01 2.46260077e-01 -1.50148183e-01 1.03618157e+00
2.69147366e-01 5.96227527e-01 -6.44791424e-01 1.57139912e-01
1.06400394e+00 1.14990741e-01 -5.32676697e-01 -2.23572645e-03
-8.33896339e-01 5.07104218e-01 3.43527973e-01 -1.05180092e-01
-1.66753799e-01 -7.67571405e-02 -3.11613560e-01 4.13952231e-01
5.77853024e-01 2.63876438e-01 2.46974006e-02 5.84060311e-01
-9.42972481e-01 4.05520499e-02 5.43659627e-01 1.09294176e+00
9.26419675e-01 6.00971758e-01 -1.00801624e-01 1.10129833e+00
8.09939206e-01 7.77418911e-01 4.04111683e-01 -6.66520000e-01
7.44244084e-02 4.77755070e-01 4.06116173e-02 -1.05760729e+00
-1.32564127e-01 -1.09626174e-01 -7.92361498e-01 6.18474007e-01
4.77477640e-01 -7.44098276e-02 -1.52101946e+00 1.44046044e+00
4.07159269e-01 6.40798450e-01 1.47984162e-01 1.14055824e+00
1.30816495e+00 5.88756382e-01 -4.60301876e-01 -4.24277365e-01
1.19510138e+00 -9.14757848e-01 -1.17725861e+00 1.16069578e-01
1.74497858e-01 -9.09896195e-01 1.27486145e+00 3.45049143e-01
-1.03356576e+00 -7.94897258e-01 -1.26149237e+00 3.79054025e-02
1.22925654e-01 6.00273013e-01 2.63066441e-01 1.23111248e+00
-1.11345088e+00 5.37825942e-01 -6.88038826e-01 -1.49268106e-01
5.67995548e-01 4.94564176e-01 -5.18315494e-01 -6.67855144e-02
-8.42762470e-01 7.60113180e-01 -9.53969732e-02 5.83057404e-01
-1.24623728e+00 -6.46520495e-01 -8.55384767e-01 -1.16121210e-01
1.40908703e-01 -3.55718434e-01 9.33799326e-01 -9.52482343e-01
-2.09655261e+00 1.02848363e+00 -3.08788985e-01 8.94180238e-02
4.44039047e-01 -1.00659125e-01 -6.33396208e-01 -1.17529944e-01
-5.07593453e-01 5.08194983e-01 1.43426120e+00 -1.50686002e+00
-1.13261834e-01 -7.34306335e-01 -1.87606007e-01 -5.94899319e-02
-5.85814595e-01 -3.01147718e-02 -4.58756357e-01 -3.44578087e-01
1.28320560e-01 -1.07812333e+00 3.62280339e-01 1.03440680e-01
-3.24726164e-01 1.55672580e-01 1.15603280e+00 -7.64197588e-01
7.78315783e-01 -2.27788043e+00 -2.81838983e-01 5.62163927e-02
1.44511908e-01 4.35465455e-01 -2.34778956e-01 -8.70890766e-02
-3.88469622e-02 -9.55399945e-02 1.23950988e-02 -4.87079799e-01
-3.11766535e-01 1.14409402e-01 -2.81789720e-01 5.23162723e-01
4.27103698e-01 6.07203782e-01 -3.70086610e-01 -3.75262111e-01
1.49079353e-01 7.23654032e-01 -4.14130926e-01 4.78000998e-01
-6.45697489e-02 7.36944795e-01 -2.67734826e-01 9.78680909e-01
1.19693673e+00 -4.14635725e-02 3.14014554e-02 -2.94139504e-01
2.74378777e-01 -2.66757965e-01 -1.16224980e+00 1.21342504e+00
-5.85342526e-01 6.84917927e-01 -3.62079889e-02 -4.38483000e-01
1.47381568e+00 4.12150085e-01 2.56166816e-01 -5.89060783e-01
3.21087182e-01 2.00893924e-01 -2.19294410e-02 -6.35790586e-01
1.63327172e-01 -3.45971823e-01 6.86563909e-01 2.70774752e-01
5.32975756e-02 -3.38079184e-01 -3.93008068e-02 -5.35450518e-01
3.69875282e-01 3.06435376e-01 -1.34214610e-01 -2.53158748e-01
5.62728226e-01 -6.96305811e-01 6.10550106e-01 8.63345638e-02
-5.99620044e-02 9.05126512e-01 2.50311583e-01 -6.79752350e-01
-1.00262487e+00 -7.48641372e-01 -4.36290205e-01 9.93591547e-02
8.38284567e-02 -3.58531363e-02 -1.16884518e+00 -5.06044149e-01
3.68016660e-02 1.48686185e-01 -6.10716939e-01 -4.17632490e-01
-4.48113441e-01 -8.12377155e-01 4.81834918e-01 2.08804533e-01
8.33386362e-01 -9.54732120e-01 -1.42369613e-01 -1.36604309e-01
2.38587663e-01 -1.14126956e+00 -5.52492917e-01 -8.30332220e-01
-7.70349801e-01 -1.12848759e+00 -9.22626972e-01 -6.92219317e-01
1.07497501e+00 3.77137899e-01 7.68063843e-01 5.07351160e-01
-3.51649761e-01 7.79504627e-02 1.70383393e-03 -4.09862250e-01
-6.13674402e-01 -5.59593081e-01 4.81065065e-01 7.04281926e-01
5.15826583e-01 -3.50816995e-01 -4.50621843e-01 6.50189817e-01
-7.20765293e-01 -1.40753418e-01 2.76834249e-01 9.37903821e-01
4.20228273e-01 3.61822955e-02 5.60989678e-01 -6.64612234e-01
4.80753213e-01 -3.85156944e-02 -1.12164986e+00 3.98194492e-01
-5.42949438e-01 -1.29002169e-01 3.97274882e-01 -3.97301495e-01
-1.13074541e+00 1.37805924e-01 -5.42910323e-02 -9.95921135e-01
-3.50476950e-02 -3.99489924e-02 -6.66931927e-01 -5.30071199e-01
6.40141845e-01 -3.93987969e-02 3.82364839e-01 -3.66997719e-01
-7.51587376e-02 9.23035800e-01 2.12777138e-01 -3.45891833e-01
8.59398127e-01 4.86977994e-01 1.62163138e-01 -1.21681619e+00
-5.54905653e-01 3.47450793e-01 -3.72995734e-01 -3.98574352e-01
7.46962011e-01 -1.02022672e+00 -1.10200012e+00 9.22506571e-01
-1.04731035e+00 -2.09892601e-01 5.78133129e-02 7.33376384e-01
-2.74547249e-01 2.60341257e-01 -4.30185258e-01 -8.40920210e-01
-1.00442350e-01 -1.55977547e+00 1.16164935e+00 5.91487467e-01
4.10040438e-01 -5.87818861e-01 -2.81923890e-01 3.90742987e-01
1.83819309e-01 3.99208888e-02 3.53825271e-01 2.61984944e-01
-9.78322685e-01 -9.02935937e-02 -2.09775463e-01 7.02335358e-01
7.01140523e-01 5.00591755e-01 -1.49346662e+00 -3.63187164e-01
2.12308899e-01 -2.80241787e-01 6.66281223e-01 4.17227894e-01
1.32977867e+00 -1.49820432e-01 -2.27017224e-01 1.00270927e+00
1.10638011e+00 3.94830495e-01 1.09229398e+00 -1.89609960e-01
8.37818027e-01 6.68355942e-01 4.72814798e-01 3.40366393e-01
-8.46006945e-02 8.04581344e-01 2.83135653e-01 -1.28817067e-01
-5.70952594e-01 -1.85273215e-01 3.64652783e-01 5.34634590e-01
-2.55341321e-01 -1.71210140e-01 -6.29936278e-01 1.66452393e-01
-1.23438382e+00 -7.69850194e-01 -4.22347002e-02 2.40362430e+00
7.49356687e-01 -4.93565351e-01 -2.25762218e-01 -9.40839127e-02
8.84260654e-01 1.90035310e-02 -4.96233314e-01 -7.19574988e-02
-2.64956683e-01 1.35574296e-01 1.15370259e-01 5.20006478e-01
-8.06539416e-01 9.75546122e-01 6.77617455e+00 3.53725851e-01
-1.51364851e+00 -1.26652405e-01 7.93201685e-01 -3.24018270e-01
-1.42084479e-01 -2.10080311e-01 -1.39931488e+00 6.38517678e-01
7.14099705e-01 3.60032022e-02 6.73937380e-01 7.39505053e-01
2.63920784e-01 1.49058416e-01 -1.03415823e+00 1.32039964e+00
4.85984087e-01 -1.25243747e+00 1.29390648e-02 3.57671559e-01
1.08897281e+00 -6.07764661e-01 3.44183028e-01 -1.00499004e-01
-4.72363904e-02 -1.24661350e+00 3.55765462e-01 6.74596727e-01
1.10571778e+00 -5.08228123e-01 3.97332221e-01 -8.23401436e-02
-7.77121484e-01 1.19821392e-01 -6.59714282e-01 2.58088678e-01
-1.82357073e-01 5.28765678e-01 -8.75140011e-01 4.29608494e-01
7.25482166e-01 5.84448338e-01 -3.99290323e-01 8.99722636e-01
-2.11433738e-01 3.06223273e-01 -1.78118423e-01 6.94096386e-02
-2.48066440e-01 -4.74167585e-01 2.44329274e-01 4.02760953e-01
6.23951674e-01 4.47774045e-02 -3.78344357e-02 1.09384835e+00
-5.45684934e-01 -5.03145903e-02 -8.42387021e-01 -1.92187577e-01
3.62267762e-01 1.07710075e+00 -3.73590112e-01 -1.95624352e-01
-5.45079887e-01 7.27721691e-01 2.27332581e-02 5.76015651e-01
-6.93264604e-01 -1.55194104e-01 9.02722120e-01 3.32346767e-01
-1.33798607e-02 -1.92213923e-01 -1.20240986e-01 -1.32214653e+00
2.76044279e-01 -1.20228529e+00 -1.37851790e-01 -7.01265872e-01
-1.11450291e+00 9.94965374e-01 -3.46378803e-01 -1.32750702e+00
-4.08077277e-02 -8.32864702e-01 -4.30287480e-01 1.09972417e+00
-1.39400756e+00 -1.14568353e+00 -7.42377877e-01 7.05062628e-01
3.48282784e-01 -6.17125273e-01 7.69989848e-01 5.04497945e-01
-8.20414841e-01 9.45678115e-01 4.13144678e-02 -9.99793634e-02
8.33213270e-01 -5.14249921e-01 4.01652366e-01 8.06457698e-01
1.76567718e-01 6.39538765e-01 4.77226168e-01 -6.50434196e-01
-1.55449855e+00 -9.80562031e-01 3.21962088e-01 -5.47863543e-01
2.49007326e-02 -3.44250500e-01 -1.10946763e+00 7.51265883e-01
2.01056346e-01 3.33148181e-01 6.11769497e-01 -1.65815383e-01
-2.04624906e-01 -4.70717520e-01 -1.37501824e+00 5.97008288e-01
1.18054974e+00 -5.31045139e-01 -2.25236222e-01 4.57530916e-01
4.87760365e-01 -5.92647552e-01 -7.95530438e-01 3.91667336e-01
8.06389570e-01 -7.74543762e-01 6.94103003e-01 -5.72603524e-01
3.62656504e-01 -4.40200061e-01 -6.32048994e-02 -1.47353327e+00
2.16586635e-01 -7.81114578e-01 -9.55917537e-02 1.32077277e+00
1.81510314e-01 -9.72390234e-01 1.04770494e+00 8.60721111e-01
-9.45376158e-02 -3.94048274e-01 -5.74844837e-01 -1.04004610e+00
-2.93613136e-01 3.22863549e-01 1.31463420e+00 8.91179085e-01
-5.39443910e-01 1.01171397e-01 -7.00186670e-01 4.75596607e-01
7.44916439e-01 2.12701216e-01 1.20496988e+00 -1.16666377e+00
1.67180295e-03 -5.36946729e-02 -4.69970077e-01 -7.49274850e-01
5.93698800e-01 -5.83887637e-01 -6.03226684e-02 -1.04880655e+00
-8.82968828e-02 -2.67622232e-01 5.68517111e-02 2.76264518e-01
-2.51893640e-01 5.83967626e-01 1.28801242e-01 6.15369976e-02
4.40401912e-01 9.25334692e-01 1.79665935e+00 -7.79817030e-02
-2.53113836e-01 1.04393087e-01 -4.01121199e-01 4.78519380e-01
7.25753188e-01 -2.40475655e-01 -4.96951073e-01 -6.62254453e-01
-4.64847535e-01 1.80890448e-02 2.91983575e-01 -1.05607736e+00
-8.84924605e-02 -1.71530440e-01 6.96396351e-01 -3.17794859e-01
7.02823162e-01 -8.06945205e-01 3.54116201e-01 1.91720814e-01
1.67528138e-01 -3.18516850e-01 3.50410253e-01 2.76520461e-01
-3.68417978e-01 -1.69162408e-01 1.00469065e+00 -4.03776765e-02
-3.07602018e-01 8.23342860e-01 1.19737402e-01 -6.16515279e-01
9.95722592e-01 -3.66025418e-01 -2.44230807e-01 -2.29501098e-01
-7.01841533e-01 -2.61210531e-01 6.84934258e-01 6.08159423e-01
1.02316689e+00 -1.54858267e+00 -7.75382876e-01 1.16371310e+00
-2.70189680e-02 -2.64542997e-01 3.22951645e-01 3.66443634e-01
-6.71251833e-01 2.32373357e-01 -5.76055050e-01 -5.22737443e-01
-1.44073093e+00 7.04368949e-01 7.61331677e-01 4.63661045e-01
-5.48981547e-01 8.06167126e-01 5.15390098e-01 -4.15298760e-01
1.71955600e-01 -8.00539479e-02 -9.52870250e-02 -3.07969749e-01
9.10881162e-01 1.10727251e-01 3.36248010e-01 -6.82964265e-01
-9.99972448e-02 9.44076896e-01 -3.36183645e-02 5.16272224e-02
1.09520304e+00 5.22121340e-02 -9.19329375e-02 1.82128605e-02
1.22088027e+00 1.59421340e-02 -1.48017955e+00 -6.78758472e-02
-4.91257578e-01 -1.32530761e+00 4.86804284e-02 -1.75965115e-01
-1.65264213e+00 1.07484388e+00 9.05210793e-01 -4.96361256e-02
1.16727102e+00 -4.70994741e-01 6.74260080e-01 5.13307571e-01
4.09833252e-01 -8.63090634e-01 1.92867741e-01 -7.42204562e-02
1.21512926e+00 -1.37397468e+00 -6.17286898e-02 -5.14532983e-01
-3.81055027e-01 1.20424998e+00 9.16195631e-01 -2.00513065e-01
8.56875598e-01 1.19393803e-01 4.07395750e-01 -1.45886362e-01
-4.61472541e-01 3.24363005e-03 3.49037826e-01 5.88627219e-01
3.21609378e-01 -4.88494858e-02 -4.25036512e-02 3.02766450e-02
-1.75189525e-01 2.21330538e-01 5.71743667e-01 4.73852038e-01
-7.78180212e-02 -1.15772033e+00 -7.75116563e-01 2.78665185e-01
-3.77056956e-01 2.29141518e-01 -3.08475405e-01 7.59201050e-01
-8.83754566e-02 1.02749217e+00 9.86188725e-02 -4.48416680e-01
4.25254315e-01 -8.89714435e-03 9.90391135e-01 -6.16168678e-01
-1.43838271e-01 3.13067958e-02 -2.20397949e-01 -4.42094028e-01
-2.72432357e-01 -4.20623690e-01 -9.82260823e-01 -5.99578023e-01
-5.54556549e-01 -2.10843056e-01 7.29475796e-01 6.66745186e-01
3.90466124e-01 3.70206386e-01 1.06218684e+00 -7.13247418e-01
-5.79837084e-01 -9.71756756e-01 -9.74565029e-01 3.57800961e-01
4.58279878e-01 -9.18587446e-01 -3.30761552e-01 3.35527509e-01] | [12.924322128295898, 0.051265325397253036] |
dcf353eb-72a7-4c21-ae88-a152739002a3 | seeing-convolution-through-the-eyes-of-finite | 1905.10901 | null | https://arxiv.org/abs/1905.10901v1 | https://arxiv.org/pdf/1905.10901v1.pdf | Seeing Convolution Through the Eyes of Finite Transformation Semigroup Theory: An Abstract Algebraic Interpretation of Convolutional Neural Networks | Researchers are actively trying to gain better insights into the representational properties of convolutional neural networks for guiding better network designs and for interpreting a network's computational nature. Gaining such insights can be an arduous task due to the number of parameters in a network and the complexity of a network's architecture. Current approaches of neural network interpretation include Bayesian probabilistic interpretations and information theoretic interpretations. In this study, we take a different approach to studying convolutional neural networks by proposing an abstract algebraic interpretation using finite transformation semigroup theory. Specifically, convolutional layers are broken up and mapped to a finite space. The state space of the proposed finite transformation semigroup is then defined as a single element within the convolutional layer, with the acting elements defined by surrounding state elements combined with convolution kernel elements. Generators of the finite transformation semigroup are defined to complete the interpretation. We leverage this approach to analyze the basic properties of the resulting finite transformation semigroup to gain insights on the representational properties of convolutional neural networks, including insights into quantized network representation. Such a finite transformation semigroup interpretation can also enable better understanding outside of the confines of fixed lattice data structures, thus useful for handling data that lie on irregular lattices. Furthermore, the proposed abstract algebraic interpretation is shown to be viable for interpreting convolutional operations within a variety of convolutional neural network architectures. | ['Alexander Wong', 'Andrew Hryniowski'] | 2019-05-26 | null | null | null | null | ['network-interpretation'] | ['computer-vision'] | [ 7.48008907e-01 6.25719249e-01 -1.94698140e-01 -2.49111056e-01
1.23921543e-01 -6.21660292e-01 7.81451285e-01 6.25090897e-02
-3.50362420e-01 3.32293898e-01 1.53493762e-01 -6.23285592e-01
-2.75509000e-01 -7.96602309e-01 -8.26304615e-01 -9.80813265e-01
-2.87247062e-01 1.98099717e-01 5.60240373e-02 -9.37428400e-02
-1.48202227e-02 5.69578707e-01 -1.55808663e+00 3.29098344e-01
3.99478644e-01 1.12479281e+00 1.50435105e-01 6.08315587e-01
-2.02261850e-01 5.77773571e-01 -3.56709033e-01 -1.54942721e-01
3.00545156e-01 -3.62039655e-01 -7.67482221e-01 1.45747721e-01
1.15542803e-02 6.64220005e-02 -3.22421104e-01 1.28046703e+00
-1.83054388e-01 3.18672694e-02 8.82149637e-01 -9.36763525e-01
-5.31519949e-01 9.92796183e-01 4.60786819e-02 9.87843573e-02
-3.87209386e-01 -2.69893765e-01 1.29995394e+00 -4.85723823e-01
1.72477156e-01 1.32147050e+00 4.98631239e-01 5.41344523e-01
-1.68624687e+00 -3.92068267e-01 3.78958918e-02 -4.05127667e-02
-1.41841340e+00 -4.73642677e-01 7.54564166e-01 -6.50642633e-01
7.10762024e-01 2.15776995e-01 7.79785156e-01 8.59913170e-01
2.28915259e-01 4.51057315e-01 8.11107695e-01 -8.32414567e-01
6.80672646e-01 -1.52665332e-01 4.15981501e-01 9.64200377e-01
4.22198296e-01 -4.77675833e-02 -5.36729813e-01 1.55670382e-02
1.20918190e+00 -1.69751421e-01 -1.06231369e-01 -3.83108407e-01
-9.88651454e-01 9.29708421e-01 6.02463961e-01 2.30974197e-01
-2.16589302e-01 6.58839881e-01 4.21025932e-01 6.67486014e-03
1.57865748e-01 3.13507736e-01 -6.61551878e-02 2.21372277e-01
-5.24080932e-01 1.78015120e-02 8.78259420e-01 7.81480789e-01
7.56132841e-01 6.88786507e-02 5.96012995e-02 4.98135954e-01
5.58833599e-01 7.77884647e-02 1.00865304e-01 -9.88650382e-01
2.57924289e-01 4.51419830e-01 -4.62586939e-01 -8.70160520e-01
-2.74292141e-01 -6.31891012e-01 -1.37352526e+00 7.99272656e-02
5.10163546e-01 1.33884594e-01 -7.29004145e-01 1.90433121e+00
-3.56612980e-01 2.01962981e-02 2.97661722e-01 5.13942182e-01
2.38353610e-01 5.99624276e-01 -1.18717581e-01 4.40391190e-02
1.62638855e+00 -4.35892433e-01 -6.16087914e-01 -1.93768308e-01
7.74356544e-01 -1.73312753e-01 7.87675440e-01 3.70913684e-01
-8.95275474e-01 -3.90302569e-01 -1.42698061e+00 8.42969418e-02
-1.36689052e-01 4.08719093e-01 6.90101981e-01 8.42634916e-01
-1.07776213e+00 4.65491384e-01 -1.14314175e+00 -9.43543538e-02
7.67594695e-01 5.68811476e-01 -5.32149300e-02 1.43583566e-01
-1.22761464e+00 6.17061257e-01 9.47477877e-01 5.03273189e-01
-5.37117064e-01 -3.22287202e-01 -8.90940130e-01 3.66279155e-01
1.85831740e-01 -5.59440851e-01 1.16527665e+00 -7.57690787e-01
-1.24619615e+00 2.47355744e-01 -1.98549271e-01 -8.28177571e-01
1.21348798e-01 3.78401488e-01 -2.43137274e-02 4.78067324e-02
-2.78744102e-01 6.42853856e-01 8.54810834e-01 -9.13576663e-01
-1.77526891e-01 -3.19640964e-01 1.97812647e-01 1.84946861e-02
-4.84924972e-01 -3.70755851e-01 -1.82133690e-01 -5.91664493e-01
7.54423738e-01 -1.08726740e+00 -4.96467710e-01 6.68516159e-02
-5.11379182e-01 -2.35950679e-01 5.64766526e-01 -1.21690020e-01
1.08484268e+00 -2.25899792e+00 1.26752585e-01 5.38692415e-01
4.22820210e-01 1.51773363e-01 1.51494160e-01 1.10586427e-01
-2.94871867e-01 4.96245503e-01 -1.60744384e-01 -1.21261224e-01
1.01890512e-01 5.08802593e-01 -5.03085554e-01 3.82899702e-01
5.25941193e-01 8.02395165e-01 -6.74458086e-01 4.78125438e-02
5.93320616e-02 5.62747777e-01 -6.67816281e-01 -2.81189978e-01
-3.88071924e-01 2.39245862e-01 -5.93641698e-01 1.61893549e-04
3.27453852e-01 -3.55293393e-01 4.87995744e-01 -3.65324050e-01
-9.06981602e-02 4.88330185e-01 -1.33669078e+00 1.41800189e+00
-4.30441529e-01 1.02176726e+00 -2.00678796e-01 -1.32009506e+00
1.09720552e+00 3.04555893e-01 -5.36623457e-03 -1.80922851e-01
2.05576465e-01 1.27168028e-02 4.37449813e-01 3.36458497e-02
4.78021055e-01 -2.69528121e-01 -1.15228325e-01 6.88500464e-01
2.99701132e-02 8.93056020e-02 -1.56388823e-02 2.11582318e-01
1.08063281e+00 -4.61898029e-01 2.94557899e-01 -6.23316467e-01
4.10547912e-01 -5.11539400e-01 3.35767090e-01 9.23447907e-01
1.63903594e-01 3.52018684e-01 1.12898910e+00 -6.01459384e-01
-1.19826615e+00 -1.24254620e+00 -3.94875377e-01 6.73247933e-01
-2.91098773e-01 -5.23124337e-01 -9.14750934e-01 -2.70123519e-02
-5.49543500e-01 3.48778933e-01 -6.81813061e-01 -4.93955642e-01
-3.27879846e-01 -7.36628771e-01 1.08809626e+00 7.01699436e-01
8.84447753e-01 -8.05976391e-01 -7.28910923e-01 9.17411894e-02
2.97191031e-02 -1.19510961e+00 -4.37956080e-02 6.78634346e-01
-1.03728640e+00 -7.50955582e-01 -1.28963754e-01 -6.92824662e-01
9.41153526e-01 -1.32584229e-01 5.85040450e-01 -7.88496882e-02
-1.77666381e-01 1.03222638e-01 2.18424872e-02 -4.54126060e-01
-8.06208253e-01 2.03791872e-01 2.52013743e-01 2.06756949e-01
2.47309525e-02 -5.43758690e-01 -1.17011964e-01 1.57342076e-01
-1.42868137e+00 2.98139542e-01 6.50443912e-01 9.79047239e-01
3.14839751e-01 5.78832686e-01 2.10297599e-01 -7.92635500e-01
6.55668736e-01 -2.24326700e-01 -7.66418934e-01 1.94392994e-01
-2.32060000e-01 9.15272355e-01 5.93794346e-01 -3.99436653e-01
-9.68266845e-01 1.87312305e-01 2.12994322e-01 -1.85604334e-01
3.24031450e-02 9.40932751e-01 -2.25511327e-01 1.18987657e-01
8.46577704e-01 5.38322143e-02 2.23327160e-01 -2.20782936e-01
3.53037328e-01 4.69885319e-01 4.96567816e-01 -7.44490921e-01
5.05861461e-01 5.30033708e-01 4.99723464e-01 -1.03991783e+00
-8.25051188e-01 -4.48432863e-02 -1.01081657e+00 -6.54686838e-02
8.97676587e-01 -6.31337345e-01 -8.80838692e-01 5.05861163e-01
-1.46259582e+00 -1.90308735e-01 -2.68768251e-01 4.00040627e-01
-5.86041808e-01 6.29296452e-02 -6.11166418e-01 -8.82050276e-01
2.05209628e-01 -1.41061783e+00 5.64439952e-01 -1.46269247e-01
-2.03312501e-01 -1.32595193e+00 -3.39767814e-01 1.81307301e-01
1.98862016e-01 -4.35986407e-02 1.49111772e+00 -7.23276079e-01
-8.63550842e-01 -1.72256857e-01 -2.04515249e-01 7.07655787e-01
-6.52549714e-02 -1.08563274e-01 -1.05505359e+00 -3.40296812e-02
1.91707686e-01 -1.54058143e-01 1.15411150e+00 2.65479356e-01
1.25533772e+00 -4.14417684e-01 -8.14751610e-02 3.83677602e-01
1.18844151e+00 7.28309900e-03 6.11954033e-01 1.52352490e-02
5.11408567e-01 6.30148768e-01 -1.49208367e-01 4.24024016e-01
-7.16318935e-02 4.58957553e-01 6.64621055e-01 2.91366160e-01
1.73959732e-01 -1.56915635e-01 3.56764019e-01 9.72189069e-01
-2.74029434e-01 -7.81610310e-02 -1.18484032e+00 2.17043683e-01
-1.83802688e+00 -8.30413461e-01 -5.85468151e-02 2.09844494e+00
6.55666649e-01 4.34521377e-01 -4.67495143e-01 4.64583874e-01
7.79021859e-01 2.71044493e-01 -4.26917821e-01 -5.43848574e-01
-1.08096033e-01 4.09562767e-01 6.06780887e-01 2.56823808e-01
-8.98344934e-01 5.72198153e-01 6.49964190e+00 8.31810594e-01
-9.43450391e-01 -1.29942641e-01 6.09518647e-01 4.37480748e-01
-2.78623611e-01 2.04284295e-01 -9.05864954e-01 -4.77709770e-02
1.08534122e+00 -4.07417044e-02 5.70256174e-01 4.54682499e-01
7.25001693e-02 -9.35858786e-02 -1.31145346e+00 7.27073789e-01
-2.06366092e-01 -1.81408226e+00 3.57661784e-01 5.08076191e-01
5.46721399e-01 -9.31742713e-02 2.30402514e-01 -2.29838625e-01
2.47379035e-01 -1.34174454e+00 9.80291545e-01 4.77197379e-01
7.66340375e-01 -8.10473919e-01 4.82365221e-01 4.69444513e-01
-1.22234929e+00 -5.39005883e-02 -5.28279722e-01 -5.55364370e-01
-3.01968932e-01 5.41654408e-01 -1.19297719e+00 1.64213568e-01
2.94488430e-01 6.56840384e-01 -3.90780270e-01 4.79561210e-01
-7.18507469e-02 7.59988368e-01 -3.80823135e-01 -5.74077219e-02
3.36030632e-01 -2.59163469e-01 2.78787047e-01 9.56108451e-01
2.47716635e-01 1.15757219e-01 -1.55883297e-01 1.38765895e+00
-8.40590298e-02 -5.08623898e-01 -7.88014710e-01 -3.63973856e-01
4.38698530e-01 1.04699874e+00 -1.18180788e+00 -6.84545115e-02
-1.21276796e-01 4.16311771e-01 3.50737870e-01 3.54336888e-01
-5.52193284e-01 -1.56112179e-01 6.54563189e-01 -4.04063091e-02
2.71166414e-01 -5.94012320e-01 -5.74544728e-01 -9.74944890e-01
-1.90945957e-02 -3.79162937e-01 -8.19562897e-02 -4.85875100e-01
-6.70313060e-01 4.97400016e-01 2.31017828e-01 -9.81713653e-01
-2.21373662e-01 -1.01702690e+00 -3.73491645e-01 9.57331836e-01
-8.52061391e-01 -8.17001939e-01 5.22234775e-02 3.33877385e-01
1.80648774e-01 -2.52102494e-01 1.06455696e+00 -2.43000507e-01
-5.36283970e-01 4.52958047e-01 2.73449063e-01 4.74568576e-01
-2.83764958e-01 -1.00700545e+00 4.13818240e-01 7.35793531e-01
4.09489244e-01 9.76058304e-01 5.19590199e-01 -3.06323558e-01
-1.34095180e+00 -1.21657407e+00 3.63636792e-01 -1.58075109e-01
8.31273735e-01 -7.55923688e-01 -9.43002343e-01 7.71263063e-01
-2.00041935e-01 1.65429771e-01 6.39226735e-01 1.75377563e-01
-5.11320055e-01 1.73670612e-02 -6.17359459e-01 7.68949509e-01
9.93826151e-01 -8.19458902e-01 -4.12469536e-01 2.19019018e-02
6.13699615e-01 -2.25427926e-01 -7.49696136e-01 2.22628444e-01
5.81239045e-01 -7.96995521e-01 8.32726121e-01 -5.89850307e-01
4.03040975e-01 -2.30929911e-01 -2.33010754e-01 -1.17665851e+00
-2.24465996e-01 -4.60521400e-01 1.89815313e-01 9.00541246e-01
3.92955661e-01 -6.80339515e-01 8.78957748e-01 4.66867268e-01
-1.32216468e-01 -6.91902220e-01 -1.01718998e+00 -7.04131007e-01
3.16324309e-02 -9.39394176e-01 3.04960907e-01 5.39001822e-01
3.35248820e-02 4.01889980e-01 1.24363236e-01 3.22907239e-01
6.33849621e-01 -5.36435306e-01 5.59000596e-02 -1.41877770e+00
-3.54976743e-01 -5.02142668e-01 -7.69845903e-01 -1.10824251e+00
4.17791039e-01 -1.24118769e+00 1.11524396e-01 -9.84301150e-01
3.90082598e-02 -8.19130063e-01 -2.23633736e-01 4.65118527e-01
7.15770423e-01 2.77269751e-01 1.80078283e-01 3.36766928e-01
-1.94758043e-01 3.72526199e-01 9.33880925e-01 -2.13401653e-02
-2.73306412e-03 2.04085067e-01 -6.58114791e-01 9.56206799e-01
7.74245024e-01 -2.63261855e-01 -6.95267797e-01 -4.56586152e-01
6.50755048e-01 -5.61053772e-03 7.27190375e-01 -1.07607567e+00
4.91193235e-01 8.69735107e-02 1.35272890e-01 -2.42509067e-01
4.22893852e-01 -5.86754441e-01 2.86266744e-01 5.34990013e-01
-8.53566468e-01 -1.97386563e-01 3.25160772e-02 8.51100445e-01
-1.40380949e-01 -6.06486499e-01 6.39956474e-01 -9.12396237e-02
-4.83545989e-01 3.40981930e-02 -7.40346909e-01 -8.91662762e-02
6.62871838e-01 -2.56705254e-01 -4.73124646e-02 -2.20528990e-01
-1.13918102e+00 -2.45686129e-01 1.45989373e-01 1.25897035e-01
6.17653012e-01 -1.41347933e+00 -1.38264790e-01 6.58160210e-01
-1.39460579e-01 4.36357796e-01 -9.18068215e-02 6.29295647e-01
-5.78155518e-01 7.15798676e-01 -2.02586398e-01 -8.68184507e-01
-9.93282139e-01 -1.45717442e-01 3.89244854e-01 -1.33623496e-01
-3.38332266e-01 6.10419273e-01 4.28458333e-01 -1.48004323e-01
2.84783334e-01 -9.11258876e-01 -9.08632278e-02 5.56947812e-02
3.63330483e-01 1.65581442e-02 1.81955487e-01 -5.65667033e-01
-9.55550671e-02 1.72543973e-01 -5.14438339e-02 -4.57791507e-01
1.11278570e+00 2.35150293e-01 -4.57063973e-01 6.60346270e-01
1.19205177e+00 -6.55525088e-01 -1.15846419e+00 -5.14911115e-01
2.22518712e-01 1.66362286e-01 1.91536769e-01 -1.21148139e-01
-8.10541213e-01 1.02164972e+00 2.49637157e-01 4.30575669e-01
9.39077973e-01 2.81662643e-01 4.39598039e-03 9.35975432e-01
2.74498940e-01 -7.31618702e-01 1.68997779e-01 9.81116712e-01
6.25841320e-01 -6.33823693e-01 -2.11044446e-01 -6.02372169e-01
-6.53335005e-02 1.63902247e+00 3.56850103e-02 -2.44859785e-01
9.49594796e-01 3.37420493e-01 -4.39424604e-01 -1.10229902e-01
-8.76794815e-01 1.88856199e-02 3.82633656e-01 3.69143993e-01
3.50511789e-01 1.77289054e-01 1.71062604e-01 7.41964102e-01
-4.42499310e-01 -2.32991397e-01 9.23162460e-01 7.28446960e-01
-4.36265111e-01 -1.01783502e+00 -2.33368844e-01 5.07510185e-01
-1.93578899e-01 -3.57459873e-01 -4.83834893e-02 4.74258125e-01
5.45003936e-02 4.96314049e-01 4.93705988e-01 -3.14232677e-01
-8.63596126e-02 1.71717137e-01 5.21944821e-01 -9.42251265e-01
2.44101956e-02 -8.74936879e-02 -8.72296765e-02 -6.23273775e-02
-5.76177597e-01 -7.78361142e-01 -1.50475657e+00 -8.53485540e-02
-4.09868211e-01 7.16935694e-02 8.75614464e-01 1.24236405e+00
1.50016367e-01 7.37894833e-01 3.99022639e-01 -7.21350431e-01
-7.97104955e-01 -6.24717951e-01 -6.73508346e-01 -2.36852348e-01
3.98276746e-01 -5.03725827e-01 -2.14237362e-01 2.31066659e-01] | [8.008162498474121, 3.574040174484253] |
c11d8c0f-baf9-454d-a161-55f69f568ddb | perfectly-accurate-membership-inference-by-a | 2203.16463 | null | https://arxiv.org/abs/2203.16463v1 | https://arxiv.org/pdf/2203.16463v1.pdf | Perfectly Accurate Membership Inference by a Dishonest Central Server in Federated Learning | Federated Learning is expected to provide strong privacy guarantees, as only gradients or model parameters but no plain text training data is ever exchanged either between the clients or between the clients and the central server. In this paper, we challenge this claim by introducing a simple but still very effective membership inference attack algorithm, which relies only on a single training step. In contrast to the popular honest-but-curious model, we investigate a framework with a dishonest central server. Our strategy is applicable to models with ReLU activations and uses the properties of this activation function to achieve perfect accuracy. Empirical evaluation on visual classification tasks with MNIST, CIFAR10, CIFAR100 and CelebA datasets show that our method provides perfect accuracy in identifying one sample in a training set with thousands of samples. Occasional failures of our method lead us to discover duplicate images in the CIFAR100 and CelebA datasets. | ['Pablo Piantanida', 'Leonardo Rey Vega', 'Marco Romanelli', 'Georg Pichler'] | 2022-03-30 | null | null | null | null | ['inference-attack', 'membership-inference-attack'] | ['adversarial', 'computer-vision'] | [-5.07189706e-02 8.88337120e-02 1.39871925e-01 -4.45300788e-01
-8.92872155e-01 -1.04313457e+00 5.00265300e-01 -1.02625638e-01
-9.04377878e-01 9.86941278e-01 -5.73280811e-01 -5.45057118e-01
2.06774892e-03 -5.44434905e-01 -1.15554249e+00 -1.09758008e+00
-2.26202950e-01 6.19516075e-01 2.48586666e-03 2.21204177e-01
-7.98260942e-02 3.60055923e-01 -9.69896853e-01 5.20822227e-01
2.52498150e-01 1.25987840e+00 -4.29238886e-01 7.99401581e-01
2.39596754e-01 9.70739365e-01 -6.48299992e-01 -1.07272959e+00
8.22107196e-01 -1.87458277e-01 -1.09019542e+00 -3.56249660e-01
5.06537020e-01 -6.48402572e-01 -3.56698841e-01 1.26354682e+00
4.13378984e-01 -3.44574898e-01 1.34728685e-01 -1.71013129e+00
-4.39091504e-01 1.07324457e+00 -2.37160087e-01 -6.88609630e-02
-2.09579051e-01 1.55477792e-01 9.65573967e-01 -3.43959719e-01
6.60169482e-01 6.38266981e-01 6.81026697e-01 6.03866100e-01
-1.27925467e+00 -9.11717176e-01 -2.19414219e-01 2.79015929e-01
-1.54107177e+00 -5.76878607e-01 2.54710406e-01 -9.35638994e-02
5.76920092e-01 7.25862980e-01 -1.34392111e-02 1.21986461e+00
-1.26432121e-01 5.77305198e-01 1.42544222e+00 -3.35505694e-01
5.00494301e-01 6.95693314e-01 2.00781643e-01 3.87290478e-01
5.65572858e-01 2.60215905e-03 -5.63620806e-01 -1.09254444e+00
3.81198764e-01 1.24723047e-01 -6.44445837e-01 -4.55482155e-01
-9.53992307e-01 9.21039462e-01 1.41747907e-01 1.54226393e-01
-1.81732222e-01 4.71600831e-01 5.12093306e-01 7.40904093e-01
3.37964863e-01 1.73686787e-01 -6.22178793e-01 2.47273445e-01
-7.78737783e-01 2.27295220e-01 1.30381835e+00 6.76306427e-01
6.79592431e-01 -1.93927824e-01 1.14336759e-01 2.05491826e-01
-4.13714759e-02 4.60272640e-01 4.12295759e-01 -1.28519309e+00
2.71619380e-01 -1.42134905e-01 3.92281562e-01 -7.86149323e-01
1.44802868e-01 -3.57609034e-01 -7.51354098e-01 4.30399597e-01
5.54315448e-01 -2.21872449e-01 -2.78877974e-01 1.79728103e+00
5.09960353e-01 1.59495458e-01 2.98381180e-01 1.00434566e+00
1.06971331e-01 2.09729359e-01 -3.01831841e-01 -2.81631667e-02
1.25613534e+00 -6.42940819e-01 -3.90165210e-01 2.83549905e-01
6.36115313e-01 -4.84210819e-01 6.40102446e-01 7.19929397e-01
-1.03011155e+00 2.32264444e-01 -8.90179038e-01 7.16489460e-03
-3.01479042e-01 -1.69170171e-01 8.39706242e-01 9.37676370e-01
-1.01893508e+00 6.56144202e-01 -7.24775136e-01 7.42899776e-02
9.53890145e-01 5.92485309e-01 -9.27799940e-01 -8.10682103e-02
-1.04248488e+00 3.51042598e-01 3.56957197e-01 -2.51065493e-01
-1.09103382e+00 -5.78239202e-01 -3.66888940e-01 1.30252525e-01
3.10872912e-01 -5.86306930e-01 1.39849377e+00 -1.17222846e+00
-1.24443984e+00 1.12723768e+00 2.48689741e-01 -1.20264614e+00
1.14700603e+00 1.98843896e-01 -1.68730877e-02 3.95756215e-01
-3.44265968e-01 1.82512090e-01 9.15215313e-01 -1.37890625e+00
-7.69333482e-01 -6.52654946e-01 6.49603084e-02 -3.63714308e-01
-4.48827833e-01 -1.52108390e-02 1.46581128e-01 -2.26376906e-01
-1.24755658e-01 -8.97067904e-01 -7.61549622e-02 3.74771029e-01
-5.82281828e-01 -1.39079556e-01 8.67869437e-01 -4.40461338e-01
4.30863559e-01 -2.22631812e+00 -6.15950108e-01 5.61054170e-01
4.75649089e-01 2.93109357e-01 6.29019067e-02 2.27008820e-01
1.65197432e-01 3.64530951e-01 -1.29925534e-01 -7.93434262e-01
2.99470305e-01 2.74247259e-01 -6.81071520e-01 1.04724741e+00
-5.49043715e-01 6.98984921e-01 -5.99573433e-01 -2.88328230e-01
-4.76682037e-01 3.82413417e-01 -4.59557295e-01 8.23660642e-02
-2.14712620e-01 1.49784401e-01 -2.33463228e-01 3.73023361e-01
1.02017736e+00 -7.57489622e-01 4.22104388e-01 2.57522464e-01
3.02870721e-01 2.40292668e-01 -1.19085991e+00 1.24924839e+00
-2.40523770e-01 3.81444871e-01 6.63223684e-01 -8.38118255e-01
4.64005858e-01 6.76267147e-01 2.30277643e-01 -4.25156981e-01
3.31727564e-01 3.74599397e-01 -2.53116101e-01 -3.42299998e-01
-4.65312414e-02 4.34453674e-02 1.54439941e-01 1.09294295e+00
1.62532330e-02 6.27474964e-01 -5.90855539e-01 4.18381572e-01
1.26641679e+00 -4.27558422e-01 -1.53804615e-01 -5.80037460e-02
3.21411788e-01 -7.78283700e-02 5.36492527e-01 1.18770230e+00
-3.86266351e-01 4.15973395e-01 5.60568273e-01 -6.69423163e-01
-1.08893478e+00 -8.39969933e-01 -3.72018293e-02 1.06853342e+00
-9.45955422e-03 -3.98763597e-01 -7.65553653e-01 -9.91610169e-01
2.42455959e-01 6.55247748e-01 -7.07179904e-01 9.21275765e-02
-9.04516354e-02 -5.97712994e-01 1.02364755e+00 -8.08693469e-02
8.07473302e-01 -5.77432632e-01 -5.88584125e-01 -1.88386530e-01
-2.31320500e-01 -1.21832788e+00 -3.39745939e-01 1.61868528e-01
-5.76626837e-01 -1.25822008e+00 -2.34396040e-01 -3.83893162e-01
6.39422238e-01 2.75412291e-01 8.45744967e-01 1.22284703e-01
-3.26273173e-01 4.34075505e-01 3.72201367e-03 -5.15073299e-01
-5.94915450e-01 -1.08109519e-01 2.12025382e-02 3.90656650e-01
3.31808269e-01 -6.38045311e-01 -6.93786144e-01 1.53987765e-01
-9.91129518e-01 -3.23277920e-01 1.93310708e-01 8.23704481e-01
4.16331351e-01 -1.14494815e-01 3.37606639e-01 -1.43828094e+00
3.07008386e-01 -5.86257100e-01 -9.59177077e-01 4.02293265e-01
-8.62823427e-01 -1.23496369e-01 9.55500066e-01 -4.60477412e-01
-6.67244673e-01 2.85580784e-01 2.62066782e-01 -8.24838221e-01
-2.27882087e-01 -1.51734307e-01 4.47840616e-02 -4.70808506e-01
7.79601693e-01 3.56176049e-01 2.74892271e-01 -6.27540469e-01
3.35129589e-01 8.14347088e-01 7.98695922e-01 -4.79054481e-01
7.21514940e-01 1.05110419e+00 9.29886103e-03 -3.61545593e-01
-2.12048322e-01 -1.39166087e-01 -1.51524201e-01 3.46442163e-01
2.91129559e-01 -8.53531957e-01 -1.58560836e+00 6.78503573e-01
-1.22310472e+00 -3.19653928e-01 -3.77414137e-01 5.11172056e-01
-3.52102935e-01 5.15193462e-01 -9.10706043e-01 -9.99692023e-01
-7.34269142e-01 -7.57878661e-01 4.67160881e-01 -1.57441050e-01
1.81578368e-01 -7.75017142e-01 -1.74929097e-01 6.48222625e-01
5.85794032e-01 9.97127816e-02 5.26969373e-01 -1.42926848e+00
-8.84618104e-01 -6.12447083e-01 -6.80491552e-02 5.57490230e-01
-2.20913634e-01 -2.65726686e-01 -1.51060748e+00 -5.30354559e-01
3.63169074e-01 -8.44985902e-01 6.76389635e-01 -1.95387870e-01
1.53087974e+00 -1.24302185e+00 -2.72536147e-02 9.29039538e-01
1.51231813e+00 -2.23238617e-01 4.88479704e-01 2.47370511e-01
3.52811277e-01 4.60351586e-01 -6.24898560e-02 7.11413741e-01
2.85313636e-01 2.72634864e-01 7.02391982e-01 1.41346306e-01
5.51828027e-01 -1.10295735e-01 1.83439940e-01 -6.69967607e-02
1.64119706e-01 -9.78814662e-02 -4.86299545e-01 4.41680998e-01
-1.72044551e+00 -1.14018476e+00 1.02988221e-02 2.58110285e+00
1.08823478e+00 -2.93193579e-01 -8.32064748e-02 -8.38541165e-02
5.00310302e-01 -1.96101800e-01 -5.42796671e-01 -3.32511425e-01
-2.60504544e-01 2.45173588e-01 1.24842525e+00 3.31579506e-01
-9.70576644e-01 5.66425622e-01 5.92913246e+00 7.69779921e-01
-1.30662405e+00 6.68663323e-01 1.07123494e+00 -4.86303389e-01
-2.20178843e-01 7.11042210e-02 -4.23518091e-01 4.72722888e-01
1.02214146e+00 -1.68293253e-01 8.43790650e-01 1.13564968e+00
-3.47576886e-01 9.24380198e-02 -1.26725709e+00 1.22060835e+00
-1.83549747e-01 -1.60676038e+00 -4.45912242e-01 3.45028371e-01
5.53649485e-01 3.93893570e-01 2.40312189e-01 -1.18288010e-01
9.65912044e-01 -1.13317513e+00 5.61586082e-01 3.31488967e-01
7.60620296e-01 -8.35396886e-01 7.11346209e-01 6.77863419e-01
-2.61058539e-01 -1.52233899e-01 -5.53086996e-01 3.20792109e-01
-4.26207393e-01 3.03723484e-01 -9.95770633e-01 4.32520747e-01
1.09120083e+00 -3.34130883e-01 -3.91513467e-01 9.75907803e-01
1.18278051e-02 8.59358907e-01 -7.89571583e-01 2.01945648e-01
1.53238401e-01 1.06885649e-01 2.85049677e-01 8.40494037e-01
-1.10259177e-02 2.80585862e-03 -9.09615383e-02 9.29474294e-01
-7.50067949e-01 1.61533430e-02 -5.99825621e-01 2.43323222e-01
6.55154228e-01 1.31082165e+00 -1.52881101e-01 -3.52651656e-01
-6.24818448e-03 1.40568805e+00 4.68918979e-01 3.29352766e-01
-6.69245243e-01 -1.91840321e-01 6.97214007e-01 -1.92126200e-01
4.86445040e-01 6.09879792e-02 -1.13146342e-01 -1.16726470e+00
1.08363979e-01 -1.10104501e+00 7.16092169e-01 -2.78117448e-01
-1.70051992e+00 6.87807977e-01 -4.20052111e-01 -6.50000513e-01
-1.81806743e-01 -3.01011741e-01 -5.66924274e-01 8.56377423e-01
-1.57094479e+00 -1.23293936e+00 7.82747343e-02 1.20745492e+00
-5.37996888e-01 -2.19436526e-01 1.05177248e+00 2.32302561e-01
-3.37501854e-01 1.26714158e+00 6.49578869e-01 5.30451715e-01
8.08706403e-01 -1.06544471e+00 1.80735275e-01 7.37516463e-01
4.73824680e-01 8.00931633e-01 6.28846824e-01 -1.87951148e-01
-1.50663865e+00 -1.08458567e+00 9.45110917e-01 -2.85549343e-01
4.95885938e-01 -6.63471282e-01 -9.29145038e-01 9.97917950e-01
1.39812469e-01 7.18239307e-01 8.57009947e-01 -3.00618052e-01
-1.02702975e+00 -4.79268610e-01 -1.63698256e+00 6.04494959e-02
6.02702498e-01 -8.33833456e-01 -2.89162658e-02 7.77265489e-01
6.66593373e-01 -2.01048270e-01 -5.77900767e-01 -2.32582301e-01
5.31051874e-01 -1.02755940e+00 6.98147118e-01 -1.01688957e+00
-2.84284204e-01 -1.06271252e-01 -3.80702347e-01 -7.64586389e-01
1.16219305e-01 -1.10099113e+00 -3.09230119e-01 1.11466348e+00
2.93320835e-01 -1.27406907e+00 1.04035008e+00 1.06642139e+00
7.57457137e-01 -2.61857003e-01 -1.50572956e+00 -8.19165707e-01
1.79252729e-01 -2.62182027e-01 7.61441469e-01 1.18110085e+00
-1.88043565e-01 -2.61332124e-01 -5.84433556e-01 4.75543201e-01
1.21444130e+00 2.38161817e-01 1.00320637e+00 -1.00689173e+00
-7.38487959e-01 2.46329866e-02 -2.26697385e-01 -5.58694065e-01
3.20494562e-01 -9.99370873e-01 -2.41574913e-01 -6.22263789e-01
2.87139803e-01 -9.58942056e-01 -3.78915191e-01 9.96781647e-01
3.87365401e-01 5.46947479e-01 2.13848963e-01 5.64340472e-01
-6.94227815e-01 1.66860774e-01 4.68963087e-01 -1.42908931e-01
4.12606180e-01 4.63130206e-01 -7.36422896e-01 1.99870199e-01
8.59125435e-01 -8.93500924e-01 -2.76145041e-01 -3.68391305e-01
3.14263225e-01 -1.29853562e-01 1.09054375e+00 -7.72194445e-01
7.20316172e-01 1.81584433e-01 2.79940754e-01 -1.06378041e-01
1.48409992e-01 -1.32826138e+00 4.84783560e-01 5.89601040e-01
-5.46763778e-01 -1.78990185e-01 -3.43431056e-01 7.45307565e-01
2.45190069e-01 -2.15559348e-01 8.36186886e-01 -2.18290761e-01
-3.33180353e-02 5.47303081e-01 1.33917540e-01 -1.50377870e-01
1.23932004e+00 2.72626132e-01 -8.59040260e-01 -5.73003411e-01
-6.73233628e-01 2.75450617e-01 7.76347637e-01 -1.33042306e-01
2.98652947e-01 -8.11799586e-01 -8.43100011e-01 2.67665118e-01
-1.16017312e-01 -1.52199790e-01 1.92080840e-01 8.53677750e-01
-4.08897072e-01 2.18586683e-01 -1.05838962e-01 -4.10610020e-01
-1.58680463e+00 7.71942616e-01 6.57343507e-01 -1.07631512e-01
-5.59235275e-01 8.69805038e-01 -7.43315890e-02 -3.96713585e-01
5.62523723e-01 2.18039617e-01 6.61792874e-01 -3.00390631e-01
8.37486982e-01 3.34245682e-01 2.62509018e-01 -4.09318924e-01
-2.90544331e-01 -4.65736210e-01 -2.49747768e-01 -3.17415535e-01
1.32110262e+00 -1.30255118e-01 -4.50182289e-01 2.26938844e-01
1.64658499e+00 7.14471936e-02 -1.13015842e+00 -4.50753182e-01
-1.96618617e-01 -7.03269601e-01 -7.19475746e-02 -7.81884313e-01
-1.19866598e+00 7.49065101e-01 8.12097430e-01 2.75123864e-01
8.22928727e-01 6.43664179e-03 8.95827651e-01 7.39573300e-01
8.97095740e-01 -7.13309944e-01 -5.33326089e-01 4.27681617e-02
3.13266635e-01 -1.24491966e+00 -2.16775835e-01 8.82840529e-02
-5.96741080e-01 9.51824546e-01 2.88384199e-01 8.38189200e-02
5.26218235e-01 3.03327680e-01 1.95729122e-01 6.64688647e-02
-1.12211549e+00 5.64148784e-01 -5.12458742e-01 4.01118070e-01
-3.38256419e-01 2.75143213e-03 -8.17877147e-03 7.79746234e-01
-2.03666892e-02 1.91310063e-01 4.61725235e-01 1.04486871e+00
-1.38554424e-01 -1.22994792e+00 -4.84327793e-01 2.08060846e-01
-1.30686617e+00 -1.13789544e-01 -3.53523016e-01 5.72147846e-01
-4.58960459e-02 7.34800756e-01 -4.51654904e-02 -1.76975608e-01
-2.60233253e-01 3.30749542e-01 1.06615163e-01 -3.77870277e-02
-1.04297781e+00 -3.24311256e-01 -2.29175061e-01 -7.89456367e-01
-3.70348208e-02 -3.77640009e-01 -1.16403401e+00 -8.85459781e-01
-3.38075221e-01 4.58014846e-01 1.00611949e+00 6.43395245e-01
6.37284279e-01 -5.33525765e-01 1.25276554e+00 -1.61893964e-01
-1.33858263e+00 -5.39372087e-01 -9.70785081e-01 5.77339888e-01
5.59274793e-01 3.34820777e-01 -7.64922738e-01 9.16299149e-02] | [5.844359874725342, 6.747485160827637] |
f00a4df0-88ae-40c8-92c6-73a6cf9342e3 | cardea-an-open-automated-machine-learning | 2010.00509 | null | https://arxiv.org/abs/2010.00509v1 | https://arxiv.org/pdf/2010.00509v1.pdf | Cardea: An Open Automated Machine Learning Framework for Electronic Health Records | An estimated 180 papers focusing on deep learning and EHR were published between 2010 and 2018. Despite the common workflow structure appearing in these publications, no trusted and verified software framework exists, forcing researchers to arduously repeat previous work. In this paper, we propose Cardea, an extensible open-source automated machine learning framework encapsulating common prediction problems in the health domain and allows users to build predictive models with their own data. This system relies on two components: Fast Healthcare Interoperability Resources (FHIR) -- a standardized data structure for electronic health systems -- and several AUTOML frameworks for automated feature engineering, model selection, and tuning. We augment these components with an adaptive data assembler and comprehensive data- and model- auditing capabilities. We demonstrate our framework via 5 prediction tasks on MIMIC-III and Kaggle datasets, which highlight Cardea's human competitiveness, flexibility in problem definition, extensive feature generation capability, adaptable automatic data assembler, and its usability. | ['Mansour Alsaleh', 'Sarah Alnegheimish', 'Kalyan Veeramachaneni', 'Dongyu Liu', 'Shahad Althobaiti', 'Faisal Aleissa', 'Najat Alrashed'] | 2020-10-01 | null | null | null | null | ['automated-feature-engineering'] | ['methodology'] | [-2.28117406e-01 -2.48950180e-02 7.74721131e-02 -6.64576411e-01
-9.79299486e-01 -6.16097033e-01 -2.25161359e-01 5.74000180e-01
-2.67864943e-01 6.05576277e-01 9.65962783e-02 -4.22422409e-01
-3.35287750e-01 -5.99862576e-01 -5.44525325e-01 -3.77820194e-01
-2.18760908e-01 7.75225461e-01 -2.65573323e-01 -1.08871400e-01
-2.17285573e-01 3.04395914e-01 -1.36567903e+00 6.78839445e-01
8.24569225e-01 8.41335773e-01 2.65038788e-01 7.16080070e-01
8.00288022e-02 8.22816014e-01 -3.31590295e-01 -3.56755286e-01
3.48225236e-01 -2.41565034e-01 -8.82389128e-01 -6.10564232e-01
-4.86737676e-02 -2.65370429e-01 2.04039574e-01 3.81676406e-01
9.89203334e-01 -7.24281371e-01 5.35693206e-02 -1.16826081e+00
-5.41598916e-01 6.37558758e-01 2.52021819e-01 -2.37100601e-01
5.67896247e-01 5.18975198e-01 9.70212340e-01 -5.93877077e-01
8.44381571e-01 5.25423765e-01 1.38796878e+00 5.87057590e-01
-1.08690631e+00 -5.28669000e-01 -5.81357121e-01 1.31427214e-01
-1.39524114e+00 -4.25690264e-01 2.06034675e-01 -9.04830754e-01
1.56291902e+00 6.51323557e-01 9.73905325e-01 1.06176329e+00
7.45575726e-01 4.79880184e-01 6.81250453e-01 -2.16261715e-01
2.73078114e-01 1.33474112e-01 4.19849902e-01 1.06415701e+00
3.98666978e-01 -2.39909470e-01 -3.89029235e-01 -7.20601320e-01
1.62363008e-01 2.69426644e-01 -6.73914701e-02 -2.91396499e-01
-1.19132447e+00 4.65415150e-01 1.07659474e-01 3.38159315e-02
-6.62237525e-01 -2.57793605e-01 7.88218975e-01 4.12037134e-01
7.46703222e-02 8.23246539e-01 -1.14478362e+00 -3.35219890e-01
-4.88892525e-01 2.56544083e-01 1.16574144e+00 1.06895304e+00
4.91108716e-01 -2.91271120e-01 -1.59424737e-01 6.37397885e-01
6.22917861e-02 8.43440518e-02 6.73224926e-01 -4.98972207e-01
-4.79976088e-02 1.02960682e+00 3.02504022e-02 -6.04668319e-01
-1.12751126e+00 -4.55725789e-01 -8.61240685e-01 -2.64092803e-01
-1.57295287e-01 -2.60286421e-01 -5.21169186e-01 1.30313349e+00
5.72416365e-01 -2.51316458e-01 2.61299640e-01 7.15645134e-01
1.21596575e+00 3.21880393e-02 1.15753099e-01 -7.40207778e-03
1.53334987e+00 -6.70952916e-01 -6.72825933e-01 4.29435223e-01
1.11622691e+00 -6.05032563e-01 1.03727055e+00 6.94385707e-01
-8.82160187e-01 -4.74723279e-01 -1.04523730e+00 -4.58957076e-01
-5.00429869e-01 1.54272377e-01 7.70502627e-01 7.26432562e-01
-9.72158492e-01 6.58982992e-01 -1.10768294e+00 -5.89291632e-01
4.50170487e-01 5.13569653e-01 -5.64769387e-01 2.86449511e-02
-8.32076013e-01 8.72806609e-01 5.15186787e-01 -3.14375848e-01
-6.48376942e-01 -1.14058161e+00 -7.31638670e-01 -3.97161953e-02
2.48385314e-02 -1.42079711e+00 1.20343006e+00 -3.80429775e-01
-1.15838802e+00 7.83529818e-01 2.77296573e-01 -5.91117144e-01
3.86351019e-01 -3.84568632e-01 -5.85763335e-01 -2.06850469e-01
-1.13611281e-01 2.49373898e-01 1.56272396e-01 -4.50854838e-01
-5.51253378e-01 -4.87457365e-01 -4.61866468e-01 -1.79852679e-01
-9.14923176e-02 1.44009516e-01 -2.74086565e-01 -4.25219655e-01
-2.66161233e-01 -7.21121848e-01 -5.19842088e-01 -3.18420082e-01
-6.24635398e-01 -3.35298176e-03 3.13386321e-01 -1.13120985e+00
1.49814463e+00 -1.90440583e+00 -2.21438415e-04 6.66868687e-02
5.21224618e-01 1.59485877e-01 -1.16195507e-01 6.62289262e-01
-1.55702636e-01 9.04717892e-02 -1.72966998e-02 1.13548085e-01
8.20446759e-02 2.17887182e-02 1.82456017e-01 3.59925658e-01
2.26291537e-01 1.16881716e+00 -6.45563185e-01 -2.08223343e-01
1.92327514e-01 5.44662058e-01 -9.54461694e-01 4.47711796e-01
-2.38481984e-01 4.24360186e-01 -4.35720980e-01 8.52562249e-01
5.68517268e-01 -6.76779330e-01 6.64949119e-01 -4.74005699e-01
-1.95916235e-01 5.28732121e-01 -9.14788723e-01 1.96679521e+00
-1.78363934e-01 -1.48022771e-01 1.79294460e-02 -4.32644129e-01
9.05478299e-01 3.40023875e-01 1.02101779e+00 -2.78523564e-01
6.32300302e-02 1.69787630e-01 9.14581399e-03 -1.10632086e+00
1.44674495e-01 3.46451670e-01 -1.43224433e-01 4.07501131e-01
2.13800415e-01 3.38042796e-01 3.02999709e-02 -9.21594054e-02
1.67133546e+00 3.54860306e-01 8.37337017e-01 -3.12055439e-01
1.77709326e-01 7.85147101e-02 8.50389719e-01 5.33974826e-01
5.91018312e-02 5.20673931e-01 2.58951694e-01 -1.16285312e+00
-1.17829204e+00 -6.74349129e-01 -3.50169688e-01 9.90995944e-01
-7.19692171e-01 -1.19130898e+00 -7.61015773e-01 -6.27467096e-01
1.75078884e-01 4.79891002e-01 -5.70405245e-01 8.00081044e-02
-2.76291698e-01 -8.98208022e-01 8.67205203e-01 3.85962844e-01
6.74856380e-02 -9.13191676e-01 -9.56355333e-01 4.95675653e-01
-1.06273368e-01 -7.44046152e-01 -2.30715215e-01 4.35767829e-01
-3.78173470e-01 -1.48935807e+00 7.69117698e-02 -7.55864918e-01
2.17422903e-01 -4.67913866e-01 1.34040761e+00 -6.88274726e-02
-1.13269508e+00 2.15256274e-01 -3.19058746e-01 -6.41567767e-01
-7.60621548e-01 3.65550160e-01 2.62183957e-02 -5.58506191e-01
7.18663335e-01 -1.86833560e-01 -5.92173338e-01 -6.60769059e-04
-7.73614943e-01 4.31942284e-01 6.68468416e-01 9.34775829e-01
8.42680693e-01 -5.43873847e-01 9.01545584e-01 -9.80352044e-01
4.90061462e-01 -6.86374903e-01 -6.22635067e-01 7.04991460e-01
-1.10952413e+00 8.26618373e-02 7.21454620e-01 3.53233039e-01
-4.37699974e-01 5.03229737e-01 -8.15392613e-01 7.22379386e-02
-3.08064491e-01 7.38092601e-01 -3.47047716e-01 2.19685331e-01
8.37581098e-01 1.22082204e-01 5.47700286e-01 -8.53304923e-01
1.20536610e-01 9.51470971e-01 1.50095344e-01 -2.33045980e-01
1.25098526e-01 -6.00421540e-02 -2.91226834e-01 -4.36241299e-01
-4.48224157e-01 -3.83837402e-01 -5.74069023e-01 4.81546015e-01
9.82744753e-01 -1.07672691e+00 -1.13317728e+00 4.49015379e-01
-9.07684147e-01 -7.11675510e-02 -2.30764091e-01 3.73429954e-01
-4.08521056e-01 2.99373746e-01 -8.21004868e-01 -2.98596203e-01
-1.28908145e+00 -1.39974034e+00 8.66445363e-01 -3.54095548e-01
-5.88453293e-01 -6.49804235e-01 5.30847907e-01 4.13567960e-01
5.90580344e-01 3.96304041e-01 1.01163721e+00 -1.15564203e+00
-2.33715892e-01 -2.74913937e-01 -1.49343833e-02 2.86737889e-01
3.50830466e-01 3.89352590e-02 -9.65245485e-01 -4.32779312e-01
-2.10985005e-01 -3.93213660e-01 3.63787413e-01 1.63298205e-01
1.35657275e+00 -6.06434584e-01 -4.94309306e-01 1.00533235e+00
1.31770122e+00 5.75955026e-02 4.19531912e-01 3.39337617e-01
6.10153973e-01 2.47666687e-01 4.58160251e-01 8.51047337e-01
6.02317572e-01 5.87255716e-01 2.59620160e-01 -3.12442511e-01
3.40059400e-01 1.29097760e-01 -5.88523783e-02 1.01863277e+00
1.66289490e-02 1.80426925e-01 -1.18878555e+00 6.70081899e-02
-1.78129971e+00 -6.67681754e-01 -1.48185268e-01 2.00350285e+00
1.07958281e+00 -5.69351912e-01 2.48027414e-01 -2.97331989e-01
1.30513534e-01 -6.62971973e-01 -6.93488061e-01 -2.46329501e-01
-4.82893456e-03 3.73759925e-01 2.52970129e-01 -2.50776596e-02
-1.09248888e+00 4.92692500e-01 6.90483904e+00 2.97604889e-01
-9.48656440e-01 2.97750860e-01 6.30897820e-01 -3.70532423e-01
-1.83931351e-01 -4.62067038e-01 -6.56498432e-01 3.36479962e-01
1.27001953e+00 -1.31900191e-01 2.81707615e-01 1.22675323e+00
2.15902284e-01 6.77050710e-01 -1.26321936e+00 9.96169329e-01
-2.52862155e-01 -1.95875919e+00 -9.42342356e-02 -4.18259911e-02
2.97807511e-02 6.18901432e-01 -2.76653826e-01 3.87781978e-01
4.52325135e-01 -1.22360444e+00 3.18162739e-01 6.86344326e-01
9.90387440e-01 -6.09189868e-01 9.73940611e-01 -6.03626333e-02
-8.20706189e-01 -2.07341120e-01 -1.00457773e-01 2.17690200e-01
-1.54740050e-01 6.76496744e-01 -1.41775823e+00 1.05624568e+00
1.06408429e+00 6.98847592e-01 -7.30842113e-01 1.04963565e+00
4.52255100e-01 4.80041653e-01 -5.37380315e-02 2.00741306e-01
-4.07166362e-01 -2.96531226e-02 4.28018384e-02 1.67582941e+00
2.89801240e-01 -9.38257575e-02 4.88556892e-01 5.86302102e-01
1.43037200e-01 4.77400362e-01 -3.24547529e-01 -2.12996006e-02
4.88841325e-01 1.59743989e+00 -1.08534604e-01 -1.87275112e-01
-4.81971651e-01 8.00749898e-01 2.55745411e-01 -1.97304308e-01
-9.24133778e-01 -3.99119258e-01 9.77705896e-01 7.66641423e-02
6.78641796e-02 2.57523000e-01 -2.43173316e-01 -1.06106389e+00
-3.42699289e-02 -1.60616064e+00 7.71876991e-01 -4.12985772e-01
-1.48243070e+00 8.14072490e-01 -5.47669649e-01 -1.10589981e+00
-3.57968748e-01 -6.14800096e-01 1.70698509e-01 7.26641417e-01
-9.45763350e-01 -1.48634708e+00 -3.95877153e-01 3.67949873e-01
1.78324610e-01 -4.73922580e-01 1.46165919e+00 5.28639197e-01
-8.62071753e-01 7.64590025e-01 2.21228138e-01 1.83872785e-02
7.44691551e-01 -1.07149994e+00 6.15453660e-01 1.99445263e-01
-4.86534119e-01 1.02386117e+00 4.68787611e-01 -7.06052244e-01
-2.16348243e+00 -1.45378709e+00 6.72197402e-01 -6.98613524e-01
4.76751179e-01 -5.34377873e-01 -7.99351096e-01 8.86090875e-01
5.56811616e-02 -2.24126205e-01 1.64167511e+00 2.40541950e-01
-4.01381999e-01 -2.94125855e-01 -1.33404970e+00 2.13730242e-02
8.00990462e-01 -2.33507916e-01 -3.23246032e-01 5.43533027e-01
7.74397254e-01 -3.97169352e-01 -1.53526592e+00 6.17778599e-01
1.03572559e+00 -8.80635977e-01 8.00795436e-01 -1.09073532e+00
3.64467762e-02 -2.90561914e-01 -1.75935626e-01 -9.36151266e-01
-6.04422331e-01 -8.62980664e-01 -1.05001755e-01 8.13410878e-01
5.98663211e-01 -7.11330712e-01 4.28717434e-01 7.37718225e-01
-7.01126277e-01 -1.00978839e+00 -8.39449525e-01 -4.06991512e-01
-3.94355386e-01 -4.35138404e-01 1.21784532e+00 1.32101333e+00
3.38452756e-01 3.50618601e-01 -2.78142959e-01 1.42312214e-01
4.26268280e-02 1.45948194e-02 1.01607943e+00 -1.27700353e+00
-8.90248060e-01 -1.11823298e-01 -5.38052142e-01 -1.05029434e-01
-5.56852818e-01 -1.22787130e+00 -2.95050740e-01 -1.60679936e+00
5.67786932e-01 -6.76716447e-01 -5.24793684e-01 1.21940899e+00
-1.48429140e-01 -1.89376175e-01 -3.20284963e-01 2.25048363e-01
-8.24373186e-01 3.50355986e-03 4.93614227e-01 1.16078354e-01
-3.28233629e-01 -3.43848735e-01 -7.92074740e-01 3.66699308e-01
7.60980844e-01 -3.68104398e-01 -8.48193914e-02 -2.98743099e-01
4.95933831e-01 -3.08813393e-01 3.05749387e-01 -1.06802607e+00
-1.50778919e-01 1.14959002e-01 5.43584168e-01 -6.26389146e-01
-7.85903484e-02 -7.67786920e-01 1.14892018e+00 9.39726353e-01
-2.57568747e-01 4.26492244e-01 3.05966169e-01 1.49876162e-01
3.26100647e-01 1.26529172e-01 3.55406523e-01 -1.64324105e-01
-3.86585504e-01 3.38509113e-01 -2.91717291e-01 -4.21519727e-01
1.12493145e+00 1.72616348e-01 -5.33630610e-01 3.82698745e-01
-7.34582663e-01 1.05583780e-01 7.43355930e-01 5.03473997e-01
5.32240987e-01 -8.04673254e-01 -8.98542941e-01 7.22572088e-01
6.06201887e-01 -2.05353856e-01 3.89793843e-01 9.17223752e-01
-9.15663242e-01 4.97417092e-01 -4.07418370e-01 -5.61648667e-01
-1.27491283e+00 8.93745184e-01 3.73425096e-01 -4.41241771e-01
-7.42548168e-01 4.30275202e-01 -2.82639772e-01 -9.92004752e-01
1.19333537e-02 -4.93452370e-01 9.01791826e-02 -1.44567087e-01
7.61546910e-01 2.79134750e-01 8.70468676e-01 -1.20635973e-02
-6.25832915e-01 -1.85066879e-01 -2.04236522e-01 5.91073394e-01
1.72935641e+00 2.34041408e-01 -4.73212779e-01 4.14734930e-01
9.52334464e-01 -2.84012139e-01 -6.29200041e-01 3.27132732e-01
1.18429944e-01 2.06814006e-01 -3.47696483e-01 -1.33533025e+00
-6.04738712e-01 2.98266709e-01 1.06624281e+00 9.13951024e-02
1.07898641e+00 -8.61925632e-02 7.78986752e-01 6.28486872e-01
4.55206692e-01 -8.08894813e-01 -4.87837076e-01 3.23360801e-01
6.89062655e-01 -1.07123911e+00 -1.79373808e-02 -9.87013429e-02
-4.80457306e-01 1.03234875e+00 4.39909518e-01 5.16287148e-01
5.33049285e-01 8.01472902e-01 2.08288670e-01 -4.18285459e-01
-1.11323428e+00 2.65333176e-01 2.13752419e-01 6.10755086e-01
1.09527934e+00 2.37051442e-01 -2.68166393e-01 1.34327173e+00
-4.00551438e-01 5.48074901e-01 6.42924309e-02 1.03207088e+00
-1.55263111e-01 -1.30279207e+00 -1.35757387e-01 1.06471848e+00
-7.15472937e-01 -3.28700334e-01 -2.01592058e-01 5.53758800e-01
2.71576613e-01 7.05173492e-01 -1.80919781e-01 -6.49905086e-01
1.95467412e-01 3.91187519e-01 2.97595412e-01 -6.56347811e-01
-1.30267334e+00 -1.56709142e-02 3.50444287e-01 -8.81916225e-01
2.60827690e-01 -6.06126249e-01 -1.26631331e+00 -3.49028893e-02
-7.43583813e-02 7.22082034e-02 8.73672187e-01 6.76022172e-01
1.47083926e+00 6.62100673e-01 1.52848452e-01 -2.12785035e-01
-5.33269703e-01 -9.48892117e-01 -3.59042197e-01 2.79674232e-01
1.98177621e-01 -8.59398115e-03 3.46578091e-01 1.78424105e-01] | [7.936159133911133, 6.282454490661621] |
7087b6b9-d27f-4e85-9de9-6ece9bf1c439 | text-classification-by-contrastive-learning | null | null | https://aclanthology.org/2020.coling-main.542 | https://aclanthology.org/2020.coling-main.542.pdf | Text Classification by Contrastive Learning and Cross-lingual Data Augmentation for Alzheimer's Disease Detection | Data scarcity is always a constraint on analyzing speech transcriptions for automatic Alzheimer{'}s disease (AD) detection, especially when the subjects are non-English speakers. To deal with this issue, this paper first proposes a contrastive learning method to obtain effective representations for text classification based on monolingual embeddings of BERT. Furthermore, a cross-lingual data augmentation method is designed by building autoencoders to learn the text representations shared by both languages. Experiments on a Mandarin AD corpus show that the contrastive learning method can achieve better detection accuracy than conventional CNN-based and BERTbased methods. Our cross-lingual data augmentation method also outperforms other compared methods when using another English AD corpus for augmentation. Finally, a best detection accuracy of 81.6{\%} is obtained by our proposed methods on the Mandarin AD corpus. | ['Yunxia Li', 'Lingjing Jin', 'Shijin Wang', 'ZhenHua Ling', 'Zhaoci Liu', 'Zhiqiang Guo'] | 2020-12-01 | null | null | null | coling-2020-8 | ['alzheimer-s-disease-detection'] | ['medical'] | [-4.93080020e-02 4.96177636e-02 -1.81836888e-01 -4.28037852e-01
-1.14221728e+00 2.53984928e-01 5.56428075e-01 -4.46499102e-02
-8.20050895e-01 7.21209943e-01 4.30088371e-01 -3.21597695e-01
4.01091993e-01 -5.36408424e-01 -2.86709875e-01 -4.02685642e-01
7.41471946e-02 4.57153052e-01 -3.49968523e-02 -8.23953375e-02
-4.55896735e-01 1.65095255e-01 -1.03712583e+00 4.67913628e-01
8.87244701e-01 9.19717908e-01 4.33124244e-01 4.80088085e-01
-4.22129393e-01 4.68955964e-01 -8.50759208e-01 -4.28608894e-01
5.09326905e-02 -2.61643529e-01 -6.99893057e-01 2.65871942e-01
4.17860419e-01 -5.18919051e-01 -5.89174569e-01 9.95524824e-01
8.31779778e-01 -3.47754061e-01 5.86930156e-01 -9.45925772e-01
-8.65629971e-01 7.90474415e-01 -3.90895545e-01 6.04072034e-01
-1.05787933e-01 -1.25463143e-01 9.03608203e-01 -9.02476072e-01
3.18913877e-01 1.17834175e+00 7.34574974e-01 8.46170008e-01
-9.55456674e-01 -7.97039568e-01 3.02700073e-01 4.26540434e-01
-1.50492668e+00 -3.86198878e-01 6.86032951e-01 -3.49510759e-01
9.03166831e-01 -1.40491817e-02 7.23647356e-01 1.51587379e+00
-8.39156751e-03 1.02905738e+00 9.69477952e-01 -6.59744859e-01
-4.01234515e-02 2.97963470e-01 2.90436327e-01 7.21094728e-01
8.48377943e-02 -2.95295894e-01 -2.73541093e-01 -3.55549067e-01
6.45248532e-01 -1.29350141e-01 -1.97517991e-01 3.11241090e-01
-1.26614082e+00 8.78178656e-01 2.54045963e-01 6.72858417e-01
-2.79450148e-01 -9.88425910e-02 7.10487366e-01 1.98525161e-01
7.65759408e-01 3.85276624e-03 -6.41998708e-01 9.67305079e-02
-7.31400490e-01 -7.61527335e-03 4.10899371e-01 1.00449800e+00
7.86869898e-02 3.72310758e-01 -5.05036302e-03 1.49596012e+00
5.20759702e-01 5.90431690e-01 1.10554850e+00 -2.45473221e-01
6.41079903e-01 6.44465506e-01 -3.60923737e-01 -4.63492364e-01
-4.60794270e-01 -6.18957579e-01 -9.79098380e-01 -2.02025473e-01
1.59635887e-01 -3.51482004e-01 -9.23833549e-01 1.59667075e+00
5.71702756e-02 -1.57660395e-01 2.38993078e-01 5.41250587e-01
8.17574501e-01 5.38362563e-01 3.38656068e-01 -3.90180618e-01
1.77012873e+00 -1.06692648e+00 -1.18938458e+00 -4.12419677e-01
9.19077396e-01 -7.94544399e-01 1.14950120e+00 3.04795265e-01
-5.41296482e-01 -4.64564890e-01 -1.05602992e+00 -1.50494576e-01
-4.36403990e-01 8.42286229e-01 4.01251376e-01 8.16665232e-01
-8.12762976e-01 -1.71230584e-01 -8.62328589e-01 -5.26291966e-01
6.24956787e-01 3.66762310e-01 -4.87147182e-01 -1.89060979e-02
-1.28966343e+00 6.70725346e-01 3.64240944e-01 1.54459160e-02
-6.72091186e-01 -2.86871672e-01 -6.63594127e-01 -2.31418282e-01
-1.73347533e-01 -4.42391634e-01 1.23470485e+00 -1.10978687e+00
-1.19507563e+00 9.78233755e-01 3.18670012e-02 -6.35294497e-01
4.21369404e-01 -3.94868463e-01 -1.03242540e+00 1.53126538e-01
2.13024989e-01 6.60050213e-01 7.09636033e-01 -6.33725464e-01
-6.48403347e-01 -5.88554263e-01 -2.65047073e-01 4.31813598e-02
-1.13414168e+00 1.82413071e-01 -2.51518548e-01 -1.06994820e+00
2.91075408e-02 -6.47961080e-01 -1.15011141e-01 2.43149519e-01
-3.63711804e-01 -5.89558244e-01 9.43844080e-01 -1.00190294e+00
1.29082990e+00 -2.12118912e+00 -3.34470928e-01 -2.60953039e-01
3.01984429e-01 4.18924153e-01 -1.24253303e-01 7.54904300e-02
-2.05845833e-01 1.30801305e-01 -2.33050168e-01 -5.39463460e-01
-2.77721316e-01 2.58896172e-01 2.70402670e-01 3.89576524e-01
3.28855634e-01 4.85031605e-01 -4.09164876e-01 -7.54984021e-01
-9.26643759e-02 5.81623077e-01 -5.80811620e-01 1.56970620e-01
-1.08753070e-01 1.20018303e-01 -5.39089918e-01 6.85372055e-01
4.60709602e-01 -1.22929111e-01 2.77390152e-01 -2.27499619e-01
1.06002338e-01 3.79568011e-01 -7.56650686e-01 1.46508133e+00
-4.18408036e-01 6.20425522e-01 4.81345579e-02 -1.20810437e+00
1.04694688e+00 6.35135412e-01 4.21669573e-01 -5.73327482e-01
3.55531752e-01 3.28969449e-01 8.72270316e-02 -7.64196575e-01
-7.25571066e-02 -9.40432400e-02 2.37014011e-01 2.26940483e-01
1.65313438e-01 6.71687603e-01 3.08462884e-02 -6.50983229e-02
1.07023752e+00 -2.82831401e-01 3.84683609e-01 -3.92163873e-01
7.53652751e-01 -9.15738754e-03 7.74616241e-01 5.09634376e-01
-5.10303259e-01 4.77547139e-01 2.34898075e-01 -5.58484197e-01
-1.21146047e+00 -6.13503635e-01 -6.65779531e-01 8.16012979e-01
-6.66032493e-01 -5.69053173e-01 -9.53043640e-01 -1.14705527e+00
-2.76349574e-01 4.41411734e-01 -5.28458953e-01 -6.46112114e-02
-6.04181767e-01 -1.31319249e+00 8.81770551e-01 8.13660026e-01
8.33902478e-01 -7.17422426e-01 2.31280372e-01 3.15229863e-01
-4.94761407e-01 -1.37018394e+00 -6.38712823e-01 1.74256548e-01
-9.02387559e-01 -1.01542723e+00 -9.46342170e-01 -1.52232099e+00
6.82880580e-01 -2.62776460e-03 6.00276589e-01 -7.77413771e-02
-4.93797660e-01 2.39447206e-01 -4.97435153e-01 -6.52577281e-01
-5.31715333e-01 2.87223935e-01 2.48874709e-01 6.06049895e-02
9.63860393e-01 -4.48688537e-01 -4.71935958e-01 1.71238199e-01
-5.14422297e-01 -1.12035871e-01 8.34484160e-01 9.72122133e-01
5.56546450e-01 -8.40168893e-02 1.09488332e+00 -5.01857758e-01
7.33543634e-01 -3.65015000e-01 -1.54165432e-01 2.74436232e-02
-6.48637950e-01 -5.52504361e-02 5.35106421e-01 -7.61492074e-01
-9.27354753e-01 2.09403753e-01 -6.76315904e-01 -8.16872790e-02
-3.57309937e-01 2.65575379e-01 -2.83963591e-01 4.29226846e-01
7.95463860e-01 4.03644532e-01 2.48108715e-01 -8.13019514e-01
3.42790969e-02 1.47083211e+00 2.44141653e-01 -1.46540716e-01
3.44995886e-01 2.61350721e-01 -7.29955018e-01 -1.22093070e+00
-6.37298822e-01 -5.12947559e-01 -7.19594955e-01 8.47007707e-02
1.32231951e+00 -1.21863186e+00 -2.68711478e-01 6.87994421e-01
-1.25493860e+00 -3.61073278e-02 4.87802066e-02 1.06999016e+00
-2.55874872e-01 3.65689844e-01 -9.34048831e-01 -7.24708080e-01
-6.43776715e-01 -1.22140801e+00 1.08320987e+00 -3.75409752e-01
-9.55259502e-02 -9.50346649e-01 3.00599374e-02 6.99734688e-01
8.62486511e-02 -4.42947477e-01 1.12190974e+00 -1.20273149e+00
3.82375438e-03 -4.16817904e-01 -3.84252191e-01 8.75312328e-01
4.05963570e-01 -5.15095413e-01 -1.02469194e+00 -1.04046069e-01
1.44726977e-01 -1.35478243e-01 7.38311708e-01 1.57486051e-01
1.08224702e+00 -5.71812510e-01 -3.13437432e-01 8.65077302e-02
1.03755975e+00 3.35007191e-01 5.88724673e-01 5.28836489e-01
7.19383299e-01 2.94006795e-01 1.91134855e-01 2.88760662e-01
4.75952327e-01 6.25149190e-01 1.32852003e-01 -2.22378507e-01
-2.36809582e-01 1.11200057e-01 5.05053401e-01 1.31544971e+00
1.83941573e-01 -1.70743555e-01 -9.36119199e-01 5.26114345e-01
-1.58406973e+00 -5.74325562e-01 -3.05451989e-01 1.88629401e+00
1.03308249e+00 5.37771359e-02 3.08674306e-01 2.30987817e-01
9.20660913e-01 -7.89896771e-02 -1.92969456e-01 -6.72317743e-02
-1.88227400e-01 1.65586099e-01 2.93788314e-01 8.67443606e-02
-1.49389946e+00 7.82091975e-01 6.79264307e+00 8.06671679e-01
-1.01821315e+00 6.12004578e-01 5.82186580e-01 1.32703185e-02
1.71705589e-01 -7.84490585e-01 -9.92978990e-01 4.62814331e-01
1.10082102e+00 2.03254119e-01 -1.68785542e-01 1.12676573e+00
2.40149200e-01 5.19200981e-01 -7.99260020e-01 1.17727995e+00
3.86911362e-01 -1.06779397e+00 1.65135026e-01 3.00833702e-01
2.64175892e-01 1.18192330e-01 -1.85977638e-01 4.05044019e-01
-2.43611503e-02 -6.82396829e-01 4.04173732e-01 2.67515630e-01
6.34538591e-01 -7.48375416e-01 1.06975067e+00 1.58515155e-01
-1.08531702e+00 -1.60560474e-01 -4.14164126e-01 3.34656656e-01
2.56963409e-02 5.39500356e-01 -1.15128648e+00 2.12722719e-01
7.64155507e-01 5.50141454e-01 -6.28012657e-01 8.09671342e-01
-3.75114344e-02 7.27277637e-01 -2.54727334e-01 -2.65025616e-01
4.76418026e-02 7.43501708e-02 4.43724304e-01 1.45566809e+00
2.24653304e-01 -2.03538954e-01 2.27919683e-01 6.93403065e-01
-3.56886655e-01 7.33677089e-01 -5.71580112e-01 -3.03173602e-01
1.75663114e-01 1.11996722e+00 -2.91173100e-01 -4.77098525e-01
-8.16594779e-01 1.05148232e+00 2.16470197e-01 1.86634153e-01
-6.61048293e-01 -3.02698761e-01 5.15233278e-01 1.55806631e-01
8.92172456e-02 -4.22281623e-01 8.18536151e-03 -1.07394874e+00
1.81578144e-01 -1.11894011e+00 3.98507357e-01 -6.00862205e-01
-1.50508571e+00 9.69196320e-01 -4.14598078e-01 -1.36423743e+00
-4.32172567e-02 -9.71347570e-01 -2.91677028e-01 5.76959729e-01
-1.24157679e+00 -1.42343724e+00 -5.28898239e-02 7.18399048e-01
8.96832585e-01 -8.46304178e-01 1.16522467e+00 1.04912651e+00
-9.73804951e-01 7.81236708e-01 1.11530840e-01 5.94301641e-01
8.13528478e-01 -1.16452897e+00 8.65269005e-02 6.50062978e-01
1.23810537e-01 4.88850683e-01 1.01329572e-01 -6.79475904e-01
-9.02553201e-01 -1.37694871e+00 9.96079445e-01 -8.11313167e-02
7.87916064e-01 -4.70673621e-01 -8.22934330e-01 8.86284113e-01
3.16412598e-01 -7.51731172e-02 9.98718798e-01 1.58053756e-01
-2.42766961e-01 -3.34068805e-01 -1.04517758e+00 6.43575191e-01
8.86566103e-01 -7.22691000e-01 -7.08265722e-01 8.46574008e-01
8.44906628e-01 2.06915289e-01 -1.02869236e+00 2.08363026e-01
4.51316267e-01 -2.60260195e-01 8.47893417e-01 -8.40918183e-01
3.13926518e-01 -7.44988173e-02 -3.25054049e-01 -9.85662520e-01
-7.75945336e-02 -8.10343176e-02 2.57047325e-01 1.30792999e+00
4.06628430e-01 -7.01507568e-01 5.47838330e-01 -4.48749624e-02
-2.82313555e-01 -5.74764252e-01 -1.05381155e+00 -8.25834394e-01
1.80347070e-01 -6.01845622e-01 2.16322362e-01 1.12505674e+00
-3.61904614e-02 4.88948971e-01 -3.27952415e-01 2.12610841e-01
4.42662925e-01 -7.41593897e-01 3.32875699e-01 -1.39383078e+00
3.52977477e-02 -1.77013204e-01 -7.28314579e-01 -8.36386323e-01
3.03101659e-01 -1.19653392e+00 -6.22126050e-02 -1.42598712e+00
2.23128065e-01 -3.32252711e-01 -2.90823817e-01 6.47351801e-01
-7.23113418e-02 2.10213110e-01 -3.21457803e-01 2.85863310e-01
-1.65375486e-01 7.30275393e-01 1.02584338e+00 -4.73805666e-01
-1.29040137e-01 -4.40244824e-02 -5.77074170e-01 8.87198031e-01
1.11482275e+00 -4.06287581e-01 -2.99494863e-01 -4.66182292e-01
-4.12177294e-01 -3.29954237e-01 2.36470968e-01 -1.05916202e+00
-5.10121398e-02 3.54088306e-01 5.53874791e-01 -4.92421210e-01
2.65195221e-01 -7.20907271e-01 -5.03480732e-01 4.26271051e-01
-4.53655571e-01 5.10034934e-02 2.04739243e-01 6.12924159e-01
-2.44370684e-01 -2.33440161e-01 8.70682836e-01 -1.93418667e-01
-2.94815630e-01 3.09641242e-01 -9.50280905e-01 7.91119132e-03
6.21230185e-01 1.93442896e-01 -3.03834468e-01 -9.14056376e-02
-1.11587596e+00 -2.84429565e-02 -3.90983284e-01 5.60232759e-01
5.58676541e-01 -1.67207825e+00 -8.18452358e-01 2.87372112e-01
3.61150384e-01 -4.34563965e-01 -3.57234967e-03 9.47654843e-01
-4.48864937e-01 5.09151280e-01 -1.17546126e-01 -4.50156182e-01
-1.40742111e+00 3.89042586e-01 6.15841985e-01 -1.80340767e-01
-8.42217147e-01 6.50791645e-01 2.22538769e-01 -4.66861069e-01
4.97435361e-01 -3.67749035e-01 -4.23866540e-01 2.07139179e-01
7.73318887e-01 1.80082172e-01 3.86378825e-01 -5.27142167e-01
-3.24291259e-01 2.32410012e-03 -5.07822335e-01 -1.50320798e-01
1.32318103e+00 -1.26056876e-02 -3.66211310e-02 4.92101073e-01
1.41914070e+00 -4.82870229e-02 -4.46085334e-01 -3.91205192e-01
4.90448773e-02 1.33592591e-01 3.48825097e-01 -6.00040555e-01
-1.17873645e+00 8.73431385e-01 1.42943656e+00 7.83925951e-02
8.06869447e-01 7.04994798e-02 9.14373219e-01 6.72750533e-01
-1.39027117e-02 -1.31894290e+00 1.56847104e-01 1.35785118e-01
1.14578509e+00 -1.28484321e+00 -3.84992272e-01 -3.31022918e-01
-3.95936757e-01 1.33889854e+00 7.49374151e-01 4.31798249e-02
8.10465395e-01 2.22948864e-01 2.69142777e-01 2.02856003e-03
-4.44489419e-01 -3.59168023e-01 2.18452364e-01 7.49121487e-01
7.05338836e-01 3.84304747e-02 -6.25844181e-01 8.95524502e-01
-2.21172616e-01 -1.14790924e-01 3.45012367e-01 8.78329515e-01
-4.26179945e-01 -1.43347788e+00 -3.77150029e-01 7.12194681e-01
-8.53221714e-01 -3.50273341e-01 -4.08010960e-01 9.28026855e-01
2.80491084e-01 8.39975834e-01 2.43511558e-01 -3.07647169e-01
3.45973998e-01 5.71208417e-01 1.76758021e-01 -7.08694220e-01
-2.92507499e-01 6.04288757e-01 4.17710513e-01 -4.57970016e-02
-4.93874162e-01 -5.89778066e-01 -1.22865653e+00 2.17376843e-01
-3.64463985e-01 -9.36532617e-02 6.19368851e-01 1.02905047e+00
1.69677407e-01 8.45363855e-01 3.31372857e-01 -4.47851047e-02
-4.04176772e-01 -1.32324755e+00 -6.79194033e-01 9.52419564e-02
3.30882460e-01 -5.36537468e-01 -3.24291199e-01 2.52423257e-01] | [13.929767608642578, 5.382170677185059] |
0af2e23a-fa2c-4adb-a2a4-6b1cd3543592 | ice-gan-identity-aware-and-capsule-enhanced | 2005.04370 | null | https://arxiv.org/abs/2005.04370v2 | https://arxiv.org/pdf/2005.04370v2.pdf | ICE-GAN: Identity-aware and Capsule-Enhanced GAN with Graph-based Reasoning for Micro-Expression Recognition and Synthesis | Micro-expressions are reflections of people's true feelings and motives, which attract an increasing number of researchers into the study of automatic facial micro-expression recognition. The short detection window, the subtle facial muscle movements, and the limited training samples make micro-expression recognition challenging. To this end, we propose a novel Identity-aware and Capsule-Enhanced Generative Adversarial Network with graph-based reasoning (ICE-GAN), introducing micro-expression synthesis as an auxiliary task to assist recognition. The generator produces synthetic faces with controllable micro-expressions and identity-aware features, whose long-ranged dependencies are captured through the graph reasoning module (GRM), and the discriminator detects the image authenticity and expression classes. Our ICE-GAN was evaluated on Micro-Expression Grand Challenge 2019 (MEGC2019) with a significant improvement (12.9%) over the winner and surpassed other state-of-the-art methods. | ['Yang song', 'Weidong Cai', 'Jianhui Yu', 'Chaoyi Zhang'] | 2020-05-09 | null | null | null | null | ['micro-expression-recognition'] | ['computer-vision'] | [ 4.10942346e-01 3.87471557e-01 6.48323894e-02 -7.34283268e-01
-4.69275057e-01 -4.08052474e-01 6.38080359e-01 -9.39943671e-01
1.24338880e-01 6.73033714e-01 4.53371219e-02 2.44278878e-01
6.26397014e-01 -6.80061519e-01 -6.46446347e-01 -9.85665202e-01
3.25526716e-03 -4.26954515e-02 -8.40442419e-01 -5.25799870e-01
-4.46389049e-01 4.89660978e-01 -1.35271263e+00 5.45326054e-01
7.97649741e-01 1.39061105e+00 -6.88943207e-01 5.38610756e-01
1.18876778e-01 1.32355809e+00 -7.69625306e-01 -1.29179108e+00
8.86924714e-02 -1.06614363e+00 -1.81306154e-01 1.45300388e-01
5.16322851e-01 -2.30258718e-01 -3.80062193e-01 1.33925235e+00
6.49331629e-01 -2.64715075e-01 4.92540359e-01 -1.80156112e+00
-1.04748285e+00 3.56735319e-01 -7.57134140e-01 -4.77375001e-01
5.04089534e-01 2.93374896e-01 6.16731942e-01 -8.85478675e-01
8.25408697e-01 1.36503243e+00 7.39722610e-01 1.25520933e+00
-1.22502530e+00 -1.22877157e+00 -1.03648826e-01 1.48828462e-01
-1.69755960e+00 -6.88772857e-01 1.11773813e+00 -2.79795676e-01
5.69862783e-01 3.65626097e-01 7.53714025e-01 1.80399227e+00
2.29371320e-02 7.02967644e-01 1.37928462e+00 -2.51136243e-01
1.43492460e-01 1.65255517e-01 -7.36653388e-01 9.46060121e-01
-3.51928711e-01 1.67135686e-01 -4.23746347e-01 -9.60015804e-02
7.04409599e-01 -1.52935639e-01 -2.76012123e-01 1.08517453e-01
-5.68867683e-01 7.97598541e-01 4.30913389e-01 1.79520369e-01
-6.15909636e-01 2.50670791e-01 3.94694775e-01 5.36840320e-01
6.84629381e-01 3.07577461e-01 1.11882284e-01 -3.18195492e-01
-8.44518006e-01 1.18818656e-01 6.33283496e-01 7.87453830e-01
6.12634003e-01 8.04738820e-01 -2.87534177e-01 8.90699029e-01
-8.34292248e-02 6.99598074e-01 4.41166461e-01 -7.29214847e-01
8.07186216e-02 7.55164206e-01 -3.43946517e-01 -1.46381330e+00
-8.67057294e-02 -4.71726418e-01 -1.28964627e+00 5.03236055e-01
1.33247659e-01 -5.45596421e-01 -7.64591098e-01 2.40645194e+00
2.18787119e-01 3.36724132e-01 2.32463062e-01 9.09920156e-01
8.14122021e-01 5.07216394e-01 1.19982764e-01 -3.67855400e-01
1.10330701e+00 -7.95462310e-01 -9.10761356e-01 -1.03050195e-01
4.89679515e-01 -4.44431573e-01 9.33814704e-01 2.05579445e-01
-8.51698697e-01 -5.58321774e-01 -1.04101312e+00 2.14093626e-01
-1.09175630e-01 3.59940916e-01 9.31367874e-01 9.21129704e-01
-1.09440708e+00 4.11121041e-01 -2.64288753e-01 2.30395615e-01
9.53802407e-01 5.10458291e-01 -8.31860065e-01 1.02986775e-01
-1.28917742e+00 5.44566512e-01 -3.53969842e-01 3.96617860e-01
-6.71045721e-01 -6.02278054e-01 -9.12993014e-01 -2.19874665e-01
1.57302871e-01 -5.13542473e-01 8.42936158e-01 -2.16989994e+00
-1.94301093e+00 1.37288880e+00 2.81450967e-03 -1.76569089e-01
8.18708479e-01 1.21352114e-01 -9.21881616e-01 1.67535812e-01
-1.79113761e-01 6.85176969e-01 1.39848745e+00 -9.80695009e-01
1.59850791e-01 -5.72539508e-01 -9.47157294e-02 -1.26376614e-01
-1.32759154e-01 1.31800458e-01 -9.49418396e-02 -9.14399862e-01
-4.91076529e-01 -1.10827482e+00 1.05792329e-01 1.82870612e-01
-4.66771603e-01 -2.14436394e-03 8.17661107e-01 -6.69815540e-01
7.58122981e-01 -2.49324608e+00 2.73985088e-01 1.77486584e-01
4.32576001e-01 3.49767894e-01 -4.85645592e-01 -2.76239542e-03
-4.04529899e-01 1.81185782e-01 -6.97909817e-02 -5.10817230e-01
1.23153374e-01 3.15091386e-02 -3.76825154e-01 5.08454204e-01
5.82948506e-01 1.42684197e+00 -6.96160495e-01 -3.03819060e-01
-1.29067719e-01 8.48636448e-01 -3.94974560e-01 3.74878556e-01
-1.12092912e-01 6.17817283e-01 -4.64627922e-01 9.86280918e-01
9.28530991e-01 1.08269099e-02 1.70133919e-01 -3.94936085e-01
5.30667067e-01 -5.33402562e-01 -5.73126793e-01 1.37229228e+00
-4.88342196e-01 7.05514371e-01 1.85918182e-01 -7.17821121e-01
1.28072214e+00 3.50503445e-01 4.35149431e-01 -8.65048885e-01
4.68157768e-01 1.22001551e-01 -8.38128850e-02 -3.32687378e-01
6.24960475e-02 -4.52064186e-01 -2.02297449e-01 2.57298797e-01
2.08254725e-01 1.13099940e-01 -2.93038100e-01 5.06791472e-02
9.10035431e-01 1.54111087e-01 1.05772302e-01 -2.92058196e-02
6.39166415e-01 -5.25820613e-01 6.90865338e-01 1.95298269e-01
-2.16333926e-01 5.36774337e-01 8.40058982e-01 -3.90306532e-01
-6.73035681e-01 -7.97549844e-01 4.41039443e-01 9.65401292e-01
-2.82503739e-02 -3.22907686e-01 -1.21702003e+00 -8.69289994e-01
-1.05023861e-01 5.72535753e-01 -1.26128197e+00 -5.21732032e-01
-2.91629612e-01 -5.50731301e-01 1.11596930e+00 4.99401599e-01
8.00158858e-01 -1.24257672e+00 -1.16914347e-01 -2.28765178e-02
-1.19491205e-01 -1.36217284e+00 -5.37706912e-01 -4.93140936e-01
-2.05966026e-01 -8.37881625e-01 -8.46550167e-01 -5.21119773e-01
8.60259831e-01 -4.82948333e-01 1.05411065e+00 -2.43588630e-02
-3.91205430e-01 1.71329141e-01 -3.23839128e-01 -4.00751710e-01
-5.58671415e-01 -3.34348440e-01 2.60205027e-02 1.01397061e+00
3.19126159e-01 -8.73146355e-01 -4.82545048e-01 1.54271081e-01
-5.73958337e-01 2.07014248e-01 7.21736550e-01 1.04764175e+00
5.77441692e-01 -4.23416615e-01 8.28951657e-01 -1.04089487e+00
6.22162879e-01 -1.57762602e-01 -2.94013947e-01 2.65969753e-01
-4.02367145e-01 -2.78115660e-01 8.30059230e-01 -7.83997059e-01
-1.24677587e+00 1.99146375e-01 -4.43841159e-01 -8.68426859e-01
1.53054684e-01 -4.93760668e-02 -7.15533495e-01 -5.44522762e-01
5.95001221e-01 4.39460158e-01 3.25467438e-01 1.84357315e-01
5.74571550e-01 4.66132134e-01 8.55044484e-01 -3.79752904e-01
7.15217054e-01 3.89457971e-01 1.39182031e-01 -6.20093465e-01
-5.21001995e-01 4.37931120e-01 -2.82443225e-01 -4.71823514e-01
7.57817149e-01 -1.20490849e+00 -1.00641835e+00 1.00423872e+00
-1.09305716e+00 -4.02774036e-01 -3.24869275e-01 -6.67169914e-02
-5.64507425e-01 -3.22403833e-02 -6.82935417e-01 -8.38156581e-01
-6.71614408e-01 -1.05846870e+00 1.10722673e+00 3.96624893e-01
-3.36638391e-01 -6.53500080e-01 -2.51083136e-01 2.98436195e-01
5.58170557e-01 1.18987596e+00 4.94071364e-01 -1.82112247e-01
-1.33714661e-01 -4.45473105e-01 -2.84733504e-01 6.26587808e-01
1.46014214e-01 1.15250245e-01 -1.29745579e+00 -4.88674790e-02
-4.03059721e-02 -8.16248119e-01 4.38473076e-01 -4.45122346e-02
1.12018836e+00 -7.98382044e-01 4.74388227e-02 1.16805971e+00
1.15392041e+00 8.05428028e-02 1.07641947e+00 -3.39520782e-01
8.39194179e-01 5.88896036e-01 1.96687058e-01 5.39406836e-01
1.08706772e-01 8.42845619e-01 4.45169955e-01 -4.65415508e-01
-1.36115044e-01 -5.96498191e-01 6.27269208e-01 4.59977388e-01
-2.78702825e-01 -1.96702629e-01 -1.45385578e-01 -2.45525725e-02
-1.50595617e+00 -1.28724086e+00 2.39419013e-01 1.45182681e+00
9.48914111e-01 -2.12021008e-01 6.92733675e-02 -1.00721896e-01
6.45226061e-01 3.66761386e-01 -8.82315278e-01 -7.13896930e-01
-6.47151470e-01 5.75609148e-01 6.73308671e-02 1.93168953e-01
-8.54818940e-01 1.08346057e+00 5.13698006e+00 9.70883548e-01
-1.69828784e+00 1.53553605e-01 1.13026679e+00 -1.93888489e-02
-3.00254554e-01 -5.34332156e-01 -3.46208483e-01 4.56185341e-01
6.80358946e-01 -2.32701436e-01 5.49003124e-01 1.15336263e+00
-1.08194262e-01 5.78972518e-01 -8.46131504e-01 1.41705549e+00
5.62909484e-01 -1.18479133e+00 6.09138608e-03 1.07635697e-02
8.22256923e-01 -4.81597513e-01 3.97123665e-01 4.11480218e-01
1.42840415e-01 -1.52712929e+00 7.38256276e-01 5.61876237e-01
1.58341718e+00 -8.81196260e-01 6.17550492e-01 -1.87107325e-01
-9.32598352e-01 1.53945848e-01 -1.03709780e-01 -9.35459044e-03
5.92866726e-02 4.11631018e-01 -4.86964911e-01 3.95559281e-01
2.46308997e-01 6.56016469e-01 -4.71019924e-01 -1.61250800e-01
-6.51711643e-01 6.89176261e-01 -1.22313544e-01 -1.28836900e-01
3.46732959e-02 -3.12831312e-01 3.05162907e-01 1.14993107e+00
2.09451884e-01 2.46380642e-01 -3.61872137e-01 1.33965147e+00
-7.09744155e-01 1.89505488e-01 -4.90942210e-01 -4.61330742e-01
7.40877539e-02 1.60922444e+00 -8.60758200e-02 -1.76142395e-01
-2.07756042e-01 1.61674249e+00 2.34603018e-01 3.68317246e-01
-1.09945309e+00 -3.36149812e-01 9.50601220e-01 -9.61395651e-02
-8.58909339e-02 3.14558774e-01 -6.70561716e-02 -1.27190232e+00
1.98840976e-01 -1.41891146e+00 -5.56984358e-02 -1.00684190e+00
-1.24428892e+00 1.15710533e+00 -6.77257061e-01 -9.03147399e-01
-7.91876435e-01 -5.39927125e-01 -6.09467685e-01 8.39496374e-01
-1.10714495e+00 -1.72601140e+00 -7.72024810e-01 7.21425116e-01
1.26282517e-02 -4.58474815e-01 1.20651960e+00 3.31133306e-01
-5.76159358e-01 1.57709384e+00 -3.53360355e-01 6.93678439e-01
5.12520134e-01 -7.82400906e-01 3.19496661e-01 6.75188601e-01
2.60821104e-01 2.41931722e-01 4.32426095e-01 -2.40418494e-01
-1.63625526e+00 -1.26335549e+00 6.08323634e-01 -8.88672471e-02
3.18651825e-01 -8.10519576e-01 -7.38267839e-01 5.66421270e-01
4.24422435e-02 5.02091825e-01 8.59296858e-01 -3.56024861e-01
-7.20181227e-01 -3.57985973e-01 -1.49332774e+00 8.04120660e-01
1.27137160e+00 -7.44103372e-01 1.28440633e-01 -1.09010460e-02
3.70189995e-01 -4.33135897e-01 -8.33460033e-01 5.19525468e-01
9.15640771e-01 -1.18097746e+00 7.55395055e-01 -5.28908551e-01
6.87689602e-01 1.66772064e-02 -1.52978137e-01 -1.29796898e+00
-1.18168019e-01 -1.25134075e+00 3.49449925e-02 1.45725632e+00
1.92134857e-01 -6.51721597e-01 1.00857973e+00 6.83239102e-01
2.24126741e-01 -9.33557987e-01 -8.66669834e-01 -5.61949074e-01
-2.10963473e-01 -1.96726486e-01 9.15464461e-01 1.15134847e+00
-9.51774642e-02 4.62010503e-01 -8.55897725e-01 -2.93693304e-01
4.17103201e-01 2.19624832e-01 1.05300534e+00 -5.59031188e-01
-4.13662761e-01 -5.77655315e-01 -9.29098785e-01 -6.57498360e-01
7.71289349e-01 -8.69671702e-01 -2.92226374e-01 -5.79288006e-01
-6.65187761e-02 -4.86402251e-02 -8.77714157e-03 8.06863904e-01
-8.46518055e-02 6.77406609e-01 2.17611298e-01 -2.36596376e-01
-3.03063482e-01 1.05713189e+00 1.45495856e+00 -3.79834175e-01
1.83594093e-01 -1.49870083e-01 -1.03324521e+00 4.90409732e-01
6.81910276e-01 -1.81505233e-01 -5.22108018e-01 -4.65226779e-03
5.08822910e-02 1.07947133e-01 5.49413085e-01 -8.13078582e-01
-1.43884495e-01 7.37819169e-03 6.72327399e-01 2.90841162e-01
6.65988207e-01 -4.21663195e-01 5.45495093e-01 1.93926170e-01
-3.77805680e-01 9.42884758e-02 2.13361785e-01 2.75861025e-01
-4.36079830e-01 4.56244498e-01 9.53078985e-01 6.61145821e-02
-7.03626812e-01 6.19405806e-01 -2.94222627e-02 2.40097761e-01
1.11161280e+00 -2.23330438e-01 -1.24137394e-01 -9.67943728e-01
-6.19785011e-01 -3.16828817e-01 5.01729846e-01 5.05796134e-01
7.82192469e-01 -1.69232905e+00 -1.03773081e+00 6.12619340e-01
2.72515833e-01 -5.04088700e-01 4.65584934e-01 6.71096981e-01
-2.13379264e-01 -4.60850775e-01 -5.59627533e-01 -2.93655872e-01
-1.38366568e+00 1.96569040e-01 5.71376801e-01 -4.97595444e-02
-3.99113715e-01 1.14675820e+00 3.07746202e-01 -1.65511116e-01
-1.99012816e-01 4.03355658e-01 9.49282497e-02 -1.10604085e-01
6.91546202e-01 4.37834300e-02 -3.22788358e-01 -1.16915989e+00
-4.05817538e-01 4.73089904e-01 1.97085172e-01 1.31017968e-01
1.11126912e+00 2.34458491e-01 -1.68784022e-01 -3.10613234e-02
1.56287158e+00 1.30607232e-01 -1.29805517e+00 6.55189082e-02
-5.39325953e-01 -5.05697370e-01 -3.10946852e-01 -9.95506942e-01
-1.64594734e+00 8.43164086e-01 6.23094141e-01 -2.87493885e-01
1.52679789e+00 -1.46575987e-01 7.61684179e-01 -2.93197632e-02
2.50847667e-01 -7.02143610e-01 2.29720324e-01 2.52315402e-02
1.33695579e+00 -1.01081908e+00 -3.33134025e-01 -5.88215232e-01
-1.07572269e+00 1.03665853e+00 8.09469581e-01 -1.80396050e-01
3.90859902e-01 4.88638043e-01 4.72332925e-01 -8.39155912e-02
-5.82449257e-01 1.87395498e-01 1.80062488e-01 8.89320970e-01
2.68557906e-01 2.86585480e-01 1.43371359e-01 1.04604650e+00
-5.08990288e-01 2.17272580e-01 3.43997553e-02 2.05000311e-01
3.85675967e-01 -9.52810287e-01 7.91321471e-02 1.99135378e-01
-7.32207716e-01 3.03207040e-02 -7.94762790e-01 7.17168450e-01
2.55304009e-01 7.17179835e-01 -7.23229349e-02 -9.97568727e-01
3.12168360e-01 5.62597765e-03 7.02840984e-01 5.18284328e-02
-5.19387245e-01 -2.75347203e-01 2.53446132e-01 -8.34562957e-01
-4.25789982e-01 -3.93338203e-01 -1.08262360e+00 -5.25614440e-01
1.71416737e-02 -1.64592208e-03 4.44688469e-01 6.33591831e-01
8.17208588e-01 3.36772710e-01 1.08543503e+00 -7.03679323e-01
-5.70414901e-01 -9.01203513e-01 -5.94612241e-01 1.08697748e+00
7.17140883e-02 -5.58374882e-01 -3.30313295e-01 2.18790919e-01] | [12.920459747314453, 0.32853963971138] |
569bb6bb-f632-4a22-8f6d-467d5efb08de | reducing-false-alarms-in-video-surveillance | 2307.04159 | null | https://arxiv.org/abs/2307.04159v1 | https://arxiv.org/pdf/2307.04159v1.pdf | Reducing False Alarms in Video Surveillance by Deep Feature Statistical Modeling | Detecting relevant changes is a fundamental problem of video surveillance. Because of the high variability of data and the difficulty of properly annotating changes, unsupervised methods dominate the field. Arguably one of the most critical issues to make them practical is to reduce their false alarm rate. In this work, we develop a method-agnostic weakly supervised a-contrario validation process, based on high dimensional statistical modeling of deep features, to reduce the number of false alarms of any change detection algorithm. We also raise the insufficiency of the conventionally used pixel-wise evaluation, as it fails to precisely capture the performance needs of most real applications. For this reason, we complement pixel-wise metrics with object-wise metrics and evaluate the impact of our approach at both pixel and object levels, on six methods and several sequences from different datasets. Experimental results reveal that the proposed a-contrario validation is able to largely reduce the number of false alarms at both pixel and object levels. | ['Rafael Grompone von Gioi', 'Jean-Michel Morel', 'Gabriele Facciolo', 'Thibaud Ehret', 'Aitor Artola', 'Xavier Bou'] | 2023-07-09 | null | null | null | null | ['change-detection'] | ['computer-vision'] | [ 5.83194435e-01 -4.10959244e-01 1.78899810e-01 -4.29243177e-01
-5.61401129e-01 -6.91250324e-01 7.76631117e-01 5.23301303e-01
-4.97072339e-01 5.55064976e-01 -1.15059108e-01 3.44967060e-02
-1.71965316e-01 -7.44769514e-01 -5.99549949e-01 -7.23112404e-01
-1.63087711e-01 -7.88381044e-03 6.52224481e-01 -1.72038570e-01
8.18652734e-02 5.23235202e-01 -2.08206201e+00 2.94829816e-01
6.74305141e-01 1.30208802e+00 -1.58259690e-01 7.09772170e-01
1.99828520e-01 6.43257558e-01 -5.84994018e-01 -3.24756622e-01
3.93846273e-01 -4.58454460e-01 -2.88034171e-01 3.37801844e-01
6.38752878e-01 -3.57082367e-01 3.10631394e-02 1.41587126e+00
2.32615367e-01 -1.45814762e-01 5.52271366e-01 -1.19558871e+00
2.96567082e-02 3.31390351e-01 -5.88152289e-01 6.67336285e-01
9.63401794e-02 4.11153942e-01 8.23587596e-01 -5.18974721e-01
4.66327906e-01 1.03640366e+00 7.77355254e-01 2.10533097e-01
-1.22556198e+00 -2.97917724e-01 2.53090650e-01 2.90954202e-01
-1.07019722e+00 -4.45544481e-01 8.20403278e-01 -8.53795767e-01
5.31517863e-01 4.03598905e-01 5.84139645e-01 1.17992306e+00
6.25006929e-02 3.82548958e-01 1.16600192e+00 -2.88227350e-01
4.94572997e-01 1.87170759e-01 1.60461754e-01 5.29586017e-01
6.84870958e-01 1.01860918e-01 -2.08703116e-01 -5.17739952e-02
4.19399858e-01 1.73240289e-01 -2.97494978e-01 -7.18001068e-01
-9.71781194e-01 5.95620096e-01 1.28276169e-01 5.71087122e-01
-3.56878966e-01 2.05097850e-02 5.81093609e-01 3.72912198e-01
4.90870953e-01 4.55632567e-01 -4.99011666e-01 -3.29521269e-01
-1.13932645e+00 9.20636058e-02 5.50268590e-01 5.81819654e-01
4.85324651e-01 -1.72567323e-01 -3.34793806e-01 4.76987839e-01
3.54447253e-02 3.21768790e-01 3.57322454e-01 -5.63875020e-01
5.23988567e-02 9.57290173e-01 2.71023244e-01 -1.42051125e+00
-4.95147854e-01 -7.03608513e-01 -1.10250461e+00 4.69436198e-01
4.06476736e-01 -7.83204362e-02 -6.36231840e-01 1.71041358e+00
3.46296072e-01 1.14513725e-01 -1.92885235e-01 8.95506382e-01
4.09215689e-01 3.79684448e-01 1.28322318e-01 -5.63898861e-01
1.20500565e+00 -5.50726473e-01 -6.23858154e-01 -1.56820834e-01
5.91690719e-01 -5.59363008e-01 9.34199512e-01 4.31135386e-01
-7.40480721e-01 -7.35333741e-01 -1.02531302e+00 4.96056795e-01
-3.85319233e-01 3.14607501e-01 2.11254597e-01 5.88672221e-01
-8.30733597e-01 6.45455301e-01 -8.48008335e-01 -6.29657149e-01
4.50153023e-01 1.23227127e-01 -4.11926419e-01 1.02043599e-01
-9.19297993e-01 8.85432065e-01 3.93310577e-01 1.87670827e-01
-8.72735500e-01 -4.77257341e-01 -6.35684073e-01 1.33259997e-01
4.87729698e-01 -3.94134045e-01 1.02270567e+00 -1.24946785e+00
-9.67617810e-01 8.57325256e-01 1.61504135e-01 -7.67140388e-01
1.00485981e+00 -3.14127296e-01 -5.42335451e-01 9.25024226e-02
-2.56514009e-02 3.85533899e-01 1.12330425e+00 -1.22562122e+00
-9.29698229e-01 -3.48894686e-01 1.07922710e-01 -2.49434307e-01
-5.50916374e-01 -8.54884014e-02 -3.25143427e-01 -7.49722183e-01
-1.07977532e-01 -8.39497626e-01 -1.96386099e-01 2.10595369e-01
-3.16075720e-02 4.10323106e-02 1.03529406e+00 -5.08607507e-01
1.24400675e+00 -2.28279924e+00 -3.99283320e-02 9.65156481e-02
3.15928102e-01 6.04764163e-01 -1.37991115e-01 1.88662574e-01
-7.99155757e-02 -3.37703265e-02 -5.06592929e-01 -9.68453065e-02
-3.86409432e-01 7.50110224e-02 -1.21358909e-01 6.18926048e-01
7.24673569e-01 4.69951391e-01 -9.47118759e-01 -2.42975011e-01
4.87777233e-01 3.43717843e-01 -4.17927951e-01 2.67071426e-01
-3.63740414e-01 4.01847094e-01 -2.74193913e-01 3.20280761e-01
5.53923309e-01 -2.38762021e-01 1.66372910e-01 -4.20656204e-01
-2.34213114e-01 -1.82109680e-02 -1.25754297e+00 1.17230368e+00
-6.76479414e-02 8.41662169e-01 -1.86379775e-01 -1.37030077e+00
7.61869431e-01 1.52700529e-01 7.02414572e-01 -5.69516480e-01
1.61850467e-01 4.62692119e-02 8.68322030e-02 -6.93379581e-01
2.61979192e-01 1.33352771e-01 4.20395099e-02 -4.82121622e-03
-1.61108300e-01 2.91273803e-01 4.22020078e-01 -1.50498018e-01
1.38792002e+00 1.00349419e-01 5.78075945e-01 -4.97327328e-01
7.73803413e-01 1.11239411e-01 7.71641731e-01 8.28066051e-01
-4.20609802e-01 5.80118835e-01 6.83553994e-01 -7.46545255e-01
-8.19255292e-01 -7.48711407e-01 -5.11364043e-02 7.61688888e-01
2.18319759e-01 -1.69500843e-01 -8.77398074e-01 -9.64699447e-01
-5.02516469e-03 3.58520299e-01 -8.77037227e-01 -3.77621531e-01
-4.27866548e-01 -9.32706058e-01 3.49971145e-01 1.82241604e-01
4.60864931e-01 -8.46047580e-01 -1.34238374e+00 2.33687982e-01
-9.85175744e-02 -1.29300129e+00 -4.23695520e-03 4.76280786e-02
-8.46542597e-01 -1.48538220e+00 -3.14218253e-01 -3.53938460e-01
5.41720450e-01 3.32841426e-01 1.12463617e+00 -5.44473492e-02
-5.68950176e-01 3.68769556e-01 -5.51223934e-01 -2.97009677e-01
-5.57158411e-01 -1.21735357e-01 7.29445517e-02 4.71833467e-01
4.39199299e-01 -3.94153595e-01 -5.41114748e-01 3.54906172e-01
-1.17397249e+00 -2.06774235e-01 6.76080883e-01 4.72803354e-01
4.06462759e-01 4.32634592e-01 4.71025467e-01 -8.09391677e-01
2.64557809e-01 -3.47786337e-01 -9.92157638e-01 3.25804323e-01
-5.63610017e-01 1.20875845e-02 5.71147025e-01 -4.41686302e-01
-6.41189337e-01 -7.95044098e-03 2.65729815e-01 -2.89767861e-01
-4.93991464e-01 2.94741124e-01 -1.88366592e-01 1.65614616e-02
8.54814529e-01 9.16967988e-02 -1.00644864e-01 -3.66361380e-01
-5.21162935e-02 3.60126615e-01 4.28197831e-01 1.03692859e-01
9.06093299e-01 6.47327423e-01 7.20993951e-02 -9.78641093e-01
-8.52555275e-01 -6.54415250e-01 -7.26430178e-01 -5.30180633e-01
8.77062321e-01 -8.20651531e-01 -3.46928716e-01 5.22395670e-01
-1.05814934e+00 -1.26627088e-01 -4.90711927e-01 3.32234353e-01
-4.14359480e-01 5.54573119e-01 -3.64990272e-02 -7.50313044e-01
-1.75545216e-01 -1.15110874e+00 1.06433129e+00 9.31824520e-02
-1.14451803e-01 -7.50952244e-01 6.10580891e-02 1.67352781e-01
5.45779049e-01 8.07590246e-01 5.42617321e-01 -4.91280437e-01
-4.12572235e-01 -3.91974300e-01 -2.99980372e-01 4.72820193e-01
4.65013504e-01 2.66461879e-01 -9.71875668e-01 -4.70176697e-01
1.59003362e-01 8.66255462e-02 9.33062017e-01 4.03270930e-01
1.00110626e+00 -2.89165586e-01 -1.85527280e-01 2.73645878e-01
1.51116502e+00 -5.57572134e-02 5.03057420e-01 4.38729078e-01
3.95798355e-01 6.24309540e-01 7.89629221e-01 7.10608184e-01
4.50999327e-02 8.59247386e-01 7.87873745e-01 -1.03048153e-01
-1.15159571e-01 1.28310889e-01 5.24626315e-01 3.42563778e-01
-2.49006841e-02 -2.97213793e-01 -9.30304229e-01 5.68453014e-01
-1.95662129e+00 -1.11520493e+00 -3.33075970e-01 2.32791257e+00
5.83146513e-01 4.22375321e-01 2.82463491e-01 5.03644228e-01
8.20657909e-01 1.83337376e-01 -3.66639823e-01 1.01709075e-01
-1.94842160e-01 -2.78413445e-01 4.17512417e-01 3.89580689e-02
-1.64139271e+00 4.79143649e-01 5.18924332e+00 3.73914629e-01
-1.17062926e+00 1.52638137e-01 5.43917596e-01 -2.92656422e-02
2.36681506e-01 -2.58596390e-01 -5.39683342e-01 6.64424121e-01
7.91956127e-01 3.47173601e-01 1.07329391e-01 9.12484467e-01
4.62797552e-01 -2.88041860e-01 -1.30324924e+00 1.05928028e+00
2.60382503e-01 -1.13817799e+00 2.71733794e-02 -1.77908108e-01
7.76846409e-01 -9.90211964e-02 -1.63071901e-01 7.48506114e-02
-2.10531190e-01 -5.95409811e-01 7.93261766e-01 6.89958036e-01
4.57371742e-01 -4.03818756e-01 9.75176632e-01 1.93176091e-01
-9.37008619e-01 -2.11098120e-01 -6.47586361e-02 -1.27104029e-01
-6.10163808e-02 8.96840990e-01 -4.91867185e-01 2.39221171e-01
1.00184262e+00 8.99569511e-01 -8.46951962e-01 1.22481775e+00
6.88319579e-02 6.26849890e-01 -2.92278767e-01 9.80823115e-02
2.08258137e-01 -2.84250285e-02 7.27742970e-01 1.42389834e+00
8.13709795e-02 -1.57702804e-01 1.00809097e-01 4.62510198e-01
1.16723992e-01 -1.43698826e-01 -7.26491928e-01 1.63630173e-01
1.10855542e-01 1.24743295e+00 -8.44021797e-01 -2.90771276e-01
-3.93255323e-01 8.85424674e-01 2.84339283e-02 1.09348871e-01
-9.66150105e-01 -2.82835159e-02 7.06562757e-01 3.75070065e-01
5.30703604e-01 -2.64775485e-01 -1.34646192e-01 -1.09507382e+00
4.81034964e-01 -9.89337921e-01 3.98720652e-01 -2.09806696e-01
-1.25639474e+00 4.71079260e-01 -9.18392558e-03 -1.62114608e+00
-1.21105827e-01 -4.93079573e-01 -5.18504620e-01 2.43607596e-01
-1.60685766e+00 -8.54150951e-01 -8.41287374e-01 4.61503059e-01
6.00010037e-01 -6.35877103e-02 5.22894919e-01 4.76294994e-01
-7.06778586e-01 4.71052170e-01 9.75043178e-02 1.32486284e-01
5.38545310e-01 -9.99394417e-01 1.46302179e-01 1.43156278e+00
1.07280642e-01 1.69234797e-01 9.56200063e-01 -4.05650467e-01
-1.19353640e+00 -1.38298738e+00 4.77341861e-01 -4.98931080e-01
5.62108696e-01 -1.98110178e-01 -9.10317242e-01 2.08692148e-01
-3.51787806e-01 2.92301208e-01 4.20260698e-01 -2.55787194e-01
-2.36609116e-01 -5.01852214e-01 -1.22149348e+00 3.64547014e-01
9.90664780e-01 -2.78970689e-01 -4.53951597e-01 2.21867859e-01
3.56784910e-01 1.31675109e-01 -7.62469292e-01 8.47929478e-01
3.48413199e-01 -1.40375340e+00 7.00090230e-01 -4.55470800e-01
2.99356818e-01 -5.33847153e-01 -8.53689089e-02 -9.49858189e-01
-2.92516887e-01 -3.39970112e-01 -2.25292802e-01 1.37492192e+00
8.23534355e-02 -4.22957897e-01 6.11592054e-01 1.94968179e-01
1.55467987e-01 -3.30088228e-01 -9.96706545e-01 -7.30950832e-01
-5.41861355e-01 -4.69388008e-01 2.87694305e-01 9.98978317e-01
-5.45456231e-01 7.75668398e-02 -2.90728241e-01 5.07985294e-01
6.62978590e-01 9.89297405e-02 8.53451252e-01 -1.52426898e+00
-2.56028801e-01 -5.61950982e-01 -9.67789829e-01 -3.85239482e-01
-3.14644843e-01 -1.98301658e-01 1.97723940e-01 -1.15851927e+00
1.46714747e-01 -3.69006485e-01 -3.94271523e-01 2.66397923e-01
-2.26165548e-01 3.99717659e-01 -4.81434129e-02 2.02921584e-01
-7.60777950e-01 3.97218466e-01 5.73635042e-01 -1.57473609e-01
-1.24779426e-01 7.18753710e-02 -1.80478171e-01 9.78064597e-01
9.18401718e-01 -5.40394485e-01 -3.04010063e-01 -3.83111089e-01
1.85405388e-01 -3.06831986e-01 7.54217327e-01 -1.48033142e+00
-3.12655754e-02 -1.77070484e-01 3.36855292e-01 -3.61238837e-01
5.82761802e-02 -1.20036316e+00 7.92021230e-02 7.06384659e-01
-2.91532844e-01 6.71913624e-02 1.63224608e-01 7.28511512e-01
-3.38305354e-01 -1.41195683e-02 1.09007514e+00 1.01217717e-01
-9.71249580e-01 1.26960844e-01 -2.60579079e-01 -1.29322959e-02
1.34136713e+00 -2.16612741e-01 -1.03159569e-01 -1.85319766e-01
-4.55027223e-01 -3.30460817e-02 5.75233519e-01 6.75021112e-01
3.20411831e-01 -8.73122692e-01 -6.55242145e-01 3.87234449e-01
5.75118959e-01 -3.21551681e-01 1.70494512e-01 1.01771498e+00
-4.83273655e-01 2.33396873e-01 -3.52024704e-01 -9.60334957e-01
-1.39233112e+00 6.77848518e-01 4.53710824e-01 -2.45229229e-01
-5.39261639e-01 4.72782612e-01 2.09083080e-01 7.47190341e-02
3.56284529e-01 -5.53145826e-01 -4.18426037e-01 2.21015990e-01
6.59698963e-01 4.18723077e-01 3.14806879e-01 -5.05886257e-01
-4.48847145e-01 4.96851593e-01 5.31553030e-02 2.89540440e-01
1.43445551e+00 3.36730182e-02 7.63874361e-03 4.17047739e-01
8.97978842e-01 -2.34825000e-01 -1.51390994e+00 -4.44545336e-02
1.87124848e-01 -4.64450508e-01 7.13015497e-02 -6.90446556e-01
-9.10693169e-01 8.13401401e-01 1.21554244e+00 5.91840267e-01
1.37173271e+00 -2.76348025e-01 3.30176443e-01 2.93866485e-01
2.48292223e-01 -1.06008852e+00 -7.22790658e-02 1.35276154e-01
7.23884761e-01 -1.66205502e+00 -3.13308537e-02 -3.47355813e-01
-2.44491935e-01 9.85230505e-01 5.82965255e-01 -9.06114355e-02
5.09316146e-01 1.25493899e-01 -2.44412795e-02 -1.22845151e-01
-6.17361784e-01 -3.15840751e-01 2.25699857e-01 6.01115584e-01
2.22666934e-01 -1.70996413e-01 -4.47184205e-01 2.65673399e-01
3.90012950e-01 3.07724718e-02 4.43795085e-01 9.02114570e-01
-4.77313131e-01 -7.80836761e-01 -3.97640586e-01 5.50267696e-01
-4.14481312e-01 3.46915871e-01 -3.31784755e-01 7.65602767e-01
4.39986467e-01 1.04374027e+00 6.82089850e-02 -3.95151615e-01
4.65568066e-01 -3.35568756e-01 1.51796117e-01 -4.02554840e-01
-5.53396404e-01 -3.51605743e-01 -4.18070592e-02 -6.64239347e-01
-7.98141658e-01 -8.57325494e-01 -6.61000848e-01 4.29534726e-02
-1.42430782e-01 -5.61901070e-02 5.55034339e-01 8.17654908e-01
4.23537254e-01 4.99249965e-01 7.11800575e-01 -6.21305287e-01
-7.41360366e-01 -8.97687018e-01 -1.46977231e-01 9.12614286e-01
5.40444672e-01 -7.98049510e-01 -4.45596367e-01 3.17358166e-01] | [7.994787216186523, 1.4120924472808838] |
633395e9-0f49-449b-9221-8cf103318836 | learning-rgb-d-feature-embeddings-for-unseen | 2007.15157 | null | https://arxiv.org/abs/2007.15157v3 | https://arxiv.org/pdf/2007.15157v3.pdf | Learning RGB-D Feature Embeddings for Unseen Object Instance Segmentation | Segmenting unseen objects in cluttered scenes is an important skill that robots need to acquire in order to perform tasks in new environments. In this work, we propose a new method for unseen object instance segmentation by learning RGB-D feature embeddings from synthetic data. A metric learning loss function is utilized to learn to produce pixel-wise feature embeddings such that pixels from the same object are close to each other and pixels from different objects are separated in the embedding space. With the learned feature embeddings, a mean shift clustering algorithm can be applied to discover and segment unseen objects. We further improve the segmentation accuracy with a new two-stage clustering algorithm. Our method demonstrates that non-photorealistic synthetic RGB and depth images can be used to learn feature embeddings that transfer well to real-world images for unseen object instance segmentation. | ['Yu Xiang', 'Dieter Fox', 'Christopher Xie', 'Arsalan Mousavian'] | 2020-07-30 | null | null | null | null | ['unseen-object-instance-segmentation'] | ['computer-vision'] | [ 4.04732466e-01 1.75468296e-01 2.43648559e-01 -7.79780984e-01
-5.42292476e-01 -6.34626091e-01 1.35728538e-01 4.75004092e-02
-6.25512779e-01 3.22434306e-01 -3.17089885e-01 3.45420182e-01
-5.60088046e-02 -7.26787329e-01 -1.08509469e+00 -7.38934875e-01
2.83450723e-01 7.94110060e-01 5.05957901e-01 4.09119949e-02
3.16482812e-01 8.92930329e-01 -1.69059789e+00 4.65882830e-02
6.43622398e-01 8.23450208e-01 5.80625355e-01 5.39624631e-01
-2.26045474e-01 4.37834486e-02 -3.97865355e-01 9.70414504e-02
7.52469659e-01 -2.34096795e-01 -9.24427748e-01 6.64557159e-01
4.23926950e-01 -2.13429525e-01 -2.76597351e-01 1.00610971e+00
1.32917434e-01 4.31044042e-01 9.41133022e-01 -1.25071299e+00
-7.95778930e-01 9.77189690e-02 -5.75248063e-01 -2.20299169e-01
2.58393228e-01 1.39420390e-01 7.46004581e-01 -9.38496292e-01
6.59327626e-01 1.20803475e+00 2.75656372e-01 8.19001496e-01
-1.48675537e+00 -2.77330011e-01 2.57413667e-02 2.27638245e-01
-1.29530692e+00 2.63542105e-02 1.01611352e+00 -5.32956243e-01
5.26249051e-01 1.18301846e-01 7.80345023e-01 9.31447327e-01
-1.04550995e-01 9.61705506e-01 9.26292777e-01 -2.27826804e-01
5.69051087e-01 3.22379917e-01 -9.47950482e-02 5.99273503e-01
3.31341982e-01 -2.64873765e-02 -2.37931892e-01 3.90883267e-01
8.36872101e-01 4.19932216e-01 -1.48237392e-01 -1.06794012e+00
-1.34798253e+00 8.74081194e-01 9.05401409e-01 2.52739996e-01
-2.50725508e-01 3.58392596e-01 -1.51528232e-02 -8.21707118e-03
1.23907581e-01 8.57261300e-01 -5.29989064e-01 2.73849331e-02
-7.55356193e-01 8.75599533e-02 6.15785956e-01 1.19851053e+00
1.27411532e+00 -2.84838200e-01 1.66478321e-01 6.53857648e-01
3.54167849e-01 4.61351126e-01 5.25985241e-01 -1.34431648e+00
-3.21958363e-02 1.01785946e+00 2.03613192e-01 -7.96093881e-01
-3.40254456e-01 8.69758949e-02 -3.95601451e-01 5.31714320e-01
3.91112387e-01 2.19908029e-01 -1.31456971e+00 1.22631836e+00
5.99237084e-01 1.60521075e-01 2.16476887e-01 1.18049490e+00
5.40328205e-01 6.22279108e-01 -4.10850585e-01 2.98676252e-01
1.06833899e+00 -1.00735509e+00 -2.54716188e-01 -4.98530716e-01
4.65069264e-01 -4.69592869e-01 1.20009720e+00 2.34841987e-01
-8.67353976e-01 -5.92273772e-01 -1.15262163e+00 -3.59952062e-01
-4.49044555e-01 -1.64389148e-01 4.06966805e-01 3.77431124e-01
-6.61267817e-01 6.00017905e-01 -1.14593887e+00 -4.88570273e-01
6.76658511e-01 4.58025873e-01 -5.60617566e-01 -3.19205821e-01
-4.94749963e-01 5.95183671e-01 8.95327985e-01 5.81233501e-02
-1.12558603e+00 -4.98457581e-01 -1.11717129e+00 -2.33464494e-01
2.20601812e-01 -4.50299948e-01 8.92662764e-01 -8.93656790e-01
-1.39858878e+00 1.12314248e+00 1.49480868e-02 -1.92406997e-01
3.78736079e-01 -2.31179878e-01 6.79639280e-02 4.83264774e-01
1.52103305e-01 1.09056091e+00 1.01390529e+00 -1.74098647e+00
-7.07265079e-01 -8.19575369e-01 -2.21996710e-01 3.12517881e-01
-2.58465767e-01 -6.02054298e-01 -4.00046855e-01 -3.64704192e-01
7.92410016e-01 -1.03120172e+00 -3.38341862e-01 4.36593115e-01
-4.30603564e-01 -1.45895526e-01 1.34392571e+00 -2.36066997e-01
3.66428673e-01 -2.19640923e+00 6.22936070e-01 5.66076152e-02
2.67733842e-01 1.41092405e-01 -1.92355886e-01 1.80410836e-02
2.06115887e-01 -8.48195553e-02 -6.56479359e-01 -2.88573176e-01
1.07970200e-01 7.91540444e-01 5.82213961e-02 5.75680435e-01
3.42501819e-01 1.07745886e+00 -1.10131645e+00 -4.56947595e-01
4.88443911e-01 4.13423270e-01 -4.03476775e-01 4.42074448e-01
-3.32468659e-01 7.31085241e-01 -4.81123328e-01 5.71510613e-01
7.05621004e-01 4.57750633e-03 -5.49269378e-01 -7.71819875e-02
1.44812882e-01 -5.45238614e-01 -1.17598510e+00 2.08968687e+00
-2.94882953e-01 7.14550495e-01 -9.04839113e-02 -1.15927708e+00
1.03979707e+00 -1.85686305e-01 3.46322596e-01 -3.73821646e-01
2.74333149e-01 2.31405348e-01 -3.68106991e-01 -9.22715306e-01
5.14575839e-01 -3.70217748e-02 -1.92016289e-01 4.63505507e-01
4.98060882e-02 -9.50857043e-01 2.58038323e-02 -7.62898922e-02
8.28706622e-01 8.48139357e-03 -4.80447084e-01 5.12564741e-02
2.60418743e-01 2.52371579e-01 5.35358727e-01 4.11332905e-01
-1.66067779e-01 1.03309524e+00 2.91324151e-03 -5.31997085e-01
-1.27071357e+00 -1.57519567e+00 -2.85223842e-01 8.12920809e-01
7.84723818e-01 5.09936631e-01 -8.15296531e-01 -7.59963334e-01
2.20548809e-01 7.36530244e-01 -7.65465498e-01 -4.18712616e-01
-5.44561744e-01 -1.56739280e-01 1.04892254e-03 7.32994378e-01
5.05278826e-01 -1.02235818e+00 -1.03587127e+00 1.79992601e-01
8.90884250e-02 -1.17975056e+00 -4.24879313e-01 4.51095760e-01
-9.87131298e-01 -1.04818869e+00 -8.53542686e-01 -1.41585672e+00
1.04657984e+00 4.28072423e-01 5.46035886e-01 -3.67797226e-01
-6.80447459e-01 7.10975170e-01 -5.28631926e-01 -2.01391265e-01
-2.59910077e-01 -5.87933399e-02 1.97472885e-01 2.57181108e-01
5.54099381e-01 -3.28648508e-01 -7.20896423e-01 5.43468177e-01
-1.26986349e+00 -8.54281858e-02 6.54011428e-01 6.39374316e-01
9.12169158e-01 -1.06684096e-01 9.47506577e-02 -4.31914330e-01
1.08687676e-01 -2.04609469e-01 -6.34826183e-01 1.41819179e-01
-2.48175234e-01 3.98793280e-01 4.39170837e-01 -7.26153255e-01
-6.10065758e-01 7.26470411e-01 3.70617151e-01 -6.99199677e-01
-4.15560365e-01 -7.30011240e-02 -1.79709509e-01 -2.41089508e-01
6.48453593e-01 3.16934317e-01 1.14924066e-01 -4.01409417e-01
7.43550539e-01 8.61346900e-01 6.30359292e-01 -4.30863529e-01
1.05974472e+00 9.46575165e-01 -9.72464979e-02 -8.74601543e-01
-7.09301591e-01 -6.96893275e-01 -1.25124204e+00 -1.28098935e-01
1.29844999e+00 -6.27801180e-01 -3.58574212e-01 3.75062138e-01
-1.03450906e+00 -3.88669908e-01 -7.73687661e-01 6.27703726e-01
-9.47499037e-01 2.33705744e-01 -2.62826681e-01 -5.15368402e-01
2.43759289e-01 -1.22900093e+00 1.30651224e+00 6.03944063e-01
1.99735146e-02 -8.35263908e-01 -1.32668018e-02 6.15440726e-01
-1.06762968e-01 5.72751820e-01 7.93487728e-01 -6.55054390e-01
-8.02853882e-01 -2.09340870e-01 -2.44912922e-01 6.78500652e-01
2.92826951e-01 -5.29717952e-02 -9.03085411e-01 -3.02391201e-01
1.24496348e-01 -4.40001398e-01 7.69442081e-01 3.58648479e-01
1.19312727e+00 5.98664172e-02 -5.66508889e-01 8.43516827e-01
1.38914323e+00 1.57612056e-01 4.63454664e-01 2.21899942e-01
9.77270424e-01 7.29886711e-01 6.83515728e-01 1.72512323e-01
1.69337019e-01 2.87297249e-01 5.89363754e-01 -1.41051024e-01
7.81919956e-02 -1.92046851e-01 1.44898808e-02 4.29789126e-01
4.90609497e-01 -2.05846746e-02 -9.59987998e-01 9.17521000e-01
-1.67759144e+00 -5.21444738e-01 -2.22803816e-01 2.11926222e+00
6.02005541e-01 6.98147416e-02 -1.73199967e-01 1.41553268e-01
7.64812350e-01 -3.00002456e-01 -1.00418973e+00 -3.29536527e-01
-5.70364557e-02 2.33470365e-01 4.61086363e-01 3.67493123e-01
-1.07932532e+00 1.23770070e+00 5.72735929e+00 2.97012866e-01
-1.04196382e+00 -1.58791333e-01 3.55035126e-01 -3.08199264e-02
-2.28189200e-01 -1.37278795e-01 -4.58630055e-01 2.16622129e-01
4.29758787e-01 -8.00468121e-03 3.07991266e-01 9.05461907e-01
3.23722102e-02 -2.53324091e-01 -1.44737244e+00 1.13236177e+00
2.32317567e-01 -9.23029602e-01 6.17494062e-02 2.50922758e-02
1.03945816e+00 1.23659149e-02 3.58863384e-01 -4.11443934e-02
1.95998877e-01 -1.10875988e+00 4.84960467e-01 4.86059368e-01
4.20575291e-01 -5.47370076e-01 4.29194778e-01 2.86196530e-01
-9.84273732e-01 -1.88688025e-01 -7.01029539e-01 2.13065416e-01
1.78371325e-01 4.10601854e-01 -1.09013557e+00 1.23794541e-01
7.73284614e-01 7.98531830e-01 -6.43493295e-01 1.18631339e+00
-3.61597836e-01 5.45836464e-02 -4.30608153e-01 -9.57590044e-02
4.20515567e-01 -4.72201794e-01 5.27374804e-01 6.35285258e-01
2.56772757e-01 8.91579017e-02 -5.04170125e-03 1.30072927e+00
-1.74798369e-02 -2.43987113e-01 -6.90158188e-01 -4.31984384e-03
3.13254803e-01 1.25849092e+00 -1.06400132e+00 -2.18926311e-01
-7.03742653e-02 1.62643433e+00 2.40789533e-01 5.76535106e-01
-7.71102369e-01 -6.81658149e-01 7.60257363e-01 -9.74386930e-03
6.73918426e-01 -6.12158120e-01 -2.43850261e-01 -8.99210095e-01
4.48848866e-02 -1.13201030e-01 -2.42181160e-02 -1.00225055e+00
-1.18869150e+00 4.02341485e-01 -1.32599086e-01 -1.34912241e+00
6.60828874e-02 -9.04641807e-01 -3.24931085e-01 3.50262403e-01
-1.40305305e+00 -9.63940740e-01 -6.48723483e-01 5.97949088e-01
6.59714222e-01 1.54252172e-01 5.97086549e-01 -1.22185595e-01
-2.54418910e-01 1.62109628e-01 5.39181292e-01 3.69937122e-01
4.37242329e-01 -1.51047039e+00 1.25025362e-01 5.57503760e-01
4.28452641e-01 1.70310140e-01 6.31080925e-01 -3.50992471e-01
-1.23670816e+00 -1.35819912e+00 2.20804513e-01 -6.61296129e-01
2.76071876e-01 -7.10263550e-01 -9.32029843e-01 6.93610966e-01
-1.44713566e-01 4.09706563e-01 6.13220036e-01 -5.58973789e-01
-2.34439164e-01 -1.10675767e-01 -1.49089587e+00 4.37259316e-01
9.62465405e-01 -3.33130032e-01 -1.07864594e+00 3.42365116e-01
1.01944792e+00 -3.59879643e-01 -8.57942402e-01 2.59168267e-01
4.14077759e-01 -7.77819753e-01 1.05896699e+00 -7.23639786e-01
1.82075337e-01 -5.73391438e-01 -3.32228661e-01 -1.52810311e+00
2.64569018e-02 -3.37023675e-01 2.22626895e-01 7.55922139e-01
3.29660505e-01 -4.87338841e-01 1.00030649e+00 8.24460268e-01
-2.98833340e-01 -5.36528707e-01 -9.25669372e-01 -1.11672914e+00
1.58636644e-01 -2.29052901e-01 4.47112858e-01 6.34106994e-01
-3.93003106e-01 4.50867787e-02 3.80938977e-01 4.30290461e-01
9.50042725e-01 4.36719388e-01 8.38106275e-01 -1.44121075e+00
2.07300872e-01 -1.98576838e-01 -1.02084184e+00 -1.17890859e+00
2.76961952e-01 -9.04146910e-01 3.87602568e-01 -1.70007837e+00
-6.17193170e-02 -4.62119013e-01 -1.46223426e-01 1.99292913e-01
1.15442514e-01 5.31828642e-01 4.19221446e-02 5.53034358e-02
-6.01580322e-01 8.42807114e-01 1.46580458e+00 -4.67521757e-01
-3.09288472e-01 -1.34288087e-01 -2.22041577e-01 6.89892828e-01
7.55915165e-01 -7.31259823e-01 -2.74477005e-01 -8.43031406e-01
-1.93342760e-01 -4.37911391e-01 3.82300973e-01 -1.27011561e+00
2.26302326e-01 -8.70351642e-02 6.58680201e-01 -4.87332284e-01
6.64202213e-01 -1.26339889e+00 -1.97545826e-01 5.30890167e-01
-2.75485843e-01 -3.67078394e-01 7.34971762e-02 8.27343464e-01
-7.70640895e-02 -4.61532354e-01 9.60588336e-01 -1.90457195e-01
-9.04295683e-01 4.18128252e-01 -2.24552631e-01 2.69429803e-01
1.75912070e+00 -8.93422544e-01 2.71379411e-01 8.91501233e-02
-9.50130999e-01 4.02976036e-01 8.89490426e-01 5.13398647e-01
1.16415203e+00 -1.34161508e+00 -5.29078186e-01 4.33833361e-01
5.40980935e-01 6.87286675e-01 -2.06070170e-02 3.77966762e-01
-9.63280737e-01 -1.62558351e-02 -3.65140229e-01 -1.13770783e+00
-9.18349862e-01 7.71104455e-01 2.98027188e-01 6.46059990e-01
-5.10990500e-01 1.18473685e+00 1.35346770e-01 -7.70529091e-01
1.82856187e-01 -6.70027614e-01 2.27237001e-01 -8.43665451e-02
1.83422759e-01 3.18233132e-01 -3.04718375e-01 -6.65830851e-01
-1.00908801e-01 1.02317011e+00 -6.66505918e-02 -1.48106664e-01
1.56008041e+00 -3.05228323e-01 -1.14399334e-03 7.80000508e-01
1.83712661e+00 -6.56749249e-01 -1.82097554e+00 -2.70047426e-01
1.90674230e-01 -8.68623257e-01 9.00741667e-03 -4.15068775e-01
-1.16837239e+00 1.05817664e+00 9.40929234e-01 3.73169919e-03
9.08405960e-01 4.87548739e-01 1.03868759e+00 5.55020452e-01
4.66846466e-01 -1.40474641e+00 6.51212037e-01 2.07889259e-01
6.29036605e-01 -1.53317547e+00 -2.86554486e-01 -8.50850865e-02
-5.70222139e-01 1.23794496e+00 5.35345137e-01 -4.53341812e-01
3.67648214e-01 -7.10760951e-02 1.77232966e-01 -3.54803801e-01
2.92988960e-02 -3.56363118e-01 1.63720042e-01 8.02297711e-01
-3.41855645e-01 1.50204701e-02 2.89772451e-01 5.22245616e-02
-8.14983174e-02 -3.90219539e-01 6.32855713e-01 1.01965117e+00
-8.10200214e-01 -9.30720925e-01 -4.19530421e-01 2.79442608e-01
3.81639123e-01 5.21932900e-01 -5.79672515e-01 7.01439559e-01
1.49442956e-01 8.49707544e-01 4.64005142e-01 -4.45859015e-01
4.08321470e-01 3.18282656e-02 7.55009711e-01 -9.48201776e-01
-1.25051141e-02 -2.34529585e-01 -8.55733156e-01 -6.33795261e-01
-3.83170485e-01 -6.20431244e-01 -1.74771702e+00 2.65611708e-01
-2.57922977e-01 -4.43903031e-03 9.19456005e-01 7.09439218e-01
2.41255030e-01 2.71017492e-01 9.14149344e-01 -1.12851620e+00
-2.11616993e-01 -6.52457356e-01 -7.75492191e-01 7.15004265e-01
3.90164465e-01 -6.59558594e-01 -5.76266408e-01 3.24212223e-01] | [7.885512351989746, -2.790205955505371] |
f8eaaffb-cbbc-4d91-b023-7567daa5aa1a | safe-risk-averse-bayesian-optimization-for | 2306.13479 | null | https://arxiv.org/abs/2306.13479v1 | https://arxiv.org/pdf/2306.13479v1.pdf | Safe Risk-averse Bayesian Optimization for Controller Tuning | Controller tuning and parameter optimization are crucial in system design to improve both the controller and underlying system performance. Bayesian optimization has been established as an efficient model-free method for controller tuning and adaptation. Standard methods, however, are not enough for high-precision systems to be robust with respect to unknown input-dependent noise and stable under safety constraints. In this work, we present a novel data-driven approach, RaGoOSE, for safe controller tuning in the presence of heteroscedastic noise, combining safe learning with risk-averse Bayesian optimization. We demonstrate the method for synthetic benchmark and compare its performance to established BO-based tuning methods. We further evaluate RaGoOSE performance on a real precision-motion system utilized in semiconductor industry applications and compare it to the built-in auto-tuning routine. | ['Alisa Rupenyan', 'Andreas Krause', 'Efe C. Balta', 'Anastasia Makarova', 'Miks Ozols', 'Christopher Koenig'] | 2023-06-23 | null | null | null | null | ['bayesian-optimization'] | ['methodology'] | [-3.37322541e-02 -2.41630241e-01 -2.22019434e-01 2.04816144e-02
-9.49742913e-01 -5.11310756e-01 3.35176349e-01 9.31666940e-02
-1.80122450e-01 1.15101039e+00 -4.82708722e-01 -4.77418363e-01
-7.38371849e-01 -4.12577599e-01 -4.76103991e-01 -9.05052483e-01
2.32659847e-01 6.21266365e-01 3.55695546e-01 -9.48895961e-02
3.35328937e-01 4.01248932e-01 -1.22129500e+00 -5.18817008e-01
6.14597857e-01 8.76572669e-01 1.56959686e-02 6.86746299e-01
7.14285791e-01 2.35038593e-01 -4.12397891e-01 2.43201226e-01
1.54997155e-01 -2.08893523e-01 -1.20519117e-01 -1.88051835e-01
-3.42391461e-01 6.14787079e-02 1.35657728e-01 7.48708069e-01
7.86354065e-01 3.15678388e-01 9.65010583e-01 -9.99627292e-01
1.10102668e-01 2.84722596e-01 -2.19257444e-01 -2.18936875e-02
-1.02388680e-01 4.87983435e-01 5.31804979e-01 -5.20834684e-01
8.25122371e-02 1.26423264e+00 4.64713484e-01 3.56394112e-01
-1.68354726e+00 -8.40850174e-01 -1.29608974e-01 -3.82995218e-01
-1.60179496e+00 -5.51744878e-01 5.31954765e-01 -9.62584913e-01
8.14231992e-01 3.30747701e-02 3.13129038e-01 9.96868372e-01
8.74850452e-01 1.86046347e-01 1.03091061e+00 -4.62580353e-01
7.10230112e-01 3.81791294e-01 -7.98458830e-02 2.88568616e-01
4.70822752e-01 1.07871711e+00 -2.75670737e-01 -3.99994284e-01
7.72598565e-01 -7.03963578e-01 4.77591413e-04 -7.76433825e-01
-1.05240786e+00 8.92248273e-01 -2.33080355e-03 -3.04332003e-03
-3.50184500e-01 6.24677658e-01 2.71094620e-01 2.32169166e-01
1.89751983e-01 8.28043818e-01 -6.57907307e-01 -2.88041979e-01
-6.40223503e-01 4.58601683e-01 7.98531532e-01 9.70013559e-01
1.09408930e-01 7.23361492e-01 -5.03321707e-01 5.26636660e-01
8.66941452e-01 7.09118426e-01 1.22950763e-01 -9.56733227e-01
1.15716355e-02 1.23048946e-01 7.03053057e-01 -8.14796567e-01
-5.11419177e-01 -6.41332686e-01 -5.90096295e-01 6.42085254e-01
3.56227040e-01 -4.02114630e-01 -7.88956463e-01 1.36928248e+00
3.25732023e-01 -6.13909550e-02 -6.67532608e-02 4.54696804e-01
-1.97368264e-01 4.78292763e-01 -1.05635419e-01 -5.69460928e-01
8.08398187e-01 -2.32986763e-01 -1.02189875e+00 5.98394731e-03
1.27786219e-01 -9.09083068e-01 9.47129786e-01 8.51107121e-01
-9.09020782e-01 -5.37579358e-01 -1.32856858e+00 8.57853055e-01
-1.36449924e-02 6.00323081e-02 8.11039284e-02 1.22156012e+00
-3.62897247e-01 7.75917709e-01 -9.93931234e-01 -4.31888066e-02
2.22338587e-01 6.43890917e-01 3.22692782e-01 7.17964292e-01
-1.13237166e+00 1.17595053e+00 7.96689153e-01 1.05017543e-01
-1.29311097e+00 -8.78817737e-01 -5.53762138e-01 -2.75672972e-01
9.66735363e-01 -6.81846142e-01 1.53237522e+00 -2.65294826e-03
-2.48702073e+00 5.37204966e-02 2.76479751e-01 -5.25496244e-01
4.02547777e-01 -3.67892176e-01 -4.03105080e-01 -5.48333526e-01
-5.38788319e-01 3.28830257e-02 1.50441945e+00 -1.13777018e+00
-5.25163531e-01 -1.38948023e-01 -4.71666783e-01 -3.05189759e-01
1.43381000e-01 -1.33319780e-01 -9.20758694e-02 -5.73985457e-01
-2.52420336e-01 -1.26047015e+00 -4.92834955e-01 -3.66167486e-01
-6.53090119e-01 1.77037671e-01 6.76890314e-01 -3.23746622e-01
1.68617046e+00 -1.88005257e+00 6.29224628e-02 8.05787146e-01
-5.12366176e-01 3.94216150e-01 4.81506437e-01 4.56273317e-01
3.08009833e-02 -8.65428001e-02 -1.34730771e-01 1.59378350e-01
2.90364563e-01 1.18388809e-01 -2.68495739e-01 6.44105673e-01
3.51180315e-01 3.92340660e-01 -5.44173777e-01 -1.98097363e-01
6.34608924e-01 2.86607832e-01 -5.80501676e-01 1.63449019e-01
-4.20618832e-01 7.74983585e-01 -7.28465617e-01 4.51391041e-01
3.95509936e-02 3.04856181e-01 -1.91891752e-02 -1.11552864e-01
-1.80899635e-01 -3.29605728e-01 -1.62772775e+00 9.93790984e-01
-7.15847611e-01 1.22975454e-01 2.87804365e-01 -6.29547477e-01
1.20851910e+00 3.53481680e-01 2.73268938e-01 -2.24344343e-01
7.92027116e-01 2.86978841e-01 1.26695499e-01 -2.05352649e-01
1.56989366e-01 -2.57009506e-01 -3.44262332e-01 -1.90043524e-02
-5.11130411e-03 -1.01667607e+00 -8.26931447e-02 -4.05643880e-01
8.39906693e-01 2.70607173e-01 6.60014570e-01 -7.81822741e-01
7.30156541e-01 -3.67135443e-02 6.56778812e-01 6.55208945e-01
-2.34301388e-01 1.71900317e-01 5.69623172e-01 1.39781103e-01
-1.01371527e+00 -1.09504974e+00 -2.96173662e-01 5.27589798e-01
-1.84728384e-01 -1.25732094e-01 -5.76860189e-01 -3.09411064e-02
2.54405171e-01 1.18844378e+00 -2.36382738e-01 -6.45840704e-01
-1.86073169e-01 -7.46434927e-01 1.90749362e-01 3.43250662e-01
-1.53560042e-01 -4.96791899e-01 -7.23795891e-01 7.81414747e-01
6.83009028e-01 -6.37633145e-01 -3.25908601e-01 5.91749251e-01
-8.04371595e-01 -9.19337213e-01 -3.36937189e-01 -1.00656450e-01
9.96531397e-02 -7.85937309e-01 9.89639819e-01 -4.22773182e-01
-5.71720362e-01 5.11932135e-01 2.91210741e-01 -9.06888127e-01
-8.47478330e-01 -1.05934076e-01 5.41196942e-01 -1.43163830e-01
-1.69897273e-01 -2.11362153e-01 -2.43973762e-01 8.66311371e-01
-4.36544925e-01 -7.05596149e-01 3.50096405e-01 1.06680012e+00
8.07501256e-01 6.48785055e-01 7.91867673e-01 -7.73785055e-01
8.17397892e-01 -8.27160478e-02 -1.79171956e+00 1.45928860e-01
-9.80933130e-01 2.43693441e-01 2.30255812e-01 -5.02115011e-01
-1.08099067e+00 7.41049409e-01 1.57275543e-01 -4.34475422e-01
1.73001990e-01 5.15073001e-01 -4.27746058e-01 -2.41862431e-01
8.01306427e-01 -4.69890028e-01 2.94565827e-01 -2.65942514e-01
1.55606210e-01 6.32683575e-01 5.29846370e-01 -8.75616193e-01
1.07852244e+00 -1.09948538e-01 4.35416341e-01 -6.48359418e-01
-5.31982303e-01 -3.64648372e-01 -5.25017738e-01 -2.51299620e-01
7.48845458e-01 -9.26523447e-01 -1.12103391e+00 4.87948298e-01
-5.52089989e-01 -6.37467563e-01 -2.54398614e-01 5.82516968e-01
-1.24339092e+00 -3.20316464e-01 1.53585477e-02 -1.42237771e+00
-1.08437151e-01 -1.38033438e+00 8.41931760e-01 1.03881933e-01
-4.42229897e-01 -1.00352502e+00 1.74968958e-01 -5.38276285e-02
6.39459908e-01 5.01616120e-01 8.31192195e-01 -4.18324143e-01
-4.97440428e-01 -5.14863491e-01 5.20533502e-01 5.05314112e-01
1.00778520e-01 4.71658975e-01 -8.34922910e-01 -4.68302310e-01
2.97225863e-01 -6.69463128e-02 2.67095506e-01 8.11663210e-01
7.56751359e-01 -3.00121903e-02 -4.44721937e-01 7.75194466e-02
1.39795828e+00 6.21281266e-01 3.64051670e-01 4.25078601e-01
3.61995220e-01 3.95991564e-01 1.18710518e+00 7.21880913e-01
-5.03664255e-01 8.97255301e-01 3.61206055e-01 4.65333253e-01
6.12920702e-01 -8.12441260e-02 3.43873590e-01 4.27772790e-01
4.62129384e-01 -1.69122413e-01 -9.16589797e-01 3.30368996e-01
-1.90856183e+00 -4.46880430e-01 -2.53533363e-01 2.72464538e+00
8.33690226e-01 6.30442679e-01 1.00923836e-01 2.22158685e-01
7.90999532e-01 -6.29825711e-01 -7.32422650e-01 -4.48007137e-01
3.83071810e-01 1.56866238e-01 8.54366720e-01 6.70193970e-01
-1.28923261e+00 5.14510334e-01 6.96791792e+00 1.21125007e+00
-6.99670196e-01 -1.84360415e-01 4.29693013e-01 2.85007153e-02
2.20409080e-01 -6.50034919e-02 -1.22540843e+00 3.10885876e-01
1.49085021e+00 -4.35704261e-01 3.74274582e-01 9.59500074e-01
8.05837750e-01 -4.78122383e-01 -1.06457126e+00 6.87239349e-01
-5.12088120e-01 -9.56321836e-01 -7.60836244e-01 -4.35352549e-02
8.43682647e-01 -5.64520657e-01 3.74106884e-01 4.39555496e-01
4.50358391e-01 -1.06751072e+00 6.20676517e-01 6.32061303e-01
4.01428401e-01 -1.11654484e+00 8.42511415e-01 3.18129152e-01
-8.15254092e-01 -4.76794481e-01 -3.33206765e-02 2.94040263e-01
4.85128134e-01 6.73838854e-01 -9.70384598e-01 3.37461472e-01
3.93881202e-01 2.86748201e-01 -2.51915395e-01 1.24237084e+00
-6.40692487e-02 6.92353308e-01 -4.70494300e-01 -1.73441961e-01
6.35342300e-02 -1.18345611e-01 9.51218665e-01 6.42735898e-01
3.87358546e-01 -2.51690775e-01 1.67082220e-01 1.06895125e+00
8.36411536e-01 -2.37137079e-01 -4.79084194e-01 -1.98839992e-01
5.57824194e-01 8.51692200e-01 -3.77701253e-01 8.08141530e-02
3.17544252e-01 -3.18993144e-02 -5.55670917e-01 2.04805970e-01
-9.98649418e-01 -4.63610530e-01 5.55006325e-01 -9.81486142e-02
4.73419338e-01 -3.64067703e-01 -3.45622033e-01 -1.77053690e-01
-4.23645198e-01 -1.07965541e+00 3.08847249e-01 -5.53685427e-01
-1.12714207e+00 1.71207905e-01 4.83500630e-01 -1.35571134e+00
-7.48337805e-01 -9.01761651e-01 -2.31564134e-01 9.45271790e-01
-1.06541955e+00 -5.95017254e-01 2.50609875e-01 4.88591641e-01
3.08690727e-01 -5.05484939e-01 7.20706403e-01 -4.95939031e-02
-9.62522745e-01 3.59312326e-01 7.55263984e-01 -6.60671294e-01
9.23641980e-01 -1.24212289e+00 -2.23600790e-01 6.16670787e-01
-5.41187167e-01 6.52385652e-01 1.38885880e+00 -7.79197335e-01
-1.53382230e+00 -1.36254954e+00 5.42617887e-02 -6.20221376e-01
1.11870408e+00 -2.22495824e-01 -7.22828984e-01 1.99983194e-01
-9.68341455e-02 -2.83716023e-01 2.31053665e-01 -8.95783231e-02
1.62129372e-01 -3.41384381e-01 -1.17380011e+00 5.74701130e-01
3.75999734e-02 -1.79915994e-01 -5.12685239e-01 1.46214753e-01
7.08944082e-01 -5.48688889e-01 -1.32092226e+00 8.44049215e-01
3.55467916e-01 -3.86554092e-01 9.96866286e-01 -1.66614935e-01
-4.81227368e-01 -6.70097709e-01 -1.50105983e-01 -1.58167183e+00
-2.02315435e-01 -1.46608710e+00 -2.25350410e-01 1.31297100e+00
6.22624755e-01 -6.27774477e-01 5.06777823e-01 6.39969409e-01
-1.28144011e-01 -5.05603909e-01 -9.80414510e-01 -1.45127308e+00
1.69786170e-01 -5.43719232e-01 3.05654794e-01 4.47537228e-02
-8.50642994e-02 3.27126205e-01 -4.47136402e-01 4.24750090e-01
7.36438990e-01 -4.02655751e-01 8.00790012e-01 -1.06520462e+00
-5.84125042e-01 -4.34451997e-01 -1.36276647e-01 -4.94556352e-02
3.20494324e-02 1.21535640e-02 7.92491972e-01 -7.23565936e-01
-2.75193065e-01 -4.61482584e-01 -3.95967662e-01 -1.71562172e-02
-1.57917678e-01 -1.15325764e-01 -3.01704019e-01 -3.56053710e-01
-1.27720773e-01 7.45297015e-01 7.93363333e-01 -8.91536772e-02
-6.09198987e-01 6.85455620e-01 -1.85805857e-01 6.80555344e-01
9.96016920e-01 -6.36968672e-01 -7.09099293e-01 4.41337913e-01
1.75544366e-01 1.88788652e-01 1.93912670e-01 -1.26022983e+00
9.34384950e-03 -3.27996820e-01 2.35976167e-02 -7.31496096e-01
2.23702654e-01 -1.24837065e+00 6.89722121e-01 7.83746243e-01
-2.03029215e-01 -1.37588188e-01 6.20470405e-01 1.05932343e+00
-1.32714540e-01 -3.32794815e-01 1.08797085e+00 3.23479116e-01
-3.98875207e-01 4.62504327e-02 -6.58321977e-01 -1.17149845e-01
1.20790815e+00 -3.39434319e-03 1.78428262e-01 -2.60408670e-01
-8.39347720e-01 3.15113544e-01 1.46929637e-01 3.54832113e-01
2.42813811e-01 -1.08180070e+00 -4.76698935e-01 8.88913348e-02
9.60953310e-02 -3.51590127e-01 1.24768773e-02 7.26493478e-01
-1.33247823e-01 7.56367981e-01 -1.55273937e-02 -8.03204358e-01
-1.14455032e+00 8.18866849e-01 5.46828330e-01 -2.05907464e-01
2.24249009e-02 5.75971961e-01 -1.74465209e-01 -3.33197296e-01
4.32595521e-01 -4.57942545e-01 6.20131008e-02 -1.61813259e-01
2.81665057e-01 4.28789079e-01 3.33885908e-01 -7.25681186e-02
-3.44120026e-01 6.50509655e-01 4.10442680e-01 -4.14261699e-01
9.06258285e-01 -1.33358181e-01 3.43016595e-01 9.31971312e-01
5.48240840e-01 -3.55013341e-01 -1.54593050e+00 1.89797759e-01
5.20771623e-01 -3.84358406e-01 5.14876604e-01 -7.91525960e-01
-7.97665656e-01 6.00644052e-01 1.04155195e+00 -8.32962431e-03
9.18535411e-01 -6.74233973e-01 7.80241340e-02 6.19810343e-01
5.95841110e-01 -1.43471825e+00 1.78185672e-01 5.09237289e-01
9.35987413e-01 -1.09433270e+00 2.48090461e-01 -2.93395758e-01
-7.11956680e-01 8.99290681e-01 5.63657999e-01 -3.10599297e-01
1.10445106e+00 7.01296747e-01 -1.55993730e-01 2.07537010e-01
-9.02894557e-01 9.39765275e-02 4.81972247e-01 7.40957618e-01
1.27498768e-02 -8.96129832e-02 7.39048496e-02 3.89703512e-01
3.55777889e-01 1.78050965e-01 2.47670516e-01 1.01522171e+00
-5.04789710e-01 -1.13499177e+00 -8.18746865e-01 3.16523761e-01
-3.83947760e-01 3.68514359e-01 1.22016214e-01 1.23096836e+00
-2.66308397e-01 1.33073807e+00 -3.00887495e-01 -2.11906061e-01
7.11324215e-01 8.55719612e-04 3.84586215e-01 -5.83541393e-01
-6.20628476e-01 5.89684010e-01 3.56002927e-01 -7.01286197e-01
7.29760230e-02 -7.34628439e-01 -7.76051879e-01 6.47253916e-02
-8.90854895e-01 -4.06395970e-03 7.47010708e-01 7.68206775e-01
2.21009344e-01 1.07061911e+00 6.74838483e-01 -7.27832377e-01
-1.28229594e+00 -7.02806234e-01 -7.04981327e-01 -4.99573797e-01
2.53233612e-01 -1.20090330e+00 -4.37738240e-01 -7.59168640e-02] | [5.032385349273682, 2.6025192737579346] |
6b76419a-acfa-4a7f-bd1c-80250bb6d9d6 | model-driven-deep-learning-for-non-coherent | 2301.00516 | null | https://arxiv.org/abs/2301.00516v1 | https://arxiv.org/pdf/2301.00516v1.pdf | Model-Driven Deep Learning for Non-Coherent Massive Machine-Type Communications | In this paper, we investigate the joint device activity and data detection in massive machine-type communications (mMTC) with a one-phase non-coherent scheme, where data bits are embedded in the pilot sequences and the base station simultaneously detects active devices and their embedded data bits without explicit channel estimation. Due to the correlated sparsity pattern introduced by the non-coherent transmission scheme, the traditional approximate message passing (AMP) algorithm cannot achieve satisfactory performance. Therefore, we propose a deep learning (DL) modified AMP network (DL-mAMPnet) that enhances the detection performance by effectively exploiting the pilot activity correlation. The DL-mAMPnet is constructed by unfolding the AMP algorithm into a feedforward neural network, which combines the principled mathematical model of the AMP algorithm with the powerful learning capability, thereby benefiting from the advantages of both techniques. Trainable parameters are introduced in the DL-mAMPnet to approximate the correlated sparsity pattern and the large-scale fading coefficient. Moreover, a refinement module is designed to further advance the performance by utilizing the spatial feature caused by the correlated sparsity pattern. Simulation results demonstrate that the proposed DL-mAMPnet can significantly outperform traditional algorithms in terms of the symbol error rate performance. | ['Shen', 'Xuemin', 'Feifei Gao', 'Wen Wu', 'Zhe Ma'] | 2023-01-02 | null | null | null | null | ['type'] | ['speech'] | [ 3.14122379e-01 -3.52083705e-02 -5.07852316e-01 2.92891711e-01
-4.60003316e-01 3.75188380e-01 3.50913852e-01 -3.13488632e-01
-2.70409346e-01 8.73893261e-01 7.22506121e-02 -4.35004294e-01
-1.39474571e-01 -4.35444862e-01 -4.86545116e-01 -1.28771925e+00
-5.24237156e-01 -3.76144111e-01 -6.42967236e-04 5.96636869e-02
-4.36013415e-02 7.70260468e-02 -7.42432415e-01 -1.47869080e-01
9.32187736e-01 1.47689211e+00 6.22350931e-01 7.36102939e-01
1.07118763e-01 8.73000801e-01 -9.26101387e-01 1.11775413e-01
1.90523729e-01 -6.65128648e-01 2.76518971e-01 2.84227282e-01
-3.66916478e-01 -4.70193923e-01 -1.08155739e+00 1.07216942e+00
8.67005646e-01 -5.30564427e-01 4.31414962e-01 -9.58882093e-01
-5.33290952e-02 6.25590146e-01 -7.97379315e-01 1.70279741e-01
7.27857500e-02 -9.16239023e-02 2.66676605e-01 -9.40820634e-01
-1.73453718e-01 8.81380677e-01 1.14565361e+00 -5.12598231e-02
-4.92142856e-01 -1.04618776e+00 -3.84752572e-01 3.32236946e-01
-1.75498176e+00 -3.97507459e-01 8.97213221e-01 1.92838199e-02
6.90012276e-01 1.28670437e-02 7.94438958e-01 7.90244341e-01
3.78860533e-01 1.11639500e+00 8.41202259e-01 -8.21234465e-01
2.38500014e-01 -2.18377709e-02 -1.70377776e-01 9.23688471e-01
6.62408471e-01 5.12576461e-01 -4.48638171e-01 -2.23708332e-01
1.03591990e+00 8.12804252e-02 -4.90771443e-01 -2.09372461e-01
-1.06266570e+00 5.64574182e-01 4.83465910e-01 9.57467377e-01
-4.90856737e-01 2.16230214e-01 1.49193779e-02 3.83133322e-01
-6.49468079e-02 2.05894157e-01 -1.17375590e-01 1.08571626e-01
-1.17938983e+00 -2.76376963e-01 9.05973434e-01 1.20452237e+00
6.22816026e-01 8.26946080e-01 -3.83921355e-01 3.42387170e-01
6.24839306e-01 9.28693950e-01 6.07171118e-01 -5.84219038e-01
4.64407235e-01 2.04321086e-01 8.31192434e-02 -9.12067771e-01
-7.39049435e-01 -1.57555723e+00 -1.83073282e+00 -3.55536342e-01
6.78836480e-02 -8.17375422e-01 -5.84793508e-01 1.47791553e+00
7.22128749e-02 7.27119505e-01 4.39300537e-01 5.77025592e-01
3.33397239e-01 8.11380506e-01 -2.93752462e-01 -8.90721440e-01
9.07787204e-01 -7.44331121e-01 -1.25289595e+00 -2.42224678e-01
8.43499780e-01 -4.51732546e-01 4.17239126e-03 4.83597726e-01
-9.30007100e-01 -7.50990510e-01 -1.94505656e+00 6.99300587e-01
4.74724285e-02 5.40467560e-01 4.37507749e-01 1.01475275e+00
-7.52573967e-01 2.12578207e-01 -5.26127994e-01 1.15872219e-01
5.79265237e-01 5.97767830e-01 4.19110417e-01 -1.02634385e-01
-1.42859530e+00 5.52324176e-01 8.36605370e-01 5.17555892e-01
-4.53729481e-01 -2.93583512e-01 -7.54105210e-01 1.17207803e-01
1.94292530e-01 -5.05181670e-01 1.31814253e+00 -9.73915219e-01
-1.50142097e+00 -1.22171380e-01 -2.63479233e-01 -9.53809440e-01
7.31372312e-02 1.83822304e-01 -9.55424786e-01 2.35657558e-01
-3.13455880e-01 -6.21062703e-02 1.07636821e+00 -8.53284478e-01
-8.74845922e-01 5.54523570e-03 -4.32049960e-01 1.03867054e-01
-6.22381270e-01 -5.52854657e-01 -5.02022266e-01 -9.79443848e-01
2.31523335e-01 -5.32166421e-01 -5.10172427e-01 -2.86231041e-01
-2.33481720e-01 2.06315681e-01 1.22583461e+00 -6.62734389e-01
1.85708725e+00 -2.34722328e+00 -2.66297162e-01 4.72889066e-01
2.13256210e-01 8.37511063e-01 7.91551024e-02 5.85406363e-01
5.02988517e-01 -6.36719525e-01 -1.23061955e-01 -2.64034003e-01
-2.05668822e-01 1.82959750e-01 -1.15520824e-02 6.15330398e-01
-2.36495566e-02 9.22625303e-01 -1.00170398e+00 -3.09134513e-01
2.26888403e-01 2.86803037e-01 -1.31902516e-01 1.34057775e-01
6.52235970e-02 4.77251321e-01 -6.18994236e-01 5.25958478e-01
1.04618514e+00 -5.66775203e-01 5.68394780e-01 -4.31944668e-01
-1.15777545e-01 -3.89226191e-02 -1.15828609e+00 1.49094141e+00
-4.98573899e-01 7.18702793e-01 5.21640182e-01 -1.26260507e+00
7.25297809e-01 7.26265192e-01 3.52805436e-01 -1.28484797e+00
3.39399993e-01 5.62598109e-01 1.26449853e-01 -3.08239311e-01
2.72963673e-01 -1.63511410e-01 -1.17237091e-01 8.54280666e-02
1.22534990e-01 3.50577116e-01 -1.53576151e-01 1.67726845e-01
1.12870848e+00 -5.64561248e-01 7.79122055e-01 -2.01260939e-01
8.55784774e-01 -5.08283138e-01 7.59856820e-01 1.19058430e+00
6.05856031e-02 -3.14683974e-01 -1.16804399e-01 -1.13943420e-01
-8.07461739e-01 -8.92119408e-01 -2.49247208e-01 3.52689147e-01
4.18700039e-01 -1.68971166e-01 -3.25840801e-01 -3.10062081e-01
-2.71017462e-01 4.40906316e-01 5.17356992e-02 -4.23181146e-01
-6.07239366e-01 -1.13635504e+00 7.07978725e-01 5.02813637e-01
1.39060557e+00 -3.57987374e-01 -1.75171182e-01 6.81017458e-01
1.63442478e-01 -1.23932838e+00 -1.84832349e-01 5.22582233e-01
-5.66242695e-01 -5.17366588e-01 -7.34692037e-01 -1.12622738e+00
5.88910520e-01 6.30464494e-01 4.37880188e-01 2.13731483e-01
2.78357267e-01 1.21966958e-01 -4.61484581e-01 -5.76151967e-01
-4.07026768e-01 -4.86757644e-02 3.00154865e-01 1.93624943e-01
4.25307244e-01 -5.11747122e-01 -4.50338006e-01 2.02589899e-01
-6.60107434e-01 8.36793557e-02 1.43617761e+00 1.05472910e+00
-4.60481830e-02 7.07283497e-01 9.34251904e-01 -3.41907263e-01
3.95977497e-01 -5.53152680e-01 -5.93891740e-01 -8.25549569e-03
-3.29067528e-01 -4.77450192e-02 6.71745241e-01 -5.35791039e-01
-8.73234928e-01 3.49022634e-02 -5.13107002e-01 -1.03200264e-01
4.54783410e-01 6.14611566e-01 -4.15767819e-01 -7.25706875e-01
3.84885669e-01 6.89513147e-01 -2.77452677e-01 -3.76539491e-02
-9.01857987e-02 1.19211948e+00 6.49797201e-01 1.25279412e-01
1.26031244e+00 3.68757904e-01 2.62270153e-01 -1.24912965e+00
-7.75473893e-01 -4.85506684e-01 -3.68669271e-01 -1.78209152e-02
2.24047273e-01 -1.46308672e+00 -4.89609897e-01 8.19458008e-01
-1.11984718e+00 -1.99196845e-01 7.57100135e-02 9.87781644e-01
-2.30649531e-01 8.31114888e-01 -7.35327244e-01 -1.00936985e+00
-3.60601008e-01 -5.85995555e-01 7.79886127e-01 2.92710602e-01
2.34397009e-01 -1.10500491e+00 -2.65680403e-01 -3.38683873e-01
7.17014849e-01 -1.54541498e-02 7.45374143e-01 -3.53827208e-01
-7.29788005e-01 -4.98199254e-01 -1.90474302e-01 2.59252250e-01
1.21616431e-01 -1.05930090e+00 -8.77411127e-01 -7.83748806e-01
5.70306718e-01 4.56182361e-02 5.54658651e-01 6.46481037e-01
9.97232795e-01 -4.81328398e-01 -7.05581367e-01 6.64683759e-01
1.40486872e+00 4.57434803e-01 8.41246545e-01 1.03544192e-02
4.28077996e-01 -4.17740464e-01 4.46508050e-01 7.63816893e-01
1.24466583e-01 6.63820863e-01 1.24648824e-01 -3.47631603e-01
-3.31729688e-02 -1.42906994e-01 4.21252310e-01 1.31273639e+00
5.34821033e-01 -4.01947021e-01 -2.32391447e-01 1.03384577e-01
-1.75008929e+00 -8.56356144e-01 -3.26342046e-01 2.05883741e+00
5.88064015e-01 4.29813683e-01 -2.73831338e-01 6.70494437e-01
5.40851951e-01 2.46611178e-01 -4.71747458e-01 1.81106240e-01
-4.82977808e-01 -3.03661674e-02 8.15683067e-01 2.73912460e-01
-1.03713751e+00 1.58027366e-01 5.72978830e+00 1.47906435e+00
-1.09635377e+00 1.61325723e-01 1.99863002e-01 4.47250515e-01
2.08471462e-01 -4.49610680e-01 -8.55079889e-01 7.28938162e-01
1.11052120e+00 5.76283894e-02 -1.11360019e-02 3.61968607e-01
3.68639648e-01 -1.46462947e-01 -7.02480495e-01 1.33376515e+00
9.47302207e-02 -1.35963678e+00 -2.81007648e-01 2.85232782e-01
8.18976283e-01 -2.19608337e-01 2.00117543e-01 6.85700774e-01
-7.37139657e-02 -6.13893867e-01 3.96241695e-01 5.89830279e-01
7.41828740e-01 -5.42469442e-01 1.13539422e+00 8.44252288e-01
-1.51518106e+00 -8.00945997e-01 -2.27539212e-01 -3.68101627e-01
1.98306844e-01 1.20000136e+00 -9.92502928e-01 9.81036007e-01
8.33991077e-03 6.80963695e-01 -5.15803173e-02 1.50628066e+00
-1.59537926e-01 8.96428943e-01 -4.60067123e-01 -3.11053455e-01
4.37600076e-01 1.38142377e-01 6.27951145e-01 1.34130740e+00
8.92317891e-01 5.99866882e-02 3.16058636e-01 3.40497941e-01
-2.06019692e-02 -3.89765590e-01 -4.93346781e-01 2.04282075e-01
9.45297897e-01 9.42610860e-01 -1.89167440e-01 -3.94363105e-01
-7.16413498e-01 7.09785938e-01 -3.03151608e-01 5.19703746e-01
-4.99820590e-01 -5.06980062e-01 -1.24414936e-01 -1.09623425e-01
8.05927753e-01 -6.77961349e-01 -2.59767354e-01 -8.13725293e-01
-3.37071382e-02 -8.30341160e-01 -1.56935126e-01 -3.40055674e-01
-1.02135909e+00 1.32605508e-01 -5.30200660e-01 -1.77932203e+00
-3.48621637e-01 -5.21242440e-01 -6.78900421e-01 7.23355412e-01
-1.69591141e+00 -9.42921937e-01 -1.51426747e-01 6.01639450e-01
2.50881642e-01 -4.24511164e-01 7.87862659e-01 6.11886024e-01
-6.58775210e-01 8.58199656e-01 7.42993295e-01 1.32165357e-01
-5.61962239e-02 -8.18061531e-01 -5.32434918e-02 1.00502908e+00
-1.74488485e-01 3.31217289e-01 5.63514173e-01 -6.81360364e-01
-1.62765431e+00 -1.22446394e+00 6.17593169e-01 4.41590369e-01
6.46887183e-01 -4.36223298e-01 -5.84783792e-01 1.79827347e-01
3.69772524e-01 -7.55122826e-02 7.91415274e-01 -4.52091843e-01
3.38046491e-01 -2.60351270e-01 -1.00284684e+00 4.07263130e-01
6.87095106e-01 -3.77377063e-01 -3.55753124e-01 2.99293816e-01
4.27491009e-01 -3.54466408e-01 -5.82790613e-01 6.21008635e-01
3.55265796e-01 -4.82113004e-01 7.03716815e-01 4.73337322e-01
-5.49230337e-01 -5.38295209e-01 -4.05716151e-01 -1.12778151e+00
-5.68418920e-01 -9.73420024e-01 -1.11606419e+00 8.48314047e-01
2.13296920e-01 -4.08490002e-01 9.45371091e-01 -5.93052626e-01
-2.57391751e-01 -7.85328627e-01 -1.21158206e+00 -8.40220928e-01
-4.27853256e-01 -3.68411094e-01 3.36028427e-01 5.68287194e-01
1.64672390e-01 4.83829468e-01 -9.53645766e-01 7.54800379e-01
7.97333241e-01 -4.18945909e-01 5.46167791e-01 -1.09379375e+00
-9.19365585e-01 -9.39738378e-02 -3.64637792e-01 -2.12678480e+00
-3.98578942e-01 -5.35565674e-01 6.12080805e-02 -1.10411084e+00
-2.02360824e-01 -4.54315990e-01 -4.82520074e-01 -1.95909530e-01
-1.63776085e-01 1.05682850e-01 -1.47279248e-01 5.28800525e-02
-9.40750659e-01 7.99945712e-01 9.86581504e-01 -1.86679110e-01
-1.20777868e-01 5.07819712e-01 -5.49386919e-01 6.99931204e-01
8.37793529e-01 -1.03621826e-01 -2.21899614e-01 -2.65114605e-01
-6.69291466e-02 4.27022189e-01 1.47020221e-01 -1.96984267e+00
6.63807690e-01 6.81677818e-01 8.15679550e-01 -9.16815042e-01
4.87716168e-01 -1.16809368e+00 -1.20755881e-01 1.16798627e+00
3.27634662e-01 -5.19282997e-01 1.44786909e-01 1.00617146e+00
-1.48711607e-01 -9.45873782e-02 6.11310840e-01 3.69042367e-01
-7.48528242e-01 1.47166654e-01 -8.67785335e-01 -6.69564843e-01
9.50467229e-01 -4.82634366e-01 -6.59554601e-02 -4.98341948e-01
-2.45561138e-01 4.05576795e-01 -3.38048637e-01 -2.12256759e-01
6.76647305e-01 -1.53221381e+00 -3.46752435e-01 6.31464541e-01
-2.33706012e-01 -3.35335791e-01 2.76576519e-01 1.21279597e+00
-1.30267139e-03 8.81111860e-01 1.99972749e-01 -7.44724691e-01
-7.22554922e-01 4.06823993e-01 4.92286712e-01 -6.89566374e-01
-4.02730525e-01 5.25897920e-01 -1.20063595e-01 7.68966898e-02
4.91788387e-01 -9.85264629e-02 -1.01004615e-01 -5.87973058e-01
7.90294051e-01 2.07608372e-01 -6.80187494e-02 -3.74394953e-01
1.50470078e-01 5.45805395e-01 -9.05837268e-02 2.48967290e-01
7.84446895e-01 -3.97125632e-01 2.39932537e-01 6.59383088e-02
1.16242480e+00 9.04505327e-02 -1.26696312e+00 -8.14861953e-01
-6.52541546e-03 -1.19403839e-01 3.68865401e-01 -4.94091064e-01
-9.00802314e-01 5.64038932e-01 1.01235914e+00 2.86790133e-01
1.46347928e+00 -5.29040277e-01 1.04988313e+00 6.92601502e-01
9.08866286e-01 -9.14608955e-01 3.09674948e-01 5.13402522e-01
1.71819270e-01 -7.18557119e-01 -1.36348689e-02 -2.49387205e-01
1.69926882e-01 1.08507299e+00 2.37749517e-01 1.14316270e-01
8.36805880e-01 7.31107116e-01 -3.13798487e-01 5.21842241e-02
-2.79500544e-01 -7.17678368e-02 -1.42539725e-01 8.90213192e-01
-1.48706973e-01 -9.93880779e-02 -4.20801997e-01 7.25023985e-01
6.52206913e-02 1.98878765e-01 2.93231636e-01 1.02712488e+00
-9.82531965e-01 -9.47103560e-01 -6.25661552e-01 7.04017997e-01
-2.61856407e-01 -7.39511997e-02 1.68166533e-01 5.42526305e-01
3.64935845e-01 1.12113512e+00 -1.22747980e-01 -7.66967773e-01
7.79946595e-02 -6.20382726e-01 2.45424077e-01 -8.92079696e-02
3.30964699e-02 3.64885390e-01 2.13032439e-02 -1.63370240e-02
-3.05448264e-01 -3.38099152e-01 -1.18644965e+00 -1.29353866e-01
-6.27370775e-01 3.43424857e-01 4.26890790e-01 1.27808404e+00
1.98340297e-01 8.67153108e-01 1.08396888e+00 -8.79949033e-01
-5.67824364e-01 -1.02433217e+00 -9.25646842e-01 -5.78453422e-01
7.66009808e-01 -4.94627357e-01 -3.57744962e-01 -5.29801548e-01] | [6.365095138549805, 1.451950192451477] |
29cd2b52-8344-46ba-b6ca-5aa6d50ea398 | the-future-is-different-large-pre-trained | 2211.00384 | null | https://arxiv.org/abs/2211.00384v2 | https://arxiv.org/pdf/2211.00384v2.pdf | The future is different: Large pre-trained language models fail in prediction tasks | Large pre-trained language models (LPLM) have shown spectacular success when fine-tuned on downstream supervised tasks. Yet, it is known that their performance can drastically drop when there is a distribution shift between the data used during training and that used at inference time. In this paper we focus on data distributions that naturally change over time and introduce four new REDDIT datasets, namely the WALLSTREETBETS, ASKSCIENCE, THE DONALD, and POLITICS sub-reddits. First, we empirically demonstrate that LPLM can display average performance drops of about 88% (in the best case!) when predicting the popularity of future posts from sub-reddits whose topic distribution changes with time. We then introduce a simple methodology that leverages neural variational dynamic topic models and attention mechanisms to infer temporal language model representations for regression tasks. Our models display performance drops of only about 40% in the worst cases (2% in the best ones) when predicting the popularity of future posts, while using only about 7% of the total number of parameters of LPLM and providing interpretable representations that offer insight into real-world events, like the GameStop short squeeze of 2021 | ['César Ojeda', 'Ramsés J. Sánchez', 'Kostadin Cvejoski'] | 2022-11-01 | null | null | null | null | ['topic-models'] | ['natural-language-processing'] | [-5.35398602e-01 2.66789436e-01 -5.41498482e-01 -8.51636082e-02
-9.40722525e-01 -4.50282454e-01 1.05917346e+00 2.71926135e-01
-4.74843502e-01 7.73294330e-01 3.40357929e-01 -5.99110484e-01
-8.26230347e-02 -7.99586773e-01 -8.44414055e-01 -5.57877660e-01
-2.63597488e-01 1.01919365e+00 2.92444438e-01 -3.60717595e-01
2.32402701e-02 -1.16605014e-01 -1.17893589e+00 8.44454393e-02
7.00810671e-01 7.24357367e-01 2.71574229e-01 6.65596128e-01
-4.50341403e-01 8.81029427e-01 -6.53921664e-01 -4.21859622e-01
-9.36226845e-02 6.61663199e-03 -7.25512743e-01 -2.76573390e-01
1.35529950e-01 -2.44852409e-01 -8.00403595e-01 5.51749885e-01
1.92705274e-01 2.50731409e-01 8.88518095e-01 -1.04174113e+00
-6.65826380e-01 1.09719563e+00 -6.98176861e-01 6.86292827e-01
-2.43524507e-01 1.08314872e-01 1.23852873e+00 -4.68220413e-01
6.95656538e-01 1.28982735e+00 5.07848024e-01 3.85854304e-01
-1.33349741e+00 -7.89941072e-01 7.78642476e-01 1.26101583e-01
-1.02420616e+00 -1.41330108e-01 4.58194017e-01 -6.70236886e-01
9.80828702e-01 -3.74134933e-03 4.83800918e-01 1.58709991e+00
4.21792805e-01 8.22679222e-01 6.90190613e-01 1.44491404e-01
5.53866476e-02 1.64015591e-01 3.06658566e-01 2.50941247e-01
-7.73667917e-02 -3.05108845e-01 -6.91528559e-01 -2.93307722e-01
3.63623321e-01 5.84052829e-03 1.91450007e-02 4.18652743e-02
-9.48569596e-01 9.95608985e-01 2.58580297e-01 4.62381572e-01
-2.29304075e-01 5.74953437e-01 4.96663630e-01 3.33875299e-01
1.21218896e+00 3.01497668e-01 -7.95483708e-01 -5.72055399e-01
-1.29692948e+00 4.50729012e-01 7.97676921e-01 6.51560783e-01
4.52045798e-01 4.29440849e-02 -2.36552358e-01 9.09660578e-01
1.10304035e-01 3.41891021e-01 8.23912144e-01 -7.01196194e-01
5.71369469e-01 5.52054606e-02 3.68363380e-01 -9.86092329e-01
-4.34636831e-01 -5.73904514e-01 -4.46705401e-01 -6.98197722e-01
5.89955688e-01 -2.72301793e-01 -8.83983672e-01 1.92845488e+00
-6.78587928e-02 2.54317701e-01 -4.15509224e-01 3.06984246e-01
5.76996922e-01 1.20314705e+00 4.59787011e-01 -3.18398565e-01
9.98524368e-01 -6.57994807e-01 -4.59938347e-01 -3.83244812e-01
6.62653863e-01 -4.58799273e-01 1.15477872e+00 4.01424140e-01
-1.04132533e+00 -3.64198536e-01 -6.03140414e-01 -2.15491235e-01
-5.34670651e-01 -3.12315285e-01 6.81626737e-01 1.77617148e-01
-9.60323453e-01 6.89579427e-01 -9.98320937e-01 -4.93262053e-01
3.49015683e-01 2.52450883e-01 2.51784474e-01 2.06075072e-01
-1.14475584e+00 7.62988091e-01 1.27720879e-02 -2.75778741e-01
-1.05155396e+00 -1.14355350e+00 -3.46977890e-01 3.88938308e-01
3.03393632e-01 -4.89123166e-01 1.45398784e+00 -5.97121239e-01
-1.22546327e+00 8.23407412e-01 -3.43351930e-01 -7.78901875e-01
6.56882882e-01 -3.90527636e-01 -4.10599351e-01 -4.69896555e-01
1.36610106e-01 4.38900441e-01 7.30679750e-01 -8.61142159e-01
-7.05022812e-01 -1.62798226e-01 9.29978862e-02 -2.24400088e-01
-5.69936275e-01 -2.39612505e-01 -5.65116942e-01 -4.76016223e-01
-3.09522837e-01 -8.78846943e-01 -2.52843648e-01 -4.03363675e-01
-4.48579937e-01 -6.52773738e-01 7.14344442e-01 -7.10743785e-01
1.57946575e+00 -2.09038806e+00 1.07805468e-01 -9.99583974e-02
2.56456614e-01 -1.45792276e-01 -1.91874176e-01 5.75225472e-01
1.97921142e-01 3.91442269e-01 2.23716706e-01 -6.35677814e-01
1.65677994e-01 2.77517319e-01 -9.45087373e-01 2.73036033e-01
-2.75861491e-02 8.20793033e-01 -9.27460730e-01 -8.29100087e-02
-7.66340569e-02 2.52412915e-01 -7.05373764e-01 -1.41119942e-01
-8.13541353e-01 2.96522290e-01 -4.42899704e-01 6.70871958e-02
2.98660398e-01 -6.42363012e-01 3.10384840e-01 2.70442039e-01
-1.41414031e-01 7.11473048e-01 -4.38468218e-01 1.31381905e+00
-7.32335508e-01 1.03246224e+00 -6.58030882e-02 -9.31116402e-01
6.06673241e-01 2.43326947e-02 6.57749057e-01 -7.17093229e-01
3.08865830e-02 6.50379434e-02 1.73963793e-02 -3.03380817e-01
7.97394395e-01 -8.52619857e-02 -2.67447978e-01 6.14939272e-01
-6.66503459e-02 6.33998662e-02 1.89642236e-01 4.37676996e-01
9.45329487e-01 -1.18013129e-01 -1.90317839e-01 -3.49825561e-01
-1.54069588e-01 -1.19278349e-01 4.31792527e-01 8.86139154e-01
6.97423667e-02 4.27722663e-01 1.16757405e+00 -5.54837584e-01
-1.27469599e+00 -1.08702171e+00 -1.21420786e-01 1.68673825e+00
-1.72091022e-01 -6.94446087e-01 -3.96473110e-01 -5.31401575e-01
1.51784211e-01 1.14972568e+00 -7.57143080e-01 -1.38429582e-01
-5.29770613e-01 -1.19397020e+00 1.25625104e-01 2.20234796e-01
9.73133147e-02 -6.68911994e-01 -1.09272487e-01 2.82913625e-01
-2.08888933e-01 -1.07392395e+00 -2.99095482e-01 7.82952830e-02
-8.45113575e-01 -7.42955387e-01 -7.36606836e-01 -3.48091662e-01
1.84600502e-01 -3.08596827e-02 1.40148044e+00 -2.91460603e-01
9.85818431e-02 3.13772559e-01 2.45043524e-02 -5.09776056e-01
-2.41345614e-01 6.50677383e-01 4.25581783e-02 -8.28889534e-02
2.81978011e-01 -1.00684893e+00 -6.53714657e-01 -1.08408416e-02
-5.44042826e-01 -1.49952143e-01 1.65937081e-01 6.24410152e-01
2.03904331e-01 -5.45624383e-02 4.49384391e-01 -1.15664184e+00
6.22146547e-01 -1.36202598e+00 -4.61474150e-01 -2.68189833e-02
-7.48179317e-01 1.32587388e-01 5.12189865e-01 -8.29986870e-01
-9.21213984e-01 -8.03929865e-01 1.26090854e-01 -3.34575236e-01
3.78706574e-01 6.60741329e-01 6.19701982e-01 9.40037370e-01
5.33623159e-01 2.27111816e-01 -3.46474707e-01 -8.37836027e-01
4.35421258e-01 5.69709420e-01 3.50449592e-01 -6.61386430e-01
7.13097692e-01 5.66899359e-01 -3.92283022e-01 -1.02803123e+00
-1.06072378e+00 -3.01158279e-01 -1.42041266e-01 4.49884348e-02
5.07009029e-01 -1.08422101e+00 -5.90845585e-01 2.88068742e-01
-9.50046897e-01 -8.79652858e-01 -2.72154540e-01 2.48537332e-01
-4.01072383e-01 1.58452000e-02 -6.66524589e-01 -9.12954628e-01
-2.17323601e-01 -7.15500116e-01 9.94557977e-01 1.93776526e-02
-4.39296544e-01 -1.29043281e+00 1.04030654e-01 7.62505904e-02
6.53080344e-01 4.80535720e-03 1.52309227e+00 -7.16476083e-01
-4.86654818e-01 -4.26829457e-02 -1.02037929e-01 -8.38856101e-02
-1.89751133e-01 3.92399222e-01 -7.43939459e-01 -2.72687554e-01
-4.95852113e-01 -3.22255790e-02 1.19232178e+00 8.73985887e-01
1.47640455e+00 -3.83610725e-01 -6.74006820e-01 3.61994505e-01
1.02126956e+00 -1.66513219e-01 3.41606170e-01 2.17780843e-01
2.61582762e-01 4.15933192e-01 1.69832975e-01 6.36563003e-01
7.52669275e-01 6.12967730e-01 3.20087343e-01 2.54186690e-01
9.44928601e-02 -7.01667905e-01 5.43884277e-01 8.39398801e-01
4.12515737e-02 -5.78586936e-01 -9.27901924e-01 9.01257753e-01
-1.89158094e+00 -8.38921368e-01 -1.14067726e-01 2.14876390e+00
8.02814543e-01 5.12997210e-01 3.38955462e-01 -5.09975910e-01
3.46175462e-01 6.63362622e-01 -7.61534572e-01 -5.25906086e-01
4.30034772e-02 1.50374532e-01 5.79261005e-01 5.83645999e-01
-8.73448491e-01 1.08043623e+00 6.97260714e+00 9.40707445e-01
-1.29462683e+00 4.13343459e-01 1.06038618e+00 -5.79385221e-01
-5.41082621e-01 -2.72550974e-02 -8.31352293e-01 8.10092211e-01
1.58647490e+00 -5.19024193e-01 2.03480378e-01 8.25568080e-01
6.27806544e-01 -1.25457808e-01 -9.17323709e-01 8.84830475e-01
-2.51550436e-01 -1.39606750e+00 -4.96280193e-02 3.70192349e-01
8.53078842e-01 6.70277119e-01 6.57993019e-01 8.57842028e-01
7.13389874e-01 -9.33271348e-01 6.52616739e-01 6.34356260e-01
6.19022608e-01 -6.13685369e-01 3.29047441e-01 8.29233766e-01
-6.54268563e-01 -3.61791968e-01 -4.82013345e-01 -1.61496922e-01
2.77843297e-01 7.90127695e-01 -8.40650141e-01 1.48607343e-02
7.08941936e-01 9.16961491e-01 -3.66533607e-01 6.16239488e-01
1.51036516e-01 1.11721206e+00 -5.44223309e-01 -1.35740966e-01
4.22819704e-01 6.31752461e-02 5.73615849e-01 1.04916131e+00
4.71827537e-01 -2.00704902e-01 -1.29454777e-01 9.16548193e-01
-4.32319492e-01 -8.67907330e-02 -4.05635208e-01 -4.16647702e-01
3.23770225e-01 8.68425071e-01 -3.44703913e-01 -2.68670082e-01
-2.43958533e-01 5.99552274e-01 4.81996983e-01 6.14188015e-01
-1.00181472e+00 3.19293320e-01 8.70779276e-01 6.62912250e-01
3.89464945e-01 -4.59746689e-01 -2.31940132e-02 -1.24060261e+00
-4.04061586e-01 -3.53213668e-01 3.30931872e-01 -6.39615119e-01
-1.45824814e+00 3.84780407e-01 1.66528627e-01 -5.44139326e-01
-5.20953476e-01 -4.61941361e-01 -8.44898760e-01 6.68959200e-01
-1.36288154e+00 -9.66862857e-01 2.12291881e-01 1.27650589e-01
9.58323240e-01 -8.01639035e-02 4.23717469e-01 2.36817941e-01
-6.96317136e-01 4.14157480e-01 6.24077916e-01 -5.98347336e-02
4.20809418e-01 -1.07754958e+00 6.81929052e-01 3.66094261e-01
3.33434671e-01 5.30059993e-01 1.04684889e+00 -4.30145204e-01
-8.58081222e-01 -9.86342609e-01 1.20948148e+00 -6.23441458e-01
1.06729734e+00 -5.95788956e-01 -7.94138670e-01 1.01078057e+00
5.47024533e-02 -3.69816124e-01 4.94064003e-01 7.05796242e-01
-3.25918764e-01 -4.21322696e-02 -4.30619687e-01 8.08325648e-01
8.18185329e-01 -4.98337686e-01 -2.14762196e-01 6.88176572e-01
8.93602431e-01 -2.27817178e-01 -6.75084054e-01 5.67817017e-02
5.14285505e-01 -7.05044866e-01 7.51870930e-01 -1.07389605e+00
6.33369803e-01 4.50718045e-01 -4.83461022e-02 -1.23426080e+00
-2.12968126e-01 -9.53679144e-01 -5.46070158e-01 1.18796527e+00
7.23474503e-01 -5.13353050e-01 8.64700615e-01 6.57832384e-01
2.83807725e-01 -8.13346028e-01 -1.00667608e+00 -6.01618111e-01
6.96346402e-01 -6.24372423e-01 1.53139979e-01 7.44333863e-01
-1.23390354e-01 3.75374049e-01 -5.40837705e-01 -2.50562698e-01
2.02040344e-01 1.78074390e-02 8.08628201e-01 -1.27248681e+00
-3.55802447e-01 -6.35856986e-01 9.73552391e-02 -1.55960858e+00
2.74068505e-01 -7.30255008e-01 -2.14606956e-01 -1.28045094e+00
4.15190905e-01 -5.71195662e-01 -3.14650029e-01 1.45226449e-01
4.39015217e-02 -1.68992728e-01 2.00306371e-01 3.40997279e-01
-5.76090932e-01 6.63003683e-01 8.74275446e-01 -2.71015555e-01
-1.97635815e-01 3.27794939e-01 -8.01797271e-01 7.00180769e-01
5.82124293e-01 -6.31503999e-01 -3.90738338e-01 -6.20275915e-01
6.50083303e-01 -6.75036386e-02 1.47985190e-01 -6.40622854e-01
-4.30350788e-02 -2.64356583e-01 3.88395451e-02 -5.52379668e-01
5.63062191e-01 -3.11821342e-01 2.88631190e-02 2.99531698e-01
-7.37227321e-01 7.22244754e-02 1.65178433e-01 1.05855024e+00
8.06489661e-02 1.75769851e-01 5.42629540e-01 -9.81018916e-02
-3.72913152e-01 5.45483470e-01 -6.78930998e-01 3.91908467e-01
7.49143898e-01 3.50611895e-01 -2.71224648e-01 -1.03959310e+00
-9.62219596e-01 3.18352222e-01 -8.63639917e-03 7.68458784e-01
-5.25466576e-02 -8.07956636e-01 -6.71471417e-01 -2.94398278e-01
-2.57301599e-01 -1.19504333e-02 3.87401611e-01 9.39022660e-01
-1.53401926e-01 5.63544631e-01 4.32853729e-01 -5.18122077e-01
-6.95773542e-01 3.81943315e-01 2.54060805e-01 -8.01868260e-01
-5.07458329e-01 9.95177805e-01 4.53995854e-01 -1.53129265e-01
1.21966526e-01 -6.37872458e-01 -1.13571532e-01 4.19174522e-01
1.75891310e-01 2.60815293e-01 -2.37871483e-01 -3.71353716e-01
2.23233942e-02 2.04790771e-01 -3.58691990e-01 -4.49922569e-02
1.57749617e+00 -2.28520185e-01 1.68265060e-01 1.01500380e+00
1.21183550e+00 5.24866544e-02 -1.24968612e+00 -4.10839230e-01
1.03621386e-01 -1.01624586e-01 1.09029360e-01 -4.74711388e-01
-8.18710983e-01 1.10315728e+00 1.73451528e-01 5.95894516e-01
4.44575459e-01 4.66490746e-01 1.12070239e+00 6.67497739e-02
2.17677578e-01 -1.13650656e+00 6.88333809e-02 6.51572406e-01
7.23968506e-01 -1.05270803e+00 -5.59072122e-02 1.42416790e-01
-5.08637369e-01 9.09036040e-01 3.57003421e-01 -2.54478335e-01
1.07022488e+00 -1.54769585e-01 -4.50847179e-01 -1.66382208e-01
-1.45472634e+00 6.61435872e-02 1.34465039e-01 2.49295115e-01
3.28405052e-01 1.25745639e-01 -2.31027737e-01 8.14510882e-01
-5.24458528e-01 -4.77705389e-01 4.28587824e-01 2.49001339e-01
-5.71654141e-01 -8.70385528e-01 2.26780236e-01 7.46852398e-01
-7.61713803e-01 -2.19146565e-01 -4.44204099e-02 9.28677261e-01
-9.40838307e-02 6.37708485e-01 5.85622549e-01 -2.47610763e-01
-3.56250294e-02 3.19287568e-01 -4.63687070e-02 -6.42906427e-01
-3.13729554e-01 8.17025751e-02 1.41616747e-01 -3.84171665e-01
6.30117487e-03 -6.86999798e-01 -9.16337788e-01 -6.60471559e-01
4.07061167e-02 9.65746343e-02 5.00765860e-01 9.68627691e-01
4.70778584e-01 4.85200793e-01 4.02496785e-01 -7.25944459e-01
-5.66369712e-01 -1.18134034e+00 -5.39375663e-01 2.59902418e-01
4.01568204e-01 -5.88890135e-01 -5.24415314e-01 -1.50109008e-01] | [10.383988380432129, 7.009603500366211] |
9cb2ebc0-0edc-4b83-aa4f-8b6d11245f61 | incomplete-multi-view-multi-label-learning | 2303.07180 | null | https://arxiv.org/abs/2303.07180v1 | https://arxiv.org/pdf/2303.07180v1.pdf | Incomplete Multi-View Multi-Label Learning via Label-Guided Masked View- and Category-Aware Transformers | As we all know, multi-view data is more expressive than single-view data and multi-label annotation enjoys richer supervision information than single-label, which makes multi-view multi-label learning widely applicable for various pattern recognition tasks. In this complex representation learning problem, three main challenges can be characterized as follows: i) How to learn consistent representations of samples across all views? ii) How to exploit and utilize category correlations of multi-label to guide inference? iii) How to avoid the negative impact resulting from the incompleteness of views or labels? To cope with these problems, we propose a general multi-view multi-label learning framework named label-guided masked view- and category-aware transformers in this paper. First, we design two transformer-style based modules for cross-view features aggregation and multi-label classification, respectively. The former aggregates information from different views in the process of extracting view-specific features, and the latter learns subcategory embedding to improve classification performance. Second, considering the imbalance of expressive power among views, an adaptively weighted view fusion module is proposed to obtain view-consistent embedding features. Third, we impose a label manifold constraint in sample-level representation learning to maximize the utilization of supervised information. Last but not least, all the modules are designed under the premise of incomplete views and labels, which makes our method adaptable to arbitrary multi-view and multi-label data. Extensive experiments on five datasets confirm that our method has clear advantages over other state-of-the-art methods. | ['Yong Xu', 'Xiaoling Luo', 'Jie Wen', 'Chengliang Liu'] | 2023-03-13 | null | null | null | null | ['multi-label-learning'] | ['methodology'] | [ 2.12022647e-01 -1.86992869e-01 -6.11705482e-01 -5.79705417e-01
-7.10442901e-01 -6.48344755e-01 4.33794171e-01 -1.65088512e-02
7.70830065e-02 2.46787772e-01 2.78804183e-01 3.14979970e-01
-3.25087219e-01 -6.05588973e-01 -2.10771278e-01 -1.06019187e+00
4.88972634e-01 2.47647986e-01 -2.92611551e-02 4.26154323e-02
-5.27342036e-02 2.45016277e-01 -1.73396826e+00 5.93108058e-01
6.69801593e-01 1.08320642e+00 3.02307811e-02 -3.26238200e-02
-1.26547024e-01 8.10185790e-01 -1.12213857e-01 -3.70225072e-01
1.80144519e-01 -4.01224554e-01 -5.47036052e-01 8.31696153e-01
4.32460159e-01 -6.01973571e-02 -5.63756004e-02 1.14862752e+00
4.23477679e-01 -5.80672733e-02 7.81472445e-01 -1.23322475e+00
-6.09295011e-01 1.99817032e-01 -8.31272840e-01 -8.64621624e-02
1.63689762e-01 -6.94246292e-02 1.28039420e+00 -1.17893946e+00
5.05238414e-01 1.28047287e+00 4.17687595e-01 4.74655658e-01
-1.29838610e+00 -6.99235201e-01 6.89175487e-01 1.48657128e-01
-1.37876117e+00 -3.48836690e-01 9.84312415e-01 -6.42437398e-01
1.93155184e-01 2.17902943e-01 4.49214071e-01 1.04885983e+00
1.91550404e-02 7.69715607e-01 1.54310238e+00 -1.77531496e-01
-6.30313531e-02 5.23900449e-01 2.34138831e-01 9.92645085e-01
1.86094537e-01 -2.09598497e-01 -5.84686041e-01 -3.28424722e-01
3.96749645e-01 6.53831303e-01 -3.52854252e-01 -9.21433747e-01
-1.46113729e+00 9.47100461e-01 3.28637421e-01 2.55845517e-01
-1.92611024e-01 -4.91806775e-01 7.08192945e-01 2.72531092e-01
6.82612479e-01 8.21968988e-02 -4.85734463e-01 5.98324180e-01
-5.43394864e-01 -1.83305711e-01 3.68818432e-01 1.05916953e+00
1.08094656e+00 -1.40131503e-01 1.48254365e-01 1.21765053e+00
3.70327711e-01 4.01145577e-01 5.87026358e-01 -6.84361219e-01
6.63607240e-01 1.10064507e+00 -3.66486073e-01 -9.37820613e-01
-4.79553282e-01 -4.58046049e-01 -1.25483835e+00 -4.78041880e-02
1.82948649e-01 8.57915059e-02 -4.89454418e-01 1.68860674e+00
5.41560292e-01 5.35389818e-02 7.86598399e-02 7.49107003e-01
1.00515962e+00 2.88864911e-01 -3.48392278e-02 -6.55829430e-01
1.73122501e+00 -9.60465550e-01 -6.37126505e-01 -6.30127415e-02
7.22631931e-01 -4.61907625e-01 7.92082369e-01 1.90380916e-01
-5.31138957e-01 -6.00533247e-01 -1.15614879e+00 9.35239196e-02
-4.33761090e-01 3.37529719e-01 6.38113678e-01 4.95192945e-01
-3.55876446e-01 1.80267289e-01 -3.69710207e-01 -2.02689648e-01
4.46036458e-01 2.17760056e-01 -8.31746280e-01 -4.48712200e-01
-8.69336903e-01 6.42994881e-01 4.73076224e-01 -5.50943427e-02
-6.31560862e-01 -6.08058691e-01 -9.67476845e-01 -1.68013796e-02
7.13876426e-01 -6.91207886e-01 7.21128583e-01 -8.74171078e-01
-1.09977114e+00 1.05695677e+00 -3.01320940e-01 4.31460679e-01
1.27730221e-01 1.82807952e-01 -7.00652540e-01 2.03596473e-01
3.50696921e-01 2.83012301e-01 7.99874306e-01 -1.53981137e+00
-8.09894025e-01 -8.51700723e-01 1.84413776e-01 4.88496155e-01
-4.27838326e-01 -3.13392699e-01 -2.65793145e-01 -5.22201121e-01
6.41425431e-01 -9.98628855e-01 3.63968238e-02 -9.79991537e-03
-2.76711345e-01 -4.05442327e-01 7.93865144e-01 -3.39632571e-01
9.67362642e-01 -2.20282292e+00 5.70893049e-01 1.84797850e-02
3.80995482e-01 -9.94228274e-02 1.49214864e-02 3.13954741e-01
-3.31464708e-01 -1.18836239e-01 -1.48545295e-01 -3.41739565e-01
-1.68888658e-01 2.50405282e-01 -2.64897883e-01 6.83868587e-01
1.55303001e-01 5.98678827e-01 -9.06327188e-01 -8.07130516e-01
2.78302461e-01 1.14393272e-01 -2.55632102e-01 3.11438918e-01
1.10671110e-01 7.02908576e-01 -5.87172568e-01 1.04051232e+00
7.02987015e-01 -5.44094384e-01 3.42673898e-01 -7.11069107e-01
8.37485865e-02 -1.88445449e-01 -1.42816186e+00 1.75094187e+00
-6.56696260e-01 -1.73609287e-01 8.93116929e-03 -1.19547713e+00
6.82413995e-01 4.51874137e-01 7.51251996e-01 -3.06325227e-01
4.96302396e-02 1.92359671e-01 -4.21758741e-01 -5.35008729e-01
-5.04153930e-02 -5.17587304e-01 -1.69807896e-01 6.08362734e-01
4.32062715e-01 2.95766979e-01 -1.93861226e-04 5.96251041e-02
5.47016144e-01 -1.07835241e-01 7.77753890e-01 -1.39527738e-01
9.76472318e-01 -4.71311122e-01 1.04663420e+00 1.56270459e-01
-7.51661882e-02 4.87812251e-01 4.88087207e-01 -4.65012878e-01
-5.34584522e-01 -8.68999958e-01 -1.78597376e-01 1.29295325e+00
3.79850090e-01 -4.03952092e-01 -1.70300066e-01 -1.37317050e+00
2.73322058e-03 2.30463982e-01 -7.04244673e-01 -2.42454801e-02
-2.62434244e-01 -8.47810805e-01 -2.85691507e-02 4.63921219e-01
4.36619878e-01 -6.44979060e-01 -2.12823711e-02 -2.89507993e-02
-3.47601086e-01 -1.07087207e+00 -5.75498343e-01 2.72803038e-01
-9.25335109e-01 -1.52742922e+00 -6.06143892e-01 -7.11544693e-01
8.92332315e-01 8.14031005e-01 8.28069866e-01 -4.27182727e-02
8.24602321e-02 5.13427138e-01 -3.68456423e-01 -7.31525943e-02
-1.40815318e-01 -1.22024119e-02 1.68885797e-01 5.94754517e-01
4.07937646e-01 -7.37936378e-01 -3.88277203e-01 6.42408133e-01
-9.63434935e-01 1.16345055e-01 7.25976884e-01 1.13046443e+00
9.44803834e-01 -2.53208680e-03 7.78966963e-01 -1.26172304e+00
1.14329182e-01 -5.75584769e-01 -3.48913401e-01 7.28255987e-01
-7.96347558e-01 2.89069321e-02 8.98853540e-01 -4.00558621e-01
-9.42157745e-01 1.68126017e-01 3.17485034e-01 -5.32023489e-01
-1.29450172e-01 2.31026441e-01 -8.05409491e-01 -1.35715026e-02
1.66087463e-01 3.39513540e-01 1.59748271e-01 -4.39211726e-01
4.56389844e-01 6.87408388e-01 1.38277207e-02 -3.92002881e-01
6.99097395e-01 7.64295638e-01 3.05590093e-01 -4.46379423e-01
-1.42208385e+00 -8.52494001e-01 -9.95600760e-01 -2.39063114e-01
8.86012137e-01 -1.36343658e+00 -7.19635367e-01 4.31249112e-01
-6.35547996e-01 5.28329492e-01 -1.14714064e-01 4.87941623e-01
-5.36042333e-01 4.99206752e-01 -4.26360697e-01 -4.61626947e-01
-1.85642496e-01 -1.29403055e+00 1.13655233e+00 1.52811736e-01
2.46893480e-01 -1.07554793e+00 -5.09412847e-02 6.93928003e-01
-1.10038266e-01 1.41858459e-01 1.03231442e+00 -7.15490997e-01
-4.22061563e-01 1.02381825e-01 -3.60126257e-01 5.37660599e-01
5.70701063e-01 -5.17200172e-01 -1.06723654e+00 -5.58831513e-01
3.36139172e-01 -5.98541319e-01 8.49201739e-01 -4.74021100e-02
1.09852624e+00 -1.58000797e-01 -3.30450833e-01 5.51049531e-01
1.50559604e+00 -1.80803135e-01 -1.90783426e-01 4.21063937e-02
1.10282373e+00 8.57641578e-01 8.64782035e-01 3.71482074e-01
6.61449730e-01 7.52632737e-01 4.10567641e-01 1.23423323e-01
8.57056603e-02 -1.80765539e-01 2.48520091e-01 1.35864437e+00
1.25114620e-01 1.01994887e-01 -3.46305102e-01 2.72481471e-01
-1.83979654e+00 -1.11157990e+00 1.65346712e-01 1.99069977e+00
4.87389445e-01 -3.26752931e-01 2.02128440e-01 -4.18322086e-02
8.84059668e-01 5.72176099e-01 -7.22580671e-01 1.16578788e-01
-1.09717458e-01 -6.52344048e-01 2.27061406e-01 -3.50418426e-02
-1.34701848e+00 3.57412457e-01 4.80861759e+00 8.43452811e-01
-9.65703487e-01 4.45410401e-01 5.48065662e-01 -5.83967157e-02
-4.74613816e-01 7.45734153e-03 -9.29728866e-01 4.36036140e-01
3.23663175e-01 9.15741324e-02 3.24855387e-01 8.95620227e-01
-2.88376540e-01 1.22550204e-01 -1.27593386e+00 1.24093354e+00
5.34633994e-01 -9.95346785e-01 2.06637472e-01 2.44542643e-01
7.55107582e-01 -2.48492181e-01 3.53284664e-02 3.86812180e-01
9.31265280e-02 -4.84794229e-01 5.12273669e-01 3.95534158e-01
8.96606028e-01 -8.41037691e-01 6.70805812e-01 5.08621037e-01
-1.59881425e+00 -3.77602398e-01 -4.10651416e-01 4.14334208e-01
7.11356625e-02 7.41565943e-01 -2.42095068e-01 1.22450900e+00
2.62365013e-01 1.22252607e+00 -7.35536516e-01 3.82080078e-01
-6.18340708e-02 9.00492370e-02 9.09462497e-02 5.05237997e-01
5.59321977e-02 -4.00261283e-01 1.77165762e-01 7.99336135e-01
2.30325341e-01 1.04734175e-01 5.81691623e-01 3.89382064e-01
-9.69353039e-03 2.64453679e-01 -8.12325835e-01 3.15700024e-01
4.88405228e-01 1.62222350e+00 -5.53115368e-01 -3.15506697e-01
-1.02637768e+00 7.79142022e-01 5.50448239e-01 2.39887878e-01
-5.88618219e-01 2.80316412e-01 5.10651350e-01 -1.72034100e-01
2.62255162e-01 1.69065863e-01 -1.30203202e-01 -1.79946899e+00
1.39866129e-01 -9.97402072e-01 7.63201296e-01 -3.60378474e-01
-1.67885268e+00 3.01507175e-01 2.09594518e-02 -1.84921122e+00
-2.16604397e-02 -5.09368956e-01 -1.56406581e-01 5.94160020e-01
-1.53798389e+00 -1.66141570e+00 -2.80435383e-01 6.07526124e-01
6.35565042e-01 -4.84668374e-01 9.34319377e-01 4.38556194e-01
-6.63330555e-01 5.39615452e-01 1.59595519e-01 3.05855591e-02
9.29554701e-01 -1.07842994e+00 -5.62212050e-01 4.90492880e-01
3.27342331e-01 4.70878541e-01 -4.12477925e-02 -3.26305777e-01
-1.40897608e+00 -1.31786776e+00 4.96503264e-01 -4.29924220e-01
4.60883051e-01 -3.14879298e-01 -9.30486679e-01 7.75284588e-01
-1.13614567e-01 3.90576184e-01 1.30713379e+00 3.71714443e-01
-8.71164262e-01 -4.32870388e-01 -1.02526641e+00 3.10993612e-01
9.76302683e-01 -8.01680982e-01 -4.40629870e-01 4.44169849e-01
5.88654995e-01 -1.50354505e-01 -1.31941473e+00 6.93107724e-01
5.39588869e-01 -1.10130036e+00 9.32558894e-01 -3.73689383e-01
2.56069124e-01 -5.29159367e-01 -5.03911376e-01 -1.43084288e+00
-5.38357139e-01 2.34651059e-01 -1.01241730e-01 1.66903031e+00
3.51465456e-02 -8.50097835e-01 3.79507214e-01 1.67694554e-01
-2.52157710e-02 -1.04628158e+00 -8.81084025e-01 -5.77910185e-01
-1.91019714e-01 2.26963535e-01 5.73677421e-01 1.28824532e+00
-1.03636146e-01 5.77238083e-01 -5.26163161e-01 2.56742477e-01
8.08327854e-01 7.69841075e-01 5.82380533e-01 -1.41586864e+00
-3.16032171e-01 -9.38426107e-02 -4.21597332e-01 -7.09291697e-01
4.77377772e-01 -1.20019841e+00 -3.88304144e-01 -1.27090001e+00
7.98325419e-01 -5.12610734e-01 -6.15579307e-01 4.60381478e-01
-3.62519145e-01 9.29652676e-02 2.92313308e-01 5.89788437e-01
-8.85804594e-01 6.69387877e-01 1.35367239e+00 -2.82197803e-01
3.00105691e-01 6.92717060e-02 -1.06050456e+00 7.85419583e-01
3.78946036e-01 -4.60129470e-01 -6.18904412e-01 -1.27038449e-01
-1.45591432e-02 1.67365775e-01 2.34018490e-01 -7.01976657e-01
1.03019416e-01 -2.97771245e-01 3.33806664e-01 -6.60520077e-01
3.63701433e-01 -1.13349950e+00 1.32321462e-01 1.69827372e-01
-2.77308494e-01 -2.81472504e-03 -3.57808203e-01 9.98729289e-01
-4.35455263e-01 -1.38976172e-01 8.98947060e-01 -3.56353253e-01
-7.25828826e-01 4.96211380e-01 -1.01072034e-02 1.31811962e-01
1.16509104e+00 7.55718723e-02 -4.19655144e-01 -1.47103341e-02
-6.77764356e-01 4.41777766e-01 5.62975168e-01 6.46147966e-01
3.87259871e-01 -1.80490923e+00 -4.96055692e-01 4.14859354e-01
7.84161091e-01 2.85662618e-02 6.86623394e-01 9.47269261e-01
3.67183059e-01 3.64747465e-01 -6.91675469e-02 -8.05171072e-01
-1.43074906e+00 9.82406318e-01 6.07937090e-02 -5.10495722e-01
-4.15421724e-01 5.08543968e-01 6.73328936e-01 -6.99050188e-01
-6.51460737e-02 2.16559678e-01 -5.80771148e-01 7.69249201e-01
5.31311333e-01 2.89111584e-01 -2.68636737e-02 -9.76433933e-01
-4.26756978e-01 1.04886174e+00 -3.11133444e-01 3.31224293e-01
1.26417267e+00 -4.83492345e-01 -2.41659641e-01 1.02765524e+00
1.48609793e+00 -1.37506202e-01 -1.05063200e+00 -6.85637236e-01
-1.79529086e-01 -3.68306279e-01 -1.99260756e-01 -3.92995983e-01
-1.25182271e+00 9.17632043e-01 5.25656223e-01 -1.81472953e-02
1.20820308e+00 1.21075340e-01 4.32011753e-01 2.04955235e-01
4.84336495e-01 -9.79657948e-01 2.83533007e-01 9.91039574e-02
6.10742092e-01 -1.55738366e+00 2.87784249e-01 -7.20308065e-01
-9.16589081e-01 1.06111741e+00 8.34788680e-01 2.57896245e-01
7.08557129e-01 -3.43101799e-01 5.78824878e-02 -4.07990247e-01
-7.66617537e-01 -1.77764639e-01 4.54104900e-01 3.26167405e-01
1.78050533e-01 1.26538724e-01 -7.47151822e-02 5.18862724e-01
5.45421779e-01 -2.85198569e-01 2.06031412e-01 6.74458861e-01
-3.46140623e-01 -1.16894925e+00 -2.94561774e-01 6.09334886e-01
-2.47534126e-01 3.40503961e-01 -1.08807720e-01 5.98302245e-01
5.70061743e-01 9.70407963e-01 -3.11124831e-01 -5.31757295e-01
3.11323345e-01 3.24206263e-01 2.95997709e-01 -8.68436933e-01
-3.77724320e-01 1.08062498e-01 -9.03780833e-02 -4.97130752e-01
-9.37865198e-01 -7.73421824e-01 -7.92111456e-01 6.64911643e-02
-6.73829377e-01 -2.89549679e-02 2.68746763e-01 1.08967066e+00
3.44404131e-01 4.86891657e-01 1.09155810e+00 -6.51223361e-01
-7.54432678e-01 -7.60160089e-01 -9.74422514e-01 6.82314992e-01
2.47582495e-01 -1.12145483e+00 -5.89932203e-01 -8.57693329e-02] | [8.564179420471191, 4.512683391571045] |
ca0ce3a4-d8cd-4a9d-b6e1-93d3c0e4b37d | look-read-and-ask-learning-to-ask-questions | 2211.12950 | null | https://arxiv.org/abs/2211.12950v1 | https://arxiv.org/pdf/2211.12950v1.pdf | Look, Read and Ask: Learning to Ask Questions by Reading Text in Images | We present a novel problem of text-based visual question generation or TextVQG in short. Given the recent growing interest of the document image analysis community in combining text understanding with conversational artificial intelligence, e.g., text-based visual question answering, TextVQG becomes an important task. TextVQG aims to generate a natural language question for a given input image and an automatically extracted text also known as OCR token from it such that the OCR token is an answer to the generated question. TextVQG is an essential ability for a conversational agent. However, it is challenging as it requires an in-depth understanding of the scene and the ability to semantically bridge the visual content with the text present in the image. To address TextVQG, we present an OCR consistent visual question generation model that Looks into the visual content, Reads the scene text, and Asks a relevant and meaningful natural language question. We refer to our proposed model as OLRA. We perform an extensive evaluation of OLRA on two public benchmarks and compare them against baselines. Our model OLRA automatically generates questions similar to the public text-based visual question answering datasets that were curated manually. Moreover, we significantly outperform baseline approaches on the performance measures popularly used in text generation literature. | ['Anand Mishra', 'Shankar Gangisetty', 'Soumya Jahagirdar'] | 2022-11-23 | null | null | null | null | ['question-generation'] | ['natural-language-processing'] | [ 5.71008503e-01 3.25081199e-01 3.35357577e-01 -2.05562279e-01
-9.08261776e-01 -8.36624205e-01 9.77036476e-01 2.54383981e-01
-1.53482065e-01 6.18152201e-01 3.57772470e-01 -6.41711593e-01
2.74102777e-01 -8.58952224e-01 -7.83727884e-01 -4.11426306e-01
7.09211469e-01 5.83370924e-01 2.60688066e-01 -2.87193239e-01
4.87479568e-01 1.04506150e-01 -1.59010768e+00 6.19860828e-01
1.00326920e+00 8.63986015e-01 5.02393126e-01 9.50359762e-01
-7.37323105e-01 1.25768697e+00 -9.95062113e-01 -5.48272789e-01
-4.92464639e-02 -9.00091767e-01 -1.52719700e+00 5.51559269e-01
8.40568423e-01 -2.14256898e-01 -9.19716582e-02 9.16078329e-01
2.27006063e-01 2.62351215e-01 8.90992165e-01 -1.67280531e+00
-9.98577535e-01 1.37673691e-01 -5.26183486e-01 1.24263263e-03
7.16227651e-01 3.57720584e-01 1.03280735e+00 -1.04771757e+00
9.77849543e-01 1.53952181e+00 -1.15155920e-01 6.32198811e-01
-1.03329504e+00 -2.31493771e-01 2.38312453e-01 3.92968297e-01
-1.12926090e+00 -1.08446389e-01 8.73367608e-01 -4.68867064e-01
6.85865998e-01 5.04235148e-01 6.03419721e-01 9.96301055e-01
1.93253905e-01 1.06605732e+00 1.13099194e+00 -5.95065475e-01
3.29537362e-01 2.54451156e-01 -1.92989141e-01 7.52004623e-01
-1.35278121e-01 -3.74751538e-01 -5.46175659e-01 -8.71622562e-03
5.80380738e-01 -1.84852287e-01 -3.95876467e-01 -2.57405579e-01
-1.34093273e+00 9.69700098e-01 5.83176851e-01 1.70246616e-01
-2.32712984e-01 4.36155230e-01 3.17761123e-01 1.15430333e-01
3.31507564e-01 6.52917802e-01 1.17574528e-01 2.60514289e-01
-6.37130141e-01 4.86505181e-01 6.32223606e-01 1.01465213e+00
9.13901269e-01 -7.81738237e-02 -7.63858140e-01 7.21984804e-01
2.85316885e-01 7.71023273e-01 2.43550971e-01 -1.04404151e+00
5.82831025e-01 9.93282318e-01 1.69030786e-01 -1.04510331e+00
1.95025608e-01 1.39188424e-01 -7.64731228e-01 3.01806569e-01
5.97664773e-01 7.01790825e-02 -1.06208420e+00 1.35802770e+00
4.44228768e-01 -3.56156260e-01 3.16392422e-01 9.76919293e-01
1.30600762e+00 1.19345129e+00 -2.10594218e-02 6.57676756e-02
1.84212005e+00 -1.32434249e+00 -9.17733073e-01 -6.48401737e-01
1.74133867e-01 -1.06327569e+00 1.47969651e+00 8.51384252e-02
-8.47284913e-01 -4.66028988e-01 -7.86304951e-01 -5.80882430e-01
-4.81980413e-01 1.53153643e-01 3.16718131e-01 3.51859361e-01
-1.37241805e+00 -3.74369770e-01 5.08271307e-02 -7.00186968e-01
1.29935935e-01 -3.87208372e-01 -1.86185479e-01 -2.84920335e-01
-9.73302543e-01 5.91085672e-01 3.47123176e-01 -4.50662943e-03
-1.13465929e+00 -2.77413845e-01 -9.19506431e-01 -2.10270092e-01
6.82240784e-01 -9.00257945e-01 1.44504178e+00 -1.19762695e+00
-1.15053475e+00 1.14193666e+00 -5.12542963e-01 -2.82082886e-01
4.85852391e-01 8.50376338e-02 5.37895150e-02 7.31261551e-01
3.37962300e-01 1.13591313e+00 1.16564643e+00 -1.74020231e+00
-5.11121392e-01 -2.18479007e-01 4.11109328e-01 3.81788373e-01
2.11660918e-02 -2.29235306e-01 -6.35871947e-01 -7.10480750e-01
-1.79221466e-01 -7.10780263e-01 -9.06138048e-02 2.90606201e-01
-5.54659188e-01 -5.77929139e-01 1.19614553e+00 -8.16476524e-01
6.32967949e-01 -1.74684131e+00 -8.75289440e-02 -8.49343091e-02
3.81985635e-01 5.15892766e-02 -4.56352770e-01 7.55982459e-01
1.08247690e-01 2.10920319e-01 -2.96875179e-01 -8.89602154e-02
-3.60060818e-02 7.26608858e-02 -7.35002995e-01 -4.99965176e-02
2.80347705e-01 1.51628077e+00 -1.09214532e+00 -9.39371049e-01
2.97469705e-01 2.36881211e-01 -1.46689981e-01 6.82102978e-01
-1.08756268e+00 2.86793560e-01 -4.05603021e-01 4.57609802e-01
4.87580627e-01 -5.41339695e-01 -1.57807559e-01 -2.49480307e-01
5.18129729e-02 -6.78527579e-02 -5.41906118e-01 1.50386703e+00
-5.03755987e-01 1.27108383e+00 -2.64361173e-01 -7.00824618e-01
1.22524369e+00 2.18398467e-01 2.00229771e-02 -9.26461399e-01
3.73927653e-02 -1.77648425e-01 -5.96034467e-01 -7.75396347e-01
9.20028627e-01 -2.66435612e-02 -5.68987466e-02 7.51728475e-01
-2.30766907e-01 -8.08245480e-01 3.26686263e-01 8.18785429e-01
7.22131491e-01 7.04833269e-02 3.01394373e-01 -4.15407270e-02
6.74783945e-01 3.90708566e-01 -3.78404677e-01 7.86848187e-01
-2.35384032e-02 8.40273559e-01 5.93858182e-01 -1.86857060e-01
-1.05702317e+00 -1.07556617e+00 4.96670544e-01 1.11652482e+00
4.59486425e-01 -3.29301327e-01 -9.28457856e-01 -7.92905271e-01
-2.83957392e-01 7.50978529e-01 -8.65431905e-01 1.85168982e-01
-3.19821060e-01 -5.48624098e-02 2.96992809e-01 1.62275970e-01
7.69232690e-01 -1.59531665e+00 -6.41661346e-01 -8.59757438e-02
-8.82976353e-01 -1.41690648e+00 -8.38513672e-01 -4.57196206e-01
-3.33078355e-01 -1.13879585e+00 -1.08165514e+00 -1.17435670e+00
8.66309226e-01 7.30452359e-01 1.48676157e+00 3.22682559e-01
-5.24762332e-01 9.28864002e-01 -8.37551355e-01 -5.23123145e-01
-6.29018962e-01 -1.61080644e-01 -1.00574589e+00 3.18938702e-01
1.12383198e-02 1.42628744e-01 -8.09384584e-01 2.57384539e-01
-1.36437476e+00 6.20951056e-01 4.97928202e-01 5.24859846e-01
5.57848454e-01 -4.63054210e-01 5.63099861e-01 -5.66983283e-01
9.51534152e-01 -1.80618495e-01 -3.80118430e-01 7.10340202e-01
-1.86065271e-01 3.46366853e-01 5.09855688e-01 -1.48439854e-01
-1.15276754e+00 -8.72134790e-02 2.40023304e-02 -1.57082438e-01
-1.10575609e-01 1.25993118e-01 -9.74600464e-02 2.08317459e-01
6.43448889e-01 5.84774613e-01 -1.07733816e-01 8.24025571e-02
8.35758746e-01 6.85574532e-01 7.00360239e-01 -4.88788873e-01
9.56125975e-01 7.10103214e-01 -2.46826351e-01 -1.19457054e+00
-8.49034190e-01 -5.45870185e-01 -9.78974998e-02 -7.02248275e-01
1.47919023e+00 -4.50613588e-01 -9.18734014e-01 2.88567394e-01
-1.85695994e+00 -2.98739910e-01 -3.90978724e-01 -3.14253598e-01
-7.20470369e-01 6.33343637e-01 9.16660652e-02 -9.91289616e-01
-8.20533812e-01 -9.72882092e-01 1.43348193e+00 3.47506315e-01
3.73980030e-02 -8.63830686e-01 -4.51902300e-02 8.69608998e-01
2.11161539e-01 4.90688562e-01 1.05322707e+00 -3.15435022e-01
-8.99944961e-01 2.10177302e-01 -8.62732053e-01 1.39777765e-01
6.39514625e-02 -1.96342170e-01 -9.72196877e-01 -2.34329477e-02
-2.78346032e-01 -6.25917733e-01 7.89795101e-01 4.92715985e-02
1.21790218e+00 -4.75315452e-01 -8.52291435e-02 -4.75726032e-04
1.37476695e+00 2.72809356e-01 7.53428817e-01 2.40106747e-01
7.54877865e-01 1.00952327e+00 7.06633687e-01 1.90620974e-01
8.35202515e-01 6.63830161e-01 7.22304165e-01 -5.78907073e-01
-4.91580129e-01 -5.00057340e-01 2.02136442e-01 2.71477014e-01
2.68000603e-01 -6.86651945e-01 -9.88167465e-01 8.27507615e-01
-1.95122302e+00 -9.42970872e-01 -5.10446012e-01 1.70694005e+00
7.54897833e-01 -4.83443111e-01 -2.74426378e-02 -4.12713811e-02
7.23470449e-01 3.66088212e-01 -5.33479989e-01 -5.64235628e-01
-2.05422938e-01 -1.62361898e-02 -3.29049006e-02 5.90980649e-01
-7.08234668e-01 1.16341054e+00 5.14490509e+00 6.64256096e-01
-8.55438232e-01 2.68530399e-02 7.77758658e-01 3.60981733e-01
-7.18208611e-01 2.18283460e-02 -3.94821346e-01 -4.15576138e-02
2.48045325e-01 -4.47956264e-01 3.98340434e-01 6.65422559e-01
4.15661275e-01 -4.70607758e-01 -1.06966233e+00 1.30583358e+00
7.78304338e-01 -1.49228060e+00 9.09234166e-01 -2.27220565e-01
8.63884807e-01 -6.50051236e-01 2.81851981e-02 -2.11061444e-02
2.65416712e-01 -1.28710091e+00 9.46383834e-01 4.58334506e-01
9.42938626e-01 -6.20128453e-01 5.17175436e-01 9.21908319e-02
-1.48045218e+00 2.22734451e-01 -2.13179648e-01 2.80519187e-01
2.30319530e-01 3.05554777e-01 -1.30279148e+00 4.48810399e-01
5.39240062e-01 3.04272115e-01 -1.07457256e+00 8.06205988e-01
-5.27420521e-01 2.23724946e-01 3.70599836e-01 -5.03459096e-01
4.59024280e-01 -7.18993600e-03 3.77027690e-01 9.96683717e-01
8.36013630e-02 1.70214251e-02 1.12866789e-01 1.06705666e+00
-2.66323090e-01 4.52637970e-01 -5.81735969e-01 -2.11273134e-01
8.79738852e-02 1.31325471e+00 -8.68896961e-01 -4.67814147e-01
-1.01989351e-01 1.49203527e+00 1.36979356e-01 7.60486662e-01
-6.36482060e-01 -6.18620217e-01 9.03162062e-02 1.70973524e-01
-7.74093857e-03 -3.26929893e-03 4.39301059e-02 -9.84897375e-01
1.40731156e-01 -1.17276192e+00 3.33003491e-01 -1.72922468e+00
-1.05672741e+00 8.65319967e-01 -8.55200142e-02 -9.68515277e-01
-4.12150145e-01 -4.37454104e-01 -6.44747615e-01 8.08103740e-01
-1.65574145e+00 -1.48546553e+00 -1.03212988e+00 7.83059657e-01
1.13128567e+00 2.05932289e-01 4.53740358e-01 -2.99027473e-01
1.84719965e-01 1.56817392e-01 -3.51877689e-01 2.45633230e-01
6.93866134e-01 -1.33875132e+00 6.99552774e-01 8.27036738e-01
4.06675696e-01 1.16317220e-01 7.70058095e-01 -5.44260323e-01
-1.31339669e+00 -1.40936160e+00 9.55757022e-01 -5.03193498e-01
4.58326250e-01 -6.04592144e-01 -7.91135967e-01 4.60476846e-01
9.16198909e-01 -5.19450366e-01 2.05494285e-01 -8.03379536e-01
-3.11957240e-01 1.78055868e-01 -7.97160387e-01 9.27340150e-01
7.98905730e-01 -5.86628437e-01 -8.54252577e-01 5.29082596e-01
9.94930267e-01 -8.48736614e-02 -5.80638982e-02 -1.39183104e-01
1.97996572e-01 -8.34291101e-01 9.57063496e-01 -4.56061721e-01
7.68708885e-01 -5.09702027e-01 1.68272555e-02 -1.07650506e+00
2.90197998e-01 -6.63741887e-01 4.82081831e-01 1.14281976e+00
3.56904596e-01 -2.82544464e-01 6.04550362e-01 2.27102324e-01
1.37463957e-01 -3.63958150e-01 -7.01728404e-01 -3.92184615e-01
-1.56113103e-01 -2.91578770e-01 4.16930050e-01 7.13974476e-01
-2.31421173e-01 7.02991903e-01 -4.89669353e-01 -5.56504987e-02
6.15617514e-01 4.18720514e-01 1.27916634e+00 -1.01568723e+00
-1.33360159e-02 -3.10860336e-01 -1.64742440e-01 -1.33836579e+00
3.67086306e-02 -9.75740731e-01 3.67108136e-01 -2.47526479e+00
5.09388745e-01 2.24589765e-01 6.41726136e-01 3.03057224e-01
-5.03190577e-01 4.52082664e-01 5.47856450e-01 -1.15925141e-01
-9.16010439e-01 6.83356285e-01 1.76869488e+00 -5.98541796e-01
-1.09118260e-01 -6.07865214e-01 -8.08006763e-01 5.10525703e-01
6.86958909e-01 -2.40453891e-02 -6.61291361e-01 -4.02860492e-01
4.99242246e-01 3.37259084e-01 8.24324727e-01 -5.97714245e-01
1.62060797e-01 -4.56574827e-01 1.70749426e-01 -8.92308116e-01
3.27396393e-01 -5.21293581e-01 -3.67373317e-01 4.08008367e-01
-5.40219843e-01 1.06759787e-01 -3.30033340e-02 5.37966251e-01
-2.85509080e-01 -2.85402685e-01 6.19826674e-01 -2.08863288e-01
-8.46495211e-01 1.27503380e-01 -5.59132040e-01 5.45774877e-01
1.07256985e+00 -2.34374538e-01 -8.34375679e-01 -1.09059966e+00
-2.84919143e-01 4.99558121e-01 4.03420299e-01 7.39243984e-01
1.22304368e+00 -1.20558548e+00 -9.62437093e-01 -2.65918136e-01
7.54911959e-01 5.31412521e-03 3.05115908e-01 2.24155843e-01
-7.47233391e-01 6.16199911e-01 1.53785363e-01 -6.19429231e-01
-1.43647945e+00 6.26506329e-01 2.77081877e-01 -1.76920950e-01
-5.06082416e-01 5.34180880e-01 8.08710992e-01 -2.29886651e-01
1.20410137e-02 -3.71792525e-01 -3.57852221e-01 4.35563289e-02
8.12258840e-01 8.27724561e-02 -2.83567667e-01 -6.63580120e-01
-1.39627472e-01 6.45371079e-01 2.14317784e-01 -4.63664055e-01
6.52440965e-01 -3.74295980e-01 -1.56635433e-01 2.51978636e-01
1.08301425e+00 -1.63461789e-01 -8.57790768e-01 -1.94294438e-01
-1.63544714e-01 -3.54722142e-01 -2.87628621e-01 -7.99482644e-01
-7.70368397e-01 9.87428367e-01 4.10207927e-01 4.74705875e-01
1.01180577e+00 5.63224137e-01 8.03498805e-01 4.90135103e-01
2.42726635e-02 -8.74831617e-01 1.08538783e+00 5.10965586e-01
1.57092071e+00 -1.33166873e+00 -1.89495251e-01 -4.63665634e-01
-9.31505799e-01 9.36718881e-01 8.15610230e-01 1.92560539e-01
-1.09407552e-01 -3.32921147e-01 4.20791894e-01 -3.49253774e-01
-7.81606138e-01 -5.47498107e-01 6.14326715e-01 8.63210380e-01
1.66668996e-01 -2.02381477e-01 -2.17027381e-01 -3.94493304e-02
-2.98341960e-01 -2.25651845e-01 7.38516629e-01 6.85102522e-01
-6.57232642e-01 -7.55152047e-01 -7.60453641e-01 2.27710858e-01
-7.16466829e-02 -1.53632775e-01 -1.00105560e+00 6.08197272e-01
-2.55630314e-01 1.46729231e+00 1.13031007e-01 7.57225789e-03
1.45979568e-01 3.72000113e-02 3.60975832e-01 -6.54779553e-01
-4.22420055e-01 -2.44720235e-01 3.61769879e-03 -4.05136734e-01
-3.87032837e-01 -1.37452126e-01 -1.34205365e+00 7.65772685e-02
2.54710764e-02 2.53622323e-01 7.87846386e-01 9.45954084e-01
1.05943024e-01 4.25123692e-01 5.08021355e-01 -4.19173956e-01
9.25074369e-02 -8.18904579e-01 -1.57786474e-01 7.41909385e-01
3.77708346e-01 -1.70883179e-01 -3.31938386e-01 5.55751503e-01] | [10.994331359863281, 1.5703485012054443] |
596ad7dd-f0c5-4b18-a2ba-627b51a4e737 | discriminative-learning-of-deep-convolutional | null | null | http://openaccess.thecvf.com/content_iccv_2015/html/Simo-Serra_Discriminative_Learning_of_ICCV_2015_paper.html | http://openaccess.thecvf.com/content_iccv_2015/papers/Simo-Serra_Discriminative_Learning_of_ICCV_2015_paper.pdf | Discriminative Learning of Deep Convolutional Feature Point Descriptors | Deep learning has revolutionalized image-level tasks such as classification, but patch-level tasks, such as correspondence, still rely on hand-crafted features, e.g. SIFT. In this paper we use Convolutional Neural Networks (CNNs) to learn discriminant patch representations and in particular train a Siamese network with pairs of (non-)corresponding patches. We deal with the large number of potential pairs with the combination of a stochastic sampling of the training set and an aggressive mining strategy biased towards patches that are hard to classify. By using the L2 distance during both training and testing we develop 128-D descriptors whose euclidean distances reflect patch similarity, and which can be used as a drop-in replacement for any task involving SIFT. We demonstrate consistent performance gains over the state of the art, and generalize well against scaling and rotation, perspective transformation, non-rigid deformation, and illumination changes. Our descriptors are efficient to compute and amenable to modern GPUs, and are publicly available. | ['Francesc Moreno-Noguer', 'Iasonas Kokkinos', 'Edgar Simo-Serra', 'Pascal Fua', 'Eduard Trulls', 'Luis Ferraz'] | 2015-12-01 | null | null | null | iccv-2015-12 | ['satellite-image-classification'] | ['computer-vision'] | [ 1.70705631e-01 -4.21855092e-01 -4.28735077e-01 -2.45722219e-01
-9.32207942e-01 -6.69183731e-01 6.79975152e-01 2.28954807e-01
-5.04440188e-01 4.22537595e-01 2.11555418e-02 -9.53668635e-03
-8.79276320e-02 -9.58583891e-01 -1.00546050e+00 -5.93257785e-01
-3.14201385e-01 5.29518962e-01 2.46912092e-01 -2.32539892e-01
4.89075661e-01 1.02214265e+00 -1.88971841e+00 1.75892219e-01
5.08461893e-01 1.29959524e+00 -2.64079869e-01 5.11905849e-01
2.26929277e-01 1.22926861e-01 -3.02680403e-01 -2.14300811e-01
8.33326042e-01 -8.40462297e-02 -6.01855040e-01 3.40050571e-02
1.19015098e+00 -2.93850839e-01 -4.84477550e-01 9.40250278e-01
4.28892821e-01 2.94431776e-01 5.95042646e-01 -1.10669434e+00
-6.92649603e-01 -1.72067627e-01 -7.73325920e-01 -1.19304880e-01
3.73902172e-01 1.53095543e-01 1.18850839e+00 -1.18172407e+00
8.52420688e-01 1.00940132e+00 1.24076712e+00 1.40574157e-01
-1.47437096e+00 -4.03198570e-01 -1.80298865e-01 1.29112631e-01
-1.55336690e+00 -4.12273437e-01 7.62693942e-01 -2.88012624e-01
1.15357554e+00 1.65383771e-01 8.57855797e-01 6.40728772e-01
3.01891208e-01 6.30962074e-01 7.14324474e-01 -3.07834208e-01
1.44491777e-01 -2.92931825e-01 -2.54083335e-01 8.33065748e-01
2.43226662e-01 1.89759210e-01 -3.90597135e-01 -4.19374377e-01
1.08861721e+00 2.51119465e-01 -6.23798221e-02 -7.50419617e-01
-1.32060623e+00 1.03190994e+00 6.54455781e-01 6.24198117e-04
-3.44901592e-01 2.79656976e-01 3.74192297e-01 5.65184653e-01
4.22699928e-01 5.15430093e-01 -4.66928720e-01 -7.89149031e-02
-1.02358913e+00 3.86745989e-01 7.62687743e-01 8.83230150e-01
1.31044984e+00 -2.62503713e-01 1.22574747e-01 9.58873034e-01
-2.88220167e-01 4.33956951e-01 3.53299022e-01 -9.75838661e-01
1.55205280e-01 4.75766182e-01 -8.12940225e-02 -1.37934649e+00
-3.07398945e-01 -3.98081928e-01 -9.24108088e-01 4.30517375e-01
3.64805520e-01 2.36493140e-01 -9.52004373e-01 1.30934274e+00
3.93035650e-01 6.43073320e-02 -3.29599470e-01 9.77526426e-01
4.49580848e-01 3.08715105e-01 -4.16101128e-01 4.08099174e-01
1.03399611e+00 -9.86660063e-01 1.38129324e-01 -1.29936919e-01
6.27787650e-01 -1.09537137e+00 9.73973215e-01 3.55407864e-01
-1.19015670e+00 -6.91316128e-01 -1.17275763e+00 -5.17174125e-01
-4.45411593e-01 8.01706612e-02 7.22422361e-01 2.75182813e-01
-1.20439541e+00 1.25594914e+00 -9.53283727e-01 -3.15038651e-01
8.38316798e-01 7.16277301e-01 -7.66779125e-01 -2.80293524e-01
-7.02090621e-01 5.55973828e-01 -1.80647701e-01 -8.82182568e-02
-4.97015685e-01 -7.14294136e-01 -8.62199068e-01 1.49693843e-02
6.06175847e-02 -5.80041885e-01 8.20883274e-01 -1.21070719e+00
-1.15605402e+00 1.08246422e+00 -1.67297736e-01 -3.83188069e-01
6.14234626e-01 -1.79061547e-01 6.03270121e-02 2.75303274e-01
3.99485886e-01 9.73444283e-01 1.13924789e+00 -6.79622233e-01
-6.34178340e-01 -3.72057706e-01 -4.46087308e-02 1.42639056e-01
-1.52860686e-01 -1.93438962e-01 -4.69415039e-01 -6.72166348e-01
3.83615971e-01 -1.24984968e+00 -2.64414161e-01 8.00181329e-01
-2.73209393e-01 -2.60814756e-01 7.58010268e-01 -2.29177356e-01
4.02296007e-01 -2.26923513e+00 -8.19943473e-02 5.63889623e-01
1.95150569e-01 1.37742475e-01 -3.97233278e-01 4.03711826e-01
-7.24881068e-02 -6.96236193e-02 -3.11202053e-02 -2.51890153e-01
-1.51450501e-03 1.58659637e-01 -8.98618773e-02 1.09711087e+00
6.98141813e-01 8.18868041e-01 -8.22467029e-01 -3.88600707e-01
2.48218954e-01 5.92329323e-01 -8.12863350e-01 -9.33672562e-02
-1.08608738e-01 1.27611816e-01 -3.76204222e-01 8.65002692e-01
8.68033051e-01 -2.96117067e-01 -2.30880320e-01 -3.87630433e-01
-7.26012513e-02 3.72274905e-01 -1.14174747e+00 2.00941849e+00
-4.26184475e-01 8.10645521e-01 -7.73001686e-02 -1.10878253e+00
1.03882444e+00 -3.76250684e-01 6.31432831e-01 -6.59530878e-01
-8.77639204e-02 4.08176690e-01 -4.73369434e-02 -6.85473457e-02
5.31683147e-01 1.96467578e-01 2.10585430e-01 5.60056567e-01
1.03368454e-01 -1.01809382e-01 -6.03560405e-03 -1.24557465e-02
1.32168031e+00 2.90227443e-01 1.67757884e-01 -5.12088120e-01
2.18692631e-01 2.62871742e-01 3.59359056e-01 8.24974298e-01
8.44690278e-02 8.78402710e-01 5.12759209e-01 -1.14352250e+00
-1.28984749e+00 -9.26508904e-01 -4.93944973e-01 9.55676556e-01
3.42860907e-01 -3.80220890e-01 -2.61734754e-01 -4.35419291e-01
4.65581924e-01 -1.29963890e-01 -7.15066731e-01 -1.39010534e-01
-7.94124186e-01 -3.25829864e-01 4.33012456e-01 6.63238049e-01
3.65083307e-01 -7.79842734e-01 -6.41387403e-01 3.49799633e-01
4.66508359e-01 -7.98091769e-01 -5.44151068e-01 4.62948799e-01
-8.86285484e-01 -1.20946562e+00 -6.11401916e-01 -1.11119425e+00
8.68857145e-01 5.61352611e-01 1.27436078e+00 3.36445838e-01
-7.41463542e-01 2.98740327e-01 -1.70110285e-01 6.56960905e-02
1.03076711e-01 1.53119594e-01 -3.88842970e-02 -1.65165886e-01
4.89017218e-01 -7.88289011e-01 -9.65478718e-01 4.87748325e-01
-6.62165642e-01 -3.21275085e-01 7.23650336e-01 1.30782139e+00
1.14802217e+00 -1.75299659e-01 -1.59915481e-02 -4.56868768e-01
3.56374621e-01 -2.33490273e-01 -6.45097256e-01 9.02277976e-02
-3.96536350e-01 1.60700113e-01 7.95921862e-01 -6.91993058e-01
-1.35335907e-01 3.68279338e-01 -8.85951146e-02 -6.93485081e-01
-8.05354565e-02 2.39048615e-01 2.77320534e-01 -9.29077744e-01
8.15638423e-01 1.32441431e-01 2.38599941e-01 -3.76022637e-01
3.13885689e-01 2.82058239e-01 4.89611924e-01 -7.99803376e-01
8.70007217e-01 6.99556530e-01 3.94240439e-01 -9.05710757e-01
-5.41672349e-01 -3.74855042e-01 -7.05573380e-01 1.96089447e-01
3.96761030e-01 -8.53660345e-01 -6.83262348e-01 3.34861159e-01
-1.08845866e+00 -2.06193224e-01 -5.57515144e-01 2.98770517e-01
-6.61158741e-01 2.19047397e-01 -5.24144888e-01 -1.98714212e-01
-3.73131782e-01 -1.21877635e+00 1.48385060e+00 1.09406516e-01
-1.90353826e-01 -7.88658321e-01 1.81163922e-01 -1.38791531e-01
5.12634575e-01 3.79777819e-01 7.65978873e-01 -6.26656234e-01
-7.57294476e-01 -6.09421551e-01 -3.64499092e-01 3.43284070e-01
1.05787724e-01 7.08571970e-02 -8.12774181e-01 -4.95766431e-01
-2.99580336e-01 -6.49009466e-01 8.93541396e-01 3.12158942e-01
1.46407676e+00 -3.68593007e-01 -3.44096959e-01 1.18814576e+00
1.45762563e+00 -3.62563998e-01 6.03079021e-01 4.42085087e-01
5.92252433e-01 4.78704989e-01 4.22527134e-01 3.04661065e-01
7.94897154e-02 6.11060143e-01 3.40616137e-01 -2.74445236e-01
-1.39807746e-01 -2.84820139e-01 -2.32153445e-01 4.27514970e-01
-9.86559466e-02 2.27670178e-01 -7.75064588e-01 5.53475797e-01
-1.65864038e+00 -8.97962213e-01 1.24087214e-01 2.28572774e+00
7.22035766e-01 9.39527676e-02 -4.85014506e-02 -1.19983964e-01
5.23325980e-01 1.29974946e-01 -7.30287254e-01 -2.41178796e-01
-3.30608279e-01 8.62151563e-01 8.44301403e-01 2.92855710e-01
-1.45059109e+00 8.47069740e-01 6.28047419e+00 8.91256213e-01
-1.39905274e+00 -2.17255995e-01 6.43431723e-01 -1.76085815e-01
-1.83367342e-01 -8.74746125e-03 -5.14084995e-01 1.23576969e-01
4.12060618e-01 1.45059139e-01 6.10356629e-01 1.01628268e+00
-3.52574438e-01 1.33387810e-02 -1.29753852e+00 1.24714577e+00
1.60334557e-01 -1.62841582e+00 -9.67979431e-02 2.29863763e-01
1.06930125e+00 5.49098909e-01 2.32692912e-01 1.75617449e-02
8.92086253e-02 -1.19524038e+00 4.54697549e-01 1.98705018e-01
9.69142318e-01 -6.90949917e-01 5.87484002e-01 -1.60904542e-01
-1.26540720e+00 1.49626866e-01 -7.93254435e-01 1.41003147e-01
-4.98196453e-01 4.12303120e-01 -4.44008023e-01 2.33703241e-01
7.49829412e-01 1.01075685e+00 -4.86542970e-01 1.15361702e+00
9.82628539e-02 3.64115648e-02 -7.70864606e-01 -1.14246182e-01
4.45195258e-01 -2.17552036e-01 2.42261395e-01 9.85701025e-01
3.53481263e-01 -2.72591203e-01 4.02877539e-01 7.43662596e-01
-2.74993598e-01 1.59182936e-01 -8.01393867e-01 2.28788435e-01
4.84658241e-01 1.21108508e+00 -8.12495112e-01 -2.27168590e-01
-5.53446412e-01 1.13508368e+00 4.08386678e-01 6.75614923e-02
-4.66411352e-01 -7.69528210e-01 1.13182175e+00 2.85373360e-01
7.50106156e-01 -3.81819367e-01 -3.68105829e-01 -1.13923144e+00
3.54204297e-01 -8.46549153e-01 2.50178985e-02 -3.52877438e-01
-1.43332541e+00 5.58368385e-01 -2.41838172e-01 -1.53650630e+00
-1.89666554e-01 -7.77165711e-01 -7.36947060e-01 7.48431444e-01
-1.62538207e+00 -1.02789378e+00 -3.79675001e-01 8.43592882e-01
3.34789127e-01 2.74034441e-02 7.88631618e-01 2.93641716e-01
6.19798228e-02 7.24694073e-01 3.21956158e-01 3.62008959e-01
7.81223118e-01 -1.00969315e+00 8.15588295e-01 3.96688163e-01
3.89204949e-01 7.06895590e-01 3.30996752e-01 -4.07587737e-01
-1.78596318e+00 -1.01470661e+00 5.90341508e-01 -7.64629543e-02
7.75789917e-01 -4.10672188e-01 -8.21401238e-01 4.16262031e-01
-1.99085966e-01 7.65154421e-01 4.74499345e-01 2.09575534e-01
-7.72532701e-01 -2.36879215e-01 -1.16860068e+00 4.88337576e-01
1.27562618e+00 -8.04801404e-01 -2.39959866e-01 5.75593114e-01
1.97713897e-01 -5.43592393e-01 -8.65025759e-01 2.30103150e-01
8.65897894e-01 -9.97265041e-01 1.13937306e+00 -7.14769721e-01
3.64949226e-01 -1.90484583e-01 -1.85115933e-01 -1.15671360e+00
-4.68986928e-01 -6.50709867e-01 1.40657440e-01 4.63307142e-01
1.66227892e-01 -6.45450652e-01 1.10154057e+00 2.54560858e-01
-9.98258665e-02 -1.01091528e+00 -1.07881904e+00 -1.02335715e+00
1.73130512e-01 -8.85572806e-02 5.84028661e-01 8.35184395e-01
-3.76204699e-01 -5.90542331e-02 -1.17977895e-01 -1.42638283e-02
6.96814835e-01 5.37235200e-01 9.94351089e-01 -1.37414837e+00
-2.52361178e-01 -5.50452292e-01 -1.09022701e+00 -1.33168769e+00
8.00501630e-02 -8.04694295e-01 4.63917516e-02 -1.09065509e+00
-1.67339414e-01 -7.46015966e-01 -9.07401815e-02 7.50789702e-01
2.69245595e-01 7.88191020e-01 -2.06720710e-01 4.13078427e-01
-4.92589980e-01 5.55646837e-01 1.10088336e+00 -3.28270853e-01
-9.46565568e-02 -2.66789019e-01 -4.35971409e-01 5.43938458e-01
6.98984563e-01 -5.31999290e-01 -2.55438555e-02 -5.39409280e-01
2.72394419e-01 -3.23260546e-01 4.93000597e-01 -1.14273763e+00
3.01653802e-01 -5.44947051e-02 8.10037374e-01 -3.76492113e-01
4.12195951e-01 -7.12525904e-01 -1.74800474e-02 3.21248114e-01
-3.43982160e-01 3.23854566e-01 1.69002384e-01 4.10413384e-01
-3.05719972e-01 -2.85693370e-02 8.49225402e-01 -7.72723034e-02
-3.88696849e-01 7.81126559e-01 1.57346085e-01 1.32080644e-01
9.08024251e-01 -5.52853107e-01 -2.25652561e-01 -2.75019884e-01
-3.57184112e-01 -1.73000112e-01 9.24774528e-01 3.01802874e-01
6.72845721e-01 -1.64772606e+00 -6.15681052e-01 7.04601884e-01
1.04046322e-01 2.11824536e-01 3.13914046e-02 9.16934311e-01
-9.20228422e-01 1.36462584e-01 -5.47578514e-01 -8.42207789e-01
-9.85849917e-01 3.38130385e-01 2.66562223e-01 7.79708475e-02
-8.22036684e-01 8.37638557e-01 -5.93944490e-02 -2.31195569e-01
2.79224455e-01 -1.44368604e-01 4.91911590e-01 -1.41155466e-01
5.67606091e-01 1.54511258e-01 3.76750648e-01 -5.84515035e-01
-5.71639061e-01 9.27161217e-01 -1.48280293e-01 3.23281944e-01
1.52634132e+00 4.60800469e-01 -2.47108459e-01 -1.32651553e-01
1.91112161e+00 -2.26006564e-02 -1.48593497e+00 -2.50966787e-01
-2.50038594e-01 -9.20177042e-01 5.31913564e-02 -2.57482469e-01
-1.20729649e+00 7.30571985e-01 5.25610507e-01 2.35034332e-01
9.61969793e-01 5.70537057e-03 9.38123882e-01 7.47259974e-01
4.12506819e-01 -1.01356566e+00 1.43887714e-01 4.46951449e-01
8.24253321e-01 -1.31736267e+00 2.82005072e-01 -7.42563382e-02
-7.27270022e-02 1.54543436e+00 4.46037263e-01 -1.07655942e+00
8.07911336e-01 1.79995060e-01 -1.98563680e-01 -1.79140776e-01
-5.80896974e-01 -2.26873830e-01 4.82270658e-01 6.46443129e-01
2.15979800e-01 -1.24451116e-01 -6.92768246e-02 -2.96040565e-01
-1.22463435e-01 -2.90609032e-01 -4.56837788e-02 1.28243208e+00
-2.60327250e-01 -1.24756539e+00 -3.24019253e-01 8.68681550e-01
-2.44699672e-01 -1.97123960e-01 -3.44661593e-01 7.93067336e-01
2.10305661e-01 3.02760690e-01 4.73271817e-01 -4.51710224e-01
2.40005374e-01 -4.17931199e-01 7.06341565e-01 -3.59521538e-01
-5.91497302e-01 -7.12614600e-03 -6.49087429e-02 -9.71431792e-01
-3.82248968e-01 -6.96402371e-01 -6.83497310e-01 -5.09204328e-01
-1.25862181e-01 -1.89787775e-01 6.84399903e-01 5.95838666e-01
7.15641201e-01 -2.32258946e-01 7.02215731e-01 -1.32446587e+00
-7.68931568e-01 -5.54332078e-01 -3.37911546e-01 5.71927369e-01
5.98393500e-01 -7.36524999e-01 -4.51629728e-01 -2.50066936e-01] | [8.24223518371582, -1.8984318971633911] |
a5af84ef-57c6-48f1-8de4-d9219d44f256 | point2mask-a-weakly-supervised-approach-for | null | null | https://link.springer.com/chapter/10.1007/978-3-031-12053-4_11 | https://www.researchgate.net/publication/361820249_Point2Mask_A_Weakly_Supervised_Approach_for_Cell_Segmentation_using_Point_Annotation | Point2Mask: A Weakly Supervised Approach for Cell Segmentation Using Point Annotation | Identifying cells in microscopic images is a crucial step toward studying image-based cell biology research. Cell instance segmentation provides an opportunity to study the shape, structure, form, and size of cells. Deep learning approaches for cell instance segmentation rely on the instance segmentation mask for each cell, which is a labor-intensive and expensive task. An ample amount of unlabeled microscopic data is available in the cell biology domain, but due to the tedious and exorbitant nature of the annotations needed for the cell instance segmentation approaches, the full potential of the data is not explored. This paper presents a weakly supervised approach, which can perform cell instance segmentation by using only point and bounding box-based annotation. This enormously reduces the annotation efforts. The proposed approach is evaluated on a benchmark dataset i.e., LIVECell, whereby only using a bounding box and randomly generated points on each cell, it achieved the mean average precision score of 43.53% which is as good as the full supervised segmentation method trained with complete segmentation mask. In addition, it is 3.71 times faster to annotate with a bounding box and point in comparison to full mask annotation. | ['Andreas Dengel & Sheraz Ahmed', 'Rickard Sjögren', 'Johan Trygg', 'Timothy R Jackson', 'Christoffer Edlund', 'Mohsin Munir', 'Mohammadmahdi Koochali', 'Fabian Schmeisser', 'Nabeel Khalid'] | 2022-07-25 | null | null | null | miua-2022-7 | ['cell-segmentation'] | ['medical'] | [ 8.24366435e-02 1.65003881e-01 1.01256154e-01 -7.15884641e-02
-8.65990460e-01 -7.01292872e-01 3.58280629e-01 7.01063991e-01
-7.92539418e-01 9.30622280e-01 -6.74016356e-01 -8.73271152e-02
5.33317208e-01 -7.44696975e-01 -7.01991558e-01 -1.14131773e+00
2.20893830e-01 7.57631660e-01 4.09472704e-01 2.43072286e-01
4.28609252e-01 5.92829108e-01 -1.23538756e+00 2.66112257e-02
7.14811444e-01 1.00141990e+00 3.08232009e-01 7.44347811e-01
-3.85636389e-01 3.49458039e-01 -5.94693124e-01 -1.74377441e-01
1.02810383e-01 -2.96228290e-01 -9.04416859e-01 3.43777448e-01
2.87663221e-01 -1.64526299e-01 2.66665757e-01 8.13320994e-01
3.66068155e-01 -2.26964787e-01 7.28761494e-01 -8.52022767e-01
-4.75000143e-02 1.41716614e-01 -6.03851736e-01 2.54600048e-01
-1.24287613e-01 8.94221738e-02 6.93872452e-01 -8.24822843e-01
8.47566545e-01 5.82754135e-01 3.93941939e-01 5.32322407e-01
-1.31121314e+00 -4.75072473e-01 -1.28685012e-01 -3.01372617e-01
-1.54635274e+00 -2.88945377e-01 4.41364944e-01 -6.89976275e-01
6.46959126e-01 8.53618085e-02 7.28261054e-01 4.13695753e-01
1.73510909e-01 6.66775405e-01 1.20422196e+00 -3.36308777e-01
6.35565937e-01 1.78244621e-01 2.67796397e-01 6.96045578e-01
2.52273530e-01 -3.56180310e-01 5.10003138e-03 1.51519459e-02
9.37589824e-01 1.52482823e-01 -1.15913004e-01 -1.07783809e-01
-1.34865284e+00 4.60777104e-01 2.30898589e-01 6.01774693e-01
-1.38465650e-02 1.69256061e-01 3.85350138e-01 -1.28962070e-01
4.61270511e-01 4.46806461e-01 -6.02789938e-01 -1.11735575e-01
-1.11281276e+00 9.01583731e-02 7.37368762e-01 7.68681884e-01
9.40276504e-01 -2.45562524e-01 -8.00196901e-02 6.27121210e-01
-8.37773755e-02 2.90229201e-01 3.12740207e-01 -1.01779222e+00
-2.53651589e-02 1.09287369e+00 1.34940013e-01 -6.79488480e-01
-5.74941278e-01 -3.02812457e-01 -8.13323438e-01 3.12074780e-01
1.20332611e+00 -1.27737507e-01 -1.09700286e+00 1.29765248e+00
6.69666409e-01 7.33944550e-02 -3.17375094e-01 9.10885394e-01
8.29890668e-01 5.45819640e-01 7.88265839e-02 -2.33477756e-01
1.41362655e+00 -6.78440511e-01 -5.63698947e-01 1.62108950e-02
1.06393027e+00 -6.22779250e-01 1.07469428e+00 3.02284241e-01
-8.90831470e-01 -3.14521730e-01 -9.50268447e-01 -2.29903236e-01
-6.61778271e-01 3.24134558e-01 6.44768953e-01 4.97482896e-01
-8.94545972e-01 3.86187792e-01 -9.96536911e-01 -4.11310464e-01
9.12046909e-01 6.30935907e-01 -6.72809184e-01 2.08618671e-01
-3.29638153e-01 4.00030524e-01 2.94278592e-01 -1.08705238e-01
-7.63748407e-01 -6.79102063e-01 -6.99372053e-01 1.68228120e-01
2.63563275e-01 -3.23046982e-01 8.10481846e-01 -5.56320965e-01
-1.39365101e+00 1.34600413e+00 -3.74575049e-01 -4.04845536e-01
5.57808340e-01 2.35093802e-01 4.19253081e-01 3.18764061e-01
-1.03383074e-02 9.80758429e-01 2.60040283e-01 -1.38130391e+00
-6.09227121e-01 -5.13554811e-01 -1.19185895e-01 -1.85135484e-01
-9.47536901e-02 -2.72528291e-01 -6.36161804e-01 -3.89451683e-01
3.21449898e-02 -8.72596979e-01 -3.14050585e-01 5.37070446e-02
-4.23932493e-01 -5.80711514e-02 9.44988310e-01 -5.93987644e-01
8.80643427e-01 -2.01809788e+00 -1.67261451e-01 6.35283664e-02
3.58942151e-01 2.55960107e-01 2.83149093e-01 1.77462950e-01
3.27475369e-01 4.12127733e-01 -5.73914170e-01 -4.23316270e-01
-3.66196990e-01 -8.17506667e-03 1.20984897e-01 6.95953608e-01
7.68775940e-02 9.93340552e-01 -6.93411767e-01 -8.36137056e-01
2.42183909e-01 5.19123018e-01 -4.50899392e-01 5.89933395e-02
-2.10368171e-01 8.08575392e-01 -2.52916574e-01 1.09733784e+00
5.90516448e-01 -5.05692184e-01 6.64583296e-02 -1.14939019e-01
-1.49310619e-01 -2.54204690e-01 -8.65017295e-01 1.38427818e+00
-1.95652932e-01 6.99310958e-01 2.52913177e-01 -1.19609845e+00
9.27508473e-01 2.86470741e-01 7.01000571e-01 -2.83741653e-01
4.14501905e-01 4.86283690e-01 -1.07976153e-01 -2.58543313e-01
-6.07817546e-02 -2.13731930e-01 -1.43088568e-02 4.26932156e-01
3.61616686e-02 -3.09091240e-01 5.30568480e-01 9.45687443e-02
1.17151690e+00 4.03669626e-02 2.69662142e-01 -5.14036953e-01
7.63355672e-01 1.66782439e-01 6.08009994e-01 4.16004449e-01
-3.92544478e-01 7.75533736e-01 5.72273910e-01 -6.72936976e-01
-1.09779119e+00 -6.10661447e-01 -3.33986908e-01 6.38834059e-01
3.90944570e-01 8.73140339e-03 -1.27805483e+00 -6.21486127e-01
-1.97289512e-01 -7.87698776e-02 -6.22999549e-01 4.21428174e-01
-6.08632505e-01 -8.53046477e-01 4.08890098e-01 4.04449701e-01
5.01153827e-01 -9.54474628e-01 -6.02923036e-01 1.70644432e-01
-7.50978887e-02 -1.32064474e+00 -2.67692864e-01 2.29953110e-01
-9.80585456e-01 -1.18758893e+00 -9.03838456e-01 -9.99967158e-01
1.24629927e+00 -1.96675267e-02 6.91657841e-01 5.11996329e-01
-5.98688602e-01 -5.97043484e-02 -1.28839985e-01 -5.10692656e-01
-5.07997237e-02 2.33582243e-01 -3.57696176e-01 -6.70454055e-02
3.24682832e-01 -2.79565811e-01 -8.12465012e-01 3.12588125e-01
-9.35281277e-01 -1.13670370e-02 5.28332889e-01 1.01534557e+00
1.28394461e+00 -4.99969311e-02 5.56372166e-01 -1.25690722e+00
-4.35175654e-03 -1.90141544e-01 -7.34909594e-01 1.00692078e-01
-3.11504811e-01 -3.87785286e-01 8.14991236e-01 -3.17305267e-01
-7.25795329e-01 5.12642503e-01 -1.08851604e-01 4.36904430e-02
-5.60865879e-01 2.48711705e-01 -6.10268861e-02 -2.74000376e-01
3.12143207e-01 1.56051382e-01 3.05957466e-01 -3.59048933e-01
-2.51606077e-01 5.26143253e-01 4.46886688e-01 -4.68983859e-01
3.72581750e-01 1.12802625e+00 3.04751605e-01 -8.76729131e-01
-6.33515298e-01 -7.14065313e-01 -9.02655900e-01 -2.05262825e-01
9.21068013e-01 -5.16378164e-01 -8.71019542e-01 6.27184749e-01
-1.09589970e+00 -6.90989614e-01 -1.91533402e-01 2.49579042e-01
-5.71942687e-01 5.11361241e-01 -8.12740445e-01 -6.29076362e-01
-3.17948073e-01 -1.20274138e+00 1.36444986e+00 2.71972835e-01
-1.68531612e-01 -1.07717347e+00 -7.43787512e-02 5.48472166e-01
9.36946198e-02 5.35207152e-01 8.81733537e-01 -8.01663280e-01
-8.01228166e-01 -6.82934999e-01 -2.52900749e-01 -1.11411318e-01
1.79568291e-01 1.84201509e-01 -1.02295399e+00 -2.30538800e-01
-1.39249310e-01 -3.18393320e-01 7.28285074e-01 6.08850360e-01
1.39521003e+00 -3.19285638e-04 -6.30577147e-01 5.38643122e-01
1.50071621e+00 2.31894806e-01 5.23712993e-01 1.70659959e-01
5.88667810e-01 7.32917786e-01 6.18546069e-01 3.88593018e-01
1.42752379e-01 4.67702657e-01 3.56862307e-01 -5.03903985e-01
5.15089035e-02 2.83958793e-01 -2.85039514e-01 4.22221303e-01
-1.54003009e-01 -3.28558564e-01 -1.16548967e+00 6.74852252e-01
-1.51164901e+00 -6.23221993e-01 -2.13041410e-01 2.12894392e+00
7.90793180e-01 8.61965567e-02 1.28084049e-01 4.10224855e-01
7.84812689e-01 -4.14928228e-01 -5.50063729e-01 -1.88847110e-01
1.69628467e-02 2.06882149e-01 4.29684818e-01 5.37282348e-01
-1.14325213e+00 9.87071991e-01 6.09175539e+00 9.96189654e-01
-1.19470727e+00 -1.74718335e-01 1.30128980e+00 5.38647324e-02
2.28604972e-01 -3.52392569e-02 -1.00553870e+00 7.12406695e-01
4.97593850e-01 1.88178807e-01 6.09688088e-02 5.49904227e-01
2.62943059e-01 -5.56879759e-01 -1.13330972e+00 9.28045869e-01
-3.27617258e-01 -1.51186228e+00 -2.50422489e-02 5.44825494e-01
7.09304690e-01 -2.03416854e-01 -2.07434133e-01 9.19317082e-02
-2.28443995e-01 -1.07598269e+00 3.38581681e-01 3.52174103e-01
8.84622574e-01 -6.68085277e-01 1.27437246e+00 6.73032701e-01
-1.06260014e+00 2.02749282e-01 -3.58397782e-01 -9.05600116e-02
1.81146145e-01 8.69472623e-01 -8.66819739e-01 -6.10083751e-02
6.18612766e-01 2.85829425e-01 -3.89702260e-01 1.21154308e+00
4.14202332e-01 7.05922663e-01 -5.89384437e-01 -2.31033396e-02
4.18100432e-02 -3.23430955e-01 2.03522816e-01 1.35551369e+00
2.61671364e-01 1.41649246e-01 3.10904205e-01 8.44949782e-01
-2.55189329e-01 2.14119434e-01 -3.83873254e-01 -1.39030352e-01
3.92335832e-01 1.65841782e+00 -1.66643381e+00 -4.54691172e-01
-7.55448863e-02 7.00820684e-01 3.81634474e-01 2.48194858e-01
-7.62356639e-01 -4.90988016e-01 1.22377299e-01 4.19235885e-01
2.58638144e-01 -1.42871603e-01 -7.36958027e-01 -5.63971400e-01
-1.70006230e-01 -3.01195920e-01 1.78543448e-01 -2.43819222e-01
-1.04578221e+00 2.53461093e-01 -3.69927973e-01 -9.18454230e-01
2.09706932e-01 -7.85442233e-01 -5.96033692e-01 6.36865139e-01
-1.35866976e+00 -1.07677495e+00 -5.95956683e-01 3.52713577e-02
4.75922495e-01 2.96671510e-01 7.29087591e-01 3.19606572e-01
-7.39192307e-01 4.06737566e-01 6.33906052e-02 3.66005272e-01
3.61614168e-01 -1.36325252e+00 -1.39024360e-02 3.26091468e-01
-3.08538292e-04 6.64675355e-01 5.19624829e-01 -4.75931674e-01
-1.10326135e+00 -9.66489732e-01 8.86086404e-01 -3.83616477e-01
2.81027853e-01 -4.56252635e-01 -1.00500357e+00 2.61247069e-01
-6.47707507e-02 4.23458934e-01 9.98162985e-01 -3.83314639e-01
2.99853861e-01 -8.88958052e-02 -1.37007904e+00 6.41574323e-01
6.04257762e-01 -2.19401196e-01 4.14077975e-02 3.89081568e-01
3.11249554e-01 -5.05662322e-01 -9.53422546e-01 4.58662063e-01
4.23552006e-01 -9.35143769e-01 7.22279966e-01 2.05626749e-02
3.16382974e-01 -4.77825612e-01 3.43540132e-01 -6.59868717e-01
-5.19091599e-02 -3.60450715e-01 -4.34639715e-02 1.36907530e+00
5.96105754e-01 -4.14259017e-01 1.26145899e+00 6.42430067e-01
-1.17259204e-01 -1.39854550e+00 -9.48284984e-01 -6.17627740e-01
3.15850645e-01 1.51012419e-02 4.66047227e-01 8.26327801e-01
1.27509788e-01 5.44603989e-02 3.76210272e-01 -2.83543319e-01
5.49363196e-01 1.40978038e-01 9.58658457e-01 -1.21179616e+00
-3.27387415e-02 -5.47563791e-01 -4.83912349e-01 -1.09514070e+00
2.49082167e-02 -7.31738508e-01 9.53606516e-02 -1.50623202e+00
2.73435980e-01 -4.67155576e-01 -1.57136351e-01 3.76059562e-01
-2.01411664e-01 6.68390870e-01 -1.96311519e-01 3.24806184e-01
-6.16862833e-01 5.90319224e-02 1.38196647e+00 -2.79981315e-01
-1.60593376e-01 -1.94945291e-01 -3.11900079e-01 8.47334504e-01
1.05146813e+00 -2.40514666e-01 -1.39766857e-01 -1.69353709e-01
-8.42416510e-02 -1.10074095e-01 3.41416925e-01 -1.05801785e+00
4.50035632e-01 -1.13513932e-01 5.26425183e-01 -7.12491870e-01
4.71983194e-01 -7.22869217e-01 1.69797812e-03 4.36131179e-01
-2.15780064e-01 -5.03309846e-01 1.58975661e-01 5.86474121e-01
-3.12891930e-01 -1.84339702e-01 1.03708863e+00 -4.53093737e-01
-2.78101683e-01 2.84401983e-01 -4.75149184e-01 6.97745942e-03
1.40690649e+00 -8.30206573e-01 -2.59826452e-01 1.35511756e-01
-6.29836440e-01 1.39924675e-01 8.13409448e-01 -5.52538931e-01
3.84892792e-01 -8.71903658e-01 -2.93986082e-01 9.81752425e-02
4.50294912e-02 5.30680954e-01 1.61057010e-01 1.07688689e+00
-1.00771070e+00 4.42705154e-01 -1.02576680e-01 -9.17118251e-01
-1.24676621e+00 4.74603057e-01 2.57987767e-01 -3.97400558e-01
-4.44257677e-01 1.00049758e+00 4.23971504e-01 -2.90428221e-01
3.34598683e-02 -4.57977831e-01 -3.33385259e-01 -7.10813776e-02
3.51477712e-01 5.17806172e-01 -7.07595125e-02 -6.82841837e-01
-2.02206284e-01 8.85417700e-01 -1.21470422e-01 6.39596209e-02
1.04916513e+00 -1.19759083e-01 -3.66495788e-01 5.20694137e-01
1.07554865e+00 -9.75110754e-02 -1.36725903e+00 2.33609661e-01
1.43722892e-01 -2.72981763e-01 2.84889713e-02 -5.08412421e-01
-1.09723175e+00 1.06512189e+00 2.06724197e-01 3.28741729e-01
7.79922247e-01 6.82661608e-02 1.07573497e+00 3.14478785e-01
4.92924571e-01 -1.19986379e+00 -1.35571927e-01 2.95466691e-01
3.54426831e-01 -1.49481893e+00 3.73189449e-02 -7.64384568e-01
-2.71836698e-01 9.25324798e-01 7.78758109e-01 -1.35106176e-01
6.58157170e-01 4.94730324e-01 1.21386468e-01 -3.23807627e-01
-5.19132733e-01 -8.42271671e-02 -1.29807815e-01 4.57848012e-01
7.51994014e-01 -1.33208498e-01 -5.14124930e-01 4.76907700e-01
1.65726706e-01 8.87960494e-02 3.40629756e-01 9.35746014e-01
-7.06978977e-01 -9.51854885e-01 -2.89377123e-01 6.11984015e-01
-9.54432249e-01 1.84863657e-01 -5.86644769e-01 7.46094465e-01
2.35596359e-01 7.86830723e-01 2.29106516e-01 9.95536968e-02
-2.50668406e-01 -1.20407961e-01 3.58719140e-01 -7.40292490e-01
-4.07706439e-01 2.55747885e-01 -3.88800651e-01 -3.64661038e-01
-3.68540734e-01 -3.63154709e-01 -2.05258751e+00 -8.79161656e-02
-4.21760559e-01 1.62419021e-01 5.93266428e-01 1.02880263e+00
3.17533165e-01 1.03720210e-01 3.72810572e-01 -9.21216786e-01
1.17712386e-01 -7.17165172e-01 -6.93115175e-01 4.03222829e-01
2.38370523e-01 -5.38122058e-01 -4.48012054e-01 5.89722395e-01] | [14.588215827941895, -3.1042420864105225] |
f5e59f77-66c6-4eba-99ec-7a1f893acd0c | burst-a-benchmark-for-unifying-object | 2209.12118 | null | https://arxiv.org/abs/2209.12118v2 | https://arxiv.org/pdf/2209.12118v2.pdf | BURST: A Benchmark for Unifying Object Recognition, Segmentation and Tracking in Video | Multiple existing benchmarks involve tracking and segmenting objects in video e.g., Video Object Segmentation (VOS) and Multi-Object Tracking and Segmentation (MOTS), but there is little interaction between them due to the use of disparate benchmark datasets and metrics (e.g. J&F, mAP, sMOTSA). As a result, published works usually target a particular benchmark, and are not easily comparable to each another. We believe that the development of generalized methods that can tackle multiple tasks requires greater cohesion among these research sub-communities. In this paper, we aim to facilitate this by proposing BURST, a dataset which contains thousands of diverse videos with high-quality object masks, and an associated benchmark with six tasks involving object tracking and segmentation in video. All tasks are evaluated using the same data and comparable metrics, which enables researchers to consider them in unison, and hence, more effectively pool knowledge from different methods across different tasks. Additionally, we demonstrate several baselines for all tasks and show that approaches for one task can be applied to another with a quantifiable and explainable performance difference. Dataset annotations and evaluation code is available at: https://github.com/Ali2500/BURST-benchmark. | ['Deva Ramanan', 'Bastian Leibe', 'Achal Dave', 'Tarasha Khurana', 'Paul Voigtlaender', 'Jonathon Luiten', 'Ali Athar'] | 2022-09-25 | null | null | null | null | ['multi-object-tracking-and-segmentation'] | ['computer-vision'] | [ 7.42690414e-02 -5.85583270e-01 -4.35013205e-01 -1.98568314e-01
-7.96195865e-01 -8.09813261e-01 4.04711902e-01 -1.15105249e-01
-4.28444624e-01 6.07511342e-01 -2.59295404e-02 1.02434553e-01
-5.36507405e-02 -1.98306710e-01 -8.43784034e-01 -5.14837742e-01
-5.62571585e-02 2.97430724e-01 9.99595225e-01 5.44913150e-02
1.68629587e-01 2.24346578e-01 -1.66169322e+00 4.61186945e-01
5.25658369e-01 9.89585996e-01 3.24465692e-01 7.53219962e-01
6.89025521e-02 8.24696422e-01 -6.32193804e-01 -5.81234515e-01
4.59573209e-01 -2.96027899e-01 -1.12919807e+00 1.13909058e-02
9.27395761e-01 -2.68939108e-01 -1.33686274e-01 1.02430212e+00
4.49089140e-01 4.56551574e-02 4.44822699e-01 -1.80389619e+00
-4.83420640e-01 5.57267785e-01 -6.06466532e-01 5.58661699e-01
3.41667950e-01 4.46083844e-01 9.71436441e-01 -5.32664835e-01
7.92353451e-01 1.20999992e+00 9.47094083e-01 6.34028554e-01
-1.20152867e+00 -6.12139523e-01 3.11790675e-01 1.98544472e-01
-1.31250656e+00 -5.55579424e-01 2.50249207e-01 -6.50932133e-01
7.29699910e-01 5.38483024e-01 5.61998665e-01 1.30181587e+00
-1.54919093e-02 1.13302004e+00 1.03480887e+00 3.38108428e-02
-1.31249771e-01 7.79175982e-02 3.00540090e-01 5.06401718e-01
4.54633623e-01 -1.10876463e-01 -4.73712057e-01 1.44986227e-01
3.62637609e-01 2.19284054e-02 -4.08333361e-01 -4.15422231e-01
-1.59144521e+00 4.01690543e-01 1.82756916e-01 3.32039446e-01
-9.49779004e-02 3.20134819e-01 7.11972773e-01 1.22640230e-01
2.30305329e-01 3.39607805e-01 -5.19687414e-01 -3.67180139e-01
-1.20717251e+00 5.62698722e-01 8.46941710e-01 1.26360750e+00
5.65073729e-01 -9.21651572e-02 -4.97736067e-01 5.01755714e-01
3.60221982e-01 3.23071182e-01 3.57222021e-01 -1.23509407e+00
4.48361695e-01 3.00368547e-01 2.59842902e-01 -9.14521337e-01
-4.03176963e-01 -8.50660652e-02 -3.62617701e-01 1.49828285e-01
6.64274752e-01 -1.59384534e-01 -1.00505853e+00 1.75525582e+00
4.45159972e-01 6.33639455e-01 -1.01336539e-01 1.05511630e+00
1.20635045e+00 4.16684777e-01 3.18556160e-01 1.42868564e-01
1.56810677e+00 -1.47515023e+00 -6.02506757e-01 -3.01940054e-01
5.26643813e-01 -8.61772180e-01 1.16288459e+00 3.13481241e-01
-1.18453467e+00 -7.41959155e-01 -8.84863257e-01 -5.13278507e-02
-5.43528438e-01 -1.21012710e-01 4.68845487e-01 5.12099981e-01
-1.07146025e+00 5.75665534e-01 -9.39779043e-01 -5.80810726e-01
4.99691576e-01 1.50106713e-01 -1.11972049e-01 4.18560766e-02
-9.97942805e-01 7.80600071e-01 5.47686994e-01 -7.81277716e-02
-1.08876717e+00 -8.29394221e-01 -5.38597703e-01 -2.73388505e-01
6.33591890e-01 -6.93035305e-01 1.38185716e+00 -9.10438657e-01
-9.22176063e-01 9.47568238e-01 1.37755141e-01 -3.56170475e-01
7.53888667e-01 -4.04832691e-01 -3.22281450e-01 2.15042189e-01
4.41672027e-01 1.03406250e+00 6.84138000e-01 -1.43350327e+00
-9.29654598e-01 -2.86053754e-02 2.66540796e-01 8.02981779e-02
-1.26961827e-01 4.41066712e-01 -1.05252635e+00 -6.81266785e-01
-4.92748380e-01 -9.66052473e-01 -4.28429805e-02 -1.40182257e-01
-5.41939437e-01 -1.67206109e-01 1.14415753e+00 -5.40599167e-01
1.32762361e+00 -2.11515927e+00 2.39608645e-01 -5.14424860e-01
4.08319026e-01 3.70498002e-01 -3.27630013e-01 2.03797713e-01
2.58474708e-01 4.31196004e-01 -2.47428805e-01 -5.19050598e-01
-3.95887159e-02 1.97362378e-01 3.56046818e-02 4.80974555e-01
1.91643551e-01 1.05638337e+00 -1.00583136e+00 -8.25059295e-01
1.62018672e-01 2.56830007e-01 -4.09657806e-01 6.64943978e-02
-4.27838892e-01 5.59762001e-01 -4.85857517e-01 1.02797735e+00
4.86679763e-01 -5.03027916e-01 -1.60849020e-01 -3.44795913e-01
-1.00126661e-01 -7.93278664e-02 -1.41811669e+00 1.71610963e+00
1.71221327e-02 7.35106587e-01 8.37815478e-02 -8.28458726e-01
2.36469120e-01 3.10351878e-01 7.87507594e-01 -5.90710819e-01
1.37321055e-01 1.38119444e-01 -5.13401255e-02 -6.40798151e-01
6.40766799e-01 4.17718679e-01 3.26290764e-02 2.95951128e-01
1.64250001e-01 7.72814825e-02 8.34522724e-01 2.50115514e-01
1.15407419e+00 3.49502444e-01 5.09364754e-02 -2.08520055e-01
2.28991002e-01 2.31256634e-01 7.46244788e-01 1.03279781e+00
-6.85576320e-01 7.37910509e-01 2.78713316e-01 -3.34186375e-01
-8.95904303e-01 -1.07289112e+00 -2.13392958e-01 1.31044376e+00
6.94030762e-01 -5.40718794e-01 -8.65971029e-01 -7.33921826e-01
2.77965963e-01 2.44781956e-01 -5.91463506e-01 2.61832356e-01
-6.07583821e-01 -7.07531393e-01 8.03860128e-01 6.01351023e-01
5.07835507e-01 -1.22143412e+00 -8.02632987e-01 1.15044318e-01
-3.63058895e-01 -1.58971107e+00 -4.97653782e-01 -1.38119951e-01
-8.18621099e-01 -1.31203973e+00 -8.40333223e-01 -6.61037385e-01
1.34624183e-01 6.10748291e-01 1.55182314e+00 2.79448241e-01
-3.05488229e-01 7.19565749e-01 -4.93593305e-01 -4.18233067e-01
-3.60658735e-01 2.82330781e-01 -7.03914911e-02 -1.46046281e-01
3.20732802e-01 -2.12747112e-01 -7.10972250e-01 8.18262756e-01
-1.01913667e+00 1.24415413e-01 4.95169759e-01 3.07287425e-01
6.52859986e-01 -4.43773568e-01 5.09623647e-01 -7.18283474e-01
3.24303687e-01 -8.01459432e-01 -5.38018346e-01 3.78970504e-01
-3.62965196e-01 -4.85656202e-01 3.13113064e-01 -5.35258293e-01
-7.24099636e-01 -9.18219388e-02 1.20899066e-01 -5.55990458e-01
-4.32060093e-01 1.86455980e-01 -5.65302521e-02 -1.53556392e-01
5.02823412e-01 9.60434824e-02 -2.07008362e-01 -5.24745226e-01
4.15353328e-01 4.39352512e-01 5.65549672e-01 -6.01796329e-01
6.59995556e-01 5.22433162e-01 -3.72978151e-01 -5.53636074e-01
-9.62625027e-01 -7.29832709e-01 -5.89701235e-01 -4.82573301e-01
1.18589246e+00 -1.01675975e+00 -5.68077683e-01 5.78082621e-01
-1.11139321e+00 -5.95645607e-01 -2.34445006e-01 4.95346963e-01
-5.22277117e-01 3.74125391e-01 -6.59379363e-01 -3.96577626e-01
-1.55765237e-02 -1.49122763e+00 1.33014059e+00 3.82033765e-01
-2.12219417e-01 -9.25982654e-01 -6.83486909e-02 7.02893496e-01
4.34859902e-01 4.94866937e-01 9.78888944e-02 -6.79910421e-01
-1.00424719e+00 8.94051492e-02 -4.26691204e-01 3.63736957e-01
8.10396522e-02 3.56517971e-01 -8.95268381e-01 -4.61209953e-01
-2.14526430e-01 -5.38228452e-01 1.03032482e+00 4.90991861e-01
1.27789474e+00 -3.21583375e-02 -6.82713687e-01 7.17109323e-01
1.29059553e+00 1.88477531e-01 5.76758742e-01 5.39997637e-01
9.30534244e-01 4.65309352e-01 7.54262209e-01 8.76241401e-02
5.88570893e-01 9.27200556e-01 4.23189729e-01 -6.23936281e-02
-4.56191897e-01 3.30099314e-01 3.32193226e-01 7.77071476e-01
-3.35209072e-01 -5.31969547e-01 -1.08753073e+00 7.34197795e-01
-1.97750854e+00 -1.08228838e+00 -3.42012852e-01 1.82521081e+00
6.78668618e-01 1.13986380e-01 5.70001185e-01 -2.28338122e-01
8.65234733e-01 2.92687118e-01 -5.57182550e-01 1.36563897e-01
-1.60369147e-02 -2.55834401e-01 6.31693542e-01 9.06578824e-02
-1.43170500e+00 9.70004618e-01 6.44414330e+00 9.24147487e-01
-9.69332278e-01 3.50201696e-01 5.83117485e-01 -3.02970797e-01
1.78911731e-01 -1.84347585e-01 -1.03333139e+00 8.30944180e-01
9.32814181e-01 -5.12983389e-02 3.31752747e-01 6.81419253e-01
8.00581947e-02 -6.92718178e-02 -1.27680838e+00 9.50089216e-01
-4.86756749e-02 -1.28584409e+00 -1.25286266e-01 -2.10808530e-01
8.47311139e-01 4.34789240e-01 -6.19864836e-02 3.09179574e-01
1.35040864e-01 -8.22017491e-01 1.17270827e+00 4.52720284e-01
4.76513952e-01 -3.94843780e-02 5.33351958e-01 -3.18152681e-02
-1.44597161e+00 -5.61633194e-03 -3.75724840e-03 2.32506588e-01
3.80261451e-01 1.11452512e-01 -3.06418449e-01 7.17105925e-01
1.29447496e+00 8.66032898e-01 -8.90896976e-01 1.46620059e+00
2.07990795e-01 6.64142013e-01 -2.22584531e-01 5.56616411e-02
3.44302714e-01 -1.21140480e-01 6.36778057e-01 1.63872755e+00
2.06796095e-01 -2.93487519e-01 5.96430957e-01 7.42190361e-01
-1.70053184e-01 -9.85466223e-03 -4.18774843e-01 -5.71484305e-02
4.53797162e-01 1.46599829e+00 -1.16706455e+00 -4.90441382e-01
-8.87065947e-01 7.35251069e-01 -1.11020962e-02 3.05291742e-01
-1.43168521e+00 -9.74172875e-02 1.00586820e+00 1.03065193e-01
4.75316316e-01 -8.58129337e-02 -8.07522312e-02 -1.06944716e+00
8.03970769e-02 -1.02276981e+00 5.13610601e-01 -6.93632901e-01
-1.32761705e+00 4.90826637e-01 3.90753657e-01 -1.26037943e+00
1.42103925e-01 -4.70968276e-01 -4.39061493e-01 3.63384575e-01
-1.63073552e+00 -1.06050122e+00 -6.68265939e-01 5.72799087e-01
6.03649080e-01 2.43690647e-02 1.85504690e-01 8.80755544e-01
-8.91963065e-01 4.61405188e-01 -3.81978638e-02 3.27248603e-01
9.03330684e-01 -1.24255586e+00 6.38646185e-01 8.57682109e-01
2.64096111e-01 3.28130126e-01 6.00328147e-01 -6.35087848e-01
-1.37155437e+00 -1.43335223e+00 2.29579777e-01 -9.30644214e-01
8.30864787e-01 -3.32360327e-01 -9.84411716e-01 8.11641872e-01
3.76969665e-01 3.25637430e-01 4.07018065e-01 -1.51582807e-01
-2.26968437e-01 6.92133829e-02 -9.21661675e-01 5.52730501e-01
1.48527598e+00 -1.15687251e-01 -3.19829553e-01 4.39012617e-01
7.85469055e-01 -6.74021780e-01 -9.96285081e-01 5.52806199e-01
5.14009356e-01 -1.19033349e+00 1.14137566e+00 -7.78944910e-01
2.89268017e-01 -5.98427176e-01 -1.47426665e-01 -9.35724854e-01
-1.20539509e-01 -4.64629978e-01 -2.71658570e-01 1.45876658e+00
2.44371802e-01 -4.33154613e-01 6.09059513e-01 3.91136825e-01
-3.37060004e-01 -7.84247994e-01 -7.01083481e-01 -1.24478281e+00
-1.10943086e-01 -4.17384446e-01 5.64076006e-01 9.88249242e-01
-8.06665599e-01 6.98391274e-02 -4.11862582e-01 2.02379405e-01
5.99599779e-01 1.88278720e-01 1.05449355e+00 -1.15041828e+00
-2.19047129e-01 -7.13629305e-01 -5.58155954e-01 -1.01652765e+00
-7.35462755e-02 -8.99993896e-01 1.27546906e-01 -1.47849619e+00
4.63123560e-01 -4.29782987e-01 -4.02007937e-01 5.30724168e-01
-4.34321553e-01 7.45013893e-01 5.47619164e-01 4.89857376e-01
-1.39418387e+00 1.02440804e-01 1.27332485e+00 -8.37372243e-02
-8.22656900e-02 -9.19684321e-02 -6.69397771e-01 8.43213201e-01
8.20386291e-01 -6.88540816e-01 -2.53235728e-01 -6.55427873e-01
-1.54320270e-01 -2.39467412e-01 7.24492908e-01 -1.32825100e+00
1.66898489e-01 -2.07154676e-01 1.23271190e-01 -3.67493033e-01
2.28985772e-01 -5.92357218e-01 3.35880280e-01 3.67948741e-01
-2.15055570e-02 2.16477558e-01 3.62931579e-01 5.81872463e-01
-3.07248741e-01 -2.26330668e-01 6.82656109e-01 -1.41866982e-01
-1.23893845e+00 5.14763474e-01 -1.45132476e-02 6.55927002e-01
1.32765806e+00 -5.16965628e-01 -5.29417634e-01 7.47337705e-03
-5.13485253e-01 5.34446299e-01 5.55447340e-01 8.14983666e-01
2.19341457e-01 -1.28349388e+00 -7.68433332e-01 -4.73160684e-01
1.91047251e-01 -6.40520453e-02 2.64277756e-01 1.16961563e+00
-5.21582425e-01 2.56632209e-01 -3.98762435e-01 -1.03027606e+00
-1.49577963e+00 5.22154093e-01 2.90569574e-01 -6.17631786e-02
-5.30535460e-01 7.42346942e-01 1.97379827e-01 -1.96662381e-01
3.51169914e-01 -4.83468264e-01 -2.11826004e-02 2.66126245e-01
4.92394149e-01 5.23295164e-01 -9.08125788e-02 -7.27974772e-01
-5.67312837e-01 5.77382386e-01 -1.73836332e-02 4.33515936e-01
1.18730974e+00 -1.88888669e-01 1.99389994e-01 4.80009168e-01
1.04056251e+00 -1.71706393e-01 -1.55740201e+00 -1.03154987e-01
3.70794445e-01 -6.80345297e-01 -3.53202671e-01 -7.02714086e-01
-1.39654303e+00 5.05337894e-01 6.31562889e-01 6.36630297e-01
1.08433068e+00 2.77298599e-01 8.37422371e-01 1.32363722e-01
2.26530954e-01 -9.63827670e-01 1.35979220e-01 2.85569847e-01
5.71215034e-01 -1.56002951e+00 1.60225239e-02 -4.23136830e-01
-7.73812175e-01 7.67197728e-01 8.61913621e-01 1.73803747e-01
2.19675556e-01 4.72044706e-01 1.34432688e-01 -3.17241907e-01
-5.72618604e-01 -5.52645802e-01 4.43485409e-01 4.37625229e-01
5.08554816e-01 -1.45732120e-01 -1.41541019e-01 3.74041826e-01
2.80155927e-01 1.52882487e-01 3.43351901e-01 1.05578220e+00
-2.47294560e-01 -9.93342817e-01 -4.13468748e-01 5.30378461e-01
-7.24528551e-01 1.75652936e-01 -2.03103870e-01 1.17500353e+00
2.98825383e-01 9.27538216e-01 -7.07787573e-02 -2.71000743e-01
3.68710339e-01 -1.51535377e-01 4.37199056e-01 -5.45630336e-01
-7.13332057e-01 -1.81304768e-01 1.08924154e-02 -9.63058829e-01
-9.09827709e-01 -8.20984542e-01 -9.67665613e-01 -3.10043246e-01
-2.27423981e-01 -8.13850686e-02 3.79908919e-01 7.94083059e-01
4.77646917e-01 7.12546229e-01 5.09691015e-02 -1.10145712e+00
-1.82485029e-01 -7.00507939e-01 -3.18260223e-01 7.99799025e-01
1.63219601e-01 -9.66321170e-01 -1.43750533e-01 2.91454077e-01] | [9.09501838684082, 0.0063134790398180485] |
905c7d16-c51e-4512-8bed-bd2881d930bd | iarm-inter-aspect-relation-modeling-with | null | null | https://aclanthology.org/D18-1377 | https://aclanthology.org/D18-1377.pdf | IARM: Inter-Aspect Relation Modeling with Memory Networks in Aspect-Based Sentiment Analysis | Sentiment analysis has immense implications in e-commerce through user feedback mining. Aspect-based sentiment analysis takes this one step further by enabling businesses to extract aspect specific sentimental information. In this paper, we present a novel approach of incorporating the neighboring aspects related information into the sentiment classification of the target aspect using memory networks. We show that our method outperforms the state of the art by 1.6{\%} on average in two distinct domains: restaurant and laptop. | ['er', 'Soujanya Poria', 'Alex Gelbukh', 'Navonil Majumder', 'Erik Cambria', 'Asif Ekbal', 'Md. Shad Akhtar'] | 2018-10-01 | null | null | null | emnlp-2018-10 | ['extract-aspect'] | ['natural-language-processing'] | [ 9.75311920e-02 4.05850448e-02 -7.46342123e-01 -7.98430562e-01
-4.68148977e-01 -6.01308882e-01 2.98481941e-01 5.70175350e-01
-3.41163665e-01 6.97261214e-01 2.74758428e-01 -4.76528704e-01
-3.45714614e-02 -1.24299049e+00 -3.14887404e-01 -4.27829355e-01
1.19666159e-01 4.06725079e-01 1.47312388e-01 -6.41694427e-01
7.97669411e-01 1.25611186e-01 -1.20110524e+00 6.68838739e-01
5.35454452e-01 1.39215147e+00 -3.93109739e-01 3.19832236e-01
-7.10089862e-01 6.21158302e-01 -6.49328768e-01 -8.13881934e-01
1.65819049e-01 -1.31966382e-01 -7.93671668e-01 1.39295191e-01
-1.49220064e-01 2.60677874e-01 5.54128706e-01 1.24620342e+00
2.27097526e-01 3.61225605e-01 3.43721688e-01 -9.33360279e-01
-8.46339762e-01 8.21299493e-01 -9.32825208e-01 2.78445810e-01
2.60625064e-01 -3.68485123e-01 1.51388872e+00 -8.34729075e-01
6.33500755e-01 7.83807397e-01 5.75235426e-01 4.94312234e-02
-7.28067517e-01 -4.51593280e-01 8.58401299e-01 1.27748892e-01
-6.24518752e-01 -2.25177128e-03 1.00887692e+00 1.46338731e-01
1.61502314e+00 2.56842762e-01 9.70625579e-01 6.02877378e-01
3.17535937e-01 9.49298084e-01 1.45339215e+00 -1.62047759e-01
5.46022713e-01 6.99287236e-01 9.40301061e-01 2.54780710e-01
7.00832784e-01 -3.26184303e-01 -8.09098363e-01 -2.82528520e-01
1.30435169e-01 2.31721610e-01 3.05115461e-01 -1.45504838e-02
-8.28314781e-01 1.11011982e+00 7.18282610e-02 2.53275186e-01
-7.41209865e-01 -3.28358352e-01 4.55129236e-01 5.94477415e-01
1.06714427e+00 6.86856866e-01 -1.32005191e+00 -3.03195208e-01
-6.58231020e-01 -1.00793310e-01 1.36157823e+00 9.20927346e-01
7.32858241e-01 -4.31569703e-02 3.60187024e-01 7.87077308e-01
4.52303469e-01 4.99983639e-01 4.37556922e-01 -3.95250440e-01
3.58240604e-01 1.11064887e+00 -1.50394022e-01 -9.66758430e-01
-4.03624386e-01 -6.50641918e-01 -4.44132864e-01 -1.57527775e-01
-1.52754068e-01 -2.31895491e-01 -8.92533064e-01 7.16570556e-01
4.25010681e-01 -5.01418829e-01 5.86079471e-02 5.87490082e-01
8.51253271e-01 4.46889341e-01 1.07532226e-01 -1.87652633e-01
1.57691336e+00 -1.39769888e+00 -7.86746204e-01 -4.38764721e-01
2.62910187e-01 -1.01509833e+00 9.89941955e-01 6.70678973e-01
-8.63205910e-01 -2.33617291e-01 -1.04238522e+00 4.64479536e-01
-1.03652942e+00 -4.18881059e-01 1.47158444e+00 1.16669226e+00
-9.85969841e-01 3.49382252e-01 -5.89794457e-01 -3.50140482e-01
4.44516838e-01 6.21695220e-01 -1.79756716e-01 1.43003419e-01
-1.08541632e+00 5.50616145e-01 -6.88173100e-02 -3.35990638e-01
6.15836633e-03 -5.04322231e-01 -9.41964269e-01 9.49856546e-03
5.24315953e-01 -5.27371109e-01 1.16680753e+00 -1.20809233e+00
-1.38773739e+00 5.61974823e-01 -4.78167504e-01 -2.76523620e-01
-2.79031903e-01 -2.42811829e-01 -9.86510813e-01 -5.30778617e-03
3.11600655e-01 2.18430147e-01 6.25324070e-01 -1.08791399e+00
-1.05142939e+00 -8.68149579e-01 5.47160923e-01 1.93878487e-01
-6.61752999e-01 2.10310921e-01 -4.46468830e-01 -6.22127175e-01
5.79532934e-03 -7.08055019e-01 -9.15121436e-01 -7.10537255e-01
-3.01623136e-01 -1.02555886e-01 5.72391689e-01 -2.68114239e-01
1.30524242e+00 -1.65639663e+00 -4.99063790e-01 6.70103908e-01
1.37243802e-02 -4.44262475e-02 -1.27603680e-01 2.00233251e-01
5.03512993e-02 2.74448574e-01 1.18627742e-01 8.29679221e-02
-5.34890629e-02 3.62460911e-02 -9.49241295e-02 -3.50436904e-02
2.46327147e-01 9.91937280e-01 -6.91759765e-01 -1.78457618e-01
-2.01192036e-01 5.62469006e-01 -6.32386029e-01 -3.78588647e-01
-3.42711926e-01 5.03103137e-02 -8.18245411e-01 1.25075173e+00
6.39404356e-01 -3.33376735e-01 4.27542478e-01 2.79504638e-02
4.13293727e-02 8.41974556e-01 -1.05525827e+00 1.28742635e+00
-5.61372459e-01 2.67009765e-01 1.83441024e-03 -9.18265522e-01
1.24534261e+00 -9.54784453e-03 5.35494924e-01 -9.31685925e-01
3.46723258e-01 -4.87627322e-03 -1.82613194e-01 -1.14336498e-01
1.00805247e+00 -3.64328504e-01 -2.99682826e-01 1.00919473e+00
-9.22566801e-02 1.33666232e-01 4.38779294e-01 -1.94440335e-01
7.49319732e-01 -1.40849084e-01 6.58848763e-01 -5.50976872e-01
5.97293437e-01 3.60542804e-01 4.60793853e-01 3.96935403e-01
-3.35613102e-01 9.44852605e-02 4.98972833e-01 -8.11005652e-01
-4.96766150e-01 -5.93777955e-01 8.85558054e-02 1.46173382e+00
8.89982134e-02 -5.17059445e-01 -4.34473902e-01 -1.30770314e+00
-1.74579427e-01 8.28086555e-01 -7.21667230e-01 1.55790895e-01
-3.05253118e-01 -1.11848438e+00 -3.34194213e-01 6.96428895e-01
6.50259793e-01 -1.12281179e+00 -1.73565418e-01 1.79371014e-01
2.44603585e-03 -9.39820290e-01 -3.68261367e-01 6.12429142e-01
-1.35292315e+00 -7.71292210e-01 -2.43647873e-01 -7.10537374e-01
6.82410419e-01 5.36825061e-01 1.58873653e+00 -1.90139115e-01
3.41977358e-01 1.76658928e-01 -5.70563555e-01 -7.35637605e-01
1.56031460e-01 6.94794297e-01 -1.17117696e-01 -1.61007002e-01
1.60264933e+00 -7.14119434e-01 -8.03819001e-01 4.66920882e-01
-7.61770308e-01 -7.65657544e-01 7.40042210e-01 4.33896810e-01
8.78105640e-01 5.06319940e-01 6.61416352e-01 -1.66810930e+00
9.29910839e-01 -6.90707624e-01 -4.26982343e-01 -1.07100494e-01
-1.40357256e+00 -1.54579476e-01 5.69991231e-01 -8.13524127e-02
-1.10948145e+00 -1.30318284e-01 -2.97911167e-01 6.61111355e-01
-2.38300875e-01 8.78796875e-01 -1.03922784e-01 1.23717785e-01
3.64521742e-01 -1.43358875e-02 -4.13826197e-01 -3.62229019e-01
1.80249974e-01 7.08842993e-01 -3.10856581e-01 -1.39602572e-01
3.14732224e-01 7.27221012e-01 -3.01356852e-01 -4.92173761e-01
-1.23372889e+00 -1.01663876e+00 -5.11031270e-01 -9.18357745e-02
4.51436967e-01 -8.25170696e-01 -6.85219526e-01 -1.01593345e-01
-8.05580318e-01 4.28719550e-01 -5.34548759e-01 3.21610928e-01
-2.54505098e-01 -7.87778273e-02 -5.62636852e-01 -6.87173784e-01
-7.19604194e-01 -9.58947301e-01 8.22712719e-01 2.84603626e-01
-5.83457172e-01 -1.19476879e+00 1.41435415e-01 8.86253178e-01
6.04766846e-01 -2.59952962e-01 6.00480855e-01 -1.18589723e+00
-4.18510824e-01 -4.39361274e-01 -5.46970218e-02 3.23319376e-01
3.29957992e-01 -4.26480830e-01 -9.08039272e-01 2.48938382e-01
4.03766751e-01 3.07637572e-01 9.75966394e-01 3.85468662e-01
4.48282510e-01 -2.73558766e-01 -1.41612798e-01 2.71140784e-01
1.50029433e+00 6.31501317e-01 5.32987654e-01 9.93866503e-01
3.75277549e-01 8.47749531e-01 1.19893742e+00 3.33161950e-01
5.01892686e-01 -4.79142414e-03 1.12907432e-01 -8.87541100e-02
4.32844818e-01 2.99415644e-02 4.84174758e-01 1.09731007e+00
-2.56031066e-01 -2.30946355e-02 -5.91220796e-01 7.69284070e-01
-1.70720375e+00 -6.70183480e-01 -9.72389430e-02 1.67771626e+00
5.34808397e-01 6.58479214e-01 1.54668063e-01 1.07950136e-01
4.85338360e-01 4.93540168e-01 -5.67786574e-01 -1.09986711e+00
-9.53954756e-02 5.06652653e-01 5.60769618e-01 2.67313808e-01
-1.37276852e+00 9.14336145e-01 6.98413658e+00 5.48331022e-01
-8.55431437e-01 1.97260410e-01 1.17866099e+00 -2.03903854e-01
-7.86771357e-01 -1.06075227e-01 -1.08040452e+00 -8.78315717e-02
1.08225155e+00 -5.35759442e-02 7.03147724e-02 1.40612650e+00
4.70905565e-02 -3.80782783e-01 -6.49432838e-01 3.40249032e-01
4.33274329e-01 -1.34358370e+00 3.15764725e-01 1.49128154e-01
1.21142328e+00 1.02005266e-01 3.36344808e-01 2.77693301e-01
4.52508062e-01 -7.36766875e-01 8.33065808e-03 1.19711295e-01
3.06204557e-01 -1.26793981e+00 1.24071908e+00 -1.38657361e-01
-1.14237833e+00 2.20069677e-01 -2.54922807e-01 -3.55657369e-01
1.74936339e-01 1.00848413e+00 -8.54790330e-01 4.87023026e-01
9.95551467e-01 9.63654041e-01 -2.64723748e-01 3.91050845e-01
-1.90984145e-01 7.02087224e-01 3.80446017e-02 -5.01524448e-01
4.26847816e-01 -4.58535641e-01 4.39524204e-01 1.13330638e+00
1.38531730e-01 -8.81355032e-02 -9.72206444e-02 4.17029917e-01
-6.94546923e-02 5.08156121e-01 -6.87893927e-01 -1.91960782e-01
-2.32813627e-01 1.64641070e+00 -1.43429768e+00 -3.53018224e-01
-8.33409131e-01 7.59121180e-01 -2.06844330e-01 2.17554525e-01
-3.99747103e-01 -4.88767415e-01 8.92148018e-01 1.86627842e-02
9.36127186e-01 1.14859775e-01 -9.98318970e-01 -1.02795470e+00
1.97643831e-01 -9.56266880e-01 3.55856359e-01 -1.68403581e-01
-1.36841977e+00 1.01347423e+00 -6.48529351e-01 -1.12433362e+00
-1.25055611e-01 -8.66232276e-01 -5.29760838e-01 5.29233813e-01
-2.03556895e+00 -1.07288849e+00 3.14314663e-01 3.46792340e-01
6.84548020e-01 -4.37105834e-01 9.49821174e-01 2.53597140e-01
-2.13437855e-01 3.93042356e-01 -2.14661077e-01 -2.28047203e-02
5.34267008e-01 -1.22688043e+00 7.81561077e-01 5.26080787e-01
1.79090902e-01 9.84910667e-01 6.95661843e-01 -7.46305943e-01
-1.29072559e+00 -9.98426259e-01 1.40970409e+00 -6.63599432e-01
1.01384342e+00 -1.63112339e-02 -2.18539447e-01 7.03914225e-01
5.23342133e-01 -3.54175687e-01 1.62808883e+00 8.46786201e-01
-5.00775337e-01 -2.48163372e-01 -1.24662781e+00 4.89667296e-01
8.26612353e-01 -4.18729067e-01 -6.01447105e-01 8.86837766e-03
9.99595404e-01 3.32038373e-01 -1.06226313e+00 3.37720126e-01
6.11429036e-01 -1.12810194e+00 8.81892920e-01 -6.78311884e-01
6.53649151e-01 -1.08365484e-01 -4.22217458e-01 -1.39034832e+00
-7.22290128e-02 -4.10892963e-01 2.60565355e-02 1.11855972e+00
1.28311670e+00 -8.93265247e-01 1.39462090e+00 6.83091283e-01
2.74023980e-01 -1.17486131e+00 -2.80649662e-01 -2.95675486e-01
-1.72249302e-01 -8.99883091e-01 1.01265609e+00 8.51190984e-01
5.01953244e-01 8.12184155e-01 8.89722407e-02 -2.27246344e-01
2.55749375e-01 8.00041020e-01 2.99189121e-01 -1.10800648e+00
-4.07409519e-01 -5.65372467e-01 -3.45954299e-01 -1.00965977e+00
-1.09969355e-01 -5.03808916e-01 -5.91114759e-01 -1.41687572e+00
4.08428043e-01 -1.56380251e-01 -9.17361736e-01 1.81626558e-01
-2.73003817e-01 6.03755414e-01 -1.41527161e-01 -2.62015343e-01
-1.17092574e+00 2.04978317e-01 1.31849515e+00 -4.29380208e-01
-3.04972529e-01 6.39389455e-01 -1.79336023e+00 9.80117023e-01
1.19566643e+00 -4.57957208e-01 -3.35641980e-01 -4.67271283e-02
1.06392312e+00 -4.72685754e-01 -6.28104210e-01 -3.04801136e-01
1.97257370e-01 -2.27628022e-01 3.09244454e-01 -9.47706103e-01
1.79747373e-01 -1.08747876e+00 -5.09920895e-01 6.43075928e-02
-3.14405233e-01 4.86033767e-01 1.75052971e-01 6.52467132e-01
-6.72327340e-01 -1.50954425e-01 5.12561761e-02 -4.04397696e-01
-9.05395210e-01 1.27653405e-01 -5.61151206e-01 -1.30848557e-01
7.91944146e-01 -1.95124567e-01 -3.36652756e-01 -5.48974395e-01
-6.70348704e-01 4.22833860e-02 1.75150573e-01 4.81361419e-01
5.25716841e-01 -1.08437288e+00 -8.57868791e-02 4.43442523e-01
2.46226564e-01 -4.80121762e-01 -1.23429969e-02 7.32004762e-01
-1.50039513e-02 7.48411775e-01 -6.00303002e-02 -9.04646814e-02
-1.26675940e+00 5.37372768e-01 -3.03861767e-01 -7.76605129e-01
1.05818883e-01 7.00250149e-01 -9.46685821e-02 -8.75457823e-01
-8.01789612e-02 -2.12677389e-01 -9.22373414e-01 4.47539389e-01
6.35138869e-01 3.19983691e-01 4.18756336e-01 -5.30944526e-01
-4.17167872e-01 5.11850297e-01 -4.83582169e-01 -1.56417981e-01
1.74941599e+00 -2.47871339e-01 -4.61243570e-01 5.71018875e-01
1.11580086e+00 4.19020861e-01 -4.08735007e-01 -3.91921252e-02
2.98011273e-01 -3.80827218e-01 1.31993845e-01 -1.20725918e+00
-1.56722677e+00 3.21436316e-01 1.27470791e-01 6.88945055e-01
1.43230915e+00 5.38616776e-02 1.05671489e+00 7.28935421e-01
6.03028119e-01 -1.57204700e+00 -1.76865876e-01 6.49059832e-01
1.61823764e-01 -1.47074437e+00 1.73202589e-01 -6.23034835e-01
-9.81954455e-01 9.00306761e-01 4.83257145e-01 -3.22473317e-01
1.28552389e+00 4.64862972e-01 5.86768389e-01 -6.75183833e-01
-9.38122988e-01 -3.58038753e-01 3.18388790e-01 6.34495378e-01
8.74236882e-01 2.59147912e-01 -5.43541431e-01 1.07146776e+00
-3.65315586e-01 -1.96273848e-01 2.89474577e-01 1.21158338e+00
-5.26267529e-01 -1.55369914e+00 -1.80558354e-01 9.62746263e-01
-1.20400679e+00 -6.60098791e-01 -6.01382375e-01 5.03513515e-01
-1.62339672e-01 1.27171028e+00 -9.57854390e-02 -6.82366431e-01
4.55478191e-01 1.59720667e-02 -3.15282911e-01 -7.42930353e-01
-1.21928930e+00 3.17719430e-01 5.97339988e-01 -8.21999371e-01
-1.03465509e+00 -6.59784913e-01 -9.74405885e-01 -2.81111747e-01
-3.75373542e-01 4.18419093e-01 9.05567467e-01 7.53336310e-01
6.90320373e-01 6.12117171e-01 6.97218597e-01 -2.34712169e-01
-1.99008994e-02 -7.68505931e-01 -6.38917446e-01 1.43647969e-01
7.14497939e-02 -2.25735098e-01 -4.14374709e-01 -1.92502424e-01] | [11.365690231323242, 6.718202590942383] |
7c4befa2-9606-48cc-837f-7b4329dc6898 | evaluation-of-rounding-functions-in-nearest | 2003.06885 | null | https://arxiv.org/abs/2003.06885v2 | https://arxiv.org/pdf/2003.06885v2.pdf | Evaluation of Rounding Functions in Nearest-Neighbor Interpolation | A novel evaluation study of the most appropriate round function for nearest-neighbor (NN) image interpolation is presented. Evaluated rounding functions are selected among the five rounding rules defined by the Institute of Electrical and Electronics Engineers (IEEE) 754-2008 standard. Both full- and no-reference image quality assessment (IQA) metrics are used to study and evaluate the influence of rounding functions on NN interpolation image quality. The concept of achieved occurrences over targeted occurrences is used to determine the percentage of achieved occurrences based on the number of test images used. Inferential statistical analysis is applied to deduce from a small number of images and draw a conclusion of the behavior of each rounding function on a bigger number of images. Under the normal distribution and at the level of confidence equals to 95%, the maximum and minimum achievable occurrences by each evaluated rounding function are both provided based on the inferential analysis-based experiments. | ['Olivier Rukundo'] | 2020-03-15 | null | null | null | null | ['no-reference-image-quality-assessment'] | ['computer-vision'] | [ 3.80228996e-01 -2.06874445e-01 -2.18485072e-01 -4.27968413e-01
-7.02577353e-01 -2.06595153e-01 2.82646865e-01 4.65998113e-01
-5.01257122e-01 8.45736086e-01 1.85342193e-01 -5.07169664e-01
-5.43610394e-01 -7.77589262e-01 -5.89880764e-01 -4.89402682e-01
-1.28920302e-01 -3.57331067e-01 1.24587841e-01 2.01832980e-01
6.60087109e-01 6.32925153e-01 -1.69620740e+00 3.92199695e-01
1.26491463e+00 9.81691897e-01 -1.07439477e-02 5.90782225e-01
2.08910465e-01 6.09484911e-01 -8.07997227e-01 -1.28576219e-01
3.21899801e-01 -5.90478420e-01 -4.98111784e-01 -1.84938312e-01
3.85179102e-01 -6.21669292e-01 2.33029529e-01 1.02978981e+00
6.62006497e-01 2.18519837e-01 1.04926920e+00 -9.29031432e-01
-8.47166777e-01 3.75483602e-01 -3.39559525e-01 3.11819673e-01
6.28233254e-01 2.70309061e-01 1.33158609e-01 -7.77156413e-01
3.36249888e-01 8.20890129e-01 6.44090414e-01 -5.86372912e-02
-1.24891078e+00 -7.15965211e-01 -5.92143714e-01 2.53930151e-01
-1.69623208e+00 -1.96481645e-01 4.00316089e-01 -5.67944825e-01
6.40966535e-01 5.00843763e-01 5.79667270e-01 2.63065934e-01
5.33532143e-01 -2.29156762e-01 1.53232217e+00 -8.93307030e-01
1.25770912e-01 4.07217205e-01 -1.89692024e-02 2.95563221e-01
3.90427977e-01 4.07521039e-01 -2.25397587e-01 -1.08142734e-01
9.52010512e-01 -5.17482877e-01 -3.40124965e-01 3.14216584e-01
-8.91604662e-01 7.79604554e-01 4.15948451e-01 4.06984627e-01
-5.25723934e-01 -2.12450981e-01 4.57756788e-01 1.64083898e-01
1.68539643e-01 1.59934238e-01 -4.34128605e-02 1.22204058e-01
-1.02698290e+00 4.81301658e-02 5.42117596e-01 5.96250594e-01
6.35835767e-01 1.95139050e-01 -5.65780580e-01 6.77777052e-01
3.32387358e-01 5.05484462e-01 3.32034975e-01 -1.18968248e+00
1.31761163e-01 4.02551293e-01 4.18730944e-01 -1.22736645e+00
-9.13114659e-03 -3.49630058e-01 -5.68626344e-01 7.93014705e-01
6.22916579e-01 -7.64099732e-02 -6.32871985e-01 1.20388472e+00
-1.37928966e-02 3.91979814e-02 -1.99136198e-01 6.31658077e-01
7.58553505e-01 6.51596010e-01 1.72726884e-01 -5.70862889e-01
1.24178600e+00 2.86107883e-03 -6.69436097e-01 5.86539090e-01
1.51072413e-01 -1.04770577e+00 1.40685678e+00 4.09070164e-01
-9.98088241e-01 -9.77593184e-01 -1.17323899e+00 -1.18090250e-02
-1.20208479e-01 1.66941538e-01 -2.08651498e-02 1.02357388e+00
-1.04547083e+00 6.13660753e-01 -2.25205213e-01 -9.42267925e-02
2.18493432e-01 1.92492336e-01 -1.18818253e-01 -1.66931763e-01
-1.09188890e+00 9.52163935e-01 -2.28259582e-02 2.75782436e-01
-4.08971220e-01 -8.44578922e-01 -5.63686550e-01 -1.22760542e-01
-4.43883985e-01 -5.74971676e-01 5.86383939e-01 -9.34480965e-01
-1.23594201e+00 6.26834989e-01 -2.10897997e-01 -3.93178582e-01
6.70564830e-01 5.74334487e-02 -4.93630856e-01 3.41881037e-01
1.96091145e-01 4.87973124e-01 5.78696251e-01 -1.42013252e+00
-5.21584690e-01 -3.15243095e-01 -1.57498002e-01 2.26405993e-01
2.49576773e-02 1.11761145e-01 1.25858858e-02 -5.86202025e-01
-4.01184987e-03 -4.81913835e-01 -1.77320793e-01 2.78978258e-01
-6.09689876e-02 7.76757002e-02 4.70018327e-01 -9.38958108e-01
1.38083291e+00 -2.15208101e+00 -7.36537039e-01 7.91430593e-01
-1.50885954e-01 -5.85686322e-03 1.87094361e-01 1.88605443e-01
-3.02467924e-02 5.10355718e-02 -2.28807271e-01 3.03630859e-01
-3.34764272e-01 -1.23587780e-01 1.45129964e-01 5.70098102e-01
-3.34799886e-01 1.16315223e-01 -5.74710548e-01 -5.96328795e-01
7.64695704e-01 9.29317594e-01 -2.27279156e-01 -1.21851992e-02
4.32842374e-01 4.51607049e-01 -2.07783654e-01 2.96355665e-01
8.61167789e-01 2.14852750e-01 -1.21108472e-01 -7.50486612e-01
-2.72958457e-01 -8.34001750e-02 -1.32254386e+00 8.63290191e-01
-5.93874514e-01 6.31614029e-01 -4.86948073e-01 -5.41354656e-01
1.28154814e+00 4.09148246e-01 9.26688090e-02 -9.44135606e-01
8.15813616e-02 4.70525086e-01 -2.25131423e-03 -6.15637124e-01
1.48065284e-01 -1.24511488e-01 5.36559284e-01 -1.10851817e-01
-3.55653495e-01 -1.02474831e-01 2.74629831e-01 -2.59135664e-01
3.45747292e-01 -8.68026167e-02 6.29913270e-01 -6.14460826e-01
7.91572213e-01 -2.60308027e-01 3.72752577e-01 7.18134224e-01
-3.81828517e-01 5.33855855e-01 2.60180771e-01 -2.31745303e-01
-9.90933418e-01 -1.14829803e+00 -7.21466422e-01 2.58998334e-01
2.27165461e-01 2.42563024e-01 -7.59013474e-01 -3.55291623e-03
-1.46070600e-01 1.03731155e+00 -6.00249231e-01 2.09100902e-01
-1.10713899e-01 -6.09578252e-01 3.06155980e-01 1.46034151e-01
6.79993331e-01 -1.08165121e+00 -9.07213390e-01 -2.37039939e-01
-5.58832698e-02 -9.38112795e-01 -1.53383628e-01 -4.09135640e-01
-1.06029737e+00 -9.97430980e-01 -8.07115912e-01 -6.32281244e-01
8.49243641e-01 -8.81967694e-02 9.16286051e-01 2.14155391e-01
-1.52904853e-01 2.26625994e-01 -2.73137838e-01 -5.84444925e-02
-6.22885525e-01 -5.44923007e-01 -2.60975212e-01 -9.48027521e-02
2.86208749e-01 -5.48174202e-01 -1.03493166e+00 3.96118402e-01
-6.58256173e-01 -4.72108573e-01 4.65185314e-01 3.32915068e-01
7.54209220e-01 4.32188243e-01 4.85856622e-01 -3.56348366e-01
8.62007380e-01 -1.62536010e-01 -7.58775592e-01 3.59410167e-01
-9.17899370e-01 -1.69548705e-01 5.93816161e-01 -1.83915064e-01
-9.61260378e-01 -4.32226181e-01 1.81953996e-01 -8.47158358e-02
-3.28595191e-01 5.22674799e-01 2.16062684e-02 -2.42702842e-01
1.08608544e+00 1.77855030e-01 1.01283878e-01 4.62236255e-02
6.19493611e-03 7.08323956e-01 6.77824676e-01 -4.99933571e-01
3.11619967e-01 2.66902387e-01 3.64566982e-01 -1.07515454e+00
-1.62595749e-01 -2.05052078e-01 -2.86166877e-01 -6.51511788e-01
9.96037602e-01 -4.90478635e-01 -9.78824854e-01 2.82374710e-01
-8.41618478e-01 -1.75861105e-01 -2.50039071e-01 9.54071283e-01
-5.60411215e-01 3.98029804e-01 -2.53367424e-01 -9.79394913e-01
-4.35142219e-01 -1.42900705e+00 3.02224547e-01 4.17313367e-01
-3.50865841e-01 -7.34959900e-01 -3.10664833e-01 3.35459292e-01
2.71870047e-01 6.44585133e-01 9.28787649e-01 5.23187891e-02
-3.27591062e-01 -4.27208036e-01 -3.35539460e-01 4.54694211e-01
3.72213662e-01 4.54890966e-01 -6.82037175e-01 -6.20080605e-02
1.78488940e-01 3.41545105e-01 2.43572503e-01 1.12423170e+00
1.13228059e+00 -4.33089107e-01 5.31839840e-02 4.55876678e-01
2.01777768e+00 6.71101689e-01 1.29215562e+00 2.57771790e-01
1.45612136e-01 6.45601273e-01 6.51798248e-01 4.43560392e-01
-7.28925094e-02 6.15966320e-01 2.35200003e-01 2.35154387e-02
-1.64266214e-01 -9.64154582e-03 2.35103779e-02 2.16305494e-01
-3.97369474e-01 -5.70176803e-02 -7.66037405e-01 4.75545555e-01
-1.02979255e+00 -1.06569791e+00 -4.61417437e-01 2.84881496e+00
6.22601390e-01 1.51545525e-01 2.19737336e-01 6.28632665e-01
9.51510429e-01 -3.01196635e-01 9.31741297e-03 -8.16327751e-01
9.90194678e-02 6.58124089e-02 6.09593928e-01 8.54239941e-01
-5.26649833e-01 -2.48823717e-01 6.55201387e+00 1.00170410e+00
-1.13110292e+00 -4.01347578e-01 9.42454159e-01 4.07465547e-01
-4.42131013e-01 1.02435157e-01 -3.74099314e-01 6.93546891e-01
1.08157825e+00 -2.63240129e-01 3.15805614e-01 3.66254956e-01
7.74932027e-01 -7.05121219e-01 -5.42984962e-01 8.12192619e-01
-9.44685340e-02 -1.14820766e+00 3.33086193e-01 2.37879734e-02
7.92342663e-01 -5.01296163e-01 2.73255676e-01 -3.97540003e-01
-3.66403878e-01 -1.30409586e+00 5.31849384e-01 8.33512127e-01
9.90958393e-01 -9.19054627e-01 9.88640547e-01 1.29925245e-02
-8.45871806e-01 -1.92643274e-02 -1.84321046e-01 -3.65644358e-02
-9.62349027e-02 8.96000564e-01 -6.53934062e-01 6.45909488e-01
7.02726066e-01 -5.67563213e-02 -4.06792492e-01 1.21702099e+00
-2.10264936e-01 6.69537187e-01 -2.51859933e-01 2.31835879e-02
-3.01516742e-01 -3.87398273e-01 3.12114179e-01 9.46860015e-01
5.53049088e-01 1.45826712e-01 -5.04549086e-01 8.68909478e-01
5.12333453e-01 4.96115595e-01 -4.49727565e-01 4.59009230e-01
9.75366771e-01 6.83445573e-01 -8.20273817e-01 -1.10204361e-01
-3.82272780e-01 6.12915814e-01 -5.32761395e-01 4.32658046e-01
-6.18054628e-01 -6.18755162e-01 -7.97947273e-02 6.20195568e-01
-2.68338080e-02 4.29301560e-02 -9.91157770e-01 -8.48833844e-02
3.04085668e-02 -7.88914323e-01 3.47190320e-01 -9.73521113e-01
-9.68565702e-01 5.29460192e-01 5.76634586e-01 -1.57068110e+00
-1.22968286e-01 -2.97149897e-01 -7.50045419e-01 1.47387207e+00
-1.10006630e+00 -6.25857532e-01 -4.99578238e-01 3.60893071e-01
9.89842564e-02 1.23635136e-01 8.55523407e-01 2.33723670e-01
-7.87730366e-02 8.38229835e-01 1.01946995e-01 3.93304117e-02
3.80576909e-01 -8.28175068e-01 -4.28460300e-01 9.05444980e-01
-5.27717412e-01 6.95600331e-01 1.05874038e+00 -5.71469307e-01
-4.45872277e-01 -5.42834878e-01 1.16225266e+00 3.78478289e-04
-1.62876233e-01 4.95286673e-01 -7.66266048e-01 2.34597877e-01
3.26031633e-02 7.36686438e-02 6.58880830e-01 -3.93294543e-01
-1.22138269e-01 -3.83723825e-01 -1.94964111e+00 6.31634116e-01
2.41198584e-01 -3.51891100e-01 -1.75308019e-01 -3.59769315e-01
2.79575616e-01 -2.43992522e-01 -1.55265343e+00 6.40757024e-01
7.24386811e-01 -1.49690831e+00 1.25897980e+00 4.61914599e-01
6.75600648e-01 -6.87947810e-01 -4.85126972e-01 -8.57854009e-01
2.76727360e-02 -2.20538482e-01 5.57719588e-01 1.15618157e+00
4.87411380e-01 -5.72296083e-01 4.46377188e-01 6.75952196e-01
2.81833678e-01 -8.87365341e-01 -1.12837780e+00 -6.89170480e-01
-1.45862460e-01 -3.38583589e-01 5.13085246e-01 5.74912488e-01
-1.07879825e-01 -1.63754985e-01 -1.26256030e-02 2.23054290e-01
1.04500365e+00 -3.28192711e-01 4.67708558e-01 -7.15434611e-01
-4.40634638e-02 -1.92023218e-01 -7.79854178e-01 -1.18072048e-01
-6.38221264e-01 -2.86618978e-01 -2.26530209e-01 -1.66566563e+00
-1.10306881e-01 -3.25886965e-01 -3.15284461e-01 -1.01909518e-01
-1.93116918e-01 4.17335510e-01 6.01189844e-02 1.67077556e-01
5.92525721e-01 1.83999941e-01 1.29041934e+00 1.54557541e-01
-2.55199105e-01 3.17334324e-01 -4.66566235e-01 4.86212134e-01
6.20314181e-01 -2.67805159e-01 -7.30041921e-01 7.87462965e-02
1.36589289e-01 5.36001921e-01 5.04547179e-01 -1.13685024e+00
-2.05847979e-01 -1.33667752e-01 6.77968085e-01 -5.25841773e-01
-1.80730343e-01 -1.07239115e+00 4.20144975e-01 6.21357620e-01
-3.85303468e-01 7.56672770e-02 1.25306427e-01 1.13795422e-01
-1.49367005e-01 -5.47107100e-01 1.07088983e+00 3.01049531e-01
-4.99417216e-01 -2.94752628e-01 -1.75450355e-01 -4.55459774e-01
1.05065656e+00 -1.06063187e+00 5.54380044e-02 -5.69385111e-01
-5.71747601e-01 -2.90538847e-01 2.69628972e-01 -3.28751862e-01
7.17122257e-01 -1.29908335e+00 -9.64513302e-01 5.17185219e-02
3.70090418e-02 -4.81333494e-01 5.87229848e-01 8.70745301e-01
-9.66951728e-01 1.97987054e-02 -3.95996809e-01 -5.31811059e-01
-1.24714327e+00 4.23227876e-01 3.73582989e-01 -6.59771636e-02
-3.45330268e-01 4.13079351e-01 -2.42804557e-01 1.16433099e-01
6.69000521e-02 -4.85872984e-01 -4.75139230e-01 -1.78245381e-01
5.61149895e-01 1.18596721e+00 8.88667777e-02 -4.98183817e-01
-2.85655946e-01 7.36737490e-01 3.82207423e-01 -3.84286702e-01
8.38052630e-01 -2.72471964e-01 -1.39854416e-01 4.35289204e-01
1.14021897e+00 2.35230416e-01 -1.07239902e+00 5.25551915e-01
-4.85886961e-01 -8.40629518e-01 -1.43130809e-01 -9.96181905e-01
-5.84983587e-01 4.76181030e-01 1.27121878e+00 1.29633080e-02
1.37488759e+00 -7.40105748e-01 1.30214006e-01 -3.67557257e-01
1.46612003e-01 -9.23136413e-01 -5.12776256e-01 -1.70457363e-01
9.94842708e-01 -9.26284194e-01 3.36472899e-01 -5.31106293e-01
-3.23569924e-01 1.23487759e+00 1.65196821e-01 -2.49838918e-01
9.55996037e-01 1.47486180e-01 2.09993139e-01 1.92895055e-01
1.71506211e-01 4.00328159e-01 4.07808453e-01 9.75573063e-01
7.38823414e-01 1.65675879e-01 -1.29258287e+00 2.32715875e-01
-1.05198003e-01 7.27446318e-01 4.56596404e-01 5.03167272e-01
-3.74123335e-01 -6.92108452e-01 -8.69285464e-01 5.52552104e-01
-5.42062759e-01 -3.18117626e-02 4.02154714e-01 8.38379741e-01
4.82382506e-01 1.20341790e+00 2.14625821e-01 -1.72622219e-01
3.91884387e-01 -5.29891610e-01 5.94918609e-01 -4.49483655e-02
-5.62229574e-01 -2.85928756e-01 3.17361541e-02 -1.43113613e-01
-5.50118744e-01 -4.98269111e-01 -1.25125110e+00 -6.28243268e-01
-3.47496420e-01 3.01524967e-01 6.57352567e-01 9.06860173e-01
3.17336135e-02 2.63177186e-01 6.90458775e-01 -6.17151558e-01
-3.26188654e-01 -8.47447872e-01 -4.68686700e-01 3.41995627e-01
8.95880610e-02 -2.57722497e-01 -6.62380636e-01 1.22618964e-02] | [11.53414535522461, -2.131709575653076] |
63786253-4f85-4a71-98f9-a5d9fe6ec199 | making-the-invisible-visible-toward-high | 2304.14894 | null | https://arxiv.org/abs/2304.14894v1 | https://arxiv.org/pdf/2304.14894v1.pdf | Making the Invisible Visible: Toward High-Quality Terahertz Tomographic Imaging via Physics-Guided Restoration | Terahertz (THz) tomographic imaging has recently attracted significant attention thanks to its non-invasive, non-destructive, non-ionizing, material-classification, and ultra-fast nature for object exploration and inspection. However, its strong water absorption nature and low noise tolerance lead to undesired blurs and distortions of reconstructed THz images. The diffraction-limited THz signals highly constrain the performances of existing restoration methods. To address the problem, we propose a novel multi-view Subspace-Attention-guided Restoration Network (SARNet) that fuses multi-view and multi-spectral features of THz images for effective image restoration and 3D tomographic reconstruction. To this end, SARNet uses multi-scale branches to extract intra-view spatio-spectral amplitude and phase features and fuse them via shared subspace projection and self-attention guidance. We then perform inter-view fusion to further improve the restoration of individual views by leveraging the redundancies between neighboring views. Here, we experimentally construct a THz time-domain spectroscopy (THz-TDS) system covering a broad frequency range from 0.1 THz to 4 THz for building up a temporal/spectral/spatial/ material THz database of hidden 3D objects. Complementary to a quantitative evaluation, we demonstrate the effectiveness of our SARNet model on 3D THz tomographic reconstruction applications. | ['Chia-Wen Lin', 'Shang-Hua Yang', 'Po-Jen Yu', 'Yi-Chun Hung', 'Weng-Tai Su'] | 2023-04-28 | null | null | null | null | ['material-classification'] | ['computer-vision'] | [ 4.89115328e-01 -2.89340973e-01 2.47550592e-01 -4.42773402e-01
-9.01128232e-01 -3.48885477e-01 3.31383288e-01 -3.87219906e-01
7.51953423e-02 3.86109769e-01 4.02257085e-01 -1.33360788e-01
-7.04126656e-01 -7.28027761e-01 -3.66493642e-01 -1.22870326e+00
3.02164704e-01 2.38466740e-01 1.43878192e-01 -1.01445690e-02
1.03948899e-01 5.02504945e-01 -1.40028381e+00 3.84614319e-01
1.25173354e+00 1.13380110e+00 4.25610185e-01 4.41611558e-02
1.08208157e-01 1.49723709e-01 -9.69629288e-02 1.32366881e-01
2.84648180e-01 -3.32833856e-01 -2.27388799e-01 5.62680401e-02
4.92855698e-01 -6.93859339e-01 -1.02045667e+00 1.00694621e+00
5.06853640e-01 -1.19093560e-01 5.65691411e-01 -7.51518369e-01
-1.09798002e+00 8.16133395e-02 -6.54581308e-01 1.09746858e-01
1.03680760e-01 3.74936104e-01 4.86191303e-01 -8.91343236e-01
7.13020504e-01 8.32385123e-01 3.32805037e-01 5.57252586e-01
-1.24103832e+00 -7.61541069e-01 -1.38479233e-01 4.27390069e-01
-8.72517407e-01 -4.23229903e-01 1.14370918e+00 -2.83262700e-01
7.57832825e-01 3.65311414e-01 8.64449084e-01 1.15568495e+00
3.16211730e-01 5.96925244e-02 1.38475108e+00 -2.06608161e-01
-1.23690508e-01 2.82698143e-02 5.41955382e-02 6.07829928e-01
2.60748774e-01 1.92150846e-01 -8.19930851e-01 -3.39093730e-02
6.96937680e-01 3.45709115e-01 -8.80081773e-01 -1.66498572e-01
-1.30224943e+00 -4.76596784e-03 6.77382231e-01 5.72527409e-01
-4.60509151e-01 -3.14810462e-02 -7.31537342e-02 2.24788249e-01
8.79927397e-01 6.11399293e-01 2.89693773e-02 4.33543682e-01
-9.47635174e-01 -3.85098439e-03 -4.78101755e-03 5.24258256e-01
8.09546173e-01 9.26493853e-02 -3.30978692e-01 6.23081923e-01
3.02676737e-01 8.66848350e-01 2.67207503e-01 -1.02989900e+00
3.34671021e-01 4.73495364e-01 7.13586360e-02 -6.34176195e-01
-2.04125583e-01 -6.41975164e-01 -1.04271102e+00 -9.61814541e-03
-4.94359955e-02 3.43029976e-01 -1.04137373e+00 1.59549856e+00
5.83184779e-01 1.77386433e-01 1.87506024e-02 1.21753681e+00
9.04329777e-01 7.83118486e-01 -3.12427610e-01 -7.93034017e-01
1.13640332e+00 -5.33714533e-01 -7.31263280e-01 -5.28596416e-02
1.33506894e-01 -5.19067764e-01 9.29044724e-01 3.14460933e-01
-8.38572800e-01 -1.58563182e-01 -1.15858698e+00 -2.36000448e-01
2.68956393e-01 1.47147313e-01 4.79857653e-01 3.77596408e-01
-6.31109357e-01 7.73612678e-01 -9.29744542e-01 -1.90743595e-01
6.57099903e-01 1.37365460e-01 -3.93123031e-01 -7.07647502e-01
-5.76957047e-01 4.30309445e-01 3.54681872e-02 9.36902221e-03
-6.49183333e-01 -1.31195652e+00 -3.34290639e-02 -2.57265836e-01
3.20290595e-01 -9.74835575e-01 6.64472938e-01 -1.34138092e-01
-1.54969466e+00 6.25314236e-01 -2.28963092e-01 -1.30264580e-01
6.60778061e-02 -4.15105559e-02 -5.90455890e-01 7.68691957e-01
2.80036688e-01 -4.15867716e-02 8.12644064e-01 -1.30274665e+00
1.69088513e-01 -9.00735497e-01 -5.02006531e-01 8.58410597e-02
-6.46538556e-01 -2.38362476e-01 4.81082452e-03 -2.62808651e-01
1.11047411e+00 -4.99296844e-01 1.07772164e-01 1.49523774e-02
-4.62543935e-01 2.50818640e-01 1.31823230e+00 -5.65210283e-01
6.67677164e-01 -2.12505889e+00 1.65209174e-01 -3.21094155e-01
6.62999868e-01 2.64013577e-02 -1.52259395e-01 7.79254794e-01
-1.33367315e-01 -1.68636337e-01 -4.70597059e-01 -6.02174066e-02
-4.41702247e-01 -4.44217920e-01 -4.23215926e-01 7.95956612e-01
-1.70310214e-01 8.95930171e-01 -6.77488327e-01 -2.08111890e-02
1.35245219e-01 5.23088336e-01 -5.03328562e-01 8.85069296e-02
-2.34588251e-01 9.62519646e-01 -9.08765256e-01 1.03061271e+00
1.22287381e+00 -4.57705081e-01 1.24868704e-02 -9.81494009e-01
-6.84570909e-01 3.07684511e-01 -2.58962721e-01 1.66451240e+00
-1.21624954e-01 4.59135741e-01 3.12691659e-01 -1.17330408e+00
6.84493840e-01 3.71895641e-01 9.91465151e-01 -1.30746007e+00
-1.46655187e-01 4.98932391e-01 -2.32712165e-01 -9.37165022e-01
2.65959978e-01 -6.30484998e-01 3.19355458e-01 6.17788911e-01
1.92362249e-01 -4.58411634e-01 -4.42323565e-01 2.95729458e-01
1.01055646e+00 -1.29604349e-02 -2.18742490e-01 -1.96472600e-01
2.88224071e-02 -3.18938009e-02 3.70737702e-01 4.44961697e-01
1.92630783e-01 7.20287502e-01 3.19814309e-02 -3.12808812e-01
-1.34289670e+00 -1.04951930e+00 -2.76294649e-01 1.81299061e-01
2.54781097e-01 -1.48593515e-01 -5.18162549e-01 -1.47052914e-01
-5.99063039e-02 4.53844726e-01 -1.70245692e-01 -2.83244759e-01
-4.64316636e-01 -1.07461095e+00 1.63411140e-01 -4.09710817e-02
9.94752765e-01 -4.11461800e-01 -2.91938573e-01 2.36625984e-01
-4.84078854e-01 -1.55573964e+00 -1.47759050e-01 -2.01902270e-01
-1.08523798e+00 -1.00282681e+00 -6.01307452e-01 -2.18694210e-01
5.73369265e-01 1.06103957e+00 4.53561932e-01 -3.28620225e-01
-4.33812022e-01 7.41501451e-01 -2.02600911e-01 -8.57801363e-02
-1.69634715e-01 -4.63538915e-01 3.90416026e-01 1.94732115e-01
-2.81610996e-01 -1.36599612e+00 -8.07791054e-01 1.65557355e-01
-6.42821968e-01 4.70668525e-01 6.38763189e-01 6.22416377e-01
6.67693496e-01 9.96408314e-02 5.47681808e-01 -5.38848817e-01
2.07688913e-01 -3.64808559e-01 -7.91235745e-01 3.29298615e-01
-2.32843131e-01 -2.15518877e-01 4.46814001e-01 -2.03161389e-01
-1.10037279e+00 -4.50093746e-01 3.58435363e-02 -6.02847636e-01
-2.21211910e-02 4.71944332e-01 -2.91852802e-01 -6.58102036e-01
5.93823731e-01 5.76308846e-01 -8.66570026e-02 -4.76600379e-01
-3.61904278e-02 5.23400664e-01 6.76848352e-01 -3.79495829e-01
1.19408929e+00 8.06859553e-01 1.63430288e-01 -1.28632641e+00
-1.00073111e+00 -3.48433644e-01 -2.34736428e-01 -3.60337496e-01
8.51890564e-01 -1.15520501e+00 -9.69831288e-01 5.03407776e-01
-1.10274827e+00 -8.96358564e-02 2.11378261e-02 7.04847574e-01
-1.07227370e-01 6.34585321e-01 -5.05023003e-01 -8.86593580e-01
-6.06804311e-01 -7.04076409e-01 1.13608134e+00 1.84106052e-01
7.21085370e-01 -2.06710935e-01 -2.82559872e-01 1.00460887e+00
5.12345672e-01 3.15908670e-01 1.26937151e+00 3.71461064e-01
-1.50689745e+00 7.18085170e-02 -4.47284013e-01 4.34678681e-02
2.02459365e-01 -4.46641475e-01 -1.10129285e+00 -2.89506674e-01
9.13843155e-01 -1.42760798e-01 1.26253092e+00 5.65896273e-01
1.36749899e+00 -1.82238832e-01 -3.44904512e-01 1.03011024e+00
1.45516634e+00 -5.84659055e-02 5.16892076e-01 -1.85826242e-01
9.86758232e-01 6.33501112e-01 4.91485357e-01 3.47059816e-01
2.29768917e-01 6.48383856e-01 6.18342280e-01 1.72805503e-01
-3.88211869e-02 6.54683262e-02 2.84943074e-01 9.68102455e-01
-3.81153494e-01 -5.52192748e-01 -6.71363235e-01 2.77900875e-01
-1.41298485e+00 -1.10395646e+00 -4.00419772e-01 2.25249314e+00
1.55669376e-01 -4.24023539e-01 -3.53303105e-01 -1.95135728e-01
2.67010033e-01 4.07499313e-01 -7.93885291e-01 7.32506752e-01
-4.33793455e-01 1.52899846e-01 3.46188724e-01 5.12457900e-02
-6.98852956e-01 6.20040536e-01 4.96536636e+00 7.23978162e-01
-1.30941570e+00 4.56328601e-01 3.77072513e-01 -3.55861038e-01
-1.07918239e+00 -8.20096210e-02 -8.14464167e-02 3.14001501e-01
5.30333579e-01 -1.02677412e-01 5.30963898e-01 -6.38197586e-02
3.75647426e-01 -1.94864228e-01 -6.42392159e-01 1.14790356e+00
-3.21977377e-01 -1.76959741e+00 7.84534886e-02 4.21747744e-01
8.07092190e-01 3.71492684e-01 2.30352551e-01 -4.86162037e-01
-3.94094586e-01 -6.25928819e-01 4.87452179e-01 7.42220342e-01
1.05108500e+00 -5.73213220e-01 1.25557020e-01 5.83881915e-01
-9.23821270e-01 -2.08366930e-01 -3.29570770e-01 4.95151222e-01
1.62116319e-01 1.36069953e+00 -3.72018993e-01 1.10222697e+00
8.60417187e-01 8.66837382e-01 4.29888666e-02 7.57645845e-01
2.17930272e-01 5.11649013e-01 -2.91460097e-01 3.86633366e-01
8.26623887e-02 -7.62230217e-01 9.28745568e-01 3.58302027e-01
1.13727820e+00 9.63481903e-01 -1.45847201e-02 1.52103746e+00
-2.53749996e-01 -4.87225950e-01 -6.75419927e-01 -4.63065803e-01
5.56335211e-01 1.32399762e+00 -5.17840385e-01 -9.39213187e-02
-2.18352169e-01 7.90609598e-01 -1.12146966e-01 7.61874199e-01
-4.70064253e-01 4.48981635e-02 7.60393202e-01 5.11528254e-01
2.59719566e-02 -6.54941022e-01 -2.01566219e-01 -1.51395845e+00
4.24832642e-01 -3.73711675e-01 -3.10995579e-01 -9.91693795e-01
-1.25999796e+00 2.74312913e-01 -6.39898479e-02 -1.32523358e+00
5.40302932e-01 -2.28063881e-01 -5.99953115e-01 8.90442133e-01
-1.74239147e+00 -1.26572084e+00 -6.25148892e-01 5.50539613e-01
4.96545024e-02 5.56918643e-02 5.68276227e-01 3.91217619e-01
-4.55129296e-01 -1.78495646e-01 4.28526938e-01 -5.99124253e-01
6.95449769e-01 -6.94100738e-01 2.92385846e-01 6.88884139e-01
-1.71067417e-01 5.02617240e-01 5.64176619e-01 -9.40465331e-01
-2.11224103e+00 -1.11537313e+00 2.78350830e-01 -4.55054119e-02
3.91713023e-01 -4.99542058e-01 -1.14425552e+00 1.55653924e-01
1.43495739e-01 4.43215966e-01 3.62875432e-01 -3.28793764e-01
-4.32712913e-01 -3.28648776e-01 -1.40061271e+00 2.11347565e-01
1.37060440e+00 -9.96746421e-01 1.39961541e-02 8.46912026e-01
1.01897776e+00 -5.25872648e-01 -8.65047872e-01 6.09166801e-01
4.11283821e-01 -1.34968472e+00 1.25551713e+00 1.07046124e-02
6.55124724e-01 -1.61607876e-01 -3.17209601e-01 -1.18156278e+00
-5.16460419e-01 -5.76626122e-01 2.07055539e-01 1.07967281e+00
-1.00527175e-01 -1.00646544e+00 8.27842832e-01 6.96290806e-02
-8.77825677e-01 -5.21262109e-01 -1.27319098e+00 -7.03252316e-01
-3.25501561e-01 -4.04013246e-01 5.91150403e-01 1.07387865e+00
-1.91599682e-01 2.73991991e-02 -4.17387784e-01 7.58451760e-01
1.28924727e+00 6.66523635e-01 1.06460340e-01 -1.25887907e+00
-1.69398680e-01 4.92058359e-02 -1.57067075e-01 -8.63862097e-01
-1.29365057e-01 -9.88652229e-01 -2.46494398e-01 -1.40090907e+00
3.99071574e-01 -4.28174525e-01 -9.15800482e-02 9.36499611e-02
3.15923989e-01 3.72125715e-01 -2.98533309e-02 5.19781291e-01
-2.68256664e-01 1.18967152e+00 1.66438973e+00 -1.78116143e-01
1.84650315e-04 -1.46378011e-01 -5.95510125e-01 2.64688402e-01
4.02933270e-01 -3.98225307e-01 -3.63757253e-01 -7.26107955e-01
-6.98732724e-03 5.65271437e-01 9.41065311e-01 -1.25889242e+00
3.31800491e-01 -2.60154009e-01 4.05304164e-01 -6.84329927e-01
7.15672076e-01 -5.16402185e-01 3.21102679e-01 2.06172168e-01
2.28436872e-01 -8.86761546e-01 -1.02068968e-01 9.40972149e-01
-1.37145087e-01 1.04499660e-01 7.84942865e-01 -3.34183633e-01
-8.65452960e-02 7.07483709e-01 -2.01827154e-01 -3.73042911e-01
6.36450529e-01 -2.13997513e-01 -7.07485139e-01 -2.25336477e-02
-2.21443728e-01 -7.16277659e-02 6.77787781e-01 -1.06458366e-01
9.12415326e-01 -1.31175756e+00 -5.10671139e-01 2.81071097e-01
2.23874047e-01 -2.20589846e-01 1.18214452e+00 1.18932033e+00
-9.65565220e-02 3.33031416e-01 -1.24311395e-01 -7.49290943e-01
-1.20942855e+00 1.09067157e-01 5.05254209e-01 1.10758446e-01
-1.01282787e+00 6.68182373e-01 4.17370856e-01 -3.40570301e-01
-6.25608444e-01 -3.50803345e-01 2.43672654e-01 -2.66740829e-01
3.26627016e-01 2.69504696e-01 3.00626665e-01 -3.72798294e-01
-2.56285667e-01 9.97618020e-01 1.01635657e-01 7.18727149e-03
2.01673341e+00 -2.32871845e-01 -3.72231245e-01 8.00423548e-02
1.21648490e+00 -7.21113235e-02 -1.21975267e+00 -5.29408216e-01
-4.81601357e-01 -6.15190923e-01 4.35624927e-01 -6.02250874e-01
-1.01253295e+00 9.55413222e-01 6.27584457e-01 2.68339694e-01
1.38944852e+00 2.10550040e-01 9.58265662e-01 4.97082621e-01
5.76840222e-01 -6.91086411e-01 3.26198190e-01 1.50289536e-01
8.86912048e-01 -1.04297554e+00 3.71642739e-01 -6.95437014e-01
-2.99736578e-02 1.10298967e+00 2.95148939e-01 1.57815352e-01
3.15618604e-01 -1.82160765e-01 -4.99255955e-01 -7.06061840e-01
-7.51995802e-01 3.08000714e-01 3.13254029e-01 7.08004653e-01
1.01271041e-01 -2.31624767e-02 3.47089916e-01 3.56934100e-01
6.56099617e-02 -1.23922892e-01 4.14322793e-01 6.29390597e-01
-5.34033716e-01 -8.79258275e-01 -3.31388175e-01 5.93567789e-01
-2.53016427e-02 7.77873304e-03 -2.00917363e-01 3.69667172e-01
-2.11531594e-02 7.76104987e-01 -3.58574122e-01 -3.78848702e-01
3.10601562e-01 -1.29572883e-01 1.02493811e+00 -5.13188601e-01
-1.66581288e-01 5.04886925e-01 -1.33857474e-01 -6.69854641e-01
-5.64738572e-01 -7.63973534e-01 -9.31734502e-01 -2.37512320e-01
-3.52734715e-01 -9.21027660e-02 7.12686300e-01 1.12506175e+00
6.04474843e-01 4.72156614e-01 8.59961390e-01 -9.29013193e-01
-2.70270169e-01 -8.36163044e-01 -7.45922863e-01 -2.41905466e-01
6.97811067e-01 -4.70325291e-01 -3.81028295e-01 -5.61073840e-01] | [10.527976036071777, -2.3128621578216553] |
99b5fc1f-7cff-48ca-aa90-57ae364d2b71 | domain-adaptive-training-bert-for-response | 1908.04812 | null | https://arxiv.org/abs/1908.04812v2 | https://arxiv.org/pdf/1908.04812v2.pdf | An Effective Domain Adaptive Post-Training Method for BERT in Response Selection | We focus on multi-turn response selection in a retrieval-based dialog system. In this paper, we utilize the powerful pre-trained language model Bi-directional Encoder Representations from Transformer (BERT) for a multi-turn dialog system and propose a highly effective post-training method on domain-specific corpus. Although BERT is easily adopted to various NLP tasks and outperforms previous baselines of each task, it still has limitations if a task corpus is too focused on a certain domain. Post-training on domain-specific corpus (e.g., Ubuntu Corpus) helps the model to train contextualized representations and words that do not appear in general corpus (e.g., English Wikipedia). Experimental results show that our approach achieves new state-of-the-art on two response selection benchmarks (i.e., Ubuntu Corpus V1, Advising Corpus) performance improvement by 5.9% and 6% on R@1. | ['Heuiseok Lim', 'Dongsuk Oh', 'Taesun Whang', 'Kisu Yang', 'Chanhee Lee', 'Dongyub Lee'] | 2019-08-13 | null | null | null | null | ['conversational-response-selection'] | ['natural-language-processing'] | [ 3.69920768e-03 1.36726931e-01 -2.29537264e-01 -6.15530074e-01
-1.49703777e+00 -7.14347422e-01 5.83408892e-01 -1.21036686e-01
-7.08201349e-01 9.80536878e-01 5.54346263e-01 -4.60686773e-01
2.18204573e-01 -5.40618122e-01 -2.44866893e-01 -2.05240130e-01
3.69824857e-01 1.13059890e+00 3.47012877e-01 -1.22921002e+00
3.64723057e-01 1.61303192e-01 -7.44341791e-01 7.58083880e-01
7.32291877e-01 5.75410247e-01 6.03978157e-01 6.31776273e-01
-2.16594249e-01 1.08733523e+00 -8.24106336e-01 -2.48514175e-01
-6.45632818e-02 -4.48265612e-01 -1.54657078e+00 -2.08393648e-01
2.66916662e-01 -6.40705287e-01 -5.42079210e-01 4.50041026e-01
8.20136726e-01 6.66477740e-01 6.00415587e-01 -5.26232004e-01
-1.01032495e+00 8.09309602e-01 -3.19135189e-01 3.37287605e-01
6.17889166e-01 1.13156922e-02 1.38672650e+00 -1.19880104e+00
7.86577106e-01 1.63906312e+00 3.84114161e-02 9.70481098e-01
-1.10894811e+00 -4.39923882e-01 2.73836166e-01 5.35996109e-02
-9.87559497e-01 -4.47998196e-01 6.01562738e-01 5.84753864e-02
1.55580807e+00 4.09166843e-01 -4.54049677e-01 1.19284272e+00
3.67229879e-02 1.28348958e+00 9.86550927e-01 -4.67061758e-01
-2.06290931e-01 3.46482873e-01 5.73709548e-01 2.82349288e-01
-7.64072478e-01 -3.22789997e-01 -4.49400842e-01 -2.71461576e-01
3.73191476e-01 -2.77652323e-01 -4.08625245e-01 9.45549235e-02
-9.94123161e-01 1.38431966e+00 3.69467944e-01 1.40324309e-01
-1.50871873e-01 -2.88433433e-01 7.47188628e-01 9.02507305e-01
6.54249847e-01 8.54551435e-01 -9.19476330e-01 -9.27543417e-02
-4.13506210e-01 4.68580216e-01 1.07991982e+00 1.37942243e+00
7.49454021e-01 -1.54722646e-01 -5.23301601e-01 1.66280103e+00
1.11860059e-01 4.93926644e-01 7.67378330e-01 -6.64739311e-01
8.86904418e-01 6.16340697e-01 2.78430313e-01 -2.15612918e-01
-5.02813697e-01 1.68435089e-02 -5.22061348e-01 -4.13305283e-01
4.06476051e-01 -5.47435820e-01 -6.28688753e-01 1.64114189e+00
1.97560385e-01 -6.74279749e-01 4.18927938e-01 1.14229786e+00
1.50642765e+00 9.47151601e-01 1.08347870e-01 -1.76547572e-01
1.50208783e+00 -1.50188148e+00 -6.31228209e-01 -6.41385138e-01
8.76351655e-01 -9.79940772e-01 1.52423441e+00 3.95825580e-02
-1.03922117e+00 -3.99536252e-01 -5.87207556e-01 -5.21999955e-01
-3.89804453e-01 4.91057634e-01 4.41461682e-01 1.89402655e-01
-9.62202728e-01 6.96862787e-02 -1.25103474e-01 -5.90359688e-01
-4.50864017e-01 3.82268906e-01 -2.72502005e-01 -1.33173570e-01
-1.82927001e+00 1.40541899e+00 -5.27714081e-02 -2.91332126e-01
-8.52620006e-01 -3.82217646e-01 -7.26874709e-01 4.88528870e-02
4.96984512e-01 -3.24244499e-01 1.85361648e+00 -5.33371687e-01
-1.90671277e+00 9.89360631e-01 -2.26573676e-01 -4.40445036e-01
1.70133904e-01 -6.18010044e-01 -3.03985447e-01 9.83430743e-02
1.02419131e-01 7.25460589e-01 4.02155071e-01 -8.29593897e-01
-5.06637037e-01 -9.84799936e-02 6.41980410e-01 6.76336944e-01
-2.51442701e-01 7.19886065e-01 -3.91266167e-01 -3.64990771e-01
-2.69165188e-01 -1.21864820e+00 -4.04793888e-01 -8.86390686e-01
-2.82410860e-01 -1.05724645e+00 6.48420453e-01 -7.09332049e-01
1.20131779e+00 -1.73585820e+00 2.03424498e-01 -4.97721195e-01
-1.42722502e-01 2.58960813e-01 -3.61754775e-01 9.90898728e-01
1.04377300e-01 -1.07283713e-02 1.81800097e-01 -1.80244595e-01
-3.30191180e-02 -8.69118646e-02 -6.02899015e-01 1.95040748e-01
2.11850762e-01 6.45515203e-01 -9.20390546e-01 -3.72003078e-01
1.10887311e-01 4.56694426e-04 -5.82189500e-01 8.05285215e-01
-4.98167455e-01 4.38147485e-01 -7.06652641e-01 4.74579453e-01
3.74432772e-01 -2.65665412e-01 4.37663794e-01 1.57593623e-01
1.67830139e-01 1.32247424e+00 -5.48595190e-01 1.93018198e+00
-1.14938116e+00 5.12130737e-01 -6.32189512e-02 -6.93213165e-01
1.19205713e+00 3.79738986e-01 2.01401606e-01 -8.70885253e-01
-7.49878725e-03 2.11309686e-01 9.46882144e-02 -3.05172116e-01
1.15050507e+00 2.23487735e-01 -6.23327971e-01 9.01792109e-01
3.53117645e-01 -1.92023322e-01 2.22944260e-01 6.17191374e-01
1.13143778e+00 1.56763513e-02 3.55973780e-01 -4.89815176e-01
7.30433345e-01 1.07976563e-01 2.55322546e-01 7.32882261e-01
-1.08085861e-02 5.73428571e-01 2.82640308e-01 -4.15446937e-01
-5.92357457e-01 -5.36955297e-01 -6.92038909e-02 2.26256895e+00
1.15172230e-01 -4.67056990e-01 -4.23399389e-01 -1.08252537e+00
-1.04660250e-01 9.62545097e-01 -5.14317930e-01 -2.01708838e-01
-1.04471350e+00 -6.31199479e-01 3.33548248e-01 4.71497595e-01
4.86006171e-01 -1.35364115e+00 -3.43218744e-01 4.82035667e-01
-5.10010123e-01 -1.11672187e+00 -7.21181214e-01 3.33359748e-01
-6.54563129e-01 -7.06747770e-01 -7.26304531e-01 -9.20582354e-01
2.14849979e-01 6.63492143e-01 1.64705575e+00 -1.14992559e-01
3.02244008e-01 3.65193725e-01 -7.82439888e-01 -1.08135402e-01
-4.11471069e-01 7.39676178e-01 -1.80722192e-01 -5.33232093e-01
8.98138463e-01 1.49160758e-01 -4.77005899e-01 8.16751361e-01
-4.61621046e-01 -2.52716720e-01 4.49010342e-01 1.30215693e+00
8.63259286e-02 -9.57742512e-01 1.17546332e+00 -1.34402907e+00
1.35731089e+00 -6.58061147e-01 -1.95855275e-01 7.77319789e-01
-5.64995050e-01 -2.69748345e-02 5.94494641e-01 -5.38135946e-01
-1.43872690e+00 -1.56833246e-01 -1.73389927e-01 1.20216727e-01
7.43313432e-02 6.71876729e-01 1.30257025e-01 3.21830094e-01
1.04974794e+00 2.15494588e-01 -3.01153988e-01 -6.92182600e-01
5.19977033e-01 1.31589770e+00 3.56088430e-02 -8.04738581e-01
1.40435342e-02 -3.02633941e-01 -9.44638073e-01 -5.34865379e-01
-7.10458696e-01 -9.53097522e-01 -4.03173655e-01 2.99952901e-03
5.46263158e-01 -1.11829638e+00 -5.52342892e-01 -1.02221467e-01
-1.29955578e+00 -5.47827959e-01 2.46633530e-01 2.09087372e-01
-3.92353147e-01 1.06135175e-01 -1.00961721e+00 -8.47885370e-01
-8.33467305e-01 -1.25001526e+00 1.17158461e+00 1.60783634e-01
-4.84980494e-01 -9.58406270e-01 3.48088712e-01 6.22564256e-01
5.14469743e-01 -7.16179848e-01 7.89874852e-01 -1.32368004e+00
-3.06754261e-01 -1.39063984e-01 -1.58046529e-01 2.05699980e-01
7.82433674e-02 -5.33767104e-01 -9.72087920e-01 -5.22334099e-01
3.54619958e-02 -1.36392653e+00 8.66452754e-01 -8.84086043e-02
8.89898479e-01 -3.02102238e-01 -2.11548045e-01 -1.39552951e-01
9.67902601e-01 3.39870095e-01 4.91483808e-01 2.57792920e-01
2.39862874e-01 1.02537775e+00 1.19409251e+00 3.67588848e-01
6.22130394e-01 1.01509440e+00 3.10999006e-02 7.68544450e-02
3.81686874e-02 -8.09501410e-02 4.52145100e-01 8.99460375e-01
3.19165766e-01 -5.78733742e-01 -8.53434741e-01 6.94251060e-01
-1.95907378e+00 -5.04408062e-01 1.64256431e-02 1.96581411e+00
1.40629780e+00 -1.79479793e-01 1.85870931e-01 -8.90809536e-01
5.16555071e-01 3.16254765e-01 -5.85825920e-01 -7.96657622e-01
-4.20115367e-02 3.48760098e-01 3.43337119e-01 7.83168554e-01
-1.05818975e+00 1.65300429e+00 5.85021591e+00 7.83071399e-01
-9.43850935e-01 2.77292281e-01 6.40343130e-01 2.23970953e-02
-2.88601786e-01 -2.14516725e-02 -1.24303222e+00 1.81882396e-01
1.14460742e+00 -4.34674174e-01 3.38311136e-01 1.07027626e+00
-2.41134524e-01 8.40654075e-02 -1.17801523e+00 6.05419517e-01
2.07297966e-01 -9.66322124e-01 -1.59496963e-01 -2.93976128e-01
3.97395551e-01 3.23387206e-01 1.84574947e-02 1.26693368e+00
1.07412827e+00 -1.10517728e+00 9.12646577e-03 -1.83604598e-01
9.17173505e-01 -5.65409005e-01 8.84205937e-01 3.99352014e-01
-6.16213262e-01 1.32172108e-01 -8.98827672e-01 1.86290026e-01
8.43824074e-02 6.03282489e-02 -1.39222407e+00 3.95628721e-01
5.04127383e-01 4.19756711e-01 -3.29093128e-01 2.19906554e-01
-2.95575857e-01 6.20385528e-01 -1.78628370e-01 -6.02147758e-01
5.27837694e-01 -2.84334291e-02 2.47337148e-01 1.52793229e+00
-2.11234540e-01 4.16146219e-01 4.36919183e-01 3.97125363e-01
-5.08295715e-01 5.45004308e-01 -5.95974147e-01 1.49547487e-01
6.05936706e-01 1.37598228e+00 -1.25731453e-02 -3.87103319e-01
-6.83028996e-01 9.02356386e-01 7.71733105e-01 4.57760096e-01
-6.99306726e-01 -3.99708658e-01 5.12940407e-01 -2.96292484e-01
-3.88608873e-03 6.29497021e-02 -3.14882211e-03 -1.16312671e+00
-2.29928419e-01 -1.45458591e+00 6.08855784e-01 -4.85353202e-01
-1.36036742e+00 1.02467275e+00 -5.91214225e-02 -1.23614061e+00
-8.54859352e-01 -5.58618903e-01 -3.50486606e-01 1.12732291e+00
-1.46983635e+00 -1.05235255e+00 2.58177012e-01 6.06717348e-01
1.46598101e+00 -4.61914212e-01 1.24309576e+00 2.67082989e-01
-4.40492749e-01 7.70861387e-01 1.73647165e-01 2.95589536e-01
1.54004121e+00 -1.42898858e+00 4.09702659e-01 2.78353631e-01
-1.42544229e-02 1.03724325e+00 5.57120323e-01 -4.01103735e-01
-1.36105156e+00 -6.76835954e-01 1.27143443e+00 -6.69819832e-01
6.89450920e-01 -4.37425345e-01 -9.04827654e-01 9.12807882e-01
9.66033399e-01 -4.62225318e-01 8.22547734e-01 7.64406621e-01
-3.52419823e-01 1.02620758e-01 -9.21469450e-01 6.98217511e-01
7.22437561e-01 -8.08990300e-01 -8.70698810e-01 9.55812573e-01
9.44843233e-01 -8.97561073e-01 -7.60671675e-01 1.87880144e-01
2.82064646e-01 -3.65052462e-01 1.00122941e+00 -1.19529843e+00
6.24653459e-01 4.15147275e-01 -2.65222281e-01 -1.61524487e+00
-2.92658269e-01 -8.41666579e-01 9.60117280e-02 1.17785227e+00
7.69380629e-01 -3.98317546e-01 4.10710633e-01 8.40133488e-01
-3.07393700e-01 -8.43867242e-01 -8.22960198e-01 -3.77788037e-01
6.85430169e-01 1.90089092e-01 4.52455014e-01 9.89413619e-01
5.29888213e-01 1.33346927e+00 -6.40851557e-01 -3.96446228e-01
-2.06660226e-01 5.60828090e-01 1.01713777e+00 -8.04246724e-01
-1.86918378e-01 -1.46583200e-01 4.63824481e-01 -2.01612926e+00
6.16269946e-01 -7.00428784e-01 2.66047508e-01 -1.30128026e+00
5.24042323e-02 -5.07798493e-01 -2.91576624e-01 3.61260772e-01
-4.91268754e-01 -3.38289261e-01 5.44660687e-02 2.64517039e-01
-8.78487468e-01 7.53284037e-01 1.06981623e+00 -4.21076030e-01
-2.79025406e-01 1.94991484e-01 -8.83491695e-01 7.95508698e-02
8.48113716e-01 -5.47325850e-01 -5.80357909e-01 -6.77731872e-01
-4.65886593e-02 6.46653831e-01 -4.68722522e-01 -1.20517552e-01
2.41587445e-01 -2.77658850e-01 -2.81287789e-01 -4.23258096e-01
4.36184704e-01 -3.89433920e-01 -8.70448291e-01 -3.20507623e-02
-1.20331430e+00 2.20520765e-01 4.51014377e-02 2.94674546e-01
-3.81092459e-01 -5.72568178e-01 6.62013710e-01 -4.55339015e-01
-8.34574401e-01 -5.50887249e-02 -5.30855477e-01 4.60166842e-01
4.55722660e-01 5.68914831e-01 -7.91442692e-01 -8.61926496e-01
-4.21191126e-01 7.17885673e-01 9.30086076e-02 7.89105177e-01
6.41704500e-01 -1.01734209e+00 -1.07699800e+00 -3.67285639e-01
4.41735089e-01 -2.15594500e-01 1.53989196e-01 4.68321830e-01
-1.72322005e-01 8.57996702e-01 -2.87968554e-02 -2.71995127e-01
-1.36774957e+00 2.45107293e-01 2.16528460e-01 -9.51172829e-01
-3.55123639e-01 1.33634543e+00 4.47471738e-01 -1.12395167e+00
1.60329998e-01 -1.48585469e-01 -7.19356000e-01 9.34114233e-02
5.71828902e-01 -1.32707050e-02 2.47960135e-01 -5.23978412e-01
-4.57689553e-01 -2.13821411e-01 -9.37942505e-01 -2.93683857e-01
1.12508047e+00 -2.49652475e-01 1.36344032e-02 2.92046189e-01
1.26269138e+00 3.13767940e-02 -8.31550598e-01 -8.27924013e-01
1.51515007e-01 -2.97093213e-01 -1.55914620e-01 -1.13758433e+00
-8.14369678e-01 9.27155077e-01 2.27791727e-01 6.73807710e-02
7.71051109e-01 4.47078831e-02 9.57730651e-01 1.18018639e+00
5.73065758e-01 -1.40912771e+00 2.84557164e-01 1.16413116e+00
1.45929503e+00 -1.52045691e+00 6.86222762e-02 -2.75235981e-01
-1.30522656e+00 1.06401205e+00 1.22624755e+00 -1.88733742e-01
3.04805905e-01 -1.67740807e-01 3.82632345e-01 -1.96816623e-01
-1.36648548e+00 -2.55869210e-01 1.94664195e-01 2.73316413e-01
1.22471929e+00 7.89743960e-02 -5.36145449e-01 5.42623103e-01
-1.49282500e-01 -5.29807627e-01 5.67556739e-01 8.72075319e-01
-4.48033243e-01 -1.33373606e+00 1.50580525e-01 3.97522599e-01
-5.47687352e-01 -4.72078681e-01 -9.96690929e-01 7.78078735e-01
-9.58619118e-01 1.35741127e+00 -3.39682221e-01 -4.87800956e-01
4.00277436e-01 3.08349907e-01 8.65090117e-02 -1.15002537e+00
-1.20392799e+00 5.41105643e-02 8.81192863e-01 -3.94066006e-01
-1.66012809e-01 -4.04013842e-01 -9.39809680e-01 -1.26638785e-01
-6.86636627e-01 3.88790369e-01 2.72990584e-01 7.33331323e-01
2.64933407e-01 3.53340775e-01 8.76562953e-01 -2.68791944e-01
-1.22925091e+00 -1.78650963e+00 -2.55748630e-01 4.25359219e-01
1.24004081e-01 -6.83819592e-01 -7.37942234e-02 -3.80629778e-01] | [12.414101600646973, 7.8055267333984375] |
74fbc1d2-f2cc-44be-949a-a99d6332c1c6 | sequential-monte-carlo-steering-of-large | 2306.03081 | null | https://arxiv.org/abs/2306.03081v1 | https://arxiv.org/pdf/2306.03081v1.pdf | Sequential Monte Carlo Steering of Large Language Models using Probabilistic Programs | Even after fine-tuning and reinforcement learning, large language models (LLMs) can be difficult, if not impossible, to control reliably with prompts alone. We propose a new inference-time approach to enforcing syntactic and semantic constraints on the outputs of LLMs, called sequential Monte Carlo (SMC) steering. The key idea is to specify language generation tasks as posterior inference problems in a class of discrete probabilistic sequence models, and replace standard decoding with sequential Monte Carlo inference. For a computational cost similar to that of beam search, SMC can steer LLMs to solve diverse tasks, including infilling, generation under syntactic constraints, and prompt intersection. To facilitate experimentation with SMC steering, we present a probabilistic programming library, LLaMPPL (https://github.com/probcomp/LLaMPPL), for concisely specifying new generation tasks as language model probabilistic programs, and automating steering of LLaMA-family Transformers. | ['Vikash K. Mansinghka', 'Gabriel Grand', 'Tan Zhi-Xuan', 'Alexander K. Lew'] | 2023-06-05 | null | null | null | null | ['probabilistic-programming'] | ['methodology'] | [ 1.10069029e-01 2.29868054e-01 2.95733269e-02 -5.68372250e-01
-1.41883230e+00 -9.62139666e-01 9.37027514e-01 -2.41934463e-01
-2.15704367e-01 8.30652237e-01 1.18333660e-01 -8.20981205e-01
3.00959498e-02 -9.37586129e-01 -9.26696479e-01 -3.55707169e-01
2.73052454e-01 1.08586180e+00 1.83467105e-01 1.99872911e-01
2.12065086e-01 2.57775128e-01 -1.37526572e+00 3.99579376e-01
8.60124886e-01 3.67975742e-01 5.46064615e-01 1.16340339e+00
-2.78500795e-01 7.64506161e-01 -4.00042623e-01 -4.61801440e-01
-3.01547527e-01 -3.02803874e-01 -7.91673839e-01 -4.87434268e-01
-7.89337680e-02 -3.11328530e-01 -5.21528572e-02 9.87571478e-01
4.17307526e-01 1.70502082e-01 8.06003869e-01 -1.43340075e+00
-3.25186044e-01 1.36860764e+00 -1.42187506e-01 -2.14110330e-01
5.58867335e-01 5.11831939e-01 1.12109053e+00 -7.16996968e-01
4.07279998e-01 1.77068913e+00 4.96024877e-01 6.01871967e-01
-1.76766455e+00 -7.78069973e-01 1.04891956e-01 -1.02711551e-01
-1.40980303e+00 -5.58382094e-01 3.25359374e-01 -5.24245441e-01
1.07286024e+00 2.59625047e-01 3.44581872e-01 1.39323843e+00
2.27058381e-02 1.08758759e+00 1.09867585e+00 -5.23942292e-01
5.30617654e-01 1.51411453e-02 5.76413274e-02 8.18069220e-01
-2.02174485e-01 3.03326339e-01 -5.91399074e-01 -7.38273144e-01
5.20152748e-01 -6.23853922e-01 1.03303604e-01 -2.48688892e-01
-1.20422900e+00 1.07575631e+00 -2.80235559e-01 -1.45829663e-01
-6.49636686e-02 8.38662386e-01 2.14116931e-01 -1.21938847e-01
-1.21469945e-02 3.68349165e-01 -6.76906407e-01 -5.49103916e-01
-9.51698363e-01 8.47003639e-01 1.13042223e+00 1.07679772e+00
6.64319396e-01 -1.08293921e-01 -4.43004638e-01 7.08775222e-01
8.37901115e-01 8.14111412e-01 1.03845656e-01 -1.17623472e+00
3.14369023e-01 -2.40117744e-01 3.50318938e-01 -1.44551918e-01
-2.46451274e-01 -1.54505089e-01 -4.96034950e-01 6.95535028e-03
7.13935375e-01 -4.36656564e-01 -8.14625740e-01 2.16176867e+00
2.67031580e-01 1.70941070e-01 -1.15614951e-01 4.64433432e-01
3.03311646e-01 1.02480471e+00 4.51428413e-01 -1.17843747e-01
1.28128564e+00 -4.52638835e-01 -3.02668601e-01 -4.38368231e-01
7.68847406e-01 -7.88845003e-01 1.36964428e+00 5.63469946e-01
-1.25995553e+00 -1.33654833e-01 -6.61455154e-01 -4.21587117e-02
-8.97665918e-02 -7.42332041e-02 9.71748054e-01 6.92571819e-01
-1.04784000e+00 4.53795671e-01 -1.28042090e+00 7.22631216e-02
2.04873607e-01 2.16449022e-01 2.14563489e-01 -1.17798328e-01
-1.27739894e+00 8.46394122e-01 6.34997070e-01 -1.30946279e-01
-1.47611129e+00 -6.89067423e-01 -8.79857063e-01 1.16002858e-01
3.29215765e-01 -7.32623279e-01 1.99758589e+00 -2.37757683e-01
-1.79896736e+00 4.96335417e-01 -4.37350124e-01 -3.78408939e-01
4.52137619e-01 -8.60919356e-02 -4.78229225e-02 -5.60250342e-01
1.73017252e-02 8.82125854e-01 6.50359631e-01 -1.01860964e+00
-6.05227768e-01 3.32357101e-02 2.69684684e-03 1.12542845e-01
3.60524774e-01 3.90727967e-01 -4.50684398e-01 -3.82699281e-01
-1.60933405e-01 -8.11405480e-01 -3.98196489e-01 -4.76345539e-01
-7.84255683e-01 -6.03826225e-01 1.16133302e-01 -4.56472427e-01
1.26222718e+00 -1.90995181e+00 3.90641332e-01 3.19095343e-01
-1.48271456e-01 -8.26857761e-02 -3.57292980e-01 3.22166562e-01
2.08741784e-01 1.37169600e-01 -4.07403618e-01 -5.26554644e-01
7.15999305e-01 4.42806244e-01 -6.07951760e-01 1.28662601e-01
1.58201948e-01 1.01531708e+00 -9.52732742e-01 -4.42783505e-01
1.04683064e-01 1.60016432e-01 -9.75874603e-01 3.62415880e-01
-1.26947737e+00 3.19208533e-01 -3.96917611e-01 4.67911631e-01
5.06203175e-01 -3.03473711e-01 5.20277798e-01 2.94095546e-01
-2.60872301e-02 7.28169858e-01 -1.52361715e+00 1.73214781e+00
-6.44275308e-01 2.32511401e-01 2.09220555e-02 -5.98505676e-01
4.81719822e-01 2.12841630e-02 -4.50007051e-01 -1.20292522e-01
-8.46215412e-02 1.98627368e-01 -2.70875722e-01 -2.06566751e-01
4.26748276e-01 -5.93435504e-02 -6.52496517e-01 8.76407564e-01
2.03156963e-01 -8.31958771e-01 4.10670996e-01 3.77560973e-01
1.00829542e+00 5.85375249e-01 1.76762447e-01 -6.22025691e-02
9.42005143e-02 3.38667706e-02 5.71316719e-01 1.49461174e+00
2.56055713e-01 1.81503847e-01 8.43810558e-01 1.82207453e-03
-1.17904270e+00 -1.60169411e+00 -6.90533146e-02 1.61059427e+00
-6.10215306e-01 -5.61192572e-01 -8.65311146e-01 -4.09681797e-01
5.90705760e-02 1.46121514e+00 1.36948908e-02 -6.20181381e-04
-5.54792285e-01 -1.02108955e+00 9.00833845e-01 4.22844738e-01
-1.10132404e-01 -1.16110241e+00 -2.31109187e-01 3.36467475e-01
-1.71039149e-01 -7.76934803e-01 -4.40341711e-01 5.06066322e-01
-5.65204978e-01 -7.34226406e-01 -2.66946375e-01 -3.80532473e-01
4.19215918e-01 -6.11952245e-01 1.35011756e+00 -1.62492111e-01
-1.39493302e-01 1.11806393e-02 1.24564707e-01 -2.41717458e-01
-1.03724003e+00 2.64911532e-01 -5.53547144e-02 -6.06605530e-01
1.70055389e-01 -6.33245051e-01 4.52980697e-02 3.11070066e-02
-7.11697996e-01 5.64875841e-01 3.04885179e-01 9.15243387e-01
3.59828562e-01 -1.20311953e-01 1.78012118e-01 -1.00029981e+00
8.42410386e-01 -4.45849568e-01 -1.28499031e+00 3.26655746e-01
-2.45432660e-01 6.07286453e-01 4.78689909e-01 -3.69058967e-01
-1.14073384e+00 1.71427950e-01 -4.12981778e-01 4.36696410e-02
-3.92876744e-01 5.06450713e-01 -2.72268593e-01 5.45390666e-01
6.64466381e-01 4.24393296e-01 -1.49691477e-01 -3.55565637e-01
8.44343901e-01 6.21234715e-01 4.54459995e-01 -1.26937020e+00
7.03191459e-01 -7.66688958e-02 -2.69158870e-01 -1.53307885e-01
-9.52711582e-01 2.13023856e-01 -2.65424754e-02 2.65792280e-01
6.92748964e-01 -8.36052239e-01 -1.01729906e+00 6.08970881e-01
-1.34273422e+00 -1.03551137e+00 1.24383979e-01 3.38049203e-01
-1.04060638e+00 1.40750170e-01 -7.24719048e-01 -1.07524788e+00
-8.30487758e-02 -1.40576208e+00 1.13685095e+00 1.01729363e-01
-7.03733146e-01 -8.28823745e-01 4.17182185e-02 1.41213939e-01
3.16385090e-01 -3.91295999e-01 1.31069696e+00 -5.65015912e-01
-8.50889266e-01 6.08297177e-02 -1.47613967e-02 6.86182454e-02
-5.50849617e-01 2.15170071e-01 -9.45469677e-01 -1.93262249e-02
-5.30164480e-01 -5.69801509e-01 6.08429492e-01 4.24984604e-01
1.35363066e+00 -3.29014778e-01 -4.59579080e-01 6.65747106e-01
1.05501783e+00 1.03863457e-03 4.02152807e-01 -1.23593919e-02
2.85594076e-01 1.24617346e-01 4.15593535e-01 7.14993060e-01
5.95072925e-01 5.88911712e-01 1.30469441e-01 5.44326365e-01
2.38197640e-01 -6.95971429e-01 6.44877970e-01 4.12893981e-01
4.65315461e-01 -2.94691712e-01 -1.12303340e+00 2.88209915e-01
-1.92210281e+00 -9.54761744e-01 5.46527505e-02 2.01605868e+00
1.51465738e+00 3.13824326e-01 5.31701371e-04 -1.91415355e-01
6.42667770e-01 -6.97762594e-02 -3.80017966e-01 -6.53111637e-01
1.67085901e-01 4.40561682e-01 3.68048996e-01 1.12268126e+00
-9.00432110e-01 1.27082050e+00 6.99851274e+00 1.35891235e+00
-4.64522958e-01 2.27600291e-01 6.32103443e-01 -1.17866188e-01
-8.16593170e-01 3.60128999e-01 -1.30396998e+00 6.81886911e-01
1.31346667e+00 -4.58933860e-02 1.10356820e+00 8.91598225e-01
3.03604245e-01 -3.13272089e-01 -1.38129914e+00 6.47916257e-01
-4.76505905e-01 -1.40250266e+00 -1.91837981e-01 -3.62463713e-01
5.61010897e-01 2.07579598e-01 -7.69024864e-02 7.13007450e-01
1.59209406e+00 -1.14957964e+00 9.94040310e-01 6.06195271e-01
7.61824787e-01 -8.17074478e-01 2.46009663e-01 8.45118999e-01
-6.74811423e-01 -4.13986072e-02 -1.01052999e-01 4.94394191e-02
5.16222358e-01 7.23710835e-01 -1.03000391e+00 -1.71780869e-01
3.46403599e-01 -3.42840552e-02 -8.36304128e-02 7.82617152e-01
-8.41607749e-01 9.72162426e-01 -6.76920354e-01 -3.41283411e-01
1.14180446e-01 1.98591873e-02 4.53948647e-01 1.41310215e+00
3.13905478e-01 -2.03426525e-01 1.49332359e-01 1.48220122e+00
1.61498353e-01 -5.56629717e-01 -3.03940922e-01 -2.81013668e-01
9.73633826e-01 1.07575810e+00 -3.53385061e-01 -5.46930730e-01
2.72545163e-02 6.72029674e-01 5.37601173e-01 3.65973622e-01
-1.15814555e+00 -2.48855308e-01 6.28823280e-01 -2.94947475e-01
2.30980232e-01 -4.83411342e-01 -4.13714081e-01 -1.05459678e+00
-4.67121691e-01 -1.15896595e+00 3.90322536e-01 -1.10619318e+00
-9.55982983e-01 5.44280522e-02 4.21712637e-01 -2.07301676e-01
-9.08610642e-01 -5.63536584e-01 -5.30863464e-01 1.25452125e+00
-1.05368006e+00 -1.01293361e+00 2.31159091e-01 2.94562191e-01
4.91589874e-01 7.76328593e-02 1.04069388e+00 -1.61360383e-01
-4.97214645e-01 5.60095012e-01 7.03548789e-02 -2.67597765e-01
3.29142660e-01 -1.42427194e+00 7.98820198e-01 7.05602348e-01
2.99422145e-02 7.92206705e-01 9.92302120e-01 -6.35047257e-01
-1.65250170e+00 -8.70880425e-01 1.11375701e+00 -6.23069882e-01
9.22738135e-01 -8.54394853e-01 -5.50864995e-01 8.64638329e-01
-5.31960391e-02 -4.85866636e-01 4.39619869e-01 3.75617594e-01
-5.32780409e-01 4.13392276e-01 -8.83694649e-01 8.36875916e-01
9.22015667e-01 -6.15800381e-01 -5.12622178e-01 6.43129528e-01
7.68322766e-01 -7.41402447e-01 -6.17137909e-01 -7.16035953e-03
5.23160160e-01 -4.86038208e-01 8.97219837e-01 -6.56393468e-01
3.83851528e-01 -5.52534521e-01 -4.26305652e-01 -1.46278179e+00
-1.98139742e-01 -9.47170317e-01 -3.60297680e-01 1.17669106e+00
8.05438042e-01 -5.35432160e-01 6.24893248e-01 6.60945475e-01
2.48420183e-02 -2.89778709e-01 -8.20737362e-01 -3.83279413e-01
3.37774485e-01 -1.26162195e+00 9.45768178e-01 4.01529342e-01
-7.46004879e-02 9.71460864e-02 -1.97175816e-01 3.65538239e-01
7.14514911e-01 1.57473251e-01 8.14379215e-01 -7.51585662e-01
-1.08864105e+00 -4.68915820e-01 3.39327246e-01 -1.04805017e+00
4.61128861e-01 -1.19445717e+00 5.26868999e-01 -1.44677496e+00
2.80351907e-01 -6.76407158e-01 2.01864079e-01 7.50382721e-01
-1.17147356e-01 -1.17923595e-01 1.24890031e-02 -1.80260256e-01
-6.00191772e-01 3.98877114e-01 8.02224696e-01 7.71380439e-02
9.30123702e-02 2.68234760e-01 -6.53616846e-01 8.12011421e-01
7.74673223e-01 -6.70619965e-01 -2.57114947e-01 -6.06254160e-01
6.71154022e-01 4.53136355e-01 4.72092062e-01 -7.03963459e-01
2.79066801e-01 -5.27108192e-01 2.85455942e-01 -7.34675646e-01
2.27652907e-01 4.29935716e-02 3.07174444e-01 2.76681602e-01
-6.97399199e-01 -1.60090551e-01 2.38099098e-01 1.29438266e-01
2.21247062e-01 -5.24506450e-01 6.95659101e-01 -3.20043772e-01
-4.36940104e-01 9.80362948e-03 -8.67879868e-01 3.27101916e-01
6.72718525e-01 4.23949808e-01 -1.49477527e-01 -3.41701537e-01
-1.04652381e+00 5.10079801e-01 3.19197983e-01 1.14403598e-01
3.03923041e-01 -1.04570830e+00 -6.34760082e-01 2.11994216e-01
-7.48281181e-02 1.84324935e-01 3.01528350e-02 4.89001423e-01
-4.03938174e-01 5.11776388e-01 4.66851234e-01 -4.42350537e-01
-8.91164482e-01 1.29022717e-01 3.91097307e-01 -4.72385019e-01
-2.07727671e-01 1.23315763e+00 3.22682634e-02 -1.01077425e+00
3.44259977e-01 -4.00622249e-01 4.20074910e-01 -4.53396171e-01
5.76802790e-01 1.07541025e-01 -2.21317232e-01 3.02653790e-01
-2.64902502e-01 -5.91503419e-02 6.14338107e-02 -6.65755749e-01
1.09907532e+00 9.00578126e-02 -2.89823055e-01 4.14220065e-01
5.47849834e-01 -1.34032607e-01 -1.29853499e+00 -1.87319443e-01
2.15608433e-01 -1.16586179e-01 -3.76853980e-02 -1.22105169e+00
-2.82022625e-01 8.28848839e-01 9.13196579e-02 1.09475330e-02
6.49457037e-01 1.98784038e-01 3.87750238e-01 5.32052338e-01
4.62471128e-01 -9.87995923e-01 -4.25778419e-01 9.12836015e-01
6.80070579e-01 -9.08213675e-01 -5.02506971e-01 -1.17621519e-01
-4.86898661e-01 8.40815604e-01 4.28335041e-01 5.14425151e-02
4.24012214e-01 8.85907829e-01 -5.23212850e-01 2.12347269e-01
-1.40308130e+00 1.36302635e-01 -1.41956791e-01 6.25872672e-01
5.00568926e-01 5.11018097e-01 9.65776443e-02 8.30780089e-01
-5.28142929e-01 2.89500564e-01 3.65840107e-01 6.74077034e-01
-4.99393314e-01 -1.47125578e+00 -5.78309000e-01 5.26987553e-01
-3.99310827e-01 -5.89381158e-01 2.51915038e-01 3.43460321e-01
-7.93706998e-02 8.48742306e-01 2.77474552e-01 7.19439052e-03
-1.97229505e-01 4.99140948e-01 6.29649341e-01 -8.15430224e-01
-1.42686859e-01 7.40905553e-02 5.66280305e-01 -4.62507218e-01
4.24961239e-01 -9.92556274e-01 -1.34270728e+00 -4.50088024e-01
-5.95698915e-02 2.89792985e-01 7.99925983e-01 9.81881917e-01
1.76712975e-01 2.67697960e-01 2.22102955e-01 -7.92417049e-01
-1.24122381e+00 -1.02887166e+00 -4.15536314e-01 -1.85330868e-01
-4.77328412e-02 -3.99584442e-01 -2.61907727e-01 -1.25682592e-01] | [8.57483196258545, 6.836208820343018] |
fec7f6f0-b2d6-4671-95bd-88ea8b25c5cb | multi-granularity-semantic-aware-graph-model-1 | null | null | https://aclanthology.org/2022.findings-acl.95 | https://aclanthology.org/2022.findings-acl.95.pdf | Multi-Granularity Semantic Aware Graph Model for Reducing Position Bias in Emotion Cause Pair Extraction | The emotion cause pair extraction (ECPE) task aims to extract emotions and causes as pairs from documents. We observe that the relative distance distribution of emotions and causes is extremely imbalanced in the typical ECPE dataset. Existing methods have set a fixed size window to capture relations between neighboring clauses. However, they neglect the effective semantic connections between distant clauses, leading to poor generalization ability towards position-insensitive data. To alleviate the problem, we propose a novel \textbf{M}ulti-\textbf{G}ranularity \textbf{S}emantic \textbf{A}ware \textbf{G}raph model (MGSAG) to incorporate fine-grained and coarse-grained semantic features jointly, without regard to distance limitation. In particular, we first explore semantic dependencies between clauses and keywords extracted from the document that convey fine-grained semantic features, obtaining keywords enhanced clause representations. Besides, a clause graph is also established to model coarse-grained semantic relations between clauses. Experimental results indicate that MGSAG surpasses the existing state-of-the-art ECPE models. Especially, MGSAG outperforms other models significantly in the condition of position-insensitive data. | ['Songlin Hu', 'Wei Zhou', 'Lingwei Wei', 'Qianwen Ma', 'Yinan Bao'] | null | null | null | null | findings-acl-2022-5 | ['emotion-cause-pair-extraction'] | ['natural-language-processing'] | [-2.76605505e-03 6.78154901e-02 -3.54824007e-01 -7.33364642e-01
-7.66733825e-01 -6.13558590e-01 5.20533860e-01 5.01452923e-01
-1.94419295e-01 6.89806402e-01 4.67298031e-01 1.76121220e-01
-6.82684898e-01 -9.08307791e-01 -5.34635782e-01 -5.95813215e-01
7.51504302e-02 3.37923825e-01 -1.68259311e-02 -5.54105699e-01
3.03297848e-01 1.39579490e-01 -1.83504021e+00 5.89992166e-01
1.08377743e+00 1.14072788e+00 4.60994504e-02 -7.23918807e-03
-6.38146281e-01 8.35906625e-01 -7.19628155e-01 -6.14201486e-01
-3.19801629e-01 -4.92479533e-01 -8.75697196e-01 -7.34469667e-02
-3.25637348e-02 2.66237736e-01 -1.81186080e-01 1.13587177e+00
1.85455441e-01 1.07275367e-01 8.08776021e-01 -1.51366997e+00
-7.40204930e-01 9.23384607e-01 -7.45605528e-01 2.63912529e-01
4.78780240e-01 -7.41098046e-01 1.53922975e+00 -8.23665261e-01
7.07769871e-01 1.55226648e+00 2.96796709e-01 3.47449929e-01
-7.74987221e-01 -9.00773048e-01 7.19094396e-01 4.18277383e-01
-1.49922276e+00 8.34645256e-02 9.94854212e-01 1.13572635e-01
1.19278526e+00 6.29079342e-01 4.37298656e-01 1.06016171e+00
2.29354605e-01 9.92520392e-01 1.05303979e+00 -2.56398410e-01
1.88337103e-01 6.01263233e-02 4.77608263e-01 4.44304436e-01
9.13071334e-02 -2.97861665e-01 -7.88893521e-01 -1.15037434e-01
3.47501755e-01 -1.36395469e-01 -4.07122344e-01 7.67276436e-02
-1.03583455e+00 9.19666231e-01 6.59907937e-01 6.58119142e-01
-1.92167401e-01 -8.77426192e-02 5.03842831e-01 2.66181499e-01
5.09226918e-01 4.67839390e-01 -6.75935805e-01 1.52562093e-02
-6.41874373e-01 3.22152108e-01 5.21813631e-01 1.17894173e+00
7.84113407e-01 -3.33222926e-01 -1.89260170e-01 9.22887743e-01
1.21362627e-01 1.45324007e-01 5.94713211e-01 -4.61166203e-01
7.11238384e-01 1.02578974e+00 -2.87790149e-01 -1.62400174e+00
-4.73382533e-01 -5.11090994e-01 -7.68771350e-01 -6.67790949e-01
-2.72930562e-01 2.42801271e-02 -5.90137601e-01 1.71491849e+00
3.48502010e-01 -1.73530802e-01 2.29042694e-01 9.71655190e-01
1.25205338e+00 8.13801348e-01 1.66879177e-01 -3.24792564e-01
1.44783056e+00 -7.21638441e-01 -9.88106668e-01 -4.75543827e-01
8.16408575e-01 -6.56572998e-01 1.04391372e+00 3.98973614e-01
-6.62075877e-01 -1.69238225e-01 -1.00138760e+00 -2.06498414e-01
-7.37677395e-01 1.34204447e-01 8.53181839e-01 1.58629090e-01
-7.33158171e-01 2.75988907e-01 -8.92574415e-02 -2.23767579e-01
1.53390497e-01 3.19888055e-01 -3.65501583e-01 -1.51927635e-01
-1.79662979e+00 5.90562582e-01 8.49333763e-01 1.82644770e-01
-5.75125590e-02 -5.81142247e-01 -1.07453847e+00 1.92560136e-01
6.50728345e-01 -2.52750427e-01 7.04945862e-01 -8.40414107e-01
-7.22276628e-01 9.47257221e-01 -3.23137552e-01 1.64864305e-02
-1.45196423e-01 -2.24305078e-01 -8.40393186e-01 5.40063120e-02
3.46529990e-01 6.98122144e-01 5.95780551e-01 -1.35675001e+00
-7.25523829e-01 -6.65241122e-01 1.70841202e-01 5.32442629e-01
-6.23796821e-01 1.86479896e-01 -5.43246090e-01 -9.55868542e-01
4.16122139e-01 -4.39093858e-01 2.33442783e-01 -6.16473377e-01
-6.96602881e-01 -6.96628392e-01 8.76512527e-01 -3.66091311e-01
1.60781860e+00 -2.09496427e+00 2.29818285e-01 3.22459966e-01
4.00935024e-01 -2.02635363e-01 -8.58390778e-02 3.77926558e-01
-3.65623534e-01 2.51217395e-01 7.12643787e-02 9.63284746e-02
2.02127352e-01 4.32815492e-01 -3.46090198e-01 -6.46829307e-02
2.78070152e-01 7.76586771e-01 -8.69447052e-01 -7.51864731e-01
-1.60931841e-01 2.42185578e-01 -4.86028790e-01 8.02130848e-02
-2.27260262e-01 -1.45340636e-01 -9.13709462e-01 7.72998691e-01
6.54056609e-01 -2.24328250e-01 1.54483199e-01 -4.31401134e-01
2.43466452e-01 4.66867626e-01 -1.07689369e+00 1.44260526e+00
-3.07472855e-01 1.74163193e-01 -6.29257113e-02 -1.11701429e+00
1.36651683e+00 1.04248747e-01 4.89377320e-01 -9.54544365e-01
3.75799775e-01 1.93037257e-01 -2.42487609e-01 -4.02408063e-01
4.54045266e-01 -2.96410173e-01 -6.48940742e-01 6.65497035e-02
-1.74681842e-02 -8.85186121e-02 2.60371655e-01 3.86388332e-01
9.61740911e-01 -1.47222161e-01 3.33497435e-01 -4.00606871e-01
5.77916026e-01 -4.44827601e-03 8.54299724e-01 1.04674682e-01
-1.98932197e-02 3.75871718e-01 9.28637743e-01 -5.54375760e-02
-2.77643144e-01 -8.87770534e-01 -3.89720313e-02 1.19581163e+00
6.47190928e-01 -8.87708485e-01 -6.49003506e-01 -9.39443171e-01
-1.18217982e-01 9.13670659e-01 -8.02330136e-01 -3.10968518e-01
-4.24755782e-01 -9.49883461e-01 4.68984425e-01 6.39312625e-01
6.43247187e-01 -9.62930620e-01 -9.72930863e-02 6.97806031e-02
-6.44686520e-01 -1.01009500e+00 -2.60893732e-01 4.22177732e-01
-3.95454764e-01 -1.05187500e+00 -1.40016839e-01 -8.51330459e-01
5.18441200e-01 -1.61288027e-02 1.34755921e+00 -3.63004347e-03
-1.96137786e-01 -1.70078561e-01 -6.88308358e-01 -4.34770733e-01
3.13009024e-01 -3.69790457e-02 -2.21832767e-01 -2.07313284e-01
1.14680600e+00 -3.37391317e-01 -4.74086136e-01 2.36567855e-01
-8.78573418e-01 1.41520910e-02 4.92260158e-01 7.41894543e-01
8.37945879e-01 5.38079858e-01 7.37403452e-01 -8.29125822e-01
7.79925048e-01 -6.80353761e-01 -4.87590134e-02 3.73077214e-01
-6.92107439e-01 8.96829925e-03 5.94828069e-01 -1.28828764e-01
-1.25265980e+00 -5.56447983e-01 -4.39715534e-02 -2.49635428e-01
-1.85360819e-01 6.95504606e-01 -6.75970376e-01 6.59438670e-01
2.01731712e-01 1.29366204e-01 -6.29217327e-01 -2.03047827e-01
3.26696575e-01 6.91309690e-01 5.06413877e-01 -7.37307966e-01
3.95604193e-01 3.43440890e-01 -1.82357177e-01 -3.78510177e-01
-1.29498100e+00 -5.66113234e-01 -3.41404229e-01 -1.09649934e-01
8.94833505e-01 -9.78454471e-01 -5.61433613e-01 -1.43661555e-02
-1.10452890e+00 3.00179839e-01 -3.14261615e-02 3.80087256e-01
-2.94074565e-01 1.48144197e-02 -7.36676097e-01 -4.56317604e-01
-1.62120745e-01 -6.81001842e-01 1.39651191e+00 3.11330378e-01
-6.01449549e-01 -8.59883785e-01 -2.75266796e-01 4.93130833e-01
-2.15496108e-01 4.77539182e-01 1.33494103e+00 -8.47743571e-01
-2.23380238e-01 -5.76522425e-02 -5.38764894e-01 2.98372433e-02
3.65474015e-01 -1.85072012e-02 -7.06943989e-01 1.85888946e-01
1.32202908e-01 -4.63430345e-01 9.01691198e-01 7.67077655e-02
1.27202928e+00 -5.54824173e-01 -5.83691120e-01 3.28095645e-01
1.37324405e+00 3.24443936e-01 4.93833840e-01 3.38615149e-01
7.20106781e-01 9.77293193e-01 1.21314573e+00 4.96266633e-01
6.65042341e-01 3.20574850e-01 4.80670959e-01 -2.34482601e-01
2.60243148e-01 -1.95762575e-01 4.95357253e-02 8.85391414e-01
1.34999648e-01 -5.69695830e-01 -6.96773529e-01 6.44563079e-01
-1.81396723e+00 -7.03453898e-01 -3.44514608e-01 1.49458539e+00
9.67406392e-01 9.03199315e-02 -3.56395215e-01 3.30516189e-01
7.91020632e-01 3.06795210e-01 -2.68626690e-01 -5.14614880e-01
-3.68552953e-01 8.06137025e-02 -8.21491182e-02 1.48379102e-01
-1.05922711e+00 1.02724969e+00 4.34594822e+00 1.40546095e+00
-8.15400958e-01 -8.41191635e-02 7.89144337e-01 -2.05099791e-01
-7.98706830e-01 -3.13377887e-01 -8.43451440e-01 4.65950847e-01
3.47865731e-01 -1.52073950e-01 1.42576560e-01 7.82223701e-01
-9.46391150e-02 6.19557910e-02 -1.03215015e+00 9.81031239e-01
3.73001158e-01 -9.67473149e-01 2.31688634e-01 -9.42595750e-02
6.54386997e-01 -5.00745893e-01 -1.14176780e-01 4.68226075e-01
5.20354174e-02 -1.09536076e+00 6.37483060e-01 2.26076797e-01
7.10247040e-01 -1.36559844e+00 1.07316005e+00 6.65562153e-02
-1.53480780e+00 4.82645147e-02 -3.73066962e-01 1.54261859e-02
-1.21931799e-01 8.88120115e-01 -2.78074056e-01 9.27509904e-01
1.05160105e+00 8.10862184e-01 -5.90661764e-01 2.29536667e-01
-3.71965468e-01 3.05440187e-01 -1.49930313e-01 -3.90244901e-01
5.58479786e-01 -1.31964371e-01 3.68518531e-01 1.28300691e+00
3.76895308e-01 5.49955547e-01 4.72232327e-02 9.37411129e-01
-2.41129413e-01 3.61970454e-01 -3.19783658e-01 -1.18764475e-01
5.25399148e-01 1.27604938e+00 -1.02391315e+00 -1.72947019e-01
-2.50547469e-01 1.02289629e+00 5.64015448e-01 3.30588490e-01
-1.02568519e+00 -7.19111204e-01 7.39553869e-01 -2.45720997e-01
3.20119083e-01 4.41307753e-01 -2.38409191e-01 -1.03536868e+00
2.53775239e-01 -7.50392497e-01 7.16561139e-01 -8.00381839e-01
-1.54585397e+00 7.14056969e-01 7.16042817e-02 -8.41522574e-01
-2.29598030e-01 -4.79195893e-01 -5.05410433e-01 5.14957190e-01
-1.41302741e+00 -1.11364532e+00 -2.79054016e-01 8.09541523e-01
5.27350008e-01 8.53328779e-02 7.21300423e-01 1.30657941e-01
-5.83194852e-01 4.44371492e-01 -2.00249657e-01 1.00132838e-01
6.68231130e-01 -1.36674035e+00 -3.04212242e-01 5.26511550e-01
4.47318405e-02 7.91910708e-01 7.45204329e-01 -8.47665846e-01
-1.02267671e+00 -1.16023767e+00 1.37853193e+00 -2.25050360e-01
5.61446488e-01 -4.63084847e-01 -9.47058082e-01 5.92428029e-01
2.52642035e-01 -1.69802025e-01 8.46386850e-01 4.65852708e-01
-5.05648196e-01 -1.88111857e-01 -1.08312821e+00 5.35491943e-01
1.01130426e+00 -4.78803486e-01 -1.02379823e+00 2.96495616e-01
9.76554334e-01 -1.79625601e-01 -8.48414421e-01 5.89155257e-01
3.38671356e-02 -1.05365336e+00 8.55522394e-01 -4.68675941e-01
8.47133517e-01 -1.77684680e-01 -2.82306194e-01 -1.19158030e+00
-2.26156190e-01 -2.09757656e-01 -8.88619125e-02 1.85016060e+00
4.22214270e-01 -3.92228693e-01 5.50767124e-01 5.15300751e-01
7.95431733e-02 -1.02918208e+00 -7.78744996e-01 -6.84312046e-01
7.71989152e-02 -3.55757266e-01 9.89836156e-01 1.36089730e+00
3.40710938e-01 6.01704776e-01 1.07015431e-01 3.34743336e-02
2.39352614e-01 5.59345603e-01 -1.42447585e-02 -1.17414355e+00
-7.79456943e-02 -4.67672825e-01 -1.75720081e-01 -6.56543016e-01
4.13914561e-01 -9.25384283e-01 1.12406656e-01 -1.61104119e+00
4.12567437e-01 -4.66359526e-01 -6.33611023e-01 5.35601377e-01
-5.61008871e-01 8.91980380e-02 -5.39360642e-02 -9.53393430e-02
-7.77593970e-01 8.86587322e-01 1.12362790e+00 -4.58866842e-02
6.42874688e-02 -4.14085001e-01 -1.08338821e+00 9.15165067e-01
6.55051172e-01 -4.90477860e-01 -8.18326056e-01 -2.00343549e-01
6.34445965e-01 -1.96657911e-01 2.32832253e-01 -5.00905812e-01
9.91961807e-02 -2.66351014e-01 4.54432726e-01 -8.13475132e-01
2.94546574e-01 -7.24072099e-01 -2.60913372e-01 -1.31345406e-01
-3.91049266e-01 1.65391684e-01 9.69725847e-02 5.73496878e-01
-7.26198256e-01 -3.17562893e-02 1.41768858e-01 -2.84095090e-02
-8.72150123e-01 9.65454674e-04 -8.45461637e-02 3.11132908e-01
9.41195905e-01 -2.79640071e-02 -4.35965866e-01 -1.69589549e-01
-3.73154908e-01 4.39566165e-01 5.37480824e-02 6.36722744e-01
7.07639754e-01 -1.47909093e+00 -4.77797985e-01 -7.36847818e-02
4.45467472e-01 1.13781661e-01 3.97067070e-01 5.33108234e-01
8.90366510e-02 6.68062449e-01 1.40025437e-01 -2.01902077e-01
-1.44343662e+00 7.68726051e-01 -3.47865932e-02 -5.22511423e-01
-2.43553132e-01 1.28007042e+00 5.67472160e-01 -5.02051890e-01
1.29141062e-01 -4.40178752e-01 -4.94671404e-01 5.11370599e-01
3.17499876e-01 1.52417943e-01 1.66587874e-01 -6.91530049e-01
-6.63905859e-01 4.39343512e-01 -2.39652485e-01 3.15913647e-01
1.26698208e+00 -3.41777861e-01 -5.47563136e-01 3.92723829e-01
1.30264878e+00 -9.82280914e-03 -6.16906166e-01 -8.34678784e-02
1.23780303e-01 -2.54595518e-01 1.43759653e-01 -9.55360472e-01
-1.15870094e+00 6.21985793e-01 1.65210217e-02 2.13895217e-01
1.52432406e+00 4.10686970e-01 7.83561647e-01 1.69194683e-01
3.95344704e-01 -1.39313173e+00 6.81281611e-02 2.94954032e-01
1.16258180e+00 -9.79271173e-01 -1.19988188e-01 -1.06603253e+00
-7.93620706e-01 9.44541037e-01 9.74054575e-01 9.25732404e-02
4.89214987e-01 3.65697831e-01 -1.95146203e-01 -6.98876262e-01
-8.05363595e-01 -3.07292461e-01 5.15775144e-01 3.49549860e-01
6.60108387e-01 1.30430087e-01 -7.13530779e-01 1.36109805e+00
-4.78097171e-01 -3.00977677e-01 2.79440470e-02 7.93656528e-01
-2.57939845e-01 -9.61117685e-01 -2.54762828e-01 4.42830861e-01
-4.00019914e-01 -3.34204882e-01 -9.16583776e-01 1.04471624e+00
5.07625163e-01 1.14910293e+00 1.50812194e-01 -3.98183763e-01
1.42794847e-01 -9.11923572e-02 1.43015102e-01 -3.75307500e-01
-6.38154328e-01 3.33582252e-01 3.24151516e-01 -7.26134181e-01
-5.05881131e-01 -5.26004374e-01 -1.89198136e+00 -2.65502214e-01
-3.32003623e-01 2.87708849e-01 1.23967722e-01 1.19729042e+00
2.83674568e-01 9.80172753e-01 5.60529590e-01 -4.17300947e-02
3.28161716e-02 -6.65388346e-01 -9.74716544e-01 5.38939536e-01
-7.96130374e-02 -6.25704467e-01 -4.49413538e-01 -3.46273303e-01] | [12.620035171508789, 6.212493419647217] |
d42916db-7612-4ee5-88f0-b24764c94044 | towards-table-to-text-generation-with | null | null | https://aclanthology.org/2021.acl-long.115 | https://aclanthology.org/2021.acl-long.115.pdf | Towards Table-to-Text Generation with Numerical Reasoning | Recent neural text generation models have shown significant improvement in generating descriptive text from structured data such as table formats. One of the remaining important challenges is generating more analytical descriptions that can be inferred from facts in a data source. The use of a template-based generator and a pointer-generator is among the potential alternatives for table-to-text generators. In this paper, we propose a framework consisting of a pre-trained model and a copy mechanism. The pre-trained models are fine-tuned to produce fluent text that is enriched with numerical reasoning. However, it still lacks fidelity to the table contents. The copy mechanism is incorporated in the fine-tuning step by using general placeholders to avoid producing hallucinated phrases that are not supported by a table while preserving high fluency. In summary, our contributions are (1) a new dataset for numerical table-to-text generation using pairs of a table and a paragraph of a table description with richer inference from scientific papers, and (2) a table-to-text generation framework enriched with numerical reasoning. | ['Hiroya Takamura', 'Manabu Okumura', 'Kotaro Funakoshi', 'Hidetaka Kamigaito', 'Lya Hulliyyatus Suadaa'] | 2021-08-01 | null | null | null | acl-2021-5 | ['table-to-text-generation'] | ['natural-language-processing'] | [ 2.03481823e-01 8.79383564e-01 -3.40624377e-02 -3.68209988e-01
-1.00146747e+00 -7.50357211e-01 9.47499454e-01 5.28262973e-01
1.75277553e-02 1.23050117e+00 7.39252627e-01 -3.17886293e-01
2.39456788e-01 -1.36014569e+00 -8.18426073e-01 9.96831805e-02
5.69841623e-01 7.86342442e-01 -2.37611964e-01 -5.40364802e-01
4.92696851e-01 3.67092609e-01 -1.45585096e+00 8.18862438e-01
1.19795346e+00 5.89691579e-01 2.45237321e-01 5.67015588e-01
-6.92875683e-01 1.16199303e+00 -1.02506840e+00 -9.28065956e-01
8.42211917e-02 -8.23238671e-01 -7.90887773e-01 -9.23454836e-02
4.08402801e-01 -4.43988770e-01 -1.61022797e-01 6.59422815e-01
5.73647261e-01 1.50681987e-01 8.90147746e-01 -1.30605042e+00
-1.01513743e+00 1.43071878e+00 6.82439730e-02 -1.94162965e-01
6.82884276e-01 2.33957574e-01 9.05472577e-01 -7.57963240e-01
1.05433714e+00 1.42239439e+00 5.25201023e-01 9.11434710e-01
-1.27127898e+00 -5.81352890e-01 -1.28593028e-01 -1.37436941e-01
-1.14508140e+00 -6.24107659e-01 5.93476892e-01 -4.43776041e-01
1.23847008e+00 2.24742115e-01 8.07093143e-01 1.43125618e+00
2.26170704e-01 4.73091096e-01 8.32711399e-01 -5.40959299e-01
3.37264001e-01 5.30654430e-01 -3.95472825e-01 5.85386276e-01
5.63516796e-01 -2.73656577e-01 -8.91980469e-01 6.78081363e-02
8.14880848e-01 -5.11770904e-01 -1.47317210e-02 2.42239475e-01
-1.46159708e+00 9.61019397e-01 1.72729865e-01 6.39005527e-02
-4.89935279e-01 -1.12520255e-01 5.07458210e-01 2.33575199e-02
2.03084573e-01 1.16426003e+00 -1.54356599e-01 -3.45045984e-01
-1.32583177e+00 9.98284996e-01 1.20919681e+00 1.62624192e+00
3.90536159e-01 1.86396316e-01 -9.91434753e-01 4.30862039e-01
5.40180504e-03 4.64629114e-01 4.97781962e-01 -8.46476614e-01
9.38455999e-01 6.71751082e-01 1.52876332e-01 -8.89415741e-01
-9.36861858e-02 -3.80096495e-01 -8.87996972e-01 3.82586382e-02
3.39743704e-01 -1.86778471e-01 -7.69338846e-01 1.80002487e+00
-1.65975705e-01 -6.35617673e-01 4.35897321e-01 2.97801316e-01
1.41905701e+00 7.39724457e-01 1.91669613e-02 1.18792499e-03
1.40535569e+00 -8.34248722e-01 -1.16237926e+00 -9.60535780e-02
6.63833141e-01 -5.35602570e-01 1.23973763e+00 7.31420144e-02
-1.76957643e+00 -6.44529402e-01 -1.06623590e+00 -7.74960935e-01
-9.80933368e-01 2.40288317e-01 4.63443458e-01 4.73870397e-01
-1.28597689e+00 5.88276625e-01 -3.39775056e-01 -2.23027855e-01
4.30666655e-01 -1.06297746e-01 -2.36967579e-01 1.53388888e-01
-1.40758967e+00 9.51363564e-01 7.25543439e-01 -2.59714633e-01
-4.25589204e-01 -1.00668490e+00 -1.18295491e+00 2.73293078e-01
2.73699045e-01 -1.31033206e+00 1.24739635e+00 -1.88800365e-01
-1.49352026e+00 7.47529924e-01 -2.36243308e-01 -4.90471721e-01
7.51779497e-01 1.49936453e-01 -2.34326765e-01 -8.35305899e-02
4.34853822e-01 1.18576968e+00 3.82463157e-01 -1.05583751e+00
-2.99027115e-01 -4.88853678e-02 9.78079587e-02 2.30771229e-01
1.71428137e-02 -2.18999490e-01 -9.30572599e-02 -1.09379208e+00
-3.27881515e-01 -4.12984639e-01 -4.49254699e-02 -1.31271064e-01
-1.02516150e+00 -3.04750264e-01 1.47570878e-01 -8.63967896e-01
1.46263707e+00 -1.38039756e+00 -8.93646851e-02 -1.17915735e-01
3.88048142e-02 -1.48098215e-01 -1.18194178e-01 9.88666236e-01
1.10775821e-01 5.90611875e-01 -2.07049504e-01 -3.25698286e-01
3.63044798e-01 -2.13842899e-01 -8.14970911e-01 -6.83180392e-01
6.50995672e-01 1.18485403e+00 -9.86415386e-01 -6.16387904e-01
-2.63850033e-01 2.82471657e-01 -7.82152355e-01 2.40489125e-01
-8.05281043e-01 2.54390389e-01 -1.87249124e-01 4.19941545e-01
5.12729824e-01 -2.27669328e-01 -1.90809444e-01 -2.78511077e-01
-2.62946159e-01 9.41760838e-01 -1.13133645e+00 1.86471140e+00
-4.67506707e-01 4.67072606e-01 -5.04218102e-01 -2.62788385e-01
1.10413611e+00 4.53094482e-01 -2.04971999e-01 -5.05147457e-01
-1.34104997e-01 2.38104910e-01 -5.26542626e-02 -4.27602500e-01
1.09422326e+00 -2.96221763e-01 -2.60950327e-01 7.26340592e-01
1.85407251e-01 -8.19497705e-01 1.01590908e+00 6.89895272e-01
9.29585278e-01 4.69278216e-01 4.91667092e-01 -1.18877068e-01
3.96963745e-01 2.74776667e-01 1.87324882e-01 1.09673369e+00
5.98318875e-01 7.70449638e-01 8.02179158e-01 -8.68323743e-02
-1.38184798e+00 -9.94390309e-01 7.97013864e-02 5.22307754e-01
-5.68490922e-01 -9.35800493e-01 -8.17715764e-01 -4.24818993e-01
-9.52179804e-02 1.47191465e+00 -7.83622146e-01 -1.41908765e-01
-4.53593642e-01 -2.01591849e-01 9.01685536e-01 8.18554461e-01
5.35937786e-01 -1.49867213e+00 -6.25320435e-01 4.00389254e-01
-6.36060953e-01 -1.02069032e+00 -3.43309999e-01 8.40154290e-02
-7.90754557e-01 -4.54827756e-01 -6.18014216e-01 -5.28895617e-01
7.73462832e-01 -5.35063922e-01 1.55419254e+00 -1.65619642e-01
4.17870060e-02 -1.56262860e-01 -2.21673414e-01 -9.52274859e-01
-1.10415602e+00 3.97254020e-01 -3.68979722e-01 -7.24343121e-01
4.53195982e-02 -4.30942476e-01 -1.14001922e-01 -4.26808387e-01
-1.02111828e+00 8.13475966e-01 6.50525451e-01 7.77717650e-01
5.15596032e-01 -1.38311416e-01 8.21462095e-01 -1.00103354e+00
1.16198552e+00 -4.29343522e-01 -3.60054910e-01 2.57718593e-01
-5.81270158e-01 5.35662711e-01 1.08068502e+00 -2.04916343e-01
-1.36235011e+00 -2.20115289e-01 -1.67076752e-01 1.74964651e-01
-1.03013806e-01 6.85637236e-01 -4.01585579e-01 8.37404072e-01
8.46812367e-01 3.98105949e-01 -3.95861603e-02 -2.41679057e-01
7.97293365e-01 4.73179638e-01 6.80526078e-01 -8.53665352e-01
7.86268771e-01 1.30235493e-01 6.79920614e-02 -2.19962820e-01
-7.69995749e-01 2.66560376e-01 -6.53205574e-01 6.90995716e-03
7.22225189e-01 -9.12342072e-01 -3.98303241e-01 1.67827398e-01
-1.44680810e+00 -4.58161116e-01 -7.16619909e-01 1.45536521e-02
-7.97688901e-01 -9.33362991e-02 -6.16418779e-01 -6.37555778e-01
-6.57493770e-01 -8.14022422e-01 1.01609707e+00 3.48660737e-01
-7.82042980e-01 -8.89789701e-01 -4.56992164e-02 1.64901346e-01
6.33544147e-01 3.97691548e-01 1.36439502e+00 -8.13431799e-01
-5.80004096e-01 -7.75120333e-02 -2.20412582e-01 -1.91619724e-01
1.86272711e-01 1.92568600e-01 -7.77309000e-01 2.58235782e-01
-3.09714526e-01 -5.74907959e-01 5.67746043e-01 4.03714143e-02
1.17771578e+00 -1.03187442e+00 -1.62686899e-01 4.50932741e-01
1.08927715e+00 2.99101174e-01 8.53255749e-01 3.08717459e-01
5.29648542e-01 7.55464435e-01 1.54070675e-01 4.98400331e-01
7.46054709e-01 4.66124266e-01 -1.25723958e-01 1.29409268e-01
-4.82411057e-01 -9.74884152e-01 1.51828513e-01 6.51542842e-01
2.04904243e-01 -5.97439051e-01 -7.32203722e-01 5.63377976e-01
-1.39366376e+00 -1.41697705e+00 1.43076051e-02 1.90292847e+00
1.54110587e+00 1.76133469e-01 5.23403566e-03 1.99398056e-01
4.01570588e-01 -9.38045010e-02 -5.00004411e-01 -6.58185005e-01
-2.72439450e-01 3.84163916e-01 -7.57884830e-02 3.87176633e-01
-4.47034657e-01 9.15166736e-01 6.33359003e+00 6.56847715e-01
-8.38691831e-01 -3.90726119e-01 5.82251310e-01 -1.08673580e-01
-8.50523949e-01 -5.26365526e-02 -1.10343039e+00 4.42281514e-01
1.08414888e+00 -7.86022484e-01 2.79201120e-01 5.89857936e-01
3.11965257e-01 1.91036642e-01 -1.43167770e+00 6.69975579e-01
2.77407199e-01 -1.79209423e+00 9.54131365e-01 -1.78955287e-01
6.69269562e-01 -8.45182180e-01 -1.15043208e-01 5.70969641e-01
3.82061452e-01 -1.34301925e+00 1.34223926e+00 7.24685669e-01
1.07812881e+00 -5.78308821e-01 5.48318982e-01 2.44888276e-01
-7.97156692e-01 2.95537382e-01 -2.94898480e-01 -5.18948548e-02
2.37511411e-01 6.71699286e-01 -1.10296905e+00 5.83115220e-01
3.01309437e-01 4.91882890e-01 -7.30708897e-01 7.03713953e-01
-5.26347399e-01 1.72176808e-01 6.37195781e-02 -4.07358736e-01
-3.01024411e-02 1.85928065e-02 3.85730654e-01 1.40514374e+00
6.34807169e-01 1.23687081e-01 -1.77756041e-01 1.84606445e+00
-4.15080696e-01 1.57168284e-01 -8.48859072e-01 -5.34927070e-01
6.97069645e-01 1.22455466e+00 -6.63428664e-01 -7.14316666e-01
-1.17217630e-01 8.53189528e-01 4.33346897e-01 1.73359469e-01
-6.04689777e-01 -9.35699463e-01 2.06278097e-02 3.09775174e-01
1.23522736e-01 4.74808402e-02 -9.09966171e-01 -1.11037016e+00
1.31400764e-01 -1.05113852e+00 2.27242991e-01 -1.43517053e+00
-1.17879987e+00 7.71112859e-01 2.39482224e-01 -8.49174440e-01
-8.58029485e-01 -3.60304207e-01 -6.03909194e-01 1.30913472e+00
-1.11304784e+00 -1.15081835e+00 -3.56221080e-01 1.59366190e-01
5.87058961e-01 -2.67040938e-01 1.08485389e+00 -2.69738823e-01
-1.83807492e-01 7.06163287e-01 -4.82933402e-01 3.46512973e-01
6.31926358e-01 -1.57558584e+00 8.30174565e-01 9.09397304e-01
-8.42552707e-02 1.15284979e+00 8.14576387e-01 -1.17178941e+00
-1.00789106e+00 -1.14708817e+00 1.48981869e+00 -6.26431346e-01
4.07115877e-01 -7.47410119e-01 -9.24585044e-01 5.93128383e-01
6.60376728e-01 -6.81706369e-01 7.12326109e-01 -1.91381246e-01
-4.00531471e-01 2.15992764e-01 -1.26552474e+00 1.10350561e+00
9.85709131e-01 -4.32591230e-01 -7.39700079e-01 3.50660861e-01
8.80339861e-01 -8.28330159e-01 -8.12563598e-01 -2.75093988e-02
4.76983160e-01 -7.13566005e-01 5.99827766e-01 -5.81478775e-01
1.41376507e+00 -2.82411367e-01 2.53437847e-01 -1.60652709e+00
-2.84059167e-01 -9.28636730e-01 -2.49420881e-01 1.60506010e+00
9.45625484e-01 -2.45522171e-01 5.79764187e-01 8.80895793e-01
-2.65610665e-01 -5.74256301e-01 -4.47632581e-01 -6.89874470e-01
4.48471278e-01 -1.06903665e-01 1.01055908e+00 7.98843741e-01
4.31600720e-01 6.98138297e-01 -7.85543323e-02 -5.47644258e-01
2.56125182e-01 1.92200616e-02 6.98829591e-01 -1.09935975e+00
-1.86572298e-01 -5.89719653e-01 1.45911217e-01 -6.64854169e-01
2.36215163e-02 -1.27540112e+00 1.44808426e-01 -2.24978447e+00
3.66303742e-01 -1.84362352e-01 4.56320494e-01 6.10028923e-01
-2.58519322e-01 -1.89761057e-01 3.30175161e-01 -1.16455361e-01
-2.13496592e-02 5.40684044e-01 1.40507805e+00 -9.76821929e-02
-1.93630055e-01 -2.89159060e-01 -1.33331096e+00 1.48056090e-01
8.74476612e-01 -5.26475847e-01 -6.09204769e-01 -2.75213689e-01
7.46258974e-01 3.90897155e-01 2.47432917e-01 -1.06287563e+00
2.84529388e-01 -1.98897675e-01 6.73683465e-01 -8.95444632e-01
1.53572768e-01 6.42510084e-03 4.63363305e-02 2.61317462e-01
-1.16299975e+00 5.45126200e-01 5.08288026e-01 1.24758311e-01
2.28993408e-02 -3.74744385e-01 3.81971985e-01 -6.82246983e-01
2.58497652e-02 -5.84442690e-02 -6.52249753e-01 5.04217207e-01
4.61080045e-01 -5.47162853e-02 -7.83178031e-01 -6.72551692e-01
-1.41232595e-01 -4.73942123e-02 4.46610570e-01 5.35554290e-01
5.86003721e-01 -1.29540026e+00 -1.07075512e+00 1.36676103e-01
1.84403867e-01 1.77768528e-01 -1.72881290e-01 1.63130596e-01
-4.92565036e-01 8.47776413e-01 -5.82890332e-01 2.23961145e-01
-6.12431288e-01 5.42824984e-01 2.16220170e-02 -5.71611345e-01
-6.11168802e-01 5.67220688e-01 -2.04713672e-01 -6.12671137e-01
3.43359597e-02 -7.61952579e-01 -1.78968042e-01 1.55353010e-01
7.60387301e-01 2.22843736e-01 1.99812874e-01 -1.93532422e-01
3.22703570e-02 -1.70101941e-01 -1.63138807e-01 -7.75835752e-01
1.09203243e+00 1.56171054e-01 -1.36929125e-01 4.79552984e-01
5.69422662e-01 4.87543434e-01 -7.56623030e-01 1.07581466e-01
-1.89135984e-01 1.01190627e-01 -3.16218346e-01 -1.59350073e+00
-5.64338803e-01 8.24854016e-01 -4.55389977e-01 1.54890567e-01
7.84780443e-01 -1.71896890e-01 8.01197112e-01 5.41581333e-01
4.76961546e-02 -8.78423691e-01 1.53411925e-01 6.87492549e-01
1.48934174e+00 -8.67702961e-01 -1.24111384e-01 -4.00674403e-01
-8.57540786e-01 1.38979936e+00 8.00743639e-01 3.24964941e-01
-2.33056918e-02 6.10458016e-01 -9.67723355e-02 -1.32257923e-01
-1.13382065e+00 -8.31019655e-02 3.65108639e-01 8.86799991e-01
8.77192557e-01 -2.57485718e-01 -3.14148366e-01 9.55994964e-01
-1.09864116e+00 2.95757979e-01 1.01242113e+00 8.44019234e-01
2.34998446e-02 -1.00833428e+00 -2.78615236e-01 6.81496620e-01
-3.21292132e-01 -6.57267272e-01 -5.95857084e-01 6.95233583e-01
-5.15234880e-02 1.01710844e+00 1.32755965e-01 -6.68515116e-02
3.15142363e-01 4.97620165e-01 5.84145069e-01 -8.85095716e-01
-9.75205839e-01 -5.09692907e-01 2.99617112e-01 -1.56918690e-01
1.27479970e-01 -5.55708528e-01 -1.32860672e+00 -4.16995555e-01
2.26873413e-01 2.64971584e-01 5.62962472e-01 6.96691811e-01
6.50986016e-01 8.88488173e-01 -1.11014552e-01 -4.61533725e-01
-4.98470902e-01 -1.17932141e+00 -2.65853196e-01 4.21545923e-01
-8.01812485e-03 -2.61682451e-01 -7.03122914e-02 4.26336348e-01] | [11.599912643432617, 8.82263469696045] |
89ce245d-3e7f-43da-9a57-e113d3f608ca | bidirectional-machine-reading-comprehension | 2103.07665 | null | https://arxiv.org/abs/2103.07665v1 | https://arxiv.org/pdf/2103.07665v1.pdf | Bidirectional Machine Reading Comprehension for Aspect Sentiment Triplet Extraction | Aspect sentiment triplet extraction (ASTE), which aims to identify aspects from review sentences along with their corresponding opinion expressions and sentiments, is an emerging task in fine-grained opinion mining. Since ASTE consists of multiple subtasks, including opinion entity extraction, relation detection, and sentiment classification, it is critical and challenging to appropriately capture and utilize the associations among them. In this paper, we transform ASTE task into a multi-turn machine reading comprehension (MTMRC) task and propose a bidirectional MRC (BMRC) framework to address this challenge. Specifically, we devise three types of queries, including non-restrictive extraction queries, restrictive extraction queries and sentiment classification queries, to build the associations among different subtasks. Furthermore, considering that an aspect sentiment triplet can derive from either an aspect or an opinion expression, we design a bidirectional MRC structure. One direction sequentially recognizes aspects, opinion expressions, and sentiments to obtain triplets, while the other direction identifies opinion expressions first, then aspects, and at last sentiments. By making the two directions complement each other, our framework can identify triplets more comprehensively. To verify the effectiveness of our approach, we conduct extensive experiments on four benchmark datasets. The experimental results demonstrate that BMRC achieves state-of-the-art performances. | ['Yuelin Wang', 'Jie Liu', 'Yu Wang', 'Shaowei Chen'] | 2021-03-13 | null | null | null | null | ['aspect-sentiment-triplet-extraction'] | ['natural-language-processing'] | [ 2.76056051e-01 -1.07149065e-01 -2.28858739e-01 -5.71367323e-01
-6.84401274e-01 -7.38258839e-01 5.47036529e-01 2.20545724e-01
-2.85970181e-01 5.31086922e-01 4.03711587e-01 -6.05137706e-01
7.73051009e-02 -7.66156852e-01 -4.22289819e-01 -5.66527188e-01
6.20457590e-01 1.55201867e-01 2.08817318e-01 -4.27535385e-01
4.33414429e-01 -8.60858411e-02 -1.50870693e+00 5.53088605e-01
9.58220184e-01 1.37777936e+00 -3.35743308e-01 3.98622125e-01
-5.87005496e-01 1.10492873e+00 -7.18271911e-01 -8.71650517e-01
-1.69706225e-01 -3.82216692e-01 -8.56084883e-01 2.63103515e-01
-1.32176712e-01 5.50786452e-03 4.38689768e-01 9.96796370e-01
2.66508669e-01 -4.04296778e-02 6.83259904e-01 -1.16168714e+00
-7.99509525e-01 5.27293026e-01 -8.08648646e-01 -1.32506892e-01
6.11693621e-01 -1.75753981e-01 1.39401042e+00 -1.01562035e+00
2.35592946e-01 1.22690761e+00 2.96563447e-01 7.75773525e-02
-4.59084213e-01 -5.57213664e-01 8.47128570e-01 9.41369981e-02
-9.20064509e-01 -2.34807879e-01 7.03241706e-01 -2.63908982e-01
9.36430395e-01 3.97570103e-01 7.53481507e-01 6.54210687e-01
3.83072475e-04 1.18594670e+00 1.46088505e+00 -3.37204218e-01
1.19820602e-01 4.75756198e-01 7.33838081e-01 5.23619115e-01
2.17186004e-01 -5.98048091e-01 -5.88078976e-01 -7.49557465e-02
6.57077804e-02 -4.92355265e-02 -3.53353083e-01 9.27517489e-02
-1.06762826e+00 5.82263887e-01 7.64418840e-02 2.18440190e-01
-3.61423284e-01 -4.48125780e-01 3.97519141e-01 4.72617120e-01
6.07886910e-01 3.17678481e-01 -1.10371482e+00 -1.13199413e-01
-3.22792083e-01 -5.60041033e-02 1.00446856e+00 1.22045231e+00
1.01810420e+00 -5.13221920e-01 -4.22344595e-01 7.67650902e-01
4.45160270e-01 7.22830296e-01 4.76253748e-01 -2.24659905e-01
7.49153376e-01 1.36198735e+00 2.27068812e-02 -1.22313988e+00
-1.86354771e-01 -4.38692689e-01 -8.47448826e-01 -6.52600408e-01
-3.33669811e-01 -4.55370754e-01 -8.08845103e-01 1.29134035e+00
6.21758997e-01 -2.28895530e-01 3.56233925e-01 6.60399795e-01
1.10424900e+00 5.79340696e-01 -5.11529623e-03 -4.90499765e-01
1.85961199e+00 -1.14884043e+00 -7.65516877e-01 -2.90776640e-01
5.97167552e-01 -9.85871315e-01 1.05930066e+00 2.95519531e-01
-5.16478419e-01 -2.78478652e-01 -8.45154941e-01 -2.43643299e-01
-5.44867218e-01 3.92657131e-01 7.46553004e-01 3.72089744e-01
-5.87139010e-01 -2.23401412e-01 -4.55736697e-01 1.40311956e-01
1.33238599e-01 3.57363075e-01 -2.28735566e-01 -2.95014717e-02
-1.37878191e+00 4.89117533e-01 1.18080668e-01 4.30063426e-01
-4.28573415e-03 -3.53727907e-01 -9.56620336e-01 -2.37714667e-02
6.67050302e-01 -7.38395393e-01 1.21875989e+00 -1.04297066e+00
-1.34942961e+00 6.13553107e-01 -7.21037865e-01 1.72728792e-01
-9.08375904e-02 -3.25692505e-01 -7.19251931e-01 -9.82574373e-02
3.89093935e-01 1.02738343e-01 5.50250649e-01 -1.13871062e+00
-1.08871913e+00 -6.45688951e-01 5.98453403e-01 4.80801642e-01
-5.18751502e-01 3.44836473e-01 -8.92400980e-01 -3.82110655e-01
9.68801230e-02 -7.86821544e-01 -1.45826340e-01 -6.90781832e-01
-8.14120591e-01 -5.35421669e-01 5.99140108e-01 -3.75787646e-01
1.55026567e+00 -1.85581219e+00 7.64721856e-02 3.91164958e-01
2.28647724e-01 2.33413249e-01 -1.01093873e-02 1.22317575e-01
-1.14815030e-02 2.07007542e-01 -2.78224558e-01 -2.45088756e-01
1.46800041e-01 6.28388524e-02 -4.39208925e-01 -3.12312245e-01
4.08007115e-01 1.04242921e+00 -7.42651045e-01 -4.84006435e-01
-2.69985378e-01 1.78532422e-01 -2.43291527e-01 2.62928486e-01
-3.23056012e-01 2.75777191e-01 -1.17029357e+00 9.63950157e-01
7.02392459e-01 -4.42630738e-01 -1.29532618e-02 -4.93029177e-01
-1.27309024e-01 4.09359366e-01 -9.86273348e-01 9.19057727e-01
-6.74557984e-01 1.27891332e-01 -1.69698432e-01 -8.11429918e-01
1.09724224e+00 2.07621783e-01 1.64040968e-01 -7.14245498e-01
2.62591928e-01 2.52038300e-01 -3.55326861e-01 -7.60257244e-01
5.91916859e-01 -2.87095070e-01 -3.11467111e-01 5.75149834e-01
-2.31181934e-01 6.83911378e-04 5.47842205e-01 8.95697325e-02
8.27809393e-01 -1.20804012e-01 4.40958351e-01 -2.98362244e-02
1.23278224e+00 4.87179495e-02 7.14258850e-01 1.39479235e-01
1.27285555e-01 2.83655077e-01 1.03675544e+00 -2.82151252e-01
-1.51496679e-01 -5.00366807e-01 2.59416759e-01 1.07737386e+00
4.34803128e-01 -7.46007264e-01 -4.30360526e-01 -1.23117304e+00
-4.13802773e-01 4.66277003e-01 -6.99846148e-01 1.25026228e-02
-3.94732445e-01 -9.84697819e-01 1.12728685e-01 4.66407984e-01
7.10491180e-01 -1.06296742e+00 -1.38899729e-01 -8.79932344e-02
-5.13339460e-01 -1.23691392e+00 -6.19109511e-01 5.79368807e-02
-3.07326913e-01 -1.34712255e+00 -4.73268598e-01 -9.37898219e-01
8.81513238e-01 4.68440503e-01 1.40369105e+00 2.02681184e-01
4.72618133e-01 1.87101617e-01 -8.57993364e-01 -6.04567409e-01
1.81114867e-01 3.99823368e-01 -2.98241228e-01 3.47333342e-01
9.08831835e-01 -4.20654416e-01 -6.73271835e-01 3.45062435e-01
-1.07350814e+00 1.09134942e-01 1.09662426e+00 5.00097752e-01
9.22446370e-01 1.22484662e-01 5.41108549e-01 -1.43646383e+00
1.06412864e+00 -5.17984271e-01 -4.10880148e-01 6.77536845e-01
-7.05737650e-01 -3.25785689e-02 8.34161639e-01 -7.62723312e-02
-1.23835778e+00 -1.78476319e-01 -2.60425806e-01 1.85116634e-01
-5.61513826e-02 1.10127604e+00 -6.42104566e-01 4.37114060e-01
-8.51431414e-02 4.63874608e-01 -4.66019869e-01 -1.99362576e-01
3.80933702e-01 1.07122386e+00 1.73703879e-01 -5.43917656e-01
6.90039277e-01 2.24839807e-01 -2.62199700e-01 -5.12626767e-01
-1.66938043e+00 -7.50503063e-01 -4.34375793e-01 -1.29886456e-02
9.06347096e-01 -9.67870593e-01 -6.50950015e-01 5.31690061e-01
-1.09237695e+00 4.61818576e-01 2.98756324e-02 1.80671021e-01
1.95448268e-02 2.35449985e-01 -3.73759300e-01 -6.92246974e-01
-7.23827302e-01 -1.33378386e+00 1.41691875e+00 6.05225325e-01
-5.38583696e-02 -9.09772396e-01 1.74314324e-02 6.95441782e-01
1.87538713e-01 -2.44802721e-02 1.09556198e+00 -7.48066723e-01
-4.23159420e-01 -1.95504308e-01 -3.49315017e-01 4.98364240e-01
4.43005532e-01 1.47105917e-01 -7.01491117e-01 1.76347017e-01
1.53774738e-01 -3.16995591e-01 7.70661533e-01 -2.55802393e-01
9.73771453e-01 -4.24303412e-01 -3.22925635e-02 3.17921519e-01
1.03506649e+00 2.36629993e-01 4.27294493e-01 4.45848286e-01
8.19146097e-01 8.02838683e-01 1.00457883e+00 2.08952665e-01
1.02350700e+00 2.68644899e-01 -7.72117730e-03 -1.09896235e-01
3.65571469e-01 -1.64205879e-01 4.76365507e-01 1.53563941e+00
6.27906248e-03 -1.86953515e-01 -4.99365926e-01 3.93411607e-01
-1.79964125e+00 -5.60470998e-01 -3.34359080e-01 1.54954875e+00
9.85554039e-01 2.17298850e-01 -2.71618009e-01 1.50654137e-01
4.90995735e-01 3.24248701e-01 -6.98768556e-01 -4.02799398e-01
-2.48448998e-01 1.61246970e-01 -6.57943860e-02 1.69712514e-01
-1.11341929e+00 9.82050538e-01 4.86208200e+00 8.05829227e-01
-9.10353899e-01 -1.97646767e-01 8.04815173e-01 3.38553846e-01
-9.78307128e-01 1.89896554e-01 -9.66770530e-01 2.76819557e-01
2.66725361e-01 -1.42935619e-01 -3.32270712e-02 7.14578927e-01
-5.53321503e-02 -3.94311436e-02 -7.78850615e-01 5.99586010e-01
2.58264065e-01 -8.19785237e-01 5.21800458e-01 -3.24825197e-01
8.92874360e-01 -4.66298938e-01 -5.08843325e-02 5.89600384e-01
1.71410680e-01 -7.59762824e-01 4.59134400e-01 4.55701709e-01
5.26125968e-01 -9.14848864e-01 1.14518082e+00 2.69461691e-01
-1.77597558e+00 1.27061352e-01 -1.48109451e-01 2.10992992e-02
1.19154118e-01 9.92845893e-01 4.64063167e-04 1.02386463e+00
5.79162598e-01 1.05107546e+00 -7.36582577e-01 4.78582680e-01
-8.10664535e-01 4.58895385e-01 1.76677927e-02 -4.71099108e-01
2.28487566e-01 -4.76453304e-01 2.01702461e-01 1.08787632e+00
9.18781757e-02 3.46030593e-01 2.01749817e-01 3.63083124e-01
-2.09611237e-01 5.77773392e-01 -1.87152267e-01 -2.96789501e-02
3.49504143e-01 1.65337622e+00 -5.78447044e-01 -5.36331952e-01
-7.59744942e-01 7.27512836e-01 3.50657582e-01 4.67438430e-01
-5.89891434e-01 -7.41710961e-01 5.42584002e-01 -5.13615251e-01
4.84925032e-01 2.67287731e-01 -1.99414015e-01 -1.65587127e+00
6.79945290e-01 -1.29770792e+00 2.83172488e-01 -7.32690513e-01
-1.34287834e+00 9.21733379e-01 -2.78366625e-01 -1.24504220e+00
7.98795652e-03 -5.51294148e-01 -8.35371375e-01 7.88154840e-01
-2.00390458e+00 -1.38250339e+00 -4.09100682e-01 6.10728621e-01
4.74720836e-01 1.27103120e-01 6.53042376e-01 1.01640530e-01
-9.38125134e-01 5.23910046e-01 -4.75577444e-01 6.52402267e-02
5.19607961e-01 -1.28171039e+00 2.38672227e-01 8.67400289e-01
-6.34106174e-02 1.02761459e+00 2.83671439e-01 -4.52016175e-01
-1.41582716e+00 -1.18471885e+00 1.13912809e+00 -3.90150607e-01
7.75076091e-01 -1.15816593e-01 -7.79877603e-01 6.62012577e-01
1.16599523e-01 -4.30213958e-01 1.06056106e+00 4.20464337e-01
-4.93300855e-01 -2.48540327e-01 -4.73236561e-01 5.60697973e-01
7.08908439e-01 -5.61699212e-01 -7.79349804e-01 2.26231307e-01
1.09081268e+00 -2.18471259e-01 -8.05674672e-01 8.77987921e-01
5.45911551e-01 -1.07026744e+00 4.81475443e-01 -4.87686098e-01
8.01000595e-01 -6.31115258e-01 -1.21089861e-01 -1.19979131e+00
3.11853021e-01 -3.61880362e-01 -1.93477169e-01 1.76595366e+00
1.00032210e+00 -7.36451685e-01 5.40005744e-01 6.10837460e-01
1.87575445e-02 -1.51785958e+00 -3.93022597e-01 -1.12359561e-01
-1.87888727e-01 -3.47504258e-01 1.07478309e+00 7.14759111e-01
7.65667111e-02 1.17049074e+00 -9.74916443e-02 1.12303689e-01
-4.22073156e-03 9.22169030e-01 7.49044180e-01 -9.30289805e-01
-3.32058400e-01 -3.93579006e-01 1.05116084e-01 -1.53754151e+00
2.51822650e-01 -4.27136958e-01 5.38255684e-02 -1.78329706e+00
3.94385159e-01 -4.01773602e-01 -3.85326207e-01 5.12937367e-01
-9.35281754e-01 -1.13318913e-01 -1.19670965e-01 1.43338308e-01
-1.08436477e+00 8.72350514e-01 1.47250247e+00 -2.58222610e-01
-8.33011046e-02 3.54919255e-01 -1.63114810e+00 1.02410233e+00
5.79421639e-01 -1.71294793e-01 -6.86275363e-01 -4.62632090e-01
8.79519403e-01 -1.27447709e-01 -3.48630190e-01 -3.27848673e-01
2.54110903e-01 -1.86801642e-01 1.75164655e-01 -8.37690294e-01
9.46578104e-03 -6.62288427e-01 -5.47712088e-01 -9.46616195e-03
-2.65120119e-01 3.58248353e-01 -1.46452144e-01 4.37189460e-01
-7.70124555e-01 -3.29235190e-05 2.26238556e-02 -8.53533372e-02
-6.74543798e-01 3.72103781e-01 -2.22158208e-01 3.56616795e-01
7.69694865e-01 1.98642433e-01 -5.39262593e-01 -2.29872227e-01
-2.76892692e-01 7.59786069e-01 2.13443130e-01 5.88377595e-01
6.31395102e-01 -1.13213611e+00 -4.83909488e-01 1.70885697e-01
5.11633158e-01 4.81835604e-01 1.36892274e-01 9.51313972e-01
3.18925828e-02 3.62610877e-01 4.40346599e-01 -2.13495582e-01
-1.39856911e+00 3.19083840e-01 9.87077430e-02 -8.09115946e-01
1.81940526e-01 8.40606868e-01 3.33028883e-01 -8.03760946e-01
-6.29742518e-02 -3.60865444e-01 -1.04207361e+00 3.52796078e-01
5.56583643e-01 -9.84088406e-02 2.08386064e-01 -7.36292481e-01
-3.93087298e-01 1.09998834e+00 -4.42283720e-01 1.62091658e-01
9.78367090e-01 -3.65937680e-01 -8.02041650e-01 3.44278365e-01
1.16342521e+00 4.12966996e-01 -5.05093634e-01 -2.57645816e-01
6.94453046e-02 -5.87688051e-02 -2.18719408e-01 -8.11651409e-01
-1.12397218e+00 6.71324611e-01 -2.65597492e-01 3.20477515e-01
1.63391328e+00 6.70262352e-02 1.18012214e+00 5.00321865e-01
1.36755124e-01 -8.49341452e-01 -6.16260525e-03 8.48594069e-01
6.28950357e-01 -1.14073336e+00 -1.36680249e-02 -7.87621081e-01
-9.78898168e-01 8.76168966e-01 9.20082927e-01 3.16289186e-01
8.16954017e-01 7.16265365e-02 3.65464717e-01 -4.34831113e-01
-9.85200047e-01 -4.76819396e-01 5.55685639e-01 1.61277682e-01
5.72265685e-01 7.15346634e-02 -4.76742327e-01 1.18242323e+00
-4.23776269e-01 -1.49959788e-01 2.59006113e-01 1.03678989e+00
-4.03030455e-01 -1.23780274e+00 9.08780321e-02 6.62557840e-01
-6.28356457e-01 -4.46167171e-01 -8.24438393e-01 3.32103014e-01
2.85156608e-01 1.47279847e+00 -3.30617070e-01 -7.13895738e-01
7.85770833e-01 -2.59054273e-01 -1.69975281e-01 -7.02549219e-01
-7.66467750e-01 -2.05268022e-02 3.17099571e-01 -3.38753045e-01
-8.11958611e-01 -3.31211358e-01 -1.22151959e+00 4.28708233e-02
-6.66973412e-01 6.96473837e-01 3.88486862e-01 1.48095489e+00
5.01755178e-01 6.85701549e-01 1.02017403e+00 1.46688506e-01
-2.62152433e-01 -9.27454174e-01 -3.08965117e-01 3.58507961e-01
2.72677392e-01 -4.21926051e-01 -3.57619315e-01 -6.21674545e-02] | [11.530445098876953, 6.568080425262451] |
13434fa9-2f6a-45fc-b7ee-ed85f5f7e867 | value-memory-graph-a-graph-structured-world | 2206.04384 | null | https://arxiv.org/abs/2206.04384v3 | https://arxiv.org/pdf/2206.04384v3.pdf | Value Memory Graph: A Graph-Structured World Model for Offline Reinforcement Learning | Reinforcement Learning (RL) methods are typically applied directly in environments to learn policies. In some complex environments with continuous state-action spaces, sparse rewards, and/or long temporal horizons, learning a good policy in the original environments can be difficult. Focusing on the offline RL setting, we aim to build a simple and discrete world model that abstracts the original environment. RL methods are applied to our world model instead of the environment data for simplified policy learning. Our world model, dubbed Value Memory Graph (VMG), is designed as a directed-graph-based Markov decision process (MDP) of which vertices and directed edges represent graph states and graph actions, separately. As state-action spaces of VMG are finite and relatively small compared to the original environment, we can directly apply the value iteration algorithm on VMG to estimate graph state values and figure out the best graph actions. VMG is trained from and built on the offline RL dataset. Together with an action translator that converts the abstract graph actions in VMG to real actions in the original environment, VMG controls agents to maximize episode returns. Our experiments on the D4RL benchmark show that VMG can outperform state-of-the-art offline RL methods in several goal-oriented tasks, especially when environments have sparse rewards and long temporal horizons. Code is available at https://github.com/TsuTikgiau/ValueMemoryGraph | ['Mohamed Elhoseiny', 'Li Erran Li', 'Deyao Zhu'] | 2022-06-09 | null | null | null | null | ['d4rl'] | ['robots'] | [-1.89506054e-01 2.81682968e-01 -5.64775050e-01 2.17312679e-01
-4.18233246e-01 -6.96867347e-01 7.02136099e-01 3.35659161e-02
-5.49652100e-01 1.05246985e+00 1.21004157e-01 -5.00335157e-01
-6.84664398e-02 -9.52978313e-01 -9.02968824e-01 -6.89725518e-01
-5.79917073e-01 6.52786672e-01 8.98443013e-02 -1.71321705e-01
1.02170803e-01 2.41207570e-01 -1.10109651e+00 -2.86327928e-01
7.47313797e-01 7.26490080e-01 5.30325234e-01 8.34067106e-01
-9.93726254e-02 1.36763310e+00 -4.98718262e-01 1.62688196e-01
5.42526901e-01 -5.51029801e-01 -6.90332055e-01 1.28126606e-01
-4.43497449e-01 -6.40244484e-01 -9.26783085e-01 1.00137639e+00
5.22090614e-01 8.39088440e-01 1.16585791e-01 -1.37232161e+00
-5.78772306e-01 9.17212784e-01 -1.60480544e-01 2.91038342e-02
4.19722885e-01 9.85709906e-01 7.50677824e-01 -1.55686215e-01
7.37652659e-01 1.44937992e+00 -1.22482799e-01 7.33886063e-01
-1.23667431e+00 -4.95205522e-01 9.72872794e-01 2.25968167e-01
-9.33531761e-01 -1.26060426e-01 5.67577422e-01 -1.12027980e-01
1.53806174e+00 -1.39382869e-01 1.17913842e+00 1.44144821e+00
4.36067313e-01 1.02566850e+00 1.21663547e+00 -6.95977733e-02
8.17342103e-01 -5.02883732e-01 -4.00726914e-01 8.56511354e-01
-1.95119027e-02 6.62212908e-01 -3.95597428e-01 -5.08833230e-02
1.01086009e+00 1.56115711e-01 -8.14376473e-02 -6.96837306e-01
-1.28035760e+00 8.01111221e-01 4.75605607e-01 -2.66026169e-01
-5.88048637e-01 7.81071961e-01 2.93285728e-01 5.22266090e-01
4.40097302e-02 5.90844810e-01 -3.18435520e-01 -4.80815858e-01
-4.94552171e-03 4.03952897e-01 9.51410353e-01 1.14370263e+00
7.89820075e-01 4.02190149e-01 -4.60274369e-01 2.38550380e-01
1.47910655e-01 8.60767663e-01 4.53285575e-01 -1.26541722e+00
7.86211848e-01 4.70643103e-01 5.23432136e-01 -5.81212878e-01
-2.88047940e-01 -2.64300287e-01 -4.05104846e-01 3.10542732e-01
3.02349150e-01 -5.09438872e-01 -1.18412423e+00 1.82551694e+00
4.17630225e-01 5.50856173e-01 3.16482693e-01 7.38177717e-01
3.36054385e-01 7.34638512e-01 2.53034830e-02 -3.01961869e-01
5.96310735e-01 -1.16307867e+00 -6.38379037e-01 -8.40853691e-01
6.99651182e-01 1.46556959e-01 1.21875191e+00 3.33092511e-01
-1.07071280e+00 -1.14346206e-01 -7.95295894e-01 4.86244678e-01
-2.15306520e-01 -2.94074237e-01 6.50049567e-01 2.90373843e-02
-1.12464750e+00 7.70634711e-01 -1.23488128e+00 -3.00495774e-02
2.18783021e-01 4.26167011e-01 -2.10414673e-05 -1.63397372e-01
-1.31925619e+00 1.03672552e+00 7.31458962e-01 8.63297135e-02
-1.99849689e+00 -2.58482788e-02 -1.06116903e+00 -4.47485633e-02
1.29159403e+00 -4.97020602e-01 1.68861878e+00 -6.95307136e-01
-2.17935991e+00 6.75837174e-02 3.11142594e-01 -7.21120298e-01
6.12877011e-01 -5.12723066e-02 -1.60557508e-01 -9.67136305e-03
-1.75134823e-01 3.19563866e-01 1.04131305e+00 -1.01594114e+00
-5.98585546e-01 -1.43910974e-01 4.67130572e-01 6.43299162e-01
1.95909575e-01 -5.32808900e-01 -4.66033936e-01 -2.24451467e-01
-3.51333469e-01 -1.30689085e+00 -7.88967252e-01 -6.58804774e-01
-2.71803349e-01 -6.16184771e-02 4.62766409e-01 -4.94315952e-01
1.24049199e+00 -1.79594135e+00 4.98229772e-01 2.62882143e-01
2.20753580e-01 8.84189904e-02 -4.55609858e-01 7.77683854e-01
3.09155315e-01 -2.20024139e-01 -3.31711359e-02 -1.92785516e-01
2.65176624e-01 6.31570995e-01 -4.90741700e-01 4.83566344e-01
-2.38826096e-01 1.21924043e+00 -1.65634263e+00 -1.72522888e-01
3.93392742e-01 -4.96051982e-02 -4.04207736e-01 4.27108198e-01
-1.01082671e+00 7.38888562e-01 -9.61059570e-01 3.80764991e-01
2.19010845e-01 -1.94412544e-01 7.28291035e-01 7.84981906e-01
4.17829379e-02 3.84983659e-01 -1.25691545e+00 1.77266157e+00
-6.29752696e-01 2.28482082e-01 -1.42870694e-01 -7.77453303e-01
8.00907433e-01 8.13222229e-02 4.20222044e-01 -1.07661450e+00
1.60611749e-01 -1.24593444e-01 4.96079922e-02 -3.44695300e-01
3.71556729e-01 2.30530739e-01 -3.24789792e-01 6.04526997e-01
-1.19496547e-02 -4.05632645e-01 3.45918953e-01 3.68287355e-01
1.58506238e+00 4.86810684e-01 4.01723683e-01 2.71339446e-01
7.34442994e-02 6.69288123e-03 6.70464694e-01 1.13768041e+00
-4.23429430e-01 -2.77342975e-01 9.17392492e-01 -3.34946841e-01
-6.73263609e-01 -1.18713534e+00 8.66567194e-01 1.15030622e+00
4.04497206e-01 -5.90872884e-01 -4.05064344e-01 -8.34281266e-01
9.97242630e-02 9.78983939e-01 -5.94497383e-01 -4.88451153e-01
-7.19830930e-01 -2.21094742e-01 1.00993007e-01 4.19011831e-01
5.56379139e-01 -1.66210628e+00 -8.96825552e-01 5.98957181e-01
1.38292879e-01 -9.63692069e-01 -6.10150456e-01 2.10408926e-01
-8.52145016e-01 -9.94107068e-01 -2.92652249e-01 -3.89673650e-01
5.87392867e-01 1.99716259e-02 1.16561043e+00 -1.25200018e-01
2.12577373e-01 8.35348308e-01 -4.15146559e-01 -3.20922166e-01
-5.50465524e-01 -6.54407218e-02 9.24005732e-02 -2.43212625e-01
-2.08853185e-01 -4.99077946e-01 -6.53793335e-01 1.18782252e-01
-6.49995923e-01 2.73021638e-01 5.58786750e-01 7.52398193e-01
6.92253351e-01 -4.92641293e-02 4.21379775e-01 -8.13306808e-01
9.05821264e-01 -5.30441701e-01 -1.24784064e+00 2.24293828e-01
-7.25764453e-01 4.62869674e-01 9.00000215e-01 -8.31813872e-01
-9.78583634e-01 1.02956332e-01 4.09447372e-01 -7.21308291e-01
2.12886527e-01 5.48022330e-01 -8.59790444e-02 3.15808237e-01
5.32498062e-01 4.25271213e-01 4.53689322e-02 -7.97065869e-02
6.82890594e-01 6.08481243e-02 2.21227482e-01 -8.67344737e-01
6.45011127e-01 4.45584282e-02 1.06832096e-02 -2.46547326e-01
-5.23003519e-01 3.23569477e-02 -1.10117473e-01 -4.22933817e-01
5.61150014e-01 -7.76675463e-01 -1.01123202e+00 3.83881837e-01
-5.92831552e-01 -1.75800598e+00 -5.18347204e-01 5.71731865e-01
-1.19753480e+00 -1.21626548e-01 -6.40966058e-01 -1.00393367e+00
5.09760641e-02 -1.31791306e+00 6.80993974e-01 2.19448805e-01
3.68583083e-01 -1.14429629e+00 4.25994039e-01 -2.31369749e-01
2.31083989e-01 2.75475174e-01 6.86903715e-01 -3.08809042e-01
-9.64692831e-01 1.66874930e-01 3.70331496e-01 3.99544798e-02
4.01362665e-02 -3.72105569e-01 -5.02080858e-01 -7.15114355e-01
-3.13812047e-01 -5.52398562e-01 7.25416243e-01 4.40725446e-01
1.22491896e+00 -6.88649118e-01 -3.62332553e-01 4.68330443e-01
1.36021507e+00 6.06643915e-01 5.12898028e-01 5.08363724e-01
6.06803000e-01 1.01351574e-01 1.03515601e+00 8.46250951e-01
5.33801615e-01 4.40762997e-01 8.70609641e-01 3.04651469e-01
2.96762362e-02 -8.76133800e-01 1.06808805e+00 4.61848289e-01
-1.44308982e-02 -4.20605004e-01 -8.69938612e-01 3.56589943e-01
-2.42580724e+00 -9.60261703e-01 4.45290297e-01 2.40797186e+00
7.68458188e-01 3.24071407e-01 1.74867749e-01 -6.03374600e-01
5.32711148e-01 4.28143919e-01 -1.35940814e+00 -4.00757104e-01
2.93470562e-01 -1.16960593e-02 8.13950837e-01 7.44546294e-01
-8.05411637e-01 1.42272162e+00 5.58449364e+00 6.39856577e-01
-1.04613066e+00 -1.35252044e-01 3.68883640e-01 -4.25273448e-01
-8.96366239e-02 2.50428289e-01 -5.83949864e-01 4.86249715e-01
1.12400842e+00 -3.63334209e-01 1.47410738e+00 9.18208778e-01
3.92265648e-01 -1.60074383e-01 -1.07719791e+00 8.60847414e-01
-5.11348307e-01 -1.13060951e+00 -3.94585937e-01 9.76399407e-02
8.03599417e-01 3.88663769e-01 1.49331346e-01 1.09917283e+00
1.40865242e+00 -1.01022112e+00 6.82845354e-01 5.08941352e-01
6.08033955e-01 -8.71185541e-01 1.45383134e-01 7.05505669e-01
-1.24143660e+00 -5.47126293e-01 -3.23159158e-01 -2.96167135e-01
4.04027142e-02 -1.17303900e-01 -1.13855839e+00 3.95958632e-01
3.82680595e-01 9.64393973e-01 -2.46533751e-01 7.21593499e-01
-9.24929082e-01 5.98707557e-01 -5.65106161e-02 -3.01816970e-01
5.88284850e-01 -5.70976198e-01 6.76052988e-01 5.26697099e-01
1.43178701e-01 1.12412825e-01 8.82510662e-01 7.84372151e-01
-8.85000266e-03 -2.99716622e-01 -1.03222239e+00 -5.43541551e-01
5.11998713e-01 1.08720517e+00 -6.31201565e-01 -4.38509554e-01
-1.96171120e-01 8.58264148e-01 7.55882084e-01 8.15898418e-01
-1.02833045e+00 -6.59734756e-02 7.37219870e-01 -2.69466847e-01
2.42187709e-01 -6.11634374e-01 3.60242605e-01 -1.20645308e+00
-3.36185366e-01 -1.20661879e+00 4.91871178e-01 -6.66247189e-01
-7.67404079e-01 4.06510621e-01 1.10604204e-01 -1.23780513e+00
-7.31797934e-01 -2.62966484e-01 -5.32024264e-01 5.60287178e-01
-1.33805776e+00 -7.14430094e-01 -5.57836983e-03 8.65702033e-01
6.20242178e-01 -2.23218143e-01 6.57589614e-01 -5.66169143e-01
-8.27893853e-01 1.29487678e-01 3.09900284e-01 6.97246566e-03
3.57644856e-01 -1.41871858e+00 6.00722849e-01 8.32931936e-01
-2.88900901e-02 3.48003298e-01 6.74329996e-01 -1.09025335e+00
-2.08863783e+00 -1.28257930e+00 -1.27113611e-01 -3.86521965e-01
9.65308666e-01 -4.44256693e-01 -5.94784439e-01 1.16057730e+00
2.05040425e-01 1.25287294e-01 -1.29482925e-01 -1.79625884e-01
1.75845519e-01 1.79459199e-01 -8.48335028e-01 1.06672573e+00
1.42764068e+00 -2.77841598e-01 -3.30072612e-01 4.94756490e-01
9.65147257e-01 -9.24168766e-01 -6.22081697e-01 -1.00211427e-01
6.35396242e-02 -4.43783581e-01 8.46627891e-01 -9.49143708e-01
2.75438558e-02 -2.52760440e-01 4.63773943e-02 -1.98489690e+00
-1.25410572e-01 -1.04296589e+00 -7.66578197e-01 5.33672810e-01
2.41305754e-01 -1.09725130e+00 5.47178507e-01 6.69735253e-01
-7.79319778e-02 -8.70391250e-01 -6.61261380e-01 -1.07290709e+00
-1.80606365e-01 -3.91438723e-01 8.52303207e-01 4.63811427e-01
1.19680576e-02 2.32656389e-01 -5.34255564e-01 1.19325727e-01
6.17830217e-01 4.17857081e-01 9.52668726e-01 -5.60085416e-01
-7.79683590e-01 -1.79874048e-01 2.13262454e-01 -1.07530367e+00
7.00642109e-01 -8.27211976e-01 3.45551193e-01 -1.85162795e+00
-7.45887235e-02 -5.10928988e-01 -3.81183773e-01 7.90558398e-01
-7.52481073e-02 -8.12422037e-01 6.16115272e-01 3.40944938e-02
-1.08668387e+00 1.03320074e+00 1.85147893e+00 -3.18623722e-01
-8.37398410e-01 1.14678271e-01 -2.89103657e-01 5.38191676e-01
1.15543735e+00 -4.26173598e-01 -1.01282227e+00 -4.52197701e-01
3.17903757e-01 7.11790204e-01 5.88975325e-02 -6.87180519e-01
1.74665406e-01 -1.10046768e+00 6.95503205e-02 -1.08929299e-01
3.98011744e-01 -5.07651329e-01 1.61857188e-01 7.83905864e-01
-5.31365693e-01 1.61961377e-01 2.02829875e-02 1.03980780e+00
2.35649765e-01 5.17571829e-02 3.80030870e-01 -5.12464762e-01
-1.13472366e+00 8.08233738e-01 -4.51534599e-01 2.87225455e-01
1.24884641e+00 1.43511072e-01 -4.76703465e-01 -8.10791612e-01
-9.71669614e-01 8.44373167e-01 5.78172147e-01 3.72482359e-01
7.78719485e-01 -1.24199581e+00 -2.30389610e-01 -1.30797759e-01
-1.10563383e-01 1.69235304e-01 1.46358877e-01 5.28720617e-01
-1.52143613e-01 1.60578802e-01 -2.88119525e-01 -2.89245341e-02
-6.09653950e-01 9.63230729e-01 5.43125212e-01 -7.99758077e-01
-9.01206613e-01 3.54488939e-01 9.93015692e-02 -4.37259853e-01
2.83452183e-01 -6.71525121e-01 -5.66242300e-02 -3.00047457e-01
3.65120232e-01 4.61456925e-01 -3.93267035e-01 1.96384043e-02
-9.59767178e-02 -1.61070421e-01 2.49760911e-01 -4.95609879e-01
1.41175771e+00 -3.76663134e-02 2.66772628e-01 4.17778909e-01
6.14835978e-01 -3.86294484e-01 -2.08162332e+00 -3.24220985e-01
-1.13366149e-01 -6.59976065e-01 1.12848267e-01 -7.89925754e-01
-9.62787807e-01 5.09862185e-01 3.89075726e-01 2.22423419e-01
8.52594554e-01 -1.50885865e-01 6.19425058e-01 7.87581861e-01
9.55613554e-01 -1.50553930e+00 4.58763957e-01 8.42963040e-01
9.78292346e-01 -1.07537937e+00 -1.96480036e-01 3.90266627e-01
-1.17204547e+00 7.11561561e-01 8.33915532e-01 -3.31162065e-01
2.42489725e-01 1.70241088e-01 -3.23162943e-01 6.55613095e-02
-1.23719573e+00 -6.21719122e-01 -2.75383145e-01 8.45161676e-01
-3.69668901e-01 5.30227900e-01 2.01138377e-01 3.41233790e-01
-1.18872784e-02 6.01377897e-02 6.80587649e-01 1.21730661e+00
-5.11308253e-01 -1.25237489e+00 -1.95518002e-01 4.92980689e-01
1.98457941e-01 1.67247251e-01 -2.40610853e-01 7.22475171e-01
-5.46002209e-01 9.50351894e-01 -9.94530022e-02 -3.85524511e-01
2.68735617e-01 -2.18471482e-01 6.65066004e-01 -7.31943488e-01
-4.11330700e-01 -1.12784341e-01 2.22989261e-01 -1.21900868e+00
1.87393576e-01 -4.51404005e-01 -1.80286825e+00 -3.10947120e-01
1.46927938e-01 2.21135002e-02 3.88201177e-01 7.74222851e-01
4.69631791e-01 7.51083910e-01 7.26015091e-01 -9.19718981e-01
-1.02741992e+00 -6.81182683e-01 -6.51802242e-01 1.81612909e-01
4.87664908e-01 -9.16391432e-01 -3.77288237e-02 -5.13668418e-01] | [4.099844932556152, 1.6367292404174805] |
f9be4830-774a-48e1-bcc9-0576061ca6fc | modelling-observation-correlations-for-active | 1401.4612 | null | http://arxiv.org/abs/1401.4612v1 | http://arxiv.org/pdf/1401.4612v1.pdf | Modelling Observation Correlations for Active Exploration and Robust Object Detection | Today, mobile robots are expected to carry out increasingly complex tasks in
multifarious, real-world environments. Often, the tasks require a certain
semantic understanding of the workspace. Consider, for example, spoken
instructions from a human collaborator referring to objects of interest; the
robot must be able to accurately detect these objects to correctly understand
the instructions. However, existing object detection, while competent, is not
perfect. In particular, the performance of detection algorithms is commonly
sensitive to the position of the sensor relative to the objects in the scene.
This paper presents an online planning algorithm which learns an explicit model
of the spatial dependence of object detection and generates plans which
maximize the expected performance of the detection, and by extension the
overall plan performance. Crucially, the learned sensor model incorporates
spatial correlations between measurements, capturing the fact that successive
measurements taken at the same or nearby locations are not independent. We show
how this sensor model can be incorporated into an efficient forward search
algorithm in the information space of detected objects, allowing the robot to
generate motion plans efficiently. We investigate the performance of our
approach by addressing the tasks of door and text detection in indoor
environments and demonstrate significant improvement in detection performance
during task execution over alternative methods in simulated and real robot
experiments. | ['Ingmar Posner', 'Albert S. Huang', 'Nicholas Roy', 'Javier Velez', 'Garrett Hemann'] | 2014-01-18 | null | null | null | null | ['robust-object-detection'] | ['computer-vision'] | [ 5.71698904e-01 1.23431668e-01 2.12808758e-01 -4.09276694e-01
-6.16401017e-01 -5.22057950e-01 4.63395685e-01 4.51490045e-01
-5.55408418e-01 4.40352678e-01 4.54639718e-02 7.77590042e-03
-5.17265439e-01 -5.83503604e-01 -8.00853550e-01 -5.18889904e-01
-2.60810941e-01 8.91690433e-01 3.86546373e-01 -2.20100313e-01
5.42311966e-01 7.02046096e-01 -1.38599980e+00 -9.54448879e-02
4.48746532e-01 7.24709153e-01 1.37683380e+00 1.08777773e+00
2.43153021e-01 6.31114721e-01 -5.38943112e-01 5.89015484e-01
2.22176060e-01 -1.25101164e-01 -8.23302984e-01 5.63971043e-01
-2.84482509e-01 -2.51447409e-01 -8.09590518e-02 8.61579835e-01
9.68254134e-02 4.04120773e-01 6.05893493e-01 -1.13088405e+00
1.95143968e-01 2.76800245e-01 -1.49641916e-01 -2.49224249e-02
7.68376887e-01 6.07741103e-02 6.83504045e-01 -7.51941860e-01
5.74828267e-01 1.11258829e+00 2.16278479e-01 4.73884381e-02
-8.73630166e-01 1.81409493e-02 2.01560661e-01 2.77096987e-01
-1.26661801e+00 -5.45456827e-01 5.07772267e-01 -3.98070127e-01
9.30195272e-01 1.47454599e-02 2.49701425e-01 6.82016015e-01
4.29140806e-01 5.64544201e-01 6.05615556e-01 -6.07988417e-01
4.57855403e-01 1.29882768e-01 4.73928191e-02 7.37543643e-01
1.61447302e-01 5.38116284e-02 -6.41360104e-01 4.01417986e-02
6.43657744e-01 1.99146569e-01 -3.31200480e-01 -9.05451596e-01
-1.43343413e+00 5.37041366e-01 4.12899166e-01 4.36649233e-01
-7.77403116e-01 8.20392966e-02 6.35766890e-03 1.51390612e-01
-2.23150820e-01 5.18459320e-01 -3.08813095e-01 -2.77404994e-01
-2.35188827e-01 1.77441552e-01 9.28290904e-01 1.34905922e+00
6.82449162e-01 -5.15350103e-01 1.28575951e-01 3.96333188e-01
2.36610621e-01 4.19254720e-01 1.55696124e-01 -1.14778054e+00
8.47187102e-01 6.02840781e-01 7.15153277e-01 -1.12086463e+00
-7.96730757e-01 -1.18535668e-01 -3.32428694e-01 1.79994881e-01
3.90237004e-01 -6.98692054e-02 -5.64536631e-01 1.46126795e+00
4.51798618e-01 -2.07178816e-01 2.61423379e-01 9.60101068e-01
-1.68642953e-01 5.28456151e-01 -2.45254278e-01 -1.79159254e-01
1.18631876e+00 -7.88619637e-01 -7.41840661e-01 -8.14084232e-01
8.21340799e-01 -7.08433926e-01 6.55054569e-01 4.23291951e-01
-7.42782831e-01 -6.82471454e-01 -1.13082576e+00 1.27902180e-01
-1.69812649e-01 2.63754606e-01 3.46732378e-01 1.71476856e-01
-7.94416845e-01 4.71831053e-01 -9.87629473e-01 -6.98051989e-01
-2.08867669e-01 4.53225017e-01 -4.32301462e-01 -4.41965759e-01
-4.89424050e-01 1.12403965e+00 4.98539865e-01 3.39539438e-01
-9.43753362e-01 1.58560634e-01 -9.48243737e-01 3.04564275e-02
8.82823467e-01 -4.25433695e-01 1.55935562e+00 -3.92765582e-01
-1.26051247e+00 2.73217499e-01 -3.80941778e-01 -3.60677093e-01
2.55136818e-01 -4.30702269e-01 7.68359974e-02 5.11022471e-02
3.92086744e-01 4.46479201e-01 6.02814198e-01 -1.27363133e+00
-8.76667082e-01 -7.29497492e-01 1.14542119e-01 8.27495992e-01
2.85718977e-01 -3.80146384e-01 -3.21433306e-01 2.39721954e-01
6.16539896e-01 -1.11841738e+00 -5.02768755e-01 -9.58092958e-02
-2.18141824e-01 -6.83109015e-02 7.44863808e-01 -2.75011778e-01
4.64096427e-01 -2.18671227e+00 2.65824616e-01 2.85970479e-01
-2.80593812e-01 -4.05063391e-01 -1.20959528e-01 6.91248655e-01
4.37731266e-01 -7.43178010e-01 -1.54087409e-01 -4.17953849e-01
-1.37541443e-02 5.25538504e-01 -3.44628721e-01 6.02945566e-01
-3.05930227e-01 5.50420225e-01 -1.08482957e+00 -1.53158516e-01
4.54888374e-01 1.19000331e-01 -1.71643093e-01 2.53524095e-01
-1.61706358e-01 6.25706911e-01 -6.46170139e-01 1.87385038e-01
3.28958966e-02 -1.10551365e-01 4.17079002e-01 4.19938117e-01
-2.09304094e-01 3.32186759e-01 -1.45870006e+00 1.93067014e+00
-6.02282047e-01 5.16845882e-01 4.28476483e-01 -9.16183829e-01
7.79007852e-01 2.43867472e-01 4.07471091e-01 -6.58417702e-01
-1.29763400e-02 2.57459342e-01 -4.25755382e-02 -5.67706287e-01
8.34114432e-01 2.60947555e-01 -2.31668711e-01 3.08669865e-01
-3.64876568e-01 -4.48685884e-01 4.77976212e-03 1.59142643e-01
1.33276534e+00 1.05884567e-01 5.10841429e-01 1.00154318e-01
1.77148387e-01 4.75418776e-01 1.00755751e-01 1.05618703e+00
-1.12259209e-01 3.83170068e-01 -7.71124065e-02 -2.03527912e-01
-7.44319081e-01 -7.69156933e-01 3.86744916e-01 9.84404743e-01
7.84653723e-01 -6.65309355e-02 -5.91042817e-01 -2.97959000e-01
-3.34871233e-01 9.41922486e-01 -3.20402324e-01 -7.08693042e-02
-6.81467116e-01 -2.67285377e-01 -1.97314888e-01 5.12655556e-01
2.77164310e-01 -1.14541471e+00 -1.67343426e+00 4.29147184e-01
-3.06071371e-01 -1.36285543e+00 -4.46887285e-01 4.73978579e-01
-7.80731022e-01 -1.30648994e+00 -1.30903497e-01 -7.92805374e-01
9.21971321e-01 8.21354389e-01 5.80432773e-01 -1.12983696e-01
-3.45788121e-01 9.75111783e-01 -4.33006167e-01 -4.53426868e-01
-4.77286398e-01 -1.56604394e-01 1.20663591e-01 -1.36636525e-01
-3.23850708e-03 -1.68753818e-01 -3.26776743e-01 3.41998011e-01
-4.16047990e-01 5.18014356e-02 6.99473202e-01 5.53829849e-01
2.96192825e-01 3.70381564e-01 3.44861473e-04 -3.51159841e-01
5.91045976e-01 -2.77482599e-01 -5.71941316e-01 2.08336949e-01
-2.88196146e-01 2.73553967e-01 2.45639428e-01 -2.55827636e-01
-1.10562027e+00 6.26761079e-01 4.60846692e-01 8.54756013e-02
-3.89337629e-01 5.01003385e-01 -4.91797887e-02 1.49656147e-01
5.41806877e-01 3.08573157e-01 -1.14497155e-01 -1.74366474e-01
2.38622859e-01 5.55141270e-01 6.29760385e-01 -4.54797268e-01
3.29895258e-01 5.61473548e-01 2.77216583e-01 -8.67168188e-01
-3.26383710e-01 -9.87685621e-01 -8.95833850e-01 -3.45769197e-01
5.87094367e-01 -6.03832126e-01 -7.52662897e-01 2.19824001e-01
-1.48646975e+00 -5.12386799e-01 -7.79129714e-02 6.12416089e-01
-9.92652118e-01 1.69638306e-01 -8.76709148e-02 -1.21464360e+00
3.31202745e-01 -1.13757741e+00 1.36124778e+00 -1.14807358e-03
-3.49930674e-01 -6.70939207e-01 -1.35076463e-01 9.82017741e-02
8.08220282e-02 1.46890646e-02 6.05232835e-01 -5.25307894e-01
-8.00725579e-01 -4.05138522e-01 4.49405573e-02 -4.46025580e-01
3.84623587e-01 -7.02904880e-01 -5.79952896e-01 -1.63129911e-01
3.55006307e-01 8.51861313e-02 3.10020238e-01 3.33292633e-01
3.92747730e-01 -1.26369879e-01 -8.97397339e-01 -2.54771709e-01
1.20796227e+00 6.26067638e-01 1.49425641e-01 3.58388007e-01
3.66599858e-01 1.02874792e+00 1.20724988e+00 5.70886850e-01
4.67081815e-01 1.02231812e+00 7.15841413e-01 4.33701396e-01
3.34085673e-01 -9.79094729e-02 2.82331496e-01 3.61470729e-01
2.42873415e-01 -3.73769522e-01 -9.74932373e-01 6.62260354e-01
-2.20386219e+00 -5.67397773e-01 -1.26798272e-01 2.26758051e+00
2.29284186e-02 8.33027139e-02 -3.32985222e-01 2.78387666e-01
5.29423594e-01 -2.69955486e-01 -5.55499732e-01 -1.32646605e-01
4.63160396e-01 -4.21270460e-01 6.76177323e-01 8.28863382e-01
-8.35963845e-01 5.92977643e-01 5.62983990e+00 5.62446006e-02
-5.54612994e-01 1.46477953e-01 -7.19780698e-02 7.13280514e-02
4.32438254e-01 -5.72865568e-02 -6.98845208e-01 1.04247211e-02
7.71463156e-01 1.81642398e-01 3.95703435e-01 1.01883984e+00
2.75613695e-01 -9.04681206e-01 -1.52116680e+00 7.23692119e-01
-1.03111090e-02 -6.79787576e-01 -6.71485186e-01 1.38526289e-02
2.84749478e-01 -1.15533039e-01 -7.76712522e-02 1.30507171e-01
4.36542898e-01 -7.35172629e-01 8.22675526e-01 5.30466735e-01
-2.26250459e-02 -6.13491178e-01 5.22821009e-01 1.24588907e+00
-1.10135829e+00 -5.78587174e-01 -2.26643234e-01 -4.90009338e-01
3.99104863e-01 1.79045901e-01 -1.66782022e+00 5.16379714e-01
4.54662949e-01 1.76065713e-01 -9.74145308e-02 9.40357625e-01
-3.18406314e-01 -2.82581508e-01 -4.56605911e-01 -4.21634644e-01
3.04064214e-01 -6.38868585e-02 6.99950218e-01 7.59948492e-01
4.90071684e-01 3.47050667e-01 4.69760478e-01 5.44121206e-01
5.84974766e-01 -2.58974314e-01 -8.97113442e-01 3.94458801e-01
6.08129323e-01 7.90933251e-01 -9.26278830e-01 -1.32909909e-01
-1.78521633e-01 1.16564894e+00 3.65538627e-01 3.49900812e-01
-4.21223372e-01 -2.10196897e-01 3.70112807e-01 7.83563033e-02
2.42144912e-01 -9.27827179e-01 -1.09852865e-01 -3.31074983e-01
4.71867383e-01 -5.17704487e-01 4.18833755e-02 -1.08326948e+00
-3.59081328e-01 2.45504543e-01 1.13956816e-01 -1.06914496e+00
-5.97840726e-01 -7.34546661e-01 -3.62873822e-02 5.63973308e-01
-1.13536549e+00 -5.57067037e-01 -4.64785486e-01 3.01091701e-01
1.06167614e+00 2.22111553e-01 8.43629181e-01 -3.85195941e-01
-1.18808709e-02 -4.19998527e-01 4.81623262e-02 -2.23735064e-01
3.99451971e-01 -9.71042991e-01 2.12365761e-01 7.86057770e-01
3.09993267e-01 5.82258701e-01 8.59310925e-01 -8.49578679e-01
-1.71642089e+00 -7.01645911e-01 7.92035103e-01 -6.11213565e-01
1.90055326e-01 -4.19316977e-01 -5.96505761e-01 8.39518189e-01
-1.90950081e-01 -2.58391708e-01 -9.38680395e-02 -8.46956447e-02
2.65334040e-01 1.58541754e-01 -9.81034935e-01 5.56944847e-01
1.10558879e+00 -2.15091452e-01 -7.11615264e-01 5.88605106e-01
4.81855452e-01 -8.37572217e-01 -1.71656355e-01 2.72253394e-01
3.51262659e-01 -8.99112284e-01 7.63394296e-01 -1.68594047e-01
-1.88350543e-01 -2.86256671e-01 -2.89376497e-01 -1.19532251e+00
-1.73972532e-01 -3.68032485e-01 6.72356114e-02 4.63608176e-01
2.51299769e-01 -3.22848827e-01 6.28856063e-01 8.68640721e-01
-1.28470510e-01 -2.89770603e-01 -1.04693460e+00 -7.60915041e-01
-9.01698530e-01 -6.76391006e-01 2.31805757e-01 3.13810110e-01
3.57928336e-01 4.19315904e-01 -1.64117217e-01 8.26579094e-01
4.10526574e-01 1.62401304e-01 8.88417542e-01 -1.08732688e+00
-1.94507837e-01 -4.38516214e-02 -3.89738262e-01 -1.54885972e+00
6.34431839e-04 -2.71425545e-01 1.06815183e+00 -1.77416325e+00
2.25156192e-02 -5.55070937e-01 2.67361075e-01 8.82589817e-02
1.01505920e-01 -6.72124624e-01 3.83110642e-01 2.81570256e-01
-8.13016713e-01 2.75473624e-01 9.71118212e-01 6.97686225e-02
-4.68266487e-01 4.10692394e-01 -6.58634901e-02 7.30667233e-01
6.50783777e-01 -4.86566216e-01 -5.75432599e-01 -7.00460911e-01
2.23722219e-01 5.46144962e-01 3.62685502e-01 -1.35376799e+00
1.01815391e+00 -2.26675749e-01 2.65204310e-01 -5.39845347e-01
8.90443265e-01 -1.37098277e+00 1.04785383e-01 7.46747553e-01
-3.75705689e-01 2.06028670e-01 -1.47470413e-02 1.00728536e+00
1.11684225e-01 -5.69575608e-01 2.25501344e-01 -4.23779100e-01
-1.12050569e+00 -2.11472645e-01 -6.39402926e-01 -5.72352171e-01
1.18919063e+00 -3.74590218e-01 6.92297369e-02 -5.66910326e-01
-6.94842339e-01 3.53049636e-01 2.86581218e-01 5.55173934e-01
8.03923428e-01 -6.94001377e-01 -2.42376253e-01 2.13286236e-01
3.73325571e-02 4.39684063e-01 -1.02823325e-01 9.34072256e-01
-1.88023135e-01 6.95973754e-01 9.89495739e-02 -8.51532102e-01
-1.20948768e+00 6.99002862e-01 5.18836491e-02 -1.31826848e-01
-5.40721536e-01 6.24289155e-01 2.12982118e-01 -4.42922741e-01
4.76231366e-01 -4.66981918e-01 -8.26058462e-02 -4.40499455e-01
3.70138943e-01 5.26794970e-01 1.69192683e-02 -4.46808934e-01
-3.37714106e-01 5.45138299e-01 1.68061137e-01 -4.91197556e-01
1.03462601e+00 -5.78838766e-01 2.01665133e-01 6.89738989e-01
6.85316801e-01 -1.56715155e-01 -1.40128100e+00 -3.43279362e-01
3.78581882e-01 -3.32938075e-01 -3.12358677e-01 -6.21947885e-01
-2.15574279e-01 6.09569013e-01 4.25374478e-01 9.73171219e-02
8.83890390e-01 9.27037597e-02 2.77801007e-01 1.07333040e+00
1.20987976e+00 -1.10265636e+00 3.91954005e-01 7.20090091e-01
8.11844766e-01 -1.23180711e+00 -1.16736954e-02 -6.15370274e-01
-6.22111261e-01 1.13570249e+00 3.99603873e-01 1.20920181e-01
1.26241088e-01 4.03605998e-01 -7.07645044e-02 -1.83436587e-01
-5.52246273e-01 -1.74156398e-01 -1.47170365e-01 5.93581676e-01
-2.39094168e-01 1.07262030e-01 1.98697716e-01 2.36678179e-02
-6.79524019e-02 -2.75937676e-01 5.15489757e-01 1.44926000e+00
-1.02157247e+00 -6.23770177e-01 -7.38494039e-01 -1.11002941e-02
2.21578062e-01 3.76648962e-01 -3.28371346e-01 7.36180365e-01
-1.43202245e-01 1.28004229e+00 1.10195950e-01 -9.30420309e-02
5.97779930e-01 2.30252948e-02 5.75826406e-01 -9.23882425e-01
9.94613096e-02 -1.35083556e-01 4.98162508e-02 -8.10135484e-01
-4.07321185e-01 -9.46934938e-01 -1.70685005e+00 4.75014746e-01
-2.17099518e-01 1.36284605e-01 1.12931097e+00 1.30242074e+00
3.90029579e-01 4.62746918e-01 4.96556342e-01 -1.17314279e+00
-6.87514305e-01 -9.11569118e-01 -3.51598650e-01 1.01474047e-01
4.79116350e-01 -7.02686608e-01 -6.60974765e-03 -3.65477875e-02] | [4.77951192855835, 0.7773662209510803] |
d883050c-59af-4999-9189-92d94aef0a3d | form-nlu-dataset-for-the-form-language | 2304.01577 | null | https://arxiv.org/abs/2304.01577v2 | https://arxiv.org/pdf/2304.01577v2.pdf | Form-NLU: Dataset for the Form Language Understanding | Compared to general document analysis tasks, form document structure understanding and retrieval are challenging. Form documents are typically made by two types of authors; A form designer, who develops the form structure and keys, and a form user, who fills out form values based on the provided keys. Hence, the form values may not be aligned with the form designer's intention (structure and keys) if a form user gets confused. In this paper, we introduce Form-NLU, the first novel dataset for form structure understanding and its key and value information extraction, interpreting the form designer's intent and the alignment of user-written value on it. It consists of 857 form images, 6k form keys and values, and 4k table keys and values. Our dataset also includes three form types: digital, printed, and handwritten, which cover diverse form appearances and layouts. We propose a robust positional and logical relation-based form key-value information extraction framework. Using this dataset, Form-NLU, we first examine strong object detection models for the form layout understanding, then evaluate the key information extraction task on the dataset, providing fine-grained results for different types of forms and keys. Furthermore, we examine it with the off-the-shelf pdf layout extraction tool and prove its feasibility in real-world cases. | ['Soyeon Caren Han', 'Hyunsuk Chung', 'Xingxiang Luo', 'Kaixuan Ren', 'Jiabin Huang', 'Siqu Long', 'Yihao Ding'] | 2023-04-04 | null | null | null | null | ['key-information-extraction'] | ['natural-language-processing'] | [ 4.64460820e-01 -2.84073591e-01 -1.38571516e-01 -1.74596384e-01
-5.04583001e-01 -1.13434863e+00 4.26416218e-01 4.58565086e-01
2.19559714e-01 3.32779557e-01 2.19989736e-02 -5.00061691e-01
-6.85532331e-01 -8.89436483e-01 -4.91906762e-01 5.39436890e-03
4.34987158e-01 6.71978235e-01 4.32949625e-02 -4.84722061e-03
7.73660898e-01 6.95965171e-01 -1.21526182e+00 6.74223006e-01
9.16767120e-01 1.21963716e+00 3.76740783e-01 7.40139127e-01
-5.80497801e-01 5.92452168e-01 -1.03712666e+00 -5.53464830e-01
3.11461657e-01 -1.77946538e-01 -6.69361889e-01 5.39913177e-01
9.10294652e-01 -9.10226822e-01 -2.23834023e-01 9.43824470e-01
1.67007282e-01 -4.17997897e-01 8.33750248e-01 -9.72592056e-01
-1.02950430e+00 6.39491677e-01 -5.14607251e-01 -5.26869655e-01
8.97418022e-01 2.70971153e-02 1.14249694e+00 -1.08728826e+00
9.88378346e-01 1.26510024e+00 3.34379464e-01 1.64535850e-01
-7.30003834e-01 -4.32275534e-01 2.28700280e-01 9.31959301e-02
-1.48332381e+00 -1.83687210e-01 8.89527738e-01 -4.29129958e-01
1.01903379e+00 3.60403866e-01 8.80846083e-01 5.71017385e-01
4.86317761e-02 1.29689121e+00 8.68867517e-01 -6.65741086e-01
1.79556504e-01 1.08157575e-01 5.80352724e-01 6.84197605e-01
6.23337328e-01 -4.68680918e-01 -5.36692321e-01 -1.72635227e-01
9.83682454e-01 -1.33171454e-01 -3.57278794e-01 -9.32054147e-02
-9.76480544e-01 1.31188080e-01 -3.04558966e-03 5.96458614e-02
-1.69267625e-01 -1.36671036e-01 -1.84930950e-01 1.62103787e-01
-3.58662516e-01 7.45789051e-01 -5.14319301e-01 -2.08660364e-01
-9.20009732e-01 4.38309282e-01 1.05950904e+00 1.81703818e+00
8.19045603e-01 -2.66157836e-01 -2.05528870e-01 7.91344881e-01
6.07797861e-01 7.99606323e-01 1.28118470e-02 -3.92375082e-01
9.08272088e-01 1.30588043e+00 3.08437139e-01 -1.05744553e+00
7.93361291e-02 -5.99682517e-02 -4.93587941e-01 -1.51704252e-01
3.96710038e-01 3.56625825e-01 -9.05612946e-01 7.89128900e-01
-2.50110626e-02 -7.50758648e-01 -2.46782809e-01 5.68251491e-01
1.00026381e+00 4.41593915e-01 -6.40915453e-01 -7.26798773e-02
1.84091914e+00 -7.98991978e-01 -1.24416256e+00 -2.37812400e-01
5.00459015e-01 -1.27366853e+00 1.30262017e+00 1.12712967e+00
-9.23983991e-01 -5.22292554e-01 -1.48413360e+00 -4.24431205e-01
-5.16883612e-01 9.13080335e-01 5.44166803e-01 5.16357899e-01
-5.07458150e-01 3.36058974e-01 -2.76246816e-01 -2.40593269e-01
4.85721558e-01 2.51818448e-01 -2.73400843e-01 -5.17418563e-01
-6.30720854e-01 3.13707143e-01 2.33937219e-01 3.85331988e-01
-3.25749397e-01 -7.65959918e-01 -7.56187916e-01 -1.64266266e-02
8.03966820e-01 -5.19233286e-01 1.35249937e+00 -1.02898516e-01
-1.15970230e+00 5.94277442e-01 -1.01474844e-01 4.93408144e-01
3.59969795e-01 -6.77807748e-01 -5.60752332e-01 1.01480268e-01
1.01686738e-01 4.05193344e-02 9.48632956e-01 -1.32154965e+00
-4.93531525e-01 -5.95166326e-01 1.02672525e-01 -1.89361855e-01
-3.49765599e-01 -8.21580216e-02 -1.03269827e+00 -8.52648497e-01
4.11061704e-01 -5.06733775e-01 4.64033335e-01 3.61388236e-01
-8.61003101e-01 -3.73785019e-01 1.11309981e+00 -9.52809691e-01
1.95927715e+00 -2.08184075e+00 -2.80504167e-01 6.78680301e-01
2.02352434e-01 2.63688147e-01 -8.72527510e-02 7.53591180e-01
2.95037419e-01 3.87973756e-01 -2.28699103e-01 -3.17605138e-02
2.98040897e-01 3.23358417e-01 -5.04810512e-01 -2.93919235e-01
2.08767548e-01 1.05113256e+00 -6.67871118e-01 -5.81928909e-01
1.02614559e-01 -5.80547377e-02 -3.10132027e-01 3.80132705e-01
-4.36250478e-01 -5.80328703e-01 -7.00660527e-01 1.46293819e+00
1.01167321e+00 -3.31690937e-01 5.48925102e-01 -7.39470005e-01
1.20702513e-01 1.37675628e-01 -1.78898585e+00 1.31647193e+00
-3.19713086e-01 4.27468210e-01 2.30661124e-01 2.55749673e-02
1.17036664e+00 -1.82497650e-01 8.23133532e-03 -7.63557971e-01
-6.62916973e-02 3.52796018e-01 -1.23612672e-01 -4.75955248e-01
1.06557095e+00 5.59227109e-01 -1.91617027e-01 7.63373077e-01
-3.74440044e-01 -7.69584656e-01 6.28023326e-01 7.29240954e-01
1.25500035e+00 1.09745868e-01 4.44640666e-01 3.26710171e-03
4.02422637e-01 -1.93358272e-01 1.67704582e-01 7.93995380e-01
4.96701956e-01 6.42420590e-01 7.59669542e-01 -2.05465138e-01
-8.22961986e-01 -1.14265919e+00 -1.43556491e-01 5.46160221e-01
3.41886640e-01 -1.37442994e+00 -6.15164518e-01 -8.61862123e-01
2.92804509e-01 4.59224015e-01 -2.70984143e-01 3.11860144e-01
-7.33019173e-01 -1.83513239e-01 4.07353252e-01 8.49925816e-01
4.82465237e-01 -1.02650166e+00 -3.34089428e-01 3.16607088e-01
3.06190550e-01 -1.20687902e+00 -7.28610873e-01 4.81676124e-02
-6.17951810e-01 -1.51013136e+00 -2.20358014e-01 -5.87142050e-01
1.20464444e+00 2.82289572e-02 1.15482426e+00 6.63400471e-01
-2.91482568e-01 6.05357468e-01 -6.37793422e-01 -3.89740944e-01
-2.47401282e-01 3.73740755e-02 -2.19834730e-01 -1.44805327e-01
3.21301043e-01 -1.37061208e-01 -3.31664771e-01 4.54582632e-01
-1.17784524e+00 1.10067204e-01 1.00130630e+00 5.20511508e-01
7.20010698e-01 3.64095032e-01 -2.97501594e-01 -8.91571522e-01
8.53047252e-01 2.76732624e-01 -5.93402445e-01 7.43458867e-01
-7.71550119e-01 1.52446836e-01 6.27170682e-01 -1.66580051e-01
-9.77433681e-01 9.32353549e-03 2.93332458e-01 1.38394594e-01
-1.04044164e-02 5.45303524e-01 -1.09982395e+00 2.34748200e-01
2.42211372e-01 2.55927444e-01 -4.81637210e-01 -1.16075504e+00
4.32761669e-01 1.02291465e+00 7.27906168e-01 -1.00027251e+00
1.01626146e+00 5.70701547e-02 -1.99858695e-01 -1.13661993e+00
-4.69386250e-01 -2.17609048e-01 -9.81830657e-01 -9.79101732e-02
3.49031776e-01 -3.49837363e-01 -4.75099862e-01 1.02420437e+00
-1.50482082e+00 -1.71800807e-01 -1.22240186e-01 -1.36751711e-01
-9.72887129e-02 5.86688936e-01 -7.13430345e-01 -6.93486094e-01
-2.55728751e-01 -1.00353897e+00 1.45635080e+00 1.47183999e-01
-3.93071741e-01 -4.66467589e-01 -5.29263079e-01 3.92509431e-01
-1.33295298e-01 -1.44479454e-01 1.39723933e+00 -1.92268744e-01
-1.11851943e+00 -3.49300325e-01 -5.90317607e-01 1.78253546e-01
5.73876619e-01 4.30043459e-01 -5.26259184e-01 4.26178090e-02
-4.58105534e-01 2.08205115e-02 3.48182321e-01 -2.68739074e-01
1.14621592e+00 -5.52726030e-01 -3.42834532e-01 4.61172909e-01
1.20832360e+00 5.00495851e-01 7.47719407e-01 1.30405456e-01
1.07706070e+00 5.65098882e-01 7.29493201e-01 7.70218849e-01
1.74836501e-01 7.57389009e-01 5.12560233e-02 3.03782791e-01
-2.54582882e-01 -8.60234857e-01 2.39230618e-01 9.07793462e-01
2.06531733e-01 -5.84481001e-01 -8.33042443e-01 3.26221406e-01
-1.59701324e+00 -2.95005351e-01 -5.98876059e-01 1.75340676e+00
8.15841854e-01 1.44027933e-01 -2.96882153e-01 4.80836749e-01
3.57361346e-01 -1.73724651e-01 -5.24590552e-01 -2.40205497e-01
-3.93525839e-01 1.85203806e-01 1.83039665e-01 3.30880642e-01
-7.24988103e-01 9.32108164e-01 6.06710815e+00 9.89171565e-01
-6.27122402e-01 -6.03768170e-01 2.34091535e-01 1.28327966e-01
-7.33765960e-01 1.72515526e-01 -1.22667432e+00 3.30321610e-01
2.84008812e-02 3.59973982e-02 3.47290874e-01 8.01521122e-01
-3.25771645e-02 -4.08645511e-01 -1.65738559e+00 1.47758007e+00
1.55951887e-01 -1.26889861e+00 6.04569554e-01 4.46715146e-01
4.47929025e-01 -1.15723419e+00 -9.04649794e-02 -2.28313088e-01
4.81779277e-02 -1.05206633e+00 1.26304889e+00 6.48896933e-01
9.76324439e-01 -2.32050508e-01 4.76456940e-01 2.69203782e-02
-1.62881577e+00 -1.25608873e-02 -9.08783078e-02 -8.69335309e-02
2.15194196e-01 7.60258198e-01 -7.34710276e-01 7.16667473e-01
4.72390324e-01 8.83406878e-01 -1.17097569e+00 7.89544165e-01
-7.02860534e-01 2.37751663e-01 -6.73404574e-01 -3.14393610e-01
-1.43916860e-01 -3.12511951e-01 7.42879659e-02 9.62217212e-01
5.95138848e-01 1.18697166e-01 1.29793927e-01 1.26081002e+00
-9.37580168e-02 2.81981438e-01 -3.95942539e-01 -5.84702849e-01
9.09204006e-01 1.30519080e+00 -9.39395785e-01 -5.48881710e-01
-2.20917091e-01 9.77253735e-01 -2.66831205e-03 3.89231026e-01
-2.31359780e-01 -7.65927374e-01 3.33509326e-01 4.80036944e-01
5.57226002e-01 -4.18379724e-01 -6.85416937e-01 -7.35772610e-01
9.03692126e-01 -1.12029219e+00 1.68429345e-01 -1.17334402e+00
-1.22177243e+00 1.68086246e-01 9.18518007e-02 -1.36900878e+00
-2.00528353e-01 -1.23604250e+00 -3.50436747e-01 6.00805581e-01
-9.93483901e-01 -1.43915939e+00 -4.30353254e-01 2.74245441e-01
6.10003471e-01 -2.00273812e-01 3.67263407e-01 2.23805651e-01
-4.92167056e-01 7.90215075e-01 -1.69412509e-01 4.28991139e-01
6.10089779e-01 -1.31079662e+00 4.73562777e-01 7.59232879e-01
3.36062193e-01 1.20075142e+00 1.49366826e-01 -1.05858433e+00
-2.19629693e+00 -6.13215685e-01 7.27735162e-01 -8.15374374e-01
5.30243874e-01 -7.99884200e-01 -8.95455539e-01 4.09665704e-01
-1.07848868e-01 -3.91051501e-01 3.37879509e-01 -2.08802998e-01
-4.83942628e-01 -7.25997090e-02 -1.03968656e+00 4.88990813e-01
1.36968756e+00 -6.99304819e-01 -6.80217206e-01 1.21447377e-01
2.96370775e-01 -4.14741844e-01 -8.54469597e-01 9.95797664e-02
9.81556594e-01 -7.81478345e-01 6.22449815e-01 -1.96789891e-01
5.20237744e-01 -7.05272377e-01 -3.29770327e-01 -7.41736174e-01
-7.76043907e-02 -8.20159376e-01 -6.82487905e-01 1.56674504e+00
3.32979172e-01 -1.07116878e-01 7.68603206e-01 8.70190740e-01
-1.69825763e-01 -8.68567526e-01 -3.38934988e-01 -8.40607822e-01
-5.43084979e-01 -7.50528395e-01 1.16804624e+00 4.89850044e-01
-9.70166549e-02 4.17073250e-01 -8.77869949e-02 -2.41899174e-02
3.67350698e-01 4.00375843e-01 1.07565367e+00 -1.14312935e+00
-4.09149647e-01 -4.54993039e-01 -1.48241863e-01 -1.75407159e+00
-3.89287114e-01 -5.89205086e-01 -2.75305361e-01 -1.72906566e+00
-2.59998739e-02 -2.61358470e-01 2.57631749e-01 6.34595037e-01
-7.78460223e-03 -2.89561152e-01 5.20790637e-01 2.20440552e-01
-4.33068126e-01 1.65474236e-01 1.31631887e+00 -7.62538135e-01
-3.37111115e-01 -2.24247560e-01 -9.13784802e-01 6.42260313e-01
9.63111222e-02 -1.01052679e-01 -4.02076185e-01 -5.08606195e-01
7.28441000e-01 -3.44180495e-01 1.33085012e-01 -6.49658740e-01
2.28800863e-01 -3.89993303e-02 8.08202863e-01 -1.26575875e+00
2.77327397e-03 -1.02768981e+00 8.74130651e-02 2.47049898e-01
1.84350267e-01 3.32104743e-01 8.03112686e-02 2.08419681e-01
8.64160284e-02 -5.90873003e-01 7.31548667e-02 -5.80711924e-02
-9.34450805e-01 2.52308249e-01 -2.31057912e-01 -1.91028237e-01
4.88370001e-01 -7.40148008e-01 -5.99029481e-01 -7.13150203e-02
-2.54555225e-01 1.03200592e-01 6.40133917e-01 5.36720574e-01
1.11729693e+00 -1.51127255e+00 -3.84850830e-01 7.57526696e-01
4.09522146e-01 1.88040584e-01 1.68937945e-03 2.26247907e-01
-7.95305908e-01 2.59006411e-01 1.72853366e-01 -4.00930703e-01
-1.40090334e+00 3.96673948e-01 -1.83678672e-01 -2.88251281e-01
-3.63946468e-01 4.24226254e-01 6.28551766e-02 -2.40931496e-01
2.96309348e-02 -1.00188375e+00 1.29792765e-02 3.21374714e-01
8.62316310e-01 3.58135730e-01 3.40231180e-01 -2.34918460e-01
-3.71471435e-01 1.20372272e+00 -4.61880565e-01 -8.74589235e-02
9.81063008e-01 1.97301865e-01 -3.61687750e-01 2.66920384e-02
9.21935678e-01 4.78705138e-01 -9.63417888e-01 -1.60027444e-01
2.97822773e-01 -8.53305697e-01 -3.12158495e-01 -1.24237788e+00
-6.60163224e-01 6.04855657e-01 2.69057989e-01 4.46127430e-02
9.21840429e-01 1.07142450e-02 1.07278526e+00 8.17213416e-01
6.51928842e-01 -1.50809038e+00 4.15717483e-01 6.35946453e-01
1.31883383e+00 -7.37823784e-01 4.25048977e-01 -1.00731766e+00
-1.32463038e-01 1.37582135e+00 8.07393193e-01 4.04954821e-01
5.52524209e-01 6.75318718e-01 2.44757496e-02 -5.06474793e-01
-4.75125879e-01 7.94180232e-05 5.99912882e-01 8.21832299e-01
2.76505470e-01 9.38496292e-02 -1.11775264e-01 1.21966219e+00
-5.92806756e-01 -7.04605952e-02 5.46917796e-01 1.43091476e+00
-1.92736104e-01 -1.50426638e+00 -6.72869027e-01 8.05838823e-01
1.24520682e-01 -1.99685186e-01 -1.23542678e+00 8.76736224e-01
2.08957985e-01 7.62472749e-01 -9.78050157e-02 -4.18548346e-01
8.17375064e-01 -1.05662778e-01 8.35890532e-01 -7.12142766e-01
-3.96370351e-01 1.97159663e-01 2.22371429e-01 -4.16008830e-01
2.82807678e-01 -5.17454505e-01 -1.18242741e+00 -1.01341598e-01
-3.21040511e-01 -2.31725544e-01 4.60449487e-01 8.19027781e-01
3.20829958e-01 4.99597698e-01 3.50963354e-01 -3.17325324e-01
-4.01758701e-01 -9.74298358e-01 -8.95387232e-01 5.66885352e-01
-5.18368892e-02 -4.25510854e-01 -2.53394265e-02 2.89684981e-01] | [11.623120307922363, 2.62036395072937] |
3ccba667-0509-4b8a-980f-64e5e80eb29c | outcast-outdoor-single-image-relighting-with | 2204.09341 | null | https://arxiv.org/abs/2204.09341v1 | https://arxiv.org/pdf/2204.09341v1.pdf | OutCast: Outdoor Single-image Relighting with Cast Shadows | We propose a relighting method for outdoor images. Our method mainly focuses on predicting cast shadows in arbitrary novel lighting directions from a single image while also accounting for shading and global effects such the sun light color and clouds. Previous solutions for this problem rely on reconstructing occluder geometry, e.g. using multi-view stereo, which requires many images of the scene. Instead, in this work we make use of a noisy off-the-shelf single-image depth map estimation as a source of geometry. Whilst this can be a good guide for some lighting effects, the resulting depth map quality is insufficient for directly ray-tracing the shadows. Addressing this, we propose a learned image space ray-marching layer that converts the approximate depth map into a deep 3D representation that is fused into occlusion queries using a learned traversal. Our proposed method achieves, for the first time, state-of-the-art relighting results, with only a single image as input. For supplementary material visit our project page at: https://dgriffiths.uk/outcast. | ['Julien Philip', 'Tobias Ritschel', 'David Griffiths'] | 2022-04-20 | null | null | null | null | ['image-relighting'] | ['computer-vision'] | [ 3.35672230e-01 6.26197900e-04 4.16341931e-01 -5.90967774e-01
-7.84499824e-01 -6.60887063e-01 6.35079920e-01 7.95277134e-02
-1.45802379e-01 6.78849161e-01 4.11702134e-02 -3.04751217e-01
4.02984679e-01 -1.16037774e+00 -9.40579116e-01 -5.82478642e-01
3.08911711e-01 6.67813838e-01 5.94722271e-01 -3.01546961e-01
2.55393893e-01 6.27172470e-01 -1.73130989e+00 2.66036838e-01
9.86783504e-01 9.22018111e-01 3.89931232e-01 1.02814639e+00
-2.61166871e-01 6.60573304e-01 -4.09211397e-01 -3.36554945e-01
3.51726562e-01 -2.42095277e-01 -4.28739160e-01 1.61280081e-01
8.43545496e-01 -9.23097730e-01 -2.16236189e-01 6.39245987e-01
3.69329542e-01 6.65481240e-02 3.47962320e-01 -8.23492467e-01
-1.09193094e-01 -3.58196318e-01 -5.18781781e-01 -2.24937603e-01
7.05621123e-01 9.84837860e-02 7.81080782e-01 -8.76067817e-01
7.58307517e-01 1.00829113e+00 5.70974886e-01 1.79276302e-01
-1.25895691e+00 -5.18191934e-01 1.21645533e-01 -2.41273493e-02
-1.13267446e+00 -3.78867090e-01 9.83005404e-01 -2.85040766e-01
7.43687272e-01 4.80403811e-01 9.00517642e-01 9.17590439e-01
8.19177404e-02 5.57909191e-01 1.52895844e+00 -3.57847363e-01
3.29479277e-01 3.00947100e-01 -2.43071556e-01 7.57913053e-01
-5.92293516e-02 1.26955286e-01 -6.96835935e-01 -9.38189626e-02
8.17498326e-01 4.12389040e-02 -4.85235929e-01 -6.79270864e-01
-7.30316460e-01 5.66405594e-01 6.25275433e-01 -2.49725461e-01
-9.06002522e-02 1.83208585e-01 -1.97321206e-01 9.77609754e-02
9.07404602e-01 2.13551857e-02 -4.11389351e-01 -5.93105815e-02
-1.07378256e+00 4.33561981e-01 7.23639905e-01 6.65551126e-01
1.38208711e+00 -2.45275736e-01 3.59264374e-01 5.43046832e-01
3.17388862e-01 6.30745113e-01 -2.67951757e-01 -1.25065732e+00
3.89461577e-01 4.73912239e-01 3.93777937e-01 -7.22906470e-01
-2.50952661e-01 -3.01156461e-01 -3.90768737e-01 7.98941135e-01
4.57693398e-01 1.21847145e-01 -9.64768231e-01 1.27583826e+00
7.98428655e-01 2.47162238e-01 -2.08526969e-01 1.01382637e+00
7.20471084e-01 7.08903491e-01 -6.99229658e-01 9.62383021e-03
1.12819862e+00 -1.00773120e+00 -3.94502372e-01 -3.72558773e-01
2.47701809e-01 -1.07371426e+00 1.13439310e+00 7.47431338e-01
-1.12865162e+00 -2.06995606e-01 -9.06508386e-01 -6.47296369e-01
-3.92459899e-01 -1.94635361e-01 6.63795829e-01 6.22945368e-01
-1.16691375e+00 6.38993621e-01 -1.00378108e+00 -3.84403557e-01
2.75602877e-01 -1.31603694e-02 -6.33463636e-02 -5.53587973e-01
-7.98272491e-01 7.22291887e-01 -2.62889266e-01 4.29994985e-02
-9.08325493e-01 -8.63901734e-01 -7.65457988e-01 -2.46041775e-01
3.91729146e-01 -7.47316480e-01 1.08721459e+00 -6.06122673e-01
-1.55150855e+00 7.90154457e-01 -5.39928377e-01 4.01832238e-02
8.78524244e-01 -4.29421186e-01 1.79194346e-01 2.32722282e-01
-6.83247223e-02 4.37329203e-01 6.80856586e-01 -2.06761193e+00
-3.40525627e-01 -5.05366623e-01 4.78418291e-01 4.40497100e-01
1.59441039e-01 -3.52693230e-01 -6.74574852e-01 -2.42493078e-01
2.20725060e-01 -7.82799423e-01 -1.91166610e-01 5.85704327e-01
-3.42758238e-01 5.13184965e-01 8.07003081e-01 -8.27408373e-01
7.51610875e-01 -1.73845649e+00 8.23044777e-02 1.22264005e-01
6.18842803e-02 -1.24305956e-01 2.25246310e-01 6.66743696e-01
3.13749462e-01 -2.77861863e-01 -4.58862782e-01 -1.04973376e+00
-1.60226822e-01 4.63169098e-01 -5.35513222e-01 6.60818934e-01
-1.69378489e-01 5.96633792e-01 -8.70474279e-01 -3.50329161e-01
7.11249769e-01 1.06836855e+00 -6.78373456e-01 4.66010034e-01
-4.73279476e-01 9.05568659e-01 -1.32137731e-01 5.72535455e-01
1.14206111e+00 3.57706137e-02 -6.20152093e-02 -4.61952016e-02
-4.79255766e-01 4.27551419e-01 -1.16630280e+00 2.15379262e+00
-9.55814838e-01 5.85531831e-01 2.94599801e-01 -3.20806175e-01
7.16251075e-01 -3.28752100e-02 2.88245738e-01 -7.05285609e-01
-7.18611926e-02 2.83669293e-01 -6.81028187e-01 -1.77416593e-01
5.43859065e-01 -1.19650073e-01 4.41287518e-01 4.69322652e-01
-3.85811746e-01 -7.59953856e-01 -2.11152658e-01 3.20039839e-01
9.55682099e-01 8.26609194e-01 -2.59298948e-03 -4.02888693e-02
1.69558659e-01 2.56530531e-02 2.73825467e-01 5.85567415e-01
4.84907418e-01 1.27676690e+00 2.82385319e-01 -4.70342845e-01
-1.08700788e+00 -1.20049751e+00 -3.56700003e-01 6.69366241e-01
3.69744629e-01 -4.77230549e-01 -7.82979846e-01 -2.68398136e-01
-5.73347248e-02 8.14437449e-01 -5.71861923e-01 5.02694011e-01
-5.84058881e-01 -5.94153821e-01 6.92891609e-03 6.66144714e-02
4.87770408e-01 -5.31515062e-01 -1.02963507e+00 -9.25317332e-02
-2.78772265e-01 -1.09088206e+00 -1.20954819e-01 2.42412776e-01
-9.69681323e-01 -1.00568545e+00 -6.55114293e-01 -9.03961062e-02
6.26225591e-01 6.64748847e-01 1.32175183e+00 2.80063391e-01
-5.54770947e-01 3.60593259e-01 -2.43350953e-01 -3.10985029e-01
-1.88309431e-01 -3.58822346e-01 -4.00871038e-01 2.64426507e-02
-2.78987199e-01 -9.99061048e-01 -1.21760499e+00 2.26004809e-01
-1.04872072e+00 6.15716517e-01 1.75405741e-01 4.47760105e-01
6.35608971e-01 -2.67598122e-01 -4.23920453e-01 -1.04090476e+00
-1.34911850e-01 -3.02525908e-01 -9.88167167e-01 -4.02499065e-02
-5.63476861e-01 -1.36257723e-01 5.79567850e-01 1.25917062e-01
-1.38736653e+00 1.83111593e-01 -2.94957012e-01 -4.38500464e-01
-3.05479944e-01 -1.28988862e-01 -9.58176479e-02 -2.79453784e-01
5.70363343e-01 1.54573813e-01 -2.96736956e-01 -7.33418643e-01
4.49024826e-01 3.95152897e-01 3.80676001e-01 -5.61733782e-01
8.95133078e-01 1.20745385e+00 2.94146001e-01 -9.60767090e-01
-9.67256010e-01 -5.82880437e-01 -7.55157948e-01 -3.67871672e-01
5.84699154e-01 -9.87993360e-01 -6.13695979e-01 5.34510434e-01
-1.25535381e+00 -7.82273412e-01 -5.22614047e-02 4.67296652e-02
-4.64720100e-01 4.79416519e-01 -4.27673727e-01 -9.43448365e-01
5.66515997e-02 -1.07595599e+00 1.79500031e+00 -4.00256775e-02
2.47621492e-01 -9.44125473e-01 2.95066267e-01 5.65826774e-01
2.12294549e-01 5.75967789e-01 5.14962077e-01 5.99816203e-01
-1.27160764e+00 1.24857150e-01 -2.49374554e-01 1.53624937e-01
1.20572649e-01 -1.05774418e-01 -1.63806224e+00 -1.74720690e-01
1.31905720e-01 -3.51843566e-01 9.15742755e-01 2.10390702e-01
1.11579370e+00 -1.04080938e-01 -1.79042786e-01 1.12586606e+00
1.98084724e+00 -1.99058577e-01 8.27205539e-01 3.11904520e-01
1.14960170e+00 7.73644865e-01 5.76305032e-01 6.64168835e-01
7.04877853e-01 9.19962883e-01 8.29800963e-01 -4.34812248e-01
-3.97164017e-01 -2.63905227e-01 -7.76424035e-02 4.26971585e-01
-2.14320317e-01 -4.79049534e-01 -9.60925698e-01 3.71602595e-01
-1.60012901e+00 -5.70578158e-01 -4.83942777e-01 2.39209938e+00
7.51012743e-01 -8.81677344e-02 -3.36125433e-01 2.67212708e-02
-2.02467595e-03 3.71176064e-01 -5.83543420e-01 -3.14518869e-01
-4.76937033e-02 3.30002964e-01 7.18540132e-01 1.32392776e+00
-5.46796143e-01 8.99517834e-01 5.00502348e+00 3.70011568e-01
-1.08979726e+00 2.50672549e-01 6.50358737e-01 -2.51865953e-01
-9.27359998e-01 4.02758062e-01 -5.23971200e-01 2.32493564e-01
5.66711724e-01 5.71084917e-01 8.02790761e-01 5.34466684e-01
4.36513990e-01 -8.64927173e-01 -7.74698913e-01 1.10920763e+00
2.00105682e-01 -1.12952459e+00 -1.87248260e-01 2.77180761e-01
8.69808197e-01 2.69011080e-01 -1.58137903e-01 -3.05332571e-01
3.14180911e-01 -8.94178927e-01 7.76520908e-01 5.65186620e-01
8.79250407e-01 -5.11810005e-01 2.10323974e-01 3.49409372e-01
-1.06239104e+00 4.05350953e-01 -2.20956430e-01 -2.11175472e-01
5.02210021e-01 1.04272866e+00 -6.36104643e-01 6.86494529e-01
7.33246386e-01 4.75717068e-01 -6.44237638e-01 9.24070060e-01
-3.57736051e-01 3.83349240e-01 -7.31307030e-01 5.18722475e-01
4.03677821e-02 -4.35624599e-01 3.68706554e-01 9.41608369e-01
3.86821598e-01 1.37880430e-01 1.40210256e-01 7.31546044e-01
9.24993232e-02 -1.94227234e-01 -7.18271732e-01 7.04340816e-01
2.17029750e-02 1.28763556e+00 -8.08183551e-01 -2.07278118e-01
-4.68009144e-01 1.45144677e+00 4.42163676e-01 5.13829529e-01
-6.99182391e-01 -1.31703585e-01 7.40842521e-01 5.44761300e-01
1.94253296e-01 -4.73215520e-01 -5.85067868e-01 -1.39313614e+00
2.47611761e-01 -3.93573463e-01 -1.48638263e-01 -1.18492258e+00
-6.64611161e-01 4.29440767e-01 1.30978762e-03 -1.12089241e+00
-1.00248702e-01 -4.88462865e-01 -4.66550171e-01 1.02786851e+00
-2.14390612e+00 -1.28259838e+00 -8.54134560e-01 5.54158688e-01
5.85477293e-01 6.96203411e-01 8.17454338e-01 1.75383121e-01
4.10987921e-02 -6.70051947e-02 3.37437540e-01 -5.44445515e-01
8.34420502e-01 -1.55210948e+00 5.36826134e-01 8.74133229e-01
1.82461858e-01 3.27831119e-01 1.02675784e+00 -5.48233926e-01
-1.66846740e+00 -8.75928998e-01 6.99358344e-01 -6.10951424e-01
2.72368640e-01 -7.92484581e-01 -8.58611286e-01 4.85482305e-01
1.50281042e-01 2.90545970e-01 2.22808048e-01 -2.04985529e-01
-3.71710777e-01 -1.64413542e-01 -1.14661431e+00 5.40989757e-01
1.28149712e+00 -5.26561081e-01 8.76572281e-02 5.46378493e-01
5.11366010e-01 -9.89182174e-01 -5.77777088e-01 1.08196124e-01
7.13856518e-01 -1.79116797e+00 1.16584337e+00 3.34078372e-01
6.61415637e-01 -5.02716601e-01 -2.45561138e-01 -1.24490929e+00
2.70930856e-01 -6.82471275e-01 -9.96429920e-02 9.24844265e-01
7.26365447e-02 -6.15069687e-01 8.94673049e-01 5.33961236e-01
-2.43716374e-01 -8.12695146e-01 -8.40622008e-01 -3.44030529e-01
-2.38061443e-01 -7.31154263e-01 5.23588121e-01 5.16245663e-01
-6.00899756e-01 7.06503689e-02 -4.18817133e-01 5.56928396e-01
9.99901891e-01 5.29579759e-01 1.11332881e+00 -1.03467834e+00
-5.73150277e-01 1.81495354e-01 -2.71535781e-03 -1.30026150e+00
-1.45268634e-01 -5.41181386e-01 1.88344985e-01 -1.84747636e+00
-3.29021513e-02 -6.43938541e-01 3.83922428e-01 4.20898870e-02
7.24300090e-03 6.59172595e-01 -1.12279998e-02 1.00417905e-01
-2.30568573e-01 5.53507626e-01 1.27560484e+00 2.75481880e-01
-1.77456826e-01 -1.73018798e-01 -2.91059405e-01 7.74177432e-01
8.57050896e-01 -3.76659572e-01 -5.38401008e-01 -9.70641196e-01
4.51410800e-01 1.28809199e-01 6.93567634e-01 -9.00804996e-01
2.81836838e-02 -1.98990360e-01 4.48640019e-01 -7.70531118e-01
1.06745350e+00 -9.37972665e-01 3.23652685e-01 2.45992690e-01
1.89177513e-01 -2.23594934e-01 1.58017263e-01 5.92033684e-01
1.27236068e-01 -1.18996747e-01 6.67218089e-01 -4.34516817e-01
-3.25572997e-01 2.20150635e-01 8.62155482e-02 -1.88167334e-01
5.48943222e-01 -3.57422084e-01 -3.64640713e-01 -5.48663676e-01
-2.81940430e-01 3.56398001e-02 1.25866103e+00 2.72267647e-02
7.35380590e-01 -8.68354738e-01 -5.15440643e-01 2.86491394e-01
4.96338867e-02 6.20048106e-01 3.10644448e-01 4.37003195e-01
-1.09682775e+00 1.51035920e-01 2.82762915e-01 -6.74479663e-01
-1.26947761e+00 1.51819214e-01 3.61305267e-01 -1.01909630e-01
-9.89056468e-01 7.22600937e-01 6.36015773e-01 -5.60611129e-01
-1.60232820e-02 -4.05235618e-01 3.07231337e-01 -2.87881851e-01
4.27061349e-01 2.97826856e-01 3.15648109e-01 -5.53313911e-01
-2.41008058e-01 1.00146949e+00 2.94251859e-01 -5.42915761e-01
1.39765608e+00 -4.62023199e-01 -2.28506908e-01 6.55664563e-01
1.22280467e+00 3.37091535e-01 -1.69937253e+00 -4.66760062e-02
-6.56011641e-01 -1.16002274e+00 4.09750223e-01 -7.86356807e-01
-8.80926609e-01 1.10781860e+00 4.69427645e-01 -2.00819731e-01
1.16810441e+00 -4.88336794e-02 9.97554243e-01 1.81320712e-01
7.35981524e-01 -7.28504181e-01 -7.15607777e-02 3.83469313e-01
8.89160037e-01 -1.38359201e+00 4.21902895e-01 -6.99872077e-01
-2.26084813e-01 1.16313517e+00 3.63091648e-01 -1.05853453e-01
6.89064026e-01 3.91263217e-01 8.91138688e-02 -1.78189546e-01
-4.42150027e-01 -2.11701855e-01 -1.12089040e-02 5.92148364e-01
3.02273273e-01 -1.50802374e-01 4.53010090e-02 -3.69381219e-01
-3.91507655e-01 -3.89023833e-02 7.33053327e-01 9.55151498e-01
-3.23065817e-01 -1.18271232e+00 -5.48383832e-01 1.99355364e-01
-1.94923028e-01 -2.20071048e-01 -2.09363431e-01 4.21763211e-01
2.10235670e-01 8.13892841e-01 -2.14975420e-02 -3.29970345e-02
2.49959841e-01 -2.37282187e-01 7.28569329e-01 -5.94282568e-01
-2.95232713e-01 1.85052246e-01 1.25519365e-01 -9.50204670e-01
-5.18142521e-01 -6.94202721e-01 -8.84214699e-01 -2.69294411e-01
-5.43267727e-02 -3.90934736e-01 1.15845108e+00 6.46454155e-01
1.56358808e-01 3.36619437e-01 6.08318388e-01 -1.46848810e+00
2.21160412e-01 -5.72466373e-01 -5.31279504e-01 2.14588687e-01
7.72368848e-01 -6.38661742e-01 -5.48786700e-01 4.00357954e-02] | [9.668317794799805, -3.0238871574401855] |
b8e33fa4-826f-49f1-9402-22d4d5022f76 | demeshnet-blind-face-inpainting-for-deep | 1611.05271 | null | http://arxiv.org/abs/1611.05271v1 | http://arxiv.org/pdf/1611.05271v1.pdf | DeMeshNet: Blind Face Inpainting for Deep MeshFace Verification | MeshFace photos have been widely used in many Chinese business organizations
to protect ID face photos from being misused. The occlusions incurred by random
meshes severely degenerate the performance of face verification systems, which
raises the MeshFace verification problem between MeshFace and daily photos.
Previous methods cast this problem as a typical low-level vision problem, i.e.
blind inpainting. They recover perceptually pleasing clear ID photos from
MeshFaces by enforcing pixel level similarity between the recovered ID images
and the ground-truth clear ID images and then perform face verification on
them. Essentially, face verification is conducted on a compact feature space
rather than the image pixel space. Therefore, this paper argues that pixel
level similarity and feature level similarity jointly offer the key to improve
the verification performance. Based on this insight, we offer a novel feature
oriented blind face inpainting framework. Specifically, we implement this by
establishing a novel DeMeshNet, which consists of three parts. The first part
addresses blind inpainting of the MeshFaces by implicitly exploiting extra
supervision from the occlusion position to enforce pixel level similarity. The
second part explicitly enforces a feature level similarity in the compact
feature space, which can explore informative supervision from the feature space
to produce better inpainting results for verification. The last part copes with
face alignment within the net via a customized spatial transformer module when
extracting deep facial features. All the three parts are implemented within an
end-to-end network that facilitates efficient optimization. Extensive
experiments on two MeshFace datasets demonstrate the effectiveness of the
proposed DeMeshNet as well as the insight of this paper. | ['Tieniu Tan', 'Shu Zhang', 'Ran He'] | 2016-11-16 | null | null | null | null | ['facial-inpainting'] | ['computer-vision'] | [ 2.73733616e-01 5.73782437e-02 -1.41502336e-01 -5.05623996e-01
-5.30058384e-01 -3.91608357e-01 3.70443761e-01 -7.89733708e-01
-5.61230443e-02 5.13788819e-01 2.16989398e-01 -1.18871639e-02
-4.24698256e-02 -6.19022548e-01 -7.80345976e-01 -7.64184654e-01
3.95084023e-01 -6.25199378e-02 -4.15556341e-01 -2.58585867e-02
2.22550079e-01 8.14939439e-01 -1.78024757e+00 1.68507859e-01
1.05612385e+00 1.22569907e+00 -2.57158671e-02 3.84709015e-02
-1.48068205e-01 4.16880429e-01 -4.76445049e-01 -7.52751231e-01
8.06833208e-01 -3.31648171e-01 -6.05093539e-01 6.42125368e-01
9.00016844e-01 -7.32709169e-01 -4.63057011e-01 1.39756978e+00
4.40534651e-01 -1.87614635e-01 3.76598269e-01 -1.63565338e+00
-9.00563121e-01 2.09777713e-01 -9.60006297e-01 -2.61200279e-01
4.78582144e-01 4.37391251e-01 6.53639972e-01 -1.25852072e+00
5.15966713e-01 1.58122253e+00 7.34838724e-01 7.04231739e-01
-1.05510902e+00 -1.00358713e+00 1.54886082e-01 1.81362495e-01
-1.62084663e+00 -8.33539248e-01 1.05627751e+00 -2.06779197e-01
3.07548076e-01 3.61612797e-01 4.62538630e-01 8.09683859e-01
-9.13001597e-02 5.35641432e-01 1.13877451e+00 -3.19167078e-01
-1.98708832e-01 2.02927366e-01 -2.56656647e-01 9.73213494e-01
2.26059496e-01 3.60249758e-01 -5.72143257e-01 -8.18041414e-02
9.28465068e-01 4.79125887e-01 -7.35487163e-01 -2.75787532e-01
-7.88120508e-01 6.14070654e-01 6.81109548e-01 3.21834488e-03
-3.00648510e-01 -1.76697299e-01 -3.46403904e-02 3.56954366e-01
3.14925194e-01 -1.43903479e-01 -6.53446987e-02 5.11176407e-01
-1.08967495e+00 8.67998451e-02 5.00349402e-01 9.84656096e-01
1.04749918e+00 6.27487078e-02 -1.68450460e-01 9.03008163e-01
7.08581626e-01 5.75024128e-01 1.74732983e-01 -9.91950631e-01
5.64989507e-01 6.47053599e-01 1.73328221e-02 -1.30449581e+00
1.44562989e-01 -9.27522704e-02 -9.87546444e-01 5.11199594e-01
2.60808736e-01 -5.05531840e-02 -9.82906640e-01 1.77991498e+00
5.52390695e-01 3.01725119e-01 -1.50696769e-01 1.16048574e+00
7.16986299e-01 4.58253741e-01 -3.07554156e-01 -3.06932181e-01
1.53101647e+00 -9.81138051e-01 -7.93492436e-01 -1.54964015e-01
-7.66237229e-02 -9.23735976e-01 1.12203729e+00 1.16912097e-01
-1.01361358e+00 -6.76760018e-01 -1.22752357e+00 -3.28422606e-01
-1.06155194e-01 2.42929608e-01 4.34665650e-01 8.40730906e-01
-1.23773193e+00 4.95985180e-01 -4.07354772e-01 -2.08511546e-01
7.89982498e-01 4.87931490e-01 -6.83156490e-01 -3.16997170e-01
-1.00568199e+00 5.20435929e-01 -1.61433190e-01 5.68920851e-01
-6.71668649e-01 -7.15009451e-01 -9.84074712e-01 8.95139650e-02
6.84811845e-02 -6.05319738e-01 9.13320363e-01 -1.36290026e+00
-1.46617305e+00 1.09159577e+00 -4.12931830e-01 -1.88434906e-02
7.24633098e-01 1.06651348e-03 -5.84909618e-01 2.39917487e-01
3.42841506e-01 6.82651043e-01 1.64359796e+00 -1.50197601e+00
-5.09955466e-01 -6.30533576e-01 -2.19905317e-01 1.12886615e-01
-5.49169004e-01 1.84701592e-01 -6.95117950e-01 -8.17897916e-01
2.29305461e-01 -5.75421512e-01 2.06725955e-01 6.28621221e-01
-3.91488135e-01 1.74439639e-01 1.17355704e+00 -1.01242197e+00
8.12216818e-01 -2.32347417e+00 -3.02160215e-02 4.10320640e-01
2.79725105e-01 2.22747877e-01 -2.57084340e-01 1.91863582e-01
-3.91218901e-01 -1.51917100e-01 -3.47269654e-01 -7.21195877e-01
-8.48098099e-03 6.82718009e-02 -5.34395218e-01 8.35666120e-01
4.98322755e-01 8.12216759e-01 -4.19798166e-01 -5.96848190e-01
1.40957415e-01 7.62434185e-01 -5.27925074e-01 3.84099633e-01
1.33918107e-01 4.75079507e-01 -1.85876310e-01 1.11249197e+00
1.58409691e+00 6.47653714e-02 5.01303412e-02 -4.64350224e-01
5.89194987e-03 -2.53488898e-01 -1.28889513e+00 1.62270749e+00
-1.92711070e-01 2.22725168e-01 7.29967058e-01 -7.55153596e-01
8.65331888e-01 2.60924906e-01 3.11204970e-01 -6.25120759e-01
2.42277369e-01 1.38836384e-01 -3.66844803e-01 -5.00817239e-01
1.68444276e-01 -1.52961820e-01 3.93834442e-01 3.33777606e-01
-1.17954724e-01 7.02192709e-02 -3.51618707e-01 -4.77162898e-02
5.74523449e-01 1.33638397e-01 -1.29924715e-01 -1.79600835e-01
9.21896517e-01 -4.73750442e-01 7.54975855e-01 1.62047520e-01
-2.52119899e-01 7.96278656e-01 2.96674013e-01 -4.03245807e-01
-7.81292260e-01 -1.01230133e+00 -2.71764666e-01 4.68823880e-01
4.17495489e-01 -1.82091475e-01 -1.07438540e+00 -8.42428088e-01
2.37120688e-01 5.49745597e-02 -6.57732964e-01 -8.36836547e-02
-5.43256462e-01 -2.60433376e-01 5.37740886e-01 3.00036162e-01
9.56354678e-01 -8.91730607e-01 -2.05176007e-02 -2.88800508e-01
-2.58506000e-01 -1.00620437e+00 -9.80890155e-01 -4.59537625e-01
-6.28812730e-01 -1.13735807e+00 -7.09380627e-01 -1.08311188e+00
1.18270671e+00 6.22523546e-01 6.72814429e-01 5.94861984e-01
-4.55156267e-01 3.00832748e-01 -4.90624160e-02 -1.54441401e-01
-1.78923845e-01 -4.35093731e-01 3.46952736e-01 7.25061774e-01
1.85402840e-01 -6.98782742e-01 -7.35643208e-01 4.26783025e-01
-1.01881456e+00 -1.10746541e-05 6.48971796e-01 9.54636991e-01
4.85245377e-01 -1.39736908e-03 3.84886205e-01 -4.56818938e-01
3.70467842e-01 -7.75379241e-02 -7.63111830e-01 3.93649846e-01
-5.65787613e-01 -1.14588797e-01 5.94505787e-01 -2.42579833e-01
-1.13271761e+00 2.29604661e-01 -1.69150621e-01 -9.95794237e-01
1.48944706e-01 -1.08490601e-01 -8.79524887e-01 -6.01991713e-01
1.78723335e-01 3.07254672e-01 6.13910317e-01 -4.27822679e-01
2.93667704e-01 8.46437454e-01 6.94493055e-01 -4.75822955e-01
1.43203366e+00 7.37172604e-01 -4.04669642e-02 -6.71048403e-01
-4.63219941e-01 -2.28340074e-01 -3.46459389e-01 -2.48474315e-01
6.97663128e-01 -9.59557712e-01 -1.02670455e+00 6.45224988e-01
-1.13123405e+00 3.18264961e-01 -1.03729710e-01 1.42701417e-01
-4.54506308e-01 7.28658080e-01 -4.99314070e-01 -6.71679080e-01
-4.42356110e-01 -1.34561682e+00 1.38059366e+00 5.41253209e-01
3.95613164e-01 -5.48786581e-01 -4.38477635e-01 5.01715899e-01
2.49730900e-01 6.81871921e-02 5.83212554e-01 1.41144887e-01
-8.86950314e-01 -1.13777541e-01 -6.41883790e-01 5.94638526e-01
4.99327987e-01 -3.13188285e-02 -1.36090279e+00 -5.60086846e-01
3.48747760e-01 -9.71859768e-02 7.52156675e-01 9.67444628e-02
1.29285276e+00 -5.64039171e-01 -2.85331756e-01 1.17114770e+00
1.44310319e+00 -1.47199646e-01 7.91194856e-01 -2.61073071e-03
8.16894889e-01 7.86160767e-01 7.06528485e-01 3.43322843e-01
1.22843385e-01 6.61892653e-01 6.65811718e-01 -3.79001081e-01
-3.27914029e-01 -4.99825478e-01 5.07594228e-01 3.88162255e-01
1.65241674e-01 1.50641546e-01 -4.41373080e-01 3.12394321e-01
-1.62186623e+00 -8.92286479e-01 3.43687743e-01 2.14094543e+00
8.57972324e-01 -4.56699431e-01 -2.08755448e-01 1.54724836e-01
9.55664694e-01 1.52519599e-01 -5.26523471e-01 -3.66325974e-02
-1.66164592e-01 1.49981394e-01 3.24021637e-01 5.58424711e-01
-9.51732814e-01 7.97114432e-01 4.78059769e+00 9.17061806e-01
-1.19190061e+00 6.12609573e-02 7.15283930e-01 1.54455990e-01
-1.87607288e-01 4.43988293e-02 -8.44420910e-01 6.10217869e-01
1.33125156e-01 5.70791140e-02 7.53816426e-01 7.02191532e-01
1.43010467e-01 2.40064427e-01 -1.06972742e+00 1.49161434e+00
3.92884851e-01 -1.31982398e+00 2.53865808e-01 3.11799377e-01
6.26863718e-01 -5.85029542e-01 3.83585960e-01 -1.48665741e-01
-1.39971241e-01 -1.22350550e+00 7.66661286e-01 5.26125610e-01
1.21882224e+00 -8.69234085e-01 4.98642951e-01 7.76507612e-03
-1.23409224e+00 -2.05473602e-01 -4.95875984e-01 2.35652551e-01
7.13409260e-02 5.03264248e-01 -6.08158946e-01 7.20447779e-01
8.13909113e-01 7.75367022e-01 -4.32638407e-01 8.58357430e-01
-2.12462246e-01 -1.13807507e-02 -7.22557306e-02 7.30376780e-01
-6.75742328e-02 -4.57364112e-01 4.53641713e-01 6.91274643e-01
3.51629019e-01 3.14567797e-02 -4.04315069e-02 1.22004104e+00
-5.46689332e-01 1.10292365e-03 -6.91972315e-01 1.31969333e-01
5.35606444e-01 1.36526918e+00 -3.57347906e-01 1.62610449e-02
-5.25662184e-01 1.31068802e+00 1.12747170e-01 3.96697998e-01
-9.57472861e-01 -4.05431598e-01 1.08770823e+00 1.56185344e-01
3.44489723e-01 2.33808964e-01 -2.51362294e-01 -1.21378887e+00
4.78370786e-01 -1.09303665e+00 6.90086335e-02 -6.51531100e-01
-1.40635490e+00 5.91634929e-01 -4.38291132e-01 -1.57028377e+00
1.37770995e-01 -4.94421005e-01 -6.22685194e-01 1.24914324e+00
-1.83058488e+00 -1.41464603e+00 -6.55562758e-01 9.77846861e-01
3.80844146e-01 -2.16063917e-01 5.80355167e-01 4.40345645e-01
-7.64294624e-01 1.10362160e+00 -9.71290693e-02 3.24794322e-01
8.63215685e-01 -6.50406301e-01 3.04514468e-01 1.13350105e+00
-1.22188605e-01 7.01533556e-01 2.41026416e-01 -6.07070804e-01
-1.70975554e+00 -1.26781321e+00 7.38865674e-01 -2.25595757e-01
1.26028717e-01 -3.58205110e-01 -8.08795393e-01 4.94652718e-01
8.63255709e-02 3.74107748e-01 2.51305073e-01 -5.56084216e-01
-5.93800008e-01 -5.08171976e-01 -1.67203486e+00 5.37069499e-01
1.19926631e+00 -8.35333288e-01 -3.35313767e-01 3.01474899e-01
7.18935490e-01 -2.02148244e-01 -6.65205717e-01 3.62968862e-01
5.25999665e-01 -1.04715693e+00 1.15680861e+00 -2.75889456e-01
3.04916382e-01 -6.03766322e-01 -1.87228158e-01 -8.89767468e-01
-1.63027376e-03 -9.13416147e-01 1.73081160e-02 1.74966860e+00
-6.84915632e-02 -7.08118260e-01 7.72817314e-01 5.33998370e-01
5.47327809e-02 -4.83951390e-01 -9.28226650e-01 -6.90943122e-01
-2.95851529e-01 -7.20039681e-02 1.06568170e+00 1.04796743e+00
-3.59424561e-01 -2.05584258e-01 -4.23798114e-01 4.76615965e-01
9.21546400e-01 3.05309266e-01 7.85662055e-01 -1.11497223e+00
1.63512323e-02 -2.97912121e-01 -4.33216214e-01 -8.76218975e-01
4.51340020e-01 -7.40157306e-01 6.67205229e-02 -1.08333278e+00
1.26627281e-01 -3.08280140e-01 6.62270561e-02 5.15353203e-01
-1.71603441e-01 5.03174305e-01 1.11236691e-01 3.23527515e-01
1.15160882e-01 6.13253772e-01 1.40646052e+00 -2.74250180e-01
7.00589567e-02 -1.54220417e-01 -9.14500713e-01 5.55790126e-01
4.83569860e-01 -2.24373147e-01 -4.39063579e-01 -5.69428921e-01
-3.01360577e-01 -3.26819569e-02 5.85613608e-01 -9.03299510e-01
2.99646467e-01 -1.53537154e-01 6.74898624e-01 -2.25769922e-01
4.12389785e-01 -1.23012424e+00 1.17921151e-01 2.48323351e-01
1.70441151e-01 -3.94098163e-02 1.15713276e-01 4.95717138e-01
-4.46222216e-01 3.76382470e-02 1.10144782e+00 9.47268605e-02
-5.64895272e-01 7.42284417e-01 4.18391496e-01 -2.09814921e-01
1.14497721e+00 -6.95442677e-01 -1.92982048e-01 -2.19152898e-01
-4.08337086e-01 9.47243795e-02 8.83989096e-01 4.83703047e-01
1.03621614e+00 -1.51231062e+00 -7.42584348e-01 1.10705757e+00
-9.22225975e-03 -9.22237560e-02 3.75902772e-01 7.44889557e-01
-5.42762995e-01 8.39153528e-02 -4.41647649e-01 -5.15113294e-01
-1.44450963e+00 7.20785022e-01 5.35941303e-01 2.80711442e-01
-5.98827302e-01 1.01880133e+00 3.80240947e-01 -3.06620091e-01
5.12208343e-01 -3.36376764e-02 3.75618301e-02 -1.48799559e-02
8.24420094e-01 1.08528703e-01 1.38886154e-01 -9.54223990e-01
-4.41843003e-01 8.97009492e-01 -2.19085842e-01 -3.31177413e-02
1.13418603e+00 -4.30683672e-01 -5.60447633e-01 -4.68415827e-01
1.36769998e+00 2.03131616e-01 -1.51984799e+00 -2.79129207e-01
-5.02615809e-01 -1.13102126e+00 -3.12892050e-02 -4.50791121e-01
-1.59895515e+00 7.01929092e-01 7.26742566e-01 -2.90715843e-01
1.56251264e+00 -2.15035647e-01 9.20421422e-01 -1.34702057e-01
2.16867030e-01 -8.80470574e-01 -2.00819075e-02 -3.73174958e-02
1.06940627e+00 -1.14424860e+00 -7.26243332e-02 -6.86202645e-01
-3.55143249e-01 9.42838788e-01 6.94016516e-01 -1.18428105e-02
7.12612092e-01 8.68878588e-02 1.19627252e-01 -7.03032538e-02
-1.72631085e-01 7.26620033e-02 3.38415384e-01 7.63445139e-01
-9.15694460e-02 -2.25422099e-01 -2.86954232e-02 6.18543088e-01
-1.80835918e-01 -2.52163839e-02 1.34469926e-01 7.69786179e-01
-1.71440318e-01 -1.14350760e+00 -8.13995838e-01 1.88271567e-01
-2.86716580e-01 -5.28519116e-02 -3.62541884e-01 5.26342034e-01
4.96733963e-01 1.03574967e+00 -3.17800231e-02 -4.36116695e-01
3.00238162e-01 -1.22779705e-01 4.34655547e-01 -3.29563767e-01
-5.18369675e-01 -1.53491601e-01 -4.49103385e-01 -8.31765175e-01
-3.36694837e-01 -5.32948494e-01 -1.05977106e+00 -7.32757568e-01
-2.12449118e-01 -1.57439522e-02 5.81482530e-01 6.55527711e-01
4.34707940e-01 5.12339696e-02 1.13487899e+00 -9.60308135e-01
-6.80444479e-01 -6.61892474e-01 -7.93259025e-01 4.04339761e-01
7.14256883e-01 -6.62667572e-01 -4.64457422e-01 1.17598437e-01] | [12.85180950164795, 0.17247916758060455] |
070e1ea8-3487-431d-ac7e-64b7599d7250 | compacting-picking-and-growing-for | 1910.06562 | null | https://arxiv.org/abs/1910.06562v3 | https://arxiv.org/pdf/1910.06562v3.pdf | Compacting, Picking and Growing for Unforgetting Continual Learning | Continual lifelong learning is essential to many applications. In this paper, we propose a simple but effective approach to continual deep learning. Our approach leverages the principles of deep model compression, critical weights selection, and progressive networks expansion. By enforcing their integration in an iterative manner, we introduce an incremental learning method that is scalable to the number of sequential tasks in a continual learning process. Our approach is easy to implement and owns several favorable characteristics. First, it can avoid forgetting (i.e., learn new tasks while remembering all previous tasks). Second, it allows model expansion but can maintain the model compactness when handling sequential tasks. Besides, through our compaction and selection/expansion mechanism, we show that the knowledge accumulated through learning previous tasks is helpful to build a better model for the new tasks compared to training the models independently with tasks. Experimental results show that our approach can incrementally learn a deep model tackling multiple tasks without forgetting, while the model compactness is maintained with the performance more satisfiable than individual task training. | ['Chu-Song Chen', 'Yi-Ming Chan', 'Chien-Hung Chen', 'Cheng-En Wu', 'Cheng-Hao Tu', 'Steven C. Y. Hung'] | 2019-10-15 | compacting-picking-and-growing-for-1 | http://papers.nips.cc/paper/9518-compacting-picking-and-growing-for-unforgetting-continual-learning | http://papers.nips.cc/paper/9518-compacting-picking-and-growing-for-unforgetting-continual-learning.pdf | neurips-2019-12 | ['age-and-gender-classification'] | ['computer-vision'] | [ 2.92096138e-01 -1.33111522e-01 -2.60597646e-01 -2.84450561e-01
-2.90551603e-01 -1.60551280e-01 2.35894665e-01 1.46743298e-01
-7.87927628e-01 9.17993009e-01 -7.65623972e-02 -8.09279606e-02
-4.13647324e-01 -6.71009600e-01 -7.96370804e-01 -7.30560184e-01
-1.03154639e-02 7.15126872e-01 6.69153750e-01 4.62783985e-02
2.66216218e-01 4.23911810e-01 -1.75101066e+00 1.45991832e-01
1.10168159e+00 8.70907664e-01 9.80398178e-01 5.06859481e-01
-1.79272830e-01 1.15791976e+00 -3.93465400e-01 -3.11421603e-01
6.68280572e-02 -1.45377636e-01 -1.05640721e+00 4.93118912e-02
3.56664687e-01 -5.00350714e-01 -3.11679631e-01 6.50057256e-01
4.04578120e-01 3.88229340e-01 4.19870824e-01 -1.05709028e+00
-5.36230266e-01 6.49495840e-01 -3.75805855e-01 2.85172313e-01
-3.15010816e-01 3.22751477e-02 7.50050247e-01 -1.02584863e+00
2.69922972e-01 8.73876691e-01 8.53680253e-01 7.87451625e-01
-1.06493795e+00 -7.90783703e-01 5.73252738e-01 3.20786208e-01
-1.11032701e+00 -7.06270039e-01 6.93872809e-01 -1.56423271e-01
1.12723112e+00 9.65496302e-02 1.03616357e+00 7.84166217e-01
1.42609686e-01 1.01379836e+00 7.07991540e-01 -6.17541909e-01
2.84116685e-01 7.82710463e-02 4.14492697e-01 8.24531853e-01
4.58425939e-01 -1.15097620e-01 -9.34836328e-01 8.27944353e-02
6.75769448e-01 6.36932433e-01 -1.22310050e-01 -3.65161300e-01
-9.04783964e-01 5.17864287e-01 1.16435438e-01 3.42322260e-01
-3.42731357e-01 4.39770907e-01 5.04129410e-01 5.57777703e-01
5.61760843e-01 3.38390112e-01 -7.06986845e-01 -2.20629737e-01
-1.16662025e+00 1.23155639e-01 5.05467534e-01 9.93205667e-01
9.62807000e-01 2.61186272e-01 4.17293645e-02 1.01623380e+00
6.68802559e-02 1.49106860e-01 1.15494096e+00 -1.00731301e+00
3.13259184e-01 4.31631625e-01 -1.87350035e-01 -5.68999171e-01
-3.65075409e-01 -7.16890335e-01 -9.95838821e-01 2.03567687e-02
-3.99681227e-03 -1.54045418e-01 -9.45373654e-01 1.82251120e+00
1.34387210e-01 2.43473917e-01 3.69402431e-02 3.80783807e-03
4.34297651e-01 3.73239517e-01 1.11830696e-01 -6.46963656e-01
9.48305547e-01 -1.39830959e+00 -9.36809599e-01 -5.48989356e-01
6.86958015e-01 -4.97935951e-01 1.25374162e+00 6.66813791e-01
-1.45080507e+00 -8.80404294e-01 -1.09738147e+00 -1.01133257e-01
-1.81291625e-01 -1.19375572e-01 6.97079778e-01 3.67387325e-01
-1.14075518e+00 7.15349793e-01 -1.01990557e+00 -8.07050318e-02
6.32188857e-01 3.73579949e-01 3.38974893e-02 -1.18319690e-01
-9.74394977e-01 9.50122237e-01 8.81981313e-01 -3.57158363e-01
-1.12855184e+00 -8.29569280e-01 -5.59643567e-01 3.71299297e-01
5.07400393e-01 -6.87256277e-01 1.62430227e+00 -7.91919172e-01
-1.18243170e+00 4.33699995e-01 -5.72322309e-01 -8.28777373e-01
4.26136792e-01 -5.74824512e-01 -1.50330260e-01 7.07471417e-03
-3.81246001e-01 7.09212184e-01 9.95402038e-01 -1.12726283e+00
-9.90440786e-01 -3.25905293e-01 3.05930078e-02 4.50740218e-01
-1.37156785e+00 -4.90385920e-01 -5.42207956e-01 -7.49365389e-01
2.62404084e-01 -7.93550134e-01 -1.37634099e-01 1.40185416e-01
9.41397026e-02 -3.05481821e-01 9.23815370e-01 -4.39486861e-01
1.79214323e+00 -2.27031255e+00 2.60498561e-02 -1.09659374e-01
7.48825073e-01 4.49812919e-01 7.67919607e-03 2.12937295e-01
2.26088524e-01 5.54074496e-02 -2.29288310e-01 -8.68525743e-01
-2.97817737e-01 4.50673908e-01 -2.74662793e-01 -2.29473561e-01
-2.15206295e-01 1.13851345e+00 -8.52996349e-01 -6.26599669e-01
-1.74190789e-01 2.28530869e-01 -4.65288311e-01 1.26343355e-01
-1.70158595e-01 -1.09967105e-01 -9.63555425e-02 4.52996194e-01
4.97851551e-01 -3.95064026e-01 2.47670770e-01 -2.82973852e-02
1.81398705e-01 2.93360949e-01 -1.06838632e+00 1.74558961e+00
-6.95075452e-01 4.71967399e-01 -2.21880361e-01 -9.75013196e-01
8.94385755e-01 2.80853093e-01 2.92717874e-01 -7.64696360e-01
-4.17058945e-01 2.45976716e-01 -1.92928329e-01 -3.38085711e-01
4.05934244e-01 -1.99299335e-01 2.79146552e-01 8.72152507e-01
3.91079098e-01 4.48633507e-02 3.05538446e-01 2.82503337e-01
8.36048186e-01 -2.05121085e-01 2.13773102e-01 -3.06652728e-02
1.54760852e-01 -1.82176888e-01 7.17480540e-01 8.74551773e-01
-1.26175225e-01 8.97088498e-02 1.27433729e-03 -8.88558805e-01
-9.66340840e-01 -8.26426148e-01 1.74517155e-01 1.54865408e+00
-9.78293940e-02 -4.27880287e-01 -4.66899574e-01 -7.39669740e-01
1.46424789e-02 6.69027925e-01 -7.11991549e-01 -5.94670236e-01
-8.75679493e-01 -6.91749752e-01 2.51641721e-01 7.68753827e-01
8.24523866e-01 -1.16248214e+00 -9.58533287e-01 3.46888781e-01
1.09242592e-02 -6.08515680e-01 -6.62373483e-01 7.48725355e-01
-1.74344027e+00 -9.86343861e-01 -6.89572930e-01 -1.26524150e+00
7.32967436e-01 7.35507131e-01 1.15097976e+00 4.18309808e-01
-8.22699964e-02 3.78992185e-02 -1.42666951e-01 -4.87338632e-01
-3.76950279e-02 4.57159013e-01 2.68394023e-01 -3.60041410e-01
3.33743811e-01 -8.46455991e-01 -4.47197080e-01 3.42013501e-02
-1.02250922e+00 3.97237659e-01 7.47717202e-01 9.70356226e-01
7.44628131e-01 5.13548553e-01 8.57516766e-01 -1.34544063e+00
8.95117640e-01 -2.86452532e-01 -1.85802400e-01 5.70286393e-01
-1.34528363e+00 7.41071180e-02 7.94749200e-01 -7.98454940e-01
-1.29465294e+00 5.15924394e-02 5.15571190e-03 -4.70111579e-01
2.06321687e-01 5.12209892e-01 2.51678348e-01 -1.02207139e-01
7.06667185e-01 7.80488193e-01 7.64823183e-02 -7.55643129e-01
2.69377083e-01 2.86324739e-01 2.96423256e-01 -4.55248445e-01
6.02933168e-01 3.32782209e-01 -3.38766903e-01 -3.82952392e-01
-8.91598582e-01 -1.98344320e-01 -7.68732667e-01 -1.54284015e-01
1.91313177e-01 -8.46743345e-01 -5.10851145e-01 7.74472177e-01
-9.87828255e-01 -7.31077135e-01 -6.22375786e-01 3.14916819e-01
-3.47438335e-01 3.07941765e-01 -6.83290958e-01 -8.41753244e-01
-6.09219491e-01 -8.66865635e-01 4.15170759e-01 9.28048715e-02
-1.85130894e-01 -1.33084035e+00 -5.87724634e-02 -3.91882546e-02
7.31059134e-01 -4.64391053e-01 1.13908601e+00 -6.68345451e-01
-3.12189281e-01 -9.75635424e-02 1.11328639e-01 5.65189362e-01
2.03909159e-01 -4.88971233e-01 -8.86960208e-01 -6.51879609e-01
3.06298345e-01 -8.46261799e-01 1.26146185e+00 1.29974142e-01
1.52213061e+00 -5.89178324e-01 -3.92045021e-01 5.99875391e-01
1.29233289e+00 4.81679201e-01 4.32326853e-01 4.27372932e-01
6.76536798e-01 2.99015194e-01 4.61332440e-01 4.41967696e-01
3.35269034e-01 1.87589884e-01 2.00990930e-01 8.01275820e-02
-3.48051071e-01 -2.60517806e-01 1.27704009e-01 1.31806135e+00
-5.18890955e-02 2.19825476e-01 -9.08459902e-01 6.45062983e-01
-1.94440687e+00 -8.02279830e-01 3.97374600e-01 2.28377938e+00
1.41896868e+00 5.53550243e-01 -1.54635057e-01 2.98750132e-01
3.42271507e-01 5.95243424e-02 -9.91472721e-01 -2.20198661e-01
1.15093447e-01 2.20414847e-01 1.59470350e-01 5.58393836e-01
-6.82007194e-01 1.09829402e+00 6.91901493e+00 1.05092311e+00
-9.63515043e-01 4.34928358e-01 7.19483972e-01 -5.89441299e-01
-1.64425075e-01 -3.55478764e-01 -1.21459293e+00 3.91269833e-01
8.20992470e-01 -4.57917035e-01 5.21391988e-01 8.99414062e-01
-1.98646620e-01 -6.53838515e-02 -1.05462098e+00 8.19528222e-01
1.79463863e-01 -1.30805099e+00 3.11181247e-01 -2.64398754e-01
1.01565623e+00 -2.13798612e-01 2.23836675e-01 6.72240078e-01
3.63809764e-01 -7.48076916e-01 6.89617336e-01 7.31684923e-01
7.67610490e-01 -9.80295658e-01 4.98739541e-01 9.80986357e-01
-1.11826241e+00 -4.93848771e-01 -5.10275364e-01 -5.86837679e-02
-6.80635571e-02 7.32497096e-01 -7.30071247e-01 4.78902981e-02
7.03492463e-01 6.48630559e-01 -7.54815459e-01 1.20549262e+00
-7.97073171e-02 5.63161612e-01 -1.43715903e-01 1.03335343e-01
-3.60651985e-02 2.89241016e-01 3.61263938e-02 1.03253758e+00
3.91224891e-01 -3.06514084e-01 2.17338622e-01 2.44424403e-01
-3.79058987e-01 -1.69053003e-02 -5.46052754e-01 2.56457865e-01
9.71993923e-01 7.48895824e-01 -4.15884256e-01 -5.96032441e-01
-1.60861850e-01 1.05043149e+00 8.18310142e-01 3.50474298e-01
-4.97679174e-01 -4.52624857e-01 1.22146897e-01 1.43417642e-01
3.33663791e-01 -3.82735878e-01 -5.09328127e-01 -8.18640053e-01
2.51422793e-01 -6.96309686e-01 4.70842063e-01 -5.73400855e-01
-1.02057576e+00 6.64561570e-01 -6.34123757e-03 -9.26136374e-01
-2.16913164e-01 -1.02636486e-01 -6.61917627e-01 6.49888694e-01
-1.89243889e+00 -8.30273569e-01 -4.68374908e-01 7.17974901e-01
1.18054283e+00 -4.45325464e-01 9.64362979e-01 1.89195320e-01
-4.41276163e-01 8.94626915e-01 2.88224936e-01 -4.38794374e-01
6.41665816e-01 -1.04032731e+00 3.29394609e-01 6.20971322e-01
6.74501285e-02 8.57188821e-01 1.84721068e-01 -7.78592110e-01
-1.14836478e+00 -1.20515025e+00 1.23618102e+00 8.41908231e-02
4.77008611e-01 -3.35904598e-01 -1.16339540e+00 9.46837306e-01
1.35676181e-02 -3.29292327e-01 7.79453158e-01 3.22434753e-01
-2.44352192e-01 -6.09214008e-01 -9.20109868e-01 4.58266318e-01
1.21171212e+00 -3.39250475e-01 -6.79518580e-01 5.57655811e-01
1.09600115e+00 -1.61276802e-01 -4.80231494e-01 3.98746550e-01
6.11682177e-01 -7.48804569e-01 9.29843366e-01 -5.15195549e-01
2.43666455e-01 2.61981696e-01 2.14487180e-01 -1.27026761e+00
-6.24417603e-01 -4.51355338e-01 -1.14598215e+00 8.86880696e-01
5.22899449e-01 -7.33208358e-01 9.01311755e-01 4.83939946e-01
-1.94186106e-01 -1.22652519e+00 -8.19994748e-01 -1.10024333e+00
5.60513185e-03 -4.04670745e-01 6.33737683e-01 8.54472041e-01
-2.35558614e-01 2.86102682e-01 -7.09332287e-01 -3.63087922e-01
6.14002168e-01 -9.16178450e-02 4.29758817e-01 -1.65278137e+00
-5.52403033e-01 -3.27521116e-01 3.80074501e-01 -1.40163684e+00
-1.13015860e-01 -8.67392600e-01 3.82929947e-03 -1.56320345e+00
4.47211355e-01 -7.85311341e-01 -8.10618401e-01 8.42104256e-01
-2.96847135e-01 3.73659283e-02 2.67667800e-01 8.15177977e-01
-1.10193861e+00 5.36766529e-01 1.39723885e+00 -7.96252489e-02
-3.95549744e-01 1.39982089e-01 -8.94201279e-01 8.42986286e-01
1.00208914e+00 -6.86718404e-01 -9.42879617e-01 -8.67765069e-01
2.22978070e-01 -3.98795485e-01 -1.03129826e-01 -1.37837577e+00
7.52373397e-01 -1.16610657e-02 5.83385050e-01 -4.90157753e-01
3.66220295e-01 -8.81965876e-01 -8.27667713e-02 8.62439156e-01
-5.93199134e-01 2.88692683e-01 2.47264177e-01 7.34080315e-01
-2.45204829e-02 -7.22586930e-01 7.70722866e-01 -4.22662258e-01
-7.40779817e-01 4.39497024e-01 -2.58822054e-01 5.49386218e-02
1.09470034e+00 -2.52155840e-01 -1.15141578e-01 -1.88116543e-02
-9.65052843e-01 4.85029131e-01 1.76911667e-01 2.70692080e-01
1.10928214e+00 -1.34269023e+00 -5.28869510e-01 3.70815396e-01
-7.41878152e-02 3.76142740e-01 2.16360345e-01 5.36988080e-01
-2.73297429e-01 3.47860873e-01 -9.21956226e-02 -3.00719976e-01
-1.47458827e+00 6.26424134e-01 -4.85639572e-02 -7.37297475e-01
-5.23957372e-01 1.15574110e+00 -3.27944942e-02 -1.83127254e-01
9.26025271e-01 -1.37950733e-01 -2.46714219e-01 2.82935590e-01
8.28302443e-01 5.11019826e-01 2.11282521e-01 3.20742667e-01
1.42601162e-01 2.41418064e-01 -8.49662006e-01 6.94018155e-02
1.64469075e+00 -1.38870865e-01 -8.18214342e-02 9.12968397e-01
9.35313821e-01 -4.26095098e-01 -1.59937000e+00 -7.77592719e-01
-7.30501339e-02 -4.06791717e-01 1.52980894e-01 -8.75596702e-01
-1.13912094e+00 9.69027400e-01 5.43653429e-01 1.46266401e-01
1.32730150e+00 -3.56369227e-01 1.14899302e+00 9.86458063e-01
4.60407168e-01 -1.41511893e+00 7.37746418e-01 8.30542266e-01
7.11909533e-01 -8.27696264e-01 4.27002944e-02 4.76048933e-03
-4.98574615e-01 1.15249717e+00 9.06753242e-01 2.14134291e-01
7.71053135e-01 2.37297311e-01 -3.35839212e-01 -6.96709231e-02
-1.14493370e+00 1.12227455e-01 -3.82302701e-02 4.26994562e-01
1.93903640e-01 -3.04408014e-01 -1.70955777e-01 6.60732448e-01
-2.61469595e-02 3.48436534e-01 8.92167911e-02 1.26903534e+00
-1.14284503e+00 -1.06707013e+00 1.29830331e-01 7.21232951e-01
-1.55228883e-01 -5.81061244e-01 3.89021486e-02 5.48321366e-01
3.37331742e-01 5.09594798e-01 1.24945492e-01 -3.24463308e-01
1.66574776e-01 6.00632966e-01 4.52473879e-01 -8.03351402e-01
-4.12899077e-01 -8.76644328e-02 -3.18048388e-01 -7.93356448e-02
-1.51986718e-01 -3.92109871e-01 -1.16177022e+00 -3.31905633e-01
-3.55514020e-01 1.10438608e-01 4.76324350e-01 9.00378108e-01
3.56457710e-01 7.29865372e-01 6.19925141e-01 -5.01101613e-01
-6.95066214e-01 -1.02847040e+00 -5.61379790e-01 1.25608802e-01
3.25595796e-01 -5.94255209e-01 -2.45464802e-01 3.18988681e-01] | [9.789528846740723, 3.4619245529174805] |
d48cb87e-6eb7-4490-908a-40bbdda901a9 | instance-neural-radiance-field | 2304.04395 | null | https://arxiv.org/abs/2304.04395v1 | https://arxiv.org/pdf/2304.04395v1.pdf | Instance Neural Radiance Field | This paper presents one of the first learning-based NeRF 3D instance segmentation pipelines, dubbed as Instance Neural Radiance Field, or Instance NeRF. Taking a NeRF pretrained from multi-view RGB images as input, Instance NeRF can learn 3D instance segmentation of a given scene, represented as an instance field component of the NeRF model. To this end, we adopt a 3D proposal-based mask prediction network on the sampled volumetric features from NeRF, which generates discrete 3D instance masks. The coarse 3D mask prediction is then projected to image space to match 2D segmentation masks from different views generated by existing panoptic segmentation models, which are used to supervise the training of the instance field. Notably, beyond generating consistent 2D segmentation maps from novel views, Instance NeRF can query instance information at any 3D point, which greatly enhances NeRF object segmentation and manipulation. Our method is also one of the first to achieve such results without ground-truth instance information during inference. Experimented on synthetic and real-world NeRF datasets with complex indoor scenes, Instance NeRF surpasses previous NeRF segmentation works and competitive 2D segmentation methods in segmentation performance on unseen views. See the demo video at https://youtu.be/wW9Bme73coI. | ['Chi-Keung Tang', 'Yu-Wing Tai', 'Yichen Liu', 'Junkai Huang', 'Benran Hu'] | 2023-04-10 | null | null | null | null | ['3d-instance-segmentation-1', 'panoptic-segmentation'] | ['computer-vision', 'computer-vision'] | [ 3.66546512e-01 4.62245435e-01 7.31446818e-02 -5.95058262e-01
-8.63909721e-01 -9.48792875e-01 6.03050113e-01 -1.92580357e-01
-1.83759630e-01 5.41206300e-01 -3.91537458e-01 -7.35562593e-02
-2.24342674e-01 -1.22527683e+00 -1.25764728e+00 -3.95849913e-01
2.00863376e-01 8.97876978e-01 4.31221426e-01 -9.77087859e-03
2.36367092e-01 8.85841846e-01 -1.67421663e+00 1.16897859e-01
8.40718389e-01 1.20343661e+00 5.62239647e-01 8.25180888e-01
-4.67848539e-01 1.07380614e-01 -2.61955768e-01 -2.22630098e-01
8.17696869e-01 -4.20400761e-02 -8.74492824e-01 4.88865644e-01
8.55790198e-01 -3.97874266e-01 3.94066311e-02 9.48117852e-01
6.25109579e-03 3.14348668e-01 7.36587882e-01 -1.03002298e+00
-3.42864305e-01 1.96555674e-01 -4.23936218e-01 -1.62457898e-01
3.62694085e-01 3.33041549e-01 7.32301950e-01 -7.57613599e-01
8.11775565e-01 1.12022817e+00 6.65156543e-01 4.80885565e-01
-1.21609402e+00 -3.20097685e-01 4.12182599e-01 -3.40717673e-01
-1.12704122e+00 4.24062870e-02 7.95800507e-01 -4.79937106e-01
9.56179678e-01 5.12652755e-01 1.03328228e+00 6.74755812e-01
-4.29396611e-03 8.11418593e-01 1.51010573e+00 1.45655125e-01
4.91768032e-01 3.88460793e-02 1.34294359e-02 7.73109913e-01
-2.08332717e-01 2.04217404e-01 -1.86057061e-01 1.06848881e-01
1.10570908e+00 2.13947684e-01 -4.56983387e-01 -2.40467936e-01
-1.33643568e+00 5.73013127e-01 8.71454418e-01 -1.74578086e-01
-5.22480547e-01 2.40703732e-01 -2.09537819e-01 -5.19168116e-02
7.84158647e-01 4.01439250e-01 -9.22550917e-01 3.08958203e-01
-1.13378811e+00 2.49625236e-01 5.92299819e-01 8.39830816e-01
1.42641830e+00 -2.48897031e-01 1.12415880e-01 5.04596233e-01
4.45745081e-01 7.49337435e-01 -7.07615539e-02 -1.53365779e+00
3.59449953e-01 9.70839858e-01 8.04196075e-02 -5.03250182e-01
-5.33708632e-01 -1.76948726e-01 -5.56489527e-01 4.82562929e-01
3.40180099e-01 9.71675366e-02 -1.55157793e+00 1.15251195e+00
9.45007086e-01 2.59570301e-01 7.03661069e-02 1.01427484e+00
1.01283908e+00 9.86297369e-01 -2.59335637e-01 2.66306281e-01
1.11422741e+00 -8.35780501e-01 -6.02967255e-02 -3.83484006e-01
2.67419189e-01 -4.39571261e-01 9.99711275e-01 5.20711064e-01
-9.27423596e-01 -6.48593843e-01 -5.58567643e-01 -6.91831186e-02
-7.26565182e-01 -1.24156259e-01 7.43277252e-01 3.32597017e-01
-9.95743990e-01 6.35057092e-01 -8.10716271e-01 -3.75034213e-01
7.86068082e-01 1.93365067e-01 -4.53318954e-01 -1.95612550e-01
-9.75908220e-01 6.11091614e-01 7.22623646e-01 3.88298333e-01
-1.24486971e+00 -9.67887163e-01 -9.25195992e-01 -2.97080696e-01
5.51011026e-01 -8.63968909e-01 9.73278999e-01 -7.82343090e-01
-1.27905643e+00 1.01230073e+00 1.58773899e-01 -4.25493360e-01
5.28546631e-01 -2.33955666e-01 1.00237109e-01 2.70598769e-01
2.21940666e-01 1.35229647e+00 7.58848429e-01 -1.83035326e+00
-7.27459013e-01 -4.62946355e-01 4.35037136e-01 3.59161049e-01
6.05556011e-01 -7.67846763e-01 -4.15874511e-01 -7.90192187e-02
3.46850038e-01 -5.64420998e-01 -6.39211059e-01 3.39148827e-02
-8.53030145e-01 -1.94283605e-01 8.57882261e-01 -5.31839252e-01
4.18135166e-01 -1.82330596e+00 2.17777804e-01 3.55365336e-01
9.40450653e-02 -1.79607928e-01 3.66388522e-02 4.26356047e-02
2.18036950e-01 3.05323780e-01 -9.18664575e-01 -2.21965611e-01
9.73236635e-02 5.01310468e-01 -1.27364635e-01 5.22014081e-01
1.70366541e-01 9.49665606e-01 -8.06019545e-01 -2.96894968e-01
7.69026101e-01 5.45656562e-01 -4.55358475e-01 4.00320768e-01
-9.60393012e-01 8.66903961e-01 -6.13123000e-01 7.10169971e-01
1.03886819e+00 -5.90761341e-02 -4.18558121e-01 -4.91862625e-01
-4.24973339e-01 -2.11803138e-01 -1.24453127e+00 2.15237451e+00
-5.09873390e-01 2.54753381e-01 8.98824334e-02 -9.86014843e-01
8.90356600e-01 -2.36635450e-02 4.78911728e-01 -5.32544553e-01
-3.24853137e-02 1.93128020e-01 -7.03202903e-01 -4.04400706e-01
3.62152755e-01 -5.73859960e-02 -1.90532669e-01 9.47626010e-02
3.41307610e-01 -1.16112971e+00 1.38280019e-01 1.19819723e-01
5.36986709e-01 7.38458633e-01 9.28454772e-02 -1.14367321e-01
4.65609282e-01 3.66155088e-01 3.51487041e-01 7.81715155e-01
1.97411478e-01 9.86410677e-01 2.34805226e-01 -5.19909263e-01
-9.72360671e-01 -1.33288753e+00 -6.15651965e-01 4.32962328e-01
4.42676514e-01 -2.03538790e-01 -1.02649128e+00 -9.91346836e-01
2.08158009e-02 7.66675770e-01 -6.83577895e-01 6.56494975e-01
-4.09934282e-01 -5.90015471e-01 1.16195269e-01 3.30389231e-01
8.61653209e-01 -9.01725888e-01 -1.07581687e+00 4.86508347e-02
-3.85549245e-03 -1.18221998e+00 -1.12045541e-01 5.45539796e-01
-8.37346375e-01 -1.34994388e+00 -6.64603710e-01 -4.14172590e-01
6.59119785e-01 -2.71683019e-02 1.19323647e+00 -1.84167892e-01
-4.69522953e-01 1.04547215e+00 -1.57578945e-01 -3.39308977e-01
-6.62475675e-02 1.03963129e-01 -4.92072761e-01 -1.11038268e-01
3.23583223e-02 -5.27382910e-01 -7.99411595e-01 1.49433061e-01
-1.04307449e+00 4.43440616e-01 3.81815106e-01 3.49862725e-01
1.43089736e+00 1.82618767e-01 -8.11789557e-02 -8.80049586e-01
-3.46084207e-01 -5.12997806e-01 -1.07543755e+00 1.98680267e-01
-2.41578758e-01 -1.96911380e-01 5.64672709e-01 1.13985971e-01
-1.20997620e+00 5.21863401e-01 -3.37722450e-01 -4.51060325e-01
-9.51869071e-01 1.75140768e-01 -2.73793846e-01 1.38773978e-01
5.55843532e-01 2.41540655e-01 -4.73862022e-01 -4.64003444e-01
6.49129987e-01 2.98476160e-01 5.16184092e-01 -5.17724454e-01
1.01051509e+00 7.20503807e-01 3.16116214e-02 -7.38769650e-01
-1.12619579e+00 -4.40187395e-01 -1.10139513e+00 -5.39284885e-01
1.54612041e+00 -9.30250227e-01 -2.93238342e-01 2.84487039e-01
-1.16832960e+00 -9.23159122e-01 -5.25106907e-01 2.27743283e-01
-7.95660496e-01 -8.48283395e-02 -3.36052865e-01 -7.76012778e-01
-1.31577224e-01 -1.23723352e+00 1.81043351e+00 4.79554921e-01
3.75880450e-01 -1.08994448e+00 -3.22161317e-02 5.83428323e-01
-3.30533311e-02 8.67718220e-01 5.37286103e-01 -2.99990654e-01
-1.29804659e+00 1.07657321e-01 -8.86954367e-02 4.92763847e-01
-4.79590483e-02 1.62648216e-01 -1.17317033e+00 3.28159630e-01
2.13887338e-02 -2.62905300e-01 9.25103426e-01 6.29573166e-01
1.29115057e+00 3.27960141e-02 -4.17176515e-01 9.96045411e-01
1.85155976e+00 7.67132416e-02 4.73951548e-01 2.96132654e-01
1.07314742e+00 7.42974520e-01 6.20444417e-01 1.21988937e-01
5.96632183e-01 4.18604583e-01 9.99291122e-01 -4.11880702e-01
2.14667246e-03 -2.79924184e-01 6.55582920e-02 1.85922831e-01
-1.18556425e-01 -3.17318678e-01 -9.72453237e-01 2.95716524e-01
-1.57317400e+00 -5.72179019e-01 -3.90994489e-01 1.86071098e+00
5.02289414e-01 7.98397139e-02 -1.20494179e-01 -1.89009085e-01
4.80348140e-01 3.43685359e-01 -8.69941711e-01 -2.00824603e-01
-3.95655483e-02 3.23990136e-01 7.07328916e-01 8.37960243e-01
-1.25189269e+00 1.03332067e+00 4.19408751e+00 5.46681643e-01
-8.66055429e-01 1.31303057e-01 9.98184919e-01 6.04764074e-02
-6.89773023e-01 1.29115120e-01 -8.21029842e-01 1.15440965e-01
6.59930825e-01 4.49493349e-01 5.90788662e-01 8.88769090e-01
1.82260886e-01 -5.59681892e-01 -9.95572031e-01 9.68588769e-01
-7.29233548e-02 -1.45688164e+00 7.32586160e-02 1.78833917e-01
8.12351286e-01 4.28045779e-01 -3.07783186e-01 5.42284362e-02
2.92826086e-01 -1.09129155e+00 7.82749236e-01 9.25534427e-01
7.04918325e-01 -6.52703881e-01 3.55434924e-01 5.77275872e-01
-1.24764979e+00 2.38173485e-01 -4.35738295e-01 4.17375743e-01
4.78315860e-01 8.61228704e-01 -6.74385786e-01 8.08199584e-01
9.55279708e-01 7.54272699e-01 -4.76222634e-01 8.89849305e-01
-3.82692277e-01 3.57938975e-01 -8.29692662e-01 4.94988918e-01
5.48297405e-01 -5.74025989e-01 4.97985840e-01 1.00026643e+00
2.28866547e-01 1.98243931e-01 4.13800806e-01 1.32036424e+00
-9.41281691e-02 -3.35339099e-01 -6.02047026e-01 3.84448469e-01
8.84827971e-02 1.66354609e+00 -1.32117081e+00 -3.76475304e-01
-1.43722534e-01 1.28928828e+00 1.15470298e-01 4.99809474e-01
-1.13057828e+00 -3.93247493e-02 3.71938705e-01 1.84808046e-01
4.57366496e-01 -1.70971632e-01 -1.01959541e-01 -1.14710641e+00
-5.44396415e-02 -1.95306271e-01 1.03974164e-01 -1.14305377e+00
-1.31904185e+00 5.54337502e-01 3.57759774e-01 -9.98150229e-01
6.35179207e-02 -7.74623454e-01 -4.38243866e-01 8.51467371e-01
-1.74201882e+00 -1.26132429e+00 -5.76739728e-01 5.75868249e-01
8.44618678e-01 5.60132086e-01 6.72363400e-01 -1.56296700e-01
-3.20913166e-01 -3.95526499e-01 -2.74432115e-02 -7.59462118e-02
1.24538772e-01 -1.77281022e+00 5.16875684e-01 6.46385849e-01
3.12739402e-01 -1.15667284e-02 1.27193898e-01 -8.45196843e-01
-1.28150129e+00 -1.57740378e+00 1.10107228e-01 -9.10776198e-01
2.29989633e-01 -6.34437561e-01 -7.49425113e-01 6.17390454e-01
2.83489712e-02 3.99676770e-01 3.50749880e-01 -5.48345387e-01
1.26775831e-01 -6.18756637e-02 -1.44347715e+00 2.06280842e-01
1.32872093e+00 -4.07057434e-01 -4.93354589e-01 6.89420521e-01
7.10832775e-01 -9.99162495e-01 -1.14002645e+00 6.19251013e-01
1.61320418e-01 -1.31645286e+00 1.19763744e+00 -1.45055845e-01
4.16398346e-01 -6.77107215e-01 -4.59624290e-01 -1.29650331e+00
2.03053951e-01 -1.76347911e-01 1.13523072e-02 1.16720569e+00
3.99273366e-01 -4.30623621e-01 8.33236396e-01 5.66445649e-01
-3.83823067e-01 -9.06502128e-01 -9.71258759e-01 -5.41656017e-01
5.52255623e-02 -8.08826447e-01 7.37490833e-01 7.68184006e-01
-1.11515689e+00 1.14685863e-01 2.76436388e-01 5.52217305e-01
8.68458688e-01 5.71530402e-01 8.65572095e-01 -1.24216700e+00
-1.68651000e-01 -1.33680031e-01 -1.43152580e-01 -1.29665077e+00
1.86692372e-01 -1.12322891e+00 1.70747593e-01 -2.19270349e+00
-2.46518657e-01 -4.55840617e-01 2.17045680e-01 4.02400225e-01
2.57442892e-02 5.24423122e-01 2.77956516e-01 1.25477582e-01
-5.03215969e-01 5.31772852e-01 1.60983169e+00 -1.34493440e-01
-1.97453633e-01 -4.76600751e-02 -2.12381825e-01 8.87427509e-01
7.45894670e-01 -3.62299323e-01 -5.60287893e-01 -4.09439862e-01
4.17747274e-02 9.15936157e-02 7.42429078e-01 -9.82288301e-01
-3.07682436e-02 -2.72508383e-01 9.49307621e-01 -1.07329607e+00
5.31785548e-01 -1.10629785e+00 3.85020792e-01 2.52251402e-02
6.98037222e-02 -2.36108199e-01 9.55464095e-02 5.76692462e-01
2.57933456e-02 -1.43691033e-01 6.01793468e-01 -9.37205732e-01
-1.06956828e+00 7.48493612e-01 -2.44121403e-02 2.35705785e-02
1.06000137e+00 -6.11320496e-01 -1.61483765e-01 1.23087332e-01
-8.59670579e-01 3.43402833e-01 6.88445210e-01 1.99918747e-02
7.82553494e-01 -9.22006726e-01 -4.31257308e-01 3.44197839e-01
-1.36343941e-01 1.27238262e+00 5.94938755e-01 6.38749599e-01
-6.42011464e-01 1.15489148e-01 1.17787138e-01 -1.21256769e+00
-5.39413750e-01 3.34637731e-01 8.44441235e-01 -1.13055276e-05
-9.01623726e-01 9.02311623e-01 5.26099205e-01 -1.20811498e+00
-1.22180767e-01 -7.56504774e-01 3.42530198e-02 6.46096095e-02
1.00830041e-01 4.73700576e-02 1.76294129e-02 -6.31386876e-01
-3.59163940e-01 9.41858292e-01 3.60649318e-01 -1.30906790e-01
1.50528157e+00 -1.95963591e-01 -1.04640469e-01 5.87229550e-01
1.12292504e+00 -1.65690988e-01 -1.92925775e+00 2.90046424e-01
-2.37997472e-01 -5.21866679e-01 2.10224286e-01 -1.01816308e+00
-1.28467286e+00 7.30481446e-01 6.65524304e-01 1.97946608e-01
1.01720321e+00 4.66203511e-01 8.09372663e-01 3.12704682e-01
6.33308411e-01 -8.80442142e-01 -3.03754270e-01 4.39759374e-01
7.49896526e-01 -1.25013089e+00 -3.11123542e-02 -4.26999360e-01
-3.72949690e-01 1.33233690e+00 8.31487775e-01 -3.02069902e-01
6.63608968e-01 1.36088029e-01 8.73350054e-02 -6.00437284e-01
-1.65856555e-01 -2.83447266e-01 4.38912004e-01 5.83277166e-01
-1.09219514e-01 6.39148653e-02 3.71085346e-01 3.14839333e-01
-3.35857481e-01 -1.83358207e-01 2.29680508e-01 6.59754992e-01
-4.58130300e-01 -7.98014164e-01 -5.74712098e-01 5.78140676e-01
1.06162451e-01 -1.85775850e-02 -2.86344409e-01 1.00401378e+00
6.68264151e-01 3.80291373e-01 3.56724560e-01 -2.00088367e-01
5.68689644e-01 -4.95443828e-02 6.30432010e-01 -7.57103503e-01
-5.53265095e-01 9.03289542e-02 -1.45901233e-01 -9.97394204e-01
-8.86741519e-01 -4.80503708e-01 -1.59277415e+00 1.60738334e-01
-8.96308571e-02 1.35989208e-03 7.64669120e-01 9.74163771e-01
9.70510580e-03 3.64696443e-01 6.03296638e-01 -1.60434914e+00
2.52597809e-01 -4.63096470e-01 -5.29324293e-01 2.20121428e-01
3.21132869e-01 -5.53474128e-01 -5.83728194e-01 1.01766534e-01] | [8.637894630432129, -2.99531626701355] |
d1c03036-a55f-41aa-a587-8ac30409f9a6 | multi-label-image-classification-using | 2301.04494 | null | https://arxiv.org/abs/2301.04494v1 | https://arxiv.org/pdf/2301.04494v1.pdf | Multi-label Image Classification using Adaptive Graph Convolutional Networks: from a Single Domain to Multiple Domains | This paper proposes an adaptive graph-based approach for multi-label image classification. Graph-based methods have been largely exploited in the field of multi-label classification, given their ability to model label correlations. Specifically, their effectiveness has been proven not only when considering a single domain but also when taking into account multiple domains. However, the topology of the used graph is not optimal as it is pre-defined heuristically. In addition, consecutive Graph Convolutional Network (GCN) aggregations tend to destroy the feature similarity. To overcome these issues, an architecture for learning the graph connectivity in an end-to-end fashion is introduced. This is done by integrating an attention-based mechanism and a similarity-preserving strategy. The proposed framework is then extended to multiple domains using an adversarial training scheme. Numerous experiments are reported on well-known single-domain and multi-domain benchmarks. The results demonstrate that our approach outperforms the state-of-the-art in terms of mean Average Precision (mAP) and model size. | ['Djamila Aouada', 'Oyebade Oyedotun', 'Enjie Ghorbel', 'Indel Pal Singh'] | 2023-01-11 | null | null | null | null | ['multi-label-image-classification'] | ['computer-vision'] | [ 3.65173817e-01 7.21296445e-02 -2.65909344e-01 -3.93675566e-01
-6.14762068e-01 -3.54980111e-01 7.19697833e-01 7.79634237e-01
-4.98562217e-01 5.96723437e-01 -3.21104556e-01 -3.44630480e-02
-3.25574487e-01 -7.29251266e-01 -5.23480415e-01 -7.18713641e-01
-3.36494995e-03 6.59587860e-01 2.86653757e-01 -1.76766455e-01
3.78743738e-01 4.37714428e-01 -1.49567711e+00 1.57721549e-01
8.80680740e-01 9.80244219e-01 -7.50391046e-03 3.53979141e-01
-1.02761880e-01 8.03426325e-01 -4.36717600e-01 -6.75471425e-01
1.25833973e-01 -4.40692812e-01 -9.15904760e-01 3.21109056e-01
6.43525362e-01 1.66781992e-01 1.59903914e-02 1.31299925e+00
4.15478647e-01 5.00813574e-02 6.94923878e-01 -1.31030524e+00
-4.90389943e-01 3.60749096e-01 -7.81565666e-01 3.18107530e-02
1.26233190e-01 -1.84592217e-01 1.27414429e+00 -5.04648089e-01
6.87589943e-01 1.17430496e+00 6.01209939e-01 2.71404624e-01
-1.35256588e+00 -5.20582795e-01 3.78854722e-01 2.44489267e-01
-1.25467932e+00 3.42242680e-02 1.05721569e+00 -3.98919761e-01
7.35676348e-01 -9.77599528e-03 3.43756706e-01 8.88761818e-01
4.04265285e-01 5.26098669e-01 1.37946296e+00 -7.13904321e-01
1.08664185e-01 1.96264073e-01 1.73577532e-01 7.49365926e-01
3.06540400e-01 -1.64549947e-01 -1.68917894e-01 -1.74558192e-01
3.70372385e-01 -1.56287894e-01 3.44860703e-02 -8.81233752e-01
-9.05289412e-01 9.92284775e-01 6.86339438e-01 7.04457879e-01
-2.32868508e-01 6.55632392e-02 7.31877148e-01 4.66085255e-01
7.81430304e-01 4.99093533e-01 -2.00993955e-01 3.64028335e-01
-9.77747381e-01 1.67379916e-01 5.99738717e-01 7.07862794e-01
6.89493179e-01 -2.39289701e-01 -9.51091275e-02 1.02080989e+00
2.33237967e-01 -1.48389041e-01 4.89970028e-01 -4.17913049e-01
5.13687253e-01 1.05001497e+00 -2.49399781e-01 -1.28679276e+00
-8.02076995e-01 -8.21573615e-01 -9.60980892e-01 3.73930037e-01
4.99781758e-01 2.75957197e-01 -7.79728234e-01 1.84594452e+00
4.43717510e-01 3.42094690e-01 -1.58004612e-01 7.55271018e-01
5.27267337e-01 2.05285400e-02 4.79608685e-01 -1.00936905e-01
1.46339738e+00 -1.27887189e+00 -5.21840990e-01 -2.10399404e-01
8.06111217e-01 -6.34837031e-01 9.33313191e-01 3.95276397e-01
-8.01155567e-01 -6.32424295e-01 -1.14924884e+00 1.14327103e-01
-7.70362794e-01 5.00588156e-02 3.47574711e-01 6.95146620e-01
-9.55360174e-01 6.74069464e-01 -5.23659408e-01 -6.52057886e-01
5.93684793e-01 5.28537273e-01 -4.73821312e-01 -2.04768017e-01
-1.15271246e+00 8.96009207e-01 6.11445248e-01 -2.08568156e-01
-5.75928152e-01 -2.86035895e-01 -6.27840996e-01 8.21652710e-02
3.51570010e-01 -7.75573313e-01 7.83055127e-01 -1.09375644e+00
-1.18327832e+00 1.06500566e+00 4.20075923e-01 -5.91604412e-01
6.30609334e-01 2.04001576e-01 -5.07359028e-01 4.14126903e-01
6.14663921e-02 4.64447826e-01 8.80331159e-01 -1.25435197e+00
-6.01173162e-01 -4.25869793e-01 4.26085770e-01 2.17512339e-01
-6.61486506e-01 -2.48882803e-03 -2.18414918e-01 -8.16281617e-01
1.47530725e-02 -1.14742839e+00 -2.88950920e-01 -2.14088455e-01
-4.24267530e-01 -1.02331974e-01 7.17258334e-01 -2.89633900e-01
1.15510130e+00 -1.89500332e+00 2.41089195e-01 3.14604372e-01
2.82012254e-01 3.46345335e-01 -1.88003987e-01 3.72212350e-01
-2.51001745e-01 3.61338221e-02 -2.02263832e-01 -4.74483222e-01
-1.46581754e-01 -1.00029357e-01 2.82714456e-01 6.39425516e-01
2.14864001e-01 7.03847349e-01 -8.53256643e-01 -7.56947994e-01
3.31486881e-01 4.24872935e-01 -5.05842149e-01 -4.99900691e-02
-3.35448772e-01 6.34716094e-01 -4.30024803e-01 4.91522610e-01
8.14692855e-01 -6.00923538e-01 6.05242789e-01 -2.24364743e-01
1.95961475e-01 -2.03551993e-01 -1.07536042e+00 1.89301956e+00
-6.44713044e-01 5.91145679e-02 -1.09434910e-01 -1.27401876e+00
9.21935499e-01 2.36784711e-01 6.98958993e-01 -5.46745121e-01
3.38507771e-01 3.96716118e-01 7.70147983e-03 -5.23190983e-02
4.83384877e-01 -7.95776099e-02 -2.91953515e-02 3.47685874e-01
2.36430451e-01 2.37232417e-01 3.67979974e-01 5.84181994e-02
1.17886043e+00 -6.53578043e-02 5.92820823e-01 -4.24796075e-01
9.86939490e-01 3.20002958e-02 1.97854787e-01 4.96819526e-01
-9.63351503e-02 4.95025128e-01 6.06947780e-01 -3.31573188e-01
-8.25294614e-01 -4.62030351e-01 -6.30517229e-02 1.08603740e+00
3.64426941e-01 -2.59800285e-01 -7.31823444e-01 -1.27465379e+00
5.37127256e-02 4.84530956e-01 -7.47119129e-01 -2.52127141e-01
-3.58410895e-01 -7.76770353e-01 4.32881981e-01 1.78094044e-01
6.54610157e-01 -8.33177209e-01 -3.58339846e-01 4.69805807e-01
-1.85123156e-03 -1.27462852e+00 -3.39521110e-01 2.23128319e-01
-8.20586741e-01 -1.35600567e+00 -7.57762730e-01 -8.05887699e-01
7.33679414e-01 1.64351121e-01 1.13222563e+00 2.47767419e-01
-1.02740072e-01 3.09075177e-01 -4.30831313e-01 2.97405291e-02
-6.37365520e-01 4.62349355e-01 -3.55314314e-01 3.70521218e-01
3.31537426e-01 -6.17211580e-01 -5.20490468e-01 4.08397228e-01
-1.08055460e+00 -1.63499177e-01 7.77964354e-01 1.09374571e+00
6.04808807e-01 1.34095863e-01 8.16652596e-01 -1.33816636e+00
7.38829613e-01 -3.96957994e-01 -6.60315394e-01 6.31622434e-01
-9.74272549e-01 8.85891020e-02 8.72101843e-01 -5.06867468e-01
-7.50890374e-01 1.45943016e-01 2.20054053e-02 -4.29635763e-01
-3.73450726e-01 4.82777148e-01 -1.04496889e-01 -6.25400722e-01
4.95402247e-01 -6.88188151e-02 2.94996183e-02 -4.37551558e-01
3.58382672e-01 6.15137577e-01 2.05478668e-01 -3.87257904e-01
5.75879455e-01 3.85570943e-01 4.50300515e-01 -4.41717148e-01
-7.41996706e-01 -6.15512490e-01 -8.64261150e-01 -2.42471516e-01
8.88733625e-01 -7.63348281e-01 -5.97762406e-01 6.23842120e-01
-1.03110862e+00 2.33158804e-02 1.61257669e-01 2.30510950e-01
-5.98839998e-01 5.56808949e-01 -5.17599940e-01 -3.48631859e-01
-3.13353091e-01 -1.36048937e+00 9.71070230e-01 9.17506404e-03
-6.92444295e-02 -1.30894744e+00 5.18629700e-02 4.10811812e-01
3.44881624e-01 4.07431722e-01 1.16302741e+00 -1.09743273e+00
-3.42648983e-01 -2.58115292e-01 -4.42159563e-01 3.17284614e-01
8.09241028e-05 -4.53178883e-01 -9.28419888e-01 -5.54427564e-01
-2.16309994e-01 -3.76047850e-01 7.38288999e-01 4.49883528e-02
1.08758366e+00 -3.83094326e-02 -5.64490318e-01 1.87499329e-01
1.94625366e+00 -1.29178777e-01 3.10475230e-01 5.72782815e-01
7.53140211e-01 5.96377134e-01 7.52227843e-01 2.74040103e-01
3.68399620e-01 1.09904802e+00 7.89316654e-01 -1.68362275e-01
-1.74975812e-01 8.33932683e-02 -1.79390371e-01 4.36854333e-01
1.49614692e-01 -5.15560806e-01 -8.25097144e-01 4.19782579e-01
-1.82996905e+00 -6.87837005e-01 -1.30228728e-01 2.12058401e+00
3.90724301e-01 3.00007939e-01 2.07860112e-01 3.05568218e-01
9.19722974e-01 3.36830765e-01 -3.02450985e-01 -5.45245111e-01
-3.88248637e-03 1.53040856e-01 6.63300753e-01 2.64268309e-01
-1.52728295e+00 8.05668533e-01 4.86902761e+00 9.49720919e-01
-1.16418290e+00 3.24366808e-01 4.79420066e-01 4.36765969e-01
6.02107197e-02 -1.31047711e-01 -5.76277375e-01 3.64872426e-01
7.82613873e-01 1.00122370e-01 2.25888640e-01 8.72983038e-01
-2.52911776e-01 -1.06756166e-02 -9.29315627e-01 7.85548210e-01
1.80973962e-01 -9.46784139e-01 -5.99933192e-02 1.48378581e-01
6.65405691e-01 -1.13295503e-01 5.44275232e-02 1.65073469e-01
1.19870469e-01 -7.64375746e-01 6.33335888e-01 1.97031349e-01
6.60520673e-01 -8.39842856e-01 8.66060436e-01 1.90266296e-01
-1.35735202e+00 -1.34398460e-01 -2.12616846e-01 2.58184999e-01
-9.72694457e-02 6.88440979e-01 -8.78658116e-01 1.01220214e+00
1.67026654e-01 6.14658475e-01 -1.07651496e+00 1.01014185e+00
-4.72200364e-02 1.80858776e-01 -4.32182178e-02 5.77063709e-02
4.14264083e-01 -1.56691954e-01 3.66311371e-01 1.26227689e+00
1.78916305e-01 -4.90013897e-01 3.74676466e-01 4.28184927e-01
-1.07918061e-01 5.27930915e-01 -6.60750091e-01 6.87530041e-02
1.62435234e-01 1.46677303e+00 -1.21569717e+00 -2.55705267e-01
-7.35722959e-01 1.02856672e+00 5.32384157e-01 -6.25199154e-02
-9.37357426e-01 -3.50350201e-01 3.20414960e-01 1.36136651e-01
3.75692248e-01 1.48000929e-03 -1.11270830e-01 -9.00940955e-01
-2.34341711e-01 -1.01178718e+00 6.31453335e-01 -2.79641479e-01
-1.49320328e+00 8.46368372e-01 -1.88943762e-02 -1.39949727e+00
-1.38706654e-01 -4.76504475e-01 -6.29519522e-02 5.57879329e-01
-1.75591278e+00 -1.50463080e+00 -3.85914236e-01 5.35983980e-01
3.13658267e-01 -2.73071855e-01 1.09942484e+00 7.69546509e-01
-4.13837671e-01 8.36855948e-01 1.49596959e-01 -9.47702453e-02
7.81008363e-01 -1.31052220e+00 8.77734050e-02 7.28703141e-01
2.85992324e-01 2.17173994e-01 6.78039908e-01 -4.52104479e-01
-9.00292099e-01 -1.25392365e+00 7.98897982e-01 -5.73620573e-02
6.26168191e-01 -2.80256122e-01 -9.71497476e-01 4.29184616e-01
2.86228210e-01 3.98739100e-01 6.15774274e-01 -5.55657335e-02
-4.55591798e-01 -1.82227939e-01 -1.33035028e+00 4.05502915e-01
8.48033965e-01 -3.51463437e-01 -1.37179598e-01 4.92764324e-01
3.94744515e-01 -3.65694761e-01 -1.11595941e+00 4.36001539e-01
4.06366110e-01 -1.10337472e+00 9.22343075e-01 -4.06059325e-01
3.79641354e-01 -1.86821237e-01 3.77515629e-02 -1.43287110e+00
-2.83429086e-01 -1.05040893e-01 1.66506782e-01 1.27203190e+00
3.47652107e-01 -6.98083043e-01 6.60571277e-01 2.16471374e-01
-5.05459011e-02 -7.60784566e-01 -9.04972374e-01 -8.28671813e-01
8.19401368e-02 -6.57435805e-02 5.58287621e-01 1.14331782e+00
-2.77227104e-01 4.61460918e-01 -4.17690337e-01 1.89992413e-01
6.60728276e-01 2.70918667e-01 6.20257854e-01 -1.53971815e+00
-2.46220618e-01 -6.36569858e-01 -7.85016358e-01 -5.87944865e-01
4.35206711e-01 -1.08164549e+00 -3.68133038e-01 -1.37846661e+00
9.14684460e-02 -6.61839306e-01 -6.73611999e-01 3.92550468e-01
-7.90856555e-02 3.85711938e-01 2.61164188e-01 1.00807652e-01
-8.88039470e-01 2.07605734e-01 1.19296157e+00 -2.34372169e-01
4.26975824e-02 -1.89360566e-02 -5.23750663e-01 4.23844784e-01
9.50517595e-01 -7.32444048e-01 -5.12815237e-01 -4.51791175e-02
1.01789786e-02 -1.04723059e-01 3.34440768e-01 -1.25309646e+00
2.22393006e-01 1.43356323e-01 -1.38050765e-01 -2.19418898e-01
3.18169147e-01 -1.17866707e+00 1.76936150e-01 4.39688027e-01
-5.28731644e-01 1.22979648e-01 6.92542195e-02 9.19677317e-01
-3.33623320e-01 -2.99400985e-01 1.15153241e+00 -6.77862838e-02
-6.46152377e-01 1.58178493e-01 9.40727592e-02 -6.53852373e-02
1.40878129e+00 -2.50292886e-02 -3.13287705e-01 1.54195284e-03
-6.49628878e-01 1.14058837e-01 5.96148312e-01 4.54574138e-01
2.25420520e-01 -1.29154336e+00 -4.23613131e-01 6.67000562e-02
5.63546240e-01 -4.35649306e-01 2.62366742e-01 1.03918183e+00
-4.21489626e-01 3.31855893e-01 -3.53744209e-01 -5.87894559e-01
-1.55598223e+00 8.32487226e-01 3.17527503e-01 -1.03193581e+00
-3.66269618e-01 5.06641984e-01 -1.04842208e-01 -3.89101863e-01
2.46165782e-01 -1.46320134e-01 -4.39282179e-01 3.09683233e-01
8.57372731e-02 2.06055284e-01 3.77074420e-01 -8.14376891e-01
-3.78403991e-01 8.04502606e-01 -2.00188085e-01 2.69006550e-01
1.19868279e+00 -2.05388755e-01 -1.83397710e-01 2.01747283e-01
1.42213655e+00 -1.67367131e-01 -7.17169940e-01 -2.28181183e-01
4.17296410e-01 -3.25774640e-01 1.73224267e-02 -7.83551812e-01
-1.19643807e+00 8.16537976e-01 7.05676854e-01 5.30229151e-01
1.15623999e+00 -2.71175444e-01 5.30449152e-01 -8.59040581e-03
4.78582442e-01 -9.80733991e-01 2.69235283e-01 1.75147906e-01
6.10046923e-01 -1.46570432e+00 7.74003193e-02 -5.50429046e-01
-6.27573431e-01 1.14630508e+00 5.05100250e-01 -1.86091542e-01
7.81786025e-01 -8.63815919e-02 -3.40697058e-02 -2.96756625e-01
-4.85232234e-01 -3.25358510e-01 1.87545374e-01 4.12367374e-01
4.73381162e-01 -7.93288425e-02 -7.21679688e-01 1.76107928e-01
4.39411253e-01 1.05161034e-01 1.57423839e-01 7.52154946e-01
-1.87425569e-01 -1.46992290e+00 -2.23133415e-01 3.56657892e-01
-8.14607918e-01 1.31068945e-01 -2.77532429e-01 1.06065953e+00
2.64635205e-01 9.71582413e-01 -3.68356198e-01 -4.19849217e-01
2.94124693e-01 1.49544537e-01 5.48740029e-01 -5.71506083e-01
-9.04882729e-01 -1.01386100e-01 9.46447328e-02 -4.22822237e-01
-8.22931468e-01 -3.76980245e-01 -9.05955613e-01 -7.81973079e-02
-4.85584527e-01 6.12399615e-02 7.02568471e-01 7.58318186e-01
3.08193088e-01 7.46202409e-01 5.73498249e-01 -5.38363278e-01
-5.84010601e-01 -1.01742828e+00 -6.61370993e-01 6.87117517e-01
-6.71516433e-02 -1.00003159e+00 -2.40695655e-01 -3.70468646e-01] | [9.726165771484375, 3.7618958950042725] |
5af707d2-1ba4-4b96-aff3-c249432e1338 | distribution-learning-based-on-evolutionary | 2207.12744 | null | https://arxiv.org/abs/2207.12744v1 | https://arxiv.org/pdf/2207.12744v1.pdf | Distribution Learning Based on Evolutionary Algorithm Assisted Deep Neural Networks for Imbalanced Image Classification | To address the trade-off problem of quality-diversity for the generated images in imbalanced classification tasks, we research on over-sampling based methods at the feature level instead of the data level and focus on searching the latent feature space for optimal distributions. On this basis, we propose an iMproved Estimation Distribution Algorithm based Latent featUre Distribution Evolution (MEDA_LUDE) algorithm, where a joint learning procedure is programmed to make the latent features both optimized and evolved by the deep neural networks and the evolutionary algorithm, respectively. We explore the effect of the Large-margin Gaussian Mixture (L-GM) loss function on distribution learning and design a specialized fitness function based on the similarities among samples to increase diversity. Extensive experiments on benchmark based imbalanced datasets validate the effectiveness of our proposed algorithm, which can generate images with both quality and diversity. Furthermore, the MEDA_LUDE algorithm is also applied to the industrial field and successfully alleviates the imbalanced issue in fabric defect classification. | ['Bing Wei', 'Chaochen Gu', 'Kuangrong Hao', 'Yudi Zhao'] | 2022-07-26 | null | null | null | null | ['imbalanced-classification'] | ['miscellaneous'] | [-9.39281061e-02 -3.69066417e-01 -2.71173269e-02 -3.60863805e-01
-2.68464953e-01 1.40777886e-01 9.45259444e-03 -3.76527719e-02
-5.99817075e-02 7.84471214e-01 -1.94697112e-01 1.07624471e-01
-5.01561224e-01 -1.13028967e+00 -3.73741537e-01 -1.14822996e+00
1.43708140e-01 4.03802514e-01 -1.11437656e-01 4.22467571e-03
6.88212276e-01 3.40888977e-01 -1.85967183e+00 1.25014439e-01
1.49846411e+00 9.32469428e-01 1.51600167e-01 3.43102247e-01
-1.55655444e-01 3.46175015e-01 -1.05543923e+00 -2.87661374e-01
2.19165817e-01 -6.40854716e-01 -2.11114794e-01 4.57076013e-01
-1.78744882e-01 -1.72535345e-01 7.71531016e-02 1.08852160e+00
9.70318794e-01 -1.39369130e-01 9.88387764e-01 -1.56943202e+00
-6.06267154e-01 2.16093212e-01 -1.13106799e+00 1.24663524e-01
-9.29816589e-02 1.63164362e-01 4.72201735e-01 -5.07888734e-01
3.62828612e-01 1.31446087e+00 5.81890225e-01 4.17749703e-01
-1.01640868e+00 -7.72369683e-01 -1.52129352e-01 3.63610595e-01
-1.30788565e+00 7.79473130e-03 1.03456640e+00 -4.71619517e-01
6.14718497e-01 1.04609258e-01 7.01849282e-01 6.80105090e-01
6.09485626e-01 9.80736911e-01 9.34754372e-01 -6.38331771e-01
3.90675694e-01 4.15643245e-01 -3.38997953e-02 6.22010887e-01
7.04002082e-01 -6.33538887e-02 -4.08108741e-01 -2.22907782e-01
6.74491763e-01 -9.92807895e-02 -2.03194052e-01 -6.14410818e-01
-7.16135085e-01 9.54404414e-01 -9.21468344e-03 1.28060430e-01
-5.00075519e-01 -2.20904902e-01 5.94548941e-01 4.26147550e-01
5.05467296e-01 5.63093603e-01 -4.83079553e-01 -8.51512104e-02
-8.19968104e-01 2.75287777e-01 5.27666986e-01 5.74797511e-01
6.70156002e-01 2.34897807e-01 -4.35111612e-01 1.19326079e+00
2.86583900e-01 2.44592816e-01 7.88770497e-01 -5.88091791e-01
3.62928599e-01 9.02307332e-01 -4.42477204e-02 -1.37566411e+00
-1.11678474e-01 -8.32952797e-01 -8.04828465e-01 4.92984325e-01
1.48324147e-02 -3.63538235e-01 -7.75564373e-01 1.47691286e+00
6.38108313e-01 -1.88152120e-01 -3.93520631e-02 7.17229843e-01
2.21839979e-01 7.37735808e-01 -1.62260592e-01 -4.65631545e-01
8.96905065e-01 -8.84466231e-01 -9.21951473e-01 2.48392388e-01
5.09451509e-01 -6.88845694e-01 1.15673554e+00 6.10920012e-01
-8.35325539e-01 -6.91429794e-01 -1.42454135e+00 6.57904148e-01
-1.18148670e-01 6.29384279e-01 4.70000952e-01 8.46859932e-01
-5.32721639e-01 6.94999397e-01 -6.31954670e-01 -8.44649225e-02
6.74723387e-01 3.97076964e-01 2.42121406e-02 5.34472689e-02
-8.55375469e-01 4.98096049e-01 5.11492908e-01 1.59900442e-01
-6.62452221e-01 -5.43300867e-01 -4.98852491e-01 -6.24935217e-02
-3.04243993e-02 -6.28129959e-01 4.98033792e-01 -1.32359946e+00
-1.81835485e+00 5.14802456e-01 4.03164625e-01 -2.06809789e-02
4.29847747e-01 2.51217652e-03 -3.55644375e-01 -1.88577875e-01
-1.42598510e-01 5.09593666e-01 9.76456702e-01 -1.15972805e+00
-6.72419071e-01 -4.91537958e-01 -5.10195136e-01 2.51859188e-01
-8.16195190e-01 -3.91863286e-01 1.42251298e-01 -7.81806827e-01
-4.32821810e-02 -6.03797793e-01 -4.94749397e-02 -1.18187718e-01
-3.03945720e-01 -3.01532656e-01 9.92838144e-01 -6.48372293e-01
1.34034944e+00 -2.20396733e+00 2.24882290e-01 3.56443852e-01
-5.80593422e-02 3.87794346e-01 -6.76225200e-02 1.41087532e-01
-1.23146959e-01 -6.21275641e-02 -3.48536700e-01 3.90151106e-02
-9.19782743e-02 4.67283130e-02 2.24150196e-01 5.14938533e-01
3.35567892e-01 3.92539501e-01 -5.33524513e-01 -5.14136553e-01
-4.97846641e-02 2.89375484e-01 -7.89149702e-01 3.93008173e-01
-1.58125818e-01 2.65919715e-01 -5.74401796e-01 6.74507737e-01
1.18839133e+00 -1.33512616e-02 -1.86753459e-02 -3.43160033e-01
5.56597523e-02 -4.08587694e-01 -1.37158728e+00 1.39929521e+00
-2.63999403e-01 2.15686738e-01 -1.44453600e-01 -1.11004448e+00
1.55417955e+00 -1.30488798e-01 4.86422628e-01 -7.46522307e-01
4.41303611e-01 4.10906047e-01 2.22706482e-01 -7.63984621e-01
3.37644845e-01 -1.29986629e-01 9.40575004e-02 3.78066242e-01
1.11110322e-01 8.37172046e-02 -2.54018907e-03 -4.41584438e-01
7.08616316e-01 2.33151466e-01 3.93466186e-03 -4.81609792e-01
7.13488400e-01 -6.65424094e-02 9.37246382e-01 3.30059081e-01
-1.75565302e-01 4.60962355e-01 7.43146539e-01 -3.71642679e-01
-1.18388855e+00 -6.58997118e-01 -2.50885218e-01 5.28302729e-01
3.65603983e-01 1.01758435e-01 -7.46137500e-01 -7.14303255e-01
1.19891666e-01 7.61414945e-01 -4.92148221e-01 -7.89302766e-01
-2.95463264e-01 -1.47063553e+00 2.65901178e-01 2.78644323e-01
7.93516994e-01 -9.15713072e-01 -6.34972572e-01 1.37358710e-01
2.60429263e-01 -1.33534119e-01 -1.22955322e-01 1.08909257e-01
-8.83204758e-01 -7.38493025e-01 -8.09183180e-01 -7.58355081e-01
7.66357124e-01 -3.13300610e-01 9.32656825e-01 2.26483271e-02
-8.03403258e-01 1.60554459e-03 -4.89229321e-01 -5.16874373e-01
-4.03218895e-01 1.09206825e-01 -1.45230770e-01 1.48206502e-01
4.36462522e-01 -7.95360923e-01 -7.25143909e-01 4.19949204e-01
-8.12205195e-01 -1.71004340e-01 7.95066714e-01 1.24335480e+00
5.02219021e-01 7.60884404e-01 9.58452344e-01 -4.42220241e-01
1.08418489e+00 -6.14905417e-01 -6.65591419e-01 4.35438663e-01
-8.11825514e-01 2.58977890e-01 4.77833718e-01 -5.75240254e-01
-1.13502443e+00 -3.36417228e-01 -9.95978992e-03 -4.19772625e-01
1.02080487e-01 1.63224936e-01 -5.55693865e-01 7.74385557e-02
3.97877574e-01 3.51569623e-01 3.71769011e-01 -3.41162443e-01
-1.12123825e-01 1.00981975e+00 3.10520474e-02 -6.74945056e-01
4.13567275e-01 7.51827136e-02 2.64411680e-02 -5.10033250e-01
-3.70237619e-01 3.78789380e-02 -2.80018985e-01 -1.94589823e-01
7.89605618e-01 -6.09051287e-01 -7.39347219e-01 8.97963107e-01
-8.73996019e-01 1.50062025e-01 -4.26773548e-01 4.98830199e-01
-6.16922975e-01 2.01817006e-01 -3.89701813e-01 -1.02608132e+00
-4.07770306e-01 -1.30547428e+00 9.21398342e-01 5.50734520e-01
2.06897333e-01 -7.04777300e-01 -5.40980697e-03 1.32370159e-01
3.92264754e-01 5.75977862e-01 1.50731456e+00 -4.32102740e-01
-3.11989456e-01 -1.06668979e-01 -7.91115910e-02 6.79571331e-01
2.15458587e-01 1.71582878e-01 -5.15677273e-01 -3.59197319e-01
3.00706744e-01 -4.51643556e-01 5.36549747e-01 4.06786561e-01
1.35225499e+00 -1.67196006e-01 -3.02925259e-01 6.07216597e-01
1.67097652e+00 6.37066007e-01 8.72880340e-01 4.97810721e-01
3.07305187e-01 6.96573794e-01 1.01131129e+00 9.50849533e-01
-6.54225349e-02 4.10549372e-01 4.24730957e-01 -1.05689457e-02
-7.06600025e-02 5.39359357e-03 9.81492624e-02 9.59124029e-01
3.53882730e-01 -5.18373191e-01 -4.72534269e-01 4.93847311e-01
-1.70920837e+00 -6.17820919e-01 1.07157379e-01 2.12536073e+00
9.17792141e-01 1.99525550e-01 1.20013282e-01 3.23237836e-01
1.22279787e+00 -1.85597256e-01 -8.37336600e-01 -4.93257970e-01
-1.40597597e-01 8.11729431e-02 2.43040636e-01 5.38714789e-03
-8.39236200e-01 3.29893857e-01 5.96240425e+00 1.48640776e+00
-1.34470797e+00 -1.90562144e-01 1.06393886e+00 -4.03397642e-02
-5.00689447e-01 -2.15685010e-01 -7.87784040e-01 8.98525476e-01
3.19241971e-01 -2.35070549e-02 2.33352914e-01 9.43665624e-01
-3.29707041e-02 -1.31406620e-01 -5.62119961e-01 1.11341310e+00
1.23987243e-01 -9.83596504e-01 2.33985975e-01 1.77516475e-01
9.66805339e-01 -7.33764172e-01 3.18684101e-01 1.33102685e-01
-1.11618340e-01 -8.97350788e-01 5.93957722e-01 6.99671090e-01
5.59304297e-01 -1.34347081e+00 9.77422595e-01 4.03312594e-01
-6.06845498e-01 -5.10910153e-01 -6.58240080e-01 3.81913245e-01
-1.32083416e-01 9.08663988e-01 -6.17053568e-01 7.30093420e-01
5.37691295e-01 4.97819155e-01 -5.42654216e-01 1.25343847e+00
1.36576742e-01 4.29467052e-01 -1.69878468e-01 -2.48770639e-01
-1.27605200e-01 -4.54404414e-01 7.20046163e-01 6.38442934e-01
5.50857961e-01 -4.04963523e-01 -6.13600835e-02 1.05497146e+00
3.22591633e-01 2.49554828e-01 -2.39657149e-01 3.37053575e-02
5.43860197e-01 1.09400296e+00 -7.45844185e-01 1.10199034e-01
2.78253347e-01 9.12401676e-01 1.05601899e-01 -6.99995691e-03
-9.36356068e-01 -8.51651251e-01 4.14398819e-01 6.01593554e-02
4.90010291e-01 9.64056887e-03 -3.54494035e-01 -7.85852492e-01
1.50372520e-01 -1.04814470e+00 2.99984783e-01 -3.97581279e-01
-1.58105624e+00 5.95368385e-01 -7.37750530e-02 -1.34075618e+00
6.70546815e-02 -4.68987256e-01 -6.56401992e-01 7.11516857e-01
-1.35442889e+00 -7.76108801e-01 -3.76472980e-01 3.26972395e-01
5.26869059e-01 -6.90537333e-01 3.93224597e-01 5.21320879e-01
-9.30049360e-01 8.50938380e-01 3.89819264e-01 -4.27033603e-01
4.59581137e-01 -9.09659624e-01 -3.31573308e-01 4.39832509e-01
-5.82360446e-01 2.30490223e-01 7.58481264e-01 -7.19713330e-01
-1.22841156e+00 -9.46880460e-01 1.12600818e-01 3.32497209e-01
-3.49762471e-04 -2.55265594e-01 -6.91818595e-01 -1.04340367e-01
2.62779176e-01 -3.30430210e-01 7.06854522e-01 -1.96021974e-01
2.77511835e-01 -5.23521006e-01 -1.64119959e+00 2.01102778e-01
7.42734969e-01 8.18101391e-02 -1.93496630e-01 1.84869975e-01
6.27291322e-01 -1.37100339e-01 -8.17762554e-01 7.93723762e-01
4.81329471e-01 -1.06293130e+00 7.39286184e-01 -2.99702764e-01
5.25609732e-01 -3.88770789e-01 -1.14626385e-01 -1.44085383e+00
-4.05405849e-01 -2.49116719e-01 1.16601236e-01 1.62504125e+00
1.16559882e-02 -7.02449262e-01 8.31098437e-01 8.62265751e-02
7.53205270e-02 -1.30170858e+00 -8.20176244e-01 -7.60179460e-01
-6.34744391e-02 3.90475094e-01 9.50480103e-01 7.71000266e-01
-6.04122400e-01 -3.16147953e-02 -2.99079716e-01 -5.35110645e-02
6.38965845e-01 2.22268894e-01 6.37647092e-01 -1.07190883e+00
-5.15401900e-01 -3.87034416e-01 -5.70657849e-01 -6.59114838e-01
8.64867866e-02 -5.12024403e-01 5.28334156e-02 -1.07971728e+00
2.13105991e-01 -8.63367856e-01 -3.56504828e-01 7.84688592e-02
-3.17935467e-01 -1.36102110e-01 -2.59272426e-01 4.31882031e-02
-4.45502609e-01 1.17745292e+00 1.42271924e+00 -1.51733950e-01
-2.15811983e-01 -8.19802731e-02 -7.18297362e-01 3.45771879e-01
7.68661857e-01 -6.49917245e-01 -7.37886846e-01 -3.76083136e-01
8.11139494e-02 1.67202000e-02 -1.80236753e-02 -1.20754242e+00
-9.24052820e-02 -1.88225716e-01 6.45852268e-01 -7.26949334e-01
-3.64063419e-02 -7.06781387e-01 2.00595140e-01 6.88342392e-01
-8.78096968e-02 3.44545245e-02 6.78623915e-02 6.09526992e-01
-3.23100030e-01 -4.11184102e-01 1.03816473e+00 7.88581967e-02
-3.16364378e-01 1.59464777e-01 -1.80017188e-01 -2.00494722e-01
1.44652700e+00 -6.08009815e-01 -1.72143847e-01 -4.85499799e-02
-2.87400752e-01 2.29016885e-01 4.57279265e-01 2.65712559e-01
6.73007965e-01 -1.53318572e+00 -7.93670952e-01 5.86409628e-01
-2.61296686e-02 -1.60773434e-02 4.64032561e-01 6.73071444e-01
-7.73803711e-01 -1.83319554e-01 -6.86588824e-01 -7.65035689e-01
-1.05351925e+00 4.43800449e-01 4.94161665e-01 -4.02297437e-01
-1.06687568e-01 1.07423401e+00 -6.91962540e-02 -6.01203799e-01
1.50077641e-01 1.32087722e-01 -3.13533783e-01 6.75971508e-02
2.67940521e-01 6.93131328e-01 8.74564275e-02 -9.78208110e-02
-2.76042312e-01 5.73221505e-01 -5.68481162e-02 2.23403111e-01
1.60981715e+00 -1.41531944e-01 -4.63980198e-01 2.17302084e-01
1.43082082e+00 -7.26531968e-02 -1.26537097e+00 4.26297575e-01
-2.30287537e-01 -9.06813323e-01 -2.26747454e-03 -8.10035408e-01
-1.40328002e+00 5.83364069e-01 1.16822910e+00 1.17625512e-01
1.54534090e+00 -5.24065912e-01 7.90736496e-01 -6.18056729e-02
2.43096948e-01 -1.24115598e+00 5.86562932e-01 1.81098171e-02
7.95901716e-01 -8.02508354e-01 1.02427691e-01 -2.23638937e-01
-6.26725495e-01 1.27830219e+00 1.02499270e+00 -4.95441794e-01
6.93264186e-01 4.13413972e-01 -9.93716791e-02 -1.46059304e-01
-6.51305079e-01 4.40540224e-01 -1.34831294e-01 6.28012240e-01
1.48028985e-01 -1.44043788e-01 -7.87390709e-01 5.60939789e-01
-6.27295151e-02 1.16542645e-01 2.82764554e-01 9.93553519e-01
-5.43979466e-01 -1.41983974e+00 -5.17619610e-01 5.60322762e-01
-1.81832910e-01 3.78237963e-01 9.51634869e-02 4.52194840e-01
8.94107878e-01 6.42049968e-01 2.58824617e-01 -5.27592361e-01
1.23625293e-01 -2.40423024e-01 6.31925106e-01 -1.41347736e-01
-4.06237721e-01 2.59599015e-02 -3.08621705e-01 -9.93075669e-02
-2.98543125e-01 -6.32072091e-01 -8.94403815e-01 -1.30988672e-01
-9.35184836e-01 3.46124649e-01 7.32008576e-01 6.08906269e-01
6.21849179e-01 8.95596683e-01 1.19134974e+00 -4.23915774e-01
-8.67391229e-01 -8.78831089e-01 -9.80439186e-01 3.00146550e-01
-1.19198196e-01 -9.16318119e-01 -3.06479990e-01 -3.41446966e-01] | [9.034748077392578, 3.7839536666870117] |
28329709-a56a-4542-9220-b3328489bf91 | complete-the-look-scene-based-complementary | 1812.01748 | null | http://arxiv.org/abs/1812.01748v2 | http://arxiv.org/pdf/1812.01748v2.pdf | Complete the Look: Scene-based Complementary Product Recommendation | Modeling fashion compatibility is challenging due to its complexity and
subjectivity. Existing work focuses on predicting compatibility between product
images (e.g. an image containing a t-shirt and an image containing a pair of
jeans). However, these approaches ignore real-world 'scene' images (e.g.
selfies); such images are hard to deal with due to their complexity, clutter,
variations in lighting and pose (etc.) but on the other hand could potentially
provide key context (e.g. the user's body type, or the season) for making more
accurate recommendations. In this work, we propose a new task called 'Complete
the Look', which seeks to recommend visually compatible products based on scene
images. We design an approach to extract training data for this task, and
propose a novel way to learn the scene-product compatibility from fashion or
interior design images. Our approach measures compatibility both globally and
locally via CNNs and attention mechanisms. Extensive experiments show that our
method achieves significant performance gains over alternative systems. Human
evaluation and qualitative analysis are also conducted to further understand
model behavior. We hope this work could lead to useful applications which link
large corpora of real-world scenes with shoppable products. | ['Jure Leskovec', 'Wang-Cheng Kang', 'Eric Kim', 'Charles Rosenberg', 'Julian McAuley'] | 2018-12-04 | complete-the-look-scene-based-complementary-1 | http://openaccess.thecvf.com/content_CVPR_2019/html/Kang_Complete_the_Look_Scene-Based_Complementary_Product_Recommendation_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Kang_Complete_the_Look_Scene-Based_Complementary_Product_Recommendation_CVPR_2019_paper.pdf | cvpr-2019-6 | ['product-recommendation'] | ['miscellaneous'] | [ 1.53545752e-01 -3.65614206e-01 -1.69497088e-01 -9.77358401e-01
-2.07048506e-01 -6.15956903e-01 4.59417373e-01 1.18789777e-01
5.74267693e-02 -6.47062138e-02 3.29607785e-01 -2.83679187e-01
-1.94971319e-02 -5.88277876e-01 -9.51597333e-01 -2.89311469e-01
2.26010382e-01 6.95678219e-02 -1.10232696e-01 -4.68118459e-01
2.39293978e-01 1.91484585e-01 -1.72025120e+00 5.82709551e-01
5.97621441e-01 1.29140043e+00 5.92263937e-01 4.36281919e-01
7.83241317e-02 5.87478876e-01 -3.11073780e-01 -6.90218270e-01
4.78815466e-01 -3.24890941e-01 -4.58024383e-01 6.29661679e-01
9.85194623e-01 -4.96600509e-01 6.33047074e-02 1.20650244e+00
2.42083222e-01 1.91938475e-01 3.20473820e-01 -1.17180181e+00
-1.50947046e+00 4.53282952e-01 -6.48582578e-01 -1.14091419e-01
6.36668801e-01 3.33154976e-01 1.01557636e+00 -6.99514151e-01
6.31592512e-01 1.39771426e+00 7.11257935e-01 2.29943752e-01
-1.15993941e+00 -3.79689425e-01 7.60752082e-01 3.25763047e-01
-1.09324872e+00 -2.96858400e-01 1.15015876e+00 -4.16836619e-01
4.65894550e-01 8.31376612e-01 8.41441393e-01 1.28974879e+00
1.39022365e-01 9.49278057e-01 1.38592935e+00 -2.57161081e-01
2.40874529e-01 4.79503781e-01 1.45252466e-01 5.97809911e-01
-4.69839014e-02 7.68712442e-03 -1.51701525e-01 1.43107608e-01
8.30743015e-01 1.90633580e-01 -6.99484274e-02 -5.09390771e-01
-1.04591238e+00 6.79485083e-01 7.13285744e-01 2.31868736e-02
-4.38271791e-01 -1.65344715e-01 1.41427502e-01 2.96587825e-01
4.23621148e-01 6.46469772e-01 -3.99751723e-01 3.36913258e-01
-5.04100204e-01 4.84469980e-01 6.71560287e-01 1.25004506e+00
3.57530385e-01 -3.45355779e-01 -1.58655733e-01 1.04701078e+00
5.02668858e-01 2.96721876e-01 6.74042478e-02 -9.01382923e-01
3.66674304e-01 4.48866755e-01 3.66537958e-01 -1.58961821e+00
-5.50557673e-01 -4.68947083e-01 -5.73035777e-01 5.43880276e-02
5.29000640e-01 1.02488555e-01 -8.15495372e-01 1.61045933e+00
3.39476705e-01 -5.62811047e-02 -4.69224691e-01 1.57016766e+00
1.08895385e+00 4.50471997e-01 1.81891605e-01 3.49249765e-02
1.85365140e+00 -1.18211532e+00 -8.83064210e-01 -3.95755380e-01
2.16323584e-01 -1.07142937e+00 1.41704535e+00 3.21360409e-01
-1.14945066e+00 -1.08042181e+00 -9.80109990e-01 -1.56756133e-01
-6.12589478e-01 3.35617810e-01 1.07221901e+00 6.08315170e-01
-9.04058814e-01 4.97031182e-01 -3.70348305e-01 -6.24186218e-01
2.94089943e-01 2.61404544e-01 -1.83539242e-01 -4.13620681e-01
-1.05644965e+00 9.26577628e-01 -1.85707286e-01 5.60053885e-01
-4.19554144e-01 -4.67188179e-01 -1.18313491e+00 -9.41639468e-02
5.27709663e-01 -8.20343614e-01 1.15392923e+00 -1.61021316e+00
-1.36345828e+00 8.76405478e-01 2.03848273e-01 7.17909709e-02
3.36193413e-01 -2.42670596e-01 -7.23371089e-01 -2.77976394e-01
1.09223127e-01 8.69453371e-01 7.90776014e-01 -1.59618449e+00
-5.20126641e-01 -5.57633400e-01 7.28163302e-01 3.19919258e-01
-3.45704824e-01 3.86989802e-01 -7.63097286e-01 -8.45829248e-01
1.00314587e-01 -1.23055053e+00 -3.62192899e-01 4.00496095e-01
-4.38571155e-01 1.35289028e-01 5.05227864e-01 -8.76716912e-01
9.05581057e-01 -2.09635425e+00 -1.45560026e-01 1.18475996e-01
1.91687029e-02 1.32978633e-01 -3.85595649e-01 9.54886973e-02
4.52203639e-02 -1.32793069e-01 3.20243925e-01 -2.61903852e-01
2.11818859e-01 1.16177976e-01 1.21299207e-01 3.81513685e-01
9.35319215e-02 1.13505077e+00 -8.23118448e-01 -2.79614270e-01
4.42494154e-01 5.70548654e-01 -6.51817620e-01 1.30076841e-01
-2.94446081e-01 3.66505742e-01 -2.55354226e-01 7.94618130e-01
9.22435701e-01 -3.26632619e-01 2.61374623e-01 -8.81823719e-01
7.30526298e-02 1.85944006e-01 -1.32793915e+00 1.75853121e+00
-5.23425043e-01 5.11120439e-01 8.35819989e-02 -6.22358143e-01
8.12209189e-01 -6.53570592e-02 1.91975743e-01 -1.12990165e+00
3.47811967e-01 -3.69301975e-01 -1.23718586e-02 -7.95708835e-01
8.76855910e-01 3.07448179e-01 -7.22738951e-02 1.17680326e-01
-4.67588127e-01 8.96381661e-02 -7.84865469e-02 -1.24185331e-01
4.56507742e-01 3.68737906e-01 9.34271235e-03 -4.15911436e-01
2.03414097e-01 -3.02975532e-02 5.63779712e-01 5.88540673e-01
-3.24240446e-01 7.21246779e-01 6.09370284e-02 -7.69986451e-01
-9.24699306e-01 -9.15862679e-01 8.08815435e-02 1.21723950e+00
7.28367686e-01 -4.77272034e-01 -6.61088645e-01 -4.78645384e-01
-3.03129889e-02 5.76933444e-01 -7.99611688e-01 -2.71691024e-01
-4.48571593e-01 -5.10381758e-01 -3.96305680e-01 8.00836265e-01
4.57618535e-01 -8.59475076e-01 -4.41563636e-01 1.01642579e-01
-2.12219328e-01 -1.27226186e+00 -7.86895514e-01 -2.34480321e-01
-6.75403178e-01 -7.76795506e-01 -4.60377455e-01 -1.14333558e+00
7.84611583e-01 7.32685506e-01 1.41605663e+00 3.78810950e-02
-1.48253679e-01 2.59032726e-01 -2.56504089e-01 -3.80568981e-01
-2.32005701e-03 -4.77697045e-01 3.72930579e-02 3.53541374e-01
4.87850845e-01 -1.82463929e-01 -1.07270181e+00 7.14891613e-01
-6.96050406e-01 4.24270451e-01 4.05324608e-01 7.07419038e-01
6.32640600e-01 2.58200437e-01 1.66333795e-01 -9.96999800e-01
7.47231185e-01 -3.96898359e-01 -3.02735984e-01 4.37198520e-01
-2.65492350e-01 -4.78754520e-01 5.15299201e-01 -6.81747437e-01
-1.22385693e+00 2.19794765e-01 4.92483787e-02 -3.60379517e-01
-4.88255382e-01 3.74324113e-01 -4.83621806e-01 1.52604625e-01
4.66661870e-01 -3.57423127e-01 -2.45281890e-01 -6.45154715e-01
6.12852395e-01 6.10214055e-01 7.01004624e-01 -5.06817043e-01
4.68805611e-01 4.03167129e-01 -4.21365947e-01 -4.19179320e-01
-9.00326014e-01 -3.60373169e-01 -5.04935682e-01 -4.99627203e-01
7.54446626e-01 -1.02425158e+00 -7.45746732e-01 2.85748214e-01
-9.42102492e-01 -9.95640606e-02 5.31974062e-02 4.66446906e-01
-3.72584939e-01 1.28191277e-01 -8.63313735e-01 -6.59977674e-01
8.41363743e-02 -1.20313096e+00 1.14074516e+00 3.60080838e-01
-5.65384030e-01 -8.79814148e-01 -3.61426502e-01 8.52204442e-01
5.13061225e-01 1.81096628e-01 7.77605474e-01 -6.33996949e-02
-4.70868260e-01 -1.58839464e-01 -3.77871960e-01 1.12110198e-01
4.34527099e-01 -5.53653762e-02 -1.01604903e+00 -1.49129361e-01
7.93638900e-02 -1.69212744e-01 4.83193994e-01 6.80789590e-01
1.40781748e+00 -4.04650509e-01 -1.52782217e-01 4.41866875e-01
1.32612169e+00 2.67651618e-01 6.53271258e-01 2.22290725e-01
1.01800215e+00 1.06346679e+00 1.02112913e+00 2.64702439e-01
7.84989476e-01 1.17498100e+00 4.57887828e-01 -6.53048337e-01
-2.51330882e-01 -3.44327539e-01 2.17832923e-01 6.49219632e-01
-1.53402209e-01 -1.86705500e-01 -2.67235458e-01 2.77408391e-01
-1.95960963e+00 -8.27844858e-01 -3.26224625e-01 1.84287810e+00
6.04757488e-01 4.57129627e-02 3.47261876e-01 -2.22824872e-01
7.02114046e-01 -6.84031546e-02 -5.63010991e-01 -6.38365805e-01
-2.24920198e-01 -2.89522141e-01 3.22608948e-01 2.60589838e-01
-1.32021248e+00 7.62366712e-01 5.73738098e+00 4.45026487e-01
-1.05009365e+00 -5.04815914e-02 1.04830956e+00 -8.93265530e-02
-5.64891338e-01 -2.67649502e-01 -4.72879380e-01 6.15388870e-01
4.66697186e-01 7.15463340e-01 5.64120770e-01 7.38840222e-01
2.59616643e-01 -1.65443331e-01 -1.26625180e+00 1.16980982e+00
3.77377152e-01 -9.53231633e-01 -2.97980905e-01 -3.31563503e-02
7.30399430e-01 -6.61895096e-01 4.55168188e-01 2.34826878e-01
3.55699360e-01 -8.45632911e-01 1.03422344e+00 3.75819594e-01
4.12061751e-01 -4.23500866e-01 5.02552450e-01 -1.71169899e-02
-1.22302151e+00 -1.49261996e-01 -4.12786424e-01 -3.91106039e-01
9.59556177e-02 3.83865714e-01 -3.19614619e-01 3.53228390e-01
9.45672572e-01 8.36700261e-01 -9.35297847e-01 8.67212117e-01
2.73040831e-01 8.74728933e-02 -1.51487648e-01 -1.46559626e-01
3.54777277e-02 -3.12685966e-01 3.57193761e-02 1.09565961e+00
9.95734483e-02 5.79582676e-02 3.84923667e-01 1.14392221e+00
1.03331558e-01 2.92068064e-01 -6.20070577e-01 1.61161408e-01
1.33044114e-02 1.48590052e+00 -8.51467013e-01 -2.08838075e-01
-8.04251730e-01 1.13963699e+00 1.68036774e-01 3.86479259e-01
-9.95041847e-01 2.50753343e-01 8.45446110e-01 2.55784512e-01
3.35098267e-01 -7.25151896e-02 -5.34518778e-01 -1.09198213e+00
2.63466328e-01 -1.01213861e+00 1.69539824e-01 -1.17250860e+00
-1.79743826e+00 4.88001287e-01 -1.68551818e-01 -1.38149512e+00
2.17361107e-01 -7.39538193e-01 -4.23582226e-01 5.53422749e-01
-1.23687446e+00 -1.52661884e+00 -5.17103791e-01 4.34953570e-01
7.47273922e-01 9.31796581e-02 6.13437772e-01 5.08357227e-01
-5.72085083e-01 7.34929204e-01 -1.39478803e-01 -1.13689609e-01
7.54215121e-01 -1.14708400e+00 6.00332856e-01 4.91683602e-01
2.25027293e-01 7.39805281e-01 9.70485747e-01 -5.74102104e-01
-1.85740423e+00 -9.01640534e-01 7.28986740e-01 -5.90797901e-01
4.07104909e-01 -6.16022706e-01 -6.29446805e-01 5.31349719e-01
3.26965928e-01 -1.20046161e-01 8.21683168e-01 4.81155872e-01
-5.87556124e-01 -2.25820288e-01 -1.08793831e+00 1.11802530e+00
1.21061277e+00 -2.83886492e-01 -2.48727232e-01 4.54324126e-01
5.46059430e-01 -4.87454027e-01 -9.23309624e-01 3.47589910e-01
9.64238882e-01 -1.00584114e+00 1.19766307e+00 -6.60533786e-01
6.63128853e-01 -2.64051646e-01 -3.50696892e-01 -1.40411270e+00
-8.23066592e-01 -1.95883483e-01 1.71706080e-01 1.37245214e+00
2.98682064e-01 -1.57871783e-01 5.26329815e-01 1.26923752e+00
-1.24144135e-02 -8.55720401e-01 -7.64662772e-02 -5.42888761e-01
-4.84686762e-01 -4.57584709e-01 9.24086571e-01 1.23526442e+00
-1.42689377e-01 4.11625922e-01 -7.20944345e-01 2.38132268e-01
2.34732240e-01 5.86796999e-01 7.20775664e-01 -9.24591064e-01
-5.54882824e-01 -4.00622725e-01 -3.68864864e-01 -1.02198136e+00
-1.12415403e-01 -5.57676911e-01 7.28679299e-02 -1.41695702e+00
3.49166006e-01 -5.87841451e-01 -3.93358350e-01 2.51821399e-01
-3.47746402e-01 5.89674234e-01 4.88138855e-01 -2.21558288e-01
-8.07150483e-01 2.16123059e-01 1.59233260e+00 -4.35387999e-01
-1.72704950e-01 -8.41323361e-02 -1.17704332e+00 7.03598857e-01
8.85343611e-01 8.13157931e-02 -5.98277807e-01 -6.04028583e-01
2.80125082e-01 -6.27411306e-02 6.58022821e-01 -6.57815158e-01
6.09543780e-03 -4.32702273e-01 8.44296217e-01 -2.95207709e-01
5.07127643e-01 -1.10094631e+00 4.60499585e-01 5.96738942e-02
-5.61553061e-01 4.35891837e-01 1.60354853e-01 5.71017206e-01
7.82206059e-02 5.44036217e-02 3.67353648e-01 -2.99538821e-01
-8.49181294e-01 1.19512439e-01 -1.10511668e-02 -5.34336150e-01
8.22933733e-01 -2.64908314e-01 -3.09076071e-01 -5.67908108e-01
-6.68146849e-01 2.09518462e-01 6.90537453e-01 1.13840580e+00
5.09767354e-01 -1.83312881e+00 -4.32869077e-01 2.21765712e-01
3.90691787e-01 -5.35839260e-01 5.28746605e-01 6.32945776e-01
-2.62852877e-01 1.65231541e-01 -6.08470678e-01 -6.34171486e-01
-1.39171183e+00 9.80670393e-01 1.01143092e-01 2.09280118e-01
-4.85526919e-01 7.43452668e-01 6.30822599e-01 -4.00191486e-01
3.51698130e-01 -5.53501487e-01 -3.67644042e-01 -1.45346507e-01
5.11231184e-01 -5.20450510e-02 -1.54644381e-02 -8.11826169e-01
-1.11719862e-01 7.30214357e-01 -2.96624333e-01 2.83188254e-01
1.04290116e+00 -5.46551168e-01 2.62313664e-01 3.45390677e-01
1.18500125e+00 -2.42935300e-01 -1.40704286e+00 -3.39783967e-01
-2.72653729e-01 -1.01832914e+00 3.82523127e-02 -1.11802697e+00
-1.39563096e+00 5.86738050e-01 1.00227773e+00 4.50236984e-02
1.25327265e+00 5.21077067e-02 9.03612316e-01 5.32821268e-02
3.03192168e-01 -1.32965100e+00 -1.71626545e-02 -2.11234272e-01
1.06742573e+00 -1.68226683e+00 -8.29398930e-02 -7.01877356e-01
-9.21710074e-01 7.42136836e-01 9.53427732e-01 8.76888856e-02
5.99595070e-01 2.70856102e-03 3.85601133e-01 -3.92681092e-01
-5.61998188e-01 -1.90140784e-01 6.78186297e-01 6.99017107e-01
5.91180921e-01 4.92285818e-01 -1.82717472e-01 9.26308751e-01
-3.15851271e-01 -1.53401718e-01 1.10032469e-01 6.78491712e-01
1.83286630e-02 -9.60785866e-01 -3.45662743e-01 5.37902772e-01
-1.48174226e-01 1.31854966e-01 -4.79762018e-01 4.35835749e-01
4.30189013e-01 1.27388644e+00 2.15777725e-01 -6.37789726e-01
7.31735945e-01 -4.93470907e-01 7.61508465e-01 -3.95132929e-01
-9.02498364e-01 2.77227789e-01 2.95764029e-01 -6.70173824e-01
-5.15322685e-01 -5.95100224e-01 -4.85591352e-01 -4.63974833e-01
-2.79203624e-01 -5.14543653e-01 8.21402133e-01 7.02860951e-01
3.42256725e-01 4.94843900e-01 5.87750494e-01 -9.36275363e-01
-1.80464804e-01 -6.09667838e-01 -5.15503943e-01 1.08824682e+00
1.94261968e-02 -6.97971702e-01 3.69898528e-01 3.27297866e-01] | [11.088159561157227, 0.08775128424167633] |
febd8ee4-6d6c-4dc3-bd42-26439e0ed80b | improving-automated-program-repair-with | 2212.11414 | null | https://arxiv.org/abs/2212.11414v1 | https://arxiv.org/pdf/2212.11414v1.pdf | Improving Automated Program Repair with Domain Adaptation | Automated Program Repair (APR) is defined as the process of fixing a bug/defect in the source code, by an automated tool. APR tools have recently experienced promising results by leveraging state-of-the-art Neural Language Processing (NLP) techniques. APR tools such as TFix and CodeXGLUE combine text-to-text transformers with software-specific techniques are outperforming alternatives, these days. However, in most APR studies the train and test sets are chosen from the same set of projects. In reality, however, APR models are meant to be generalizable to new and different projects. Therefore, there is a potential threat that reported APR models with high effectiveness perform poorly when the characteristics of the new project or its bugs are different than the training set's(Domain Shift). In this study, we first define and measure the domain shift problem in automated program repair. Then, we then propose a domain adaptation framework that can adapt an APR model for a given target project. We conduct an empirical study with three domain adaptation methods FullFineTuning, TuningWithLightWeightAdapterLayers, and CurriculumLearning using two state-of-the-art domain adaptation tools (TFix and CodeXGLUE) and two APR models on 611 bugs from 19 projects. The results show that our proposed framework can improve the effectiveness of TFix by 13.05% and CodeXGLUE by 23.4%. Another contribution of this study is the proposal of a data synthesis method to address the lack of labelled data in APR. We leverage transformers to create a bug generator model. We use the generated synthetic data to domain adapt TFix and CodeXGLUE on the projects with no data (Zero-shot learning), which results in an average improvement of 5.76% and 24.42% for TFix and CodeXGLUE, respectively. | ['Hadi Hemati', 'Armin Zirak'] | 2022-12-21 | null | null | null | null | ['program-repair', 'program-repair'] | ['computer-code', 'reasoning'] | [-9.69913974e-02 7.79239759e-02 -1.80598393e-01 -2.07200095e-01
-7.13968992e-01 -4.29836094e-01 1.68193504e-01 1.12275295e-01
-1.21467710e-01 6.87913239e-01 -5.63057661e-02 -3.29160959e-01
-5.89732267e-02 -7.84364879e-01 -9.81221139e-01 -2.40266532e-01
5.64658865e-02 2.82658070e-01 3.63569319e-01 -3.58815223e-01
5.17823040e-01 -1.21544860e-01 -1.50010288e+00 4.50225502e-01
1.19454706e+00 3.65328074e-01 5.25663257e-01 6.29471600e-01
-4.07301992e-01 5.03036737e-01 -1.05498922e+00 -3.42101485e-01
1.71680197e-01 -2.89580286e-01 -7.84483850e-01 -4.66302842e-01
1.95428386e-01 2.07296461e-02 -7.83653185e-02 1.22391987e+00
5.66435337e-01 -1.14537753e-01 4.03293312e-01 -1.21409917e+00
-1.06598175e+00 9.10638630e-01 -6.25104368e-01 3.29441458e-01
4.96231824e-01 2.35744834e-01 6.52814925e-01 -9.90006685e-01
6.37234986e-01 1.01254845e+00 1.00842464e+00 8.09056282e-01
-1.21915889e+00 -8.95661771e-01 -4.93509881e-02 1.59104049e-01
-1.27728760e+00 -1.14059508e-01 7.58702278e-01 -9.08836663e-01
1.59005237e+00 -2.79153824e-01 2.43661970e-01 1.35760391e+00
6.23971879e-01 3.12670082e-01 6.96187913e-01 -8.30940545e-01
3.96678895e-01 3.26644540e-01 4.61637497e-01 7.63772070e-01
2.91947186e-01 1.83454975e-01 -3.61499369e-01 -3.55662435e-01
3.28581274e-01 -8.39334875e-02 -3.28268111e-01 -6.75208960e-03
-8.88902366e-01 7.32358456e-01 5.03788479e-02 5.37014604e-01
-3.12573463e-01 5.92847504e-02 6.88922167e-01 6.99329495e-01
2.97787160e-01 9.15137649e-01 -8.43040586e-01 -4.52453226e-01
-6.55551136e-01 4.38052237e-01 7.70966530e-01 1.40749621e+00
8.24037552e-01 1.99501500e-01 -1.57101452e-01 9.38308001e-01
3.05521667e-01 2.91641086e-01 1.05282784e+00 -5.67158401e-01
7.33979285e-01 9.37196851e-01 -2.90608536e-02 -9.13153291e-01
-1.70024171e-01 -1.49792090e-01 -3.63153279e-01 2.04730168e-01
5.21142967e-02 -4.15902734e-01 -1.07454467e+00 1.60565984e+00
-2.43323427e-02 2.16256246e-01 1.57987624e-01 3.38536680e-01
8.70133221e-01 8.12708437e-01 2.47586216e-03 7.74330739e-03
1.09191477e+00 -9.59412396e-01 -5.80063820e-01 -4.35590506e-01
9.43061531e-01 -6.93354249e-01 1.59923422e+00 6.85612559e-01
-6.81016386e-01 -6.35223567e-01 -1.32278717e+00 4.47746962e-01
-4.10964638e-01 1.24282241e-01 3.36124271e-01 7.43657470e-01
-9.02098656e-01 5.18542886e-01 -7.32810318e-01 -6.15404785e-01
1.16233476e-01 2.22658023e-01 -2.41158336e-01 -2.60382712e-01
-1.08856606e+00 8.59251201e-01 7.24454761e-01 -5.41834891e-01
-1.18621826e+00 -1.20418131e+00 -8.24124157e-01 1.85446054e-01
4.26304281e-01 -3.95479649e-01 1.56114793e+00 -8.18142951e-01
-1.42575598e+00 4.27700907e-01 2.41800517e-01 -3.63135219e-01
-1.10105336e-01 -2.01785818e-01 -5.53619683e-01 -3.18873107e-01
1.67873979e-01 1.68417305e-01 6.57890975e-01 -9.56409693e-01
-6.58438087e-01 -1.42640891e-02 3.97369415e-02 -5.14691949e-01
-6.16077781e-01 4.35532749e-01 -2.37031907e-01 -7.35004187e-01
-4.06550616e-01 -6.58797979e-01 -4.28351350e-02 -5.37730038e-01
-6.59585297e-02 -1.97687641e-01 6.10813618e-01 -9.79007185e-01
1.78212786e+00 -2.20443559e+00 2.03041062e-01 -1.03015229e-01
-5.90878539e-02 4.83991295e-01 -5.45503080e-01 5.49359798e-01
-5.03066778e-01 2.40783036e-01 -4.65018958e-01 -8.18877760e-03
-4.31120992e-02 1.80796206e-01 -3.09727490e-01 -2.14524996e-02
5.32976687e-01 4.91765171e-01 -9.06533122e-01 -3.49981189e-01
-2.80187070e-01 -5.50729148e-02 -9.28903520e-01 3.85995746e-01
-4.95087147e-01 -3.10366936e-02 -4.16756123e-01 9.09656227e-01
5.20681262e-01 2.75069594e-01 -2.35707983e-01 4.23322499e-01
-1.62513867e-01 4.06848863e-02 -1.00434959e+00 1.96801758e+00
-5.98163009e-01 3.47700000e-01 -4.21162337e-01 -8.71585846e-01
1.46379530e+00 4.81629789e-01 8.05258825e-02 -5.15951991e-01
-1.88345730e-01 2.52834052e-01 7.97170997e-02 -9.66971099e-01
4.45104122e-01 -5.80762560e-03 -5.02871573e-01 3.66081476e-01
6.06438220e-01 -1.04680583e-01 2.40093589e-01 -8.93136859e-02
1.85589981e+00 2.39567503e-01 2.90373534e-01 -1.21172123e-01
4.34808612e-01 1.78854525e-01 8.60024035e-01 6.65878594e-01
-1.01998277e-01 5.94381452e-01 7.38640249e-01 -4.79952902e-01
-1.12409842e+00 -7.11282790e-01 1.98232681e-01 1.04630566e+00
-4.08143133e-01 -5.41096628e-01 -9.60778832e-01 -9.25979555e-01
-8.92637894e-02 1.07571828e+00 -4.90594566e-01 -6.40203357e-01
-5.27867556e-01 -7.35492587e-01 8.56989980e-01 4.01836544e-01
3.43463033e-01 -1.25962150e+00 -5.54230392e-01 4.69767958e-01
1.57162745e-03 -6.11127019e-01 -4.00042921e-01 3.11474234e-01
-6.97057903e-01 -1.02364969e+00 -4.57404584e-01 -9.67559040e-01
6.21285558e-01 -4.29983884e-02 1.10505128e+00 1.99378923e-01
-2.67205864e-01 1.39641121e-01 -7.72863448e-01 -4.71613050e-01
-8.78028512e-01 2.01410949e-01 7.74501935e-02 -5.55355608e-01
6.36226416e-01 -5.18891871e-01 1.40911579e-01 1.27395660e-01
-9.37445998e-01 -5.32237291e-01 6.93555593e-01 1.05162287e+00
1.67307943e-01 2.75589377e-01 8.34319651e-01 -8.81892502e-01
1.16244876e+00 -8.10100853e-01 -6.73479736e-01 6.06998324e-01
-9.60575044e-01 1.59697652e-01 7.68165052e-01 -6.72246993e-01
-1.28737438e+00 -5.92953749e-02 -4.00307626e-02 -4.86772597e-01
-1.85636550e-01 9.26628530e-01 -1.41793698e-01 -1.65065750e-01
1.36954451e+00 4.80049886e-02 -2.82562107e-01 -5.92831612e-01
1.79068577e-02 9.06296015e-01 4.48003560e-01 -1.04280639e+00
7.46911645e-01 -6.39693677e-01 -6.81206822e-01 -4.14269120e-01
-2.37843037e-01 -3.33632141e-01 -3.62413496e-01 1.18959174e-01
5.89206755e-01 -7.17392027e-01 -3.33678313e-02 5.13529181e-01
-1.62378633e+00 -3.78030181e-01 -1.05304800e-01 2.11990878e-01
-3.76734465e-01 3.43915820e-01 -4.58539516e-01 -4.09625977e-01
-4.99476165e-01 -1.24572206e+00 5.97236454e-01 2.24619865e-01
-4.07716751e-01 -6.64740860e-01 5.25380492e-01 -8.02887008e-02
3.73521030e-01 2.52406567e-01 1.41909456e+00 -8.68353665e-01
-2.06424549e-01 -2.23709151e-01 8.22285786e-02 6.41700268e-01
-7.52461283e-03 3.30487788e-01 -6.82418883e-01 -1.64907143e-01
2.00721607e-01 -1.90125912e-01 3.04137588e-01 7.12191761e-02
8.72469306e-01 -2.43088782e-01 -2.89104730e-01 3.90557021e-01
1.68950856e+00 7.27659643e-01 6.82496846e-01 6.01139069e-01
3.33279401e-01 4.21202689e-01 1.00642514e+00 5.30181468e-01
1.16187908e-01 6.36233985e-01 2.99039364e-01 5.14013708e-01
-9.69963670e-02 -3.02742481e-01 7.57549822e-01 9.44447756e-01
1.61438882e-01 -4.32214998e-02 -1.64895296e+00 9.54866111e-01
-1.73699760e+00 -7.21916139e-01 1.23534426e-02 2.12964249e+00
1.08510172e+00 2.48202905e-01 -2.20322907e-01 -3.49113606e-02
7.28035629e-01 -5.82158029e-01 -3.07827622e-01 -8.28327596e-01
4.60059911e-01 5.34029186e-01 1.62943795e-01 2.11529404e-01
-7.59767771e-01 1.00328267e+00 5.62645197e+00 6.69811964e-01
-8.61766636e-01 3.62769634e-01 -1.19685829e-02 1.61216035e-01
-1.44535869e-01 6.82546571e-02 -8.77799451e-01 5.87976754e-01
1.31540000e+00 -5.21381378e-01 5.03846586e-01 1.29298294e+00
6.42345520e-03 1.27478048e-01 -1.25857842e+00 6.46831751e-01
2.30019629e-01 -1.13014400e+00 -2.99315959e-01 -3.08218718e-01
8.72752428e-01 -7.61736259e-02 -9.56080481e-02 1.09775162e+00
3.49532396e-01 -8.49099338e-01 5.30861974e-01 5.26349664e-01
6.00571454e-01 -8.07319999e-01 1.12099588e+00 5.68775773e-01
-7.62595594e-01 -5.66474617e-01 -6.91072464e-01 -5.13999686e-02
-3.68826509e-01 4.68611062e-01 -1.26253140e+00 5.86339474e-01
1.11065471e+00 5.79116762e-01 -7.71661103e-01 1.24064422e+00
-3.74885917e-01 7.21046805e-01 1.25383094e-01 -9.35554504e-02
-7.01970607e-02 2.23567724e-01 4.67215091e-01 1.35218322e+00
8.27647209e-01 -1.84444949e-01 3.68882529e-02 1.29642713e+00
2.22641453e-02 -9.47511494e-02 -8.55023921e-01 -1.73385516e-01
8.00390363e-01 8.14354002e-01 -6.90122098e-02 -1.31785274e-01
-5.56745887e-01 5.17208636e-01 2.94763923e-01 3.71740937e-01
-9.18929100e-01 -9.60383534e-01 6.19762659e-01 -1.29364943e-02
3.08916539e-01 4.95919026e-02 -4.23914850e-01 -8.59777510e-01
2.69238561e-01 -1.35673630e+00 1.25836492e-01 -7.34290421e-01
-1.17483318e+00 6.51534975e-01 2.91864365e-01 -1.22023416e+00
-2.46187225e-01 -3.69359285e-01 -9.06439006e-01 9.59014833e-01
-1.25807786e+00 -8.65358353e-01 -3.91194880e-01 1.43797338e-01
8.05812657e-01 -8.59706521e-01 9.08395350e-01 4.57504153e-01
-7.16372013e-01 9.43448901e-01 4.49882783e-02 -1.03554532e-01
8.95029128e-01 -1.16945267e+00 6.34636462e-01 1.22095883e+00
-3.49051505e-01 1.00125074e+00 7.59282112e-01 -9.33358133e-01
-1.32676220e+00 -1.26167893e+00 1.00850439e+00 -5.22762537e-01
8.12192261e-01 -3.89642537e-01 -1.36004972e+00 7.19492495e-01
2.29452044e-01 -1.43640906e-01 3.79585177e-01 5.09487614e-02
-3.82670224e-01 -4.38999608e-02 -1.48745346e+00 3.80044997e-01
7.09586024e-01 -3.35501611e-01 -1.02422988e+00 1.89528137e-01
1.17703402e+00 -3.70251000e-01 -1.02762437e+00 2.31897920e-01
1.96978170e-02 -8.31417918e-01 5.96976221e-01 -6.00508153e-01
7.44748414e-01 -3.65317911e-01 -5.05969822e-02 -1.53492165e+00
-4.40426171e-01 -5.17358899e-01 -1.42108127e-01 1.61462700e+00
5.21461666e-01 -5.92621803e-01 5.62297583e-01 5.50558269e-01
-6.77968860e-01 -4.65163618e-01 -8.89740229e-01 -1.00419307e+00
3.02128196e-01 -4.89489287e-01 8.59238684e-01 1.06462765e+00
3.00525457e-01 1.01949513e-01 -1.38751995e-02 2.88704664e-01
1.22758541e-02 -5.40083885e-01 8.21411669e-01 -1.16963267e+00
-5.90940356e-01 -1.61447451e-01 -4.63347346e-01 -3.13791364e-01
3.47288013e-01 -7.47967601e-01 4.11151737e-01 -1.10746098e+00
1.58296689e-01 -2.72907525e-01 -2.81323373e-01 9.98127699e-01
-8.88576806e-02 -6.22650146e-01 -1.17232755e-01 -1.40649542e-01
-1.46278311e-02 4.03930902e-01 6.65791214e-01 -4.63638514e-01
-5.75942755e-01 -2.72565018e-02 -6.76014781e-01 5.54788053e-01
8.34123731e-01 -9.85098720e-01 -5.68627417e-01 -5.71979463e-01
3.09921741e-01 1.44579625e-02 6.54441640e-02 -1.21854186e+00
2.82164216e-01 -8.27311203e-02 -1.90491319e-01 -5.93950525e-02
-4.19768602e-01 -7.17686832e-01 2.23599821e-01 5.69018364e-01
-1.35313541e-01 2.75719404e-01 5.73519289e-01 5.04734755e-01
-2.47589469e-01 -1.08387685e+00 5.25352418e-01 -1.28821731e-01
-9.81783092e-01 -1.33941874e-01 -4.05265123e-01 -3.60061526e-02
1.28160179e+00 -2.61813253e-01 -5.70194364e-01 2.95450330e-01
-2.22232774e-01 1.07758418e-02 4.90435064e-01 8.13204169e-01
8.25715601e-01 -1.15996337e+00 -6.14405930e-01 2.89810449e-01
6.40573800e-01 3.42300236e-02 1.32274985e-01 4.91101891e-01
-5.16628265e-01 3.36520255e-01 -3.43038708e-01 -3.12818617e-01
-1.21053815e+00 7.61897087e-01 2.20507115e-01 -3.25325221e-01
-4.68985826e-01 8.54878128e-01 -1.04419649e-01 -8.93687129e-01
2.37028897e-01 -7.83198178e-01 -1.11079559e-01 -3.81524533e-01
4.81831729e-01 2.30701968e-01 3.86825472e-01 2.30089352e-01
-2.41398737e-01 4.29389626e-01 -3.22290719e-01 2.20164493e-01
1.47928703e+00 6.52146816e-01 -1.00743636e-01 2.20154583e-01
8.55269134e-01 -2.62190193e-01 -9.00730252e-01 -2.04701096e-01
5.63078642e-01 -4.80656326e-01 -1.73286259e-01 -1.07148373e+00
-8.00578356e-01 7.81713545e-01 7.68434107e-01 -1.39547572e-01
1.18074894e+00 -1.74061760e-01 5.58557808e-01 5.54144681e-01
6.94619834e-01 -1.15911913e+00 2.83779114e-01 6.96474254e-01
1.02639592e+00 -1.11904204e+00 -3.09239179e-01 -1.62634000e-01
-5.64195633e-01 1.24756360e+00 1.11338651e+00 -6.41132295e-02
3.44423145e-01 4.80751067e-01 -2.93050706e-01 -4.97599840e-02
-1.02359450e+00 5.19879818e-01 -5.75445481e-02 1.04416585e+00
4.92636144e-01 -2.43747085e-01 -3.51649731e-01 1.23197901e+00
-1.17195416e-02 3.75632554e-01 9.29891348e-01 1.18205059e+00
-5.96972227e-01 -1.42701221e+00 -5.04835010e-01 2.82523990e-01
-2.10037753e-01 -1.45557389e-01 -2.63480306e-01 8.55274796e-01
4.66597468e-01 8.61385345e-01 -4.15104568e-01 -8.21420133e-01
7.83623993e-01 3.28599542e-01 2.41560966e-01 -1.39934874e+00
-1.04627979e+00 -3.65622461e-01 4.99647111e-02 -2.57018089e-01
1.41393766e-01 -5.03667891e-01 -1.21365547e+00 -4.79698069e-02
-3.25759411e-01 1.61696583e-01 6.63066447e-01 6.57944322e-01
5.57610095e-01 8.49853456e-01 3.53353769e-01 -2.36991569e-01
-8.97796631e-01 -1.20434785e+00 -2.47282356e-01 -4.52967398e-02
7.60231242e-02 -7.55270898e-01 -1.43388242e-01 2.56799251e-01] | [7.633476734161377, 7.752171516418457] |
6b26dbf7-ca59-4c21-9386-75289ce37f34 | towards-better-dermoscopic-image-feature | 2207.07303 | null | https://arxiv.org/abs/2207.07303v1 | https://arxiv.org/pdf/2207.07303v1.pdf | Towards Better Dermoscopic Image Feature Representation Learning for Melanoma Classification | Deep learning-based melanoma classification with dermoscopic images has recently shown great potential in automatic early-stage melanoma diagnosis. However, limited by the significant data imbalance and obvious extraneous artifacts, i.e., the hair and ruler markings, discriminative feature extraction from dermoscopic images is very challenging. In this study, we seek to resolve these problems respectively towards better representation learning for lesion features. Specifically, a GAN-based data augmentation (GDA) strategy is adapted to generate synthetic melanoma-positive images, in conjunction with the proposed implicit hair denoising (IHD) strategy. Wherein the hair-related representations are implicitly disentangled via an auxiliary classifier network and reversely sent to the melanoma-feature extraction backbone for better melanoma-specific representation learning. Furthermore, to train the IHD module, the hair noises are additionally labeled on the ISIC2020 dataset, making it the first large-scale dermoscopic dataset with annotation of hair-like artifacts. Extensive experiments demonstrate the superiority of the proposed framework as well as the effectiveness of each component. The improved dataset publicly avaliable at https://github.com/kirtsy/DermoscopicDataset. | ['Xiu Li', 'Xiao-jun Zeng', 'Yu Yang', 'HanMo Chen', 'Jiangpeng Yan', 'Zhe Xu', 'Mingqing Wang', 'ShengGe Yang', 'Mingkang Tang', 'Chenghui Yu'] | 2022-07-15 | null | null | null | null | ['melanoma-diagnosis'] | ['computer-vision'] | [ 6.96221650e-01 2.50289381e-01 -1.45155370e-01 -1.36758834e-01
-9.36045349e-01 -1.76568896e-01 6.12737298e-01 1.08005717e-01
-6.52386099e-02 6.91369653e-01 1.34125546e-01 2.76913159e-02
-5.02162129e-02 -6.78692818e-01 -4.95822847e-01 -1.12207675e+00
4.49923456e-01 -5.50719649e-02 -2.67566949e-01 -6.74009100e-02
9.02131572e-02 2.91559815e-01 -1.18006134e+00 1.69655383e-01
1.34089255e+00 1.02901006e+00 -2.98275263e-03 4.65542346e-01
6.04303814e-02 7.39233375e-01 -3.42996269e-01 -6.64301157e-01
1.70978248e-01 -5.30234456e-01 -5.92953503e-01 4.30059612e-01
3.54424685e-01 -3.15027177e-01 -2.80907631e-01 1.10727382e+00
5.89022458e-01 -4.67122912e-01 7.07481205e-01 -1.01883805e+00
-5.79412580e-01 -7.18768612e-02 -1.07947135e+00 -2.41084397e-01
1.14922509e-01 4.40350920e-01 6.45899475e-01 -6.61302567e-01
7.68837929e-01 7.14144945e-01 6.20896041e-01 8.63689244e-01
-1.23605299e+00 -6.69119000e-01 -9.93001163e-02 1.16837405e-01
-1.26035488e+00 -3.19773734e-01 9.75461841e-01 -4.93759900e-01
6.20594621e-02 5.90403438e-01 8.12046289e-01 1.56652439e+00
7.32280463e-02 8.81909192e-01 1.49483514e+00 -3.39390099e-01
-1.69766620e-02 2.68438518e-01 -3.24465297e-02 8.22522104e-01
2.80344695e-01 2.07257476e-02 -3.80765021e-01 3.21881287e-02
7.79001951e-01 1.64079979e-01 -2.51664519e-01 -1.26257792e-01
-7.61631131e-01 6.18991733e-01 6.68070495e-01 -7.30395922e-03
-3.69642586e-01 2.99627297e-02 4.35751349e-01 1.33906445e-02
6.79851413e-01 1.98997557e-01 2.32683718e-01 3.79616559e-01
-8.14455688e-01 1.01414192e-02 3.83566916e-01 5.52322447e-01
5.49362600e-01 -1.03681326e-01 -2.61237651e-01 1.04883742e+00
2.03815028e-01 2.65832216e-01 3.63039583e-01 -4.45523202e-01
3.75325710e-01 8.07279348e-01 -1.94077432e-01 -8.31894457e-01
-2.71387368e-01 -6.35617971e-01 -1.33676636e+00 1.70637742e-01
4.11614776e-01 5.30329440e-03 -1.07770896e+00 1.47623956e+00
6.45856977e-01 2.60490716e-01 -1.27344161e-01 1.04612398e+00
7.72814810e-01 -1.57060623e-02 3.28109533e-01 -7.10947216e-02
1.34299076e+00 -8.02255332e-01 -8.08339894e-01 4.79685888e-02
5.53257287e-01 -7.81737387e-01 1.00979459e+00 4.60256755e-01
-7.66790271e-01 -3.13776344e-01 -1.02125502e+00 -1.97525457e-01
-5.90065382e-02 4.70269948e-01 8.23691964e-01 5.04372716e-01
-5.87760210e-01 3.31649572e-01 -9.72691953e-01 -3.51623327e-01
8.26056659e-01 2.32653528e-01 -4.19530004e-01 -2.18391836e-01
-9.93164718e-01 6.16425455e-01 -6.96566552e-02 3.94693494e-01
-9.11503077e-01 -7.65749693e-01 -6.96203589e-01 -6.58434033e-01
2.81808972e-01 -7.07638562e-01 7.18507886e-01 -1.11167204e+00
-1.38632131e+00 1.01300991e+00 -4.60646339e-02 -3.24049681e-01
7.68227160e-01 -9.74945817e-03 -4.86944944e-01 2.92722523e-01
-9.51713547e-02 5.03060222e-01 1.10594964e+00 -1.36391270e+00
-3.53062123e-01 -7.13110745e-01 -1.33334547e-01 1.35400400e-01
-5.39565802e-01 -4.29236889e-01 -3.30638260e-01 -8.67093384e-01
-1.63665429e-01 -9.51946199e-01 -2.95946449e-01 4.82885182e-01
-9.50725555e-01 1.74400777e-01 4.22150344e-01 -1.22260106e+00
1.18161821e+00 -2.16427588e+00 1.61989927e-01 3.99502069e-01
4.85378265e-01 3.32287371e-01 -2.93513328e-01 2.86760777e-01
-8.62988308e-02 1.23705857e-01 -3.07550520e-01 -3.73448819e-01
-3.60748112e-01 -1.97929382e-01 2.26417676e-01 6.24771953e-01
7.77147710e-01 9.03418243e-01 -7.20784903e-01 -5.43435872e-01
2.59119064e-01 6.97197855e-01 -1.73795134e-01 6.06498495e-02
-1.07048564e-02 6.95230126e-01 -4.65413064e-01 9.98877227e-01
9.54326093e-01 -1.70428723e-01 1.86330840e-01 -5.24381042e-01
2.16338232e-01 2.04463899e-02 -7.65969515e-01 1.95080113e+00
-6.77552402e-01 1.78943649e-01 1.84008911e-01 -4.67429489e-01
7.94722855e-01 3.04386199e-01 3.41631562e-01 -7.84723341e-01
-5.42340893e-03 3.15250516e-01 6.00101873e-02 -8.33021700e-01
2.88250744e-02 -2.73756713e-01 2.62405127e-01 9.54411700e-02
-7.93054774e-02 8.95028189e-03 -1.36680361e-02 1.63574740e-01
1.03272355e+00 8.33477750e-02 2.89879799e-01 -1.59541275e-02
5.52647889e-01 2.14677360e-02 4.25794423e-01 4.90753986e-02
-1.64413169e-01 7.13973463e-01 4.28247571e-01 -1.56921104e-01
-8.90420794e-01 -9.10263896e-01 -3.09637249e-01 4.30552989e-01
2.13677570e-01 -9.87707675e-02 -8.10198605e-01 -9.04127061e-01
-1.40197471e-01 3.93399537e-01 -1.10508168e+00 -2.39107192e-01
-1.31882817e-01 -1.05838192e+00 6.12105548e-01 4.09077317e-01
4.32606369e-01 -6.17733240e-01 -1.83652967e-01 -1.45753488e-01
-5.48139811e-02 -4.91934508e-01 -3.53532761e-01 -1.11181559e-02
-6.05133832e-01 -1.45558393e+00 -9.56665695e-01 -4.87457573e-01
1.06778514e+00 1.15689129e-01 5.18671393e-01 1.80125043e-01
-1.02571988e+00 1.45154297e-02 -2.99313039e-01 -2.85826117e-01
-4.45759654e-01 -3.69681641e-02 -4.22377914e-01 4.74815547e-01
2.69372076e-01 -3.60986888e-01 -1.09658968e+00 -1.40727118e-01
-1.02896559e+00 3.94606918e-01 1.03834724e+00 1.17059898e+00
8.28879654e-01 -1.37996003e-01 5.53714097e-01 -1.18559563e+00
5.91298699e-01 -5.86978614e-01 -1.40367463e-01 2.61135280e-01
-4.27053839e-01 -3.45531523e-01 5.14110327e-01 -3.47508013e-01
-1.40294087e+00 1.59587696e-01 -3.71647865e-01 -3.30438465e-01
-2.66565770e-01 3.15422654e-01 -2.56876349e-01 -2.49005631e-01
6.85193777e-01 2.35108599e-01 4.13216382e-01 -3.59764040e-01
1.34012297e-01 7.26685762e-01 3.33903641e-01 -4.69905175e-02
9.16126311e-01 5.47668338e-01 1.77368104e-01 -7.35596716e-01
-9.28880453e-01 -4.04510677e-01 -4.93207723e-01 -1.21752262e-01
7.34163642e-01 -1.05885649e+00 -2.56596833e-01 8.48133862e-01
-7.85934567e-01 -1.59115091e-01 -3.47197086e-01 1.62880287e-01
-2.13382065e-01 4.85272735e-01 -7.89381027e-01 -9.61421072e-01
-5.87058485e-01 -9.47075605e-01 1.13887334e+00 3.30316603e-01
-2.68285692e-01 -1.00684679e+00 1.39471754e-01 7.98681617e-01
1.88182190e-01 7.89295495e-01 8.99643719e-01 -3.48859578e-01
-3.31372559e-01 -4.24248636e-01 -2.94663966e-01 6.09325647e-01
3.70076716e-01 8.34578276e-02 -1.24679947e+00 -3.21479023e-01
-4.17609513e-03 -6.73298538e-01 1.03411078e+00 2.04483755e-02
1.46750820e+00 -3.67723346e-01 -1.87096760e-01 8.00833881e-01
1.55352485e+00 -1.71584442e-01 7.24789023e-01 -6.84602559e-03
8.53462338e-01 8.24462175e-01 7.25610197e-01 3.80383432e-01
2.17956096e-01 3.61083031e-01 5.99756122e-01 -6.58544123e-01
-7.32301354e-01 -4.40274179e-01 -8.02836940e-02 7.65670002e-01
-2.26417273e-01 -1.19787723e-01 -5.71666658e-01 4.87907499e-01
-1.35739970e+00 -3.85528684e-01 -3.34246814e-01 2.21720433e+00
9.02294159e-01 -1.67384788e-01 9.25347805e-02 2.43207887e-01
6.65075541e-01 1.04664423e-01 -8.88966382e-01 1.39834061e-01
-1.57423273e-01 4.12151545e-01 2.43317828e-01 2.81981796e-01
-1.02042115e+00 5.14224887e-01 4.42225838e+00 1.18944025e+00
-1.19165444e+00 2.81269997e-01 8.28929901e-01 -1.42927140e-01
-6.49629831e-01 -2.67193913e-01 -3.56725365e-01 4.67890114e-01
3.40164036e-01 2.51496792e-01 3.33367676e-01 2.57225692e-01
9.50880274e-02 -3.18399966e-02 -6.53224051e-01 8.47413659e-01
9.14020017e-02 -1.10011518e+00 1.26634613e-01 3.25220495e-01
5.92773557e-01 -4.36779946e-01 4.11605150e-01 -2.64071196e-01
-3.13489228e-01 -1.06982148e+00 2.14148358e-01 8.45904112e-01
1.35324466e+00 -7.94770420e-01 7.25052118e-01 -3.18721542e-03
-7.76242554e-01 1.06915951e-01 -1.29258841e-01 3.88267189e-01
-2.45711893e-01 8.83988678e-01 -9.16155040e-01 8.29602897e-01
1.22095808e-01 8.20662379e-01 -9.05183077e-01 9.83198166e-01
-3.57969850e-01 6.76802337e-01 -4.69169905e-03 1.18192390e-01
-1.57531332e-02 -2.56349593e-01 3.41665715e-01 9.09621000e-01
8.94874409e-02 -2.76015610e-01 -3.64606857e-01 9.89860415e-01
-1.06850147e-01 2.00029626e-01 -4.65493202e-01 -1.75009593e-01
2.57853657e-01 1.80969083e+00 -3.85131299e-01 1.36619121e-01
-2.59739041e-01 1.12440419e+00 2.73071915e-01 4.09072340e-01
-7.72519946e-01 -3.31271678e-01 4.06138361e-01 3.98084760e-01
-3.01468164e-01 3.79610389e-01 -2.65582681e-01 -1.16233397e+00
1.46441877e-01 -9.42590475e-01 1.78594097e-01 -3.20663393e-01
-1.61167419e+00 3.53752822e-01 -6.91829324e-01 -1.44148302e+00
6.50950894e-02 -3.90413791e-01 -8.15569222e-01 9.40695405e-01
-1.65067542e+00 -1.86699545e+00 -8.10342610e-01 5.88720977e-01
2.91933984e-01 1.04437940e-01 9.53195393e-01 2.61119038e-01
-1.01549768e+00 8.80425274e-01 -1.96570121e-02 2.48185955e-02
8.31463873e-01 -1.23519599e+00 4.79273312e-02 6.19996786e-01
-2.37917557e-01 4.90272790e-01 3.39000314e-01 -7.96121836e-01
-1.39797568e+00 -1.39080751e+00 2.17190817e-01 -1.08627923e-01
7.56465912e-01 -4.72947538e-01 -8.77650917e-01 2.51451015e-01
2.18386337e-01 -1.41129829e-02 1.09952557e+00 -1.66754767e-01
-3.07517946e-01 -3.20072055e-01 -1.21151483e+00 5.94792187e-01
9.16934013e-01 -6.31149232e-01 -2.75202282e-02 6.03558481e-01
2.60420054e-01 -2.51303881e-01 -1.12721610e+00 5.36138654e-01
5.64053953e-01 -7.31248558e-01 7.17242837e-01 -3.60913694e-01
7.93207943e-01 -1.18235700e-01 1.88091755e-01 -1.19993412e+00
-6.81283250e-02 -3.54250401e-01 -1.08033270e-01 1.48906815e+00
2.25381836e-01 -4.94154483e-01 9.23834026e-01 2.31312230e-01
9.48239490e-02 -1.13772273e+00 -7.82082677e-01 -3.97397965e-01
-2.39241347e-02 4.09147749e-03 2.08215520e-01 8.46393585e-01
-2.79533565e-01 1.90814272e-01 -5.27062118e-01 3.13786753e-02
8.28883529e-01 6.91275299e-02 6.72444165e-01 -9.15955365e-01
-3.31613362e-01 -6.23405613e-02 -3.24967712e-01 -5.38114429e-01
-4.38780375e-02 -1.02618062e+00 -2.24031702e-01 -1.28756225e+00
4.94527251e-01 -5.39369226e-01 -4.58777338e-01 4.43677723e-01
-5.56325197e-01 5.96601307e-01 -2.27792300e-02 2.16251060e-01
-3.69449914e-01 5.80228150e-01 1.68541062e+00 -4.64981645e-01
1.18071698e-01 1.37405202e-01 -8.71953964e-01 5.85178614e-01
8.56813073e-01 -2.23904610e-01 -3.78983170e-01 -2.16537446e-01
3.98194082e-02 3.64964344e-02 8.54375124e-01 -8.58377039e-01
1.28957912e-01 9.95145664e-02 5.92287838e-01 -3.65588337e-01
4.81126398e-01 -5.89951694e-01 2.68467963e-01 7.53320873e-01
-3.42069477e-01 -7.29785919e-01 -2.29061451e-02 7.85068274e-01
-3.59629244e-01 -9.45274085e-02 8.83395493e-01 -6.93130940e-02
-3.36809635e-01 3.77910078e-01 -6.77433014e-02 -1.34187639e-01
1.24543917e+00 -1.92758083e-01 -3.65739703e-01 -4.49198969e-02
-7.12312937e-01 1.31436139e-01 6.17676437e-01 1.34936333e-01
7.51441658e-01 -1.16958189e+00 -8.29281271e-01 3.27480167e-01
4.46835667e-01 1.53065920e-01 8.99801135e-01 1.33800256e+00
-4.05520618e-01 -1.10884301e-01 -2.97967792e-01 -3.57928365e-01
-1.23304343e+00 3.13269675e-01 2.59611636e-01 -4.15855139e-01
-4.23700780e-01 9.45351660e-01 2.11320430e-01 -3.33722025e-01
4.64552641e-02 3.57128419e-02 -3.46797705e-02 6.78023621e-02
4.53448951e-01 3.69636476e-01 1.05449289e-01 -2.78461069e-01
-8.46808329e-02 4.01131690e-01 -3.74926180e-01 3.66839409e-01
1.18622541e+00 -1.48337200e-01 -1.70165449e-01 1.38078466e-01
9.72579122e-01 1.25133201e-01 -1.29044235e+00 1.04677960e-01
-2.40400314e-01 -4.79565561e-01 -1.29179463e-01 -9.79033053e-01
-1.27879715e+00 1.12619126e+00 9.20157969e-01 -1.01805255e-01
1.35271895e+00 -3.53738636e-01 8.96039426e-01 -2.11552531e-01
2.06429005e-01 -9.29055035e-01 1.40085742e-01 -4.08295810e-01
1.02888072e+00 -1.33822453e+00 2.18101069e-02 -8.60233486e-01
-7.54609466e-01 9.12958503e-01 6.66949034e-01 -2.13816822e-01
3.15206856e-01 2.53090829e-01 2.51451135e-01 -1.10621147e-01
-5.56383252e-01 -1.24044307e-01 3.04631501e-01 7.22015679e-01
4.65424478e-01 2.40804389e-01 -4.22629893e-01 6.38005376e-01
1.93807110e-01 6.15060404e-02 3.06598604e-01 5.78916132e-01
2.37902194e-01 -1.10264564e+00 -8.07131380e-02 8.68885338e-01
-4.12854701e-01 -1.16628364e-01 -5.84398389e-01 9.24648345e-01
3.94304067e-01 8.16791236e-01 -3.53022844e-01 -4.20629919e-01
9.37173739e-02 -1.06513038e-01 4.51801091e-01 -5.70810080e-01
-7.03419864e-01 2.16487691e-01 5.67202568e-02 -5.96215546e-01
-4.61769611e-01 -4.25252289e-01 -7.84167349e-01 1.21127196e-01
-4.82617795e-01 -2.80737847e-01 6.66902184e-01 6.27222776e-01
2.23536968e-01 7.38416016e-01 7.15979278e-01 -3.95820677e-01
-6.93037450e-01 -1.07090676e+00 -8.40607822e-01 7.62411892e-01
2.38195896e-01 -3.89337271e-01 -3.01548570e-01 5.52397082e-03] | [15.51213550567627, -2.8235180377960205] |
26154b30-476b-44b2-a35a-5906ffce42d6 | lp-ioanet-efficient-high-resolution-document | 2303.12862 | null | https://arxiv.org/abs/2303.12862v1 | https://arxiv.org/pdf/2303.12862v1.pdf | LP-IOANet: Efficient High Resolution Document Shadow Removal | Document shadow removal is an integral task in document enhancement pipelines, as it improves visibility, readability and thus the overall quality. Assuming that the majority of practical document shadow removal scenarios require real-time, accurate models that can produce high-resolution outputs in-the-wild, we propose Laplacian Pyramid with Input/Output Attention Network (LP-IOANet), a novel pipeline with a lightweight architecture and an upsampling module. Furthermore, we propose three new datasets which cover a wide range of lighting conditions, images, shadow shapes and viewpoints. Our results show that we outperform the state-of-the-art by a 35% relative improvement in mean average error (MAE), while running real-time in four times the resolution (of the state-of-the-art method) on a mobile device. | ['Bruno Manganelli', 'Albert Saa-Garriga', 'Anastasios Drosou', 'Valia Dimaridou', 'Evangelos Skartados', 'M. Kerim Yucel', 'Konstantinos Georgiadis'] | 2023-03-22 | null | null | null | null | ['document-enhancement', 'shadow-removal'] | ['computer-vision', 'computer-vision'] | [ 6.99793875e-01 -2.88763344e-01 6.29438281e-01 -1.77323371e-01
-5.74921191e-01 -5.37154198e-01 5.65625548e-01 -2.68923819e-01
-1.97411269e-01 3.72772723e-01 3.25540304e-01 -4.58655953e-01
3.09664607e-01 -5.59595048e-01 -8.36545587e-01 -5.58528543e-01
1.71807051e-01 -9.04671028e-02 6.03445768e-01 -8.35694224e-02
1.31855488e-01 4.80547577e-01 -1.58199871e+00 7.52250671e-01
1.11319518e+00 9.08349514e-01 5.76418400e-01 1.34396720e+00
4.00402583e-02 6.64364040e-01 -8.30819070e-01 -7.16272354e-01
1.00086346e-01 6.86955005e-02 -1.52282834e-01 1.48790762e-01
9.84359026e-01 -6.48035884e-01 -4.54970092e-01 9.33497846e-01
7.25124657e-01 -7.67752528e-02 3.89617324e-01 -7.34317899e-01
-1.05692160e+00 6.02326617e-02 -9.70023215e-01 -7.23594353e-02
2.60640413e-01 4.07489240e-01 5.55029571e-01 -1.21510077e+00
5.14236569e-01 1.33826506e+00 9.97461200e-01 1.58980638e-01
-1.21000111e+00 -4.84121919e-01 2.13386357e-01 7.40574002e-02
-1.16846836e+00 -5.67438900e-01 2.27630615e-01 -6.01438060e-02
1.08456528e+00 4.44391966e-01 4.51185048e-01 1.14639235e+00
4.69899684e-01 8.39506030e-01 1.24270105e+00 -4.74309683e-01
-9.53556132e-03 2.18270168e-01 1.77644156e-02 6.93090260e-01
4.95452732e-01 -3.11663330e-01 -8.34671557e-01 2.17180848e-01
5.63383639e-01 5.54650351e-02 -5.94703674e-01 1.70955747e-01
-8.59260440e-01 1.33854553e-01 5.44951975e-01 -1.72576513e-02
-3.73274952e-01 3.34165901e-01 -5.82160726e-02 -4.82718386e-02
5.30754566e-01 2.61760782e-02 -3.76519263e-01 1.15020208e-01
-1.24854922e+00 1.31471083e-01 5.61317265e-01 8.88483763e-01
3.35410714e-01 9.24554616e-02 -6.99320376e-01 5.08387864e-01
2.29053333e-01 1.08078027e+00 8.80761445e-02 -7.92180181e-01
5.15653610e-01 4.49400187e-01 3.32609713e-01 -9.46397960e-01
-2.49355316e-01 -6.87681258e-01 -8.71142805e-01 5.72434902e-01
2.98275143e-01 2.32994910e-02 -1.17876661e+00 1.06678808e+00
1.62573382e-02 2.14685430e-03 -9.35415104e-02 8.55868995e-01
6.09030843e-01 8.03975046e-01 -3.01953673e-01 -1.61541626e-01
1.64722121e+00 -1.24382424e+00 -9.93017673e-01 -4.67922449e-01
-3.69064510e-02 -1.15572071e+00 1.57592773e+00 9.45278347e-01
-1.11725307e+00 -6.71943665e-01 -1.34747076e+00 -5.56634605e-01
-3.89816195e-01 4.27944541e-01 6.07210517e-01 1.02279198e+00
-1.19819558e+00 5.35358429e-01 -7.01605678e-01 -3.16604704e-01
6.14627957e-01 1.56198278e-01 -7.92085752e-02 -3.44091952e-01
-6.35531545e-01 6.97779834e-01 -2.48473600e-01 3.62883471e-02
-8.03691864e-01 -8.49034369e-01 -3.13014567e-01 4.36272949e-01
3.31462175e-01 -4.89588886e-01 1.21390295e+00 -6.94343507e-01
-1.63888681e+00 5.08813143e-01 -4.46997464e-01 -3.84436786e-01
7.38759816e-01 -8.41226041e-01 -5.10659695e-01 7.76022524e-02
-3.54499131e-01 3.21749061e-01 1.05321634e+00 -1.48257434e+00
-5.86931527e-01 -4.44717914e-01 1.25344545e-01 2.38349870e-01
-6.61339343e-01 4.25219610e-02 -1.05883467e+00 -5.17298400e-01
-2.82979280e-01 -7.86351681e-01 1.35074914e-01 4.50604379e-01
-4.51175541e-01 4.13619816e-01 1.42461526e+00 -1.24478757e+00
1.17993391e+00 -2.15505695e+00 -3.46651256e-01 -1.47964224e-01
1.09620549e-01 5.94508469e-01 -2.06154250e-02 1.59836456e-01
3.60260189e-01 -4.11574580e-02 -3.21852863e-01 -9.20546472e-01
-4.25987132e-02 -2.10496083e-01 -5.50580382e-01 5.19427061e-01
1.07365474e-02 8.51953149e-01 -5.15350819e-01 -2.51820058e-01
4.69548672e-01 1.16832042e+00 -2.73781270e-01 1.73795789e-01
-1.87696189e-01 1.06611028e-01 1.61663666e-01 6.47321403e-01
1.15396321e+00 -3.00699055e-01 2.04593748e-01 -3.24024111e-01
-2.79098928e-01 -1.39914379e-01 -1.29147553e+00 1.69563293e+00
-8.15769553e-01 1.42099893e+00 2.65897453e-01 2.72179663e-01
5.49474239e-01 1.03682995e-01 -2.58820266e-01 -8.02545428e-01
-5.73540479e-02 2.93811988e-02 -3.61669958e-01 -3.84118319e-01
1.08290124e+00 6.69506669e-01 5.19085228e-01 2.99888730e-01
-4.13181573e-01 1.39594957e-01 -2.58453260e-03 3.64748895e-01
1.24308181e+00 2.79972643e-01 -1.40559107e-01 -2.04735920e-01
3.22694510e-01 -5.47144234e-01 1.49391860e-01 8.16222966e-01
1.14794843e-01 1.01518762e+00 3.71444613e-01 -2.71383643e-01
-1.14980006e+00 -1.10737646e+00 -6.80487975e-02 1.36590958e+00
5.50105348e-02 -3.98252636e-01 -1.19343233e+00 -2.75028169e-01
-2.02000886e-01 8.43882620e-01 -4.98440385e-01 4.57196921e-01
-5.01811385e-01 -8.99241865e-01 4.43418771e-01 7.12542295e-01
8.97652388e-01 -9.05637562e-01 -1.10076141e+00 -1.31958023e-01
5.72152920e-02 -1.52146626e+00 -5.48111856e-01 1.64923310e-01
-4.96018171e-01 -9.56073940e-01 -7.76303887e-01 -2.97741145e-01
6.80302799e-01 7.44908512e-01 1.17457807e+00 1.37269929e-01
-5.24402320e-01 2.51091234e-02 -1.39774099e-01 -6.42550528e-01
-3.76879163e-02 -1.92852616e-01 -2.13616535e-01 9.63083655e-02
-1.03716664e-02 -3.29280764e-01 -9.88688350e-01 -7.68849179e-02
-1.02744389e+00 3.54727060e-01 7.27050424e-01 6.50025666e-01
3.03583384e-01 1.75382093e-01 -1.79951936e-01 -1.12844288e+00
7.02866077e-01 3.19503248e-01 -8.39280784e-01 3.57857615e-01
-8.48171592e-01 -1.18495248e-01 7.03166246e-01 -2.75417715e-01
-1.63936508e+00 -1.41976729e-01 1.97494864e-01 -2.68488616e-01
6.25700057e-02 -2.09178567e-01 -4.80259746e-01 -5.25671523e-03
8.34923983e-01 1.53945848e-01 -5.41148305e-01 -6.01927936e-01
5.56265891e-01 9.12375510e-01 9.85529840e-01 -6.49797469e-02
7.85940170e-01 9.03715014e-01 1.04258526e-02 -9.48870957e-01
-7.59108424e-01 -3.32956165e-01 -7.40943253e-01 -1.42883331e-01
7.51272559e-01 -8.98391545e-01 -7.39213228e-01 8.08551073e-01
-1.46710110e+00 -4.80161667e-01 1.35259047e-01 -1.78658411e-01
1.01222947e-01 3.61810446e-01 -4.38373625e-01 -1.14477897e+00
-8.36194158e-01 -1.21739054e+00 1.36650646e+00 5.50310016e-01
2.42380932e-01 -5.33181429e-01 -4.23931807e-01 2.00425789e-01
7.92457759e-01 1.12131558e-01 5.56132555e-01 1.76364690e-01
-8.41424108e-01 6.90021738e-02 -9.87678349e-01 2.60257721e-01
-3.35768983e-02 1.41560242e-01 -1.88710928e+00 -4.97218668e-01
-3.53929967e-01 1.65281802e-01 1.25198066e+00 5.67746103e-01
1.37145615e+00 -2.12311566e-01 -3.66663426e-01 9.05009687e-01
1.62315965e+00 -1.46172360e-01 1.11311162e+00 2.19343692e-01
8.20323050e-01 3.24518979e-01 3.22547674e-01 5.35678446e-01
2.01044217e-01 6.73677802e-01 4.35386062e-01 -6.04646325e-01
-8.97547603e-01 2.71294229e-02 4.39371735e-01 3.11994463e-01
-9.03604925e-02 -7.44093060e-01 -7.29985476e-01 4.16600585e-01
-1.77216661e+00 -7.62411594e-01 -4.15152639e-01 2.25833893e+00
7.21579969e-01 2.69162506e-01 -3.68583858e-01 2.50772059e-01
5.55829704e-01 3.88236821e-01 -5.17103195e-01 -5.01482069e-01
-4.11338270e-01 3.24671507e-01 8.82126570e-01 6.39021218e-01
-9.63379085e-01 1.03088903e+00 6.44979334e+00 5.21237791e-01
-1.07531464e+00 2.07706898e-01 6.42659605e-01 -1.25944629e-01
-2.58768022e-01 -2.39138469e-01 -7.80965626e-01 3.97940904e-01
1.13478279e+00 3.95624965e-01 6.59677982e-01 6.21232331e-01
2.57789701e-01 -2.72237688e-01 -8.43818843e-01 9.14126992e-01
5.23472965e-01 -1.20936418e+00 -2.94361293e-01 1.15060769e-01
8.97014499e-01 -7.06837475e-02 4.49595958e-01 -9.11848992e-02
2.27837488e-01 -1.17865157e+00 6.98215246e-01 7.03201771e-01
1.16122055e+00 -7.60165274e-01 4.87082243e-01 1.42928734e-02
-1.21625960e+00 -6.64149523e-02 -3.71075898e-01 1.46250173e-01
2.36219049e-01 1.09974432e+00 -7.95153558e-01 2.03092232e-01
1.22027791e+00 2.61579186e-01 -8.57620776e-01 1.05205798e+00
-5.18863559e-01 5.27467012e-01 -1.52494028e-01 1.37699321e-01
-1.23268172e-01 1.76735148e-01 3.57269943e-01 1.89431584e+00
3.05736035e-01 -1.51401266e-01 -2.78341889e-01 5.51342785e-01
-3.77254397e-01 -3.08629453e-01 -1.75731540e-01 3.46129388e-01
3.05760473e-01 1.44912970e+00 -7.12559760e-01 -3.09594661e-01
-4.47819948e-01 1.73280299e+00 -2.60277949e-02 5.83567321e-01
-9.44384277e-01 -7.06763387e-01 5.99718809e-01 2.53426999e-01
4.99220163e-01 -1.61457881e-01 -5.94657004e-01 -8.72704268e-01
3.96106631e-01 -9.15572405e-01 -3.89048070e-01 -1.30345345e+00
-7.90490270e-01 6.76283956e-01 -5.27498603e-01 -7.54739881e-01
3.70477647e-01 -1.12492561e+00 -5.87751269e-01 1.03746057e+00
-1.92930496e+00 -1.52648568e+00 -1.03142488e+00 3.12199920e-01
7.42223442e-01 3.18393931e-02 8.69123638e-01 2.62379140e-01
-4.90032136e-01 5.89987457e-01 5.53395391e-01 -1.89099014e-01
9.75364506e-01 -1.38126969e+00 9.50755179e-01 1.34173656e+00
1.60508201e-01 5.67730606e-01 9.30317283e-01 -4.96196806e-01
-1.65960562e+00 -1.09592211e+00 4.84727770e-01 -4.18962657e-01
2.16115654e-01 -8.17110419e-01 -9.29143071e-01 5.48708677e-01
7.47442424e-01 -6.60237391e-03 3.10700834e-01 1.13256969e-01
-5.58907390e-01 -2.02827007e-01 -1.06279325e+00 9.85006630e-01
9.75819349e-01 -6.18272603e-01 3.18967775e-02 5.01187861e-01
8.86578143e-01 -7.85393059e-01 -3.79341036e-01 4.38650399e-02
9.26348746e-01 -1.41911936e+00 1.08223593e+00 1.89282164e-01
4.04279202e-01 -4.58279938e-01 -2.93630868e-01 -1.13646758e+00
-3.74108851e-01 -8.97733688e-01 -5.60086370e-01 1.22639227e+00
2.95474917e-01 -1.87411726e-01 6.71898067e-01 4.05385792e-01
-1.50664300e-01 -6.16196334e-01 -5.56678295e-01 -4.02099699e-01
-4.75429982e-01 -6.54343545e-01 5.87508619e-01 1.96072102e-01
-7.08451927e-01 2.04144567e-01 -6.38852358e-01 5.92050672e-01
7.78754473e-01 -7.18259951e-03 8.60933959e-01 -1.16552532e+00
-5.06825805e-01 -9.87340286e-02 2.80990638e-03 -1.11635303e+00
-3.46865833e-01 -3.18910092e-01 4.76882346e-02 -1.95631266e+00
4.61537182e-01 1.52268067e-01 2.33110748e-02 4.25465316e-01
-3.00882578e-01 5.46350181e-01 4.69763160e-01 3.27314660e-02
-7.36148179e-01 2.25413397e-01 9.75310266e-01 -2.69383699e-01
-1.56476781e-01 -2.26143748e-01 -7.31366515e-01 9.20340955e-01
2.93038875e-01 -1.19135730e-01 -3.27823550e-01 -1.12353432e+00
1.63470626e-01 -5.78139365e-01 3.67758155e-01 -1.18670607e+00
3.62472802e-01 1.51361555e-01 1.09182787e+00 -8.78944576e-01
6.65276766e-01 -8.75838339e-01 -8.92274603e-02 4.95440900e-01
-5.28156683e-02 -1.39098121e-02 4.23778683e-01 6.74093544e-01
4.67085898e-01 3.30690630e-02 7.83392429e-01 2.00368270e-01
-5.84354520e-01 -9.29307342e-02 -6.03032000e-02 -3.83835256e-01
7.38813221e-01 -3.15810293e-01 -1.05293715e+00 -3.38185042e-01
-9.57426876e-02 -1.97405860e-01 5.97299755e-01 2.66560644e-01
7.70584285e-01 -7.80593336e-01 -7.82646060e-01 2.47348994e-01
-9.49507728e-02 -5.45107983e-02 3.44173312e-01 6.03222013e-01
-7.78858900e-01 4.22281802e-01 5.32939658e-02 -5.48477709e-01
-1.56227469e+00 2.67210990e-01 1.64333478e-01 -4.43525940e-01
-7.81783700e-01 9.17986095e-01 3.10015678e-01 5.04329503e-02
4.15755868e-01 -2.96889395e-01 1.30329311e-01 -3.58139396e-01
1.08028662e+00 5.68751454e-01 5.63388109e-01 -1.46825925e-01
-2.19773695e-01 5.08752286e-01 -8.62229466e-02 -1.30584881e-01
1.18339562e+00 -1.45585269e-01 8.36304054e-02 1.16861515e-01
7.82514691e-01 5.26003659e-01 -1.76818085e+00 -1.77683860e-01
-4.26074445e-01 -1.02521443e+00 5.30984581e-01 -1.21452641e+00
-9.46634829e-01 1.07697523e+00 1.20192134e+00 1.25856511e-02
1.41197002e+00 -5.82194924e-01 7.49287188e-01 3.06578577e-01
8.41011852e-02 -1.13989401e+00 1.88395366e-01 3.95209849e-01
1.16137254e+00 -1.23306143e+00 5.08621573e-01 -4.34396297e-01
-7.65317261e-01 8.48169148e-01 4.70055521e-01 1.95599869e-01
3.30155015e-01 9.20688808e-01 9.56094265e-02 1.75660104e-01
-7.09688485e-01 2.59699002e-02 4.43832934e-01 8.65969718e-01
7.30514526e-01 -9.63620692e-02 2.51120716e-01 4.72007751e-01
-1.12137228e-01 -1.67018592e-01 5.22555232e-01 7.25802183e-01
-5.25788605e-01 -6.96838498e-01 -6.31826878e-01 2.98109204e-01
-5.42665601e-01 -5.39187491e-01 -2.76969850e-01 4.31688935e-01
5.61010540e-02 1.11383235e+00 4.44946140e-02 -1.76772296e-01
4.59077924e-01 -3.49491626e-01 4.64442074e-01 -2.95908511e-01
-5.87689638e-01 2.94271380e-01 -1.09278157e-01 -7.21090853e-01
1.17808413e-02 -2.92122155e-01 -1.13856864e+00 -5.65475702e-01
-4.69691962e-01 -6.78900242e-01 1.04369855e+00 6.18135393e-01
6.69158459e-01 1.19293141e+00 3.29824239e-01 -1.12884545e+00
-1.25660866e-01 -1.20532131e+00 -5.10420322e-01 2.06952363e-01
4.13445234e-01 -2.34813184e-01 -1.44292906e-01 3.72360259e-01] | [10.879341125488281, -3.977834701538086] |
5411d75f-7163-48e8-a3cb-b937fc688f97 | object-centered-image-stitching-1 | 2011.11789 | null | https://arxiv.org/abs/2011.11789v1 | https://arxiv.org/pdf/2011.11789v1.pdf | Object-centered image stitching | Image stitching is typically decomposed into three phases: registration, which aligns the source images with a common target image; seam finding, which determines for each target pixel the source image it should come from; and blending, which smooths transitions over the seams. As described in [1], the seam finding phase attempts to place seams between pixels where the transition between source images is not noticeable. Here, we observe that the most problematic failures of this approach occur when objects are cropped, omitted, or duplicated. We therefore take an object-centered approach to the problem, leveraging recent advances in object detection [2,3,4]. We penalize candidate solutions with this class of error by modifying the energy function used in the seam finding stage. This produces substantially more realistic stitching results on challenging imagery. In addition, these methods can be used to determine when there is non-recoverable occlusion in the input data, and also suggest a simple evaluation metric that can be used to evaluate the output of stitching algorithms. | ['Ramin Zabih', 'Emil Keyder', 'Richard Strong Bowen', 'Chen Wang', 'Charles Herrmann'] | 2020-11-23 | object-centered-image-stitching | http://openaccess.thecvf.com/content_ECCV_2018/html/Charles_Herrmann_Object-centered_image_stitching_ECCV_2018_paper.html | http://openaccess.thecvf.com/content_ECCV_2018/papers/Charles_Herrmann_Object-centered_image_stitching_ECCV_2018_paper.pdf | eccv-2018-9 | ['image-stitching'] | ['computer-vision'] | [ 9.68048215e-01 -1.68716878e-01 1.43660724e-01 -8.65716115e-02
-6.87287867e-01 -8.27035666e-01 7.15004921e-01 2.59946913e-01
-1.72858939e-01 3.15649778e-01 3.92329246e-02 -9.78716835e-02
7.43204802e-02 -3.25463653e-01 -5.16098440e-01 -7.33700573e-01
8.72055907e-03 1.78213254e-01 5.28283715e-01 -9.08124149e-02
6.51145875e-01 3.98519963e-01 -1.39024377e+00 -3.37920967e-03
8.84901106e-01 8.30079257e-01 5.29816188e-02 6.57099187e-01
1.18536726e-01 3.27543736e-01 -6.64992630e-01 -1.86969489e-01
5.86769164e-01 -5.12633145e-01 -7.99092114e-01 7.50384152e-01
9.29552376e-01 -2.44265750e-01 1.18111916e-01 1.16593218e+00
3.09440009e-02 2.13815421e-01 5.31531096e-01 -1.25655842e+00
-1.05042025e-01 1.70055583e-01 -1.16172409e+00 7.67543912e-02
4.12814558e-01 2.71110117e-01 7.93689013e-01 -8.10815871e-01
7.19977856e-01 1.13749051e+00 6.97420955e-01 5.58723435e-02
-1.51372612e+00 -4.95095730e-01 -2.46318325e-01 -2.14085758e-01
-1.15640819e+00 -7.46089578e-01 7.99500227e-01 -5.05898833e-01
3.88950258e-01 3.85729045e-01 4.59443450e-01 4.85359073e-01
2.96181619e-01 4.71499383e-01 9.63535070e-01 -7.49919653e-01
1.20214544e-01 -1.18975952e-01 -9.89650376e-03 3.66583854e-01
2.93703943e-01 3.01169634e-01 -4.28845614e-01 -4.05361056e-01
7.47972429e-01 -2.12283075e-01 -4.96086895e-01 -4.65926617e-01
-1.25578856e+00 5.25727630e-01 4.53304261e-01 2.36310020e-01
-7.73443058e-02 2.45703071e-01 1.28140971e-01 2.17087522e-01
4.31767374e-01 7.04330087e-01 2.61479542e-02 1.84895754e-01
-1.38100183e+00 1.80607393e-01 4.55759674e-01 6.98067725e-01
1.01444626e+00 -7.77595490e-02 1.24728628e-01 7.31746078e-01
1.65274180e-02 9.86019149e-02 -4.69490588e-02 -1.22705352e+00
3.00976068e-01 3.30022514e-01 2.52755463e-01 -1.09530628e+00
-5.40414266e-02 -2.20340103e-01 -4.29266214e-01 6.13014638e-01
5.70702434e-01 -4.14253287e-02 -1.11420107e+00 1.41781116e+00
5.89061677e-01 5.03383577e-01 1.32995797e-02 8.92297506e-01
3.05944681e-01 5.04839897e-01 -3.54691893e-01 -1.75016969e-01
1.25897002e+00 -9.90339398e-01 -5.64657509e-01 -6.50401890e-01
3.29699636e-01 -1.40417528e+00 6.16046131e-01 2.36557901e-01
-1.14457357e+00 -4.49847549e-01 -1.25730574e+00 3.31229381e-02
-5.85647300e-02 -3.89403440e-02 3.63633186e-01 4.64273065e-01
-1.12678099e+00 8.22207868e-01 -7.01876640e-01 -3.78948420e-01
6.89701140e-02 3.13257158e-01 -2.26578891e-01 -1.77705631e-01
-5.79936028e-01 9.69538927e-01 3.30660760e-01 9.61562395e-02
-4.27475363e-01 -6.26728833e-01 -8.37047756e-01 -1.64987773e-01
6.06959283e-01 -5.54566860e-01 1.10949588e+00 -1.43350446e+00
-9.23764527e-01 9.67983961e-01 -3.20345938e-01 -2.36106962e-01
4.09316152e-01 -5.61975017e-02 -1.38992250e-01 1.00819573e-01
2.65156388e-01 8.69898617e-01 1.45498657e+00 -1.68022585e+00
-6.78629816e-01 -2.72874445e-01 -3.41591597e-01 2.73993939e-01
1.76792666e-01 1.16168976e-01 -7.26370335e-01 -9.63803053e-01
4.43281949e-01 -1.10669518e+00 -2.28033483e-01 2.09103778e-01
-5.15070021e-01 3.73912334e-01 1.17067993e+00 -7.65064478e-01
9.81291652e-01 -2.40630007e+00 1.45952821e-01 4.56154048e-01
1.30414367e-01 1.25545278e-01 -2.76419222e-01 5.49113095e-01
-1.65115505e-01 -1.17678463e-01 -7.38180161e-01 -6.04701340e-01
-5.04986823e-01 1.53627414e-02 -5.30264139e-01 5.44765711e-01
2.71710068e-01 4.72082257e-01 -9.12851393e-01 -4.22144502e-01
2.69909978e-01 2.44130656e-01 -2.96598703e-01 6.63532168e-02
-6.33136854e-02 4.16340500e-01 8.44198912e-02 6.60555542e-01
8.95620108e-01 -4.39563692e-02 -6.94688549e-03 -4.03906167e-01
-4.65072513e-01 4.76753106e-03 -1.34023356e+00 1.52999926e+00
2.94482112e-02 9.47042048e-01 3.29262346e-01 -5.21688402e-01
7.71035969e-01 8.11744928e-02 6.46637440e-01 -2.26058424e-01
5.06700352e-02 2.83583045e-01 -6.39168173e-02 -2.58453399e-01
9.17280555e-01 8.10040683e-02 1.61160082e-01 5.51543415e-01
-2.84525573e-01 -5.30892491e-01 3.32698166e-01 2.48603433e-01
1.11939275e+00 1.18086994e-01 1.08663157e-01 -2.93612719e-01
2.39880234e-01 4.52307612e-01 4.56775546e-01 6.90073669e-01
-1.16790324e-01 1.20114684e+00 2.84734368e-01 -2.66979307e-01
-1.09394670e+00 -8.74334276e-01 -2.37471256e-02 5.39440393e-01
7.53614426e-01 -3.07521790e-01 -8.43248963e-01 -4.07233030e-01
-5.89945503e-02 5.64865172e-01 -6.49787605e-01 -1.39830872e-01
-6.68593764e-01 -4.47605729e-01 1.98757634e-01 1.76691353e-01
4.36970711e-01 -8.10135663e-01 -8.77807140e-01 1.19034365e-01
-3.28508168e-01 -9.65705276e-01 -1.11801231e+00 3.71631473e-01
-1.13354063e+00 -1.26103771e+00 -6.88558936e-01 -7.88703263e-01
1.05606854e+00 8.96704197e-01 9.57092702e-01 5.71315408e-01
-4.29638714e-01 3.30634207e-01 -2.78438717e-01 6.97701573e-02
-5.71547389e-01 -2.78416663e-01 -3.17112476e-01 1.48178831e-01
-2.85614103e-01 -2.94860452e-01 -6.90318763e-01 5.38171470e-01
-1.30044985e+00 1.51365057e-01 3.47322106e-01 7.44776011e-01
4.32202399e-01 2.40421489e-01 -1.24018788e-01 -5.88513911e-01
4.21224624e-01 -1.45262897e-01 -6.58492744e-01 3.29331994e-01
-5.21581531e-01 1.32760882e-01 1.98870704e-01 -4.92025405e-01
-7.54854023e-01 3.68783981e-01 3.32458705e-01 -5.90973794e-01
9.26820710e-02 2.57125884e-01 2.69618571e-01 -4.93803769e-01
5.87626040e-01 1.38818592e-01 2.04677671e-01 -3.76888007e-01
1.71823770e-01 3.45275819e-01 9.42626536e-01 -2.87931025e-01
1.15306628e+00 6.51848674e-01 -4.14741002e-02 -7.58298099e-01
-5.58313608e-01 -6.04424953e-01 -6.14593923e-01 -3.79520506e-01
7.36814141e-01 -4.82756525e-01 -1.57500044e-01 5.55706084e-01
-1.22639155e+00 -3.06882024e-01 -2.40159139e-01 1.20556235e-01
-3.93673986e-01 7.23852396e-01 -1.70191988e-01 -6.32542610e-01
-1.90486073e-01 -1.35072434e+00 1.15803909e+00 3.47474039e-01
-3.95372719e-01 -8.80787671e-01 1.67001337e-01 2.44845971e-01
3.56614590e-01 4.59237069e-01 7.22381353e-01 -1.49774596e-01
-6.28008485e-01 -8.31734687e-02 -1.98362574e-01 1.46838591e-01
5.00679076e-01 4.31389570e-01 -7.61050761e-01 -5.15551805e-01
1.66610293e-02 2.85697952e-02 8.79855633e-01 4.49452668e-01
4.59590018e-01 -1.83476120e-01 -5.11127293e-01 6.11900568e-01
1.42221165e+00 3.28405052e-01 7.65187085e-01 4.99654412e-01
4.58346158e-01 7.43768156e-01 7.21521735e-01 -4.75947447e-02
-3.06786709e-02 1.12284672e+00 4.63962674e-01 -5.54523647e-01
-3.01350594e-01 -1.02475323e-02 2.60637671e-01 5.98770790e-02
1.13620527e-01 -2.10584044e-01 -7.41362453e-01 5.53860605e-01
-1.78393352e+00 -8.29138458e-01 -2.38285407e-01 2.58674145e+00
6.61046922e-01 1.46569684e-01 1.25416994e-01 1.74102813e-01
9.25808132e-01 3.01295787e-01 -4.14697111e-01 -1.87950328e-01
-6.46296814e-02 1.11884996e-02 7.32921541e-01 7.69487560e-01
-1.16341102e+00 9.33538437e-01 6.77982998e+00 6.68499470e-01
-1.11115575e+00 -3.35449427e-01 5.48797488e-01 1.83269218e-01
-2.33966529e-01 6.84843242e-01 -3.38608205e-01 4.59604651e-01
1.64478704e-01 1.81755051e-01 4.93816853e-01 3.18620294e-01
1.14903755e-01 -7.63013422e-01 -1.04948795e+00 8.25375378e-01
1.25518173e-01 -1.23752403e+00 -3.25487494e-01 7.48101100e-02
9.02615666e-01 -2.59874314e-01 6.61068708e-02 -5.74669182e-01
1.66849256e-01 -6.49275899e-01 8.90539050e-01 1.36284426e-01
5.15977144e-01 -5.36488533e-01 2.90376782e-01 1.81425866e-02
-1.31433058e+00 2.03516454e-01 -2.53540669e-02 3.80301923e-01
3.67002487e-01 5.82509577e-01 -8.16628814e-01 4.68666434e-01
5.14120102e-01 5.17967343e-01 -5.55777550e-01 1.43645501e+00
-6.37563020e-02 2.87606806e-01 -4.45199370e-01 6.64699376e-01
1.02697045e-01 -5.03678083e-01 8.25321794e-01 9.81999040e-01
3.71746212e-01 -6.45367280e-02 2.97821641e-01 8.53569508e-01
2.89064050e-01 -1.97357327e-01 -6.13567352e-01 2.63354301e-01
5.30887544e-01 1.19579923e+00 -1.31347072e+00 -1.74028009e-01
-2.13067681e-01 1.28549433e+00 -1.63356900e-01 4.42572117e-01
-5.73164999e-01 -3.60844165e-01 6.83929563e-01 1.70611501e-01
3.36052060e-01 -3.91925693e-01 -3.66517484e-01 -9.49211180e-01
2.49531940e-02 -9.81984258e-01 2.59894937e-01 -9.25606251e-01
-7.92309999e-01 4.80316341e-01 1.02048449e-01 -1.44403613e+00
-2.08527163e-01 -1.46696597e-01 -9.60274756e-01 6.78518951e-01
-1.28367567e+00 -9.24797177e-01 -3.75103980e-01 3.73165339e-01
6.68371797e-01 4.83363271e-01 2.92848021e-01 1.33803142e-02
-5.09936333e-01 2.87762582e-01 4.43522334e-02 -1.32315367e-01
7.86442220e-01 -9.01209891e-01 5.99447489e-01 1.36977613e+00
1.78592205e-01 4.85106468e-01 1.08224857e+00 -9.18101907e-01
-1.14652336e+00 -7.27753699e-01 6.21359468e-01 -1.44513875e-01
4.24158722e-01 -1.02821052e-01 -1.02821457e+00 5.36869228e-01
4.03249979e-01 -2.21142724e-01 6.70908168e-02 -3.91609579e-01
-3.65951389e-01 3.34733725e-02 -1.14590871e+00 7.03625083e-01
6.63538158e-01 -9.51360315e-02 -3.18456531e-01 7.48108253e-02
3.94151419e-01 -5.62869191e-01 -5.12022495e-01 2.45060802e-01
4.27478105e-01 -1.18100262e+00 9.86530364e-01 -4.33442853e-02
3.85804355e-01 -6.91260219e-01 1.61450237e-01 -1.34400177e+00
-1.26382127e-01 -1.07384789e+00 2.51634300e-01 1.23557532e+00
2.86603570e-01 -4.55506235e-01 6.17015302e-01 7.43187070e-01
-2.10574374e-01 -1.94897756e-01 -8.30581784e-01 -7.14185059e-01
-3.82785916e-01 1.41755238e-01 3.61640543e-01 9.74068344e-01
-1.60742968e-01 -1.21527053e-02 -4.11637217e-01 2.42741257e-01
9.83964622e-01 2.36631081e-01 9.33319747e-01 -1.05044913e+00
-1.40299991e-01 -5.15319407e-01 -2.07675919e-01 -9.55226302e-01
-3.28862399e-01 -3.79807889e-01 4.69341159e-01 -1.21928346e+00
7.69013120e-03 -6.55730367e-01 2.98671424e-01 5.33291638e-01
-4.05668795e-01 4.31968570e-01 2.79159427e-01 6.08958960e-01
-5.14895059e-02 1.69002637e-01 1.14675438e+00 -8.71458873e-02
-3.29520196e-01 -1.04343191e-01 -5.04536867e-01 6.29248917e-01
6.24002278e-01 -5.90082645e-01 -2.06871033e-01 -4.66514200e-01
-1.78670660e-01 1.27719089e-01 4.38209087e-01 -1.04676163e+00
3.56160283e-01 -3.75649929e-01 2.26362914e-01 -5.09736180e-01
4.14117634e-01 -9.42979991e-01 6.62988007e-01 5.64092934e-01
-2.17442974e-01 1.57613069e-01 2.61232942e-01 2.72446126e-01
-2.10128754e-01 -6.46616817e-01 1.06162858e+00 2.09953468e-02
-5.36839306e-01 -4.07712683e-02 -2.55327314e-01 -1.52047247e-01
9.92374778e-01 -7.73469031e-01 -3.24041456e-01 -4.09047037e-01
-2.88521349e-01 1.15493104e-01 1.08128905e+00 2.30345666e-01
6.13811135e-01 -1.10439014e+00 -5.01186848e-01 3.85131955e-01
-1.27292257e-02 2.82712318e-02 4.40150092e-04 9.75639284e-01
-6.87419534e-01 -2.73764282e-01 -1.29566342e-01 -7.17304111e-01
-1.56119359e+00 4.88584518e-01 3.44137937e-01 -5.15500344e-02
-6.65830195e-01 6.48752213e-01 8.58008787e-02 2.57841140e-01
1.48047864e-01 -1.92961976e-01 1.98677272e-01 -6.15541153e-02
4.74213779e-01 3.74281973e-01 1.68321922e-01 -7.73533523e-01
-3.25360805e-01 9.47527885e-01 -2.11106569e-01 -5.67000329e-01
1.05781913e+00 -2.35481158e-01 -9.48753506e-02 6.21384196e-02
1.09166420e+00 1.84535682e-01 -1.68118703e+00 -1.92761213e-01
1.36666536e-01 -9.62611020e-01 1.66076764e-01 -3.78939182e-01
-1.26723826e+00 3.72333646e-01 5.48978925e-01 3.21356237e-01
1.29383206e+00 -1.03825033e-01 8.27869296e-01 -1.91321492e-01
9.78251249e-02 -9.49350178e-01 1.89258352e-01 1.84733152e-01
8.38970542e-01 -1.06814933e+00 2.82723427e-01 -5.36886334e-01
-4.43831533e-01 1.21414089e+00 4.37313855e-01 -1.34963349e-01
3.11813354e-01 4.62889075e-01 1.68262824e-01 -2.56782770e-01
-4.07944918e-01 -9.51784700e-02 2.83612728e-01 4.42819864e-01
4.09076512e-02 -4.22179818e-01 -1.18677422e-01 -5.74968278e-01
5.48857413e-02 -2.35822275e-01 6.72463894e-01 1.29005682e+00
-6.74384832e-01 -1.18673027e+00 -9.72661555e-01 1.18929766e-01
-1.79633528e-01 -1.90545619e-01 -6.69399083e-01 5.77446878e-01
-2.69448347e-02 1.09019637e+00 1.23290978e-01 -3.41291994e-01
1.37545303e-01 -3.09297889e-01 5.79867005e-01 -4.68086839e-01
-7.97465622e-01 4.65108901e-01 -5.77935167e-02 -4.83167887e-01
-5.06413221e-01 -8.86088610e-01 -1.05559421e+00 -1.99021414e-01
-6.85066819e-01 -2.31403466e-02 6.28037572e-01 8.03303778e-01
2.70504534e-01 -4.96803410e-02 8.27951312e-01 -1.29760766e+00
-2.35971108e-01 -4.81454253e-01 -3.33186299e-01 4.92107719e-01
6.73714459e-01 -5.87921262e-01 -6.33588612e-01 5.22616804e-01] | [9.413105964660645, -2.3511838912963867] |
b74755a3-f775-46e4-b13c-22871725e4ee | diffusion-based-generation-optimization-and | 2301.06015 | null | https://arxiv.org/abs/2301.06015v1 | https://arxiv.org/pdf/2301.06015v1.pdf | Diffusion-based Generation, Optimization, and Planning in 3D Scenes | We introduce SceneDiffuser, a conditional generative model for 3D scene understanding. SceneDiffuser provides a unified model for solving scene-conditioned generation, optimization, and planning. In contrast to prior works, SceneDiffuser is intrinsically scene-aware, physics-based, and goal-oriented. With an iterative sampling strategy, SceneDiffuser jointly formulates the scene-aware generation, physics-based optimization, and goal-oriented planning via a diffusion-based denoising process in a fully differentiable fashion. Such a design alleviates the discrepancies among different modules and the posterior collapse of previous scene-conditioned generative models. We evaluate SceneDiffuser with various 3D scene understanding tasks, including human pose and motion generation, dexterous grasp generation, path planning for 3D navigation, and motion planning for robot arms. The results show significant improvements compared with previous models, demonstrating the tremendous potential of SceneDiffuser for the broad community of 3D scene understanding. | ['Song-Chun Zhu', 'Wei Liang', 'Yixin Zhu', 'Tengyu Liu', 'Baoxiong Jia', 'Puhao Li', 'Zan Wang', 'Siyuan Huang'] | 2023-01-15 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Huang_Diffusion-Based_Generation_Optimization_and_Planning_in_3D_Scenes_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Huang_Diffusion-Based_Generation_Optimization_and_Planning_in_3D_Scenes_CVPR_2023_paper.pdf | cvpr-2023-1 | ['grasp-generation', 'motion-planning'] | ['computer-vision', 'robots'] | [ 8.57294425e-02 2.22055197e-01 9.40242633e-02 -4.04138297e-01
-5.49745977e-01 -5.31217217e-01 7.38965511e-01 -1.92500189e-01
2.81974405e-01 2.14466691e-01 2.61289924e-01 -3.26173037e-01
-3.03188175e-01 -8.40666294e-01 -7.04046786e-01 -5.79041243e-01
4.20997776e-02 9.24010694e-01 -1.67605802e-01 -3.09233785e-01
4.61837679e-01 6.74490154e-01 -1.18894386e+00 -1.63722098e-01
1.13150406e+00 5.30765712e-01 1.02125120e+00 9.20660257e-01
-7.62812272e-02 7.22040951e-01 -6.31880090e-02 6.73205405e-02
1.16489887e-01 -3.50358248e-01 -7.98116982e-01 5.52680433e-01
3.05073351e-01 -3.50095481e-01 -5.12316346e-01 9.10602868e-01
3.12507182e-01 3.94857079e-01 1.04682457e+00 -1.09283745e+00
-6.52364731e-01 1.69688940e-01 -3.69719267e-01 -3.89768749e-01
7.26912081e-01 3.48404884e-01 7.14975178e-01 -9.36688960e-01
8.85518610e-01 1.67684710e+00 4.49777454e-01 5.16327202e-01
-1.29994154e+00 -1.14163183e-01 3.94946784e-01 -1.02069058e-01
-1.03511178e+00 -4.60843258e-02 8.17719758e-01 -6.71367824e-01
1.03709114e+00 -1.84408411e-01 7.66256630e-01 1.16778135e+00
7.93555260e-01 1.09291637e+00 7.40224123e-01 -6.48011193e-02
4.99422669e-01 -4.58734065e-01 -2.07271595e-02 9.41582680e-01
-5.27941845e-02 3.58025938e-01 -6.84836686e-01 -2.35069320e-02
1.44626963e+00 -1.10660732e-01 -1.35362402e-01 -1.04279399e+00
-1.31104076e+00 7.04999685e-01 5.02045929e-01 -3.48413378e-01
-6.15928292e-01 5.36497533e-01 -1.85640991e-01 -2.27366969e-01
5.37169397e-01 6.30195081e-01 -3.98365289e-01 -1.73271701e-01
-5.97052753e-01 7.00263798e-01 1.06468534e+00 1.62490857e+00
6.90055370e-01 3.15058619e-01 -3.62271927e-02 4.65648651e-01
9.32899952e-01 8.74773979e-01 -1.94565371e-01 -1.46228695e+00
2.88916200e-01 3.80534202e-01 1.79771706e-01 -9.96817648e-01
-4.37727928e-01 -3.82675260e-01 -7.28309155e-01 3.37824672e-01
8.18536580e-02 -6.78554550e-02 -1.45962238e+00 1.63928318e+00
4.80468184e-01 1.45746812e-01 1.98387504e-02 9.66190755e-01
9.21085060e-01 7.66848266e-01 2.00012699e-01 2.76430488e-01
8.20630729e-01 -1.11861193e+00 -4.86439079e-01 -4.98737365e-01
3.56817007e-01 -6.90714419e-01 8.24720919e-01 4.97755140e-01
-1.16193581e+00 -5.40185094e-01 -7.47740090e-01 -1.69579461e-01
1.81229953e-02 -1.00167245e-01 1.19085848e+00 4.03467566e-01
-1.15117931e+00 5.48010290e-01 -1.12604821e+00 -3.51617247e-01
2.55106539e-01 -6.07421808e-02 -3.25928926e-01 -3.35429192e-01
-4.31008786e-01 1.06503236e+00 2.94853568e-01 -3.97470361e-03
-1.71542442e+00 -7.99084425e-01 -1.30101657e+00 -2.91535884e-01
2.63202578e-01 -1.46704543e+00 1.43304479e+00 2.13434100e-01
-1.78471804e+00 8.18309605e-01 -3.20178956e-01 -1.89736411e-01
2.37181187e-01 -5.18988550e-01 4.43517536e-01 -4.44823541e-02
2.82282978e-01 7.46309757e-01 8.41000080e-01 -1.55034256e+00
-1.67902470e-01 -4.35373545e-01 2.09059000e-01 8.39561939e-01
7.00925648e-01 -7.49049962e-01 -7.61772335e-01 -2.99226254e-01
9.05136526e-01 -8.63400638e-01 -9.12293434e-01 -4.27667014e-02
-6.84857905e-01 -6.87574446e-02 6.48197472e-01 -3.47204834e-01
3.53971392e-01 -1.92308414e+00 9.30043399e-01 -1.12371584e-02
2.40339682e-01 -6.08899355e-01 -3.59195203e-01 5.21928012e-01
1.83928266e-01 -1.50006413e-01 -3.56137753e-01 -7.66095817e-01
4.01925325e-01 4.62277651e-01 -3.08260739e-01 3.51870149e-01
2.07336143e-01 1.10293686e+00 -1.30333710e+00 -2.57858008e-01
8.59167039e-01 4.02759939e-01 -1.00355828e+00 4.21307951e-01
-8.81305993e-01 9.62479651e-01 -8.81026030e-01 7.38651454e-01
7.00822532e-01 -2.79067308e-01 6.20359890e-02 -1.57950670e-01
-1.12465382e-01 3.31226885e-01 -9.90464270e-01 2.76324844e+00
-6.31226897e-01 1.75357535e-01 4.25849289e-01 -6.01191759e-01
8.61653328e-01 8.51588603e-03 6.14653587e-01 -3.29312474e-01
-3.51459570e-02 -3.85038145e-02 -6.55892372e-01 -3.85891050e-01
8.65813792e-01 -1.02076747e-01 -3.11242551e-01 3.76253128e-01
3.77571702e-01 -1.60853064e+00 -1.69565022e-01 5.77633858e-01
9.54814434e-01 1.09123790e+00 -1.71831790e-02 -4.26417261e-01
-2.36909568e-01 4.85690206e-01 1.83785602e-01 8.29379141e-01
1.19855680e-01 8.19409668e-01 4.55019623e-03 5.43110780e-02
-9.37443912e-01 -1.64136028e+00 9.17722359e-02 6.49576366e-01
7.96765029e-01 -3.39255333e-01 -5.68306267e-01 -2.08524391e-01
1.33028492e-01 1.13337493e+00 -3.31879526e-01 -1.51034355e-01
-3.71297210e-01 -6.48979604e-01 -6.98831454e-02 3.30318749e-01
4.36428070e-01 -8.95928204e-01 -5.73701918e-01 3.89947981e-01
-2.97195792e-01 -1.03236091e+00 -2.40383402e-01 2.45629743e-01
-1.15687835e+00 -1.04029989e+00 -6.13197863e-01 -5.67841709e-01
6.59923315e-01 5.01009524e-01 1.37598801e+00 -5.28842390e-01
-4.15492743e-01 1.15431201e+00 -1.60682008e-01 -3.12567383e-01
-4.56391573e-01 -1.55464366e-01 -5.21734841e-02 -7.46987045e-01
-2.11328879e-01 -6.83418572e-01 -4.86960977e-01 9.51953530e-02
-5.31465471e-01 4.06383514e-01 5.64149201e-01 6.31515682e-01
6.97195590e-01 -1.67520285e-01 1.89970285e-02 -4.05936718e-01
6.08525038e-01 -6.49962842e-01 -4.12445545e-01 -7.71880671e-02
-2.64523387e-01 1.33097405e-02 -5.75483143e-02 -1.60413831e-01
-1.49532378e+00 2.82114953e-01 -1.91585317e-01 -4.35592979e-01
-5.31522691e-01 4.13998455e-01 -3.30033928e-01 -2.09695790e-02
5.55750787e-01 2.91878253e-01 -1.49958864e-01 -4.89369035e-01
9.71920013e-01 -1.40885308e-01 7.58960545e-01 -9.60983574e-01
6.78160489e-01 4.88991290e-01 1.09684609e-01 -1.06748235e+00
-6.79318368e-01 -4.18253094e-01 -7.77684748e-01 -2.76729167e-01
1.11581624e+00 -9.12810802e-01 -5.51644564e-01 6.95840299e-01
-1.42170393e+00 -7.68419445e-01 -4.80568856e-01 5.51598251e-01
-1.22057402e+00 5.35503209e-01 -5.47501385e-01 -9.98173237e-01
7.83490688e-02 -1.25343812e+00 1.61776161e+00 1.17090039e-01
-3.36339206e-01 -1.09970903e+00 1.59807112e-02 2.16359258e-01
6.85750842e-02 5.68901360e-01 9.37529564e-01 2.31032625e-01
-1.35507143e+00 9.30593237e-02 -1.13151610e-01 -1.92032501e-01
1.39758185e-01 -3.25593412e-01 -7.52093256e-01 1.14888668e-01
9.61617529e-02 -3.07521969e-01 7.12011158e-01 8.99565697e-01
7.74316609e-01 2.00414851e-01 -4.37460035e-01 8.32240283e-01
1.23199928e+00 3.18617553e-01 5.36858499e-01 -3.28849293e-02
7.77115643e-01 3.55947971e-01 7.56977081e-01 5.56758702e-01
6.45903289e-01 3.83037835e-01 9.70043004e-01 1.67641025e-02
-1.54171839e-01 -7.41442859e-01 2.21628308e-01 6.85714543e-01
-1.00114448e-02 -5.19045293e-01 -7.98210919e-01 6.25636101e-01
-1.81803262e+00 -5.39934099e-01 -3.93737704e-01 1.66913128e+00
3.18738699e-01 -1.76748753e-01 -4.32624340e-01 -4.32755589e-01
1.40569165e-01 2.44683728e-01 -8.33916724e-01 -1.79832578e-01
1.08154200e-01 9.89367291e-02 2.67028213e-01 7.63673007e-01
-9.66423333e-01 1.29464781e+00 7.32846594e+00 6.31133020e-01
-4.52632189e-01 -8.54145065e-02 1.19892024e-01 2.77314037e-01
-7.66787052e-01 2.39250898e-01 -6.66181028e-01 -2.22439826e-01
1.40443221e-01 -4.90091229e-03 3.25772256e-01 1.08253968e+00
3.11180502e-01 -4.61002916e-01 -1.25250351e+00 8.99333775e-01
-5.33082150e-02 -1.22327435e+00 -3.37669849e-02 1.14235617e-01
8.73526871e-01 1.66505143e-01 4.05937396e-02 4.71668877e-02
1.08645093e+00 -1.00002348e+00 9.84844267e-01 7.31505930e-01
3.16406280e-01 -4.47961569e-01 9.76145491e-02 6.34086788e-01
-1.10802495e+00 3.56408507e-01 -2.86665589e-01 1.45376278e-02
1.03298581e+00 7.87866056e-01 -6.99179471e-01 8.43388081e-01
5.45187652e-01 9.70300019e-01 2.17464283e-01 9.39223349e-01
-5.26394427e-01 2.14032844e-01 -3.91620636e-01 2.15426534e-02
4.07088012e-01 -5.16885936e-01 1.19502592e+00 8.36161792e-01
2.80123800e-01 2.22233489e-01 5.41115105e-01 1.49750125e+00
5.17230809e-01 -4.57287878e-01 -9.06152904e-01 -6.12983108e-02
1.43185064e-01 8.75762284e-01 -6.58492029e-01 -1.26773119e-01
3.14645737e-01 1.24970055e+00 -3.64100225e-02 6.32024705e-01
-6.67725861e-01 1.10745654e-02 7.31210053e-01 -2.06301972e-01
7.56608918e-02 -1.34799290e+00 -6.71324074e-01 -1.03469944e+00
-4.56035763e-01 -3.02255541e-01 -2.54929751e-01 -1.09680462e+00
-1.10333574e+00 1.17109500e-01 4.73395020e-01 -7.81986415e-01
-3.48894119e-01 -7.83658206e-01 -4.11808878e-01 8.40377152e-01
-1.24976170e+00 -1.15875483e+00 -4.39151376e-01 3.85149151e-01
9.68877137e-01 2.33074069e-01 8.04462194e-01 -3.04364562e-01
-1.04917459e-01 -4.48722839e-01 -2.93258369e-01 -5.29889166e-01
2.77088374e-01 -1.47876894e+00 8.80791008e-01 8.05273831e-01
-1.17344119e-01 7.30714023e-01 8.01812172e-01 -1.14161360e+00
-2.02496648e+00 -8.79436255e-01 2.98670262e-01 -7.41157055e-01
3.82874697e-01 -3.75966758e-01 -3.51216674e-01 6.47422373e-01
1.04107754e-02 -8.50052118e-01 1.73795119e-01 9.00151730e-02
1.21987581e-01 7.13935077e-01 -1.04248154e+00 7.91855514e-01
1.72835112e+00 -2.45922551e-01 -6.22623384e-01 5.28062999e-01
8.33061993e-01 -1.03122771e+00 -7.91330338e-01 5.09771287e-01
4.46218938e-01 -8.48379910e-01 1.33808124e+00 -3.41319233e-01
5.85007846e-01 -2.60602385e-01 -4.16532278e-01 -1.41678965e+00
-4.58493680e-01 -8.05709600e-01 -3.34346443e-01 5.01044631e-01
2.24457428e-01 -3.58335167e-01 8.90918553e-01 7.08436191e-01
-8.87142837e-01 -4.46205050e-01 -4.85041648e-01 -5.48527896e-01
3.05222776e-02 -8.49713027e-01 4.33787197e-01 4.80776578e-01
-2.95058340e-01 2.54370540e-01 -3.08172941e-01 4.05129552e-01
1.13097370e+00 4.18737739e-01 1.15244257e+00 -1.03749168e+00
-3.75632554e-01 -2.65429616e-01 -9.22780856e-02 -2.23626208e+00
8.64355564e-02 -8.52581680e-01 7.57740259e-01 -2.32434845e+00
4.43550907e-02 -7.04335690e-01 6.41323805e-01 2.01601133e-01
1.45879462e-01 -4.13097262e-01 2.17326775e-01 -6.65079057e-02
-3.94264340e-01 1.01683676e+00 2.07435155e+00 -8.56759325e-02
-4.19470787e-01 -9.48433056e-02 -5.13068974e-01 8.84484708e-01
4.21193242e-01 -5.02741523e-02 -8.29634607e-01 -7.58593500e-01
1.16468675e-01 2.91563183e-01 7.50168741e-01 -7.20910132e-01
3.24066997e-01 -6.37581408e-01 2.27591947e-01 -1.13322639e+00
8.21755707e-01 -5.12224674e-01 1.62243783e-01 4.38011616e-01
9.59131420e-02 -2.97232240e-01 3.00180882e-01 9.35687482e-01
8.84321705e-02 2.53719985e-02 4.07192141e-01 -5.63642025e-01
-1.25081265e+00 5.40351927e-01 -6.95290089e-01 -1.23641886e-01
7.10272610e-01 -4.63541389e-01 3.98118496e-02 -6.56928480e-01
-1.07857203e+00 3.78912598e-01 5.06605566e-01 5.19748807e-01
1.13185573e+00 -1.08038270e+00 -5.39906025e-01 1.94220364e-01
-1.57617539e-01 8.32800150e-01 5.26986361e-01 3.91370028e-01
-6.15479708e-01 2.77842939e-01 1.12880327e-01 -1.12479150e+00
-5.72311819e-01 9.10005569e-02 1.75202593e-01 -8.95097852e-02
-7.17224836e-01 1.20693040e+00 7.03453600e-01 -1.07752419e+00
-7.94583410e-02 -5.85327923e-01 3.41675282e-01 -7.99843371e-01
-1.81889892e-01 4.67926949e-01 -5.32354712e-01 -4.00377512e-01
-1.49214879e-01 7.85071671e-01 4.20907259e-01 -3.18876952e-01
1.20816970e+00 -4.02152032e-01 -2.49622948e-03 2.98571974e-01
5.83592117e-01 -5.31322658e-01 -1.67182791e+00 2.33748972e-01
-4.02662188e-01 -2.56215572e-01 8.18969086e-02 -6.03363216e-01
-5.82641840e-01 9.97574091e-01 8.12112093e-02 -2.72299796e-01
6.38464749e-01 2.48539761e-01 6.36724174e-01 3.62747282e-01
1.09798849e+00 -7.88314521e-01 4.07449394e-01 1.20232010e+00
1.21112347e+00 -9.70786154e-01 1.00038826e-01 -9.11185443e-01
-4.46425110e-01 7.82876730e-01 7.33496845e-01 -3.11289281e-01
8.43727648e-01 3.70118141e-01 -2.58536220e-01 -6.10870063e-01
-5.00638545e-01 -2.44863346e-01 5.55209816e-01 8.88321936e-01
-2.60960460e-02 1.44170359e-01 4.17616695e-01 2.44010046e-01
-4.04479414e-01 -1.16811238e-01 1.01032801e-01 1.13846397e+00
-5.17540276e-01 -9.21842754e-01 -2.26063311e-01 -1.10665761e-01
4.76326734e-01 4.35585761e-03 -3.31799060e-01 7.07239628e-01
-1.16038263e-01 1.04344249e+00 -1.11990213e-01 -3.41186613e-01
2.43261576e-01 -2.27965340e-01 1.02382910e+00 -9.89809811e-01
1.89072251e-01 2.08096310e-01 7.30051100e-02 -1.08735836e+00
-3.82312387e-01 -4.42862749e-01 -1.40400696e+00 -2.80795936e-02
-4.04374272e-01 -2.23660693e-01 1.02885497e+00 9.26152408e-01
5.23987055e-01 6.57678068e-01 2.77980953e-01 -1.55721533e+00
-2.10249305e-01 -7.88574755e-01 -4.17366832e-01 1.55171439e-01
7.03085437e-02 -9.82096195e-01 -1.40606046e-01 1.37782633e-01] | [8.911052703857422, -3.053144693374634] |
ac1ae2a3-d7d2-4f92-b54d-aa9f9bc0c61b | unsupervised-domain-adaptation-by-adversarial | 1807.11284 | null | http://arxiv.org/abs/1807.11284v1 | http://arxiv.org/pdf/1807.11284v1.pdf | Unsupervised Domain Adaptation by Adversarial Learning for Robust Speech Recognition | In this paper, we investigate the use of adversarial learning for
unsupervised adaptation to unseen recording conditions, more specifically,
single microphone far-field speech. We adapt neural networks based acoustic
models trained with close-talk clean speech to the new recording conditions
using untranscribed adaptation data. Our experimental results on Italian
SPEECON data set show that our proposed method achieves 19.8% relative word
error rate (WER) reduction compared to the unadapted models. Furthermore, this
adaptation method is beneficial even when performed on data from another
language (i.e. French) giving 12.6% relative WER reduction. | ['Ngoc Thang Vu', 'Pavel Denisov', 'Marc Ferras Font'] | 2018-07-30 | null | null | null | null | ['robust-speech-recognition'] | ['speech'] | [ 3.74788642e-01 2.15851769e-01 6.77180529e-01 -4.08737719e-01
-1.35752285e+00 -6.09417021e-01 2.01820329e-01 -4.22101080e-01
-7.36991107e-01 7.95945883e-01 2.95255333e-01 -2.23619089e-01
3.49500299e-01 -3.01674277e-01 -9.30030763e-01 -8.72351289e-01
-2.00411826e-02 -8.28128681e-02 9.49827395e-03 -1.57953113e-01
-3.46522152e-01 4.93778706e-01 -1.02984381e+00 -1.20683156e-01
5.09554267e-01 6.42342627e-01 2.15676203e-01 1.17772090e+00
2.78594226e-01 3.24513942e-01 -1.26355660e+00 -3.42057139e-01
3.30105841e-01 -2.22175047e-01 -4.51533943e-01 -2.39233240e-01
5.08573353e-01 -2.48841599e-01 -5.46971858e-01 1.09278238e+00
9.42551851e-01 6.88706160e-01 3.37537974e-01 -3.60068917e-01
-6.06425047e-01 8.49503517e-01 1.46360099e-01 5.68541825e-01
2.35010564e-01 -2.73232460e-02 4.78664219e-01 -1.09793949e+00
3.83757390e-02 1.07319558e+00 5.43224514e-01 9.49430645e-01
-9.30709898e-01 -9.07116830e-01 9.59665850e-02 9.96919442e-03
-1.58783531e+00 -1.12694263e+00 7.34901130e-01 -6.02766033e-03
1.22874475e+00 5.82620502e-01 -1.29630998e-01 1.50360692e+00
1.53370332e-02 2.78374970e-01 9.47661400e-01 -6.22605741e-01
3.19217652e-01 5.07950336e-02 -1.94787145e-01 -5.21454550e-02
-2.20750093e-01 3.21360439e-01 -4.69329715e-01 2.71220244e-02
1.79444626e-01 -6.65699422e-01 -5.44867814e-01 4.02971417e-01
-9.29429948e-01 4.61261839e-01 1.78615272e-01 5.22631466e-01
-2.49257907e-01 -2.99079008e-02 4.07074213e-01 4.54221517e-01
7.13501096e-01 3.37194264e-01 -6.95764959e-01 -3.91568363e-01
-6.52888238e-01 -3.43098819e-01 8.63415897e-01 8.64571154e-01
3.50430518e-01 1.05711234e+00 2.56263256e-01 1.49726927e+00
2.81240612e-01 1.08119535e+00 7.89698720e-01 -6.43576920e-01
5.56615293e-01 -6.28986299e-01 -1.70063019e-01 -4.61644232e-01
-1.27367586e-01 -6.85187161e-01 -7.58300066e-01 -5.91561720e-02
1.19051009e-01 -6.93298638e-01 -1.28398263e+00 1.87142003e+00
7.45098591e-02 3.76460522e-01 7.09831774e-01 6.77873373e-01
7.99416900e-01 1.01153302e+00 -1.84068158e-02 -4.33031857e-01
4.26041722e-01 -1.00068033e+00 -1.14080203e+00 -4.28835154e-01
1.93079129e-01 -1.02630234e+00 9.73568380e-01 6.17397964e-01
-1.02823532e+00 -9.08135474e-01 -1.02834737e+00 5.07270992e-01
-3.73834521e-01 -1.58352658e-01 -2.66452402e-01 1.18641901e+00
-1.12360275e+00 2.22119406e-01 -7.92319715e-01 -2.58809954e-01
-2.01539069e-01 5.47593653e-01 -4.43844616e-01 1.12615079e-01
-1.42534959e+00 5.53281844e-01 4.46521044e-01 3.68151009e-01
-1.04832518e+00 -3.02848369e-01 -9.67461050e-01 -2.12350488e-02
2.39743069e-01 8.44910741e-02 1.46581590e+00 -9.68276381e-01
-2.29658151e+00 2.93993711e-01 -2.38716274e-01 -6.32486582e-01
7.17340633e-02 -5.80178320e-01 -1.22925413e+00 -3.95148508e-02
-4.86479610e-01 7.87434056e-02 9.87355232e-01 -1.24962687e+00
-2.13785097e-01 -3.74837257e-02 -3.54207039e-01 2.95937896e-01
-6.32728040e-01 2.82503009e-01 -2.15279892e-01 -1.03866577e+00
-5.03304191e-02 -9.06565428e-01 -1.17066406e-01 -7.54139721e-01
-4.19781268e-01 4.77927513e-02 9.79783297e-01 -1.14935803e+00
1.28443336e+00 -2.47613907e+00 4.73619998e-02 3.64337742e-01
-4.17650938e-01 8.82512689e-01 -1.40278116e-01 2.41011962e-01
-2.02384904e-01 4.67428863e-02 -4.22703683e-01 -4.36533421e-01
-1.84852004e-01 4.33459461e-01 -1.71648011e-01 2.35346988e-01
1.16005510e-01 4.16799575e-01 -6.79744005e-01 7.12376684e-02
3.54344666e-01 6.79352522e-01 -3.86660188e-01 7.03314185e-01
4.69634056e-01 7.63958931e-01 -5.93806989e-02 5.11275232e-01
6.36323392e-01 8.11876178e-01 -1.71903357e-01 1.03460021e-01
2.21422464e-01 2.27208748e-01 -9.40988779e-01 1.73061228e+00
-1.01179469e+00 7.67114639e-01 5.19461811e-01 -1.03466284e+00
1.24536741e+00 6.92752600e-01 -1.56267419e-01 -4.06777591e-01
1.18545793e-01 3.79707247e-01 2.61848986e-01 -4.90492851e-01
2.55246222e-01 -3.70389998e-01 -1.92029998e-01 -2.19468921e-01
3.71454000e-01 -1.55178413e-01 -4.46626186e-01 -3.12342733e-01
1.11791122e+00 -2.43823186e-01 1.87804386e-01 -9.93928984e-02
7.95204997e-01 -7.97389030e-01 4.85583335e-01 7.71955669e-01
-4.29225147e-01 6.49600208e-01 -6.38214231e-01 -3.58107835e-02
-7.22495735e-01 -1.51437533e+00 -1.74962983e-01 1.05142987e+00
-4.17091101e-01 -1.11736685e-01 -9.57107842e-01 -5.03819823e-01
-4.43788230e-01 1.00890386e+00 -2.19281986e-01 -2.74729013e-01
-1.01967895e+00 -4.02822196e-01 1.00004458e+00 5.72531283e-01
3.42909873e-01 -9.84493315e-01 1.99445158e-01 4.14517522e-01
-7.69389942e-02 -1.44286919e+00 -6.21525645e-01 5.47552586e-01
-4.11453009e-01 -3.46063346e-01 -8.22117388e-01 -9.78211403e-01
3.61199349e-01 -2.71681279e-01 7.52368391e-01 -5.35618663e-01
4.82926577e-01 4.90252405e-01 -5.04342735e-01 -5.80044687e-01
-1.04657149e+00 4.99357358e-02 6.81650400e-01 1.97179973e-01
3.81136090e-01 -6.19253516e-01 -1.53346628e-01 3.49943966e-01
-8.12570333e-01 -8.56548131e-01 3.60656321e-01 9.93478537e-01
5.86935282e-01 1.39545470e-01 9.79135513e-01 -6.37189567e-01
6.37680888e-01 -3.17970842e-01 -3.51554543e-01 1.04801990e-01
-3.31808835e-01 -2.22828493e-01 1.11489034e+00 -6.70363784e-01
-1.47175741e+00 4.84000891e-02 -1.03827405e+00 -4.40891832e-01
-6.73678339e-01 2.14398861e-01 -5.73111475e-01 -1.24313310e-01
8.25390995e-01 8.02022293e-02 -4.87375379e-01 -8.08939695e-01
1.43357605e-01 1.43236172e+00 1.01960909e+00 -2.77181387e-01
9.82361317e-01 -1.10541604e-01 -6.58300042e-01 -1.16383445e+00
-1.99295789e-01 -3.90488744e-01 -5.57853103e-01 -4.56138849e-02
7.81754196e-01 -8.38591754e-01 -3.92775796e-02 6.36246681e-01
-9.75070655e-01 -2.97850162e-01 -2.74470150e-01 9.17963743e-01
-3.49478185e-01 3.43658984e-01 -5.60476124e-01 -1.06968236e+00
-5.99341452e-01 -9.76832986e-01 8.03080499e-01 1.29123077e-01
-1.82094231e-01 -1.14739943e+00 1.98499143e-01 2.84794003e-01
6.36672795e-01 -8.21813494e-02 3.01369011e-01 -1.20184481e+00
1.81458920e-01 -5.11421144e-01 7.31330752e-01 1.35627568e+00
4.50288683e-01 -3.10069352e-01 -1.58694100e+00 -5.07813811e-01
3.38554978e-01 -1.25504345e-01 4.31403846e-01 3.10947329e-01
1.14729464e+00 -4.92901742e-01 1.65380150e-01 6.92928731e-01
1.14405346e+00 7.86465108e-01 5.22445142e-01 -1.12928450e-01
6.68333113e-01 1.50696993e-01 2.95635581e-01 -5.34026958e-02
-3.63312632e-01 6.95421576e-01 1.35123402e-01 -1.93780065e-02
-3.96860540e-01 -2.46735960e-01 6.79643095e-01 1.73509121e+00
2.67441660e-01 -7.78003514e-01 -9.30935979e-01 5.63955605e-01
-1.03811574e+00 -6.23869002e-01 1.84001833e-01 2.39089561e+00
8.19086909e-01 2.48970032e-01 -2.88762838e-01 3.26566905e-01
8.53708267e-01 1.08698010e-01 -3.85433286e-01 -1.13604689e+00
-2.17250884e-01 7.57348418e-01 5.45939863e-01 9.86786485e-01
-1.07522058e+00 8.78012300e-01 6.84246922e+00 9.46691036e-01
-1.13835764e+00 5.40364027e-01 4.29899305e-01 -8.18372220e-02
-1.40865799e-03 -5.83362162e-01 -3.82604808e-01 3.68149549e-01
1.72167039e+00 8.04806501e-02 6.42819703e-01 6.16368294e-01
2.36382291e-01 3.60584348e-01 -8.79096210e-01 9.53388453e-01
3.32198232e-01 -4.07876819e-01 -2.88018614e-01 -3.83830249e-01
6.93335712e-01 3.53740603e-01 8.98786485e-02 4.45318878e-01
5.89770898e-02 -1.08321190e+00 3.59744221e-01 2.69630671e-01
1.13657737e+00 -7.95399845e-01 9.81630325e-01 3.70602846e-01
-8.42385530e-01 1.09984517e-01 -3.00614804e-01 2.90342923e-02
2.81133533e-01 2.92387158e-01 -1.17207658e+00 4.45707887e-01
8.09376955e-01 1.58289984e-01 -1.35683000e-01 7.98776686e-01
-1.54802859e-01 1.41667747e+00 -4.56593692e-01 4.95796129e-02
1.56763270e-02 3.17110956e-01 9.83876348e-01 1.58444285e+00
5.34057975e-01 -1.15780011e-02 -2.45686099e-01 1.50355607e-01
-2.22441480e-01 1.05852507e-01 -7.76342452e-01 1.88907325e-01
5.66701829e-01 6.46413624e-01 1.95659362e-02 -8.31908435e-02
-3.16461116e-01 1.20928609e+00 1.90922860e-02 9.01017964e-01
-6.02524519e-01 -8.01497817e-01 5.72646976e-01 -3.65760803e-01
3.69429976e-01 -1.99505121e-01 5.98288327e-02 -9.21982050e-01
1.54196650e-01 -9.08076942e-01 -8.09463263e-02 -7.24754632e-01
-1.26402926e+00 1.13619614e+00 -3.47208716e-02 -1.00994551e+00
-6.66549146e-01 -5.36199391e-01 -6.01070166e-01 1.24811518e+00
-1.25597405e+00 -6.77496314e-01 1.82205722e-01 5.56319356e-01
9.85536277e-01 -6.14673793e-01 1.42966962e+00 4.95175004e-01
-5.19147575e-01 1.23866165e+00 6.56372309e-01 2.14785367e-01
8.23492467e-01 -1.23179567e+00 7.49605834e-01 1.25955904e+00
3.50684732e-01 4.03871596e-01 8.45133781e-01 -2.61542916e-01
-1.06933713e+00 -1.13981783e+00 8.71062756e-01 -4.23285514e-01
4.96269703e-01 -6.97387993e-01 -1.14672804e+00 6.39546275e-01
5.07415593e-01 4.23582345e-02 9.49655056e-01 -1.13954157e-01
-1.88272059e-01 -3.94802868e-01 -1.31995225e+00 3.13463420e-01
7.79749751e-01 -7.40916133e-01 -7.65375853e-01 1.38315856e-01
1.09420729e+00 -6.53207958e-01 -9.35447991e-01 5.51911235e-01
2.83616662e-01 -4.62993175e-01 9.61957455e-01 -3.21915835e-01
-3.67107242e-01 -1.11721337e-01 -7.89675415e-01 -1.85983682e+00
-3.95268342e-03 -1.10206628e+00 -2.23963439e-01 1.43560731e+00
6.85291588e-01 -7.65131295e-01 2.71024793e-01 2.85411268e-01
-6.50015891e-01 -3.67025465e-01 -1.43282962e+00 -1.12153828e+00
4.36705083e-01 -9.03739750e-01 4.08907294e-01 6.24781311e-01
-2.93591440e-01 3.49903315e-01 -5.86743236e-01 6.53715491e-01
2.32190162e-01 -9.61330295e-01 6.23623908e-01 -4.80114490e-01
-5.16315818e-01 1.58082545e-01 -4.39569563e-01 -9.62126076e-01
4.05530870e-01 -5.77590823e-01 6.22211397e-01 -9.95012164e-01
-6.98606610e-01 -8.21398646e-02 -8.20982933e-01 2.18392581e-01
-2.89128780e-01 4.03614879e-01 5.22030257e-02 -3.02163124e-01
-1.79393262e-01 5.18054605e-01 6.94214642e-01 -2.29390651e-01
-1.91009894e-01 3.67225051e-01 -3.29760998e-01 6.87152803e-01
1.17181599e+00 -5.43308616e-01 -3.82100165e-01 -8.01179588e-01
-5.57384074e-01 -1.17054552e-01 -1.42535731e-01 -1.41878247e+00
-4.00048718e-02 1.70279443e-01 1.48499116e-01 -4.78661209e-02
6.07341051e-01 -1.09141123e+00 4.98836599e-02 2.38592476e-01
-3.52122754e-01 -8.34600031e-02 7.31433392e-01 7.47679830e-01
-4.37845200e-01 -3.21206182e-01 7.01454341e-01 3.27314764e-01
-4.83768731e-01 -2.73354113e-01 -4.82838184e-01 8.68861303e-02
5.94029486e-01 -6.21017590e-02 8.31012949e-02 -7.20257580e-01
-1.18585527e+00 -3.89816642e-01 -5.12249805e-02 5.95134974e-01
7.02720940e-01 -1.29139233e+00 -1.03173590e+00 6.41436160e-01
1.75895598e-02 -2.10194096e-01 2.92836070e-01 1.92055881e-01
-1.77630395e-01 3.80043715e-01 3.77707511e-01 -4.03662294e-01
-1.36609077e+00 3.59509826e-01 5.81008255e-01 1.56008631e-01
-2.97631949e-01 1.20064604e+00 8.49032849e-02 -6.68668330e-01
5.62863529e-01 -3.43192875e-01 -4.82777320e-02 -7.78187156e-01
4.45348442e-01 2.79762328e-01 5.00346303e-01 -9.50573444e-01
-4.59853768e-01 4.27651912e-01 2.23416656e-01 -4.05824035e-01
9.75282729e-01 -2.49897435e-01 4.67511475e-01 7.38524079e-01
1.51582134e+00 7.37466335e-01 -9.98132229e-01 -3.92436951e-01
-3.47527713e-01 -3.20919394e-01 7.01825619e-02 -1.15105510e+00
-9.30086076e-01 8.99620473e-01 1.03391087e+00 2.21143559e-01
1.44692576e+00 -1.30671442e-01 9.23664570e-01 7.21292257e-01
1.25598326e-01 -1.19717944e+00 -4.41976607e-01 5.63450813e-01
9.44386959e-01 -1.05156505e+00 -7.14160204e-01 -1.38748348e-01
-3.93070102e-01 9.12817240e-01 4.23400581e-01 -3.22642140e-02
8.54740739e-01 5.44526994e-01 6.34822011e-01 5.27733982e-01
-4.06378329e-01 1.62568942e-01 3.72310609e-01 1.06397450e+00
3.46895605e-01 2.81039536e-01 6.87722936e-02 5.87443352e-01
-6.17709339e-01 -5.69047928e-01 4.45388913e-01 5.60446918e-01
-2.36910224e-01 -9.86182630e-01 -6.91395938e-01 1.52455807e-01
-9.24558580e-01 -2.88314492e-01 -2.87624806e-01 4.93722469e-01
-5.72974831e-02 1.52042735e+00 -6.19640313e-02 -6.99047387e-01
6.70115948e-01 3.68673056e-01 -2.09109951e-02 -6.79166436e-01
-6.55537844e-01 4.04632509e-01 3.31535846e-01 -9.15590674e-02
-4.44117963e-01 -4.03035581e-01 -1.12898636e+00 1.28383771e-01
-4.45341498e-01 3.44566464e-01 9.52643037e-01 9.27067757e-01
7.00414628e-02 9.24939752e-01 9.56872106e-01 -4.70035911e-01
-6.83405697e-01 -1.40586841e+00 -3.42888176e-01 2.75415301e-01
7.57779181e-01 -9.15275589e-02 -7.65340805e-01 2.14209378e-01] | [14.737072944641113, 6.2922444343566895] |
b8e614c5-5afa-4abc-8df5-b43d3917544e | data2vec-a-general-framework-for-self-1 | 2202.03555 | null | https://arxiv.org/abs/2202.03555v3 | https://arxiv.org/pdf/2202.03555v3.pdf | data2vec: A General Framework for Self-supervised Learning in Speech, Vision and Language | While the general idea of self-supervised learning is identical across modalities, the actual algorithms and objectives differ widely because they were developed with a single modality in mind. To get us closer to general self-supervised learning, we present data2vec, a framework that uses the same learning method for either speech, NLP or computer vision. The core idea is to predict latent representations of the full input data based on a masked view of the input in a self-distillation setup using a standard Transformer architecture. Instead of predicting modality-specific targets such as words, visual tokens or units of human speech which are local in nature, data2vec predicts contextualized latent representations that contain information from the entire input. Experiments on the major benchmarks of speech recognition, image classification, and natural language understanding demonstrate a new state of the art or competitive performance to predominant approaches. | ['Michael Auli', 'Jiatao Gu', 'Arun Babu', 'Qiantong Xu', 'Wei-Ning Hsu', 'Alexei Baevski'] | 2022-02-07 | null | null | null | preprint-2022-1 | ['paraphrase-identification', 'linguistic-acceptability'] | ['natural-language-processing', 'natural-language-processing'] | [ 4.41257179e-01 4.30253655e-01 -4.93446648e-01 -5.55244029e-01
-6.03839219e-01 -4.70512003e-01 1.39564729e+00 -1.38172880e-01
-4.05477807e-02 5.03839135e-01 8.87456954e-01 -2.06811488e-01
4.35774833e-01 -5.99042952e-01 -6.80590928e-01 -8.40083182e-01
3.25437605e-01 4.33704942e-01 -2.11317435e-01 -1.62265763e-01
-1.14008501e-01 1.37482047e-01 -1.75852633e+00 9.66041565e-01
3.26987594e-01 9.36079681e-01 2.04018250e-01 6.22458994e-01
-6.03842974e-01 9.77712154e-01 -2.91116416e-01 -2.99304388e-02
-2.66267490e-02 -4.22949970e-01 -9.52965379e-01 4.95018363e-01
5.74480236e-01 1.19560704e-01 -8.56911480e-01 9.45039749e-01
3.29463482e-01 -4.97912206e-02 7.85378277e-01 -1.13725293e+00
-1.12527084e+00 7.89093375e-01 -5.26814461e-02 1.67137057e-01
5.10361075e-01 7.76586309e-02 1.07629263e+00 -1.17033303e+00
5.85698247e-01 1.42045319e+00 1.20800734e-01 8.52154970e-01
-1.59540308e+00 -2.90659457e-01 2.31689692e-01 4.00441319e-01
-1.14754689e+00 -8.15433919e-01 9.02148902e-01 -5.55786729e-01
1.12631929e+00 1.10256001e-01 1.65078551e-01 1.89198935e+00
1.01352975e-01 1.06758916e+00 1.43963146e+00 -8.01542640e-01
2.16475785e-01 6.44707263e-01 1.62616804e-01 4.99135345e-01
-4.30368125e-01 3.34636152e-01 -1.02442229e+00 1.00984767e-01
4.51476306e-01 8.15731138e-02 -1.58539921e-01 -7.62996793e-01
-1.77323103e+00 8.65693808e-01 2.52566338e-01 6.03102565e-01
-1.84543669e-01 -1.07027933e-01 4.98156041e-01 6.40666664e-01
5.72969377e-01 8.54654834e-02 -5.10144949e-01 -4.34475318e-02
-6.92395329e-01 -3.83240700e-01 7.22153902e-01 8.43516886e-01
7.73805141e-01 5.03257394e-01 -1.67484418e-01 6.89364552e-01
2.27489159e-01 6.68798029e-01 1.00224364e+00 -6.19606972e-01
4.80093777e-01 6.03601038e-01 -3.58104497e-01 -4.85109031e-01
-4.67133671e-02 -3.17296058e-01 -9.27750051e-01 1.79895967e-01
3.21819410e-02 1.55985728e-01 -1.22040057e+00 1.75130749e+00
-1.19098529e-01 1.47244349e-01 7.36098886e-01 6.13211572e-01
1.29783368e+00 1.00240612e+00 1.22088920e-02 -2.45135248e-01
1.24262452e+00 -1.22346509e+00 -7.44586527e-01 -6.40460730e-01
2.32885033e-01 -3.96811217e-01 1.02933836e+00 2.18787670e-01
-7.85513520e-01 -9.42115545e-01 -8.38401377e-01 -2.20130086e-01
-9.16051388e-01 1.16426861e-02 3.47951829e-01 4.68526393e-01
-1.18413424e+00 1.34436667e-01 -6.01274490e-01 -7.27348924e-01
1.68433413e-01 -1.68025903e-02 -8.86444449e-01 6.84183538e-02
-1.04483438e+00 1.05592394e+00 5.21155596e-01 -4.84205037e-01
-1.17318833e+00 -5.99541605e-01 -1.27822697e+00 1.02528229e-01
1.83675304e-01 -5.29410243e-01 1.14450586e+00 -1.34163451e+00
-1.46845007e+00 1.21599853e+00 -5.70601761e-01 -6.58005118e-01
1.80963844e-01 9.07332376e-02 -7.32551157e-01 1.52095445e-02
1.11073451e-02 7.13546336e-01 1.30228555e+00 -1.46462190e+00
-3.26585352e-01 -3.22609603e-01 -1.91141233e-01 2.63721466e-01
-5.76304972e-01 -1.82890907e-01 -5.91014549e-02 -5.56184828e-01
1.35985583e-01 -6.06168211e-01 9.71812904e-02 -2.67603725e-01
-4.02167052e-01 -3.16555738e-01 1.00098932e+00 -3.79954934e-01
6.41519964e-01 -2.38268805e+00 4.44638312e-01 -1.72417790e-01
2.55208284e-01 1.24279097e-01 -4.73656565e-01 7.68636405e-01
-6.56244636e-01 -1.53275549e-01 -2.50511393e-02 -5.95443487e-01
1.02677606e-01 5.26595771e-01 -1.00486219e+00 3.80960494e-01
6.41691685e-02 1.21186435e+00 -1.01187444e+00 -3.47274244e-01
6.22784078e-01 4.99719828e-01 -2.13720556e-02 3.88811558e-01
-2.25479782e-01 3.79267901e-01 2.05136612e-02 5.22928059e-01
3.38694066e-01 -3.84635538e-01 1.99107975e-01 -4.07305568e-01
-1.61428586e-01 6.96621478e-01 -8.68338585e-01 2.08573556e+00
-5.83368480e-01 1.01108444e+00 -1.41647726e-01 -1.64401495e+00
7.57608533e-01 7.37194717e-01 4.03888494e-01 -9.24371898e-01
-9.94761437e-02 -2.90982962e-01 -4.34989095e-01 -4.79969859e-01
1.66146711e-01 -3.23453933e-01 -8.48913491e-02 6.54855072e-01
7.87983060e-01 6.90412819e-02 -1.41188309e-01 3.30536336e-01
9.07595932e-01 -5.35241552e-02 4.79270846e-01 3.39806229e-02
3.85859162e-01 -1.66032638e-03 1.74488109e-02 8.67638409e-01
1.85701437e-02 6.89825475e-01 1.91197053e-01 -4.84002173e-01
-1.01486087e+00 -1.58476484e+00 -1.24019139e-01 1.33787203e+00
-1.59133255e-01 -6.03068948e-01 -1.26681075e-01 -8.89600456e-01
-1.16936758e-01 8.08130026e-01 -8.17620456e-01 -2.29146764e-01
-3.43080759e-02 -2.22034082e-01 2.78645664e-01 6.26653373e-01
2.41731629e-01 -1.25902307e+00 -2.66034693e-01 -7.91723505e-02
-1.82141140e-01 -1.41456711e+00 -1.36838518e-02 4.55563515e-01
-7.50019729e-01 -6.37868583e-01 -5.51836908e-01 -9.68295395e-01
6.07609332e-01 4.17309761e-01 1.35341287e+00 -5.05297422e-01
-9.66534242e-02 8.54195595e-01 -3.12029779e-01 -3.97098154e-01
-6.92094505e-01 -2.69381523e-01 3.68424326e-01 2.93520957e-01
4.48511094e-01 -4.64990318e-01 -3.74179892e-02 -1.43527612e-01
-8.36481452e-01 1.35879621e-01 6.31923556e-01 1.08471274e+00
3.40620369e-01 -1.27643391e-01 3.06839913e-01 -6.45713389e-01
3.46035123e-01 -5.94893277e-01 5.66424541e-02 3.72393996e-01
-5.93384802e-01 2.30664223e-01 6.51701152e-01 -5.77579260e-01
-1.06070578e+00 1.22494027e-01 1.52050659e-01 -7.22151935e-01
-8.40028763e-01 4.94042277e-01 -1.64787903e-01 3.69787902e-01
6.61573887e-01 1.02979791e+00 2.48154253e-01 -4.85289752e-01
8.59177530e-01 7.27824748e-01 5.88600636e-01 -2.01148242e-01
7.89344072e-01 7.51785219e-01 -5.39367318e-01 -1.10403371e+00
-8.09911966e-01 -4.26930755e-01 -7.91420221e-01 4.66135750e-03
7.59865880e-01 -1.16651678e+00 -4.59891558e-01 2.11794630e-01
-1.07319748e+00 -1.33442253e-01 -8.10796380e-01 4.47770655e-01
-6.03574574e-01 4.30296808e-01 -1.86285585e-01 -7.47817397e-01
-3.73557918e-02 -7.31841266e-01 1.08951747e+00 -2.65282467e-02
-1.57353923e-01 -1.32512784e+00 2.69311935e-01 4.63399589e-01
3.87997717e-01 -3.69046837e-01 7.94532716e-01 -1.09433424e+00
-2.88554847e-01 -2.13622358e-02 -5.50663061e-02 7.29727387e-01
6.07097208e-01 -3.89053613e-01 -1.60445130e+00 -3.04186761e-01
1.32744424e-02 -7.10768104e-01 1.10070419e+00 1.95570320e-01
9.21083450e-01 -4.27359998e-01 -3.68731260e-01 3.85298103e-01
1.15730059e+00 -1.05023332e-01 3.45614940e-01 -2.68073026e-02
5.05167782e-01 7.46236980e-01 7.83336610e-02 2.36707702e-01
3.54285747e-01 4.55168635e-01 3.98170382e-01 -3.12718511e-01
-3.96685034e-01 -5.84637582e-01 7.85842001e-01 9.19851363e-01
3.23282957e-01 -4.80892174e-02 -8.17002237e-01 8.33681762e-01
-1.67645407e+00 -1.32791698e+00 3.07446331e-01 1.96185350e+00
8.17880392e-01 -1.41695724e-03 2.62092557e-02 6.76702484e-02
4.37345743e-01 5.18347383e-01 -4.65143055e-01 -3.88406545e-01
-5.46312988e-01 1.80012763e-01 1.59649514e-02 3.94078314e-01
-1.12651098e+00 1.01324654e+00 7.93009138e+00 6.04929507e-01
-1.44872022e+00 2.20184281e-01 2.75618136e-01 -3.28022875e-02
-4.89313424e-01 1.50878467e-02 -4.66010749e-01 1.46476820e-01
1.08425605e+00 -2.90435910e-01 4.05047625e-01 7.75355697e-01
3.76156047e-02 2.64373630e-01 -1.46985149e+00 1.15629506e+00
4.97974515e-01 -1.58856761e+00 3.26717526e-01 -3.62653136e-02
6.88396394e-01 4.11305398e-01 4.53864694e-01 4.28929746e-01
3.44193190e-01 -1.17723703e+00 6.24848664e-01 4.66502756e-01
7.94803917e-01 -9.17942971e-02 3.18843007e-01 5.24978995e-01
-8.63881767e-01 -3.26960459e-02 -3.33102137e-01 -1.86236784e-01
-6.73753917e-02 2.67719746e-01 -9.65478241e-01 3.73235285e-01
6.56908154e-01 9.91347194e-01 -5.52673876e-01 2.66582876e-01
-1.24150455e-01 7.24632323e-01 -5.02307154e-02 3.31779927e-01
1.90723732e-01 1.62582994e-01 5.56334555e-01 1.46613693e+00
-1.32411430e-02 -3.54355156e-01 6.38174787e-02 8.98560107e-01
-3.36238258e-02 7.45513886e-02 -1.25688434e+00 -3.80193293e-01
2.37559795e-01 1.17358148e+00 -1.89309567e-01 -6.53284907e-01
-9.43072617e-01 1.01560307e+00 2.64730573e-01 6.16847754e-01
-4.38257605e-01 1.26427263e-01 7.15529680e-01 -7.19554201e-02
4.39427078e-01 -1.36050418e-01 -3.15684468e-01 -1.56986558e+00
-1.08731404e-01 -9.13278520e-01 3.07907164e-01 -9.76624668e-01
-1.50384367e+00 6.88029468e-01 -7.62514845e-02 -1.36968219e+00
-5.99206746e-01 -7.69290090e-01 -7.09704459e-01 7.40113735e-01
-1.60295987e+00 -1.53869259e+00 -1.64579034e-01 1.06617486e+00
9.80755866e-01 -8.41722250e-01 1.26879716e+00 -3.77546459e-01
-1.55281246e-01 4.26981270e-01 3.69595975e-01 2.36555055e-01
7.77724326e-01 -1.38306391e+00 3.79402429e-01 6.62679136e-01
9.91061509e-01 5.92965543e-01 7.58381486e-01 -3.32085460e-01
-1.55185533e+00 -7.49839604e-01 1.12500787e+00 -5.84314108e-01
1.01270962e+00 -6.33349478e-01 -7.16510355e-01 8.99616659e-01
9.83540654e-01 1.78831831e-01 8.55477452e-01 2.48587668e-01
-9.55350220e-01 -1.56321511e-01 -7.44852722e-01 4.91117686e-01
8.18545818e-01 -1.24676120e+00 -1.15187120e+00 4.75228071e-01
6.44670665e-01 1.35602811e-02 -5.76069176e-01 2.11969316e-01
3.56700033e-01 -7.80097008e-01 1.15460062e+00 -1.17077482e+00
4.93813336e-01 -1.13567092e-01 -5.74709773e-01 -1.47716284e+00
-3.16249549e-01 -3.91612619e-01 -4.29209292e-01 1.11753953e+00
3.99507284e-01 -5.12649775e-01 6.21370375e-01 2.10813671e-01
-8.36898834e-02 -2.65899032e-01 -1.07281351e+00 -7.12458491e-01
8.16135332e-02 -5.86679995e-01 1.73353925e-01 1.09137571e+00
4.02576804e-01 8.25761735e-01 -4.80263859e-01 1.29797742e-01
8.00118446e-01 3.99503589e-01 6.97875202e-01 -1.10807419e+00
-1.73857331e-01 -2.05859378e-01 -5.22216976e-01 -1.17705619e+00
7.47754514e-01 -1.08244097e+00 -1.05768442e-01 -1.57732785e+00
2.93773830e-01 3.54207546e-01 -4.82828587e-01 7.04668045e-01
3.00738543e-01 2.04614088e-01 7.50671774e-02 2.89175808e-01
-5.33985734e-01 6.76039875e-01 9.12048519e-01 -7.30492413e-01
3.18150371e-02 -1.34935066e-01 -6.71632588e-01 6.11352146e-01
4.92996842e-01 -7.38964379e-02 -6.55506968e-01 -5.87921679e-01
-1.61211580e-01 -8.94075558e-02 7.03170776e-01 -7.41355896e-01
4.39594924e-01 -2.75855571e-01 4.25754160e-01 -6.16320550e-01
6.86023474e-01 -8.91954601e-01 -3.53917301e-01 1.43773675e-01
-7.43423998e-01 -2.12909758e-01 8.42692778e-02 6.50195301e-01
-6.38481081e-01 1.33254945e-01 5.44001877e-01 -2.33291522e-01
-9.94167030e-01 6.42576292e-02 -6.58063352e-01 -1.73426062e-01
7.55111516e-01 -3.55558008e-01 -4.85148102e-01 -7.11936891e-01
-1.22400784e+00 -1.85992882e-01 2.28149191e-01 1.04899907e+00
8.68034720e-01 -1.46081316e+00 -7.70012081e-01 5.87397218e-01
6.79205418e-01 -6.53051615e-01 1.98541626e-01 5.22504449e-01
4.63683128e-01 8.76460254e-01 -1.96807861e-01 -8.02919149e-01
-1.16598749e+00 9.88816857e-01 1.80354431e-01 6.67967573e-02
-6.72596037e-01 6.37464941e-01 6.31562352e-01 -5.28694034e-01
4.06190485e-01 -1.82065234e-01 -3.33554894e-01 1.29524410e-01
6.16050482e-01 -1.61181301e-01 -6.27095699e-02 -9.88904595e-01
-4.28972125e-01 1.85837269e-01 -6.56050295e-02 -3.68736893e-01
1.27321839e+00 -2.90295690e-01 4.72958013e-02 1.17826653e+00
1.21843338e+00 -1.59353480e-01 -9.60475147e-01 -9.03514981e-01
-2.16372550e-01 -1.88722014e-01 1.74043939e-01 -7.03827858e-01
-9.44375157e-01 1.10565305e+00 7.60109365e-01 4.99903679e-01
9.27013516e-01 5.81049442e-01 2.52278566e-01 5.07017016e-01
1.12904802e-01 -1.12734914e+00 3.16935837e-01 6.89061761e-01
9.94279385e-01 -1.51984417e+00 -2.75340706e-01 5.95339835e-02
-9.62420404e-01 1.08751929e+00 4.97930557e-01 1.66518882e-01
7.78894007e-01 2.58685797e-01 3.35504234e-01 -1.09585486e-01
-1.21109319e+00 -5.01027048e-01 4.43223417e-01 9.42267120e-01
4.66268092e-01 -2.46571284e-02 4.48163867e-01 3.71983141e-01
-5.30128591e-02 -2.43964091e-01 2.40775749e-01 8.90293360e-01
-2.89445281e-01 -1.12774074e+00 -2.96422154e-01 2.44603828e-01
5.88405691e-03 -2.79137999e-01 -6.37415767e-01 3.90060335e-01
-1.22395143e-01 1.10223317e+00 2.19016656e-01 -5.88935137e-01
1.66937649e-01 6.43202484e-01 4.22380000e-01 -9.28625703e-01
-2.06018597e-01 2.49736309e-02 -1.04179062e-01 -4.88801628e-01
-7.92872310e-01 -6.35392725e-01 -8.51326883e-01 9.00503621e-03
-5.88227762e-03 8.52045491e-02 7.40669847e-01 1.24672520e+00
3.69842559e-01 2.81067193e-01 7.79083252e-01 -7.28056252e-01
-5.47503948e-01 -1.13951254e+00 -4.16729182e-01 6.66596413e-01
7.67845571e-01 -5.31769216e-01 -4.32392895e-01 5.89133143e-01] | [10.476151466369629, 2.2121756076812744] |
d29a0219-2f70-4c73-8089-ee6d60edd7d9 | cascade-evidential-learning-for-open-world | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Chen_Cascade_Evidential_Learning_for_Open-World_Weakly-Supervised_Temporal_Action_Localization_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Chen_Cascade_Evidential_Learning_for_Open-World_Weakly-Supervised_Temporal_Action_Localization_CVPR_2023_paper.pdf | Cascade Evidential Learning for Open-World Weakly-Supervised Temporal Action Localization | Targeting at recognizing and localizing action instances with only video-level labels during training, Weakly-supervised Temporal Action Localization (WTAL) has achieved significant progress in recent years. However, living in the dynamically changing open world where unknown actions constantly spring up, the closed-set assumption of existing WTAL methods is invalid. Compared with traditional open-set recognition tasks, Open-world WTAL (OWTAL) is challenging since not only are the annotations of unknown samples unavailable, but also the fine-grained annotations of known action instances can only be inferred ambiguously from the video category labels. To address this problem, we propose a Cascade Evidential Learning framework at an evidence level, which targets at OWTAL for the first time. Our method jointly leverages multi-scale temporal contexts and knowledge-guided prototype information to progressively collect cascade and enhanced evidence for known action, unknown action, and background separation. Extensive experiments conducted on THUMOS-14 and ActivityNet-v1.3 verify the effectiveness of our method. Besides the classification metrics adopted by previous open-set recognition methods, we also evaluate our method on localization metrics which are more reasonable for OWTAL. | ['Changsheng Xu', 'Junyu Gao', 'Mengyuan Chen'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['weakly-supervised-temporal-action', 'action-localization', 'action-recognition', 'open-set-learning'] | ['computer-vision', 'computer-vision', 'computer-vision', 'miscellaneous'] | [ 5.26732922e-01 -4.94176410e-02 -6.66644454e-01 -1.74735442e-01
-8.28569710e-01 -6.06487274e-01 6.43380582e-01 -2.87051678e-01
-2.87038386e-01 7.93908834e-01 4.25593108e-01 1.66480213e-01
2.26150528e-02 -2.22769469e-01 -6.90268397e-01 -6.52217269e-01
-1.36759534e-01 1.63119197e-01 6.95898950e-01 2.90188432e-01
-1.38960481e-01 5.29008731e-02 -1.52210033e+00 6.50291622e-01
5.74328780e-01 1.35695827e+00 -2.46190708e-02 4.93637979e-01
-5.15377661e-03 1.43642604e+00 -3.35282177e-01 -1.17880061e-01
3.63878578e-01 -3.48495513e-01 -9.18888032e-01 4.94599223e-01
9.22273159e-01 -4.95091975e-01 -5.93399763e-01 8.96487594e-01
1.17016532e-01 2.50830740e-01 3.08958918e-01 -1.53355384e+00
-6.34024918e-01 5.55716932e-01 -6.37476325e-01 4.62584406e-01
4.18652475e-01 4.84255075e-01 1.07207215e+00 -1.01817417e+00
7.09020197e-01 1.13628662e+00 6.08544588e-01 5.43857396e-01
-9.44346189e-01 -6.61622763e-01 9.46255684e-01 6.96721673e-01
-1.23655796e+00 -5.66160679e-01 5.92114747e-01 -4.08626825e-01
7.45071232e-01 6.76715747e-02 8.60675991e-01 1.68181658e+00
-2.19684049e-01 1.31066835e+00 1.15501821e+00 -1.29162729e-01
3.13406259e-01 -3.72100234e-01 3.51764983e-03 8.27815950e-01
3.53073105e-02 -3.57491784e-02 -9.23270404e-01 1.49077680e-02
8.66593063e-01 3.41876298e-01 -3.52256715e-01 -4.85871404e-01
-1.53537083e+00 2.17111409e-01 3.57764751e-01 3.74018818e-01
-3.05419326e-01 3.62649411e-01 4.37943161e-01 1.00928977e-01
5.87578058e-01 4.03023604e-03 -5.93359888e-01 -5.23972511e-01
-8.26633930e-01 -2.86074549e-01 5.41632831e-01 8.33642602e-01
5.29091477e-01 3.88849862e-02 -4.73823935e-01 7.43928015e-01
5.93323633e-02 3.93707722e-01 5.66214025e-01 -1.15841830e+00
5.54371715e-01 7.24052966e-01 2.24972963e-01 -6.02921546e-01
-3.38198282e-02 -2.92787910e-01 -4.50379580e-01 8.62637237e-02
7.36715496e-01 2.09496189e-02 -1.08319032e+00 1.71246493e+00
4.84931558e-01 9.07443166e-01 -1.01469737e-03 9.95825350e-01
4.52515811e-01 3.50542486e-01 1.42505124e-01 -2.60766387e-01
1.11840963e+00 -1.32518291e+00 -8.18489492e-01 -4.12530601e-01
7.23519087e-01 -1.76301435e-01 1.03727126e+00 2.96767890e-01
-5.54235697e-01 -5.25725305e-01 -8.79386246e-01 1.51327729e-01
-1.65683672e-01 3.52858573e-01 7.14980543e-01 2.92257398e-01
-7.07399130e-01 3.93076479e-01 -1.19534409e+00 -4.63082016e-01
9.30900395e-01 -3.19119380e-03 -6.36292517e-01 -3.96751523e-01
-8.89081597e-01 5.97464502e-01 2.90136367e-01 4.03333992e-01
-1.31374216e+00 -4.23386008e-01 -7.95837998e-01 -1.29203260e-01
1.18244290e+00 -2.64274329e-01 1.23551810e+00 -1.32380033e+00
-1.30678523e+00 4.64199573e-01 -1.26546979e-01 -5.34670591e-01
6.26158714e-01 -4.58110958e-01 -4.95273650e-01 3.20779353e-01
2.53073961e-01 5.30077279e-01 8.53536665e-01 -1.02039254e+00
-9.41543818e-01 -3.29829901e-01 4.92121369e-01 2.51891881e-01
-4.45704252e-01 -1.36298925e-01 -5.71259260e-01 -5.43979824e-01
1.42646089e-01 -8.90788078e-01 -1.33959547e-01 4.13882226e-01
-1.67905867e-01 -3.95228058e-01 1.07474411e+00 -6.06844306e-01
1.16341674e+00 -2.27115583e+00 9.79205687e-03 -8.31647739e-02
2.37908006e-01 1.52576238e-01 -9.66047645e-02 1.29326627e-01
7.35356137e-02 -2.22421631e-01 -1.97770953e-01 -1.66055322e-01
3.62652317e-02 5.65940082e-01 -2.58160830e-01 4.58840251e-01
2.27231994e-01 1.01197803e+00 -1.22505367e+00 -7.83217013e-01
9.34176221e-02 2.05491796e-01 -3.53423804e-01 -7.01813474e-02
-3.89763325e-01 5.07745206e-01 -5.50985456e-01 1.18006349e+00
6.11388832e-02 -4.93877798e-01 1.89140782e-01 -2.67725080e-01
2.30985731e-02 -1.23350292e-01 -1.27599609e+00 1.93531823e+00
-2.24144369e-01 6.68390572e-01 -1.29644200e-01 -1.02793944e+00
2.55805075e-01 3.56294572e-01 8.34748209e-01 -5.31091213e-01
-1.37204707e-01 4.82558571e-02 -4.64650691e-02 -7.19578862e-01
2.21146047e-01 1.86077729e-01 1.26763489e-02 3.46110195e-01
3.43071878e-01 6.39108181e-01 2.26281300e-01 3.22916061e-01
1.65418494e+00 8.39457512e-01 2.36861050e-01 2.58369684e-01
3.84028226e-01 -3.97704653e-02 1.10601771e+00 8.01527739e-01
-8.10244560e-01 3.84128362e-01 3.61469924e-01 -3.43214333e-01
-3.72979313e-01 -9.69292343e-01 1.13566555e-01 1.26939380e+00
3.23071003e-01 -4.99515325e-01 -5.87256491e-01 -1.36993957e+00
-1.08930573e-01 2.08967552e-01 -7.92573452e-01 -5.19548096e-02
-4.83535320e-01 -3.38195801e-01 6.26079619e-01 8.79359186e-01
7.15390265e-01 -9.80034590e-01 -5.93404114e-01 1.51047051e-01
-5.18332124e-01 -1.73628056e+00 -6.14959240e-01 4.08129767e-02
-7.43388951e-01 -1.42503881e+00 -4.40865070e-01 -4.74997491e-01
6.82062566e-01 3.68853301e-01 8.33803535e-01 -7.73586556e-02
-2.59863168e-01 8.30811441e-01 -5.87380350e-01 -9.01666507e-02
1.66002400e-02 -4.41629380e-01 1.75451860e-01 7.12052286e-01
4.30272847e-01 -5.32008767e-01 -7.04566717e-01 5.95141888e-01
-6.93755805e-01 -1.11477915e-03 7.72669733e-01 7.17140734e-01
7.55314469e-01 6.08151890e-02 4.72227305e-01 -5.90659320e-01
-1.91296786e-01 -3.84808600e-01 -2.78109282e-01 5.25620699e-01
-2.50119627e-01 -1.82995722e-01 2.16494113e-01 -1.01013601e+00
-1.21785820e+00 3.19852203e-01 2.70487696e-01 -9.41206574e-01
-3.43616039e-01 2.82512337e-01 -3.56205195e-01 -1.72029566e-02
5.32530129e-01 2.09034875e-01 -3.50326478e-01 -3.22015792e-01
3.84756565e-01 4.30368662e-01 6.98504448e-01 -5.34689665e-01
6.28157854e-01 9.03732896e-01 -4.25347179e-01 -5.77966988e-01
-1.35616279e+00 -6.21937275e-01 -8.41030777e-01 -6.15125775e-01
9.33851719e-01 -1.16583240e+00 -3.60545397e-01 5.09730756e-01
-7.46621728e-01 -6.29219890e-01 -6.56552196e-01 5.07207811e-01
-4.59470063e-01 4.76005048e-01 -4.65350062e-01 -7.92008162e-01
1.62881494e-01 -9.42035615e-01 1.25333691e+00 -1.65203154e-01
-1.81567594e-01 -8.32544506e-01 -2.21537828e-01 7.80488551e-01
-6.15108816e-04 3.40810657e-01 2.19841644e-01 -5.65628886e-01
-8.47567558e-01 -2.36460030e-01 -1.20506950e-01 4.53610569e-01
2.59505183e-01 -2.02659622e-01 -1.05115306e+00 -2.03727782e-02
-1.49589121e-01 -6.72794521e-01 1.10176492e+00 1.30270645e-01
1.20580125e+00 -2.82231838e-01 -4.34198350e-01 2.28368133e-01
9.61644590e-01 -3.95902358e-02 2.69283473e-01 1.00185433e-02
9.71944511e-01 2.59444594e-01 1.00770581e+00 5.40740311e-01
2.76738614e-01 7.37219810e-01 4.95919853e-01 -1.81816835e-02
-4.13510084e-01 -4.01476383e-01 8.76194417e-01 3.68983924e-01
-2.79485852e-01 -7.98652619e-02 -6.70901716e-01 5.76363027e-01
-2.22964787e+00 -1.31277180e+00 8.65266919e-02 1.97529709e+00
7.77286232e-01 2.81145453e-01 1.13564804e-01 1.06909864e-01
6.66705012e-01 4.08243984e-01 -8.37385416e-01 5.15949607e-01
-1.93158999e-01 -1.61344886e-01 4.20867711e-01 1.62321925e-01
-1.36643934e+00 9.43355322e-01 5.40529823e+00 8.33419621e-01
-8.65398586e-01 6.43278480e-01 3.22560638e-01 -5.31401098e-01
2.88184404e-01 2.22090602e-01 -6.99490249e-01 4.31927860e-01
4.87293035e-01 3.09248984e-01 1.63626984e-01 9.29165423e-01
2.15653226e-01 -2.70309716e-01 -1.46217859e+00 1.07168722e+00
2.27804139e-01 -1.17055452e+00 -1.88893497e-01 2.05893349e-02
7.90452778e-01 2.49186054e-01 -1.37175217e-01 6.30117714e-01
4.13086861e-01 -4.74679500e-01 9.41147327e-01 4.56432283e-01
6.91677809e-01 -1.02218218e-01 4.02392149e-01 3.75586927e-01
-1.50397968e+00 -6.93747699e-01 3.52201909e-02 -2.77250946e-01
2.04384699e-01 4.66191083e-01 -4.56518918e-01 3.87134045e-01
8.04721832e-01 1.41505897e+00 -5.28995693e-01 8.65058303e-01
-6.21566236e-01 9.57549036e-01 -3.63873810e-01 3.68051618e-01
3.57207179e-01 1.62062019e-01 3.71891916e-01 8.74008298e-01
-2.85772001e-03 2.41279155e-01 7.30478525e-01 5.06008625e-01
-9.76639464e-02 -3.23425144e-01 -4.66658771e-01 -2.10478932e-01
2.25602075e-01 1.13154638e+00 -7.78916597e-01 -4.34562206e-01
-7.06576049e-01 1.19956040e+00 3.22229505e-01 6.13709986e-01
-1.17355382e+00 3.82480502e-01 7.26251781e-01 8.05696622e-02
4.44853127e-01 -3.23059529e-01 3.89132738e-01 -1.67084181e+00
3.64488930e-01 -7.26278245e-01 7.37400651e-01 -6.63861156e-01
-1.27656198e+00 1.48079410e-01 7.98891485e-02 -1.51870358e+00
3.54083180e-02 -4.92341459e-01 -4.39535767e-01 -1.80901866e-02
-1.38327932e+00 -1.43757737e+00 -3.29652846e-01 8.28309178e-01
1.04588950e+00 6.18837848e-02 5.57457864e-01 4.00099069e-01
-8.53417158e-01 3.80632937e-01 -1.17751487e-01 2.99538523e-01
7.03651547e-01 -1.03821969e+00 -1.08478121e-01 1.13234854e+00
7.29208350e-01 1.69809967e-01 2.79572517e-01 -7.31840730e-01
-1.43146968e+00 -1.34671748e+00 3.08236659e-01 -8.17963958e-01
1.15379155e+00 -5.38228869e-01 -8.30219507e-01 9.01115477e-01
-2.06051707e-01 8.20942163e-01 5.87463498e-01 1.80507198e-01
-6.02960229e-01 -1.85093492e-01 -6.00200355e-01 4.37045753e-01
1.77759635e+00 -7.10695863e-01 -5.07591128e-01 5.99016607e-01
5.92260838e-01 -4.31911051e-01 -8.08253765e-01 4.39762563e-01
6.77734971e-01 -6.39689147e-01 8.44377160e-01 -7.28303552e-01
2.55281568e-01 -4.71823573e-01 -2.79956430e-01 -8.35984826e-01
-1.20162323e-01 -4.47747052e-01 -6.74641252e-01 1.09128118e+00
1.00058444e-01 -3.55866134e-01 8.49553406e-01 5.58717072e-01
-1.37725785e-01 -6.60608530e-01 -1.22134399e+00 -1.02783835e+00
-6.22550964e-01 -7.21577942e-01 1.41060278e-01 9.87809539e-01
-5.50482713e-04 1.79657280e-01 -5.83345473e-01 3.10231864e-01
6.11773372e-01 1.93880275e-01 7.20761240e-01 -1.01950276e+00
-5.76673388e-01 -1.04930483e-01 -6.71251476e-01 -1.20937347e+00
4.26091343e-01 -6.18065119e-01 3.12465996e-01 -1.44945800e+00
3.02043587e-01 -3.98949832e-01 -7.97391593e-01 1.12065411e+00
-1.51003346e-01 4.29172099e-01 1.25880137e-01 3.68174970e-01
-1.46876156e+00 6.86600685e-01 1.13197172e+00 -3.47698927e-01
4.22327071e-02 -1.74632922e-01 -3.63538027e-01 9.82824802e-01
3.60795647e-01 -3.63694459e-01 -6.96669519e-01 -4.20758992e-01
-1.40928686e-01 4.29604352e-02 7.79209852e-01 -1.27109230e+00
1.33774444e-01 -4.77518618e-01 2.87286997e-01 -3.29728454e-01
5.34069180e-01 -9.71355379e-01 -7.42450133e-02 2.07170054e-01
-3.67095619e-01 -5.88377178e-01 -1.39378503e-01 1.26950228e+00
-2.64395177e-01 2.59376973e-01 3.94738495e-01 -5.27187288e-02
-1.44403327e+00 6.71984792e-01 -2.56369323e-01 1.26761809e-01
1.22296596e+00 -5.24256468e-01 -3.50415558e-01 -1.92795098e-01
-9.49044287e-01 4.24738914e-01 3.02282423e-01 7.05909908e-01
6.27650440e-01 -1.33468676e+00 -4.29660589e-01 -9.21290517e-02
5.91464520e-01 -1.04990557e-01 4.24195319e-01 1.30480385e+00
2.69463301e-01 8.86514783e-02 -3.43041611e-03 -7.51529872e-01
-1.29553223e+00 4.18120176e-01 3.38364303e-01 -2.44732842e-01
-8.80026162e-01 7.62151182e-01 3.78764004e-01 5.74311726e-02
4.55339015e-01 -4.13146973e-01 -1.43867314e-01 1.32834530e-02
5.55018425e-01 2.03297734e-01 -2.06224337e-01 -7.01505542e-01
-4.01643276e-01 2.13144436e-01 2.76519153e-02 4.23788130e-02
1.19425833e+00 -1.73195183e-01 2.95371354e-01 7.50806987e-01
7.83999860e-01 -3.08947235e-01 -1.85142386e+00 -6.69547260e-01
6.29737899e-02 -5.63298881e-01 2.06093471e-02 -8.55214536e-01
-1.09149325e+00 7.37795413e-01 5.97310364e-01 -3.23302597e-01
9.99220550e-01 2.05269963e-01 7.23895609e-01 5.22015512e-01
5.87142110e-01 -1.29709435e+00 6.37959838e-01 3.17361355e-01
6.35605156e-01 -1.54758620e+00 -8.22589323e-02 -3.56894076e-01
-7.22926617e-01 6.98213935e-01 1.03489482e+00 2.59030074e-01
4.00962710e-01 1.89640120e-01 -9.38016102e-02 -7.56519586e-02
-9.77124035e-01 -4.91267353e-01 1.13517828e-01 5.29609621e-01
-6.02044314e-02 -1.28166199e-01 1.50765345e-01 5.81049919e-01
7.00755894e-01 2.97823191e-01 3.10432047e-01 1.26148963e+00
-3.65212739e-01 -8.40290368e-01 -2.22036824e-01 4.08023626e-01
-4.21750605e-01 1.71147048e-01 -4.67659265e-01 6.40621781e-01
4.26692337e-01 9.49800432e-01 -1.98811099e-01 -4.00026798e-01
8.51861313e-02 1.85085118e-01 5.19062519e-01 -6.06522381e-01
-5.16383201e-02 -2.42626686e-02 3.34198445e-01 -1.14163840e+00
-9.47501719e-01 -8.89768243e-01 -1.33204937e+00 4.14565593e-01
-3.85385752e-01 -2.47480422e-02 1.34108558e-01 1.31052375e+00
3.07507962e-01 5.71527064e-01 3.18432182e-01 -8.09934139e-01
-4.78440166e-01 -8.32166433e-01 -5.00311732e-01 5.71346760e-01
2.81004548e-01 -1.12114501e+00 -2.78139532e-01 5.15090764e-01] | [8.560032844543457, 0.6972261667251587] |
b0e8d0f3-2652-41b8-a14e-11a83e7f875f | a-dual-modality-approach-for-zero-shot-multi | 2208.09562 | null | https://arxiv.org/abs/2208.09562v1 | https://arxiv.org/pdf/2208.09562v1.pdf | A Dual Modality Approach For (Zero-Shot) Multi-Label Classification | In computer vision, multi-label classification, including zero-shot multi-label classification are important tasks with many real-world applications. In this paper, we propose a novel algorithm, Aligned Dual moDality ClaSsifier (ADDS), which includes a Dual-Modal decoder (DM-decoder) with alignment between visual and textual features, for multi-label classification tasks. Moreover, we design a simple and yet effective method called Pyramid-Forwarding to enhance the performance for inputs with high resolutions. Extensive experiments conducted on standard multi-label benchmark datasets, MS-COCO and NUS-WIDE, demonstrate that our approach significantly outperforms previous methods and provides state-of-the-art performance for conventional multi-label classification, zero-shot multi-label classification, and an extreme case called single-to-multi label classification where models trained on single-label datasets (ImageNet-1k, ImageNet-21k) are tested on multi-label ones (MS-COCO and NUS-WIDE). We also analyze how visual-textual alignment contributes to the proposed approach, validate the significance of the DM-decoder, and demonstrate the effectiveness of Pyramid-Forwarding on vision transformer. | ['Zhu Qi', 'Chiuman Ho', 'Jenhao Hsiao', 'Yikang Li', 'Shichao Xu'] | 2022-08-19 | null | null | null | null | ['multi-label-zero-shot-learning'] | ['computer-vision'] | [ 6.07253134e-01 -1.95774361e-01 -3.13216120e-01 -3.60910058e-01
-1.23637509e+00 -4.13863957e-01 6.93588316e-01 2.34846994e-01
-3.89054000e-01 5.03889859e-01 -1.94342896e-01 -1.10258989e-01
3.53098549e-02 -2.44479477e-01 -4.93209779e-01 -7.28483379e-01
6.83439016e-01 3.37554485e-01 2.84932256e-01 8.60203505e-02
1.47446170e-01 -1.32801741e-01 -1.96459758e+00 6.95978284e-01
4.16755766e-01 1.19524252e+00 2.22729281e-01 6.04943991e-01
5.18700816e-02 9.63871658e-01 -2.65610777e-03 -6.73331439e-01
-2.17291545e-02 -3.33866030e-01 -9.12289798e-01 3.52646947e-01
8.67846906e-01 -1.00736268e-01 5.32216020e-02 1.30428171e+00
6.01793945e-01 -1.14842437e-01 1.12827969e+00 -1.38687456e+00
-8.06123972e-01 2.04509005e-01 -1.10371816e+00 -1.92336962e-01
2.35180870e-01 2.74718683e-02 9.46565509e-01 -1.08294678e+00
7.16744423e-01 1.40641212e+00 9.24584091e-01 6.87735498e-01
-1.13126135e+00 -6.03964150e-01 -4.60889004e-02 3.49201530e-01
-1.32780886e+00 -3.17426324e-01 3.41392815e-01 -5.59610486e-01
5.52988052e-01 8.30389559e-02 1.25588551e-01 1.29661119e+00
2.73816735e-01 9.10768151e-01 1.71383071e+00 -7.89019346e-01
-7.26554617e-02 1.72354028e-01 3.26498926e-01 1.18584645e+00
-1.23444691e-01 -1.11857496e-01 -4.66246217e-01 -3.21563274e-01
3.50860983e-01 1.16909601e-01 -1.05429344e-01 -2.85420567e-01
-1.44570673e+00 7.96322286e-01 -3.10502406e-02 3.11563969e-01
-1.05920717e-01 4.37502600e-02 8.62638295e-01 1.84582055e-01
5.13517678e-01 -2.42632851e-01 -3.29986244e-01 2.97583759e-01
-8.53355110e-01 -4.68002521e-02 3.03779393e-01 1.15460372e+00
6.46860540e-01 -3.75765383e-01 -5.23794532e-01 1.34734821e+00
6.44441724e-01 5.88003814e-01 7.07189202e-01 -8.64982963e-01
2.71332681e-01 5.21649003e-01 -3.43763113e-01 -5.57309628e-01
-7.30160117e-01 -4.01589483e-01 -8.95969272e-01 9.53925401e-02
2.31016546e-01 1.19488016e-01 -1.16169930e+00 1.59465170e+00
4.79070693e-01 3.85556668e-01 3.76629442e-01 7.17893362e-01
1.23441875e+00 3.07105303e-01 5.85316539e-01 -4.75705594e-01
1.86607635e+00 -1.46117985e+00 -6.87261999e-01 -1.50230452e-01
1.03610313e+00 -9.34089363e-01 8.89220297e-01 5.18723354e-02
-7.75033355e-01 -6.13952875e-01 -8.85359585e-01 -2.27651805e-01
-3.63243073e-01 3.24033231e-01 1.90339267e-01 4.04213548e-01
-8.32055509e-01 2.84051538e-01 -6.45429790e-02 -7.18009889e-01
4.26124990e-01 5.21996953e-02 -6.80532455e-01 -4.24987584e-01
-1.11411619e+00 9.63297844e-01 5.87505281e-01 -4.52399045e-01
-1.02329195e+00 -4.09771264e-01 -9.05099630e-01 -2.24530712e-01
3.93088371e-01 -6.70909941e-01 1.37090313e+00 -9.49842691e-01
-9.51890349e-01 1.61828816e+00 -7.36549944e-02 -3.35224979e-02
2.56846100e-01 5.87297857e-01 -6.58711493e-01 5.20551503e-01
5.25421083e-01 1.11301756e+00 8.67250860e-01 -1.53620684e+00
-1.11242723e+00 -5.32944143e-01 8.57852027e-02 2.68807441e-01
-3.60579938e-01 5.06864525e-02 -3.03378552e-01 -4.37646717e-01
1.98470473e-01 -1.11425889e+00 6.68571740e-02 -2.25678161e-01
-5.06199002e-01 -4.08818007e-01 7.78904498e-01 -4.63016689e-01
1.00339437e+00 -2.27203107e+00 -1.25434147e-02 -2.00674847e-01
2.38510221e-01 1.71579435e-01 -4.10489619e-01 1.71168089e-01
-8.44458267e-02 -5.37525862e-02 -1.07414320e-01 -6.71353400e-01
-3.87289450e-02 1.38745427e-01 2.86752909e-01 5.80308199e-01
-1.06674038e-01 1.01882637e+00 -8.58744144e-01 -1.11152017e+00
2.68801689e-01 1.92465588e-01 -1.94613971e-02 -7.10321218e-02
-1.14649527e-01 4.69035566e-01 -1.53930679e-01 1.22481024e+00
5.11056840e-01 -7.51789927e-01 2.22203881e-01 -6.40169919e-01
2.82153506e-02 -8.03350925e-01 -1.01180351e+00 1.71820986e+00
-4.59759057e-01 2.30768561e-01 -1.58089802e-01 -8.96108925e-01
4.52750772e-01 6.43301964e-01 5.47221839e-01 -9.31396365e-01
4.42633390e-01 2.41159528e-01 -5.74499786e-01 -7.70901740e-01
3.46248686e-01 -3.61904979e-01 -1.25391245e-01 4.46265370e-01
4.35690373e-01 5.49387395e-01 5.13655007e-01 1.88574165e-01
7.47520089e-01 -7.70461932e-02 5.66891849e-01 -1.02549866e-01
7.28181362e-01 -1.20590866e-01 5.62864900e-01 6.39996111e-01
-5.50812423e-01 7.59386301e-01 4.16476339e-01 2.68175802e-03
-1.00788975e+00 -6.85641527e-01 -3.59419346e-01 1.46835542e+00
3.36552262e-01 -2.45152563e-01 -6.35706306e-01 -1.16991055e+00
-9.27438214e-02 5.76617122e-01 -7.29081929e-01 9.74655896e-02
1.41202539e-01 -8.68337095e-01 8.08739722e-01 2.81245440e-01
6.54399037e-01 -7.45048344e-01 -3.66573751e-01 2.75179353e-02
-5.32764792e-01 -1.45857978e+00 -5.62764108e-01 2.43909970e-01
-5.90371251e-01 -1.35366988e+00 -9.80144799e-01 -1.19320571e+00
5.02919972e-01 5.91397822e-01 7.82167375e-01 -2.03234658e-01
-3.94295931e-01 7.23963380e-01 -4.75768000e-01 1.05813332e-02
-6.21510506e-01 -2.97602326e-01 -1.92254215e-01 5.20107210e-01
3.69982988e-01 -2.24886060e-01 -3.69065344e-01 5.05022168e-01
-9.77427959e-01 3.72742057e-01 7.77138710e-01 1.10063517e+00
9.23039913e-01 -4.01415020e-01 7.66030073e-01 -1.02532065e+00
2.85817683e-01 -6.17819428e-01 -1.37025952e-01 9.56756651e-01
-9.21439350e-01 -1.64878398e-01 3.14313114e-01 -5.10600924e-01
-9.27918136e-01 3.68479908e-01 -4.40313891e-02 -7.94881284e-01
-3.71572971e-01 3.73443544e-01 1.20413169e-01 -3.83226544e-01
5.93985081e-01 3.06312770e-01 -7.97272287e-03 -4.44326043e-01
6.66021347e-01 1.01711035e+00 4.45589721e-01 -1.86528847e-01
8.98489431e-02 5.88472247e-01 2.89175004e-01 -5.48016548e-01
-1.14339864e+00 -9.36443508e-01 -7.59332120e-01 -5.77040851e-01
1.23319864e+00 -1.17381895e+00 -6.37556374e-01 7.65851319e-01
-1.14676929e+00 1.34116843e-01 2.06346691e-01 3.07134867e-01
-7.01623499e-01 5.46587706e-01 -7.59758115e-01 -6.02874100e-01
-3.85405600e-01 -1.46426725e+00 1.48462749e+00 2.39956558e-01
2.38113239e-01 -1.16568100e+00 -5.63541353e-02 9.99094129e-01
-9.02898461e-02 1.51279852e-01 1.18932188e+00 -6.03154778e-01
-1.46560475e-01 2.25510895e-01 -8.28124285e-01 3.73602092e-01
-3.26658100e-01 -2.85209805e-01 -1.27194762e+00 -3.85104567e-01
-4.16655421e-01 -9.77941453e-01 9.97606754e-01 2.98464417e-01
7.95645416e-01 3.18320125e-01 -5.44164956e-01 2.85048813e-01
1.87786722e+00 1.35167111e-02 1.73556924e-01 2.52797425e-01
8.36424530e-01 4.85094458e-01 9.67369616e-01 2.92740166e-01
7.62007356e-01 7.88117886e-01 6.13765836e-01 -1.37906834e-01
-5.64154983e-01 1.94064349e-01 1.19878955e-01 1.01979148e+00
1.91466063e-01 -1.34290591e-01 -7.26165533e-01 4.39436764e-01
-2.15175128e+00 -8.24657381e-01 -5.46484828e-01 1.67377388e+00
6.23550177e-01 -2.99958676e-01 2.54231021e-02 -4.42448556e-02
1.06942666e+00 1.80184111e-01 -5.23423791e-01 -8.42069685e-02
-2.08282977e-01 -2.96531200e-01 6.37505710e-01 1.65314868e-01
-1.56796706e+00 7.99097598e-01 5.96033621e+00 1.25839531e+00
-9.71072793e-01 9.18676019e-01 6.50642991e-01 9.56154168e-02
1.10647254e-01 -3.05148661e-01 -1.16894138e+00 3.90127391e-01
8.97781909e-01 2.33948305e-01 -7.02884719e-02 8.33600700e-01
-3.19329768e-01 -3.04894239e-01 -1.16749322e+00 1.32621670e+00
6.06633544e-01 -1.14013112e+00 3.26242559e-02 -2.05664337e-02
9.14126873e-01 -6.12257235e-02 1.57066450e-01 5.41101873e-01
-1.19992897e-01 -5.51294446e-01 8.51920903e-01 3.57140303e-01
1.33729959e+00 -5.11931717e-01 6.72018290e-01 3.21061075e-01
-1.40792072e+00 -2.61612803e-01 -2.30937913e-01 4.95225966e-01
1.85520381e-01 5.25455832e-01 -4.27374154e-01 7.88478076e-01
3.07643086e-01 9.26932335e-01 -9.49065745e-01 6.87499225e-01
2.09029466e-01 2.12269515e-01 3.26238394e-01 2.32278168e-01
4.83607620e-01 2.58413017e-01 1.11737087e-01 1.37347078e+00
1.04027197e-01 -1.86762065e-01 5.19564033e-01 2.44362488e-01
-9.05848145e-02 4.19155687e-01 -6.06739759e-01 1.56598613e-01
2.36700401e-01 1.53758931e+00 -9.67014670e-01 -5.14522851e-01
-8.80736232e-01 1.03821266e+00 4.06227142e-01 3.13531905e-01
-1.10413826e+00 -3.65759395e-02 1.31533638e-01 -4.85239685e-01
2.15777829e-01 3.31025660e-01 -1.60272315e-01 -1.11700463e+00
-2.93055981e-01 -9.02546704e-01 7.16626823e-01 -7.84598887e-01
-1.32982802e+00 4.57415134e-01 -1.60452664e-01 -1.45668399e+00
2.70125568e-02 -5.53198159e-01 1.34440437e-01 4.21331942e-01
-1.74755991e+00 -1.97380686e+00 -2.19061196e-01 7.97032416e-01
7.01355219e-01 -3.97692889e-01 9.67241287e-01 8.73667955e-01
-5.62511802e-01 8.71384382e-01 4.33694929e-01 -7.69686028e-02
1.04162979e+00 -9.04282808e-01 -3.75368267e-01 3.73301536e-01
2.21397385e-01 -7.16939867e-02 1.50565118e-01 -4.52303529e-01
-9.48611915e-01 -1.37657857e+00 7.80131876e-01 -1.98214397e-01
4.82691795e-01 3.43886577e-02 -5.33546448e-01 6.44363999e-01
2.36933634e-01 6.42331988e-02 9.66591358e-01 -1.92578852e-01
-6.26429856e-01 1.51658922e-01 -1.34841371e+00 3.22780073e-01
9.18264985e-01 -6.90648556e-01 -3.71485740e-01 7.55070925e-01
6.84524298e-01 -2.40174428e-01 -1.06546855e+00 6.58445597e-01
7.89064109e-01 -9.16405082e-01 9.10964847e-01 -3.28919858e-01
7.69614398e-01 -1.95057392e-01 -6.13005936e-01 -1.15326190e+00
-4.55539376e-01 4.33815747e-01 -1.66259950e-03 1.29682219e+00
2.93797195e-01 -3.88925999e-01 2.51655281e-01 -1.26614153e-01
-2.58281212e-02 -1.01559186e+00 -9.55330312e-01 -5.73228240e-01
-2.82410443e-01 -1.71136141e-01 -7.57284686e-02 1.38034785e+00
-2.87929207e-01 6.60568535e-01 -6.62761331e-01 4.54788990e-02
1.10923016e+00 6.13284148e-02 1.68904617e-01 -1.18452227e+00
-2.08552793e-01 -3.35167676e-01 -4.74113792e-01 -5.54838955e-01
3.80941838e-01 -1.20300531e+00 -2.13028938e-01 -1.65320230e+00
7.79523909e-01 -2.82382011e-01 -6.39388025e-01 7.07104623e-01
-1.41245753e-01 1.01170409e+00 4.42300946e-01 4.42292511e-01
-1.30719280e+00 2.05558553e-01 1.38104808e+00 -4.07802284e-01
5.80532432e-01 -4.37905103e-01 -4.84329164e-01 7.43471920e-01
2.24343494e-01 -5.26547909e-01 -3.06900203e-01 -1.30073592e-01
-1.68390185e-01 2.44860217e-01 2.78622687e-01 -1.03447092e+00
3.07881057e-01 -2.39242297e-02 1.03610538e-01 -6.58578277e-01
4.23396975e-01 -8.45156550e-01 1.55946449e-03 4.90197212e-01
-5.71482122e-01 -2.65689135e-01 -1.44615725e-01 8.11297297e-01
-2.83164233e-01 -7.22703278e-01 1.24561727e+00 -1.88381344e-01
-1.31401670e+00 2.56937474e-01 -2.55438954e-01 1.47788813e-02
1.66110194e+00 -8.29677060e-02 -6.20963097e-01 1.35311872e-01
-9.18953598e-01 4.58230436e-01 2.73409575e-01 3.84251386e-01
6.00830853e-01 -1.69475377e+00 -6.41898930e-01 3.76553610e-02
8.63389373e-01 -6.23554528e-01 7.59836435e-01 1.27117085e+00
-3.32547605e-01 4.49402153e-01 -3.08229595e-01 -8.97310257e-01
-1.98301232e+00 7.07272112e-01 2.02519313e-01 -4.27129209e-01
-4.29394841e-01 8.36912215e-01 3.32511276e-01 -4.40851748e-01
2.25611746e-01 2.29198024e-01 -5.40308654e-01 6.22814476e-01
6.06521189e-01 3.86055529e-01 -3.02000213e-02 -1.15694404e+00
-2.20743090e-01 1.01376247e+00 -2.71593541e-01 6.82816729e-02
7.87050426e-01 -4.83206451e-01 -2.31900394e-01 8.19460392e-01
1.57976139e+00 -7.34933138e-01 -8.13684165e-01 -5.21075666e-01
8.00594501e-03 -1.16825297e-01 1.32766113e-01 -8.37968886e-01
-9.88783002e-01 8.15701663e-01 1.22068167e+00 -1.52749894e-02
9.96097565e-01 5.80254272e-02 7.87820339e-01 1.51719093e-01
5.33646345e-01 -1.26457512e+00 3.57514530e-01 4.76818264e-01
3.20053637e-01 -1.82054257e+00 -2.80393213e-01 -3.50573421e-01
-9.25325394e-01 1.03361738e+00 5.71145952e-01 5.69227636e-01
6.94099784e-01 -9.27256495e-02 3.77878428e-01 -2.04849064e-01
-7.97996700e-01 -5.93773484e-01 2.07969218e-01 3.34119737e-01
2.58583635e-01 4.39844802e-02 -4.82340187e-01 2.06161633e-01
8.22069883e-01 2.55159885e-02 3.11067462e-01 9.38960135e-01
-6.53191566e-01 -9.89248574e-01 -4.67952639e-01 4.73116785e-01
-5.05818486e-01 -6.75875992e-02 9.52387527e-02 5.14538586e-01
5.89855433e-01 1.15944922e+00 -3.39959949e-01 -6.10392034e-01
1.92004919e-01 6.43167198e-01 6.35681689e-01 -5.87985396e-01
-5.23101866e-01 2.05158979e-01 7.50716105e-02 -4.22202200e-01
-9.02285516e-01 -5.60520113e-01 -9.82401490e-01 -7.02916756e-02
-4.49310571e-01 -3.53909791e-01 9.38525200e-01 1.12227631e+00
2.61001319e-01 7.90324330e-01 5.56482017e-01 -5.87378085e-01
-5.43606818e-01 -1.05973363e+00 -9.40087974e-01 7.68049598e-01
1.17203720e-01 -1.06018758e+00 -1.45898640e-01 2.19952628e-01] | [9.75380802154541, 3.996514320373535] |
a735dee1-ff60-4a46-9212-dec5a5f413bf | an-embedding-for-eeg-signals-learned-using-a | 2304.06495 | null | https://arxiv.org/abs/2304.06495v1 | https://arxiv.org/pdf/2304.06495v1.pdf | An embedding for EEG signals learned using a triplet loss | Neurophysiological time series recordings like the electroencephalogram (EEG) or local field potentials are obtained from multiple sensors. They can be decoded by machine learning models in order to estimate the ongoing brain state of a patient or healthy user. In a brain-computer interface (BCI), this decoded brain state information can be used with minimal time delay to either control an application, e.g., for communication or for rehabilitation after stroke, or to passively monitor the ongoing brain state of the subject, e.g., in a demanding work environment. A specific challenge in such decoding tasks is posed by the small dataset sizes in BCI compared to other domains of machine learning like computer vision or natural language processing. A possibility to tackle classification or regression problems in BCI despite small training data sets is through transfer learning, which utilizes data from other sessions, subjects or even datasets to train a model. In this exploratory study, we propose novel domain-specific embeddings for neurophysiological data. Our approach is based on metric learning and builds upon the recently proposed ladder loss. Using embeddings allowed us to benefit, both from the good generalisation abilities and robustness of deep learning and from the fast training of classical machine learning models for subject-specific calibration. In offline analyses using EEG data of 14 subjects, we tested the embeddings' feasibility and compared their efficiency with state-of-the-art deep learning models and conventional machine learning pipelines. In summary, we propose the use of metric learning to obtain pre-trained embeddings of EEG-BCI data as a means to incorporate domain knowledge and to reach competitive performance on novel subjects with minimal calibration requirements. | ['Michael Tangermann', 'Théodore Papadopoulo', 'Pierre Guetschel'] | 2023-03-23 | null | null | null | null | ['metric-learning', 'eeg', 'metric-learning', 'eeg'] | ['computer-vision', 'methodology', 'methodology', 'time-series'] | [ 4.24473226e-01 1.39483800e-02 2.81389236e-01 -3.75906169e-01
-4.41552043e-01 -3.00895840e-01 5.64636886e-01 4.32471871e-01
-8.64784420e-01 9.06262040e-01 -1.93580147e-02 -1.80524647e-01
-4.19852465e-01 -5.53006113e-01 -5.55340707e-01 -5.51924765e-01
-2.94174373e-01 6.07008576e-01 1.33456200e-01 -1.03247263e-01
3.02634627e-01 8.23833764e-01 -1.53095341e+00 2.34318197e-01
6.96926892e-01 1.14706993e+00 5.07362723e-01 5.68100691e-01
1.51448026e-01 1.61007777e-01 -7.89478004e-01 9.44523960e-02
-1.49566367e-01 -2.64843196e-01 -4.70412761e-01 -3.71694475e-01
6.48143515e-02 -1.75323904e-01 -3.77584100e-01 8.19100916e-01
9.23697233e-01 -4.95051146e-02 7.21991003e-01 -1.10358989e+00
-7.06557482e-02 1.18776895e-01 1.57346755e-01 4.27310526e-01
4.17857260e-01 2.38949284e-01 3.33122134e-01 -4.43807125e-01
4.46609557e-01 6.23445868e-01 5.68268836e-01 7.56859660e-01
-1.51469946e+00 -7.38055229e-01 -1.79508790e-01 7.61640728e-01
-1.25961590e+00 -4.20347005e-01 6.61020517e-01 -6.14930212e-01
1.07532048e+00 1.57386452e-01 8.65533948e-01 1.59734702e+00
4.72812712e-01 4.50346291e-01 1.21257424e+00 -3.70916426e-01
5.93064070e-01 3.95900071e-01 1.95235033e-02 1.42519280e-01
-1.17295524e-02 1.39395386e-01 -6.07692778e-01 -6.44650087e-02
4.87779140e-01 1.53189758e-02 -6.29289031e-01 -3.39063138e-01
-1.37716413e+00 4.59353805e-01 4.69777256e-01 7.37181723e-01
-5.82554340e-01 1.04837306e-02 4.96913671e-01 6.20721042e-01
3.04852486e-01 6.69952393e-01 -7.62832880e-01 -5.87585926e-01
-1.03769207e+00 3.10606789e-02 7.96967924e-01 5.24856448e-01
4.92939413e-01 -2.15739325e-01 -2.91471690e-01 7.39133537e-01
-7.64046311e-02 1.80696860e-01 1.02139461e+00 -3.57559264e-01
2.63586462e-01 5.23357868e-01 -1.28597870e-01 -6.41680539e-01
-7.48364329e-01 -3.39936405e-01 -6.84586287e-01 4.06294942e-01
3.80826920e-01 -1.60248876e-01 -7.64321744e-01 1.76436758e+00
-5.31343445e-02 4.42653835e-01 -2.49153659e-01 7.77183056e-01
2.64023840e-01 2.36404240e-01 -1.01041906e-01 -2.46410787e-01
1.28107822e+00 -1.10904120e-01 -5.63283801e-01 -2.58644253e-01
6.73925042e-01 -2.45973200e-01 9.26632047e-01 8.33440840e-01
-8.18791151e-01 -4.30342078e-01 -1.19720232e+00 3.39050651e-01
-7.28383601e-01 9.29341987e-02 3.31880361e-01 6.02588296e-01
-1.04459476e+00 9.53687072e-01 -1.06137729e+00 -5.24179459e-01
4.95626420e-01 8.26545238e-01 -6.58387005e-01 9.40366387e-02
-1.18578327e+00 1.34317386e+00 5.03569663e-01 1.01313300e-01
-9.84688818e-01 -7.94908464e-01 -6.54717565e-01 9.11019444e-02
-1.22111447e-01 -3.83650273e-01 7.11553991e-01 -6.84374928e-01
-1.70032179e+00 8.50936055e-01 2.68455762e-02 -7.01360941e-01
6.05291367e-01 -9.34219807e-02 -4.41638887e-01 8.26704279e-02
-3.08218628e-01 4.20981348e-01 9.82704282e-01 -5.28999686e-01
-2.66846597e-01 -6.53835595e-01 -1.88619494e-01 -2.88093895e-01
-7.34321892e-01 -8.64076987e-02 1.69653054e-02 -2.56444335e-01
-2.43990377e-01 -9.54130411e-01 1.74067006e-01 -3.97491679e-02
9.08380672e-02 -1.59025028e-01 7.28461921e-01 -7.18342304e-01
9.43274796e-01 -2.14496946e+00 4.75356013e-01 2.54741192e-01
7.29327276e-02 5.24800062e-01 -1.54998362e-01 2.31135607e-01
-4.82296675e-01 -4.02194113e-01 -3.08144897e-01 -7.70040676e-02
-1.54959530e-01 8.35759491e-02 1.85070261e-01 7.58172691e-01
3.14587295e-01 8.17253232e-01 -9.14327621e-01 -1.68495215e-02
4.65909243e-01 7.07389116e-01 -2.75485337e-01 4.56182182e-01
2.10258022e-01 8.72766495e-01 1.68755099e-01 9.16945934e-02
2.28533819e-01 3.39454800e-01 -5.70535334e-03 -3.82567704e-01
-3.17247957e-03 4.55664247e-01 -1.03171158e+00 2.14657831e+00
-8.33404183e-01 9.58192766e-01 -1.06943389e-02 -1.77304745e+00
9.33133662e-01 6.57482088e-01 6.76559567e-01 -9.88213420e-01
5.65697193e-01 2.93820947e-01 3.54272783e-01 -6.17229342e-01
-2.98347384e-01 -9.35160965e-02 1.31903723e-01 4.31270272e-01
4.54324961e-01 -1.54184207e-01 9.69877932e-03 -2.87797064e-01
1.34724486e+00 1.54102966e-01 1.32679373e-01 -4.32825744e-01
6.17842138e-01 -4.20747221e-01 2.27045029e-01 3.23724538e-01
-2.64641732e-01 5.27876794e-01 4.03639823e-01 -2.42495045e-01
-7.09765315e-01 -1.02897632e+00 -5.84298730e-01 6.77995801e-01
-3.62688690e-01 -2.24892259e-01 -7.58345246e-01 -3.11165363e-01
-1.01742908e-01 7.17851460e-01 -7.69437373e-01 -7.66474247e-01
-4.99788404e-01 -5.44364035e-01 2.67712086e-01 5.10402083e-01
3.63043062e-02 -1.23707414e+00 -9.76314783e-01 5.82803369e-01
2.52087545e-02 -9.02660966e-01 4.56927940e-02 7.31602967e-01
-1.03852057e+00 -9.52866793e-01 -7.83997893e-01 -6.04161084e-01
3.52174520e-01 -3.76701623e-01 7.09767818e-01 -4.11409497e-01
-6.58666193e-01 6.01450920e-01 -2.71087646e-01 -6.54771864e-01
-2.42014453e-01 1.27324894e-01 5.13453245e-01 1.25480190e-01
6.17140889e-01 -1.12216604e+00 -7.15810359e-01 2.61756331e-01
-8.25602889e-01 -4.04328853e-01 5.56865454e-01 8.50605428e-01
3.47874373e-01 -1.68210357e-01 7.46826649e-01 -4.37408298e-01
1.02645898e+00 -4.21039164e-01 -4.09018636e-01 2.60338277e-01
-6.00471795e-01 1.24973714e-01 6.14831805e-01 -8.35801363e-01
-4.94395167e-01 -1.11446105e-01 -1.50797606e-01 -3.29516947e-01
-3.78582150e-01 4.69443321e-01 -1.90670937e-01 -1.12175010e-01
9.33297873e-01 2.02059597e-01 1.26831234e-01 -4.54510063e-01
-1.72535777e-02 1.06537557e+00 7.78713524e-01 -4.68088001e-01
2.85944283e-01 1.09325752e-01 -6.68401718e-02 -1.05709505e+00
-1.14590198e-01 -4.76960391e-01 -1.02333856e+00 -1.83194250e-01
7.19459236e-01 -7.23804355e-01 -6.85493767e-01 3.18346292e-01
-1.19293880e+00 -5.01752019e-01 -1.79392189e-01 8.54023933e-01
-7.88607121e-01 -3.12224124e-02 -1.04566552e-01 -6.15458667e-01
-3.31614375e-01 -1.03238940e+00 9.05275822e-01 -5.30628748e-02
-5.32623112e-01 -1.02134252e+00 2.04645872e-01 -9.12311394e-03
5.80363393e-01 3.72057498e-01 7.98791289e-01 -8.18788826e-01
3.28685082e-02 -6.56446934e-01 1.96375549e-02 6.39952362e-01
3.47225755e-01 -6.85340524e-01 -1.23522246e+00 -5.14447749e-01
2.72456259e-01 -2.99243152e-01 4.05647874e-01 2.54736573e-01
1.31911170e+00 2.28760824e-01 -2.57682562e-01 5.98887205e-01
1.04343796e+00 1.47744164e-01 9.85243618e-01 3.67608994e-01
2.87831068e-01 6.52977228e-01 2.30003953e-01 4.38889831e-01
-8.79425853e-02 7.54914641e-01 2.10502073e-01 3.43055546e-01
2.56443769e-02 1.77557364e-01 3.74231398e-01 6.90290987e-01
-9.85846817e-02 3.06286216e-01 -8.08846831e-01 6.85713172e-01
-1.61902332e+00 -7.03278005e-01 6.97503686e-02 2.63940358e+00
7.70055234e-01 2.00540617e-01 -2.42790626e-03 5.59885621e-01
2.90216923e-01 -5.05102932e-01 -6.64814889e-01 -5.68650782e-01
3.12209487e-01 7.76914477e-01 2.52753854e-01 5.33189215e-02
-6.93561494e-01 3.21441323e-01 5.22961378e+00 3.86863708e-01
-1.55859077e+00 5.11952937e-01 6.31967783e-02 -6.03275180e-01
2.32032642e-01 -3.92048568e-01 -4.08702165e-01 7.38462627e-01
1.70888698e+00 -8.35692361e-02 7.84287155e-01 4.66562212e-01
4.20707464e-01 -5.49079925e-02 -1.75212884e+00 1.55752051e+00
-3.04958620e-03 -1.06865644e+00 -6.62343979e-01 5.27500920e-02
1.24069951e-01 3.54480982e-01 -2.36257136e-01 4.19270992e-01
-6.19348824e-01 -1.05643106e+00 5.03689528e-01 6.66439772e-01
9.40266073e-01 -5.16320884e-01 6.77493572e-01 4.91742879e-01
-6.62497163e-01 -2.46388808e-01 -2.13962123e-01 -2.18001306e-01
1.70755852e-02 4.27172750e-01 -8.07771623e-01 1.80753097e-01
7.48665154e-01 7.68706739e-01 -4.80299532e-01 1.20907712e+00
-2.27417126e-01 5.65223992e-01 -4.75944340e-01 -5.04665934e-02
1.73782818e-02 -8.97790343e-02 4.37756240e-01 1.26565611e+00
4.85341877e-01 -1.69017404e-01 -4.32586521e-01 8.52215886e-01
1.35525718e-01 -7.50982910e-02 -7.17827320e-01 -3.15688662e-02
9.35022309e-02 1.22177744e+00 -4.17291999e-01 -2.06103316e-03
-2.23998562e-01 1.24469841e+00 4.55723554e-01 2.46217847e-01
-6.38993025e-01 -6.89489663e-01 7.61520684e-01 2.72799939e-01
1.08055986e-01 -4.05497551e-01 -3.99952084e-01 -1.18736625e+00
3.90994519e-01 -6.61279202e-01 -5.84347267e-03 -6.06924295e-01
-9.30044353e-01 7.26503968e-01 2.00727597e-01 -1.24691522e+00
-4.36100334e-01 -1.04263449e+00 -7.21647501e-01 1.12571323e+00
-1.47914350e+00 -6.40623510e-01 -3.57679635e-01 9.31347191e-01
9.52802524e-02 -2.92593956e-01 1.12886763e+00 5.52731693e-01
-3.99076700e-01 4.20269191e-01 8.75142887e-02 -1.24179885e-01
8.30063999e-01 -1.16507304e+00 -9.29054841e-02 5.44476330e-01
2.37996727e-01 5.79893112e-01 4.97388482e-01 -1.58066779e-01
-1.24207318e+00 -6.74968123e-01 7.94835269e-01 -5.25530398e-01
7.31774092e-01 -8.20684612e-01 -1.06188500e+00 3.03221524e-01
1.18916728e-01 1.57242090e-01 7.72710204e-01 3.30750234e-02
6.37007579e-02 -4.79687154e-01 -1.25438011e+00 3.81638139e-01
9.01681185e-01 -8.18003714e-01 -8.95826936e-01 3.15805495e-01
1.27412915e-01 -8.28415900e-02 -1.08712125e+00 1.75757647e-01
7.51709282e-01 -7.35864341e-01 7.50834107e-01 -6.17191792e-01
-5.97856343e-02 4.34900075e-02 4.70318832e-02 -1.84219623e+00
-1.16645239e-01 -5.95504105e-01 -2.10880533e-01 9.53509033e-01
3.00307125e-01 -8.40259612e-01 5.26981592e-01 9.75556970e-01
-6.25460297e-02 -6.06212854e-01 -1.30729914e+00 -7.90744483e-01
1.17148951e-01 -7.91660786e-01 4.70030129e-01 5.55815339e-01
4.81599838e-01 1.90813139e-01 -1.67686313e-01 -3.81029584e-02
2.37111881e-01 -3.61049205e-01 4.77624655e-01 -1.56433749e+00
-2.19743937e-01 -4.20277685e-01 -1.16032064e+00 -3.57990116e-01
3.85020047e-01 -1.18070817e+00 3.52417044e-02 -1.48639286e+00
-2.49458835e-01 -2.96400100e-01 -9.57031369e-01 5.07478058e-01
2.79665023e-01 2.16271713e-01 -3.08773667e-02 -1.28497452e-01
4.65719067e-02 5.00196815e-01 8.24175835e-01 -3.40844363e-01
-4.42103475e-01 3.84638906e-02 -3.47936064e-01 3.57318521e-01
8.60246599e-01 -6.22267902e-01 -4.88215029e-01 -4.82871056e-01
2.14719251e-02 -9.92420167e-02 4.18893278e-01 -1.44250667e+00
4.18267548e-01 4.63509351e-01 5.51309586e-01 -3.42748426e-02
4.56121415e-01 -1.18381166e+00 -1.53956771e-01 5.32965124e-01
-1.53327033e-01 -2.73937643e-01 4.14043933e-01 5.11116207e-01
-2.12904155e-01 -2.62415975e-01 8.03549409e-01 8.68194923e-02
-6.20439708e-01 2.51716852e-01 -5.45544565e-01 -1.46373376e-01
1.10811806e+00 -4.59618121e-01 1.55713916e-01 -1.16206869e-01
-9.98400748e-01 -1.20223656e-01 4.78920229e-02 6.24953866e-01
7.58294046e-01 -1.23206341e+00 -6.16075933e-01 6.84353352e-01
3.36702228e-01 -3.33501488e-01 -1.17167726e-01 1.27681446e+00
-1.77617610e-01 4.65158224e-01 -7.89081872e-01 -7.19890296e-01
-1.01468360e+00 3.66825968e-01 3.86067033e-01 5.78266568e-02
-7.10390270e-01 5.25703251e-01 -3.24570030e-01 -2.99933940e-01
4.20997977e-01 -6.20428205e-01 -2.25966483e-01 3.00327957e-01
7.23957956e-01 1.91466108e-01 7.90533125e-01 -2.39173934e-01
-7.22315192e-01 1.61147863e-01 -2.85016205e-02 -1.68882623e-01
1.66577280e+00 3.43032479e-01 -1.45695070e-02 6.39266729e-01
1.55333447e+00 -6.56545818e-01 -1.11430585e+00 -8.23169500e-02
9.07546803e-02 -2.74520725e-01 4.99709070e-01 -9.35746908e-01
-8.38735402e-01 1.28742921e+00 1.31358004e+00 -1.24936022e-01
1.24784303e+00 -1.88588232e-01 5.85936189e-01 5.12460947e-01
8.48443627e-01 -1.27097702e+00 -1.31810009e-01 1.74542606e-01
9.73090649e-01 -1.10332513e+00 -1.28469422e-01 5.71803749e-01
-3.61121118e-01 1.39449334e+00 1.25601962e-01 -1.37768969e-01
9.57252026e-01 2.82363266e-01 -1.26122564e-01 -1.22820474e-01
-5.68724751e-01 -4.71952893e-02 3.67507786e-01 9.85398650e-01
4.64983642e-01 7.17382580e-02 -3.87177378e-01 6.43212795e-01
-1.47976145e-01 4.97377098e-01 3.56805861e-01 8.06921899e-01
-8.20277482e-02 -1.24956977e+00 -1.13084577e-01 7.74892509e-01
-2.06028715e-01 1.33269370e-01 -1.63120493e-01 4.77653444e-01
1.73362911e-01 9.09420311e-01 4.33925763e-02 -4.04551029e-01
6.00959063e-01 3.88231814e-01 8.95457268e-01 -6.96181357e-01
-5.63511729e-01 -4.36090499e-01 -1.88180372e-01 -7.54060566e-01
-3.01873237e-01 -8.24592113e-01 -8.09022486e-01 2.97854334e-01
-1.60133436e-01 -1.32697374e-01 1.26442814e+00 1.14863718e+00
4.17464048e-01 8.25455189e-01 3.41663688e-01 -1.14354944e+00
-4.17691618e-01 -1.32917404e+00 -6.92431450e-01 4.42348421e-01
3.60824287e-01 -8.53604138e-01 -1.71799213e-01 4.05175202e-02] | [13.123900413513184, 3.4431936740875244] |
9c4a0e12-54b7-458f-9e38-8be82f92352c | optimal-investment-under-partial-observations | 2212.04394 | null | https://arxiv.org/abs/2212.04394v1 | https://arxiv.org/pdf/2212.04394v1.pdf | Optimal investment under partial observations and robust VaR-type constraint | The present paper extends the literature on utility maximization by combining the framework of partial information and (robust) regulatory constraints. Partial information is characterized by the fact that the stock price itself is observable to the optimizing financial institution, but the outcome of the market price of risk $\theta$ is unknown to the institution. The regulator builds the same or a different belief about the market price of risk as the financial institution. The solution to our optimization problem takes the same form as in the full information case: the optimal wealth can be expressed as a decreasing function of the state price density, and the regulatory threshold is ensured in the intermediate economic states. The main difference lies in the terminal state price density depending on the entire evolution of the estimated market price of risk $\hat{\theta}(s)$. The subjective evaluation of the regulatory constraint influences the width of the ensured region. | ['An Chen', 'Nicole Bäuerle'] | 2022-12-08 | null | null | null | null | ['type'] | ['speech'] | [-3.04665416e-01 6.28484666e-01 -6.39406025e-01 6.42128363e-02
-2.20105439e-01 -7.80518115e-01 1.26755014e-01 -2.24478170e-01
-7.08609521e-01 9.08707857e-01 7.50240162e-02 -4.43001062e-01
-6.03921890e-01 -9.00111675e-01 -2.72145569e-01 -9.27816570e-01
2.23887518e-01 2.97642320e-01 -8.53706617e-03 -1.21183693e-01
3.68917644e-01 5.40265068e-02 -8.09677482e-01 -4.61859852e-01
6.42320454e-01 1.48365545e+00 7.75950029e-02 2.04458237e-01
1.58574022e-02 6.05168402e-01 -6.59255445e-01 -4.30898696e-01
7.75607944e-01 -2.09459841e-01 -3.32663894e-01 1.53443322e-01
-8.67467642e-01 -3.93268704e-01 -2.98125185e-02 1.45374274e+00
3.08712889e-02 -9.07245129e-02 6.45133138e-01 -9.31397676e-01
-5.60485899e-01 7.94073582e-01 -2.34195367e-01 2.07683995e-01
1.45812947e-02 -1.28953401e-02 1.08650613e+00 -2.54887491e-01
5.16481102e-01 7.87485719e-01 2.19109431e-02 4.19527233e-01
-1.05797756e+00 -4.29869056e-01 4.36892003e-01 -1.51351899e-01
-1.31595504e+00 -1.59398854e-01 4.97772127e-01 -6.64851189e-01
4.03750867e-01 -1.36787165e-02 5.35706878e-01 3.97822261e-01
4.87501949e-01 8.16469863e-02 1.01075947e+00 -3.06980312e-01
6.67408228e-01 5.19942462e-01 -1.15032224e-02 3.58626107e-03
6.85549736e-01 3.23616832e-01 -2.10041866e-01 -1.64547071e-01
7.47192383e-01 -3.08861196e-01 -6.62128925e-01 -3.96472365e-01
-5.27706027e-01 8.76993954e-01 -3.88834029e-02 2.35366911e-01
-5.35949945e-01 -6.89876080e-02 -2.36380175e-02 7.53385901e-01
2.58187324e-01 3.85307521e-01 -5.74249804e-01 -4.19058427e-02
-5.42539597e-01 8.39686170e-02 1.02121115e+00 3.84688228e-01
3.96657437e-01 1.95835024e-01 3.08929831e-02 -5.59077924e-03
6.66309476e-01 7.22308874e-01 1.06640063e-01 -1.23995733e+00
7.91168034e-01 2.24469692e-01 9.33306038e-01 -5.15475512e-01
-1.13784475e-02 -7.02817440e-01 -6.52674958e-02 6.39934182e-01
9.12742019e-01 -8.98088157e-01 -2.50382632e-01 2.06350350e+00
-7.78778866e-02 -5.50824642e-01 2.61921287e-01 6.90510690e-01
-2.43450984e-01 8.18603158e-01 -3.87509346e-01 -9.71181989e-01
1.07826591e+00 -2.33156607e-01 -9.97416437e-01 2.51594614e-02
1.56491861e-01 -2.17344344e-01 2.55791664e-01 3.59365046e-01
-1.37990284e+00 1.69742301e-01 -8.69965136e-01 8.70899856e-01
-1.16356067e-01 -4.16816682e-01 1.64773285e-01 9.09506559e-01
-1.01467812e+00 7.41073787e-01 -4.78291899e-01 5.42064130e-01
4.44360375e-02 3.90667528e-01 1.30900055e-01 5.86228788e-01
-1.37258863e+00 1.19946253e+00 2.20238611e-01 2.76706368e-01
-5.93713939e-01 -5.66136122e-01 -7.28256762e-01 5.78390718e-01
7.34618127e-01 -4.11890358e-01 1.28120387e+00 -1.21427965e+00
-1.45802498e+00 4.64892000e-01 4.12419647e-01 -4.93120730e-01
7.47878492e-01 5.44962399e-02 -1.68267727e-01 1.40354991e-01
2.44185165e-01 -3.80881608e-01 4.71378624e-01 -1.00343359e+00
-4.24144089e-01 -6.10395014e-01 1.10476203e-01 4.34847802e-01
1.38069093e-01 2.45467618e-01 2.43148416e-01 -3.92242253e-01
-2.52705235e-02 -6.06199205e-01 -4.44889039e-01 -3.47059399e-01
-1.56262472e-01 1.04176834e-01 1.07379913e-01 -5.25613189e-01
1.29256845e+00 -1.84900010e+00 -6.14691004e-02 5.76899648e-01
-2.91736335e-01 -2.03470334e-01 5.90079010e-01 3.58533919e-01
-3.59848976e-01 3.30024838e-01 -4.17613089e-01 2.38334835e-01
4.48412955e-01 -6.69063209e-03 -3.29531819e-01 7.23196208e-01
-5.14763072e-02 4.03413206e-01 -4.32320744e-01 1.65458888e-01
-2.16684148e-01 -1.66414231e-01 -2.09991246e-01 -1.19960152e-01
-9.79540721e-02 1.44657925e-01 -9.22759652e-01 3.18918496e-01
3.27531248e-01 -1.15462922e-01 4.17239934e-01 6.49796069e-01
-6.84935510e-01 1.65965959e-01 -1.63569593e+00 9.08824980e-01
8.53797197e-02 6.22783303e-02 5.14281988e-01 -1.00564122e+00
5.83095372e-01 7.09856153e-01 3.92795861e-01 -3.40258509e-01
4.71171558e-01 4.28860709e-02 1.98340595e-01 -1.98801830e-01
2.94039637e-01 -7.57685304e-01 -2.61296868e-01 8.30399275e-01
-3.28510255e-01 -1.06333166e-01 8.12581703e-02 2.48967502e-02
6.82398081e-01 4.80518118e-03 5.51570535e-01 -6.32024288e-01
2.02237263e-01 -3.61435592e-01 8.72867525e-01 5.60653687e-01
-2.34129012e-01 -3.09377946e-02 1.49926555e+00 2.89863318e-01
-8.65829289e-01 -1.13839173e+00 -4.24375206e-01 3.50897402e-01
-8.00609589e-03 3.86772186e-01 -8.58402133e-01 -3.61263305e-01
2.37420365e-01 9.83318567e-01 -6.45195961e-01 3.04987617e-02
-7.46859014e-02 -1.02387226e+00 -3.56494606e-01 8.52373466e-02
4.99098629e-01 -8.33056509e-01 -8.30243587e-01 1.25066534e-01
2.03093842e-01 -5.82190990e-01 -5.06666780e-01 1.77505583e-01
-6.60838783e-01 -8.12648416e-01 -5.59928179e-01 1.67463556e-01
6.46609426e-01 -6.14113510e-01 7.17729986e-01 -3.41050535e-01
4.30211216e-01 4.82049912e-01 9.77972895e-02 -8.05339336e-01
-3.50083441e-01 -3.98741305e-01 -1.64557040e-01 4.34826553e-01
-2.68453956e-01 -1.79251909e-01 -5.47311246e-01 2.17345998e-01
-8.18391323e-01 -7.63171613e-01 -1.18258208e-01 3.41089368e-01
2.94266462e-01 7.70327270e-01 1.04623532e+00 -6.54099584e-01
7.04264700e-01 -5.20599306e-01 -1.40644693e+00 2.99793601e-01
-9.48104501e-01 2.03332454e-01 7.83078372e-02 -1.08409196e-03
-1.70374084e+00 -1.50279373e-01 2.66223043e-01 9.72064883e-02
1.48337305e-01 7.73738801e-01 -7.03473747e-01 3.53235096e-01
-3.78653593e-02 -1.57728955e-01 9.66682434e-02 -6.59742653e-01
7.47439638e-02 3.89042616e-01 3.57388347e-01 -5.45582533e-01
7.52678692e-01 4.34420735e-01 1.38599932e-01 -4.72403914e-02
-6.28991365e-01 2.60375421e-02 -3.54697466e-01 -2.00197205e-01
8.26295316e-01 -8.31564248e-01 -1.00732148e+00 4.38811749e-01
-7.64917791e-01 -1.30410001e-01 -8.52519691e-01 6.32707298e-01
-7.16029763e-01 3.15520436e-01 -3.36665332e-01 -1.59492445e+00
7.96588883e-02 -1.12331128e+00 -2.09144987e-02 2.32674509e-01
1.05077036e-01 -9.63471174e-01 5.60626499e-02 3.55692953e-01
4.30640787e-01 4.19407099e-01 8.50713491e-01 -4.52907950e-01
-8.93841088e-01 -4.22111005e-01 1.55331925e-01 3.15881222e-01
1.39207183e-03 -3.41834016e-02 -4.54197735e-01 -3.43533531e-02
9.28903639e-01 1.83614910e-01 6.00672185e-01 7.99085021e-01
1.88970119e-01 -6.17932081e-01 1.53221816e-01 1.26284570e-01
1.73900747e+00 8.14339638e-01 5.61446905e-01 5.62263072e-01
-3.52836877e-01 8.24744344e-01 5.89177907e-01 8.40510428e-01
2.48703346e-01 6.13295496e-01 6.24503732e-01 4.52762723e-01
9.29255366e-01 5.29078431e-02 6.22915268e-01 2.59177461e-02
-2.63740629e-01 9.87724680e-03 -6.69328570e-01 5.85220695e-01
-1.80234241e+00 -1.15661836e+00 4.92343694e-01 2.59359217e+00
7.79808283e-01 4.14820552e-01 2.24999681e-01 2.13631600e-01
8.96998823e-01 1.16885677e-01 -6.71748042e-01 -6.85046077e-01
-2.65393257e-01 -9.35911089e-02 8.99485230e-01 8.97577882e-01
-4.36648220e-01 2.06726074e-01 7.00292397e+00 4.71913278e-01
-8.01191926e-01 -1.81650057e-01 9.98759508e-01 -3.32895160e-01
-5.41266084e-01 4.92368639e-01 -7.09471047e-01 8.73059750e-01
8.09920251e-01 -1.10251570e+00 1.64810643e-01 7.07506776e-01
5.59649765e-01 -6.22989297e-01 -6.95883870e-01 4.02227901e-02
-4.46801186e-01 -1.02002549e+00 -6.08869493e-01 6.85911775e-01
6.33785963e-01 -3.34506571e-01 4.78882253e-01 -2.74950832e-01
6.34458363e-01 -6.22844934e-01 1.26751852e+00 7.45629668e-01
5.30521631e-01 -1.11584079e+00 9.23623085e-01 6.88573658e-01
-9.94064510e-01 -6.60629392e-01 -3.97259369e-02 -6.29403532e-01
5.92906833e-01 7.05989182e-01 -3.23754698e-01 6.00284636e-01
3.95615876e-01 -6.44469932e-02 1.44693002e-01 6.95938051e-01
-3.66735131e-01 2.85490751e-01 -2.65851557e-01 2.15610951e-01
1.42501369e-01 -9.50287402e-01 5.41387498e-01 3.68385047e-01
4.27072555e-01 4.45957989e-01 -1.31956458e-01 1.18793786e+00
-1.41547620e-02 1.40045201e-02 -5.69004238e-01 -8.18480253e-02
2.57941216e-01 6.26847148e-01 -5.79465270e-01 -1.15428306e-01
-5.11986792e-01 6.52707070e-02 -1.79279864e-01 5.42685747e-01
-3.45100790e-01 -8.54434967e-02 5.17575443e-01 1.06199577e-01
3.42007399e-01 1.28084779e-01 -6.55698001e-01 -1.07511294e+00
2.57967234e-01 -2.90288419e-01 7.21757352e-01 -4.15083379e-01
-7.91398644e-01 -2.24430636e-02 2.56960630e-01 -8.51588666e-01
-4.51815605e-01 -4.98845398e-01 -7.81970620e-01 1.19996178e+00
-1.54881513e+00 3.41782272e-02 1.02877700e+00 3.04322690e-01
-1.33312240e-01 -1.67064071e-01 4.00025010e-01 -3.79928499e-01
-7.59900033e-01 -4.24854048e-02 5.11816144e-01 1.74065471e-01
-2.40467072e-01 -1.36229646e+00 -2.47101039e-01 1.02104211e+00
-5.42755663e-01 4.86591220e-01 6.45464420e-01 -8.00562501e-01
-9.99416709e-01 -4.56510425e-01 9.97125089e-01 -3.65750432e-01
1.17208791e+00 1.54055655e-01 -4.78978157e-01 6.26859188e-01
3.80177945e-01 -4.93354023e-01 4.64771748e-01 -4.22684401e-01
1.71995517e-02 -1.99920312e-01 -1.44968092e+00 4.53548789e-01
2.99942255e-01 -2.99570769e-01 -5.80635607e-01 -2.13491350e-01
7.00288475e-01 2.04906270e-01 -1.08195019e+00 2.62516677e-01
5.89394450e-01 -8.48264873e-01 6.18363321e-01 -4.02279854e-01
-1.20976031e-01 -9.52260122e-02 -1.49230868e-01 -1.10123420e+00
-1.99629381e-01 -9.17829871e-01 4.05586958e-01 1.06287313e+00
7.36710787e-01 -1.03326559e+00 6.41334176e-01 1.20077860e+00
3.63892794e-01 -6.60661042e-01 -1.68869710e+00 -7.77910888e-01
5.63052654e-01 -7.02333897e-02 5.72830856e-01 6.25005543e-01
3.66664201e-01 -2.78465301e-01 -1.24087483e-01 2.12582380e-01
1.04537296e+00 3.56963485e-01 -1.05292633e-01 -1.10030901e+00
-5.12810349e-01 -4.72795129e-01 2.50675559e-01 -5.51636457e-01
1.59822971e-01 -4.16496396e-01 -2.09996134e-01 -9.69510019e-01
5.08396067e-02 -1.55798637e-03 -6.72117293e-01 -1.24584585e-01
2.57323682e-01 -6.49236679e-01 6.49026692e-01 -5.28024547e-02
-9.91922542e-02 3.96637738e-01 1.06506455e+00 2.09531888e-01
-2.19497889e-01 5.94642401e-01 -1.04130232e+00 1.04918778e+00
8.39423001e-01 -2.86661506e-01 -4.29047585e-01 2.19626771e-03
7.02216089e-01 9.42151964e-01 8.04189891e-02 -1.42288759e-01
1.02026455e-01 -8.11274946e-01 -1.77808508e-01 -5.69810450e-01
1.31013274e-01 -8.73534918e-01 5.53645074e-01 6.33760989e-01
-3.82823735e-01 -6.25024140e-02 -4.52224702e-01 4.09436464e-01
2.49588508e-02 -8.84180307e-01 9.25860703e-01 -6.21543109e-01
2.09627435e-01 8.43967423e-02 -6.98416233e-01 -5.66406287e-02
9.43763673e-01 -2.21590057e-01 -3.39655519e-01 -7.61065900e-01
-9.76831138e-01 3.75410020e-01 7.55326033e-01 -1.00486822e-01
2.03955576e-01 -8.34723115e-01 -6.06143415e-01 -2.74638683e-01
-4.40759391e-01 -2.39471987e-01 8.87251794e-02 4.68579710e-01
1.87547863e-01 5.32526493e-01 5.42526357e-02 2.73208380e-01
-5.55350244e-01 5.56125998e-01 8.65168393e-01 -4.13433850e-01
-4.82623242e-02 4.04195905e-01 4.49998826e-01 3.99807453e-01
1.08704932e-01 -9.97936055e-02 -2.26751506e-01 4.79267836e-01
4.91401404e-01 5.08303106e-01 -5.02708375e-01 -6.02668822e-01
-3.40417139e-02 4.74862218e-01 3.60809684e-01 -7.57756948e-01
1.28778934e+00 -7.37845778e-01 -2.03833148e-01 5.83649337e-01
5.76077819e-01 -4.19987045e-04 -1.52104843e+00 -2.03691781e-01
4.64876503e-01 -4.17310834e-01 1.04846954e-02 -8.06993306e-01
-1.31487501e+00 2.20495522e-01 7.96615109e-02 5.71339965e-01
8.90865862e-01 -8.66032168e-02 -1.30366487e-02 1.00026451e-01
4.78964299e-01 -1.44346535e+00 -4.32521343e-01 6.78914115e-02
8.15879464e-01 -5.83053350e-01 -5.45795495e-03 -2.50272483e-01
-7.99846768e-01 4.32888985e-01 -6.15054034e-02 -1.51402041e-01
1.15112782e+00 5.32911301e-01 -3.05075292e-02 2.77267378e-02
-8.12943816e-01 4.63180505e-02 6.84779435e-02 3.53957385e-01
-1.26740500e-01 2.01845109e-01 -8.59000742e-01 1.13974869e+00
-1.42508119e-01 -8.03264454e-02 1.04750538e+00 7.88292944e-01
-7.22390831e-01 -9.24413860e-01 -6.29923940e-01 7.48552978e-02
-1.11759329e+00 2.66456068e-03 -2.66531825e-01 5.88824153e-01
2.02811360e-01 7.40151048e-01 1.32667542e-01 7.28997827e-01
4.33079749e-01 1.87403306e-01 3.81435342e-02 -6.23899639e-01
-2.95310855e-01 4.71372187e-01 2.10010171e-01 -2.59026706e-01
-2.99458653e-01 -9.99583483e-01 -1.21189892e+00 5.58061600e-02
-5.16431808e-01 5.74607790e-01 6.01125836e-01 1.09008920e+00
-3.59333128e-01 -7.68041536e-02 1.08829391e+00 -2.58974165e-01
-1.11633074e+00 -3.73385668e-01 -1.53258896e+00 -5.07339202e-02
4.26002860e-01 -3.71121019e-01 -9.04608250e-01 -1.72566846e-01] | [4.876237392425537, 3.827347755432129] |
12033d29-4c57-4745-aa0d-529f78293c9b | multi-modal-siamese-network-for-entity | null | null | https://dl.acm.org/doi/10.1145/3534678.3539244 | https://dl.acm.org/doi/10.1145/3534678.3539244 | Multi-modal Siamese Network for Entity Alignment | The booming of multi-modal knowledge graphs (MMKGs) has raised the imperative demand for multi-modal entity alignment techniques, which facilitate the integration of multiple MMKGs from separate data sources. Unfortunately, prior arts harness multi-modal knowledge only via the heuristic merging of uni-modal feature embeddings. Therefore, inter-modal cues concealed in multi-modal knowledge could be largely ignored. To deal with that problem, in this paper, we propose a novel Multi-modal Siamese Network for Entity Alignment (MSNEA) to align entities in different MMKGs, in which multi-modal knowledge could be comprehensively leveraged by the exploitation of inter-modal effect. Specifically, we first devise a multi-modal knowledge embedding module to extract visual, relational, and attribute features of entities to generate holistic entity representations for distinct MMKGs. During this procedure, we employ inter-modal enhancement mechanisms to integrate visual features to guide relational feature learning and adaptively assign attention weights to capture valuable attributes for alignment. Afterwards, we design a multi-modal contrastive learning module to achieve inter-modal enhancement fusion with avoiding the overwhelming impact of weak modalities. Experimental results on two public datasets demonstrate that our proposed MSNEA provides state-of-the-art performance with a large margin compared with competitive baselines. | ['Enhong Chen', 'Nicholas Jing Yuan', 'Zhefeng Wang', 'Han Wu', 'Tong Xu', 'Zhi Li', 'Liyi Chen'] | 2022-08-14 | null | null | null | kdd-2022-8 | ['multi-modal-entity-alignment', 'entity-alignment', 'multi-modal-knowledge-graph', 'entity-alignment'] | ['knowledge-base', 'knowledge-base', 'knowledge-base', 'natural-language-processing'] | [-1.61287218e-01 6.40243962e-02 -3.39210123e-01 -1.34720027e-01
-7.26059198e-01 -5.86744905e-01 5.62270284e-01 2.48832881e-01
-3.36647004e-01 2.54189998e-01 5.25393724e-01 2.12890655e-01
-4.27701324e-01 -8.67797554e-01 -5.85627377e-01 -6.74581170e-01
1.11839108e-01 8.86333808e-02 -3.10582500e-02 -2.24815518e-01
2.55593527e-02 3.59415293e-01 -1.41370344e+00 3.00522566e-01
1.15822434e+00 9.45167601e-01 -2.88000591e-02 1.74071491e-01
-4.47577357e-01 5.10664225e-01 -1.36685580e-01 -8.63454878e-01
1.10856891e-01 -1.36298448e-01 -6.85992539e-01 7.58378580e-02
6.04018331e-01 -1.41379610e-02 -4.65963125e-01 1.08658373e+00
5.96144795e-01 1.55756310e-01 7.17103839e-01 -1.50815523e+00
-1.23379385e+00 7.53034711e-01 -8.97240043e-01 1.08789131e-01
2.71838784e-01 1.43194720e-01 1.61531270e+00 -1.38319409e+00
5.74114382e-01 1.15638137e+00 5.49416542e-01 9.66026485e-02
-1.24612808e+00 -5.35443842e-01 3.40334207e-01 4.04933006e-01
-1.77458966e+00 -3.17819625e-01 1.08384800e+00 -2.51499206e-01
6.84005797e-01 1.09162554e-01 6.89617455e-01 9.05947447e-01
-2.82650322e-01 9.16160762e-01 6.90131843e-01 -3.74385625e-01
-3.24939311e-01 1.89028963e-01 -1.19712520e-02 8.79534900e-01
3.68944287e-01 -2.03448460e-01 -7.69485831e-01 -8.19634087e-03
6.06647372e-01 8.84945467e-02 -2.38824844e-01 -8.22530091e-01
-1.53434622e+00 6.69219434e-01 9.15600359e-01 3.60125095e-01
-5.74162424e-01 -1.05021432e-01 3.62105906e-01 3.96967456e-02
1.25491112e-01 4.30261701e-01 -2.58921653e-01 2.86033958e-01
-6.14025772e-01 -5.14127426e-02 1.93218872e-01 1.07961190e+00
1.13935065e+00 -1.07208595e-01 -3.20008069e-01 8.84545445e-01
5.08670807e-01 5.69639862e-01 3.03902656e-01 -6.08887732e-01
8.70899260e-01 1.32558870e+00 -2.30776757e-01 -1.38726580e+00
-4.61786389e-01 -4.27830547e-01 -1.00105464e+00 -2.95289904e-01
6.19335473e-02 2.51763612e-01 -7.21611857e-01 1.91338944e+00
6.47815526e-01 9.19432268e-02 2.90064275e-01 8.71090651e-01
1.02591133e+00 4.20077980e-01 2.75585383e-01 1.40197441e-01
1.52358150e+00 -1.01012909e+00 -5.25103152e-01 -1.10555135e-01
6.50402367e-01 -5.67058563e-01 1.18603861e+00 -2.46919945e-01
-8.08051348e-01 -5.73439717e-01 -1.05526447e+00 -2.89966106e-01
-8.33455861e-01 1.19279347e-01 6.68310404e-01 4.56682980e-01
-5.60550690e-01 1.37137637e-01 -5.97957313e-01 -2.43509829e-01
5.82281172e-01 1.97836995e-01 -7.37109244e-01 -2.23808467e-01
-1.35482633e+00 8.30120921e-01 7.37072051e-01 2.74556160e-01
-3.08264554e-01 -8.94291997e-01 -1.17131031e+00 5.83489761e-02
4.93575543e-01 -9.20147121e-01 3.79165709e-01 -7.08080232e-01
-9.15365219e-01 8.09589326e-01 4.45802603e-03 1.48272678e-01
7.24715814e-02 -2.05837578e-01 -8.18973541e-01 2.87785947e-01
2.68213242e-01 7.23987520e-01 7.53020763e-01 -1.54079592e+00
-6.28360033e-01 -4.95102674e-01 2.85694331e-01 6.35629773e-01
-8.56551349e-01 -3.76501560e-01 -7.53288627e-01 -7.36281693e-01
1.67686254e-01 -6.57006383e-01 1.58289328e-01 -1.65570214e-01
-5.80877542e-01 -2.62828022e-01 5.62481880e-01 -5.00861645e-01
1.36284268e+00 -2.15250444e+00 6.98960781e-01 3.76527905e-01
4.83173758e-01 8.43463540e-02 -3.58410627e-01 4.81911600e-01
-2.68606953e-02 5.49729727e-02 -1.07666165e-01 -4.14911300e-01
2.73874372e-01 1.35040209e-01 -1.59957021e-01 2.76050508e-01
5.29689789e-01 1.37021649e+00 -9.59478378e-01 -8.52176428e-01
4.64637987e-02 5.75768352e-01 -4.69388694e-01 1.82726637e-01
1.95574448e-01 1.75022200e-01 -5.75674176e-01 1.23514616e+00
7.22484589e-01 -5.71316957e-01 1.13714918e-01 -9.76815760e-01
1.49856731e-01 -3.13915223e-01 -1.14212000e+00 2.05064797e+00
-3.05934519e-01 5.27072176e-02 -2.10497528e-01 -9.04082954e-01
5.92206717e-01 -1.12793073e-02 4.95242715e-01 -6.74880326e-01
8.90537128e-02 1.06661394e-01 -1.91116899e-01 -4.74377185e-01
8.28882217e-01 -8.31058696e-02 -4.03567344e-01 1.21535532e-01
3.06096733e-01 4.60514158e-01 1.35768071e-01 5.46698868e-01
6.90348148e-01 2.10799769e-01 1.94057792e-01 9.13259685e-02
6.90342069e-01 -1.16147339e-01 5.06532013e-01 3.26609999e-01
-2.67703176e-01 3.88679355e-01 1.63601488e-01 -1.24878392e-01
-6.99515402e-01 -1.26601458e+00 1.58625618e-02 1.24100184e+00
6.88065708e-01 -4.79913056e-01 -2.59181052e-01 -9.61442769e-01
2.55409420e-01 2.69243926e-01 -8.28408301e-01 -5.32745838e-01
-4.23590124e-01 -8.15937102e-01 5.79960108e-01 6.80937052e-01
4.53903198e-01 -9.53616261e-01 -1.36904106e-01 -2.01546494e-02
-3.63861084e-01 -1.29053307e+00 -5.94371855e-01 -4.04567085e-02
-4.39739823e-01 -1.04780710e+00 -7.53511131e-01 -8.07055593e-01
7.99685597e-01 4.02981251e-01 1.08920789e+00 6.79557025e-03
-1.77765638e-01 7.08683193e-01 -4.64658320e-01 1.05653167e-01
1.67616442e-01 4.40104932e-01 4.88603227e-02 3.29913914e-01
5.38506269e-01 -7.61238754e-01 -6.79447591e-01 5.16044535e-02
-9.77470338e-01 1.05260648e-01 1.05897069e+00 8.76897633e-01
7.17849433e-01 -1.34827957e-01 8.64079058e-01 -7.05139458e-01
2.69786894e-01 -7.26348400e-01 -1.77092776e-01 6.79446220e-01
-5.48383355e-01 1.81616127e-01 6.93141580e-01 -4.72271740e-01
-9.13912058e-01 -2.03945369e-01 3.62264276e-01 -7.41260231e-01
1.09926209e-01 9.38967824e-01 -7.15324998e-01 -7.52030239e-02
2.78878540e-01 3.47706884e-01 -4.29780185e-01 -2.54843831e-01
1.06433177e+00 3.78051043e-01 6.65981472e-01 -7.46680260e-01
1.17884195e+00 4.15077597e-01 -3.09905037e-02 -4.39521253e-01
-8.85659039e-01 -4.85548139e-01 -7.60613084e-01 -1.25756428e-01
8.90987396e-01 -1.22217917e+00 -3.54203910e-01 4.90619689e-01
-7.58003652e-01 3.95128131e-01 -2.69788355e-01 3.08707356e-01
-1.52732328e-01 5.71718276e-01 -2.71770239e-01 -3.57305646e-01
-2.78085947e-01 -8.47799540e-01 1.06591916e+00 6.01655543e-01
8.40290934e-02 -9.95971203e-01 8.39992166e-02 6.11413002e-01
2.64492512e-01 1.93843499e-01 1.00850213e+00 -6.86795056e-01
-8.06954503e-01 -8.75898078e-02 -6.80839539e-01 -4.98490557e-02
3.62300366e-01 -1.31477654e-01 -8.58062029e-01 -2.89906830e-01
-8.52144182e-01 -5.19743800e-01 8.44592869e-01 -1.71449184e-01
8.98150086e-01 -1.85619861e-01 -3.67194563e-01 7.16832936e-01
1.43310034e+00 -3.96406710e-01 4.30227071e-01 4.09239352e-01
1.51435506e+00 5.69597483e-01 7.00402975e-01 3.83351386e-01
1.18606615e+00 6.09690070e-01 4.98414010e-01 -1.09829366e-01
-1.36595428e-01 -4.42446321e-01 2.19344944e-01 1.27897286e+00
7.23904893e-02 -3.17658335e-02 -7.87531912e-01 9.79485512e-01
-1.84559000e+00 -8.03395510e-01 2.66470999e-01 1.95217335e+00
9.33209777e-01 -1.43387452e-01 1.44310519e-01 -2.58149296e-01
7.56359875e-01 3.28343570e-01 -5.18952847e-01 2.62664586e-01
-4.35427487e-01 -2.59839803e-01 3.07873607e-01 -2.47710617e-03
-1.29178369e+00 9.02099371e-01 4.14138842e+00 1.01765084e+00
-8.19617033e-01 -3.24364640e-02 7.96826631e-02 3.21950652e-02
-7.89627135e-01 -8.58031362e-02 -7.69301474e-01 4.28402543e-01
4.85538363e-01 -1.74688026e-01 2.47944951e-01 5.10814130e-01
-4.39144880e-01 2.77576149e-01 -9.03777719e-01 1.08468926e+00
1.07762344e-01 -1.13832712e+00 3.49922210e-01 4.91111539e-02
8.18864644e-01 -2.07861558e-01 1.68052092e-01 6.79135501e-01
8.30873251e-02 -7.80394912e-01 4.74430531e-01 6.42114401e-01
7.97959268e-01 -1.06682408e+00 6.14707232e-01 -1.54653057e-01
-1.85387659e+00 6.94356710e-02 -2.05971718e-01 6.65111959e-01
1.41771600e-01 3.91323507e-01 -3.12195390e-01 1.45112729e+00
6.82594836e-01 8.82545114e-01 -8.15769076e-01 8.92971098e-01
8.99108797e-02 -3.04606203e-02 -1.86792687e-01 3.47243488e-01
2.90472150e-01 -1.06004834e-01 4.46644485e-01 1.18134511e+00
2.60520637e-01 -7.53371269e-02 1.86604515e-01 7.71637440e-01
-3.49414229e-01 3.89506489e-01 -5.50657213e-01 -3.71232390e-01
8.26057196e-01 1.54388809e+00 -4.90017772e-01 -1.87973723e-01
-7.51648784e-01 1.05698323e+00 8.18679512e-01 3.63784790e-01
-9.70206082e-01 -5.34741223e-01 6.32246733e-01 -3.09308201e-01
6.09378338e-01 -3.08841318e-02 -2.75598913e-01 -1.52338719e+00
1.56956613e-01 -6.96177304e-01 7.51776040e-01 -5.64821720e-01
-1.75466716e+00 5.43609381e-01 -2.27204263e-02 -1.37405241e+00
2.11450130e-01 -3.49472851e-01 -2.80445397e-01 8.92577708e-01
-1.88928688e+00 -1.94301438e+00 -4.34293002e-01 8.41049850e-01
-8.74921121e-03 -2.79919863e-01 7.22739637e-01 5.84103584e-01
-8.32435787e-01 1.05049264e+00 -1.90027356e-01 3.28505784e-01
8.78696620e-01 -1.35866296e+00 -1.27517760e-01 9.34975684e-01
3.28519374e-01 8.09815586e-01 1.95035756e-01 -5.46904385e-01
-1.70746565e+00 -1.14551115e+00 7.75989592e-01 -4.33778226e-01
8.80476296e-01 -6.19814731e-03 -1.18958163e+00 6.08724833e-01
2.05570847e-01 2.92761862e-01 1.05455852e+00 4.16570425e-01
-8.97184789e-01 -1.80767685e-01 -8.61593962e-01 7.32207239e-01
1.00987184e+00 -9.18559551e-01 -7.47822106e-01 -2.36391217e-01
8.12401652e-01 -2.77996927e-01 -1.38490129e+00 7.86973953e-01
5.32554567e-01 -5.55029511e-01 1.20879281e+00 -7.53691435e-01
5.46084464e-01 -6.93676233e-01 -4.02887762e-01 -1.23382723e+00
-4.14197654e-01 -2.11684912e-01 -6.00852430e-01 1.69363892e+00
3.22020650e-01 -4.37683523e-01 4.68746454e-01 5.28774858e-01
-2.16407269e-01 -8.83791804e-01 -9.00025427e-01 -5.19235969e-01
-1.05334170e-01 -9.54383016e-02 7.04482377e-01 1.49969149e+00
6.71977773e-02 4.65723604e-01 -4.47577387e-01 6.13426745e-01
6.89476252e-01 5.54371476e-01 7.11742103e-01 -9.28198457e-01
-7.87953958e-02 -6.50238991e-01 -5.91626108e-01 -8.78908992e-01
2.05466494e-01 -1.16078436e+00 -3.41853142e-01 -1.49517334e+00
4.90400851e-01 -5.30001998e-01 -9.06080544e-01 7.23072112e-01
-6.86465800e-01 3.22948128e-01 1.92734078e-01 1.62020922e-01
-9.49889004e-01 1.10187733e+00 1.17534292e+00 -2.74022400e-01
-5.08094728e-02 -7.24381268e-01 -1.14305484e+00 4.82896984e-01
3.28188509e-01 -2.13193029e-01 -4.92319822e-01 -4.24898803e-01
4.66702908e-01 -3.93495321e-01 4.45477515e-01 -7.60850608e-01
3.75599474e-01 -6.58839718e-02 4.75451827e-01 -7.02094972e-01
2.99799770e-01 -1.07550728e+00 -1.82435736e-02 -1.90999895e-01
-1.00573860e-01 1.31600320e-01 2.08419293e-01 9.27573919e-01
-5.22417843e-01 2.04211950e-01 4.50901657e-01 2.25498646e-01
-1.14449108e+00 4.83670324e-01 3.87761682e-01 2.66672611e-01
9.68893468e-01 -2.12781250e-01 -5.86782336e-01 1.46781430e-01
-6.05123937e-01 6.64591253e-01 5.30314326e-01 7.05754042e-01
6.48796082e-01 -1.99373984e+00 -5.33576667e-01 1.22132726e-01
8.02958012e-01 3.86860073e-02 6.94667757e-01 1.03804219e+00
5.70038408e-02 1.01857342e-01 -2.48130769e-01 -4.67703789e-01
-9.12108958e-01 7.28129745e-01 1.09173439e-01 -4.44061279e-01
-3.76300484e-01 8.27751875e-01 2.89132655e-01 -6.38090491e-01
-3.35739441e-02 1.50197506e-01 -4.34068978e-01 6.60323203e-01
2.93512702e-01 1.75440148e-01 -4.04593088e-02 -1.16539407e+00
-6.40038311e-01 8.33777547e-01 -2.41316721e-01 3.13292235e-01
1.21330631e+00 -5.17511368e-01 -3.78134325e-02 4.73646164e-01
1.19299364e+00 2.89553493e-01 -9.10165787e-01 -7.50400543e-01
-2.58179177e-02 -4.26399171e-01 -6.93212673e-02 -7.83786774e-01
-1.30952680e+00 6.33683026e-01 3.97984207e-01 6.75138906e-02
1.33898354e+00 1.85548633e-01 9.21832085e-01 1.76537141e-01
1.58331335e-01 -8.90679479e-01 2.61415988e-01 2.49878675e-01
7.08744287e-01 -1.41202974e+00 -6.43687695e-02 -4.05504555e-01
-9.30576682e-01 8.71569395e-01 9.66612637e-01 1.13099486e-01
5.42208672e-01 -3.53333235e-01 -3.93761322e-02 -4.94995862e-01
-3.51382494e-01 -5.50659895e-01 9.85590756e-01 4.71405864e-01
8.33991766e-02 5.57950772e-02 9.61022601e-02 9.79907930e-01
2.30782211e-01 -3.94390970e-01 2.59825643e-02 8.17485750e-01
-2.17423171e-01 -1.01085925e+00 -9.76459309e-02 2.42740840e-01
-2.12550908e-01 -3.80035669e-01 -2.73220539e-01 8.33922207e-01
2.93560147e-01 4.66149539e-01 -2.55432546e-01 -5.90195537e-01
4.11364406e-01 1.19787924e-01 3.99778694e-01 -2.67111272e-01
-4.81615007e-01 -1.19547240e-01 -1.90532748e-02 -3.99856120e-01
-7.53739059e-01 -4.93137002e-01 -1.16029382e+00 -1.75552577e-01
-2.83128113e-01 -2.60875762e-01 9.97556895e-02 8.97209644e-01
7.50169754e-01 5.97776711e-01 6.34555399e-01 -6.97391272e-01
-2.08426073e-01 -7.40036666e-01 -5.20425856e-01 6.49310231e-01
1.27822846e-01 -8.69857490e-01 -3.13216597e-01 -1.91269085e-01] | [8.698372840881348, 7.719961643218994] |
d0b04baa-19ee-495c-be6d-bad30a0d2c03 | end-to-end-spatio-temporal-action | 2304.12160 | null | https://arxiv.org/abs/2304.12160v1 | https://arxiv.org/pdf/2304.12160v1.pdf | End-to-End Spatio-Temporal Action Localisation with Video Transformers | The most performant spatio-temporal action localisation models use external person proposals and complex external memory banks. We propose a fully end-to-end, purely-transformer based model that directly ingests an input video, and outputs tubelets -- a sequence of bounding boxes and the action classes at each frame. Our flexible model can be trained with either sparse bounding-box supervision on individual frames, or full tubelet annotations. And in both cases, it predicts coherent tubelets as the output. Moreover, our end-to-end model requires no additional pre-processing in the form of proposals, or post-processing in terms of non-maximal suppression. We perform extensive ablation experiments, and significantly advance the state-of-the-art results on four different spatio-temporal action localisation benchmarks with both sparse keyframes and full tubelet annotations. | ['Anurag Arnab', 'Cordelia Schmid', 'Mario Lučić', 'Chen Sun', 'Mostafa Dehghani', 'Josip Djolonga', 'Xuehan Xiong', 'Alexey Gritsenko'] | 2023-04-24 | null | null | null | null | ['action-recognition-in-videos', 'spatio-temporal-action-localization'] | ['computer-vision', 'computer-vision'] | [ 3.14143032e-01 1.08686879e-01 -2.23882347e-01 -3.43138397e-01
-8.19466889e-01 -3.23451579e-01 7.94998705e-01 -2.95701712e-01
-7.36637890e-01 5.71952999e-01 7.53541827e-01 1.74627349e-01
3.27783287e-01 -4.16692555e-01 -6.78198159e-01 -4.38190728e-01
-1.80706635e-01 5.42008758e-01 6.47685111e-01 -1.08718939e-01
-7.11937025e-02 -3.92792746e-02 -1.34758389e+00 9.70884204e-01
2.85088241e-01 1.26332164e+00 -1.00576309e-02 8.03696752e-01
3.06999117e-01 1.08915198e+00 -1.81316316e-01 -5.78622401e-01
5.58424115e-01 -5.25836349e-01 -7.68890798e-01 2.96830863e-01
5.97346902e-01 -5.09748399e-01 -8.18128824e-01 5.23741841e-01
6.00632250e-01 2.96578288e-01 3.25764239e-01 -1.04756916e+00
-3.29207629e-01 2.35560834e-01 -4.90890592e-01 3.01632196e-01
6.55819654e-01 7.02097237e-01 9.47147608e-01 -1.19634604e+00
7.72905946e-01 1.36463368e+00 9.81323004e-01 4.59191918e-01
-1.25484467e+00 -3.32846135e-01 3.79805624e-01 3.00129145e-01
-1.12766910e+00 -7.98099458e-01 4.41014856e-01 -3.67594421e-01
1.53203356e+00 4.50899033e-03 9.29869652e-01 1.39484298e+00
2.70188916e-02 1.05787802e+00 8.95727038e-01 -1.88861355e-01
1.09912969e-01 -4.59991276e-01 -3.63734424e-01 9.43844378e-01
-5.46860516e-01 -2.66098119e-02 -8.35410178e-01 -8.52246881e-02
1.11681783e+00 1.93647981e-01 -2.86960214e-01 -6.03819907e-01
-1.61690998e+00 4.34830517e-01 3.92655075e-01 2.29060367e-01
-4.97829229e-01 4.75057065e-01 6.50048256e-01 1.88024789e-01
6.12266183e-01 7.01940507e-02 -5.79284370e-01 -4.36530262e-01
-1.19187737e+00 3.29423279e-01 3.80910665e-01 7.54868686e-01
5.24935901e-01 -3.39200944e-01 -7.24253476e-01 4.33973134e-01
-1.07688993e-01 3.45209956e-01 4.10764396e-01 -1.20193458e+00
8.79209816e-01 5.98927915e-01 2.78491765e-01 -6.42944813e-01
-5.27146220e-01 -2.16707900e-01 -7.15795934e-01 -2.23014969e-02
5.34254491e-01 3.45541835e-02 -1.20658016e+00 1.68279088e+00
3.33224803e-01 5.93816161e-01 -2.79514670e-01 1.12722671e+00
6.06266201e-01 3.75785917e-01 2.18053594e-01 -6.76731318e-02
1.31530619e+00 -1.46060359e+00 -5.54533124e-01 -6.07453644e-01
8.58069837e-01 -4.06041265e-01 1.08781838e+00 2.17307508e-01
-1.34211445e+00 -7.38602281e-01 -5.19222617e-01 -4.36262161e-01
-4.60415743e-02 6.48500025e-01 4.79580373e-01 1.27745852e-01
-1.03758061e+00 6.57362103e-01 -1.26046562e+00 -5.01629233e-01
6.33924186e-01 3.49077940e-01 -8.50393951e-01 -1.25740036e-01
-8.26551735e-01 6.53234422e-01 2.28114530e-01 8.52423683e-02
-9.64063585e-01 -5.54524720e-01 -1.10761189e+00 6.17459463e-03
7.03681707e-01 -9.98915553e-01 1.31633425e+00 -1.01081765e+00
-1.36394250e+00 9.08626616e-01 -4.43089694e-01 -6.25475585e-01
8.44010115e-01 -6.03275299e-01 -3.22161452e-03 3.23876351e-01
2.92163283e-01 1.01976347e+00 1.04478717e+00 -4.39547598e-01
-6.54273689e-01 -4.36344236e-01 1.55357257e-01 2.43413299e-01
-1.99475750e-01 3.15820634e-01 -8.24992597e-01 -8.35166037e-01
1.75678656e-01 -9.25291359e-01 -3.73543978e-01 3.21829408e-01
-1.16613135e-01 -3.10253173e-01 6.19279087e-01 -6.71078503e-01
9.51206744e-01 -2.14845490e+00 3.52909297e-01 -1.76269516e-01
6.40565827e-02 -5.98027818e-02 -3.09838772e-01 2.04267517e-01
-1.05915777e-01 -3.21854889e-01 -2.01449454e-01 -9.08383846e-01
1.93051532e-01 5.85635714e-02 -3.99970740e-01 7.02517807e-01
4.10851002e-01 1.13289964e+00 -9.52629268e-01 -6.42530203e-01
4.81511354e-01 4.93892550e-01 -7.98399627e-01 1.58215702e-01
-2.79154480e-01 6.36735082e-01 -2.86511362e-01 5.64555407e-01
4.80991304e-02 -3.04455191e-01 6.63586184e-02 -4.02656168e-01
3.32794990e-03 5.72447836e-01 -1.12504208e+00 2.34377193e+00
-2.31746361e-01 5.54988325e-01 -1.38089135e-01 -8.61970067e-01
3.29845071e-01 5.08284092e-01 7.17286587e-01 -7.91064262e-01
-4.21947241e-03 3.86612676e-02 -4.44155902e-01 -3.87712210e-01
3.18848759e-01 -2.51141917e-02 -1.90750435e-01 5.23343205e-01
4.12478060e-01 4.97479022e-01 2.63531804e-01 9.97610986e-02
1.46102548e+00 8.77060294e-01 2.34504327e-01 2.31989816e-01
4.43580270e-01 -1.70846537e-01 6.99666202e-01 6.46420658e-01
-2.40862995e-01 9.10868645e-01 5.39336383e-01 -8.55516195e-01
-1.14705133e+00 -7.85631776e-01 4.44027781e-01 1.50821507e+00
-1.57504037e-01 -8.88823390e-01 -5.86491048e-01 -1.06705558e+00
-3.02394271e-01 2.02962786e-01 -7.16799796e-01 1.37695119e-01
-8.84487748e-01 -8.02094787e-02 6.43048465e-01 9.60932970e-01
6.90037549e-01 -1.34861910e+00 -1.00665712e+00 2.32907191e-01
-5.98596990e-01 -1.44259930e+00 -7.66022801e-01 7.43077770e-02
-9.01660681e-01 -1.09152615e+00 -9.66708064e-01 -5.03559053e-01
5.77648699e-01 3.57691534e-02 1.24751139e+00 -3.60928103e-02
-2.20915347e-01 2.43081987e-01 -3.35495472e-01 1.81370214e-01
3.29504251e-01 -1.67413317e-02 -7.28031546e-02 1.74135745e-01
2.99323201e-01 -4.87915039e-01 -8.06373537e-01 5.67280948e-01
-3.47157925e-01 2.71610588e-01 5.51330924e-01 6.91501200e-01
6.78348124e-01 -2.89794534e-01 -1.67559478e-02 -3.51242036e-01
-2.92286985e-02 9.36567187e-02 -3.41983080e-01 2.45436728e-01
2.73607671e-01 6.39613420e-02 3.65881026e-01 -3.25990111e-01
-8.85225058e-01 5.84572852e-01 -5.18712774e-02 -8.35779965e-01
-2.95228839e-01 1.41258165e-01 2.54999455e-02 2.47988805e-01
8.06180835e-01 2.32665479e-01 -2.76875824e-01 -5.48101485e-01
5.51573396e-01 1.86468989e-01 9.21926022e-01 -2.86302537e-01
6.51410460e-01 8.37863326e-01 -2.31833518e-01 -3.82250458e-01
-1.03601170e+00 -7.05206871e-01 -1.18997622e+00 -1.72997922e-01
1.12619364e+00 -1.21770120e+00 -6.58716381e-01 4.46850449e-01
-1.13491595e+00 -8.73475552e-01 -4.54523265e-01 4.71897304e-01
-1.00087488e+00 3.71272415e-01 -5.90115011e-01 -7.15126574e-01
-2.36807957e-01 -9.04589295e-01 1.65363431e+00 -2.98953503e-01
-4.13180143e-01 -5.43884814e-01 7.04205856e-02 4.76303279e-01
8.19649845e-02 1.26289368e-01 1.60191044e-01 -2.33976379e-01
-6.00777984e-01 -2.69397587e-01 -1.11004040e-01 -5.21268398e-02
-1.54359847e-01 -6.20323062e-01 -9.83809888e-01 -4.46669281e-01
-2.26171628e-01 -7.54449546e-01 1.44281912e+00 3.99253309e-01
9.87855673e-01 -4.09456372e-01 -5.21222115e-01 7.12156534e-01
7.80017793e-01 -4.28076625e-01 7.75396764e-01 2.71576941e-01
6.34479702e-01 5.32274723e-01 8.30848098e-01 6.06240869e-01
3.24483186e-01 1.06165922e+00 2.62966812e-01 -2.75623024e-01
-1.85376391e-01 -3.90296936e-01 7.26977706e-01 -9.25675035e-02
-4.90603328e-01 -1.88221380e-01 -7.13562489e-01 6.56768084e-01
-2.37014604e+00 -1.55135894e+00 3.24221849e-01 2.07216573e+00
6.71413779e-01 1.79893732e-01 6.91679060e-01 9.00019035e-02
4.43001777e-01 5.68026066e-01 -5.07017136e-01 3.43448579e-01
3.18107232e-02 3.23180377e-01 4.45256710e-01 2.83315301e-01
-1.72070599e+00 1.33120191e+00 6.21820354e+00 4.40296441e-01
-8.76300395e-01 4.25616056e-01 4.87109154e-01 -6.79806054e-01
3.06320190e-01 -1.99949164e-02 -4.56717640e-01 1.64294094e-01
6.25024319e-01 4.80546415e-01 2.44105071e-01 8.08653355e-01
3.16076398e-01 -1.31338269e-01 -1.24681354e+00 1.19048572e+00
1.17546059e-01 -1.37564838e+00 -3.29175383e-01 -8.01448599e-02
4.61294830e-01 2.61784881e-01 -1.50472954e-01 3.58000875e-01
1.61937758e-01 -9.65628445e-01 1.07906735e+00 5.85589409e-01
9.40558314e-01 -4.16312695e-01 6.09277666e-01 3.07208270e-01
-1.63827002e+00 -4.28361148e-01 -2.36550704e-01 -4.09259140e-01
4.79470015e-01 1.13979563e-01 -4.75086033e-01 2.61329025e-01
8.73144209e-01 1.13845670e+00 -6.37896001e-01 1.00357974e+00
-3.71772200e-01 5.17860293e-01 -5.08603394e-01 4.99690831e-01
4.23131287e-01 1.24766901e-01 3.36922675e-01 1.38446891e+00
2.88497955e-01 4.42331940e-01 6.58286631e-01 5.51115692e-01
1.58525839e-01 -2.51435846e-01 -3.82977992e-01 9.56750214e-02
1.51530737e-02 1.04721236e+00 -6.75135434e-01 -5.16895711e-01
-6.15133166e-01 1.46589184e+00 4.51922983e-01 3.54855448e-01
-9.78849173e-01 1.59983024e-01 6.99432075e-01 5.20728707e-01
6.46500885e-01 -3.42215180e-01 -5.96519969e-02 -1.31467593e+00
1.98520482e-01 -8.77262473e-01 6.12352550e-01 -8.64234030e-01
-9.43980753e-01 3.85523230e-01 -2.73511440e-01 -1.28469801e+00
-4.86369729e-01 -3.82221311e-01 -5.69724321e-01 5.42492688e-01
-1.01492417e+00 -1.52029955e+00 -3.24382812e-01 1.07016325e+00
6.59748495e-01 -4.45734300e-02 7.35887766e-01 2.28284508e-01
-4.96459037e-01 5.09829342e-01 -5.44041336e-01 4.90022629e-01
8.47950041e-01 -1.01655042e+00 8.37838471e-01 9.53137696e-01
5.18320501e-01 2.97443360e-01 5.46921551e-01 -7.10390210e-01
-1.00211239e+00 -1.24754798e+00 1.09102345e+00 -8.01232159e-01
4.19182926e-01 -6.81979239e-01 -4.89548206e-01 1.02003527e+00
4.01489027e-02 5.72216570e-01 2.66330063e-01 2.39900962e-01
-2.18756497e-01 1.58355206e-01 -6.93320096e-01 6.49314106e-01
1.71723711e+00 -7.75053501e-01 -6.57827139e-01 5.57625413e-01
4.29097831e-01 -6.72167659e-01 -6.27092957e-01 2.78571784e-01
6.71501160e-01 -1.23479056e+00 1.17634881e+00 -8.35252821e-01
3.70814800e-01 -3.19961995e-01 2.49218885e-02 -6.90429091e-01
-6.76869929e-01 -7.53567696e-01 -2.88127869e-01 6.71891868e-01
2.57774085e-01 8.78784731e-02 1.08079350e+00 6.79575086e-01
-2.24399462e-01 -7.83572912e-01 -1.20807767e+00 -6.20048761e-01
-6.88672960e-01 -7.63293803e-01 1.84008732e-01 5.04313111e-01
1.13405071e-01 2.37789914e-01 -8.65874350e-01 -3.58801447e-02
3.47156346e-01 -3.88559476e-02 1.12577784e+00 -8.74450743e-01
-5.62903047e-01 -1.71553671e-01 -6.64063573e-01 -1.38075936e+00
2.45714903e-01 -5.22636831e-01 2.10373923e-01 -1.57888389e+00
1.55118480e-01 5.11637628e-02 -3.01730216e-01 1.16733551e+00
-1.07768260e-01 6.78435504e-01 2.75747985e-01 3.17980886e-01
-1.29624987e+00 8.18950117e-01 9.43674147e-01 -7.00642765e-02
-2.30823651e-01 -1.93865765e-02 -1.00452214e-01 1.06057906e+00
3.78956139e-01 -5.21767557e-01 -4.21983153e-01 -4.41403419e-01
8.14141929e-02 6.93204328e-02 1.03615522e+00 -1.33248603e+00
2.41262197e-01 -2.78441012e-02 7.41282225e-01 -6.70373380e-01
8.01517248e-01 -5.66750586e-01 -1.57557115e-01 2.84660995e-01
-4.36703265e-01 1.21780992e-01 -1.08894855e-01 6.67705894e-01
8.37321952e-03 4.41022664e-01 5.93746662e-01 -4.19839382e-01
-8.13793123e-01 5.10929048e-01 -3.71262766e-02 1.90430418e-01
9.68935966e-01 -5.50142586e-01 -9.65087116e-02 -4.08230633e-01
-1.15509689e+00 1.49664789e-01 5.97614348e-01 4.11505222e-01
5.50345123e-01 -1.45464694e+00 -6.34151578e-01 1.34599403e-01
3.60915326e-02 -1.45898655e-01 2.75642753e-01 1.27634895e+00
-6.64595608e-03 5.83956182e-01 -1.85625911e-01 -7.32060432e-01
-1.35170996e+00 5.16162455e-01 4.34246778e-01 -5.45755327e-01
-1.10903382e+00 1.01113808e+00 5.15447378e-01 -2.48158544e-01
4.53119159e-01 -4.62326586e-01 8.81240293e-02 -1.46848947e-01
6.87354267e-01 1.44702151e-01 -2.21975058e-01 -8.34935904e-01
-5.44009984e-01 5.76671839e-01 2.33980536e-01 -4.84719753e-01
1.28975892e+00 8.17994475e-02 2.01746166e-01 1.83577031e-01
8.44371378e-01 -4.88835663e-01 -1.91460276e+00 -3.86053652e-01
-6.23338930e-02 -6.93610549e-01 -1.05414622e-01 -6.55533850e-01
-9.30453837e-01 7.58509457e-01 4.14325237e-01 -3.73858511e-01
1.14345157e+00 2.14587256e-01 9.25905526e-01 5.65208793e-01
4.34285671e-01 -1.25473905e+00 4.89349484e-01 5.16095340e-01
1.08044910e+00 -1.09698570e+00 -1.09437192e-02 -2.23403379e-01
-7.15801001e-01 8.18035841e-01 6.71762943e-01 -1.27643228e-01
3.11819375e-01 1.96577296e-01 -1.19671464e-01 -2.30485961e-01
-9.84370232e-01 -4.90668416e-01 4.18799013e-01 5.31021535e-01
5.11106610e-01 -3.17454815e-01 -1.11778255e-03 5.28605342e-01
-4.36541699e-02 4.03222680e-01 -1.35588452e-01 9.07819331e-01
-2.75265664e-01 -8.28088105e-01 -2.05376685e-01 2.38698289e-01
-2.46768683e-01 -1.43913373e-01 -5.48944890e-01 5.34822106e-01
4.86977249e-01 7.98358858e-01 1.93041787e-01 -3.44094247e-01
5.11668503e-01 3.12555730e-01 6.40270829e-01 -5.98523080e-01
-6.45023167e-01 2.86363721e-01 2.17391387e-01 -1.37215436e+00
-6.01872802e-01 -8.17773044e-01 -1.33834505e+00 -1.14333816e-01
-1.16945237e-01 -2.35199943e-01 -8.21922794e-02 1.21266413e+00
5.70154130e-01 2.84904957e-01 -9.17913020e-02 -1.36449075e+00
-1.52080894e-01 -1.15260303e+00 6.20693993e-03 5.62521100e-01
2.19269559e-01 -7.27357626e-01 1.96739420e-01 3.10162306e-01] | [8.34424114227295, 0.4236794412136078] |
17aad19b-146f-402f-a221-4ffc6e799fe0 | long-range-feature-propagating-for-natural | 2109.12252 | null | https://arxiv.org/abs/2109.12252v1 | https://arxiv.org/pdf/2109.12252v1.pdf | Long-Range Feature Propagating for Natural Image Matting | Natural image matting estimates the alpha values of unknown regions in the trimap. Recently, deep learning based methods propagate the alpha values from the known regions to unknown regions according to the similarity between them. However, we find that more than 50\% pixels in the unknown regions cannot be correlated to pixels in known regions due to the limitation of small effective reception fields of common convolutional neural networks, which leads to inaccurate estimation when the pixels in the unknown regions cannot be inferred only with pixels in the reception fields. To solve this problem, we propose Long-Range Feature Propagating Network (LFPNet), which learns the long-range context features outside the reception fields for alpha matte estimation. Specifically, we first design the propagating module which extracts the context features from the downsampled image. Then, we present Center-Surround Pyramid Pooling (CSPP) that explicitly propagates the context features from the surrounding context image patch to the inner center image patch. Finally, we use the matting module which takes the image, trimap and context features to estimate the alpha matte. Experimental results demonstrate that the proposed method performs favorably against the state-of-the-art methods on the AlphaMatting and Adobe Image Matting datasets. | ['Rongrong Ji', 'Bineng Zhong', 'Shengping Zhang', 'Haozhe Xie', 'Qinglin Liu'] | 2021-09-25 | null | null | null | null | ['image-matting'] | ['computer-vision'] | [ 2.53762513e-01 -1.32515863e-01 3.57421488e-02 -6.28406882e-01
-5.57317376e-01 -1.40299127e-01 2.30705321e-01 -2.08574489e-01
-3.83086860e-01 8.39881241e-01 1.38426587e-01 2.06454098e-01
1.29769355e-01 -1.12957275e+00 -1.17209351e+00 -9.14023697e-01
1.11381583e-01 -5.23016416e-02 4.03240621e-01 2.15886394e-03
3.14072490e-01 2.69224107e-01 -1.41966701e+00 1.11806834e+00
7.73104250e-01 1.40572631e+00 2.89092511e-01 4.00063992e-01
-3.41894269e-01 9.96412814e-01 -6.72327638e-01 -1.57916605e-01
3.16159368e-01 -2.81883627e-01 -4.89233643e-01 5.44818640e-02
9.93724227e-01 -5.84212065e-01 -5.50180614e-01 1.26932335e+00
-2.59707458e-02 -2.93011546e-01 4.39681441e-01 -1.21996093e+00
-8.28922212e-01 8.59991133e-01 -1.17014408e+00 3.39028239e-01
9.22700018e-02 2.99790531e-01 7.10583389e-01 -1.05118382e+00
3.75890523e-01 1.09652746e+00 6.32867098e-01 -1.51722714e-01
-8.35563242e-01 -8.97333980e-01 2.96737194e-01 4.95212674e-01
-1.69818938e+00 -5.15758023e-02 7.23739982e-01 -2.44045898e-01
6.38519824e-01 1.22006528e-01 7.68830419e-01 6.35448873e-01
5.86353779e-01 8.01818609e-01 1.34471154e+00 -1.33794963e-01
8.22554007e-02 2.90881153e-02 -3.11106294e-02 8.56439114e-01
1.99649520e-02 -1.41791686e-01 -7.90814340e-01 1.24957345e-01
1.09414279e+00 -5.54844853e-04 -3.35395128e-01 4.32219468e-02
-1.67584324e+00 4.77072626e-01 1.16685081e+00 3.41076612e-01
-5.85391402e-01 3.35902959e-01 -2.11464033e-01 2.32362866e-01
3.71038496e-01 1.85105443e-01 -3.64796996e-01 2.48106733e-01
-1.15597367e+00 9.53432322e-02 3.88734311e-01 1.13163102e+00
1.49182022e+00 -5.36242463e-02 -1.73597768e-01 5.21579087e-01
2.64393449e-01 5.07063150e-01 1.74414456e-01 -9.87041771e-01
8.12672973e-01 9.53206480e-01 -1.35889323e-02 -1.17814529e+00
-1.71329193e-02 -4.75325435e-01 -9.84350681e-01 3.50047559e-01
6.30944431e-01 -8.61729756e-02 -1.24148905e+00 1.41914701e+00
2.32742563e-01 5.36241949e-01 -1.87585860e-01 1.04357195e+00
7.35030413e-01 1.00199020e+00 -2.51520663e-01 -9.37473923e-02
1.31530368e+00 -9.91631031e-01 -6.70191050e-01 -5.26879907e-01
1.71940118e-01 -9.45563138e-01 9.69016910e-01 4.17067617e-01
-1.07843947e+00 -8.13246489e-01 -1.47302258e+00 -2.36186564e-01
-4.72693533e-01 3.29093456e-01 4.97636914e-01 2.39825994e-01
-1.11389995e+00 4.19005632e-01 -5.25929570e-01 2.28748381e-01
7.23178804e-01 3.51358771e-01 -3.29719186e-01 -2.44585112e-01
-1.00331354e+00 6.54000998e-01 4.12076592e-01 6.29812598e-01
-9.54556763e-01 -8.26494515e-01 -8.94573987e-01 2.85381433e-02
8.77187327e-02 -7.06234097e-01 7.06301272e-01 -1.56262743e+00
-7.90855467e-01 5.58247864e-01 -2.67203122e-01 -4.20622796e-01
3.37617546e-01 -2.92143792e-01 -3.26743752e-01 2.09978551e-01
1.68147296e-01 1.00920784e+00 1.05752826e+00 -1.42393303e+00
-9.07710075e-01 -2.83227891e-01 8.44676197e-02 2.72198498e-01
-2.49185115e-01 -3.77293378e-01 -4.71511245e-01 -7.83727169e-01
5.22372365e-01 -5.41608155e-01 -1.23937957e-01 3.91142517e-01
-5.06104350e-01 3.18028778e-01 1.06895924e+00 -8.13587606e-01
1.12450886e+00 -2.35974121e+00 -9.77624953e-02 5.03638923e-01
2.13767126e-01 -1.15058884e-01 -3.00941728e-02 -2.74991572e-01
-1.18861981e-01 -3.68494958e-01 -5.04329264e-01 1.36484012e-01
-3.78572106e-01 3.17714036e-01 -5.64885497e-01 4.06488836e-01
1.74040839e-01 7.81453848e-01 -8.17565680e-01 -6.83809578e-01
2.58989364e-01 5.58593154e-01 -2.78559983e-01 1.88305840e-01
-2.28311166e-01 2.45230317e-01 -1.31048098e-01 7.03505695e-01
1.20790958e+00 -1.69273332e-01 -2.98860967e-01 -5.68680346e-01
-3.22695076e-01 -1.40488269e-02 -1.12198412e+00 1.76952720e+00
-4.16081667e-01 7.97612369e-01 1.16116991e-02 -6.53478086e-01
9.69919980e-01 -6.29104376e-02 2.66972035e-01 -5.24680197e-01
1.19585797e-01 6.07475489e-02 -8.78964178e-03 -4.42243725e-01
5.36668897e-01 1.92650303e-01 1.86335951e-01 2.95843929e-01
-3.97328809e-02 8.91780406e-02 -1.74537301e-01 2.35243529e-01
9.97504175e-01 -1.48738306e-02 -2.18278076e-02 -3.43954675e-02
4.65459406e-01 -5.25464416e-02 6.76610351e-01 6.86234295e-01
-1.83025301e-01 8.56366754e-01 3.85098755e-01 -6.52765870e-01
-9.78326619e-01 -1.45724320e+00 -9.29670110e-02 8.84214222e-01
4.46143776e-01 -2.83102900e-01 -8.10158730e-01 -6.17484093e-01
-8.10760856e-02 3.44298244e-01 -1.12420952e+00 5.06102555e-02
-7.96985567e-01 -6.14510596e-01 2.50695348e-01 8.15342546e-01
1.50154912e+00 -1.04061842e+00 -2.82585561e-01 3.66394892e-02
-3.05719793e-01 -1.04876566e+00 -5.51079035e-01 6.56816512e-02
-8.34667683e-01 -8.58774006e-01 -6.67570949e-01 -1.03540206e+00
1.02975845e+00 2.89037049e-01 8.86460364e-01 1.13418140e-01
-4.79260311e-02 -1.61752403e-01 -3.39945517e-02 -1.35463893e-01
-1.12091258e-01 -1.67251527e-01 -4.53331172e-01 3.03947121e-01
6.09446764e-01 -6.52502179e-01 -8.95569265e-01 4.48371291e-01
-7.53990829e-01 4.04070616e-01 1.14412105e+00 7.29718804e-01
8.66974235e-01 1.87870622e-01 3.01297575e-01 -9.10699069e-01
-6.91140369e-02 -4.80784535e-01 -3.56381476e-01 2.83290088e-01
-2.42713988e-01 1.04691282e-01 4.34387743e-01 -4.43793476e-01
-1.14563203e+00 3.14792961e-01 9.20534432e-02 -3.88784438e-01
-1.44443437e-01 3.81753474e-01 -5.10325670e-01 -4.93190400e-02
7.77037024e-01 3.63452137e-01 -2.05381811e-01 -1.45874023e-01
3.71647418e-01 5.60446322e-01 9.28960979e-01 -2.56462157e-01
1.00586534e+00 7.17899799e-01 -1.61172196e-01 -5.85550725e-01
-1.01822615e+00 -2.52853602e-01 -6.15707517e-01 -2.32230201e-01
9.85799670e-01 -1.21311629e+00 -2.60892689e-01 5.81336796e-01
-1.10838795e+00 -2.29459792e-01 -2.03201815e-01 3.73990715e-01
-4.99048084e-01 4.02218759e-01 -8.61334920e-01 -3.52208018e-01
-3.62850606e-01 -1.17132831e+00 8.94680679e-01 4.77823466e-01
1.15006931e-01 -5.30156255e-01 -4.64321285e-01 3.65286589e-01
3.69727254e-01 1.06833190e-01 9.31325078e-01 2.40700006e-01
-1.17046034e+00 5.79968952e-02 -6.94537461e-01 4.86257643e-01
1.21442497e-01 7.23032132e-02 -1.30683911e+00 -2.79298313e-02
-1.20993331e-02 1.60650060e-01 1.30223250e+00 6.80010855e-01
1.38736331e+00 -4.02350426e-01 -1.84000820e-01 8.55750144e-01
1.29666257e+00 -1.08069614e-01 1.21205425e+00 4.28041637e-01
8.79217386e-01 2.86967725e-01 9.02257800e-01 3.00115138e-01
2.72869200e-01 2.68801093e-01 6.02120817e-01 -2.21724123e-01
-4.49552745e-01 -2.19319493e-01 3.65977049e-01 5.59397042e-01
-6.78320751e-02 2.13867769e-01 -4.66068536e-01 6.14052176e-01
-1.78340435e+00 -8.93911064e-01 -1.47555813e-01 2.11391473e+00
1.16205251e+00 3.59318435e-01 -3.37332070e-01 -4.53748321e-03
9.07763243e-01 1.92504585e-01 -5.25072873e-01 -1.75830424e-01
-4.67005014e-01 1.04963072e-01 5.90990961e-01 4.12730306e-01
-1.16065204e+00 7.63451636e-01 6.01168489e+00 8.15240264e-01
-9.52076614e-01 -5.06210476e-02 9.33237910e-01 1.20450966e-01
-3.37844998e-01 1.22192182e-01 -7.55209923e-01 5.47409177e-01
5.31289935e-01 1.91919237e-01 4.25760657e-01 8.52167904e-01
-6.18722811e-02 -4.68271345e-01 -1.03573716e+00 9.88705575e-01
1.82137117e-01 -1.36549270e+00 1.70594826e-01 -1.56048745e-01
9.73621547e-01 -1.40614361e-01 3.71038884e-01 9.54486653e-02
-1.09653682e-01 -1.00559568e+00 7.14448631e-01 6.70554221e-01
8.20050001e-01 -8.48285496e-01 9.15128767e-01 1.05086416e-01
-1.32882535e+00 -1.57614395e-01 -8.32985222e-01 -1.38488278e-01
-2.53063262e-01 1.05708146e+00 -1.15029323e+00 1.79013565e-01
8.79280329e-01 6.38605475e-01 -7.24532187e-01 1.28085887e+00
-4.92239356e-01 5.24426877e-01 -2.70681888e-01 2.65018672e-01
1.84808478e-01 -1.82898983e-01 1.18539639e-01 1.07001805e+00
4.03093845e-01 -6.13597482e-02 -1.69994429e-01 1.20256519e+00
-3.07874024e-01 -2.05226943e-01 -3.94929230e-01 4.31779504e-01
4.44109410e-01 1.31280243e+00 -6.74028277e-01 -5.95463514e-01
-3.72824848e-01 9.83715653e-01 2.72041559e-01 6.07155740e-01
-1.00331175e+00 -5.29490292e-01 5.06377757e-01 3.32496673e-01
3.47053230e-01 -1.91972852e-01 -8.57556701e-01 -9.30218339e-01
2.41652757e-01 -6.19651258e-01 3.86398673e-01 -1.16957593e+00
-1.19441187e+00 6.20016336e-01 -5.88114560e-03 -1.47948432e+00
1.96800679e-01 -4.94335413e-01 -6.95285916e-01 8.76141965e-01
-1.46270823e+00 -1.42400050e+00 -8.92911017e-01 7.22122729e-01
5.19894481e-01 2.14859843e-02 3.66987139e-01 2.48076454e-01
-1.82933182e-01 5.71219623e-01 -1.90645114e-01 4.21565652e-01
9.06690955e-01 -1.18936312e+00 2.51181930e-01 1.13181758e+00
8.90519097e-02 5.48018217e-01 4.90561277e-01 -7.68619776e-01
-1.02920365e+00 -1.30984104e+00 5.67656875e-01 -2.83402771e-01
5.03210425e-01 -4.43950981e-01 -1.07499146e+00 8.32762837e-01
4.52600002e-01 3.65164191e-01 3.47822964e-01 -3.13009471e-01
-6.67269766e-01 -5.33412039e-01 -1.05049038e+00 5.29145122e-01
8.42933536e-01 -4.64300543e-01 -5.75697780e-01 1.06577478e-01
6.57118320e-01 -4.82566774e-01 -7.29884148e-01 2.91697919e-01
5.38842857e-01 -9.37070966e-01 9.07179832e-01 1.13110833e-01
7.59876072e-01 -9.68005478e-01 -2.02627569e-01 -1.21607816e+00
-2.31041223e-01 -3.89778428e-02 2.54348032e-02 1.14845681e+00
6.82721496e-01 -3.51333320e-01 1.08160639e+00 3.76316339e-01
-3.41713935e-01 -5.93423903e-01 -9.64803040e-01 -1.30428210e-01
-9.46881324e-02 -3.04913729e-01 6.61713719e-01 9.45343852e-01
-2.30213404e-01 -4.42509986e-02 -3.34505945e-01 6.59490347e-01
6.98179603e-01 5.25280774e-01 7.01869428e-01 -5.50018907e-01
-3.38891774e-01 5.68999164e-02 -6.56793714e-01 -1.27605045e+00
-1.75951451e-01 -4.91286874e-01 3.94615173e-01 -1.47871399e+00
3.05828393e-01 -5.75467050e-01 -3.74693215e-01 6.25041008e-01
-3.34576458e-01 7.23132730e-01 -1.42157286e-01 9.43560824e-02
-4.06170368e-01 3.86758000e-01 1.48684692e+00 -6.44577146e-01
5.47906905e-02 -1.21703930e-01 -4.59831208e-01 7.96369374e-01
7.22245336e-01 -3.29825759e-01 -3.57863396e-01 -7.26777911e-01
3.23653132e-01 -2.77488440e-01 3.89083713e-01 -1.46119630e+00
3.62522870e-01 -2.05785617e-01 1.34092951e+00 -1.01810336e+00
5.07778883e-01 -8.94616544e-01 2.74412110e-02 1.69456050e-01
-3.59236032e-01 3.20764482e-02 5.22092432e-02 4.63110209e-01
-2.40656227e-01 -1.45918086e-01 6.27533257e-01 -2.95044303e-01
-8.80515754e-01 3.29337895e-01 -2.05000579e-01 -2.49501169e-01
9.21139181e-01 -4.30985987e-01 -7.95759797e-01 -2.82607794e-01
-4.93309557e-01 -4.86461781e-02 4.25046831e-01 1.71754941e-01
1.17156196e+00 -1.26494753e+00 -6.25417709e-01 5.93013704e-01
2.49572657e-02 4.25572604e-01 4.57116514e-01 8.72312903e-01
-7.98640907e-01 -2.84967363e-01 -5.14922380e-01 -7.08759904e-01
-1.08540404e+00 6.10159993e-01 2.43801028e-01 2.68794358e-01
-5.60231507e-01 1.06910324e+00 5.81906140e-01 1.15325108e-01
7.91785419e-02 -6.59021854e-01 -3.82756032e-02 -2.29938969e-01
8.79594862e-01 -7.84887448e-02 -6.19063899e-02 -6.84990644e-01
-2.14384302e-01 5.86742520e-01 -4.78464901e-01 -1.80895235e-02
1.01730907e+00 -1.68374330e-01 -4.89388108e-01 3.79778802e-01
1.15582860e+00 -4.32699956e-02 -1.62668872e+00 -2.51255572e-01
-3.46075088e-01 -9.43055034e-01 2.53692836e-01 -7.51804411e-01
-1.42130876e+00 9.44360614e-01 8.08912218e-01 -2.99223483e-01
1.43023658e+00 -1.20176725e-01 8.25674772e-01 2.64873266e-01
2.47307032e-01 -1.07926285e+00 1.63664863e-01 2.08047614e-01
8.38393569e-01 -1.07682800e+00 1.85858667e-01 -6.94589019e-01
-5.63409328e-01 1.27863741e+00 1.16480041e+00 -2.90751040e-01
6.63869262e-01 6.13266945e-01 3.74338537e-01 -6.07840866e-02
-6.85486078e-01 4.04627137e-02 9.12490562e-02 6.17230117e-01
2.29500324e-01 -5.88984452e-02 2.50557184e-01 3.01896542e-01
-3.38075548e-01 -8.30493122e-02 6.78235590e-01 9.42480683e-01
-7.98171043e-01 -5.82294941e-01 -6.44774437e-01 5.95977783e-01
-1.70151666e-01 -2.70702213e-01 -2.18675032e-01 5.05359888e-01
6.24831200e-01 7.30486453e-01 3.92047584e-01 -6.74434781e-01
-5.47756115e-03 -3.61784965e-01 4.44752306e-01 -2.87916273e-01
-6.16435111e-01 5.28919026e-02 -1.49538204e-01 -4.95741814e-01
-4.32789773e-01 -2.56658316e-01 -1.40865672e+00 -3.47938806e-01
-2.91305006e-01 -7.92109296e-02 5.52228749e-01 8.03670406e-01
1.08465180e-01 4.03564453e-01 6.26482666e-01 -6.14043176e-01
9.83006880e-03 -1.06459403e+00 -6.32468522e-01 3.16613972e-01
3.15140754e-01 -4.25022095e-01 -4.07833457e-01 2.94718802e-01] | [10.666590690612793, -0.9101769924163818] |
0382f802-fd14-479c-8938-7fc2aa204e13 | human-alignment-of-neural-network | 2211.01201 | null | https://arxiv.org/abs/2211.01201v4 | https://arxiv.org/pdf/2211.01201v4.pdf | Human alignment of neural network representations | Today's computer vision models achieve human or near-human level performance across a wide variety of vision tasks. However, their architectures, data, and learning algorithms differ in numerous ways from those that give rise to human vision. In this paper, we investigate the factors that affect the alignment between the representations learned by neural networks and human mental representations inferred from behavioral responses. We find that model scale and architecture have essentially no effect on the alignment with human behavioral responses, whereas the training dataset and objective function both have a much larger impact. These findings are consistent across three datasets of human similarity judgments collected using two different tasks. Linear transformations of neural network representations learned from behavioral responses from one dataset substantially improve alignment with human similarity judgments on the other two datasets. In addition, we find that some human concepts such as food and animals are well-represented by neural networks whereas others such as royal or sports-related objects are not. Overall, although models trained on larger, more diverse datasets achieve better alignment with humans than models trained on ImageNet alone, our results indicate that scaling alone is unlikely to be sufficient to train neural networks with conceptual representations that match those used by humans. | ['Simon Kornblith', 'Robert A. Vandermeulen', 'Lorenz Linhardt', 'Jonas Dippel', 'Lukas Muttenthaler'] | 2022-11-02 | null | null | null | null | ['odd-one-out'] | ['reasoning'] | [ 6.82023540e-02 -2.32661784e-01 8.25936869e-02 -7.45111883e-01
1.00333236e-01 -5.30869246e-01 8.66340935e-01 4.17991221e-01
-9.15690184e-01 2.01801360e-01 2.41008699e-01 -2.03855410e-01
-1.22010812e-01 -6.16483688e-01 -5.44860959e-01 -1.97035626e-01
4.84385043e-01 3.95957112e-01 2.24297613e-01 -3.02255094e-01
5.04714131e-01 3.25260371e-01 -1.74519467e+00 3.82340908e-01
6.44318163e-01 8.82491231e-01 2.25495964e-01 5.17920554e-01
1.76974878e-01 1.03722632e+00 -3.87378842e-01 -3.44223350e-01
3.73062074e-01 -4.64721292e-01 -4.35874403e-01 3.07604894e-02
1.13249815e+00 -4.92199689e-01 -6.09816730e-01 8.78726125e-01
2.52009153e-01 4.01413351e-01 1.16666543e+00 -1.15109277e+00
-1.41051054e+00 4.03312653e-01 -3.94615442e-01 3.82201076e-01
1.75711617e-01 7.05511093e-01 9.04007912e-01 -7.75570095e-01
2.90510058e-01 1.61431289e+00 7.88711548e-01 3.98440182e-01
-1.58915734e+00 -6.57382429e-01 1.04621299e-01 2.04756871e-01
-1.09958100e+00 -4.81641024e-01 4.85118181e-01 -8.33161294e-01
1.12067032e+00 -2.31194392e-01 9.19851601e-01 9.01759684e-01
5.18880337e-02 5.98343499e-02 1.34811819e+00 -3.45786840e-01
1.18776061e-01 2.85665125e-01 3.47825795e-01 5.96653223e-01
7.12730527e-01 2.23313019e-01 -2.61142135e-01 -1.12521067e-01
9.44195151e-01 1.92185551e-01 -2.51569580e-02 -2.67515332e-01
-1.34055197e+00 9.95604575e-01 9.98538911e-01 2.66035527e-01
-6.92932665e-01 3.33133131e-01 1.88653767e-01 2.31256217e-01
-7.96917155e-02 9.88338351e-01 -1.70594648e-01 2.71415323e-01
-6.59562826e-01 2.56680638e-01 4.86478388e-01 4.76478189e-01
8.48540545e-01 4.14088458e-01 1.17033273e-01 1.10858309e+00
2.02608839e-01 6.33020759e-01 7.39391863e-01 -1.24894667e+00
8.30363184e-02 5.07011473e-01 1.14305124e-01 -1.51571679e+00
-6.28838837e-01 -2.70041436e-01 -3.75857025e-01 2.71507919e-01
8.21964085e-01 3.39434966e-02 -1.12917686e+00 2.03957105e+00
-2.67666161e-01 -4.76340801e-01 6.67359605e-02 1.14717889e+00
8.38817000e-01 5.16897798e-01 6.95256054e-01 2.79179960e-01
1.41994047e+00 -7.78091431e-01 -1.07902795e-01 -1.09068441e+00
3.64925802e-01 -7.29999065e-01 1.26237786e+00 5.86788617e-02
-1.11427951e+00 -1.10255134e+00 -1.15309083e+00 -1.57489315e-01
-4.86721456e-01 -1.71029791e-02 7.16930032e-01 4.56278592e-01
-1.20230055e+00 3.30286592e-01 -3.35015714e-01 -8.54004085e-01
4.57983851e-01 1.22782603e-01 -3.47115993e-01 -1.73824772e-01
-7.81575799e-01 1.56717038e+00 5.31915188e-01 -2.67503440e-01
-6.64124668e-01 -4.21848863e-01 -8.44350040e-01 1.12300888e-01
-1.38278352e-02 -8.62448692e-01 1.39854157e+00 -1.53999233e+00
-7.45929241e-01 1.19716489e+00 9.08751041e-02 -6.28242791e-01
-1.07752025e-01 1.86981589e-01 -2.88705558e-01 1.03821315e-01
3.83475982e-02 1.30693746e+00 6.80609882e-01 -1.26819575e+00
-5.27107537e-01 -4.61333662e-01 1.49462312e-01 2.81980932e-01
-4.09554958e-01 2.78162491e-02 -5.74419834e-02 -5.57724893e-01
1.19352147e-01 -1.08413649e+00 -4.17885818e-02 3.30127597e-01
3.72168899e-01 -3.31056178e-01 1.21269256e-01 -5.93816400e-01
7.43926466e-01 -2.10694790e+00 -6.65170699e-02 8.24656188e-02
3.56426448e-01 2.39434391e-01 -4.64732170e-01 2.61339486e-01
-9.87850577e-02 1.03351854e-01 2.32246183e-02 1.83164358e-01
1.11525029e-01 2.42100805e-01 -2.38825187e-01 3.47239196e-01
1.87255040e-01 1.01518619e+00 -7.60438085e-01 -2.14822128e-01
2.87637293e-01 2.19569221e-01 -6.47215366e-01 1.20613612e-01
5.69120571e-02 -3.32012475e-02 5.04973456e-02 2.47762114e-01
3.09063524e-01 -3.60387653e-01 1.51887223e-01 -2.01825768e-01
6.34654015e-02 2.12545902e-01 -4.70797211e-01 1.27172589e+00
-2.91704357e-01 1.08457375e+00 -3.81795913e-01 -9.09904182e-01
1.03997362e+00 -6.96039349e-02 -6.40325993e-02 -1.24999321e+00
1.79540757e-02 1.35489637e-02 9.48870778e-01 -5.84823847e-01
5.91277957e-01 -4.87050712e-01 2.60257274e-01 6.82904780e-01
1.47970291e-02 -4.75835711e-01 1.52077168e-01 -1.86200887e-02
7.36906111e-01 -1.82176694e-01 5.15676022e-01 -2.39568919e-01
-1.36568246e-03 3.05990726e-01 5.84380209e-01 9.54412162e-01
-3.86031657e-01 4.31238383e-01 1.97979838e-01 -7.50904977e-01
-1.36389577e+00 -1.14769959e+00 1.61645431e-02 1.33505905e+00
1.57953147e-02 -4.00559828e-02 -5.35028696e-01 -1.71755478e-01
1.77659094e-01 1.07973993e+00 -8.50813627e-01 -5.05018592e-01
-2.66203642e-01 -4.51638430e-01 4.82668936e-01 9.56625462e-01
2.81447381e-01 -1.25615191e+00 -1.10870183e+00 7.32927248e-02
1.73886374e-01 -1.14633012e+00 -1.76311225e-01 -6.88266084e-02
-8.72678399e-01 -1.02992952e+00 -4.08638656e-01 -9.37714696e-01
8.22189987e-01 7.34475493e-01 1.36192214e+00 3.01710993e-01
-3.72756124e-01 6.51658297e-01 -5.36831655e-02 -7.63510704e-01
-3.32971036e-01 -3.61507773e-01 1.96334481e-01 -3.25060487e-01
7.76749551e-01 -2.84228474e-01 -5.93822598e-01 3.85732085e-01
-8.59974563e-01 -8.27727690e-02 6.91017151e-01 7.08679020e-01
6.02125674e-02 -2.83987880e-01 6.07370496e-01 -4.93626684e-01
7.92834401e-01 -5.04820287e-01 -2.32162446e-01 1.93310410e-01
-7.10227191e-01 -1.08065046e-02 4.21276182e-01 -8.12721252e-01
-1.01480448e+00 -1.68259025e-01 4.71952140e-01 -3.51079106e-01
-4.50844526e-01 4.29421604e-01 6.58005953e-01 1.78506430e-02
1.21029747e+00 1.72825679e-02 3.91073227e-02 -1.32004783e-01
2.72616923e-01 4.53055054e-01 6.02985561e-01 -5.90988696e-01
6.97864473e-01 3.45192194e-01 -3.39234769e-01 -8.33049417e-01
-8.19267869e-01 -1.07045755e-01 -5.15587449e-01 -1.55694097e-01
1.12709486e+00 -1.08375776e+00 -5.58532178e-01 4.31094199e-01
-1.02696657e+00 -5.02525628e-01 -1.70031384e-01 9.73388016e-01
-4.87238079e-01 -6.85076863e-02 -4.83482093e-01 -6.00012660e-01
5.04571609e-02 -1.03345156e+00 4.70251352e-01 4.17749673e-01
-8.37755084e-01 -9.04402494e-01 3.03053800e-02 4.77119058e-01
7.19087899e-01 -3.61922085e-02 1.41890371e+00 -8.18436503e-01
-2.00538471e-01 -7.17906207e-02 -6.51410580e-01 3.32628548e-01
2.59511024e-01 2.70355996e-02 -9.45766211e-01 -1.78728506e-01
9.04066786e-02 -8.57632637e-01 1.07350719e+00 5.73015928e-01
8.41686308e-01 -2.13051611e-03 7.64021277e-02 4.50794190e-01
1.33576775e+00 4.54952359e-01 5.93357384e-01 2.64668971e-01
5.74423194e-01 1.00037885e+00 2.75494307e-01 3.44591737e-02
6.08940840e-01 4.83736515e-01 1.25945389e-01 -3.30697447e-02
-8.25777352e-02 -2.66440928e-01 3.27395439e-01 3.57500285e-01
-2.07508311e-01 3.77088748e-02 -1.28664780e+00 5.04470289e-01
-1.75434697e+00 -1.24569392e+00 1.00149773e-01 2.20687079e+00
5.88988304e-01 3.09874475e-01 2.97323257e-01 -3.67408901e-01
6.52814031e-01 8.27417001e-02 -8.67816091e-01 -6.38751805e-01
-5.78309298e-02 -7.34253079e-02 3.30696076e-01 1.51906416e-01
-7.17483401e-01 9.61073160e-01 7.69687414e+00 1.05173379e-01
-1.17901456e+00 -1.43998623e-01 5.36792338e-01 -2.43504450e-01
-1.73339024e-01 -9.25967172e-02 -2.47378856e-01 1.65849730e-01
1.09000838e+00 -2.43999779e-01 5.55204093e-01 7.72388220e-01
2.11786807e-01 -9.12479982e-02 -1.40224099e+00 1.21929741e+00
4.83580738e-01 -9.97087419e-01 2.51826882e-01 -4.65096012e-02
6.87235236e-01 1.73783512e-03 3.17710280e-01 4.59982455e-01
5.47994375e-01 -1.50371504e+00 9.51569140e-01 6.45721018e-01
4.43443626e-01 -3.59773099e-01 3.50426108e-01 2.91665465e-01
-7.54638970e-01 -2.90332705e-01 -8.18876207e-01 -6.08409405e-01
-3.08261245e-01 -4.94024977e-02 -8.28596175e-01 -5.61629295e-01
7.59597063e-01 6.03514016e-01 -1.08227861e+00 8.70085359e-01
-9.75341722e-02 4.84461755e-01 -1.03733733e-01 -9.15456489e-02
2.28676826e-01 -6.50975248e-03 -1.46195769e-01 1.05917656e+00
1.07576467e-01 3.01630199e-01 2.38030832e-02 1.13015604e+00
5.11018187e-02 2.43751705e-02 -9.22353864e-01 -3.59342754e-01
6.24403715e-01 9.39599037e-01 -6.98565543e-01 -4.66751307e-01
-8.59554708e-01 4.63517338e-01 5.23911893e-01 5.26491225e-01
-7.52862155e-01 -4.09365743e-02 8.61797571e-01 2.79333383e-01
1.55741245e-01 -3.98813039e-01 -4.57066953e-01 -9.07487750e-01
-2.78070182e-01 -1.09226871e+00 2.89894313e-01 -1.35307956e+00
-1.52886307e+00 4.03282762e-01 5.69845885e-02 -9.03852522e-01
-3.90927255e-01 -9.85180080e-01 -5.52813351e-01 8.93450081e-01
-9.26079690e-01 -9.43941295e-01 -5.15404284e-01 5.66868484e-01
4.92298365e-01 -2.71377951e-01 7.12011039e-01 -2.95964032e-01
-1.88410386e-01 3.71959567e-01 -3.42149138e-01 3.30148935e-01
1.02963769e+00 -8.43385816e-01 5.76348066e-01 5.19231319e-01
3.02468151e-01 1.12538683e+00 5.80581129e-01 -4.26117361e-01
-9.95160460e-01 -4.76106435e-01 3.70727986e-01 -6.04219079e-01
4.84376431e-01 2.21764371e-01 -1.00944006e+00 8.61553490e-01
4.70050246e-01 -2.78195709e-01 7.73884952e-01 2.15350553e-01
-1.03667474e+00 1.54038880e-03 -1.02642858e+00 7.45210409e-01
9.95790660e-01 -9.08795714e-01 -1.18550158e+00 5.71593273e-06
4.30028081e-01 2.44623587e-01 -6.88063502e-01 2.35653967e-01
7.55363584e-01 -1.05335486e+00 1.20991421e+00 -1.15660870e+00
6.26141667e-01 -6.11093640e-02 -5.03682017e-01 -1.65694094e+00
-9.93180990e-01 3.85290235e-01 3.20741236e-01 6.80609524e-01
3.66647780e-01 -8.47653508e-01 3.91790688e-01 1.02882552e+00
6.62597641e-02 -3.05815011e-01 -2.44970798e-01 -6.34238422e-01
2.66514808e-01 -1.91641659e-01 3.87977004e-01 1.09851491e+00
-2.82823294e-01 6.23416603e-01 9.77048948e-02 -2.35225514e-01
7.18137085e-01 2.98775956e-02 6.58579588e-01 -1.51841378e+00
-2.53713936e-01 -8.13319802e-01 -6.48936093e-01 -6.29929721e-01
1.17839709e-01 -7.50017583e-01 1.13307405e-02 -1.65035677e+00
4.49917465e-01 -2.29513690e-01 -2.68698156e-01 5.15689433e-01
-1.42210767e-01 2.71563411e-01 6.52538955e-01 3.91731322e-01
-6.86000437e-02 1.49234891e-01 1.00270057e+00 -1.83923513e-01
1.02961503e-01 -5.01836479e-01 -1.25549650e+00 1.19692802e+00
9.26608860e-01 -1.75846919e-01 -6.57606602e-01 -9.49661136e-01
9.51707065e-02 -2.35787690e-01 6.20642781e-01 -1.41696215e+00
3.13904613e-01 -5.08812904e-01 1.03013074e+00 8.14303532e-02
5.34373641e-01 -6.64228320e-01 -1.29721195e-01 6.66679323e-01
-7.52544165e-01 5.40063620e-01 4.54280287e-01 4.71271276e-01
7.54344463e-02 -2.57939041e-01 1.13382912e+00 -3.18785608e-01
-1.10280359e+00 -1.07564576e-01 -5.15972197e-01 2.67311424e-01
7.55221963e-01 -4.88812506e-01 -6.92401052e-01 -5.81742942e-01
-4.66715425e-01 -1.18175060e-01 7.98234463e-01 7.01134384e-01
6.83782101e-01 -1.29555678e+00 -7.86432624e-01 3.66087779e-02
3.54184508e-01 -6.42509639e-01 -8.87382403e-02 4.85375106e-01
-4.57789421e-01 4.91924286e-01 -8.61541331e-01 -4.49299634e-01
-1.02047205e+00 6.33112609e-01 4.44654256e-01 4.63469088e-01
-1.24507464e-01 6.22094274e-01 6.61907017e-01 -3.50933522e-01
-4.45202962e-02 -4.24608380e-01 -1.64067164e-01 1.52726308e-01
5.68978548e-01 1.75073475e-01 -5.13470829e-01 -8.87066007e-01
-1.53794646e-01 6.93165839e-01 -2.99773365e-01 -1.71795368e-01
1.31982100e+00 1.41748801e-01 1.16703503e-01 5.82942665e-01
7.70359159e-01 -5.43920219e-01 -1.43880653e+00 -4.22724873e-01
-3.52019697e-01 -5.35125077e-01 -7.62646496e-02 -8.25066447e-01
-8.40485692e-01 1.00475550e+00 5.89733064e-01 3.44589874e-02
8.82346988e-01 -2.31743581e-03 3.52745771e-01 8.32576931e-01
3.30165595e-01 -1.23467612e+00 4.82928544e-01 4.44116056e-01
9.95941401e-01 -1.47482145e+00 1.22739971e-01 1.92000940e-01
-9.82109666e-01 8.81952643e-01 1.17345548e+00 -2.77884662e-01
2.50041127e-01 -3.67439061e-01 6.08672202e-02 -2.80748338e-01
-9.11122620e-01 -3.22719932e-01 3.52395177e-01 7.85587251e-01
5.99920988e-01 1.69121578e-01 3.36544439e-02 2.81411409e-01
-3.61119717e-01 -1.67803153e-01 4.12795722e-01 6.55308545e-01
-9.24184442e-01 -3.70327801e-01 -5.50924063e-01 6.56238675e-01
5.79607189e-02 -2.03047886e-01 -8.11844349e-01 8.93027365e-01
1.53179318e-01 9.77102816e-01 6.67978644e-01 -4.88453448e-01
3.86915237e-01 2.53825873e-01 7.63310611e-01 -8.77874255e-01
-5.43409824e-01 -4.46721017e-01 6.17347769e-02 -4.32929784e-01
-3.96325737e-01 -6.12959385e-01 -1.27792668e+00 -5.55390298e-01
1.57987148e-01 -4.16737884e-01 5.53717136e-01 8.03755939e-01
7.20547587e-02 1.90764964e-01 7.61034638e-02 -8.67398858e-01
-7.44365096e-01 -1.17657030e+00 -3.78924042e-01 8.86438429e-01
1.66837021e-03 -7.33287990e-01 -2.83132166e-01 1.99973360e-01] | [10.404619216918945, 2.3119213581085205] |
7281aa87-002f-4a30-bb27-48852ce3b941 | membership-privacy-protection-for-image | 2203.05212 | null | https://arxiv.org/abs/2203.05212v1 | https://arxiv.org/pdf/2203.05212v1.pdf | Membership Privacy Protection for Image Translation Models via Adversarial Knowledge Distillation | Image-to-image translation models are shown to be vulnerable to the Membership Inference Attack (MIA), in which the adversary's goal is to identify whether a sample is used to train the model or not. With daily increasing applications based on image-to-image translation models, it is crucial to protect the privacy of these models against MIAs. We propose adversarial knowledge distillation (AKD) as a defense method against MIAs for image-to-image translation models. The proposed method protects the privacy of the training samples by improving the generalizability of the model. We conduct experiments on the image-to-image translation models and show that AKD achieves the state-of-the-art utility-privacy tradeoff by reducing the attack performance up to 38.9% compared with the regular training model at the cost of a slight drop in the quality of the generated output images. The experimental results also indicate that the models trained by AKD generalize better than the regular training models. Furthermore, compared with existing defense methods, the results show that at the same privacy protection level, image translation models trained by AKD generate outputs with higher quality; while at the same quality of outputs, AKD enhances the privacy protection over 30%. | ['Yong Zhang', 'Jian Pei', 'Lanjun Wang', 'Saeed Ranjbar Alvar'] | 2022-03-10 | null | null | null | null | ['inference-attack', 'membership-inference-attack'] | ['adversarial', 'computer-vision'] | [ 3.87036085e-01 2.79236436e-01 -5.58402911e-02 -2.96066552e-01
-1.03622413e+00 -9.39659178e-01 5.70254564e-01 -2.63707012e-01
-2.71387249e-01 5.82846105e-01 -2.02990174e-01 -6.92382336e-01
4.88165081e-01 -8.30291867e-01 -1.10659885e+00 -8.74358773e-01
3.13911259e-01 8.88680853e-03 -7.46093094e-02 3.08066428e-01
-8.28792304e-02 5.47960162e-01 -8.49225104e-01 5.52534342e-01
7.48216689e-01 1.22120333e+00 -5.28791070e-01 5.64330697e-01
3.38114738e-01 4.57829267e-01 -7.16512620e-01 -1.14701128e+00
9.56689417e-01 -5.19400775e-01 -6.24223590e-01 -2.68344209e-02
5.31031787e-01 -7.41106272e-01 -6.32833660e-01 1.43171608e+00
3.47972184e-01 -3.00088108e-01 5.02532601e-01 -1.63807487e+00
-9.52744126e-01 2.19791815e-01 -5.35520792e-01 -3.67901534e-01
1.03226967e-01 2.63440549e-01 5.46401203e-01 -7.19142675e-01
4.77659345e-01 1.11415887e+00 2.62443155e-01 7.73520410e-01
-1.42077279e+00 -1.23138642e+00 -2.28304639e-01 -2.55192090e-02
-1.74198198e+00 -4.33348328e-01 2.57559359e-01 -2.14303479e-01
4.55529630e-01 5.99600136e-01 1.48098392e-03 1.24324358e+00
2.52925783e-01 5.93855500e-01 1.38109946e+00 -2.38699630e-01
1.58377990e-01 8.67277145e-01 -1.68881819e-01 6.87524438e-01
4.78805333e-01 3.29365343e-01 -3.38004470e-01 -7.73979008e-01
8.28180969e-01 -1.67716473e-01 -4.24358904e-01 -3.89738649e-01
-7.20295846e-01 8.22939098e-01 2.65017241e-01 -2.98907995e-01
-3.25742960e-02 1.50897011e-01 3.38096499e-01 5.12473524e-01
3.03732336e-01 2.93439001e-01 -1.25475511e-01 4.72669095e-01
-6.22027695e-01 2.35167652e-01 8.43036711e-01 1.18144572e+00
6.77467525e-01 -3.00412495e-02 -4.85049784e-01 1.90249011e-01
1.75987512e-01 8.69877040e-01 1.58699244e-01 -9.72299039e-01
6.44371927e-01 3.31676334e-01 1.78378671e-01 -1.14116883e+00
6.75121248e-01 -3.37485254e-01 -9.66134608e-01 4.05745864e-01
2.92691559e-01 -2.71979213e-01 -7.97867179e-01 1.83401942e+00
1.94795445e-01 1.26132041e-01 4.01398808e-01 6.79907501e-01
2.47198120e-01 8.48951936e-01 8.38926584e-02 -2.72129416e-01
1.30141544e+00 -6.44459367e-01 -6.51545763e-01 9.38995630e-02
4.57004666e-01 -6.95349336e-01 9.93764400e-01 7.51020089e-02
-1.10627782e+00 -3.71625602e-01 -1.10973060e+00 1.55846238e-01
-1.94306776e-01 3.88987884e-02 3.28384966e-01 1.19890738e+00
-8.75708282e-01 2.50818163e-01 -5.55142403e-01 -6.38038889e-02
8.93639386e-01 5.85103452e-01 -6.78291500e-01 -1.73138961e-01
-1.27236247e+00 6.21475697e-01 1.03885412e-01 -2.56685883e-01
-1.19914150e+00 -7.55272806e-01 -7.92911887e-01 2.56497175e-01
1.05428457e-01 -7.93922603e-01 9.70160723e-01 -1.05796587e+00
-1.40554285e+00 1.07425022e+00 -7.96037242e-02 -8.26222599e-01
1.11831284e+00 -2.04149988e-02 -3.83445770e-01 2.96442837e-01
-9.80909467e-02 5.19216359e-01 1.13674760e+00 -1.47006011e+00
-5.17791748e-01 -4.97643441e-01 2.28235468e-01 2.24973317e-02
-5.82531810e-01 1.26388252e-01 -4.09948051e-01 -5.07016659e-01
-3.40362966e-01 -1.29533613e+00 -1.33198753e-01 4.84381020e-01
-5.98275006e-01 4.84564543e-01 1.25464904e+00 -5.54099023e-01
9.95130360e-01 -2.43715930e+00 -3.82557601e-01 4.61295545e-01
5.50300488e-03 6.21767640e-01 4.05126475e-02 2.41440386e-01
1.13785736e-01 4.61672992e-01 -2.39715680e-01 -5.54270089e-01
-2.65548807e-02 2.93409705e-01 -7.70395279e-01 6.31151259e-01
4.64148708e-02 9.14062500e-01 -5.13538122e-01 -4.14660901e-01
-5.35220280e-02 4.03407872e-01 -4.42421138e-01 5.11147320e-01
5.44075966e-02 4.79427069e-01 -3.71001184e-01 4.25950289e-01
1.18986034e+00 -1.40920088e-01 1.66658908e-01 8.90533254e-02
5.04681408e-01 -2.60071099e-01 -7.03121364e-01 1.10198259e+00
-1.70359537e-01 5.40799379e-01 7.36569613e-02 -4.73503888e-01
7.61101902e-01 6.12963557e-01 -3.76177356e-02 -4.13304746e-01
1.69019327e-01 1.15881130e-01 -8.39330703e-02 -3.39234471e-01
8.12122822e-02 -1.22192502e-01 -2.39425659e-01 4.48259503e-01
-3.76702428e-01 4.92234863e-02 -6.50758982e-01 1.90105632e-01
7.05431461e-01 -2.48439595e-01 -4.14407291e-02 -1.35665506e-01
6.97358489e-01 -2.92976201e-01 6.10200822e-01 8.59245956e-01
-2.26951659e-01 5.51898241e-01 4.61932659e-01 -2.78682977e-01
-1.11191010e+00 -1.28378487e+00 -1.40921950e-01 5.29940367e-01
2.53924698e-01 -1.09121300e-01 -1.15641987e+00 -1.04544270e+00
1.81166604e-01 7.63405919e-01 -6.04575872e-01 -3.13664496e-01
-1.72386736e-01 -4.62892771e-01 1.20424044e+00 1.50400266e-01
1.19239175e+00 -3.94178301e-01 -2.37191636e-02 -4.96954650e-01
-2.44155556e-01 -1.38190532e+00 -8.22510719e-01 -5.33964276e-01
-7.16820300e-01 -7.99246430e-01 -6.76820755e-01 -6.70921087e-01
1.24418557e+00 3.62027138e-01 4.15522039e-01 -6.34326339e-02
1.10979835e-02 2.41793349e-01 -1.37781724e-02 -4.45064276e-01
-9.76195991e-01 -2.03371152e-01 1.47488683e-01 4.87309188e-01
2.50105977e-01 -4.69606757e-01 -5.53948641e-01 6.11178100e-01
-1.12466252e+00 -3.23147848e-02 4.48754042e-01 7.56436288e-01
5.60583889e-01 1.11816555e-01 3.58019680e-01 -1.14575481e+00
5.52263141e-01 -3.94921154e-01 -7.63809621e-01 4.44664299e-01
-9.06897664e-01 -8.22432637e-02 8.83256197e-01 -7.19597638e-01
-9.91190553e-01 1.54257789e-01 1.11157514e-01 -8.95702183e-01
-6.40453696e-02 -1.30232215e-01 -5.65800726e-01 -5.87153077e-01
6.53137684e-01 5.48895299e-01 3.97079468e-01 -6.41974881e-02
3.63133758e-01 8.67315114e-01 4.92787987e-01 -5.37212551e-01
1.21210861e+00 6.08914673e-01 2.44061336e-01 -4.13072348e-01
-3.41909260e-01 6.85827807e-02 -1.28913283e-01 1.16952971e-01
8.06145012e-01 -1.07087076e+00 -8.29153240e-01 6.80130124e-01
-9.83380198e-01 3.11482009e-02 1.83020849e-02 3.62878889e-01
-5.60860574e-01 6.72899246e-01 -7.10668147e-01 -7.75177419e-01
-5.92498362e-01 -1.28254843e+00 7.97715008e-01 -2.16705408e-02
1.11023091e-01 -7.32413709e-01 -5.78074694e-01 7.83483148e-01
3.64517450e-01 4.71121311e-01 1.03341246e+00 -6.46688402e-01
-9.52263176e-01 -6.62141204e-01 -1.84489205e-01 8.46622169e-01
1.76110608e-03 -2.96531290e-01 -1.17436874e+00 -6.05517507e-01
4.67200339e-01 -2.54276037e-01 4.72741514e-01 -7.20312893e-02
1.36585891e+00 -1.11204994e+00 -1.02196380e-01 8.65365326e-01
1.42929006e+00 1.52663738e-01 1.07534456e+00 9.85608902e-03
4.28689420e-01 2.93410003e-01 5.22036791e-01 1.93188190e-01
-1.19245417e-01 4.90922362e-01 5.43136477e-01 -2.40633622e-01
3.19619149e-01 -6.08317435e-01 5.89634359e-01 1.26855746e-01
3.79239529e-01 -2.49151528e-01 -1.81419343e-01 3.25770587e-01
-1.83074689e+00 -9.18485999e-01 -1.12897903e-02 2.57048607e+00
8.19087446e-01 -2.12054551e-02 -1.38303980e-01 -2.25288913e-01
6.38994575e-01 -5.74089028e-03 -5.13333023e-01 -7.75777757e-01
-7.11387694e-02 1.60868969e-02 9.79882598e-01 4.42822307e-01
-9.82189536e-01 1.02004671e+00 6.66019011e+00 1.03466845e+00
-1.00211835e+00 1.82878718e-01 1.11524677e+00 -3.25092338e-02
-2.67250896e-01 1.40313551e-01 -5.94870925e-01 5.03111839e-01
8.68945599e-01 -5.34218669e-01 5.04898906e-01 1.01117516e+00
7.31801465e-02 2.97980726e-01 -1.09375811e+00 7.77815223e-01
2.45457273e-02 -1.27209973e+00 3.97580892e-01 5.34063160e-01
8.11955273e-01 -5.16956210e-01 5.94389975e-01 8.00315812e-02
2.20816508e-01 -1.08981144e+00 4.30985183e-01 1.22040763e-01
1.11469853e+00 -1.03213155e+00 7.22990870e-01 5.15591741e-01
-5.33865392e-01 6.69207424e-02 -6.10076070e-01 1.98459387e-01
-1.86224431e-01 2.67601967e-01 -9.44270551e-01 6.59960508e-01
4.53329295e-01 -1.03653230e-01 -2.61423796e-01 3.35680932e-01
-4.26473945e-01 7.22934544e-01 -1.62381902e-01 2.20107615e-01
2.30980828e-01 -2.65855163e-01 4.81972128e-01 1.02375734e+00
3.69112819e-01 6.16495498e-02 2.22694054e-02 1.10115397e+00
-5.59399962e-01 2.41241641e-02 -1.01551330e+00 4.89911735e-02
5.97598672e-01 8.19956124e-01 1.20652914e-01 -3.15382361e-01
-2.03505710e-01 1.58513558e+00 -1.19099997e-01 4.52260256e-01
-1.06395817e+00 -2.64010578e-01 1.11068630e+00 1.92621320e-01
1.62130415e-01 -8.32057595e-02 -3.14416260e-01 -9.99607265e-01
1.99542031e-01 -1.10574400e+00 2.42391035e-01 -5.06410241e-01
-1.21717703e+00 5.33678055e-01 -1.46822453e-01 -1.28628445e+00
-1.80389613e-01 -2.88553655e-01 -5.35351872e-01 1.20998216e+00
-1.26230264e+00 -1.44321227e+00 1.15565717e-01 8.53600800e-01
-8.45548809e-02 -2.86151946e-01 1.28142595e+00 5.09808809e-02
-3.23605865e-01 1.45302022e+00 3.02189857e-01 5.76709628e-01
6.36733353e-01 -6.78936481e-01 4.38207924e-01 1.21904206e+00
-2.89019174e-03 8.72620702e-01 4.49343204e-01 -4.68553931e-01
-1.57612860e+00 -1.35185039e+00 8.25998545e-01 -4.84302431e-01
2.19645917e-01 -5.93857646e-01 -8.31145883e-01 7.95772016e-01
2.59280413e-01 1.58184141e-01 9.58032250e-01 -5.93607843e-01
-7.71416008e-01 -2.01651484e-01 -1.95833826e+00 7.01444447e-01
6.03334785e-01 -9.88383293e-01 -2.38709133e-02 3.17103744e-01
1.08125627e+00 -3.00869763e-01 -9.36730862e-01 8.79741982e-02
6.12275779e-01 -8.46652269e-01 9.88238454e-01 -7.01998413e-01
2.42435694e-01 -4.22390848e-01 -4.53008026e-01 -8.06226194e-01
-2.77871728e-01 -9.28655922e-01 -1.07925557e-01 1.32688582e+00
4.58867162e-01 -8.49513829e-01 1.01267159e+00 1.11084044e+00
6.54760182e-01 -3.81410688e-01 -1.10045910e+00 -1.12741661e+00
2.19952524e-01 -7.05798492e-02 7.51175880e-01 9.70988989e-01
-2.41119146e-01 2.89775264e-02 -8.57000291e-01 8.68425071e-01
9.65659201e-01 -4.67664339e-02 1.10410774e+00 -3.79994184e-01
-3.77391607e-01 4.93285544e-02 -5.08939385e-01 -9.10576880e-01
1.96758464e-01 -9.55500603e-01 -3.10393184e-01 -8.56972992e-01
2.57219464e-01 -1.84331313e-01 -3.15310031e-01 4.30647671e-01
-2.24451721e-01 3.24039251e-01 4.27521348e-01 3.05729032e-01
2.63814330e-02 2.87175566e-01 1.15657926e+00 -2.17715457e-01
2.09826618e-01 5.00871062e-01 -9.38733220e-01 2.69099623e-01
7.61381149e-01 -7.41504312e-01 -6.84223771e-01 -3.65835607e-01
-2.36497387e-01 -3.04668825e-02 4.11037982e-01 -7.80790091e-01
1.71675712e-01 -2.81269491e-01 2.28800371e-01 -7.81753436e-02
3.05783868e-01 -1.25666153e+00 5.28610587e-01 6.76631689e-01
-4.72993433e-01 -2.62371093e-01 1.83201313e-01 7.40543425e-01
-3.84340137e-02 -5.82772167e-03 9.85986471e-01 -1.04100868e-01
-1.98865160e-02 5.77966630e-01 -2.96113610e-01 -2.79000223e-01
1.28656924e+00 -1.77329510e-01 -4.88687634e-01 -7.01223791e-01
-4.02155310e-01 1.69979841e-01 7.83668339e-01 2.21732646e-01
6.25756800e-01 -1.35215783e+00 -7.47128606e-01 5.74876666e-01
2.44449288e-01 -2.71184534e-01 1.04780048e-02 4.00936484e-01
-3.47318590e-01 1.49685547e-01 -1.11080892e-01 -3.79325926e-01
-1.74970877e+00 8.22055578e-01 2.56976157e-01 -2.04386070e-01
-4.10957217e-01 6.92557633e-01 6.40439510e-01 -1.79367691e-01
3.99107963e-01 4.09893803e-02 6.28382564e-01 -7.14440286e-01
5.90120077e-01 2.04198837e-01 -2.62970090e-01 -5.68315148e-01
-1.57259420e-01 1.84107661e-01 -2.41995290e-01 -2.70136088e-01
7.57578254e-01 -8.84447619e-02 -2.44424827e-02 -3.51474077e-01
1.61870170e+00 1.55366316e-01 -1.30644298e+00 -2.93724120e-01
-6.03105307e-01 -1.14230609e+00 -1.86851785e-01 -8.97597849e-01
-1.03790128e+00 9.45285320e-01 7.75288343e-01 5.54606989e-02
1.14670599e+00 -4.47752267e-01 1.10626137e+00 1.57657936e-01
5.40813863e-01 -7.52063632e-01 -1.76966503e-01 -9.92183387e-03
7.15339601e-01 -1.18980968e+00 -2.41089184e-02 -6.04591906e-01
-1.01960504e+00 6.46741867e-01 3.67644072e-01 -3.59660238e-02
5.19285083e-01 2.48261496e-01 2.15237036e-01 4.76067007e-01
-4.62172598e-01 4.87274408e-01 6.26118258e-02 8.26464236e-01
-4.08270776e-01 2.78636724e-01 -1.43046007e-02 5.54458261e-01
8.17978829e-02 8.81392974e-03 4.27711487e-01 7.49050260e-01
7.23451599e-02 -1.41738272e+00 -4.19387192e-01 -2.47055804e-03
-9.77237642e-01 -1.49794882e-02 -6.31918609e-01 5.60144842e-01
-3.46658863e-02 9.89641130e-01 -2.41457820e-01 -6.68923855e-01
2.62110561e-01 5.18542975e-02 2.02597648e-01 -2.18009070e-01
-8.60549092e-01 -2.23181650e-01 -2.71671768e-02 -5.19303501e-01
3.61797288e-02 -3.18205744e-01 -9.98425603e-01 -8.55032146e-01
-3.43952179e-01 1.58389419e-01 6.63149714e-01 4.05163944e-01
8.24066460e-01 -1.05738752e-01 1.40905643e+00 -6.58722147e-02
-1.03241420e+00 -4.66743290e-01 -5.24524629e-01 6.49046063e-01
2.42578179e-01 1.63019136e-01 -4.69001263e-01 1.56979740e-01] | [5.904544830322266, 7.141457557678223] |
20d6b8d7-ed37-43f5-bd57-7dcd9b4b8da5 | learning-phone-recognition-from-unpaired | 2207.14568 | null | https://arxiv.org/abs/2207.14568v1 | https://arxiv.org/pdf/2207.14568v1.pdf | Learning Phone Recognition from Unpaired Audio and Phone Sequences Based on Generative Adversarial Network | ASR has been shown to achieve great performance recently. However, most of them rely on massive paired data, which is not feasible for low-resource languages worldwide. This paper investigates how to learn directly from unpaired phone sequences and speech utterances. We design a two-stage iterative framework. GAN training is adopted in the first stage to find the mapping relationship between unpaired speech and phone sequence. In the second stage, another HMM model is introduced to train from the generator's output, which boosts the performance and provides a better segmentation for the next iteration. In the experiment, we first investigate different choices of model designs. Then we compare the framework to different types of baselines: (i) supervised methods (ii) acoustic unit discovery based methods (iii) methods learning from unpaired data. Our framework performs consistently better than all acoustic unit discovery methods and previous methods learning from unpaired data based on the TIMIT dataset. | ['Hung-Yi Lee', 'Da-Yi Wu', 'Shun-Po Chuang', 'Sung-Feng Huang', 'Yi-Chen Chen', 'Po-chun Hsu', 'Da-Rong Liu'] | 2022-07-29 | null | null | null | null | ['acoustic-unit-discovery'] | ['speech'] | [ 5.56913197e-01 7.18166307e-02 5.12207225e-02 -3.96527588e-01
-1.27965200e+00 -6.12706304e-01 6.46640718e-01 -3.45833778e-01
-4.68781918e-01 7.38023818e-01 2.54060954e-01 -5.23572326e-01
5.25358021e-01 -4.33164537e-01 -6.67015135e-01 -8.47239554e-01
4.02590871e-01 6.85423315e-01 2.02473670e-01 1.27154440e-01
7.94078410e-02 -9.09187421e-02 -1.38984442e+00 1.86798796e-01
9.23475683e-01 8.45526099e-01 6.56246603e-01 7.94909358e-01
-3.36702824e-01 4.18897718e-01 -5.83351076e-01 -8.40868950e-02
2.58260578e-01 -1.12405729e+00 -6.57836437e-01 -2.71970052e-02
-4.37526405e-02 -1.29390761e-01 -6.66345954e-02 1.01449251e+00
7.42758930e-01 2.60108620e-01 6.62196696e-01 -8.05471897e-01
-2.65155613e-01 9.77429032e-01 -4.18609321e-01 6.95884004e-02
2.50381917e-01 -4.05328870e-02 8.52686882e-01 -1.01325226e+00
3.20393950e-01 1.19335806e+00 5.25244772e-01 8.23082924e-01
-1.04762554e+00 -6.93417251e-01 1.10932095e-02 5.95245995e-02
-1.42218864e+00 -9.50067461e-01 7.63563156e-01 -1.86250240e-01
8.20463955e-01 2.86228806e-01 2.97684342e-01 1.38288128e+00
-4.33353901e-01 9.04133320e-01 1.24439943e+00 -8.80236804e-01
3.95675838e-01 1.57685995e-01 -1.38773039e-01 4.56616610e-01
-4.55608815e-01 2.23034754e-01 -5.28092384e-01 3.04643270e-02
6.75771117e-01 -5.36813140e-01 -1.24626100e-01 2.73077935e-01
-1.12201107e+00 8.09588969e-01 -7.25474954e-02 5.79275548e-01
-2.53267407e-01 -6.61697909e-02 2.44984969e-01 1.88116163e-01
3.84649783e-01 3.40968579e-01 -3.20115864e-01 -5.97189844e-01
-1.24347425e+00 -1.82785794e-01 6.95597947e-01 1.06398988e+00
5.84915996e-01 2.55115002e-01 -1.43639043e-01 1.20558202e+00
5.76333761e-01 3.52963507e-01 6.51440382e-01 -6.24178767e-01
7.23154783e-01 -9.30832922e-02 -2.11613253e-01 -2.13345468e-01
5.59209883e-02 -3.82677197e-01 -7.59677470e-01 -3.02307844e-01
4.54357207e-01 -3.24451774e-01 -1.31681275e+00 1.61809874e+00
1.30863875e-01 4.74203080e-01 1.03330970e-01 7.05038130e-01
7.08907247e-01 9.98739481e-01 -3.09806585e-01 -4.58497465e-01
1.11950088e+00 -1.31235206e+00 -8.03440809e-01 -2.04052716e-01
6.22616172e-01 -9.40116942e-01 1.06742573e+00 4.32632208e-01
-9.93161321e-01 -8.27431262e-01 -9.46485758e-01 1.87344566e-01
-1.44979343e-01 3.76879513e-01 5.38578741e-02 8.84571493e-01
-1.05488729e+00 2.89309889e-01 -1.03455222e+00 -2.76132315e-01
2.66765188e-02 2.25971580e-01 9.49599035e-03 4.26055714e-02
-1.22484493e+00 6.49933517e-01 4.32115704e-01 3.71470690e-01
-1.06997621e+00 -3.40379775e-01 -8.40673983e-01 -1.30467430e-01
3.21057081e-01 -3.81402582e-01 1.60202038e+00 -8.43488216e-01
-2.09832764e+00 6.23584151e-01 -5.69700778e-01 -4.55966890e-01
4.79988545e-01 -1.83648065e-01 -3.29910070e-01 -1.40540555e-01
-1.53019175e-01 7.19646037e-01 8.61210167e-01 -1.38249135e+00
-7.39221990e-01 -8.06926936e-02 -4.07819480e-01 2.16253579e-01
-4.11035791e-02 -8.17128494e-02 -8.35610390e-01 -7.70121872e-01
1.10748969e-01 -1.04111195e+00 -1.99305937e-01 -8.86480987e-01
-7.97107458e-01 -2.69117683e-01 7.10196972e-01 -9.53381360e-01
1.36683524e+00 -2.29129243e+00 1.05166450e-01 3.26840729e-01
-3.63236248e-01 3.97350281e-01 7.64897326e-03 4.00643051e-01
1.19925275e-01 2.17170388e-01 -4.91224706e-01 -9.58055079e-01
5.65956011e-02 3.60748708e-01 -2.99851984e-01 3.22145503e-03
1.02488674e-01 7.41277397e-01 -7.32501566e-01 -4.98358965e-01
2.63837904e-01 3.58343512e-01 -3.42206955e-01 6.55068636e-01
-7.49749765e-02 9.14467990e-01 -1.84437767e-01 5.70814133e-01
4.50097203e-01 2.34718531e-01 3.73320043e-01 -3.53294946e-02
-2.69034624e-01 7.38374472e-01 -8.66107345e-01 1.91050696e+00
-6.32002652e-01 5.36951363e-01 8.49079937e-02 -1.14720345e+00
1.12832594e+00 7.62296796e-01 6.52221590e-02 -7.12010562e-01
1.59827903e-01 4.43999261e-01 2.54319459e-01 -3.66974115e-01
3.11062038e-01 -2.90292531e-01 -5.41407391e-02 2.37653702e-01
1.40380695e-01 -1.42163187e-01 1.21675916e-02 -1.20311633e-01
1.01604247e+00 2.02582255e-01 1.32127434e-01 4.95134294e-02
4.50244725e-01 -3.52681994e-01 6.94299638e-01 8.61331999e-01
-7.12681329e-03 9.68526483e-01 1.03001617e-01 1.97370455e-01
-1.05681121e+00 -1.10318601e+00 2.63714939e-02 8.72614443e-01
-6.83519244e-02 -2.43411362e-01 -1.13706684e+00 -6.33203030e-01
-6.13599300e-01 8.97281706e-01 -4.50444251e-01 1.03613943e-01
-8.62260818e-01 -6.69937193e-01 7.41925299e-01 7.45933473e-01
5.87794960e-01 -1.15197551e+00 -2.87954770e-02 2.42939353e-01
-4.48549747e-01 -1.18588901e+00 -6.64637685e-01 5.13628304e-01
-8.20457339e-01 -3.98462087e-01 -7.38713205e-01 -1.00553882e+00
5.60590565e-01 -1.96546808e-01 8.31981361e-01 -1.04604229e-01
2.29066342e-01 -4.97544967e-02 -6.99722350e-01 -1.72021776e-01
-7.62292325e-01 4.65826154e-01 1.05941564e-01 1.42046422e-01
2.92460769e-01 -5.23251772e-01 -2.02679366e-01 3.20910662e-01
-6.87541127e-01 3.49639021e-02 8.25683594e-01 9.28417325e-01
7.09542453e-01 5.07314727e-02 7.72808254e-01 -1.01606214e+00
3.47719014e-01 -4.37508166e-01 -4.74152923e-01 2.09039107e-01
-5.14132142e-01 2.22193375e-01 7.57098913e-01 -2.92043090e-01
-1.36263514e+00 3.10478956e-01 -7.80243933e-01 -3.89794141e-01
-4.35616583e-01 5.20957232e-01 -6.10534370e-01 4.78367746e-01
3.53816271e-01 3.26772302e-01 -8.36818367e-02 -9.12722349e-01
3.45775038e-01 1.09819996e+00 7.87538350e-01 -5.82659483e-01
7.17438161e-01 -9.55354609e-03 -7.21767306e-01 -1.00432253e+00
-5.77786684e-01 -4.81568635e-01 -7.43528843e-01 1.25601277e-01
9.71947193e-01 -6.80008292e-01 -1.32817388e-01 5.34333706e-01
-1.16649115e+00 -5.26786268e-01 -3.85798886e-02 7.18542159e-01
-5.74548006e-01 3.08195174e-01 -7.28666663e-01 -1.03269625e+00
-2.42084995e-01 -1.16517115e+00 1.01198995e+00 2.34938622e-01
-1.52797326e-01 -9.49625790e-01 3.79380643e-01 4.16206479e-01
2.97168642e-01 -1.71290666e-01 6.80767059e-01 -8.39341462e-01
-3.36487651e-01 8.02098066e-02 2.72702962e-01 6.86947525e-01
3.40847880e-01 -1.27933929e-02 -1.30564272e+00 -1.88302740e-01
1.11563010e-02 -1.13377370e-01 7.04783499e-01 4.00153399e-01
1.06841171e+00 -2.15980962e-01 -3.99675965e-01 5.16716301e-01
9.72638190e-01 8.43846679e-01 9.85025167e-01 -7.68637005e-03
7.13753045e-01 4.37453985e-01 6.73800707e-01 -4.30527963e-02
2.56208152e-01 8.02399158e-01 -1.21383239e-02 -1.69694468e-01
-1.34439453e-01 -6.19417250e-01 5.91836572e-01 1.66239953e+00
2.96322443e-02 -5.52989662e-01 -9.16608155e-01 6.36191368e-01
-1.71403170e+00 -7.10857987e-01 1.19619131e-01 2.35939026e+00
9.62794125e-01 2.51623243e-01 2.08337858e-01 2.31085077e-01
9.02748227e-01 -2.66323145e-02 -2.76325405e-01 -4.52435225e-01
-1.31384395e-02 5.42464972e-01 4.58352361e-03 6.18384421e-01
-8.94584060e-01 9.90583122e-01 6.70499611e+00 1.12672353e+00
-9.70778823e-01 3.50569189e-01 6.08547091e-01 1.36585861e-01
-3.90210539e-01 1.29540980e-01 -9.15060580e-01 7.34775424e-01
1.31750822e+00 3.89572501e-01 4.33184594e-01 4.01566058e-01
1.68379769e-01 -1.87814012e-02 -1.17117012e+00 1.02740991e+00
-1.41151687e-02 -1.01854360e+00 -2.30798498e-01 -4.98464033e-02
6.91876948e-01 -1.18728383e-02 -9.84833613e-02 4.23278928e-01
2.56787390e-01 -1.11920023e+00 7.56452084e-01 3.69612932e-01
8.34219754e-01 -7.84449339e-01 8.92398298e-01 4.88629401e-01
-1.26729095e+00 2.45652646e-01 -8.64709169e-02 1.78779930e-01
6.02286279e-01 5.65136492e-01 -1.14430523e+00 6.43802464e-01
4.63308662e-01 3.39430869e-01 -2.13224590e-01 9.70241129e-01
-4.69446391e-01 1.38925338e+00 -3.60752493e-01 1.19792670e-01
3.19010049e-01 -3.69240433e-01 4.52092469e-01 1.44195724e+00
5.26157856e-01 -4.96024676e-02 2.04909787e-01 7.46144891e-01
9.53567494e-03 4.52869087e-02 -5.65033436e-01 -2.55249619e-01
8.16527188e-01 1.07823062e+00 -9.06429172e-01 -3.29097807e-01
-3.88132632e-01 1.13658035e+00 8.98480266e-02 3.68133545e-01
-8.31073582e-01 -4.32046264e-01 1.52857646e-01 -1.80557698e-01
5.05369544e-01 -3.53303254e-01 -1.34377018e-01 -9.10603464e-01
-4.46104109e-02 -1.00984287e+00 1.57590836e-01 -4.57859427e-01
-1.17487025e+00 8.10909152e-01 -1.58423364e-01 -1.04180324e+00
-7.85010099e-01 -2.54574597e-01 -7.35542655e-01 9.54982638e-01
-1.19700396e+00 -9.84944463e-01 9.85189900e-02 2.33767882e-01
7.80479789e-01 -1.75268680e-01 8.61072540e-01 4.54087943e-01
-6.82832658e-01 8.84361982e-01 1.85124129e-01 3.44470620e-01
5.34154475e-01 -1.27108335e+00 7.33467042e-01 1.02990735e+00
5.24096966e-01 4.89163160e-01 5.45196295e-01 -4.88648832e-01
-9.51985121e-01 -8.00527513e-01 8.51651073e-01 -4.17525679e-01
3.22590947e-01 -7.59082437e-01 -9.21956658e-01 6.16250277e-01
3.50848645e-01 -3.25650454e-01 8.14376414e-01 7.35192597e-02
-7.45394826e-02 -2.54743430e-03 -6.97096348e-01 2.92325705e-01
9.23523188e-01 -7.40566015e-01 -6.71390295e-01 -1.37477219e-01
8.54001403e-01 -4.23452586e-01 -6.65680528e-01 2.97558725e-01
4.86227512e-01 -6.88430309e-01 5.41752636e-01 -2.03356951e-01
7.23437518e-02 -4.71425116e-01 -4.30182219e-01 -1.51998997e+00
9.65605974e-02 -9.65742767e-01 -7.40833059e-02 1.86006224e+00
9.39187288e-01 -5.90780377e-01 5.42130649e-01 2.20255982e-02
-6.56213820e-01 -5.83319724e-01 -1.01885438e+00 -9.46667135e-01
2.03138683e-02 -4.59782362e-01 5.48289716e-01 6.47828937e-01
-2.71447927e-01 6.17357671e-01 -4.41023469e-01 1.48736030e-01
4.41104412e-01 1.04458528e-02 6.84485495e-01 -6.75870776e-01
-6.58937454e-01 -1.88481972e-01 -8.83317590e-02 -1.45413423e+00
6.06958941e-02 -6.56014681e-01 7.91188002e-01 -1.27283263e+00
-4.52723838e-02 -7.62210667e-01 -3.00555885e-01 4.60230738e-01
-3.08327168e-01 2.80055940e-01 1.31329689e-02 -2.89496426e-02
-2.72943974e-01 6.37201250e-01 8.10461938e-01 4.81579229e-02
-3.59374344e-01 2.52251118e-01 -4.58870471e-01 4.96341735e-01
8.88873279e-01 -4.73006666e-01 -5.00385404e-01 -4.09530878e-01
-2.27423266e-01 8.69926214e-02 -4.32497226e-02 -9.58422840e-01
2.12372318e-01 1.78719267e-01 -1.80641282e-02 -7.90478528e-01
3.49651933e-01 -4.68277872e-01 1.90461114e-01 1.29519030e-01
-2.94082105e-01 -6.78463057e-02 3.45942788e-02 4.96244073e-01
-5.74940145e-01 -4.84913588e-01 7.10738420e-01 3.21018323e-02
-4.88451600e-01 4.12503220e-02 -4.05447543e-01 1.46576911e-01
5.74092031e-01 -1.96150810e-01 9.78915170e-02 -4.19490039e-01
-8.29303503e-01 -9.24748555e-03 2.96885431e-01 4.43272799e-01
4.15444732e-01 -1.37119746e+00 -6.52871192e-01 2.08072275e-01
-1.63765237e-01 1.06626168e-01 -1.73488781e-02 8.76660705e-01
-1.60712928e-01 3.57478857e-01 3.58432621e-01 -6.83424592e-01
-1.11329329e+00 3.03672343e-01 1.11149885e-01 -2.07406342e-01
-3.97902668e-01 1.00714505e+00 2.23517925e-01 -6.49382293e-01
3.41920108e-01 -2.40251824e-01 -1.08638823e-01 1.23050781e-02
3.91970068e-01 2.27858633e-01 1.97280765e-01 -7.92771697e-01
-3.73398662e-01 4.12018478e-01 -1.51058495e-01 -7.49073744e-01
1.07648325e+00 -3.31167608e-01 1.33391812e-01 8.77889216e-01
1.22994852e+00 9.03608352e-02 -1.19715798e+00 -1.96011975e-01
2.13708848e-01 -3.00596863e-01 -7.97300190e-02 -6.39926910e-01
-9.07771409e-01 1.07353032e+00 5.45166492e-01 1.78873688e-01
1.09714043e+00 4.57712114e-02 1.09810579e+00 -2.68808883e-02
4.15863127e-01 -1.32203770e+00 -2.59572621e-02 4.95117217e-01
5.36153257e-01 -1.10972095e+00 -6.90022826e-01 -4.30976540e-01
-6.19409442e-01 8.63551438e-01 4.87617701e-01 2.45570973e-01
4.55992669e-01 4.71657246e-01 2.45644107e-01 2.45182157e-01
-5.91074586e-01 -3.56508523e-01 2.50872254e-01 6.45763159e-01
6.72970116e-01 1.81376934e-01 -1.64197415e-01 6.50018334e-01
-4.69598293e-01 -3.63351554e-01 2.55953252e-01 7.95394957e-01
-3.49276662e-01 -1.59691942e+00 -2.91979939e-01 3.01673055e-01
-4.31100577e-01 -3.25712472e-01 -4.26075131e-01 4.67769980e-01
1.03857420e-01 1.35541773e+00 1.34442732e-01 -6.14413440e-01
1.87834874e-01 3.48869085e-01 4.83488262e-01 -8.88564408e-01
-4.04314697e-01 4.57248092e-01 2.71182567e-01 -2.42705345e-01
-4.16247636e-01 -8.34118068e-01 -1.28534329e+00 1.62422001e-01
-6.91375077e-01 5.10688722e-01 7.29018450e-01 1.21271503e+00
1.63546816e-01 5.82330644e-01 8.15175831e-01 -7.78182983e-01
-3.65988076e-01 -1.28420019e+00 -4.63882267e-01 -1.87160773e-03
2.63364106e-01 -5.30769706e-01 -4.62607145e-01 2.70019650e-01] | [14.532453536987305, 6.709886074066162] |
672549bf-ecc0-4ca3-8149-f0fffb2c5628 | tackling-heavy-tailed-rewards-in | 2306.06836 | null | https://arxiv.org/abs/2306.06836v1 | https://arxiv.org/pdf/2306.06836v1.pdf | Tackling Heavy-Tailed Rewards in Reinforcement Learning with Function Approximation: Minimax Optimal and Instance-Dependent Regret Bounds | While numerous works have focused on devising efficient algorithms for reinforcement learning (RL) with uniformly bounded rewards, it remains an open question whether sample or time-efficient algorithms for RL with large state-action space exist when the rewards are \emph{heavy-tailed}, i.e., with only finite $(1+\epsilon)$-th moments for some $\epsilon\in(0,1]$. In this work, we address the challenge of such rewards in RL with linear function approximation. We first design an algorithm, \textsc{Heavy-OFUL}, for heavy-tailed linear bandits, achieving an \emph{instance-dependent} $T$-round regret of $\tilde{O}\big(d T^{\frac{1-\epsilon}{2(1+\epsilon)}} \sqrt{\sum_{t=1}^T \nu_t^2} + d T^{\frac{1-\epsilon}{2(1+\epsilon)}}\big)$, the \emph{first} of this kind. Here, $d$ is the feature dimension, and $\nu_t^{1+\epsilon}$ is the $(1+\epsilon)$-th central moment of the reward at the $t$-th round. We further show the above bound is minimax optimal when applied to the worst-case instances in stochastic and deterministic linear bandits. We then extend this algorithm to the RL settings with linear function approximation. Our algorithm, termed as \textsc{Heavy-LSVI-UCB}, achieves the \emph{first} computationally efficient \emph{instance-dependent} $K$-episode regret of $\tilde{O}(d \sqrt{H \mathcal{U}^*} K^\frac{1}{1+\epsilon} + d \sqrt{H \mathcal{V}^* K})$. Here, $H$ is length of the episode, and $\mathcal{U}^*, \mathcal{V}^*$ are instance-dependent quantities scaling with the central moment of reward and value functions, respectively. We also provide a matching minimax lower bound $\Omega(d H K^{\frac{1}{1+\epsilon}} + d \sqrt{H^3 K})$ to demonstrate the optimality of our algorithm in the worst case. Our result is achieved via a novel robust self-normalized concentration inequality that may be of independent interest in handling heavy-tailed noise in general online regression problems. | ['Lin F. Yang', 'LiWei Wang', 'Han Zhong', 'Jiayi Huang'] | 2023-06-12 | null | null | null | null | ['open-question'] | ['natural-language-processing'] | [-6.70390204e-04 3.37468863e-01 -3.42996448e-01 -1.94255516e-01
-1.32193255e+00 -6.90590143e-01 -1.27126664e-01 1.75848931e-01
-1.16070676e+00 1.22498953e+00 -5.15701830e-01 -7.58749485e-01
-8.74575675e-01 -8.99582267e-01 -9.43578362e-01 -9.35163438e-01
-6.66934609e-01 5.45338035e-01 -1.09120153e-01 -3.24791551e-01
2.51048923e-01 1.06825009e-01 -1.57396924e+00 -3.85370255e-01
1.04150772e+00 1.74639571e+00 1.05599374e-01 7.50164807e-01
3.16385999e-02 7.40547538e-01 -6.45100057e-01 -5.01969635e-01
4.73010451e-01 -8.44829917e-01 -5.71132302e-01 -2.74979770e-01
-2.33124733e-01 -2.76275992e-01 -7.21755773e-02 1.22849023e+00
4.99209046e-01 4.41819072e-01 5.46608329e-01 -1.06111979e+00
-2.81164289e-01 7.27409124e-01 -9.10990596e-01 1.66244745e-01
1.66791212e-03 2.17077971e-01 1.15761149e+00 -3.13777775e-02
2.55202502e-01 8.68528485e-01 2.95139194e-01 5.90907931e-01
-9.76436436e-01 -1.03864908e+00 5.26165545e-01 -4.60601509e-01
-9.47615623e-01 1.69642180e-01 3.20337266e-01 -2.94058621e-01
7.52069116e-01 5.33949256e-01 5.37907243e-01 2.92659163e-01
6.94379956e-02 9.30896044e-01 1.32882714e+00 -4.75955248e-01
6.55893564e-01 1.08747557e-01 -3.00578866e-02 6.33056819e-01
2.03677416e-02 4.26241338e-01 -1.62062213e-01 2.95747034e-02
9.74826753e-01 1.42637461e-01 -4.43865173e-03 3.07949185e-01
-6.71698332e-01 1.26019394e+00 1.65194005e-01 9.03294012e-02
-4.00386453e-01 8.27572465e-01 1.88862994e-01 5.01094043e-01
3.41895401e-01 2.73613781e-01 -6.99613333e-01 -6.43487394e-01
-8.71660352e-01 4.49767888e-01 5.10363042e-01 1.23142147e+00
6.51799917e-01 1.19676962e-01 -4.12227660e-01 6.42085433e-01
-1.38844147e-01 1.05907917e+00 4.14166879e-03 -1.39700842e+00
8.01128507e-01 1.51518196e-01 5.71708918e-01 -5.54223321e-02
-4.11896020e-01 -6.05534732e-01 -7.00618982e-01 2.00028658e-01
8.79299879e-01 -7.52836347e-01 -4.67692763e-01 2.15776348e+00
8.07839409e-02 -5.48533916e-01 -1.81878954e-01 7.02025712e-01
-1.50931971e-02 6.60240710e-01 -5.26329651e-02 -1.08641100e+00
1.12450612e+00 -3.62247437e-01 -2.91158020e-01 -2.29034737e-01
5.75852275e-01 -5.77227950e-01 1.44057357e+00 4.47026074e-01
-1.66633892e+00 -4.59395498e-02 -4.95760262e-01 5.66029489e-01
2.00950772e-01 -3.65078390e-01 7.88112998e-01 1.00908101e+00
-5.80166936e-01 7.74323225e-01 -4.48173195e-01 6.46991074e-01
4.01767373e-01 6.37451768e-01 1.32328838e-01 2.31365524e-02
-1.07468903e+00 2.34873265e-01 6.67715743e-02 -1.24953404e-01
-7.97658205e-01 -7.04816341e-01 -4.92840946e-01 4.95817959e-02
8.80764604e-01 1.21316180e-01 1.40773118e+00 -6.96872711e-01
-1.42094874e+00 4.51913744e-01 2.51811504e-01 -6.64213657e-01
6.22967839e-01 2.63461947e-01 1.09471709e-01 -4.57713157e-02
2.82282382e-01 3.46993566e-01 4.76626992e-01 -8.12069535e-01
-1.08701754e+00 -6.98297977e-01 4.42358822e-01 1.86177567e-02
-2.01885670e-01 5.00677228e-02 1.15021849e-02 -5.42174339e-01
-1.90435410e-01 -1.13334429e+00 -3.43696177e-01 -7.09907889e-01
-8.66714343e-02 -2.03277752e-01 1.33762602e-02 -2.46176377e-01
1.43390858e+00 -2.01660919e+00 -7.93325305e-02 4.60241139e-01
-1.84783548e-01 -2.83171888e-03 2.75539160e-01 3.28681171e-01
1.59134701e-01 2.08476484e-01 -1.25055641e-01 -1.29033387e-01
4.35006917e-01 2.29033232e-01 -3.34788263e-01 3.61491561e-01
-5.13227522e-01 5.36137819e-01 -7.50410497e-01 3.62044647e-02
-1.96875453e-01 4.54928167e-02 -6.01334214e-01 -5.89454770e-02
-5.39811611e-01 3.89918745e-01 -8.20440173e-01 3.61329257e-01
5.44691384e-01 -2.56793261e-01 5.10722697e-02 5.71011424e-01
-4.31676686e-01 -8.80233943e-02 -1.53767920e+00 1.14737725e+00
-3.89264047e-01 -2.72311538e-01 2.06894949e-01 -1.24237502e+00
6.07650995e-01 -9.28029045e-03 7.58046567e-01 -1.14876008e+00
4.74918813e-01 3.98735344e-01 -2.55473644e-01 4.24639061e-02
2.37025797e-01 -9.26613033e-01 -5.18653274e-01 7.37191439e-01
-3.25350195e-01 -2.32065856e-01 4.17816609e-01 -8.11406523e-02
1.19274437e+00 -3.46096046e-02 -1.19038463e-01 -3.11832190e-01
1.83407441e-01 -3.55313271e-01 7.32791007e-01 1.03679478e+00
-3.38825613e-01 -1.38520598e-01 1.45095491e+00 -2.59805530e-01
-8.78192842e-01 -7.87799954e-01 -1.97820753e-01 1.57117760e+00
2.02887148e-01 -3.02036782e-03 -7.91005135e-01 -6.29222810e-01
8.53110105e-02 1.06542337e+00 -9.10734117e-01 -4.34587374e-02
-3.87673795e-01 -8.29745889e-01 6.43595338e-01 6.85244679e-01
5.04083395e-01 -1.25203490e+00 -1.04605722e+00 1.56751484e-01
1.56532556e-01 -5.30477524e-01 -6.48234963e-01 9.56825137e-01
-5.97175360e-01 -5.71586430e-01 -9.68687654e-01 -1.55960411e-01
6.49708867e-01 -3.02382469e-01 6.53087556e-01 -3.62397254e-01
-2.08238214e-01 3.63526106e-01 -3.94056499e-01 -8.13379884e-01
2.16122389e-01 -3.48295778e-01 3.14589180e-02 -3.33418369e-01
-1.05826510e-02 -4.15072083e-01 -1.02209139e+00 1.12645805e-01
-8.81461084e-01 -4.07401860e-01 4.26060200e-01 9.56176758e-01
1.15341818e+00 8.17461759e-02 6.82566762e-01 -7.93390274e-01
6.57604814e-01 -6.83331192e-02 -1.35141325e+00 -2.03620326e-02
-6.94809318e-01 3.78046274e-01 1.00940096e+00 -6.07116699e-01
-5.78931212e-01 -2.19126716e-01 -1.15961298e-01 -4.74454343e-01
3.28051776e-01 3.29025686e-01 3.65043461e-01 2.01996714e-01
5.28642416e-01 4.10628438e-01 -1.23114809e-01 -3.43796790e-01
2.49470010e-01 3.32417071e-01 9.79551598e-02 -1.03804541e+00
4.06735599e-01 1.63300827e-01 2.98300177e-01 -2.67066419e-01
-1.05026031e+00 2.33138934e-01 3.93841922e-01 1.16523020e-02
6.16407871e-01 -5.86863220e-01 -1.94509244e+00 -2.76838839e-01
-2.40670770e-01 -8.13260674e-01 -1.13180971e+00 6.48618639e-01
-1.29346526e+00 1.24970824e-01 -4.35599595e-01 -1.80410910e+00
-4.08900559e-01 -1.09224021e+00 6.20877862e-01 4.56038058e-01
2.90410459e-01 -3.91851306e-01 -2.84133613e-01 5.30298829e-01
2.80890405e-01 7.29950517e-02 1.06433308e+00 -3.55176210e-01
-4.85056400e-01 -2.81498611e-01 -1.06153779e-01 7.35830069e-01
-3.75728905e-01 -5.80940366e-01 -3.28691572e-01 -5.05295157e-01
-6.68863654e-02 -6.13485098e-01 5.85173607e-01 7.71521449e-01
1.43467879e+00 -6.89233661e-01 1.67109087e-01 2.20436111e-01
1.43876517e+00 7.72683263e-01 5.35433412e-01 2.63152599e-01
-2.59158701e-01 1.59650803e-01 9.70801711e-01 1.33409357e+00
2.36582130e-01 3.09748083e-01 7.31833100e-01 2.93482065e-01
7.37565875e-01 -2.38081682e-02 4.14011508e-01 -9.98752844e-03
-3.00970465e-01 -4.91586067e-02 -4.72438812e-01 6.79825723e-01
-1.65273738e+00 -9.67764676e-01 2.61938810e-01 2.93406987e+00
1.12252975e+00 5.16777515e-01 4.96103823e-01 1.73436239e-01
5.13337612e-01 -2.03988433e-01 -7.17049837e-01 -9.31799173e-01
-1.33010047e-02 1.09292984e+00 9.63518322e-01 3.46374750e-01
-6.39475405e-01 8.21142077e-01 3.12053823e+00 1.54358923e+00
-9.27255452e-01 6.71041850e-03 8.38973939e-01 -8.56366754e-01
-1.79238230e-01 2.25688055e-01 -9.02845383e-01 1.04506063e+00
9.94188428e-01 -9.07087699e-02 8.43873978e-01 1.13215828e+00
1.67553172e-01 -6.66570425e-01 -9.61586237e-01 7.68701077e-01
-6.70772076e-01 -1.11052036e+00 -7.31887519e-01 4.58627969e-01
9.40674961e-01 -4.11997348e-01 5.97408652e-01 7.14143336e-01
8.15965950e-01 -1.07421398e+00 8.58697712e-01 -3.88316587e-02
1.34937596e+00 -1.40056169e+00 5.20773888e-01 7.16548204e-01
-1.18708730e+00 -7.04916060e-01 -2.64045030e-01 -1.60027355e-01
5.40825278e-02 6.26151562e-01 -3.13176364e-01 4.62755919e-01
9.65825200e-01 -5.03965020e-01 6.34326577e-01 6.83528543e-01
-2.97198649e-02 2.77675629e-01 -5.74836850e-01 -4.47611094e-01
7.56560445e-01 -2.53705084e-01 1.15094200e-01 6.86526895e-01
4.99920160e-01 6.59191847e-01 5.28327703e-01 5.82751632e-01
-1.60730928e-01 2.35138744e-01 6.56637456e-03 -1.33375347e-01
3.89736921e-01 6.95005774e-01 -6.68674052e-01 3.03938519e-02
4.37609926e-02 4.59255576e-01 2.31630012e-01 2.57209539e-01
-1.27833986e+00 -8.66031408e-01 5.72347999e-01 1.93399280e-01
7.16994822e-01 -9.95475277e-02 -2.43713170e-01 -6.02895141e-01
2.40748271e-01 -5.35415590e-01 7.39366531e-01 -1.89394355e-01
-9.45032597e-01 1.80741578e-01 -3.76575254e-02 -9.59936261e-01
-2.47149676e-01 -2.82317311e-01 -1.89564347e-01 8.65748465e-01
-1.23398292e+00 -6.02173328e-01 4.70616996e-01 9.54427600e-01
1.26996100e-01 -9.50489044e-02 6.56834483e-01 1.51775032e-01
-3.72678220e-01 1.01199353e+00 5.95228612e-01 -1.68781400e-01
1.36744693e-01 -1.11220717e+00 -6.83099985e-01 1.97883815e-01
-5.22689819e-01 3.30956399e-01 7.58300960e-01 -4.05727863e-01
-1.37355280e+00 -7.93736815e-01 3.65677714e-01 1.44675151e-01
5.31688929e-01 -1.36511713e-01 -1.12166367e-01 7.90935755e-01
-3.15672576e-01 1.73909560e-01 6.56350970e-01 -1.37059003e-01
-6.64576292e-02 -4.39601332e-01 -1.53130138e+00 6.36407256e-01
8.47790241e-01 -1.56799614e-01 6.50246888e-02 4.31416094e-01
6.59886658e-01 -3.82486582e-01 -1.31158078e+00 4.67056870e-01
4.47069108e-01 -1.02885294e+00 5.01503408e-01 -6.26996875e-01
4.07193750e-02 2.04835221e-01 -4.98644143e-01 -7.91470289e-01
9.31199417e-02 -1.15507603e+00 -3.63435224e-02 6.76691771e-01
7.76733279e-01 -6.05979860e-01 9.63512778e-01 8.81811142e-01
-1.21927047e-02 -1.23344493e+00 -1.57757866e+00 -9.29308832e-01
8.98934066e-01 -4.22859192e-01 2.54583746e-01 4.02231276e-01
2.11231112e-01 -3.30423713e-01 -5.30432343e-01 -3.34243894e-01
6.20994329e-01 2.34966755e-01 3.64884913e-01 -8.77215624e-01
-8.55086565e-01 -4.86872941e-01 3.31912369e-01 -1.05830526e+00
-1.60496265e-01 -4.43241388e-01 -6.25545830e-02 -9.93980408e-01
3.00720513e-01 -1.12230122e+00 -5.15248179e-01 7.41735816e-01
2.46674284e-01 -4.33654010e-01 3.71982843e-01 -2.76946634e-01
-7.73099422e-01 5.99556684e-01 1.40078914e+00 9.60709602e-02
-3.85177076e-01 3.58504176e-01 -8.36622238e-01 4.24775958e-01
6.01705134e-01 -5.72182059e-01 -5.68748593e-01 -2.30382696e-01
6.26703143e-01 1.01801288e+00 1.39150899e-02 -5.31547368e-01
-2.66636945e-02 -7.91149735e-01 3.85547429e-02 -5.48565805e-01
3.86008203e-01 -4.33337301e-01 -1.04037389e-01 6.19056702e-01
-4.97275591e-01 -1.86963454e-02 6.71591684e-02 5.75844824e-01
3.93923163e-01 -3.00983518e-01 1.05318832e+00 -4.63564843e-01
2.61681706e-01 3.28530461e-01 -2.25068942e-01 4.66146529e-01
1.36205816e+00 1.91976894e-02 -2.86803961e-01 -6.51604772e-01
-6.04309142e-01 4.60423291e-01 2.04870198e-03 -2.48814002e-01
9.75856185e-02 -8.73180926e-01 -3.46361607e-01 -4.02551033e-02
-3.45983744e-01 4.18463141e-01 6.79850817e-01 1.03363621e+00
1.21637676e-02 3.89843166e-01 -2.98205912e-02 -1.53547168e-01
-5.42721272e-01 6.09849691e-01 2.96402574e-01 -7.50175476e-01
-1.22645438e-01 1.39122343e+00 -1.38009772e-01 -3.02844252e-02
4.96718407e-01 -4.69286114e-01 5.66753030e-01 3.33484374e-02
3.95285606e-01 7.06832707e-01 -1.19178191e-01 1.09460987e-01
-1.92866921e-01 1.97089240e-01 -1.84233502e-01 -6.37431026e-01
1.38249791e+00 1.08140171e-01 6.38041273e-02 2.14752883e-01
9.69260871e-01 -1.74370110e-01 -1.65496445e+00 5.42937368e-02
-3.30640256e-01 -2.66885817e-01 -3.19998294e-01 -1.10752892e+00
-1.02301407e+00 8.53204012e-01 6.88806534e-01 4.22852635e-01
1.11728716e+00 -3.04143000e-02 7.45228648e-01 1.70388639e-01
8.99158478e-01 -1.83415866e+00 2.07177609e-01 6.71851397e-01
4.19313341e-01 -8.18368435e-01 -1.37476891e-01 3.12937409e-01
-9.57272768e-01 7.62963057e-01 4.12333161e-01 -1.61779359e-01
4.33763117e-01 2.04825833e-01 -7.05236316e-01 9.64453369e-02
-4.13670957e-01 -4.37897146e-01 -2.96804547e-01 -1.19487315e-01
3.78741890e-01 4.32818264e-01 -8.48529816e-01 1.23798442e+00
-4.26401496e-01 2.86271237e-02 2.98139334e-01 1.31878316e+00
-7.51985073e-01 -1.23473334e+00 -1.53571278e-01 6.45334005e-01
-8.29753518e-01 8.05375874e-02 2.31131271e-01 7.29387641e-01
1.91441134e-01 9.53715086e-01 4.93302904e-02 1.93463452e-02
3.10743749e-01 8.12379420e-02 7.45399594e-01 -1.12873100e-01
-6.65813029e-01 5.37269711e-01 -2.79839076e-02 -3.83103192e-01
1.71068728e-01 -5.83255470e-01 -1.89650035e+00 -5.83578646e-01
-1.27090946e-01 5.82177997e-01 6.68459654e-01 9.72682297e-01
1.36213452e-02 1.72343060e-01 1.02551854e+00 -4.22488302e-01
-1.13539863e+00 -5.99076867e-01 -1.14554358e+00 1.39739454e-01
1.30362827e-02 -6.79421425e-01 -4.54556584e-01 -7.56523609e-01] | [4.43297815322876, 3.058152914047241] |
a8e45b60-b9d9-44c4-b748-12059a1e49aa | supervised-dimensionality-reduction-and-1 | 2208.12152 | null | https://arxiv.org/abs/2208.12152v4 | https://arxiv.org/pdf/2208.12152v4.pdf | Supervised Dimensionality Reduction and Image Classification Utilizing Convolutional Autoencoders | The joint optimization of the reconstruction and classification error is a hard non convex problem, especially when a non linear mapping is utilized. In order to overcome this obstacle, a novel optimization strategy is proposed, in which a Convolutional Autoencoder for dimensionality reduction and a classifier composed by a Fully Connected Network, are combined to simultaneously produce supervised dimensionality reduction and predictions. It turned out that this methodology can also be greatly beneficial in enforcing explainability of deep learning architectures. Additionally, the resulting Latent Space, optimized for the classification task, can be utilized to improve traditional, interpretable classification algorithms. The experimental results, showed that the proposed methodology achieved competitive results against the state of the art deep learning methods, while being much more efficient in terms of parameter count. Finally, it was empirically justified that the proposed methodology introduces advanced explainability regarding, not only the data structure through the produced latent space, but also about the classification behaviour. | ['Spiros V. Georgakopoulos', 'Vassilis P. Plagianakos', 'Sotiris K. Tasoulis', 'Ioannis A. Nellas'] | 2022-08-25 | null | null | null | null | ['supervised-dimensionality-reduction', 'classification'] | ['computer-vision', 'methodology'] | [-3.48674692e-02 4.73640978e-01 -6.36998862e-02 -4.31516290e-01
1.62638584e-03 -7.73449615e-02 7.86734998e-01 -6.82864189e-02
-2.72900760e-01 7.74830580e-01 -1.58822220e-02 -1.31165370e-01
-7.31445789e-01 -7.10749388e-01 -5.36567748e-01 -8.98905635e-01
2.17613176e-01 4.72626060e-01 -6.42382145e-01 2.79065043e-01
3.06715041e-01 7.85273194e-01 -1.96025956e+00 5.78418486e-02
9.47916389e-01 1.10463393e+00 2.83164412e-01 1.63188815e-01
-1.87782466e-01 7.68038869e-01 -2.44701281e-01 -3.09525877e-01
4.94413227e-02 -2.13081539e-01 -7.30324745e-01 3.65319461e-01
-4.35050018e-02 -2.21578896e-01 -3.47543927e-03 8.71845007e-01
2.53724400e-02 1.82494894e-01 9.04786944e-01 -9.92959678e-01
-6.59676313e-01 5.23763955e-01 -2.03280728e-02 -1.75287798e-01
-9.55714583e-02 -5.95060438e-02 1.00075901e+00 -9.82425511e-01
3.96623760e-01 1.04656410e+00 4.21276778e-01 3.76430362e-01
-1.41409767e+00 -4.65470195e-01 -3.19066793e-02 2.74500340e-01
-1.16364419e+00 -2.36188158e-01 1.11402285e+00 -6.59326613e-01
9.98440802e-01 2.39251509e-01 7.98954070e-01 1.15269542e+00
4.11225855e-01 4.26102459e-01 1.10515416e+00 -6.63330138e-01
5.08867562e-01 6.65064216e-01 3.47520739e-01 7.11628020e-01
4.20484066e-01 1.87392652e-01 -2.34240592e-01 7.49332011e-02
5.49341619e-01 8.99630338e-02 -1.61064714e-01 -8.07830453e-01
-7.56263375e-01 9.91993070e-01 4.14571613e-01 7.19451904e-01
-7.81689882e-01 -2.99126785e-02 1.97480485e-01 1.51244968e-01
6.50885761e-01 7.38571227e-01 -4.97880191e-01 -7.64886895e-03
-8.69223833e-01 1.07639832e-02 7.36955822e-01 4.60690975e-01
9.37239885e-01 3.68645966e-01 2.68558472e-01 4.48479801e-01
6.41633093e-01 2.93121091e-03 6.60524845e-01 -7.00212717e-01
3.40279698e-01 1.10619664e+00 -1.29807279e-01 -1.40752435e+00
-5.88616133e-01 -8.71740997e-01 -1.09138024e+00 4.65442598e-01
1.54811576e-01 5.54299243e-02 -7.10556448e-01 1.55670393e+00
2.48203799e-01 -2.73794591e-01 3.12823802e-01 9.05917108e-01
3.55500102e-01 5.69340587e-01 2.15297341e-01 -2.25927800e-01
1.17649209e+00 -7.28463173e-01 -1.02934015e+00 -3.93457990e-03
6.44970477e-01 -6.02918565e-01 1.05137277e+00 5.57106614e-01
-9.68094289e-01 -8.56709838e-01 -1.21329010e+00 -1.04902834e-02
-4.80483264e-01 7.96943128e-01 8.57716501e-01 4.96468782e-01
-8.54086637e-01 8.56914520e-01 -1.00988483e+00 -2.13450238e-01
3.40830654e-01 5.63957274e-01 -4.89292055e-01 3.65148872e-01
-9.77556944e-01 1.09073913e+00 8.98752928e-01 5.36092103e-01
-7.51576185e-01 -3.68345350e-01 -7.02859640e-01 4.80600893e-01
-8.71821642e-02 -6.90277517e-01 4.85785902e-01 -1.08492577e+00
-1.61200285e+00 7.05787957e-01 1.16880447e-01 -6.79688573e-01
5.58438659e-01 -3.44108731e-01 -2.67193139e-01 -1.97071414e-02
-3.67254674e-01 4.94270533e-01 1.04304743e+00 -1.08872962e+00
-2.72027612e-01 -4.39637303e-01 -9.23761651e-02 7.66282231e-02
-9.31243658e-01 -5.76626599e-01 4.99170981e-02 -4.57459003e-01
3.79901528e-01 -8.71506989e-01 -1.12478659e-01 -1.00107141e-01
-3.67536873e-01 -2.89751977e-01 8.96593571e-01 -6.48374796e-01
1.17206311e+00 -2.12090445e+00 6.88372314e-01 4.76982296e-01
2.99224854e-01 2.78157920e-01 1.63179144e-01 2.78788239e-01
-3.92067492e-01 3.66652608e-02 -2.76531041e-01 -5.99763513e-01
1.00099586e-01 3.70772660e-01 -2.23285839e-01 5.90454996e-01
2.38787428e-01 5.17854571e-01 -2.84735978e-01 -1.82905331e-01
5.70765913e-01 8.05233657e-01 -5.94418883e-01 3.07819337e-01
-1.66944697e-01 3.57465863e-01 -4.89721268e-01 1.05738133e-01
5.46621799e-01 -3.48395228e-01 2.38089815e-01 -8.12965557e-02
-1.73900619e-01 -3.46422307e-02 -1.15016615e+00 1.58460641e+00
-6.02219880e-01 7.77690172e-01 -2.39272490e-01 -1.39004779e+00
1.52672267e+00 3.91001672e-01 6.17218435e-01 -3.88689846e-01
3.54852706e-01 3.45750719e-01 -9.94968861e-02 -5.88388741e-01
4.37725991e-01 2.06486378e-02 3.48347008e-01 4.30412084e-01
9.69707146e-02 2.74771601e-01 -1.08447865e-01 -4.03596401e-01
5.67041576e-01 2.74695605e-01 4.30253714e-01 -4.65546548e-01
7.98371494e-01 -4.78182593e-03 2.53702641e-01 2.86402673e-01
3.56126815e-01 7.04753771e-02 4.72757280e-01 -9.03671026e-01
-1.28034186e+00 -6.56105638e-01 -3.23451638e-01 4.34638798e-01
-1.03842661e-01 4.91917543e-02 -1.00355649e+00 -5.56621253e-01
-1.72765210e-01 8.38912606e-01 -6.57584548e-01 -3.56427282e-01
-3.84444296e-01 -6.81546986e-01 2.35145316e-01 2.43478432e-01
5.68733275e-01 -1.12437499e+00 -7.86919653e-01 1.72192037e-01
1.72325492e-01 -8.70112002e-01 4.19961303e-01 4.70667481e-01
-1.37730050e+00 -9.24565852e-01 -3.26494217e-01 -5.42933285e-01
8.13415051e-01 -1.99991465e-01 8.15384924e-01 1.79123133e-01
-1.37891382e-01 5.45472652e-02 -3.12175572e-01 -2.57583827e-01
-5.40133238e-01 4.05074626e-01 1.85062885e-01 3.80220085e-01
5.71079552e-01 -7.57463932e-01 -4.72574145e-01 6.22079968e-02
-9.76920307e-01 1.81578726e-01 8.85273039e-01 1.03632414e+00
5.26213408e-01 3.62246305e-01 5.11977255e-01 -6.55312181e-01
5.59118867e-01 -4.13937449e-01 -6.55334115e-01 9.88778397e-02
-1.17082417e+00 6.31750047e-01 8.74746561e-01 -1.94466457e-01
-1.23590004e+00 3.04203808e-01 -7.11856857e-02 -3.78663063e-01
-4.67132747e-01 4.49711978e-01 -3.02473605e-01 -6.23759143e-02
7.34634340e-01 2.98640221e-01 2.56928355e-01 -8.96224737e-01
3.08111578e-01 7.26352990e-01 7.65370354e-02 -1.77030742e-01
6.36257946e-01 4.76738513e-01 2.43952855e-01 -7.20391512e-01
-4.77153629e-01 -1.49175242e-01 -8.87755752e-01 -1.12499833e-01
1.06574988e+00 -6.68640196e-01 -5.63339114e-01 1.92125618e-01
-1.43061924e+00 1.93580613e-01 -2.81295270e-01 8.86360943e-01
-6.62940025e-01 7.00567737e-02 -2.72142947e-01 -7.83222198e-01
-2.62751162e-01 -1.23463023e+00 6.14800930e-01 9.61733013e-02
-3.56853634e-01 -1.39191699e+00 -1.73501730e-01 4.58596647e-01
3.39420795e-01 2.33787388e-01 1.33638656e+00 -8.88163626e-01
-6.41826570e-01 -2.48705849e-01 -1.73512727e-01 7.19336748e-01
5.15217036e-02 -1.42092392e-01 -1.14795613e+00 -1.75193191e-01
4.09793764e-01 -1.16416909e-01 6.92594469e-01 1.98915139e-01
1.38551259e+00 -3.75412166e-01 -8.49764720e-02 5.78283668e-01
1.41913903e+00 1.35666177e-01 5.69036782e-01 5.14224946e-01
5.69322467e-01 7.71158397e-01 4.80889738e-01 6.51796937e-01
4.47589271e-02 7.59675205e-01 7.25612521e-01 -3.23905468e-01
-1.19522095e-01 -3.41744125e-02 6.91156313e-02 9.93470311e-01
-2.39979118e-01 -1.46372303e-01 -8.37565839e-01 1.02191068e-01
-1.76719391e+00 -7.06532300e-01 -2.65984297e-01 1.91102231e+00
2.42439970e-01 5.95956445e-02 -4.38242108e-01 6.67642295e-01
4.54383522e-01 -1.92971975e-02 -3.64383131e-01 -6.66690886e-01
-4.27982174e-02 1.37110218e-01 1.23604752e-01 4.84568119e-01
-1.03146493e+00 6.91689909e-01 5.94836998e+00 6.15867794e-01
-1.27017581e+00 2.31244080e-02 4.60071623e-01 2.54508466e-01
-5.19179881e-01 -6.60734028e-02 -6.38364971e-01 4.37758982e-01
8.60089302e-01 2.37332612e-01 4.21328366e-01 1.14361405e+00
4.16210443e-01 1.63706079e-01 -1.13860404e+00 9.09224868e-01
5.58748133e-02 -1.31882787e+00 1.68469593e-01 4.17723626e-01
5.60166419e-01 -5.59099376e-01 2.97571838e-01 1.75394878e-01
-4.06622469e-01 -1.04260194e+00 5.12617528e-01 9.56659615e-01
1.61962137e-01 -9.56027329e-01 9.21454370e-01 5.76685548e-01
-8.01683307e-01 -2.57475436e-01 -5.37557006e-01 -3.92947465e-01
-4.17048097e-01 6.35788620e-01 -8.23966384e-01 5.92209458e-01
4.62893009e-01 6.99785054e-01 -5.83926260e-01 6.90082371e-01
-2.73120403e-01 3.81843239e-01 -1.46052931e-02 -1.91410318e-01
3.44598442e-01 -2.73528337e-01 5.76544702e-01 8.62842023e-01
4.15820986e-01 -3.23711872e-01 -2.36652151e-01 1.25035787e+00
2.07914576e-01 1.56562373e-01 -7.64458895e-01 -1.13932211e-02
1.13199897e-01 1.27983999e+00 -6.87057972e-01 -1.46625906e-01
-3.45504098e-02 9.01238680e-01 4.31004256e-01 3.63541692e-01
-8.47478569e-01 -1.28490418e-01 4.93025661e-01 2.75395531e-02
3.42864156e-01 -1.88052371e-01 -6.49790943e-01 -1.27525258e+00
2.15982527e-01 -6.67511582e-01 3.36873271e-02 -5.39544642e-01
-9.59206104e-01 8.86625886e-01 -1.86858140e-03 -1.26133335e+00
-5.05863249e-01 -8.97002935e-01 -1.57503828e-01 8.35023284e-01
-1.43163168e+00 -1.08510351e+00 -3.29529583e-01 4.01228756e-01
5.12894332e-01 -6.07601881e-01 1.17748308e+00 3.58741641e-01
-7.53611803e-01 3.49777788e-01 3.07089984e-01 -4.35775012e-01
-1.00392029e-01 -1.14320254e+00 -3.14327031e-01 7.69398868e-01
2.64364451e-01 5.81684113e-01 7.47222066e-01 -3.46700877e-01
-1.33846045e+00 -8.89219522e-01 9.69167411e-01 -7.35304579e-02
4.49969739e-01 -2.96669096e-01 -1.02504134e+00 3.56338799e-01
5.41937761e-02 -2.31993183e-01 6.71850443e-01 2.72290587e-01
-4.39989045e-02 -2.05762506e-01 -1.02097893e+00 4.03763801e-01
5.73790014e-01 -5.46282530e-01 -7.87660003e-01 1.97859868e-01
5.08822560e-01 -6.34257421e-02 -1.03961051e+00 2.43812472e-01
6.05963647e-01 -1.14573431e+00 7.61926115e-01 -6.47453487e-01
6.96005821e-01 -6.73318803e-02 -4.65566777e-02 -1.17855871e+00
-3.62053335e-01 -1.23509020e-01 -5.42135119e-01 1.21699274e+00
4.43370104e-01 -6.13640368e-01 1.07769907e+00 7.38266706e-01
-2.18298182e-01 -9.78693128e-01 -9.77946341e-01 -7.14765251e-01
-1.82149738e-01 -3.13165039e-01 4.79631871e-01 9.28175867e-01
-3.97708058e-01 1.89629346e-01 -6.03553951e-01 2.52940357e-01
6.18067503e-01 9.76209268e-02 7.20834672e-01 -1.54586363e+00
-3.78271639e-01 -4.04299557e-01 -7.34831631e-01 -6.61185861e-01
4.26591069e-01 -8.41847539e-01 -2.88318843e-01 -1.18555522e+00
-6.18543662e-02 -3.57528925e-01 -5.58866203e-01 3.54098380e-01
2.28049785e-01 1.75097510e-02 1.04514383e-01 5.53762138e-01
-6.19098544e-02 7.55051076e-01 1.17672431e+00 7.35712275e-02
-3.35752875e-01 1.01001918e-01 -4.50426847e-01 9.09908175e-01
7.86638558e-01 -6.59143269e-01 -6.29294455e-01 -4.78643537e-01
1.25140056e-01 -4.52018529e-02 4.12138820e-01 -1.24669516e+00
1.07373409e-01 2.38831908e-01 5.65135658e-01 -5.42206883e-01
5.69405377e-01 -1.34772253e+00 3.59076619e-01 7.10349560e-01
-5.01375616e-01 -1.95507362e-01 4.62451316e-02 5.37144482e-01
-4.61476803e-01 -4.20748413e-01 6.82581246e-01 2.46091709e-01
-5.70506394e-01 3.03363032e-03 -2.57655591e-01 -8.03932667e-01
1.21273553e+00 -4.35963809e-01 1.38329804e-01 2.81116506e-03
-9.04163659e-01 -2.87228197e-01 1.46936059e-01 2.51925856e-01
6.31960869e-01 -1.35975897e+00 -3.25379848e-01 4.61547434e-01
-9.73986164e-02 -2.98431933e-01 3.98182273e-01 7.55559206e-01
-5.92798531e-01 7.77439356e-01 -5.43618023e-01 -5.24376512e-01
-1.04483771e+00 5.77152789e-01 2.17223585e-01 -4.26092267e-01
-6.35482490e-01 4.45217758e-01 -3.09494697e-02 -4.42476600e-01
4.46589112e-01 -4.70345080e-01 -4.86679286e-01 7.69220963e-02
1.68341935e-01 4.30499107e-01 1.57671198e-01 -5.63734412e-01
-1.48778424e-01 5.35533488e-01 9.71578658e-02 1.96334675e-01
1.52616036e+00 -2.82920212e-01 -2.03480318e-01 3.82137865e-01
1.31979811e+00 -4.24820483e-01 -1.13726413e+00 -1.37461806e-02
1.79552689e-01 -2.58202493e-01 3.84976447e-01 -5.37175417e-01
-8.48747373e-01 1.06386161e+00 7.32482553e-01 5.50075531e-01
1.35552800e+00 -3.03679258e-01 3.16070378e-01 6.04658782e-01
6.86534643e-02 -1.02786326e+00 3.11406013e-02 2.39078358e-01
9.38540220e-01 -1.25831056e+00 1.12089284e-01 -1.66669711e-01
-3.00531060e-01 1.69218993e+00 4.64177281e-01 -2.44278267e-01
5.66261351e-01 -7.02263713e-02 -2.55375385e-01 -3.41907948e-01
-5.74778318e-01 1.83899134e-01 5.53467989e-01 2.14966774e-01
4.18129414e-01 5.48514090e-02 -5.50778806e-01 6.36591673e-01
-2.80587643e-01 3.64184491e-02 1.49856612e-01 3.88837785e-01
-3.89752537e-01 -1.11708951e+00 -3.78926039e-01 2.28492588e-01
-2.79759556e-01 2.19666183e-01 -2.62298405e-01 1.09374058e+00
4.99690771e-01 6.71607196e-01 6.63576052e-02 -4.91951197e-01
1.61472633e-01 3.73088159e-02 1.64529420e-02 -2.74380296e-01
-4.53791022e-01 -2.34671086e-01 -1.26328319e-01 -6.15895331e-01
-4.21048224e-01 -3.65222484e-01 -1.04157770e+00 -1.27299592e-01
-4.80123371e-01 3.66153449e-01 1.25060940e+00 1.27608907e+00
3.10916871e-01 6.55750930e-01 8.50099862e-01 -6.91683948e-01
-8.54093075e-01 -9.51065123e-01 -4.65509862e-01 3.21987391e-01
2.02643603e-01 -8.62777233e-01 -5.76787829e-01 2.68286243e-02] | [8.647355079650879, 3.325664758682251] |
230535dd-b5fb-4bcc-a045-045c8b19c8ef | deep-multitask-learning-for-semantic | 1704.06855 | null | http://arxiv.org/abs/1704.06855v2 | http://arxiv.org/pdf/1704.06855v2.pdf | Deep Multitask Learning for Semantic Dependency Parsing | We present a deep neural architecture that parses sentences into three
semantic dependency graph formalisms. By using efficient, nearly arc-factored
inference and a bidirectional-LSTM composed with a multi-layer perceptron, our
base system is able to significantly improve the state of the art for semantic
dependency parsing, without using hand-engineered features or syntax. We then
explore two multitask learning approaches---one that shares parameters across
formalisms, and one that uses higher-order structures to predict the graphs
jointly. We find that both approaches improve performance across formalisms on
average, achieving a new state of the art. Our code is open-source and
available at https://github.com/Noahs-ARK/NeurboParser. | ['Sam Thomson', 'Noah A. Smith', 'Hao Peng'] | 2017-04-22 | deep-multitask-learning-for-semantic-1 | https://aclanthology.org/P17-1186 | https://aclanthology.org/P17-1186.pdf | acl-2017-7 | ['semantic-dependency-parsing'] | ['natural-language-processing'] | [ 2.85889432e-02 6.29321516e-01 -1.61806270e-01 -8.00449073e-01
-1.00313938e+00 -6.43658876e-01 3.69659513e-01 2.67736644e-01
-4.07444984e-01 7.44819045e-01 3.65207344e-01 -9.32376206e-01
5.86812794e-02 -8.45511317e-01 -7.94796109e-01 -3.16406041e-01
4.51091938e-02 8.26827049e-01 2.29267910e-01 -4.01864856e-01
6.48065954e-02 2.12041348e-01 -8.23113680e-01 5.36330879e-01
5.03974497e-01 7.63815880e-01 3.44430596e-01 1.08229768e+00
-4.29285496e-01 1.11763358e+00 -3.99336457e-01 -8.19756746e-01
-1.38308972e-01 -1.92285046e-01 -1.47342503e+00 -6.86109602e-01
3.71707708e-01 -7.10343495e-02 -4.25396711e-01 9.58067894e-01
1.42501205e-01 5.26988320e-02 2.22883463e-01 -8.84875953e-01
-9.35498655e-01 1.01172888e+00 -6.92201406e-02 3.62705320e-01
2.37324104e-01 -1.83789834e-01 1.49287558e+00 -3.81054223e-01
5.77201426e-01 1.46067882e+00 6.85269892e-01 7.06433415e-01
-1.21113789e+00 -5.31699181e-01 4.12993491e-01 8.05787295e-02
-8.44352543e-01 -5.40842235e-01 8.47074866e-01 -2.75237560e-01
1.75002313e+00 -1.20826192e-01 3.48040640e-01 1.34564757e+00
4.03259635e-01 8.08580637e-01 1.04641998e+00 -4.55741554e-01
-1.69970505e-02 -3.15580249e-01 7.35163867e-01 1.32872260e+00
1.23984545e-01 -1.14996970e-01 -5.01023412e-01 -9.69355553e-02
4.65940863e-01 -4.18716818e-01 2.01965630e-01 9.14113671e-02
-8.27099025e-01 9.85017836e-01 4.49578494e-01 5.44638991e-01
4.42449749e-02 4.10776198e-01 4.06509101e-01 3.78594428e-01
5.96904457e-01 2.95550972e-01 -1.04367065e+00 -4.11314182e-02
-5.05299270e-01 1.40398487e-01 1.31534219e+00 9.93175030e-01
7.57188618e-01 -1.08271576e-01 3.45752388e-02 7.43953288e-01
4.65020508e-01 1.64810166e-01 9.57363173e-02 -9.42050040e-01
7.39945948e-01 4.64669853e-01 -2.98740417e-01 -6.39130056e-01
-8.66141677e-01 -9.44289938e-02 -4.32492763e-01 -1.22397006e-01
6.35541737e-01 -4.13984448e-01 -1.02572691e+00 1.88548791e+00
-1.32557765e-01 -1.33854836e-01 2.32083440e-01 6.35905087e-01
1.11524880e+00 4.43801492e-01 6.79268301e-01 3.02117020e-01
1.66229069e+00 -1.16180396e+00 -6.72776461e-01 -7.13532150e-01
9.62458909e-01 -3.20765704e-01 8.55157971e-01 1.10310704e-01
-1.27027524e+00 -3.85247976e-01 -1.05525053e+00 -6.96946800e-01
-6.56401157e-01 -1.05118833e-01 1.23276484e+00 5.56342304e-01
-1.40884244e+00 8.64131272e-01 -1.14719331e+00 -4.67974603e-01
4.06596392e-01 4.64491814e-01 -5.52709997e-01 6.25220314e-02
-1.39111352e+00 1.17475867e+00 7.23289013e-01 -3.10190096e-02
-6.59022510e-01 -4.61257160e-01 -1.23553705e+00 2.75196314e-01
1.06536500e-01 -1.03498173e+00 1.48494172e+00 -7.42570937e-01
-1.64388824e+00 1.01130271e+00 -4.93323982e-01 -3.52055043e-01
-9.32829902e-02 -3.16012114e-01 -3.56544524e-01 -8.33148137e-02
1.59439221e-02 6.53785169e-01 3.07889193e-01 -9.33594227e-01
-5.32211483e-01 -3.78520668e-01 3.45612943e-01 3.13206576e-02
-1.69715405e-01 4.71093595e-01 -4.45384443e-01 -3.25061381e-01
1.22055188e-01 -9.37726021e-01 -3.35597843e-01 -5.17828703e-01
-6.53237939e-01 -5.64417601e-01 4.21238869e-01 -9.28264737e-01
8.55140388e-01 -1.74019194e+00 3.33286613e-01 -2.01905966e-01
2.91044682e-01 1.36376798e-01 -3.24529171e-01 4.57326442e-01
-1.69231981e-01 4.21641678e-01 -4.24508601e-01 -7.88966656e-01
1.03818394e-01 4.93848801e-01 1.45137003e-02 3.70022580e-02
5.92249274e-01 1.25803113e+00 -9.05132174e-01 -4.29835260e-01
3.21836062e-02 4.28158581e-01 -2.96877056e-01 3.41902077e-01
-5.92283189e-01 3.75918776e-01 -4.50034708e-01 5.08929968e-01
5.80746353e-01 -6.91308737e-01 7.92633176e-01 -1.59637079e-01
8.93530175e-02 9.15373385e-01 -5.36286950e-01 2.23942733e+00
-5.46115577e-01 5.25551200e-01 1.18827403e-01 -1.20918477e+00
8.69589269e-01 3.76715183e-01 -1.37365505e-01 -6.78236663e-01
3.01300287e-01 2.07453012e-01 -8.24774504e-02 -3.40338647e-01
1.67835802e-01 -1.68143272e-01 -3.37994635e-01 2.79728651e-01
8.22189331e-01 4.67167459e-02 3.26014519e-01 2.42855996e-01
1.52767992e+00 3.54920208e-01 2.35348552e-01 -4.75558251e-01
2.01218903e-01 1.65043641e-02 6.37962222e-01 7.95591056e-01
-1.01673588e-01 1.70080647e-01 8.17418277e-01 -7.45753586e-01
-7.79296279e-01 -1.19657302e+00 6.60783798e-02 1.61456871e+00
-2.26081505e-01 -4.19278264e-01 -6.51334405e-01 -9.90180075e-01
-4.79803495e-02 9.70225990e-01 -4.79097217e-01 1.30137131e-01
-9.20559227e-01 -7.33748496e-01 8.86270106e-01 8.59270155e-01
2.23213300e-01 -1.34670258e+00 -2.46260568e-01 2.44033232e-01
-3.46075147e-01 -1.52022922e+00 1.60908476e-01 9.00804102e-01
-9.04570758e-01 -8.20695221e-01 -2.00517625e-01 -1.14656663e+00
3.29730242e-01 -4.11788166e-01 1.66209257e+00 2.49728739e-01
4.62527536e-02 -1.09918691e-01 -2.62773782e-01 -3.74025881e-01
-5.93824089e-01 5.04537821e-01 -5.20168722e-01 -6.97664380e-01
4.88935471e-01 -7.64347494e-01 -1.35220110e-01 -4.77977514e-01
-5.01965642e-01 2.77195424e-01 5.91549993e-01 7.22264588e-01
1.48813620e-01 -5.05262434e-01 3.09113026e-01 -1.32483375e+00
6.13871753e-01 -5.05989492e-01 -6.62318766e-01 2.91582286e-01
-4.20121253e-01 6.22629762e-01 7.48884082e-01 2.73236483e-01
-1.16452014e+00 1.05267748e-01 -5.69415927e-01 7.27550033e-03
-6.78176701e-01 5.49041510e-01 -2.35674419e-02 3.40351537e-02
2.76134431e-01 -1.92528263e-01 -3.87895137e-01 -6.20484948e-01
5.47713280e-01 1.65522933e-01 5.77422678e-01 -8.62621009e-01
4.03754264e-01 2.56555170e-01 1.71870753e-01 -5.88690996e-01
-1.34235871e+00 -1.54094324e-01 -8.38328421e-01 2.51150250e-01
1.42689431e+00 -7.87853658e-01 -7.83820808e-01 3.67786556e-01
-1.70780396e+00 -8.65447521e-01 3.88414770e-01 2.45301321e-01
-4.62574631e-01 3.28837097e-01 -1.39134085e+00 -3.99111480e-01
-4.50853795e-01 -9.76214409e-01 1.06259179e+00 1.28019407e-01
-8.46255347e-02 -1.51547766e+00 2.80179441e-01 5.33422709e-01
3.12748522e-01 1.38705641e-01 1.35860276e+00 -1.07499576e+00
-4.54809934e-01 2.46289939e-01 -4.73572671e-01 1.00341357e-01
-1.45857543e-01 -1.41396061e-01 -1.10869670e+00 -5.62900156e-02
-3.31789404e-01 -6.01744294e-01 1.12562823e+00 4.26657200e-01
9.45697904e-01 -1.60736680e-01 -4.32750225e-01 7.73788333e-01
1.33701193e+00 6.56441525e-02 2.73566097e-01 3.97384912e-01
8.84958982e-01 6.14194393e-01 -3.99385951e-03 -1.83109939e-02
1.07331431e+00 2.83245146e-01 4.57560033e-01 -1.24649853e-01
-9.63597670e-02 -2.08127975e-01 8.34640861e-02 1.01335418e+00
-1.29139557e-01 -5.18664360e-01 -1.19450533e+00 6.50675356e-01
-2.04004145e+00 -5.51135004e-01 -1.48630828e-01 1.30757821e+00
6.81827962e-01 3.64865869e-01 -1.98507220e-01 -3.33394259e-01
6.20448411e-01 4.03303295e-01 -2.72155076e-01 -9.78842139e-01
-1.24163747e-01 8.01389098e-01 6.82295859e-01 8.62324357e-01
-1.38555670e+00 1.63846159e+00 6.83043575e+00 2.99875259e-01
-8.33916128e-01 2.96461493e-01 5.49051106e-01 3.44030142e-01
-4.45874721e-01 3.72501016e-01 -8.40175509e-01 1.07068695e-01
1.49892938e+00 2.61323869e-01 5.76677859e-01 6.00186348e-01
-3.22798014e-01 1.35578990e-01 -9.91506517e-01 3.58686388e-01
-1.37032673e-01 -1.51273644e+00 -1.20327473e-01 -3.29824686e-01
1.24325089e-01 6.59779251e-01 -3.88649970e-01 4.34664071e-01
1.23452306e+00 -1.11161661e+00 5.55154920e-01 3.97971630e-01
2.86655396e-01 -3.82266700e-01 6.60871863e-01 1.95398167e-01
-1.22272086e+00 -3.18937153e-02 -2.82309383e-01 -3.43037337e-01
4.67051566e-01 4.99305636e-01 -3.76538545e-01 8.25041354e-01
8.65567148e-01 6.51757777e-01 -5.61850369e-01 1.23966537e-01
-7.54425228e-01 6.52498782e-01 -2.19020844e-01 -1.81382895e-01
2.09003597e-01 -6.64894879e-02 2.79023796e-01 1.67997444e+00
-8.17083195e-02 1.47670433e-01 1.37260661e-01 8.33930552e-01
-1.27207994e-01 -2.42077067e-01 -6.06499016e-01 -1.23037837e-01
4.13994730e-01 1.32138669e+00 -6.57082200e-01 -3.14881325e-01
-6.85046911e-01 9.81033623e-01 1.15173066e+00 3.95753056e-01
-7.67389894e-01 -4.66668546e-01 5.18974721e-01 -6.25739396e-01
4.68881309e-01 -5.44144452e-01 -5.13433993e-01 -1.19461024e+00
-2.49999408e-02 -4.51247185e-01 8.55461597e-01 -8.11186254e-01
-1.40095973e+00 9.43665028e-01 -5.09470738e-02 -8.50379542e-02
-3.49293917e-01 -1.09952247e+00 -5.66628277e-01 9.21701670e-01
-1.73365593e+00 -1.49732757e+00 5.29331006e-02 5.83411753e-01
2.85617620e-01 -1.50973126e-02 1.32187223e+00 1.96185157e-01
-5.84791064e-01 5.28870404e-01 -5.20808220e-01 6.11079216e-01
3.71279240e-01 -1.53470063e+00 1.21042717e+00 7.94975460e-01
3.26993197e-01 5.45618474e-01 4.82316583e-01 -4.68483955e-01
-1.39754355e+00 -8.95470321e-01 1.48625243e+00 -6.56868279e-01
1.00935304e+00 -6.89477861e-01 -9.11866903e-01 1.44116569e+00
5.71552277e-01 1.05276473e-01 6.49704337e-01 8.00676227e-01
-7.11184025e-01 1.42417222e-01 -7.89886057e-01 4.52423006e-01
1.33713210e+00 -6.05572701e-01 -8.18390012e-01 3.81587535e-01
1.06158876e+00 -5.12234211e-01 -1.13827264e+00 4.80514139e-01
3.60098511e-01 -9.55316186e-01 7.58893013e-01 -1.05993366e+00
4.75776821e-01 3.06633592e-01 -2.79231399e-01 -1.31047940e+00
-6.72108650e-01 -4.56627041e-01 -2.77483523e-01 9.48321342e-01
9.43305552e-01 -9.79500592e-01 8.08993697e-01 6.42615616e-01
-5.00205934e-01 -7.01986670e-01 -9.08665299e-01 -5.81037402e-01
4.44525212e-01 -4.27683115e-01 3.23461145e-01 1.03113556e+00
1.86470747e-01 9.43804801e-01 9.59339179e-03 4.01498735e-01
4.64190274e-01 1.65087834e-01 3.46039951e-01 -1.24762952e+00
-4.32428986e-01 -4.78595644e-01 -8.87663737e-02 -1.04330754e+00
9.28499281e-01 -1.23906267e+00 -4.90147471e-02 -1.93734288e+00
3.62904295e-02 -2.17710942e-01 -5.37570775e-01 1.13915277e+00
-1.15540162e-01 -5.17003946e-02 1.44251794e-01 -1.17923148e-01
-6.92949057e-01 1.71217427e-01 9.28093135e-01 -4.90741134e-02
6.09509014e-02 -3.38108629e-01 -8.94379854e-01 8.63806605e-01
1.25200438e+00 -7.28124261e-01 -7.55416900e-02 -1.15808594e+00
2.92485893e-01 3.50151211e-01 2.71822453e-01 -7.64067411e-01
3.24426144e-01 4.43197452e-02 1.92009300e-01 -2.54605234e-01
4.83245105e-01 -3.08177590e-01 -2.39462882e-01 2.12274745e-01
-4.52644438e-01 3.78214687e-01 5.34365535e-01 2.69481778e-01
8.52164999e-03 -1.91709682e-01 5.54898739e-01 -5.17040074e-01
-6.55493319e-01 4.54609506e-02 -3.70763987e-01 2.03303918e-01
3.39353621e-01 4.43824530e-01 -9.10277128e-01 -8.10113549e-02
-9.37015235e-01 1.48218021e-01 -4.06024531e-02 5.56144834e-01
2.02409118e-01 -7.90096581e-01 -7.23169208e-01 2.67586894e-02
-1.59074143e-01 -1.71173289e-01 2.30755135e-02 5.52414715e-01
-4.99523401e-01 6.66640818e-01 -1.38172373e-01 -2.94341415e-01
-1.01426458e+00 3.05057585e-01 2.89080888e-01 -7.57424891e-01
-5.86315453e-01 1.10004365e+00 2.17690859e-02 -9.90094900e-01
-6.22951724e-02 -4.26552892e-01 -1.85476452e-01 -4.75111246e-01
3.95485759e-02 -6.35529086e-02 2.88244188e-01 -3.75649214e-01
-5.63547075e-01 2.99221635e-01 -9.38162953e-02 -1.36374692e-02
1.38703930e+00 2.53821939e-01 -5.03048182e-01 4.21977550e-01
1.25951016e+00 -2.42878377e-01 -9.71188486e-01 -8.72402713e-02
4.43436444e-01 2.10973546e-01 2.46936008e-02 -1.00019157e+00
-1.01520503e+00 9.08729076e-01 -6.48407787e-02 3.74903619e-01
9.26503897e-01 4.87968445e-01 9.24540222e-01 8.73503447e-01
2.62714535e-01 -6.91588461e-01 -2.84372747e-01 1.10496151e+00
4.13629979e-01 -1.22079146e+00 -2.48165578e-01 -4.49672490e-01
-5.47015548e-01 1.18475592e+00 5.35916626e-01 -4.10664499e-01
8.78994882e-01 7.47969508e-01 2.68146455e-01 -4.39261794e-01
-1.09793043e+00 -4.15670633e-01 -4.31510713e-03 5.34261584e-01
7.67058551e-01 2.34013185e-01 -1.56202286e-01 8.78273129e-01
-3.37725937e-01 -3.47788870e-01 1.65409610e-01 8.99586201e-01
-2.32075125e-01 -1.38218021e+00 1.93400756e-01 3.30326766e-01
-7.04358816e-01 -5.73522627e-01 -4.42624986e-01 8.00842285e-01
-3.55225891e-01 1.11949801e+00 1.53266057e-01 -2.97443092e-01
3.01329076e-01 4.50325638e-01 6.63227081e-01 -6.57902241e-01
-6.78563654e-01 -2.92501241e-01 8.32011580e-01 -7.96540856e-01
-4.34501737e-01 -4.50959712e-01 -1.41108608e+00 -2.96677113e-01
-5.48069365e-02 6.61828667e-02 7.13387132e-01 1.19946218e+00
4.33871537e-01 7.14922667e-01 9.60147306e-02 -8.33720684e-01
-2.02241361e-01 -9.83603835e-01 -1.79830521e-01 1.17303796e-01
2.26639137e-01 -4.47393030e-01 -1.31246567e-01 -3.17632817e-02] | [10.401695251464844, 9.508073806762695] |
9f8357e5-51c6-4de0-bf20-ada854058c3e | r2d2-reliable-and-repeatable-detector-and | null | null | http://papers.nips.cc/paper/9407-r2d2-reliable-and-repeatable-detector-and-descriptor | http://papers.nips.cc/paper/9407-r2d2-reliable-and-repeatable-detector-and-descriptor.pdf | R2D2: Reliable and Repeatable Detector and Descriptor | Interest point detection and local feature description are fundamental steps in many computer vision applications. Classical approaches are based on a detect-then-describe paradigm where separate handcrafted methods are used to first identify repeatable keypoints and then represent them with a local descriptor. Neural networks trained with metric learning losses have recently caught up with these techniques, focusing on learning repeatable saliency maps for keypoint detection or learning descriptors at the detected keypoint locations. In this work, we argue that repeatable regions are not necessarily discriminative and can therefore lead to select suboptimal keypoints. Furthermore, we claim that descriptors should be learned only in regions for which matching can be performed with high confidence.
We thus propose to jointly learn keypoint detection and description together with a predictor of the local descriptor discriminativeness. This allows to avoid ambiguous areas, thus leading to reliable keypoint detection and description. Our detection-and-description approach simultaneously outputs sparse, repeatable and reliable keypoints that outperforms state-of-the-art detectors and descriptors on the HPatches dataset and on the recent Aachen Day-Night localization benchmark. | ['Martin Humenberger', 'Jerome Revaud', 'Cesar De Souza', 'Philippe Weinzaepfel'] | 2019-12-01 | null | null | null | neurips-2019-12 | ['interest-point-detection', 'camera-localization', 'image-matching'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [-3.58733423e-02 -1.59434125e-01 -4.01491106e-01 -1.05759121e-01
-1.22729063e+00 -5.90667605e-01 9.34405208e-01 7.65876889e-01
-7.01518714e-01 6.08731091e-01 -1.86985999e-01 3.29467624e-01
-3.65730941e-01 -3.94671947e-01 -7.85032272e-01 -8.08043599e-01
-2.67863691e-01 4.14756924e-01 5.93389213e-01 -3.75086186e-03
6.22405708e-01 9.13102627e-01 -1.69979668e+00 -1.79018285e-02
5.60504854e-01 1.21296263e+00 5.07594883e-01 6.42858148e-01
2.24739864e-01 6.06887281e-01 -4.74718690e-01 -2.12067347e-02
3.17460001e-01 -3.78940552e-01 -4.08580601e-01 -2.83853650e-01
8.64686728e-01 -1.31652966e-01 -2.08979517e-01 1.04758561e+00
4.99872565e-01 1.75738290e-01 6.94208205e-01 -1.26521206e+00
-3.62273693e-01 1.67560980e-01 -5.30753493e-01 3.06802630e-01
3.03386390e-01 -1.88349113e-01 1.38241029e+00 -1.28625679e+00
8.23627532e-01 8.05819392e-01 6.37197018e-01 1.94754437e-01
-1.27991509e+00 -5.07480800e-01 1.95201635e-01 5.81319571e-01
-1.84637594e+00 -6.26574278e-01 7.11584806e-01 -2.36995384e-01
7.85872161e-01 3.04401219e-01 5.14857590e-01 8.30870867e-01
2.18919203e-01 1.02234173e+00 6.33902252e-01 -4.35135514e-01
3.66836905e-01 1.97923020e-01 -1.22547127e-01 8.74581397e-01
2.90944129e-01 2.72974521e-01 -6.99913144e-01 -1.86039150e-01
7.20492184e-01 3.45171720e-01 -2.96180993e-01 -9.24984634e-01
-1.59985709e+00 9.86570776e-01 9.80910957e-01 4.45049107e-01
-6.54182494e-01 2.53689468e-01 2.85294712e-01 1.29691646e-01
2.73090005e-01 7.47864306e-01 -1.20250292e-01 1.29811153e-01
-1.32011974e+00 5.06653428e-01 5.33114731e-01 8.89037907e-01
1.10762942e+00 -3.28691423e-01 -6.18196905e-01 5.95498383e-01
4.86593284e-02 5.77621579e-01 4.32152003e-01 -6.72309279e-01
1.75283626e-01 5.01899779e-01 3.86154711e-01 -1.28216112e+00
-4.33751434e-01 -5.90533137e-01 -5.76554656e-01 2.22498685e-01
1.66308567e-01 3.56021166e-01 -1.20874763e+00 1.47995079e+00
2.56884694e-01 3.22065175e-01 -1.24498107e-01 1.04796982e+00
3.24234754e-01 4.73528713e-01 -1.14854865e-01 7.37657547e-02
1.07467842e+00 -8.04597199e-01 -3.02755445e-01 -3.13537121e-01
5.38658202e-01 -6.24061584e-01 5.78186333e-01 3.21089149e-01
-8.64322305e-01 -5.15761793e-01 -1.12279880e+00 -1.84253398e-02
-4.20885354e-01 3.92566442e-01 2.76566178e-01 1.70808643e-01
-1.08515143e+00 6.87710047e-01 -8.22976589e-01 -4.57076371e-01
3.70940924e-01 2.71589637e-01 -4.29118842e-01 5.34600541e-02
-8.60347450e-01 1.13539910e+00 6.47489846e-01 9.79536399e-02
-1.05029178e+00 -5.27983308e-01 -8.63307595e-01 3.71676125e-02
3.82875860e-01 -4.04566288e-01 9.96666133e-01 -9.42188859e-01
-1.03040147e+00 8.57027292e-01 -2.57335126e-01 -9.94222701e-01
4.60326552e-01 -3.81769091e-01 -1.72508702e-01 5.47928035e-01
4.25161064e-01 9.61976826e-01 1.25410831e+00 -1.24729025e+00
-1.31858730e+00 -9.78706926e-02 -2.20006481e-01 9.24020484e-02
4.41545807e-02 -2.16501057e-01 -3.11028898e-01 -4.49149191e-01
4.52755779e-01 -9.47714508e-01 -1.32130817e-01 3.03702950e-01
-4.66488123e-01 -4.26872700e-01 8.22530389e-01 -3.77278924e-01
7.63017416e-01 -1.96915388e+00 2.26032600e-01 3.82934183e-01
2.90034592e-01 2.43567958e-01 -1.45612851e-01 3.57657552e-01
6.70616850e-02 -3.39794606e-01 5.67274466e-02 -3.52558821e-01
1.93019137e-02 -9.67407897e-02 -4.30888772e-01 9.08684254e-01
4.47845995e-01 9.45817053e-01 -9.52094972e-01 -4.58410859e-01
5.62755764e-01 4.31284696e-01 -3.62824529e-01 6.25771210e-02
3.21257785e-02 8.30597207e-02 -6.27554536e-01 7.92563796e-01
5.40837765e-01 -3.94837186e-02 -4.98328090e-01 -2.65132934e-01
-4.04366940e-01 -1.89398210e-02 -1.25291443e+00 1.76376760e+00
-4.35712725e-01 8.96691382e-01 -1.36669070e-01 -1.11219049e+00
1.09178424e+00 -3.27937007e-02 3.33826214e-01 -8.02127242e-01
-2.17159793e-01 6.64719045e-01 -4.88510400e-01 -4.15351838e-02
6.33080840e-01 1.12194285e-01 1.15604550e-02 -1.20897949e-01
1.98664054e-01 -8.32014829e-02 1.29583791e-01 7.24060386e-02
1.09039915e+00 -7.28733316e-02 4.83224213e-01 -2.67108202e-01
6.12869799e-01 5.28746471e-03 3.93761516e-01 1.12276208e+00
-2.51096457e-01 8.99578750e-01 2.51064837e-01 -5.15864313e-01
-1.04450214e+00 -1.06061864e+00 -1.50129408e-01 9.03678775e-01
4.81973261e-01 -3.32451969e-01 -3.69381785e-01 -8.08295012e-01
9.57459658e-02 5.38155198e-01 -6.96638942e-01 -3.08655322e-01
-5.34374118e-01 9.65814218e-02 2.07196191e-01 3.21965635e-01
3.36352795e-01 -1.05293131e+00 -1.26249576e+00 2.88480818e-01
1.78520769e-01 -9.06427801e-01 -3.15982550e-01 7.34950840e-01
-4.38767523e-01 -1.09123015e+00 -1.18162167e+00 -7.95985043e-01
7.96997488e-01 4.79396790e-01 9.91967559e-01 1.29776429e-02
-4.40098703e-01 4.32886302e-01 -5.74110150e-01 -1.95236176e-01
-3.86831947e-02 2.50549853e-01 2.61470862e-02 1.31558284e-01
2.16676071e-01 -1.54893637e-01 -9.52677965e-01 2.76693374e-01
-6.31032467e-01 -2.82842994e-01 1.02583706e+00 1.10112548e+00
8.68950069e-01 -3.35562617e-01 3.90860558e-01 -2.60130912e-01
1.80861041e-01 -2.00671047e-01 -8.33437920e-01 5.15090346e-01
-5.96648097e-01 3.81533444e-01 5.68616033e-01 -1.91365287e-01
-2.80128181e-01 5.10069013e-01 4.26470749e-02 -6.70645654e-01
-2.10275054e-01 2.05800220e-01 2.82872736e-01 -5.21980166e-01
8.38076353e-01 5.33798814e-01 -3.13492984e-01 -2.78644353e-01
4.05958116e-01 4.13974553e-01 4.59829390e-01 -2.41448760e-01
9.72164273e-01 6.12664163e-01 1.99534610e-01 -8.53145063e-01
-6.03071630e-01 -9.25000429e-01 -7.42568552e-01 -3.18511538e-02
6.03641272e-01 -1.02487350e+00 -3.49928916e-01 3.57544944e-02
-1.15258789e+00 1.57434583e-01 -5.18918395e-01 5.16839921e-01
-6.80917263e-01 1.39864281e-01 -1.35641909e-04 -7.98670411e-01
-3.24177593e-01 -1.13491642e+00 1.70138395e+00 3.74098837e-01
-1.02741964e-01 -6.17303252e-01 5.34080863e-02 -5.39468467e-01
4.90765512e-01 2.11108044e-01 2.58846909e-01 -7.77039528e-01
-9.31992531e-01 -7.49028087e-01 -4.24855351e-01 9.72773954e-02
-2.87840217e-02 -2.39120498e-01 -9.74419415e-01 -5.99938750e-01
-3.39235723e-01 -2.10688621e-01 1.30669999e+00 4.69731659e-01
7.98808873e-01 -1.50270030e-01 -8.08319449e-01 6.70101881e-01
1.51822805e+00 3.76728177e-02 3.36109906e-01 4.62983251e-01
5.40000737e-01 3.66993457e-01 8.45737159e-01 3.51352304e-01
1.01987056e-01 9.98317957e-01 5.10877788e-01 -1.42062455e-01
8.25303346e-02 -4.65476722e-01 1.49763539e-01 -9.60794538e-02
3.16480488e-01 -4.81142960e-02 -8.12072396e-01 1.06640685e+00
-2.01094127e+00 -8.08746636e-01 2.80027181e-01 2.27294421e+00
4.34771657e-01 1.11031778e-01 1.55835703e-01 4.97847656e-03
8.13417435e-01 2.60665089e-01 -4.60363925e-01 1.31888539e-02
-9.70018134e-02 1.82143599e-01 9.21820998e-01 2.85547227e-01
-1.38633001e+00 9.69709635e-01 5.36455154e+00 8.93385470e-01
-1.34043169e+00 5.02612554e-02 3.22011352e-01 2.21258793e-02
-3.76747474e-02 1.33796588e-01 -1.06940150e+00 2.44965300e-01
6.80120587e-01 -4.33883704e-02 2.81873029e-02 1.04679167e+00
-6.62906691e-02 -3.89027268e-01 -1.18106639e+00 1.25092125e+00
2.07483739e-01 -1.44410717e+00 2.94769313e-02 -1.19283542e-01
6.59341216e-01 1.89722940e-01 1.86675727e-01 3.89447738e-03
-1.58555895e-01 -8.24356496e-01 9.95050251e-01 6.92693889e-01
3.30365777e-01 -9.07896042e-01 6.08567953e-01 2.12477833e-01
-1.27234709e+00 -2.36723825e-01 -6.63484991e-01 3.80480886e-01
8.43382906e-03 4.38640416e-01 -1.02219915e+00 4.63671714e-01
5.46735704e-01 8.89214039e-01 -6.64451420e-01 1.71482098e+00
-4.66530114e-01 1.50262594e-01 -4.46078062e-01 -3.42812419e-01
5.96345186e-01 1.88392371e-01 8.31910193e-01 1.24536383e+00
6.80383205e-01 -4.82594341e-01 3.08839947e-01 1.06912422e+00
2.44781360e-01 1.81888223e-01 -5.72064698e-01 1.82199612e-01
6.12304807e-01 1.26286125e+00 -1.08977520e+00 -1.85544118e-01
-1.24890834e-01 1.23231626e+00 4.25117135e-01 2.03011185e-01
-5.56320071e-01 -6.40510499e-01 5.51395953e-01 1.64898291e-01
7.88950741e-01 -3.56725186e-01 3.23557824e-01 -9.99936402e-01
6.26530051e-02 -3.62376571e-01 3.08639616e-01 -5.24991214e-01
-9.14249480e-01 4.71011430e-01 -1.99299324e-02 -1.41857660e+00
-4.17674571e-01 -5.07938921e-01 -5.59585452e-01 7.67504096e-01
-1.86447299e+00 -1.34261429e+00 -3.39712441e-01 5.31597555e-01
5.06693482e-01 -9.76453051e-02 6.83180511e-01 3.95220928e-02
-1.45713106e-01 4.83037144e-01 3.85657787e-01 -2.32463982e-02
7.87721813e-01 -1.40507865e+00 3.89034212e-01 9.68178689e-01
7.76144087e-01 5.23127496e-01 8.81788075e-01 -3.42253774e-01
-1.24402666e+00 -1.00075197e+00 8.33827138e-01 -3.82967353e-01
5.57975888e-01 -5.14036179e-01 -8.19870532e-01 2.85381436e-01
-1.62911922e-01 4.95639771e-01 -2.15722159e-01 -1.00460455e-01
-1.58411235e-01 -4.14927989e-01 -9.11880553e-01 4.60878521e-01
6.27250075e-01 -6.85268998e-01 -5.01227617e-01 4.33977902e-01
3.71584207e-01 -2.89945930e-01 -2.38611534e-01 1.64775893e-01
4.37999487e-01 -9.86043036e-01 1.13194191e+00 -4.02696766e-02
-2.13134646e-01 -4.82943565e-01 -1.27210602e-01 -1.28783536e+00
-3.33288819e-01 -5.05055249e-01 -1.29770758e-02 8.28281581e-01
2.42683634e-01 -5.13821185e-01 7.62469292e-01 -2.27239560e-02
-1.12170810e-02 -4.29260105e-01 -1.37714076e+00 -9.66897249e-01
-2.76400387e-01 -3.41937356e-02 2.75705278e-01 5.00341833e-01
-2.54449904e-01 5.34924604e-02 -4.17063266e-01 2.44818449e-01
7.02343524e-01 3.28004330e-01 6.29809141e-01 -1.26823390e+00
9.65631902e-02 -4.92314249e-01 -1.10209000e+00 -1.03606153e+00
2.84584105e-01 -7.68131077e-01 4.15703505e-01 -1.33419442e+00
6.08695634e-02 -4.00365680e-01 -7.06496775e-01 4.54446226e-01
-4.18113507e-02 3.23033661e-01 1.85078517e-01 4.00327951e-01
-1.00220823e+00 4.44290161e-01 6.91627324e-01 -2.56054670e-01
-3.00754607e-01 5.18436264e-03 -3.91933531e-01 3.04400563e-01
4.24725950e-01 -5.66188812e-01 -9.26546678e-02 9.13382024e-02
1.95365012e-01 -2.09076643e-01 1.11279845e+00 -1.33814299e+00
6.63680136e-01 1.50470063e-01 5.12106180e-01 -8.11051130e-01
3.49650443e-01 -8.46598864e-01 -3.33516717e-01 5.21749735e-01
-4.64784294e-01 -7.54141584e-02 3.39109302e-02 8.04315150e-01
-3.58612001e-01 -3.09912115e-01 9.43473041e-01 7.59900175e-03
-1.35832644e+00 2.38166630e-01 1.44341305e-01 -2.40709424e-01
1.24262547e+00 -2.88806409e-01 3.48066702e-03 -3.39471996e-01
-5.42648494e-01 7.92104751e-03 4.01520818e-01 4.65611488e-01
9.25653279e-01 -1.26289403e+00 -7.82592595e-01 3.46781135e-01
5.64238071e-01 -1.93692699e-01 -4.24845703e-02 8.85499001e-01
-4.74307954e-01 7.41365433e-01 -1.53267041e-01 -9.63338137e-01
-1.16777849e+00 6.99504554e-01 5.03331542e-01 1.58900350e-01
-7.72867680e-01 9.99533355e-01 8.71344581e-02 1.58217803e-01
1.99817896e-01 -5.85776806e-01 -1.58048980e-02 3.27830970e-01
4.59765434e-01 1.05476394e-01 2.01642528e-01 -8.79280686e-01
-7.05661416e-01 8.92832577e-01 -2.35575274e-01 1.27381384e-01
1.22275436e+00 -1.88247129e-01 2.18115956e-01 3.41118455e-01
1.42578328e+00 -1.08081527e-01 -1.36537540e+00 -4.37112093e-01
4.53752935e-01 -5.28449833e-01 3.56480092e-01 -5.31160295e-01
-7.16983676e-01 7.37578273e-01 1.07504559e+00 1.27074271e-01
8.72887194e-01 1.49088338e-01 5.08129478e-01 6.28531039e-01
5.82277179e-01 -1.18151033e+00 -5.20331785e-02 3.08669150e-01
9.89422798e-01 -1.52018344e+00 1.24503322e-01 6.73645586e-02
-3.82956654e-01 1.12426949e+00 9.42952931e-02 -4.65215474e-01
3.63047808e-01 -9.43538770e-02 -1.88397005e-01 -2.76821196e-01
-4.75216061e-01 -6.49129629e-01 7.23303735e-01 5.83182693e-01
-1.39251994e-02 -4.97220457e-02 -6.98981360e-02 -7.04010874e-02
1.29811972e-01 -3.08230668e-01 3.76644507e-02 9.88395512e-01
-8.84621382e-01 -8.45267892e-01 -4.22506124e-01 3.38520557e-01
-1.52893320e-01 -5.75147867e-02 -3.85611862e-01 9.75417972e-01
5.85936010e-02 6.30548060e-01 7.49662444e-02 -1.16462089e-01
4.28397983e-01 -1.31287202e-01 3.28302115e-01 -5.18439174e-01
-2.13020459e-01 7.98738748e-02 -3.79604876e-01 -8.43290985e-01
-2.77689815e-01 -6.78757966e-01 -1.04253066e+00 3.27578813e-01
-5.72368443e-01 4.25536245e-01 7.47207642e-01 8.38112652e-01
4.16607529e-01 1.47294626e-01 6.52201056e-01 -1.34198701e+00
-7.15136290e-01 -5.35811841e-01 -7.33873308e-01 2.37005472e-01
9.59759176e-01 -8.49942565e-01 -5.01593888e-01 -3.16638976e-01] | [7.896174430847168, -2.0176913738250732] |
e437afc4-28ea-48d3-b9bc-2d2c13aaa18b | personalized-federated-domain-adaptation-for | 2306.03191 | null | https://arxiv.org/abs/2306.03191v1 | https://arxiv.org/pdf/2306.03191v1.pdf | Personalized Federated Domain Adaptation for Item-to-Item Recommendation | Item-to-Item (I2I) recommendation is an important function in most recommendation systems, which generates replacement or complement suggestions for a particular item based on its semantic similarities to other cataloged items. Given that subsets of items in a recommendation system might be co-interacted with by the same set of customers, graph-based models, such as graph neural networks (GNNs), provide a natural framework to combine, ingest and extract valuable insights from such high-order relational interactions between cataloged items, as well as their metadata features, as has been shown in many recent studies. However, learning GNNs effectively for I2I requires ingesting a large amount of relational data, which might not always be available, especially in new, emerging market segments. To mitigate this data bottleneck, we postulate that recommendation patterns learned from existing mature market segments (with private data) could be adapted to build effective warm-start models for emerging ones. To achieve this, we propose and investigate a personalized federated modeling framework based on GNNs to summarize, assemble and adapt recommendation patterns across market segments with heterogeneous customer behaviors into effective local models. Our key contribution is a personalized graph adaptation model that bridges the gap between recent literature on federated GNNs and (non-graph) personalized federated learning, which either does not optimize for the adaptability of the federated model or is restricted to local models with homogeneous parameterization, excluding GNNs with heterogeneous local graphs. | ['Trong Nghia Hoang', 'Anoop Deoras', 'Hao Ding', 'Ziwei Fan'] | 2023-06-05 | null | null | null | null | ['personalized-federated-learning'] | ['methodology'] | [-3.74642432e-01 9.73395333e-02 -6.63790405e-01 -2.94496894e-01
2.41580233e-01 -8.34126592e-01 3.44377548e-01 3.99712771e-01
1.76518142e-01 4.67489064e-01 4.90206301e-01 -2.40778774e-01
-7.45792270e-01 -1.28469265e+00 -7.92271793e-01 -1.29216626e-01
-3.21681976e-01 8.69610727e-01 -5.87564558e-02 -6.62120879e-01
-7.66081363e-02 4.05320674e-01 -1.45181489e+00 4.22874242e-01
9.98890400e-01 9.36053157e-01 1.61648870e-01 2.64361590e-01
-4.15887684e-01 5.85951686e-01 -1.81466460e-01 -8.31102610e-01
6.99634433e-01 -2.40638062e-01 -4.48259413e-01 5.98980077e-02
3.02817166e-01 -1.85799718e-01 -6.19163632e-01 9.04930890e-01
9.23545063e-02 5.28065085e-01 5.28283000e-01 -1.10047209e+00
-1.33552372e+00 1.39015472e+00 -7.89851695e-02 4.27201390e-02
2.35061333e-01 6.23447597e-02 1.48160446e+00 -6.35903478e-01
9.15538311e-01 8.19669187e-01 5.16634166e-01 1.85188755e-01
-1.07600427e+00 -3.62588137e-01 8.15292060e-01 1.49307460e-01
-1.04084945e+00 1.75735921e-01 6.46112502e-01 -1.28061637e-01
9.74260211e-01 3.87179255e-01 1.15598273e+00 1.10717976e+00
1.79226547e-01 8.20579112e-01 2.76288033e-01 -1.35675315e-02
2.10455388e-01 2.02892214e-01 4.35703665e-01 2.38684312e-01
8.47046912e-01 -5.07426001e-02 -6.65372670e-01 -2.07815036e-01
6.57702625e-01 6.83441997e-01 -1.93705499e-01 -7.11455762e-01
-8.51676941e-01 9.47299242e-01 6.96742892e-01 3.65101486e-01
-7.60493279e-01 -2.58507729e-01 2.05935255e-01 8.05890799e-01
5.60802281e-01 8.81128371e-01 -8.02299976e-01 1.63785070e-01
-5.15307844e-01 1.53402790e-01 1.18924868e+00 1.46016669e+00
9.77549195e-01 2.28362679e-01 -2.83455420e-02 7.11783290e-01
2.71479100e-01 3.44555259e-01 5.54535627e-01 -2.84839243e-01
3.80841821e-01 1.02844250e+00 -1.87831372e-01 -9.95473981e-01
-4.93868500e-01 -1.13027048e+00 -9.27823365e-01 -5.57877958e-01
2.29978919e-01 -1.49805114e-01 -5.20276904e-01 1.54840600e+00
2.93373376e-01 1.17084995e-01 -2.14112774e-01 8.79030704e-01
7.45037913e-01 3.59470189e-01 -2.14380354e-01 -5.09325713e-02
9.40851748e-01 -1.15584350e+00 -3.32417369e-01 1.01561816e-02
7.44757414e-01 -3.98806959e-01 8.92315388e-01 5.69614828e-01
-8.94052148e-01 -6.21289313e-01 -7.94421375e-01 3.71856689e-01
-6.25750005e-01 -4.82280135e-01 1.10130143e+00 4.26787406e-01
-1.00716937e+00 8.51333618e-01 -3.15887153e-01 -6.50412738e-01
1.64250746e-01 6.32029533e-01 -2.10900277e-01 -2.37737626e-01
-1.29236436e+00 4.33652759e-01 3.28823298e-01 -1.13725826e-01
-4.08982217e-01 -8.04145038e-01 -4.49147671e-01 4.54115212e-01
8.33427668e-01 -1.20509529e+00 8.19513142e-01 -1.12971210e+00
-1.33638585e+00 -9.19339210e-02 5.50997198e-01 -5.90010822e-01
7.01490184e-03 -8.41500461e-02 -8.83148909e-01 -4.59413677e-01
-3.73288780e-01 -1.30398035e-01 6.78319871e-01 -1.21458745e+00
-7.42859304e-01 -6.28821373e-01 3.71543407e-01 2.23000690e-01
-5.87634265e-01 -4.14715976e-01 -5.24289608e-01 -6.13103807e-01
-4.48187515e-02 -9.39147830e-01 -3.23129028e-01 -7.86691546e-01
-3.65052551e-01 -3.85765195e-01 2.64876962e-01 -2.39248067e-01
1.36185181e+00 -1.87362480e+00 1.76255077e-01 8.09891403e-01
4.54672486e-01 1.84864283e-01 -4.80777949e-01 8.89876723e-01
1.36886165e-01 1.11378066e-01 7.43489921e-01 -1.39892891e-01
1.76686272e-01 3.76769930e-01 -2.08589360e-01 2.03691125e-02
-5.75911582e-01 1.23699212e+00 -9.40038562e-01 2.23754317e-01
-7.13800043e-02 -2.51871236e-02 -8.65687609e-01 2.97501892e-01
-5.77021301e-01 3.92366350e-01 -6.03385389e-01 5.05687416e-01
6.10730529e-01 -6.43686235e-01 6.59718633e-01 -3.87022674e-01
2.74596870e-01 2.37192988e-01 -1.28868186e+00 1.61162484e+00
-3.43655616e-01 -4.08252031e-01 6.41950071e-02 -1.05307269e+00
1.00579000e+00 -1.48833357e-02 8.78235459e-01 -7.91534722e-01
-9.74766165e-02 1.48985162e-01 7.87633732e-02 -2.40995139e-01
8.32735002e-01 3.66186619e-01 -5.51956892e-02 7.89370537e-01
4.49631304e-01 5.58171690e-01 3.59679490e-01 5.42484820e-01
1.15890145e+00 -1.24328732e-01 3.19178641e-01 -1.16825864e-01
1.76672339e-01 -1.47959769e-01 4.64063734e-01 1.15662241e+00
5.32920837e-01 3.05362135e-01 8.92666429e-02 -5.95890105e-01
-8.07235539e-01 -8.36948514e-01 4.65421200e-01 1.50871003e+00
2.71283329e-01 -7.58014321e-01 -1.95951432e-01 -1.08583927e+00
4.99016911e-01 7.17554927e-01 -3.92368525e-01 -4.00483936e-01
-3.93824875e-01 -5.11002779e-01 -2.25492418e-01 4.84233081e-01
-1.65392414e-01 -1.00984037e+00 4.41077381e-01 4.64195639e-01
2.49042526e-01 -7.19683647e-01 -8.84733438e-01 2.60213107e-01
-9.43257809e-01 -1.16568005e+00 -2.76073188e-01 -3.80634487e-01
5.55864334e-01 7.93298542e-01 1.49207008e+00 3.26456904e-01
4.85184193e-01 8.09747338e-01 -8.70681405e-01 -8.01796541e-02
-3.51479173e-01 6.26782358e-01 4.37906086e-01 3.28745067e-01
4.65841830e-01 -9.03963566e-01 -8.05287659e-01 5.02892435e-01
-1.01825011e+00 -1.47275060e-01 6.49574697e-01 7.92992294e-01
5.44887245e-01 7.80994445e-02 9.53963995e-01 -1.50850892e+00
7.76969790e-01 -1.13371515e+00 -3.29232007e-01 5.77153862e-01
-1.29508936e+00 5.38806990e-03 1.18806398e+00 -5.76099873e-01
-8.25494051e-01 -4.75697458e-01 4.68647517e-02 -4.35654461e-01
9.01456401e-02 1.11855471e+00 -1.28659904e-01 -6.17919713e-02
6.38267040e-01 4.58511151e-02 -3.59044820e-02 -9.45150733e-01
9.87029433e-01 4.44233268e-01 2.10037708e-01 -6.61423087e-01
7.47557640e-01 6.08653054e-02 -1.64740458e-01 -1.84556186e-01
-6.62752628e-01 -6.76113307e-01 -5.01537859e-01 -7.34097930e-03
1.11827582e-01 -8.09539497e-01 -6.56733990e-01 -2.26135249e-03
-4.00237590e-01 -3.26131523e-01 -8.59796882e-01 6.38560116e-01
-3.08685005e-01 3.31603318e-01 -6.73884630e-01 -2.71486640e-01
-5.94344199e-01 -7.47010291e-01 3.75057578e-01 2.56413519e-01
-9.30606052e-02 -1.17106080e+00 6.86627552e-02 2.44273096e-01
8.15889955e-01 -4.79121119e-01 1.21175158e+00 -1.66932142e+00
-9.02348280e-01 -2.22400337e-01 8.97361115e-02 2.50684619e-01
3.93469274e-01 -2.97954798e-01 -1.23870939e-01 -6.36417031e-01
-2.66022831e-01 9.88117754e-02 5.29383063e-01 2.30905145e-01
9.79923069e-01 -7.32049465e-01 -2.66446322e-01 7.54181027e-01
1.33693993e+00 5.24158254e-02 1.36210859e-01 2.17083797e-01
1.04305732e+00 4.28399920e-01 4.36082900e-01 5.79388320e-01
7.58693397e-01 7.67076194e-01 5.59900224e-01 1.88143566e-01
-2.26669237e-01 -6.69301689e-01 1.55047625e-01 1.18480992e+00
-3.33112597e-01 -7.25385666e-01 -2.44860992e-01 3.77670705e-01
-2.29072690e+00 -7.23735690e-01 -1.19723760e-01 2.49486828e+00
3.24207902e-01 -9.56280828e-02 4.49643970e-01 -5.79821765e-01
5.48452318e-01 -2.29163423e-01 -8.49591017e-01 -2.20217481e-01
-1.97807640e-01 2.01833338e-01 6.24056041e-01 -2.44031586e-02
-4.76921320e-01 7.66003013e-01 5.39947176e+00 7.10033596e-01
-9.17262793e-01 1.06766194e-01 3.04035574e-01 -2.80501872e-01
-8.86525095e-01 1.28920004e-01 -7.67796695e-01 5.11656880e-01
1.07020199e+00 -5.11218548e-01 1.13059473e+00 1.08869922e+00
-1.37472078e-01 6.13391101e-01 -1.29739273e+00 7.26723254e-01
7.07066432e-02 -1.57971025e+00 5.95835984e-01 4.03625578e-01
1.02236021e+00 3.82669002e-01 8.56691673e-02 7.21419036e-01
8.39987814e-01 -6.21160686e-01 3.55905145e-01 8.10145497e-01
1.88269943e-01 -7.59637117e-01 6.43513799e-01 3.63514543e-01
-1.30629945e+00 -5.35342455e-01 -5.60514092e-01 1.21710800e-01
2.15255674e-02 5.26507616e-01 -6.47277355e-01 1.36355078e+00
6.50053978e-01 9.88789797e-01 -6.59396291e-01 1.09668112e+00
1.98928371e-01 6.63613617e-01 -3.10247570e-01 6.68600260e-04
4.01501097e-02 -7.34985352e-01 4.72883642e-01 7.40682721e-01
7.71956205e-01 -1.14015944e-01 4.21372443e-01 6.43093824e-01
-3.97757113e-01 6.56590581e-01 -7.62551188e-01 6.91033527e-03
4.77887899e-01 1.55496681e+00 -4.09833968e-01 -2.96642542e-01
-9.26296234e-01 5.91653883e-01 4.65599239e-01 4.82144117e-01
-4.25639391e-01 1.97779089e-01 7.02788711e-01 5.02384603e-01
4.85909581e-01 -3.27368989e-03 6.84499070e-02 -1.58045244e+00
-2.77929217e-01 -1.32001162e+00 8.23923945e-01 -4.26498502e-01
-2.13474917e+00 5.24351299e-01 -9.40042585e-02 -1.23007858e+00
-3.56931120e-01 -3.87156278e-01 -5.08306861e-01 4.72421020e-01
-1.12369227e+00 -1.51090896e+00 -1.29477829e-01 9.69667614e-01
2.38142416e-01 -5.99473417e-01 4.31652039e-01 4.08885628e-01
-3.75920117e-01 8.87040257e-01 4.42456275e-01 -3.02397162e-01
6.73708022e-01 -1.17896640e+00 4.28857863e-01 6.63176537e-01
7.03463256e-01 1.04992163e+00 2.18038842e-01 -9.31883335e-01
-2.05163193e+00 -1.39197755e+00 4.57199156e-01 -3.08899850e-01
8.76945257e-01 -2.96463609e-01 -1.05814588e+00 1.05988169e+00
6.25697002e-02 -2.75216699e-02 8.39842021e-01 8.55468452e-01
-3.29024762e-01 -4.93660778e-01 -8.52406263e-01 5.95385551e-01
1.65183520e+00 -9.51462388e-02 -1.40875533e-01 5.35749614e-01
9.41127360e-01 -6.80016056e-02 -1.44204247e+00 2.38966376e-01
3.27044249e-01 -8.49257827e-01 8.99190605e-01 -9.28492427e-01
-3.31380963e-01 -1.46880308e-02 -1.38189107e-01 -1.71776628e+00
-9.29481030e-01 -8.31910133e-01 -6.22515202e-01 1.26861703e+00
5.74216187e-01 -1.02170432e+00 8.05184901e-01 7.27248490e-01
-3.61951977e-01 -7.50245571e-01 -4.68909919e-01 -8.58172178e-01
-2.09234282e-01 -2.70264953e-01 1.48707175e+00 8.72391701e-01
1.99801639e-01 3.50837409e-01 -5.37944794e-01 7.62010515e-02
2.87349641e-01 5.69142461e-01 1.11391270e+00 -1.59162903e+00
-8.08494627e-01 -4.64489222e-01 -1.08955652e-01 -1.32722402e+00
-8.53264630e-02 -1.53582740e+00 -7.99522817e-01 -1.57336426e+00
1.63637549e-01 -7.91794360e-01 -8.20497870e-01 4.21307474e-01
1.96704380e-02 -1.62633359e-01 2.92737126e-01 4.80868161e-01
-9.32680011e-01 4.53823090e-01 1.37847984e+00 -1.42043695e-01
-5.45465231e-01 5.26057839e-01 -1.39933157e+00 2.13020429e-01
4.61112887e-01 -1.95989653e-01 -7.61898160e-01 -2.53946304e-01
9.96010602e-01 -3.00352704e-02 -1.53405949e-01 -5.39045572e-01
4.31652725e-01 -3.18457127e-01 -2.45491453e-02 -3.49634498e-01
-2.04807833e-01 -9.92046177e-01 7.98479736e-01 -8.87889415e-02
-2.73105055e-01 2.07492024e-01 -3.55703741e-01 1.00232303e+00
-3.95262688e-02 -1.45145416e-01 -9.99303237e-02 -2.66183764e-01
-6.72890902e-01 8.94661903e-01 3.39957364e-02 -2.47755200e-01
6.75302088e-01 -1.26201719e-01 -5.56407094e-01 -6.94916427e-01
-8.95935893e-01 2.34142601e-01 5.59863448e-01 7.67033875e-01
3.13084096e-01 -1.23394322e+00 -5.10341465e-01 3.51699531e-01
3.12677920e-01 -1.74511299e-01 4.39971924e-01 5.64641774e-01
2.03298271e-01 3.96898419e-01 4.19794880e-02 -1.27318114e-01
-6.51172280e-01 1.24132597e+00 2.22482383e-02 -7.33029008e-01
-6.17415190e-01 5.60312808e-01 1.02245376e-01 -6.82988703e-01
-1.38984203e-01 -4.23428953e-01 -3.63730937e-01 -1.58772487e-02
9.16800648e-02 3.13781232e-01 3.03478599e-01 -3.82268965e-01
5.68839386e-02 1.73733801e-01 -4.15408283e-01 7.44227648e-01
1.63459635e+00 -3.70750397e-01 -1.30039945e-01 1.97469026e-01
7.07193971e-01 2.88667470e-01 -8.78036976e-01 -6.95651174e-01
-2.10096762e-01 -3.65677357e-01 -3.18052143e-01 -7.78868616e-01
-1.59499466e+00 -7.54929781e-02 1.09330982e-01 7.76060522e-01
9.80763376e-01 1.42513946e-01 9.45033193e-01 4.80263621e-01
8.52226853e-01 -9.94055092e-01 1.76210850e-02 3.29872578e-01
6.18409038e-01 -8.87067139e-01 1.14442326e-01 -2.15731457e-01
-7.16135859e-01 1.03920448e+00 7.01660275e-01 -1.61736131e-01
1.05603766e+00 -3.13235074e-01 -3.99404198e-01 -4.37500924e-01
-9.45157111e-01 -2.23445728e-01 6.04137957e-01 6.33874714e-01
1.32525086e-01 2.60870457e-01 -2.33830214e-01 1.12313747e+00
-1.68858021e-01 -5.91653734e-02 2.94495314e-01 3.71728778e-01
-9.52642690e-03 -1.50193369e+00 3.05013925e-01 1.14077806e+00
-1.06269918e-01 -1.58212617e-01 -5.16600460e-02 6.47159636e-01
1.47768602e-01 1.03530955e+00 -1.23039614e-02 -7.64271855e-01
4.94834661e-01 -3.29168551e-02 2.01204434e-01 -8.80558133e-01
-1.12008083e+00 -9.17467400e-02 1.42992288e-01 -5.81517339e-01
-1.12401426e-01 -3.99575561e-01 -7.75847197e-01 -5.52063406e-01
-5.88536263e-01 3.38846892e-01 4.38986093e-01 7.58400440e-01
1.08638275e+00 4.40764666e-01 8.59116912e-01 -5.95297575e-01
-7.56121933e-01 -8.98920417e-01 -1.16441107e+00 8.56332600e-01
-2.35021323e-01 -5.60196519e-01 -2.53509849e-01 -4.46836919e-01] | [10.218926429748535, 5.6145453453063965] |
2005444c-5735-414f-894f-60ff5189e237 | a-deep-learning-approach-for-semantic | 2201.07342 | null | https://arxiv.org/abs/2201.07342v1 | https://arxiv.org/pdf/2201.07342v1.pdf | A Deep Learning Approach for Semantic Segmentation of Unbalanced Data in Electron Tomography of Catalytic Materials | Heterogeneous catalysts possess complex surface and bulk structures, relatively poor intrinsic contrast, and often a sparse distribution of the catalytic nanoparticles (NPs), posing a significant challenge for image segmentation, including the current state-of-the-art deep learning methods. To tackle this problem, we apply a deep learning-based approach for the multi-class semantic segmentation of a $\gamma$-Alumina/Pt catalytic material in a class imbalance situation. Specifically, we used the weighted focal loss as a loss function and attached it to the U-Net's fully convolutional network architecture. We assessed the accuracy of our results using Dice similarity coefficient (DSC), recall, precision, and Hausdorff distance (HD) metrics on the overlap between the ground-truth and predicted segmentations. Our adopted U-Net model with the weighted focal loss function achieved an average DSC score of 0.96 $\pm$ 0.003 in the $\gamma$-Alumina support material and 0.84 $\pm$ 0.03 in the Pt NPs segmentation tasks. We report an average boundary-overlap error of less than 2 nm at the 90th percentile of HD for $\gamma$-Alumina and Pt NPs segmentations. The complex surface morphology of the $\gamma$-Alumina and its relation to the Pt NPs were visualized in 3D by the deep learning-assisted automatic segmentation of a large data set of high-angle annular dark-field (HAADF) scanning transmission electron microscopy (STEM) tomography reconstructions. | ['Hamish L. Fraser', 'Libor Kovarik', 'Arda Genc'] | 2022-01-18 | null | null | null | null | ['electron-tomography'] | ['medical'] | [ 4.10710782e-01 4.37925309e-01 3.62922370e-01 1.03733085e-01
-9.64632928e-01 -2.57810801e-01 3.57395172e-01 3.41677636e-01
-6.36150241e-01 8.03534746e-01 -6.42442107e-01 -2.45466605e-01
-2.06890374e-01 -1.13832998e+00 -9.52524781e-01 -1.12456596e+00
8.48336669e-04 9.48700249e-01 3.85498852e-01 -7.98528567e-02
3.16616118e-01 6.53407753e-01 -1.32468092e+00 4.88035709e-01
9.70194221e-01 1.77696109e+00 6.09985530e-01 2.63231069e-01
-1.42364264e-01 2.14308754e-01 -4.56427395e-01 -3.52343112e-01
2.53788531e-01 8.93508866e-02 -7.05024838e-01 -7.16378614e-02
1.70604900e-01 2.10522249e-01 3.46713094e-03 1.01188028e+00
3.14916700e-01 6.42158836e-02 1.34372473e+00 -7.63385177e-01
-3.77805024e-01 9.75938141e-02 -3.76361936e-01 1.80360645e-01
1.97507083e-01 2.44614020e-01 8.87922287e-01 -7.95143008e-01
7.66916454e-01 7.95646846e-01 9.46986794e-01 1.86709076e-01
-9.51147079e-01 -5.21540701e-01 -2.30942070e-01 6.45985901e-02
-1.38019538e+00 6.06804527e-02 6.86241746e-01 -9.51406002e-01
1.10887861e+00 2.67879069e-02 5.83805919e-01 6.40510440e-01
5.92209995e-01 2.95321465e-01 9.11426127e-01 -3.30698818e-01
3.04163456e-01 -1.75383389e-01 -2.80470550e-01 9.73273516e-01
3.80187541e-01 -2.49214068e-01 4.87091452e-01 2.73263678e-02
8.82477403e-01 5.13070961e-03 -5.05614281e-02 -3.28818820e-02
-7.22763896e-01 7.28420198e-01 5.93163848e-01 5.69252908e-01
-4.23713148e-01 -1.15110919e-01 3.51677001e-01 -1.66527063e-01
7.51293600e-01 6.67398989e-01 -3.93885970e-01 2.02498630e-01
-8.61599386e-01 1.94383219e-01 6.96480572e-01 6.19033515e-01
7.32426524e-01 -1.41277641e-01 8.80530290e-03 8.58531892e-01
1.67412579e-01 3.02241743e-01 2.27643922e-01 -9.50962722e-01
3.64483535e-01 9.89136040e-01 1.58821642e-01 -1.01790476e+00
-4.72513556e-01 -9.62447375e-02 -8.47704232e-01 5.07759690e-01
5.61388791e-01 -5.10997958e-02 -7.95496345e-01 1.18151915e+00
3.05299968e-01 -3.54169190e-01 -6.22398257e-02 6.51919961e-01
8.07601631e-01 4.81146216e-01 -1.50863752e-01 -3.85889381e-01
1.33965957e+00 -7.69936740e-01 -3.58813703e-01 -1.38759445e-02
4.01410997e-01 -6.15033805e-01 1.17654693e+00 1.83789045e-01
-1.23481119e+00 -2.93480903e-01 -1.36824298e+00 1.94745839e-01
-5.04496872e-01 1.30607426e-01 1.94587812e-01 3.79113048e-01
-5.88539004e-01 1.09824395e+00 -8.14690590e-01 -1.02163278e-01
1.12855947e+00 5.52987993e-01 -3.64884526e-01 1.03107048e-02
-8.35727632e-01 7.29229152e-01 2.12759688e-01 -2.22035050e-01
-8.24812472e-01 -8.07463288e-01 -5.69236755e-01 7.98140913e-02
2.96860486e-01 -4.22727525e-01 9.01164949e-01 -6.78515851e-01
-1.45936668e+00 1.12496865e+00 2.31272668e-01 -3.50677848e-01
8.10366631e-01 2.12464720e-01 -3.51276109e-03 3.47909987e-01
2.81486303e-01 3.37342739e-01 3.72571766e-01 -1.33708096e+00
-5.12206316e-01 -5.20181000e-01 -3.75608236e-01 7.04243183e-02
6.25874400e-02 -4.49040413e-01 1.24724254e-01 -2.84952074e-01
2.44574800e-01 -4.86351103e-01 -2.62050092e-01 -4.65039462e-02
-4.06681269e-01 -2.48196512e-01 7.99239516e-01 -6.24217391e-01
7.64964700e-01 -1.99140251e+00 -1.95742175e-01 3.78528208e-01
3.53353560e-01 2.05906838e-01 2.17610255e-01 -5.18542528e-03
2.55222499e-01 -8.41857027e-03 -8.60874057e-01 -1.43908337e-01
-9.84379724e-02 -1.50583491e-01 5.14967859e-01 7.07300782e-01
-6.12355694e-02 7.33640909e-01 -6.25254631e-01 -3.59619737e-01
1.30525470e-01 4.30470347e-01 -2.76747823e-01 3.87435146e-02
-7.32556462e-01 1.99847475e-01 -3.12958449e-01 6.90705121e-01
7.66485989e-01 -3.20880830e-01 6.47087544e-02 -4.35618460e-01
-4.95504886e-01 -1.88730434e-01 -6.90287828e-01 1.34969282e+00
-2.25185350e-01 4.89423394e-01 5.45369089e-01 -1.10299706e+00
1.19969773e+00 2.86057055e-01 1.06827986e+00 -1.15801179e+00
4.39543694e-01 6.84197128e-01 2.15948708e-02 -3.98903728e-01
5.67581393e-02 -6.04551435e-01 1.74516350e-01 3.00865918e-01
-2.18326986e-01 -3.61340940e-01 2.31609985e-01 -4.53541309e-01
1.02244508e+00 -3.97960573e-01 9.31650996e-02 -8.39456618e-01
6.13163292e-01 2.54586965e-01 3.35934669e-01 4.78525132e-01
-4.08891924e-02 1.00555444e+00 6.25059426e-01 -6.73383117e-01
-1.54797304e+00 -9.89515245e-01 -4.80846614e-01 4.35368389e-01
6.18543811e-02 2.93377757e-01 -1.27953625e+00 -4.65487599e-01
-1.00005470e-01 3.99730742e-01 -5.72208643e-01 5.95735833e-02
-3.63963097e-01 -9.36592817e-01 4.66378301e-01 4.47243840e-01
6.74642980e-01 -1.29790866e+00 -6.48791254e-01 4.38434064e-01
1.01486899e-01 -1.16746151e+00 -8.03456903e-02 3.52285326e-01
-6.60138488e-01 -1.61167455e+00 -7.66865075e-01 -9.67479825e-01
6.81726933e-01 -3.91098171e-01 1.10600495e+00 -2.18097642e-01
-4.29263979e-01 3.14670295e-01 1.80778559e-02 -3.82155836e-01
-4.65617627e-01 9.91347358e-02 -2.42172718e-01 -3.24330956e-01
3.43848318e-02 -5.22015929e-01 -9.40432668e-01 5.16195714e-01
-9.73054051e-01 -2.03179210e-01 3.29344690e-01 7.29693711e-01
1.12974191e+00 3.29153687e-01 4.83703822e-01 -8.46577108e-01
7.06609428e-01 -2.91022450e-01 -6.59215093e-01 1.68177396e-01
-5.81072927e-01 -4.26703304e-01 8.86852324e-01 -2.47518420e-01
-8.75330925e-01 -6.24528974e-02 -4.80303913e-01 -3.45848620e-01
-3.13976228e-01 2.57269681e-01 -2.54121482e-01 -8.83315355e-02
7.21931338e-01 1.05842957e-02 2.63560534e-01 2.97457706e-02
-1.56223685e-01 6.04174614e-01 3.85058016e-01 -8.77301872e-01
1.51782222e-02 8.97363842e-01 1.78271517e-01 -1.12336290e+00
-7.42246151e-01 -1.54803589e-01 -4.03082550e-01 -2.32973740e-01
1.29392004e+00 -4.91596311e-01 -9.76072252e-01 5.37649155e-01
-1.12399745e+00 -7.32107460e-01 -3.89356434e-01 1.73989892e-01
-8.00567746e-01 4.20963377e-01 -6.72994375e-01 -3.78667563e-01
-9.11093950e-01 -1.56306684e+00 8.31211925e-01 1.63911074e-01
1.15811191e-02 -1.09503078e+00 -8.60107467e-02 6.25583172e-01
4.63469803e-01 5.88686168e-01 1.32154751e+00 -7.34367967e-01
-8.01503062e-02 -3.43621731e-01 -5.04698038e-01 5.48453927e-01
1.04493029e-01 -2.05676511e-01 -8.65534127e-01 -5.89619055e-02
3.69384110e-01 1.84117854e-02 6.90817714e-01 9.52432692e-01
1.32855546e+00 -1.48144796e-01 -3.69764954e-01 4.74792868e-01
1.49450791e+00 6.69199407e-01 9.03697014e-01 5.58880031e-01
7.83096373e-01 2.31570646e-01 3.84575576e-01 5.02905250e-01
-1.78717300e-02 4.83542204e-01 8.03933322e-01 -1.31164223e-01
6.75415322e-02 3.65300775e-01 -8.10152963e-02 3.48789960e-01
-1.76971808e-01 -2.85157710e-01 -1.04012668e+00 3.72634888e-01
-1.30150366e+00 -6.30497217e-01 -7.76862800e-02 1.98333180e+00
4.41996187e-01 3.01580876e-01 1.43227512e-02 1.77781507e-01
9.08924818e-01 -1.87626645e-01 -7.30824411e-01 -2.89804667e-01
-3.35897356e-01 4.55995500e-01 4.17957217e-01 4.17653978e-01
-1.06885445e+00 6.38283968e-01 5.20563650e+00 1.13763165e+00
-1.27556682e+00 -9.81328823e-03 1.10961127e+00 1.66290581e-01
-2.23866299e-01 -5.49856484e-01 -4.44799840e-01 6.57489419e-01
5.03365874e-01 2.15167746e-01 2.06585407e-01 5.98802865e-01
-7.42573813e-02 -2.08864838e-01 -8.28411877e-01 7.82212794e-01
-2.03097627e-01 -1.73887813e+00 -1.76537409e-01 3.25653672e-01
8.89199972e-01 3.63727331e-01 3.48731130e-02 -2.01243833e-01
2.77015641e-02 -1.23235452e+00 7.42083609e-01 4.21595097e-01
1.26778972e+00 -7.10687339e-01 9.57463264e-01 1.44232064e-01
-1.32786024e+00 -1.44284457e-01 -4.90251362e-01 1.27630204e-01
-7.08821565e-02 8.59850228e-01 -8.99074972e-01 3.35009247e-01
7.39687026e-01 2.79429317e-01 9.77873653e-02 5.49011648e-01
4.89070654e-01 -1.85899287e-02 -6.33274198e-01 3.99088189e-02
4.97358829e-01 -7.14128137e-01 3.63731086e-01 1.11997962e+00
5.99819481e-01 1.13535076e-01 3.61567549e-02 1.12860310e+00
-2.82795936e-01 2.16873810e-01 -5.43878555e-01 2.59701777e-02
3.57918501e-01 1.12508106e+00 -1.35279393e+00 -2.31394380e-01
-2.31590569e-01 4.28386152e-01 2.05161110e-01 1.39900045e-02
-6.98928833e-01 -2.90475637e-01 4.19578552e-01 6.75192118e-01
4.21571106e-01 1.62009597e-02 -7.13607192e-01 -5.76666474e-01
1.79046959e-01 -3.80785704e-01 4.11462307e-01 -5.48159599e-01
-1.50873828e+00 6.25455558e-01 -2.74402857e-01 -6.90578461e-01
5.75279176e-01 -1.12173188e+00 -7.96266615e-01 5.26399553e-01
-1.16664600e+00 -6.58744514e-01 -3.29922974e-01 2.58849978e-01
4.41550553e-01 -3.76619846e-01 5.89219332e-01 3.28649193e-01
-4.86704826e-01 1.83983222e-01 5.75081646e-01 -7.92137682e-02
-3.67404260e-02 -1.02789927e+00 4.52818312e-02 7.78864473e-02
-5.93950868e-01 -4.67403829e-02 5.66862702e-01 -6.15923285e-01
-9.62263048e-01 -1.12442577e+00 2.28038684e-01 2.05129152e-03
4.83381778e-01 -4.92519476e-02 -9.72515166e-01 2.97518849e-01
6.91502094e-02 8.03693086e-02 5.98720491e-01 -5.97533584e-01
2.99122483e-01 5.92126362e-02 -1.87955499e+00 5.24083793e-01
6.86657786e-01 -2.35160187e-01 -8.58011618e-02 5.13396204e-01
4.53026563e-01 -2.88446337e-01 -1.38956892e+00 5.92661500e-01
6.30889177e-01 -1.12632418e+00 1.01775491e+00 -9.44594517e-02
6.40125215e-01 9.03154612e-02 -3.67928565e-01 -8.86299253e-01
-2.92777926e-01 1.26635700e-01 4.03360009e-01 6.87515378e-01
6.31641924e-01 -5.40792406e-01 1.13415468e+00 4.84190851e-01
-8.59120429e-01 -1.34739244e+00 -1.32136869e+00 -4.55530941e-01
5.96234024e-01 -2.35658973e-01 4.20487851e-01 7.04685330e-01
-2.32940555e-01 -6.59044087e-02 7.03340232e-01 -1.95798978e-01
6.73346698e-01 -1.30375475e-01 3.87898050e-02 -1.65382969e+00
4.49763358e-01 -7.12598920e-01 -2.36226410e-01 -4.62982982e-01
2.88682997e-01 -9.81083453e-01 8.34670514e-02 -1.87702191e+00
3.03605497e-02 -5.66780865e-01 -1.06342018e-01 6.10382333e-02
4.07467037e-01 3.90158445e-02 -3.04013580e-01 2.75672257e-01
-6.04967117e-01 6.92404687e-01 1.55177486e+00 -5.14232516e-01
7.46635487e-03 -1.03522405e-01 -3.28057021e-01 8.71159315e-01
9.37711835e-01 -4.08798397e-01 4.84495834e-02 -1.96304843e-01
2.90519923e-01 9.56219509e-02 8.23069736e-02 -1.47833729e+00
9.72330645e-02 -6.26946017e-02 4.46769178e-01 -3.68401974e-01
2.69195557e-01 -1.07312822e+00 1.62771031e-01 5.99803388e-01
6.10998571e-02 -2.51005113e-01 1.29321828e-01 2.07481548e-01
1.85028985e-02 -2.20541075e-01 1.30680609e+00 -7.10683286e-01
2.92827804e-02 4.73387569e-01 -3.78569663e-01 -4.25886065e-02
1.20595956e+00 -6.28077924e-01 -3.19360018e-01 3.16510111e-01
-8.00119758e-01 -1.81172088e-01 8.32569957e-01 -4.25267458e-01
6.43876731e-01 -1.01738214e+00 -2.52109319e-01 1.36753276e-01
-1.56067759e-01 6.80696189e-01 4.54020709e-01 8.51634443e-01
-1.04938042e+00 1.91510335e-01 -3.95186335e-01 -7.85242736e-01
-7.90530026e-01 5.02649806e-02 9.00141716e-01 -4.93861705e-01
-6.08371317e-01 8.11530769e-01 2.66472697e-01 -4.84833449e-01
-2.33145431e-01 -4.21475947e-01 -2.71992445e-01 -1.54896621e-02
-8.47001597e-02 8.42165649e-01 3.86097610e-01 -5.22615492e-01
-1.57214984e-01 6.53904617e-01 -1.09246885e-02 2.37089872e-01
1.54813004e+00 2.79768229e-01 -2.69345522e-01 1.77120805e-01
1.22530687e+00 -5.85648835e-01 -1.78508592e+00 1.72037676e-01
-2.55346924e-01 2.51516551e-01 -1.51003450e-01 -4.74064082e-01
-1.33709383e+00 7.42440581e-01 7.33417511e-01 4.80305165e-01
5.82747936e-01 2.88426101e-01 9.27253604e-01 2.63635248e-01
-2.88781412e-02 -1.34971046e+00 1.68646827e-01 7.01632798e-01
7.00146854e-01 -1.21109843e+00 1.05455287e-01 -4.09825027e-01
-2.29377344e-01 1.22243834e+00 5.86302817e-01 -2.23366424e-01
5.79318583e-01 3.34253937e-01 -1.23024479e-01 -8.93104732e-01
-3.29571575e-01 2.62652904e-01 4.57165055e-02 6.00820184e-01
4.30469960e-01 1.91971079e-01 -2.07830235e-01 1.01685584e+00
-1.38674706e-01 -2.31072754e-01 1.07364379e-01 8.18831921e-01
-7.48876870e-01 -4.21604097e-01 -3.82135332e-01 8.34041059e-01
-6.00634992e-01 1.60029158e-01 -1.95912823e-01 6.96471751e-01
3.11481118e-01 6.06290162e-01 5.15580952e-01 -8.39922875e-02
5.12744725e-01 -1.11307785e-01 6.53495610e-01 -2.59179741e-01
-5.61767399e-01 -4.44307830e-03 -2.27150306e-01 -2.13197291e-01
-1.89865664e-01 -5.39115012e-01 -1.72215533e+00 -2.64722332e-02
-3.20942998e-02 2.10989416e-01 8.04117084e-01 1.11899638e+00
1.68723822e-01 6.23081446e-01 3.86420369e-01 -1.20758903e+00
1.87615380e-02 -9.28032041e-01 -7.98141003e-01 4.93254334e-01
-1.69746950e-01 -8.99980843e-01 -5.70063591e-01 -3.32793146e-01] | [14.211243629455566, -2.7426204681396484] |
ae0cd526-7dc3-4169-96e0-88a6b2093125 | fewclue-a-chinese-few-shot-learning | 2107.07498 | null | https://arxiv.org/abs/2107.07498v2 | https://arxiv.org/pdf/2107.07498v2.pdf | FewCLUE: A Chinese Few-shot Learning Evaluation Benchmark | Pretrained Language Models (PLMs) have achieved tremendous success in natural language understanding tasks. While different learning schemes -- fine-tuning, zero-shot, and few-shot learning -- have been widely explored and compared for languages such as English, there is comparatively little work in Chinese to fairly and comprehensively evaluate and compare these methods and thus hinders cumulative progress. In this paper, we introduce the Chinese Few-shot Learning Evaluation Benchmark (FewCLUE), the first comprehensive few-shot evaluation benchmark in Chinese. It includes nine tasks, ranging from single-sentence and sentence-pair classification tasks to machine reading comprehension tasks. We systematically evaluate five state-of-the-art (SOTA) few-shot learning methods (including PET, ADAPET, LM-BFF, P-tuning and EFL), and compare their performance with fine-tuning and zero-shot learning schemes on the newly constructed FewCLUE benchmark. Experimental results reveal that: 1) The effect of different few-shot learning methods is sensitive to the pre-trained model to which the methods are applied; 2) PET and P-tuning achieve the best overall performance with RoBERTa and ERNIE respectively. Our benchmark is used in the few-shot learning contest of NLPCC 2021. In addition, we provide a user-friendly toolkit, as well as an online leaderboard to help facilitate further progress on Chinese few-shot learning. We provide a baseline performance on different learning methods, a reference for future research. | ['Hu Hai', 'Libo Qin', 'Xin Tian', 'Hu Yuan', 'Huilin Xu', 'Xiang Pan', 'Guoao Wei', 'Xuanwei Zhang', 'Chenyang Yuan', 'Xiaojing Lu', 'Liang Xu'] | 2021-07-15 | null | null | null | null | ['sentence-pair-classification'] | ['natural-language-processing'] | [ 1.40972719e-01 -2.60095477e-01 -4.38341260e-01 -3.18202466e-01
-1.36864209e+00 -1.64270639e-01 7.35942483e-01 2.15648204e-01
-7.80075967e-01 8.34150910e-01 5.93743742e-01 -4.09543782e-01
1.11145094e-01 -6.23690367e-01 -3.61959964e-01 -3.81186485e-01
2.56643564e-01 4.64206338e-01 4.44120616e-01 -5.58226049e-01
3.75181466e-01 -3.97967964e-01 -1.41415834e+00 3.48921180e-01
1.04064715e+00 5.62406242e-01 5.33059776e-01 9.83306885e-01
-4.38858241e-01 8.33075583e-01 -5.42961240e-01 -2.76052386e-01
-2.24648118e-01 -6.19974732e-01 -9.93705869e-01 -4.46718276e-01
2.57200003e-01 -1.33827180e-01 -2.95715015e-02 9.29123342e-01
1.10365939e+00 7.31160343e-01 7.44584739e-01 -7.12454379e-01
-9.14873183e-01 8.38732243e-01 -3.82947415e-01 5.48019350e-01
4.53858376e-01 2.53511548e-01 1.10718274e+00 -1.14091015e+00
6.69141591e-01 1.18214130e+00 5.59863329e-01 1.13034797e+00
-7.45789170e-01 -5.14175296e-01 -8.20164159e-02 6.41328216e-01
-1.11481869e+00 -6.06090367e-01 4.46462572e-01 -2.72087008e-01
1.50002241e+00 -1.04889013e-01 9.22288597e-02 1.40843475e+00
3.56889546e-01 1.10707748e+00 9.18587506e-01 -1.06672072e+00
5.53556681e-01 1.03336252e-01 7.50778615e-01 4.41044003e-01
-3.18143547e-01 -1.85073659e-01 -6.51690125e-01 -1.11248996e-02
-7.04562664e-02 -6.36563227e-02 -3.10567796e-01 6.39433973e-03
-9.43978131e-01 1.03648615e+00 -1.50835603e-01 7.43908763e-01
4.61874567e-02 -7.90678710e-02 7.57116079e-01 5.49181819e-01
7.86978126e-01 4.69269484e-01 -8.20511580e-01 -8.25403154e-01
-8.38010490e-01 5.09929024e-02 9.08763349e-01 1.07790089e+00
6.84973836e-01 7.71899596e-02 -8.35562348e-01 1.37363064e+00
-2.46700391e-01 8.99532139e-02 1.02451766e+00 -5.46374917e-01
6.43362284e-01 7.81344473e-02 -9.24281683e-03 -3.44775766e-01
-2.66154319e-01 1.33226067e-01 -8.01509738e-01 -4.22420859e-01
8.31052735e-02 -4.74833459e-01 -8.94480348e-01 1.50396705e+00
-2.18876734e-01 6.17596507e-01 2.33258978e-01 4.76997972e-01
1.31030190e+00 1.01683569e+00 2.90457815e-01 -4.08988953e-01
1.22070229e+00 -1.36575592e+00 -8.38443816e-01 -2.82024294e-01
1.25111926e+00 -7.09514380e-01 1.69542468e+00 1.09559372e-01
-9.26176190e-01 -7.52188623e-01 -9.78055060e-01 -2.15481207e-01
-6.97851121e-01 -1.90159485e-01 2.92195022e-01 7.87311137e-01
-8.19350481e-01 7.79919922e-01 -6.43050671e-01 -7.07725763e-01
3.13536108e-01 -1.47099406e-01 -6.65241629e-02 -4.26948458e-01
-1.73116624e+00 1.23462117e+00 5.59444547e-01 -7.71196961e-01
-8.74890983e-01 -1.08489299e+00 -1.04317796e+00 3.23475927e-01
5.01624644e-01 -2.89924562e-01 1.67326295e+00 -3.74045789e-01
-1.81341469e+00 8.83584857e-01 -2.39593774e-01 -6.25065863e-01
1.47019014e-01 -4.60289896e-01 -5.30702889e-01 -1.30783305e-01
9.89575908e-02 5.95842302e-01 4.79178399e-01 -5.81677794e-01
-4.95293349e-01 6.18593469e-02 -9.49359462e-02 1.10052258e-01
-5.65113783e-01 4.83494550e-01 -5.40125072e-01 -5.24078667e-01
-7.58758962e-01 -4.04243380e-01 -1.57271490e-01 -4.17356312e-01
-1.88189909e-01 -7.90380776e-01 7.60626733e-01 -3.88869554e-01
1.62772179e+00 -2.02934408e+00 -9.63465646e-02 -7.07722068e-01
-2.94594675e-01 8.40292394e-01 -5.28317869e-01 6.95592105e-01
9.26393196e-02 2.00011432e-01 -2.18199641e-01 -4.95611012e-01
1.22351777e-02 1.88934952e-01 -2.31508330e-01 2.12234393e-01
1.54621899e-01 1.21206748e+00 -1.30201983e+00 -5.54010451e-01
5.04561782e-01 2.45243177e-01 -3.84486943e-01 2.81923741e-01
-3.56179476e-01 1.01029873e-03 -2.41014287e-01 6.14475608e-01
2.97459632e-01 -2.22582579e-01 -3.01410615e-01 3.59879047e-01
-1.19083241e-01 3.37876439e-01 -7.55556285e-01 2.12030053e+00
-6.34659588e-01 5.70039034e-01 -5.49153209e-01 -9.08426046e-01
9.10576761e-01 5.27419567e-01 1.86388772e-02 -8.03062022e-01
1.80864021e-01 -2.12074869e-04 -9.37969144e-03 -5.51587045e-01
5.93786299e-01 -2.21522883e-01 -2.48198524e-01 6.11802757e-01
7.16770232e-01 -2.51408160e-01 5.26658237e-01 3.83282393e-01
1.12642109e+00 2.95337606e-02 9.02578294e-01 -4.01512802e-01
4.28285390e-01 -1.49414569e-01 5.48741043e-01 1.05500352e+00
-8.04307163e-01 8.10046732e-01 3.82204741e-01 -1.90363050e-01
-8.76689076e-01 -9.39915359e-01 -5.87599725e-02 1.81212234e+00
-1.75420538e-01 -7.38632143e-01 -8.28612983e-01 -5.72460830e-01
-4.52719241e-01 1.48158062e+00 -7.27980793e-01 -3.13801318e-01
-2.40427837e-01 -6.97222650e-01 5.01414835e-01 5.90984464e-01
4.16830093e-01 -1.31383443e+00 -6.09037280e-01 3.14236939e-01
-1.48084700e-01 -8.87241125e-01 -8.32324922e-01 3.32915962e-01
-5.39956212e-01 -8.81808698e-01 -1.08573639e+00 -1.05012572e+00
-8.87233706e-04 4.72728819e-01 1.26287794e+00 -1.72002614e-01
-4.30980921e-01 4.71303523e-01 -9.60955739e-01 -8.07112157e-01
-2.69943148e-01 2.17494473e-01 7.28376955e-02 -4.42390203e-01
9.97372389e-01 -1.78189874e-01 -6.66179657e-02 4.53538336e-02
-4.56854790e-01 -1.47277877e-01 1.84584081e-01 1.20733356e+00
2.09665671e-01 -3.08655322e-01 7.48738587e-01 -1.18710446e+00
1.11118531e+00 -4.75174636e-01 -1.49079338e-01 7.80321121e-01
-4.22826946e-01 -1.65915281e-01 6.01484895e-01 -3.99167597e-01
-1.40649605e+00 -5.25279224e-01 -3.15330148e-01 -3.89079332e-01
-1.74102962e-01 5.86230993e-01 8.24553594e-02 2.34718949e-01
9.10909057e-01 2.83109665e-01 -4.76366222e-01 -6.44890487e-01
5.87052107e-01 8.86386156e-01 2.90610582e-01 -5.55075824e-01
3.26579869e-01 -2.64045954e-01 -7.76581287e-01 -1.23869622e+00
-1.24212587e+00 -1.05863917e+00 -8.85610521e-01 6.60789385e-02
1.02113199e+00 -9.91448522e-01 -2.88474746e-02 3.97719681e-01
-1.25265074e+00 -4.72287774e-01 -2.88109601e-01 3.96544904e-01
-5.68711638e-01 4.07215357e-01 -7.61860132e-01 -8.01285088e-01
-7.76831865e-01 -8.16429198e-01 9.46428597e-01 4.64259416e-01
-3.09390545e-01 -1.28290498e+00 5.32444835e-01 1.85443252e-01
6.16912246e-01 -3.74519914e-01 1.08174336e+00 -1.02825022e+00
2.11771756e-01 3.41248177e-02 -8.51990283e-02 2.76140273e-01
-7.24860607e-03 -1.06948391e-01 -1.10746706e+00 -3.28887731e-01
-3.72497886e-02 -1.08867395e+00 1.08536148e+00 5.22535205e-01
1.10381091e+00 1.29101768e-01 -2.64296353e-01 3.25703859e-01
1.44465005e+00 9.26254466e-02 6.17062867e-01 1.59554303e-01
4.49167132e-01 3.98595244e-01 7.53857076e-01 5.57047129e-01
1.85025156e-01 4.66730535e-01 -2.80905247e-01 3.90949935e-01
-1.98861659e-01 -2.64501452e-01 4.21123862e-01 1.47476256e+00
-1.00063436e-01 -3.05901080e-01 -1.06485236e+00 6.23696387e-01
-1.87046313e+00 -1.14380562e+00 2.76166767e-01 2.02156401e+00
1.13010275e+00 2.99130172e-01 -1.73021510e-01 -2.93090522e-01
6.19691253e-01 6.02693141e-01 -3.14342678e-01 -8.02035451e-01
-1.48786098e-01 7.18920529e-01 -2.76973154e-02 4.25077468e-01
-1.19817650e+00 1.37106919e+00 6.21795034e+00 1.43118465e+00
-8.05984378e-01 6.62157834e-01 6.47113919e-01 -1.17407195e-01
-3.34013812e-02 -3.37877013e-02 -1.18148363e+00 3.11951965e-01
1.41933191e+00 -6.61343038e-01 9.57994983e-02 1.10888648e+00
2.36234143e-01 -1.06554702e-01 -9.26918745e-01 8.75306785e-01
4.64795053e-01 -1.47898662e+00 -9.07317549e-02 -6.25386655e-01
1.12142956e+00 5.90323806e-01 -3.88579339e-01 1.31856036e+00
4.23212916e-01 -9.36001241e-01 4.69514169e-02 4.02650177e-01
9.04502094e-01 -8.03279996e-01 7.60377645e-01 7.15233147e-01
-1.05696261e+00 -4.30373698e-02 -1.06250429e+00 -2.93136239e-01
3.85670900e-01 2.43148431e-01 -5.77659428e-01 4.08011347e-01
5.72362602e-01 9.02998090e-01 -3.87825221e-01 9.88431871e-01
-3.89422268e-01 1.02534235e+00 3.83040845e-01 -6.31601632e-01
5.92119217e-01 1.27880499e-01 3.05794060e-01 1.59540761e+00
7.80642405e-02 2.92690456e-01 2.52714664e-01 6.00246489e-01
-2.45020628e-01 4.53426123e-01 -5.62504649e-01 -2.90962964e-01
4.96510446e-01 1.09189892e+00 -2.45497018e-01 -7.12274909e-01
-9.83529449e-01 7.92405784e-01 6.71278238e-01 2.52090901e-01
-5.20151317e-01 -8.34466517e-01 5.49250185e-01 -2.98424512e-01
3.11198801e-01 -6.27986640e-02 -1.20508619e-01 -1.52262259e+00
-5.40766537e-01 -7.79854238e-01 4.30255353e-01 -6.45500958e-01
-1.58110893e+00 4.27565426e-01 -5.01103373e-03 -1.17219806e+00
-2.98408419e-01 -5.74022651e-01 -1.47533786e+00 9.03420210e-01
-1.55450916e+00 -8.67920220e-01 -2.33441629e-02 4.84135866e-01
1.53357267e+00 -6.25066400e-01 1.28388703e+00 1.84543151e-02
-7.18495727e-01 7.87485123e-01 4.16187078e-01 1.38404861e-01
1.07896745e+00 -1.19347191e+00 7.09303796e-01 7.27972150e-01
2.43770421e-01 5.17718732e-01 5.16367376e-01 -5.87898850e-01
-1.26624513e+00 -1.04813361e+00 1.45280647e+00 -4.29157197e-01
9.09497321e-01 -3.79277974e-01 -1.04422021e+00 5.56065321e-01
5.69767714e-01 -2.29702927e-02 9.14376974e-01 4.23285156e-01
-2.01456651e-01 2.82560706e-01 -7.07718372e-01 5.09040058e-01
6.28222466e-01 -7.63239086e-01 -1.10805655e+00 5.16220629e-01
1.13904309e+00 -5.81463650e-02 -6.94611907e-01 2.48266786e-01
1.97224677e-01 -7.28231132e-01 6.68908775e-01 -1.04470611e+00
6.29120708e-01 4.83424008e-01 -2.85505503e-01 -1.72492051e+00
-4.33673501e-01 -6.34036064e-01 -1.79128915e-01 1.22455800e+00
4.93857086e-01 -1.25528008e-01 6.09414518e-01 4.97783720e-01
-3.04699957e-01 -9.88591492e-01 -8.34318578e-01 -1.04248261e+00
6.34302437e-01 -5.26730418e-01 1.36827022e-01 1.04434001e+00
4.45541680e-01 8.46385598e-01 -7.52090931e-01 -6.58215821e-01
3.53294820e-01 -2.49986395e-01 7.33845770e-01 -8.80301237e-01
-5.07396817e-01 -4.32975680e-01 5.59736341e-02 -8.52699876e-01
3.38422984e-01 -8.12106788e-01 2.73556739e-01 -1.43262827e+00
7.31711626e-01 1.88068777e-01 -4.41190660e-01 5.18915474e-01
-7.32492507e-01 -2.27979288e-01 2.81021833e-01 -7.76155740e-02
-9.99778628e-01 8.65846097e-01 1.09270978e+00 -3.59770596e-01
-2.16854990e-01 4.57126014e-02 -4.16921586e-01 5.53175092e-01
8.50742698e-01 -2.90885985e-01 -5.00647604e-01 -3.57888609e-01
-4.64456290e-01 -1.95683688e-01 -3.70324016e-01 -8.82995427e-01
4.53097343e-01 -9.18008983e-02 1.06920116e-01 -5.87878764e-01
2.05131039e-01 1.85085863e-01 -9.27364767e-01 5.43863356e-01
-5.13241172e-01 -2.02220172e-01 1.42483011e-01 4.94279653e-01
-2.51216382e-01 -8.53514850e-01 9.94809628e-01 -5.80863297e-01
-1.45213151e+00 3.58646154e-01 -2.99995124e-01 8.05614829e-01
9.62572575e-01 5.83976135e-02 -3.82972032e-01 -3.94809216e-01
-4.85178173e-01 4.04879093e-01 4.65816818e-02 8.20633054e-01
6.29540384e-01 -1.01690066e+00 -9.90544915e-01 -4.78385156e-03
6.29372656e-01 -5.43905795e-01 4.95920837e-01 6.83186233e-01
-8.58408660e-02 7.64291763e-01 -1.18355148e-01 -2.89346725e-01
-1.17463267e+00 5.99500120e-01 -8.11052546e-02 -5.43605924e-01
-6.51311994e-01 1.15966535e+00 -1.46705611e-02 -7.61858463e-01
4.97861832e-01 1.48118250e-02 -4.33636159e-01 1.78333879e-01
1.16022539e+00 5.44242263e-01 -5.77256456e-02 -2.17775792e-01
-2.56887019e-01 4.09318715e-01 -4.11210656e-01 5.69687895e-02
1.34547448e+00 -2.90043298e-02 2.58121282e-01 1.07678878e+00
1.27364969e+00 -5.98886609e-01 -9.16959763e-01 -5.91457963e-01
1.65611014e-01 -1.80891290e-01 2.00853534e-02 -8.38837266e-01
-2.64019638e-01 1.43831909e+00 2.66390026e-01 -2.19464391e-01
7.63034940e-01 -6.87071830e-02 9.79016602e-01 6.25582278e-01
5.61405540e-01 -1.58650362e+00 2.46178225e-01 1.23156190e+00
5.16701937e-01 -1.63762915e+00 -1.24455512e-01 1.16040170e-01
-9.46764886e-01 1.04885983e+00 9.00454223e-01 3.00197545e-02
8.79496217e-01 6.54327124e-02 -3.16900685e-02 1.91271767e-01
-1.34471846e+00 -2.69402832e-01 3.61550927e-01 6.01376414e-01
8.21508765e-01 1.21336259e-01 -5.60969651e-01 9.01851118e-01
4.55109887e-02 1.95137188e-01 3.78650963e-01 9.97836947e-01
-9.95905399e-01 -1.10115981e+00 2.11515501e-01 6.44892931e-01
-2.60756344e-01 -5.15936017e-01 -2.79276997e-01 6.11330092e-01
-1.87723741e-01 1.07569289e+00 -9.28334668e-02 -2.30585426e-01
1.77347496e-01 3.11350167e-01 3.45704705e-01 -1.38175285e+00
-5.72308302e-01 -5.56628481e-02 2.60081828e-01 -4.09984797e-01
-1.79862842e-01 -4.10711527e-01 -9.53085959e-01 -1.80721506e-01
-5.29968381e-01 2.20109239e-01 3.36786091e-01 1.30171525e+00
1.83554441e-01 4.27735656e-01 3.09207827e-01 -8.15447748e-01
-9.80794549e-01 -1.59358811e+00 -6.98732734e-01 2.08139643e-01
-1.09066904e-01 -5.01649618e-01 -3.56901646e-01 -2.52346992e-01] | [10.723665237426758, 7.968075275421143] |
120665ad-1cdf-4ad7-9b15-2d5a522f705b | delad-deep-landweber-guided-deconvolution | 2209.15377 | null | https://arxiv.org/abs/2209.15377v1 | https://arxiv.org/pdf/2209.15377v1.pdf | DELAD: Deep Landweber-guided deconvolution with Hessian and sparse prior | We present a model for non-blind image deconvolution that incorporates the classic iterative method into a deep learning application. Instead of using large over-parameterised generative networks to create sharp picture representations, we build our network based on the iterative Landweber deconvolution algorithm, which is integrated with trainable convolutional layers to enhance the recovered image structures and details. Additional to the data fidelity term, we also add Hessian and sparse constraints as regularization terms to improve the image reconstruction quality. Our proposed model is \textit{self-supervised} and converges to a solution based purely on the input blurred image and respective blur kernel without the requirement of any pre-training. We evaluate our technique using standard computer vision benchmarking datasets as well as real microscope images obtained by our enhanced depth-of-field (EDOF) underwater microscope, demonstrating the capabilities of our model in a real-world application. The quantitative results demonstrate that our approach is competitive with state-of-the-art non-blind image deblurring methods despite having a fraction of the parameters and not being pre-trained, demonstrating the efficiency and efficacy of embedding a classic deconvolution approach inside a deep network. | ['Tingying Peng', 'Jan Taucher', 'Anton Theileis', 'Tomas Chobola'] | 2022-09-30 | null | null | null | null | ['blind-image-deblurring', 'image-deconvolution'] | ['computer-vision', 'computer-vision'] | [ 2.36493990e-01 1.94504887e-01 8.03524494e-01 -3.51081401e-01
-4.67185199e-01 -5.54536283e-01 6.11613214e-01 -6.71850979e-01
-7.38542616e-01 6.59790576e-01 4.29574579e-01 -7.54787326e-02
-1.95499063e-01 -3.54438394e-01 -9.99191225e-01 -7.86306918e-01
-3.68530191e-02 2.50813574e-01 1.87249810e-01 -1.43265188e-01
4.35565978e-01 3.48777562e-01 -1.33379745e+00 2.90750004e-02
9.09054220e-01 8.01023960e-01 5.13081253e-01 8.56832504e-01
2.41755262e-01 9.12005603e-01 -2.95095533e-01 -3.60826224e-01
5.66138923e-01 -4.82742161e-01 -9.45974350e-01 4.25494611e-01
6.03930771e-01 -9.48624372e-01 -6.46433532e-01 1.22611046e+00
5.15741765e-01 1.32380292e-01 4.94330525e-01 -4.65721846e-01
-1.10346401e+00 2.81740338e-01 -3.24751884e-01 2.30404034e-01
5.03685102e-02 3.44817460e-01 5.40184557e-01 -1.15341008e+00
4.96549040e-01 9.09872055e-01 1.04070294e+00 4.72841561e-01
-1.45417178e+00 -3.04365665e-01 -3.75481248e-01 1.90885931e-01
-1.05270004e+00 -8.64367962e-01 5.54184437e-01 -4.85395521e-01
8.58519912e-01 -1.35728016e-01 3.62680644e-01 7.50396073e-01
-3.18723992e-02 3.58643174e-01 1.37864423e+00 -4.11040157e-01
3.27354282e-01 -2.22895611e-02 -1.23634011e-01 7.20226467e-01
1.11598656e-01 3.31325620e-01 -4.56347227e-01 5.20594455e-02
1.16530979e+00 1.31502539e-01 -8.96661878e-01 -4.04599935e-01
-1.17422330e+00 6.25098765e-01 7.62763023e-01 1.70927167e-01
-4.31623220e-01 3.94528568e-01 -1.58952430e-01 2.32361257e-01
4.73203957e-01 5.11001647e-01 -2.30730131e-01 9.79385078e-02
-1.35428011e+00 1.47755509e-02 7.74844825e-01 5.76883197e-01
1.24407506e+00 6.31244928e-02 2.37276945e-02 7.36703992e-01
3.85095477e-01 2.67832786e-01 6.50280654e-01 -1.55013251e+00
-1.20727934e-01 3.74963939e-01 5.12355387e-01 -5.64573884e-01
-7.72454813e-02 -5.31980753e-01 -8.99340630e-01 6.10703707e-01
3.14357907e-01 -1.15151480e-01 -1.21093881e+00 1.48194516e+00
-9.91177186e-02 5.54678082e-01 2.75292039e-01 1.31149173e+00
7.30002582e-01 4.37539577e-01 -6.35936141e-01 -1.11113796e-02
1.07665145e+00 -1.17383981e+00 -6.25328183e-01 -4.64065135e-01
-2.07400112e-03 -7.53241360e-01 8.38098526e-01 3.45880091e-01
-1.25186956e+00 -2.96507269e-01 -1.06588733e+00 -3.62944901e-01
-4.51133214e-02 1.92578763e-01 2.85889477e-01 3.78438890e-01
-1.65730143e+00 9.90089238e-01 -8.54219079e-01 -3.41521084e-01
5.84244490e-01 3.58670682e-01 -6.80986404e-01 -3.53524417e-01
-7.24644601e-01 9.69173849e-01 1.53109476e-01 4.37393963e-01
-1.30668914e+00 -8.11736107e-01 -7.43247330e-01 4.96928766e-02
-2.06355259e-01 -9.70043361e-01 1.11798823e+00 -8.87546301e-01
-1.88422894e+00 7.76848376e-01 -2.21299566e-02 -4.70193595e-01
6.91684783e-01 -4.93962407e-01 2.03028664e-01 5.15759289e-01
-1.63861722e-01 6.34510934e-01 1.20762050e+00 -1.59257746e+00
-1.23075277e-01 -3.57180461e-02 1.88782156e-01 1.83542013e-01
-2.31631026e-01 -4.99937683e-02 -4.64212149e-01 -5.74819684e-01
1.30182445e-01 -7.75892317e-01 -2.91404337e-01 3.28166395e-01
-2.17742413e-01 6.40314519e-01 7.56756961e-01 -9.81390417e-01
6.58369303e-01 -2.02376246e+00 4.27003473e-01 -1.82575852e-01
4.42678928e-01 3.15749407e-01 -1.39601275e-01 3.76014650e-01
-5.14997654e-02 -2.86806881e-01 -8.71971905e-01 -8.50664496e-01
-2.08376661e-01 3.80637199e-01 -1.98204443e-01 9.39862609e-01
1.35995641e-01 8.05595696e-01 -8.67714107e-01 5.32609895e-02
4.03088808e-01 8.38605046e-01 -6.88465238e-01 6.57870054e-01
1.72938973e-01 6.35574520e-01 2.91552156e-01 1.29877090e-01
9.33643162e-01 -5.39718688e-01 -1.57927722e-01 -3.93594980e-01
-3.05529863e-01 4.54382449e-02 -9.85960424e-01 1.99227238e+00
-4.89479035e-01 7.33982861e-01 6.58887684e-01 -1.17854810e+00
5.19827187e-01 2.51520336e-01 4.70985249e-02 -3.47937047e-01
2.64270604e-02 5.21358192e-01 -2.82012969e-01 -6.79333150e-01
5.01037538e-01 -3.30790728e-01 7.33651817e-01 5.36072016e-01
5.08708656e-01 -2.28726283e-01 -1.14908218e-01 2.43588567e-01
1.17037034e+00 3.19478691e-01 -1.86815143e-01 -5.06517708e-01
4.09907609e-01 -1.75568074e-01 4.30996008e-02 7.37907529e-01
1.28410608e-01 1.33161616e+00 -5.17182797e-02 -2.96815127e-01
-1.22970116e+00 -9.27005589e-01 -1.81014270e-01 7.26140261e-01
2.62663752e-01 1.61264226e-01 -1.05931127e+00 -2.58032858e-01
-3.11649293e-01 3.51587415e-01 -7.79926538e-01 5.93808182e-02
-3.64370376e-01 -1.01502013e+00 5.34296155e-01 1.78102210e-01
8.02666545e-01 -8.95732939e-01 -5.91554880e-01 2.08252668e-01
-1.92716435e-01 -1.18326545e+00 -6.01341546e-01 2.11155266e-01
-9.41804528e-01 -1.04565966e+00 -1.31072223e+00 -1.02743864e+00
1.05348825e+00 5.68965316e-01 9.29769754e-01 1.12589031e-01
-1.71669468e-01 5.45070589e-01 -1.52389556e-01 2.69820094e-01
-5.68556130e-01 -5.52292109e-01 -7.50461221e-02 1.49653539e-01
-2.43738726e-01 -9.80485260e-01 -1.22937644e+00 1.76782042e-01
-1.26821959e+00 1.22514918e-01 8.79527926e-01 1.23847556e+00
1.93473995e-02 -1.11051902e-01 9.36348438e-02 -3.63175213e-01
4.80981112e-01 -3.19918454e-01 -5.64333081e-01 -1.62831713e-02
-6.91923857e-01 3.52664709e-01 3.90340894e-01 -4.41539407e-01
-1.08170581e+00 1.83559060e-01 8.67259353e-02 -6.64093852e-01
-5.03059588e-02 4.44060743e-01 4.26304847e-01 -6.24758661e-01
7.94356525e-01 6.39355779e-01 5.20124495e-01 -7.52111316e-01
3.97153944e-01 6.09373450e-01 1.10410631e+00 -1.10479884e-01
9.96602654e-01 8.53528023e-01 -2.29238480e-01 -7.28864610e-01
-6.93018854e-01 -4.93879467e-01 -5.97043335e-01 -9.85462964e-02
8.32864523e-01 -1.16399169e+00 -7.62577832e-01 1.00652635e+00
-1.34736955e+00 -5.56156754e-01 -4.09678109e-02 5.23581564e-01
-5.31144738e-01 9.90972579e-01 -9.63062167e-01 -5.79011440e-01
-4.58420128e-01 -1.19941664e+00 9.59713280e-01 2.46534035e-01
4.93690431e-01 -1.11595452e+00 1.09890699e-01 4.42116112e-01
9.61288750e-01 -1.85651898e-01 3.80988836e-01 -3.90394107e-02
-7.21655369e-01 1.19520463e-01 -7.52218246e-01 9.45992887e-01
1.61874428e-01 -5.84213316e-01 -1.30820441e+00 -5.25685787e-01
4.30231392e-01 -3.31203818e-01 1.26955843e+00 4.38910425e-01
7.68495500e-01 -6.28257036e-01 2.47762665e-01 1.02339053e+00
1.73272133e+00 -6.90732241e-01 9.31311011e-01 4.67000633e-01
7.22870469e-01 5.84565401e-01 -3.86700302e-01 2.39891917e-01
4.95372683e-01 4.35733527e-01 8.71158063e-01 -3.73866618e-01
-5.13259053e-01 1.11237615e-01 4.95417267e-01 6.02366507e-01
-4.02350813e-01 8.45409781e-02 -7.19989359e-01 8.55950773e-01
-1.74901772e+00 -8.44014645e-01 -1.57898545e-01 2.08876705e+00
1.13677692e+00 -4.86048579e-01 -3.42892766e-01 -1.51683614e-01
5.57409585e-01 -1.48751214e-01 -3.69382620e-01 1.47135615e-01
-1.39594480e-01 1.90888897e-01 6.83395565e-01 8.43456149e-01
-9.72199142e-01 7.11878240e-01 6.54236031e+00 1.93630099e-01
-1.27461720e+00 1.91460773e-01 2.92890638e-01 1.29345745e-01
-1.51965275e-01 1.60203327e-03 -9.47841257e-02 4.54301059e-01
6.62429512e-01 2.09880859e-01 1.15179598e+00 3.37104440e-01
2.33148649e-01 -7.15664849e-02 -9.92214024e-01 1.08711433e+00
9.89149511e-02 -1.59763932e+00 -1.92869261e-01 1.48621976e-01
9.02256131e-01 5.08503854e-01 4.67580073e-02 -3.08806121e-01
4.81837869e-01 -1.23086941e+00 8.71740222e-01 8.03036153e-01
7.75149643e-01 -1.15031317e-01 8.95195842e-01 2.95574188e-01
-4.43438768e-01 -5.95184974e-02 -5.34890115e-01 -8.76154378e-02
1.75993994e-01 6.36862874e-01 -6.16661549e-01 4.84230489e-01
1.09984386e+00 7.75492609e-01 -4.92880583e-01 1.25156951e+00
-1.40082568e-01 5.34671187e-01 -1.95396721e-01 7.92846799e-01
2.53360659e-01 -3.17314535e-01 6.10583723e-01 1.23152626e+00
5.81445038e-01 1.07213855e-01 -4.00150657e-01 1.20131171e+00
-2.06623927e-01 -5.02393484e-01 -2.18965590e-01 2.68053800e-01
-1.32215060e-02 1.13455677e+00 -3.25763732e-01 -2.69042224e-01
-4.71231252e-01 1.56501639e+00 2.15208098e-01 6.69574440e-01
-3.93038809e-01 -3.09753031e-01 5.85635841e-01 6.95918351e-02
7.78020740e-01 -2.58942664e-01 -1.03232250e-01 -1.40740228e+00
-4.09342870e-02 -6.72563493e-01 -2.83557087e-01 -1.14831865e+00
-1.45169032e+00 7.05551326e-01 -3.27643156e-01 -1.21432352e+00
-3.68420593e-02 -7.11172104e-01 -7.09141791e-01 1.17079639e+00
-2.17379093e+00 -1.13927341e+00 -7.71063328e-01 6.90084517e-01
1.73301905e-01 5.53885885e-02 7.54958332e-01 2.86075830e-01
-3.20625871e-01 1.67625830e-01 4.90684450e-01 9.32330415e-02
8.70880246e-01 -1.38773918e+00 2.52848506e-01 1.19227028e+00
-3.46756607e-01 9.51789320e-01 1.05527008e+00 -2.52737939e-01
-1.42024219e+00 -9.71866548e-01 4.66280609e-01 -2.62471974e-01
8.06943834e-01 -6.36202469e-02 -1.16487443e+00 6.16227746e-01
7.01397240e-01 4.71592665e-01 2.31228977e-01 -6.07225478e-01
-2.57292956e-01 7.96074048e-02 -1.25683129e+00 1.79276139e-01
8.68765116e-01 -5.60717762e-01 -8.60954523e-01 1.51877210e-01
4.71699476e-01 -4.93024349e-01 -5.91223836e-01 1.28527313e-01
3.70431989e-01 -1.34308720e+00 1.10518217e+00 8.94885417e-03
7.63874531e-01 -5.18058896e-01 7.54630715e-02 -1.47980177e+00
-4.52838331e-01 -9.71809387e-01 -1.51559427e-01 9.33042467e-01
1.53429776e-01 -6.50721729e-01 5.70410907e-01 3.36623907e-01
-4.71037090e-01 -1.95052102e-01 -8.47913086e-01 -5.66642284e-01
-5.31324334e-02 1.77363940e-02 1.37480482e-01 7.97066033e-01
-3.70407164e-01 6.68438673e-02 -6.77056968e-01 6.02641702e-01
1.20607769e+00 -1.00828268e-01 5.36958933e-01 -1.09957969e+00
-5.13827264e-01 -2.92527497e-01 -2.78033406e-01 -1.41274929e+00
-7.98398629e-02 -6.08731508e-01 5.78732610e-01 -1.66185117e+00
2.41351262e-01 -1.41751781e-01 -3.53063606e-02 4.92569685e-01
-8.58659670e-02 8.30601215e-01 -1.82560757e-01 7.14138389e-01
-2.90144622e-01 6.71142519e-01 1.24831259e+00 1.48285162e-02
-2.38301270e-02 -2.78799504e-01 -7.44467497e-01 6.25517130e-01
3.35641623e-01 -3.66482347e-01 -1.64972752e-01 -9.55222547e-01
5.58751151e-02 -2.06036925e-01 9.79231715e-01 -9.87482071e-01
5.60488164e-01 2.67533749e-01 3.20408672e-01 1.36844933e-01
3.87144923e-01 -7.32284784e-01 1.48440495e-01 4.47509289e-01
-6.33569136e-02 -4.31669146e-01 1.54544562e-01 6.72638059e-01
-1.54951483e-01 -3.61217946e-01 1.04693782e+00 -1.88128740e-01
-6.24724150e-01 -1.38509059e-02 -3.01940501e-01 -3.11557889e-01
3.94702554e-01 -2.92296290e-01 -5.72149396e-01 -4.14505631e-01
-5.45534134e-01 -1.41170472e-01 9.69125688e-01 -4.16630432e-02
6.72155082e-01 -8.05333972e-01 -7.77318716e-01 2.91076452e-01
-7.59487003e-02 2.42330045e-01 2.31506810e-01 1.01564503e+00
-1.02916229e+00 -4.71998937e-02 -3.06326121e-01 -5.46501994e-01
-7.55502403e-01 3.20051730e-01 7.17135906e-01 1.05023190e-01
-1.00717795e+00 9.09657121e-01 3.56282175e-01 -4.94704127e-01
1.56022385e-01 -2.35976189e-01 -1.95729788e-02 -5.13339520e-01
6.63937688e-01 2.64127225e-01 7.18024075e-02 -5.09440064e-01
-1.97593823e-01 5.60015023e-01 1.79486841e-01 -2.34379783e-01
1.82518637e+00 -6.29841089e-01 -5.43020248e-01 -1.69518232e-01
1.06307685e+00 -4.29010503e-02 -1.96666884e+00 -3.55709493e-01
-3.44112158e-01 -3.72824460e-01 7.18509495e-01 -9.05602992e-01
-1.23747540e+00 7.99396753e-01 7.49576509e-01 -3.56789082e-02
1.14617479e+00 -7.97074959e-02 6.80032194e-01 4.02155310e-01
1.47866264e-01 -5.45847416e-01 3.46366644e-01 5.08753657e-01
1.11723232e+00 -1.38844121e+00 9.41870362e-02 -5.92038371e-02
-3.57839108e-01 1.23100686e+00 7.75861274e-03 -6.03934944e-01
6.94553256e-01 2.52245575e-01 2.95795530e-01 -1.05178967e-01
-2.25370690e-01 -1.44056305e-01 1.95482016e-01 6.33872867e-01
1.50829613e-01 -6.40590727e-01 8.20042863e-02 2.38753349e-01
2.08820608e-02 2.75240362e-01 9.21920657e-01 6.54577255e-01
-4.83187914e-01 -7.13242829e-01 -4.24080610e-01 -2.95756515e-02
-4.84998375e-01 -4.42937762e-01 9.54406187e-02 1.98968574e-01
-1.37357756e-01 8.49171877e-01 -4.34058495e-02 -7.70860165e-02
-6.42052218e-02 -2.62578338e-01 5.88554084e-01 -4.35982198e-01
-4.67083037e-01 5.34965061e-02 -3.29755276e-01 -4.51918632e-01
-8.41557801e-01 -3.45327258e-01 -1.05255938e+00 -3.60478610e-01
-2.99668103e-01 4.64554504e-03 7.22966850e-01 1.02057719e+00
5.06884336e-01 4.36769992e-01 4.67456549e-01 -1.64329052e+00
-5.65593839e-01 -1.20987713e+00 -4.38848585e-01 3.30700547e-01
9.54246938e-01 -4.47909117e-01 -8.78173053e-01 5.46995521e-01] | [11.683362007141113, -2.669924259185791] |
cf0dd41b-9ec0-4eae-a053-47fb50d70029 | deep-learning-based-end-to-end-spoken | null | null | https://aclanthology.org/2022.lrec-1.798 | https://aclanthology.org/2022.lrec-1.798.pdf | Deep learning-based end-to-end spoken language identification system for domain-mismatched scenario | Domain mismatch is a critical issue when it comes to spoken language identification. To overcome the domain mismatch problem, we have applied several architectures and deep learning strategies which have shown good results in cross-domain speaker verification tasks to spoken language identification. Our systems were evaluated on the Oriental Language Recognition (OLR) Challenge 2021 Task 1 dataset, which provides a set of cross-domain language identification trials. Among our experimented systems, the best performance was achieved by using the mel frequency cepstral coefficient (MFCC) and pitch features as input and training the ECAPA-TDNN system with a flow-based regularization technique, which resulted in a Cavg of 0.0631 on the OLR 2021 progress set. | ['Abderrahim Fathan', 'Md Jahangir Alam', 'Woohyun Kang'] | null | null | null | null | lrec-2022-6 | ['spoken-language-identification'] | ['speech'] | [-5.55577762e-02 -4.01937157e-01 6.08606860e-02 -5.18091083e-01
-1.01929975e+00 -6.61460400e-01 7.99338222e-01 -1.51771083e-01
-9.42299783e-01 6.15361810e-01 2.48096541e-01 -5.06648600e-01
1.00972347e-01 1.63386464e-01 -3.87111902e-02 -4.65979278e-01
-1.53796449e-01 4.06845778e-01 -1.51769057e-01 -3.48763704e-01
2.74572462e-01 5.41164160e-01 -1.28795648e+00 2.83369750e-01
9.20143366e-01 1.06897306e+00 -1.64907500e-01 9.16010857e-01
6.13265447e-02 5.01917303e-01 -9.12616551e-01 -1.83338895e-01
7.17086047e-02 -4.97133583e-01 -9.43592072e-01 -2.52314150e-01
4.56373453e-01 -2.25446224e-01 -3.00531656e-01 7.94961870e-01
1.11902463e+00 3.82971227e-01 6.96698010e-01 -9.61701035e-01
-6.06226742e-01 7.56155133e-01 -1.86526343e-01 6.58707857e-01
5.73865771e-01 1.75694507e-02 6.64059103e-01 -1.36842692e+00
2.72931784e-01 1.55291903e+00 6.08358800e-01 9.72990572e-01
-1.20935142e+00 -8.28542173e-01 5.13586290e-02 3.71064842e-01
-1.88413322e+00 -1.10672843e+00 6.08664393e-01 -2.07635179e-01
1.48824143e+00 3.11112702e-02 1.16370089e-01 1.18605924e+00
-1.42894238e-01 7.62048066e-01 9.76293445e-01 -6.60744488e-01
2.01455012e-01 3.47337782e-01 2.79955328e-01 1.02294095e-01
-4.31712151e-01 4.26656038e-01 -9.34216142e-01 6.77382248e-03
2.70713508e-01 -1.07496691e+00 -3.30038309e-01 2.15688616e-01
-1.12823081e+00 1.00205958e+00 4.40682769e-02 7.30143070e-01
-2.83073902e-01 -3.61022919e-01 7.97428846e-01 6.17776632e-01
5.22799671e-01 3.46707642e-01 -4.87283200e-01 -1.68662667e-01
-8.04637313e-01 -1.26632452e-01 9.37335551e-01 3.40909004e-01
8.41107816e-02 6.47995353e-01 -1.13021515e-01 1.48872232e+00
3.55192393e-01 4.88821238e-01 1.04192162e+00 -5.57351887e-01
4.79218423e-01 -2.15369705e-02 -2.55571604e-02 -6.65885389e-01
-3.80391091e-01 -2.04636902e-01 -7.79097319e-01 4.93063293e-02
7.14755118e-01 -4.11789775e-01 -7.92473316e-01 1.68665028e+00
-9.17116031e-02 8.36765021e-02 7.64530182e-01 1.07598495e+00
1.04982424e+00 6.56464458e-01 1.86073229e-01 -1.36038557e-01
1.22903121e+00 -6.34055912e-01 -9.02197182e-01 -2.98790276e-01
6.51812911e-01 -9.77867901e-01 9.90126014e-01 5.87721586e-01
-9.23367321e-01 -8.20321858e-01 -1.19368184e+00 1.66640267e-01
-5.41281939e-01 4.41684961e-01 -1.52762473e-01 9.70083416e-01
-1.18325281e+00 2.20200285e-01 -2.86843807e-01 -4.52007532e-01
-2.74167717e-01 4.64039236e-01 -6.72117591e-01 2.29376778e-02
-1.75288677e+00 1.16806424e+00 4.96882319e-01 2.39005730e-01
-7.62647390e-01 -5.55235267e-01 -8.46608520e-01 -7.14538693e-02
-2.24464700e-01 8.60314593e-02 1.32636666e+00 -9.80410397e-01
-1.91288412e+00 1.02477980e+00 -2.07569852e-01 -8.81588876e-01
3.47870082e-01 -1.21618308e-01 -1.11578739e+00 2.62212511e-02
-3.69562000e-01 7.84084320e-01 8.47739100e-01 -8.14073861e-01
-5.74559033e-01 -2.29937002e-01 -6.75877452e-01 4.42411453e-01
-3.55880588e-01 5.20591557e-01 1.69535875e-01 -4.64398146e-01
-1.10887356e-01 -6.71020687e-01 2.43106604e-01 -7.22173274e-01
-1.23977505e-01 -6.49636626e-01 6.27046466e-01 -1.06930351e+00
1.35997832e+00 -2.41989541e+00 2.23955289e-02 1.61360633e-02
-3.99422705e-01 7.81040013e-01 -3.90193492e-01 1.38266966e-01
-2.52964467e-01 1.32788882e-01 -9.67067331e-02 -5.13390899e-01
8.57881606e-02 -5.75661249e-02 -3.69810164e-01 5.26722789e-01
3.91333610e-01 3.50553542e-01 -5.27648628e-01 -1.32241830e-01
3.19287360e-01 7.38780737e-01 -1.11277170e-01 3.28459978e-01
1.88572690e-01 3.44580352e-01 2.79655725e-01 3.34717661e-01
8.40888262e-01 5.28415918e-01 8.29467252e-02 2.05972902e-02
-3.25130671e-01 8.85568619e-01 -1.23314977e+00 1.51909137e+00
-5.01756489e-01 9.19196904e-01 4.18346018e-01 -7.89738953e-01
1.34764612e+00 7.60794938e-01 1.44942984e-01 -8.16756666e-01
2.62824088e-01 5.21033466e-01 4.73483145e-01 -1.22017086e-01
3.52035314e-01 -7.94096589e-02 -1.13014571e-01 1.98244169e-01
3.67434591e-01 -8.81807953e-02 -6.61561117e-02 -1.89251199e-01
5.80967307e-01 -5.92693388e-01 3.66039991e-01 -4.55436975e-01
1.23610079e+00 -1.39278263e-01 4.11988020e-01 6.41143501e-01
-6.96325064e-01 6.60732687e-01 3.61942537e-02 -2.76735842e-01
-7.76695609e-01 -9.60718453e-01 -3.56090277e-01 1.02225530e+00
-4.80785966e-01 -8.21060017e-02 -7.40885854e-01 -4.72638905e-01
-6.09236099e-02 9.61647928e-01 -2.72314578e-01 -2.34603986e-01
-6.60617948e-01 -3.68891031e-01 1.20394540e+00 3.91196489e-01
5.08418441e-01 -1.11234581e+00 9.71292481e-02 3.85749519e-01
-1.52955428e-01 -1.35403967e+00 -9.11404967e-01 3.28144640e-01
-3.07349056e-01 -7.86186099e-01 -1.05569172e+00 -1.11656141e+00
-2.95616910e-02 -6.33080378e-02 8.66256118e-01 -5.13391793e-01
-1.74069300e-01 4.39412266e-01 -3.17201018e-01 -1.99973434e-01
-8.19032967e-01 2.98433870e-01 7.13570237e-01 1.36494145e-01
7.74992049e-01 1.95272285e-02 -2.21705168e-01 5.00510812e-01
-4.51798469e-01 -6.07374787e-01 8.76508206e-02 1.23565507e+00
5.78847043e-02 -2.01440766e-01 1.24320972e+00 -8.36058408e-02
1.17089367e+00 -1.02667948e-02 -7.66131639e-01 1.79706514e-01
-4.37368125e-01 -2.15778351e-01 6.03350878e-01 -7.14058638e-01
-1.02216470e+00 1.54348120e-01 -6.69775426e-01 -2.92414397e-01
-4.02922690e-01 4.03510273e-01 -2.94328511e-01 -5.47119379e-02
7.64496684e-01 5.81087172e-01 3.11135888e-01 -4.92335737e-01
1.85694173e-01 1.40425587e+00 6.97186470e-01 -2.13239327e-01
1.97887897e-01 -2.84862220e-01 -7.76430070e-01 -1.41530526e+00
-2.86820769e-01 -6.87963665e-01 -6.72432363e-01 3.56376208e-02
9.05353487e-01 -1.14369237e+00 -8.93301845e-01 1.00508070e+00
-1.29692662e+00 -2.46263936e-01 2.17936978e-01 8.88067722e-01
-3.25325221e-01 3.09480578e-01 -4.44544256e-01 -1.13859832e+00
-5.54963768e-01 -1.02515709e+00 5.79750896e-01 1.28981426e-01
-4.99764204e-01 -1.09309232e+00 1.31217346e-01 8.57136920e-02
8.90674174e-01 -5.51506758e-01 6.07532203e-01 -1.43810308e+00
3.28546792e-01 -7.81592131e-02 9.91076813e-04 9.55537975e-01
1.28219500e-01 -3.37615401e-01 -1.35184729e+00 -5.40589809e-01
4.39089984e-02 -4.42775428e-01 6.29431188e-01 2.21601009e-01
6.70979559e-01 -1.83100998e-02 1.33072719e-01 4.09735650e-01
8.09205115e-01 7.38118529e-01 3.46780270e-01 -3.78916189e-02
3.18340629e-01 5.88406384e-01 3.28714579e-01 3.81213337e-01
2.07548425e-01 9.33936059e-01 -3.38871688e-01 2.62746066e-02
-3.15877318e-01 -1.42768592e-01 5.58115065e-01 9.64626491e-01
6.43617690e-01 -3.20759267e-01 -1.28548050e+00 7.02107131e-01
-1.32439780e+00 -7.98174381e-01 1.64544165e-01 2.24840903e+00
7.62443364e-01 -1.07469201e-01 2.44635239e-01 3.34794402e-01
8.41559708e-01 1.26290731e-02 -4.91169482e-01 -8.57362628e-01
-4.21759784e-01 1.85983777e-01 1.59551367e-01 8.95561159e-01
-1.16284525e+00 1.13308573e+00 7.17847157e+00 8.43992293e-01
-1.53222537e+00 -2.23717373e-02 3.80584627e-01 2.71193564e-01
3.22781771e-01 -6.83329046e-01 -1.25819695e+00 3.21292132e-01
1.63323939e+00 -3.80433977e-01 6.15079045e-01 5.95623791e-01
2.31064886e-01 4.31752093e-02 -1.12566292e+00 1.23050284e+00
2.68936872e-01 -8.01022351e-01 -2.39206433e-01 -2.64594942e-01
1.16449833e-01 2.64826983e-01 2.19013497e-01 5.36036015e-01
1.19691990e-01 -1.27029943e+00 3.45142633e-01 -7.80722052e-02
8.97806406e-01 -8.72472048e-01 7.70174384e-01 2.29285434e-01
-9.82095540e-01 -4.82425876e-02 -1.36498928e-01 1.01375423e-01
4.90750410e-02 -1.49204582e-02 -1.56296945e+00 2.28226170e-01
5.97061574e-01 5.29313207e-01 -1.97567046e-01 8.90120327e-01
1.02572277e-01 8.82869899e-01 -4.38004464e-01 -9.77646858e-02
1.01808645e-01 1.92514554e-01 5.97687781e-01 1.60216665e+00
1.70091391e-01 -2.99389690e-01 -1.93888083e-01 5.01697779e-01
-4.20246758e-02 1.70569822e-01 -6.71232820e-01 -1.97878540e-01
6.30868673e-01 8.10300052e-01 -6.98286816e-02 -1.36626989e-01
-2.82185584e-01 9.17717159e-01 7.00869858e-02 4.20205981e-01
-4.59389359e-01 -6.18689120e-01 1.03116179e+00 -2.51571804e-01
1.32065967e-01 -3.67777646e-01 -1.16776899e-01 -1.09056914e+00
-3.40397060e-01 -1.22121382e+00 4.21982318e-01 -3.58742595e-01
-1.52822518e+00 1.22245610e+00 -1.50998622e-01 -1.03935385e+00
-8.12808692e-01 -8.32407653e-01 -3.60210627e-01 1.59362185e+00
-1.69558382e+00 -7.73504674e-01 2.28563994e-01 7.42770135e-01
7.18502700e-01 -9.62682009e-01 1.19592881e+00 5.83056808e-01
-4.94006217e-01 1.15291035e+00 6.03117533e-02 3.77387404e-01
1.10864925e+00 -9.89989936e-01 6.42590404e-01 8.17577302e-01
7.27468878e-02 4.68285501e-01 3.44025493e-01 -1.13157257e-01
-1.06277668e+00 -7.73111582e-01 1.25067949e+00 -1.81320265e-01
5.82284749e-01 -6.08646810e-01 -1.20384741e+00 2.79647321e-01
5.71431696e-01 -5.86927757e-02 8.89605045e-01 -5.11343963e-02
-4.12944883e-01 -1.59706265e-01 -1.12902105e+00 1.51781231e-01
5.25372446e-01 -9.38172936e-01 -8.27059865e-01 -9.53671187e-02
6.88706875e-01 -3.10450286e-01 -8.53316963e-01 3.84864718e-01
3.35953474e-01 -6.71927214e-01 9.22228158e-01 -5.80505729e-01
-4.08176184e-01 -2.80568361e-01 -3.97855073e-01 -1.76421928e+00
2.98223142e-02 -5.94191015e-01 3.04247677e-01 1.58883548e+00
6.00598812e-01 -8.03515494e-01 1.80746287e-01 4.60101575e-01
-6.37506768e-02 6.34079501e-02 -1.44442713e+00 -9.23573792e-01
3.92680734e-01 -3.48296613e-01 3.17901194e-01 9.01663721e-01
-6.14777915e-02 6.26058877e-01 -2.90050983e-01 5.29592521e-02
1.57602668e-01 -5.55230260e-01 3.06288570e-01 -1.07616460e+00
1.15833648e-01 -6.64651155e-01 -1.44492954e-01 -8.92502964e-01
7.39275455e-01 -8.06995630e-01 1.67916059e-01 -8.91066968e-01
-5.66120684e-01 -2.00999796e-01 -4.33399618e-01 3.25973928e-01
1.44648135e-01 -7.08384588e-02 3.17220062e-01 -1.19902797e-01
-2.39604607e-01 5.92792869e-01 5.50698102e-01 -2.42330015e-01
-4.87863928e-01 2.37382025e-01 -3.32136124e-01 4.01979357e-01
8.76508236e-01 -1.32698014e-01 -9.86777842e-02 -4.51216608e-01
-6.87534869e-01 1.75600246e-01 8.19796473e-02 -1.00660443e+00
3.68305922e-01 3.58887553e-01 2.55233467e-01 -6.52248681e-01
6.14074528e-01 -5.45313060e-01 -2.37786740e-01 5.65742314e-01
-6.32645905e-01 6.98645040e-02 6.91774726e-01 -8.06497335e-02
-7.44550169e-01 -1.21766798e-01 1.14304423e+00 2.74064779e-01
-9.88950431e-01 -2.73417979e-01 -8.91131878e-01 1.78467706e-01
5.59982181e-01 1.72351189e-02 -1.03269204e-01 -3.74759048e-01
-8.47135365e-01 2.60145307e-01 -1.36905178e-01 8.73533785e-01
7.14779735e-01 -1.35731685e+00 -1.12419975e+00 7.50183463e-01
9.59613919e-02 -5.64009368e-01 1.92997664e-01 4.86034155e-01
-1.23995356e-01 9.49299932e-01 -1.61807358e-01 -6.41531587e-01
-1.52812171e+00 1.10131837e-01 8.38062406e-01 2.44056825e-02
-7.83125684e-02 1.07392442e+00 -2.12881967e-01 -8.14639270e-01
7.12841630e-01 -9.57331210e-02 -4.32928532e-01 3.99897099e-01
7.65748501e-01 3.03280175e-01 5.32724202e-01 -9.10823286e-01
-1.01074255e+00 4.54791784e-01 -2.12030455e-01 -5.04719734e-01
8.41037571e-01 -3.26404065e-01 2.70120561e-01 3.19537431e-01
1.42967546e+00 -1.64528877e-01 -8.94208729e-01 -4.33666974e-01
2.26623565e-01 -1.66588992e-01 2.89256752e-01 -1.27955759e+00
-5.72218895e-01 1.21169829e+00 1.03709924e+00 2.57696509e-01
1.09215260e+00 -3.95894021e-01 5.45067132e-01 2.60416508e-01
-1.28950505e-02 -1.16349030e+00 -1.86400205e-01 1.18670285e+00
1.15781248e+00 -1.61217988e+00 -7.43592262e-01 2.07241967e-01
-9.97900009e-01 1.18303156e+00 6.25517905e-01 -1.42376433e-04
7.77716994e-01 2.26470426e-01 5.99777341e-01 5.95055640e-01
-7.65700698e-01 -1.39736235e-01 5.05220592e-01 8.77544582e-01
8.05351615e-01 8.03864002e-02 -3.90918255e-01 6.83602810e-01
-1.79146171e-01 -2.16233179e-01 2.33457506e-01 3.48487854e-01
-2.11263418e-01 -1.09260702e+00 -5.95578134e-01 -1.67729720e-01
-4.92273331e-01 -1.40326932e-01 -6.55881047e-01 7.47187734e-01
-3.45112354e-01 1.30741167e+00 1.73565298e-01 -4.71893102e-01
4.88392979e-01 5.91198444e-01 9.16816965e-02 -2.31742442e-01
-9.06450570e-01 1.37015209e-01 2.60557383e-01 -2.19849333e-01
-1.36436731e-01 -7.30218172e-01 -1.17388320e+00 -1.91347271e-01
-9.72251371e-02 3.00905824e-01 9.41556334e-01 7.41078138e-01
3.81969154e-01 4.22953427e-01 7.87947536e-01 -5.71191907e-01
-1.00128746e+00 -1.41055691e+00 -5.71475089e-01 7.18081445e-02
8.12657654e-01 -3.94814014e-01 -6.14760220e-01 -7.31330663e-02] | [14.202017784118652, 6.562345027923584] |
360f3c0c-a454-4cf7-9a6e-621759979195 | exploring-transformers-for-behavioural | 2206.01441 | null | https://arxiv.org/abs/2206.01441v1 | https://arxiv.org/pdf/2206.01441v1.pdf | Exploring Transformers for Behavioural Biometrics: A Case Study in Gait Recognition | Biometrics on mobile devices has attracted a lot of attention in recent years as it is considered a user-friendly authentication method. This interest has also been motivated by the success of Deep Learning (DL). Architectures based on Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) have been established to be convenient for the task, improving the performance and robustness in comparison to traditional machine learning techniques. However, some aspects must still be revisited and improved. To the best of our knowledge, this is the first article that intends to explore and propose novel gait biometric recognition systems based on Transformers, which currently obtain state-of-the-art performance in many applications. Several state-of-the-art architectures (Vanilla, Informer, Autoformer, Block-Recurrent Transformer, and THAT) are considered in the experimental framework. In addition, new configurations of the Transformers are proposed to further increase the performance. Experiments are carried out using the two popular public databases whuGAIT and OU-ISIR. The results achieved prove the high ability of the proposed Transformer, outperforming state-of-the-art CNN and RNN architectures. | ['Ruben Vera-Rodriguez', 'Farzin Deravi', 'Richard Guest', 'Ruben Tolosana', 'Paula Delgado-Santos'] | 2022-06-03 | null | null | null | null | ['gait-recognition'] | ['computer-vision'] | [ 1.07146800e-02 -2.77258128e-01 1.60125315e-01 -2.19445810e-01
-6.38395101e-02 1.19939007e-01 4.87053186e-01 -3.30917954e-01
-6.12957478e-01 6.28242314e-01 -3.21317576e-02 -1.57216266e-01
-1.78460017e-01 -6.39759302e-01 -2.37221688e-01 -7.70184100e-01
-1.81661651e-01 1.43924132e-01 4.64083217e-02 -3.25662225e-01
1.29265904e-01 6.47761166e-01 -1.78998446e+00 -2.30815669e-04
7.48722374e-01 1.20430291e+00 -1.85496897e-01 1.65687069e-01
5.85953146e-02 6.49652541e-01 -4.46949959e-01 -9.61632252e-01
3.20833549e-02 -2.23393172e-01 -5.73026478e-01 -3.43524605e-01
1.35139093e-01 -6.22044913e-02 -4.68562514e-01 7.80795038e-01
8.93344998e-01 -5.46321087e-02 4.65189874e-01 -8.29714000e-01
-6.76010311e-01 6.28213882e-01 -3.89077663e-01 1.79821074e-01
2.80354738e-01 1.16277404e-01 6.88529551e-01 -8.23418379e-01
2.14929402e-01 9.39752758e-01 1.00854599e+00 4.95975524e-01
-7.76929498e-01 -8.48150611e-01 -1.13307431e-01 7.16863632e-01
-1.41963422e+00 -3.78237963e-01 9.87460554e-01 -4.66173775e-02
1.13340545e+00 1.98820345e-02 8.33403468e-01 1.53833485e+00
3.66026849e-01 7.89510012e-01 1.06182945e+00 -4.35538590e-01
5.88459671e-02 9.45329480e-03 3.11390817e-01 7.62643874e-01
5.27176678e-01 -7.65210763e-02 -6.55789077e-01 1.62490204e-01
6.41985774e-01 1.49147660e-01 -2.84916371e-01 -1.82059020e-01
-8.56852889e-01 4.78610367e-01 5.85194290e-01 1.12540174e+00
-8.22434902e-01 7.92630538e-02 5.47294378e-01 5.76856025e-02
2.01468214e-01 9.62092876e-02 -1.76120266e-01 -5.82855642e-01
-9.12324250e-01 -1.15807123e-01 6.95689559e-01 3.76104236e-01
1.94782227e-01 6.07706547e-01 3.83567661e-02 8.62961233e-01
3.17518353e-01 6.64094687e-01 7.71015406e-01 -1.32737039e-02
3.41450840e-01 6.88349366e-01 -2.97741950e-01 -1.12534726e+00
-5.93012452e-01 -8.84757102e-01 -1.56906152e+00 -2.18754366e-01
3.65584731e-01 1.51317837e-02 -7.85678029e-01 1.31866562e+00
-2.59385139e-01 4.43338692e-01 -1.37300463e-02 7.54958987e-01
7.81889021e-01 4.54006046e-01 -2.61879973e-02 -7.51020312e-02
1.53674090e+00 -3.91815394e-01 -9.27393198e-01 1.78833723e-01
1.36656612e-01 -7.51464963e-01 6.82287991e-01 6.67229295e-01
-7.62184739e-01 -9.74282026e-01 -1.07550216e+00 2.92692393e-01
-5.14973700e-01 6.31787539e-01 5.40719748e-01 1.29452968e+00
-9.88928258e-01 7.66021609e-01 -1.01493335e+00 -7.92172551e-01
3.63736808e-01 6.60751998e-01 -3.47831726e-01 4.01815504e-01
-1.44915986e+00 1.08934712e+00 2.06151307e-01 9.83723462e-01
-3.68192852e-01 -1.71799853e-01 -5.14081359e-01 2.61928011e-02
2.50185616e-02 -4.81549889e-01 8.08911681e-01 -5.94293058e-01
-1.84483790e+00 6.54546857e-01 -8.57381448e-02 -8.79538238e-01
2.93954581e-01 -3.69321764e-01 -7.15986311e-01 -1.31411701e-01
-3.73091012e-01 2.41306815e-02 5.50744891e-01 -6.47097230e-01
-3.35154593e-01 -6.05264664e-01 -2.68768370e-01 -2.56816715e-01
-9.46423113e-01 -8.90550986e-02 -2.39069134e-01 -6.75673604e-01
1.03837818e-01 -1.03448987e+00 7.59100169e-02 -4.53865051e-01
-4.70662832e-01 -2.19070375e-01 8.30520391e-01 -8.40137303e-01
1.32884467e+00 -1.88578045e+00 4.90658693e-02 3.98997158e-01
-1.08454682e-01 9.53004003e-01 1.17083266e-01 4.56594497e-01
-7.82570466e-02 -1.91346675e-01 7.06129298e-02 -3.94170731e-01
-7.45676234e-02 9.67586190e-02 -1.09507754e-01 7.02918053e-01
1.06783748e-01 9.06028628e-01 -1.38952926e-01 -1.38997987e-01
5.53863525e-01 9.33638990e-01 -7.34907165e-02 6.59682602e-02
5.23328722e-01 4.01931226e-01 -4.09607291e-01 8.23357105e-01
5.99390805e-01 1.70816053e-02 1.55520201e-01 -6.57108247e-01
-1.72528043e-01 3.81109789e-02 -1.26931310e+00 1.32005024e+00
-3.58183175e-01 5.00627875e-01 -4.72753048e-01 -1.32213461e+00
1.34454608e+00 5.82464635e-01 5.99429131e-01 -9.14744854e-01
4.75199044e-01 5.32642186e-01 2.02564597e-01 -5.94245076e-01
4.82904196e-01 3.65851000e-02 2.27884352e-01 1.83060065e-01
3.65175694e-01 5.29819071e-01 -7.31936246e-02 -4.84834403e-01
7.58833468e-01 1.38258651e-01 2.59030223e-01 -1.00353256e-01
1.06490552e+00 -8.07205439e-01 4.58498567e-01 5.44286430e-01
-1.46952882e-01 2.41970077e-01 2.41679654e-01 -6.16649032e-01
-8.93397450e-01 -7.55560338e-01 -1.37779191e-01 3.90718788e-01
-1.63927659e-01 -1.57749370e-01 -7.71553934e-01 -2.21422032e-01
-3.43177229e-01 1.95110261e-01 -6.24518871e-01 -5.99726848e-03
-7.31241941e-01 -9.22963023e-01 1.27027833e+00 6.69709444e-01
1.30876255e+00 -1.15696442e+00 -8.56510639e-01 4.12568867e-01
-2.38042191e-01 -1.29407370e+00 1.18042223e-01 8.79217219e-03
-9.87164736e-01 -9.93494153e-01 -1.14867175e+00 -5.62633812e-01
1.61084205e-01 -1.98818862e-01 6.32967293e-01 -8.40609446e-02
-1.64404392e-01 1.77189857e-01 -4.85906333e-01 -2.31749803e-01
1.06639236e-01 6.40009642e-01 3.47075939e-01 5.83236694e-01
4.81977075e-01 -6.93088770e-01 -4.50312406e-01 4.36146349e-01
-5.81900358e-01 -2.87797451e-01 9.07002568e-01 9.36531305e-01
3.10490225e-02 -1.03116453e-01 4.80212539e-01 -6.18509173e-01
6.32417738e-01 1.57365456e-01 -4.40521955e-01 2.94385344e-01
-5.01521885e-01 7.46457502e-02 7.14695454e-01 -2.59138376e-01
-1.00911891e+00 -1.15696967e-01 -6.13193333e-01 -2.44470671e-01
-1.07830912e-01 6.12901866e-01 -1.30898744e-01 -2.97779441e-01
4.84919220e-01 5.49684525e-01 2.41280291e-02 -5.59190869e-01
-1.13473453e-01 8.21623981e-01 4.92146879e-01 -3.82285416e-01
6.77178144e-01 2.52779156e-01 1.76139355e-01 -1.10442293e+00
-7.68149421e-02 -3.29731137e-01 -5.57693362e-01 -4.57511187e-01
5.99074662e-01 -5.65987468e-01 -1.36379111e+00 1.21427786e+00
-1.05085874e+00 2.83649802e-01 6.04459755e-02 6.44280851e-01
-1.62601158e-01 5.67292452e-01 -7.81680644e-01 -1.25392556e+00
-9.16237712e-01 -1.32110441e+00 6.22844040e-01 4.22005653e-01
-1.69885289e-02 -7.65358627e-01 -2.51214683e-01 1.83379620e-01
1.03291535e+00 1.52949214e-01 4.06955302e-01 -7.28051364e-01
-3.32132846e-01 -4.39820111e-01 -3.35791707e-02 3.82266164e-01
1.43918261e-01 6.26583695e-02 -1.34604132e+00 -2.29011670e-01
-3.89962317e-03 3.30390478e-03 7.59109735e-01 4.24652159e-01
7.98306346e-01 -1.87616915e-01 -2.18741223e-01 3.47988009e-01
1.26723826e+00 4.50086981e-01 1.27134430e+00 3.96953642e-01
3.87163192e-01 2.73847073e-01 9.32034552e-02 4.30226326e-01
2.34426811e-01 9.07908380e-01 3.11783552e-01 -1.66730508e-01
1.47164762e-01 5.21402434e-02 3.64528030e-01 1.00633371e+00
-8.68478775e-01 -1.39216959e-01 -9.66059089e-01 2.14802280e-01
-1.84148204e+00 -1.17354226e+00 -1.74843475e-01 2.16340327e+00
1.16535597e-01 1.66626438e-01 3.09051126e-01 8.19226980e-01
7.12620676e-01 2.24049762e-01 -2.60776550e-01 -2.77062416e-01
-2.24904731e-01 8.20497155e-01 2.98951477e-01 2.68233493e-02
-1.18165457e+00 7.55610526e-01 5.04825974e+00 7.54175484e-01
-1.60481095e+00 -2.38565989e-02 5.17058372e-01 3.60278189e-01
3.88369948e-01 -6.01055622e-01 -8.18819284e-01 4.22970235e-01
1.14244831e+00 3.01045567e-01 1.59968004e-01 6.53027534e-01
1.35879934e-01 1.64925069e-01 -4.74919200e-01 1.27247846e+00
2.86915928e-01 -1.01762915e+00 -5.47941439e-02 2.04374552e-01
9.75082740e-02 -2.19377369e-01 4.07931000e-01 2.64210314e-01
-4.31189448e-01 -9.05534029e-01 3.57541025e-01 7.38466203e-01
6.52549565e-01 -8.78684044e-01 1.62793791e+00 -1.03430480e-01
-1.61326456e+00 -1.89832374e-01 -2.74757385e-01 -4.70948592e-02
1.31181821e-01 7.62516499e-01 -4.58709985e-01 1.01077688e+00
8.13895226e-01 9.25727367e-01 -8.13947320e-01 1.19790924e+00
-2.60161217e-02 6.55910015e-01 -2.31204718e-01 -5.96669495e-01
1.92599863e-01 -1.03532843e-01 1.70232624e-01 1.19920230e+00
6.09856069e-01 -1.55137196e-01 -4.91261423e-01 5.28183281e-01
7.34117925e-02 2.75206089e-01 -5.88619769e-01 -4.72685508e-02
6.00396842e-02 1.18636453e+00 -6.95605576e-01 -1.64712727e-01
-2.09839612e-01 8.46999466e-01 -2.12691888e-01 1.51655734e-01
-8.51994395e-01 -5.54601431e-01 4.25086498e-01 -1.29471428e-03
2.90045798e-01 -1.81338966e-01 -1.71113521e-01 -1.42108023e+00
2.29782760e-01 -9.00665879e-01 1.65605173e-01 -4.22013909e-01
-1.21598232e+00 9.37846899e-01 -4.33072209e-01 -1.52976823e+00
-3.24550629e-01 -8.66317332e-01 -2.18267322e-01 8.70259643e-01
-1.31021500e+00 -1.37673450e+00 -3.43513608e-01 6.93998158e-01
1.33892849e-01 -5.87108970e-01 1.12082267e+00 6.69958830e-01
-6.45383537e-01 8.96787703e-01 -5.15311807e-02 4.91362840e-01
3.93881321e-01 -7.14746714e-01 3.99965346e-01 1.00780571e+00
1.18349969e-01 9.25322413e-01 5.25152028e-01 -2.55600512e-01
-1.53157353e+00 -5.76806188e-01 9.35713649e-01 1.95559084e-01
4.40815628e-01 -1.63903818e-01 -8.21474910e-01 2.95454115e-01
3.80343199e-01 -2.03366220e-01 7.39565194e-01 3.10925424e-01
-3.28054130e-01 -6.04179204e-01 -1.16031098e+00 6.41525328e-01
8.60014260e-01 -5.72211683e-01 -3.44346374e-01 -4.13734049e-01
-3.19418281e-01 -2.67680705e-01 -7.80369699e-01 7.06303000e-01
1.18790066e+00 -1.28999889e+00 8.77343714e-01 -8.58450010e-02
-1.36029467e-01 -2.63478339e-01 -2.39108894e-02 -1.04894495e+00
-6.06400669e-02 -4.65582967e-01 -1.90310121e-01 1.30738735e+00
1.18763894e-01 -9.73912895e-01 9.93003547e-01 3.14110398e-01
1.77593648e-01 -8.50750268e-01 -1.27577090e+00 -8.37401927e-01
-4.27916437e-01 -4.46181715e-01 4.80579615e-01 6.94126368e-01
6.90050200e-02 4.36361194e-01 -7.69559741e-01 -8.63327309e-02
5.61621487e-01 -3.16376448e-01 6.56285286e-01 -1.49966216e+00
-1.32279962e-01 -5.77024162e-01 -1.17151761e+00 -8.10129166e-01
-1.52142152e-01 -3.34520131e-01 -4.52697664e-01 -1.24904525e+00
-1.70375153e-01 -6.07568957e-02 -4.55120116e-01 6.95090771e-01
2.28174210e-01 5.56571662e-01 3.00205439e-01 -1.62031148e-02
-2.48760864e-01 6.34954572e-01 7.73384571e-01 -2.48875216e-01
-3.86735238e-02 2.76018679e-01 -3.60833973e-01 4.18563575e-01
9.08974290e-01 1.46245420e-01 -7.64754862e-02 1.36543484e-02
-1.08333565e-01 -8.05751830e-02 1.89074486e-01 -1.63764215e+00
5.09221554e-01 7.43076265e-01 6.28006756e-01 -6.23980939e-01
6.95555806e-01 -9.84365225e-01 4.33257282e-01 9.07369137e-01
4.72271862e-03 4.43409324e-01 2.31964305e-01 4.91417535e-02
-3.98485988e-01 2.73123235e-02 8.02393854e-01 2.06194326e-01
-5.14882982e-01 2.43083894e-01 -5.88386834e-01 -6.24767125e-01
7.73631155e-01 -5.66513717e-01 -1.74297825e-01 -3.48940194e-01
-6.76512361e-01 -4.91736978e-01 -2.60577619e-01 4.47441161e-01
8.29878092e-01 -1.19089758e+00 -5.36586642e-01 3.82614762e-01
-8.38337094e-02 -7.92076647e-01 3.99299294e-01 9.95065928e-01
-4.44561690e-01 9.04976487e-01 -7.20732093e-01 -5.32066464e-01
-1.40266454e+00 2.24746078e-01 6.54326499e-01 -5.51157773e-01
-4.97683793e-01 4.34177488e-01 -7.39517093e-01 -3.37386727e-01
4.49383736e-01 -3.07337433e-01 -7.35175490e-01 1.67104185e-01
3.67856473e-01 5.39585412e-01 5.13494611e-01 -8.15926552e-01
-6.11385405e-01 9.71652150e-01 2.47314852e-02 -5.95697463e-02
1.40635550e+00 3.71881068e-01 -6.76884130e-02 3.34937274e-01
8.69709373e-01 -3.10766101e-01 -4.12549049e-01 -1.20416120e-01
1.37322679e-01 -1.26303717e-01 -2.56744534e-01 -7.35149145e-01
-1.47955787e+00 1.16538000e+00 1.18039668e+00 2.36438334e-01
1.23379695e+00 -7.98297346e-01 8.09909642e-01 5.58450401e-01
8.58981907e-01 -9.28162158e-01 4.56177406e-02 5.44230163e-01
5.59510350e-01 -5.96391976e-01 -3.28896195e-01 -7.56866261e-02
-3.56467038e-01 1.58152449e+00 5.12659729e-01 -1.18359968e-01
8.74709845e-01 2.26989284e-01 -7.34146461e-02 4.79346961e-02
-1.73018366e-01 -4.63866413e-01 2.49285743e-01 6.29043043e-01
8.90221119e-01 -1.50630429e-01 -5.93775451e-01 5.74122488e-01
-7.65963346e-02 4.74155337e-01 3.04764241e-01 6.75524235e-01
4.34851199e-02 -1.25363290e+00 -4.11935717e-01 4.59960401e-01
-7.55262434e-01 7.82157108e-02 -9.70672369e-02 8.03804100e-01
1.25702322e-01 8.92754197e-01 -3.79846752e-01 -8.10042024e-01
5.74262559e-01 8.32102224e-02 5.24165928e-01 1.02201790e-01
-1.01735592e+00 -6.22393377e-02 -5.33582047e-02 -3.55866045e-01
-7.72898138e-01 -5.39897561e-01 -6.82551146e-01 -5.04418194e-01
-5.34327030e-01 2.67432570e-01 7.16969490e-01 9.96306598e-01
2.34919265e-01 6.43556535e-01 3.77095670e-01 -8.36451232e-01
-2.73856908e-01 -1.24722624e+00 -6.66010797e-01 3.94297615e-02
-9.36360061e-02 -7.24078953e-01 1.91353023e-01 -2.33211577e-01] | [13.945267677307129, 1.508736491203308] |
be7cb7ff-bc99-4605-b328-f69cf6f6818f | person-search-by-multi-scale-matching-1 | null | null | http://openaccess.thecvf.com/content_ECCV_2018/html/Xu_Lan_Person_Search_by_ECCV_2018_paper.html | http://openaccess.thecvf.com/content_ECCV_2018/papers/Xu_Lan_Person_Search_by_ECCV_2018_paper.pdf | Person Search by Multi-Scale Matching | We consider the problem of person search in unconstrained scene images. Existing methods usually focus on improving the person detection accuracy to mitigate negative effects imposed by misalignment, mis-detections, and false alarms resulted from noisy people auto-detection. In contrast to previous studies, we show that sufficiently reliable person instance cropping is achievable by slightly improved state-of-the-art deep learning object detectors (e.g. Faster-RCNN), and the under-studied multi-scale matching problem in person search is a more severe barrier. In this work, we address this multi-scale person search challenge by proposing a Cross-Level Semantic Alignment (CLSA) deep learning approach capable of learning more discriminative identity feature representations in a unified end-to-end model. This is realised by exploiting the in-network feature pyramid structure of a deep neural network enhanced by a novel cross pyramid-level semantic alignment loss function. This favourably eliminates the need for constructing a computationally expensive image pyramid and a complex multi-branch network architecture. Extensive experiments show the modelling advantages and performance superiority of CLSA over the state-of-the-art person search and multi-scale matching methods on two large person search benchmarking datasets: CUHK-SYSU and PRW. | ['Shaogang Gong ', 'Xiatian Zhu ', 'Xu Lan '] | 2018-09-01 | null | null | null | eccv-2018-9 | ['person-search'] | ['computer-vision'] | [ 2.13098690e-01 -3.41136009e-01 3.51216137e-01 -2.69237906e-01
-8.79197598e-01 -3.31712753e-01 7.41561651e-01 5.61189801e-02
-1.14151263e+00 4.74678934e-01 2.39749521e-01 5.02181649e-01
-3.29900235e-01 -6.90435827e-01 -7.22631812e-01 -3.57997924e-01
1.81002125e-01 8.74074876e-01 3.05141300e-01 -2.66036242e-01
-7.41537735e-02 4.29442763e-01 -1.99754322e+00 3.41841161e-01
6.73279881e-01 9.22460556e-01 6.60258755e-02 6.07209563e-01
4.21531528e-01 2.47536883e-01 -8.11298668e-01 -9.35959220e-01
5.98078191e-01 -1.17248341e-01 -7.84226716e-01 -3.64778899e-02
1.25185406e+00 -2.80729413e-01 -3.89501184e-01 1.11525071e+00
1.18916261e+00 2.10998788e-01 3.61145824e-01 -1.21895111e+00
-6.38424993e-01 1.33052677e-01 -5.30423999e-01 3.95935386e-01
6.29813373e-01 3.86576533e-01 9.47495043e-01 -8.69996190e-01
5.25571883e-01 1.75839543e+00 1.17781770e+00 5.25311708e-01
-1.34687436e+00 -5.87651491e-01 3.03341132e-02 4.65662062e-01
-1.73518765e+00 -2.73623079e-01 3.29043716e-01 -3.16490084e-01
1.04952645e+00 3.63733858e-01 5.74158430e-01 1.41043520e+00
-3.35264504e-01 1.03454041e+00 9.97453868e-01 -5.59664249e-01
-3.09685051e-01 1.94415510e-01 1.60666630e-01 8.43329430e-01
5.21198928e-01 4.63270307e-01 -6.36154652e-01 -3.15686613e-01
7.19797313e-01 9.21200553e-04 -9.55737159e-02 -4.74168360e-01
-1.06946945e+00 7.76464045e-01 7.81051278e-01 4.71713573e-01
-5.19100428e-01 1.71136394e-01 6.06631875e-01 1.86598644e-01
1.86794192e-01 4.39366311e-01 -3.06427538e-01 2.46864915e-01
-1.21908796e+00 8.90687227e-01 5.41117966e-01 6.78972960e-01
3.37854475e-01 -2.04326674e-01 -7.72822201e-01 1.01185477e+00
3.36109847e-02 5.77553213e-01 5.11336863e-01 -5.25347888e-01
3.73599201e-01 6.56985521e-01 1.98630512e-01 -9.04056311e-01
-6.10002697e-01 -9.97665167e-01 -7.84465730e-01 1.71153396e-01
7.54241347e-01 2.14916334e-01 -9.19238567e-01 1.79579568e+00
2.46309608e-01 1.38425738e-01 -2.10212916e-01 1.27688456e+00
8.60170484e-01 -9.12210345e-02 2.98444092e-01 5.51245928e-01
2.01885486e+00 -1.21093726e+00 -3.32497448e-01 -4.29607540e-01
2.48447567e-01 -6.77386761e-01 9.61537600e-01 1.01005271e-01
-1.18435383e+00 -1.09333813e+00 -8.67068291e-01 -1.64877191e-01
-7.41915345e-01 4.51675177e-01 4.58282232e-01 9.89331007e-01
-1.11253679e+00 4.78670925e-01 -2.75621951e-01 -6.43125296e-01
6.37933314e-01 5.27480721e-01 -5.03382504e-01 -1.26942113e-01
-1.31044447e+00 1.13060021e+00 3.42622548e-01 1.73398212e-01
-4.36105430e-01 -7.67174900e-01 -8.33122969e-01 1.92126483e-01
4.08065975e-01 -1.32077849e+00 1.06866670e+00 -9.07361984e-01
-8.28390718e-01 1.41992199e+00 -6.68076724e-02 -6.53961837e-01
1.07092786e+00 -6.05053425e-01 -3.70773733e-01 1.16726987e-01
4.88625467e-01 7.45769083e-01 7.95044184e-01 -1.02088130e+00
-8.13460290e-01 -6.37213290e-01 -1.62332520e-01 1.94564015e-01
-1.74405694e-01 6.17697716e-01 -7.85510778e-01 -8.12533259e-01
-3.90416294e-01 -8.98576498e-01 -3.18472207e-01 6.04487322e-02
-1.84185863e-01 -4.37290162e-01 3.24774683e-01 -9.03724730e-01
9.17414725e-01 -1.69634819e+00 1.05865747e-01 2.42209986e-01
-1.58432692e-01 5.21351278e-01 -3.58593643e-01 7.95490965e-02
4.44551222e-02 -4.28330302e-01 9.68506262e-02 -7.89366424e-01
2.16135159e-01 -1.03353433e-01 1.51796222e-01 6.52812481e-01
3.62719558e-02 1.17214727e+00 -6.63091600e-01 -4.26201999e-01
2.60215431e-01 6.36210859e-01 -4.32402253e-01 -3.11166886e-03
3.32216084e-01 1.84804782e-01 -7.72354528e-02 9.27277923e-01
5.43909788e-01 -2.74083227e-01 -3.23832452e-01 -2.70867348e-01
1.13323756e-01 -1.09521158e-01 -1.51162481e+00 1.89715910e+00
-2.47941807e-01 2.93377578e-01 3.89524102e-02 -8.34156334e-01
6.10805213e-01 1.45064712e-01 2.04771712e-01 -9.58631158e-01
1.12924345e-01 2.08138749e-01 -1.89391807e-01 -2.30768427e-01
4.80906785e-01 3.86023939e-01 -2.65200973e-01 -8.27802531e-03
1.70936108e-01 5.02136469e-01 1.69351891e-01 -2.37736236e-02
8.38851929e-01 9.05696079e-02 2.20854074e-01 -4.69617575e-01
9.86286104e-01 -1.31104305e-01 3.35290939e-01 1.39805686e+00
-5.04143178e-01 6.81276798e-01 -2.29853421e-01 -8.36336315e-01
-1.09110856e+00 -1.00445223e+00 -6.80048466e-02 1.44115460e+00
2.19408289e-01 -3.08323592e-01 -8.54769766e-01 -4.64228511e-01
3.84096175e-01 2.34157249e-01 -6.38825297e-01 -1.15084596e-01
-8.66511524e-01 -1.01427805e+00 1.05841780e+00 6.19130790e-01
7.77580619e-01 -1.12379146e+00 -8.28154981e-01 2.63470918e-01
-2.61134297e-01 -1.39514554e+00 -6.69844627e-01 -2.43550748e-01
-1.21120185e-01 -1.17200506e+00 -1.11419225e+00 -8.92755151e-01
4.09542024e-01 9.46950614e-02 1.36219418e+00 1.13658525e-01
-1.11631966e+00 5.79858840e-01 -1.45505304e-02 -3.39914173e-01
1.73161533e-02 1.17239520e-01 2.84681290e-01 1.07200798e-02
7.94389844e-01 -2.53148645e-01 -9.60341692e-01 3.66208702e-01
-3.86186957e-01 -2.57814258e-01 7.22598314e-01 9.84214604e-01
3.10094863e-01 -1.81762248e-01 3.22476447e-01 -1.93755984e-01
6.37067676e-01 3.91361788e-02 -6.40372455e-01 5.15372634e-01
-5.72866976e-01 -1.54244397e-02 1.08046383e-01 -3.86461049e-01
-8.94995511e-01 2.09831566e-01 -1.95178509e-01 -2.68291712e-01
-4.79000062e-01 -2.90129662e-01 7.98866600e-02 -3.86758059e-01
8.14660251e-01 3.64382654e-01 -1.52769417e-01 -5.51669061e-01
3.17867488e-01 4.65062261e-01 1.06120574e+00 -5.34455776e-01
8.63789976e-01 5.82880795e-01 -2.75701378e-02 -4.91818517e-01
-7.86758900e-01 -1.08087468e+00 -8.50975096e-01 -1.91231206e-01
9.34341311e-01 -1.33851039e+00 -9.80857670e-01 7.27943480e-01
-1.16554308e+00 1.33776903e-01 -1.18588217e-01 1.55299217e-01
-3.52133363e-01 5.35230100e-01 -5.46216905e-01 -8.14909101e-01
-7.74763048e-01 -9.83557224e-01 1.60730529e+00 4.18834239e-01
-2.95652747e-01 -5.90628684e-01 -6.02040626e-02 8.35031927e-01
4.61707652e-01 2.22295046e-01 9.40720215e-02 -8.19705248e-01
-5.22256851e-01 -5.83957434e-01 -6.13708138e-01 4.10575084e-02
-3.04461002e-01 -9.52546120e-01 -1.12435436e+00 -5.84850430e-01
-5.21744370e-01 -3.99953336e-01 1.22021365e+00 2.57102609e-01
9.80564356e-01 -3.42568918e-03 -5.20599723e-01 6.50592566e-01
1.44043088e+00 -4.82968271e-01 3.29404742e-01 6.53549671e-01
6.97813153e-01 6.13495827e-01 3.14535856e-01 3.02104324e-01
5.33568978e-01 1.40110719e+00 7.66922683e-02 -3.41168553e-01
-5.23254514e-01 -1.32568449e-01 -3.46897170e-02 -4.69338655e-01
-3.36810380e-01 3.64927389e-02 -8.70458782e-01 7.37644315e-01
-2.22400093e+00 -1.29745412e+00 -8.23131353e-02 2.09539676e+00
6.46769702e-01 -1.25964116e-02 6.81229651e-01 1.19686536e-02
8.10718596e-01 -8.88433009e-02 -1.63227946e-01 8.34644139e-02
-5.33791006e-01 3.78048152e-01 6.73795581e-01 2.99909800e-01
-1.63831902e+00 9.24024343e-01 5.43125057e+00 1.08776307e+00
-3.85097355e-01 3.86831731e-01 2.37129748e-01 -3.38585615e-01
5.18891454e-01 -5.74027002e-01 -1.28792512e+00 4.33429927e-01
4.48424101e-01 3.59319657e-01 3.10457319e-01 9.32128012e-01
-1.31164566e-01 -5.77676035e-02 -1.18370497e+00 1.62580979e+00
5.00283837e-01 -1.14317334e+00 -8.73698592e-02 -1.31392166e-01
4.42306042e-01 2.25166008e-02 5.67847900e-02 4.62529510e-01
7.60764480e-02 -9.91837382e-01 8.24360311e-01 4.63771611e-01
4.20567870e-01 -6.13177061e-01 8.89341891e-01 1.88919768e-01
-1.42392397e+00 -4.34498101e-01 -2.73405313e-01 1.40795097e-01
2.58849353e-01 1.97368473e-01 -4.44531649e-01 6.24901950e-01
1.25554550e+00 1.15411907e-01 -1.09021711e+00 1.40255308e+00
1.72764510e-01 -1.20216042e-01 -5.35809398e-01 1.55537009e-01
2.67293155e-01 2.05015376e-01 5.46777308e-01 1.75483048e+00
4.91592921e-02 -3.86445463e-01 4.05458778e-01 8.82559836e-01
1.24036744e-01 -3.55836749e-02 -1.75007924e-01 6.95598841e-01
1.66867599e-01 1.27688336e+00 -5.04370034e-01 -4.94227678e-01
-4.39724863e-01 1.49119079e+00 3.46095055e-01 1.29331231e-01
-8.11296701e-01 2.82763317e-03 7.71050274e-01 1.54767513e-01
5.19177318e-01 2.24765122e-01 -1.67901933e-01 -1.04685533e+00
3.02609295e-01 -9.77393031e-01 8.63638222e-01 -4.43074107e-01
-1.62534189e+00 4.87927139e-01 -9.94529128e-02 -6.95370197e-01
-2.44827613e-01 -5.70880949e-01 -3.84909600e-01 1.15587318e+00
-1.61964619e+00 -1.81426787e+00 -4.38096970e-01 8.93291593e-01
6.10094786e-01 -5.11524737e-01 8.33167255e-01 7.75879502e-01
-3.76183152e-01 1.25782669e+00 -3.58183682e-01 4.06864941e-01
8.11786771e-01 -1.18343091e+00 6.53663695e-01 9.35254991e-01
2.07775384e-01 5.16613841e-01 6.83258176e-01 -6.26641929e-01
-9.70770657e-01 -8.66489172e-01 1.16624415e+00 -8.13323677e-01
3.08278531e-01 -5.61147273e-01 -6.54733777e-01 2.59154409e-01
3.79680693e-02 1.03336334e-01 5.03874242e-01 2.51158237e-01
-6.78106248e-01 1.06150582e-02 -1.20089555e+00 4.84168500e-01
1.52099013e+00 -5.91526687e-01 -6.34226859e-01 4.86332476e-01
2.35855967e-01 -2.84267783e-01 -5.28877556e-01 4.68684912e-01
7.80634284e-01 -1.07800674e+00 1.95415461e+00 -5.94589889e-01
-2.58031368e-01 -2.35392630e-01 -3.78412046e-02 -8.02207768e-01
-6.22633755e-01 -3.58483136e-01 9.80704129e-02 9.14953291e-01
4.76346817e-03 -3.12550455e-01 8.08475852e-01 6.16298378e-01
2.62848556e-01 -3.71604919e-01 -1.21000481e+00 -1.10787499e+00
-3.38648826e-01 -1.45845506e-02 4.01034504e-01 5.63697517e-01
-5.46390593e-01 1.71978384e-01 -6.50358498e-01 4.41153407e-01
1.23443615e+00 3.79210263e-02 8.82140279e-01 -1.41459525e+00
-5.10287046e-01 -7.78598487e-01 -6.53326094e-01 -6.47917867e-01
7.43008107e-02 -7.62053728e-01 -1.74303398e-01 -1.34232807e+00
4.66828912e-01 -1.71193287e-01 -3.87790382e-01 2.37518549e-01
-5.97690761e-01 5.21572471e-01 4.75874752e-01 2.13066027e-01
-8.81886899e-01 3.63162041e-01 7.27770448e-01 -1.40838772e-01
4.31364290e-02 1.42450854e-01 -4.67746228e-01 6.08573258e-01
4.55582023e-01 -2.72023290e-01 1.53331056e-01 -3.78955752e-01
-9.64904390e-03 -4.87690747e-01 1.29495645e+00 -1.32053077e+00
5.97934842e-01 6.70552433e-01 8.88337076e-01 -5.64984024e-01
6.49748385e-01 -7.09141493e-01 1.21021755e-02 8.75141740e-01
-1.95080101e-01 1.33138731e-01 1.89316556e-01 4.76798773e-01
-1.27716988e-01 -3.62451673e-02 8.72232258e-01 -4.33397591e-01
-8.77792537e-01 2.53004640e-01 1.89172849e-01 -1.31989062e-01
7.41715729e-01 -4.71870780e-01 -1.87288299e-01 -6.39035087e-03
-6.67401314e-01 3.25871319e-01 1.13372304e-01 8.31626654e-01
2.96694010e-01 -1.29871452e+00 -9.54751551e-01 2.66797543e-01
1.96386084e-01 -4.53641623e-01 3.39719892e-01 5.89976251e-01
-1.58649191e-01 5.94261646e-01 -2.83368915e-01 -6.00784600e-01
-1.66621530e+00 4.79317486e-01 6.39625192e-01 -6.39732778e-01
-5.86226463e-01 1.27842820e+00 6.84230477e-02 -5.96232116e-01
4.79615450e-01 2.88739473e-01 -1.32636458e-01 8.31387714e-02
6.37950420e-01 4.85905528e-01 5.71412370e-02 -7.96892524e-01
-7.20395207e-01 7.82072008e-01 -1.55136526e-01 1.44507676e-01
1.10662651e+00 1.14880232e-02 1.54443190e-01 -4.20657426e-01
8.27456474e-01 -3.52134138e-01 -9.15256083e-01 -5.64283907e-01
1.66437551e-01 -4.44787323e-01 -2.88146168e-01 -1.03720248e+00
-6.56665683e-01 4.36396450e-01 1.14677477e+00 -1.73221275e-01
8.90992999e-01 2.49267310e-01 8.43027294e-01 4.51321393e-01
3.41406405e-01 -1.50048542e+00 1.28161743e-01 1.77188337e-01
9.55482423e-01 -1.54518902e+00 4.82086949e-02 -2.88744509e-01
-3.82290035e-01 6.53797984e-01 7.42173791e-01 -2.01272845e-01
2.59656966e-01 -1.15514852e-01 -9.20228362e-02 -3.09885532e-01
-4.16389219e-02 -7.73390532e-01 6.93090737e-01 6.54942930e-01
-8.14700276e-02 -1.41594941e-02 -3.73470969e-02 7.04231918e-01
-2.62502432e-01 -5.40559217e-02 -3.46792936e-01 5.71043015e-01
-2.46280089e-01 -1.04302895e+00 -8.12149823e-01 1.43597081e-01
-5.58915198e-01 -1.70112222e-01 -3.24253350e-01 7.84963191e-01
5.80110908e-01 7.25458682e-01 -2.17900947e-02 4.41277996e-02
7.11352944e-01 2.18326449e-01 7.69612551e-01 -1.76712334e-01
-1.24521160e+00 -2.20493019e-01 1.72505453e-01 -7.94291496e-01
-5.93303859e-01 -8.86294901e-01 -5.04507244e-01 -1.02620721e-01
-2.51372606e-01 -3.78798932e-01 4.49673980e-01 9.13660467e-01
3.88033956e-01 4.59205687e-01 -1.78850457e-01 -9.45022821e-01
-8.22437882e-01 -8.74533951e-01 -1.13794699e-01 9.54833329e-01
2.33523905e-01 -8.03857625e-01 9.02903378e-02 -1.90248936e-01] | [14.827638626098633, 0.8070068955421448] |
59b46d4a-2442-4203-8013-27bbd288ac3c | selective-pseudo-label-clustering | 2107.10692 | null | https://arxiv.org/abs/2107.10692v1 | https://arxiv.org/pdf/2107.10692v1.pdf | Selective Pseudo-label Clustering | Deep neural networks (DNNs) offer a means of addressing the challenging task of clustering high-dimensional data. DNNs can extract useful features, and so produce a lower dimensional representation, which is more amenable to clustering techniques. As clustering is typically performed in a purely unsupervised setting, where no training labels are available, the question then arises as to how the DNN feature extractor can be trained. The most accurate existing approaches combine the training of the DNN with the clustering objective, so that information from the clustering process can be used to update the DNN to produce better features for clustering. One problem with this approach is that these ``pseudo-labels'' produced by the clustering algorithm are noisy, and any errors that they contain will hurt the training of the DNN. In this paper, we propose selective pseudo-label clustering, which uses only the most confident pseudo-labels for training the~DNN. We formally prove the performance gains under certain conditions. Applied to the task of image clustering, the new approach achieves a state-of-the-art performance on three popular image datasets. Code is available at https://github.com/Lou1sM/clustering. | ['Thomas Lukasiewicz', 'Louis Mahon'] | 2021-07-22 | null | null | null | null | ['image-clustering'] | ['computer-vision'] | [ 9.61230770e-02 -6.71252087e-02 -2.47727096e-01 -5.57396889e-01
-7.68248618e-01 -6.87987268e-01 3.59050721e-01 1.27578348e-01
-5.65151274e-01 3.62090021e-01 7.42949098e-02 3.34347598e-02
-1.59763172e-01 -5.17819464e-01 -5.91445982e-01 -1.26020062e+00
-3.13195959e-02 7.41847515e-01 -9.88785550e-02 4.41464365e-01
8.99800435e-02 3.77386928e-01 -1.76888394e+00 2.82365739e-01
7.50251472e-01 1.03908515e+00 2.96236157e-01 5.10943711e-01
-1.57276437e-01 7.32951939e-01 -5.87838888e-01 -8.34697634e-02
3.86515498e-01 -4.09488648e-01 -8.64128530e-01 3.75972092e-01
1.25006912e-02 -2.12951601e-01 -3.31285179e-01 1.36233747e+00
4.38641489e-01 2.77818382e-01 9.29372370e-01 -1.16236937e+00
-4.51850593e-01 8.64720166e-01 -5.08227229e-01 -1.35664389e-01
-1.94252059e-01 -8.16940367e-02 1.00736117e+00 -8.29237878e-01
6.24493182e-01 1.14697278e+00 4.69801486e-01 7.83968806e-01
-1.39653647e+00 -7.15990901e-01 6.69096187e-02 1.32038221e-01
-1.72316098e+00 -5.27444959e-01 8.12471688e-01 -4.62571442e-01
3.65488410e-01 -3.68832052e-02 3.40997130e-01 9.47380006e-01
-4.23353910e-01 9.99800682e-01 7.63432682e-01 -3.47375065e-01
5.86343169e-01 4.11464386e-02 2.68822402e-01 3.79184127e-01
1.72728494e-01 -7.48441219e-02 -9.22039300e-02 -1.79888550e-02
3.90545368e-01 2.99977243e-01 -1.85910836e-01 -5.86599946e-01
-9.61999416e-01 1.02615607e+00 6.70799792e-01 5.16436100e-01
-4.41283226e-01 1.82827815e-01 3.05561900e-01 5.13833351e-02
4.26302522e-01 5.84125459e-01 -4.20539945e-01 -6.00409545e-02
-1.08578312e+00 -1.11515298e-01 6.80900395e-01 8.21910441e-01
9.42251801e-01 -3.22710544e-01 1.41911983e-01 9.61811841e-01
3.98798078e-01 1.31214768e-01 4.51044172e-01 -1.39585710e+00
9.67737511e-02 6.11209333e-01 -1.87314674e-01 -7.81554580e-01
-4.90843266e-01 -4.13131952e-01 -1.22734499e+00 1.15190931e-01
2.70596296e-01 -2.65418023e-01 -1.27039611e+00 1.86075163e+00
3.88689667e-01 2.03272149e-01 1.18145078e-01 8.58397722e-01
6.09117627e-01 6.49532497e-01 -1.22460000e-01 -2.08941206e-01
8.51236999e-01 -7.75642455e-01 -7.64431953e-01 -6.24640696e-02
9.12912786e-01 -6.54159069e-01 6.15067780e-01 5.07460833e-01
-7.69286633e-01 -4.95492488e-01 -8.47517729e-01 8.40596557e-02
-3.91987294e-01 3.83600928e-02 5.12252450e-01 3.37049216e-01
-1.25159872e+00 7.34568298e-01 -1.02376258e+00 -3.60625237e-01
7.79628992e-01 6.77291393e-01 -1.82899624e-01 -4.67885375e-01
-9.63740945e-01 5.06934643e-01 6.96277797e-01 1.65995926e-01
-9.03569043e-01 -2.87724406e-01 -6.82774842e-01 2.02398766e-02
4.49042320e-01 -2.49862552e-01 1.28729713e+00 -1.03084683e+00
-1.04230678e+00 7.86796391e-01 -2.11129874e-01 -2.89141446e-01
2.03454211e-01 8.38734210e-02 -1.16488807e-01 3.88388783e-01
2.30070189e-01 1.13169849e+00 6.94625914e-01 -1.79199171e+00
-7.85234034e-01 -4.04510736e-01 -4.05673951e-01 2.98149854e-01
-3.75049919e-01 -2.09104434e-01 -8.15406561e-01 -4.49802548e-01
3.07832241e-01 -1.13871813e+00 -4.97510284e-01 -2.16099024e-01
-5.88898957e-01 -4.88367796e-01 9.49850261e-01 -2.77186126e-01
1.17986393e+00 -2.43067741e+00 2.41802469e-01 5.23755312e-01
5.99495292e-01 4.23792899e-01 -1.44102395e-01 2.23346710e-01
-1.53001577e-01 2.53446072e-01 -5.01976132e-01 -6.19654655e-01
1.46706954e-01 4.16365325e-01 1.02045313e-01 5.49890697e-01
1.09253824e-01 6.38572872e-01 -7.27654219e-01 -4.96440381e-01
2.60727227e-01 5.19802272e-01 -4.43558782e-01 3.14899892e-01
-1.78164765e-01 4.64972138e-01 -2.35163212e-01 5.10991395e-01
6.21320546e-01 -5.16870022e-01 3.02271456e-01 -7.16537088e-02
1.27842531e-01 1.73093811e-01 -1.33153379e+00 1.59720206e+00
7.52860233e-02 6.95198596e-01 1.91459730e-01 -1.34859526e+00
6.93198562e-01 2.71176219e-01 8.44978333e-01 -2.92918324e-01
3.87424439e-01 5.84998541e-02 2.07844540e-01 -3.05154085e-01
6.27348199e-02 -5.46638714e-03 -1.09488051e-02 6.64093912e-01
1.29230961e-01 1.38534561e-01 3.89612883e-01 4.31830555e-01
1.10473156e+00 -4.15876448e-01 -8.93819407e-02 -2.55343914e-01
1.01656452e-01 1.02430828e-01 7.26482749e-01 7.82319486e-01
-2.09304497e-01 6.62474513e-01 4.63317841e-01 -2.98347265e-01
-1.12431860e+00 -1.01146686e+00 -2.27187440e-01 9.51458037e-01
7.43806362e-02 -2.63183892e-01 -9.47601378e-01 -7.25590467e-01
-1.21300593e-01 3.78999114e-01 -6.61463559e-01 -3.37435514e-01
-3.15167993e-01 -8.74567449e-01 4.22681302e-01 5.46850920e-01
3.25160861e-01 -8.75853956e-01 -1.89921901e-01 2.24044666e-01
-1.68502286e-01 -1.05946386e+00 -1.91734046e-01 6.82782888e-01
-8.34305048e-01 -1.01699686e+00 -5.55497229e-01 -1.12874973e+00
9.78717089e-01 2.80334532e-01 1.03232241e+00 3.29863578e-01
-1.45483717e-01 9.09738317e-02 -5.25491953e-01 -1.50770083e-01
-4.69661295e-01 3.08573216e-01 2.19979838e-01 -3.19404006e-02
7.77148485e-01 -5.24049640e-01 -6.09752893e-01 9.42009166e-02
-1.00931251e+00 -1.64566502e-01 6.58055127e-01 7.08281219e-01
6.71937048e-01 6.91713929e-01 5.51284134e-01 -1.07144606e+00
4.00390923e-01 -4.83696967e-01 -3.47459108e-01 -1.37948468e-01
-5.25802374e-01 1.58197880e-01 8.27424824e-01 -5.07889032e-01
-5.57745695e-01 5.70317984e-01 -1.95850730e-01 -6.70264363e-01
-5.52108884e-01 4.00069952e-01 -3.18958193e-01 2.74393946e-01
5.31920910e-01 2.25037965e-03 3.80277261e-02 -5.81561387e-01
4.80170518e-01 1.00063097e+00 6.36756659e-01 -4.17073369e-01
7.22737074e-01 5.57398498e-01 -1.90460607e-01 -5.79095244e-01
-1.03181922e+00 -8.05168569e-01 -1.02837431e+00 -6.87478855e-02
1.04988265e+00 -8.71104240e-01 -6.12664580e-01 4.05197114e-01
-9.14771378e-01 -3.91763031e-01 -1.94232166e-01 4.74678546e-01
-3.75916570e-01 2.39233628e-01 -6.44156754e-01 -5.72812736e-01
7.60222040e-03 -1.28144634e+00 7.78512359e-01 2.13450760e-01
-1.83960512e-01 -9.47275579e-01 -1.45116508e-01 1.04989439e-01
-4.78989445e-02 1.15206875e-01 9.79139149e-01 -1.07633901e+00
-3.95371407e-01 -1.96949363e-01 -1.89486653e-01 7.33855247e-01
2.14120388e-01 5.00759892e-02 -1.11098242e+00 -3.61464262e-01
1.16230309e-01 -3.98820579e-01 1.08625472e+00 6.25334084e-01
1.43034375e+00 -2.62063384e-01 -4.55579549e-01 5.96874952e-01
1.33316660e+00 3.04252297e-01 4.32918340e-01 7.06352144e-02
1.01199305e+00 5.62348843e-01 3.70469630e-01 3.09972018e-01
4.50604498e-01 2.86270231e-01 4.11363989e-01 -2.07429841e-01
6.35788366e-02 2.10138217e-01 2.31540054e-02 1.11744654e+00
3.10472190e-01 -3.22898924e-01 -1.16282737e+00 6.03561044e-01
-2.11301589e+00 -8.08698356e-01 1.67974946e-03 2.09022379e+00
9.60001051e-01 5.20552248e-02 1.13902852e-01 3.37049454e-01
1.05276155e+00 -1.12802193e-01 -7.37677991e-01 -5.03550097e-03
6.89435154e-02 -3.34470756e-02 4.00975168e-01 3.76197457e-01
-1.44227684e+00 8.93886447e-01 5.84706259e+00 7.57552564e-01
-9.16488290e-01 -1.19667038e-01 8.13724041e-01 -1.40410662e-01
4.68724966e-02 -1.86980382e-01 -7.05161393e-01 6.23144865e-01
9.84203577e-01 2.52317458e-01 5.05024374e-01 8.26599300e-01
1.74811706e-01 -2.22787559e-01 -1.36615396e+00 1.01777828e+00
-4.53229621e-02 -1.23896897e+00 7.87771831e-04 3.43576998e-01
9.81216550e-01 2.43297964e-01 2.26391867e-01 1.52743205e-01
8.24904263e-01 -9.95702624e-01 2.86126167e-01 2.71487027e-01
5.53763270e-01 -1.25069940e+00 9.01472926e-01 4.90845561e-01
-9.30151582e-01 -1.99956074e-01 -6.09765530e-01 1.89676523e-01
-1.71611160e-01 1.07966709e+00 -9.37206388e-01 1.75711721e-01
8.68343592e-01 8.81933272e-01 -4.79743659e-01 1.10472965e+00
-2.09430575e-01 6.55302882e-01 -4.45147902e-01 1.44105136e-01
4.63343114e-01 -2.57419586e-01 9.56431180e-02 1.20973861e+00
1.49470866e-01 9.48898941e-02 4.60718334e-01 6.85891747e-01
-5.74138105e-01 -1.82823434e-01 -6.09594107e-01 -2.64876753e-01
8.17011595e-01 1.32611859e+00 -1.19295073e+00 -5.45825124e-01
-7.89716989e-02 8.10060680e-01 5.71798086e-01 3.92821193e-01
-4.81075704e-01 -3.57578605e-01 7.68120348e-01 -2.03268126e-01
5.29221773e-01 -2.42137551e-01 -3.08859110e-01 -8.98089468e-01
-1.75377429e-01 -8.51390719e-01 5.50648153e-01 -3.71086210e-01
-1.38229156e+00 4.61651564e-01 -2.59312183e-01 -1.11811936e+00
-2.87473470e-01 -5.25315285e-01 -2.05118045e-01 4.45434868e-01
-1.18244588e+00 -6.63047493e-01 -1.67690068e-01 5.66743553e-01
3.23897600e-01 8.83031487e-02 6.93228185e-01 6.26168370e-01
-6.89919233e-01 3.85039479e-01 7.70898938e-01 5.52335918e-01
7.43001699e-01 -1.47518575e+00 -4.19379957e-02 9.21892405e-01
1.56414747e-01 5.38139164e-01 4.94096458e-01 -5.10955334e-01
-1.06797469e+00 -1.41266930e+00 6.19078577e-01 -5.55205822e-01
3.54003072e-01 -6.97491229e-01 -9.03228402e-01 5.41695178e-01
-3.78112458e-02 -3.00249662e-02 1.00022638e+00 6.56916276e-02
-2.55672455e-01 1.54937636e-02 -1.13617170e+00 4.24158186e-01
8.88354957e-01 -4.12946016e-01 -3.42115372e-01 3.85855168e-01
8.25708926e-01 1.95352361e-02 -9.21433449e-01 1.82732403e-01
2.58440971e-02 -8.07547867e-01 8.48084748e-01 -2.97542274e-01
3.49927425e-01 -4.78662699e-01 -2.17644885e-01 -1.35984468e+00
-4.22332972e-01 -2.64755696e-01 -7.86378458e-02 1.27344537e+00
4.02613878e-01 -2.26832196e-01 8.27023268e-01 5.64631462e-01
-1.71415165e-01 -6.10444486e-01 -7.38925636e-01 -7.12740004e-01
4.28184196e-02 -4.49965984e-01 4.64434117e-01 1.18771803e+00
-2.13739872e-01 4.26450461e-01 -1.99794024e-01 2.73154765e-01
8.72090340e-01 -3.58776040e-02 5.04669130e-01 -1.57406271e+00
6.93300515e-02 -4.97027874e-01 -3.82683545e-01 -1.06109226e+00
4.02842164e-01 -1.03672779e+00 5.81211567e-01 -1.63874590e+00
2.79245317e-01 -4.89744574e-01 -3.45634371e-01 5.16818702e-01
-1.15333587e-01 3.59098822e-01 2.01486588e-01 5.25757074e-01
-1.14726543e+00 3.96222889e-01 9.42871988e-01 -5.67162298e-02
-3.17299068e-01 7.37146214e-02 -8.57756317e-01 7.75944710e-01
9.44730878e-01 -9.41264808e-01 -3.96763414e-01 -4.53894138e-01
-1.32486880e-01 -5.65276206e-01 4.88996413e-03 -1.05384445e+00
5.92444420e-01 1.92073464e-01 6.35936439e-01 -7.34643221e-01
2.56419152e-01 -9.46573138e-01 -9.05653462e-02 3.88263196e-01
-4.68756199e-01 -2.58981943e-01 -2.19265893e-01 4.85121995e-01
-2.84119636e-01 -3.26406002e-01 1.05989957e+00 -2.42721841e-01
-4.91440684e-01 5.16665459e-01 -6.63517892e-01 5.54032512e-02
9.56053317e-01 -1.95834368e-01 -6.82279542e-02 -4.40694124e-01
-8.50416005e-01 4.90876853e-01 6.60930455e-01 9.67087150e-02
4.03511167e-01 -1.37927258e+00 -4.14413214e-01 2.05871407e-02
-7.78919756e-02 6.88219845e-01 -3.12539823e-02 6.56150520e-01
-2.10579455e-01 1.91534743e-01 1.74805000e-01 -9.18890238e-01
-1.00208211e+00 8.97170544e-01 2.24770263e-01 3.08616497e-02
-5.61078310e-01 8.60537648e-01 2.97988772e-01 -6.40198350e-01
6.09557569e-01 -7.76904374e-02 -1.06382728e-01 2.55534083e-01
5.31186223e-01 1.70991451e-01 9.34817642e-02 -6.01468742e-01
-3.05133790e-01 3.56092751e-01 -3.32631916e-01 -8.12661126e-02
1.60230494e+00 -3.50349545e-01 -3.68582577e-01 4.57574606e-01
1.63813198e+00 -6.91873193e-01 -1.28742111e+00 -4.60559517e-01
3.46775591e-01 -1.71607986e-01 7.79087245e-02 -4.31050777e-01
-1.27422976e+00 9.47827101e-01 5.59979320e-01 4.24344897e-01
1.22573972e+00 3.91921818e-01 6.56499207e-01 6.72511876e-01
-8.71144384e-02 -1.48840725e+00 1.19890869e-01 4.96128291e-01
1.94062307e-01 -1.35991764e+00 -2.32199445e-01 -3.03642731e-03
-5.96578836e-01 8.72051656e-01 5.75605690e-01 -1.38272136e-01
8.24300647e-01 2.48397499e-01 3.32387954e-01 -3.71123970e-01
-6.80007815e-01 -3.86333704e-01 6.12129569e-02 6.30063534e-01
1.31018564e-01 5.54529540e-02 1.07292399e-01 4.24146324e-01
-5.40688969e-02 -4.12982166e-01 4.84960616e-01 9.24114347e-01
-6.79189026e-01 -1.01335180e+00 -3.78117919e-01 8.40336502e-01
-3.63337249e-01 3.53329480e-02 -6.80229008e-01 4.46197391e-01
3.34208071e-01 1.13170028e+00 2.63648689e-01 -5.54773211e-01
-1.55837834e-01 1.68076843e-01 -5.66537166e-03 -8.80837202e-01
-2.48372555e-01 4.73318934e-01 -2.57094920e-01 -5.22255659e-01
-8.04117620e-01 -7.75113463e-01 -1.54634035e+00 -3.37295592e-01
-3.54335219e-01 4.07364905e-01 6.88755810e-01 1.00933933e+00
3.86000037e-01 3.82750005e-01 9.34329033e-01 -9.17622089e-01
-3.61900121e-01 -8.03771079e-01 -7.10637510e-01 5.72620928e-01
4.76677120e-01 -5.25785983e-01 -5.62196314e-01 2.05514863e-01] | [9.173064231872559, 3.2370235919952393] |
1c1e2373-3227-4e75-b00f-8be0649965a6 | epro-pnp-generalized-end-to-end-probabilistic | 2203.13254 | null | https://arxiv.org/abs/2203.13254v4 | https://arxiv.org/pdf/2203.13254v4.pdf | EPro-PnP: Generalized End-to-End Probabilistic Perspective-n-Points for Monocular Object Pose Estimation | Locating 3D objects from a single RGB image via Perspective-n-Points (PnP) is a long-standing problem in computer vision. Driven by end-to-end deep learning, recent studies suggest interpreting PnP as a differentiable layer, so that 2D-3D point correspondences can be partly learned by backpropagating the gradient w.r.t. object pose. Yet, learning the entire set of unrestricted 2D-3D points from scratch fails to converge with existing approaches, since the deterministic pose is inherently non-differentiable. In this paper, we propose the EPro-PnP, a probabilistic PnP layer for general end-to-end pose estimation, which outputs a distribution of pose on the SE(3) manifold, essentially bringing categorical Softmax to the continuous domain. The 2D-3D coordinates and corresponding weights are treated as intermediate variables learned by minimizing the KL divergence between the predicted and target pose distribution. The underlying principle unifies the existing approaches and resembles the attention mechanism. EPro-PnP significantly outperforms competitive baselines, closing the gap between PnP-based method and the task-specific leaders on the LineMOD 6DoF pose estimation and nuScenes 3D object detection benchmarks. | ['Hao Li', 'Lu Xiong', 'Wei Tian', 'Fan Wang', 'Pichao Wang', 'Hansheng Chen'] | 2022-03-24 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Chen_EPro-PnP_Generalized_End-to-End_Probabilistic_Perspective-N-Points_for_Monocular_Object_Pose_Estimation_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Chen_EPro-PnP_Generalized_End-to-End_Probabilistic_Perspective-N-Points_for_Monocular_Object_Pose_Estimation_CVPR_2022_paper.pdf | cvpr-2022-1 | ['6d-pose-estimation'] | ['computer-vision'] | [-1.39934570e-01 7.10282177e-02 -2.81154782e-01 -5.78837574e-01
-1.01182306e+00 -6.88860595e-01 5.89096308e-01 -2.52019376e-01
-6.23901188e-01 1.15185522e-01 7.58019909e-02 1.84225738e-01
1.64182764e-02 -5.32347739e-01 -1.31270933e+00 -6.19887829e-01
1.24418557e-01 7.35913992e-01 2.91281015e-01 7.88446739e-02
3.88790399e-01 8.15350831e-01 -1.30856156e+00 8.74096341e-03
2.10970759e-01 1.40586269e+00 1.73502758e-01 5.90539455e-01
-9.63996872e-02 2.15612024e-01 -4.33891058e-01 -3.10816824e-01
5.80637276e-01 1.64772406e-01 -4.85609651e-01 5.83567247e-02
1.10409236e+00 -4.85632479e-01 -4.58391756e-01 1.05931175e+00
5.76745868e-01 6.81525245e-02 8.33231390e-01 -1.38945842e+00
-1.05040359e+00 4.71140556e-02 -9.17810619e-01 -2.38063484e-01
4.18698698e-01 2.83940643e-01 1.19487274e+00 -1.25899422e+00
7.31859028e-01 1.61463630e+00 8.54018450e-01 4.94729280e-01
-1.13227689e+00 -2.69439638e-01 3.34616899e-01 -3.67860794e-02
-1.41478801e+00 1.63975865e-01 1.02617610e+00 -5.35942733e-01
1.05202329e+00 -1.25048449e-02 6.25563622e-01 1.37402606e+00
1.31600469e-01 1.19514358e+00 7.77094126e-01 -1.49454817e-01
-2.81852894e-02 -1.90388590e-01 -4.35608327e-02 7.64322579e-01
9.61018205e-02 3.11089665e-01 -6.50224268e-01 8.33304971e-02
1.04016757e+00 2.20211685e-01 -4.82917689e-02 -1.07695103e+00
-1.36499131e+00 5.53822935e-01 1.00821304e+00 -3.27665538e-01
-3.20704252e-01 4.98602152e-01 2.11160686e-02 -6.10577874e-02
3.86538982e-01 3.34210068e-01 -7.75438786e-01 1.77303683e-02
-4.85723227e-01 4.67388421e-01 4.30707514e-01 1.25049531e+00
6.17257774e-01 -4.01424378e-01 -2.49115199e-01 5.33153594e-01
7.78287232e-01 7.88676620e-01 9.90283117e-02 -1.05726862e+00
6.55457973e-01 6.63214862e-01 3.87804508e-01 -7.74481475e-01
-4.01806682e-01 -4.80895042e-01 -2.74942428e-01 5.67320943e-01
6.03107214e-01 -4.47718725e-02 -1.01657605e+00 1.74191809e+00
6.04461312e-01 -4.10621092e-02 -2.74706692e-01 1.32917786e+00
8.72475266e-01 6.10904872e-01 -1.48481146e-01 5.26302457e-01
1.08135533e+00 -1.14124215e+00 -1.36503860e-01 -4.43602771e-01
1.62456334e-01 -6.24658644e-01 1.30904138e+00 1.15915887e-01
-1.04859841e+00 -6.69533253e-01 -7.49299884e-01 -7.13513494e-01
-4.31050837e-01 3.25667977e-01 4.04125571e-01 1.98241532e-01
-9.37389016e-01 4.77787763e-01 -9.17159915e-01 -2.98105955e-01
8.42421651e-01 5.60400307e-01 -5.81631303e-01 9.88722872e-03
-5.67765951e-01 1.04744565e+00 3.31614703e-01 4.28331077e-01
-8.61573040e-01 -8.07148874e-01 -8.93567324e-01 -4.84881066e-02
3.92333150e-01 -1.02451015e+00 1.18406641e+00 -4.32671607e-01
-1.64767849e+00 1.32710242e+00 -5.81566729e-02 -1.35763094e-01
9.26095963e-01 -7.36807168e-01 4.54447657e-01 -1.54385105e-01
1.15035094e-01 1.26700342e+00 8.49085569e-01 -1.39467967e+00
-6.12944126e-01 -8.36511493e-01 1.42539859e-01 5.33593893e-01
5.17334901e-02 -3.33644420e-01 -7.24580646e-01 -3.47766370e-01
5.32128155e-01 -9.67347622e-01 -1.03740245e-01 8.10333192e-01
-5.99599659e-01 -7.36209750e-01 8.22035432e-01 -3.83860111e-01
2.57035434e-01 -2.09988880e+00 4.96249497e-01 -1.17080986e-01
4.36084978e-02 -1.45285413e-01 -3.22176844e-01 5.91455400e-02
1.02667533e-01 -1.84958160e-01 2.42338553e-02 -6.98344111e-01
4.77121294e-01 -2.16523767e-03 -2.95269698e-01 8.36996257e-01
4.81802106e-01 1.46678376e+00 -1.02224815e+00 -1.52704120e-01
4.23860729e-01 7.73981631e-01 -4.31744963e-01 2.95463771e-01
-5.56949496e-01 3.45746279e-01 -3.71507168e-01 7.94625938e-01
8.48016679e-01 -1.79832891e-01 -6.46916807e-01 -3.82806122e-01
-4.30473536e-02 2.16046035e-01 -9.56725955e-01 2.24884439e+00
-1.77167788e-01 6.06197953e-01 -2.42387608e-01 -6.24193132e-01
1.06748927e+00 -1.54063597e-01 5.28027058e-01 -3.53465706e-01
2.90969372e-01 1.37227148e-01 -3.40317160e-01 -3.60564113e-01
2.72788584e-01 5.52753620e-02 -1.91545799e-01 1.61868352e-02
3.53697658e-01 -4.21831191e-01 -3.93524140e-01 -3.54422688e-01
6.08345985e-01 8.14914346e-01 1.16503030e-01 8.15260634e-02
2.30141163e-01 -7.24183861e-03 2.81125247e-01 7.25045860e-01
-3.22704136e-01 1.01874495e+00 6.40289783e-01 -4.89048451e-01
-1.02328920e+00 -1.55693471e+00 -1.83643833e-01 8.53817761e-01
4.39168245e-01 9.05745998e-02 -4.75373924e-01 -1.10122085e+00
5.08004665e-01 3.51863980e-01 -7.82200217e-01 -7.40623176e-02
-4.88303781e-01 -1.92448467e-01 2.19438374e-01 7.34019756e-01
3.84120792e-01 -9.64825988e-01 -6.79178774e-01 -8.01169351e-02
1.42543852e-01 -1.09955513e+00 -6.24758065e-01 3.92325729e-01
-8.49041581e-01 -1.03210568e+00 -9.98066485e-01 -8.90577734e-01
6.88763678e-01 1.84878543e-01 1.05775821e+00 -6.57085657e-01
-1.24402799e-01 5.17695129e-01 -8.59813020e-02 -7.69280553e-01
2.63529152e-01 1.04541041e-01 2.37891320e-02 -1.38553768e-01
8.23314428e-01 -3.30141276e-01 -7.05241501e-01 2.16867045e-01
-2.89301962e-01 -6.95538372e-02 6.29000068e-01 6.00650191e-01
1.16840661e+00 -5.85996926e-01 1.36732802e-01 -3.90418351e-01
1.15217373e-01 -1.39767379e-01 -8.82492006e-01 2.17865497e-01
-1.16809458e-01 1.06640644e-01 1.09033987e-01 -5.75483859e-01
-6.23209298e-01 6.44243121e-01 -2.22410560e-01 -1.05778015e+00
-4.16961253e-01 6.36729179e-04 -4.01107639e-01 -1.95893675e-01
5.92299163e-01 1.56871080e-01 -1.43851042e-01 -6.05021954e-01
6.22234702e-01 3.99911314e-01 6.87274694e-01 -5.17619431e-01
9.89979684e-01 6.27456903e-01 1.31524131e-01 -4.17706013e-01
-1.32761490e+00 -6.46557868e-01 -1.01778674e+00 -1.01350620e-01
1.05892944e+00 -1.04594505e+00 -1.00558829e+00 5.19073725e-01
-1.54398715e+00 -3.95383030e-01 -6.01519823e-01 6.05883062e-01
-8.89655530e-01 4.10635024e-02 -2.84089714e-01 -6.57901168e-01
-2.70606756e-01 -1.06847095e+00 1.94102788e+00 1.34952784e-01
-1.16981328e-01 -6.27160788e-01 -1.95816290e-02 9.29929614e-02
-1.32161662e-01 5.88916242e-01 4.69897956e-01 -2.23574445e-01
-8.33420396e-01 -4.76363957e-01 -4.23379987e-01 3.39459866e-01
-1.63577363e-01 -8.68469775e-02 -1.07185280e+00 -2.28836894e-01
1.16673090e-01 -5.31639576e-01 7.74724424e-01 7.01037109e-01
1.29412794e+00 1.35685503e-01 -2.46510193e-01 1.04006052e+00
1.35910642e+00 -3.40872228e-01 2.36125499e-01 3.00406277e-01
9.64370608e-01 4.64038044e-01 6.05120599e-01 1.05724365e-01
5.13736546e-01 7.06494153e-01 9.86804485e-01 6.91340491e-02
-1.38410717e-01 -6.47523880e-01 2.79697061e-01 1.04508407e-01
6.12568520e-02 -6.18637837e-02 -7.57465303e-01 3.20467860e-01
-1.73939705e+00 -5.31986475e-01 -8.99205655e-02 2.13501072e+00
5.58239460e-01 3.79209012e-01 1.33918673e-01 -3.91203344e-01
5.66675127e-01 1.85225099e-01 -1.04232073e+00 1.86628863e-01
-8.76363441e-02 -1.65936053e-01 6.47187114e-01 5.02184987e-01
-1.26459455e+00 9.85740364e-01 5.52592707e+00 4.21916366e-01
-1.15822423e+00 -6.04519844e-02 3.50796282e-01 -4.09203529e-01
3.35562937e-02 -2.26746172e-01 -1.15563965e+00 3.26515764e-01
1.88396722e-01 4.03686106e-01 2.46181905e-01 1.21005964e+00
-1.02657042e-01 1.84855834e-01 -1.63471925e+00 1.36739099e+00
1.07227065e-01 -1.13757062e+00 -1.08808190e-01 5.45836389e-02
6.82840645e-01 7.29409516e-01 4.13109183e-01 3.00925791e-01
2.26568021e-02 -8.87652099e-01 1.22674036e+00 5.30987382e-01
7.45649219e-01 -4.59185570e-01 5.42008340e-01 5.19856393e-01
-8.61041844e-01 -9.23978686e-02 -5.88133812e-01 2.25676477e-01
1.18309371e-01 2.79736429e-01 -7.85319686e-01 1.03974558e-01
8.86835456e-01 9.07492936e-01 -4.51652527e-01 1.22486377e+00
-6.37311459e-01 -5.11772605e-03 -7.32443392e-01 -1.70909435e-01
4.55049604e-01 -6.35547191e-02 7.01932371e-01 9.96583641e-01
3.29376549e-01 -3.38295698e-01 1.16625488e-01 1.23470592e+00
-3.84862542e-01 -3.40063721e-01 -4.89982069e-01 2.69326717e-01
3.03694814e-01 1.13978457e+00 -6.45740271e-01 1.05154999e-01
-2.64868200e-01 1.05059671e+00 4.85339612e-01 3.35298747e-01
-8.36696446e-01 -2.64980495e-01 6.74873233e-01 5.44452816e-02
5.80035686e-01 -4.45620924e-01 -4.95485872e-01 -1.04504514e+00
4.73235428e-01 -3.85447174e-01 1.15488850e-01 -1.04294455e+00
-1.71517396e+00 3.75985086e-01 -7.39053413e-02 -1.29642069e+00
9.84846652e-02 -1.17901516e+00 -3.67658585e-01 1.01314843e+00
-1.69486248e+00 -1.35625947e+00 -3.38457197e-01 4.13651288e-01
5.55985034e-01 2.04811871e-01 5.28303027e-01 -8.36209580e-03
-1.00417174e-01 6.67182863e-01 -1.37082040e-01 2.80804843e-01
6.80056930e-01 -1.52986395e+00 6.63185954e-01 4.38650459e-01
4.33450729e-01 5.00975251e-01 4.23069268e-01 -3.27183157e-01
-1.54462731e+00 -1.01666486e+00 8.02334905e-01 -1.23013067e+00
4.71233040e-01 -7.34045208e-01 -6.51681244e-01 6.95526361e-01
-2.82579213e-01 6.38459027e-01 3.43075469e-02 1.25664830e-01
-5.73926806e-01 -5.98349907e-02 -1.21394801e+00 6.23311937e-01
1.43789768e+00 -6.24101043e-01 -7.04474866e-01 5.66220045e-01
8.91225398e-01 -1.01751673e+00 -8.01318288e-01 3.25446248e-01
5.62752306e-01 -6.54007673e-01 1.41854548e+00 -6.38082325e-01
3.91046733e-01 -4.42486852e-01 -3.54290158e-01 -1.07933950e+00
-1.89184114e-01 -4.11873281e-01 -4.37669188e-01 7.89433002e-01
2.14724287e-01 -2.49083266e-01 1.21684337e+00 4.44738865e-01
-1.24964304e-01 -1.07431948e+00 -1.12504494e+00 -8.11547577e-01
2.28980601e-01 -4.95207518e-01 5.15997350e-01 4.48027909e-01
-6.03182077e-01 4.78901207e-01 -7.78120458e-02 5.32862425e-01
8.60131562e-01 2.55507052e-01 9.74404573e-01 -1.19883382e+00
-4.18373168e-01 -6.26073420e-01 -6.02702260e-01 -1.75389111e+00
1.34272993e-01 -1.04775321e+00 3.48938167e-01 -1.54890001e+00
-3.65325436e-02 -2.35869274e-01 -1.55061424e-01 4.34704065e-01
-5.84064052e-02 2.17455700e-01 2.87439972e-01 1.66071609e-01
-6.30507350e-01 7.17803895e-01 1.46378660e+00 -1.38741270e-01
-1.30750149e-01 7.80135915e-02 -4.37085360e-01 8.54103565e-01
4.09881115e-01 -6.23766184e-01 -2.32437342e-01 -8.34117115e-01
1.46744922e-01 -2.07691580e-01 8.39826941e-01 -8.35851073e-01
2.27177873e-01 -9.19101387e-02 7.65136003e-01 -1.15857518e+00
6.97977960e-01 -1.03525305e+00 -3.38946581e-01 2.51700789e-01
-4.32699561e-01 -8.76204893e-02 -1.72548488e-01 7.20638514e-01
4.92289104e-03 -7.83306807e-02 7.39380956e-01 -1.86165363e-01
-7.11438119e-01 8.95909965e-01 5.08206725e-01 6.49227202e-02
1.01695383e+00 -4.55822885e-01 -1.12856224e-01 -1.07687553e-02
-6.75050259e-01 3.67092699e-01 5.18133461e-01 7.38260090e-01
7.78506160e-01 -1.41937256e+00 -6.49043143e-01 2.73694843e-01
2.35667542e-01 8.90219152e-01 2.85510663e-02 5.94233334e-01
-4.82313126e-01 3.04859191e-01 -2.02328876e-01 -1.26758635e+00
-8.64323318e-01 5.82659304e-01 5.14505088e-01 1.32771537e-01
-7.52698541e-01 1.44929731e+00 1.65542141e-01 -9.85877693e-01
7.50451982e-01 -5.93481123e-01 2.92348593e-01 -1.49076313e-01
1.17013961e-01 9.24128294e-02 -9.57861543e-02 -5.71954906e-01
-3.68071437e-01 9.77668107e-01 -2.39821747e-02 4.21350934e-02
1.65619040e+00 -1.16960024e-02 1.08375929e-01 5.97147584e-01
1.71030855e+00 -3.79547775e-01 -2.19126439e+00 -4.04385686e-01
-1.43811386e-02 -6.81309164e-01 -6.57391325e-02 -8.17466676e-01
-9.82017040e-01 1.12146640e+00 7.14547336e-01 -2.62536705e-01
5.48570216e-01 4.03981596e-01 6.28121078e-01 6.93749607e-01
3.81061852e-01 -8.47809911e-01 3.29587311e-01 6.64289773e-01
1.19418657e+00 -1.55217016e+00 -1.72836691e-01 -1.72427624e-01
-4.64973569e-01 1.10282910e+00 8.78171623e-01 -5.08159280e-01
5.87979734e-01 1.42109483e-01 5.61120100e-02 -2.64414281e-01
-2.54795671e-01 -8.96507204e-02 6.59962296e-01 7.95298636e-01
1.31098077e-01 -8.32090303e-02 2.72350967e-01 4.35849428e-01
-2.03311235e-01 -1.11735158e-01 -9.92408693e-02 7.76741624e-01
-3.26833755e-01 -9.02117372e-01 -4.46750551e-01 2.35502288e-01
-1.31134436e-01 3.25900555e-01 -5.77449441e-01 8.46245587e-01
3.71460587e-01 1.84468642e-01 1.95869178e-01 -1.79138482e-01
6.43418193e-01 -5.21872304e-02 7.88755059e-01 -6.87760353e-01
-2.40162238e-01 3.15064332e-03 -5.77113509e-01 -7.18370914e-01
-2.57460177e-01 -7.30606198e-01 -1.21629012e+00 1.25418350e-01
-3.14707130e-01 -4.63034898e-01 8.31005991e-01 6.60320401e-01
3.53039414e-01 1.98638648e-01 4.16997880e-01 -1.45192087e+00
-1.01098514e+00 -7.45077848e-01 -3.10517192e-01 4.95025367e-01
5.11694014e-01 -7.73064554e-01 -1.56862602e-01 -2.39897132e-01] | [7.600008964538574, -2.646291971206665] |
4b9f21db-f72b-49df-93fa-0be1adcb285e | autoregressive-neural-network-wavefunctions | 2109.12606 | null | https://arxiv.org/abs/2109.12606v2 | https://arxiv.org/pdf/2109.12606v2.pdf | Autoregressive neural-network wavefunctions for ab initio quantum chemistry | In recent years, neural network quantum states (NNQS) have emerged as powerful tools for the study of quantum many-body systems. Electronic structure calculations are one such canonical many-body problem that have attracted significant research efforts spanning multiple decades, whilst only recently being attempted with NNQS. However, the complex non-local interactions and high sample complexity are significant challenges that call for bespoke solutions. Here, we parameterise the electronic wavefunction with a novel autoregressive neural network (ARN) that permits highly efficient and scalable sampling, whilst also embedding physical priors reflecting the structure of molecular systems without sacrificing expressibility. This allows us to perform electronic structure calculations on molecules with up to 30 spin-orbitals -- at least an order of magnitude more Slater determinants than previous applications of conventional NNQS -- and we find that our ansatz can outperform the de-facto gold-standard coupled cluster methods even in the presence of strong quantum correlations. With a highly expressive neural network for which sampling is no longer a computational bottleneck, we conclude that the barriers to further scaling are not associated with the wavefunction ansatz itself, but rather are inherent to any variational Monte Carlo approach. | ['A. I. Lvovsky', 'Aleksei Malyshev', 'Thomas D. Barrett'] | 2021-09-26 | null | null | null | null | ['variational-monte-carlo'] | ['miscellaneous'] | [ 3.96761715e-01 -2.23236516e-01 3.13821831e-03 -2.19523653e-01
-9.62001860e-01 -1.97996438e-01 6.64352596e-01 1.07517101e-01
-6.48312092e-01 1.07184148e+00 1.53450817e-01 -2.82408834e-01
-2.76260197e-01 -7.68610716e-01 -6.98445916e-01 -1.25788569e+00
-6.87995255e-02 6.63579345e-01 -2.90528566e-01 -4.71144170e-01
3.54167700e-01 6.24904156e-01 -1.44309413e+00 -9.65278894e-02
7.14215517e-01 8.54340732e-01 -3.65153641e-01 4.71659988e-01
4.14447427e-01 6.51463091e-01 -4.16119881e-02 -2.94155985e-01
7.72646815e-02 -4.85619336e-01 -7.85437226e-01 -5.72558463e-01
3.76809478e-01 1.34672612e-01 -5.99975049e-01 1.14732945e+00
7.74741530e-01 7.76513577e-01 9.07918513e-01 -4.64813560e-01
-7.42732882e-01 3.71221304e-01 1.18383896e-02 -6.50420636e-02
-1.24838091e-01 5.20013213e-01 1.21948755e+00 -7.97341526e-01
4.26240563e-01 1.02609563e+00 9.62285161e-01 6.13008916e-01
-1.86261189e+00 -4.34817106e-01 -7.45897889e-01 2.06541002e-01
-1.60091317e+00 -7.50242352e-01 8.07308733e-01 -3.46950799e-01
1.55941200e+00 1.15369484e-01 6.36379719e-01 1.10996139e+00
6.25247121e-01 -7.98014700e-02 1.00800645e+00 -7.22525954e-01
5.89947641e-01 -9.13466662e-02 2.90004253e-01 5.24700224e-01
2.27228701e-01 5.82764387e-01 -4.53009039e-01 -6.20998442e-01
5.81760585e-01 -3.66006196e-02 -1.29268439e-02 -7.62419760e-01
-1.12071121e+00 1.20012796e+00 5.27953863e-01 1.32072106e-01
-8.20629597e-01 7.65554488e-01 6.32893085e-01 -6.71939403e-02
1.99455142e-01 9.26148772e-01 -2.31008112e-01 -2.75459856e-01
-9.77196693e-01 4.06227648e-01 7.90132701e-01 2.84571260e-01
9.19941187e-01 2.18815491e-01 1.74831584e-01 3.84711802e-01
1.99673101e-01 4.05880600e-01 -6.96296394e-02 -1.15287673e+00
-1.73144981e-01 -8.07498395e-02 2.32747257e-01 -6.59266591e-01
-4.22335953e-01 -4.15078640e-01 -1.39976740e+00 9.80460197e-02
1.93488210e-01 -3.52474272e-01 -3.59128475e-01 1.60934353e+00
3.34773362e-01 -2.44145229e-01 6.96952268e-02 4.87597972e-01
5.78279912e-01 8.88748765e-01 1.19383678e-01 -4.09504563e-01
1.10664129e+00 -5.22622108e-01 -5.77265501e-01 2.02705577e-01
6.45827413e-01 -4.28917021e-01 6.74007475e-01 5.13574600e-01
-1.21964180e+00 -3.08329135e-01 -1.17695022e+00 -2.00947359e-01
-3.46222699e-01 -3.84290218e-01 1.44898880e+00 8.64194512e-01
-1.08304226e+00 1.38034058e+00 -1.00183582e+00 -3.66546065e-02
2.13826209e-01 8.53579938e-01 -6.01777673e-01 1.33863896e-01
-1.24985015e+00 1.09321260e+00 6.24845028e-01 2.39827499e-01
-4.01087254e-01 -5.84096551e-01 -6.73985541e-01 7.95083120e-02
3.20645690e-01 -7.66264915e-01 1.12609887e+00 -4.93501037e-01
-1.87841451e+00 2.23587006e-01 -4.22866106e-01 -3.34540695e-01
-1.46238208e-01 4.02121067e-01 -3.58784586e-01 4.34918702e-02
-1.41793296e-01 5.20119667e-01 3.49592030e-01 -6.50317371e-01
4.69070375e-01 -2.34030008e-01 -1.91136196e-01 1.24960676e-01
-7.05846027e-02 9.52313766e-02 9.86336470e-02 9.92147103e-02
3.66841890e-02 -1.12074375e+00 -7.47476816e-01 -3.72665137e-01
-2.53767043e-01 -4.46200788e-01 -9.31503549e-02 -6.15000874e-02
1.08290315e+00 -1.70545363e+00 3.28159302e-01 3.99165660e-01
1.82885081e-01 2.25591540e-01 1.15322601e-02 9.52636898e-01
-3.94901514e-01 -5.73174208e-02 -4.01445657e-01 -1.42956555e-01
2.33309135e-01 2.06822067e-01 1.65933117e-01 6.55210614e-01
1.85885400e-01 1.01829839e+00 -7.04704702e-01 -1.49019182e-01
1.48623005e-01 9.91065025e-01 -9.23042119e-01 -2.63802022e-01
-1.73136517e-01 3.46206397e-01 -3.07954222e-01 1.66672394e-02
4.15004611e-01 -7.64528275e-01 1.85961932e-01 -3.62767726e-01
-3.49571079e-01 6.63129508e-01 -1.28032434e+00 1.56748950e+00
-1.88363701e-01 3.98515016e-01 -1.28536671e-02 -1.26694643e+00
5.51727653e-01 3.63056540e-01 5.91747701e-01 -6.75307095e-01
1.25699431e-01 2.56372780e-01 5.97532272e-01 -2.85419643e-01
6.77181959e-01 -7.87373662e-01 -2.25873943e-02 3.40567559e-01
1.67683974e-01 -3.48182589e-01 2.61940956e-01 -5.89407869e-02
8.77657354e-01 1.49534464e-01 4.89822030e-01 -5.79270661e-01
3.65459591e-01 -4.43799756e-02 3.74438167e-01 1.03185534e+00
-2.82094777e-01 3.93419206e-01 4.77918416e-01 -6.83050275e-01
-1.45393562e+00 -6.64993346e-01 -7.06676245e-01 8.69208872e-01
-3.62409383e-01 -6.18021905e-01 -6.66758418e-01 2.06784591e-01
-1.98643416e-01 6.13506734e-01 -5.30443728e-01 -2.89900154e-01
-3.74354661e-01 -1.52871597e+00 4.08452868e-01 1.77463070e-01
-3.26039913e-06 -1.29665470e+00 -2.84411162e-01 6.45540774e-01
3.90297562e-01 -7.38087416e-01 -2.70840060e-02 7.73078620e-01
-7.42959440e-01 -6.54042840e-01 -4.16122139e-01 -8.66992548e-02
2.05021441e-01 -2.66071856e-01 1.15636384e+00 -3.35989267e-01
-4.66438204e-01 4.31861654e-02 2.64189929e-01 -2.25111023e-02
-7.30558515e-01 1.17747620e-01 4.13211793e-01 -2.82795638e-01
6.68316960e-01 -8.21852386e-01 -6.91426039e-01 -3.94640386e-01
-7.34901130e-01 -1.23054907e-01 7.37707615e-01 1.12898493e+00
6.36197090e-01 1.09560817e-01 3.97248268e-01 -7.98933387e-01
7.26204038e-01 -2.86846250e-01 -6.83526874e-01 -6.10238500e-02
-5.85991383e-01 5.92997372e-01 8.09364259e-01 4.33519594e-02
-8.89813542e-01 -6.69675553e-03 -5.02733529e-01 1.65586129e-01
-2.60821760e-01 7.36759722e-01 1.90507755e-01 -5.06512582e-01
1.00936544e+00 2.59544432e-01 1.56200351e-02 -2.07678691e-01
3.45974177e-01 4.97574121e-01 2.45517194e-01 -9.09394860e-01
3.81027341e-01 3.15756053e-01 9.33468282e-01 -1.16762459e+00
-6.40602410e-01 -3.66964608e-01 -8.42994809e-01 3.35760772e-01
1.10063398e+00 -8.28932762e-01 -1.46392751e+00 5.48900738e-02
-1.24545014e+00 -1.10705920e-01 -1.60501167e-01 7.01597750e-01
-4.92241174e-01 5.81964135e-01 -9.45961237e-01 -1.03611243e+00
-4.85441446e-01 -1.49110985e+00 7.38092184e-01 -8.57406631e-02
-2.00465024e-01 -1.03806353e+00 3.92160416e-01 2.37912565e-01
7.42403686e-01 3.63710999e-01 1.11781085e+00 -3.47473711e-01
-6.17119491e-01 -5.00023186e-01 -1.86926931e-01 2.65515924e-01
-2.54658312e-01 7.27416128e-02 -9.73635733e-01 -4.13090825e-01
-6.61615133e-02 -5.98155916e-01 7.99090028e-01 6.41126692e-01
9.60059822e-01 -1.45040199e-01 -1.45296469e-01 5.45725226e-01
1.49166501e+00 9.10742208e-02 4.48651254e-01 -2.16324270e-01
7.77721763e-01 2.99306422e-01 -3.12260747e-01 4.10830975e-01
1.04945026e-01 6.10634089e-01 3.03535044e-01 1.75056085e-01
5.49158216e-01 4.35811877e-02 3.12118590e-01 1.19413137e+00
-6.62231982e-01 1.13995880e-01 -1.00016391e+00 -4.72073369e-02
-1.50846195e+00 -1.32095397e+00 -4.07935649e-01 2.33029747e+00
9.23110247e-01 1.68539491e-02 -6.04953878e-02 -1.95040897e-01
2.30357692e-01 3.22952539e-01 -7.82416940e-01 -6.07412040e-01
1.13061210e-02 9.28156734e-01 5.38826108e-01 5.65243661e-01
-9.86301780e-01 8.12166512e-01 7.14122772e+00 8.38823378e-01
-1.10069740e+00 1.71100244e-01 5.53420484e-01 -1.69580787e-01
-1.31034672e-01 3.02008390e-01 -8.39418948e-01 3.63736510e-01
1.55285430e+00 2.58092433e-01 9.04307902e-01 7.03337491e-01
2.67448068e-01 -2.57517304e-02 -1.09113944e+00 1.00224602e+00
-2.77397543e-01 -1.74761760e+00 -5.02301604e-02 4.62086350e-01
8.64107430e-01 6.39005899e-01 6.08195476e-02 2.80113339e-01
2.47375324e-01 -1.49572527e+00 2.55709141e-01 6.72226787e-01
1.00137579e+00 -9.85923171e-01 5.84869504e-01 3.52642596e-01
-6.54390752e-01 3.18430692e-01 -8.17993164e-01 -4.15054142e-01
5.78449517e-02 5.82218468e-01 -1.84552178e-01 2.84333110e-01
2.22720340e-01 5.84369361e-01 -9.05914158e-02 7.72435069e-01
2.67844349e-01 6.05187535e-01 -4.22892451e-01 -5.94211519e-01
5.42095006e-01 -8.14645886e-01 1.63396761e-01 8.13579798e-01
3.15691620e-01 1.56221256e-01 -1.76621705e-01 1.25272083e+00
-8.19274187e-02 -2.23901663e-02 -5.92558503e-01 -4.25716728e-01
2.94676960e-01 1.19541061e+00 -2.95007378e-01 -2.69419849e-01
-3.11063021e-01 6.51961029e-01 4.30138260e-01 3.70262116e-01
-4.95495737e-01 -3.67786974e-01 5.87329030e-01 -1.14399716e-01
3.27957422e-01 -5.12824833e-01 1.08443573e-01 -1.30567956e+00
-3.00360292e-01 -8.94261718e-01 -2.60485321e-01 -4.91731614e-01
-1.38046122e+00 3.87750208e-01 -1.71517193e-01 -3.83887261e-01
-4.15199459e-01 -1.09773529e+00 -5.09493828e-01 1.07851911e+00
-1.16099858e+00 -6.79484308e-01 5.07717729e-01 2.32372433e-01
-4.25247639e-01 3.69257256e-02 1.44585276e+00 1.61167115e-01
-8.00982535e-01 2.15622351e-01 1.03889298e+00 -2.60597587e-01
3.16637695e-01 -1.25184929e+00 4.02651250e-01 4.25711125e-01
1.75257593e-01 1.19617355e+00 8.82037342e-01 -2.93960452e-01
-1.84995437e+00 -5.16961277e-01 6.81563854e-01 -6.03295267e-01
7.68311262e-01 -5.47583580e-01 -8.88154924e-01 3.40077102e-01
2.68667370e-01 3.23327154e-01 1.00881541e+00 5.78621924e-01
-1.25986710e-01 1.60010159e-01 -8.29190016e-01 4.27596599e-01
7.79817522e-01 -9.87034261e-01 -2.89770097e-01 6.18411481e-01
3.03993851e-01 1.08576547e-02 -1.11672032e+00 1.40162066e-01
5.63898802e-01 -1.23611057e+00 1.15136385e+00 -7.85502970e-01
2.66542286e-01 -2.63877697e-02 -2.52657652e-01 -1.18444526e+00
-6.20518446e-01 -1.11741078e+00 1.33592291e-02 5.02686262e-01
3.27546328e-01 -6.23328805e-01 9.10940826e-01 1.10116041e+00
-1.67693093e-01 -6.57611012e-01 -1.18378150e+00 -4.37662780e-01
6.65910602e-01 -6.19119227e-01 5.04697919e-01 8.46748829e-01
1.23621121e-01 7.56126642e-01 -5.93294978e-01 -5.29913828e-02
8.20386291e-01 -1.70296326e-01 4.01791543e-01 -1.32142937e+00
-6.69068277e-01 -5.33104897e-01 -1.73686385e-01 -7.66805887e-01
3.39231670e-01 -1.05935860e+00 -1.01010583e-01 -9.53503847e-01
4.99285311e-01 -2.31631055e-01 -4.57075924e-01 1.58500761e-01
1.12619199e-01 2.44086474e-01 -3.07006449e-01 1.22651838e-01
-7.25569963e-01 9.73531067e-01 1.02219617e+00 1.51106372e-01
-4.57559852e-03 -2.43925810e-01 -5.83950579e-01 5.65084696e-01
4.56245184e-01 -5.24659991e-01 5.72557636e-02 1.18727706e-01
8.58739853e-01 2.69114107e-01 5.20933032e-01 -1.32883692e+00
3.76470685e-01 3.48718558e-03 3.76734972e-01 -3.67597133e-01
7.85967469e-01 -2.94185728e-01 3.16135317e-01 4.85163659e-01
-3.54139447e-01 -9.65053886e-02 6.18317127e-02 6.25531018e-01
-1.16662942e-01 -5.19408226e-01 9.31221604e-01 -4.11292106e-01
-1.43187538e-01 4.66382116e-01 -5.45285106e-01 -1.86479926e-01
3.89438719e-01 3.27737145e-02 1.56759888e-01 -1.70832217e-01
-5.91964126e-01 -4.22470748e-01 5.15686572e-01 -5.97534895e-01
-1.32050049e-02 -1.22187507e+00 -3.53643686e-01 1.63349912e-01
2.77900491e-02 -8.66811723e-02 5.70757508e-01 7.66454518e-01
-6.06756568e-01 8.49889815e-01 -5.05998209e-02 -4.59966481e-01
-6.40069366e-01 4.04089987e-01 5.71771920e-01 -3.02057236e-01
-3.95557135e-01 6.94775879e-01 -1.55654266e-01 -6.59878135e-01
-3.14619809e-01 -2.90429872e-02 2.48921305e-01 -1.98088154e-01
2.30394900e-01 2.95600265e-01 9.38745886e-02 -8.54540288e-01
-2.95002535e-02 3.85716945e-01 -4.89313342e-02 1.28919154e-01
1.67086577e+00 4.46946353e-01 -4.74624068e-01 3.86822253e-01
1.34610248e+00 -1.20133318e-01 -1.23522842e+00 -2.33974114e-01
-2.02836737e-01 3.00128043e-01 5.18482268e-01 -4.65616673e-01
-3.22719723e-01 1.38434279e+00 4.42593902e-01 2.54959643e-01
1.50984526e-01 -2.05122679e-01 6.41054571e-01 1.20502460e+00
4.39044744e-01 -1.00407171e+00 -3.93190593e-01 6.92758679e-01
3.73114288e-01 -1.51714444e+00 1.68259472e-01 2.01562107e-01
-1.54074073e-01 1.22768188e+00 -4.28756848e-02 -3.05449158e-01
4.98365521e-01 -2.24198222e-01 -6.72267199e-01 -4.62965995e-01
-7.25970924e-01 1.18368492e-01 3.86182755e-01 2.42077902e-01
7.71655500e-01 1.73922613e-01 1.14757381e-01 1.60510629e-01
-1.92511544e-01 -9.21884552e-02 5.77208757e-01 5.33160925e-01
-3.31570297e-01 -1.39342880e+00 -1.16428539e-01 5.81089675e-01
-4.57671076e-01 -5.91476142e-01 4.15752493e-02 7.87166059e-01
-7.38641769e-02 5.95480323e-01 -9.61804464e-02 1.42196685e-01
-8.45176727e-02 4.19362634e-01 6.20867908e-01 -6.10171020e-01
-4.71283615e-01 -1.18941829e-01 6.51691034e-02 -5.81366062e-01
-5.57093859e-01 -6.88589454e-01 -1.02212870e+00 -6.81920052e-01
-5.78564346e-01 6.25582993e-01 7.08555460e-01 9.63178635e-01
6.32818997e-01 2.37277612e-01 2.06697106e-01 -1.39186132e+00
-1.14255571e+00 -9.31817412e-01 -6.51868224e-01 1.76631108e-01
4.28398967e-01 -6.17314994e-01 -3.31542224e-01 -5.88732481e-01] | [5.4013352394104, 5.130925178527832] |
6b422503-4aad-4aa7-a817-2a07f04a1c49 | zero-episode-few-shot-contrastive-predictive | 2205.01924 | null | https://arxiv.org/abs/2205.01924v1 | https://arxiv.org/pdf/2205.01924v1.pdf | Zero-Episode Few-Shot Contrastive Predictive Coding: Solving intelligence tests without prior training | Video prediction models often combine three components: an encoder from pixel space to a small latent space, a latent space prediction model, and a generative model back to pixel space. However, the large and unpredictable pixel space makes training such models difficult, requiring many training examples. We argue that finding a predictive latent variable and using it to evaluate the consistency of a future image enables data-efficient predictions because it precludes the necessity of a generative model training. To demonstrate it, we created sequence completion intelligence tests in which the task is to identify a predictably changing feature in a sequence of images and use this prediction to select the subsequent image. We show that a one-dimensional Markov Contrastive Predictive Coding (M-CPC_1D) model solves these tests efficiently, with only five examples. Finally, we demonstrate the usefulness of M-CPC_1D in solving two tasks without prior training: anomaly detection and stochastic movement video prediction. | ['Y. Loewenstein', 'T. Barak'] | 2022-05-04 | null | null | null | null | ['video-prediction'] | ['computer-vision'] | [ 8.23894322e-01 5.55359647e-02 -2.50379920e-01 -4.16359603e-01
-5.90681553e-01 -2.99788475e-01 6.61574900e-01 -4.94709402e-01
2.38151308e-02 6.38859451e-01 1.65316880e-01 -2.58860469e-01
-2.03439761e-02 -4.86932814e-01 -9.49204445e-01 -7.47054636e-01
-2.40581214e-01 3.85375530e-01 2.71159351e-01 5.48266172e-01
4.29052800e-01 2.04586715e-01 -1.71191919e+00 4.90162462e-01
7.07077920e-01 1.01527238e+00 5.05134881e-01 1.02051842e+00
2.80308515e-01 1.16787672e+00 -1.57012552e-01 -1.49309888e-01
1.91847533e-01 -8.14768434e-01 -6.86280370e-01 5.94149709e-01
2.97414571e-01 -4.61989820e-01 -6.27897382e-02 9.54486251e-01
-1.99659973e-01 2.69069284e-01 9.32936966e-01 -1.57959783e+00
-7.17382550e-01 1.55454932e-03 -2.63076693e-01 1.34197861e-01
6.00883961e-01 3.90147626e-01 9.31905627e-01 -6.78042293e-01
8.52136850e-01 1.00443852e+00 6.95351958e-01 6.84915662e-01
-1.51308179e+00 -3.51233780e-01 2.55581677e-01 4.02908564e-01
-1.16815460e+00 -4.66896087e-01 5.42654276e-01 -8.59670997e-01
1.20387352e+00 2.53728688e-01 1.01329350e+00 1.29123724e+00
4.56119776e-01 8.69816005e-01 8.00813615e-01 -3.18335027e-01
4.27247733e-01 -2.25098655e-01 -2.52486229e-01 9.62541521e-01
-1.45833030e-01 2.77593613e-01 -7.47779608e-01 -1.62392393e-01
8.28889370e-01 2.13771939e-01 -1.88112766e-01 -6.01400375e-01
-9.50219810e-01 6.55067384e-01 -2.20791593e-01 -7.11521134e-02
-5.57228029e-01 4.23006237e-01 -8.47359374e-02 4.14238781e-01
4.81041342e-01 3.08977008e-01 -4.03511822e-01 -6.31950200e-01
-1.16768909e+00 1.81456015e-01 6.46100700e-01 9.28247869e-01
6.75349355e-01 8.55533183e-02 8.08863267e-02 3.06520641e-01
5.03238261e-01 3.17952335e-01 7.25656748e-01 -1.60549128e+00
2.86212176e-01 1.55673847e-01 1.45895228e-01 -9.74741101e-01
8.42064153e-03 1.36776222e-02 -4.09582585e-01 4.68947589e-01
1.36990011e-01 -9.00509059e-02 -9.23558712e-01 1.91855657e+00
-1.96477875e-01 5.64732790e-01 6.80796951e-02 6.41223967e-01
-2.45468035e-01 8.86917889e-01 1.08902231e-01 -7.42699385e-01
5.76508105e-01 -9.77011263e-01 -3.74240875e-01 -3.48333269e-01
7.69129515e-01 -3.91827464e-01 9.26840067e-01 4.97135371e-01
-1.22413158e+00 -5.89914203e-01 -1.02809274e+00 1.06882900e-01
1.66816428e-01 -1.22163028e-01 7.10169435e-01 3.49208504e-01
-1.24942899e+00 8.97610784e-01 -1.25193405e+00 -3.17272961e-01
3.31754088e-01 3.35649192e-01 -3.31836969e-01 3.54299415e-03
-6.94740117e-01 6.87149525e-01 2.28959277e-01 -3.84610116e-01
-1.12222457e+00 -2.72533864e-01 -8.35053682e-01 9.24046189e-02
8.20652172e-02 -6.95338428e-01 1.17728436e+00 -1.55852008e+00
-1.34789300e+00 7.58775234e-01 -5.97035706e-01 -7.29755282e-01
6.20864093e-01 -3.80360782e-01 -4.43248630e-01 1.67401090e-01
2.68596202e-01 9.24482465e-01 1.33506000e+00 -8.45762253e-01
-7.76187479e-01 -1.78741530e-01 -4.62543756e-01 3.84669602e-01
1.33707732e-01 -2.78432876e-01 -5.24188101e-01 -6.16203606e-01
4.10042405e-01 -1.11525249e+00 -2.05577612e-01 2.00945679e-02
-1.05033651e-01 2.20871600e-03 7.75892079e-01 -7.23214388e-01
1.01338100e+00 -2.26675034e+00 2.96438307e-01 3.18561196e-01
1.22089960e-01 -2.92772472e-01 -1.89523399e-01 -4.89289723e-02
-3.10923934e-01 3.67404670e-02 -1.26682624e-01 -3.27605307e-01
-1.90560594e-01 2.65725136e-01 -5.14969230e-01 4.25112754e-01
2.38397941e-01 8.59144270e-01 -7.82207429e-01 -5.11337638e-01
1.76585352e-04 1.97313815e-01 -9.16807294e-01 6.62160069e-02
-6.01082385e-01 3.68716240e-01 -1.99269071e-01 4.63988543e-01
1.12155594e-01 -6.16624236e-01 2.90970743e-01 3.25840384e-01
6.15055598e-02 1.70101434e-01 -7.96196878e-01 1.55549097e+00
1.34820104e-01 1.02197635e+00 -7.39790261e-01 -9.22125995e-01
7.62859762e-01 2.98935205e-01 7.75734484e-01 -6.27650619e-01
-4.22737271e-01 -6.64517209e-02 -1.37932241e-01 -6.61209047e-01
4.52729762e-01 -1.52129531e-01 2.58494169e-01 6.58451319e-01
6.57821894e-02 1.94466293e-01 1.99119747e-01 2.04028443e-01
1.36981261e+00 7.46019244e-01 7.38525018e-02 3.05382669e-01
1.87741578e-01 6.87226057e-02 9.24797773e-01 8.97079051e-01
-2.43127763e-01 6.37514353e-01 7.42280960e-01 -5.40698230e-01
-1.27269578e+00 -1.18309176e+00 1.84592918e-01 9.27299798e-01
2.97370907e-02 -4.76141512e-01 -6.42398357e-01 -5.71032047e-01
-2.12188095e-01 1.04212868e+00 -5.67447722e-01 -2.05751881e-01
-4.02845442e-01 -3.54436785e-01 -4.81027691e-03 6.28639519e-01
1.99844807e-01 -1.05004084e+00 -8.18069875e-01 1.36919186e-01
-3.12653631e-01 -7.78684378e-01 -3.48129034e-01 2.45293573e-01
-1.14013386e+00 -9.56456721e-01 -3.88419747e-01 -7.18889892e-01
8.50928545e-01 -5.79878176e-03 1.18219066e+00 1.17420601e-02
-1.25490934e-01 7.76545942e-01 -1.96844518e-01 1.95213575e-02
-5.50200462e-01 -4.73829299e-01 2.10978419e-01 -1.96860820e-01
6.72328830e-01 -6.76490545e-01 -4.13350165e-01 2.65392959e-01
-4.34848428e-01 6.68412805e-01 5.82708597e-01 8.98720205e-01
9.88279223e-01 8.93917456e-02 4.40938734e-02 -7.05608904e-01
3.06004345e-01 -5.30139923e-01 -7.14218616e-01 3.64929229e-01
-7.99743652e-01 1.50290072e-01 1.81148887e-01 -6.37929738e-01
-8.88030946e-01 6.75226867e-01 1.81171432e-01 -6.81427538e-01
-1.01906337e-01 2.89450347e-01 4.43057828e-02 3.69581938e-01
4.82191741e-01 7.02438831e-01 1.07135475e-01 -1.18196055e-01
1.49876580e-01 -2.40416322e-02 6.66265965e-01 -1.63203716e-01
5.02113461e-01 1.52532876e-01 9.64310393e-03 -5.93666077e-01
-6.08925045e-01 -1.32980913e-01 -8.13698053e-01 -2.27963895e-01
1.22100258e+00 -8.61511528e-01 -4.47051853e-01 6.17251061e-02
-9.49205399e-01 -5.64227760e-01 -5.49425364e-01 6.39772296e-01
-1.44495893e+00 3.41664940e-01 -5.07042587e-01 -7.97386467e-01
2.04688028e-01 -1.21907723e+00 7.49240756e-01 -1.80766776e-01
-6.75016403e-01 -8.67621958e-01 1.27359256e-01 1.68180406e-01
-3.69404107e-02 -9.77914110e-02 9.92814422e-01 -4.34106857e-01
-1.01362717e+00 -1.32510111e-01 1.96313441e-01 1.96463659e-01
-2.19509657e-02 6.12026155e-02 -6.71949625e-01 2.05396354e-04
3.04566085e-01 -2.35275507e-01 8.26233685e-01 6.64544940e-01
1.48375309e+00 -3.45834702e-01 -4.67110395e-01 8.11170697e-01
1.14886105e+00 5.78245997e-01 7.92528331e-01 2.41930827e-01
3.68629217e-01 1.15086980e-01 5.74126303e-01 4.51422870e-01
1.15730509e-01 6.08542740e-01 1.74454361e-01 3.41987699e-01
1.04194470e-01 -5.16810954e-01 6.54785991e-01 5.42646706e-01
-2.71135062e-01 -2.89675891e-01 -7.93285489e-01 3.64085793e-01
-2.06762695e+00 -1.56779993e+00 1.65201619e-01 1.96562338e+00
5.82782090e-01 2.90487468e-01 -1.09687001e-01 -5.59761422e-03
5.14594078e-01 -1.21596761e-01 -8.21487188e-01 -1.72265589e-01
9.82292965e-02 -1.81339383e-01 3.29014570e-01 5.57500482e-01
-1.16948271e+00 8.87468040e-01 7.70556259e+00 4.47841257e-01
-8.18123817e-01 -4.58273031e-02 1.08365846e+00 -2.06849828e-01
-4.38284695e-01 2.28760093e-01 -5.23807824e-01 6.64929926e-01
1.04257727e+00 2.89132670e-02 4.77080375e-01 1.09015393e+00
1.79962382e-01 -3.44072372e-01 -1.51318502e+00 1.19637263e+00
2.52326936e-01 -1.42615116e+00 1.70969993e-01 1.81323498e-01
8.65478694e-01 -1.22992627e-01 3.86699319e-01 9.05090496e-02
2.98776925e-01 -1.15439308e+00 7.03103721e-01 1.01404309e+00
6.57448053e-01 -4.39223349e-01 1.50695130e-01 5.09994090e-01
-7.13101685e-01 -2.44692191e-01 -4.09043044e-01 -2.22188011e-01
1.78105086e-01 3.00101228e-02 -7.56806850e-01 -3.89354020e-01
4.78741527e-01 8.60532522e-01 -4.26288724e-01 9.92351174e-01
-3.44777316e-01 7.53932536e-01 -6.64827460e-03 2.62266427e-01
9.41165313e-02 -3.15275192e-01 6.01159036e-01 7.89465666e-01
7.67443836e-01 1.24466926e-01 3.52730863e-02 8.98777843e-01
2.37696797e-01 -3.79952848e-01 -7.12294579e-01 -7.44413361e-02
1.50221437e-01 5.18612146e-01 -7.13410079e-01 -4.59455788e-01
-4.25222993e-01 1.60366726e+00 1.40679091e-01 5.37568748e-01
-8.51636410e-01 4.09104049e-01 6.86098218e-01 1.11698039e-01
5.26442111e-01 -4.23088193e-01 -3.37625325e-01 -1.30353761e+00
-1.00931071e-01 -7.28929520e-01 2.86169350e-01 -1.18132985e+00
-9.39464211e-01 3.31600159e-01 -2.33925864e-01 -1.44543648e+00
-1.10734713e+00 -3.38273942e-01 -4.29044336e-01 7.81008840e-01
-9.54219222e-01 -7.63757586e-01 -9.77728814e-02 6.32670462e-01
8.50449085e-01 -5.16891420e-01 9.82918680e-01 -2.77629316e-01
-3.02314609e-01 3.11307520e-01 1.39639869e-01 -6.12259805e-02
3.14726830e-01 -1.22498357e+00 5.17426074e-01 1.14569283e+00
3.35450143e-01 3.65232259e-01 7.74643898e-01 -9.83270168e-01
-1.29136097e+00 -8.78372669e-01 9.54931498e-01 -6.23703897e-01
5.35302639e-01 -8.56949985e-02 -7.81474769e-01 1.19039929e+00
-2.70183086e-01 -1.22107513e-01 8.24984848e-01 4.62806644e-03
-8.91931802e-02 3.34685594e-01 -8.29700589e-01 6.80975795e-01
1.23280692e+00 -6.56221151e-01 -4.81637567e-01 3.51152152e-01
6.23364270e-01 -1.22166231e-01 -5.23918211e-01 -4.78210039e-02
6.99295819e-01 -1.04113638e+00 7.91761398e-01 -7.25940287e-01
8.16966653e-01 -1.05765276e-01 -4.01874751e-01 -9.37694192e-01
-6.84797585e-01 -6.85169160e-01 -6.05534673e-01 6.74753904e-01
4.77340937e-01 -1.90761402e-01 1.24988675e+00 1.27038050e+00
3.59470099e-02 -5.10908365e-01 -7.32107997e-01 -7.00768173e-01
-3.76785934e-01 -9.23607409e-01 2.40693226e-01 7.61709273e-01
6.93246052e-02 1.80014998e-01 -7.27863491e-01 1.31955341e-01
5.05756795e-01 2.65660733e-01 5.62073708e-01 -1.13160884e+00
-6.83371186e-01 -2.07854301e-01 -8.86104524e-01 -1.58618319e+00
1.40290558e-01 -5.26693940e-01 1.98073253e-01 -1.17833090e+00
3.48989397e-01 -3.79233271e-01 -1.08747967e-01 3.93007427e-01
-8.11441839e-02 5.43754473e-02 9.99839753e-02 6.43371820e-01
-9.58073258e-01 3.99902910e-01 8.56029689e-01 -5.48035139e-03
-3.09954882e-01 1.49446249e-01 -3.70178312e-01 9.57087219e-01
5.55353999e-01 -4.43765014e-01 -7.89085925e-01 -3.81460965e-01
3.26700836e-01 5.32471359e-01 4.12566990e-01 -1.29992127e+00
2.73637146e-01 -4.27313238e-01 1.09510934e+00 -7.03638434e-01
6.19322538e-01 -7.82215774e-01 3.53003263e-01 4.89534557e-01
-5.53722799e-01 2.30810165e-01 -2.16718808e-01 7.97504127e-01
-1.43215075e-01 -2.58089453e-01 5.18670797e-01 -1.39912516e-01
-1.43315172e+00 3.69140983e-01 -6.77109003e-01 -3.34622949e-01
1.30823863e+00 -7.25572348e-01 1.09926246e-01 -7.76207447e-01
-1.31356728e+00 -1.53588213e-03 7.11137712e-01 2.74652421e-01
1.01176417e+00 -1.32636642e+00 -2.59688795e-01 7.10276246e-01
-1.26609296e-01 -5.34601092e-01 2.38951638e-01 6.67759717e-01
-4.99690235e-01 3.30209404e-01 -3.81308377e-01 -8.36543977e-01
-1.34912932e+00 6.93861842e-01 2.67144769e-01 -4.58893627e-02
-9.26357210e-01 8.63952637e-01 3.14371079e-01 2.79190928e-01
2.86742151e-01 -8.87261629e-02 9.92701650e-02 -3.59085470e-01
5.50010085e-01 1.17208458e-01 -6.12906516e-01 -5.36225319e-01
-5.14043309e-02 2.28653803e-01 5.45713715e-02 -4.78494436e-01
1.40625358e+00 -3.38066012e-01 1.21321686e-01 6.16980076e-01
9.78031516e-01 -5.10625422e-01 -2.05192828e+00 7.55648362e-03
1.64608628e-01 -6.49484932e-01 -1.37804970e-01 -5.33380151e-01
-5.90030313e-01 7.03106046e-01 8.21310103e-01 6.09671660e-02
1.27498591e+00 -1.02326214e-01 4.42458302e-01 5.03040969e-01
2.10234165e-01 -1.11257517e+00 3.33482146e-01 5.53993225e-01
5.44408917e-01 -1.17439866e+00 2.25865319e-02 -1.21768147e-01
-9.69544768e-01 1.08825862e+00 5.89433908e-01 3.32506448e-02
6.54532492e-01 5.17913066e-02 -3.13906491e-01 -1.84317082e-01
-1.33724642e+00 7.90595040e-02 2.89529532e-01 7.43628085e-01
2.73401827e-01 2.71658245e-02 3.54933083e-01 1.03882939e-01
-2.01281935e-01 3.34680021e-01 2.55497038e-01 8.37655306e-01
-5.88973641e-01 -9.05856013e-01 -1.21899374e-01 7.35961735e-01
-1.76686540e-01 -1.31738558e-01 1.21762902e-02 2.75702178e-01
2.42036194e-01 5.12159467e-01 6.18298233e-01 -5.67407012e-01
-3.83581460e-01 5.47056913e-01 5.10575116e-01 -3.59971821e-01
2.81293869e-01 7.92686343e-02 -1.91886112e-01 -8.95018339e-01
-4.60260421e-01 -1.14750636e+00 -1.03069735e+00 -2.56917864e-01
6.75389245e-02 -2.64332853e-02 5.92997730e-01 1.08555412e+00
3.54111910e-01 2.52579570e-01 3.80265236e-01 -6.54626250e-01
-3.11047643e-01 -5.82328975e-01 -4.73560601e-01 4.89391685e-01
2.31667206e-01 -4.43864077e-01 -8.53722245e-02 7.72138178e-01] | [8.502057075500488, 0.3568885922431946] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.